pax_global_header00006660000000000000000000000064126753063400014520gustar00rootroot0000000000000052 comment=dab9465c40b0a8fb7a59b461e516f27ff074274b rocksdb-4.5.1/000077500000000000000000000000001267530634000131565ustar00rootroot00000000000000rocksdb-4.5.1/.arcconfig000066400000000000000000000010641267530634000151130ustar00rootroot00000000000000{ "project_id" : "rocksdb", "conduit_uri" : "https://reviews.facebook.net/", "copyright_holder" : "Facebook", "load" : [ "arcanist_util" ], "lint.engine" : "FacebookFbcodeLintEngine", "lint.engine.single.linter" : "FbcodeCppLinter", "unit.engine" : "FacebookFbcodeUnitTestEngine", "arcanist_configuration" : "FacebookArcanistConfiguration", "base" : "git:HEAD^, hg:.^", "git.default-relative-commit" : "HEAD^", "git:arc.feature.start.default" : "origin/master", "arc.feature.start.default" : "master", "history.immutable" : false } rocksdb-4.5.1/.clang-format000066400000000000000000000002121267530634000155240ustar00rootroot00000000000000# Complete list of style options can be found at: # http://clang.llvm.org/docs/ClangFormatStyleOptions.html --- BasedOnStyle: Google ... rocksdb-4.5.1/.gitignore000066400000000000000000000011561267530634000151510ustar00rootroot00000000000000TARGETS make_config.mk *.a *.arc *.d *.dylib* *.gcda *.gcno *.o *.so *.so.* *_test *_bench *_stress *.out *.class *.jar *.*jnilib* *.d-e *.o-* *.swp *~ *.vcxproj *.vcxproj.filters *.sln *.cmake CMakeCache.txt CMakeFiles/ build/ ldb manifest_dump sst_dump util/build_version.cc build_tools/VALGRIND_LOGS/ coverage/COVERAGE_REPORT .gdbhistory package/ .phutil_module_cache unity.a tags rocksdb_dump rocksdb_undump java/out java/target java/test-libs java/*.log java/include/org_rocksdb_*.h .idea/ *.iml rocksdb.cc rocksdb.h unity.cc java/crossbuild/.vagrant .vagrant/ java/**.asc java/javadoc scan_build_report/ t LOG rocksdb-4.5.1/.travis.yml000066400000000000000000000036161267530634000152750ustar00rootroot00000000000000sudo: false language: cpp matrix: include: - os: linux compiler: clang env: COMPILER=clang++-3.6 addons: apt: sources: ['ubuntu-toolchain-r-test', 'llvm-toolchain-precise-3.6'] packages: ['clang-3.6', 'clang-format-3.6', 'zlib1g-dev', 'libbz2-dev', 'libsnappy-dev', 'curl'] - os: osx compiler: clang install: # Build gflags # TODO(noetzli): Remove when gflags available through Travis - pushd /tmp/ && curl -L https://github.com/gflags/gflags/archive/v2.1.2.tar.gz -o gflags.tar.gz && tar xfz gflags.tar.gz && cd gflags-2.1.2 && cmake . && make && popd # Download clang-format-diff.py to check source code formatting - pushd /tmp/ && curl -L http://llvm.org/svn/llvm-project/cfe/trunk/tools/clang-format/clang-format-diff.py -o clang-format-diff.py && chmod +x clang-format-diff.py && popd before_script: # Add gflags to include/library paths # TODO(noetzli): Remove when gflags available through Travis - export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/tmp/gflags-2.1.2/lib" - export LIBRARY_PATH="$LIBRARY_PATH:/tmp/gflags-2.1.2/lib" - export CPLUS_INCLUDE_PATH="$CPLUS_INCLUDE_PATH:/tmp/gflags-2.1.2/include" - if [ -n "${COMPILER}" ]; then CXX=${COMPILER}; fi - if [[ "${TRAVIS_OS_NAME}" == 'osx' ]]; then brew install gflags snappy; fi - ulimit -n 2000 || true # Lousy hack to disable use and testing of fallocate, which doesn't behave quite # as EnvPosixTest::AllocateTest expects within the Travis OpenVZ environment. script: - if [[ "${TRAVIS_OS_NAME}" == 'linux' ]]; then OPT=-DTRAVIS CLANG_FORMAT_DIFF=/tmp/clang-format-diff.py make format || true; fi - OPT=-DTRAVIS V=1 make -j4 check && OPT=-DTRAVIS V=1 make clean jclean rocksdbjava jtest && make clean && OPT="-DTRAVIS -DROCKSDB_LITE" V=1 make -j4 static_lib notifications: email: - leveldb@fb.com webhooks: - https://buildtimetrend.herokuapp.com/travis rocksdb-4.5.1/AUTHORS000066400000000000000000000003671267530634000142340ustar00rootroot00000000000000Facebook Inc. Facebook Engineering Team Google Inc. # Initial version authors: Jeffrey Dean Sanjay Ghemawat # Partial list of contributors: Kevin Regan Johan Bilien rocksdb-4.5.1/CMakeLists.txt000066400000000000000000000412761267530634000157300ustar00rootroot00000000000000# This cmake build is for Windows 64-bit only. # # Prerequisites: # You must have Visual Studio 2013 Update 4 installed. Start the Developer Command Prompt window that is a part of Visual Studio installation. # Run the build commands from within the Developer Command Prompt window to have paths to the compiler and runtime libraries set. # You must have git.exe in your %PATH% environment variable. # # To build Rocksdb for Windows is as easy as 1-2-3-4-5: # # 1. Update paths to third-party libraries in thirdparty.inc file # 2. Create a new directory for build artifacts # mkdir build # cd build # 3. Run cmake to generate project files for Windows, add more options to enable required third-party libraries. # See thirdparty.inc for more information. # sample command: cmake -G "Visual Studio 12 Win64" -DGFLAGS=1 -DSNAPPY=1 -DJEMALLOC=1 -DJNI=1 .. # OR for VS Studio 15 cmake -G "Visual Studio 14 Win64" -DGFLAGS=1 -DSNAPPY=1 -DJEMALLOC=1 -DJNI=1 .. # 4. Then build the project in debug mode (you may want to add /m[:] flag to run msbuild in parallel threads # or simply /m ot use all avail cores) # msbuild rocksdb.sln # # rocksdb.sln build features exclusions of test only code in Release. If you build ALL_BUILD then everything # will be attempted but test only code does not build in Release mode. # # 5. And release mode (/m[:] is also supported) # msbuild rocksdb.sln /p:Configuration=Release # cmake_minimum_required(VERSION 2.6) project(rocksdb) include(${CMAKE_CURRENT_SOURCE_DIR}/thirdparty.inc) execute_process(COMMAND powershell -Command "Get-Date -format MM_dd_yyyy" OUTPUT_VARIABLE DATE) execute_process(COMMAND powershell -Command "Get-Date -format HH:mm:ss" OUTPUT_VARIABLE TIME) string(REGEX REPLACE "(..)_(..)_..(..).*" "\\1/\\2/\\3" DATE ${DATE}) string(REGEX REPLACE "(..):(.....).*" " \\1:\\2" TIME ${TIME}) string(CONCAT GIT_DATE_TIME ${DATE} ${TIME}) find_package(Git) if (GIT_FOUND AND EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/.git") execute_process(COMMAND $ENV{COMSPEC} /C ${GIT_EXECUTABLE} -C ${CMAKE_CURRENT_SOURCE_DIR} rev-parse HEAD OUTPUT_VARIABLE GIT_SHA) else() set(GIT_SHA 0) endif() string(REGEX REPLACE "[^0-9a-f]+" "" GIT_SHA ${GIT_SHA}) set(BUILD_VERSION_CC ${CMAKE_CURRENT_SOURCE_DIR}/util/build_version.cc) add_custom_command(OUTPUT ${BUILD_VERSION_CC} COMMAND echo "#include \"build_version.h\"" > ${BUILD_VERSION_CC} COMMAND echo "const char* rocksdb_build_git_sha = \"rocksdb_build_git_sha:${GIT_SHA}\";" >> ${BUILD_VERSION_CC} COMMAND echo "const char* rocksdb_build_git_datetime = \"rocksdb_build_git_datetime:${GIT_DATE_TIME}\";" >> ${BUILD_VERSION_CC} COMMAND echo const char* rocksdb_build_compile_date = __DATE__\; >> ${BUILD_VERSION_CC} ) add_custom_target(GenerateBuildVersion DEPENDS ${BUILD_VERSION_CC}) set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue") set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W3 /WX /wd4127 /wd4800 /wd4996 /wd4351") # Used to run CI build and tests so we can run faster set(OPTIMIZE_DEBUG_DEFAULT 0) # Debug build is unoptimized by default use -DOPTDBG=1 to optimize if(DEFINED OPTDBG) set(OPTIMIZE_DEBUG ${OPTDBG}) else() set(OPTIMIZE_DEBUG ${OPTIMIZE_DEBUG_DEFAULT}) endif() if((${OPTIMIZE_DEBUG} EQUAL 1)) message(STATUS "Debug optimization is enabled") set(CMAKE_CXX_FLAGS_DEBUG "/Oxt /MDd") else() set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /Od /RTC1 /Gm /MDd") endif() set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Oxt /Zp8 /Gm- /Gy /MD") set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} /DEBUG") set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} /DEBUG") add_definitions(-DWIN32 -DOS_WIN -D_MBCS -DWIN64) include_directories(${PROJECT_SOURCE_DIR}) include_directories(${PROJECT_SOURCE_DIR}/include) include_directories(${PROJECT_SOURCE_DIR}/third-party/gtest-1.7.0/fused-src) set(ROCKSDB_LIBS rocksdblib${ARTIFACT_SUFFIX}) set(THIRDPARTY_LIBS ${THIRDPARTY_LIBS} gtest) set(SYSTEM_LIBS Shlwapi.lib Rpcrt4.lib) set(LIBS ${ROCKSDB_LIBS} ${THIRDPARTY_LIBS} ${SYSTEM_LIBS}) add_subdirectory(third-party/gtest-1.7.0/fused-src/gtest) # Main library source code set(SOURCES db/auto_roll_logger.cc db/builder.cc db/c.cc db/column_family.cc db/compacted_db_impl.cc db/compaction.cc db/compaction_iterator.cc db/compaction_job.cc db/compaction_picker.cc db/convenience.cc db/dbformat.cc db/db_filesnapshot.cc db/db_impl.cc db/db_impl_debug.cc db/db_impl_experimental.cc db/db_impl_readonly.cc db/db_info_dumper.cc db/db_iter.cc db/event_helpers.cc db/experimental.cc db/filename.cc db/file_indexer.cc db/flush_job.cc db/flush_scheduler.cc db/forward_iterator.cc db/internal_stats.cc db/log_reader.cc db/log_writer.cc db/managed_iterator.cc db/memtable.cc db/memtable_allocator.cc db/memtable_list.cc db/merge_helper.cc db/merge_operator.cc db/repair.cc db/slice.cc db/snapshot_impl.cc db/table_cache.cc db/table_properties_collector.cc db/transaction_log_impl.cc db/version_builder.cc db/version_edit.cc db/version_set.cc db/wal_manager.cc db/write_batch.cc db/write_batch_base.cc db/write_controller.cc db/write_thread.cc db/xfunc_test_points.cc memtable/hash_cuckoo_rep.cc memtable/hash_linklist_rep.cc memtable/hash_skiplist_rep.cc memtable/skiplistrep.cc memtable/vectorrep.cc port/stack_trace.cc port/win/env_win.cc port/win/port_win.cc port/win/win_logger.cc table/adaptive_table_factory.cc table/block.cc table/block_based_filter_block.cc table/block_based_table_builder.cc table/block_based_table_factory.cc table/block_based_table_reader.cc table/block_builder.cc table/block_hash_index.cc table/block_prefix_index.cc table/bloom_block.cc table/cuckoo_table_builder.cc table/cuckoo_table_factory.cc table/cuckoo_table_reader.cc table/flush_block_policy.cc table/format.cc table/full_filter_block.cc table/get_context.cc table/iterator.cc table/merger.cc table/sst_file_writer.cc table/meta_blocks.cc table/plain_table_builder.cc table/plain_table_factory.cc table/plain_table_index.cc table/plain_table_key_coding.cc table/plain_table_reader.cc table/table_properties.cc table/two_level_iterator.cc tools/sst_dump_tool.cc tools/dump/db_dump_tool.cc util/arena.cc util/bloom.cc util/build_version.cc util/cache.cc util/coding.cc util/compaction_job_stats_impl.cc util/comparator.cc util/concurrent_arena.cc util/crc32c.cc util/delete_scheduler.cc util/dynamic_bloom.cc util/env.cc util/env_hdfs.cc util/event_logger.cc util/file_util.cc util/file_reader_writer.cc util/sst_file_manager_impl.cc util/filter_policy.cc util/hash.cc util/histogram.cc util/instrumented_mutex.cc util/iostats_context.cc tools/ldb_cmd.cc tools/ldb_tool.cc util/logging.cc util/log_buffer.cc util/memenv.cc util/murmurhash.cc util/mutable_cf_options.cc util/options.cc util/options_builder.cc util/options_helper.cc util/options_parser.cc util/options_sanity_check.cc util/perf_context.cc util/perf_level.cc util/random.cc util/rate_limiter.cc util/slice.cc util/statistics.cc util/status.cc util/status_message.cc util/string_util.cc util/sync_point.cc util/testharness.cc util/testutil.cc util/thread_local.cc util/thread_status_impl.cc util/thread_status_updater.cc util/thread_status_util.cc util/thread_status_util_debug.cc util/xfunc.cc util/xxhash.cc utilities/backupable/backupable_db.cc utilities/checkpoint/checkpoint.cc utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc utilities/document/document_db.cc utilities/document/json_document.cc utilities/document/json_document_builder.cc utilities/env_mirror.cc utilities/flashcache/flashcache.cc utilities/geodb/geodb_impl.cc utilities/leveldb_options/leveldb_options.cc utilities/memory/memory_util.cc utilities/merge_operators/string_append/stringappend.cc utilities/merge_operators/string_append/stringappend2.cc utilities/merge_operators/put.cc utilities/merge_operators/uint64add.cc utilities/options/options_util.cc utilities/redis/redis_lists.cc utilities/spatialdb/spatial_db.cc utilities/table_properties_collectors/compact_on_deletion_collector.cc utilities/transactions/optimistic_transaction_impl.cc utilities/transactions/optimistic_transaction_db_impl.cc utilities/transactions/transaction_base.cc utilities/transactions/transaction_impl.cc utilities/transactions/transaction_db_impl.cc utilities/transactions/transaction_db_mutex_impl.cc utilities/transactions/transaction_lock_mgr.cc utilities/transactions/transaction_util.cc utilities/ttl/db_ttl_impl.cc utilities/write_batch_with_index/write_batch_with_index.cc utilities/write_batch_with_index/write_batch_with_index_internal.cc ) # For test util library that is build only in DEBUG mode # and linked to tests. Add test only code that is not #ifdefed for Release here. set(TESTUTIL_SOURCE db/db_test_util.cc table/mock_table.cc util/mock_env.cc util/thread_status_updater_debug.cc ) add_library(rocksdblib${ARTIFACT_SUFFIX} ${SOURCES}) set_target_properties(rocksdblib${ARTIFACT_SUFFIX} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/rocksdblib${ARTIFACT_SUFFIX}.pdb") add_dependencies(rocksdblib${ARTIFACT_SUFFIX} GenerateBuildVersion) add_library(rocksdb${ARTIFACT_SUFFIX} SHARED ${SOURCES}) set_target_properties(rocksdb${ARTIFACT_SUFFIX} PROPERTIES COMPILE_FLAGS "-DROCKSDB_DLL -DROCKSDB_LIBRARY_EXPORTS /Fd${CMAKE_CFG_INTDIR}/rocksdb${ARTIFACT_SUFFIX}.pdb") add_dependencies(rocksdb${ARTIFACT_SUFFIX} GenerateBuildVersion) target_link_libraries(rocksdb${ARTIFACT_SUFFIX} ${LIBS}) if (DEFINED JNI) if (${JNI} EQUAL 1) message(STATUS "JNI library is enabled") add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/java) else() message(STATUS "JNI library is disabled") endif() else() message(STATUS "JNI library is disabled") endif() set(APPS db/db_bench.cc db/memtablerep_bench.cc table/table_reader_bench.cc tools/db_stress.cc tools/write_stress.cc tools/db_repl_stress.cc tools/sst_dump.cc tools/dump/rocksdb_dump.cc tools/dump/rocksdb_undump.cc util/cache_bench.cc ) set(C_TESTS db/c_test.c) set(TESTS db/auto_roll_logger_test.cc db/column_family_test.cc db/compact_files_test.cc db/compaction_iterator_test.cc db/compaction_job_test.cc db/compaction_job_stats_test.cc db/compaction_picker_test.cc db/comparator_db_test.cc db/corruption_test.cc db/cuckoo_table_db_test.cc db/db_compaction_filter_test.cc db/db_compaction_test.cc db/db_dynamic_level_test.cc db/db_inplace_update_test.cc db/db_iter_test.cc db/db_log_iter_test.cc db/db_properties_test.cc db/db_table_properties_test.cc db/db_tailing_iter_test.cc db/db_test.cc db/db_universal_compaction_test.cc db/db_wal_test.cc db/dbformat_test.cc db/deletefile_test.cc db/fault_injection_test.cc db/file_indexer_test.cc db/filename_test.cc db/flush_job_test.cc db/inlineskiplist_test.cc db/listener_test.cc db/log_test.cc db/manual_compaction_test.cc db/memtable_list_test.cc db/merge_test.cc db/merge_helper_test.cc db/options_file_test.cc db/perf_context_test.cc db/plain_table_db_test.cc db/prefix_test.cc db/skiplist_test.cc db/table_properties_collector_test.cc db/version_builder_test.cc db/version_edit_test.cc db/version_set_test.cc db/wal_manager_test.cc db/write_batch_test.cc db/write_callback_test.cc db/write_controller_test.cc table/block_based_filter_block_test.cc table/block_hash_index_test.cc table/block_test.cc table/cuckoo_table_builder_test.cc table/cuckoo_table_reader_test.cc table/full_filter_block_test.cc table/merger_test.cc table/table_test.cc tools/db_sanity_test.cc tools/ldb_cmd_test.cc tools/reduce_levels_test.cc tools/sst_dump_test.cc util/arena_test.cc util/autovector_test.cc util/bloom_test.cc util/cache_test.cc util/coding_test.cc util/crc32c_test.cc util/delete_scheduler_test.cc util/dynamic_bloom_test.cc util/env_test.cc util/event_logger_test.cc util/filelock_test.cc util/file_reader_writer_test.cc util/heap_test.cc util/histogram_test.cc util/memenv_test.cc util/mock_env_test.cc util/options_test.cc util/rate_limiter_test.cc util/slice_transform_test.cc util/thread_list_test.cc util/thread_local_test.cc utilities/backupable/backupable_db_test.cc utilities/checkpoint/checkpoint_test.cc utilities/document/document_db_test.cc utilities/document/json_document_test.cc utilities/geodb/geodb_test.cc utilities/memory/memory_test.cc utilities/merge_operators/string_append/stringappend_test.cc utilities/options/options_util_test.cc utilities/redis/redis_lists_test.cc utilities/spatialdb/spatial_db_test.cc utilities/table_properties_collectors/compact_on_deletion_collector_test.cc utilities/transactions/optimistic_transaction_test.cc utilities/transactions/transaction_test.cc utilities/ttl/ttl_test.cc utilities/write_batch_with_index/write_batch_with_index_test.cc ) set(EXES ${APPS}) foreach(sourcefile ${EXES}) string(REPLACE ".cc" "" exename ${sourcefile}) string(REGEX REPLACE "^((.+)/)+" "" exename ${exename}) add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}) target_link_libraries(${exename}${ARTIFACT_SUFFIX} ${LIBS}) endforeach(sourcefile ${EXES}) # test utilities are only build in debug set(TESTUTILLIB testutillib${ARTIFACT_SUFFIX}) add_library(${TESTUTILLIB} STATIC ${TESTUTIL_SOURCE}) set_target_properties(${TESTUTILLIB} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/testutillib${ARTIFACT_SUFFIX}.pdb") set_target_properties(${TESTUTILLIB} PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1 EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1 EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1 ) # Tests are excluded from Release builds set(TEST_EXES ${TESTS}) foreach(sourcefile ${TEST_EXES}) string(REPLACE ".cc" "" exename ${sourcefile}) string(REGEX REPLACE "^((.+)/)+" "" exename ${exename}) add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}) set_target_properties(${exename}${ARTIFACT_SUFFIX} PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1 EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1 EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1 ) target_link_libraries(${exename}${ARTIFACT_SUFFIX} ${LIBS} testutillib${ARTIFACT_SUFFIX}) endforeach(sourcefile ${TEST_EXES}) # C executables must link to a shared object set(C_TEST_EXES ${C_TESTS}) foreach(sourcefile ${C_TEST_EXES}) string(REPLACE ".c" "" exename ${sourcefile}) string(REGEX REPLACE "^((.+)/)+" "" exename ${exename}) add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}) set_target_properties(${exename}${ARTIFACT_SUFFIX} PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1 EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1 EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1 ) target_link_libraries(${exename}${ARTIFACT_SUFFIX} rocksdb${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX}) endforeach(sourcefile ${C_TEST_EXES}) rocksdb-4.5.1/CONTRIBUTING.md000066400000000000000000000013351267530634000154110ustar00rootroot00000000000000# Contributing to RocksDB ## Contributor License Agreement ("CLA") In order to accept your pull request, we need you to submit a CLA. You only need to do this once, so if you've done this for another Facebook open source project, you're good to go. If you are submitting a pull request for the first time, just let us know that you have completed the CLA and we can cross-check with your GitHub username. Complete your CLA here: If you prefer to sign a paper copy, we can send you a PDF. Send us an e-mail or create a new github issue to request the CLA in PDF format. ## License By contributing to RocksDB, you agree that your contributions will be licensed under the [BSD License](LICENSE). rocksdb-4.5.1/DUMP_FORMAT.md000066400000000000000000000013731267530634000153210ustar00rootroot00000000000000## RocksDB dump format The version 1 RocksDB dump format is fairly simple: 1) The dump starts with the magic 8 byte identifier "ROCKDUMP" 2) The magic is followed by an 8 byte big-endian version which is 0x00000001. 3) Next are arbitrarily sized chunks of bytes prepended by 4 byte little endian number indicating how large each chunk is. 4) The first chunk is special and is a json string indicating some things about the creation of this dump. It contains the following keys: * database-path: The path of the database this dump was created from. * hostname: The hostname of the machine where the dump was created. * creation-time: Unix seconds since epoc when this dump was created. 5) Following the info dump the slices paired into are key/value pairs. rocksdb-4.5.1/HISTORY.md000066400000000000000000000745501267530634000146540ustar00rootroot00000000000000# Rocksdb Change Log ## Unreleased ## 4.5.1 (3/25/2016) ### Bug Fixes * Fix failures caused by the destorying order of singleton objects. ## 4.5.0 (2/5/2016) ### Public API Changes * Add a new perf context level between kEnableCount and kEnableTime. Level 2 now does not include timers for mutexes. * Statistics of mutex operation durations will not be measured by default. If you want to have them enabled, you need to set Statistics::stats_level_ to kAll. * DBOptions::delete_scheduler and NewDeleteScheduler() are removed, please use DBOptions::sst_file_manager and NewSstFileManager() instead ### New Features * ldb tool now supports operations to non-default column families. * Add kPersistedTier to ReadTier. This option allows Get and MultiGet to read only the persited data and skip mem-tables if writes were done with disableWAL = true. * Add DBOptions::sst_file_manager. Use NewSstFileManager() in include/rocksdb/sst_file_manager.h to create a SstFileManager that can be used to track the total size of SST files and control the SST files deletion rate. ## 4.4.0 (1/14/2016) ### Public API Changes * Change names in CompactionPri and add a new one. * Deprecate options.soft_rate_limit and add options.soft_pending_compaction_bytes_limit. * If options.max_write_buffer_number > 3, writes will be slowed down when writing to the last write buffer to delay a full stop. * Introduce CompactionJobInfo::compaction_reason, this field include the reason to trigger the compaction. * After slow down is triggered, if estimated pending compaction bytes keep increasing, slowdown more. * Increase default options.delayed_write_rate to 2MB/s. * Added a new parameter --path to ldb tool. --path accepts the name of either MANIFEST, SST or a WAL file. Either --db or --path can be used when calling ldb. ## 4.3.0 (12/8/2015) ### New Features * CompactionFilter has new member function called IgnoreSnapshots which allows CompactionFilter to be called even if there are snapshots later than the key. * RocksDB will now persist options under the same directory as the RocksDB database on successful DB::Open, CreateColumnFamily, DropColumnFamily, and SetOptions. * Introduce LoadLatestOptions() in rocksdb/utilities/options_util.h. This function can construct the latest DBOptions / ColumnFamilyOptions used by the specified RocksDB intance. * Introduce CheckOptionsCompatibility() in rocksdb/utilities/options_util.h. This function checks whether the input set of options is able to open the specified DB successfully. ### Public API Changes * When options.db_write_buffer_size triggers, only the column family with the largest column family size will be flushed, not all the column families. ## 4.2.0 (11/9/2015) ### New Features * Introduce CreateLoggerFromOptions(), this function create a Logger for provided DBOptions. * Add GetAggregatedIntProperty(), which returns the sum of the GetIntProperty of all the column families. * Add MemoryUtil in rocksdb/utilities/memory.h. It currently offers a way to get the memory usage by type from a list rocksdb instances. ### Public API Changes * CompactionFilter::Context includes information of Column Family ID * The need-compaction hint given by TablePropertiesCollector::NeedCompact() will be persistent and recoverable after DB recovery. This introduces a breaking format change. If you use this experimental feature, including NewCompactOnDeletionCollectorFactory() in the new version, you may not be able to directly downgrade the DB back to version 4.0 or lower. * TablePropertiesCollectorFactory::CreateTablePropertiesCollector() now takes an option Context, containing the information of column family ID for the file being written. * Remove DefaultCompactionFilterFactory. ## 4.1.0 (10/8/2015) ### New Features * Added single delete operation as a more efficient way to delete keys that have not been overwritten. * Added experimental AddFile() to DB interface that allow users to add files created by SstFileWriter into an empty Database, see include/rocksdb/sst_file_writer.h and DB::AddFile() for more info. * Added support for opening SST files with .ldb suffix which enables opening LevelDB databases. * CompactionFilter now supports filtering of merge operands and merge results. ### Public API Changes * Added SingleDelete() to the DB interface. * Added AddFile() to DB interface. * Added SstFileWriter class. * CompactionFilter has a new method FilterMergeOperand() that RocksDB applies to every merge operand during compaction to decide whether to filter the operand. * We removed CompactionFilterV2 interfaces from include/rocksdb/compaction_filter.h. The functionality was deprecated already in version 3.13. ## 4.0.0 (9/9/2015) ### New Features * Added support for transactions. See include/rocksdb/utilities/transaction.h for more info. * DB::GetProperty() now accepts "rocksdb.aggregated-table-properties" and "rocksdb.aggregated-table-properties-at-levelN", in which case it returns aggregated table properties of the target column family, or the aggregated table properties of the specified level N if the "at-level" version is used. * Add compression option kZSTDNotFinalCompression for people to experiment ZSTD although its format is not finalized. * We removed the need for LATEST_BACKUP file in BackupEngine. We still keep writing it when we create new backups (because of backward compatibility), but we don't read it anymore. ### Public API Changes * Removed class Env::RandomRWFile and Env::NewRandomRWFile(). * Renamed DBOptions.num_subcompactions to DBOptions.max_subcompactions to make the name better match the actual functionality of the option. * Added Equal() method to the Comparator interface that can optionally be overwritten in cases where equality comparisons can be done more efficiently than three-way comparisons. * Previous 'experimental' OptimisticTransaction class has been replaced by Transaction class. ## 3.13.0 (8/6/2015) ### New Features * RollbackToSavePoint() in WriteBatch/WriteBatchWithIndex * Add NewCompactOnDeletionCollectorFactory() in utilities/table_properties_collectors, which allows rocksdb to mark a SST file as need-compaction when it observes at least D deletion entries in any N consecutive entries in that SST file. Note that this feature depends on an experimental NeedCompact() API --- the result of this API will not persist after DB restart. * Add DBOptions::delete_scheduler. Use NewDeleteScheduler() in include/rocksdb/delete_scheduler.h to create a DeleteScheduler that can be shared among multiple RocksDB instances to control the file deletion rate of SST files that exist in the first db_path. ### Public API Changes * Deprecated WriteOptions::timeout_hint_us. We no longer support write timeout. If you really need this option, talk to us and we might consider returning it. * Deprecated purge_redundant_kvs_while_flush option. * Removed BackupEngine::NewBackupEngine() and NewReadOnlyBackupEngine() that were deprecated in RocksDB 3.8. Please use BackupEngine::Open() instead. * Deprecated Compaction Filter V2. We are not aware of any existing use-cases. If you use this filter, your compile will break with RocksDB 3.13. Please let us know if you use it and we'll put it back in RocksDB 3.14. * Env::FileExists now returns a Status instead of a boolean * Add statistics::getHistogramString() to print detailed distribution of a histogram metric. * Add DBOptions::skip_stats_update_on_db_open. When it is on, DB::Open() will run faster as it skips the random reads required for loading necessary stats from SST files to optimize compaction. ## 3.12.0 (7/2/2015) ### New Features * Added experimental support for optimistic transactions. See include/rocksdb/utilities/optimistic_transaction.h for more info. * Added a new way to report QPS from db_bench (check out --report_file and --report_interval_seconds) * Added a cache for individual rows. See DBOptions::row_cache for more info. * Several new features on EventListener (see include/rocksdb/listener.h): - OnCompationCompleted() now returns per-compaciton job statistics, defined in include/rocksdb/compaction_job_stats.h. - Added OnTableFileCreated() and OnTableFileDeleted(). * Add compaction_options_universal.enable_trivial_move to true, to allow trivial move while performing universal compaction. Trivial move will happen only when all the input files are non overlapping. ### Public API changes * EventListener::OnFlushCompleted() now passes FlushJobInfo instead of a list of parameters. * DB::GetDbIdentity() is now a const function. If this function is overridden in your application, be sure to also make GetDbIdentity() const to avoid compile error. * Move listeners from ColumnFamilyOptions to DBOptions. * Add max_write_buffer_number_to_maintain option * DB::CompactRange()'s parameter reduce_level is changed to change_level, to allow users to move levels to lower levels if allowed. It can be used to migrate a DB from options.level_compaction_dynamic_level_bytes=false to options.level_compaction_dynamic_level_bytes.true. * Change default value for options.compaction_filter_factory and options.compaction_filter_factory_v2 to nullptr instead of DefaultCompactionFilterFactory and DefaultCompactionFilterFactoryV2. * If CancelAllBackgroundWork is called without doing a flush after doing loads with WAL disabled, the changes which haven't been flushed before the call to CancelAllBackgroundWork will be lost. * WBWIIterator::Entry() now returns WriteEntry instead of `const WriteEntry&` * options.hard_rate_limit is deprecated. * When options.soft_rate_limit or options.level0_slowdown_writes_trigger is triggered, the way to slow down writes is changed to: write rate to DB is limited to to options.delayed_write_rate. * DB::GetApproximateSizes() adds a parameter to allow the estimation to include data in mem table, with default to be not to include. It is now only supported in skip list mem table. * DB::CompactRange() now accept CompactRangeOptions instead of multiple paramters. CompactRangeOptions is defined in include/rocksdb/options.h. * CompactRange() will now skip bottommost level compaction for level based compaction if there is no compaction filter, bottommost_level_compaction is introduced in CompactRangeOptions to control when it's possbile to skip bottommost level compaction. This mean that if you want the compaction to produce a single file you need to set bottommost_level_compaction to BottommostLevelCompaction::kForce. * Add Cache.GetPinnedUsage() to get the size of memory occupied by entries that are in use by the system. * DB:Open() will fail if the compression specified in Options is not linked with the binary. If you see this failure, recompile RocksDB with compression libraries present on your system. Also, previously our default compression was snappy. This behavior is now changed. Now, the default compression is snappy only if it's available on the system. If it isn't we change the default to kNoCompression. * We changed how we account for memory used in block cache. Previously, we only counted the sum of block sizes currently present in block cache. Now, we count the actual memory usage of the blocks. For example, a block of size 4.5KB will use 8KB memory with jemalloc. This might decrease your memory usage and possibly decrease performance. Increase block cache size if you see this happening after an upgrade. * Add BackupEngineImpl.options_.max_background_operations to specify the maximum number of operations that may be performed in parallel. Add support for parallelized backup and restore. * Add DB::SyncWAL() that does a WAL sync without blocking writers. ## 3.11.0 (5/19/2015) ### New Features * Added a new API Cache::SetCapacity(size_t capacity) to dynamically change the maximum configured capacity of the cache. If the new capacity is less than the existing cache usage, the implementation will try to lower the usage by evicting the necessary number of elements following a strict LRU policy. * Added an experimental API for handling flashcache devices (blacklists background threads from caching their reads) -- NewFlashcacheAwareEnv * If universal compaction is used and options.num_levels > 1, compact files are tried to be stored in none-L0 with smaller files based on options.target_file_size_base. The limitation of DB size when using universal compaction is greatly mitigated by using more levels. You can set num_levels = 1 to make universal compaction behave as before. If you set num_levels > 1 and want to roll back to a previous version, you need to compact all files to a big file in level 0 (by setting target_file_size_base to be large and CompactRange(, nullptr, nullptr, true, 0) and reopen the DB with the same version to rewrite the manifest, and then you can open it using previous releases. * More information about rocksdb background threads are available in Env::GetThreadList(), including the number of bytes read / written by a compaction job, mem-table size and current number of bytes written by a flush job and many more. Check include/rocksdb/thread_status.h for more detail. ### Public API changes * TablePropertiesCollector::AddUserKey() is added to replace TablePropertiesCollector::Add(). AddUserKey() exposes key type, sequence number and file size up to now to users. * DBOptions::bytes_per_sync used to apply to both WAL and table files. As of 3.11 it applies only to table files. If you want to use this option to sync WAL in the background, please use wal_bytes_per_sync ## 3.10.0 (3/24/2015) ### New Features * GetThreadStatus() is now able to report detailed thread status, including: - Thread Operation including flush and compaction. - The stage of the current thread operation. - The elapsed time in micros since the current thread operation started. More information can be found in include/rocksdb/thread_status.h. In addition, when running db_bench with --thread_status_per_interval, db_bench will also report thread status periodically. * Changed the LRU caching algorithm so that referenced blocks (by iterators) are never evicted. This change made parameter removeScanCountLimit obsolete. Because of that NewLRUCache doesn't take three arguments anymore. table_cache_remove_scan_limit option is also removed * By default we now optimize the compilation for the compilation platform (using -march=native). If you want to build portable binary, use 'PORTABLE=1' before the make command. * We now allow level-compaction to place files in different paths by specifying them in db_paths along with the target_size. Lower numbered levels will be placed earlier in the db_paths and higher numbered levels will be placed later in the db_paths vector. * Potentially big performance improvements if you're using RocksDB with lots of column families (100-1000) * Added BlockBasedTableOptions.format_version option, which allows user to specify which version of block based table he wants. As a general guidline, newer versions have more features, but might not be readable by older versions of RocksDB. * Added new block based table format (version 2), which you can enable by setting BlockBasedTableOptions.format_version = 2. This format changes how we encode size information in compressed blocks and should help with memory allocations if you're using Zlib or BZip2 compressions. * MemEnv (env that stores data in memory) is now available in default library build. You can create it by calling NewMemEnv(). * Add SliceTransform.SameResultWhenAppended() to help users determine it is safe to apply prefix bloom/hash. * Block based table now makes use of prefix bloom filter if it is a full fulter. * Block based table remembers whether a whole key or prefix based bloom filter is supported in SST files. Do a sanity check when reading the file with users' configuration. * Fixed a bug in ReadOnlyBackupEngine that deleted corrupted backups in some cases, even though the engine was ReadOnly * options.level_compaction_dynamic_level_bytes, a feature to allow RocksDB to pick dynamic base of bytes for levels. With this feature turned on, we will automatically adjust max bytes for each level. The goal of this feature is to have lower bound on size amplification. For more details, see comments in options.h. * Added an abstract base class WriteBatchBase for write batches * Fixed a bug where we start deleting files of a dropped column families even if there are still live references to it ### Public API changes * Deprecated skip_log_error_on_recovery and table_cache_remove_scan_count_limit options. * Logger method logv with log level parameter is now virtual ### RocksJava * Added compression per level API. * MemEnv is now available in RocksJava via RocksMemEnv class. * lz4 compression is now included in rocksjava static library when running `make rocksdbjavastatic`. * Overflowing a size_t when setting rocksdb options now throws an IllegalArgumentException, which removes the necessity for a developer to catch these Exceptions explicitly. ## 3.9.0 (12/8/2014) ### New Features * Add rocksdb::GetThreadList(), which in the future will return the current status of all rocksdb-related threads. We will have more code instruments in the following RocksDB releases. * Change convert function in rocksdb/utilities/convenience.h to return Status instead of boolean. Also add support for nested options in convert function ### Public API changes * New API to create a checkpoint added. Given a directory name, creates a new database which is an image of the existing database. * New API LinkFile added to Env. If you implement your own Env class, an implementation of the API LinkFile will have to be provided. * MemTableRep takes MemTableAllocator instead of Arena ### Improvements * RocksDBLite library now becomes smaller and will be compiled with -fno-exceptions flag. ## 3.8.0 (11/14/2014) ### Public API changes * BackupEngine::NewBackupEngine() was deprecated; please use BackupEngine::Open() from now on. * BackupableDB/RestoreBackupableDB have new GarbageCollect() methods, which will clean up files from corrupt and obsolete backups. * BackupableDB/RestoreBackupableDB have new GetCorruptedBackups() methods which list corrupt backups. ### Cleanup * Bunch of code cleanup, some extra warnings turned on (-Wshadow, -Wshorten-64-to-32, -Wnon-virtual-dtor) ### New features * CompactFiles and EventListener, although they are still in experimental state * Full ColumnFamily support in RocksJava. ## 3.7.0 (11/6/2014) ### Public API changes * Introduce SetOptions() API to allow adjusting a subset of options dynamically online * Introduce 4 new convenient functions for converting Options from string: GetColumnFamilyOptionsFromMap(), GetColumnFamilyOptionsFromString(), GetDBOptionsFromMap(), GetDBOptionsFromString() * Remove WriteBatchWithIndex.Delete() overloads using SliceParts * When opening a DB, if options.max_background_compactions is larger than the existing low pri pool of options.env, it will enlarge it. Similarly, options.max_background_flushes is larger than the existing high pri pool of options.env, it will enlarge it. ## 3.6.0 (10/7/2014) ### Disk format changes * If you're using RocksDB on ARM platforms and you're using default bloom filter, there is a disk format change you need to be aware of. There are three steps you need to do when you convert to new release: 1. turn off filter policy, 2. compact the whole database, 3. turn on filter policy ### Behavior changes * We have refactored our system of stalling writes. Any stall-related statistics' meanings are changed. Instead of per-write stall counts, we now count stalls per-epoch, where epochs are periods between flushes and compactions. You'll find more information in our Tuning Perf Guide once we release RocksDB 3.6. * When disableDataSync=true, we no longer sync the MANIFEST file. * Add identity_as_first_hash property to CuckooTable. SST file needs to be rebuilt to be opened by reader properly. ### Public API changes * Change target_file_size_base type to uint64_t from int. * Remove allow_thread_local. This feature was proved to be stable, so we are turning it always-on. ## 3.5.0 (9/3/2014) ### New Features * Add include/utilities/write_batch_with_index.h, providing a utilitiy class to query data out of WriteBatch when building it. * Move BlockBasedTable related options to BlockBasedTableOptions from Options. Change corresponding JNI interface. Options affected include: no_block_cache, block_cache, block_cache_compressed, block_size, block_size_deviation, block_restart_interval, filter_policy, whole_key_filtering. filter_policy is changed to shared_ptr from a raw pointer. * Remove deprecated options: disable_seek_compaction and db_stats_log_interval * OptimizeForPointLookup() takes one parameter for block cache size. It now builds hash index, bloom filter, and block cache. ### Public API changes * The Prefix Extractor used with V2 compaction filters is now passed user key to SliceTransform::Transform instead of unparsed RocksDB key. ## 3.4.0 (8/18/2014) ### New Features * Support Multiple DB paths in universal style compactions * Add feature of storing plain table index and bloom filter in SST file. * CompactRange() will never output compacted files to level 0. This used to be the case when all the compaction input files were at level 0. * Added iterate_upper_bound to define the extent upto which the forward iterator will return entries. This will prevent iterating over delete markers and overwritten entries for edge cases where you want to break out the iterator anyways. This may improve performance in case there are a large number of delete markers or overwritten entries. ### Public API changes * DBOptions.db_paths now is a vector of a DBPath structure which indicates both of path and target size * NewPlainTableFactory instead of bunch of parameters now accepts PlainTableOptions, which is defined in include/rocksdb/table.h * Moved include/utilities/*.h to include/rocksdb/utilities/*.h * Statistics APIs now take uint32_t as type instead of Tickers. Also make two access functions getTickerCount and histogramData const * Add DB property rocksdb.estimate-num-keys, estimated number of live keys in DB. * Add DB::GetIntProperty(), which returns DB properties that are integer as uint64_t. * The Prefix Extractor used with V2 compaction filters is now passed user key to SliceTransform::Transform instead of unparsed RocksDB key. ## 3.3.0 (7/10/2014) ### New Features * Added JSON API prototype. * HashLinklist reduces performance outlier caused by skewed bucket by switching data in the bucket from linked list to skip list. Add parameter threshold_use_skiplist in NewHashLinkListRepFactory(). * RocksDB is now able to reclaim storage space more effectively during the compaction process. This is done by compensating the size of each deletion entry by the 2X average value size, which makes compaction to be triggerred by deletion entries more easily. * Add TimeOut API to write. Now WriteOptions have a variable called timeout_hint_us. With timeout_hint_us set to non-zero, any write associated with this timeout_hint_us may be aborted when it runs longer than the specified timeout_hint_us, and it is guaranteed that any write completes earlier than the specified time-out will not be aborted due to the time-out condition. * Add a rate_limiter option, which controls total throughput of flush and compaction. The throughput is specified in bytes/sec. Flush always has precedence over compaction when available bandwidth is constrained. ### Public API changes * Removed NewTotalOrderPlainTableFactory because it is not used and implemented semantically incorrect. ## 3.2.0 (06/20/2014) ### Public API changes * We removed seek compaction as a concept from RocksDB because: 1) It makes more sense for spinning disk workloads, while RocksDB is primarily designed for flash and memory, 2) It added some complexity to the important code-paths, 3) None of our internal customers were really using it. Because of that, Options::disable_seek_compaction is now obsolete. It is still a parameter in Options, so it does not break the build, but it does not have any effect. We plan to completely remove it at some point, so we ask users to please remove this option from your code base. * Add two paramters to NewHashLinkListRepFactory() for logging on too many entries in a hash bucket when flushing. * Added new option BlockBasedTableOptions::hash_index_allow_collision. When enabled, prefix hash index for block-based table will not store prefix and allow hash collision, reducing memory consumption. ### New Features * PlainTable now supports a new key encoding: for keys of the same prefix, the prefix is only written once. It can be enabled through encoding_type paramter of NewPlainTableFactory() * Add AdaptiveTableFactory, which is used to convert from a DB of PlainTable to BlockBasedTabe, or vise versa. It can be created using NewAdaptiveTableFactory() ### Performance Improvements * Tailing Iterator re-implemeted with ForwardIterator + Cascading Search Hint , see ~20% throughput improvement. ## 3.1.0 (05/21/2014) ### Public API changes * Replaced ColumnFamilyOptions::table_properties_collectors with ColumnFamilyOptions::table_properties_collector_factories ### New Features * Hash index for block-based table will be materialized and reconstructed more efficiently. Previously hash index is constructed by scanning the whole table during every table open. * FIFO compaction style ## 3.0.0 (05/05/2014) ### Public API changes * Added _LEVEL to all InfoLogLevel enums * Deprecated ReadOptions.prefix and ReadOptions.prefix_seek. Seek() defaults to prefix-based seek when Options.prefix_extractor is supplied. More detail is documented in https://github.com/facebook/rocksdb/wiki/Prefix-Seek-API-Changes * MemTableRepFactory::CreateMemTableRep() takes info logger as an extra parameter. ### New Features * Column family support * Added an option to use different checksum functions in BlockBasedTableOptions * Added ApplyToAllCacheEntries() function to Cache ## 2.8.0 (04/04/2014) * Removed arena.h from public header files. * By default, checksums are verified on every read from database * Change default value of several options, including: paranoid_checks=true, max_open_files=5000, level0_slowdown_writes_trigger=20, level0_stop_writes_trigger=24, disable_seek_compaction=true, max_background_flushes=1 and allow_mmap_writes=false * Added is_manual_compaction to CompactionFilter::Context * Added "virtual void WaitForJoin()" in class Env. Default operation is no-op. * Removed BackupEngine::DeleteBackupsNewerThan() function * Added new option -- verify_checksums_in_compaction * Changed Options.prefix_extractor from raw pointer to shared_ptr (take ownership) Changed HashSkipListRepFactory and HashLinkListRepFactory constructor to not take SliceTransform object (use Options.prefix_extractor implicitly) * Added Env::GetThreadPoolQueueLen(), which returns the waiting queue length of thread pools * Added a command "checkconsistency" in ldb tool, which checks if file system state matches DB state (file existence and file sizes) * Separate options related to block based table to a new struct BlockBasedTableOptions. * WriteBatch has a new function Count() to return total size in the batch, and Data() now returns a reference instead of a copy * Add more counters to perf context. * Supports several more DB properties: compaction-pending, background-errors and cur-size-active-mem-table. ### New Features * If we find one truncated record at the end of the MANIFEST or WAL files, we will ignore it. We assume that writers of these records were interrupted and that we can safely ignore it. * A new SST format "PlainTable" is added, which is optimized for memory-only workloads. It can be created through NewPlainTableFactory() or NewTotalOrderPlainTableFactory(). * A new mem table implementation hash linked list optimizing for the case that there are only few keys for each prefix, which can be created through NewHashLinkListRepFactory(). * Merge operator supports a new function PartialMergeMulti() to allow users to do partial merges against multiple operands. * Now compaction filter has a V2 interface. It buffers the kv-pairs sharing the same key prefix, process them in batches, and return the batched results back to DB. The new interface uses a new structure CompactionFilterContext for the same purpose as CompactionFilter::Context in V1. * Geo-spatial support for locations and radial-search. ## 2.7.0 (01/28/2014) ### Public API changes * Renamed `StackableDB::GetRawDB()` to `StackableDB::GetBaseDB()`. * Renamed `WriteBatch::Data()` `const std::string& Data() const`. * Renamed class `TableStats` to `TableProperties`. * Deleted class `PrefixHashRepFactory`. Please use `NewHashSkipListRepFactory()` instead. * Supported multi-threaded `EnableFileDeletions()` and `DisableFileDeletions()`. * Added `DB::GetOptions()`. * Added `DB::GetDbIdentity()`. ### New Features * Added [BackupableDB](https://github.com/facebook/rocksdb/wiki/How-to-backup-RocksDB%3F) * Implemented [TailingIterator](https://github.com/facebook/rocksdb/wiki/Tailing-Iterator), a special type of iterator that doesn't create a snapshot (can be used to read newly inserted data) and is optimized for doing sequential reads. * Added property block for table, which allows (1) a table to store its metadata and (2) end user to collect and store properties they are interested in. * Enabled caching index and filter block in block cache (turned off by default). * Supported error report when doing manual compaction. * Supported additional Linux platform flavors and Mac OS. * Put with `SliceParts` - Variant of `Put()` that gathers output like `writev(2)` * Bug fixes and code refactor for compatibility with upcoming Column Family feature. ### Performance Improvements * Huge benchmark performance improvements by multiple efforts. For example, increase in readonly QPS from about 530k in 2.6 release to 1.1 million in 2.7 [1] * Speeding up a way RocksDB deleted obsolete files - no longer listing the whole directory under a lock -- decrease in p99 * Use raw pointer instead of shared pointer for statistics: [5b825d](https://github.com/facebook/rocksdb/commit/5b825d6964e26ec3b4bb6faa708ebb1787f1d7bd) -- huge increase in performance -- shared pointers are slow * Optimized locking for `Get()` -- [1fdb3f](https://github.com/facebook/rocksdb/commit/1fdb3f7dc60e96394e3e5b69a46ede5d67fb976c) -- 1.5x QPS increase for some workloads * Cache speedup - [e8d40c3](https://github.com/facebook/rocksdb/commit/e8d40c31b3cca0c3e1ae9abe9b9003b1288026a9) * Implemented autovector, which allocates first N elements on stack. Most of vectors in RocksDB are small. Also, we never want to allocate heap objects while holding a mutex. -- [c01676e4](https://github.com/facebook/rocksdb/commit/c01676e46d3be08c3c140361ef1f5884f47d3b3c) * Lots of efforts to move malloc, memcpy and IO outside of locks rocksdb-4.5.1/INSTALL.md000066400000000000000000000076611267530634000146200ustar00rootroot00000000000000## Compilation **Important**: If you plan to run RocksDB in production, don't compile using default `make` or `make all`. That will compile RocksDB in debug mode, which is much slower than release mode. RocksDB's library should be able to compile without any dependency installed, although we recommend installing some compression libraries (see below). We do depend on newer gcc/clang with C++11 support. There are few options when compiling RocksDB: * [recommended] `make static_lib` will compile librocksdb.a, RocksDB static library. Compiles static library in release mode. * `make shared_lib` will compile librocksdb.so, RocksDB shared library. Compiles shared library in release mode. * `make check` will compile and run all the unit tests. `make check` will compile RocksDB in debug mode. * `make all` will compile our static library, and all our tools and unit tests. Our tools depend on gflags. You will need to have gflags installed to run `make all`. This will compile RocksDB in debug mode. Don't use binaries compiled by `make all` in production. * By default the binary we produce is optimized for the platform you're compiling on (-march=native or the equivalent). If you want to build a portable binary, add 'PORTABLE=1' before your make commands, like this: `PORTABLE=1 make static_lib` ## Dependencies * You can link RocksDB with following compression libraries: - [zlib](http://www.zlib.net/) - a library for data compression. - [bzip2](http://www.bzip.org/) - a library for data compression. - [snappy](https://code.google.com/p/snappy/) - a library for fast data compression. * All our tools depend on: - [gflags](https://gflags.github.io/gflags/) - a library that handles command line flags processing. You can compile rocksdb library even if you don't have gflags installed. ## Supported platforms * **Linux - Ubuntu** * Upgrade your gcc to version at least 4.7 to get C++11 support. * Install gflags. First, try: `sudo apt-get install libgflags-dev` If this doesn't work and you're using Ubuntu, here's a nice tutorial: (http://askubuntu.com/questions/312173/installing-gflags-12-04) * Install snappy. This is usually as easy as: `sudo apt-get install libsnappy-dev`. * Install zlib. Try: `sudo apt-get install zlib1g-dev`. * Install bzip2: `sudo apt-get install libbz2-dev`. * **Linux - CentOS** * Upgrade your gcc to version at least 4.7 to get C++11 support: `yum install gcc47-c++` * Install gflags: wget https://gflags.googlecode.com/files/gflags-2.0-no-svn-files.tar.gz tar -xzvf gflags-2.0-no-svn-files.tar.gz cd gflags-2.0 ./configure && make && sudo make install * Install snappy: wget https://snappy.googlecode.com/files/snappy-1.1.1.tar.gz tar -xzvf snappy-1.1.1.tar.gz cd snappy-1.1.1 ./configure && make && sudo make install * Install zlib: sudo yum install zlib sudo yum install zlib-devel * Install bzip2: sudo yum install bzip2 sudo yum install bzip2-devel * **OS X**: * Install latest C++ compiler that supports C++ 11: * Update XCode: run `xcode-select --install` (or install it from XCode App's settting). * Install via [homebrew](http://brew.sh/). * If you're first time developer in MacOS, you still need to run: `xcode-select --install` in your command line. * run `brew tap homebrew/versions; brew install gcc47 --use-llvm` to install gcc 4.7 (or higher). * run `brew install rocksdb` * **iOS**: * Run: `TARGET_OS=IOS make static_lib`. When building the project which uses rocksdb iOS library, make sure to define two important pre-processing macros: `ROCKSDB_LITE` and `IOS_CROSS_COMPILE`. * **Windows**: * For building with MS Visual Studio 13 you will need Update 4 installed. * Read and follow the instructions at CMakeLists.txt rocksdb-4.5.1/LICENSE000066400000000000000000000031561267530634000141700ustar00rootroot00000000000000BSD License For rocksdb software Copyright (c) 2014, Facebook, Inc. All rights reserved. --------------------------------------------------------------------- Copyright (c) 2011 The LevelDB Authors. All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. * Neither the name of Google Inc. nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. rocksdb-4.5.1/Makefile000066400000000000000000001217461267530634000146310ustar00rootroot00000000000000# Copyright (c) 2011 The LevelDB Authors. All rights reserved. # Use of this source code is governed by a BSD-style license that can be # found in the LICENSE file. See the AUTHORS file for names of contributors. # Inherit some settings from environment variables, if available #----------------------------------------------- CLEAN_FILES = # deliberately empty, so we can append below. CFLAGS += ${EXTRA_CFLAGS} CXXFLAGS += ${EXTRA_CXXFLAGS} LDFLAGS += $(EXTRA_LDFLAGS) MACHINE ?= $(shell uname -m) ARFLAGS = rs # Transform parallel LOG output into something more readable. perl_command = perl -n \ -e '@a=split("\t",$$_,-1); $$t=$$a[8]; $$t =~ s,^\./,,;' \ -e '$$t =~ s, >.*,,; chomp $$t;' \ -e '$$t =~ /.*--gtest_filter=(.*?\.[\w\/]+)/ and $$t=$$1;' \ -e 'printf "%7.3f %s %s\n", $$a[3], $$a[6] == 0 ? "PASS" : "FAIL", $$t' quoted_perl_command = $(subst ','\'',$(perl_command)) # DEBUG_LEVEL can have three values: # * DEBUG_LEVEL=2; this is the ultimate debug mode. It will compile rocksdb # without any optimizations. To compile with level 2, issue `make dbg` # * DEBUG_LEVEL=1; debug level 1 enables all assertions and debug code, but # compiles rocksdb with -O2 optimizations. this is the default debug level. # `make all` or `make ` compile RocksDB with debug level 1. # We use this debug level when developing RocksDB. # * DEBUG_LEVEL=0; this is the debug level we use for release. If you're # running rocksdb in production you most definitely want to compile RocksDB # with debug level 0. To compile with level 0, run `make shared_lib`, # `make install-shared`, `make static_lib`, `make install-static` or # `make install` # Set the default DEBUG_LEVEL to 1 DEBUG_LEVEL?=1 ifeq ($(MAKECMDGOALS),dbg) DEBUG_LEVEL=2 endif ifeq ($(MAKECMDGOALS),clean) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),release) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),shared_lib) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),install-shared) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),static_lib) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),install-static) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),install) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),rocksdbjavastatic) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),rocksdbjavastaticrelease) DEBUG_LEVEL=0 endif ifeq ($(MAKECMDGOALS),rocksdbjavastaticpublish) DEBUG_LEVEL=0 endif # compile with -O2 if debug level is not 2 ifneq ($(DEBUG_LEVEL), 2) OPT += -O2 -fno-omit-frame-pointer ifeq (,$(findstring ppc64,$(MACHINE))) # ppc64[le] doesn't support -momit-leaf-frame-pointer ifneq ($(MACHINE),armv7l) OPT += -momit-leaf-frame-pointer endif endif endif # if we're compiling for release, compile without debug code (-DNDEBUG) and # don't treat warnings as errors ifeq ($(DEBUG_LEVEL),0) OPT += -DNDEBUG DISABLE_WARNING_AS_ERROR=1 else $(warning Warning: Compiling in debug mode. Don't use the resulting binary in production) endif #----------------------------------------------- include src.mk AM_DEFAULT_VERBOSITY = 0 AM_V_GEN = $(am__v_GEN_$(V)) am__v_GEN_ = $(am__v_GEN_$(AM_DEFAULT_VERBOSITY)) am__v_GEN_0 = @echo " GEN " $@; am__v_GEN_1 = AM_V_at = $(am__v_at_$(V)) am__v_at_ = $(am__v_at_$(AM_DEFAULT_VERBOSITY)) am__v_at_0 = @ am__v_at_1 = AM_V_CC = $(am__v_CC_$(V)) am__v_CC_ = $(am__v_CC_$(AM_DEFAULT_VERBOSITY)) am__v_CC_0 = @echo " CC " $@; am__v_CC_1 = CCLD = $(CC) LINK = $(CCLD) $(AM_CFLAGS) $(CFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@ AM_V_CCLD = $(am__v_CCLD_$(V)) am__v_CCLD_ = $(am__v_CCLD_$(AM_DEFAULT_VERBOSITY)) am__v_CCLD_0 = @echo " CCLD " $@; am__v_CCLD_1 = AM_V_AR = $(am__v_AR_$(V)) am__v_AR_ = $(am__v_AR_$(AM_DEFAULT_VERBOSITY)) am__v_AR_0 = @echo " AR " $@; am__v_AR_1 = AM_LINK = $(AM_V_CCLD)$(CXX) $^ $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS) # detect what platform we're building on dummy := $(shell (export ROCKSDB_ROOT="$(CURDIR)"; "$(CURDIR)/build_tools/build_detect_platform" "$(CURDIR)/make_config.mk")) # this file is generated by the previous line to set build flags and sources include make_config.mk CLEAN_FILES += make_config.mk ifneq ($(PLATFORM), IOS) CFLAGS += -g CXXFLAGS += -g else # no debug info for IOS, that will make our library big OPT += -DNDEBUG endif ifeq ($(PLATFORM), OS_SOLARIS) PLATFORM_CXXFLAGS += -D _GLIBCXX_USE_C99 endif ifneq ($(filter -DROCKSDB_LITE,$(OPT)),) # found CFLAGS += -fno-exceptions CXXFLAGS += -fno-exceptions endif # ASAN doesn't work well with jemalloc. If we're compiling with ASAN, we should use regular malloc. ifdef COMPILE_WITH_ASAN DISABLE_JEMALLOC=1 EXEC_LDFLAGS += -fsanitize=address PLATFORM_CCFLAGS += -fsanitize=address PLATFORM_CXXFLAGS += -fsanitize=address endif # TSAN doesn't work well with jemalloc. If we're compiling with TSAN, we should use regular malloc. ifdef COMPILE_WITH_TSAN DISABLE_JEMALLOC=1 EXEC_LDFLAGS += -fsanitize=thread -pie PLATFORM_CCFLAGS += -fsanitize=thread -fPIC -DROCKSDB_TSAN_RUN PLATFORM_CXXFLAGS += -fsanitize=thread -fPIC -DROCKSDB_TSAN_RUN # Turn off -pg when enabling TSAN testing, because that induces # a link failure. TODO: find the root cause pg = else pg = -pg endif ifndef DISABLE_JEMALLOC EXEC_LDFLAGS := $(JEMALLOC_LIB) $(EXEC_LDFLAGS) PLATFORM_CXXFLAGS += $(JEMALLOC_INCLUDE) PLATFORM_CCFLAGS += $(JEMALLOC_INCLUDE) endif export GTEST_THROW_ON_FAILURE=1 GTEST_HAS_EXCEPTIONS=1 GTEST_DIR = ./third-party/gtest-1.7.0/fused-src PLATFORM_CCFLAGS += -isystem $(GTEST_DIR) PLATFORM_CXXFLAGS += -isystem $(GTEST_DIR) # This (the first rule) must depend on "all". default: all WARNING_FLAGS = -W -Wextra -Wall -Wsign-compare -Wshadow \ -Wno-unused-parameter ifndef DISABLE_WARNING_AS_ERROR WARNING_FLAGS += -Werror endif CFLAGS += $(WARNING_FLAGS) -I. -I./include $(PLATFORM_CCFLAGS) $(OPT) CXXFLAGS += $(WARNING_FLAGS) -I. -I./include $(PLATFORM_CXXFLAGS) $(OPT) -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers LDFLAGS += $(PLATFORM_LDFLAGS) date := $(shell date +%F) ifdef FORCE_GIT_SHA git_sha := $(FORCE_GIT_SHA) else git_sha := $(shell git rev-parse HEAD 2>/dev/null) endif gen_build_version = \ printf '%s\n' \ '\#include "build_version.h"' \ 'const char* rocksdb_build_git_sha = \ "rocksdb_build_git_sha:$(git_sha)";' \ 'const char* rocksdb_build_git_date = \ "rocksdb_build_git_date:$(date)";' \ 'const char* rocksdb_build_compile_date = __DATE__;' # Record the version of the source that we are compiling. # We keep a record of the git revision in this file. It is then built # as a regular source file as part of the compilation process. # One can run "strings executable_filename | grep _build_" to find # the version of the source that we used to build the executable file. CLEAN_FILES += util/build_version.cc: FORCE: util/build_version.cc: FORCE $(AM_V_GEN)rm -f $@-t $(AM_V_at)$(gen_build_version) > $@-t $(AM_V_at)if test -f $@; then \ cmp -s $@-t $@ && rm -f $@-t || mv -f $@-t $@; \ else mv -f $@-t $@; fi LIBOBJECTS = $(LIB_SOURCES:.cc=.o) LIBOBJECTS += $(TOOL_SOURCES:.cc=.o) MOCKOBJECTS = $(MOCK_SOURCES:.cc=.o) GTEST = $(GTEST_DIR)/gtest/gtest-all.o TESTUTIL = ./util/testutil.o TESTHARNESS = ./util/testharness.o $(TESTUTIL) $(MOCKOBJECTS) $(GTEST) VALGRIND_ERROR = 2 VALGRIND_VER := $(join $(VALGRIND_VER),valgrind) VALGRIND_OPTS = --error-exitcode=$(VALGRIND_ERROR) --leak-check=full TESTS = \ db_test \ db_iter_test \ db_log_iter_test \ db_compaction_filter_test \ db_compaction_test \ db_dynamic_level_test \ db_inplace_update_test \ db_tailing_iter_test \ db_universal_compaction_test \ db_wal_test \ db_properties_test \ db_table_properties_test \ block_hash_index_test \ autovector_test \ column_family_test \ table_properties_collector_test \ arena_test \ auto_roll_logger_test \ block_test \ bloom_test \ dynamic_bloom_test \ c_test \ cache_test \ checkpoint_test \ coding_test \ corruption_test \ crc32c_test \ slice_transform_test \ dbformat_test \ env_test \ fault_injection_test \ filelock_test \ filename_test \ file_reader_writer_test \ block_based_filter_block_test \ full_filter_block_test \ histogram_test \ inlineskiplist_test \ log_test \ manual_compaction_test \ memenv_test \ mock_env_test \ memtable_list_test \ merge_helper_test \ memory_test \ merge_test \ merger_test \ options_file_test \ redis_test \ reduce_levels_test \ plain_table_db_test \ comparator_db_test \ prefix_test \ skiplist_test \ stringappend_test \ ttl_test \ backupable_db_test \ document_db_test \ json_document_test \ spatial_db_test \ version_edit_test \ version_set_test \ compaction_picker_test \ version_builder_test \ file_indexer_test \ write_batch_test \ write_batch_with_index_test \ write_controller_test\ deletefile_test \ table_test \ thread_local_test \ geodb_test \ rate_limiter_test \ delete_scheduler_test \ options_test \ options_util_test \ event_logger_test \ cuckoo_table_builder_test \ cuckoo_table_reader_test \ cuckoo_table_db_test \ flush_job_test \ wal_manager_test \ listener_test \ compaction_iterator_test \ compaction_job_test \ thread_list_test \ sst_dump_test \ compact_files_test \ perf_context_test \ optimistic_transaction_test \ write_callback_test \ heap_test \ compact_on_deletion_collector_test \ compaction_job_stats_test \ transaction_test \ ldb_cmd_test SUBSET := $(shell echo $(TESTS) |sed s/^.*$(ROCKSDBTESTS_START)/$(ROCKSDBTESTS_START)/) TOOLS = \ sst_dump \ db_sanity_test \ db_stress \ write_stress \ ldb \ db_repl_stress \ rocksdb_dump \ rocksdb_undump # TODO: add back forward_iterator_bench, after making it build in all environemnts. BENCHMARKS = db_bench table_reader_bench cache_bench memtablerep_bench # if user didn't config LIBNAME, set the default ifeq ($(LIBNAME),) # we should only run rocksdb in production with DEBUG_LEVEL 0 ifeq ($(DEBUG_LEVEL),0) LIBNAME=librocksdb else LIBNAME=librocksdb_debug endif endif LIBRARY = ${LIBNAME}.a ROCKSDB_MAJOR = $(shell egrep "ROCKSDB_MAJOR.[0-9]" include/rocksdb/version.h | cut -d ' ' -f 3) ROCKSDB_MINOR = $(shell egrep "ROCKSDB_MINOR.[0-9]" include/rocksdb/version.h | cut -d ' ' -f 3) ROCKSDB_PATCH = $(shell egrep "ROCKSDB_PATCH.[0-9]" include/rocksdb/version.h | cut -d ' ' -f 3) default: all #----------------------------------------------- # Create platform independent shared libraries. #----------------------------------------------- ifneq ($(PLATFORM_SHARED_EXT),) ifneq ($(PLATFORM_SHARED_VERSIONED),true) SHARED1 = ${LIBNAME}.$(PLATFORM_SHARED_EXT) SHARED2 = $(SHARED1) SHARED3 = $(SHARED1) SHARED4 = $(SHARED1) SHARED = $(SHARED1) else SHARED_MAJOR = $(ROCKSDB_MAJOR) SHARED_MINOR = $(ROCKSDB_MINOR) SHARED_PATCH = $(ROCKSDB_PATCH) SHARED1 = ${LIBNAME}.$(PLATFORM_SHARED_EXT) ifeq ($(PLATFORM), OS_MACOSX) SHARED_OSX = $(LIBNAME).$(SHARED_MAJOR) SHARED2 = $(SHARED_OSX).$(PLATFORM_SHARED_EXT) SHARED3 = $(SHARED_OSX).$(SHARED_MINOR).$(PLATFORM_SHARED_EXT) SHARED4 = $(SHARED_OSX).$(SHARED_MINOR).$(SHARED_PATCH).$(PLATFORM_SHARED_EXT) else SHARED2 = $(SHARED1).$(SHARED_MAJOR) SHARED3 = $(SHARED1).$(SHARED_MAJOR).$(SHARED_MINOR) SHARED4 = $(SHARED1).$(SHARED_MAJOR).$(SHARED_MINOR).$(SHARED_PATCH) endif SHARED = $(SHARED1) $(SHARED2) $(SHARED3) $(SHARED4) $(SHARED1): $(SHARED4) ln -fs $(SHARED4) $(SHARED1) $(SHARED2): $(SHARED4) ln -fs $(SHARED4) $(SHARED2) $(SHARED3): $(SHARED4) ln -fs $(SHARED4) $(SHARED3) endif $(SHARED4): $(CXX) $(PLATFORM_SHARED_LDFLAGS)$(SHARED3) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(LIB_SOURCES) $(TOOL_SOURCES) \ $(LDFLAGS) -o $@ endif # PLATFORM_SHARED_EXT .PHONY: blackbox_crash_test check clean coverage crash_test ldb_tests package \ release tags valgrind_check whitebox_crash_test format static_lib shared_lib all \ dbg rocksdbjavastatic rocksdbjava install install-static install-shared uninstall \ analyze tools all: $(LIBRARY) $(BENCHMARKS) tools $(TESTS) static_lib: $(LIBRARY) shared_lib: $(SHARED) tools: $(TOOLS) dbg: $(LIBRARY) $(BENCHMARKS) tools $(TESTS) # creates static library and programs release: $(MAKE) clean DEBUG_LEVEL=0 $(MAKE) static_lib tools db_bench coverage: $(MAKE) clean COVERAGEFLAGS="-fprofile-arcs -ftest-coverage" LDFLAGS+="-lgcov" $(MAKE) J=1 all check cd coverage && ./coverage_test.sh # Delete intermediate files find . -type f -regex ".*\.\(\(gcda\)\|\(gcno\)\)" -exec rm {} \; # Extract the names of its tests by running db_test with --gtest_list_tests. # This filter removes the "#"-introduced comments, and expands to # fully-qualified names by changing input like this: # # DBTest. # Empty # WriteEmptyBatch # MultiThreaded/MultiThreadedDBTest. # MultiThreaded/0 # GetParam() = 0 # MultiThreaded/1 # GetParam() = 1 # # into this: # # DBTest.Empty # DBTest.WriteEmptyBatch # MultiThreaded/MultiThreadedDBTest.MultiThreaded/0 # MultiThreaded/MultiThreadedDBTest.MultiThreaded/1 # test_names = \ ./db_test --gtest_list_tests \ | perl -n \ -e 's/ *\#.*//;' \ -e '/^(\s*)(\S+)/; !$$1 and do {$$p=$$2; break};' \ -e 'print qq! $$p$$2!' ifneq (,$(filter check parallel_check,$(MAKECMDGOALS)),) # Use /dev/shm if it has the sticky bit set (otherwise, /tmp), # and create a randomly-named rocksdb.XXXX directory therein. # We'll use that directory in the "make check" rules. ifeq ($(TMPD),) TMPD := $(shell f=/dev/shm; test -k $$f || f=/tmp; \ perl -le 'use File::Temp "tempdir";' \ -e 'print tempdir("'$$f'/rocksdb.XXXX", CLEANUP => 0)') endif endif ifneq ($(T),) # Run all tests in parallel, accumulating per-test logs in t/log-*. # t_sanitized is each $(T) with "-" in place of each "/". t_sanitized = $(subst /,-,$(T)) # t_run is each sanitized name with a leading "t/". t_run = $(patsubst %,t/%,$(t_sanitized)) # Each t_run file is a tiny generated bourne shell script # that invokes one of db_tests's sub-tests. Why use a file # for this? Because that makes the invocation of parallel # below simpler, which in turn makes the parsing of parallel's # LOG simpler (the latter is for live monitoring as parallel # tests run). filter = --gtest_filter=$(subst -,/,$(@F)) $(t_run): Makefile db_test $(AM_V_GEN)mkdir -p t $(AM_V_at)rm -f $@ $@-t $(AM_V_at)printf '%s\n' \ '#!/bin/sh' \ 'd=$(TMPD)/$(@F)' \ 'mkdir -p $$d' \ 'TEST_TMPDIR=$$d ./db_test $(filter)' \ > $@-t $(AM_V_at)chmod a=rx $@-t $(AM_V_at)mv $@-t $@ # Reorder input lines (which are one per test) so that the # longest-running tests appear first in the output. # Do this by prefixing each selected name with its duration, # sort the resulting names, and remove the leading numbers. # FIXME: the "100" we prepend is a fake time, for now. # FIXME: squirrel away timings from each run and use them # (when present) on subsequent runs to order these tests. # # Without this reordering, these two tests would happen to start only # after almost all other tests had completed, thus adding 100 seconds # to the duration of parallel "make check". That's the difference # between 4 minutes (old) and 2m20s (new). # # 152.120 PASS t/DBTest.FileCreationRandomFailure # 107.816 PASS t/DBTest.EncodeDecompressedBlockSizeTest # slow_test_regexp = \ ^t/DBTest\.(?:FileCreationRandomFailure|EncodeDecompressedBlockSizeTest)$$ prioritize_long_running_tests = \ perl -pe 's,($(slow_test_regexp)),100 $$1,' \ | sort -k1,1gr \ | sed 's/^[.0-9]* //' # "make check" uses # Run with "make J=1 check" to disable parallelism in "make check". # Run with "make J=200% check" to run two parallel jobs per core. # The default is to run one job per core (J=100%). # See "man parallel" for its "-j ..." option. J = 100% # Use this regexp to select the subset of tests whose names match. tests-regexp = . .PHONY: check_0 check_0: $(t_run) $(AM_V_GEN)export TEST_TMPDIR=$(TMPD); \ printf '%s\n' '' \ 'To monitor subtest ,' \ ' run "make watch-log" in a separate window' ''; \ test -t 1 && eta=--eta || eta=; \ { \ printf './%s\n' $(filter-out db_test, $(TESTS)); \ printf '%s\n' $(t_run); \ } \ | $(prioritize_long_running_tests) \ | grep -E '$(tests-regexp)' \ | parallel -j$(J) --joblog=LOG $$eta --gnu '{} >& t/log-{/}' endif CLEAN_FILES += t LOG $(TMPD) # When running parallel "make check", you can monitor its progress # from another window. # Run "make watch_LOG" to show the duration,PASS/FAIL,name of parallel # tests as they are being run. We sort them so that longer-running ones # appear at the top of the list and any failing tests remain at the top # regardless of their duration. As with any use of "watch", hit ^C to # interrupt. watch-log: watch --interval=0 'sort -k7,7nr -k4,4gr LOG|$(quoted_perl_command)' # If J != 1 and GNU parallel is installed, run the tests in parallel, # via the check_0 rule above. Otherwise, run them sequentially. check: all $(AM_V_GEN)if test "$(J)" != 1 \ && (parallel --gnu --help 2>/dev/null) | \ grep -q 'GNU Parallel'; \ then \ t=$$($(test_names)); \ $(MAKE) T="$$t" TMPD=$(TMPD) check_0; \ else \ for t in $(TESTS); do \ echo "===== Running $$t"; ./$$t || exit 1; done; \ fi rm -rf $(TMPD) ifeq ($(filter -DROCKSDB_LITE,$(OPT)),) python tools/ldb_test.py sh tools/rocksdb_dump_test.sh endif check_some: $(SUBSET) ldb_tests for t in $(SUBSET); do echo "===== Running $$t"; ./$$t || exit 1; done .PHONY: ldb_tests ldb_tests: ldb python tools/ldb_test.py crash_test: whitebox_crash_test blackbox_crash_test blackbox_crash_test: db_stress python -u tools/db_crashtest.py --simple blackbox python -u tools/db_crashtest.py blackbox whitebox_crash_test: db_stress python -u tools/db_crashtest.py --simple whitebox python -u tools/db_crashtest.py whitebox asan_check: $(MAKE) clean COMPILE_WITH_ASAN=1 $(MAKE) check -j32 $(MAKE) clean asan_crash_test: $(MAKE) clean COMPILE_WITH_ASAN=1 $(MAKE) crash_test $(MAKE) clean valgrind_check: $(TESTS) for t in $(filter-out %skiplist_test,$(TESTS)); do \ $(VALGRIND_VER) $(VALGRIND_OPTS) ./$$t; \ ret_code=$$?; \ if [ $$ret_code -ne 0 ]; then \ exit $$ret_code; \ fi; \ done ifneq ($(PAR_TEST),) parloop: ret_bad=0; \ for t in $(PAR_TEST); do \ echo "===== Running $$t in parallel $(NUM_PAR)";\ if [ $(db_test) -eq 1 ]; then \ seq $(J) | v="$$t" parallel --gnu 's=$(TMPD)/rdb-{}; export TEST_TMPDIR=$$s;' \ 'timeout 2m ./db_test --gtest_filter=$$v >> $$s/log-{} 2>1'; \ else\ seq $(J) | v="./$$t" parallel --gnu 's=$(TMPD)/rdb-{};' \ 'export TEST_TMPDIR=$$s; timeout 10m $$v >> $$s/log-{} 2>1'; \ fi; \ ret_code=$$?; \ if [ $$ret_code -ne 0 ]; then \ ret_bad=$$ret_code; \ echo $$t exited with $$ret_code; \ fi; \ done; \ exit $$ret_bad; endif all_tests:=$(shell $(test_names)) parallel_check: $(TESTS) $(AM_V_GEN)if test "$(J)" > 1 \ && (parallel --gnu --help 2>/dev/null) | \ grep -q 'GNU Parallel'; \ then \ echo Running in parallel $(J); \ else \ echo "Need to have GNU Parallel and J > 1"; exit 1; \ fi; \ ret_bad=0; \ echo $(J);\ echo Test Dir: $(TMPD); \ seq $(J) | parallel --gnu 's=$(TMPD)/rdb-{}; rm -rf $$s; mkdir $$s'; \ $(MAKE) PAR_TEST="$(all_tests)" TMPD=$(TMPD) \ J=$(J) db_test=1 parloop; \ $(MAKE) PAR_TEST="$(filter-out db_test, $(TESTS))" \ TMPD=$(TMPD) J=$(J) db_test=0 parloop; analyze: clean $(CLANG_SCAN_BUILD) --use-analyzer=$(CLANG_ANALYZER) \ --use-c++=$(CXX) --use-cc=$(CC) --status-bugs \ -o $(CURDIR)/scan_build_report \ $(MAKE) dbg CLEAN_FILES += unity.cc unity.cc: Makefile rm -f $@ $@-t for source_file in $(LIB_SOURCES); do \ echo "#include \"$$source_file\"" >> $@-t; \ done chmod a=r $@-t mv $@-t $@ unity.a: unity.o $(AM_V_AR)rm -f $@ $(AM_V_at)$(AR) $(ARFLAGS) $@ unity.o # try compiling db_test with unity unity_test: db/db_test.o db/db_test_util.o $(TESTHARNESS) unity.a $(AM_LINK) ./unity_test rocksdb.h rocksdb.cc: build_tools/amalgamate.py Makefile $(LIB_SOURCES) unity.cc build_tools/amalgamate.py -I. -i./include unity.cc -x include/rocksdb/c.h -H rocksdb.h -o rocksdb.cc clean: rm -f $(BENCHMARKS) $(TOOLS) $(TESTS) $(LIBRARY) $(SHARED) rm -rf $(CLEAN_FILES) ios-x86 ios-arm scan_build_report find . -name "*.[oda]" -exec rm -f {} \; find . -type f -regex ".*\.\(\(gcda\)\|\(gcno\)\)" -exec rm {} \; rm -rf bzip2* snappy* zlib* lz4* tags: ctags * -R cscope -b `find . -name '*.cc'` `find . -name '*.h'` `find . -name '*.c'` format: build_tools/format-diff.sh package: bash build_tools/make_package.sh $(SHARED_MAJOR).$(SHARED_MINOR) # --------------------------------------------------------------------------- # Unit tests and tools # --------------------------------------------------------------------------- $(LIBRARY): $(LIBOBJECTS) $(AM_V_AR)rm -f $@ $(AM_V_at)$(AR) $(ARFLAGS) $@ $(LIBOBJECTS) db_bench: db/db_bench.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) cache_bench: util/cache_bench.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) memtablerep_bench: db/memtablerep_bench.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) block_hash_index_test: table/block_hash_index_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_stress: tools/db_stress.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) write_stress: tools/write_stress.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) db_sanity_test: tools/db_sanity_test.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) db_repl_stress: tools/db_repl_stress.o $(LIBOBJECTS) $(TESTUTIL) $(AM_LINK) arena_test: util/arena_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) autovector_test: util/autovector_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) column_family_test: db/column_family_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) table_properties_collector_test: db/table_properties_collector_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) bloom_test: util/bloom_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) dynamic_bloom_test: util/dynamic_bloom_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) c_test: db/c_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) cache_test: util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) stringappend_test: utilities/merge_operators/string_append/stringappend_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) redis_test: utilities/redis/redis_lists_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) histogram_test: util/histogram_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) thread_local_test: util/thread_local_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) corruption_test: db/corruption_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) crc32c_test: util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) slice_transform_test: util/slice_transform_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_test: db/db_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_log_iter_test: db/db_log_iter_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_compaction_filter_test: db/db_compaction_filter_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_compaction_test: db/db_compaction_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_dynamic_level_test: db/db_dynamic_level_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_inplace_update_test: db/db_inplace_update_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_tailing_iter_test: db/db_tailing_iter_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_iter_test: db/db_iter_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_universal_compaction_test: db/db_universal_compaction_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_wal_test: db/db_wal_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_properties_test: db/db_properties_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) db_table_properties_test: db/db_table_properties_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) log_write_bench: util/log_write_bench.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) $(pg) plain_table_db_test: db/plain_table_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) comparator_db_test: db/comparator_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) table_reader_bench: table/table_reader_bench.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) $(pg) perf_context_test: db/perf_context_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_V_CCLD)$(CXX) $^ $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) prefix_test: db/prefix_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_V_CCLD)$(CXX) $^ $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) backupable_db_test: utilities/backupable/backupable_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) checkpoint_test: utilities/checkpoint/checkpoint_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) document_db_test: utilities/document/document_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) json_document_test: utilities/document/json_document_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) spatial_db_test: utilities/spatialdb/spatial_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) env_mirror_test: utilities/env_mirror_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) ttl_test: utilities/ttl/ttl_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) write_batch_with_index_test: utilities/write_batch_with_index/write_batch_with_index_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) flush_job_test: db/flush_job_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) compaction_iterator_test: db/compaction_iterator_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) compaction_job_test: db/compaction_job_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) compaction_job_stats_test: db/compaction_job_stats_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) compact_on_deletion_collector_test: utilities/table_properties_collectors/compact_on_deletion_collector_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) wal_manager_test: db/wal_manager_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) dbformat_test: db/dbformat_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) env_test: util/env_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) fault_injection_test: db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) rate_limiter_test: util/rate_limiter_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) delete_scheduler_test: util/delete_scheduler_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) filename_test: db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) file_reader_writer_test: util/file_reader_writer_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) block_based_filter_block_test: table/block_based_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) full_filter_block_test: table/full_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) log_test: db/log_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) table_test: table/table_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) block_test: table/block_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) inlineskiplist_test: db/inlineskiplist_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) skiplist_test: db/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) version_edit_test: db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) version_set_test: db/version_set_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) compaction_picker_test: db/compaction_picker_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) version_builder_test: db/version_builder_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) file_indexer_test: db/file_indexer_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) reduce_levels_test: tools/reduce_levels_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) write_batch_test: db/write_batch_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) write_controller_test: db/write_controller_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) merge_helper_test: db/merge_helper_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) memory_test: utilities/memory/memory_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) merge_test: db/merge_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) merger_test: table/merger_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) options_file_test: db/options_file_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) deletefile_test: db/deletefile_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) geodb_test: utilities/geodb/geodb_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) rocksdb_dump: tools/dump/rocksdb_dump.o $(LIBOBJECTS) $(AM_LINK) rocksdb_undump: tools/dump/rocksdb_undump.o $(LIBOBJECTS) $(AM_LINK) cuckoo_table_builder_test: table/cuckoo_table_builder_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) cuckoo_table_reader_test: table/cuckoo_table_reader_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) cuckoo_table_db_test: db/cuckoo_table_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) listener_test: db/listener_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) thread_list_test: util/thread_list_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) compact_files_test: db/compact_files_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) options_test: util/options_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) options_util_test: utilities/options/options_util_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) event_logger_test: util/event_logger_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) sst_dump_test: tools/sst_dump_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) memenv_test : util/memenv_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) optimistic_transaction_test: utilities/transactions/optimistic_transaction_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) mock_env_test : util/mock_env_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) manual_compaction_test: db/manual_compaction_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) filelock_test: util/filelock_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) auto_roll_logger_test: db/auto_roll_logger_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) memtable_list_test: db/memtable_list_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) write_callback_test: db/write_callback_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) heap_test: util/heap_test.o $(GTEST) $(AM_LINK) transaction_test: utilities/transactions/transaction_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) sst_dump: tools/sst_dump.o $(LIBOBJECTS) $(AM_LINK) ldb_cmd_test: tools/ldb_cmd_test.o $(LIBOBJECTS) $(TESTHARNESS) $(AM_LINK) ldb: tools/ldb.o $(LIBOBJECTS) $(AM_LINK) #------------------------------------------------- # make install related stuff INSTALL_PATH ?= /usr/local uninstall: rm -rf $(INSTALL_PATH)/include/rocksdb \ $(INSTALL_PATH)/lib/$(LIBRARY) \ $(INSTALL_PATH)/lib/$(SHARED4) \ $(INSTALL_PATH)/lib/$(SHARED3) \ $(INSTALL_PATH)/lib/$(SHARED2) \ $(INSTALL_PATH)/lib/$(SHARED1) install-headers: install -d $(INSTALL_PATH)/lib for header_dir in `find "include/rocksdb" -type d`; do \ install -d $(INSTALL_PATH)/$$header_dir; \ done for header in `find "include/rocksdb" -type f -name *.h`; do \ install -C -m 644 $$header $(INSTALL_PATH)/$$header; \ done install-static: install-headers $(LIBRARY) install -C -m 755 $(LIBRARY) $(INSTALL_PATH)/lib install-shared: install-headers $(SHARED4) install -C -m 755 $(SHARED4) $(INSTALL_PATH)/lib && \ ln -fs $(SHARED4) $(INSTALL_PATH)/lib/$(SHARED3) && \ ln -fs $(SHARED4) $(INSTALL_PATH)/lib/$(SHARED2) && \ ln -fs $(SHARED4) $(INSTALL_PATH)/lib/$(SHARED1) # install static by default + install shared if it exists install: install-static [ -e $(SHARED4) ] && $(MAKE) install-shared || : #------------------------------------------------- # --------------------------------------------------------------------------- # Jni stuff # --------------------------------------------------------------------------- JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/linux ifeq ($(PLATFORM), OS_SOLARIS) ARCH := $(shell isainfo -b) else ARCH := $(shell getconf LONG_BIT) endif ROCKSDBJNILIB = librocksdbjni-linux$(ARCH).so ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-linux$(ARCH).jar ROCKSDB_JAR_ALL = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH).jar ROCKSDB_JAVADOCS_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-javadoc.jar ROCKSDB_SOURCES_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-sources.jar ifeq ($(PLATFORM), OS_MACOSX) ROCKSDBJNILIB = librocksdbjni-osx.jnilib ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-osx.jar ifneq ("$(wildcard $(JAVA_HOME)/include/darwin)","") JAVA_INCLUDE = -I$(JAVA_HOME)/include -I $(JAVA_HOME)/include/darwin else JAVA_INCLUDE = -I/System/Library/Frameworks/JavaVM.framework/Headers/ endif endif ifeq ($(PLATFORM), OS_SOLARIS) ROCKSDBJNILIB = librocksdbjni-solaris$(ARCH).so ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-solaris$(ARCH).jar JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/solaris endif libz.a: -rm -rf zlib-1.2.8 curl -O http://zlib.net/zlib-1.2.8.tar.gz tar xvzf zlib-1.2.8.tar.gz cd zlib-1.2.8 && CFLAGS='-fPIC' ./configure --static && make cp zlib-1.2.8/libz.a . libbz2.a: -rm -rf bzip2-1.0.6 curl -O http://www.bzip.org/1.0.6/bzip2-1.0.6.tar.gz tar xvzf bzip2-1.0.6.tar.gz cd bzip2-1.0.6 && make CFLAGS='-fPIC -O2 -g -D_FILE_OFFSET_BITS=64' cp bzip2-1.0.6/libbz2.a . libsnappy.a: -rm -rf snappy-1.1.1 curl -O https://snappy.googlecode.com/files/snappy-1.1.1.tar.gz tar xvzf snappy-1.1.1.tar.gz cd snappy-1.1.1 && ./configure --with-pic --enable-static cd snappy-1.1.1 && make cp snappy-1.1.1/.libs/libsnappy.a . liblz4.a: -rm -rf lz4-r127 curl -O https://codeload.github.com/Cyan4973/lz4/tar.gz/r127 mv r127 lz4-r127.tar.gz tar xvzf lz4-r127.tar.gz cd lz4-r127/lib && make CFLAGS='-fPIC' all cp lz4-r127/lib/liblz4.a . # A version of each $(LIBOBJECTS) compiled with -fPIC and a fixed set of static compression libraries java_static_libobjects = $(patsubst %,jls/%,$(LIBOBJECTS)) CLEAN_FILES += jls JAVA_STATIC_FLAGS = -DZLIB -DBZIP2 -DSNAPPY -DLZ4 JAVA_STATIC_INCLUDES = -I./zlib-1.2.8 -I./bzip2-1.0.6 -I./snappy-1.1.1 -I./lz4-r127/lib $(java_static_libobjects): jls/%.o: %.cc libz.a libbz2.a libsnappy.a liblz4.a $(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) $(JAVA_STATIC_FLAGS) $(JAVA_STATIC_INCLUDES) -fPIC -c $< -o $@ $(COVERAGEFLAGS) rocksdbjavastatic: $(java_static_libobjects) cd java;$(MAKE) javalib; rm -f ./java/target/$(ROCKSDBJNILIB) $(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC \ -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) \ $(java_static_libobjects) $(COVERAGEFLAGS) \ libz.a libbz2.a libsnappy.a liblz4.a $(JAVA_STATIC_LDFLAGS) cd java/target;strip -S -x $(ROCKSDBJNILIB) cd java;jar -cf target/$(ROCKSDB_JAR) HISTORY*.md cd java/target;jar -uf $(ROCKSDB_JAR) $(ROCKSDBJNILIB) cd java/target/classes;jar -uf ../$(ROCKSDB_JAR) org/rocksdb/*.class org/rocksdb/util/*.class cd java/target/apidocs;jar -cf ../$(ROCKSDB_JAVADOCS_JAR) * cd java/src/main/java;jar -cf ../../../target/$(ROCKSDB_SOURCES_JAR) org rocksdbjavastaticrelease: rocksdbjavastatic cd java/crossbuild && vagrant destroy -f && vagrant up linux32 && vagrant halt linux32 && vagrant up linux64 && vagrant halt linux64 cd java;jar -cf target/$(ROCKSDB_JAR_ALL) HISTORY*.md cd java/target;jar -uf $(ROCKSDB_JAR_ALL) librocksdbjni-*.so librocksdbjni-*.jnilib cd java/target/classes;jar -uf ../$(ROCKSDB_JAR_ALL) org/rocksdb/*.class org/rocksdb/util/*.class rocksdbjavastaticpublish: rocksdbjavastaticrelease mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-javadoc.jar -Dclassifier=javadoc mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-sources.jar -Dclassifier=sources mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-linux64.jar -Dclassifier=linux64 mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-linux32.jar -Dclassifier=linux32 mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-osx.jar -Dclassifier=osx mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH).jar # A version of each $(LIBOBJECTS) compiled with -fPIC java_libobjects = $(patsubst %,jl/%,$(LIBOBJECTS)) CLEAN_FILES += jl $(java_libobjects): jl/%.o: %.cc $(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) -fPIC -c $< -o $@ $(COVERAGEFLAGS) rocksdbjava: $(java_libobjects) $(AM_V_GEN)cd java;$(MAKE) javalib; $(AM_V_at)rm -f ./java/target/$(ROCKSDBJNILIB) $(AM_V_at)$(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) $(java_libobjects) $(JAVA_LDFLAGS) $(COVERAGEFLAGS) $(AM_V_at)cd java;jar -cf target/$(ROCKSDB_JAR) HISTORY*.md $(AM_V_at)cd java/target;jar -uf $(ROCKSDB_JAR) $(ROCKSDBJNILIB) $(AM_V_at)cd java/target/classes;jar -uf ../$(ROCKSDB_JAR) org/rocksdb/*.class org/rocksdb/util/*.class jclean: cd java;$(MAKE) clean; jtest: rocksdbjava cd java;$(MAKE) sample;$(MAKE) test; jdb_bench: cd java;$(MAKE) db_bench; commit_prereq: build_tools/rocksdb-lego-determinator \ build_tools/precommit_checker.py build_tools/precommit_checker.py unit unit_481 clang_unit tsan asan lite $(MAKE) clean && $(MAKE) jclean && $(MAKE) rocksdbjava; xfunc: for xftest in $(XFUNC_TESTS); do \ echo "===== Running xftest $$xftest"; \ make check ROCKSDB_XFUNC_TEST="$$xftest" tests-regexp="DBTest" ;\ done # --------------------------------------------------------------------------- # Platform-specific compilation # --------------------------------------------------------------------------- ifeq ($(PLATFORM), IOS) # For iOS, create universal object files to be used on both the simulator and # a device. PLATFORMSROOT=/Applications/Xcode.app/Contents/Developer/Platforms SIMULATORROOT=$(PLATFORMSROOT)/iPhoneSimulator.platform/Developer DEVICEROOT=$(PLATFORMSROOT)/iPhoneOS.platform/Developer IOSVERSION=$(shell defaults read $(PLATFORMSROOT)/iPhoneOS.platform/version CFBundleShortVersionString) .cc.o: mkdir -p ios-x86/$(dir $@) $(CXX) $(CXXFLAGS) -isysroot $(SIMULATORROOT)/SDKs/iPhoneSimulator$(IOSVERSION).sdk -arch i686 -arch x86_64 -c $< -o ios-x86/$@ mkdir -p ios-arm/$(dir $@) xcrun -sdk iphoneos $(CXX) $(CXXFLAGS) -isysroot $(DEVICEROOT)/SDKs/iPhoneOS$(IOSVERSION).sdk -arch armv6 -arch armv7 -arch armv7s -arch arm64 -c $< -o ios-arm/$@ lipo ios-x86/$@ ios-arm/$@ -create -output $@ .c.o: mkdir -p ios-x86/$(dir $@) $(CC) $(CFLAGS) -isysroot $(SIMULATORROOT)/SDKs/iPhoneSimulator$(IOSVERSION).sdk -arch i686 -arch x86_64 -c $< -o ios-x86/$@ mkdir -p ios-arm/$(dir $@) xcrun -sdk iphoneos $(CC) $(CFLAGS) -isysroot $(DEVICEROOT)/SDKs/iPhoneOS$(IOSVERSION).sdk -arch armv6 -arch armv7 -arch armv7s -arch arm64 -c $< -o ios-arm/$@ lipo ios-x86/$@ ios-arm/$@ -create -output $@ else .cc.o: $(AM_V_CC)$(CXX) $(CXXFLAGS) -c $< -o $@ $(COVERAGEFLAGS) .c.o: $(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@ endif # --------------------------------------------------------------------------- # Source files dependencies detection # --------------------------------------------------------------------------- all_sources = $(LIB_SOURCES) $(TEST_BENCH_SOURCES) $(MOCK_SOURCES) DEPFILES = $(all_sources:.cc=.d) # Add proper dependency support so changing a .h file forces a .cc file to # rebuild. # The .d file indicates .cc file's dependencies on .h files. We generate such # dependency by g++'s -MM option, whose output is a make dependency rule. $(DEPFILES): %.d: %.cc @$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) \ -MM -MT'$@' -MT'$(<:.cc=.o)' "$<" -o '$@' depend: $(DEPFILES) # if the make goal is either "clean" or "format", we shouldn't # try to import the *.d files. # TODO(kailiu) The unfamiliarity of Make's conditions leads to the ugly # working solution. ifneq ($(MAKECMDGOALS),clean) ifneq ($(MAKECMDGOALS),format) ifneq ($(MAKECMDGOALS),jclean) ifneq ($(MAKECMDGOALS),jtest) ifneq ($(MAKECMDGOALS),package) ifneq ($(MAKECMDGOALS),analyze) -include $(DEPFILES) endif endif endif endif endif endif rocksdb-4.5.1/PATENTS000066400000000000000000000036721267530634000142270ustar00rootroot00000000000000Additional Grant of Patent Rights Version 2 "Software" means the RocksDB software distributed by Facebook, Inc. Facebook, Inc. ("Facebook") hereby grants to each recipient of the Software ("you") a perpetual, worldwide, royalty-free, non-exclusive, irrevocable (subject to the termination provision below) license under any Necessary Claims, to make, have made, use, sell, offer to sell, import, and otherwise transfer the Software. For avoidance of doubt, no license is granted under Facebook’s rights in any patent claims that are infringed by (i) modifications to the Software made by you or any third party or (ii) the Software in combination with any software or other technology. The license granted hereunder will terminate, automatically and without notice, if you (or any of your subsidiaries, corporate affiliates or agents) initiate directly or indirectly, or take a direct financial interest in, any Patent Assertion: (i) against Facebook or any of its subsidiaries or corporate affiliates, (ii) against any party if such Patent Assertion arises in whole or in part from any software, technology, product or service of Facebook or any of its subsidiaries or corporate affiliates, or (iii) against any party relating to the Software. Notwithstanding the foregoing, if Facebook or any of its subsidiaries or corporate affiliates files a lawsuit alleging patent infringement against you in the first instance, and you respond by filing a patent infringement counterclaim in that lawsuit against that party that is unrelated to the Software, the license granted hereunder will not terminate under section (i) of this paragraph due to such counterclaim. A "Necessary Claim" is a claim of a patent owned by Facebook that is necessarily infringed by the Software standing alone. A "Patent Assertion" is any lawsuit or other action alleging direct, indirect, or contributory infringement or inducement to infringe any patent, including a cross-claim or counterclaim. rocksdb-4.5.1/README.md000066400000000000000000000023641267530634000144420ustar00rootroot00000000000000## RocksDB: A Persistent Key-Value Store for Flash and RAM Storage [![Build Status](https://travis-ci.org/facebook/rocksdb.svg?branch=master)](https://travis-ci.org/facebook/rocksdb) RocksDB is developed and maintained by Facebook Database Engineering Team. It is built on earlier work on LevelDB by Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com) This code is a library that forms the core building block for a fast key value server, especially suited for storing data on flash drives. It has a Log-Structured-Merge-Database (LSM) design with flexible tradeoffs between Write-Amplification-Factor (WAF), Read-Amplification-Factor (RAF) and Space-Amplification-Factor (SAF). It has multi-threaded compactions, making it specially suitable for storing multiple terabytes of data in a single database. Start with example usage here: https://github.com/facebook/rocksdb/tree/master/examples See the [github wiki](https://github.com/facebook/rocksdb/wiki) for more explanation. The public interface is in `include/`. Callers should not include or rely on the details of any other header files in this package. Those internal APIs may be changed without warning. Design discussions are conducted in https://www.facebook.com/groups/rocksdb.dev/ rocksdb-4.5.1/ROCKSDB_LITE.md000066400000000000000000000020161267530634000154030ustar00rootroot00000000000000# RocksDBLite RocksDBLite is a project focused on mobile use cases, which don't need a lot of fancy things we've built for server workloads and they are very sensitive to binary size. For that reason, we added a compile flag ROCKSDB_LITE that comments out a lot of the nonessential code and keeps the binary lean. Some examples of the features disabled by ROCKSDB_LITE: * compiled-in support for LDB tool * No backupable DB * No support for replication (which we provide in form of TrasactionalIterator) * No advanced monitoring tools * No special-purpose memtables that are highly optimized for specific use cases * No Transactions When adding a new big feature to RocksDB, please add ROCKSDB_LITE compile guard if: * Nobody from mobile really needs your feature, * Your feature is adding a lot of weight to the binary. Don't add ROCKSDB_LITE compile guard if: * It would introduce a lot of code complexity. Compile guards make code harder to read. It's a trade-off. * Your feature is not adding a lot of weight. If unsure, ask. :) rocksdb-4.5.1/USERS.md000066400000000000000000000055261267530634000144110ustar00rootroot00000000000000This document lists users of RocksDB and their use cases. If you are using RocksDB, please open a pull request and add yourself to the list. ## Facebook At Facebook, we use RocksDB as a backend for many different stateful services. We're also experimenting with running RocksDB as a storage engine for two databases: 1. MyRocks -- https://github.com/MySQLOnRocksDB/mysql-5.6 2. MongoRocks -- https://github.com/mongodb-partners/mongo-rocks ## LinkedIn Two different use cases at Linkedin are using RocksDB as a storage engine: 1. LinkedIn's follow feed for storing user's activities 2. Apache Samza, open source framework for stream processing Learn more about those use cases in a Tech Talk by Ankit Gupta and Naveen Somasundaram: http://www.youtube.com/watch?v=plqVp_OnSzg ## Yahoo Yahoo is using RocksDB as a storage engine for their biggest distributed data store Sherpa. Learn more about it here: http://yahooeng.tumblr.com/post/120730204806/sherpa-scales-new-heights ## CockroachDB CockroachDB is an open-source geo-replicated transactional database (still in development). They are using RocksDB as their storage engine. Check out their github: https://github.com/cockroachdb/cockroach ## DNANexus DNANexus is using RocksDB to speed up processing of genomics data. You can learn more from this great blog post by Mike Lin: http://devblog.dnanexus.com/faster-bam-sorting-with-samtools-and-rocksdb/ ## Iron.io Iron.io is using RocksDB as a storage engine for their distributed queueing system. Learn more from Tech Talk by Reed Allman: http://www.youtube.com/watch?v=HTjt6oj-RL4 ## Tango Me Tango is using RocksDB as a graph storage to store all users' connection data and other social activity data. ## Turn Turn is using RocksDB as a storage layer for their key/value store, serving at peak 2.4MM QPS out of different datacenters. Check out our RocksDB Protobuf merge operator at: https://github.com/vladb38/rocksdb_protobuf ## Santanader UK/Cloudera Profession Services Check out their blog post: http://blog.cloudera.com/blog/2015/08/inside-santanders-near-real-time-data-ingest-architecture/ ## Airbnb Airbnb is using RocksDB as a storage engine for their personalized search service. You can learn more about it here: https://www.youtube.com/watch?v=ASQ6XMtogMs ## Pinterest Pinterest's Object Retrieval System uses RocksDB for storage: https://www.youtube.com/watch?v=MtFEVEs_2Vo ## Smyte [Smyte](https://www.smyte.com/) uses RocksDB as the storage layer for their core key-value storage, high-performance counters and time-windowed HyperLogLog services. ## Rakuten Marketing [Rakuten Marketing](https://marketing.rakuten.com/) uses RocksDB as the disk cache layer for the real-time bidding service in their Performance DSP. ## VWO, Wingify [VWO's](https://vwo.com/) Smart Code checker and URL helper uses RocksDB to store all the URLs where VWO's Smart Code is installed. rocksdb-4.5.1/Vagrantfile000066400000000000000000000015361267530634000153500ustar00rootroot00000000000000Vagrant.configure("2") do |config| config.vm.provider "virtualbox" do |v| v.memory = 4096 v.cpus = 2 end config.vm.define "ubuntu14" do |box| box.vm.box = "ubuntu/trusty64" end config.vm.define "centos65" do |box| box.vm.box = "chef/centos-6.5" end config.vm.define "FreeBSD10" do |box| box.vm.guest = :freebsd box.vm.box = "robin/freebsd-10" # FreeBSD does not support 'mount_virtualbox_shared_folder', use NFS box.vm.synced_folder ".", "/vagrant", :nfs => true, id: "vagrant-root" box.vm.network "private_network", ip: "10.0.1.10" # build everything after creating VM, skip using --no-provision box.vm.provision "shell", inline: <<-SCRIPT pkg install -y gmake clang35 export CXX=/usr/local/bin/clang++35 cd /vagrant gmake clean gmake all OPT=-g SCRIPT end end rocksdb-4.5.1/WINDOWS_PORT.md000066400000000000000000000310431267530634000154770ustar00rootroot00000000000000# Microsoft Contribution Notes ## Contributors * Alexander Zinoviev https://github.com/zinoale * Dmitri Smirnov https://github.com/yuslepukhin * Praveen Rao https://github.com/PraveenSinghRao * Sherlock Huang https://github.com/SherlockNoMad ## Introduction RocksDB is a well proven open source key-value persistent store, optimized for fast storage. It provides scalability with number of CPUs and storage IOPS, to support IO-bound, in-memory and write-once workloads, most importantly, to be flexible to allow for innovation. As Microsoft Bing team we have been continuously pushing hard to improve the scalability, efficiency of platform and eventually benefit Bing end-user satisfaction. We would like to explore the opportunity to embrace open source, RocksDB here, to use, enhance and customize for our usage, and also contribute back to the RocksDB community. Herein, we are pleased to offer this RocksDB port for Windows platform. These notes describe some decisions and changes we had to make with regards to porting RocksDB on Windows. We hope this will help both reviewers and users of the Windows port. We are open for comments and improvements. ## OS specifics All of the porting, testing and benchmarking was done on Windows Server 2012 R2 Datacenter 64-bit but to the best of our knowledge there is not a specific API we used during porting that is unsupported on other Windows OS after Vista. ## Porting goals We strive to achieve the following goals: * make use of the existing porting interface of RocksDB * make minimum [WY2]modifications within platform independent code. * make all unit test pass both in debug and release builds. * Note: latest introduction of SyncPoint seems to disable running db_test in Release. * make performance on par with published benchmarks accounting for HW differences * we would like to keep the port code inline with the master branch with no forking ## Build system We have chosen CMake as a widely accepted build system to build the Windows port. It is very fast and convenient. At the same time it generates Visual Studio projects that are both usable from a command line and IDE. The top-level CMakeLists.txt file contains description of all targets and build rules. It also provides brief instructions on how to build the software for Windows. One more build related file is thirdparty.inc that also resides on the top level. This file must be edited to point to actual third party libraries location. We think that it would be beneficial to merge the existing make-based build system and the new cmake-based build system into a single one to use on all platforms. All building and testing was done for 64-bit. We have not conducted any testing for 32-bit and early reports indicate that it will not run on 32-bit. ## C++ and STL notes We had to make some minimum changes within the portable files that either account for OS differences or the shortcomings of C++11 support in the current version of the MS compiler. Most or all of them are expected to be fixed in the upcoming compiler releases. We plan to use this port for our business purposes here at Bing and this provided business justification for this port. This also means, we do not have at present to choose the compiler version at will. * Certain headers that are not present and not necessary on Windows were simply `#ifndef OS_WIN` in a few places (`unistd.h`) * All posix specific headers were replaced to port/port.h which worked well * Replaced `dirent.h` for `port/dirent.h` (very few places) with the implementation of the relevant interfaces within `rocksdb::port` namespace * Replaced `sys/time.h` to `port/sys_time.h` (few places) implemented equivalents within `rocksdb::port` * `printf %z` specification is not supported on Windows. To imitate existing standards we came up with a string macro `ROCKSDB_PRIszt` which expands to `%z` on posix systems and to Iu on windows. * in class member initialization were moved to a __ctors in some cases * `constexpr` is not supported. We had to replace `std::numeric_limits<>::max/min()` to its C macros for constants. Sometimes we had to make class members `static const` and place a definition within a .cc file. * `constexpr` for functions was replaced to a template specialization (1 place) * Union members that have non-trivial constructors were replaced to `char[]` in one place along with bug fixes (spatial experimental feature) * Zero-sized arrays are deemed a non-standard extension which we converted to 1 size array and that should work well for the purposes of these classes. * `std::chrono` lacks nanoseconds support (fixed in the upcoming release of the STL) and we had to use `QueryPerfCounter()` within env_win.cc * Function local statics initialization is still not safe. Used `std::once` to mitigate within WinEnv. ## Windows Environments notes We endeavored to make it functionally on par with posix_env. This means we replicated the functionality of the thread pool and other things as precise as possible, including: * Replicate posix logic using std:thread primitives. * Implement all posix_env disk access functionality. * Set `use_os_buffer=false` to disable OS disk buffering for WinWritableFile and WinRandomAccessFile. * Replace `pread/pwrite` with `WriteFile/ReadFile` with `OVERLAPPED` structure. * Use `SetFileInformationByHandle` to compensate absence of `fallocate`. ### In detail Even though Windows provides its own efficient thread-pool implementation we chose to replicate posix logic using `std::thread` primitives. This allows anyone to quickly detect any changes within the posix source code and replicate them within windows env. This has proven to work very well. At the same time for anyone who wishes to replace the built-in thread-pool can do so using RocksDB stackable environments. For disk access we implemented all of the functionality present within the posix_env which includes memory mapped files, random access, rate-limiter support etc. The `use_os_buffer` flag on Posix platforms currently denotes disabling read-ahead log via `fadvise` mechanism. Windows does not have `fadvise` system call. What is more, it implements disk cache in a way that differs from Linux greatly. It’s not an uncommon practice on Windows to perform un-buffered disk access to gain control of the memory consumption. We think that in our use case this may also be a good configuration option at the expense of disk throughput. To compensate one may increase the configured in-memory cache size instead. Thus we have chosen `use_os_buffer=false` to disable OS disk buffering for `WinWritableFile` and `WinRandomAccessFile`. The OS imposes restrictions on the alignment of the disk offsets, buffers used and the amount of data that is read/written when accessing files in un-buffered mode. When the option is true, the classes behave in a standard way. This allows to perform writes and reads in cases when un-buffered access does not make sense such as WAL and MANIFEST. We have replaced `pread/pwrite` with `WriteFile/ReadFile` with `OVERLAPPED` structure so we can atomically seek to the position of the disk operation but still perform the operation synchronously. Thus we able to emulate that functionality of `pread/pwrite` reasonably well. The only difference is that the file pointer is not returned to its original position but that hardly matters given the random nature of access. We used `SetFileInformationByHandle` both to truncate files after writing a full final page to disk and to pre-allocate disk space for faster I/O thus compensating for the absence of `fallocate` although some differences remain. For example, the pre-allocated space is not filled with zeros like on Linux, however, on a positive note, the end of file position is also not modified after pre-allocation. RocksDB renames, copies and deletes files at will even though they may be opened with another handle at the same time. We had to relax and allow nearly all the concurrent access permissions possible. ## Thread-Local Storage Thread-Local storage plays a significant role for RocksDB performance. Rather than creating a separate implementation we chose to create inline wrappers that forward `pthread_specific` calls to Windows `Tls` interfaces within `rocksdb::port` namespace. This leaves the existing meat of the logic in tact and unchanged and just as maintainable. To mitigate the lack of thread local storage cleanup on thread-exit we added a limited amount of windows specific code within the same thread_local.cc file that injects a cleanup callback into a `"__tls"` structure within `".CRT$XLB"` data segment. This approach guarantees that the callback is invoked regardless of whether RocksDB used within an executable, standalone DLL or within another DLL. ## Jemalloc usage When RocksDB is used with Jemalloc the latter needs to be initialized before any of the C++ globals or statics. To accomplish that we injected an initialization routine into `".CRT$XCT"` that is automatically invoked by the runtime before initializing static objects. je-uninit is queued to `atexit()`. The jemalloc redirecting `new/delete` global operators are used by the linker providing certain conditions are met. See build section in these notes. ## Stack Trace and Unhandled Exception Handler We decided not to implement these two features because the hosting program as a rule has these two things in it. We experienced no inconveniences debugging issues in the debugger or analyzing process dumps if need be and thus we did not see this as a priority. ## Performance results ### Setup All of the benchmarks are run on the same set of machines. Here are the details of the test setup: * 2 Intel(R) Xeon(R) E5 2450 0 @ 2.10 GHz (total 16 cores) * 2 XK0480GDQPH SSD Device, total 894GB free disk * Machine has 128 GB of RAM * Operating System: Windows Server 2012 R2 Datacenter * 100 Million keys; each key is of size 10 bytes, each value is of size 800 bytes * total database size is ~76GB * The performance result is based on RocksDB 3.11. * The parameters used, unless specified, were exactly the same as published in the GitHub Wiki page. ### RocksDB on flash storage #### Test 1. Bulk Load of keys in Random Order Version 3.11 * Total Run Time: 17.6 min * Fillrandom: 5.480 micros/op 182465 ops/sec; 142.0 MB/s * Compact: 486056544.000 micros/op 0 ops/sec Version 3.10 * Total Run Time: 16.2 min * Fillrandom: 5.018 micros/op 199269 ops/sec; 155.1 MB/s * Compact: 441313173.000 micros/op 0 ops/sec; #### Test 2. Bulk Load of keys in Sequential Order Version 3.11 * Fillseq: 4.944 micros/op 202k ops/sec; 157.4 MB/s Version 3.10 * Fillseq: 4.105 micros/op 243.6k ops/sec; 189.6 MB/s #### Test 3. Random Write Version 3.11 * Unbuffered I/O enabled * Overwrite: 52.661 micros/op 18.9k ops/sec; 14.8 MB/s Version 3.10 * Unbuffered I/O enabled * Overwrite: 52.661 micros/op 18.9k ops/sec; #### Test 4. Random Read Version 3.11 * Unbuffered I/O enabled * Readrandom: 15.716 micros/op 63.6k ops/sec; 49.5 MB/s Version 3.10 * Unbuffered I/O enabled * Readrandom: 15.548 micros/op 64.3k ops/sec; #### Test 5. Multi-threaded read and single-threaded write Version 3.11 * Unbuffered I/O enabled * Readwhilewriting: 25.128 micros/op 39.7k ops/sec; Version 3.10 * Unbuffered I/O enabled * Readwhilewriting: 24.854 micros/op 40.2k ops/sec; ### RocksDB In Memory #### Test 1. Point Lookup Version 3.11 80K writes/sec * Write Rate Achieved: 40.5k write/sec; * Readwhilewriting: 0.314 micros/op 3187455 ops/sec; 364.8 MB/s (715454999 of 715454999 found) Version 3.10 * Write Rate Achieved: 50.6k write/sec * Readwhilewriting: 0.316 micros/op 3162028 ops/sec; (719576999 of 719576999 found) *10K writes/sec* Version 3.11 * Write Rate Achieved: 5.8k/s write/sec * Readwhilewriting: 0.246 micros/op 4062669 ops/sec; 464.9 MB/s (915481999 of 915481999 found) Version 3.10 * Write Rate Achieved: 5.8k/s write/sec * Readwhilewriting: 0.244 micros/op 4106253 ops/sec; (927986999 of 927986999 found) #### Test 2. Prefix Range Query Version 3.11 80K writes/sec * Write Rate Achieved: 46.3k/s write/sec * Readwhilewriting: 0.362 micros/op 2765052 ops/sec; 316.4 MB/s (611549999 of 611549999 found) Version 3.10 * Write Rate Achieved: 45.8k/s write/sec * Readwhilewriting: 0.317 micros/op 3154941 ops/sec; (708158999 of 708158999 found) Version 3.11 10K writes/sec * Write Rate Achieved: 5.78k write/sec * Readwhilewriting: 0.269 micros/op 3716692 ops/sec; 425.3 MB/s (837401999 of 837401999 found) Version 3.10 * Write Rate Achieved: 5.7k write/sec * Readwhilewriting: 0.261 micros/op 3830152 ops/sec; (863482999 of 863482999 found) We think that there is still big room to improve the performance, which will be an ongoing effort for us. rocksdb-4.5.1/appveyor.yml000066400000000000000000000007161267530634000155520ustar00rootroot00000000000000version: 1.0.{build} before_build: - md %APPVEYOR_BUILD_FOLDER%\build - cd %APPVEYOR_BUILD_FOLDER%\build - cmake -G "Visual Studio 12 Win64" -DOPTDBG=1 .. - cd .. build: project: build\rocksdb.sln parallel: true verbosity: minimal test: test_script: - ps: build_tools\run_ci_db_test.ps1 -EnableRerun -Run db_test -Exclude DBTest.Randomized,DBTest.FileCreationRandomFailure -Concurrency 18 - ps: build_tools\run_ci_db_test.ps1 -Run env_test -Concurrency 1 rocksdb-4.5.1/arcanist_util/000077500000000000000000000000001267530634000160175ustar00rootroot00000000000000rocksdb-4.5.1/arcanist_util/__phutil_library_init__.php000066400000000000000000000000731267530634000234000ustar00rootroot00000000000000 2, 'class' => array( 'ArcanistCpplintLinter' => 'cpp_linter/ArcanistCpplintLinter.php', 'BaseDirectoryScopedFormatLinter' => 'cpp_linter/BaseDirectoryScopedFormatLinter.php', 'FacebookArcanistConfiguration' => 'config/FacebookArcanistConfiguration.php', 'FacebookFbcodeLintEngine' => 'lint_engine/FacebookFbcodeLintEngine.php', 'FacebookFbcodeUnitTestEngine' => 'unit_engine/FacebookFbcodeUnitTestEngine.php', 'FacebookHowtoevenLintEngine' => 'lint_engine/FacebookHowtoevenLintEngine.php', 'FacebookHowtoevenLinter' => 'cpp_linter/FacebookHowtoevenLinter.php', 'FbcodeClangFormatLinter' => 'cpp_linter/FbcodeClangFormatLinter.php', 'FbcodeCppLinter' => 'cpp_linter/FbcodeCppLinter.php', ), 'function' => array( ), 'xmap' => array( 'ArcanistCpplintLinter' => 'ArcanistLinter', 'BaseDirectoryScopedFormatLinter' => 'ArcanistLinter', 'FacebookArcanistConfiguration' => 'ArcanistConfiguration', 'FacebookFbcodeLintEngine' => 'ArcanistLintEngine', 'FacebookFbcodeUnitTestEngine' => 'ArcanistBaseUnitTestEngine', 'FacebookHowtoevenLintEngine' => 'ArcanistLintEngine', 'FacebookHowtoevenLinter' => 'ArcanistLinter', 'FbcodeClangFormatLinter' => 'BaseDirectoryScopedFormatLinter', 'FbcodeCppLinter' => 'ArcanistLinter', ), )); rocksdb-4.5.1/arcanist_util/config/000077500000000000000000000000001267530634000172645ustar00rootroot00000000000000rocksdb-4.5.1/arcanist_util/config/FacebookArcanistConfiguration.php000066400000000000000000000156371267530634000257370ustar00rootroot00000000000000isRawDiffSource()) { $this->startTestsInJenkins($workflow); $this->startTestsInSandcastle($workflow); } } ////////////////////////////////////////////////////////////////////// /* Run tests in sandcastle */ function postURL($diffID, $url) { $cmd = 'echo \'{"diff_id": "' . $diffID . '", ' . '"name":"click here for sandcastle tests for D' . $diffID . '", ' . '"link":"' . $url . '"}\' | ' . 'http_proxy=fwdproxy.any.facebook.com:8080 ' . 'https_proxy=fwdproxy.any.facebook.com:8080 arc call-conduit ' . 'differential.updateunitresults'; shell_exec($cmd); } function updateTestCommand($diffID, $test, $status) { $cmd = 'echo \'{"diff_id": "' . $diffID . '", ' . '"name":"' . $test . '", ' . '"result":"' . $status . '"}\' | ' . 'http_proxy=fwdproxy.any.facebook.com:8080 ' . 'https_proxy=fwdproxy.any.facebook.com:8080 arc call-conduit ' . 'differential.updateunitresults'; return $cmd; } function updateTest($diffID, $test) { shell_exec($this->updateTestCommand($diffID, $test, "waiting")); } function getSteps($diffID, $username, $test) { $arcrc_content = exec("cat ~/.arcrc | gzip -f | base64 -w0"); // Sandcastle machines don't have arc setup. We copy the user certificate // and authenticate using that in sandcastle $setup = array( "name" => "Setup arcrc", "shell" => "echo " . $arcrc_content . " | base64 --decode" . " | gzip -d > ~/.arcrc", "user" => "root" ); // arc demands certain permission on its config $fix_permission = array( "name" => "Fix environment", "shell" => "chmod 600 ~/.arcrc", "user" => "root" ); // fbcode is a sub-repo. We cannot patch until we add it to ignore otherwise // git thinks it is uncommited change $fix_git_ignore = array( "name" => "Fix git ignore", "shell" => "echo fbcode >> .git/info/exclude", "user" => "root" ); // Patch the code (keep your fingures crossed) $patch = array( "name" => "Patch " . $diffID, "shell" => "HTTPS_PROXY=fwdproxy:8080 arc --arcrc-file ~/.arcrc " . "patch --diff " . $diffID, "user" => "root" ); // Clean up the user arc config we are using $cleanup = array( "name" => "Arc cleanup", "shell" => "rm -f ~/.arcrc", "user" => "root" ); // Construct the steps in the order of execution $steps[] = $setup; $steps[] = $fix_permission; $steps[] = $fix_git_ignore; $steps[] = $patch; // Run the actual command $this->updateTest($diffID, $test); $cmd = $this->updateTestCommand($diffID, $test, "running") . ";" . "(./build_tools/precommit_checker.py " . $test . "&& " . $this->updateTestCommand($diffID, $test, "pass") . ")" . "|| " . $this->updateTestCommand($diffID, $test, "fail") . "; cat /tmp/precommit-check.log" . "; for f in `ls t/log-*`; do echo \$f; cat \$f; done"; $run_test = array( "name" => "Run " . $test, "shell" => $cmd, "user" => "root", ); $steps[] = $run_test; $steps[] = $cleanup; return $steps; } function startTestsInSandcastle($workflow) { // extract information we need from workflow or CLI $diffID = $workflow->getDiffId(); $username = exec("whoami"); if ($diffID == null || $username == null) { // there is no diff and we can't extract username // we cannot schedule sandcasstle job return; } if (strcmp(getenv("ROCKSDB_CHECK_ALL"), 1) == 0) { // extract all tests from the CI definition $output = file_get_contents("build_tools/rocksdb-lego-determinator"); preg_match_all('/[ ]{2}([a-zA-Z0-9_]+)[\)]{1}/', $output, $matches); $tests = $matches[1]; } else { // manually list of tests we want to run in sandcastle $tests = array( "unit", "unit_481", "clang_unit", "tsan", "asan", "lite", "valgrind" ); } // construct a job definition for each test and add it to the master plan foreach ($tests as $test) { $arg[] = array( "name" => "RocksDB diff " . $diffID . " test " . $test, "steps" => $this->getSteps($diffID, $username, $test) ); } // we cannot submit the parallel execution master plan to sandcastle // we need supply the job plan as a determinator // so we construct a small job that will spit out the master job plan // which sandcastle will parse and execute // Why compress ? Otherwise we run over the max string size. $cmd = "echo " . base64_encode(json_encode($arg)) . " | gzip -f | base64 -w0"; $arg_encoded = shell_exec($cmd); $command = array( "name" => "Run diff " . $diffID . "for user " . $username, "steps" => array() ); $command["steps"][] = array( "name" => "Generate determinator", "shell" => "echo " . $arg_encoded . " | base64 --decode | gzip -d" . " | base64 --decode", "determinator" => true, "user" => "root" ); // submit to sandcastle $url = 'https://interngraph.intern.facebook.com/sandcastle/generate?' .'command=SandcastleUniversalCommand' .'&vcs=rocksdb-git&revision=origin%2Fmaster&type=lego' .'&user=krad&alias=rocksdb-precommit' .'&command-args=' . urlencode(json_encode($command)); $cmd = 'https_proxy= HTTPS_PROXY= curl -s -k -F app=659387027470559 ' . '-F token=AeO_3f2Ya3TujjnxGD4 "' . $url . '"'; $output = shell_exec($cmd); // extract sandcastle URL from the response preg_match('/url": "(.+)"/', $output, $sandcastle_url); echo "\nSandcastle URL: " . $sandcastle_url[1] . "\n"; // Ask phabricator to display it on the diff UI $this->postURL($diffID, $sandcastle_url[1]); } ////////////////////////////////////////////////////////////////////// /* Send off builds to jenkins */ function startTestsInJenkins($workflow) { $diffID = $workflow->getDiffID(); if ($diffID === null) { return; } $results = $workflow->getTestResults(); if (!$results) { return; } $url = "https://ci-builds.fb.com/view/rocksdb/job/rocksdb_diff_check/" ."buildWithParameters?token=AUTH&DIFF_ID=$diffID"; system("curl --noproxy '*' \"$url\" > /dev/null 2>&1"); } } rocksdb-4.5.1/arcanist_util/cpp_linter/000077500000000000000000000000001267530634000201565ustar00rootroot00000000000000rocksdb-4.5.1/arcanist_util/cpp_linter/ArcanistCpplintLinter.php000066400000000000000000000047341267530634000251530ustar00rootroot00000000000000linterDir(), $bin); if (!$err) { return $this->linterDir().'/'.$bin; } // Look for globally installed cpplint.py list($err) = exec_manual('which %s', $bin); if ($err) { throw new ArcanistUsageException( "cpplint.py does not appear to be installed on this system. Install ". "it (e.g., with 'wget \"http://google-styleguide.googlecode.com/". "svn/trunk/cpplint/cpplint.py\"') ". "in your .arcconfig to point to the directory where it resides. ". "Also don't forget to chmod a+x cpplint.py!"); } return $bin; } public function lintPath($path) { $bin = $this->getLintPath(); $path = $this->rocksdbDir().'/'.$path; $f = new ExecFuture("%C $path", $bin); list($err, $stdout, $stderr) = $f->resolve(); if ($err === 2) { throw new Exception("cpplint failed to run correctly:\n".$stderr); } $lines = explode("\n", $stderr); $messages = array(); foreach ($lines as $line) { $line = trim($line); $matches = null; $regex = '/^[^:]+:(\d+):\s*(.*)\s*\[(.*)\] \[(\d+)\]$/'; if (!preg_match($regex, $line, $matches)) { continue; } foreach ($matches as $key => $match) { $matches[$key] = trim($match); } $message = new ArcanistLintMessage(); $message->setPath($path); $message->setLine($matches[1]); $message->setCode($matches[3]); $message->setName($matches[3]); $message->setDescription($matches[2]); $message->setSeverity(ArcanistLintSeverity::SEVERITY_WARNING); $this->addLintMessage($message); } } // The path of this linter private function linterDir() { return dirname(__FILE__); } // TODO(kaili) a quick and dirty way to figure out rocksdb's root dir. private function rocksdbDir() { return $this->linterDir()."/../.."; } } rocksdb-4.5.1/arcanist_util/cpp_linter/BaseDirectoryScopedFormatLinter.php000066400000000000000000000040171267530634000271150ustar00rootroot00000000000000getPathsToLint() as $p) { // check if $path starts with $p if (strncmp($path, $p, strlen($p)) === 0) { return true; } } return false; } // API to tell this linter which lines were changed final public function setPathChangedLines($path, $changed) { $this->changedLines[$path] = $changed; } final public function willLintPaths(array $paths) { $futures = array(); foreach ($paths as $path) { if (!$this->shouldLintPath($path)) { continue; } $changed = $this->changedLines[$path]; if (!isset($changed)) { // do not run linter if there are no changes continue; } $futures[$path] = $this->getFormatFuture($path, $changed); } foreach (Futures($futures)->limit(8) as $p => $f) { $this->rawLintOutput[$p] = $f->resolvex(); } } abstract protected function getFormatFuture($path, array $changed); abstract protected function getLintMessage($diff); final public function lintPath($path) { if (!isset($this->rawLintOutput[$path])) { return; } list($new_content) = $this->rawLintOutput[$path]; $old_content = $this->getData($path); if ($new_content != $old_content) { $diff = ArcanistDiffUtils::renderDifferences($old_content, $new_content); $this->raiseLintAtOffset( 0, self::LINT_FORMATTING, $this->getLintMessage($diff), $old_content, $new_content); } } } rocksdb-4.5.1/arcanist_util/cpp_linter/FacebookHowtoevenLinter.php000066400000000000000000000146131267530634000254620ustar00rootroot00000000000000localExecx("rm -rf _build/_lint"); // Build compilation database. $lintable_paths = $this->getLintablePaths($paths); $interesting_paths = $this->getInterestingPaths($lintable_paths); if (!$lintable_paths) { return; } // Run lint. try { $this->localExecx( "%C %C -p _build/dev/ %Ls", $this->getBinaryPath(), $this->getFilteredIssues(), $lintable_paths); } catch (CommandException $exception) { PhutilConsole::getConsole()->writeErr($exception->getMessage()); } // Load results. $result = id( new SQLite3( $this->getProjectRoot().'/_build/_lint/lint.db', SQLITE3_OPEN_READONLY)) ->query("SELECT * FROM raised_issues"); while ($issue = $result->fetchArray(SQLITE3_ASSOC)) { // Skip issues not part of the linted file. if (in_array($issue['file'], $interesting_paths)) { $this->addLintMessage(id(new ArcanistLintMessage()) ->setPath($issue['file']) ->setLine($issue['line']) ->setChar($issue['column']) ->setCode('Howtoeven') ->setSeverity($this->getSeverity($issue['severity'])) ->setName('Hte-'.$issue['name']) ->setDescription( sprintf( "%s\n\n%s", ($issue['message']) ? $issue['message'] : $issue['description'], $issue['explanation'])) ->setOriginalText(idx($issue, 'original', '')) ->setReplacementText(idx($issue, 'replacement', ''))); } } } public function lintPath($path) { } /** * Get the paths that we know how to lint. * * The strategy is to first look whether there's an existing compilation * database and use that if it's exhaustive. We generate our own only if * necessary. */ private function getLintablePaths($paths) { // Replace headers with existing sources. for ($i = 0; $i < count($paths); $i++) { if (preg_match("/\.h$/", $paths[$i])) { $header = preg_replace("/\.h$/", ".cpp", $paths[$i]); if (file_exists($header)) { $paths[$i] = $header; } } } // Check if database exists and is exhaustive. $available_paths = $this->getAvailablePaths(); $lintable_paths = array_intersect($paths, $available_paths); if ($paths === $lintable_paths) { return $lintable_paths; } // Generate our own database. $targets = $this->getTargetsFor($paths); if (!$targets) { PhutilConsole::getConsole()->writeErr( "No build targets found for %s\n", implode(', ', $paths)); return array(); } $this->localExecx("./tools/build/bin/fbconfig.par -r %Ls", $targets); $this->localExecx("./tools/build/bin/fbmake.par gen_cdb"); $available_paths = $this->getAvailablePaths(); $lintable_paths = array_intersect($paths, $available_paths); if ($paths != $lintable_paths) { PhutilConsole::getConsole()->writeErr( "Can't lint %s\n", implode(', ', array_diff($paths, $available_paths))); } // Return what we know how to lint. return $lintable_paths; } /** * Get the available paths in the current compilation database. */ private function getAvailablePaths() { $database_path = $this->getProjectRoot() .'/_build/dev/compile_commands.json'; if (!file_exists($database_path)) { return array(); } $entries = json_decode(file_get_contents($database_path), true); $paths = array(); foreach ($entries as $entry) { $paths[] = $entry['file']; } return $paths; } /** * Search for the targets directories for the given files. */ private static function getTargetsFor($paths) { $targets = array(); foreach ($paths as $path) { while (($path = dirname($path)) !== '.') { if (in_array('TARGETS', scandir($path))) { $contents = file_get_contents($path.'/TARGETS'); if (strpos($contents, 'cpp_binary') !== false) { $targets[] = $path; break; } } } } return array_unique($targets); } /** * The paths that we actually want to report on. */ private function getInterestingPaths($paths) { $headers = array(); foreach ($paths as $path) { $headers[] = preg_replace("/\.cpp$/", ".h", $path); } return array_merge($paths, $headers); } /** * The path where the binary is located. Will return the current dewey binary * unless the `HOWTOEVEN_BUILD` environment variable is set. */ private function getBinaryPath() { $path = sprintf( "/mnt/dewey/fbcode/.commits/%s/builds/howtoeven/client", self::VERSION); $build = getenv('HOWTOEVEN_BUILD'); if ($build) { $path = sprintf( "./_build/%s/tools/howtoeven/client", $build); if (!file_exists($path)) { PhutilConsole::getConsole()->writeErr(">> %s does not exist\n", $path); exit(1); } } return $path; } /** * Execute the command in the root directory. */ private function localExecx($command /* , ... */) { $arguments = func_get_args(); return newv('ExecFuture', $arguments) ->setCWD($this->getProjectRoot()) ->resolvex(); } /** * The root of the project. */ private function getProjectRoot() { return $this->getEngine()->getWorkingCopy()->getProjectRoot(); } private function getFilteredIssues() { $issues = getenv('HOWTOEVEN_ISSUES'); return ($issues) ? csprintf('-issues %s', $issues) : ''; } } rocksdb-4.5.1/arcanist_util/cpp_linter/FbcodeClangFormatLinter.php000066400000000000000000000031151267530634000253450ustar00rootroot00000000000000 ArcanistLintSeverity::SEVERITY_ADVICE, ); } public function getLintNameMap() { return array( self::LINT_FORMATTING => pht('Changes are not clang-formatted'), ); } protected function getFormatFuture($path, array $changed) { $args = ""; foreach ($changed as $key => $value) { $args .= " --lines=$key:$key"; } $binary = self::CLANG_FORMAT_BINARY; if (!file_exists($binary)) { // trust the $PATH $binary = "clang-format"; } return new ExecFuture( "%s %s $args", $binary, $this->getEngine()->getFilePathOnDisk($path)); } protected function getLintMessage($diff) { $link_to_clang_format = "[[ http://fburl.com/clang-format | clang-format ]]"; return <<getEngine()->getFilePathOnDisk($p); $lpath_file = file($lpath); if (preg_match('/\.(c)$/', $lpath) || preg_match('/-\*-.*Mode: C[; ].*-\*-/', $lpath_file[0]) || preg_match('/vim(:.*)*:\s*(set\s+)?filetype=c\s*:/', $lpath_file[0]) ) { $futures[$p] = new ExecFuture("%s %s %s 2>&1", self::FLINT, self::C_FLAG, $this->getEngine()->getFilePathOnDisk($p)); } else { $futures[$p] = new ExecFuture("%s %s 2>&1", self::FLINT, $this->getEngine()->getFilePathOnDisk($p)); } } foreach (Futures($futures)->limit(8) as $p => $f) { $this->rawLintOutput[$p] = $f->resolvex(); } return; } public function getLinterName() { return "FBCPP"; } public function lintPath($path) { $this->runCppLint($path); } private function runCppLint($path) { $msgs = $this->getCppLintOutput($path); foreach ($msgs as $m) { $this->raiseLintAtLine($m['line'], 0, $m['severity'], $m['msg']); } } private function adviseOnEachPattern( $path, $regex, $message, $lint_type = self::LINT_ADVICE, $match_idx = 0) { $file_data = $this->getData($path); $matches = array(); if (!preg_match_all($regex, $file_data, $matches, PREG_OFFSET_CAPTURE)) { return; } foreach ($matches[$match_idx] as $match) { list($match_str, $offset) = $match; $this->raiseLintAtOffset($offset, $lint_type, $message, $match_str); } } public function getLintSeverityMap() { return array( self::LINT_WARNING => ArcanistLintSeverity::SEVERITY_WARNING, self::LINT_ADVICE => ArcanistLintSeverity::SEVERITY_ADVICE, self::LINT_ERROR => ArcanistLintSeverity::SEVERITY_ERROR ); } public function getLintNameMap() { return array( self::LINT_ADVICE => "CppLint Advice", self::LINT_WARNING => "CppLint Warning", self::LINT_ERROR => "CppLint Error" ); } private function getCppLintOutput($path) { if (!array_key_exists($path, $this->rawLintOutput)) { return array(); } list($output) = $this->rawLintOutput[$path]; $msgs = array(); $current = null; $matches = array(); foreach (explode("\n", $output) as $line) { if (preg_match('/.*?:(\d+):(.*)/', $line, $matches)) { if ($current) { $msgs[] = $current; } $line = $matches[1]; $text = $matches[2]; if (preg_match('/.*Warning.*/', $text)) { $sev = self::LINT_WARNING; } else if (preg_match('/.*Advice.*/', $text)) { $sev = self::LINT_ADVICE; } else { $sev = self::LINT_ERROR; } $current = array('line' => $line, 'msg' => $text, 'severity' => $sev); } else if ($current) { $current['msg'] .= ' ' . $line; } } if ($current) { $msgs[] = $current; } return $msgs; } } rocksdb-4.5.1/arcanist_util/cpp_linter/cpplint.py000077500000000000000000005463621267530634000222240ustar00rootroot00000000000000#!/usr/bin/python # Copyright (c) 2013, Facebook, Inc. All rights reserved. # This source code is licensed under the BSD-style license found in the # LICENSE file in the root directory of this source tree. An additional grant # of patent rights can be found in the PATENTS file in the same directory. # Copyright (c) 2011 The LevelDB Authors. All rights reserved. # Use of this source code is governed by a BSD-style license that can be # found in the LICENSE file. See the AUTHORS file for names of contributors. # # Copyright (c) 2009 Google Inc. All rights reserved. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are # met: # * Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # * Redistributions in binary form must reproduce the above # copyright notice, this list of conditions and the following disclaimer # in the documentation and/or other materials provided with the # distribution. # * Neither the name of Google Inc. nor the names of its # contributors may be used to endorse or promote products derived from # this software without specific prior written permission. # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. """Does google-lint on c++ files. The goal of this script is to identify places in the code that *may* be in non-compliance with google style. It does not attempt to fix up these problems -- the point is to educate. It does also not attempt to find all problems, or to ensure that everything it does find is legitimately a problem. In particular, we can get very confused by /* and // inside strings! We do a small hack, which is to ignore //'s with "'s after them on the same line, but it is far from perfect (in either direction). """ import codecs import copy import getopt import math # for log import os import re import sre_compile import string import sys import unicodedata _USAGE = """ Syntax: cpplint.py [--verbose=#] [--output=vs7] [--filter=-x,+y,...] [--counting=total|toplevel|detailed] [--root=subdir] [--linelength=digits] [file] ... The style guidelines this tries to follow are those in http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml Every problem is given a confidence score from 1-5, with 5 meaning we are certain of the problem, and 1 meaning it could be a legitimate construct. This will miss some errors, and is not a substitute for a code review. To suppress false-positive errors of a certain category, add a 'NOLINT(category)' comment to the line. NOLINT or NOLINT(*) suppresses errors of all categories on that line. The files passed in will be linted; at least one file must be provided. Default linted extensions are .cc, .cpp, .cu, .cuh and .h. Change the extensions with the --extensions flag. Flags: output=vs7 By default, the output is formatted to ease emacs parsing. Visual Studio compatible output (vs7) may also be used. Other formats are unsupported. verbose=# Specify a number 0-5 to restrict errors to certain verbosity levels. filter=-x,+y,... Specify a comma-separated list of category-filters to apply: only error messages whose category names pass the filters will be printed. (Category names are printed with the message and look like "[whitespace/indent]".) Filters are evaluated left to right. "-FOO" and "FOO" means "do not print categories that start with FOO". "+FOO" means "do print categories that start with FOO". Examples: --filter=-whitespace,+whitespace/braces --filter=whitespace,runtime/printf,+runtime/printf_format --filter=-,+build/include_what_you_use To see a list of all the categories used in cpplint, pass no arg: --filter= counting=total|toplevel|detailed The total number of errors found is always printed. If 'toplevel' is provided, then the count of errors in each of the top-level categories like 'build' and 'whitespace' will also be printed. If 'detailed' is provided, then a count is provided for each category like 'build/class'. root=subdir The root directory used for deriving header guard CPP variable. By default, the header guard CPP variable is calculated as the relative path to the directory that contains .git, .hg, or .svn. When this flag is specified, the relative path is calculated from the specified directory. If the specified directory does not exist, this flag is ignored. Examples: Assuing that src/.git exists, the header guard CPP variables for src/chrome/browser/ui/browser.h are: No flag => CHROME_BROWSER_UI_BROWSER_H_ --root=chrome => BROWSER_UI_BROWSER_H_ --root=chrome/browser => UI_BROWSER_H_ linelength=digits This is the allowed line length for the project. The default value is 80 characters. Examples: --linelength=120 extensions=extension,extension,... The allowed file extensions that cpplint will check Examples: --extensions=hpp,cpp """ # We categorize each error message we print. Here are the categories. # We want an explicit list so we can list them all in cpplint --filter=. # If you add a new error message with a new category, add it to the list # here! cpplint_unittest.py should tell you if you forget to do this. _ERROR_CATEGORIES = [ 'build/class', 'build/deprecated', 'build/endif_comment', 'build/explicit_make_pair', 'build/forward_decl', 'build/header_guard', 'build/include', 'build/include_alpha', 'build/include_order', 'build/include_what_you_use', 'build/namespaces', 'build/printf_format', 'build/storage_class', 'legal/copyright', 'readability/alt_tokens', 'readability/braces', 'readability/casting', 'readability/check', 'readability/constructors', 'readability/fn_size', 'readability/function', 'readability/multiline_comment', 'readability/multiline_string', 'readability/namespace', 'readability/nolint', 'readability/nul', 'readability/streams', 'readability/todo', 'readability/utf8', 'runtime/arrays', 'runtime/casting', 'runtime/explicit', 'runtime/int', 'runtime/init', 'runtime/invalid_increment', 'runtime/member_string_references', 'runtime/memset', 'runtime/operator', 'runtime/printf', 'runtime/printf_format', 'runtime/references', 'runtime/string', 'runtime/threadsafe_fn', 'runtime/vlog', 'whitespace/blank_line', 'whitespace/braces', 'whitespace/comma', 'whitespace/comments', 'whitespace/empty_conditional_body', 'whitespace/empty_loop_body', 'whitespace/end_of_line', 'whitespace/ending_newline', 'whitespace/forcolon', 'whitespace/indent', 'whitespace/line_length', 'whitespace/newline', 'whitespace/operators', 'whitespace/parens', 'whitespace/semicolon', 'whitespace/tab', 'whitespace/todo' ] # The default state of the category filter. This is overrided by the --filter= # flag. By default all errors are on, so only add here categories that should be # off by default (i.e., categories that must be enabled by the --filter= flags). # All entries here should start with a '-' or '+', as in the --filter= flag. _DEFAULT_FILTERS = [] # We used to check for high-bit characters, but after much discussion we # decided those were OK, as long as they were in UTF-8 and didn't represent # hard-coded international strings, which belong in a separate i18n file. # C++ headers _CPP_HEADERS = frozenset([ # Legacy 'algobase.h', 'algo.h', 'alloc.h', 'builtinbuf.h', 'bvector.h', 'complex.h', 'defalloc.h', 'deque.h', 'editbuf.h', 'fstream.h', 'function.h', 'hash_map', 'hash_map.h', 'hash_set', 'hash_set.h', 'hashtable.h', 'heap.h', 'indstream.h', 'iomanip.h', 'iostream.h', 'istream.h', 'iterator.h', 'list.h', 'map.h', 'multimap.h', 'multiset.h', 'ostream.h', 'pair.h', 'parsestream.h', 'pfstream.h', 'procbuf.h', 'pthread_alloc', 'pthread_alloc.h', 'rope', 'rope.h', 'ropeimpl.h', 'set.h', 'slist', 'slist.h', 'stack.h', 'stdiostream.h', 'stl_alloc.h', 'stl_relops.h', 'streambuf.h', 'stream.h', 'strfile.h', 'strstream.h', 'tempbuf.h', 'tree.h', 'type_traits.h', 'vector.h', # 17.6.1.2 C++ library headers 'algorithm', 'array', 'atomic', 'bitset', 'chrono', 'codecvt', 'complex', 'condition_variable', 'deque', 'exception', 'forward_list', 'fstream', 'functional', 'future', 'initializer_list', 'iomanip', 'ios', 'iosfwd', 'iostream', 'istream', 'iterator', 'limits', 'list', 'locale', 'map', 'memory', 'mutex', 'new', 'numeric', 'ostream', 'queue', 'random', 'ratio', 'regex', 'set', 'sstream', 'stack', 'stdexcept', 'streambuf', 'string', 'strstream', 'system_error', 'thread', 'tuple', 'typeindex', 'typeinfo', 'type_traits', 'unordered_map', 'unordered_set', 'utility', 'valarray', 'vector', # 17.6.1.2 C++ headers for C library facilities 'cassert', 'ccomplex', 'cctype', 'cerrno', 'cfenv', 'cfloat', 'cinttypes', 'ciso646', 'climits', 'clocale', 'cmath', 'csetjmp', 'csignal', 'cstdalign', 'cstdarg', 'cstdbool', 'cstddef', 'cstdint', 'cstdio', 'cstdlib', 'cstring', 'ctgmath', 'ctime', 'cuchar', 'cwchar', 'cwctype', ]) # Assertion macros. These are defined in base/logging.h and # testing/base/gunit.h. Note that the _M versions need to come first # for substring matching to work. _CHECK_MACROS = [ 'DCHECK', 'CHECK', 'EXPECT_TRUE_M', 'EXPECT_TRUE', 'ASSERT_TRUE_M', 'ASSERT_TRUE', 'EXPECT_FALSE_M', 'EXPECT_FALSE', 'ASSERT_FALSE_M', 'ASSERT_FALSE', ] # Replacement macros for CHECK/DCHECK/EXPECT_TRUE/EXPECT_FALSE _CHECK_REPLACEMENT = dict([(m, {}) for m in _CHECK_MACROS]) for op, replacement in [('==', 'EQ'), ('!=', 'NE'), ('>=', 'GE'), ('>', 'GT'), ('<=', 'LE'), ('<', 'LT')]: _CHECK_REPLACEMENT['DCHECK'][op] = 'DCHECK_%s' % replacement _CHECK_REPLACEMENT['CHECK'][op] = 'CHECK_%s' % replacement _CHECK_REPLACEMENT['EXPECT_TRUE'][op] = 'EXPECT_%s' % replacement _CHECK_REPLACEMENT['ASSERT_TRUE'][op] = 'ASSERT_%s' % replacement _CHECK_REPLACEMENT['EXPECT_TRUE_M'][op] = 'EXPECT_%s_M' % replacement _CHECK_REPLACEMENT['ASSERT_TRUE_M'][op] = 'ASSERT_%s_M' % replacement for op, inv_replacement in [('==', 'NE'), ('!=', 'EQ'), ('>=', 'LT'), ('>', 'LE'), ('<=', 'GT'), ('<', 'GE')]: _CHECK_REPLACEMENT['EXPECT_FALSE'][op] = 'EXPECT_%s' % inv_replacement _CHECK_REPLACEMENT['ASSERT_FALSE'][op] = 'ASSERT_%s' % inv_replacement _CHECK_REPLACEMENT['EXPECT_FALSE_M'][op] = 'EXPECT_%s_M' % inv_replacement _CHECK_REPLACEMENT['ASSERT_FALSE_M'][op] = 'ASSERT_%s_M' % inv_replacement # Alternative tokens and their replacements. For full list, see section 2.5 # Alternative tokens [lex.digraph] in the C++ standard. # # Digraphs (such as '%:') are not included here since it's a mess to # match those on a word boundary. _ALT_TOKEN_REPLACEMENT = { 'and': '&&', 'bitor': '|', 'or': '||', 'xor': '^', 'compl': '~', 'bitand': '&', 'and_eq': '&=', 'or_eq': '|=', 'xor_eq': '^=', 'not': '!', 'not_eq': '!=' } # Compile regular expression that matches all the above keywords. The "[ =()]" # bit is meant to avoid matching these keywords outside of boolean expressions. # # False positives include C-style multi-line comments and multi-line strings # but those have always been troublesome for cpplint. _ALT_TOKEN_REPLACEMENT_PATTERN = re.compile( r'[ =()](' + ('|'.join(_ALT_TOKEN_REPLACEMENT.keys())) + r')(?=[ (]|$)') # These constants define types of headers for use with # _IncludeState.CheckNextIncludeOrder(). _C_SYS_HEADER = 1 _CPP_SYS_HEADER = 2 _LIKELY_MY_HEADER = 3 _POSSIBLE_MY_HEADER = 4 _OTHER_HEADER = 5 # These constants define the current inline assembly state _NO_ASM = 0 # Outside of inline assembly block _INSIDE_ASM = 1 # Inside inline assembly block _END_ASM = 2 # Last line of inline assembly block _BLOCK_ASM = 3 # The whole block is an inline assembly block # Match start of assembly blocks _MATCH_ASM = re.compile(r'^\s*(?:asm|_asm|__asm|__asm__)' r'(?:\s+(volatile|__volatile__))?' r'\s*[{(]') _regexp_compile_cache = {} # Finds occurrences of NOLINT or NOLINT(...). _RE_SUPPRESSION = re.compile(r'\bNOLINT\b(\([^)]*\))?') # {str, set(int)}: a map from error categories to sets of linenumbers # on which those errors are expected and should be suppressed. _error_suppressions = {} # The root directory used for deriving header guard CPP variable. # This is set by --root flag. _root = None # The allowed line length of files. # This is set by --linelength flag. _line_length = 80 # The allowed extensions for file names # This is set by --extensions flag. _valid_extensions = set(['cc', 'h', 'cpp', 'cu', 'cuh']) def ParseNolintSuppressions(filename, raw_line, linenum, error): """Updates the global list of error-suppressions. Parses any NOLINT comments on the current line, updating the global error_suppressions store. Reports an error if the NOLINT comment was malformed. Args: filename: str, the name of the input file. raw_line: str, the line of input text, with comments. linenum: int, the number of the current line. error: function, an error handler. """ # FIXME(adonovan): "NOLINT(" is misparsed as NOLINT(*). matched = _RE_SUPPRESSION.search(raw_line) if matched: category = matched.group(1) if category in (None, '(*)'): # => "suppress all" _error_suppressions.setdefault(None, set()).add(linenum) else: if category.startswith('(') and category.endswith(')'): category = category[1:-1] if category in _ERROR_CATEGORIES: _error_suppressions.setdefault(category, set()).add(linenum) else: error(filename, linenum, 'readability/nolint', 5, 'Unknown NOLINT error category: %s' % category) def ResetNolintSuppressions(): "Resets the set of NOLINT suppressions to empty." _error_suppressions.clear() def IsErrorSuppressedByNolint(category, linenum): """Returns true if the specified error category is suppressed on this line. Consults the global error_suppressions map populated by ParseNolintSuppressions/ResetNolintSuppressions. Args: category: str, the category of the error. linenum: int, the current line number. Returns: bool, True iff the error should be suppressed due to a NOLINT comment. """ return (linenum in _error_suppressions.get(category, set()) or linenum in _error_suppressions.get(None, set())) def Match(pattern, s): """Matches the string with the pattern, caching the compiled regexp.""" # The regexp compilation caching is inlined in both Match and Search for # performance reasons; factoring it out into a separate function turns out # to be noticeably expensive. if pattern not in _regexp_compile_cache: _regexp_compile_cache[pattern] = sre_compile.compile(pattern) return _regexp_compile_cache[pattern].match(s) def ReplaceAll(pattern, rep, s): """Replaces instances of pattern in a string with a replacement. The compiled regex is kept in a cache shared by Match and Search. Args: pattern: regex pattern rep: replacement text s: search string Returns: string with replacements made (or original string if no replacements) """ if pattern not in _regexp_compile_cache: _regexp_compile_cache[pattern] = sre_compile.compile(pattern) return _regexp_compile_cache[pattern].sub(rep, s) def Search(pattern, s): """Searches the string for the pattern, caching the compiled regexp.""" if pattern not in _regexp_compile_cache: _regexp_compile_cache[pattern] = sre_compile.compile(pattern) return _regexp_compile_cache[pattern].search(s) class _IncludeState(dict): """Tracks line numbers for includes, and the order in which includes appear. As a dict, an _IncludeState object serves as a mapping between include filename and line number on which that file was included. Call CheckNextIncludeOrder() once for each header in the file, passing in the type constants defined above. Calls in an illegal order will raise an _IncludeError with an appropriate error message. """ # self._section will move monotonically through this set. If it ever # needs to move backwards, CheckNextIncludeOrder will raise an error. _INITIAL_SECTION = 0 _MY_H_SECTION = 1 _C_SECTION = 2 _CPP_SECTION = 3 _OTHER_H_SECTION = 4 _TYPE_NAMES = { _C_SYS_HEADER: 'C system header', _CPP_SYS_HEADER: 'C++ system header', _LIKELY_MY_HEADER: 'header this file implements', _POSSIBLE_MY_HEADER: 'header this file may implement', _OTHER_HEADER: 'other header', } _SECTION_NAMES = { _INITIAL_SECTION: "... nothing. (This can't be an error.)", _MY_H_SECTION: 'a header this file implements', _C_SECTION: 'C system header', _CPP_SECTION: 'C++ system header', _OTHER_H_SECTION: 'other header', } def __init__(self): dict.__init__(self) self.ResetSection() def ResetSection(self): # The name of the current section. self._section = self._INITIAL_SECTION # The path of last found header. self._last_header = '' def SetLastHeader(self, header_path): self._last_header = header_path def CanonicalizeAlphabeticalOrder(self, header_path): """Returns a path canonicalized for alphabetical comparison. - replaces "-" with "_" so they both cmp the same. - removes '-inl' since we don't require them to be after the main header. - lowercase everything, just in case. Args: header_path: Path to be canonicalized. Returns: Canonicalized path. """ return header_path.replace('-inl.h', '.h').replace('-', '_').lower() def IsInAlphabeticalOrder(self, clean_lines, linenum, header_path): """Check if a header is in alphabetical order with the previous header. Args: clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. header_path: Canonicalized header to be checked. Returns: Returns true if the header is in alphabetical order. """ # If previous section is different from current section, _last_header will # be reset to empty string, so it's always less than current header. # # If previous line was a blank line, assume that the headers are # intentionally sorted the way they are. if (self._last_header > header_path and not Match(r'^\s*$', clean_lines.elided[linenum - 1])): return False return True def CheckNextIncludeOrder(self, header_type): """Returns a non-empty error message if the next header is out of order. This function also updates the internal state to be ready to check the next include. Args: header_type: One of the _XXX_HEADER constants defined above. Returns: The empty string if the header is in the right order, or an error message describing what's wrong. """ error_message = ('Found %s after %s' % (self._TYPE_NAMES[header_type], self._SECTION_NAMES[self._section])) last_section = self._section if header_type == _C_SYS_HEADER: if self._section <= self._C_SECTION: self._section = self._C_SECTION else: self._last_header = '' return error_message elif header_type == _CPP_SYS_HEADER: if self._section <= self._CPP_SECTION: self._section = self._CPP_SECTION else: self._last_header = '' return error_message elif header_type == _LIKELY_MY_HEADER: if self._section <= self._MY_H_SECTION: self._section = self._MY_H_SECTION else: self._section = self._OTHER_H_SECTION elif header_type == _POSSIBLE_MY_HEADER: if self._section <= self._MY_H_SECTION: self._section = self._MY_H_SECTION else: # This will always be the fallback because we're not sure # enough that the header is associated with this file. self._section = self._OTHER_H_SECTION else: assert header_type == _OTHER_HEADER self._section = self._OTHER_H_SECTION if last_section != self._section: self._last_header = '' return '' class _CppLintState(object): """Maintains module-wide state..""" def __init__(self): self.verbose_level = 1 # global setting. self.error_count = 0 # global count of reported errors # filters to apply when emitting error messages self.filters = _DEFAULT_FILTERS[:] self.counting = 'total' # In what way are we counting errors? self.errors_by_category = {} # string to int dict storing error counts # output format: # "emacs" - format that emacs can parse (default) # "vs7" - format that Microsoft Visual Studio 7 can parse self.output_format = 'emacs' def SetOutputFormat(self, output_format): """Sets the output format for errors.""" self.output_format = output_format def SetVerboseLevel(self, level): """Sets the module's verbosity, and returns the previous setting.""" last_verbose_level = self.verbose_level self.verbose_level = level return last_verbose_level def SetCountingStyle(self, counting_style): """Sets the module's counting options.""" self.counting = counting_style def SetFilters(self, filters): """Sets the error-message filters. These filters are applied when deciding whether to emit a given error message. Args: filters: A string of comma-separated filters (eg "+whitespace/indent"). Each filter should start with + or -; else we die. Raises: ValueError: The comma-separated filters did not all start with '+' or '-'. E.g. "-,+whitespace,-whitespace/indent,whitespace/badfilter" """ # Default filters always have less priority than the flag ones. self.filters = _DEFAULT_FILTERS[:] for filt in filters.split(','): clean_filt = filt.strip() if clean_filt: self.filters.append(clean_filt) for filt in self.filters: if not (filt.startswith('+') or filt.startswith('-')): raise ValueError('Every filter in --filters must start with + or -' ' (%s does not)' % filt) def ResetErrorCounts(self): """Sets the module's error statistic back to zero.""" self.error_count = 0 self.errors_by_category = {} def IncrementErrorCount(self, category): """Bumps the module's error statistic.""" self.error_count += 1 if self.counting in ('toplevel', 'detailed'): if self.counting != 'detailed': category = category.split('/')[0] if category not in self.errors_by_category: self.errors_by_category[category] = 0 self.errors_by_category[category] += 1 def PrintErrorCounts(self): """Print a summary of errors by category, and the total.""" for category, count in self.errors_by_category.iteritems(): sys.stderr.write('Category \'%s\' errors found: %d\n' % (category, count)) sys.stderr.write('Total errors found: %d\n' % self.error_count) _cpplint_state = _CppLintState() def _OutputFormat(): """Gets the module's output format.""" return _cpplint_state.output_format def _SetOutputFormat(output_format): """Sets the module's output format.""" _cpplint_state.SetOutputFormat(output_format) def _VerboseLevel(): """Returns the module's verbosity setting.""" return _cpplint_state.verbose_level def _SetVerboseLevel(level): """Sets the module's verbosity, and returns the previous setting.""" return _cpplint_state.SetVerboseLevel(level) def _SetCountingStyle(level): """Sets the module's counting options.""" _cpplint_state.SetCountingStyle(level) def _Filters(): """Returns the module's list of output filters, as a list.""" return _cpplint_state.filters def _SetFilters(filters): """Sets the module's error-message filters. These filters are applied when deciding whether to emit a given error message. Args: filters: A string of comma-separated filters (eg "whitespace/indent"). Each filter should start with + or -; else we die. """ _cpplint_state.SetFilters(filters) class _FunctionState(object): """Tracks current function name and the number of lines in its body.""" _NORMAL_TRIGGER = 250 # for --v=0, 500 for --v=1, etc. _TEST_TRIGGER = 400 # about 50% more than _NORMAL_TRIGGER. def __init__(self): self.in_a_function = False self.lines_in_function = 0 self.current_function = '' def Begin(self, function_name): """Start analyzing function body. Args: function_name: The name of the function being tracked. """ self.in_a_function = True self.lines_in_function = 0 self.current_function = function_name def Count(self): """Count line in current function body.""" if self.in_a_function: self.lines_in_function += 1 def Check(self, error, filename, linenum): """Report if too many lines in function body. Args: error: The function to call with any errors found. filename: The name of the current file. linenum: The number of the line to check. """ if Match(r'T(EST|est)', self.current_function): base_trigger = self._TEST_TRIGGER else: base_trigger = self._NORMAL_TRIGGER trigger = base_trigger * 2**_VerboseLevel() if self.lines_in_function > trigger: error_level = int(math.log(self.lines_in_function / base_trigger, 2)) # 50 => 0, 100 => 1, 200 => 2, 400 => 3, 800 => 4, 1600 => 5, ... if error_level > 5: error_level = 5 error(filename, linenum, 'readability/fn_size', error_level, 'Small and focused functions are preferred:' ' %s has %d non-comment lines' ' (error triggered by exceeding %d lines).' % ( self.current_function, self.lines_in_function, trigger)) def End(self): """Stop analyzing function body.""" self.in_a_function = False class _IncludeError(Exception): """Indicates a problem with the include order in a file.""" pass class FileInfo: """Provides utility functions for filenames. FileInfo provides easy access to the components of a file's path relative to the project root. """ def __init__(self, filename): self._filename = filename def FullName(self): """Make Windows paths like Unix.""" return os.path.abspath(self._filename).replace('\\', '/') def RepositoryName(self): """FullName after removing the local path to the repository. If we have a real absolute path name here we can try to do something smart: detecting the root of the checkout and truncating /path/to/checkout from the name so that we get header guards that don't include things like "C:\Documents and Settings\..." or "/home/username/..." in them and thus people on different computers who have checked the source out to different locations won't see bogus errors. """ fullname = self.FullName() if os.path.exists(fullname): project_dir = os.path.dirname(fullname) if os.path.exists(os.path.join(project_dir, ".svn")): # If there's a .svn file in the current directory, we recursively look # up the directory tree for the top of the SVN checkout root_dir = project_dir one_up_dir = os.path.dirname(root_dir) while os.path.exists(os.path.join(one_up_dir, ".svn")): root_dir = os.path.dirname(root_dir) one_up_dir = os.path.dirname(one_up_dir) prefix = os.path.commonprefix([root_dir, project_dir]) return fullname[len(prefix) + 1:] # Not SVN <= 1.6? Try to find a git, hg, or svn top level directory by # searching up from the current path. root_dir = os.path.dirname(fullname) while (root_dir != os.path.dirname(root_dir) and not os.path.exists(os.path.join(root_dir, ".git")) and not os.path.exists(os.path.join(root_dir, ".hg")) and not os.path.exists(os.path.join(root_dir, ".svn"))): root_dir = os.path.dirname(root_dir) if (os.path.exists(os.path.join(root_dir, ".git")) or os.path.exists(os.path.join(root_dir, ".hg")) or os.path.exists(os.path.join(root_dir, ".svn"))): prefix = os.path.commonprefix([root_dir, project_dir]) return fullname[len(prefix) + 1:] # Don't know what to do; header guard warnings may be wrong... return fullname def Split(self): """Splits the file into the directory, basename, and extension. For 'chrome/browser/browser.cc', Split() would return ('chrome/browser', 'browser', '.cc') Returns: A tuple of (directory, basename, extension). """ googlename = self.RepositoryName() project, rest = os.path.split(googlename) return (project,) + os.path.splitext(rest) def BaseName(self): """File base name - text after the final slash, before the final period.""" return self.Split()[1] def Extension(self): """File extension - text following the final period.""" return self.Split()[2] def NoExtension(self): """File has no source file extension.""" return '/'.join(self.Split()[0:2]) def IsSource(self): """File has a source file extension.""" return self.Extension()[1:] in ('c', 'cc', 'cpp', 'cxx') def _ShouldPrintError(category, confidence, linenum): """If confidence >= verbose, category passes filter and is not suppressed.""" # There are three ways we might decide not to print an error message: # a "NOLINT(category)" comment appears in the source, # the verbosity level isn't high enough, or the filters filter it out. if IsErrorSuppressedByNolint(category, linenum): return False if confidence < _cpplint_state.verbose_level: return False is_filtered = False for one_filter in _Filters(): if one_filter.startswith('-'): if category.startswith(one_filter[1:]): is_filtered = True elif one_filter.startswith('+'): if category.startswith(one_filter[1:]): is_filtered = False else: assert False # should have been checked for in SetFilter. if is_filtered: return False return True def Error(filename, linenum, category, confidence, message): """Logs the fact we've found a lint error. We log where the error was found, and also our confidence in the error, that is, how certain we are this is a legitimate style regression, and not a misidentification or a use that's sometimes justified. False positives can be suppressed by the use of "cpplint(category)" comments on the offending line. These are parsed into _error_suppressions. Args: filename: The name of the file containing the error. linenum: The number of the line containing the error. category: A string used to describe the "category" this bug falls under: "whitespace", say, or "runtime". Categories may have a hierarchy separated by slashes: "whitespace/indent". confidence: A number from 1-5 representing a confidence score for the error, with 5 meaning that we are certain of the problem, and 1 meaning that it could be a legitimate construct. message: The error message. """ if _ShouldPrintError(category, confidence, linenum): _cpplint_state.IncrementErrorCount(category) if _cpplint_state.output_format == 'vs7': sys.stderr.write('%s(%s): %s [%s] [%d]\n' % ( filename, linenum, message, category, confidence)) elif _cpplint_state.output_format == 'eclipse': sys.stderr.write('%s:%s: warning: %s [%s] [%d]\n' % ( filename, linenum, message, category, confidence)) else: sys.stderr.write('%s:%s: %s [%s] [%d]\n' % ( filename, linenum, message, category, confidence)) # Matches standard C++ escape sequences per 2.13.2.3 of the C++ standard. _RE_PATTERN_CLEANSE_LINE_ESCAPES = re.compile( r'\\([abfnrtv?"\\\']|\d+|x[0-9a-fA-F]+)') # Matches strings. Escape codes should already be removed by ESCAPES. _RE_PATTERN_CLEANSE_LINE_DOUBLE_QUOTES = re.compile(r'"[^"]*"') # Matches characters. Escape codes should already be removed by ESCAPES. _RE_PATTERN_CLEANSE_LINE_SINGLE_QUOTES = re.compile(r"'.'") # Matches multi-line C++ comments. # This RE is a little bit more complicated than one might expect, because we # have to take care of space removals tools so we can handle comments inside # statements better. # The current rule is: We only clear spaces from both sides when we're at the # end of the line. Otherwise, we try to remove spaces from the right side, # if this doesn't work we try on left side but only if there's a non-character # on the right. _RE_PATTERN_CLEANSE_LINE_C_COMMENTS = re.compile( r"""(\s*/\*.*\*/\s*$| /\*.*\*/\s+| \s+/\*.*\*/(?=\W)| /\*.*\*/)""", re.VERBOSE) def IsCppString(line): """Does line terminate so, that the next symbol is in string constant. This function does not consider single-line nor multi-line comments. Args: line: is a partial line of code starting from the 0..n. Returns: True, if next character appended to 'line' is inside a string constant. """ line = line.replace(r'\\', 'XX') # after this, \\" does not match to \" return ((line.count('"') - line.count(r'\"') - line.count("'\"'")) & 1) == 1 def CleanseRawStrings(raw_lines): """Removes C++11 raw strings from lines. Before: static const char kData[] = R"( multi-line string )"; After: static const char kData[] = "" (replaced by blank line) ""; Args: raw_lines: list of raw lines. Returns: list of lines with C++11 raw strings replaced by empty strings. """ delimiter = None lines_without_raw_strings = [] for line in raw_lines: if delimiter: # Inside a raw string, look for the end end = line.find(delimiter) if end >= 0: # Found the end of the string, match leading space for this # line and resume copying the original lines, and also insert # a "" on the last line. leading_space = Match(r'^(\s*)\S', line) line = leading_space.group(1) + '""' + line[end + len(delimiter):] delimiter = None else: # Haven't found the end yet, append a blank line. line = '' else: # Look for beginning of a raw string. # See 2.14.15 [lex.string] for syntax. matched = Match(r'^(.*)\b(?:R|u8R|uR|UR|LR)"([^\s\\()]*)\((.*)$', line) if matched: delimiter = ')' + matched.group(2) + '"' end = matched.group(3).find(delimiter) if end >= 0: # Raw string ended on same line line = (matched.group(1) + '""' + matched.group(3)[end + len(delimiter):]) delimiter = None else: # Start of a multi-line raw string line = matched.group(1) + '""' lines_without_raw_strings.append(line) # TODO(unknown): if delimiter is not None here, we might want to # emit a warning for unterminated string. return lines_without_raw_strings def FindNextMultiLineCommentStart(lines, lineix): """Find the beginning marker for a multiline comment.""" while lineix < len(lines): if lines[lineix].strip().startswith('/*'): # Only return this marker if the comment goes beyond this line if lines[lineix].strip().find('*/', 2) < 0: return lineix lineix += 1 return len(lines) def FindNextMultiLineCommentEnd(lines, lineix): """We are inside a comment, find the end marker.""" while lineix < len(lines): if lines[lineix].strip().endswith('*/'): return lineix lineix += 1 return len(lines) def RemoveMultiLineCommentsFromRange(lines, begin, end): """Clears a range of lines for multi-line comments.""" # Having // dummy comments makes the lines non-empty, so we will not get # unnecessary blank line warnings later in the code. for i in range(begin, end): lines[i] = '// dummy' def RemoveMultiLineComments(filename, lines, error): """Removes multiline (c-style) comments from lines.""" lineix = 0 while lineix < len(lines): lineix_begin = FindNextMultiLineCommentStart(lines, lineix) if lineix_begin >= len(lines): return lineix_end = FindNextMultiLineCommentEnd(lines, lineix_begin) if lineix_end >= len(lines): error(filename, lineix_begin + 1, 'readability/multiline_comment', 5, 'Could not find end of multi-line comment') return RemoveMultiLineCommentsFromRange(lines, lineix_begin, lineix_end + 1) lineix = lineix_end + 1 def CleanseComments(line): """Removes //-comments and single-line C-style /* */ comments. Args: line: A line of C++ source. Returns: The line with single-line comments removed. """ commentpos = line.find('//') if commentpos != -1 and not IsCppString(line[:commentpos]): line = line[:commentpos].rstrip() # get rid of /* ... */ return _RE_PATTERN_CLEANSE_LINE_C_COMMENTS.sub('', line) class CleansedLines(object): """Holds 3 copies of all lines with different preprocessing applied to them. 1) elided member contains lines without strings and comments, 2) lines member contains lines without comments, and 3) raw_lines member contains all the lines without processing. All these three members are of , and of the same length. """ def __init__(self, lines): self.elided = [] self.lines = [] self.raw_lines = lines self.num_lines = len(lines) self.lines_without_raw_strings = CleanseRawStrings(lines) for linenum in range(len(self.lines_without_raw_strings)): self.lines.append(CleanseComments( self.lines_without_raw_strings[linenum])) elided = self._CollapseStrings(self.lines_without_raw_strings[linenum]) self.elided.append(CleanseComments(elided)) def NumLines(self): """Returns the number of lines represented.""" return self.num_lines @staticmethod def _CollapseStrings(elided): """Collapses strings and chars on a line to simple "" or '' blocks. We nix strings first so we're not fooled by text like '"http://"' Args: elided: The line being processed. Returns: The line with collapsed strings. """ if not _RE_PATTERN_INCLUDE.match(elided): # Remove escaped characters first to make quote/single quote collapsing # basic. Things that look like escaped characters shouldn't occur # outside of strings and chars. elided = _RE_PATTERN_CLEANSE_LINE_ESCAPES.sub('', elided) elided = _RE_PATTERN_CLEANSE_LINE_SINGLE_QUOTES.sub("''", elided) elided = _RE_PATTERN_CLEANSE_LINE_DOUBLE_QUOTES.sub('""', elided) return elided def FindEndOfExpressionInLine(line, startpos, depth, startchar, endchar): """Find the position just after the matching endchar. Args: line: a CleansedLines line. startpos: start searching at this position. depth: nesting level at startpos. startchar: expression opening character. endchar: expression closing character. Returns: On finding matching endchar: (index just after matching endchar, 0) Otherwise: (-1, new depth at end of this line) """ for i in xrange(startpos, len(line)): if line[i] == startchar: depth += 1 elif line[i] == endchar: depth -= 1 if depth == 0: return (i + 1, 0) return (-1, depth) def CloseExpression(clean_lines, linenum, pos): """If input points to ( or { or [ or <, finds the position that closes it. If lines[linenum][pos] points to a '(' or '{' or '[' or '<', finds the linenum/pos that correspond to the closing of the expression. Args: clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. pos: A position on the line. Returns: A tuple (line, linenum, pos) pointer *past* the closing brace, or (line, len(lines), -1) if we never find a close. Note we ignore strings and comments when matching; and the line we return is the 'cleansed' line at linenum. """ line = clean_lines.elided[linenum] startchar = line[pos] if startchar not in '({[<': return (line, clean_lines.NumLines(), -1) if startchar == '(': endchar = ')' if startchar == '[': endchar = ']' if startchar == '{': endchar = '}' if startchar == '<': endchar = '>' # Check first line (end_pos, num_open) = FindEndOfExpressionInLine( line, pos, 0, startchar, endchar) if end_pos > -1: return (line, linenum, end_pos) # Continue scanning forward while linenum < clean_lines.NumLines() - 1: linenum += 1 line = clean_lines.elided[linenum] (end_pos, num_open) = FindEndOfExpressionInLine( line, 0, num_open, startchar, endchar) if end_pos > -1: return (line, linenum, end_pos) # Did not find endchar before end of file, give up return (line, clean_lines.NumLines(), -1) def FindStartOfExpressionInLine(line, endpos, depth, startchar, endchar): """Find position at the matching startchar. This is almost the reverse of FindEndOfExpressionInLine, but note that the input position and returned position differs by 1. Args: line: a CleansedLines line. endpos: start searching at this position. depth: nesting level at endpos. startchar: expression opening character. endchar: expression closing character. Returns: On finding matching startchar: (index at matching startchar, 0) Otherwise: (-1, new depth at beginning of this line) """ for i in xrange(endpos, -1, -1): if line[i] == endchar: depth += 1 elif line[i] == startchar: depth -= 1 if depth == 0: return (i, 0) return (-1, depth) def ReverseCloseExpression(clean_lines, linenum, pos): """If input points to ) or } or ] or >, finds the position that opens it. If lines[linenum][pos] points to a ')' or '}' or ']' or '>', finds the linenum/pos that correspond to the opening of the expression. Args: clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. pos: A position on the line. Returns: A tuple (line, linenum, pos) pointer *at* the opening brace, or (line, 0, -1) if we never find the matching opening brace. Note we ignore strings and comments when matching; and the line we return is the 'cleansed' line at linenum. """ line = clean_lines.elided[linenum] endchar = line[pos] if endchar not in ')}]>': return (line, 0, -1) if endchar == ')': startchar = '(' if endchar == ']': startchar = '[' if endchar == '}': startchar = '{' if endchar == '>': startchar = '<' # Check last line (start_pos, num_open) = FindStartOfExpressionInLine( line, pos, 0, startchar, endchar) if start_pos > -1: return (line, linenum, start_pos) # Continue scanning backward while linenum > 0: linenum -= 1 line = clean_lines.elided[linenum] (start_pos, num_open) = FindStartOfExpressionInLine( line, len(line) - 1, num_open, startchar, endchar) if start_pos > -1: return (line, linenum, start_pos) # Did not find startchar before beginning of file, give up return (line, 0, -1) def CheckForCopyright(filename, lines, error): """Logs an error if no Copyright message appears at the top of the file.""" # We'll say it should occur by line 10. Don't forget there's a # dummy line at the front. for line in xrange(1, min(len(lines), 11)): if re.search(r'Copyright', lines[line], re.I): break else: # means no copyright line was found error(filename, 0, 'legal/copyright', 5, 'No copyright message found. ' 'You should have a line: "Copyright [year] "') def GetHeaderGuardCPPVariable(filename): """Returns the CPP variable that should be used as a header guard. Args: filename: The name of a C++ header file. Returns: The CPP variable that should be used as a header guard in the named file. """ # Restores original filename in case that cpplint is invoked from Emacs's # flymake. filename = re.sub(r'_flymake\.h$', '.h', filename) filename = re.sub(r'/\.flymake/([^/]*)$', r'/\1', filename) fileinfo = FileInfo(filename) file_path_from_root = fileinfo.RepositoryName() if _root: file_path_from_root = re.sub('^' + _root + os.sep, '', file_path_from_root) return re.sub(r'[-./\s]', '_', file_path_from_root).upper() + '_' def CheckForHeaderGuard(filename, lines, error): """Checks that the file contains a header guard. Logs an error if no #ifndef header guard is present. For other headers, checks that the full pathname is used. Args: filename: The name of the C++ header file. lines: An array of strings, each representing a line of the file. error: The function to call with any errors found. """ cppvar = GetHeaderGuardCPPVariable(filename) ifndef = None ifndef_linenum = 0 define = None endif = None endif_linenum = 0 for linenum, line in enumerate(lines): # Already been well guarded, no need for further checking. if line.strip() == "#pragma once": return linesplit = line.split() if len(linesplit) >= 2: # find the first occurrence of #ifndef and #define, save arg if not ifndef and linesplit[0] == '#ifndef': # set ifndef to the header guard presented on the #ifndef line. ifndef = linesplit[1] ifndef_linenum = linenum if not define and linesplit[0] == '#define': define = linesplit[1] # find the last occurrence of #endif, save entire line if line.startswith('#endif'): endif = line endif_linenum = linenum if not ifndef: error(filename, 0, 'build/header_guard', 5, 'No #ifndef header guard found, suggested CPP variable is: %s' % cppvar) return if not define: error(filename, 0, 'build/header_guard', 5, 'No #define header guard found, suggested CPP variable is: %s' % cppvar) return # The guard should be PATH_FILE_H_, but we also allow PATH_FILE_H__ # for backward compatibility. if ifndef != cppvar: error_level = 0 if ifndef != cppvar + '_': error_level = 5 ParseNolintSuppressions(filename, lines[ifndef_linenum], ifndef_linenum, error) error(filename, ifndef_linenum, 'build/header_guard', error_level, '#ifndef header guard has wrong style, please use: %s' % cppvar) if define != ifndef: error(filename, 0, 'build/header_guard', 5, '#ifndef and #define don\'t match, suggested CPP variable is: %s' % cppvar) return if endif != ('#endif // %s' % cppvar): error_level = 0 if endif != ('#endif // %s' % (cppvar + '_')): error_level = 5 ParseNolintSuppressions(filename, lines[endif_linenum], endif_linenum, error) error(filename, endif_linenum, 'build/header_guard', error_level, '#endif line should be "#endif // %s"' % cppvar) def CheckForBadCharacters(filename, lines, error): """Logs an error for each line containing bad characters. Two kinds of bad characters: 1. Unicode replacement characters: These indicate that either the file contained invalid UTF-8 (likely) or Unicode replacement characters (which it shouldn't). Note that it's possible for this to throw off line numbering if the invalid UTF-8 occurred adjacent to a newline. 2. NUL bytes. These are problematic for some tools. Args: filename: The name of the current file. lines: An array of strings, each representing a line of the file. error: The function to call with any errors found. """ for linenum, line in enumerate(lines): if u'\ufffd' in line: error(filename, linenum, 'readability/utf8', 5, 'Line contains invalid UTF-8 (or Unicode replacement character).') if '\0' in line: error(filename, linenum, 'readability/nul', 5, 'Line contains NUL byte.') def CheckForNewlineAtEOF(filename, lines, error): """Logs an error if there is no newline char at the end of the file. Args: filename: The name of the current file. lines: An array of strings, each representing a line of the file. error: The function to call with any errors found. """ # The array lines() was created by adding two newlines to the # original file (go figure), then splitting on \n. # To verify that the file ends in \n, we just have to make sure the # last-but-two element of lines() exists and is empty. if len(lines) < 3 or lines[-2]: error(filename, len(lines) - 2, 'whitespace/ending_newline', 5, 'Could not find a newline character at the end of the file.') def CheckForMultilineCommentsAndStrings(filename, clean_lines, linenum, error): """Logs an error if we see /* ... */ or "..." that extend past one line. /* ... */ comments are legit inside macros, for one line. Otherwise, we prefer // comments, so it's ok to warn about the other. Likewise, it's ok for strings to extend across multiple lines, as long as a line continuation character (backslash) terminates each line. Although not currently prohibited by the C++ style guide, it's ugly and unnecessary. We don't do well with either in this lint program, so we warn about both. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] # Remove all \\ (escaped backslashes) from the line. They are OK, and the # second (escaped) slash may trigger later \" detection erroneously. line = line.replace('\\\\', '') if line.count('/*') > line.count('*/'): error(filename, linenum, 'readability/multiline_comment', 5, 'Complex multi-line /*...*/-style comment found. ' 'Lint may give bogus warnings. ' 'Consider replacing these with //-style comments, ' 'with #if 0...#endif, ' 'or with more clearly structured multi-line comments.') if (line.count('"') - line.count('\\"')) % 2: error(filename, linenum, 'readability/multiline_string', 5, 'Multi-line string ("...") found. This lint script doesn\'t ' 'do well with such strings, and may give bogus warnings. ' 'Use C++11 raw strings or concatenation instead.') threading_list = ( ('asctime(', 'asctime_r('), ('ctime(', 'ctime_r('), ('getgrgid(', 'getgrgid_r('), ('getgrnam(', 'getgrnam_r('), ('getlogin(', 'getlogin_r('), ('getpwnam(', 'getpwnam_r('), ('getpwuid(', 'getpwuid_r('), ('gmtime(', 'gmtime_r('), ('localtime(', 'localtime_r('), ('rand(', 'rand_r('), ('strtok(', 'strtok_r('), ('ttyname(', 'ttyname_r('), ) def CheckPosixThreading(filename, clean_lines, linenum, error): """Checks for calls to thread-unsafe functions. Much code has been originally written without consideration of multi-threading. Also, engineers are relying on their old experience; they have learned posix before threading extensions were added. These tests guide the engineers to use thread-safe functions (when using posix directly). Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] for single_thread_function, multithread_safe_function in threading_list: ix = line.find(single_thread_function) # Comparisons made explicit for clarity -- pylint: disable=g-explicit-bool-comparison if ix >= 0 and (ix == 0 or (not line[ix - 1].isalnum() and line[ix - 1] not in ('_', '.', '>'))): error(filename, linenum, 'runtime/threadsafe_fn', 2, 'Consider using ' + multithread_safe_function + '...) instead of ' + single_thread_function + '...) for improved thread safety.') def CheckVlogArguments(filename, clean_lines, linenum, error): """Checks that VLOG() is only used for defining a logging level. For example, VLOG(2) is correct. VLOG(INFO), VLOG(WARNING), VLOG(ERROR), and VLOG(FATAL) are not. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] if Search(r'\bVLOG\((INFO|ERROR|WARNING|DFATAL|FATAL)\)', line): error(filename, linenum, 'runtime/vlog', 5, 'VLOG() should be used with numeric verbosity level. ' 'Use LOG() if you want symbolic severity levels.') # Matches invalid increment: *count++, which moves pointer instead of # incrementing a value. _RE_PATTERN_INVALID_INCREMENT = re.compile( r'^\s*\*\w+(\+\+|--);') def CheckInvalidIncrement(filename, clean_lines, linenum, error): """Checks for invalid increment *count++. For example following function: void increment_counter(int* count) { *count++; } is invalid, because it effectively does count++, moving pointer, and should be replaced with ++*count, (*count)++ or *count += 1. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] if _RE_PATTERN_INVALID_INCREMENT.match(line): error(filename, linenum, 'runtime/invalid_increment', 5, 'Changing pointer instead of value (or unused value of operator*).') class _BlockInfo(object): """Stores information about a generic block of code.""" def __init__(self, seen_open_brace): self.seen_open_brace = seen_open_brace self.open_parentheses = 0 self.inline_asm = _NO_ASM def CheckBegin(self, filename, clean_lines, linenum, error): """Run checks that applies to text up to the opening brace. This is mostly for checking the text after the class identifier and the "{", usually where the base class is specified. For other blocks, there isn't much to check, so we always pass. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ pass def CheckEnd(self, filename, clean_lines, linenum, error): """Run checks that applies to text after the closing brace. This is mostly used for checking end of namespace comments. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ pass class _ClassInfo(_BlockInfo): """Stores information about a class.""" def __init__(self, name, class_or_struct, clean_lines, linenum): _BlockInfo.__init__(self, False) self.name = name self.starting_linenum = linenum self.is_derived = False if class_or_struct == 'struct': self.access = 'public' self.is_struct = True else: self.access = 'private' self.is_struct = False # Remember initial indentation level for this class. Using raw_lines here # instead of elided to account for leading comments. initial_indent = Match(r'^( *)\S', clean_lines.raw_lines[linenum]) if initial_indent: self.class_indent = len(initial_indent.group(1)) else: self.class_indent = 0 # Try to find the end of the class. This will be confused by things like: # class A { # } *x = { ... # # But it's still good enough for CheckSectionSpacing. self.last_line = 0 depth = 0 for i in range(linenum, clean_lines.NumLines()): line = clean_lines.elided[i] depth += line.count('{') - line.count('}') if not depth: self.last_line = i break def CheckBegin(self, filename, clean_lines, linenum, error): # Look for a bare ':' if Search('(^|[^:]):($|[^:])', clean_lines.elided[linenum]): self.is_derived = True def CheckEnd(self, filename, clean_lines, linenum, error): # Check that closing brace is aligned with beginning of the class. # Only do this if the closing brace is indented by only whitespaces. # This means we will not check single-line class definitions. indent = Match(r'^( *)\}', clean_lines.elided[linenum]) if indent and len(indent.group(1)) != self.class_indent: if self.is_struct: parent = 'struct ' + self.name else: parent = 'class ' + self.name error(filename, linenum, 'whitespace/indent', 3, 'Closing brace should be aligned with beginning of %s' % parent) class _NamespaceInfo(_BlockInfo): """Stores information about a namespace.""" def __init__(self, name, linenum): _BlockInfo.__init__(self, False) self.name = name or '' self.starting_linenum = linenum def CheckEnd(self, filename, clean_lines, linenum, error): """Check end of namespace comments.""" line = clean_lines.raw_lines[linenum] # Check how many lines is enclosed in this namespace. Don't issue # warning for missing namespace comments if there aren't enough # lines. However, do apply checks if there is already an end of # namespace comment and it's incorrect. # # TODO(unknown): We always want to check end of namespace comments # if a namespace is large, but sometimes we also want to apply the # check if a short namespace contained nontrivial things (something # other than forward declarations). There is currently no logic on # deciding what these nontrivial things are, so this check is # triggered by namespace size only, which works most of the time. if (linenum - self.starting_linenum < 10 and not Match(r'};*\s*(//|/\*).*\bnamespace\b', line)): return # Look for matching comment at end of namespace. # # Note that we accept C style "/* */" comments for terminating # namespaces, so that code that terminate namespaces inside # preprocessor macros can be cpplint clean. # # We also accept stuff like "// end of namespace ." with the # period at the end. # # Besides these, we don't accept anything else, otherwise we might # get false negatives when existing comment is a substring of the # expected namespace. if self.name: # Named namespace if not Match((r'};*\s*(//|/\*).*\bnamespace\s+' + re.escape(self.name) + r'[\*/\.\\\s]*$'), line): error(filename, linenum, 'readability/namespace', 5, 'Namespace should be terminated with "// namespace %s"' % self.name) else: # Anonymous namespace if not Match(r'};*\s*(//|/\*).*\bnamespace[\*/\.\\\s]*$', line): error(filename, linenum, 'readability/namespace', 5, 'Namespace should be terminated with "// namespace"') class _PreprocessorInfo(object): """Stores checkpoints of nesting stacks when #if/#else is seen.""" def __init__(self, stack_before_if): # The entire nesting stack before #if self.stack_before_if = stack_before_if # The entire nesting stack up to #else self.stack_before_else = [] # Whether we have already seen #else or #elif self.seen_else = False class _NestingState(object): """Holds states related to parsing braces.""" def __init__(self): # Stack for tracking all braces. An object is pushed whenever we # see a "{", and popped when we see a "}". Only 3 types of # objects are possible: # - _ClassInfo: a class or struct. # - _NamespaceInfo: a namespace. # - _BlockInfo: some other type of block. self.stack = [] # Stack of _PreprocessorInfo objects. self.pp_stack = [] def SeenOpenBrace(self): """Check if we have seen the opening brace for the innermost block. Returns: True if we have seen the opening brace, False if the innermost block is still expecting an opening brace. """ return (not self.stack) or self.stack[-1].seen_open_brace def InNamespaceBody(self): """Check if we are currently one level inside a namespace body. Returns: True if top of the stack is a namespace block, False otherwise. """ return self.stack and isinstance(self.stack[-1], _NamespaceInfo) def UpdatePreprocessor(self, line): """Update preprocessor stack. We need to handle preprocessors due to classes like this: #ifdef SWIG struct ResultDetailsPageElementExtensionPoint { #else struct ResultDetailsPageElementExtensionPoint : public Extension { #endif We make the following assumptions (good enough for most files): - Preprocessor condition evaluates to true from #if up to first #else/#elif/#endif. - Preprocessor condition evaluates to false from #else/#elif up to #endif. We still perform lint checks on these lines, but these do not affect nesting stack. Args: line: current line to check. """ if Match(r'^\s*#\s*(if|ifdef|ifndef)\b', line): # Beginning of #if block, save the nesting stack here. The saved # stack will allow us to restore the parsing state in the #else case. self.pp_stack.append(_PreprocessorInfo(copy.deepcopy(self.stack))) elif Match(r'^\s*#\s*(else|elif)\b', line): # Beginning of #else block if self.pp_stack: if not self.pp_stack[-1].seen_else: # This is the first #else or #elif block. Remember the # whole nesting stack up to this point. This is what we # keep after the #endif. self.pp_stack[-1].seen_else = True self.pp_stack[-1].stack_before_else = copy.deepcopy(self.stack) # Restore the stack to how it was before the #if self.stack = copy.deepcopy(self.pp_stack[-1].stack_before_if) else: # TODO(unknown): unexpected #else, issue warning? pass elif Match(r'^\s*#\s*endif\b', line): # End of #if or #else blocks. if self.pp_stack: # If we saw an #else, we will need to restore the nesting # stack to its former state before the #else, otherwise we # will just continue from where we left off. if self.pp_stack[-1].seen_else: # Here we can just use a shallow copy since we are the last # reference to it. self.stack = self.pp_stack[-1].stack_before_else # Drop the corresponding #if self.pp_stack.pop() else: # TODO(unknown): unexpected #endif, issue warning? pass def Update(self, filename, clean_lines, linenum, error): """Update nesting state with current line. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] # Update pp_stack first self.UpdatePreprocessor(line) # Count parentheses. This is to avoid adding struct arguments to # the nesting stack. if self.stack: inner_block = self.stack[-1] depth_change = line.count('(') - line.count(')') inner_block.open_parentheses += depth_change # Also check if we are starting or ending an inline assembly block. if inner_block.inline_asm in (_NO_ASM, _END_ASM): if (depth_change != 0 and inner_block.open_parentheses == 1 and _MATCH_ASM.match(line)): # Enter assembly block inner_block.inline_asm = _INSIDE_ASM else: # Not entering assembly block. If previous line was _END_ASM, # we will now shift to _NO_ASM state. inner_block.inline_asm = _NO_ASM elif (inner_block.inline_asm == _INSIDE_ASM and inner_block.open_parentheses == 0): # Exit assembly block inner_block.inline_asm = _END_ASM # Consume namespace declaration at the beginning of the line. Do # this in a loop so that we catch same line declarations like this: # namespace proto2 { namespace bridge { class MessageSet; } } while True: # Match start of namespace. The "\b\s*" below catches namespace # declarations even if it weren't followed by a whitespace, this # is so that we don't confuse our namespace checker. The # missing spaces will be flagged by CheckSpacing. namespace_decl_match = Match(r'^\s*namespace\b\s*([:\w]+)?(.*)$', line) if not namespace_decl_match: break new_namespace = _NamespaceInfo(namespace_decl_match.group(1), linenum) self.stack.append(new_namespace) line = namespace_decl_match.group(2) if line.find('{') != -1: new_namespace.seen_open_brace = True line = line[line.find('{') + 1:] # Look for a class declaration in whatever is left of the line # after parsing namespaces. The regexp accounts for decorated classes # such as in: # class LOCKABLE API Object { # }; # # Templates with class arguments may confuse the parser, for example: # template , # class Vector = vector > # class HeapQueue { # # Because this parser has no nesting state about templates, by the # time it saw "class Comparator", it may think that it's a new class. # Nested templates have a similar problem: # template < # typename ExportedType, # typename TupleType, # template class ImplTemplate> # # To avoid these cases, we ignore classes that are followed by '=' or '>' class_decl_match = Match( r'\s*(template\s*<[\w\s<>,:]*>\s*)?' r'(class|struct)\s+([A-Z_]+\s+)*(\w+(?:::\w+)*)' r'(([^=>]|<[^<>]*>|<[^<>]*<[^<>]*>\s*>)*)$', line) if (class_decl_match and (not self.stack or self.stack[-1].open_parentheses == 0)): self.stack.append(_ClassInfo( class_decl_match.group(4), class_decl_match.group(2), clean_lines, linenum)) line = class_decl_match.group(5) # If we have not yet seen the opening brace for the innermost block, # run checks here. if not self.SeenOpenBrace(): self.stack[-1].CheckBegin(filename, clean_lines, linenum, error) # Update access control if we are inside a class/struct if self.stack and isinstance(self.stack[-1], _ClassInfo): classinfo = self.stack[-1] access_match = Match( r'^(.*)\b(public|private|protected|signals)(\s+(?:slots\s*)?)?' r':(?:[^:]|$)', line) if access_match: classinfo.access = access_match.group(2) # Check that access keywords are indented +1 space. Skip this # check if the keywords are not preceded by whitespaces. indent = access_match.group(1) if (len(indent) != classinfo.class_indent + 1 and Match(r'^\s*$', indent)): if classinfo.is_struct: parent = 'struct ' + classinfo.name else: parent = 'class ' + classinfo.name slots = '' if access_match.group(3): slots = access_match.group(3) error(filename, linenum, 'whitespace/indent', 3, '%s%s: should be indented +1 space inside %s' % ( access_match.group(2), slots, parent)) # Consume braces or semicolons from what's left of the line while True: # Match first brace, semicolon, or closed parenthesis. matched = Match(r'^[^{;)}]*([{;)}])(.*)$', line) if not matched: break token = matched.group(1) if token == '{': # If namespace or class hasn't seen a opening brace yet, mark # namespace/class head as complete. Push a new block onto the # stack otherwise. if not self.SeenOpenBrace(): self.stack[-1].seen_open_brace = True else: self.stack.append(_BlockInfo(True)) if _MATCH_ASM.match(line): self.stack[-1].inline_asm = _BLOCK_ASM elif token == ';' or token == ')': # If we haven't seen an opening brace yet, but we already saw # a semicolon, this is probably a forward declaration. Pop # the stack for these. # # Similarly, if we haven't seen an opening brace yet, but we # already saw a closing parenthesis, then these are probably # function arguments with extra "class" or "struct" keywords. # Also pop these stack for these. if not self.SeenOpenBrace(): self.stack.pop() else: # token == '}' # Perform end of block checks and pop the stack. if self.stack: self.stack[-1].CheckEnd(filename, clean_lines, linenum, error) self.stack.pop() line = matched.group(2) def InnermostClass(self): """Get class info on the top of the stack. Returns: A _ClassInfo object if we are inside a class, or None otherwise. """ for i in range(len(self.stack), 0, -1): classinfo = self.stack[i - 1] if isinstance(classinfo, _ClassInfo): return classinfo return None def CheckCompletedBlocks(self, filename, error): """Checks that all classes and namespaces have been completely parsed. Call this when all lines in a file have been processed. Args: filename: The name of the current file. error: The function to call with any errors found. """ # Note: This test can result in false positives if #ifdef constructs # get in the way of brace matching. See the testBuildClass test in # cpplint_unittest.py for an example of this. for obj in self.stack: if isinstance(obj, _ClassInfo): error(filename, obj.starting_linenum, 'build/class', 5, 'Failed to find complete declaration of class %s' % obj.name) elif isinstance(obj, _NamespaceInfo): error(filename, obj.starting_linenum, 'build/namespaces', 5, 'Failed to find complete declaration of namespace %s' % obj.name) def CheckForNonStandardConstructs(filename, clean_lines, linenum, nesting_state, error): r"""Logs an error if we see certain non-ANSI constructs ignored by gcc-2. Complain about several constructs which gcc-2 accepts, but which are not standard C++. Warning about these in lint is one way to ease the transition to new compilers. - put storage class first (e.g. "static const" instead of "const static"). - "%lld" instead of %qd" in printf-type functions. - "%1$d" is non-standard in printf-type functions. - "\%" is an undefined character escape sequence. - text after #endif is not allowed. - invalid inner-style forward declaration. - >? and ?= and )\?=?\s*(\w+|[+-]?\d+)(\.\d*)?', line): error(filename, linenum, 'build/deprecated', 3, '>? and ))?' # r'\s*const\s*' + type_name + '\s*&\s*\w+\s*;' error(filename, linenum, 'runtime/member_string_references', 2, 'const string& members are dangerous. It is much better to use ' 'alternatives, such as pointers or simple constants.') # Everything else in this function operates on class declarations. # Return early if the top of the nesting stack is not a class, or if # the class head is not completed yet. classinfo = nesting_state.InnermostClass() if not classinfo or not classinfo.seen_open_brace: return # The class may have been declared with namespace or classname qualifiers. # The constructor and destructor will not have those qualifiers. base_classname = classinfo.name.split('::')[-1] # Look for single-argument constructors that aren't marked explicit. # Technically a valid construct, but against style. args = Match(r'\s+(?:inline\s+)?%s\s*\(([^,()]+)\)' % re.escape(base_classname), line) if (args and args.group(1) != 'void' and not Match(r'(const\s+)?%s(\s+const)?\s*(?:<\w+>\s*)?&' % re.escape(base_classname), args.group(1).strip())): error(filename, linenum, 'runtime/explicit', 5, 'Single-argument constructors should be marked explicit.') def CheckSpacingForFunctionCall(filename, line, linenum, error): """Checks for the correctness of various spacing around function calls. Args: filename: The name of the current file. line: The text of the line to check. linenum: The number of the line to check. error: The function to call with any errors found. """ # Since function calls often occur inside if/for/while/switch # expressions - which have their own, more liberal conventions - we # first see if we should be looking inside such an expression for a # function call, to which we can apply more strict standards. fncall = line # if there's no control flow construct, look at whole line for pattern in (r'\bif\s*\((.*)\)\s*{', r'\bfor\s*\((.*)\)\s*{', r'\bwhile\s*\((.*)\)\s*[{;]', r'\bswitch\s*\((.*)\)\s*{'): match = Search(pattern, line) if match: fncall = match.group(1) # look inside the parens for function calls break # Except in if/for/while/switch, there should never be space # immediately inside parens (eg "f( 3, 4 )"). We make an exception # for nested parens ( (a+b) + c ). Likewise, there should never be # a space before a ( when it's a function argument. I assume it's a # function argument when the char before the whitespace is legal in # a function name (alnum + _) and we're not starting a macro. Also ignore # pointers and references to arrays and functions coz they're too tricky: # we use a very simple way to recognize these: # " (something)(maybe-something)" or # " (something)(maybe-something," or # " (something)[something]" # Note that we assume the contents of [] to be short enough that # they'll never need to wrap. if ( # Ignore control structures. not Search(r'\b(if|for|while|switch|return|new|delete|catch|sizeof)\b', fncall) and # Ignore pointers/references to functions. not Search(r' \([^)]+\)\([^)]*(\)|,$)', fncall) and # Ignore pointers/references to arrays. not Search(r' \([^)]+\)\[[^\]]+\]', fncall)): if Search(r'\w\s*\(\s(?!\s*\\$)', fncall): # a ( used for a fn call error(filename, linenum, 'whitespace/parens', 4, 'Extra space after ( in function call') elif Search(r'\(\s+(?!(\s*\\)|\()', fncall): error(filename, linenum, 'whitespace/parens', 2, 'Extra space after (') if (Search(r'\w\s+\(', fncall) and not Search(r'#\s*define|typedef', fncall) and not Search(r'\w\s+\((\w+::)*\*\w+\)\(', fncall)): error(filename, linenum, 'whitespace/parens', 4, 'Extra space before ( in function call') # If the ) is followed only by a newline or a { + newline, assume it's # part of a control statement (if/while/etc), and don't complain if Search(r'[^)]\s+\)\s*[^{\s]', fncall): # If the closing parenthesis is preceded by only whitespaces, # try to give a more descriptive error message. if Search(r'^\s+\)', fncall): error(filename, linenum, 'whitespace/parens', 2, 'Closing ) should be moved to the previous line') else: error(filename, linenum, 'whitespace/parens', 2, 'Extra space before )') def IsBlankLine(line): """Returns true if the given line is blank. We consider a line to be blank if the line is empty or consists of only white spaces. Args: line: A line of a string. Returns: True, if the given line is blank. """ return not line or line.isspace() def CheckForFunctionLengths(filename, clean_lines, linenum, function_state, error): """Reports for long function bodies. For an overview why this is done, see: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Write_Short_Functions Uses a simplistic algorithm assuming other style guidelines (especially spacing) are followed. Only checks unindented functions, so class members are unchecked. Trivial bodies are unchecked, so constructors with huge initializer lists may be missed. Blank/comment lines are not counted so as to avoid encouraging the removal of vertical space and comments just to get through a lint check. NOLINT *on the last line of a function* disables this check. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. function_state: Current function name and lines in body so far. error: The function to call with any errors found. """ lines = clean_lines.lines line = lines[linenum] raw = clean_lines.raw_lines raw_line = raw[linenum] joined_line = '' starting_func = False regexp = r'(\w(\w|::|\*|\&|\s)*)\(' # decls * & space::name( ... match_result = Match(regexp, line) if match_result: # If the name is all caps and underscores, figure it's a macro and # ignore it, unless it's TEST or TEST_F. function_name = match_result.group(1).split()[-1] if function_name == 'TEST' or function_name == 'TEST_F' or ( not Match(r'[A-Z_]+$', function_name)): starting_func = True if starting_func: body_found = False for start_linenum in xrange(linenum, clean_lines.NumLines()): start_line = lines[start_linenum] joined_line += ' ' + start_line.lstrip() if Search(r'(;|})', start_line): # Declarations and trivial functions body_found = True break # ... ignore elif Search(r'{', start_line): body_found = True function = Search(r'((\w|:)*)\(', line).group(1) if Match(r'TEST', function): # Handle TEST... macros parameter_regexp = Search(r'(\(.*\))', joined_line) if parameter_regexp: # Ignore bad syntax function += parameter_regexp.group(1) else: function += '()' function_state.Begin(function) break if not body_found: # No body for the function (or evidence of a non-function) was found. error(filename, linenum, 'readability/fn_size', 5, 'Lint failed to find start of function body.') elif Match(r'^\}\s*$', line): # function end function_state.Check(error, filename, linenum) function_state.End() elif not Match(r'^\s*$', line): function_state.Count() # Count non-blank/non-comment lines. _RE_PATTERN_TODO = re.compile(r'^//(\s*)TODO(\(.+?\))?:?(\s|$)?') def CheckComment(comment, filename, linenum, error): """Checks for common mistakes in TODO comments. Args: comment: The text of the comment from the line in question. filename: The name of the current file. linenum: The number of the line to check. error: The function to call with any errors found. """ match = _RE_PATTERN_TODO.match(comment) if match: # One whitespace is correct; zero whitespace is handled elsewhere. leading_whitespace = match.group(1) if len(leading_whitespace) > 1: error(filename, linenum, 'whitespace/todo', 2, 'Too many spaces before TODO') username = match.group(2) if not username: error(filename, linenum, 'readability/todo', 2, 'Missing username in TODO; it should look like ' '"// TODO(my_username): Stuff."') middle_whitespace = match.group(3) # Comparisons made explicit for correctness -- pylint: disable=g-explicit-bool-comparison if middle_whitespace != ' ' and middle_whitespace != '': error(filename, linenum, 'whitespace/todo', 2, 'TODO(my_username) should be followed by a space') def CheckAccess(filename, clean_lines, linenum, nesting_state, error): """Checks for improper use of DISALLOW* macros. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. nesting_state: A _NestingState instance which maintains information about the current stack of nested blocks being parsed. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] # get rid of comments and strings matched = Match((r'\s*(DISALLOW_COPY_AND_ASSIGN|' r'DISALLOW_EVIL_CONSTRUCTORS|' r'DISALLOW_IMPLICIT_CONSTRUCTORS)'), line) if not matched: return if nesting_state.stack and isinstance(nesting_state.stack[-1], _ClassInfo): if nesting_state.stack[-1].access != 'private': error(filename, linenum, 'readability/constructors', 3, '%s must be in the private: section' % matched.group(1)) else: # Found DISALLOW* macro outside a class declaration, or perhaps it # was used inside a function when it should have been part of the # class declaration. We could issue a warning here, but it # probably resulted in a compiler error already. pass def FindNextMatchingAngleBracket(clean_lines, linenum, init_suffix): """Find the corresponding > to close a template. Args: clean_lines: A CleansedLines instance containing the file. linenum: Current line number. init_suffix: Remainder of the current line after the initial <. Returns: True if a matching bracket exists. """ line = init_suffix nesting_stack = ['<'] while True: # Find the next operator that can tell us whether < is used as an # opening bracket or as a less-than operator. We only want to # warn on the latter case. # # We could also check all other operators and terminate the search # early, e.g. if we got something like this "a(),;\[\]]*([<>(),;\[\]])(.*)$', line) if match: # Found an operator, update nesting stack operator = match.group(1) line = match.group(2) if nesting_stack[-1] == '<': # Expecting closing angle bracket if operator in ('<', '(', '['): nesting_stack.append(operator) elif operator == '>': nesting_stack.pop() if not nesting_stack: # Found matching angle bracket return True elif operator == ',': # Got a comma after a bracket, this is most likely a template # argument. We have not seen a closing angle bracket yet, but # it's probably a few lines later if we look for it, so just # return early here. return True else: # Got some other operator. return False else: # Expecting closing parenthesis or closing bracket if operator in ('<', '(', '['): nesting_stack.append(operator) elif operator in (')', ']'): # We don't bother checking for matching () or []. If we got # something like (] or [), it would have been a syntax error. nesting_stack.pop() else: # Scan the next line linenum += 1 if linenum >= len(clean_lines.elided): break line = clean_lines.elided[linenum] # Exhausted all remaining lines and still no matching angle bracket. # Most likely the input was incomplete, otherwise we should have # seen a semicolon and returned early. return True def FindPreviousMatchingAngleBracket(clean_lines, linenum, init_prefix): """Find the corresponding < that started a template. Args: clean_lines: A CleansedLines instance containing the file. linenum: Current line number. init_prefix: Part of the current line before the initial >. Returns: True if a matching bracket exists. """ line = init_prefix nesting_stack = ['>'] while True: # Find the previous operator match = Search(r'^(.*)([<>(),;\[\]])[^<>(),;\[\]]*$', line) if match: # Found an operator, update nesting stack operator = match.group(2) line = match.group(1) if nesting_stack[-1] == '>': # Expecting opening angle bracket if operator in ('>', ')', ']'): nesting_stack.append(operator) elif operator == '<': nesting_stack.pop() if not nesting_stack: # Found matching angle bracket return True elif operator == ',': # Got a comma before a bracket, this is most likely a # template argument. The opening angle bracket is probably # there if we look for it, so just return early here. return True else: # Got some other operator. return False else: # Expecting opening parenthesis or opening bracket if operator in ('>', ')', ']'): nesting_stack.append(operator) elif operator in ('(', '['): nesting_stack.pop() else: # Scan the previous line linenum -= 1 if linenum < 0: break line = clean_lines.elided[linenum] # Exhausted all earlier lines and still no matching angle bracket. return False def CheckSpacing(filename, clean_lines, linenum, nesting_state, error): """Checks for the correctness of various spacing issues in the code. Things we check for: spaces around operators, spaces after if/for/while/switch, no spaces around parens in function calls, two spaces between code and comment, don't start a block with a blank line, don't end a function with a blank line, don't add a blank line after public/protected/private, don't have too many blank lines in a row. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. nesting_state: A _NestingState instance which maintains information about the current stack of nested blocks being parsed. error: The function to call with any errors found. """ # Don't use "elided" lines here, otherwise we can't check commented lines. # Don't want to use "raw" either, because we don't want to check inside C++11 # raw strings, raw = clean_lines.lines_without_raw_strings line = raw[linenum] # Before nixing comments, check if the line is blank for no good # reason. This includes the first line after a block is opened, and # blank lines at the end of a function (ie, right before a line like '}' # # Skip all the blank line checks if we are immediately inside a # namespace body. In other words, don't issue blank line warnings # for this block: # namespace { # # } # # A warning about missing end of namespace comments will be issued instead. if IsBlankLine(line) and not nesting_state.InNamespaceBody(): elided = clean_lines.elided prev_line = elided[linenum - 1] prevbrace = prev_line.rfind('{') # TODO(unknown): Don't complain if line before blank line, and line after, # both start with alnums and are indented the same amount. # This ignores whitespace at the start of a namespace block # because those are not usually indented. if prevbrace != -1 and prev_line[prevbrace:].find('}') == -1: # OK, we have a blank line at the start of a code block. Before we # complain, we check if it is an exception to the rule: The previous # non-empty line has the parameters of a function header that are indented # 4 spaces (because they did not fit in a 80 column line when placed on # the same line as the function name). We also check for the case where # the previous line is indented 6 spaces, which may happen when the # initializers of a constructor do not fit into a 80 column line. exception = False if Match(r' {6}\w', prev_line): # Initializer list? # We are looking for the opening column of initializer list, which # should be indented 4 spaces to cause 6 space indentation afterwards. search_position = linenum-2 while (search_position >= 0 and Match(r' {6}\w', elided[search_position])): search_position -= 1 exception = (search_position >= 0 and elided[search_position][:5] == ' :') else: # Search for the function arguments or an initializer list. We use a # simple heuristic here: If the line is indented 4 spaces; and we have a # closing paren, without the opening paren, followed by an opening brace # or colon (for initializer lists) we assume that it is the last line of # a function header. If we have a colon indented 4 spaces, it is an # initializer list. exception = (Match(r' {4}\w[^\(]*\)\s*(const\s*)?(\{\s*$|:)', prev_line) or Match(r' {4}:', prev_line)) if not exception: error(filename, linenum, 'whitespace/blank_line', 2, 'Redundant blank line at the start of a code block ' 'should be deleted.') # Ignore blank lines at the end of a block in a long if-else # chain, like this: # if (condition1) { # // Something followed by a blank line # # } else if (condition2) { # // Something else # } if linenum + 1 < clean_lines.NumLines(): next_line = raw[linenum + 1] if (next_line and Match(r'\s*}', next_line) and next_line.find('} else ') == -1): error(filename, linenum, 'whitespace/blank_line', 3, 'Redundant blank line at the end of a code block ' 'should be deleted.') matched = Match(r'\s*(public|protected|private):', prev_line) if matched: error(filename, linenum, 'whitespace/blank_line', 3, 'Do not leave a blank line after "%s:"' % matched.group(1)) # Next, we complain if there's a comment too near the text commentpos = line.find('//') if commentpos != -1: # Check if the // may be in quotes. If so, ignore it # Comparisons made explicit for clarity -- pylint: disable=g-explicit-bool-comparison if (line.count('"', 0, commentpos) - line.count('\\"', 0, commentpos)) % 2 == 0: # not in quotes # Allow one space for new scopes, two spaces otherwise: if (not Match(r'^\s*{ //', line) and ((commentpos >= 1 and line[commentpos-1] not in string.whitespace) or (commentpos >= 2 and line[commentpos-2] not in string.whitespace))): error(filename, linenum, 'whitespace/comments', 2, 'At least two spaces is best between code and comments') # There should always be a space between the // and the comment commentend = commentpos + 2 if commentend < len(line) and not line[commentend] == ' ': # but some lines are exceptions -- e.g. if they're big # comment delimiters like: # //---------------------------------------------------------- # or are an empty C++ style Doxygen comment, like: # /// # or C++ style Doxygen comments placed after the variable: # ///< Header comment # //!< Header comment # or they begin with multiple slashes followed by a space: # //////// Header comment match = (Search(r'[=/-]{4,}\s*$', line[commentend:]) or Search(r'^/$', line[commentend:]) or Search(r'^!< ', line[commentend:]) or Search(r'^/< ', line[commentend:]) or Search(r'^/+ ', line[commentend:])) if not match: error(filename, linenum, 'whitespace/comments', 4, 'Should have a space between // and comment') CheckComment(line[commentpos:], filename, linenum, error) line = clean_lines.elided[linenum] # get rid of comments and strings # Don't try to do spacing checks for operator methods line = re.sub(r'operator(==|!=|<|<<|<=|>=|>>|>)\(', 'operator\(', line) # We allow no-spaces around = within an if: "if ( (a=Foo()) == 0 )". # Otherwise not. Note we only check for non-spaces on *both* sides; # sometimes people put non-spaces on one side when aligning ='s among # many lines (not that this is behavior that I approve of...) if Search(r'[\w.]=[\w.]', line) and not Search(r'\b(if|while) ', line): error(filename, linenum, 'whitespace/operators', 4, 'Missing spaces around =') # It's ok not to have spaces around binary operators like + - * /, but if # there's too little whitespace, we get concerned. It's hard to tell, # though, so we punt on this one for now. TODO. # You should always have whitespace around binary operators. # # Check <= and >= first to avoid false positives with < and >, then # check non-include lines for spacing around < and >. match = Search(r'[^<>=!\s](==|!=|<=|>=)[^<>=!\s]', line) if match: error(filename, linenum, 'whitespace/operators', 3, 'Missing spaces around %s' % match.group(1)) # We allow no-spaces around << when used like this: 10<<20, but # not otherwise (particularly, not when used as streams) # Also ignore using ns::operator<<; match = Search(r'(operator|\S)(?:L|UL|ULL|l|ul|ull)?<<(\S)', line) if (match and not (match.group(1).isdigit() and match.group(2).isdigit()) and not (match.group(1) == 'operator' and match.group(2) == ';')): error(filename, linenum, 'whitespace/operators', 3, 'Missing spaces around <<') elif not Match(r'#.*include', line): # Avoid false positives on -> reduced_line = line.replace('->', '') # Look for < that is not surrounded by spaces. This is only # triggered if both sides are missing spaces, even though # technically should should flag if at least one side is missing a # space. This is done to avoid some false positives with shifts. match = Search(r'[^\s<]<([^\s=<].*)', reduced_line) if (match and not FindNextMatchingAngleBracket(clean_lines, linenum, match.group(1))): error(filename, linenum, 'whitespace/operators', 3, 'Missing spaces around <') # Look for > that is not surrounded by spaces. Similar to the # above, we only trigger if both sides are missing spaces to avoid # false positives with shifts. match = Search(r'^(.*[^\s>])>[^\s=>]', reduced_line) if (match and not FindPreviousMatchingAngleBracket(clean_lines, linenum, match.group(1))): error(filename, linenum, 'whitespace/operators', 3, 'Missing spaces around >') # We allow no-spaces around >> for almost anything. This is because # C++11 allows ">>" to close nested templates, which accounts for # most cases when ">>" is not followed by a space. # # We still warn on ">>" followed by alpha character, because that is # likely due to ">>" being used for right shifts, e.g.: # value >> alpha # # When ">>" is used to close templates, the alphanumeric letter that # follows would be part of an identifier, and there should still be # a space separating the template type and the identifier. # type> alpha match = Search(r'>>[a-zA-Z_]', line) if match: error(filename, linenum, 'whitespace/operators', 3, 'Missing spaces around >>') # There shouldn't be space around unary operators match = Search(r'(!\s|~\s|[\s]--[\s;]|[\s]\+\+[\s;])', line) if match: error(filename, linenum, 'whitespace/operators', 4, 'Extra space for operator %s' % match.group(1)) # A pet peeve of mine: no spaces after an if, while, switch, or for match = Search(r' (if\(|for\(|while\(|switch\()', line) if match: error(filename, linenum, 'whitespace/parens', 5, 'Missing space before ( in %s' % match.group(1)) # For if/for/while/switch, the left and right parens should be # consistent about how many spaces are inside the parens, and # there should either be zero or one spaces inside the parens. # We don't want: "if ( foo)" or "if ( foo )". # Exception: "for ( ; foo; bar)" and "for (foo; bar; )" are allowed. match = Search(r'\b(if|for|while|switch)\s*' r'\(([ ]*)(.).*[^ ]+([ ]*)\)\s*{\s*$', line) if match: if len(match.group(2)) != len(match.group(4)): if not (match.group(3) == ';' and len(match.group(2)) == 1 + len(match.group(4)) or not match.group(2) and Search(r'\bfor\s*\(.*; \)', line)): error(filename, linenum, 'whitespace/parens', 5, 'Mismatching spaces inside () in %s' % match.group(1)) if len(match.group(2)) not in [0, 1]: error(filename, linenum, 'whitespace/parens', 5, 'Should have zero or one spaces inside ( and ) in %s' % match.group(1)) # You should always have a space after a comma (either as fn arg or operator) # # This does not apply when the non-space character following the # comma is another comma, since the only time when that happens is # for empty macro arguments. # # We run this check in two passes: first pass on elided lines to # verify that lines contain missing whitespaces, second pass on raw # lines to confirm that those missing whitespaces are not due to # elided comments. if Search(r',[^,\s]', line) and Search(r',[^,\s]', raw[linenum]): error(filename, linenum, 'whitespace/comma', 3, 'Missing space after ,') # You should always have a space after a semicolon # except for few corner cases # TODO(unknown): clarify if 'if (1) { return 1;}' is requires one more # space after ; if Search(r';[^\s};\\)/]', line): error(filename, linenum, 'whitespace/semicolon', 3, 'Missing space after ;') # Next we will look for issues with function calls. CheckSpacingForFunctionCall(filename, line, linenum, error) # Except after an opening paren, or after another opening brace (in case of # an initializer list, for instance), you should have spaces before your # braces. And since you should never have braces at the beginning of a line, # this is an easy test. match = Match(r'^(.*[^ ({]){', line) if match: # Try a bit harder to check for brace initialization. This # happens in one of the following forms: # Constructor() : initializer_list_{} { ... } # Constructor{}.MemberFunction() # Type variable{}; # FunctionCall(type{}, ...); # LastArgument(..., type{}); # LOG(INFO) << type{} << " ..."; # map_of_type[{...}] = ...; # # We check for the character following the closing brace, and # silence the warning if it's one of those listed above, i.e. # "{.;,)<]". # # To account for nested initializer list, we allow any number of # closing braces up to "{;,)<". We can't simply silence the # warning on first sight of closing brace, because that would # cause false negatives for things that are not initializer lists. # Silence this: But not this: # Outer{ if (...) { # Inner{...} if (...){ // Missing space before { # }; } # # There is a false negative with this approach if people inserted # spurious semicolons, e.g. "if (cond){};", but we will catch the # spurious semicolon with a separate check. (endline, endlinenum, endpos) = CloseExpression( clean_lines, linenum, len(match.group(1))) trailing_text = '' if endpos > -1: trailing_text = endline[endpos:] for offset in xrange(endlinenum + 1, min(endlinenum + 3, clean_lines.NumLines() - 1)): trailing_text += clean_lines.elided[offset] if not Match(r'^[\s}]*[{.;,)<\]]', trailing_text): error(filename, linenum, 'whitespace/braces', 5, 'Missing space before {') # Make sure '} else {' has spaces. if Search(r'}else', line): error(filename, linenum, 'whitespace/braces', 5, 'Missing space before else') # You shouldn't have spaces before your brackets, except maybe after # 'delete []' or 'new char * []'. if Search(r'\w\s+\[', line) and not Search(r'delete\s+\[', line): error(filename, linenum, 'whitespace/braces', 5, 'Extra space before [') # You shouldn't have a space before a semicolon at the end of the line. # There's a special case for "for" since the style guide allows space before # the semicolon there. if Search(r':\s*;\s*$', line): error(filename, linenum, 'whitespace/semicolon', 5, 'Semicolon defining empty statement. Use {} instead.') elif Search(r'^\s*;\s*$', line): error(filename, linenum, 'whitespace/semicolon', 5, 'Line contains only semicolon. If this should be an empty statement, ' 'use {} instead.') elif (Search(r'\s+;\s*$', line) and not Search(r'\bfor\b', line)): error(filename, linenum, 'whitespace/semicolon', 5, 'Extra space before last semicolon. If this should be an empty ' 'statement, use {} instead.') # In range-based for, we wanted spaces before and after the colon, but # not around "::" tokens that might appear. if (Search('for *\(.*[^:]:[^: ]', line) or Search('for *\(.*[^: ]:[^:]', line)): error(filename, linenum, 'whitespace/forcolon', 2, 'Missing space around colon in range-based for loop') def CheckSectionSpacing(filename, clean_lines, class_info, linenum, error): """Checks for additional blank line issues related to sections. Currently the only thing checked here is blank line before protected/private. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. class_info: A _ClassInfo objects. linenum: The number of the line to check. error: The function to call with any errors found. """ # Skip checks if the class is small, where small means 25 lines or less. # 25 lines seems like a good cutoff since that's the usual height of # terminals, and any class that can't fit in one screen can't really # be considered "small". # # Also skip checks if we are on the first line. This accounts for # classes that look like # class Foo { public: ... }; # # If we didn't find the end of the class, last_line would be zero, # and the check will be skipped by the first condition. if (class_info.last_line - class_info.starting_linenum <= 24 or linenum <= class_info.starting_linenum): return matched = Match(r'\s*(public|protected|private):', clean_lines.lines[linenum]) if matched: # Issue warning if the line before public/protected/private was # not a blank line, but don't do this if the previous line contains # "class" or "struct". This can happen two ways: # - We are at the beginning of the class. # - We are forward-declaring an inner class that is semantically # private, but needed to be public for implementation reasons. # Also ignores cases where the previous line ends with a backslash as can be # common when defining classes in C macros. prev_line = clean_lines.lines[linenum - 1] if (not IsBlankLine(prev_line) and not Search(r'\b(class|struct)\b', prev_line) and not Search(r'\\$', prev_line)): # Try a bit harder to find the beginning of the class. This is to # account for multi-line base-specifier lists, e.g.: # class Derived # : public Base { end_class_head = class_info.starting_linenum for i in range(class_info.starting_linenum, linenum): if Search(r'\{\s*$', clean_lines.lines[i]): end_class_head = i break if end_class_head < linenum - 1: error(filename, linenum, 'whitespace/blank_line', 3, '"%s:" should be preceded by a blank line' % matched.group(1)) def GetPreviousNonBlankLine(clean_lines, linenum): """Return the most recent non-blank line and its line number. Args: clean_lines: A CleansedLines instance containing the file contents. linenum: The number of the line to check. Returns: A tuple with two elements. The first element is the contents of the last non-blank line before the current line, or the empty string if this is the first non-blank line. The second is the line number of that line, or -1 if this is the first non-blank line. """ prevlinenum = linenum - 1 while prevlinenum >= 0: prevline = clean_lines.elided[prevlinenum] if not IsBlankLine(prevline): # if not a blank line... return (prevline, prevlinenum) prevlinenum -= 1 return ('', -1) def CheckBraces(filename, clean_lines, linenum, error): """Looks for misplaced braces (e.g. at the end of line). Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] # get rid of comments and strings if Match(r'\s*{\s*$', line): # We allow an open brace to start a line in the case where someone is using # braces in a block to explicitly create a new scope, which is commonly used # to control the lifetime of stack-allocated variables. Braces are also # used for brace initializers inside function calls. We don't detect this # perfectly: we just don't complain if the last non-whitespace character on # the previous non-blank line is ',', ';', ':', '(', '{', or '}', or if the # previous line starts a preprocessor block. prevline = GetPreviousNonBlankLine(clean_lines, linenum)[0] if (not Search(r'[,;:}{(]\s*$', prevline) and not Match(r'\s*#', prevline)): error(filename, linenum, 'whitespace/braces', 4, '{ should almost always be at the end of the previous line') # An else clause should be on the same line as the preceding closing brace. if Match(r'\s*else\s*', line): prevline = GetPreviousNonBlankLine(clean_lines, linenum)[0] if Match(r'\s*}\s*$', prevline): error(filename, linenum, 'whitespace/newline', 4, 'An else should appear on the same line as the preceding }') # If braces come on one side of an else, they should be on both. # However, we have to worry about "else if" that spans multiple lines! if Search(r'}\s*else[^{]*$', line) or Match(r'[^}]*else\s*{', line): if Search(r'}\s*else if([^{]*)$', line): # could be multi-line if # find the ( after the if pos = line.find('else if') pos = line.find('(', pos) if pos > 0: (endline, _, endpos) = CloseExpression(clean_lines, linenum, pos) if endline[endpos:].find('{') == -1: # must be brace after if error(filename, linenum, 'readability/braces', 5, 'If an else has a brace on one side, it should have it on both') else: # common case: else not followed by a multi-line if error(filename, linenum, 'readability/braces', 5, 'If an else has a brace on one side, it should have it on both') # Likewise, an else should never have the else clause on the same line if Search(r'\belse [^\s{]', line) and not Search(r'\belse if\b', line): error(filename, linenum, 'whitespace/newline', 4, 'Else clause should never be on same line as else (use 2 lines)') # In the same way, a do/while should never be on one line if Match(r'\s*do [^\s{]', line): error(filename, linenum, 'whitespace/newline', 4, 'do/while clauses should not be on a single line') # Block bodies should not be followed by a semicolon. Due to C++11 # brace initialization, there are more places where semicolons are # required than not, so we use a whitelist approach to check these # rather than a blacklist. These are the places where "};" should # be replaced by just "}": # 1. Some flavor of block following closing parenthesis: # for (;;) {}; # while (...) {}; # switch (...) {}; # Function(...) {}; # if (...) {}; # if (...) else if (...) {}; # # 2. else block: # if (...) else {}; # # 3. const member function: # Function(...) const {}; # # 4. Block following some statement: # x = 42; # {}; # # 5. Block at the beginning of a function: # Function(...) { # {}; # } # # Note that naively checking for the preceding "{" will also match # braces inside multi-dimensional arrays, but this is fine since # that expression will not contain semicolons. # # 6. Block following another block: # while (true) {} # {}; # # 7. End of namespaces: # namespace {}; # # These semicolons seems far more common than other kinds of # redundant semicolons, possibly due to people converting classes # to namespaces. For now we do not warn for this case. # # Try matching case 1 first. match = Match(r'^(.*\)\s*)\{', line) if match: # Matched closing parenthesis (case 1). Check the token before the # matching opening parenthesis, and don't warn if it looks like a # macro. This avoids these false positives: # - macro that defines a base class # - multi-line macro that defines a base class # - macro that defines the whole class-head # # But we still issue warnings for macros that we know are safe to # warn, specifically: # - TEST, TEST_F, TEST_P, MATCHER, MATCHER_P # - TYPED_TEST # - INTERFACE_DEF # - EXCLUSIVE_LOCKS_REQUIRED, SHARED_LOCKS_REQUIRED, LOCKS_EXCLUDED: # # We implement a whitelist of safe macros instead of a blacklist of # unsafe macros, even though the latter appears less frequently in # google code and would have been easier to implement. This is because # the downside for getting the whitelist wrong means some extra # semicolons, while the downside for getting the blacklist wrong # would result in compile errors. # # In addition to macros, we also don't want to warn on compound # literals. closing_brace_pos = match.group(1).rfind(')') opening_parenthesis = ReverseCloseExpression( clean_lines, linenum, closing_brace_pos) if opening_parenthesis[2] > -1: line_prefix = opening_parenthesis[0][0:opening_parenthesis[2]] macro = Search(r'\b([A-Z_]+)\s*$', line_prefix) if ((macro and macro.group(1) not in ( 'TEST', 'TEST_F', 'MATCHER', 'MATCHER_P', 'TYPED_TEST', 'EXCLUSIVE_LOCKS_REQUIRED', 'SHARED_LOCKS_REQUIRED', 'LOCKS_EXCLUDED', 'INTERFACE_DEF')) or Search(r'\s+=\s*$', line_prefix)): match = None # Whitelist lambda function definition which also requires a ";" after # closing brace if match: if Match(r'^.*\[.*\]\s*(.*\)\s*)\{', line): match = None else: # Try matching cases 2-3. match = Match(r'^(.*(?:else|\)\s*const)\s*)\{', line) if not match: # Try matching cases 4-6. These are always matched on separate lines. # # Note that we can't simply concatenate the previous line to the # current line and do a single match, otherwise we may output # duplicate warnings for the blank line case: # if (cond) { # // blank line # } prevline = GetPreviousNonBlankLine(clean_lines, linenum)[0] if prevline and Search(r'[;{}]\s*$', prevline): match = Match(r'^(\s*)\{', line) # Check matching closing brace if match: (endline, endlinenum, endpos) = CloseExpression( clean_lines, linenum, len(match.group(1))) if endpos > -1 and Match(r'^\s*;', endline[endpos:]): # Current {} pair is eligible for semicolon check, and we have found # the redundant semicolon, output warning here. # # Note: because we are scanning forward for opening braces, and # outputting warnings for the matching closing brace, if there are # nested blocks with trailing semicolons, we will get the error # messages in reversed order. error(filename, endlinenum, 'readability/braces', 4, "You don't need a ; after a }") def CheckEmptyBlockBody(filename, clean_lines, linenum, error): """Look for empty loop/conditional body with only a single semicolon. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ # Search for loop keywords at the beginning of the line. Because only # whitespaces are allowed before the keywords, this will also ignore most # do-while-loops, since those lines should start with closing brace. # # We also check "if" blocks here, since an empty conditional block # is likely an error. line = clean_lines.elided[linenum] matched = Match(r'\s*(for|while|if)\s*\(', line) if matched: # Find the end of the conditional expression (end_line, end_linenum, end_pos) = CloseExpression( clean_lines, linenum, line.find('(')) # Output warning if what follows the condition expression is a semicolon. # No warning for all other cases, including whitespace or newline, since we # have a separate check for semicolons preceded by whitespace. if end_pos >= 0 and Match(r';', end_line[end_pos:]): if matched.group(1) == 'if': error(filename, end_linenum, 'whitespace/empty_conditional_body', 5, 'Empty conditional bodies should use {}') else: error(filename, end_linenum, 'whitespace/empty_loop_body', 5, 'Empty loop bodies should use {} or continue') def CheckCheck(filename, clean_lines, linenum, error): """Checks the use of CHECK and EXPECT macros. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ # Decide the set of replacement macros that should be suggested lines = clean_lines.elided check_macro = None start_pos = -1 for macro in _CHECK_MACROS: i = lines[linenum].find(macro) if i >= 0: check_macro = macro # Find opening parenthesis. Do a regular expression match here # to make sure that we are matching the expected CHECK macro, as # opposed to some other macro that happens to contain the CHECK # substring. matched = Match(r'^(.*\b' + check_macro + r'\s*)\(', lines[linenum]) if not matched: continue start_pos = len(matched.group(1)) break if not check_macro or start_pos < 0: # Don't waste time here if line doesn't contain 'CHECK' or 'EXPECT' return # Find end of the boolean expression by matching parentheses (last_line, end_line, end_pos) = CloseExpression( clean_lines, linenum, start_pos) if end_pos < 0: return if linenum == end_line: expression = lines[linenum][start_pos + 1:end_pos - 1] else: expression = lines[linenum][start_pos + 1:] for i in xrange(linenum + 1, end_line): expression += lines[i] expression += last_line[0:end_pos - 1] # Parse expression so that we can take parentheses into account. # This avoids false positives for inputs like "CHECK((a < 4) == b)", # which is not replaceable by CHECK_LE. lhs = '' rhs = '' operator = None while expression: matched = Match(r'^\s*(<<|<<=|>>|>>=|->\*|->|&&|\|\||' r'==|!=|>=|>|<=|<|\()(.*)$', expression) if matched: token = matched.group(1) if token == '(': # Parenthesized operand expression = matched.group(2) (end, _) = FindEndOfExpressionInLine(expression, 0, 1, '(', ')') if end < 0: return # Unmatched parenthesis lhs += '(' + expression[0:end] expression = expression[end:] elif token in ('&&', '||'): # Logical and/or operators. This means the expression # contains more than one term, for example: # CHECK(42 < a && a < b); # # These are not replaceable with CHECK_LE, so bail out early. return elif token in ('<<', '<<=', '>>', '>>=', '->*', '->'): # Non-relational operator lhs += token expression = matched.group(2) else: # Relational operator operator = token rhs = matched.group(2) break else: # Unparenthesized operand. Instead of appending to lhs one character # at a time, we do another regular expression match to consume several # characters at once if possible. Trivial benchmark shows that this # is more efficient when the operands are longer than a single # character, which is generally the case. matched = Match(r'^([^-=!<>()&|]+)(.*)$', expression) if not matched: matched = Match(r'^(\s*\S)(.*)$', expression) if not matched: break lhs += matched.group(1) expression = matched.group(2) # Only apply checks if we got all parts of the boolean expression if not (lhs and operator and rhs): return # Check that rhs do not contain logical operators. We already know # that lhs is fine since the loop above parses out && and ||. if rhs.find('&&') > -1 or rhs.find('||') > -1: return # At least one of the operands must be a constant literal. This is # to avoid suggesting replacements for unprintable things like # CHECK(variable != iterator) # # The following pattern matches decimal, hex integers, strings, and # characters (in that order). lhs = lhs.strip() rhs = rhs.strip() match_constant = r'^([-+]?(\d+|0[xX][0-9a-fA-F]+)[lLuU]{0,3}|".*"|\'.*\')$' if Match(match_constant, lhs) or Match(match_constant, rhs): # Note: since we know both lhs and rhs, we can provide a more # descriptive error message like: # Consider using CHECK_EQ(x, 42) instead of CHECK(x == 42) # Instead of: # Consider using CHECK_EQ instead of CHECK(a == b) # # We are still keeping the less descriptive message because if lhs # or rhs gets long, the error message might become unreadable. error(filename, linenum, 'readability/check', 2, 'Consider using %s instead of %s(a %s b)' % ( _CHECK_REPLACEMENT[check_macro][operator], check_macro, operator)) def CheckAltTokens(filename, clean_lines, linenum, error): """Check alternative keywords being used in boolean expressions. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] # Avoid preprocessor lines if Match(r'^\s*#', line): return # Last ditch effort to avoid multi-line comments. This will not help # if the comment started before the current line or ended after the # current line, but it catches most of the false positives. At least, # it provides a way to workaround this warning for people who use # multi-line comments in preprocessor macros. # # TODO(unknown): remove this once cpplint has better support for # multi-line comments. if line.find('/*') >= 0 or line.find('*/') >= 0: return for match in _ALT_TOKEN_REPLACEMENT_PATTERN.finditer(line): error(filename, linenum, 'readability/alt_tokens', 2, 'Use operator %s instead of %s' % ( _ALT_TOKEN_REPLACEMENT[match.group(1)], match.group(1))) def GetLineWidth(line): """Determines the width of the line in column positions. Args: line: A string, which may be a Unicode string. Returns: The width of the line in column positions, accounting for Unicode combining characters and wide characters. """ if isinstance(line, unicode): width = 0 for uc in unicodedata.normalize('NFC', line): if unicodedata.east_asian_width(uc) in ('W', 'F'): width += 2 elif not unicodedata.combining(uc): width += 1 return width else: return len(line) def CheckStyle(filename, clean_lines, linenum, file_extension, nesting_state, error): """Checks rules from the 'C++ style rules' section of cppguide.html. Most of these rules are hard to test (naming, comment style), but we do what we can. In particular we check for 2-space indents, line lengths, tab usage, spaces inside code, etc. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. file_extension: The extension (without the dot) of the filename. nesting_state: A _NestingState instance which maintains information about the current stack of nested blocks being parsed. error: The function to call with any errors found. """ # Don't use "elided" lines here, otherwise we can't check commented lines. # Don't want to use "raw" either, because we don't want to check inside C++11 # raw strings, raw_lines = clean_lines.lines_without_raw_strings line = raw_lines[linenum] if line.find('\t') != -1: error(filename, linenum, 'whitespace/tab', 1, 'Tab found; better to use spaces') # One or three blank spaces at the beginning of the line is weird; it's # hard to reconcile that with 2-space indents. # NOTE: here are the conditions rob pike used for his tests. Mine aren't # as sophisticated, but it may be worth becoming so: RLENGTH==initial_spaces # if(RLENGTH > 20) complain = 0; # if(match($0, " +(error|private|public|protected):")) complain = 0; # if(match(prev, "&& *$")) complain = 0; # if(match(prev, "\\|\\| *$")) complain = 0; # if(match(prev, "[\",=><] *$")) complain = 0; # if(match($0, " <<")) complain = 0; # if(match(prev, " +for \\(")) complain = 0; # if(prevodd && match(prevprev, " +for \\(")) complain = 0; initial_spaces = 0 cleansed_line = clean_lines.elided[linenum] while initial_spaces < len(line) and line[initial_spaces] == ' ': initial_spaces += 1 if line and line[-1].isspace(): error(filename, linenum, 'whitespace/end_of_line', 4, 'Line ends in whitespace. Consider deleting these extra spaces.') # There are certain situations we allow one space, notably for section labels elif ((initial_spaces == 1 or initial_spaces == 3) and not Match(r'\s*\w+\s*:\s*$', cleansed_line)): error(filename, linenum, 'whitespace/indent', 3, 'Weird number of spaces at line-start. ' 'Are you using a 2-space indent?') # Check if the line is a header guard. is_header_guard = False if file_extension == 'h': cppvar = GetHeaderGuardCPPVariable(filename) if (line.startswith('#ifndef %s' % cppvar) or line.startswith('#define %s' % cppvar) or line.startswith('#endif // %s' % cppvar)): is_header_guard = True # #include lines and header guards can be long, since there's no clean way to # split them. # # URLs can be long too. It's possible to split these, but it makes them # harder to cut&paste. # # The "$Id:...$" comment may also get very long without it being the # developers fault. if (not line.startswith('#include') and not is_header_guard and not Match(r'^\s*//.*http(s?)://\S*$', line) and not Match(r'^// \$Id:.*#[0-9]+ \$$', line)): line_width = GetLineWidth(line) extended_length = int((_line_length * 1.25)) if line_width > extended_length: error(filename, linenum, 'whitespace/line_length', 4, 'Lines should very rarely be longer than %i characters' % extended_length) elif line_width > _line_length: error(filename, linenum, 'whitespace/line_length', 2, 'Lines should be <= %i characters long' % _line_length) if (cleansed_line.count(';') > 1 and # for loops are allowed two ;'s (and may run over two lines). cleansed_line.find('for') == -1 and (GetPreviousNonBlankLine(clean_lines, linenum)[0].find('for') == -1 or GetPreviousNonBlankLine(clean_lines, linenum)[0].find(';') != -1) and # It's ok to have many commands in a switch case that fits in 1 line not ((cleansed_line.find('case ') != -1 or cleansed_line.find('default:') != -1) and cleansed_line.find('break;') != -1)): error(filename, linenum, 'whitespace/newline', 0, 'More than one command on the same line') # Some more style checks CheckBraces(filename, clean_lines, linenum, error) CheckEmptyBlockBody(filename, clean_lines, linenum, error) CheckAccess(filename, clean_lines, linenum, nesting_state, error) CheckSpacing(filename, clean_lines, linenum, nesting_state, error) CheckCheck(filename, clean_lines, linenum, error) CheckAltTokens(filename, clean_lines, linenum, error) classinfo = nesting_state.InnermostClass() if classinfo: CheckSectionSpacing(filename, clean_lines, classinfo, linenum, error) _RE_PATTERN_INCLUDE_NEW_STYLE = re.compile(r'#include +"[^/]+\.h"') _RE_PATTERN_INCLUDE = re.compile(r'^\s*#\s*include\s*([<"])([^>"]*)[>"].*$') # Matches the first component of a filename delimited by -s and _s. That is: # _RE_FIRST_COMPONENT.match('foo').group(0) == 'foo' # _RE_FIRST_COMPONENT.match('foo.cc').group(0) == 'foo' # _RE_FIRST_COMPONENT.match('foo-bar_baz.cc').group(0) == 'foo' # _RE_FIRST_COMPONENT.match('foo_bar-baz.cc').group(0) == 'foo' _RE_FIRST_COMPONENT = re.compile(r'^[^-_.]+') def _DropCommonSuffixes(filename): """Drops common suffixes like _test.cc or -inl.h from filename. For example: >>> _DropCommonSuffixes('foo/foo-inl.h') 'foo/foo' >>> _DropCommonSuffixes('foo/bar/foo.cc') 'foo/bar/foo' >>> _DropCommonSuffixes('foo/foo_internal.h') 'foo/foo' >>> _DropCommonSuffixes('foo/foo_unusualinternal.h') 'foo/foo_unusualinternal' Args: filename: The input filename. Returns: The filename with the common suffix removed. """ for suffix in ('test.cc', 'regtest.cc', 'unittest.cc', 'inl.h', 'impl.h', 'internal.h'): if (filename.endswith(suffix) and len(filename) > len(suffix) and filename[-len(suffix) - 1] in ('-', '_')): return filename[:-len(suffix) - 1] return os.path.splitext(filename)[0] def _IsTestFilename(filename): """Determines if the given filename has a suffix that identifies it as a test. Args: filename: The input filename. Returns: True if 'filename' looks like a test, False otherwise. """ if (filename.endswith('_test.cc') or filename.endswith('_unittest.cc') or filename.endswith('_regtest.cc')): return True else: return False def _ClassifyInclude(fileinfo, include, is_system): """Figures out what kind of header 'include' is. Args: fileinfo: The current file cpplint is running over. A FileInfo instance. include: The path to a #included file. is_system: True if the #include used <> rather than "". Returns: One of the _XXX_HEADER constants. For example: >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'stdio.h', True) _C_SYS_HEADER >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'string', True) _CPP_SYS_HEADER >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'foo/foo.h', False) _LIKELY_MY_HEADER >>> _ClassifyInclude(FileInfo('foo/foo_unknown_extension.cc'), ... 'bar/foo_other_ext.h', False) _POSSIBLE_MY_HEADER >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'foo/bar.h', False) _OTHER_HEADER """ # This is a list of all standard c++ header files, except # those already checked for above. is_cpp_h = include in _CPP_HEADERS if is_system: if is_cpp_h: return _CPP_SYS_HEADER else: return _C_SYS_HEADER # If the target file and the include we're checking share a # basename when we drop common extensions, and the include # lives in . , then it's likely to be owned by the target file. target_dir, target_base = ( os.path.split(_DropCommonSuffixes(fileinfo.RepositoryName()))) include_dir, include_base = os.path.split(_DropCommonSuffixes(include)) if target_base == include_base and ( include_dir == target_dir or include_dir == os.path.normpath(target_dir + '/../public')): return _LIKELY_MY_HEADER # If the target and include share some initial basename # component, it's possible the target is implementing the # include, so it's allowed to be first, but we'll never # complain if it's not there. target_first_component = _RE_FIRST_COMPONENT.match(target_base) include_first_component = _RE_FIRST_COMPONENT.match(include_base) if (target_first_component and include_first_component and target_first_component.group(0) == include_first_component.group(0)): return _POSSIBLE_MY_HEADER return _OTHER_HEADER def CheckIncludeLine(filename, clean_lines, linenum, include_state, error): """Check rules that are applicable to #include lines. Strings on #include lines are NOT removed from elided line, to make certain tasks easier. However, to prevent false positives, checks applicable to #include lines in CheckLanguage must be put here. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. include_state: An _IncludeState instance in which the headers are inserted. error: The function to call with any errors found. """ fileinfo = FileInfo(filename) line = clean_lines.lines[linenum] # "include" should use the new style "foo/bar.h" instead of just "bar.h" if _RE_PATTERN_INCLUDE_NEW_STYLE.search(line): error(filename, linenum, 'build/include', 4, 'Include the directory when naming .h files') # we shouldn't include a file more than once. actually, there are a # handful of instances where doing so is okay, but in general it's # not. match = _RE_PATTERN_INCLUDE.search(line) if match: include = match.group(2) is_system = (match.group(1) == '<') if include in include_state: error(filename, linenum, 'build/include', 4, '"%s" already included at %s:%s' % (include, filename, include_state[include])) else: include_state[include] = linenum # We want to ensure that headers appear in the right order: # 1) for foo.cc, foo.h (preferred location) # 2) c system files # 3) cpp system files # 4) for foo.cc, foo.h (deprecated location) # 5) other google headers # # We classify each include statement as one of those 5 types # using a number of techniques. The include_state object keeps # track of the highest type seen, and complains if we see a # lower type after that. error_message = include_state.CheckNextIncludeOrder( _ClassifyInclude(fileinfo, include, is_system)) if error_message: error(filename, linenum, 'build/include_order', 4, '%s. Should be: %s.h, c system, c++ system, other.' % (error_message, fileinfo.BaseName())) canonical_include = include_state.CanonicalizeAlphabeticalOrder(include) if not include_state.IsInAlphabeticalOrder( clean_lines, linenum, canonical_include): error(filename, linenum, 'build/include_alpha', 4, 'Include "%s" not in alphabetical order' % include) include_state.SetLastHeader(canonical_include) # Look for any of the stream classes that are part of standard C++. match = _RE_PATTERN_INCLUDE.match(line) if match: include = match.group(2) if Match(r'(f|ind|io|i|o|parse|pf|stdio|str|)?stream$', include): # Many unit tests use cout, so we exempt them. if not _IsTestFilename(filename): error(filename, linenum, 'readability/streams', 3, 'Streams are highly discouraged.') def _GetTextInside(text, start_pattern): r"""Retrieves all the text between matching open and close parentheses. Given a string of lines and a regular expression string, retrieve all the text following the expression and between opening punctuation symbols like (, [, or {, and the matching close-punctuation symbol. This properly nested occurrences of the punctuations, so for the text like printf(a(), b(c())); a call to _GetTextInside(text, r'printf\(') will return 'a(), b(c())'. start_pattern must match string having an open punctuation symbol at the end. Args: text: The lines to extract text. Its comments and strings must be elided. It can be single line and can span multiple lines. start_pattern: The regexp string indicating where to start extracting the text. Returns: The extracted text. None if either the opening string or ending punctuation could not be found. """ # TODO(sugawarayu): Audit cpplint.py to see what places could be profitably # rewritten to use _GetTextInside (and use inferior regexp matching today). # Give opening punctuations to get the matching close-punctuations. matching_punctuation = {'(': ')', '{': '}', '[': ']'} closing_punctuation = set(matching_punctuation.itervalues()) # Find the position to start extracting text. match = re.search(start_pattern, text, re.M) if not match: # start_pattern not found in text. return None start_position = match.end(0) assert start_position > 0, ( 'start_pattern must ends with an opening punctuation.') assert text[start_position - 1] in matching_punctuation, ( 'start_pattern must ends with an opening punctuation.') # Stack of closing punctuations we expect to have in text after position. punctuation_stack = [matching_punctuation[text[start_position - 1]]] position = start_position while punctuation_stack and position < len(text): if text[position] == punctuation_stack[-1]: punctuation_stack.pop() elif text[position] in closing_punctuation: # A closing punctuation without matching opening punctuations. return None elif text[position] in matching_punctuation: punctuation_stack.append(matching_punctuation[text[position]]) position += 1 if punctuation_stack: # Opening punctuations left without matching close-punctuations. return None # punctuations match. return text[start_position:position - 1] # Patterns for matching call-by-reference parameters. # # Supports nested templates up to 2 levels deep using this messy pattern: # < (?: < (?: < [^<>]* # > # | [^<>] )* # > # | [^<>] )* # > _RE_PATTERN_IDENT = r'[_a-zA-Z]\w*' # =~ [[:alpha:]][[:alnum:]]* _RE_PATTERN_TYPE = ( r'(?:const\s+)?(?:typename\s+|class\s+|struct\s+|union\s+|enum\s+)?' r'(?:\w|' r'\s*<(?:<(?:<[^<>]*>|[^<>])*>|[^<>])*>|' r'::)+') # A call-by-reference parameter ends with '& identifier'. _RE_PATTERN_REF_PARAM = re.compile( r'(' + _RE_PATTERN_TYPE + r'(?:\s*(?:\bconst\b|[*]))*\s*' r'&\s*' + _RE_PATTERN_IDENT + r')\s*(?:=[^,()]+)?[,)]') # A call-by-const-reference parameter either ends with 'const& identifier' # or looks like 'const type& identifier' when 'type' is atomic. _RE_PATTERN_CONST_REF_PARAM = ( r'(?:.*\s*\bconst\s*&\s*' + _RE_PATTERN_IDENT + r'|const\s+' + _RE_PATTERN_TYPE + r'\s*&\s*' + _RE_PATTERN_IDENT + r')') def CheckLanguage(filename, clean_lines, linenum, file_extension, include_state, nesting_state, error): """Checks rules from the 'C++ language rules' section of cppguide.html. Some of these rules are hard to test (function overloading, using uint32 inappropriately), but we do the best we can. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. file_extension: The extension (without the dot) of the filename. include_state: An _IncludeState instance in which the headers are inserted. nesting_state: A _NestingState instance which maintains information about the current stack of nested blocks being parsed. error: The function to call with any errors found. """ # If the line is empty or consists of entirely a comment, no need to # check it. line = clean_lines.elided[linenum] if not line: return match = _RE_PATTERN_INCLUDE.search(line) if match: CheckIncludeLine(filename, clean_lines, linenum, include_state, error) return # Reset include state across preprocessor directives. This is meant # to silence warnings for conditional includes. if Match(r'^\s*#\s*(?:ifdef|elif|else|endif)\b', line): include_state.ResetSection() # Make Windows paths like Unix. fullname = os.path.abspath(filename).replace('\\', '/') # TODO(unknown): figure out if they're using default arguments in fn proto. # Check to see if they're using an conversion function cast. # I just try to capture the most common basic types, though there are more. # Parameterless conversion functions, such as bool(), are allowed as they are # probably a member operator declaration or default constructor. match = Search( r'(\bnew\s+)?\b' # Grab 'new' operator, if it's there r'(int|float|double|bool|char|int32|uint32|int64|uint64)' r'(\([^)].*)', line) if match: matched_new = match.group(1) matched_type = match.group(2) matched_funcptr = match.group(3) # gMock methods are defined using some variant of MOCK_METHODx(name, type) # where type may be float(), int(string), etc. Without context they are # virtually indistinguishable from int(x) casts. Likewise, gMock's # MockCallback takes a template parameter of the form return_type(arg_type), # which looks much like the cast we're trying to detect. # # std::function<> wrapper has a similar problem. # # Return types for function pointers also look like casts if they # don't have an extra space. if (matched_new is None and # If new operator, then this isn't a cast not (Match(r'^\s*MOCK_(CONST_)?METHOD\d+(_T)?\(', line) or Search(r'\bMockCallback<.*>', line) or Search(r'\bstd::function<.*>', line)) and not (matched_funcptr and Match(r'\((?:[^() ]+::\s*\*\s*)?[^() ]+\)\s*\(', matched_funcptr))): # Try a bit harder to catch gmock lines: the only place where # something looks like an old-style cast is where we declare the # return type of the mocked method, and the only time when we # are missing context is if MOCK_METHOD was split across # multiple lines. The missing MOCK_METHOD is usually one or two # lines back, so scan back one or two lines. # # It's not possible for gmock macros to appear in the first 2 # lines, since the class head + section name takes up 2 lines. if (linenum < 2 or not (Match(r'^\s*MOCK_(?:CONST_)?METHOD\d+(?:_T)?\((?:\S+,)?\s*$', clean_lines.elided[linenum - 1]) or Match(r'^\s*MOCK_(?:CONST_)?METHOD\d+(?:_T)?\(\s*$', clean_lines.elided[linenum - 2]))): error(filename, linenum, 'readability/casting', 4, 'Using deprecated casting style. ' 'Use static_cast<%s>(...) instead' % matched_type) CheckCStyleCast(filename, linenum, line, clean_lines.raw_lines[linenum], 'static_cast', r'\((int|float|double|bool|char|u?int(16|32|64))\)', error) # This doesn't catch all cases. Consider (const char * const)"hello". # # (char *) "foo" should always be a const_cast (reinterpret_cast won't # compile). if CheckCStyleCast(filename, linenum, line, clean_lines.raw_lines[linenum], 'const_cast', r'\((char\s?\*+\s?)\)\s*"', error): pass else: # Check pointer casts for other than string constants CheckCStyleCast(filename, linenum, line, clean_lines.raw_lines[linenum], 'reinterpret_cast', r'\((\w+\s?\*+\s?)\)', error) # In addition, we look for people taking the address of a cast. This # is dangerous -- casts can assign to temporaries, so the pointer doesn't # point where you think. match = Search( r'(?:&\(([^)]+)\)[\w(])|' r'(?:&(static|dynamic|down|reinterpret)_cast\b)', line) if match and match.group(1) != '*': error(filename, linenum, 'runtime/casting', 4, ('Are you taking an address of a cast? ' 'This is dangerous: could be a temp var. ' 'Take the address before doing the cast, rather than after')) # Create an extended_line, which is the concatenation of the current and # next lines, for more effective checking of code that may span more than one # line. if linenum + 1 < clean_lines.NumLines(): extended_line = line + clean_lines.elided[linenum + 1] else: extended_line = line # Check for people declaring static/global STL strings at the top level. # This is dangerous because the C++ language does not guarantee that # globals with constructors are initialized before the first access. match = Match( r'((?:|static +)(?:|const +))string +([a-zA-Z0-9_:]+)\b(.*)', line) # Make sure it's not a function. # Function template specialization looks like: "string foo(...". # Class template definitions look like: "string Foo::Method(...". # # Also ignore things that look like operators. These are matched separately # because operator names cross non-word boundaries. If we change the pattern # above, we would decrease the accuracy of matching identifiers. if (match and not Search(r'\boperator\W', line) and not Match(r'\s*(<.*>)?(::[a-zA-Z0-9_]+)?\s*\(([^"]|$)', match.group(3))): error(filename, linenum, 'runtime/string', 4, 'For a static/global string constant, use a C style string instead: ' '"%schar %s[]".' % (match.group(1), match.group(2))) if Search(r'\b([A-Za-z0-9_]*_)\(\1\)', line): error(filename, linenum, 'runtime/init', 4, 'You seem to be initializing a member variable with itself.') if file_extension == 'h': # TODO(unknown): check that 1-arg constructors are explicit. # How to tell it's a constructor? # (handled in CheckForNonStandardConstructs for now) # TODO(unknown): check that classes have DISALLOW_EVIL_CONSTRUCTORS # (level 1 error) pass # Check if people are using the verboten C basic types. The only exception # we regularly allow is "unsigned short port" for port. if Search(r'\bshort port\b', line): if not Search(r'\bunsigned short port\b', line): error(filename, linenum, 'runtime/int', 4, 'Use "unsigned short" for ports, not "short"') else: match = Search(r'\b(short|long(?! +double)|long long)\b', line) if match: error(filename, linenum, 'runtime/int', 4, 'Use int16/int64/etc, rather than the C type %s' % match.group(1)) # When snprintf is used, the second argument shouldn't be a literal. match = Search(r'snprintf\s*\(([^,]*),\s*([0-9]*)\s*,', line) if match and match.group(2) != '0': # If 2nd arg is zero, snprintf is used to calculate size. error(filename, linenum, 'runtime/printf', 3, 'If you can, use sizeof(%s) instead of %s as the 2nd arg ' 'to snprintf.' % (match.group(1), match.group(2))) # Check if some verboten C functions are being used. if Search(r'\bsprintf\b', line): error(filename, linenum, 'runtime/printf', 5, 'Never use sprintf. Use snprintf instead.') match = Search(r'\b(strcpy|strcat)\b', line) if match: error(filename, linenum, 'runtime/printf', 4, 'Almost always, snprintf is better than %s' % match.group(1)) # Check if some verboten operator overloading is going on # TODO(unknown): catch out-of-line unary operator&: # class X {}; # int operator&(const X& x) { return 42; } // unary operator& # The trick is it's hard to tell apart from binary operator&: # class Y { int operator&(const Y& x) { return 23; } }; // binary operator& if Search(r'\boperator\s*&\s*\(\s*\)', line): error(filename, linenum, 'runtime/operator', 4, 'Unary operator& is dangerous. Do not use it.') # Check for suspicious usage of "if" like # } if (a == b) { if Search(r'\}\s*if\s*\(', line): error(filename, linenum, 'readability/braces', 4, 'Did you mean "else if"? If not, start a new line for "if".') # Check for potential format string bugs like printf(foo). # We constrain the pattern not to pick things like DocidForPrintf(foo). # Not perfect but it can catch printf(foo.c_str()) and printf(foo->c_str()) # TODO(sugawarayu): Catch the following case. Need to change the calling # convention of the whole function to process multiple line to handle it. # printf( # boy_this_is_a_really_long_variable_that_cannot_fit_on_the_prev_line); printf_args = _GetTextInside(line, r'(?i)\b(string)?printf\s*\(') if printf_args: match = Match(r'([\w.\->()]+)$', printf_args) if match and match.group(1) != '__VA_ARGS__': function_name = re.search(r'\b((?:string)?printf)\s*\(', line, re.I).group(1) error(filename, linenum, 'runtime/printf', 4, 'Potential format string bug. Do %s("%%s", %s) instead.' % (function_name, match.group(1))) # Check for potential memset bugs like memset(buf, sizeof(buf), 0). match = Search(r'memset\s*\(([^,]*),\s*([^,]*),\s*0\s*\)', line) if match and not Match(r"^''|-?[0-9]+|0x[0-9A-Fa-f]$", match.group(2)): error(filename, linenum, 'runtime/memset', 4, 'Did you mean "memset(%s, 0, %s)"?' % (match.group(1), match.group(2))) if Search(r'\busing namespace\b', line): error(filename, linenum, 'build/namespaces', 5, 'Do not use namespace using-directives. ' 'Use using-declarations instead.') # Detect variable-length arrays. match = Match(r'\s*(.+::)?(\w+) [a-z]\w*\[(.+)];', line) if (match and match.group(2) != 'return' and match.group(2) != 'delete' and match.group(3).find(']') == -1): # Split the size using space and arithmetic operators as delimiters. # If any of the resulting tokens are not compile time constants then # report the error. tokens = re.split(r'\s|\+|\-|\*|\/|<<|>>]', match.group(3)) is_const = True skip_next = False for tok in tokens: if skip_next: skip_next = False continue if Search(r'sizeof\(.+\)', tok): continue if Search(r'arraysize\(\w+\)', tok): continue tok = tok.lstrip('(') tok = tok.rstrip(')') if not tok: continue if Match(r'\d+', tok): continue if Match(r'0[xX][0-9a-fA-F]+', tok): continue if Match(r'k[A-Z0-9]\w*', tok): continue if Match(r'(.+::)?k[A-Z0-9]\w*', tok): continue if Match(r'(.+::)?[A-Z][A-Z0-9_]*', tok): continue # A catch all for tricky sizeof cases, including 'sizeof expression', # 'sizeof(*type)', 'sizeof(const type)', 'sizeof(struct StructName)' # requires skipping the next token because we split on ' ' and '*'. if tok.startswith('sizeof'): skip_next = True continue is_const = False break if not is_const: error(filename, linenum, 'runtime/arrays', 1, 'Do not use variable-length arrays. Use an appropriately named ' "('k' followed by CamelCase) compile-time constant for the size.") # If DISALLOW_EVIL_CONSTRUCTORS, DISALLOW_COPY_AND_ASSIGN, or # DISALLOW_IMPLICIT_CONSTRUCTORS is present, then it should be the last thing # in the class declaration. match = Match( (r'\s*' r'(DISALLOW_(EVIL_CONSTRUCTORS|COPY_AND_ASSIGN|IMPLICIT_CONSTRUCTORS))' r'\(.*\);$'), line) if match and linenum + 1 < clean_lines.NumLines(): next_line = clean_lines.elided[linenum + 1] # We allow some, but not all, declarations of variables to be present # in the statement that defines the class. The [\w\*,\s]* fragment of # the regular expression below allows users to declare instances of # the class or pointers to instances, but not less common types such # as function pointers or arrays. It's a tradeoff between allowing # reasonable code and avoiding trying to parse more C++ using regexps. if not Search(r'^\s*}[\w\*,\s]*;', next_line): error(filename, linenum, 'readability/constructors', 3, match.group(1) + ' should be the last thing in the class') # Check for use of unnamed namespaces in header files. Registration # macros are typically OK, so we allow use of "namespace {" on lines # that end with backslashes. if (file_extension == 'h' and Search(r'\bnamespace\s*{', line) and line[-1] != '\\'): error(filename, linenum, 'build/namespaces', 4, 'Do not use unnamed namespaces in header files. See ' 'http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Namespaces' ' for more information.') def CheckForNonConstReference(filename, clean_lines, linenum, nesting_state, error): """Check for non-const references. Separate from CheckLanguage since it scans backwards from current line, instead of scanning forward. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. nesting_state: A _NestingState instance which maintains information about the current stack of nested blocks being parsed. error: The function to call with any errors found. """ # Do nothing if there is no '&' on current line. line = clean_lines.elided[linenum] if '&' not in line: return # Long type names may be broken across multiple lines, usually in one # of these forms: # LongType # ::LongTypeContinued &identifier # LongType:: # LongTypeContinued &identifier # LongType< # ...>::LongTypeContinued &identifier # # If we detected a type split across two lines, join the previous # line to current line so that we can match const references # accordingly. # # Note that this only scans back one line, since scanning back # arbitrary number of lines would be expensive. If you have a type # that spans more than 2 lines, please use a typedef. if linenum > 1: previous = None if Match(r'\s*::(?:[\w<>]|::)+\s*&\s*\S', line): # previous_line\n + ::current_line previous = Search(r'\b((?:const\s*)?(?:[\w<>]|::)+[\w<>])\s*$', clean_lines.elided[linenum - 1]) elif Match(r'\s*[a-zA-Z_]([\w<>]|::)+\s*&\s*\S', line): # previous_line::\n + current_line previous = Search(r'\b((?:const\s*)?(?:[\w<>]|::)+::)\s*$', clean_lines.elided[linenum - 1]) if previous: line = previous.group(1) + line.lstrip() else: # Check for templated parameter that is split across multiple lines endpos = line.rfind('>') if endpos > -1: (_, startline, startpos) = ReverseCloseExpression( clean_lines, linenum, endpos) if startpos > -1 and startline < linenum: # Found the matching < on an earlier line, collect all # pieces up to current line. line = '' for i in xrange(startline, linenum + 1): line += clean_lines.elided[i].strip() # Check for non-const references in function parameters. A single '&' may # found in the following places: # inside expression: binary & for bitwise AND # inside expression: unary & for taking the address of something # inside declarators: reference parameter # We will exclude the first two cases by checking that we are not inside a # function body, including one that was just introduced by a trailing '{'. # TODO(unknwon): Doesn't account for preprocessor directives. # TODO(unknown): Doesn't account for 'catch(Exception& e)' [rare]. check_params = False if not nesting_state.stack: check_params = True # top level elif (isinstance(nesting_state.stack[-1], _ClassInfo) or isinstance(nesting_state.stack[-1], _NamespaceInfo)): check_params = True # within class or namespace elif Match(r'.*{\s*$', line): if (len(nesting_state.stack) == 1 or isinstance(nesting_state.stack[-2], _ClassInfo) or isinstance(nesting_state.stack[-2], _NamespaceInfo)): check_params = True # just opened global/class/namespace block # We allow non-const references in a few standard places, like functions # called "swap()" or iostream operators like "<<" or ">>". Do not check # those function parameters. # # We also accept & in static_assert, which looks like a function but # it's actually a declaration expression. whitelisted_functions = (r'(?:[sS]wap(?:<\w:+>)?|' r'operator\s*[<>][<>]|' r'static_assert|COMPILE_ASSERT' r')\s*\(') if Search(whitelisted_functions, line): check_params = False elif not Search(r'\S+\([^)]*$', line): # Don't see a whitelisted function on this line. Actually we # didn't see any function name on this line, so this is likely a # multi-line parameter list. Try a bit harder to catch this case. for i in xrange(2): if (linenum > i and Search(whitelisted_functions, clean_lines.elided[linenum - i - 1])): check_params = False break if check_params: decls = ReplaceAll(r'{[^}]*}', ' ', line) # exclude function body for parameter in re.findall(_RE_PATTERN_REF_PARAM, decls): if not Match(_RE_PATTERN_CONST_REF_PARAM, parameter): error(filename, linenum, 'runtime/references', 2, 'Is this a non-const reference? ' 'If so, make const or use a pointer: ' + ReplaceAll(' *<', '<', parameter)) def CheckCStyleCast(filename, linenum, line, raw_line, cast_type, pattern, error): """Checks for a C-style cast by looking for the pattern. Args: filename: The name of the current file. linenum: The number of the line to check. line: The line of code to check. raw_line: The raw line of code to check, with comments. cast_type: The string for the C++ cast to recommend. This is either reinterpret_cast, static_cast, or const_cast, depending. pattern: The regular expression used to find C-style casts. error: The function to call with any errors found. Returns: True if an error was emitted. False otherwise. """ match = Search(pattern, line) if not match: return False # Exclude lines with sizeof, since sizeof looks like a cast. sizeof_match = Match(r'.*sizeof\s*$', line[0:match.start(1) - 1]) if sizeof_match: return False # operator++(int) and operator--(int) if (line[0:match.start(1) - 1].endswith(' operator++') or line[0:match.start(1) - 1].endswith(' operator--')): return False # A single unnamed argument for a function tends to look like old # style cast. If we see those, don't issue warnings for deprecated # casts, instead issue warnings for unnamed arguments where # appropriate. # # These are things that we want warnings for, since the style guide # explicitly require all parameters to be named: # Function(int); # Function(int) { # ConstMember(int) const; # ConstMember(int) const { # ExceptionMember(int) throw (...); # ExceptionMember(int) throw (...) { # PureVirtual(int) = 0; # # These are functions of some sort, where the compiler would be fine # if they had named parameters, but people often omit those # identifiers to reduce clutter: # (FunctionPointer)(int); # (FunctionPointer)(int) = value; # Function((function_pointer_arg)(int)) # ; # <(FunctionPointerTemplateArgument)(int)>; remainder = line[match.end(0):] if Match(r'^\s*(?:;|const\b|throw\b|=|>|\{|\))', remainder): # Looks like an unnamed parameter. # Don't warn on any kind of template arguments. if Match(r'^\s*>', remainder): return False # Don't warn on assignments to function pointers, but keep warnings for # unnamed parameters to pure virtual functions. Note that this pattern # will also pass on assignments of "0" to function pointers, but the # preferred values for those would be "nullptr" or "NULL". matched_zero = Match(r'^\s=\s*(\S+)\s*;', remainder) if matched_zero and matched_zero.group(1) != '0': return False # Don't warn on function pointer declarations. For this we need # to check what came before the "(type)" string. if Match(r'.*\)\s*$', line[0:match.start(0)]): return False # Don't warn if the parameter is named with block comments, e.g.: # Function(int /*unused_param*/); if '/*' in raw_line: return False # Passed all filters, issue warning here. error(filename, linenum, 'readability/function', 3, 'All parameters should be named in a function') return True # At this point, all that should be left is actual casts. error(filename, linenum, 'readability/casting', 4, 'Using C-style cast. Use %s<%s>(...) instead' % (cast_type, match.group(1))) return True _HEADERS_CONTAINING_TEMPLATES = ( ('', ('deque',)), ('', ('unary_function', 'binary_function', 'plus', 'minus', 'multiplies', 'divides', 'modulus', 'negate', 'equal_to', 'not_equal_to', 'greater', 'less', 'greater_equal', 'less_equal', 'logical_and', 'logical_or', 'logical_not', 'unary_negate', 'not1', 'binary_negate', 'not2', 'bind1st', 'bind2nd', 'pointer_to_unary_function', 'pointer_to_binary_function', 'ptr_fun', 'mem_fun_t', 'mem_fun', 'mem_fun1_t', 'mem_fun1_ref_t', 'mem_fun_ref_t', 'const_mem_fun_t', 'const_mem_fun1_t', 'const_mem_fun_ref_t', 'const_mem_fun1_ref_t', 'mem_fun_ref', )), ('', ('numeric_limits',)), ('', ('list',)), ('', ('map', 'multimap',)), ('', ('allocator',)), ('', ('queue', 'priority_queue',)), ('', ('set', 'multiset',)), ('', ('stack',)), ('', ('char_traits', 'basic_string',)), ('', ('pair',)), ('', ('vector',)), # gcc extensions. # Note: std::hash is their hash, ::hash is our hash ('', ('hash_map', 'hash_multimap',)), ('', ('hash_set', 'hash_multiset',)), ('', ('slist',)), ) _RE_PATTERN_STRING = re.compile(r'\bstring\b') _re_pattern_algorithm_header = [] for _template in ('copy', 'max', 'min', 'min_element', 'sort', 'swap', 'transform'): # Match max(..., ...), max(..., ...), but not foo->max, foo.max or # type::max(). _re_pattern_algorithm_header.append( (re.compile(r'[^>.]\b' + _template + r'(<.*?>)?\([^\)]'), _template, '')) _re_pattern_templates = [] for _header, _templates in _HEADERS_CONTAINING_TEMPLATES: for _template in _templates: _re_pattern_templates.append( (re.compile(r'(\<|\b)' + _template + r'\s*\<'), _template + '<>', _header)) def FilesBelongToSameModule(filename_cc, filename_h): """Check if these two filenames belong to the same module. The concept of a 'module' here is a as follows: foo.h, foo-inl.h, foo.cc, foo_test.cc and foo_unittest.cc belong to the same 'module' if they are in the same directory. some/path/public/xyzzy and some/path/internal/xyzzy are also considered to belong to the same module here. If the filename_cc contains a longer path than the filename_h, for example, '/absolute/path/to/base/sysinfo.cc', and this file would include 'base/sysinfo.h', this function also produces the prefix needed to open the header. This is used by the caller of this function to more robustly open the header file. We don't have access to the real include paths in this context, so we need this guesswork here. Known bugs: tools/base/bar.cc and base/bar.h belong to the same module according to this implementation. Because of this, this function gives some false positives. This should be sufficiently rare in practice. Args: filename_cc: is the path for the .cc file filename_h: is the path for the header path Returns: Tuple with a bool and a string: bool: True if filename_cc and filename_h belong to the same module. string: the additional prefix needed to open the header file. """ if not filename_cc.endswith('.cc'): return (False, '') filename_cc = filename_cc[:-len('.cc')] if filename_cc.endswith('_unittest'): filename_cc = filename_cc[:-len('_unittest')] elif filename_cc.endswith('_test'): filename_cc = filename_cc[:-len('_test')] filename_cc = filename_cc.replace('/public/', '/') filename_cc = filename_cc.replace('/internal/', '/') if not filename_h.endswith('.h'): return (False, '') filename_h = filename_h[:-len('.h')] if filename_h.endswith('-inl'): filename_h = filename_h[:-len('-inl')] filename_h = filename_h.replace('/public/', '/') filename_h = filename_h.replace('/internal/', '/') files_belong_to_same_module = filename_cc.endswith(filename_h) common_path = '' if files_belong_to_same_module: common_path = filename_cc[:-len(filename_h)] return files_belong_to_same_module, common_path def UpdateIncludeState(filename, include_state, io=codecs): """Fill up the include_state with new includes found from the file. Args: filename: the name of the header to read. include_state: an _IncludeState instance in which the headers are inserted. io: The io factory to use to read the file. Provided for testability. Returns: True if a header was successfully added. False otherwise. """ headerfile = None try: headerfile = io.open(filename, 'r', 'utf8', 'replace') except IOError: return False linenum = 0 for line in headerfile: linenum += 1 clean_line = CleanseComments(line) match = _RE_PATTERN_INCLUDE.search(clean_line) if match: include = match.group(2) # The value formatting is cute, but not really used right now. # What matters here is that the key is in include_state. include_state.setdefault(include, '%s:%d' % (filename, linenum)) return True def CheckForIncludeWhatYouUse(filename, clean_lines, include_state, error, io=codecs): """Reports for missing stl includes. This function will output warnings to make sure you are including the headers necessary for the stl containers and functions that you use. We only give one reason to include a header. For example, if you use both equal_to<> and less<> in a .h file, only one (the latter in the file) of these will be reported as a reason to include the . Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. include_state: An _IncludeState instance. error: The function to call with any errors found. io: The IO factory to use to read the header file. Provided for unittest injection. """ required = {} # A map of header name to linenumber and the template entity. # Example of required: { '': (1219, 'less<>') } for linenum in xrange(clean_lines.NumLines()): line = clean_lines.elided[linenum] if not line or line[0] == '#': continue # String is special -- it is a non-templatized type in STL. matched = _RE_PATTERN_STRING.search(line) if matched: # Don't warn about strings in non-STL namespaces: # (We check only the first match per line; good enough.) prefix = line[:matched.start()] if prefix.endswith('std::') or not prefix.endswith('::'): required[''] = (linenum, 'string') for pattern, template, header in _re_pattern_algorithm_header: if pattern.search(line): required[header] = (linenum, template) # The following function is just a speed up, no semantics are changed. if not '<' in line: # Reduces the cpu time usage by skipping lines. continue for pattern, template, header in _re_pattern_templates: if pattern.search(line): required[header] = (linenum, template) # The policy is that if you #include something in foo.h you don't need to # include it again in foo.cc. Here, we will look at possible includes. # Let's copy the include_state so it is only messed up within this function. include_state = include_state.copy() # Did we find the header for this file (if any) and successfully load it? header_found = False # Use the absolute path so that matching works properly. abs_filename = FileInfo(filename).FullName() # For Emacs's flymake. # If cpplint is invoked from Emacs's flymake, a temporary file is generated # by flymake and that file name might end with '_flymake.cc'. In that case, # restore original file name here so that the corresponding header file can be # found. # e.g. If the file name is 'foo_flymake.cc', we should search for 'foo.h' # instead of 'foo_flymake.h' abs_filename = re.sub(r'_flymake\.cc$', '.cc', abs_filename) # include_state is modified during iteration, so we iterate over a copy of # the keys. header_keys = include_state.keys() for header in header_keys: (same_module, common_path) = FilesBelongToSameModule(abs_filename, header) fullpath = common_path + header if same_module and UpdateIncludeState(fullpath, include_state, io): header_found = True # If we can't find the header file for a .cc, assume it's because we don't # know where to look. In that case we'll give up as we're not sure they # didn't include it in the .h file. # TODO(unknown): Do a better job of finding .h files so we are confident that # not having the .h file means there isn't one. if filename.endswith('.cc') and not header_found: return # All the lines have been processed, report the errors found. for required_header_unstripped in required: template = required[required_header_unstripped][1] if required_header_unstripped.strip('<>"') not in include_state: error(filename, required[required_header_unstripped][0], 'build/include_what_you_use', 4, 'Add #include ' + required_header_unstripped + ' for ' + template) _RE_PATTERN_EXPLICIT_MAKEPAIR = re.compile(r'\bmake_pair\s*<') def CheckMakePairUsesDeduction(filename, clean_lines, linenum, error): """Check that make_pair's template arguments are deduced. G++ 4.6 in C++0x mode fails badly if make_pair's template arguments are specified explicitly, and such use isn't intended in any case. Args: filename: The name of the current file. clean_lines: A CleansedLines instance containing the file. linenum: The number of the line to check. error: The function to call with any errors found. """ line = clean_lines.elided[linenum] match = _RE_PATTERN_EXPLICIT_MAKEPAIR.search(line) if match: error(filename, linenum, 'build/explicit_make_pair', 4, # 4 = high confidence 'For C++11-compatibility, omit template arguments from make_pair' ' OR use pair directly OR if appropriate, construct a pair directly') def ProcessLine(filename, file_extension, clean_lines, line, include_state, function_state, nesting_state, error, extra_check_functions=[]): """Processes a single line in the file. Args: filename: Filename of the file that is being processed. file_extension: The extension (dot not included) of the file. clean_lines: An array of strings, each representing a line of the file, with comments stripped. line: Number of line being processed. include_state: An _IncludeState instance in which the headers are inserted. function_state: A _FunctionState instance which counts function lines, etc. nesting_state: A _NestingState instance which maintains information about the current stack of nested blocks being parsed. error: A callable to which errors are reported, which takes 4 arguments: filename, line number, error level, and message extra_check_functions: An array of additional check functions that will be run on each source line. Each function takes 4 arguments: filename, clean_lines, line, error """ raw_lines = clean_lines.raw_lines ParseNolintSuppressions(filename, raw_lines[line], line, error) nesting_state.Update(filename, clean_lines, line, error) if nesting_state.stack and nesting_state.stack[-1].inline_asm != _NO_ASM: return CheckForFunctionLengths(filename, clean_lines, line, function_state, error) CheckForMultilineCommentsAndStrings(filename, clean_lines, line, error) CheckStyle(filename, clean_lines, line, file_extension, nesting_state, error) CheckLanguage(filename, clean_lines, line, file_extension, include_state, nesting_state, error) CheckForNonConstReference(filename, clean_lines, line, nesting_state, error) CheckForNonStandardConstructs(filename, clean_lines, line, nesting_state, error) CheckVlogArguments(filename, clean_lines, line, error) CheckPosixThreading(filename, clean_lines, line, error) CheckInvalidIncrement(filename, clean_lines, line, error) CheckMakePairUsesDeduction(filename, clean_lines, line, error) for check_fn in extra_check_functions: check_fn(filename, clean_lines, line, error) def ProcessFileData(filename, file_extension, lines, error, extra_check_functions=[]): """Performs lint checks and reports any errors to the given error function. Args: filename: Filename of the file that is being processed. file_extension: The extension (dot not included) of the file. lines: An array of strings, each representing a line of the file, with the last element being empty if the file is terminated with a newline. error: A callable to which errors are reported, which takes 4 arguments: filename, line number, error level, and message extra_check_functions: An array of additional check functions that will be run on each source line. Each function takes 4 arguments: filename, clean_lines, line, error """ lines = (['// marker so line numbers and indices both start at 1'] + lines + ['// marker so line numbers end in a known way']) include_state = _IncludeState() function_state = _FunctionState() nesting_state = _NestingState() ResetNolintSuppressions() CheckForCopyright(filename, lines, error) if file_extension == 'h': CheckForHeaderGuard(filename, lines, error) RemoveMultiLineComments(filename, lines, error) clean_lines = CleansedLines(lines) for line in xrange(clean_lines.NumLines()): ProcessLine(filename, file_extension, clean_lines, line, include_state, function_state, nesting_state, error, extra_check_functions) nesting_state.CheckCompletedBlocks(filename, error) CheckForIncludeWhatYouUse(filename, clean_lines, include_state, error) # We check here rather than inside ProcessLine so that we see raw # lines rather than "cleaned" lines. CheckForBadCharacters(filename, lines, error) CheckForNewlineAtEOF(filename, lines, error) def ProcessFile(filename, vlevel, extra_check_functions=[]): """Does google-lint on a single file. Args: filename: The name of the file to parse. vlevel: The level of errors to report. Every error of confidence >= verbose_level will be reported. 0 is a good default. extra_check_functions: An array of additional check functions that will be run on each source line. Each function takes 4 arguments: filename, clean_lines, line, error """ _SetVerboseLevel(vlevel) try: # Support the UNIX convention of using "-" for stdin. Note that # we are not opening the file with universal newline support # (which codecs doesn't support anyway), so the resulting lines do # contain trailing '\r' characters if we are reading a file that # has CRLF endings. # If after the split a trailing '\r' is present, it is removed # below. If it is not expected to be present (i.e. os.linesep != # '\r\n' as in Windows), a warning is issued below if this file # is processed. if filename == '-': lines = codecs.StreamReaderWriter(sys.stdin, codecs.getreader('utf8'), codecs.getwriter('utf8'), 'replace').read().split('\n') else: lines = codecs.open(filename, 'r', 'utf8', 'replace').read().split('\n') carriage_return_found = False # Remove trailing '\r'. for linenum in range(len(lines)): if lines[linenum].endswith('\r'): lines[linenum] = lines[linenum].rstrip('\r') carriage_return_found = True except IOError: sys.stderr.write( "Skipping input '%s': Can't open for reading\n" % filename) return # Note, if no dot is found, this will give the entire filename as the ext. file_extension = filename[filename.rfind('.') + 1:] # When reading from stdin, the extension is unknown, so no cpplint tests # should rely on the extension. if filename != '-' and file_extension not in _valid_extensions: sys.stderr.write('Ignoring %s; not a valid file name ' '(%s)\n' % (filename, ', '.join(_valid_extensions))) else: ProcessFileData(filename, file_extension, lines, Error, extra_check_functions) if carriage_return_found and os.linesep != '\r\n': # Use 0 for linenum since outputting only one error for potentially # several lines. Error(filename, 0, 'whitespace/newline', 1, 'One or more unexpected \\r (^M) found;' 'better to use only a \\n') sys.stderr.write('Done processing %s\n' % filename) def PrintUsage(message): """Prints a brief usage string and exits, optionally with an error message. Args: message: The optional error message. """ sys.stderr.write(_USAGE) if message: sys.exit('\nFATAL ERROR: ' + message) else: sys.exit(1) def PrintCategories(): """Prints a list of all the error-categories used by error messages. These are the categories used to filter messages via --filter. """ sys.stderr.write(''.join(' %s\n' % cat for cat in _ERROR_CATEGORIES)) sys.exit(0) def ParseArguments(args): """Parses the command line arguments. This may set the output format and verbosity level as side-effects. Args: args: The command line arguments: Returns: The list of filenames to lint. """ try: (opts, filenames) = getopt.getopt(args, '', ['help', 'output=', 'verbose=', 'counting=', 'filter=', 'root=', 'linelength=', 'extensions=']) except getopt.GetoptError: PrintUsage('Invalid arguments.') verbosity = _VerboseLevel() output_format = _OutputFormat() filters = '' counting_style = '' for (opt, val) in opts: if opt == '--help': PrintUsage(None) elif opt == '--output': if val not in ('emacs', 'vs7', 'eclipse'): PrintUsage('The only allowed output formats are emacs, vs7 and eclipse.') output_format = val elif opt == '--verbose': verbosity = int(val) elif opt == '--filter': filters = val if not filters: PrintCategories() elif opt == '--counting': if val not in ('total', 'toplevel', 'detailed'): PrintUsage('Valid counting options are total, toplevel, and detailed') counting_style = val elif opt == '--root': global _root _root = val elif opt == '--linelength': global _line_length try: _line_length = int(val) except ValueError: PrintUsage('Line length must be digits.') elif opt == '--extensions': global _valid_extensions try: _valid_extensions = set(val.split(',')) except ValueError: PrintUsage('Extensions must be comma separated list.') if not filenames: PrintUsage('No files were specified.') _SetOutputFormat(output_format) _SetVerboseLevel(verbosity) _SetFilters(filters) _SetCountingStyle(counting_style) return filenames def main(): filenames = ParseArguments(sys.argv[1:]) # Change stderr to write with replacement characters so we don't die # if we try to print something containing non-ASCII characters. sys.stderr = codecs.StreamReaderWriter(sys.stderr, codecs.getreader('utf8'), codecs.getwriter('utf8'), 'replace') _cpplint_state.ResetErrorCounts() for filename in filenames: ProcessFile(filename, _cpplint_state.verbose_level) _cpplint_state.PrintErrorCounts() sys.exit(_cpplint_state.error_count > 0) if __name__ == '__main__': main() rocksdb-4.5.1/arcanist_util/lint_engine/000077500000000000000000000000001267530634000203125ustar00rootroot00000000000000rocksdb-4.5.1/arcanist_util/lint_engine/FacebookFbcodeLintEngine.php000066400000000000000000000103351267530634000256160ustar00rootroot00000000000000getPaths(); // Remove all deleted files, which are not checked by the // following linters. foreach ($paths as $key => $path) { if (!Filesystem::pathExists($this->getFilePathOnDisk($path))) { unset($paths[$key]); } } $generated_linter = new ArcanistGeneratedLinter(); $linters[] = $generated_linter; $nolint_linter = new ArcanistNoLintLinter(); $linters[] = $nolint_linter; $text_linter = new ArcanistTextLinter(); $text_linter->setCustomSeverityMap(array( ArcanistTextLinter::LINT_LINE_WRAP => ArcanistLintSeverity::SEVERITY_ADVICE, )); $linters[] = $text_linter; $java_text_linter = new ArcanistTextLinter(); $java_text_linter->setMaxLineLength(100); $java_text_linter->setCustomSeverityMap(array( ArcanistTextLinter::LINT_LINE_WRAP => ArcanistLintSeverity::SEVERITY_ADVICE, )); $linters[] = $java_text_linter; $python_linter = new ArcanistPEP8Linter(); $linters[] = $python_linter; if (!$this->getCommitHookMode()) { $cpp_linters = array(); $cpp_linters[] = $linters[] = new ArcanistCpplintLinter(); $cpp_linters[] = $linters[] = new FbcodeCppLinter(); $clang_format_linter = new FbcodeClangFormatLinter(); $linters[] = $clang_format_linter; } $spelling_linter = new ArcanistSpellingLinter(); $linters[] = $spelling_linter; foreach ($paths as $path) { $is_text = false; $text_extensions = ( '/\.('. 'cpp|cxx|c|cc|h|hpp|hxx|tcc|'. 'py|rb|hs|pl|pm|tw|'. 'php|phpt|css|js|'. 'java|'. 'thrift|'. 'lua|'. 'siv|'. 'txt'. ')$/' ); if (preg_match($text_extensions, $path)) { $is_text = true; } if ($is_text) { $nolint_linter->addPath($path); $generated_linter->addPath($path); $generated_linter->addData($path, $this->loadData($path)); if (preg_match('/\.java$/', $path)) { $java_text_linter->addPath($path); $java_text_linter->addData($path, $this->loadData($path)); } else { $text_linter->addPath($path); $text_linter->addData($path, $this->loadData($path)); } $spelling_linter->addPath($path); $spelling_linter->addData($path, $this->loadData($path)); } if (preg_match('/\.(cpp|c|cc|cxx|h|hh|hpp|hxx|tcc)$/', $path) && !preg_match('/third-party/', $path)) { foreach ($cpp_linters as &$linter) { $linter->addPath($path); $linter->addData($path, $this->loadData($path)); } $clang_format_linter->addPath($path); $clang_format_linter->addData($path, $this->loadData($path)); $clang_format_linter->setPathChangedLines( $path, $this->getPathChangedLines($path)); } // Match *.py and contbuild config files if (preg_match('/(\.(py|tw|smcprops)|^contbuild\/configs\/[^\/]*)$/', $path)) { $space_count = 4; $real_path = $this->getFilePathOnDisk($path); $dir = dirname($real_path); do { if (file_exists($dir.'/.python2space')) { $space_count = 2; break; } $dir = dirname($dir); } while ($dir != '/' && $dir != '.'); $cur_path_linter = $python_linter; $cur_path_linter->addPath($path); $cur_path_linter->addData($path, $this->loadData($path)); if (preg_match('/\.tw$/', $path)) { $cur_path_linter->setCustomSeverityMap(array( 'E251' => ArcanistLintSeverity::SEVERITY_DISABLED, )); } } } $name_linter = new ArcanistFilenameLinter(); $linters[] = $name_linter; foreach ($paths as $path) { $name_linter->addPath($path); } return $linters; } } rocksdb-4.5.1/arcanist_util/lint_engine/FacebookHowtoevenLintEngine.php000066400000000000000000000015161267530634000264130ustar00rootroot00000000000000getPaths() as $path) { // Don't try to lint deleted files or changed directories. if (!Filesystem::pathExists($path) || is_dir($path)) { continue; } if (preg_match('/\.(cpp|c|cc|cxx|h|hh|hpp|hxx|tcc)$/', $path)) { $paths[] = $path; } } $howtoeven = new FacebookHowtoevenLinter(); $howtoeven->setPaths($paths); return array($howtoeven); } } rocksdb-4.5.1/arcanist_util/unit_engine/000077500000000000000000000000001267530634000203235ustar00rootroot00000000000000rocksdb-4.5.1/arcanist_util/unit_engine/FacebookFbcodeUnitTestEngine.php000066400000000000000000000016611267530634000265020ustar00rootroot00000000000000setName("jenkins_async_test"); $result->setResult(ArcanistUnitTestResult::RESULT_POSTPONED); return array($result); } } rocksdb-4.5.1/build_tools/000077500000000000000000000000001267530634000154755ustar00rootroot00000000000000rocksdb-4.5.1/build_tools/amalgamate.py000077500000000000000000000111341267530634000201430ustar00rootroot00000000000000#!/usr/bin/python # amalgamate.py creates an amalgamation from a unity build. # It can be run with either Python 2 or 3. # An amalgamation consists of a header that includes the contents of all public # headers and a source file that includes the contents of all source files and # private headers. # # This script works by starting with the unity build file and recursively expanding # #include directives. If the #include is found in a public include directory, # that header is expanded into the amalgamation header. # # A particular header is only expanded once, so this script will # break if there are multiple inclusions of the same header that are expected to # expand differently. Similarly, this type of code causes issues: # # #ifdef FOO # #include "bar.h" # // code here # #else # #include "bar.h" // oops, doesn't get expanded # // different code here # #endif # # The solution is to move the include out of the #ifdef. from __future__ import print_function import argparse from os import path import re import sys include_re = re.compile('^[ \t]*#include[ \t]+"(.*)"[ \t]*$') included = set() excluded = set() def find_header(name, abs_path, include_paths): samedir = path.join(path.dirname(abs_path), name) if path.exists(samedir): return samedir for include_path in include_paths: include_path = path.join(include_path, name) if path.exists(include_path): return include_path return None def expand_include(include_path, f, abs_path, source_out, header_out, include_paths, public_include_paths): if include_path in included: return False included.add(include_path) with open(include_path) as f: print('#line 1 "{}"'.format(include_path), file=source_out) process_file(f, include_path, source_out, header_out, include_paths, public_include_paths) return True def process_file(f, abs_path, source_out, header_out, include_paths, public_include_paths): for (line, text) in enumerate(f): m = include_re.match(text) if m: filename = m.groups()[0] # first check private headers include_path = find_header(filename, abs_path, include_paths) if include_path: if include_path in excluded: source_out.write(text) expanded = False else: expanded = expand_include(include_path, f, abs_path, source_out, header_out, include_paths, public_include_paths) else: # now try public headers include_path = find_header(filename, abs_path, public_include_paths) if include_path: # found public header expanded = False if include_path in excluded: source_out.write(text) else: expand_include(include_path, f, abs_path, header_out, None, public_include_paths, []) else: sys.exit("unable to find {}, included in {} on line {}".format(filename, abs_path, line)) if expanded: print('#line {} "{}"'.format(line+1, abs_path), file=source_out) elif text != "#pragma once\n": source_out.write(text) def main(): parser = argparse.ArgumentParser(description="Transform a unity build into an amalgamation") parser.add_argument("source", help="source file") parser.add_argument("-I", action="append", dest="include_paths", help="include paths for private headers") parser.add_argument("-i", action="append", dest="public_include_paths", help="include paths for public headers") parser.add_argument("-x", action="append", dest="excluded", help="excluded header files") parser.add_argument("-o", dest="source_out", help="output C++ file", required=True) parser.add_argument("-H", dest="header_out", help="output C++ header file", required=True) args = parser.parse_args() include_paths = list(map(path.abspath, args.include_paths or [])) public_include_paths = list(map(path.abspath, args.public_include_paths or [])) excluded.update(map(path.abspath, args.excluded or [])) filename = args.source abs_path = path.abspath(filename) with open(filename) as f, open(args.source_out, 'w') as source_out, open(args.header_out, 'w') as header_out: print('#line 1 "{}"'.format(filename), file=source_out) print('#include "{}"'.format(header_out.name), file=source_out) process_file(f, abs_path, source_out, header_out, include_paths, public_include_paths) if __name__ == "__main__": main() rocksdb-4.5.1/build_tools/build_detect_platform000077500000000000000000000333001267530634000217550ustar00rootroot00000000000000#!/bin/sh # # Detects OS we're compiling on and outputs a file specified by the first # argument, which in turn gets read while processing Makefile. # # The output will set the following variables: # CC C Compiler path # CXX C++ Compiler path # PLATFORM_LDFLAGS Linker flags # JAVA_LDFLAGS Linker flags for RocksDBJava # JAVA_STATIC_LDFLAGS Linker flags for RocksDBJava static build # PLATFORM_SHARED_EXT Extension for shared libraries # PLATFORM_SHARED_LDFLAGS Flags for building shared library # PLATFORM_SHARED_CFLAGS Flags for compiling objects for shared library # PLATFORM_CCFLAGS C compiler flags # PLATFORM_CXXFLAGS C++ compiler flags. Will contain: # PLATFORM_SHARED_VERSIONED Set to 'true' if platform supports versioned # shared libraries, empty otherwise. # # The PLATFORM_CCFLAGS and PLATFORM_CXXFLAGS might include the following: # # -DLEVELDB_PLATFORM_POSIX if cstdatomic is present # -DLEVELDB_PLATFORM_NOATOMIC if it is not # -DSNAPPY if the Snappy library is present # -DLZ4 if the LZ4 library is present # -DZSTD if the ZSTD library is present # -DNUMA if the NUMA library is present # # Using gflags in rocksdb: # Our project depends on gflags, which requires users to take some extra steps # before they can compile the whole repository: # 1. Install gflags. You may download it from here: # https://code.google.com/p/gflags/ # 2. Once install, add the include path/lib path for gflags to CPATH and # LIBRARY_PATH respectively. If installed with default mode, the # lib and include path will be /usr/local/lib and /usr/local/include # Mac user can do this by having brew installed and running brew install gflags OUTPUT=$1 if test -z "$OUTPUT"; then echo "usage: $0 " >&2 exit 1 fi # we depend on C++11 PLATFORM_CXXFLAGS="-std=c++11" # we currently depend on POSIX platform COMMON_FLAGS="-DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX" # Default to fbcode gcc on internal fb machines if [ -z "$ROCKSDB_NO_FBCODE" -a -d /mnt/gvfs/third-party ]; then FBCODE_BUILD="true" # If we're compiling with TSAN we need pic build PIC_BUILD=$COMPILE_WITH_TSAN if [ -z "$ROCKSDB_FBCODE_BUILD_WITH_481" ]; then source "$PWD/build_tools/fbcode_config.sh" else # we need this to build with MySQL. Don't use for other purposes. source "$PWD/build_tools/fbcode_config4.8.1.sh" fi fi # Delete existing output, if it exists rm -f "$OUTPUT" touch "$OUTPUT" if test -z "$CC"; then CC=cc fi if test -z "$CXX"; then CXX=g++ fi # Detect OS if test -z "$TARGET_OS"; then TARGET_OS=`uname -s` fi if test -z "$TARGET_ARCHITECTURE"; then TARGET_ARCHITECTURE=`uname -m` fi if test -z "$CLANG_SCAN_BUILD"; then CLANG_SCAN_BUILD=scan-build fi if test -z "$CLANG_ANALYZER"; then CLANG_ANALYZER=$(which clang++ 2> /dev/null) fi COMMON_FLAGS="$COMMON_FLAGS ${CFLAGS}" CROSS_COMPILE= PLATFORM_CCFLAGS= PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS" PLATFORM_SHARED_EXT="so" PLATFORM_SHARED_LDFLAGS="-Wl,--no-as-needed -shared -Wl,-soname -Wl," PLATFORM_SHARED_CFLAGS="-fPIC" PLATFORM_SHARED_VERSIONED=true # generic port files (working on all platform by #ifdef) go directly in /port GENERIC_PORT_FILES=`cd "$ROCKSDB_ROOT"; find port -name '*.cc' | tr "\n" " "` # On GCC, we pick libc's memcmp over GCC's memcmp via -fno-builtin-memcmp case "$TARGET_OS" in Darwin) PLATFORM=OS_MACOSX COMMON_FLAGS="$COMMON_FLAGS -DOS_MACOSX" PLATFORM_SHARED_EXT=dylib PLATFORM_SHARED_LDFLAGS="-dynamiclib -install_name " # PORT_FILES=port/darwin/darwin_specific.cc ;; IOS) PLATFORM=IOS COMMON_FLAGS="$COMMON_FLAGS -DOS_MACOSX -DIOS_CROSS_COMPILE -DROCKSDB_LITE" PLATFORM_SHARED_EXT=dylib PLATFORM_SHARED_LDFLAGS="-dynamiclib -install_name " CROSS_COMPILE=true PLATFORM_SHARED_VERSIONED= ;; Linux) PLATFORM=OS_LINUX COMMON_FLAGS="$COMMON_FLAGS -DOS_LINUX" if [ -z "$USE_CLANG" ]; then COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp" fi PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt" # PORT_FILES=port/linux/linux_specific.cc ;; SunOS) PLATFORM=OS_SOLARIS COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_SOLARIS" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt" # PORT_FILES=port/sunos/sunos_specific.cc ;; FreeBSD) PLATFORM=OS_FREEBSD COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_FREEBSD" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread" # PORT_FILES=port/freebsd/freebsd_specific.cc ;; NetBSD) PLATFORM=OS_NETBSD COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_NETBSD" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lgcc_s" # PORT_FILES=port/netbsd/netbsd_specific.cc ;; OpenBSD) PLATFORM=OS_OPENBSD COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_OPENBSD" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -pthread" # PORT_FILES=port/openbsd/openbsd_specific.cc ;; DragonFly) PLATFORM=OS_DRAGONFLYBSD COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_DRAGONFLYBSD" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread" # PORT_FILES=port/dragonfly/dragonfly_specific.cc ;; Cygwin) PLATFORM=CYGWIN PLATFORM_SHARED_CFLAGS="" PLATFORM_CXXFLAGS="-std=gnu++11" COMMON_FLAGS="$COMMON_FLAGS -DCYGWIN" if [ -z "$USE_CLANG" ]; then COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp" fi PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt" # PORT_FILES=port/linux/linux_specific.cc ;; OS_ANDROID_CROSSCOMPILE) PLATFORM=OS_ANDROID COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_ANDROID -DLEVELDB_PLATFORM_POSIX" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS " # All pthread features are in the Android C library # PORT_FILES=port/android/android.cc CROSS_COMPILE=true ;; *) echo "Unknown platform!" >&2 exit 1 esac PLATFORM_CXXFLAGS="$PLATFORM_CXXFLAGS ${CXXFLAGS}" JAVA_LDFLAGS="$PLATFORM_LDFLAGS" JAVA_STATIC_LDFLAGS="$PLATFORM_LDFLAGS" if [ "$CROSS_COMPILE" = "true" -o "$FBCODE_BUILD" = "true" ]; then # Cross-compiling; do not try any compilation tests. # Also don't need any compilation tests if compiling on fbcode true else # Test whether fallocate is available $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null < #include int main() { int fd = open("/dev/null", 0); fallocate(fd, FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE, 0, 1024); } EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_FALLOCATE_PRESENT" fi # Test whether Snappy library is installed # http://code.google.com/p/snappy/ $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null < int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DSNAPPY" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lsnappy" JAVA_LDFLAGS="$JAVA_LDFLAGS -lsnappy" fi # Test whether gflags library is installed # http://gflags.github.io/gflags/ # check if the namespace is gflags $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null << EOF #include using namespace gflags; int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DGFLAGS=gflags" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lgflags" else # check if namespace is google $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null << EOF #include using namespace google; int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DGFLAGS=google" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lgflags" fi fi # Test whether zlib library is installed $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null < int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DZLIB" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lz" JAVA_LDFLAGS="$JAVA_LDFLAGS -lz" fi # Test whether bzip library is installed $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null < int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DBZIP2" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lbz2" JAVA_LDFLAGS="$JAVA_LDFLAGS -lbz2" fi # Test whether lz4 library is installed $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null < #include int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DLZ4" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -llz4" JAVA_LDFLAGS="$JAVA_LDFLAGS -llz4" fi # Test whether zstd library is installed $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null < int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DZSTD" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lzstd" JAVA_LDFLAGS="$JAVA_LDFLAGS -lzstd" fi # Test whether numa is available $CXX $CFLAGS -x c++ - -o /dev/null -lnuma 2>/dev/null < #inlcude int main() {} EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DNUMA" PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lnuma" JAVA_LDFLAGS="$JAVA_LDFLAGS -lnuma" fi # Test whether jemalloc is available if echo 'int main() {}' | $CXX $CFLAGS -x c++ - -o /dev/null -ljemalloc \ 2>/dev/null; then PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -ljemalloc" JAVA_LDFLAGS="$JAVA_LDFLAGS -ljemalloc" else # jemalloc is not available. Let's try tcmalloc if echo 'int main() {}' | $CXX $CFLAGS -x c++ - -o /dev/null \ -ltcmalloc 2>/dev/null; then PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -ltcmalloc" JAVA_LDFLAGS="$JAVA_LDFLAGS -ltcmalloc" fi fi # Test whether malloc_usable_size is available $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null < int main() { size_t res = malloc_usable_size(0); return 0; } EOF if [ "$?" = 0 ]; then COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_MALLOC_USABLE_SIZE" fi fi # TODO(tec): Fix -Wshorten-64-to-32 errors on FreeBSD and enable the warning. # -Wshorten-64-to-32 breaks compilation on FreeBSD i386 if ! [ "$TARGET_OS" = FreeBSD -a "$TARGET_ARCHITECTURE" = i386 ]; then # Test whether -Wshorten-64-to-32 is available $CXX $CFLAGS -x c++ - -o /dev/null -Wshorten-64-to-32 2>/dev/null <> "$OUTPUT" echo "CXX=$CXX" >> "$OUTPUT" echo "PLATFORM=$PLATFORM" >> "$OUTPUT" echo "PLATFORM_LDFLAGS=$PLATFORM_LDFLAGS" >> "$OUTPUT" echo "JAVA_LDFLAGS=$JAVA_LDFLAGS" >> "$OUTPUT" echo "JAVA_STATIC_LDFLAGS=$JAVA_STATIC_LDFLAGS" >> "$OUTPUT" echo "VALGRIND_VER=$VALGRIND_VER" >> "$OUTPUT" echo "PLATFORM_CCFLAGS=$PLATFORM_CCFLAGS" >> "$OUTPUT" echo "PLATFORM_CXXFLAGS=$PLATFORM_CXXFLAGS" >> "$OUTPUT" echo "PLATFORM_SHARED_CFLAGS=$PLATFORM_SHARED_CFLAGS" >> "$OUTPUT" echo "PLATFORM_SHARED_EXT=$PLATFORM_SHARED_EXT" >> "$OUTPUT" echo "PLATFORM_SHARED_LDFLAGS=$PLATFORM_SHARED_LDFLAGS" >> "$OUTPUT" echo "PLATFORM_SHARED_VERSIONED=$PLATFORM_SHARED_VERSIONED" >> "$OUTPUT" echo "EXEC_LDFLAGS=$EXEC_LDFLAGS" >> "$OUTPUT" echo "JEMALLOC_INCLUDE=$JEMALLOC_INCLUDE" >> "$OUTPUT" echo "JEMALLOC_LIB=$JEMALLOC_LIB" >> "$OUTPUT" echo "ROCKSDB_MAJOR=$ROCKSDB_MAJOR" >> "$OUTPUT" echo "ROCKSDB_MINOR=$ROCKSDB_MINOR" >> "$OUTPUT" echo "ROCKSDB_PATCH=$ROCKSDB_PATCH" >> "$OUTPUT" echo "CLANG_SCAN_BUILD=$CLANG_SCAN_BUILD" >> "$OUTPUT" echo "CLANG_ANALYZER=$CLANG_ANALYZER" >> "$OUTPUT" rocksdb-4.5.1/build_tools/dependencies.sh000066400000000000000000000034401267530634000204600ustar00rootroot00000000000000GCC_BASE=/mnt/vol/engshare/fbcode/third-party2/gcc/4.9.x/centos6-native/1317bc4/ CLANG_BASE=/mnt/vol/engshare/fbcode/third-party2/clang/3.7.1/centos6-native/9d9ecb9/ LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/5710d6a0fb0d12820aac0bffcd7fcb8646e7fff7/4.9.x/gcc-4.9-glibc-2.20/024dbc3 GLIBC_BASE=/mnt/gvfs/third-party2/glibc/0600c95b31226b5e535614c590677d87c62d8016/2.20/gcc-4.9-glibc-2.20/500e281 SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/cbf6f1f209e5bd160bdc5d971744e039f36b1566/1.1.3/gcc-4.9-glibc-2.20/e9936bf ZLIB_BASE=/mnt/gvfs/third-party2/zlib/6d39cb54708049f527e713ad19f2aadb9d3667e8/1.2.8/gcc-4.9-glibc-2.20/e9936bf BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/2ddd45f0853bfc8bb1c27f0f447236a1a26c338a/1.0.6/gcc-4.9-glibc-2.20/e9936bf LZ4_BASE=/mnt/gvfs/third-party2/lz4/6858fac689e0f92e584224d91bdb0e39f6c8320d/r131/gcc-4.9-glibc-2.20/e9936bf ZSTD_BASE=/mnt/gvfs/third-party2/zstd/69d56740ffb89d8bc81ded8ec428c01a813ea948/0.4.7/gcc-4.9-glibc-2.20/e9936bf GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/c7275a4ceae0aca0929e56964a31dafc53c1ee96/2.1.1/gcc-4.8.1-glibc-2.17/c3f970a JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/5f0be745ddc0f86f22c8c8bb64b6b1163c93df91/4.0.3/gcc-4.9-glibc-2.20/a6c5e1e NUMA_BASE=/mnt/gvfs/third-party2/numa/ae54a5ed22cdabb1c6446dce4e8ffae5b4446d73/2.0.8/gcc-4.9-glibc-2.20/e9936bf LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/121f1a75c4414683aea8c70b761bfaf187f7c1a3/trunk/gcc-4.9-glibc-2.20/12266b1 KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/1a48835975c66d30e47770ec419758ed3b9ba010/3.10.62-62_fbk17_03959_ge29cc63/gcc-4.9-glibc-2.20/da39a3e BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/75670d0d8ef4891fd1ec2a7513ef01cd002c823b/2.25/centos6-native/da39a3e VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/af85c56f424cd5edfc2c97588299b44ecdec96bb/3.10.0/gcc-4.9-glibc-2.20/e9936bf rocksdb-4.5.1/build_tools/dependencies_4.8.1.sh000066400000000000000000000034671267530634000212210ustar00rootroot00000000000000GCC_BASE=/mnt/vol/engshare/fbcode/third-party2/gcc/4.8.1/centos6-native/cc6c9dc/ CLANG_BASE=/mnt/vol/engshare/fbcode/third-party2/clang/3.7.1/centos6-native/9d9ecb9/ LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/5710d6a0fb0d12820aac0bffcd7fcb8646e7fff7/4.8.1/gcc-4.8.1-glibc-2.17/8aac7fc GLIBC_BASE=/mnt/gvfs/third-party2/glibc/0600c95b31226b5e535614c590677d87c62d8016/2.17/gcc-4.8.1-glibc-2.17/99df8fc SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/cbf6f1f209e5bd160bdc5d971744e039f36b1566/1.1.3/gcc-4.8.1-glibc-2.17/c3f970a ZLIB_BASE=/mnt/gvfs/third-party2/zlib/6d39cb54708049f527e713ad19f2aadb9d3667e8/1.2.8/gcc-4.8.1-glibc-2.17/c3f970a BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/2ddd45f0853bfc8bb1c27f0f447236a1a26c338a/1.0.6/gcc-4.8.1-glibc-2.17/c3f970a LZ4_BASE=/mnt/gvfs/third-party2/lz4/6858fac689e0f92e584224d91bdb0e39f6c8320d/r131/gcc-4.8.1-glibc-2.17/c3f970a ZSTD_BASE=/mnt/gvfs/third-party2/zstd/69d56740ffb89d8bc81ded8ec428c01a813ea948/0.4.7/gcc-4.8.1-glibc-2.17/c3f970a GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/c7275a4ceae0aca0929e56964a31dafc53c1ee96/2.1.1/gcc-4.8.1-glibc-2.17/c3f970a JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/5f0be745ddc0f86f22c8c8bb64b6b1163c93df91/4.0.3/gcc-4.8.1-glibc-2.17/8d31e51 NUMA_BASE=/mnt/gvfs/third-party2/numa/ae54a5ed22cdabb1c6446dce4e8ffae5b4446d73/2.0.8/gcc-4.8.1-glibc-2.17/c3f970a LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/121f1a75c4414683aea8c70b761bfaf187f7c1a3/trunk/gcc-4.8.1-glibc-2.17/675d945 KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/1a48835975c66d30e47770ec419758ed3b9ba010/3.10.62-62_fbk17_03959_ge29cc63/gcc-4.8.1-glibc-2.17/da39a3e BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/75670d0d8ef4891fd1ec2a7513ef01cd002c823b/2.25/centos6-native/da39a3e VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/af85c56f424cd5edfc2c97588299b44ecdec96bb/3.8.1/gcc-4.8.1-glibc-2.17/c3f970a rocksdb-4.5.1/build_tools/dockerbuild.sh000077500000000000000000000001041267530634000203160ustar00rootroot00000000000000#!/bin/bash docker run -v $PWD:/rocks -w /rocks buildpack-deps make rocksdb-4.5.1/build_tools/fb_compile_mongo.sh000077500000000000000000000026661267530634000213440ustar00rootroot00000000000000#!/bin/sh # fail early set -e if test -z $ROCKSDB_PATH; then ROCKSDB_PATH=~/rocksdb fi source $ROCKSDB_PATH/build_tools/fbcode_config4.8.1.sh EXTRA_LDFLAGS="" if test -z $ALLOC; then # default ALLOC=tcmalloc elif [[ $ALLOC == "jemalloc" ]]; then ALLOC=system EXTRA_LDFLAGS+=" -Wl,--whole-archive $JEMALLOC_LIB -Wl,--no-whole-archive" fi # we need to force mongo to use static library, not shared STATIC_LIB_DEP_DIR='build/static_library_dependencies' test -d $STATIC_LIB_DEP_DIR || mkdir $STATIC_LIB_DEP_DIR test -h $STATIC_LIB_DEP_DIR/`basename $SNAPPY_LIBS` || ln -s $SNAPPY_LIBS $STATIC_LIB_DEP_DIR test -h $STATIC_LIB_DEP_DIR/`basename $LZ4_LIBS` || ln -s $LZ4_LIBS $STATIC_LIB_DEP_DIR EXTRA_LDFLAGS+=" -L $STATIC_LIB_DEP_DIR" set -x EXTRA_CMD="" if ! test -e version.json; then # this is Mongo 3.0 EXTRA_CMD="--rocksdb \ --variant-dir=linux2/norm --cxx=${CXX} \ --cc=${CC} \ --use-system-zlib" # add this line back to normal code path # when https://jira.mongodb.org/browse/SERVER-19123 is resolved fi scons \ LINKFLAGS="$EXTRA_LDFLAGS $EXEC_LDFLAGS $PLATFORM_LDFLAGS" \ CCFLAGS="$CXXFLAGS -L $STATIC_LIB_DEP_DIR" \ LIBS="lz4 gcc stdc++" \ LIBPATH="$ROCKSDB_PATH" \ CPPPATH="$ROCKSDB_PATH/include" \ -j32 \ --allocator=$ALLOC \ --nostrip \ --opt=on \ --disable-minimum-compiler-version-enforcement \ --use-system-snappy \ --disable-warnings-as-errors \ $EXTRA_CMD $* rocksdb-4.5.1/build_tools/fbcode_config.sh000066400000000000000000000075501267530634000206070ustar00rootroot00000000000000#!/bin/sh # # Set environment variables so that we can compile rocksdb using # fbcode settings. It uses the latest g++ and clang compilers and also # uses jemalloc # Environment variables that change the behavior of this script: # PIC_BUILD -- if true, it will only take pic versions of libraries from fbcode. libraries that don't have pic variant will not be included BASEDIR=`dirname $BASH_SOURCE` source "$BASEDIR/dependencies.sh" CFLAGS="" # libgcc LIBGCC_INCLUDE="$LIBGCC_BASE/include" LIBGCC_LIBS=" -L $LIBGCC_BASE/libs" # glibc GLIBC_INCLUDE="$GLIBC_BASE/include" GLIBC_LIBS=" -L $GLIB_BASE/lib" # snappy SNAPPY_INCLUDE=" -I $SNAPPY_BASE/include/" if test -z $PIC_BUILD; then SNAPPY_LIBS=" $SNAPPY_BASE/lib/libsnappy.a" else SNAPPY_LIBS=" $SNAPPY_BASE/lib/libsnappy_pic.a" fi CFLAGS+=" -DSNAPPY" if test -z $PIC_BUILD; then # location of zlib headers and libraries ZLIB_INCLUDE=" -I $ZLIB_BASE/include/" ZLIB_LIBS=" $ZLIB_BASE/lib/libz.a" CFLAGS+=" -DZLIB" # location of bzip headers and libraries BZIP_INCLUDE=" -I $BZIP2_BASE/include/" BZIP_LIBS=" $BZIP2_BASE/lib/libbz2.a" CFLAGS+=" -DBZIP2" LZ4_INCLUDE=" -I $LZ4_BASE/include/" LZ4_LIBS=" $LZ4_BASE/lib/liblz4.a" CFLAGS+=" -DLZ4" ZSTD_INCLUDE=" -I $ZSTD_BASE/include/" ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a" CFLAGS+=" -DZSTD" fi # location of gflags headers and libraries GFLAGS_INCLUDE=" -I $GFLAGS_BASE/include/" if test -z $PIC_BUILD; then GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags.a" else GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags_pic.a" fi CFLAGS+=" -DGFLAGS=google" # location of jemalloc JEMALLOC_INCLUDE=" -I $JEMALLOC_BASE/include/" JEMALLOC_LIB=" $JEMALLOC_BASE/lib/libjemalloc.a" if test -z $PIC_BUILD; then # location of numa NUMA_INCLUDE=" -I $NUMA_BASE/include/" NUMA_LIB=" $NUMA_BASE/lib/libnuma.a" CFLAGS+=" -DNUMA" # location of libunwind LIBUNWIND="$LIBUNWIND_BASE/lib/libunwind.a" fi # use Intel SSE support for checksum calculations export USE_SSE=1 BINUTILS="$BINUTILS_BASE/bin" AR="$BINUTILS/ar" DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE" STDLIBS="-L $GCC_BASE/lib64" CLANG_BIN="$CLANG_BASE/bin" CLANG_LIB="$CLANG_BASE/lib" CLANG_SRC="$CLANG_BASE/../../src" CLANG_ANALYZER="$CLANG_BIN/clang++" CLANG_SCAN_BUILD="$CLANG_SRC/clang/tools/scan-build/scan-build" if [ -z "$USE_CLANG" ]; then # gcc CC="$GCC_BASE/bin/gcc" CXX="$GCC_BASE/bin/g++" CFLAGS+=" -B$BINUTILS/gold" CFLAGS+=" -isystem $GLIBC_INCLUDE" CFLAGS+=" -isystem $LIBGCC_INCLUDE" else # clang CLANG_INCLUDE="$CLANG_LIB/clang/*/include" CC="$CLANG_BIN/clang" CXX="$CLANG_BIN/clang++" KERNEL_HEADERS_INCLUDE="$KERNEL_HEADERS_BASE/include" CFLAGS+=" -B$BINUTILS/gold -nostdinc -nostdlib" CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.9.x " CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.9.x/x86_64-facebook-linux " CFLAGS+=" -isystem $GLIBC_INCLUDE" CFLAGS+=" -isystem $LIBGCC_INCLUDE" CFLAGS+=" -isystem $CLANG_INCLUDE" CFLAGS+=" -isystem $KERNEL_HEADERS_INCLUDE/linux " CFLAGS+=" -isystem $KERNEL_HEADERS_INCLUDE " CXXFLAGS="-nostdinc++" fi CFLAGS+=" $DEPS_INCLUDE" CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE" CXXFLAGS+=" $CFLAGS" EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB" EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/gcc-4.9-glibc-2.20/lib/ld.so" EXEC_LDFLAGS+=" $LIBUNWIND" EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/gcc-4.9-glibc-2.20/lib" PLATFORM_LDFLAGS="$LIBGCC_LIBS $GLIBC_LIBS $STDLIBS -lgcc -lstdc++" EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS" VALGRIND_VER="$VALGRIND_BASE/bin/" export CC CXX AR CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE CLANG_ANALYZER CLANG_SCAN_BUILD rocksdb-4.5.1/build_tools/fbcode_config4.8.1.sh000066400000000000000000000063351267530634000212000ustar00rootroot00000000000000#!/bin/sh # # Set environment variables so that we can compile rocksdb using # fbcode settings. It uses the latest g++ compiler and also # uses jemalloc BASEDIR=`dirname $BASH_SOURCE` source "$BASEDIR/dependencies_4.8.1.sh" # location of libgcc LIBGCC_INCLUDE="$LIBGCC_BASE/include" LIBGCC_LIBS=" -L $LIBGCC_BASE/libs" # location of glibc GLIBC_INCLUDE="$GLIBC_BASE/include" GLIBC_LIBS=" -L $GLIBC_BASE/lib" # location of snappy headers and libraries SNAPPY_INCLUDE=" -I $SNAPPY_BASE/include" SNAPPY_LIBS=" $SNAPPY_BASE/lib/libsnappy.a" # location of zlib headers and libraries ZLIB_INCLUDE=" -I $ZLIB_BASE/include" ZLIB_LIBS=" $ZLIB_BASE/lib/libz.a" # location of bzip headers and libraries BZIP2_INCLUDE=" -I $BZIP2_BASE/include/" BZIP2_LIBS=" $BZIP2_BASE/lib/libbz2.a" LZ4_INCLUDE=" -I $LZ4_BASE/include" LZ4_LIBS=" $LZ4_BASE/lib/liblz4.a" ZSTD_INCLUDE=" -I $ZSTD_BASE/include" ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a" # location of gflags headers and libraries GFLAGS_INCLUDE=" -I $GFLAGS_BASE/include/" GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags.a" # location of jemalloc JEMALLOC_INCLUDE=" -I $JEMALLOC_BASE/include" JEMALLOC_LIB="$JEMALLOC_BASE/lib/libjemalloc.a" # location of numa NUMA_INCLUDE=" -I $NUMA_BASE/include/" NUMA_LIB=" $NUMA_BASE/lib/libnuma.a" # location of libunwind LIBUNWIND="$LIBUNWIND_BASE/lib/libunwind.a" # use Intel SSE support for checksum calculations export USE_SSE=1 BINUTILS="$BINUTILS_BASE/bin" AR="$BINUTILS/ar" DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP2_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE" STDLIBS="-L $GCC_BASE/lib64" if [ -z "$USE_CLANG" ]; then # gcc CC="$GCC_BASE/bin/gcc" CXX="$GCC_BASE/bin/g++" CFLAGS="-B$BINUTILS/gold -m64 -mtune=generic" CFLAGS+=" -isystem $GLIBC_INCLUDE" CFLAGS+=" -isystem $LIBGCC_INCLUDE" else # clang CLANG_BIN="$CLANG_BASE/bin" CLANG_LIB="$CLANG_BASE/lib" CLANG_INCLUDE="$CLANG_LIB/clang/*/include" CC="$CLANG_BIN/clang" CXX="$CLANG_BIN/clang++" KERNEL_HEADERS_INCLUDE="$KERNEL_HEADERS_BASE/include/" CFLAGS="-B$BINUTILS/gold -nostdinc -nostdlib" CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.8.1 " CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.8.1/x86_64-facebook-linux " CFLAGS+=" -isystem $GLIBC_INCLUDE" CFLAGS+=" -isystem $LIBGCC_INCLUDE" CFLAGS+=" -isystem $CLANG_INCLUDE" CFLAGS+=" -isystem $KERNEL_HEADERS_INCLUDE/linux " CFLAGS+=" -isystem $KERNEL_HEADERS_INCLUDE " CXXFLAGS="-nostdinc++" fi CFLAGS+=" $DEPS_INCLUDE" CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE" CFLAGS+=" -DSNAPPY -DGFLAGS=google -DZLIB -DBZIP2 -DLZ4 -DZSTD -DNUMA" CXXFLAGS+=" $CFLAGS" EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP2_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB" EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/gcc-4.8.1-glibc-2.17/lib/ld.so" EXEC_LDFLAGS+=" $LIBUNWIND" EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/gcc-4.8.1-glibc-2.17/lib" PLATFORM_LDFLAGS="$LIBGCC_LIBS $GLIBC_LIBS $STDLIBS -lgcc -lstdc++" EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP2_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS" VALGRIND_VER="$VALGRIND_BASE/bin/" export CC CXX AR CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE rocksdb-4.5.1/build_tools/format-diff.sh000077500000000000000000000065541267530634000202440ustar00rootroot00000000000000#!/bin/bash # If clang_format_diff.py command is not specfied, we assume we are able to # access directly without any path. if [ -z $CLANG_FORMAT_DIFF ] then CLANG_FORMAT_DIFF="clang-format-diff.py" fi # Check clang-format-diff.py if ! which $CLANG_FORMAT_DIFF &> /dev/null then echo "You didn't have clang-format-diff.py available in your computer!" echo "You can download it by running: " echo " curl http://goo.gl/iUW1u2" exit 128 fi # Check argparse, a library that clang-format-diff.py requires. python 2>/dev/null << EOF import argparse EOF if [ "$?" != 0 ] then echo "To run clang-format-diff.py, we'll need the library "argparse" to be" echo "installed. You can try either of the follow ways to install it:" echo " 1. Manually download argparse: https://pypi.python.org/pypi/argparse" echo " 2. easy_install argparse (if you have easy_install)" echo " 3. pip install argparse (if you have pip)" exit 129 fi # TODO(kailiu) following work is not complete since we still need to figure # out how to add the modified files done pre-commit hook to git's commit index. # # Check if this script has already been added to pre-commit hook. # Will suggest user to add this script to pre-commit hook if their pre-commit # is empty. # PRE_COMMIT_SCRIPT_PATH="`git rev-parse --show-toplevel`/.git/hooks/pre-commit" # if ! ls $PRE_COMMIT_SCRIPT_PATH &> /dev/null # then # echo "Would you like to add this script to pre-commit hook, which will do " # echo -n "the format check for all the affected lines before you check in (y/n):" # read add_to_hook # if [ "$add_to_hook" == "y" ] # then # ln -s `git rev-parse --show-toplevel`/build_tools/format-diff.sh $PRE_COMMIT_SCRIPT_PATH # fi # fi set -e uncommitted_code=`git diff HEAD` # If there's no uncommitted changes, we assume user are doing post-commit # format check, in which case we'll check the modified lines from latest commit. # Otherwise, we'll check format of the uncommitted code only. if [ -z "$uncommitted_code" ] then # Check the format of last commit diffs=$(git diff -U0 HEAD^ | $CLANG_FORMAT_DIFF -p 1) else # Check the format of uncommitted lines, diffs=$(git diff -U0 HEAD | $CLANG_FORMAT_DIFF -p 1) fi if [ -z "$diffs" ] then echo "Nothing needs to be reformatted!" exit 0 fi # Highlight the insertion/deletion from the clang-format-diff.py's output COLOR_END="\033[0m" COLOR_RED="\033[0;31m" COLOR_GREEN="\033[0;32m" echo -e "Detect lines that doesn't follow the format rules:\r" # Add the color to the diff. lines added will be green; lines removed will be red. echo "$diffs" | sed -e "s/\(^-.*$\)/`echo -e \"$COLOR_RED\1$COLOR_END\"`/" | sed -e "s/\(^+.*$\)/`echo -e \"$COLOR_GREEN\1$COLOR_END\"`/" if [[ "$OPT" == *"-DTRAVIS"* ]] then exit 1 fi echo -e "Would you like to fix the format automatically (y/n): \c" # Make sure under any mode, we can read user input. exec < /dev/tty read to_fix if [ "$to_fix" != "y" ] then exit 1 fi # Do in-place format adjustment. git diff -U0 HEAD^ | $CLANG_FORMAT_DIFF -i -p 1 echo "Files reformatted!" # Amend to last commit if user do the post-commit format check if [ -z "$uncommitted_code" ]; then echo -e "Would you like to amend the changes to last commit (`git log HEAD --oneline | head -1`)? (y/n): \c" read to_amend if [ "$to_amend" == "y" ] then git commit -a --amend --reuse-message HEAD echo "Amended to last commit" fi fi rocksdb-4.5.1/build_tools/make_new_version.sh000077500000000000000000000022301267530634000213640ustar00rootroot00000000000000#!/bin/bash # Copyright (c) 2013, Facebook, Inc. All rights reserved. # This source code is licensed under the BSD-style license found in the # LICENSE file in the root directory of this source tree. An additional grant # of patent rights can be found in the PATENTS file in the same directory. set -e if [ -z "$GIT" ] then GIT="git" fi # Print out the colored progress info so that it can be brainlessly # distinguished by users. function title() { echo -e "\033[1;32m$*\033[0m" } usage="Create new RocksDB version and prepare it for the release process\n" usage+="USAGE: ./make_new_version.sh " # -- Pre-check if [[ $# < 1 ]]; then echo -e $usage exit 1 fi ROCKSDB_VERSION=$1 GIT_BRANCH=`git rev-parse --abbrev-ref HEAD` echo $GIT_BRANCH if [ $GIT_BRANCH != "master" ]; then echo "Error: Current branch is '$GIT_BRANCH', Please switch to master branch." exit 1 fi title "Adding new tag for this release ..." BRANCH="$ROCKSDB_VERSION.fb" $GIT checkout -b $BRANCH # Setting up the proxy for remote repo access title "Pushing new branch to remote repo ..." git push origin --set-upstream $BRANCH title "Branch $BRANCH is pushed to github;" rocksdb-4.5.1/build_tools/make_package.sh000077500000000000000000000052311267530634000204250ustar00rootroot00000000000000#/usr/bin/env bash set -e function log() { echo "[+] $1" } function fatal() { echo "[!] $1" exit 1 } function platform() { local __resultvar=$1 if [[ -f "/etc/yum.conf" ]]; then eval $__resultvar="centos" elif [[ -f "/etc/dpkg/dpkg.cfg" ]]; then eval $__resultvar="ubuntu" else fatal "Unknwon operating system" fi } platform OS function package() { if [[ $OS = "ubuntu" ]]; then if dpkg --get-selections | grep --quiet $1; then log "$1 is already installed. skipping." else apt-get install $@ -y fi elif [[ $OS = "centos" ]]; then if rpm -qa | grep --quiet $1; then log "$1 is already installed. skipping." else yum install $@ -y fi fi } function detect_fpm_output() { if [[ $OS = "ubuntu" ]]; then export FPM_OUTPUT=deb elif [[ $OS = "centos" ]]; then export FPM_OUTPUT=rpm fi } detect_fpm_output function gem_install() { if gem list | grep --quiet $1; then log "$1 is already installed. skipping." else gem install $@ fi } function main() { if [[ $# -ne 1 ]]; then fatal "Usage: $0 " else log "using rocksdb version: $1" fi if [[ -d /vagrant ]]; then if [[ $OS = "ubuntu" ]]; then package g++-4.7 export CXX=g++-4.7 # the deb would depend on libgflags2, but the static lib is the only thing # installed by make install package libgflags-dev package ruby-all-dev elif [[ $OS = "centos" ]]; then pushd /etc/yum.repos.d if [[ ! -f /etc/yum.repos.d/devtools-1.1.repo ]]; then wget http://people.centos.org/tru/devtools-1.1/devtools-1.1.repo fi package devtoolset-1.1-gcc --enablerepo=testing-1.1-devtools-6 package devtoolset-1.1-gcc-c++ --enablerepo=testing-1.1-devtools-6 export CC=/opt/centos/devtoolset-1.1/root/usr/bin/gcc export CPP=/opt/centos/devtoolset-1.1/root/usr/bin/cpp export CXX=/opt/centos/devtoolset-1.1/root/usr/bin/c++ export PATH=$PATH:/opt/centos/devtoolset-1.1/root/usr/bin popd if ! rpm -qa | grep --quiet gflags; then rpm -i https://github.com/schuhschuh/gflags/releases/download/v2.1.0/gflags-devel-2.1.0-1.amd64.rpm fi package ruby package ruby-devel package rubygems package rpm-build fi fi gem_install fpm make static_lib make install INSTALL_PATH=package fpm \ -s dir \ -t $FPM_OUTPUT \ -n rocksdb \ -v $1 \ --prefix /usr \ --url http://rocksdb.org/ \ -m rocksdb@fb.com \ --license BSD \ --vendor Facebook \ --description "RocksDB is an embeddable persistent key-value store for fast storage." \ package } main $@ rocksdb-4.5.1/build_tools/precommit_checker.py000077500000000000000000000120101267530634000215270ustar00rootroot00000000000000#!/usr/local/fbcode/gcc-4.8.1-glibc-2.17-fb/bin/python2.7 from __future__ import absolute_import from __future__ import division from __future__ import print_function from __future__ import unicode_literals import argparse import commands import subprocess import sys import re import os import time # # Simple logger # class Log: LOG_FILE = "/tmp/precommit-check.log" def __init__(self): self.filename = Log.LOG_FILE self.f = open(self.filename, 'w+', 0) def caption(self, str): line = "\n##### %s #####\n" % str if self.f: self.f.write("%s \n" % line) else: print(line) def error(self, str): data = "\n\n##### ERROR ##### %s" % str if self.f: self.f.write("%s \n" % data) else: print(data) def log(self, str): if self.f: self.f.write("%s \n" % str) else: print(str) # # Shell Environment # class Env(object): def __init__(self, tests): self.tests = tests self.log = Log() def shell(self, cmd, path=os.getcwd()): if path: os.chdir(path) self.log.log("==== shell session ===========================") self.log.log("%s> %s" % (path, cmd)) status = subprocess.call("cd %s; %s" % (path, cmd), shell=True, stdout=self.log.f, stderr=self.log.f) self.log.log("status = %s" % status) self.log.log("============================================== \n\n") return status def GetOutput(self, cmd, path=os.getcwd()): if path: os.chdir(path) self.log.log("==== shell session ===========================") self.log.log("%s> %s" % (path, cmd)) status, out = commands.getstatusoutput(cmd) self.log.log("status = %s" % status) self.log.log("out = %s" % out) self.log.log("============================================== \n\n") return status, out # # Pre-commit checker # class PreCommitChecker(Env): def __init__(self, tests): Env.__init__(self, tests) # # Get commands for a given job from the determinator file # def get_commands(self, test): status, out = self.GetOutput( "build_tools/rocksdb-lego-determinator %s" % test, ".") return status, out # # Run a specific CI job # def run_test(self, test): self.log.caption("Running test %s locally" % test) # get commands for the CI job determinator status, cmds = self.get_commands(test) if status != 0: self.log.error("Error getting commands for test %s" % test) return False # Parse the JSON to extract the commands to run cmds = re.findall("'shell':'([^\']*)'", cmds) if len(cmds) == 0: self.log.log("No commands found") return False # Run commands for cmd in cmds: # Replace J=<..> with the local environment variable if "J" in os.environ: cmd = cmd.replace("J=1", "J=%s" % os.environ["J"]) cmd = cmd.replace("make ", "make -j%s " % os.environ["J"]) # Run the command status = self.shell(cmd, ".") if status != 0: self.log.error("Error running command %s for test %s" % (cmd, test)) return False return True # # Run specified CI jobs # def run_tests(self): if not self.tests: self.log.error("Invalid args. Please provide tests") return False self.print_separator() self.print_row("TEST", "RESULT") self.print_separator() for test in self.tests: start_time = time.time() self.print_test(test) result = self.run_test(test) elapsed_min = (time.time() - start_time) / 60 if not result: self.log.error("Error running test %s" % test) self.print_result("FAIL (%dm)" % elapsed_min) return False self.print_result("PASS (%dm)" % elapsed_min) self.print_separator() return True # # Print a line # def print_separator(self): print("".ljust(60, "-")) # # Print two colums # def print_row(self, c0, c1): print("%s%s" % (c0.ljust(40), c1.ljust(20))) def print_test(self, test): print(test.ljust(40), end="") sys.stdout.flush() def print_result(self, result): print(result.ljust(20)) # # Main # parser = argparse.ArgumentParser(description='RocksDB pre-commit checker.') # parser.add_argument('test', nargs='+', help='CI test(s) to run. e.g: unit punit asan tsan') print("Please follow log %s" % Log.LOG_FILE) args = parser.parse_args() checker = PreCommitChecker(args.test) if not checker.run_tests(): print("Error running tests. Please check log file %s" % Log.LOG_FILE) sys.exit(1) sys.exit(0) rocksdb-4.5.1/build_tools/regression_build_test.sh000077500000000000000000000272401267530634000224370ustar00rootroot00000000000000#!/bin/bash set -e NUM=10000000 if [ $# -eq 1 ];then DATA_DIR=$1 elif [ $# -eq 2 ];then DATA_DIR=$1 STAT_FILE=$2 fi # On the production build servers, set data and stat # files/directories not in /tmp or else the tempdir cleaning # scripts will make you very unhappy. DATA_DIR=${DATA_DIR:-$(mktemp -t -d rocksdb_XXXX)} STAT_FILE=${STAT_FILE:-$(mktemp -t -u rocksdb_test_stats_XXXX)} function cleanup { rm -rf $DATA_DIR rm -f $STAT_FILE.fillseq rm -f $STAT_FILE.readrandom rm -f $STAT_FILE.overwrite rm -f $STAT_FILE.memtablefillreadrandom } trap cleanup EXIT if [ -z $GIT_BRANCH ]; then git_br=`git rev-parse --abbrev-ref HEAD` else git_br=$(basename $GIT_BRANCH) fi if [ $git_br == "master" ]; then git_br="" else git_br="."$git_br fi make release # measure fillseq + fill up the DB for overwrite benchmark ./db_bench \ --benchmarks=fillseq \ --db=$DATA_DIR \ --use_existing_db=0 \ --bloom_bits=10 \ --num=$NUM \ --writes=$NUM \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 > ${STAT_FILE}.fillseq # measure overwrite performance ./db_bench \ --benchmarks=overwrite \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$NUM \ --writes=$((NUM / 10)) \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=8 > ${STAT_FILE}.overwrite # fill up the db for readrandom benchmark (1GB total size) ./db_bench \ --benchmarks=fillseq \ --db=$DATA_DIR \ --use_existing_db=0 \ --bloom_bits=10 \ --num=$NUM \ --writes=$NUM \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=1 > /dev/null # measure readrandom with 6GB block cache ./db_bench \ --benchmarks=readrandom \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$NUM \ --reads=$((NUM / 5)) \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > ${STAT_FILE}.readrandom # measure readrandom with 6GB block cache and tailing iterator ./db_bench \ --benchmarks=readrandom \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$NUM \ --reads=$((NUM / 5)) \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --use_tailing_iterator=1 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > ${STAT_FILE}.readrandomtailing # measure readrandom with 100MB block cache ./db_bench \ --benchmarks=readrandom \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$NUM \ --reads=$((NUM / 5)) \ --cache_size=104857600 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > ${STAT_FILE}.readrandomsmallblockcache # measure readrandom with 8k data in memtable ./db_bench \ --benchmarks=overwrite,readrandom \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$NUM \ --reads=$((NUM / 5)) \ --writes=512 \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --write_buffer_size=1000000000 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > ${STAT_FILE}.readrandom_mem_sst # fill up the db for readrandom benchmark with filluniquerandom (1GB total size) ./db_bench \ --benchmarks=filluniquerandom \ --db=$DATA_DIR \ --use_existing_db=0 \ --bloom_bits=10 \ --num=$((NUM / 4)) \ --writes=$((NUM / 4)) \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=1 > /dev/null # dummy test just to compact the data ./db_bench \ --benchmarks=readrandom \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$((NUM / 1000)) \ --reads=$((NUM / 1000)) \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > /dev/null # measure readrandom after load with filluniquerandom with 6GB block cache ./db_bench \ --benchmarks=readrandom \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$((NUM / 4)) \ --reads=$((NUM / 4)) \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --disable_auto_compactions=1 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > ${STAT_FILE}.readrandom_filluniquerandom # measure readwhilewriting after load with filluniquerandom with 6GB block cache ./db_bench \ --benchmarks=readwhilewriting \ --db=$DATA_DIR \ --use_existing_db=1 \ --bloom_bits=10 \ --num=$((NUM / 4)) \ --reads=$((NUM / 4)) \ --benchmark_write_rate_limit=$(( 110 * 1024 )) \ --write_buffer_size=100000000 \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=16 > ${STAT_FILE}.readwhilewriting # measure memtable performance -- none of the data gets flushed to disk ./db_bench \ --benchmarks=fillrandom,readrandom, \ --db=$DATA_DIR \ --use_existing_db=0 \ --num=$((NUM / 10)) \ --reads=$NUM \ --cache_size=6442450944 \ --cache_numshardbits=6 \ --table_cache_numshardbits=4 \ --write_buffer_size=1000000000 \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --value_size=10 \ --threads=16 > ${STAT_FILE}.memtablefillreadrandom common_in_mem_args="--db=/dev/shm/rocksdb \ --num_levels=6 \ --key_size=20 \ --prefix_size=12 \ --keys_per_prefix=10 \ --value_size=100 \ --compression_type=none \ --compression_ratio=1 \ --hard_rate_limit=2 \ --write_buffer_size=134217728 \ --max_write_buffer_number=4 \ --level0_file_num_compaction_trigger=8 \ --level0_slowdown_writes_trigger=16 \ --level0_stop_writes_trigger=24 \ --target_file_size_base=134217728 \ --max_bytes_for_level_base=1073741824 \ --disable_wal=0 \ --wal_dir=/dev/shm/rocksdb \ --sync=0 \ --disable_data_sync=1 \ --verify_checksum=1 \ --delete_obsolete_files_period_micros=314572800 \ --max_grandparent_overlap_factor=10 \ --use_plain_table=1 \ --open_files=-1 \ --mmap_read=1 \ --mmap_write=0 \ --memtablerep=prefix_hash \ --bloom_bits=10 \ --bloom_locality=1 \ --perf_level=0" # prepare a in-memory DB with 50M keys, total DB size is ~6G ./db_bench \ $common_in_mem_args \ --statistics=0 \ --max_background_compactions=16 \ --max_background_flushes=16 \ --benchmarks=filluniquerandom \ --use_existing_db=0 \ --num=52428800 \ --threads=1 > /dev/null # Readwhilewriting ./db_bench \ $common_in_mem_args \ --statistics=1 \ --max_background_compactions=4 \ --max_background_flushes=0 \ --benchmarks=readwhilewriting\ --use_existing_db=1 \ --duration=600 \ --threads=32 \ --benchmark_write_rate_limit=9502720 > ${STAT_FILE}.readwhilewriting_in_ram # Seekrandomwhilewriting ./db_bench \ $common_in_mem_args \ --statistics=1 \ --max_background_compactions=4 \ --max_background_flushes=0 \ --benchmarks=seekrandomwhilewriting \ --use_existing_db=1 \ --use_tailing_iterator=1 \ --duration=600 \ --threads=32 \ --benchmark_write_rate_limit=9502720 > ${STAT_FILE}.seekwhilewriting_in_ram # measure fillseq with bunch of column families ./db_bench \ --benchmarks=fillseq \ --num_column_families=500 \ --write_buffer_size=1048576 \ --db=$DATA_DIR \ --use_existing_db=0 \ --num=$NUM \ --writes=$NUM \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 > ${STAT_FILE}.fillseq_lots_column_families # measure overwrite performance with bunch of column families ./db_bench \ --benchmarks=overwrite \ --num_column_families=500 \ --write_buffer_size=1048576 \ --db=$DATA_DIR \ --use_existing_db=1 \ --num=$NUM \ --writes=$((NUM / 10)) \ --open_files=55000 \ --statistics=1 \ --histogram=1 \ --disable_data_sync=1 \ --disable_wal=1 \ --sync=0 \ --threads=8 > ${STAT_FILE}.overwrite_lots_column_families # send data to ods function send_to_ods { key="$1" value="$2" if [ -z $JENKINS_HOME ]; then # running on devbox, just print out the values echo $1 $2 return fi if [ -z "$value" ];then echo >&2 "ERROR: Key $key doesn't have a value." return fi curl -s "https://www.intern.facebook.com/intern/agent/ods_set.php?entity=rocksdb_build$git_br&key=$key&value=$value" \ --connect-timeout 60 } function send_benchmark_to_ods { bench="$1" bench_key="$2" file="$3" QPS=$(grep $bench $file | awk '{print $5}') P50_MICROS=$(grep $bench $file -A 6 | grep "Percentiles" | awk '{print $3}' ) P75_MICROS=$(grep $bench $file -A 6 | grep "Percentiles" | awk '{print $5}' ) P99_MICROS=$(grep $bench $file -A 6 | grep "Percentiles" | awk '{print $7}' ) send_to_ods rocksdb.build.$bench_key.qps $QPS send_to_ods rocksdb.build.$bench_key.p50_micros $P50_MICROS send_to_ods rocksdb.build.$bench_key.p75_micros $P75_MICROS send_to_ods rocksdb.build.$bench_key.p99_micros $P99_MICROS } send_benchmark_to_ods overwrite overwrite $STAT_FILE.overwrite send_benchmark_to_ods fillseq fillseq $STAT_FILE.fillseq send_benchmark_to_ods readrandom readrandom $STAT_FILE.readrandom send_benchmark_to_ods readrandom readrandom_tailing $STAT_FILE.readrandomtailing send_benchmark_to_ods readrandom readrandom_smallblockcache $STAT_FILE.readrandomsmallblockcache send_benchmark_to_ods readrandom readrandom_memtable_sst $STAT_FILE.readrandom_mem_sst send_benchmark_to_ods readrandom readrandom_fillunique_random $STAT_FILE.readrandom_filluniquerandom send_benchmark_to_ods fillrandom memtablefillrandom $STAT_FILE.memtablefillreadrandom send_benchmark_to_ods readrandom memtablereadrandom $STAT_FILE.memtablefillreadrandom send_benchmark_to_ods readwhilewriting readwhilewriting $STAT_FILE.readwhilewriting send_benchmark_to_ods readwhilewriting readwhilewriting_in_ram ${STAT_FILE}.readwhilewriting_in_ram send_benchmark_to_ods seekrandomwhilewriting seekwhilewriting_in_ram ${STAT_FILE}.seekwhilewriting_in_ram send_benchmark_to_ods fillseq fillseq_lots_column_families ${STAT_FILE}.fillseq_lots_column_families send_benchmark_to_ods overwrite overwrite_lots_column_families ${STAT_FILE}.overwrite_lots_column_families rocksdb-4.5.1/build_tools/rocksdb-lego-determinator000077500000000000000000000363661267530634000225070ustar00rootroot00000000000000#!/bin/bash # This script is executed by Sandcastle # to determine next steps to run # Usage: # EMAIL= ONCALL= TRIGGER= SUBSCRIBER= rocks_ci.py # # Input Value # ------------------------------------------------------------------------- # EMAIL Email address to report on trigger conditions # ONCAL Email address to raise a task on failure # TRIGGER Trigger conditions for email. Valid values are fail, warn, all # SUBSCRIBER Email addresss to add as subscriber for task # # # Report configuration # REPORT_EMAIL= if [ ! -z $EMAIL ]; then if [ -z $TRIGGER ]; then TRIGGER="fail" fi REPORT_EMAIL=" { 'type':'email', 'triggers': [ '$TRIGGER' ], 'emails':['$EMAIL'] }," fi CREATE_TASK= if [ ! -z $ONCALL ]; then CREATE_TASK=" { 'type':'task', 'triggers':[ 'fail' ], 'priority':0, 'subscribers':[ '$SUBSCRIBER' ], 'tags':[ 'rocksdb', 'ci' ], }," fi REPORT= if [[ ! -z $REPORT_EMAIL || ! -z $CREATE_TASK ]]; then REPORT="'report': [ $REPORT_EMAIL $CREATE_TASK ]" fi # # Helper variables # CLEANUP_ENV=" { 'name':'Cleanup environment', 'shell':'rm -rf /dev/shm/rocksdb && mkdir /dev/shm/rocksdb && make clean', 'user':'root' }" DEBUG="OPT=-g" SHM="TEST_TMPDIR=/dev/shm/rocksdb" GCC_481="ROCKSDB_FBCODE_BUILD_WITH_481=1" ASAN="COMPILE_WITH_ASAN=1" CLANG="USE_CLANG=1" LITE="OPT=-DROCKSDB_LITE" TSAN="COMPILE_WITH_TSAN=1" DISABLE_JEMALLOC="DISABLE_JEMALLOC=1" PARSER="'parser':'egrep \'Failure|^#|Abort|Expected|Actual|GoogleTestFailure|^==\''" ARTIFACTS=" 'artifacts': [ { 'name':'database', 'paths':[ '/dev/shm/rocksdb' ], } ]" # # A mechanism to disable tests temporarily # DISABLE_COMMANDS="[ { 'name':'Disable test', 'oncall':'$ONCALL', 'steps': [ { 'name':'Job disabled. Please contact test owner', 'shell':'exit 1', 'user':'root' }, ], } ]" # # RocksDB unit test in parallel # Currently we always have noise in our parallel runs. This job is to help # manage the noise # PARALLEL_UNIT_TEST_COMMANDS="[ { 'name':'Rocksdb Parallel Unit Test', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build and test RocksDB debug version', 'shell':'$DEBUG make -j$(nproc) all && $SHM make check > /dev/null 2>&1 || cat t/log-*', 'user':'root', $PARSER }, $CLEANUP_ENV, { 'name':'Build and test RocksDB debug version under gcc-4.8.1', 'shell':'$GCC_481 $DEBUG make -j$(nproc) all && $SHM make check > /dev/null 2>&1 || cat t/log-*', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB unit test # UNIT_TEST_COMMANDS="[ { 'name':'Rocksdb Unit Test', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build and test RocksDB debug version', 'shell':'$SHM $DEBUG make J=1 check', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB unit test on gcc-4.8.1 # UNIT_TEST_COMMANDS_481="[ { 'name':'Rocksdb Unit Test on GCC 4.8.1', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build and test RocksDB debug version', 'shell':'$SHM $GCC_481 $DEBUG make J=1 check', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB unit test with CLANG # CLANG_UNIT_TEST_COMMANDS="[ { 'name':'Rocksdb Unit Test', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build and test RocksDB debug', 'shell':'$CLANG $SHM $DEBUG make J=1 check', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB analyze # CLANG_ANALYZE_COMMANDS="[ { 'name':'Rocksdb analyze', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'RocksDB build and analyze', 'shell':'$CLANG $SHM $DEBUG make J=1 analyze', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB code coverage # CODE_COV_COMMANDS="[ { 'name':'Rocksdb Unit Test Code Coverage', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build, test and collect code coverage info', 'shell':'$SHM $DEBUG make J=1 coverage', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB unity # UNITY_COMMANDS="[ { 'name':'Rocksdb Unity', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build, test unity test', 'shell':'$SHM $DEBUG V=1 make J=1 unity_test', 'user':'root', $PARSER }, ], $REPORT } ]" # # Build RocksDB lite # LITE_BUILD_COMMANDS="[ { 'name':'Rocksdb Lite build', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build RocksDB debug version', 'shell':'$LITE $DEBUG make J=1 static_lib', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB lite tests # LITE_UNIT_TEST_COMMANDS="[ { 'name':'Rocksdb Lite Unit Test', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build RocksDB debug version', 'shell':'$SHM $LITE $DEBUG make J=1 check', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB stress/crash test # STRESS_CRASH_TEST_COMMANDS="[ { 'name':'Rocksdb Stress/Crash Test', 'oncall':'$ONCALL', 'timeout': 86400, 'steps': [ $CLEANUP_ENV, { 'name':'Build and run RocksDB debug stress tests', 'shell':'$SHM $DEBUG make J=1 db_stress', 'user':'root', $PARSER }, { 'name':'Build and run RocksDB debug crash tests', 'timeout': 86400, 'shell':'$SHM $DEBUG make J=1 crash_test', 'user':'root', $PARSER } ], $ARTIFACTS, $REPORT } ]" # RocksDB write stress test. # We run on disk device on purpose (i.e. no $SHM) # because we want to add some randomness to fsync commands WRITE_STRESS_COMMANDS="[ { 'name':'Rocksdb Write Stress Test', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Build and run RocksDB write stress tests', 'shell':'make write_stress && python tools/write_stress_runner.py --runtime_sec=3600 --db=/tmp/rocksdb_write_stress', 'user':'root', $PARSER } ], 'artifacts': [{'name': 'database', 'paths': ['/tmp/rocksdb_write_stress']}], $REPORT } ]" # # RocksDB test under address sanitizer # ASAN_TEST_COMMANDS="[ { 'name':'Rocksdb Unit Test under ASAN', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Test RocksDB debug under ASAN', 'shell':'set -o pipefail && $SHM $ASAN $DEBUG make J=1 asan_check |& /usr/facebook/ops/scripts/asan_symbolize.py -d', 'user':'root', $PARSER } ], $REPORT } ]" # # RocksDB crash testing under address sanitizer # ASAN_CRASH_TEST_COMMANDS="[ { 'name':'Rocksdb crash test under ASAN', 'oncall':'$ONCALL', 'timeout': 86400, 'steps': [ $CLEANUP_ENV, { 'name':'Build and run RocksDB debug asan_crash_test', 'timeout': 86400, 'shell':'$SHM $DEBUG make J=1 asan_crash_test', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB unit test under valgrind # VALGRIND_TEST_COMMANDS="[ { 'name':'Rocksdb Unit Test under valgrind', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Run RocksDB debug unit tests', 'shell':'$DISABLE_JEMALLOC $SHM $DEBUG make valgrind_check', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB test under TSAN # TSAN_UNIT_TEST_COMMANDS="[ { 'name':'Rocksdb Unit Test under TSAN', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Run RocksDB debug unit test', 'shell':'set -o pipefail && $SHM $DEBUG $TSAN make J=1 check', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB crash test under TSAN # TSAN_CRASH_TEST_COMMANDS="[ { 'name':'Rocksdb Crash Test under TSAN', 'oncall':'$ONCALL', 'timeout': 86400, 'steps': [ $CLEANUP_ENV, { 'name':'Compile and run', 'timeout': 86400, 'shell':'set -o pipefail && $SHM $DEBUG $TSAN make J=1 crash_test', 'user':'root', $PARSER }, ], $REPORT } ]" TSAN_CRASH_TEST_COMMANDS=$DISABLE_COMMANDS # # RocksDB format compatible # run_format_compatible() { export TEST_TMPDIR=/dev/shm/rocksdb rm -rf /dev/shm/rocksdb mkdir /dev/shm/rocksdb echo ' if [ -e "build_tools/build_detect_platform" ] then sed "s/tcmalloc/nothingnothingnothing/g" build_tools/build_detect_platform > $TEST_TMPDIR/temp_build_file rm -rf build_tools/build_detect_platform cp $TEST_TMPDIR/temp_build_file build_tools/build_detect_platform chmod +x build_tools/build_detect_platform fi if [ -e "build_detect_platform" ] then sed "s/tcmalloc/nothingnothingnothing/g" build_detect_platform > $TEST_TMPDIR/temp_build_file rm -rf build_detect_platform cp $TEST_TMPDIR/temp_build_file build_detect_platform chmod +x build_detect_platform fi make ldb -j32 if [ -e "build_detect_platform" ] then git checkout -- build_detect_platform fi if [ -e "build_tools/build_detect_platform" ] then git checkout -- build_tools/build_detect_platform fi ' > temp_build_ldb.sh sed "s/make ldb -j32/source temp_build_ldb.sh/g" tools/check_format_compatible.sh > tools/temp_check_format_compatible.sh chmod +x tools/temp_check_format_compatible.sh tools/temp_check_format_compatible.sh } FORMAT_COMPATIBLE_COMMANDS="[ { 'name':'Rocksdb Format Compatible tests', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Run RocksDB debug unit test', 'shell':'build_tools/rocksdb-lego-determinator run_format_compatible', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB no compression # run_no_compression() { export TEST_TMPDIR=/dev/shm/rocksdb rm -rf /dev/shm/rocksdb mkdir /dev/shm/rocksdb make clean cat build_tools/fbcode_config.sh | grep -iv dzlib | grep -iv dlz4 | grep -iv dsnappy | grep -iv dbzip2 > .tmp.fbcode_config.sh mv .tmp.fbcode_config.sh build_tools/fbcode_config.sh cat Makefile | grep -v tools/ldb_test.py > .tmp.Makefile mv .tmp.Makefile Makefile make $DEBUG J=1 check } NO_COMPRESSION_COMMANDS="[ { 'name':'Rocksdb No Compression tests', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Run RocksDB debug unit test', 'shell':'build_tools/rocksdb-lego-determinator run_no_compression', 'user':'root', $PARSER }, ], $REPORT } ]" # # RocksDB regression # run_regression() { time -v bash -vx ./build_tools/regression_build_test.sh $(mktemp -d $WORKSPACE/leveldb.XXXX) $(mktemp leveldb_test_stats.XXXX) # ======= report size to ODS ======== # parameters: $1 -- key, $2 -- value function send_size_to_ods { curl -s "https://www.intern.facebook.com/intern/agent/ods_set.php?entity=rocksdb_build&key=rocksdb.build_size.$1&value=$2" \ --connect-timeout 60 } # === normal build === make clean make -j$(nproc) static_lib send_size_to_ods static_lib $(stat --printf="%s" librocksdb.a) strip librocksdb.a send_size_to_ods static_lib_stripped $(stat --printf="%s" librocksdb.a) make -j$(nproc) shared_lib send_size_to_ods shared_lib $(stat --printf="%s" `readlink -f librocksdb.so`) strip `readlink -f librocksdb.so` send_size_to_ods shared_lib_stripped $(stat --printf="%s" `readlink -f librocksdb.so`) # === lite build === make clean OPT=-DROCKSDB_LITE make -j$(nproc) static_lib send_size_to_ods static_lib_lite $(stat --printf="%s" librocksdb.a) strip librocksdb.a send_size_to_ods static_lib_lite_stripped $(stat --printf="%s" librocksdb.a) OPT=-DROCKSDB_LITE make -j$(nproc) shared_lib send_size_to_ods shared_lib_lite $(stat --printf="%s" `readlink -f librocksdb.so`) strip `readlink -f librocksdb.so` send_size_to_ods shared_lib_lite_stripped $(stat --printf="%s" `readlink -f librocksdb.so`) } REGRESSION_COMMANDS="[ { 'name':'Rocksdb regression commands', 'oncall':'$ONCALL', 'steps': [ $CLEANUP_ENV, { 'name':'Make and run script', 'shell':'build_tools/rocksdb-lego-determinator run_regression', 'user':'root', $PARSER }, ], $REPORT } ]" case $1 in punit) echo $PARALLEL_UNIT_TEST_COMMANDS ;; unit) echo $UNIT_TEST_COMMANDS ;; unit_481) echo $UNIT_TEST_COMMANDS_481 ;; clang_unit) echo $CLANG_UNIT_TEST_COMMANDS ;; clang_analyze) echo $CLANG_ANALYZE_COMMANDS ;; code_cov) echo $CODE_COV_COMMANDS ;; unity) echo $UNITY_COMMANDS ;; lite) echo $LITE_BUILD_COMMANDS ;; lite_test) echo $LITE_UNIT_TEST_COMMANDS ;; stress_crash) echo $STRESS_CRASH_TEST_COMMANDS ;; write_stress) echo $WRITE_STRESS_COMMANDS ;; asan) echo $ASAN_TEST_COMMANDS ;; asan_crash) echo $ASAN_CRASH_TEST_COMMANDS ;; valgrind) echo $VALGRIND_TEST_COMMANDS ;; tsan) echo $TSAN_UNIT_TEST_COMMANDS ;; tsan_crash) echo $TSAN_CRASH_TEST_COMMANDS ;; format_compatible) echo $FORMAT_COMPATIBLE_COMMANDS ;; run_format_compatible) run_format_compatible ;; no_compression) echo $NO_COMPRESSION_COMMANDS ;; run_no_compression) run_no_compression ;; regression) echo $REGRESSION_COMMANDS ;; run_regression) run_regression ;; *) echo "Invalid determinator command" ;; esac rocksdb-4.5.1/build_tools/run_ci_db_test.ps1000066400000000000000000000221351267530634000211100ustar00rootroot00000000000000# This script enables you running RocksDB tests by running # All the tests in parallel and utilizing all the cores # For db_test the script first lists and parses the tests # and then fires them up in parallel using async PS Job functionality # Run the script from the enlistment Param( [switch]$EnableJE = $false, # Use je executable [switch]$EnableRerun = $false, # Rerun failed tests sequentially at the end [string]$WorkFolder = "", # Direct tests to use that folder [int]$Limit = -1, # -1 means run all otherwise limit for testing purposes [string]$Exclude = "", # Expect a comma separated list, no spaces [string]$Run = "db_test", # Run db_test|tests|testname1,testname2... # Number of async tasks that would run concurrently. Recommend a number below 64. # However, CPU utlization really depends on the storage media. Recommend ram based disk. # a value of 1 will run everything serially [int]$Concurrency = 16 ) # Folders and commands must be fullpath to run assuming # the current folder is at the root of the git enlistment Get-Date # If running under Appveyor assume that root [string]$Appveyor = $Env:APPVEYOR_BUILD_FOLDER if($Appveyor -ne "") { $RootFolder = $Appveyor } else { $RootFolder = $PSScriptRoot -replace '\\build_tools', '' } $LogFolder = -Join($RootFolder, "\db_logs\") $BinariesFolder = -Join($RootFolder, "\build\Debug\") if($WorkFolder -eq "") { # If TEST_TMPDIR is set use it [string]$var = $Env:TEST_TMPDIR if($var -eq "") { $WorkFolder = -Join($RootFolder, "\db_tests\") $Env:TEST_TMPDIR = $WorkFolder } else { $WorkFolder = $var } } else { # Override from a command line $Env:TEST_TMPDIR = $WorkFolder } # Use JEMALLOC executables if($EnableJE) { $db_test = -Join ($BinariesFolder, "db_test_je.exe") } else { $db_test = -Join ($BinariesFolder, "db_test.exe") } Write-Output "Root: $RootFolder, WorkFolder: $WorkFolder" Write-Output "Binaries: $BinariesFolder db_test: $db_test" #Exclusions that we do not want to run $ExcludeTests = New-Object System.Collections.Generic.HashSet[string] if($Exclude -ne "") { Write-Host "Exclude: $Exclude" $l = $Exclude -split ' ' ForEach($t in $l) { $ExcludeTests.Add($t) | Out-Null } } # Create test directories in the current folder md -Path $WorkFolder -ErrorAction Ignore | Out-Null md -Path $LogFolder -ErrorAction Ignore | Out-Null # Extract the names of its tests by running db_test with --gtest_list_tests. # This filter removes the "#"-introduced comments, and expands to # fully-qualified names by changing input like this: # # DBTest. # Empty # WriteEmptyBatch # MultiThreaded/MultiThreadedDBTest. # MultiThreaded/0 # GetParam() = 0 # MultiThreaded/1 # GetParam() = 1 # # into this: # # DBTest.Empty # DBTest.WriteEmptyBatch # MultiThreaded/MultiThreadedDBTest.MultiThreaded/0 # MultiThreaded/MultiThreadedDBTest.MultiThreaded/1 # Output into the parameter in a form TestName -> Log File Name function Normalize-DbTests($HashTable) { $Tests = @() # Run db_test to get a list of tests and store it into $a array &$db_test --gtest_list_tests | tee -Variable Tests | Out-Null # Current group $Group="" ForEach( $l in $Tests) { # Trailing dot is a test group if( $l -match "\.$") { $Group = $l } else { # Otherwise it is a test name, remove leading space $test = $l -replace '^\s+','' # remove trailing comment if any and create a log name $test = $test -replace '\s+\#.*','' $test = "$Group$test" if($ExcludeTests.Contains($test)) { continue } $test_log = $test -replace '[\./]','_' $test_log += ".log" $log_path = -join ($LogFolder, $test_log) # Add to a hashtable $HashTable.Add($test, $log_path); } } } # The function removes trailing .exe siffix if any, # creates a name for the log file function MakeAndAdd([string]$token, $HashTable) { $test_name = $token -replace '.exe$', '' $log_name = -join ($test_name, ".log") $log_path = -join ($LogFolder, $log_name) if(!$ExcludeTests.Contains($test_name)) { $HashTable.Add($test_name, $log_path) } else { Write-Warning "Test $test_name is excluded" } } # The function scans build\Debug folder to discover # Test executables. It then populates a table with # Test executable name -> Log file function Discover-TestBinaries([string]$Pattern, $HashTable) { $Exclusions = @("db_test*", "db_sanity_test*") $p = -join ($BinariesFolder, $pattern) Write-Host "Path: $p" dir -Path $p -Exclude $Exclusions | ForEach-Object { MakeAndAdd -token ($_.Name) -HashTable $HashTable } } $TestsToRun = [ordered]@{} if($Run -ceq "db_test") { Normalize-DbTests -HashTable $TestsToRun } elseif($Run -ceq "tests") { if($EnableJE) { $pattern = "*_test_je.exe" } else { $pattern = "*_test.exe" } Discover-TestBinaries -Pattern $pattern -HashTable $TestsToRun } else { $test_list = $Run -split ' ' ForEach($t in $test_list) { MakeAndAdd -token $t -HashTable $TestsToRun } } $NumTestsToStart = $TestsToRun.Count if($Limit -ge 0 -and $NumTestsToStart -gt $Limit) { $NumTestsToStart = $Limit } Write-Host "Attempting to start: $NumTestsToStart tests" # Invoke a test with a filter and redirect all output $InvokeTestCase = { param($exe, $test, $log); &$exe --gtest_filter=$test > $log 2>&1 } # Invoke all tests and redirect output $InvokeTestAsync = { param($exe, $log) &$exe > $log 2>&1 } # Hash that contains tests to rerun if any failed # Those tests will be rerun sequentially $Rerun = [ordered]@{} # Test limiting factor here $count = 0 # Overall status [bool]$success = $true; function RunJobs($TestToLog, [int]$ConcurrencyVal, [bool]$AddForRerun) { # Array to wait for any of the running jobs $jobs = @() # Hash JobToLog $JobToLog = @{} # Wait for all to finish and get the results while(($JobToLog.Count -gt 0) -or ($TestToLog.Count -gt 0)) { # Make sure we have maximum concurrent jobs running if anything # and the $Limit either not set or allows to proceed while(($JobToLog.Count -lt $ConcurrencyVal) -and (($TestToLog.Count -gt 0) -and (($Limit -lt 0) -or ($count -lt $Limit)))) { # We only need the first key foreach($key in $TestToLog.keys) { $k = $key break } Write-Host "Starting $k" $log_path = ($TestToLog.$k) if($Run -ceq "db_test") { $job = Start-Job -Name $k -ScriptBlock $InvokeTestCase -ArgumentList @($db_test,$k,$log_path) } else { [string]$Exe = -Join ($BinariesFolder, $k) $job = Start-Job -Name $k -ScriptBlock $InvokeTestAsync -ArgumentList @($exe,$log_path) } $JobToLog.Add($job, $log_path) $TestToLog.Remove($k) ++$count } if($JobToLog.Count -lt 1) { break } $jobs = @() foreach($k in $JobToLog.Keys) { $jobs += $k } $completed = Wait-Job -Job $jobs -Any $log = $JobToLog[$completed] $JobToLog.Remove($completed) $message = -join @($completed.Name, " State: ", ($completed.State)) $log_content = @(Get-Content $log) if($completed.State -ne "Completed") { $success = $false Write-Warning $message $log_content | Write-Warning } else { # Scan the log. If we find PASSED and no occurrence of FAILED # then it is a success [bool]$pass_found = $false ForEach($l in $log_content) { if(($l -match "^\[\s+FAILED") -or ($l -match "Assertion failed:")) { $pass_found = $false break } if(($l -match "^\[\s+PASSED") -or ($l -match " : PASSED$") -or ($l -match "^PASS$") -or # Special c_test case ($l -match "Passed all tests!") ) { $pass_found = $true } } if(!$pass_found) { $success = $false; Write-Warning $message $log_content | Write-Warning if($AddForRerun) { $Rerun.Add($completed.Name, $log) } } else { Write-Host $message } } # Remove cached job info from the system # Should be no output Receive-Job -Job $completed | Out-Null } } RunJobs -TestToLog $TestsToRun -ConcurrencyVal $Concurrency -AddForRerun $EnableRerun if($Rerun.Count -gt 0) { Write-Host "Rerunning " ($Rerun.Count) " tests sequentially" $success = $true $count = 0 RunJobs -TestToLog $Rerun -ConcurrencyVal 1 -AddForRerun $false } Get-Date if(!$success) { # This does not succeed killing off jobs quick # So we simply exit # Remove-Job -Job $jobs -Force # indicate failure using this exit code exit 12345 } rocksdb-4.5.1/build_tools/update_dependencies.sh000077500000000000000000000067221267530634000220330ustar00rootroot00000000000000#!/bin/sh # # Update dependencies.sh file with the latest avaliable versions BASEDIR=$(dirname $0) OUTPUT="" function log_variable() { echo "$1=${!1}" >> "$OUTPUT" } TP2_LATEST="/mnt/vol/engshare/fbcode/third-party2" ## $1 => lib name ## $2 => lib version (if not provided, will try to pick latest) ## $3 => platform (if not provided, will try to pick latest gcc) ## ## get_lib_base will set a variable named ${LIB_NAME}_BASE to the lib location function get_lib_base() { local lib_name=$1 local lib_version=$2 local lib_platform=$3 local result="$TP2_LATEST/$lib_name/" # Lib Version if [ -z "$lib_version" ] || [ "$lib_version" = "LATEST" ]; then # version is not provided, use latest result=`ls -dr1v $result/*/ | head -n1` else result="$result/$lib_version/" fi # Lib Platform if [ -z "$lib_platform" ]; then # platform is not provided, use latest gcc result=`ls -dr1v $result/gcc-*[^fb]/ | head -n1` else result="$result/$lib_platform/" fi result=`ls -1d $result/*/ | head -n1` # lib_name => LIB_NAME_BASE local __res_var=${lib_name^^}"_BASE" __res_var=`echo $__res_var | tr - _` # LIB_NAME_BASE=$result eval $__res_var=`readlink -f $result` log_variable $__res_var } ########################################################### # 4.9.x dependencies # ########################################################### OUTPUT="$BASEDIR/dependencies.sh" rm -f "$OUTPUT" touch "$OUTPUT" echo "Writing dependencies to $OUTPUT" # Compilers locations GCC_BASE=`ls -d1 $TP2_LATEST/gcc/4.9.x/centos6-native/*/ | head -n1` CLANG_BASE=`ls -d1 $TP2_LATEST/clang/3.7.1/centos6-native/*/ | head -n1` log_variable GCC_BASE log_variable CLANG_BASE # Libraries locations get_lib_base libgcc 4.9.x get_lib_base glibc 2.20 get_lib_base snappy LATEST get_lib_base zlib LATEST get_lib_base bzip2 LATEST get_lib_base lz4 LATEST get_lib_base zstd LATEST get_lib_base gflags LATEST get_lib_base jemalloc LATEST get_lib_base numa LATEST get_lib_base libunwind LATEST get_lib_base kernel-headers LATEST get_lib_base binutils LATEST centos6-native get_lib_base valgrind LATEST git diff $OUTPUT ########################################################### # 4.8.1 dependencies # ########################################################### OUTPUT="$BASEDIR/dependencies_4.8.1.sh" rm -f "$OUTPUT" touch "$OUTPUT" echo "Writing 4.8.1 dependencies to $OUTPUT" # Compilers locations GCC_BASE=`ls -d1 $TP2_LATEST/gcc/4.8.1/centos6-native/*/ | head -n1` CLANG_BASE=`ls -d1 $TP2_LATEST/clang/3.7.1/centos6-native/*/ | head -n1` log_variable GCC_BASE log_variable CLANG_BASE # Libraries locations get_lib_base libgcc 4.8.1 gcc-4.8.1-glibc-2.17 get_lib_base glibc 2.17 gcc-4.8.1-glibc-2.17 get_lib_base snappy LATEST gcc-4.8.1-glibc-2.17 get_lib_base zlib LATEST gcc-4.8.1-glibc-2.17 get_lib_base bzip2 LATEST gcc-4.8.1-glibc-2.17 get_lib_base lz4 LATEST gcc-4.8.1-glibc-2.17 get_lib_base zstd LATEST gcc-4.8.1-glibc-2.17 get_lib_base gflags LATEST gcc-4.8.1-glibc-2.17 get_lib_base jemalloc LATEST gcc-4.8.1-glibc-2.17 get_lib_base numa LATEST gcc-4.8.1-glibc-2.17 get_lib_base libunwind LATEST gcc-4.8.1-glibc-2.17 get_lib_base kernel-headers LATEST gcc-4.8.1-glibc-2.17 get_lib_base binutils LATEST centos6-native get_lib_base valgrind 3.8.1 gcc-4.8.1-glibc-2.17 git diff $OUTPUT rocksdb-4.5.1/build_tools/version.sh000077500000000000000000000006241267530634000175230ustar00rootroot00000000000000#!/bin/sh if [ "$#" = "0" ]; then echo "Usage: $0 major|minor|patch" exit 1 fi if [ "$1" = "major" ]; then cat include/rocksdb/version.h | grep MAJOR | head -n1 | awk '{print $3}' fi if [ "$1" = "minor" ]; then cat include/rocksdb/version.h | grep MINOR | head -n1 | awk '{print $3}' fi if [ "$1" = "patch" ]; then cat include/rocksdb/version.h | grep PATCH | head -n1 | awk '{print $3}' fi rocksdb-4.5.1/coverage/000077500000000000000000000000001267530634000147515ustar00rootroot00000000000000rocksdb-4.5.1/coverage/coverage_test.sh000077500000000000000000000042651267530634000201510ustar00rootroot00000000000000#!/bin/bash # Exit on error. set -e if [ -n "$USE_CLANG" ]; then echo "Error: Coverage test is supported only for gcc." exit 1 fi ROOT=".." # Fetch right version of gcov if [ -d /mnt/gvfs/third-party -a -z "$CXX" ]; then source $ROOT/build_tools/fbcode_config.sh GCOV=$GCC_BASE/bin/gcov else GCOV=$(which gcov) fi COVERAGE_DIR="$PWD/COVERAGE_REPORT" mkdir -p $COVERAGE_DIR # Find all gcno files to generate the coverage report GCNO_FILES=`find $ROOT -name "*.gcno"` $GCOV --preserve-paths --relative-only --no-output $GCNO_FILES 2>/dev/null | # Parse the raw gcov report to more human readable form. python $ROOT/coverage/parse_gcov_output.py | # Write the output to both stdout and report file. tee $COVERAGE_DIR/coverage_report_all.txt && echo -e "Generated coverage report for all files: $COVERAGE_DIR/coverage_report_all.txt\n" # TODO: we also need to get the files of the latest commits. # Get the most recently committed files. LATEST_FILES=` git show --pretty="format:" --name-only HEAD | grep -v "^$" | paste -s -d,` RECENT_REPORT=$COVERAGE_DIR/coverage_report_recent.txt echo -e "Recently updated files: $LATEST_FILES\n" > $RECENT_REPORT $GCOV --preserve-paths --relative-only --no-output $GCNO_FILES 2>/dev/null | python $ROOT/coverage/parse_gcov_output.py -interested-files $LATEST_FILES | tee -a $RECENT_REPORT && echo -e "Generated coverage report for recently updated files: $RECENT_REPORT\n" # Unless otherwise specified, we'll not generate html report by default if [ -z "$HTML" ]; then exit 0 fi # Generate the html report. If we cannot find lcov in this machine, we'll simply # skip this step. echo "Generating the html coverage report..." LCOV=$(which lcov || true 2>/dev/null) if [ -z $LCOV ] then echo "Skip: Cannot find lcov to generate the html report." exit 0 fi LCOV_VERSION=$(lcov -v | grep 1.1 || true) if [ $LCOV_VERSION ] then echo "Not supported lcov version. Expect lcov 1.1." exit 0 fi (cd $ROOT; lcov --no-external \ --capture \ --directory $PWD \ --gcov-tool $GCOV \ --output-file $COVERAGE_DIR/coverage.info) genhtml $COVERAGE_DIR/coverage.info -o $COVERAGE_DIR echo "HTML Coverage report is generated in $COVERAGE_DIR" rocksdb-4.5.1/coverage/parse_gcov_output.py000066400000000000000000000101601267530634000210710ustar00rootroot00000000000000import optparse import re import sys from optparse import OptionParser # the gcov report follows certain pattern. Each file will have two lines # of report, from which we can extract the file name, total lines and coverage # percentage. def parse_gcov_report(gcov_input): per_file_coverage = {} total_coverage = None for line in sys.stdin: line = line.strip() # --First line of the coverage report (with file name in it)? match_obj = re.match("^File '(.*)'$", line) if match_obj: # fetch the file name from the first line of the report. current_file = match_obj.group(1) continue # -- Second line of the file report (with coverage percentage) match_obj = re.match("^Lines executed:(.*)% of (.*)", line) if match_obj: coverage = float(match_obj.group(1)) lines = int(match_obj.group(2)) if current_file is not None: per_file_coverage[current_file] = (coverage, lines) current_file = None else: # If current_file is not set, we reach the last line of report, # which contains the summarized coverage percentage. total_coverage = (coverage, lines) continue # If the line's pattern doesn't fall into the above categories. We # can simply ignore them since they're either empty line or doesn't # find executable lines of the given file. current_file = None return per_file_coverage, total_coverage def get_option_parser(): usage = "Parse the gcov output and generate more human-readable code " +\ "coverage report." parser = OptionParser(usage) parser.add_option( "--interested-files", "-i", dest="filenames", help="Comma separated files names. if specified, we will display " + "the coverage report only for interested source files. " + "Otherwise we will display the coverage report for all " + "source files." ) return parser def display_file_coverage(per_file_coverage, total_coverage): # To print out auto-adjustable column, we need to know the longest # length of file names. max_file_name_length = max( len(fname) for fname in per_file_coverage.keys() ) # -- Print header # size of separator is determined by 3 column sizes: # file name, coverage percentage and lines. header_template = \ "%" + str(max_file_name_length) + "s\t%s\t%s" separator = "-" * (max_file_name_length + 10 + 20) print header_template % ("Filename", "Coverage", "Lines") print separator # -- Print body # template for printing coverage report for each file. record_template = "%" + str(max_file_name_length) + "s\t%5.2f%%\t%10d" for fname, coverage_info in per_file_coverage.items(): coverage, lines = coverage_info print record_template % (fname, coverage, lines) # -- Print footer if total_coverage: print separator print record_template % ("Total", total_coverage[0], total_coverage[1]) def report_coverage(): parser = get_option_parser() (options, args) = parser.parse_args() interested_files = set() if options.filenames is not None: interested_files = set(f.strip() for f in options.filenames.split(',')) # To make things simple, right now we only read gcov report from the input per_file_coverage, total_coverage = parse_gcov_report(sys.stdin) # Check if we need to display coverage info for interested files. if len(interested_files): per_file_coverage = dict( (fname, per_file_coverage[fname]) for fname in interested_files if fname in per_file_coverage ) # If we only interested in several files, it makes no sense to report # the total_coverage total_coverage = None if not len(per_file_coverage): print >> sys.stderr, "Cannot find coverage info for the given files." return display_file_coverage(per_file_coverage, total_coverage) if __name__ == "__main__": report_coverage() rocksdb-4.5.1/db/000077500000000000000000000000001267530634000135435ustar00rootroot00000000000000rocksdb-4.5.1/db/auto_roll_logger.cc000066400000000000000000000125461267530634000174210ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include "db/auto_roll_logger.h" #include "util/mutexlock.h" using namespace std; namespace rocksdb { // -- AutoRollLogger Status AutoRollLogger::ResetLogger() { TEST_SYNC_POINT("AutoRollLogger::ResetLogger:BeforeNewLogger"); status_ = env_->NewLogger(log_fname_, &logger_); TEST_SYNC_POINT("AutoRollLogger::ResetLogger:AfterNewLogger"); if (!status_.ok()) { return status_; } if (logger_->GetLogFileSize() == Logger::kDoNotSupportGetLogFileSize) { status_ = Status::NotSupported( "The underlying logger doesn't support GetLogFileSize()"); } if (status_.ok()) { cached_now = static_cast(env_->NowMicros() * 1e-6); ctime_ = cached_now; cached_now_access_count = 0; } return status_; } void AutoRollLogger::RollLogFile() { // This function is called when log is rotating. Two rotations // can happen quickly (NowMicro returns same value). To not overwrite // previous log file we increment by one micro second and try again. uint64_t now = env_->NowMicros(); std::string old_fname; do { old_fname = OldInfoLogFileName( dbname_, now, db_absolute_path_, db_log_dir_); now++; } while (env_->FileExists(old_fname).ok()); env_->RenameFile(log_fname_, old_fname); } string AutoRollLogger::ValistToString(const char* format, va_list args) const { // Any log messages longer than 1024 will get truncated. // The user is responsible for chopping longer messages into multi line log static const int MAXBUFFERSIZE = 1024; char buffer[MAXBUFFERSIZE]; int count = vsnprintf(buffer, MAXBUFFERSIZE, format, args); (void) count; assert(count >= 0); return buffer; } void AutoRollLogger::LogInternal(const char* format, ...) { mutex_.AssertHeld(); va_list args; va_start(args, format); logger_->Logv(format, args); va_end(args); } void AutoRollLogger::Logv(const char* format, va_list ap) { assert(GetStatus().ok()); std::shared_ptr logger; { MutexLock l(&mutex_); if ((kLogFileTimeToRoll > 0 && LogExpired()) || (kMaxLogFileSize > 0 && logger_->GetLogFileSize() >= kMaxLogFileSize)) { RollLogFile(); Status s = ResetLogger(); if (!s.ok()) { // can't really log the error if creating a new LOG file failed return; } WriteHeaderInfo(); } // pin down the current logger_ instance before releasing the mutex. logger = logger_; } // Another thread could have put a new Logger instance into logger_ by now. // However, since logger is still hanging on to the previous instance // (reference count is not zero), we don't have to worry about it being // deleted while we are accessing it. // Note that logv itself is not mutex protected to allow maximum concurrency, // as thread safety should have been handled by the underlying logger. logger->Logv(format, ap); } void AutoRollLogger::WriteHeaderInfo() { mutex_.AssertHeld(); for (auto& header : headers_) { LogInternal("%s", header.c_str()); } } void AutoRollLogger::LogHeader(const char* format, va_list args) { // header message are to be retained in memory. Since we cannot make any // assumptions about the data contained in va_list, we will retain them as // strings va_list tmp; va_copy(tmp, args); string data = ValistToString(format, tmp); va_end(tmp); MutexLock l(&mutex_); headers_.push_back(data); // Log the original message to the current log logger_->Logv(format, args); } bool AutoRollLogger::LogExpired() { if (cached_now_access_count >= call_NowMicros_every_N_records_) { cached_now = static_cast(env_->NowMicros() * 1e-6); cached_now_access_count = 0; } ++cached_now_access_count; return cached_now >= ctime_ + kLogFileTimeToRoll; } Status CreateLoggerFromOptions(const std::string& dbname, const DBOptions& options, std::shared_ptr* logger) { if (options.info_log) { *logger = options.info_log; return Status::OK(); } Env* env = options.env; std::string db_absolute_path; env->GetAbsolutePath(dbname, &db_absolute_path); std::string fname = InfoLogFileName(dbname, db_absolute_path, options.db_log_dir); env->CreateDirIfMissing(dbname); // In case it does not exist // Currently we only support roll by time-to-roll and log size if (options.log_file_time_to_roll > 0 || options.max_log_file_size > 0) { AutoRollLogger* result = new AutoRollLogger( env, dbname, options.db_log_dir, options.max_log_file_size, options.log_file_time_to_roll, options.info_log_level); Status s = result->GetStatus(); if (!s.ok()) { delete result; } else { logger->reset(result); } return s; } else { // Open a log file in the same directory as the db env->RenameFile( fname, OldInfoLogFileName(dbname, env->NowMicros(), db_absolute_path, options.db_log_dir)); auto s = env->NewLogger(fname, logger); if (logger->get() != nullptr) { (*logger)->SetInfoLogLevel(options.info_log_level); } return s; } } } // namespace rocksdb rocksdb-4.5.1/db/auto_roll_logger.h000066400000000000000000000076031267530634000172610ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Logger implementation that can be shared by all environments // where enough posix functionality is available. #pragma once #include #include #include "db/filename.h" #include "port/port.h" #include "port/util_logger.h" #include "util/sync_point.h" namespace rocksdb { // Rolls the log file by size and/or time class AutoRollLogger : public Logger { public: AutoRollLogger(Env* env, const std::string& dbname, const std::string& db_log_dir, size_t log_max_size, size_t log_file_time_to_roll, const InfoLogLevel log_level = InfoLogLevel::INFO_LEVEL) : Logger(log_level), dbname_(dbname), db_log_dir_(db_log_dir), env_(env), status_(Status::OK()), kMaxLogFileSize(log_max_size), kLogFileTimeToRoll(log_file_time_to_roll), cached_now(static_cast(env_->NowMicros() * 1e-6)), ctime_(cached_now), cached_now_access_count(0), call_NowMicros_every_N_records_(100), mutex_() { env->GetAbsolutePath(dbname, &db_absolute_path_); log_fname_ = InfoLogFileName(dbname_, db_absolute_path_, db_log_dir_); RollLogFile(); ResetLogger(); } using Logger::Logv; void Logv(const char* format, va_list ap) override; // Write a header entry to the log. All header information will be written // again every time the log rolls over. virtual void LogHeader(const char* format, va_list ap) override; // check if the logger has encountered any problem. Status GetStatus() { return status_; } size_t GetLogFileSize() const override { std::shared_ptr logger; { MutexLock l(&mutex_); // pin down the current logger_ instance before releasing the mutex. logger = logger_; } return logger->GetLogFileSize(); } void Flush() override { std::shared_ptr logger; { MutexLock l(&mutex_); // pin down the current logger_ instance before releasing the mutex. logger = logger_; } TEST_SYNC_POINT("AutoRollLogger::Flush:PinnedLogger"); if (logger) { logger->Flush(); } } virtual ~AutoRollLogger() { } void SetCallNowMicrosEveryNRecords(uint64_t call_NowMicros_every_N_records) { call_NowMicros_every_N_records_ = call_NowMicros_every_N_records; } // Expose the log file path for testing purpose std::string TEST_log_fname() const { return log_fname_; } private: bool LogExpired(); Status ResetLogger(); void RollLogFile(); // Log message to logger without rolling void LogInternal(const char* format, ...); // Serialize the va_list to a string std::string ValistToString(const char* format, va_list args) const; // Write the logs marked as headers to the new log file void WriteHeaderInfo(); std::string log_fname_; // Current active info log's file name. std::string dbname_; std::string db_log_dir_; std::string db_absolute_path_; Env* env_; std::shared_ptr logger_; // current status of the logger Status status_; const size_t kMaxLogFileSize; const size_t kLogFileTimeToRoll; // header information std::list headers_; // to avoid frequent env->NowMicros() calls, we cached the current time uint64_t cached_now; uint64_t ctime_; uint64_t cached_now_access_count; uint64_t call_NowMicros_every_N_records_; mutable port::Mutex mutex_; }; // Facade to craete logger automatically Status CreateLoggerFromOptions(const std::string& dbname, const DBOptions& options, std::shared_ptr* logger); } // namespace rocksdb rocksdb-4.5.1/db/auto_roll_logger_test.cc000066400000000000000000000403471267530634000204600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include #include #include #include #include #include #include #include #include "db/auto_roll_logger.h" #include "port/port.h" #include "util/sync_point.h" #include "util/testharness.h" #include "rocksdb/db.h" #include #include using namespace std; namespace rocksdb { class AutoRollLoggerTest : public testing::Test { public: static void InitTestDb() { #ifdef OS_WIN // Replace all slashes in the path so windows CompSpec does not // become confused std::string testDir(kTestDir); std::replace_if(testDir.begin(), testDir.end(), [](char ch) { return ch == '/'; }, '\\'); std::string deleteCmd = "if exist " + testDir + " rd /s /q " + testDir; #else std::string deleteCmd = "rm -rf " + kTestDir; #endif ASSERT_TRUE(system(deleteCmd.c_str()) == 0); Env::Default()->CreateDir(kTestDir); } void RollLogFileBySizeTest(AutoRollLogger* logger, size_t log_max_size, const string& log_message); uint64_t RollLogFileByTimeTest(AutoRollLogger* logger, size_t time, const string& log_message); static const string kSampleMessage; static const string kTestDir; static const string kLogFile; static Env* env; }; const string AutoRollLoggerTest::kSampleMessage( "this is the message to be written to the log file!!"); const string AutoRollLoggerTest::kTestDir(test::TmpDir() + "/db_log_test"); const string AutoRollLoggerTest::kLogFile(test::TmpDir() + "/db_log_test/LOG"); Env* AutoRollLoggerTest::env = Env::Default(); // In this test we only want to Log some simple log message with // no format. LogMessage() provides such a simple interface and // avoids the [format-security] warning which occurs when you // call Log(logger, log_message) directly. namespace { void LogMessage(Logger* logger, const char* message) { Log(logger, "%s", message); } void LogMessage(const InfoLogLevel log_level, Logger* logger, const char* message) { Log(log_level, logger, "%s", message); } } // namespace namespace { void GetFileCreateTime(const std::string& fname, uint64_t* file_ctime) { struct stat s; if (stat(fname.c_str(), &s) != 0) { *file_ctime = (uint64_t)0; } *file_ctime = static_cast(s.st_ctime); } } // namespace void AutoRollLoggerTest::RollLogFileBySizeTest(AutoRollLogger* logger, size_t log_max_size, const string& log_message) { logger->SetInfoLogLevel(InfoLogLevel::INFO_LEVEL); // measure the size of each message, which is supposed // to be equal or greater than log_message.size() LogMessage(logger, log_message.c_str()); size_t message_size = logger->GetLogFileSize(); size_t current_log_size = message_size; // Test the cases when the log file will not be rolled. while (current_log_size + message_size < log_max_size) { LogMessage(logger, log_message.c_str()); current_log_size += message_size; ASSERT_EQ(current_log_size, logger->GetLogFileSize()); } // Now the log file will be rolled LogMessage(logger, log_message.c_str()); // Since rotation is checked before actual logging, we need to // trigger the rotation by logging another message. LogMessage(logger, log_message.c_str()); ASSERT_TRUE(message_size == logger->GetLogFileSize()); } uint64_t AutoRollLoggerTest::RollLogFileByTimeTest( AutoRollLogger* logger, size_t time, const string& log_message) { uint64_t expected_create_time; uint64_t actual_create_time; uint64_t total_log_size; EXPECT_OK(env->GetFileSize(kLogFile, &total_log_size)); GetFileCreateTime(kLogFile, &expected_create_time); logger->SetCallNowMicrosEveryNRecords(0); // -- Write to the log for several times, which is supposed // to be finished before time. for (int i = 0; i < 10; ++i) { LogMessage(logger, log_message.c_str()); EXPECT_OK(logger->GetStatus()); // Make sure we always write to the same log file (by // checking the create time); GetFileCreateTime(kLogFile, &actual_create_time); // Also make sure the log size is increasing. EXPECT_EQ(expected_create_time, actual_create_time); EXPECT_GT(logger->GetLogFileSize(), total_log_size); total_log_size = logger->GetLogFileSize(); } // -- Make the log file expire #ifdef OS_WIN Sleep(static_cast(time) * 1000); #else sleep(static_cast(time)); #endif LogMessage(logger, log_message.c_str()); // At this time, the new log file should be created. GetFileCreateTime(kLogFile, &actual_create_time); EXPECT_GT(actual_create_time, expected_create_time); EXPECT_LT(logger->GetLogFileSize(), total_log_size); expected_create_time = actual_create_time; return expected_create_time; } TEST_F(AutoRollLoggerTest, RollLogFileBySize) { InitTestDb(); size_t log_max_size = 1024 * 5; AutoRollLogger logger(Env::Default(), kTestDir, "", log_max_size, 0); RollLogFileBySizeTest(&logger, log_max_size, kSampleMessage + ":RollLogFileBySize"); } TEST_F(AutoRollLoggerTest, RollLogFileByTime) { size_t time = 2; size_t log_size = 1024 * 5; InitTestDb(); // -- Test the existence of file during the server restart. ASSERT_EQ(Status::NotFound(), env->FileExists(kLogFile)); AutoRollLogger logger(Env::Default(), kTestDir, "", log_size, time); ASSERT_OK(env->FileExists(kLogFile)); RollLogFileByTimeTest(&logger, time, kSampleMessage + ":RollLogFileByTime"); } TEST_F(AutoRollLoggerTest, OpenLogFilesMultipleTimesWithOptionLog_max_size) { // If only 'log_max_size' options is specified, then every time // when rocksdb is restarted, a new empty log file will be created. InitTestDb(); // WORKAROUND: // avoid complier's complaint of "comparison between signed // and unsigned integer expressions" because literal 0 is // treated as "singed". size_t kZero = 0; size_t log_size = 1024; AutoRollLogger* logger = new AutoRollLogger( Env::Default(), kTestDir, "", log_size, 0); LogMessage(logger, kSampleMessage.c_str()); ASSERT_GT(logger->GetLogFileSize(), kZero); delete logger; // reopens the log file and an empty log file will be created. logger = new AutoRollLogger( Env::Default(), kTestDir, "", log_size, 0); ASSERT_EQ(logger->GetLogFileSize(), kZero); delete logger; } TEST_F(AutoRollLoggerTest, CompositeRollByTimeAndSizeLogger) { size_t time = 2, log_max_size = 1024 * 5; InitTestDb(); AutoRollLogger logger(Env::Default(), kTestDir, "", log_max_size, time); // Test the ability to roll by size RollLogFileBySizeTest( &logger, log_max_size, kSampleMessage + ":CompositeRollByTimeAndSizeLogger"); // Test the ability to roll by Time RollLogFileByTimeTest( &logger, time, kSampleMessage + ":CompositeRollByTimeAndSizeLogger"); } #ifndef OS_WIN // TODO: does not build for Windows because of PosixLogger use below. Need to // port TEST_F(AutoRollLoggerTest, CreateLoggerFromOptions) { DBOptions options; shared_ptr logger; // Normal logger ASSERT_OK(CreateLoggerFromOptions(kTestDir, options, &logger)); ASSERT_TRUE(dynamic_cast(logger.get())); // Only roll by size InitTestDb(); options.max_log_file_size = 1024; ASSERT_OK(CreateLoggerFromOptions(kTestDir, options, &logger)); AutoRollLogger* auto_roll_logger = dynamic_cast(logger.get()); ASSERT_TRUE(auto_roll_logger); RollLogFileBySizeTest( auto_roll_logger, options.max_log_file_size, kSampleMessage + ":CreateLoggerFromOptions - size"); // Only roll by Time InitTestDb(); options.max_log_file_size = 0; options.log_file_time_to_roll = 2; ASSERT_OK(CreateLoggerFromOptions(kTestDir, options, &logger)); auto_roll_logger = dynamic_cast(logger.get()); RollLogFileByTimeTest( auto_roll_logger, options.log_file_time_to_roll, kSampleMessage + ":CreateLoggerFromOptions - time"); // roll by both Time and size InitTestDb(); options.max_log_file_size = 1024 * 5; options.log_file_time_to_roll = 2; ASSERT_OK(CreateLoggerFromOptions(kTestDir, options, &logger)); auto_roll_logger = dynamic_cast(logger.get()); RollLogFileBySizeTest( auto_roll_logger, options.max_log_file_size, kSampleMessage + ":CreateLoggerFromOptions - both"); RollLogFileByTimeTest( auto_roll_logger, options.log_file_time_to_roll, kSampleMessage + ":CreateLoggerFromOptions - both"); } TEST_F(AutoRollLoggerTest, LogFlushWhileRolling) { DBOptions options; shared_ptr logger; InitTestDb(); options.max_log_file_size = 1024 * 5; ASSERT_OK(CreateLoggerFromOptions(kTestDir, options, &logger)); AutoRollLogger* auto_roll_logger = dynamic_cast(logger.get()); ASSERT_TRUE(auto_roll_logger); std::thread flush_thread; rocksdb::SyncPoint::GetInstance()->LoadDependency({ // Need to pin the old logger before beginning the roll, as rolling grabs // the mutex, which would prevent us from accessing the old logger. {"AutoRollLogger::Flush:PinnedLogger", "AutoRollLoggerTest::LogFlushWhileRolling:PreRollAndPostThreadInit"}, // Need to finish the flush thread init before this callback because the // callback accesses flush_thread.get_id() in order to apply certain sync // points only to the flush thread. {"AutoRollLoggerTest::LogFlushWhileRolling:PreRollAndPostThreadInit", "AutoRollLoggerTest::LogFlushWhileRolling:FlushCallbackBegin"}, // Need to reset logger at this point in Flush() to exercise a race // condition case, which is executing the flush with the pinned (old) // logger after the roll has cut over to a new logger. {"AutoRollLoggerTest::LogFlushWhileRolling:FlushCallback1", "AutoRollLogger::ResetLogger:BeforeNewLogger"}, {"AutoRollLogger::ResetLogger:AfterNewLogger", "AutoRollLoggerTest::LogFlushWhileRolling:FlushCallback2"}, }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "PosixLogger::Flush:BeginCallback", [&](void* arg) { TEST_SYNC_POINT( "AutoRollLoggerTest::LogFlushWhileRolling:FlushCallbackBegin"); if (std::this_thread::get_id() == flush_thread.get_id()) { TEST_SYNC_POINT( "AutoRollLoggerTest::LogFlushWhileRolling:FlushCallback1"); TEST_SYNC_POINT( "AutoRollLoggerTest::LogFlushWhileRolling:FlushCallback2"); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); flush_thread = std::thread([&]() { auto_roll_logger->Flush(); }); TEST_SYNC_POINT( "AutoRollLoggerTest::LogFlushWhileRolling:PreRollAndPostThreadInit"); RollLogFileBySizeTest(auto_roll_logger, options.max_log_file_size, kSampleMessage + ":LogFlushWhileRolling"); flush_thread.join(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } #endif // OS_WIN TEST_F(AutoRollLoggerTest, InfoLogLevel) { InitTestDb(); size_t log_size = 8192; size_t log_lines = 0; // an extra-scope to force the AutoRollLogger to flush the log file when it // becomes out of scope. { AutoRollLogger logger(Env::Default(), kTestDir, "", log_size, 0); for (int log_level = InfoLogLevel::HEADER_LEVEL; log_level >= InfoLogLevel::DEBUG_LEVEL; log_level--) { logger.SetInfoLogLevel((InfoLogLevel)log_level); for (int log_type = InfoLogLevel::DEBUG_LEVEL; log_type <= InfoLogLevel::HEADER_LEVEL; log_type++) { // log messages with log level smaller than log_level will not be // logged. LogMessage((InfoLogLevel)log_type, &logger, kSampleMessage.c_str()); } log_lines += InfoLogLevel::HEADER_LEVEL - log_level + 1; } for (int log_level = InfoLogLevel::HEADER_LEVEL; log_level >= InfoLogLevel::DEBUG_LEVEL; log_level--) { logger.SetInfoLogLevel((InfoLogLevel)log_level); // again, messages with level smaller than log_level will not be logged. Log(InfoLogLevel::HEADER_LEVEL, &logger, "%s", kSampleMessage.c_str()); Debug(&logger, "%s", kSampleMessage.c_str()); Info(&logger, "%s", kSampleMessage.c_str()); Warn(&logger, "%s", kSampleMessage.c_str()); Error(&logger, "%s", kSampleMessage.c_str()); Fatal(&logger, "%s", kSampleMessage.c_str()); log_lines += InfoLogLevel::HEADER_LEVEL - log_level + 1; } } std::ifstream inFile(AutoRollLoggerTest::kLogFile.c_str()); size_t lines = std::count(std::istreambuf_iterator(inFile), std::istreambuf_iterator(), '\n'); ASSERT_EQ(log_lines, lines); inFile.close(); } // Test the logger Header function for roll over logs // We expect the new logs creates as roll over to carry the headers specified static std::vector GetOldFileNames(const string& path) { std::vector ret; const string dirname = path.substr(/*start=*/ 0, path.find_last_of("/")); const string fname = path.substr(path.find_last_of("/") + 1); std::vector children; Env::Default()->GetChildren(dirname, &children); // We know that the old log files are named [path] // Return all entities that match the pattern for (auto& child : children) { if (fname != child && child.find(fname) == 0) { ret.push_back(dirname + "/" + child); } } return ret; } // Return the number of lines where a given pattern was found in the file static size_t GetLinesCount(const string& fname, const string& pattern) { stringstream ssbuf; string line; size_t count = 0; ifstream inFile(fname.c_str()); ssbuf << inFile.rdbuf(); while (getline(ssbuf, line)) { if (line.find(pattern) != std::string::npos) { count++; } } return count; } TEST_F(AutoRollLoggerTest, LogHeaderTest) { static const size_t MAX_HEADERS = 10; static const size_t LOG_MAX_SIZE = 1024 * 5; static const std::string HEADER_STR = "Log header line"; // test_num == 0 -> standard call to Header() // test_num == 1 -> call to Log() with InfoLogLevel::HEADER_LEVEL for (int test_num = 0; test_num < 2; test_num++) { InitTestDb(); AutoRollLogger logger(Env::Default(), kTestDir, /*db_log_dir=*/ "", LOG_MAX_SIZE, /*log_file_time_to_roll=*/ 0); if (test_num == 0) { // Log some headers explicitly using Header() for (size_t i = 0; i < MAX_HEADERS; i++) { Header(&logger, "%s %d", HEADER_STR.c_str(), i); } } else if (test_num == 1) { // HEADER_LEVEL should make this behave like calling Header() for (size_t i = 0; i < MAX_HEADERS; i++) { Log(InfoLogLevel::HEADER_LEVEL, &logger, "%s %d", HEADER_STR.c_str(), i); } } const string newfname = logger.TEST_log_fname(); // Log enough data to cause a roll over int i = 0; for (size_t iter = 0; iter < 2; iter++) { while (logger.GetLogFileSize() < LOG_MAX_SIZE) { Info(&logger, (kSampleMessage + ":LogHeaderTest line %d").c_str(), i); ++i; } Info(&logger, "Rollover"); } // Flush the log for the latest file LogFlush(&logger); const auto oldfiles = GetOldFileNames(newfname); ASSERT_EQ(oldfiles.size(), (size_t) 2); for (auto& oldfname : oldfiles) { // verify that the files rolled over ASSERT_NE(oldfname, newfname); // verify that the old log contains all the header logs ASSERT_EQ(GetLinesCount(oldfname, HEADER_STR), MAX_HEADERS); } } } TEST_F(AutoRollLoggerTest, LogFileExistence) { rocksdb::DB* db; rocksdb::Options options; string deleteCmd = "rm -rf " + kTestDir; ASSERT_EQ(system(deleteCmd.c_str()), 0); options.max_log_file_size = 100 * 1024 * 1024; options.create_if_missing = true; ASSERT_OK(rocksdb::DB::Open(options, kTestDir, &db)); ASSERT_OK(env->FileExists(kLogFile)); delete db; } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/builder.cc000066400000000000000000000140721267530634000155040ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/builder.h" #include #include #include #include "db/compaction_iterator.h" #include "db/dbformat.h" #include "db/filename.h" #include "db/internal_stats.h" #include "db/merge_helper.h" #include "db/table_cache.h" #include "db/version_edit.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "rocksdb/table.h" #include "table/block_based_table_builder.h" #include "table/internal_iterator.h" #include "util/file_reader_writer.h" #include "util/iostats_context_imp.h" #include "util/stop_watch.h" #include "util/thread_status_util.h" namespace rocksdb { class TableFactory; TableBuilder* NewTableBuilder( const ImmutableCFOptions& ioptions, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* file, const CompressionType compression_type, const CompressionOptions& compression_opts, const bool skip_filters) { return ioptions.table_factory->NewTableBuilder( TableBuilderOptions(ioptions, internal_comparator, int_tbl_prop_collector_factories, compression_type, compression_opts, skip_filters), column_family_id, file); } Status BuildTable( const std::string& dbname, Env* env, const ImmutableCFOptions& ioptions, const EnvOptions& env_options, TableCache* table_cache, InternalIterator* iter, FileMetaData* meta, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, std::vector snapshots, SequenceNumber earliest_write_conflict_snapshot, const CompressionType compression, const CompressionOptions& compression_opts, bool paranoid_file_checks, InternalStats* internal_stats, const Env::IOPriority io_priority, TableProperties* table_properties) { // Reports the IOStats for flush for every following bytes. const size_t kReportFlushIOStatsEvery = 1048576; Status s; meta->fd.file_size = 0; iter->SeekToFirst(); std::string fname = TableFileName(ioptions.db_paths, meta->fd.GetNumber(), meta->fd.GetPathId()); if (iter->Valid()) { TableBuilder* builder; unique_ptr file_writer; { unique_ptr file; s = NewWritableFile(env, fname, &file, env_options); if (!s.ok()) { return s; } file->SetIOPriority(io_priority); file_writer.reset(new WritableFileWriter(std::move(file), env_options)); builder = NewTableBuilder( ioptions, internal_comparator, int_tbl_prop_collector_factories, column_family_id, file_writer.get(), compression, compression_opts); } MergeHelper merge(env, internal_comparator.user_comparator(), ioptions.merge_operator, nullptr, ioptions.info_log, ioptions.min_partial_merge_operands, true /* internal key corruption is not ok */, snapshots.empty() ? 0 : snapshots.back()); CompactionIterator c_iter(iter, internal_comparator.user_comparator(), &merge, kMaxSequenceNumber, &snapshots, earliest_write_conflict_snapshot, env, true /* internal key corruption is not ok */); c_iter.SeekToFirst(); for (; c_iter.Valid(); c_iter.Next()) { const Slice& key = c_iter.key(); const Slice& value = c_iter.value(); builder->Add(key, value); meta->UpdateBoundaries(key, c_iter.ikey().sequence); // TODO(noetzli): Update stats after flush, too. if (io_priority == Env::IO_HIGH && IOSTATS(bytes_written) >= kReportFlushIOStatsEvery) { ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::FLUSH_BYTES_WRITTEN, IOSTATS(bytes_written)); } } // Finish and check for builder errors bool empty = builder->NumEntries() == 0; s = c_iter.status(); if (!s.ok() || empty) { builder->Abandon(); } else { s = builder->Finish(); } if (s.ok() && !empty) { meta->fd.file_size = builder->FileSize(); meta->marked_for_compaction = builder->NeedCompact(); assert(meta->fd.GetFileSize() > 0); if (table_properties) { *table_properties = builder->GetTableProperties(); } } delete builder; // Finish and check for file errors if (s.ok() && !empty && !ioptions.disable_data_sync) { StopWatch sw(env, ioptions.statistics, TABLE_SYNC_MICROS); file_writer->Sync(ioptions.use_fsync); } if (s.ok() && !empty) { s = file_writer->Close(); } if (s.ok() && !empty) { // Verify that the table is usable std::unique_ptr it(table_cache->NewIterator( ReadOptions(), env_options, internal_comparator, meta->fd, nullptr, (internal_stats == nullptr) ? nullptr : internal_stats->GetFileReadHist(0), false)); s = it->status(); if (s.ok() && paranoid_file_checks) { for (it->SeekToFirst(); it->Valid(); it->Next()) { } s = it->status(); } } } // Check for input iterator errors if (!iter->status().ok()) { s = iter->status(); } if (!s.ok() || meta->fd.GetFileSize() == 0) { env->DeleteFile(fname); } return s; } } // namespace rocksdb rocksdb-4.5.1/db/builder.h000066400000000000000000000050431267530634000153440ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include "db/table_properties_collector.h" #include "rocksdb/comparator.h" #include "rocksdb/env.h" #include "rocksdb/status.h" #include "rocksdb/types.h" #include "rocksdb/options.h" #include "rocksdb/immutable_options.h" #include "rocksdb/table_properties.h" namespace rocksdb { struct Options; struct FileMetaData; class Env; struct EnvOptions; class Iterator; class TableCache; class VersionEdit; class TableBuilder; class WritableFileWriter; class InternalStats; class InternalIterator; TableBuilder* NewTableBuilder( const ImmutableCFOptions& options, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* file, const CompressionType compression_type, const CompressionOptions& compression_opts, const bool skip_filters = false); // Build a Table file from the contents of *iter. The generated file // will be named according to number specified in meta. On success, the rest of // *meta will be filled with metadata about the generated table. // If no data is present in *iter, meta->file_size will be set to // zero, and no Table file will be produced. extern Status BuildTable( const std::string& dbname, Env* env, const ImmutableCFOptions& options, const EnvOptions& env_options, TableCache* table_cache, InternalIterator* iter, FileMetaData* meta, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, std::vector snapshots, SequenceNumber earliest_write_conflict_snapshot, const CompressionType compression, const CompressionOptions& compression_opts, bool paranoid_file_checks, InternalStats* internal_stats, const Env::IOPriority io_priority = Env::IO_HIGH, TableProperties* table_properties = nullptr); } // namespace rocksdb rocksdb-4.5.1/db/c.cc000066400000000000000000002142651267530634000143060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "rocksdb/c.h" #include #include "port/port.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/comparator.h" #include "rocksdb/convenience.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/iterator.h" #include "rocksdb/merge_operator.h" #include "rocksdb/options.h" #include "rocksdb/status.h" #include "rocksdb/write_batch.h" #include "rocksdb/memtablerep.h" #include "rocksdb/universal_compaction.h" #include "rocksdb/statistics.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "rocksdb/utilities/backupable_db.h" #include "utilities/merge_operators.h" using rocksdb::Cache; using rocksdb::ColumnFamilyDescriptor; using rocksdb::ColumnFamilyHandle; using rocksdb::ColumnFamilyOptions; using rocksdb::CompactionFilter; using rocksdb::CompactionFilterFactory; using rocksdb::CompactionFilterContext; using rocksdb::CompactionOptionsFIFO; using rocksdb::Comparator; using rocksdb::CompressionType; using rocksdb::DB; using rocksdb::DBOptions; using rocksdb::Env; using rocksdb::InfoLogLevel; using rocksdb::FileLock; using rocksdb::FilterPolicy; using rocksdb::FlushOptions; using rocksdb::Iterator; using rocksdb::Logger; using rocksdb::MergeOperator; using rocksdb::MergeOperators; using rocksdb::NewBloomFilterPolicy; using rocksdb::NewLRUCache; using rocksdb::Options; using rocksdb::BlockBasedTableOptions; using rocksdb::CuckooTableOptions; using rocksdb::RandomAccessFile; using rocksdb::Range; using rocksdb::ReadOptions; using rocksdb::SequentialFile; using rocksdb::Slice; using rocksdb::SliceParts; using rocksdb::SliceTransform; using rocksdb::Snapshot; using rocksdb::Status; using rocksdb::WritableFile; using rocksdb::WriteBatch; using rocksdb::WriteOptions; using rocksdb::LiveFileMetaData; using rocksdb::BackupEngine; using rocksdb::BackupableDBOptions; using rocksdb::BackupInfo; using rocksdb::RestoreOptions; using rocksdb::CompactRangeOptions; using std::shared_ptr; extern "C" { struct rocksdb_t { DB* rep; }; struct rocksdb_backup_engine_t { BackupEngine* rep; }; struct rocksdb_backup_engine_info_t { std::vector rep; }; struct rocksdb_restore_options_t { RestoreOptions rep; }; struct rocksdb_iterator_t { Iterator* rep; }; struct rocksdb_writebatch_t { WriteBatch rep; }; struct rocksdb_snapshot_t { const Snapshot* rep; }; struct rocksdb_flushoptions_t { FlushOptions rep; }; struct rocksdb_fifo_compaction_options_t { CompactionOptionsFIFO rep; }; struct rocksdb_readoptions_t { ReadOptions rep; Slice upper_bound; // stack variable to set pointer to in ReadOptions }; struct rocksdb_writeoptions_t { WriteOptions rep; }; struct rocksdb_options_t { Options rep; }; struct rocksdb_block_based_table_options_t { BlockBasedTableOptions rep; }; struct rocksdb_cuckoo_table_options_t { CuckooTableOptions rep; }; struct rocksdb_seqfile_t { SequentialFile* rep; }; struct rocksdb_randomfile_t { RandomAccessFile* rep; }; struct rocksdb_writablefile_t { WritableFile* rep; }; struct rocksdb_filelock_t { FileLock* rep; }; struct rocksdb_logger_t { shared_ptr rep; }; struct rocksdb_cache_t { shared_ptr rep; }; struct rocksdb_livefiles_t { std::vector rep; }; struct rocksdb_column_family_handle_t { ColumnFamilyHandle* rep; }; struct rocksdb_compactionfiltercontext_t { CompactionFilter::Context rep; }; struct rocksdb_compactionfilter_t : public CompactionFilter { void* state_; void (*destructor_)(void*); unsigned char (*filter_)( void*, int level, const char* key, size_t key_length, const char* existing_value, size_t value_length, char** new_value, size_t *new_value_length, unsigned char* value_changed); const char* (*name_)(void*); virtual ~rocksdb_compactionfilter_t() { (*destructor_)(state_); } virtual bool Filter(int level, const Slice& key, const Slice& existing_value, std::string* new_value, bool* value_changed) const override { char* c_new_value = nullptr; size_t new_value_length = 0; unsigned char c_value_changed = 0; unsigned char result = (*filter_)( state_, level, key.data(), key.size(), existing_value.data(), existing_value.size(), &c_new_value, &new_value_length, &c_value_changed); if (c_value_changed) { new_value->assign(c_new_value, new_value_length); *value_changed = true; } return result; } virtual const char* Name() const override { return (*name_)(state_); } }; struct rocksdb_compactionfilterfactory_t : public CompactionFilterFactory { void* state_; void (*destructor_)(void*); rocksdb_compactionfilter_t* (*create_compaction_filter_)( void*, rocksdb_compactionfiltercontext_t* context); const char* (*name_)(void*); virtual ~rocksdb_compactionfilterfactory_t() { (*destructor_)(state_); } virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { rocksdb_compactionfiltercontext_t ccontext; ccontext.rep = context; CompactionFilter* cf = (*create_compaction_filter_)(state_, &ccontext); return std::unique_ptr(cf); } virtual const char* Name() const override { return (*name_)(state_); } }; struct rocksdb_comparator_t : public Comparator { void* state_; void (*destructor_)(void*); int (*compare_)( void*, const char* a, size_t alen, const char* b, size_t blen); const char* (*name_)(void*); virtual ~rocksdb_comparator_t() { (*destructor_)(state_); } virtual int Compare(const Slice& a, const Slice& b) const override { return (*compare_)(state_, a.data(), a.size(), b.data(), b.size()); } virtual const char* Name() const override { return (*name_)(state_); } // No-ops since the C binding does not support key shortening methods. virtual void FindShortestSeparator(std::string*, const Slice&) const override {} virtual void FindShortSuccessor(std::string* key) const override {} }; struct rocksdb_filterpolicy_t : public FilterPolicy { void* state_; void (*destructor_)(void*); const char* (*name_)(void*); char* (*create_)( void*, const char* const* key_array, const size_t* key_length_array, int num_keys, size_t* filter_length); unsigned char (*key_match_)( void*, const char* key, size_t length, const char* filter, size_t filter_length); void (*delete_filter_)( void*, const char* filter, size_t filter_length); virtual ~rocksdb_filterpolicy_t() { (*destructor_)(state_); } virtual const char* Name() const override { return (*name_)(state_); } virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const override { std::vector key_pointers(n); std::vector key_sizes(n); for (int i = 0; i < n; i++) { key_pointers[i] = keys[i].data(); key_sizes[i] = keys[i].size(); } size_t len; char* filter = (*create_)(state_, &key_pointers[0], &key_sizes[0], n, &len); dst->append(filter, len); if (delete_filter_ != nullptr) { (*delete_filter_)(state_, filter, len); } else { free(filter); } } virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const override { return (*key_match_)(state_, key.data(), key.size(), filter.data(), filter.size()); } }; struct rocksdb_mergeoperator_t : public MergeOperator { void* state_; void (*destructor_)(void*); const char* (*name_)(void*); char* (*full_merge_)( void*, const char* key, size_t key_length, const char* existing_value, size_t existing_value_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length); char* (*partial_merge_)(void*, const char* key, size_t key_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length); void (*delete_value_)( void*, const char* value, size_t value_length); virtual ~rocksdb_mergeoperator_t() { (*destructor_)(state_); } virtual const char* Name() const override { return (*name_)(state_); } virtual bool FullMerge(const Slice& key, const Slice* existing_value, const std::deque& operand_list, std::string* new_value, Logger* logger) const override { size_t n = operand_list.size(); std::vector operand_pointers(n); std::vector operand_sizes(n); for (size_t i = 0; i < n; i++) { Slice operand(operand_list[i]); operand_pointers[i] = operand.data(); operand_sizes[i] = operand.size(); } const char* existing_value_data = nullptr; size_t existing_value_len = 0; if (existing_value != nullptr) { existing_value_data = existing_value->data(); existing_value_len = existing_value->size(); } unsigned char success; size_t new_value_len; char* tmp_new_value = (*full_merge_)( state_, key.data(), key.size(), existing_value_data, existing_value_len, &operand_pointers[0], &operand_sizes[0], static_cast(n), &success, &new_value_len); new_value->assign(tmp_new_value, new_value_len); if (delete_value_ != nullptr) { (*delete_value_)(state_, tmp_new_value, new_value_len); } else { free(tmp_new_value); } return success; } virtual bool PartialMergeMulti(const Slice& key, const std::deque& operand_list, std::string* new_value, Logger* logger) const override { size_t operand_count = operand_list.size(); std::vector operand_pointers(operand_count); std::vector operand_sizes(operand_count); for (size_t i = 0; i < operand_count; ++i) { Slice operand(operand_list[i]); operand_pointers[i] = operand.data(); operand_sizes[i] = operand.size(); } unsigned char success; size_t new_value_len; char* tmp_new_value = (*partial_merge_)( state_, key.data(), key.size(), &operand_pointers[0], &operand_sizes[0], static_cast(operand_count), &success, &new_value_len); new_value->assign(tmp_new_value, new_value_len); if (delete_value_ != nullptr) { (*delete_value_)(state_, tmp_new_value, new_value_len); } else { free(tmp_new_value); } return success; } }; struct rocksdb_env_t { Env* rep; bool is_default; }; struct rocksdb_slicetransform_t : public SliceTransform { void* state_; void (*destructor_)(void*); const char* (*name_)(void*); char* (*transform_)( void*, const char* key, size_t length, size_t* dst_length); unsigned char (*in_domain_)( void*, const char* key, size_t length); unsigned char (*in_range_)( void*, const char* key, size_t length); virtual ~rocksdb_slicetransform_t() { (*destructor_)(state_); } virtual const char* Name() const override { return (*name_)(state_); } virtual Slice Transform(const Slice& src) const override { size_t len; char* dst = (*transform_)(state_, src.data(), src.size(), &len); return Slice(dst, len); } virtual bool InDomain(const Slice& src) const override { return (*in_domain_)(state_, src.data(), src.size()); } virtual bool InRange(const Slice& src) const override { return (*in_range_)(state_, src.data(), src.size()); } }; struct rocksdb_universal_compaction_options_t { rocksdb::CompactionOptionsUniversal *rep; }; static bool SaveError(char** errptr, const Status& s) { assert(errptr != nullptr); if (s.ok()) { return false; } else if (*errptr == nullptr) { *errptr = strdup(s.ToString().c_str()); } else { // TODO(sanjay): Merge with existing error? // This is a bug if *errptr is not created by malloc() free(*errptr); *errptr = strdup(s.ToString().c_str()); } return true; } static char* CopyString(const std::string& str) { char* result = reinterpret_cast(malloc(sizeof(char) * str.size())); memcpy(result, str.data(), sizeof(char) * str.size()); return result; } rocksdb_t* rocksdb_open( const rocksdb_options_t* options, const char* name, char** errptr) { DB* db; if (SaveError(errptr, DB::Open(options->rep, std::string(name), &db))) { return nullptr; } rocksdb_t* result = new rocksdb_t; result->rep = db; return result; } rocksdb_t* rocksdb_open_for_read_only( const rocksdb_options_t* options, const char* name, unsigned char error_if_log_file_exist, char** errptr) { DB* db; if (SaveError(errptr, DB::OpenForReadOnly(options->rep, std::string(name), &db, error_if_log_file_exist))) { return nullptr; } rocksdb_t* result = new rocksdb_t; result->rep = db; return result; } rocksdb_backup_engine_t* rocksdb_backup_engine_open( const rocksdb_options_t* options, const char* path, char** errptr) { BackupEngine* be; if (SaveError(errptr, BackupEngine::Open(options->rep.env, BackupableDBOptions(path), &be))) { return nullptr; } rocksdb_backup_engine_t* result = new rocksdb_backup_engine_t; result->rep = be; return result; } void rocksdb_backup_engine_create_new_backup(rocksdb_backup_engine_t* be, rocksdb_t* db, char** errptr) { SaveError(errptr, be->rep->CreateNewBackup(db->rep)); } rocksdb_restore_options_t* rocksdb_restore_options_create() { return new rocksdb_restore_options_t; } void rocksdb_restore_options_destroy(rocksdb_restore_options_t* opt) { delete opt; } void rocksdb_restore_options_set_keep_log_files(rocksdb_restore_options_t* opt, int v) { opt->rep.keep_log_files = v; } void rocksdb_backup_engine_restore_db_from_latest_backup( rocksdb_backup_engine_t* be, const char* db_dir, const char* wal_dir, const rocksdb_restore_options_t* restore_options, char** errptr) { SaveError(errptr, be->rep->RestoreDBFromLatestBackup(std::string(db_dir), std::string(wal_dir), restore_options->rep)); } const rocksdb_backup_engine_info_t* rocksdb_backup_engine_get_backup_info( rocksdb_backup_engine_t* be) { rocksdb_backup_engine_info_t* result = new rocksdb_backup_engine_info_t; be->rep->GetBackupInfo(&result->rep); return result; } int rocksdb_backup_engine_info_count(const rocksdb_backup_engine_info_t* info) { return static_cast(info->rep.size()); } int64_t rocksdb_backup_engine_info_timestamp( const rocksdb_backup_engine_info_t* info, int index) { return info->rep[index].timestamp; } uint32_t rocksdb_backup_engine_info_backup_id( const rocksdb_backup_engine_info_t* info, int index) { return info->rep[index].backup_id; } uint64_t rocksdb_backup_engine_info_size( const rocksdb_backup_engine_info_t* info, int index) { return info->rep[index].size; } uint32_t rocksdb_backup_engine_info_number_files( const rocksdb_backup_engine_info_t* info, int index) { return info->rep[index].number_files; } void rocksdb_backup_engine_info_destroy( const rocksdb_backup_engine_info_t* info) { delete info; } void rocksdb_backup_engine_close(rocksdb_backup_engine_t* be) { delete be->rep; delete be; } void rocksdb_close(rocksdb_t* db) { delete db->rep; delete db; } void rocksdb_options_set_uint64add_merge_operator(rocksdb_options_t* opt) { opt->rep.merge_operator = rocksdb::MergeOperators::CreateUInt64AddOperator(); } rocksdb_t* rocksdb_open_column_families( const rocksdb_options_t* db_options, const char* name, int num_column_families, const char** column_family_names, const rocksdb_options_t** column_family_options, rocksdb_column_family_handle_t** column_family_handles, char** errptr) { std::vector column_families; for (int i = 0; i < num_column_families; i++) { column_families.push_back(ColumnFamilyDescriptor( std::string(column_family_names[i]), ColumnFamilyOptions(column_family_options[i]->rep))); } DB* db; std::vector handles; if (SaveError(errptr, DB::Open(DBOptions(db_options->rep), std::string(name), column_families, &handles, &db))) { return nullptr; } for (size_t i = 0; i < handles.size(); i++) { rocksdb_column_family_handle_t* c_handle = new rocksdb_column_family_handle_t; c_handle->rep = handles[i]; column_family_handles[i] = c_handle; } rocksdb_t* result = new rocksdb_t; result->rep = db; return result; } rocksdb_t* rocksdb_open_for_read_only_column_families( const rocksdb_options_t* db_options, const char* name, int num_column_families, const char** column_family_names, const rocksdb_options_t** column_family_options, rocksdb_column_family_handle_t** column_family_handles, unsigned char error_if_log_file_exist, char** errptr) { std::vector column_families; for (int i = 0; i < num_column_families; i++) { column_families.push_back(ColumnFamilyDescriptor( std::string(column_family_names[i]), ColumnFamilyOptions(column_family_options[i]->rep))); } DB* db; std::vector handles; if (SaveError(errptr, DB::OpenForReadOnly(DBOptions(db_options->rep), std::string(name), column_families, &handles, &db, error_if_log_file_exist))) { return nullptr; } for (size_t i = 0; i < handles.size(); i++) { rocksdb_column_family_handle_t* c_handle = new rocksdb_column_family_handle_t; c_handle->rep = handles[i]; column_family_handles[i] = c_handle; } rocksdb_t* result = new rocksdb_t; result->rep = db; return result; } char** rocksdb_list_column_families( const rocksdb_options_t* options, const char* name, size_t* lencfs, char** errptr) { std::vector fams; SaveError(errptr, DB::ListColumnFamilies(DBOptions(options->rep), std::string(name), &fams)); *lencfs = fams.size(); char** column_families = static_cast(malloc(sizeof(char*) * fams.size())); for (size_t i = 0; i < fams.size(); i++) { column_families[i] = strdup(fams[i].c_str()); } return column_families; } void rocksdb_list_column_families_destroy(char** list, size_t len) { for (size_t i = 0; i < len; ++i) { free(list[i]); } free(list); } rocksdb_column_family_handle_t* rocksdb_create_column_family( rocksdb_t* db, const rocksdb_options_t* column_family_options, const char* column_family_name, char** errptr) { rocksdb_column_family_handle_t* handle = new rocksdb_column_family_handle_t; SaveError(errptr, db->rep->CreateColumnFamily(ColumnFamilyOptions(column_family_options->rep), std::string(column_family_name), &(handle->rep))); return handle; } void rocksdb_drop_column_family( rocksdb_t* db, rocksdb_column_family_handle_t* handle, char** errptr) { SaveError(errptr, db->rep->DropColumnFamily(handle->rep)); } void rocksdb_column_family_handle_destroy(rocksdb_column_family_handle_t* handle) { delete handle->rep; delete handle; } void rocksdb_put( rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr) { SaveError(errptr, db->rep->Put(options->rep, Slice(key, keylen), Slice(val, vallen))); } void rocksdb_put_cf( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr) { SaveError(errptr, db->rep->Put(options->rep, column_family->rep, Slice(key, keylen), Slice(val, vallen))); } void rocksdb_delete( rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key, size_t keylen, char** errptr) { SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen))); } void rocksdb_delete_cf( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, char** errptr) { SaveError(errptr, db->rep->Delete(options->rep, column_family->rep, Slice(key, keylen))); } void rocksdb_merge( rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr) { SaveError(errptr, db->rep->Merge(options->rep, Slice(key, keylen), Slice(val, vallen))); } void rocksdb_merge_cf( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr) { SaveError(errptr, db->rep->Merge(options->rep, column_family->rep, Slice(key, keylen), Slice(val, vallen))); } void rocksdb_write( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_writebatch_t* batch, char** errptr) { SaveError(errptr, db->rep->Write(options->rep, &batch->rep)); } char* rocksdb_get( rocksdb_t* db, const rocksdb_readoptions_t* options, const char* key, size_t keylen, size_t* vallen, char** errptr) { char* result = nullptr; std::string tmp; Status s = db->rep->Get(options->rep, Slice(key, keylen), &tmp); if (s.ok()) { *vallen = tmp.size(); result = CopyString(tmp); } else { *vallen = 0; if (!s.IsNotFound()) { SaveError(errptr, s); } } return result; } char* rocksdb_get_cf( rocksdb_t* db, const rocksdb_readoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, size_t* vallen, char** errptr) { char* result = nullptr; std::string tmp; Status s = db->rep->Get(options->rep, column_family->rep, Slice(key, keylen), &tmp); if (s.ok()) { *vallen = tmp.size(); result = CopyString(tmp); } else { *vallen = 0; if (!s.IsNotFound()) { SaveError(errptr, s); } } return result; } void rocksdb_multi_get( rocksdb_t* db, const rocksdb_readoptions_t* options, size_t num_keys, const char* const* keys_list, const size_t* keys_list_sizes, char** values_list, size_t* values_list_sizes, char** errs) { std::vector keys(num_keys); for (size_t i = 0; i < num_keys; i++) { keys[i] = Slice(keys_list[i], keys_list_sizes[i]); } std::vector values(num_keys); std::vector statuses = db->rep->MultiGet(options->rep, keys, &values); for (size_t i = 0; i < num_keys; i++) { if (statuses[i].ok()) { values_list[i] = CopyString(values[i]); values_list_sizes[i] = values[i].size(); errs[i] = nullptr; } else { values_list[i] = nullptr; values_list_sizes[i] = 0; if (!statuses[i].IsNotFound()) { errs[i] = strdup(statuses[i].ToString().c_str()); } else { errs[i] = nullptr; } } } } void rocksdb_multi_get_cf( rocksdb_t* db, const rocksdb_readoptions_t* options, const rocksdb_column_family_handle_t* const* column_families, size_t num_keys, const char* const* keys_list, const size_t* keys_list_sizes, char** values_list, size_t* values_list_sizes, char** errs) { std::vector keys(num_keys); std::vector cfs(num_keys); for (size_t i = 0; i < num_keys; i++) { keys[i] = Slice(keys_list[i], keys_list_sizes[i]); cfs[i] = column_families[i]->rep; } std::vector values(num_keys); std::vector statuses = db->rep->MultiGet(options->rep, cfs, keys, &values); for (size_t i = 0; i < num_keys; i++) { if (statuses[i].ok()) { values_list[i] = CopyString(values[i]); values_list_sizes[i] = values[i].size(); errs[i] = nullptr; } else { values_list[i] = nullptr; values_list_sizes[i] = 0; if (!statuses[i].IsNotFound()) { errs[i] = strdup(statuses[i].ToString().c_str()); } else { errs[i] = nullptr; } } } } rocksdb_iterator_t* rocksdb_create_iterator( rocksdb_t* db, const rocksdb_readoptions_t* options) { rocksdb_iterator_t* result = new rocksdb_iterator_t; result->rep = db->rep->NewIterator(options->rep); return result; } rocksdb_iterator_t* rocksdb_create_iterator_cf( rocksdb_t* db, const rocksdb_readoptions_t* options, rocksdb_column_family_handle_t* column_family) { rocksdb_iterator_t* result = new rocksdb_iterator_t; result->rep = db->rep->NewIterator(options->rep, column_family->rep); return result; } const rocksdb_snapshot_t* rocksdb_create_snapshot( rocksdb_t* db) { rocksdb_snapshot_t* result = new rocksdb_snapshot_t; result->rep = db->rep->GetSnapshot(); return result; } void rocksdb_release_snapshot( rocksdb_t* db, const rocksdb_snapshot_t* snapshot) { db->rep->ReleaseSnapshot(snapshot->rep); delete snapshot; } char* rocksdb_property_value( rocksdb_t* db, const char* propname) { std::string tmp; if (db->rep->GetProperty(Slice(propname), &tmp)) { // We use strdup() since we expect human readable output. return strdup(tmp.c_str()); } else { return nullptr; } } char* rocksdb_property_value_cf( rocksdb_t* db, rocksdb_column_family_handle_t* column_family, const char* propname) { std::string tmp; if (db->rep->GetProperty(column_family->rep, Slice(propname), &tmp)) { // We use strdup() since we expect human readable output. return strdup(tmp.c_str()); } else { return nullptr; } } void rocksdb_approximate_sizes( rocksdb_t* db, int num_ranges, const char* const* range_start_key, const size_t* range_start_key_len, const char* const* range_limit_key, const size_t* range_limit_key_len, uint64_t* sizes) { Range* ranges = new Range[num_ranges]; for (int i = 0; i < num_ranges; i++) { ranges[i].start = Slice(range_start_key[i], range_start_key_len[i]); ranges[i].limit = Slice(range_limit_key[i], range_limit_key_len[i]); } db->rep->GetApproximateSizes(ranges, num_ranges, sizes); delete[] ranges; } void rocksdb_approximate_sizes_cf( rocksdb_t* db, rocksdb_column_family_handle_t* column_family, int num_ranges, const char* const* range_start_key, const size_t* range_start_key_len, const char* const* range_limit_key, const size_t* range_limit_key_len, uint64_t* sizes) { Range* ranges = new Range[num_ranges]; for (int i = 0; i < num_ranges; i++) { ranges[i].start = Slice(range_start_key[i], range_start_key_len[i]); ranges[i].limit = Slice(range_limit_key[i], range_limit_key_len[i]); } db->rep->GetApproximateSizes(column_family->rep, ranges, num_ranges, sizes); delete[] ranges; } void rocksdb_delete_file( rocksdb_t* db, const char* name) { db->rep->DeleteFile(name); } const rocksdb_livefiles_t* rocksdb_livefiles( rocksdb_t* db) { rocksdb_livefiles_t* result = new rocksdb_livefiles_t; db->rep->GetLiveFilesMetaData(&result->rep); return result; } void rocksdb_compact_range( rocksdb_t* db, const char* start_key, size_t start_key_len, const char* limit_key, size_t limit_key_len) { Slice a, b; db->rep->CompactRange( CompactRangeOptions(), // Pass nullptr Slice if corresponding "const char*" is nullptr (start_key ? (a = Slice(start_key, start_key_len), &a) : nullptr), (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : nullptr)); } void rocksdb_compact_range_cf( rocksdb_t* db, rocksdb_column_family_handle_t* column_family, const char* start_key, size_t start_key_len, const char* limit_key, size_t limit_key_len) { Slice a, b; db->rep->CompactRange( CompactRangeOptions(), column_family->rep, // Pass nullptr Slice if corresponding "const char*" is nullptr (start_key ? (a = Slice(start_key, start_key_len), &a) : nullptr), (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : nullptr)); } void rocksdb_flush( rocksdb_t* db, const rocksdb_flushoptions_t* options, char** errptr) { SaveError(errptr, db->rep->Flush(options->rep)); } void rocksdb_disable_file_deletions( rocksdb_t* db, char** errptr) { SaveError(errptr, db->rep->DisableFileDeletions()); } void rocksdb_enable_file_deletions( rocksdb_t* db, unsigned char force, char** errptr) { SaveError(errptr, db->rep->EnableFileDeletions(force)); } void rocksdb_destroy_db( const rocksdb_options_t* options, const char* name, char** errptr) { SaveError(errptr, DestroyDB(name, options->rep)); } void rocksdb_repair_db( const rocksdb_options_t* options, const char* name, char** errptr) { SaveError(errptr, RepairDB(name, options->rep)); } void rocksdb_iter_destroy(rocksdb_iterator_t* iter) { delete iter->rep; delete iter; } unsigned char rocksdb_iter_valid(const rocksdb_iterator_t* iter) { return iter->rep->Valid(); } void rocksdb_iter_seek_to_first(rocksdb_iterator_t* iter) { iter->rep->SeekToFirst(); } void rocksdb_iter_seek_to_last(rocksdb_iterator_t* iter) { iter->rep->SeekToLast(); } void rocksdb_iter_seek(rocksdb_iterator_t* iter, const char* k, size_t klen) { iter->rep->Seek(Slice(k, klen)); } void rocksdb_iter_next(rocksdb_iterator_t* iter) { iter->rep->Next(); } void rocksdb_iter_prev(rocksdb_iterator_t* iter) { iter->rep->Prev(); } const char* rocksdb_iter_key(const rocksdb_iterator_t* iter, size_t* klen) { Slice s = iter->rep->key(); *klen = s.size(); return s.data(); } const char* rocksdb_iter_value(const rocksdb_iterator_t* iter, size_t* vlen) { Slice s = iter->rep->value(); *vlen = s.size(); return s.data(); } void rocksdb_iter_get_error(const rocksdb_iterator_t* iter, char** errptr) { SaveError(errptr, iter->rep->status()); } rocksdb_writebatch_t* rocksdb_writebatch_create() { return new rocksdb_writebatch_t; } rocksdb_writebatch_t* rocksdb_writebatch_create_from(const char* rep, size_t size) { rocksdb_writebatch_t* b = new rocksdb_writebatch_t; b->rep = WriteBatch(std::string(rep, size)); return b; } void rocksdb_writebatch_destroy(rocksdb_writebatch_t* b) { delete b; } void rocksdb_writebatch_clear(rocksdb_writebatch_t* b) { b->rep.Clear(); } int rocksdb_writebatch_count(rocksdb_writebatch_t* b) { return b->rep.Count(); } void rocksdb_writebatch_put( rocksdb_writebatch_t* b, const char* key, size_t klen, const char* val, size_t vlen) { b->rep.Put(Slice(key, klen), Slice(val, vlen)); } void rocksdb_writebatch_put_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, const char* key, size_t klen, const char* val, size_t vlen) { b->rep.Put(column_family->rep, Slice(key, klen), Slice(val, vlen)); } void rocksdb_writebatch_putv( rocksdb_writebatch_t* b, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes) { std::vector key_slices(num_keys); for (int i = 0; i < num_keys; i++) { key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]); } std::vector value_slices(num_values); for (int i = 0; i < num_values; i++) { value_slices[i] = Slice(values_list[i], values_list_sizes[i]); } b->rep.Put(SliceParts(key_slices.data(), num_keys), SliceParts(value_slices.data(), num_values)); } void rocksdb_writebatch_putv_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes) { std::vector key_slices(num_keys); for (int i = 0; i < num_keys; i++) { key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]); } std::vector value_slices(num_values); for (int i = 0; i < num_values; i++) { value_slices[i] = Slice(values_list[i], values_list_sizes[i]); } b->rep.Put(column_family->rep, SliceParts(key_slices.data(), num_keys), SliceParts(value_slices.data(), num_values)); } void rocksdb_writebatch_merge( rocksdb_writebatch_t* b, const char* key, size_t klen, const char* val, size_t vlen) { b->rep.Merge(Slice(key, klen), Slice(val, vlen)); } void rocksdb_writebatch_merge_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, const char* key, size_t klen, const char* val, size_t vlen) { b->rep.Merge(column_family->rep, Slice(key, klen), Slice(val, vlen)); } void rocksdb_writebatch_mergev( rocksdb_writebatch_t* b, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes) { std::vector key_slices(num_keys); for (int i = 0; i < num_keys; i++) { key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]); } std::vector value_slices(num_values); for (int i = 0; i < num_values; i++) { value_slices[i] = Slice(values_list[i], values_list_sizes[i]); } b->rep.Merge(SliceParts(key_slices.data(), num_keys), SliceParts(value_slices.data(), num_values)); } void rocksdb_writebatch_mergev_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes) { std::vector key_slices(num_keys); for (int i = 0; i < num_keys; i++) { key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]); } std::vector value_slices(num_values); for (int i = 0; i < num_values; i++) { value_slices[i] = Slice(values_list[i], values_list_sizes[i]); } b->rep.Merge(column_family->rep, SliceParts(key_slices.data(), num_keys), SliceParts(value_slices.data(), num_values)); } void rocksdb_writebatch_delete( rocksdb_writebatch_t* b, const char* key, size_t klen) { b->rep.Delete(Slice(key, klen)); } void rocksdb_writebatch_delete_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, const char* key, size_t klen) { b->rep.Delete(column_family->rep, Slice(key, klen)); } void rocksdb_writebatch_deletev( rocksdb_writebatch_t* b, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes) { std::vector key_slices(num_keys); for (int i = 0; i < num_keys; i++) { key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]); } b->rep.Delete(SliceParts(key_slices.data(), num_keys)); } void rocksdb_writebatch_deletev_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes) { std::vector key_slices(num_keys); for (int i = 0; i < num_keys; i++) { key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]); } b->rep.Delete(column_family->rep, SliceParts(key_slices.data(), num_keys)); } void rocksdb_writebatch_put_log_data( rocksdb_writebatch_t* b, const char* blob, size_t len) { b->rep.PutLogData(Slice(blob, len)); } void rocksdb_writebatch_iterate( rocksdb_writebatch_t* b, void* state, void (*put)(void*, const char* k, size_t klen, const char* v, size_t vlen), void (*deleted)(void*, const char* k, size_t klen)) { class H : public WriteBatch::Handler { public: void* state_; void (*put_)(void*, const char* k, size_t klen, const char* v, size_t vlen); void (*deleted_)(void*, const char* k, size_t klen); virtual void Put(const Slice& key, const Slice& value) override { (*put_)(state_, key.data(), key.size(), value.data(), value.size()); } virtual void Delete(const Slice& key) override { (*deleted_)(state_, key.data(), key.size()); } }; H handler; handler.state_ = state; handler.put_ = put; handler.deleted_ = deleted; b->rep.Iterate(&handler); } const char* rocksdb_writebatch_data(rocksdb_writebatch_t* b, size_t* size) { *size = b->rep.GetDataSize(); return b->rep.Data().c_str(); } rocksdb_block_based_table_options_t* rocksdb_block_based_options_create() { return new rocksdb_block_based_table_options_t; } void rocksdb_block_based_options_destroy( rocksdb_block_based_table_options_t* options) { delete options; } void rocksdb_block_based_options_set_block_size( rocksdb_block_based_table_options_t* options, size_t block_size) { options->rep.block_size = block_size; } void rocksdb_block_based_options_set_block_size_deviation( rocksdb_block_based_table_options_t* options, int block_size_deviation) { options->rep.block_size_deviation = block_size_deviation; } void rocksdb_block_based_options_set_block_restart_interval( rocksdb_block_based_table_options_t* options, int block_restart_interval) { options->rep.block_restart_interval = block_restart_interval; } void rocksdb_block_based_options_set_filter_policy( rocksdb_block_based_table_options_t* options, rocksdb_filterpolicy_t* filter_policy) { options->rep.filter_policy.reset(filter_policy); } void rocksdb_block_based_options_set_no_block_cache( rocksdb_block_based_table_options_t* options, unsigned char no_block_cache) { options->rep.no_block_cache = no_block_cache; } void rocksdb_block_based_options_set_block_cache( rocksdb_block_based_table_options_t* options, rocksdb_cache_t* block_cache) { if (block_cache) { options->rep.block_cache = block_cache->rep; } } void rocksdb_block_based_options_set_block_cache_compressed( rocksdb_block_based_table_options_t* options, rocksdb_cache_t* block_cache_compressed) { if (block_cache_compressed) { options->rep.block_cache_compressed = block_cache_compressed->rep; } } void rocksdb_block_based_options_set_whole_key_filtering( rocksdb_block_based_table_options_t* options, unsigned char v) { options->rep.whole_key_filtering = v; } void rocksdb_block_based_options_set_format_version( rocksdb_block_based_table_options_t* options, int v) { options->rep.format_version = v; } void rocksdb_block_based_options_set_index_type( rocksdb_block_based_table_options_t* options, int v) { options->rep.index_type = static_cast(v); } void rocksdb_block_based_options_set_hash_index_allow_collision( rocksdb_block_based_table_options_t* options, unsigned char v) { options->rep.hash_index_allow_collision = v; } void rocksdb_block_based_options_set_cache_index_and_filter_blocks( rocksdb_block_based_table_options_t* options, unsigned char v) { options->rep.cache_index_and_filter_blocks = v; } void rocksdb_block_based_options_set_skip_table_builder_flush( rocksdb_block_based_table_options_t* options, unsigned char v) { options->rep.skip_table_builder_flush = v; } void rocksdb_options_set_block_based_table_factory( rocksdb_options_t *opt, rocksdb_block_based_table_options_t* table_options) { if (table_options) { opt->rep.table_factory.reset( rocksdb::NewBlockBasedTableFactory(table_options->rep)); } } rocksdb_cuckoo_table_options_t* rocksdb_cuckoo_options_create() { return new rocksdb_cuckoo_table_options_t; } void rocksdb_cuckoo_options_destroy( rocksdb_cuckoo_table_options_t* options) { delete options; } void rocksdb_cuckoo_options_set_hash_ratio( rocksdb_cuckoo_table_options_t* options, double v) { options->rep.hash_table_ratio = v; } void rocksdb_cuckoo_options_set_max_search_depth( rocksdb_cuckoo_table_options_t* options, uint32_t v) { options->rep.max_search_depth = v; } void rocksdb_cuckoo_options_set_cuckoo_block_size( rocksdb_cuckoo_table_options_t* options, uint32_t v) { options->rep.cuckoo_block_size = v; } void rocksdb_cuckoo_options_set_identity_as_first_hash( rocksdb_cuckoo_table_options_t* options, unsigned char v) { options->rep.identity_as_first_hash = v; } void rocksdb_cuckoo_options_set_use_module_hash( rocksdb_cuckoo_table_options_t* options, unsigned char v) { options->rep.use_module_hash = v; } void rocksdb_options_set_cuckoo_table_factory( rocksdb_options_t *opt, rocksdb_cuckoo_table_options_t* table_options) { if (table_options) { opt->rep.table_factory.reset( rocksdb::NewCuckooTableFactory(table_options->rep)); } } rocksdb_options_t* rocksdb_options_create() { return new rocksdb_options_t; } void rocksdb_options_destroy(rocksdb_options_t* options) { delete options; } void rocksdb_options_increase_parallelism( rocksdb_options_t* opt, int total_threads) { opt->rep.IncreaseParallelism(total_threads); } void rocksdb_options_optimize_for_point_lookup( rocksdb_options_t* opt, uint64_t block_cache_size_mb) { opt->rep.OptimizeForPointLookup(block_cache_size_mb); } void rocksdb_options_optimize_level_style_compaction( rocksdb_options_t* opt, uint64_t memtable_memory_budget) { opt->rep.OptimizeLevelStyleCompaction(memtable_memory_budget); } void rocksdb_options_optimize_universal_style_compaction( rocksdb_options_t* opt, uint64_t memtable_memory_budget) { opt->rep.OptimizeUniversalStyleCompaction(memtable_memory_budget); } void rocksdb_options_set_compaction_filter( rocksdb_options_t* opt, rocksdb_compactionfilter_t* filter) { opt->rep.compaction_filter = filter; } void rocksdb_options_set_compaction_filter_factory( rocksdb_options_t* opt, rocksdb_compactionfilterfactory_t* factory) { opt->rep.compaction_filter_factory = std::shared_ptr(factory); } void rocksdb_options_set_comparator( rocksdb_options_t* opt, rocksdb_comparator_t* cmp) { opt->rep.comparator = cmp; } void rocksdb_options_set_merge_operator( rocksdb_options_t* opt, rocksdb_mergeoperator_t* merge_operator) { opt->rep.merge_operator = std::shared_ptr(merge_operator); } void rocksdb_options_set_create_if_missing( rocksdb_options_t* opt, unsigned char v) { opt->rep.create_if_missing = v; } void rocksdb_options_set_create_missing_column_families( rocksdb_options_t* opt, unsigned char v) { opt->rep.create_missing_column_families = v; } void rocksdb_options_set_error_if_exists( rocksdb_options_t* opt, unsigned char v) { opt->rep.error_if_exists = v; } void rocksdb_options_set_paranoid_checks( rocksdb_options_t* opt, unsigned char v) { opt->rep.paranoid_checks = v; } void rocksdb_options_set_env(rocksdb_options_t* opt, rocksdb_env_t* env) { opt->rep.env = (env ? env->rep : nullptr); } void rocksdb_options_set_info_log(rocksdb_options_t* opt, rocksdb_logger_t* l) { if (l) { opt->rep.info_log = l->rep; } } void rocksdb_options_set_info_log_level( rocksdb_options_t* opt, int v) { opt->rep.info_log_level = static_cast(v); } void rocksdb_options_set_db_write_buffer_size(rocksdb_options_t* opt, size_t s) { opt->rep.db_write_buffer_size = s; } void rocksdb_options_set_write_buffer_size(rocksdb_options_t* opt, size_t s) { opt->rep.write_buffer_size = s; } void rocksdb_options_set_max_open_files(rocksdb_options_t* opt, int n) { opt->rep.max_open_files = n; } void rocksdb_options_set_max_total_wal_size(rocksdb_options_t* opt, uint64_t n) { opt->rep.max_total_wal_size = n; } void rocksdb_options_set_target_file_size_base( rocksdb_options_t* opt, uint64_t n) { opt->rep.target_file_size_base = n; } void rocksdb_options_set_target_file_size_multiplier( rocksdb_options_t* opt, int n) { opt->rep.target_file_size_multiplier = n; } void rocksdb_options_set_max_bytes_for_level_base( rocksdb_options_t* opt, uint64_t n) { opt->rep.max_bytes_for_level_base = n; } void rocksdb_options_set_max_bytes_for_level_multiplier( rocksdb_options_t* opt, int n) { opt->rep.max_bytes_for_level_multiplier = n; } void rocksdb_options_set_expanded_compaction_factor( rocksdb_options_t* opt, int n) { opt->rep.expanded_compaction_factor = n; } void rocksdb_options_set_max_grandparent_overlap_factor( rocksdb_options_t* opt, int n) { opt->rep.max_grandparent_overlap_factor = n; } void rocksdb_options_set_max_bytes_for_level_multiplier_additional( rocksdb_options_t* opt, int* level_values, size_t num_levels) { opt->rep.max_bytes_for_level_multiplier_additional.resize(num_levels); for (size_t i = 0; i < num_levels; ++i) { opt->rep.max_bytes_for_level_multiplier_additional[i] = level_values[i]; } } void rocksdb_options_enable_statistics(rocksdb_options_t* opt) { opt->rep.statistics = rocksdb::CreateDBStatistics(); } void rocksdb_options_set_num_levels(rocksdb_options_t* opt, int n) { opt->rep.num_levels = n; } void rocksdb_options_set_level0_file_num_compaction_trigger( rocksdb_options_t* opt, int n) { opt->rep.level0_file_num_compaction_trigger = n; } void rocksdb_options_set_level0_slowdown_writes_trigger( rocksdb_options_t* opt, int n) { opt->rep.level0_slowdown_writes_trigger = n; } void rocksdb_options_set_level0_stop_writes_trigger( rocksdb_options_t* opt, int n) { opt->rep.level0_stop_writes_trigger = n; } void rocksdb_options_set_max_mem_compaction_level(rocksdb_options_t* opt, int n) {} void rocksdb_options_set_compression(rocksdb_options_t* opt, int t) { opt->rep.compression = static_cast(t); } void rocksdb_options_set_compression_per_level(rocksdb_options_t* opt, int* level_values, size_t num_levels) { opt->rep.compression_per_level.resize(num_levels); for (size_t i = 0; i < num_levels; ++i) { opt->rep.compression_per_level[i] = static_cast(level_values[i]); } } void rocksdb_options_set_compression_options( rocksdb_options_t* opt, int w_bits, int level, int strategy) { opt->rep.compression_opts.window_bits = w_bits; opt->rep.compression_opts.level = level; opt->rep.compression_opts.strategy = strategy; } void rocksdb_options_set_prefix_extractor( rocksdb_options_t* opt, rocksdb_slicetransform_t* prefix_extractor) { opt->rep.prefix_extractor.reset(prefix_extractor); } void rocksdb_options_set_disable_data_sync( rocksdb_options_t* opt, int disable_data_sync) { opt->rep.disableDataSync = disable_data_sync; } void rocksdb_options_set_use_fsync( rocksdb_options_t* opt, int use_fsync) { opt->rep.use_fsync = use_fsync; } void rocksdb_options_set_db_log_dir( rocksdb_options_t* opt, const char* db_log_dir) { opt->rep.db_log_dir = db_log_dir; } void rocksdb_options_set_wal_dir( rocksdb_options_t* opt, const char* v) { opt->rep.wal_dir = v; } void rocksdb_options_set_WAL_ttl_seconds(rocksdb_options_t* opt, uint64_t ttl) { opt->rep.WAL_ttl_seconds = ttl; } void rocksdb_options_set_WAL_size_limit_MB( rocksdb_options_t* opt, uint64_t limit) { opt->rep.WAL_size_limit_MB = limit; } void rocksdb_options_set_manifest_preallocation_size( rocksdb_options_t* opt, size_t v) { opt->rep.manifest_preallocation_size = v; } // noop void rocksdb_options_set_purge_redundant_kvs_while_flush(rocksdb_options_t* opt, unsigned char v) {} void rocksdb_options_set_allow_os_buffer(rocksdb_options_t* opt, unsigned char v) { opt->rep.allow_os_buffer = v; } void rocksdb_options_set_allow_mmap_reads( rocksdb_options_t* opt, unsigned char v) { opt->rep.allow_mmap_reads = v; } void rocksdb_options_set_allow_mmap_writes( rocksdb_options_t* opt, unsigned char v) { opt->rep.allow_mmap_writes = v; } void rocksdb_options_set_is_fd_close_on_exec( rocksdb_options_t* opt, unsigned char v) { opt->rep.is_fd_close_on_exec = v; } void rocksdb_options_set_skip_log_error_on_recovery( rocksdb_options_t* opt, unsigned char v) { opt->rep.skip_log_error_on_recovery = v; } void rocksdb_options_set_stats_dump_period_sec( rocksdb_options_t* opt, unsigned int v) { opt->rep.stats_dump_period_sec = v; } void rocksdb_options_set_advise_random_on_open( rocksdb_options_t* opt, unsigned char v) { opt->rep.advise_random_on_open = v; } void rocksdb_options_set_access_hint_on_compaction_start( rocksdb_options_t* opt, int v) { switch(v) { case 0: opt->rep.access_hint_on_compaction_start = rocksdb::Options::NONE; break; case 1: opt->rep.access_hint_on_compaction_start = rocksdb::Options::NORMAL; break; case 2: opt->rep.access_hint_on_compaction_start = rocksdb::Options::SEQUENTIAL; break; case 3: opt->rep.access_hint_on_compaction_start = rocksdb::Options::WILLNEED; break; } } void rocksdb_options_set_use_adaptive_mutex( rocksdb_options_t* opt, unsigned char v) { opt->rep.use_adaptive_mutex = v; } void rocksdb_options_set_bytes_per_sync( rocksdb_options_t* opt, uint64_t v) { opt->rep.bytes_per_sync = v; } void rocksdb_options_set_verify_checksums_in_compaction( rocksdb_options_t* opt, unsigned char v) { opt->rep.verify_checksums_in_compaction = v; } void rocksdb_options_set_filter_deletes( rocksdb_options_t* opt, unsigned char v) { opt->rep.filter_deletes = v; } void rocksdb_options_set_max_sequential_skip_in_iterations( rocksdb_options_t* opt, uint64_t v) { opt->rep.max_sequential_skip_in_iterations = v; } void rocksdb_options_set_max_write_buffer_number(rocksdb_options_t* opt, int n) { opt->rep.max_write_buffer_number = n; } void rocksdb_options_set_min_write_buffer_number_to_merge(rocksdb_options_t* opt, int n) { opt->rep.min_write_buffer_number_to_merge = n; } void rocksdb_options_set_max_write_buffer_number_to_maintain( rocksdb_options_t* opt, int n) { opt->rep.max_write_buffer_number_to_maintain = n; } void rocksdb_options_set_max_background_compactions(rocksdb_options_t* opt, int n) { opt->rep.max_background_compactions = n; } void rocksdb_options_set_max_background_flushes(rocksdb_options_t* opt, int n) { opt->rep.max_background_flushes = n; } void rocksdb_options_set_max_log_file_size(rocksdb_options_t* opt, size_t v) { opt->rep.max_log_file_size = v; } void rocksdb_options_set_log_file_time_to_roll(rocksdb_options_t* opt, size_t v) { opt->rep.log_file_time_to_roll = v; } void rocksdb_options_set_keep_log_file_num(rocksdb_options_t* opt, size_t v) { opt->rep.keep_log_file_num = v; } void rocksdb_options_set_recycle_log_file_num(rocksdb_options_t* opt, size_t v) { opt->rep.recycle_log_file_num = v; } void rocksdb_options_set_soft_rate_limit(rocksdb_options_t* opt, double v) { opt->rep.soft_rate_limit = v; } void rocksdb_options_set_hard_rate_limit(rocksdb_options_t* opt, double v) { opt->rep.hard_rate_limit = v; } void rocksdb_options_set_rate_limit_delay_max_milliseconds( rocksdb_options_t* opt, unsigned int v) { opt->rep.rate_limit_delay_max_milliseconds = v; } void rocksdb_options_set_max_manifest_file_size( rocksdb_options_t* opt, size_t v) { opt->rep.max_manifest_file_size = v; } void rocksdb_options_set_table_cache_numshardbits( rocksdb_options_t* opt, int v) { opt->rep.table_cache_numshardbits = v; } void rocksdb_options_set_table_cache_remove_scan_count_limit( rocksdb_options_t* opt, int v) { // this option is deprecated } void rocksdb_options_set_arena_block_size( rocksdb_options_t* opt, size_t v) { opt->rep.arena_block_size = v; } void rocksdb_options_set_disable_auto_compactions(rocksdb_options_t* opt, int disable) { opt->rep.disable_auto_compactions = disable; } void rocksdb_options_set_delete_obsolete_files_period_micros( rocksdb_options_t* opt, uint64_t v) { opt->rep.delete_obsolete_files_period_micros = v; } void rocksdb_options_set_source_compaction_factor( rocksdb_options_t* opt, int n) { opt->rep.expanded_compaction_factor = n; } void rocksdb_options_prepare_for_bulk_load(rocksdb_options_t* opt) { opt->rep.PrepareForBulkLoad(); } void rocksdb_options_set_memtable_vector_rep(rocksdb_options_t *opt) { static rocksdb::VectorRepFactory* factory = 0; if (!factory) { factory = new rocksdb::VectorRepFactory; } opt->rep.memtable_factory.reset(factory); } void rocksdb_options_set_memtable_prefix_bloom_bits( rocksdb_options_t* opt, uint32_t v) { opt->rep.memtable_prefix_bloom_bits = v; } void rocksdb_options_set_memtable_prefix_bloom_probes( rocksdb_options_t* opt, uint32_t v) { opt->rep.memtable_prefix_bloom_probes = v; } void rocksdb_options_set_memtable_prefix_bloom_huge_page_tlb_size( rocksdb_options_t* opt, size_t v) { opt->rep.memtable_prefix_bloom_huge_page_tlb_size = v; } void rocksdb_options_set_hash_skip_list_rep( rocksdb_options_t *opt, size_t bucket_count, int32_t skiplist_height, int32_t skiplist_branching_factor) { static rocksdb::MemTableRepFactory* factory = 0; if (!factory) { factory = rocksdb::NewHashSkipListRepFactory( bucket_count, skiplist_height, skiplist_branching_factor); } opt->rep.memtable_factory.reset(factory); } void rocksdb_options_set_hash_link_list_rep( rocksdb_options_t *opt, size_t bucket_count) { static rocksdb::MemTableRepFactory* factory = 0; if (!factory) { factory = rocksdb::NewHashLinkListRepFactory(bucket_count); } opt->rep.memtable_factory.reset(factory); } void rocksdb_options_set_plain_table_factory( rocksdb_options_t *opt, uint32_t user_key_len, int bloom_bits_per_key, double hash_table_ratio, size_t index_sparseness) { static rocksdb::TableFactory* factory = 0; if (!factory) { rocksdb::PlainTableOptions options; options.user_key_len = user_key_len; options.bloom_bits_per_key = bloom_bits_per_key; options.hash_table_ratio = hash_table_ratio; options.index_sparseness = index_sparseness; factory = rocksdb::NewPlainTableFactory(options); } opt->rep.table_factory.reset(factory); } void rocksdb_options_set_max_successive_merges( rocksdb_options_t* opt, size_t v) { opt->rep.max_successive_merges = v; } void rocksdb_options_set_min_partial_merge_operands( rocksdb_options_t* opt, uint32_t v) { opt->rep.min_partial_merge_operands = v; } void rocksdb_options_set_bloom_locality( rocksdb_options_t* opt, uint32_t v) { opt->rep.bloom_locality = v; } void rocksdb_options_set_inplace_update_support( rocksdb_options_t* opt, unsigned char v) { opt->rep.inplace_update_support = v; } void rocksdb_options_set_inplace_update_num_locks( rocksdb_options_t* opt, size_t v) { opt->rep.inplace_update_num_locks = v; } void rocksdb_options_set_compaction_style(rocksdb_options_t *opt, int style) { opt->rep.compaction_style = static_cast(style); } void rocksdb_options_set_universal_compaction_options(rocksdb_options_t *opt, rocksdb_universal_compaction_options_t *uco) { opt->rep.compaction_options_universal = *(uco->rep); } void rocksdb_options_set_fifo_compaction_options( rocksdb_options_t* opt, rocksdb_fifo_compaction_options_t* fifo) { opt->rep.compaction_options_fifo = fifo->rep; } char *rocksdb_options_statistics_get_string(rocksdb_options_t *opt) { rocksdb::Statistics *statistics = opt->rep.statistics.get(); if (statistics) { return strdup(statistics->ToString().c_str()); } return nullptr; } /* TODO: DB::OpenForReadOnly DB::KeyMayExist DB::GetOptions DB::GetSortedWalFiles DB::GetLatestSequenceNumber DB::GetUpdatesSince DB::GetDbIdentity DB::RunManualCompaction custom cache table_properties_collectors */ rocksdb_compactionfilter_t* rocksdb_compactionfilter_create( void* state, void (*destructor)(void*), unsigned char (*filter)( void*, int level, const char* key, size_t key_length, const char* existing_value, size_t value_length, char** new_value, size_t *new_value_length, unsigned char* value_changed), const char* (*name)(void*)) { rocksdb_compactionfilter_t* result = new rocksdb_compactionfilter_t; result->state_ = state; result->destructor_ = destructor; result->filter_ = filter; result->name_ = name; return result; } void rocksdb_compactionfilter_destroy(rocksdb_compactionfilter_t* filter) { delete filter; } unsigned char rocksdb_compactionfiltercontext_is_full_compaction( rocksdb_compactionfiltercontext_t* context) { return context->rep.is_full_compaction; } unsigned char rocksdb_compactionfiltercontext_is_manual_compaction( rocksdb_compactionfiltercontext_t* context) { return context->rep.is_manual_compaction; } rocksdb_compactionfilterfactory_t* rocksdb_compactionfilterfactory_create( void* state, void (*destructor)(void*), rocksdb_compactionfilter_t* (*create_compaction_filter)( void*, rocksdb_compactionfiltercontext_t* context), const char* (*name)(void*)) { rocksdb_compactionfilterfactory_t* result = new rocksdb_compactionfilterfactory_t; result->state_ = state; result->destructor_ = destructor; result->create_compaction_filter_ = create_compaction_filter; result->name_ = name; return result; } void rocksdb_compactionfilterfactory_destroy( rocksdb_compactionfilterfactory_t* factory) { delete factory; } rocksdb_comparator_t* rocksdb_comparator_create( void* state, void (*destructor)(void*), int (*compare)( void*, const char* a, size_t alen, const char* b, size_t blen), const char* (*name)(void*)) { rocksdb_comparator_t* result = new rocksdb_comparator_t; result->state_ = state; result->destructor_ = destructor; result->compare_ = compare; result->name_ = name; return result; } void rocksdb_comparator_destroy(rocksdb_comparator_t* cmp) { delete cmp; } rocksdb_filterpolicy_t* rocksdb_filterpolicy_create( void* state, void (*destructor)(void*), char* (*create_filter)( void*, const char* const* key_array, const size_t* key_length_array, int num_keys, size_t* filter_length), unsigned char (*key_may_match)( void*, const char* key, size_t length, const char* filter, size_t filter_length), void (*delete_filter)( void*, const char* filter, size_t filter_length), const char* (*name)(void*)) { rocksdb_filterpolicy_t* result = new rocksdb_filterpolicy_t; result->state_ = state; result->destructor_ = destructor; result->create_ = create_filter; result->key_match_ = key_may_match; result->delete_filter_ = delete_filter; result->name_ = name; return result; } void rocksdb_filterpolicy_destroy(rocksdb_filterpolicy_t* filter) { delete filter; } rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom_format(int bits_per_key, bool original_format) { // Make a rocksdb_filterpolicy_t, but override all of its methods so // they delegate to a NewBloomFilterPolicy() instead of user // supplied C functions. struct Wrapper : public rocksdb_filterpolicy_t { const FilterPolicy* rep_; ~Wrapper() { delete rep_; } const char* Name() const override { return rep_->Name(); } void CreateFilter(const Slice* keys, int n, std::string* dst) const override { return rep_->CreateFilter(keys, n, dst); } bool KeyMayMatch(const Slice& key, const Slice& filter) const override { return rep_->KeyMayMatch(key, filter); } static void DoNothing(void*) { } }; Wrapper* wrapper = new Wrapper; wrapper->rep_ = NewBloomFilterPolicy(bits_per_key, original_format); wrapper->state_ = nullptr; wrapper->delete_filter_ = nullptr; wrapper->destructor_ = &Wrapper::DoNothing; return wrapper; } rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom_full(int bits_per_key) { return rocksdb_filterpolicy_create_bloom_format(bits_per_key, false); } rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom(int bits_per_key) { return rocksdb_filterpolicy_create_bloom_format(bits_per_key, true); } rocksdb_mergeoperator_t* rocksdb_mergeoperator_create( void* state, void (*destructor)(void*), char* (*full_merge)(void*, const char* key, size_t key_length, const char* existing_value, size_t existing_value_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length), char* (*partial_merge)(void*, const char* key, size_t key_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length), void (*delete_value)(void*, const char* value, size_t value_length), const char* (*name)(void*)) { rocksdb_mergeoperator_t* result = new rocksdb_mergeoperator_t; result->state_ = state; result->destructor_ = destructor; result->full_merge_ = full_merge; result->partial_merge_ = partial_merge; result->delete_value_ = delete_value; result->name_ = name; return result; } void rocksdb_mergeoperator_destroy(rocksdb_mergeoperator_t* merge_operator) { delete merge_operator; } rocksdb_readoptions_t* rocksdb_readoptions_create() { return new rocksdb_readoptions_t; } void rocksdb_readoptions_destroy(rocksdb_readoptions_t* opt) { delete opt; } void rocksdb_readoptions_set_verify_checksums( rocksdb_readoptions_t* opt, unsigned char v) { opt->rep.verify_checksums = v; } void rocksdb_readoptions_set_fill_cache( rocksdb_readoptions_t* opt, unsigned char v) { opt->rep.fill_cache = v; } void rocksdb_readoptions_set_snapshot( rocksdb_readoptions_t* opt, const rocksdb_snapshot_t* snap) { opt->rep.snapshot = (snap ? snap->rep : nullptr); } void rocksdb_readoptions_set_iterate_upper_bound( rocksdb_readoptions_t* opt, const char* key, size_t keylen) { if (key == nullptr) { opt->upper_bound = Slice(); opt->rep.iterate_upper_bound = nullptr; } else { opt->upper_bound = Slice(key, keylen); opt->rep.iterate_upper_bound = &opt->upper_bound; } } void rocksdb_readoptions_set_read_tier( rocksdb_readoptions_t* opt, int v) { opt->rep.read_tier = static_cast(v); } void rocksdb_readoptions_set_tailing( rocksdb_readoptions_t* opt, unsigned char v) { opt->rep.tailing = v; } rocksdb_writeoptions_t* rocksdb_writeoptions_create() { return new rocksdb_writeoptions_t; } void rocksdb_writeoptions_destroy(rocksdb_writeoptions_t* opt) { delete opt; } void rocksdb_writeoptions_set_sync( rocksdb_writeoptions_t* opt, unsigned char v) { opt->rep.sync = v; } void rocksdb_writeoptions_disable_WAL(rocksdb_writeoptions_t* opt, int disable) { opt->rep.disableWAL = disable; } rocksdb_flushoptions_t* rocksdb_flushoptions_create() { return new rocksdb_flushoptions_t; } void rocksdb_flushoptions_destroy(rocksdb_flushoptions_t* opt) { delete opt; } void rocksdb_flushoptions_set_wait( rocksdb_flushoptions_t* opt, unsigned char v) { opt->rep.wait = v; } rocksdb_cache_t* rocksdb_cache_create_lru(size_t capacity) { rocksdb_cache_t* c = new rocksdb_cache_t; c->rep = NewLRUCache(capacity); return c; } void rocksdb_cache_destroy(rocksdb_cache_t* cache) { delete cache; } rocksdb_env_t* rocksdb_create_default_env() { rocksdb_env_t* result = new rocksdb_env_t; result->rep = Env::Default(); result->is_default = true; return result; } void rocksdb_env_set_background_threads(rocksdb_env_t* env, int n) { env->rep->SetBackgroundThreads(n); } void rocksdb_env_set_high_priority_background_threads(rocksdb_env_t* env, int n) { env->rep->SetBackgroundThreads(n, Env::HIGH); } void rocksdb_env_join_all_threads(rocksdb_env_t* env) { env->rep->WaitForJoin(); } void rocksdb_env_destroy(rocksdb_env_t* env) { if (!env->is_default) delete env->rep; delete env; } rocksdb_slicetransform_t* rocksdb_slicetransform_create( void* state, void (*destructor)(void*), char* (*transform)( void*, const char* key, size_t length, size_t* dst_length), unsigned char (*in_domain)( void*, const char* key, size_t length), unsigned char (*in_range)( void*, const char* key, size_t length), const char* (*name)(void*)) { rocksdb_slicetransform_t* result = new rocksdb_slicetransform_t; result->state_ = state; result->destructor_ = destructor; result->transform_ = transform; result->in_domain_ = in_domain; result->in_range_ = in_range; result->name_ = name; return result; } void rocksdb_slicetransform_destroy(rocksdb_slicetransform_t* st) { delete st; } rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t prefixLen) { struct Wrapper : public rocksdb_slicetransform_t { const SliceTransform* rep_; ~Wrapper() { delete rep_; } const char* Name() const override { return rep_->Name(); } Slice Transform(const Slice& src) const override { return rep_->Transform(src); } bool InDomain(const Slice& src) const override { return rep_->InDomain(src); } bool InRange(const Slice& src) const override { return rep_->InRange(src); } static void DoNothing(void*) { } }; Wrapper* wrapper = new Wrapper; wrapper->rep_ = rocksdb::NewFixedPrefixTransform(prefixLen); wrapper->state_ = nullptr; wrapper->destructor_ = &Wrapper::DoNothing; return wrapper; } rocksdb_slicetransform_t* rocksdb_slicetransform_create_noop() { struct Wrapper : public rocksdb_slicetransform_t { const SliceTransform* rep_; ~Wrapper() { delete rep_; } const char* Name() const override { return rep_->Name(); } Slice Transform(const Slice& src) const override { return rep_->Transform(src); } bool InDomain(const Slice& src) const override { return rep_->InDomain(src); } bool InRange(const Slice& src) const override { return rep_->InRange(src); } static void DoNothing(void*) { } }; Wrapper* wrapper = new Wrapper; wrapper->rep_ = rocksdb::NewNoopTransform(); wrapper->state_ = nullptr; wrapper->destructor_ = &Wrapper::DoNothing; return wrapper; } rocksdb_universal_compaction_options_t* rocksdb_universal_compaction_options_create() { rocksdb_universal_compaction_options_t* result = new rocksdb_universal_compaction_options_t; result->rep = new rocksdb::CompactionOptionsUniversal; return result; } void rocksdb_universal_compaction_options_set_size_ratio( rocksdb_universal_compaction_options_t* uco, int ratio) { uco->rep->size_ratio = ratio; } void rocksdb_universal_compaction_options_set_min_merge_width( rocksdb_universal_compaction_options_t* uco, int w) { uco->rep->min_merge_width = w; } void rocksdb_universal_compaction_options_set_max_merge_width( rocksdb_universal_compaction_options_t* uco, int w) { uco->rep->max_merge_width = w; } void rocksdb_universal_compaction_options_set_max_size_amplification_percent( rocksdb_universal_compaction_options_t* uco, int p) { uco->rep->max_size_amplification_percent = p; } void rocksdb_universal_compaction_options_set_compression_size_percent( rocksdb_universal_compaction_options_t* uco, int p) { uco->rep->compression_size_percent = p; } void rocksdb_universal_compaction_options_set_stop_style( rocksdb_universal_compaction_options_t* uco, int style) { uco->rep->stop_style = static_cast(style); } void rocksdb_universal_compaction_options_destroy( rocksdb_universal_compaction_options_t* uco) { delete uco->rep; delete uco; } rocksdb_fifo_compaction_options_t* rocksdb_fifo_compaction_options_create() { rocksdb_fifo_compaction_options_t* result = new rocksdb_fifo_compaction_options_t; result->rep = CompactionOptionsFIFO(); return result; } void rocksdb_fifo_compaction_options_set_max_table_files_size( rocksdb_fifo_compaction_options_t* fifo_opts, uint64_t size) { fifo_opts->rep.max_table_files_size = size; } void rocksdb_fifo_compaction_options_destroy( rocksdb_fifo_compaction_options_t* fifo_opts) { delete fifo_opts; } void rocksdb_options_set_min_level_to_compress(rocksdb_options_t* opt, int level) { if (level >= 0) { assert(level <= opt->rep.num_levels); opt->rep.compression_per_level.resize(opt->rep.num_levels); for (int i = 0; i < level; i++) { opt->rep.compression_per_level[i] = rocksdb::kNoCompression; } for (int i = level; i < opt->rep.num_levels; i++) { opt->rep.compression_per_level[i] = opt->rep.compression; } } } int rocksdb_livefiles_count( const rocksdb_livefiles_t* lf) { return static_cast(lf->rep.size()); } const char* rocksdb_livefiles_name( const rocksdb_livefiles_t* lf, int index) { return lf->rep[index].name.c_str(); } int rocksdb_livefiles_level( const rocksdb_livefiles_t* lf, int index) { return lf->rep[index].level; } size_t rocksdb_livefiles_size( const rocksdb_livefiles_t* lf, int index) { return lf->rep[index].size; } const char* rocksdb_livefiles_smallestkey( const rocksdb_livefiles_t* lf, int index, size_t* size) { *size = lf->rep[index].smallestkey.size(); return lf->rep[index].smallestkey.data(); } const char* rocksdb_livefiles_largestkey( const rocksdb_livefiles_t* lf, int index, size_t* size) { *size = lf->rep[index].largestkey.size(); return lf->rep[index].largestkey.data(); } extern void rocksdb_livefiles_destroy( const rocksdb_livefiles_t* lf) { delete lf; } void rocksdb_get_options_from_string(const rocksdb_options_t* base_options, const char* opts_str, rocksdb_options_t* new_options, char** errptr) { SaveError(errptr, GetOptionsFromString(base_options->rep, std::string(opts_str), &new_options->rep)); } void rocksdb_free(void* ptr) { free(ptr); } } // end extern "C" #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/c_test.c000066400000000000000000000723111267530634000151740ustar00rootroot00000000000000/* Copyright (c) 2011 The LevelDB Authors. All rights reserved. Use of this source code is governed by a BSD-style license that can be found in the LICENSE file. See the AUTHORS file for names of contributors. */ #ifndef ROCKSDB_LITE // Lite does not support C API #include "rocksdb/c.h" #include #include #include #include #include #ifndef OS_WIN # include #endif #include // Can not use port/port.h macros as this is a c file #ifdef OS_WIN #include # define snprintf _snprintf // Ok for uniqueness int geteuid() { int result = 0; result = ((int)GetCurrentProcessId() << 16); result |= (int)GetCurrentThreadId(); return result; } #endif const char* phase = ""; static char dbname[200]; static char dbbackupname[200]; static void StartPhase(const char* name) { fprintf(stderr, "=== Test %s\n", name); phase = name; } static const char* GetTempDir(void) { const char* ret = getenv("TEST_TMPDIR"); if (ret == NULL || ret[0] == '\0') ret = "/tmp"; return ret; } #define CheckNoError(err) \ if ((err) != NULL) { \ fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, (err)); \ abort(); \ } #define CheckCondition(cond) \ if (!(cond)) { \ fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, #cond); \ abort(); \ } static void CheckEqual(const char* expected, const char* v, size_t n) { if (expected == NULL && v == NULL) { // ok } else if (expected != NULL && v != NULL && n == strlen(expected) && memcmp(expected, v, n) == 0) { // ok return; } else { fprintf(stderr, "%s: expected '%s', got '%s'\n", phase, (expected ? expected : "(null)"), (v ? v : "(null")); abort(); } } static void Free(char** ptr) { if (*ptr) { free(*ptr); *ptr = NULL; } } static void CheckGet( rocksdb_t* db, const rocksdb_readoptions_t* options, const char* key, const char* expected) { char* err = NULL; size_t val_len; char* val; val = rocksdb_get(db, options, key, strlen(key), &val_len, &err); CheckNoError(err); CheckEqual(expected, val, val_len); Free(&val); } static void CheckGetCF( rocksdb_t* db, const rocksdb_readoptions_t* options, rocksdb_column_family_handle_t* handle, const char* key, const char* expected) { char* err = NULL; size_t val_len; char* val; val = rocksdb_get_cf(db, options, handle, key, strlen(key), &val_len, &err); CheckNoError(err); CheckEqual(expected, val, val_len); Free(&val); } static void CheckIter(rocksdb_iterator_t* iter, const char* key, const char* val) { size_t len; const char* str; str = rocksdb_iter_key(iter, &len); CheckEqual(key, str, len); str = rocksdb_iter_value(iter, &len); CheckEqual(val, str, len); } // Callback from rocksdb_writebatch_iterate() static void CheckPut(void* ptr, const char* k, size_t klen, const char* v, size_t vlen) { int* state = (int*) ptr; CheckCondition(*state < 2); switch (*state) { case 0: CheckEqual("bar", k, klen); CheckEqual("b", v, vlen); break; case 1: CheckEqual("box", k, klen); CheckEqual("c", v, vlen); break; } (*state)++; } // Callback from rocksdb_writebatch_iterate() static void CheckDel(void* ptr, const char* k, size_t klen) { int* state = (int*) ptr; CheckCondition(*state == 2); CheckEqual("bar", k, klen); (*state)++; } static void CmpDestroy(void* arg) { } static int CmpCompare(void* arg, const char* a, size_t alen, const char* b, size_t blen) { size_t n = (alen < blen) ? alen : blen; int r = memcmp(a, b, n); if (r == 0) { if (alen < blen) r = -1; else if (alen > blen) r = +1; } return r; } static const char* CmpName(void* arg) { return "foo"; } // Custom filter policy static unsigned char fake_filter_result = 1; static void FilterDestroy(void* arg) { } static const char* FilterName(void* arg) { return "TestFilter"; } static char* FilterCreate( void* arg, const char* const* key_array, const size_t* key_length_array, int num_keys, size_t* filter_length) { *filter_length = 4; char* result = malloc(4); memcpy(result, "fake", 4); return result; } static unsigned char FilterKeyMatch( void* arg, const char* key, size_t length, const char* filter, size_t filter_length) { CheckCondition(filter_length == 4); CheckCondition(memcmp(filter, "fake", 4) == 0); return fake_filter_result; } // Custom compaction filter static void CFilterDestroy(void* arg) {} static const char* CFilterName(void* arg) { return "foo"; } static unsigned char CFilterFilter(void* arg, int level, const char* key, size_t key_length, const char* existing_value, size_t value_length, char** new_value, size_t* new_value_length, unsigned char* value_changed) { if (key_length == 3) { if (memcmp(key, "bar", key_length) == 0) { return 1; } else if (memcmp(key, "baz", key_length) == 0) { *value_changed = 1; *new_value = "newbazvalue"; *new_value_length = 11; return 0; } } return 0; } static void CFilterFactoryDestroy(void* arg) {} static const char* CFilterFactoryName(void* arg) { return "foo"; } static rocksdb_compactionfilter_t* CFilterCreate( void* arg, rocksdb_compactionfiltercontext_t* context) { return rocksdb_compactionfilter_create(NULL, CFilterDestroy, CFilterFilter, CFilterName); } static rocksdb_t* CheckCompaction(rocksdb_t* db, rocksdb_options_t* options, rocksdb_readoptions_t* roptions, rocksdb_writeoptions_t* woptions) { char* err = NULL; db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo", 3, "foovalue", 8, &err); CheckNoError(err); CheckGet(db, roptions, "foo", "foovalue"); rocksdb_put(db, woptions, "bar", 3, "barvalue", 8, &err); CheckNoError(err); CheckGet(db, roptions, "bar", "barvalue"); rocksdb_put(db, woptions, "baz", 3, "bazvalue", 8, &err); CheckNoError(err); CheckGet(db, roptions, "baz", "bazvalue"); // Force compaction rocksdb_compact_range(db, NULL, 0, NULL, 0); // should have filtered bar, but not foo CheckGet(db, roptions, "foo", "foovalue"); CheckGet(db, roptions, "bar", NULL); CheckGet(db, roptions, "baz", "newbazvalue"); return db; } // Custom merge operator static void MergeOperatorDestroy(void* arg) { } static const char* MergeOperatorName(void* arg) { return "TestMergeOperator"; } static char* MergeOperatorFullMerge( void* arg, const char* key, size_t key_length, const char* existing_value, size_t existing_value_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length) { *new_value_length = 4; *success = 1; char* result = malloc(4); memcpy(result, "fake", 4); return result; } static char* MergeOperatorPartialMerge( void* arg, const char* key, size_t key_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length) { *new_value_length = 4; *success = 1; char* result = malloc(4); memcpy(result, "fake", 4); return result; } int main(int argc, char** argv) { rocksdb_t* db; rocksdb_comparator_t* cmp; rocksdb_cache_t* cache; rocksdb_env_t* env; rocksdb_options_t* options; rocksdb_block_based_table_options_t* table_options; rocksdb_readoptions_t* roptions; rocksdb_writeoptions_t* woptions; char* err = NULL; int run = -1; snprintf(dbname, sizeof(dbname), "%s/rocksdb_c_test-%d", GetTempDir(), ((int) geteuid())); snprintf(dbbackupname, sizeof(dbbackupname), "%s/rocksdb_c_test-%d-backup", GetTempDir(), ((int) geteuid())); StartPhase("create_objects"); cmp = rocksdb_comparator_create(NULL, CmpDestroy, CmpCompare, CmpName); env = rocksdb_create_default_env(); cache = rocksdb_cache_create_lru(100000); options = rocksdb_options_create(); rocksdb_options_set_comparator(options, cmp); rocksdb_options_set_error_if_exists(options, 1); rocksdb_options_set_env(options, env); rocksdb_options_set_info_log(options, NULL); rocksdb_options_set_write_buffer_size(options, 100000); rocksdb_options_set_paranoid_checks(options, 1); rocksdb_options_set_max_open_files(options, 10); table_options = rocksdb_block_based_options_create(); rocksdb_block_based_options_set_block_cache(table_options, cache); rocksdb_options_set_block_based_table_factory(options, table_options); rocksdb_options_set_compression(options, rocksdb_no_compression); rocksdb_options_set_compression_options(options, -14, -1, 0); int compression_levels[] = {rocksdb_no_compression, rocksdb_no_compression, rocksdb_no_compression, rocksdb_no_compression}; rocksdb_options_set_compression_per_level(options, compression_levels, 4); roptions = rocksdb_readoptions_create(); rocksdb_readoptions_set_verify_checksums(roptions, 1); rocksdb_readoptions_set_fill_cache(roptions, 0); woptions = rocksdb_writeoptions_create(); rocksdb_writeoptions_set_sync(woptions, 1); StartPhase("destroy"); rocksdb_destroy_db(options, dbname, &err); Free(&err); StartPhase("open_error"); rocksdb_open(options, dbname, &err); CheckCondition(err != NULL); Free(&err); StartPhase("open"); rocksdb_options_set_create_if_missing(options, 1); db = rocksdb_open(options, dbname, &err); CheckNoError(err); CheckGet(db, roptions, "foo", NULL); StartPhase("put"); rocksdb_put(db, woptions, "foo", 3, "hello", 5, &err); CheckNoError(err); CheckGet(db, roptions, "foo", "hello"); StartPhase("backup_and_restore"); { rocksdb_destroy_db(options, dbbackupname, &err); CheckNoError(err); rocksdb_backup_engine_t *be = rocksdb_backup_engine_open(options, dbbackupname, &err); CheckNoError(err); rocksdb_backup_engine_create_new_backup(be, db, &err); CheckNoError(err); rocksdb_delete(db, woptions, "foo", 3, &err); CheckNoError(err); rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); CheckNoError(err); rocksdb_restore_options_t *restore_options = rocksdb_restore_options_create(); rocksdb_restore_options_set_keep_log_files(restore_options, 0); rocksdb_backup_engine_restore_db_from_latest_backup(be, dbname, dbname, restore_options, &err); CheckNoError(err); rocksdb_restore_options_destroy(restore_options); rocksdb_options_set_error_if_exists(options, 0); db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_options_set_error_if_exists(options, 1); CheckGet(db, roptions, "foo", "hello"); rocksdb_backup_engine_close(be); } StartPhase("compactall"); rocksdb_compact_range(db, NULL, 0, NULL, 0); CheckGet(db, roptions, "foo", "hello"); StartPhase("compactrange"); rocksdb_compact_range(db, "a", 1, "z", 1); CheckGet(db, roptions, "foo", "hello"); StartPhase("writebatch"); { rocksdb_writebatch_t* wb = rocksdb_writebatch_create(); rocksdb_writebatch_put(wb, "foo", 3, "a", 1); rocksdb_writebatch_clear(wb); rocksdb_writebatch_put(wb, "bar", 3, "b", 1); rocksdb_writebatch_put(wb, "box", 3, "c", 1); rocksdb_writebatch_delete(wb, "bar", 3); rocksdb_write(db, woptions, wb, &err); CheckNoError(err); CheckGet(db, roptions, "foo", "hello"); CheckGet(db, roptions, "bar", NULL); CheckGet(db, roptions, "box", "c"); int pos = 0; rocksdb_writebatch_iterate(wb, &pos, CheckPut, CheckDel); CheckCondition(pos == 3); rocksdb_writebatch_destroy(wb); } StartPhase("writebatch_vectors"); { rocksdb_writebatch_t* wb = rocksdb_writebatch_create(); const char* k_list[2] = { "z", "ap" }; const size_t k_sizes[2] = { 1, 2 }; const char* v_list[3] = { "x", "y", "z" }; const size_t v_sizes[3] = { 1, 1, 1 }; rocksdb_writebatch_putv(wb, 2, k_list, k_sizes, 3, v_list, v_sizes); rocksdb_write(db, woptions, wb, &err); CheckNoError(err); CheckGet(db, roptions, "zap", "xyz"); rocksdb_writebatch_delete(wb, "zap", 3); rocksdb_write(db, woptions, wb, &err); CheckNoError(err); CheckGet(db, roptions, "zap", NULL); rocksdb_writebatch_destroy(wb); } StartPhase("writebatch_rep"); { rocksdb_writebatch_t* wb1 = rocksdb_writebatch_create(); rocksdb_writebatch_put(wb1, "baz", 3, "d", 1); rocksdb_writebatch_put(wb1, "quux", 4, "e", 1); rocksdb_writebatch_delete(wb1, "quux", 4); size_t repsize1 = 0; const char* rep = rocksdb_writebatch_data(wb1, &repsize1); rocksdb_writebatch_t* wb2 = rocksdb_writebatch_create_from(rep, repsize1); CheckCondition(rocksdb_writebatch_count(wb1) == rocksdb_writebatch_count(wb2)); size_t repsize2 = 0; CheckCondition( memcmp(rep, rocksdb_writebatch_data(wb2, &repsize2), repsize1) == 0); rocksdb_writebatch_destroy(wb1); rocksdb_writebatch_destroy(wb2); } StartPhase("iter"); { rocksdb_iterator_t* iter = rocksdb_create_iterator(db, roptions); CheckCondition(!rocksdb_iter_valid(iter)); rocksdb_iter_seek_to_first(iter); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "box", "c"); rocksdb_iter_next(iter); CheckIter(iter, "foo", "hello"); rocksdb_iter_prev(iter); CheckIter(iter, "box", "c"); rocksdb_iter_prev(iter); CheckCondition(!rocksdb_iter_valid(iter)); rocksdb_iter_seek_to_last(iter); CheckIter(iter, "foo", "hello"); rocksdb_iter_seek(iter, "b", 1); CheckIter(iter, "box", "c"); rocksdb_iter_get_error(iter, &err); CheckNoError(err); rocksdb_iter_destroy(iter); } StartPhase("multiget"); { const char* keys[3] = { "box", "foo", "notfound" }; const size_t keys_sizes[3] = { 3, 3, 8 }; char* vals[3]; size_t vals_sizes[3]; char* errs[3]; rocksdb_multi_get(db, roptions, 3, keys, keys_sizes, vals, vals_sizes, errs); int i; for (i = 0; i < 3; i++) { CheckEqual(NULL, errs[i], 0); switch (i) { case 0: CheckEqual("c", vals[i], vals_sizes[i]); break; case 1: CheckEqual("hello", vals[i], vals_sizes[i]); break; case 2: CheckEqual(NULL, vals[i], vals_sizes[i]); break; } Free(&vals[i]); } } StartPhase("approximate_sizes"); { int i; int n = 20000; char keybuf[100]; char valbuf[100]; uint64_t sizes[2]; const char* start[2] = { "a", "k00000000000000010000" }; size_t start_len[2] = { 1, 21 }; const char* limit[2] = { "k00000000000000010000", "z" }; size_t limit_len[2] = { 21, 1 }; rocksdb_writeoptions_set_sync(woptions, 0); for (i = 0; i < n; i++) { snprintf(keybuf, sizeof(keybuf), "k%020d", i); snprintf(valbuf, sizeof(valbuf), "v%020d", i); rocksdb_put(db, woptions, keybuf, strlen(keybuf), valbuf, strlen(valbuf), &err); CheckNoError(err); } rocksdb_approximate_sizes(db, 2, start, start_len, limit, limit_len, sizes); CheckCondition(sizes[0] > 0); CheckCondition(sizes[1] > 0); } StartPhase("property"); { char* prop = rocksdb_property_value(db, "nosuchprop"); CheckCondition(prop == NULL); prop = rocksdb_property_value(db, "rocksdb.stats"); CheckCondition(prop != NULL); Free(&prop); } StartPhase("snapshot"); { const rocksdb_snapshot_t* snap; snap = rocksdb_create_snapshot(db); rocksdb_delete(db, woptions, "foo", 3, &err); CheckNoError(err); rocksdb_readoptions_set_snapshot(roptions, snap); CheckGet(db, roptions, "foo", "hello"); rocksdb_readoptions_set_snapshot(roptions, NULL); CheckGet(db, roptions, "foo", NULL); rocksdb_release_snapshot(db, snap); } StartPhase("repair"); { // If we do not compact here, then the lazy deletion of // files (https://reviews.facebook.net/D6123) would leave // around deleted files and the repair process will find // those files and put them back into the database. rocksdb_compact_range(db, NULL, 0, NULL, 0); rocksdb_close(db); rocksdb_options_set_create_if_missing(options, 0); rocksdb_options_set_error_if_exists(options, 0); rocksdb_repair_db(options, dbname, &err); CheckNoError(err); db = rocksdb_open(options, dbname, &err); CheckNoError(err); CheckGet(db, roptions, "foo", NULL); CheckGet(db, roptions, "bar", NULL); CheckGet(db, roptions, "box", "c"); rocksdb_options_set_create_if_missing(options, 1); rocksdb_options_set_error_if_exists(options, 1); } StartPhase("filter"); for (run = 0; run < 2; run++) { // First run uses custom filter, second run uses bloom filter CheckNoError(err); rocksdb_filterpolicy_t* policy; if (run == 0) { policy = rocksdb_filterpolicy_create( NULL, FilterDestroy, FilterCreate, FilterKeyMatch, NULL, FilterName); } else { policy = rocksdb_filterpolicy_create_bloom(10); } rocksdb_block_based_options_set_filter_policy(table_options, policy); // Create new database rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); rocksdb_options_set_block_based_table_factory(options, table_options); db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo", 3, "foovalue", 8, &err); CheckNoError(err); rocksdb_put(db, woptions, "bar", 3, "barvalue", 8, &err); CheckNoError(err); rocksdb_compact_range(db, NULL, 0, NULL, 0); fake_filter_result = 1; CheckGet(db, roptions, "foo", "foovalue"); CheckGet(db, roptions, "bar", "barvalue"); if (phase == 0) { // Must not find value when custom filter returns false fake_filter_result = 0; CheckGet(db, roptions, "foo", NULL); CheckGet(db, roptions, "bar", NULL); fake_filter_result = 1; CheckGet(db, roptions, "foo", "foovalue"); CheckGet(db, roptions, "bar", "barvalue"); } // Reset the policy rocksdb_block_based_options_set_filter_policy(table_options, NULL); rocksdb_options_set_block_based_table_factory(options, table_options); } StartPhase("compaction_filter"); { rocksdb_options_t* options_with_filter = rocksdb_options_create(); rocksdb_options_set_create_if_missing(options_with_filter, 1); rocksdb_compactionfilter_t* cfilter; cfilter = rocksdb_compactionfilter_create(NULL, CFilterDestroy, CFilterFilter, CFilterName); // Create new database rocksdb_close(db); rocksdb_destroy_db(options_with_filter, dbname, &err); rocksdb_options_set_compaction_filter(options_with_filter, cfilter); db = CheckCompaction(db, options_with_filter, roptions, woptions); rocksdb_options_set_compaction_filter(options_with_filter, NULL); rocksdb_compactionfilter_destroy(cfilter); rocksdb_options_destroy(options_with_filter); } StartPhase("compaction_filter_factory"); { rocksdb_options_t* options_with_filter_factory = rocksdb_options_create(); rocksdb_options_set_create_if_missing(options_with_filter_factory, 1); rocksdb_compactionfilterfactory_t* factory; factory = rocksdb_compactionfilterfactory_create( NULL, CFilterFactoryDestroy, CFilterCreate, CFilterFactoryName); // Create new database rocksdb_close(db); rocksdb_destroy_db(options_with_filter_factory, dbname, &err); rocksdb_options_set_compaction_filter_factory(options_with_filter_factory, factory); db = CheckCompaction(db, options_with_filter_factory, roptions, woptions); rocksdb_options_set_compaction_filter_factory( options_with_filter_factory, NULL); rocksdb_options_destroy(options_with_filter_factory); } StartPhase("merge_operator"); { rocksdb_mergeoperator_t* merge_operator; merge_operator = rocksdb_mergeoperator_create( NULL, MergeOperatorDestroy, MergeOperatorFullMerge, MergeOperatorPartialMerge, NULL, MergeOperatorName); // Create new database rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); rocksdb_options_set_merge_operator(options, merge_operator); db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo", 3, "foovalue", 8, &err); CheckNoError(err); CheckGet(db, roptions, "foo", "foovalue"); rocksdb_merge(db, woptions, "foo", 3, "barvalue", 8, &err); CheckNoError(err); CheckGet(db, roptions, "foo", "fake"); // Merge of a non-existing value rocksdb_merge(db, woptions, "bar", 3, "barvalue", 8, &err); CheckNoError(err); CheckGet(db, roptions, "bar", "fake"); } StartPhase("columnfamilies"); { rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); CheckNoError(err) rocksdb_options_t* db_options = rocksdb_options_create(); rocksdb_options_set_create_if_missing(db_options, 1); db = rocksdb_open(db_options, dbname, &err); CheckNoError(err) rocksdb_column_family_handle_t* cfh; cfh = rocksdb_create_column_family(db, db_options, "cf1", &err); rocksdb_column_family_handle_destroy(cfh); CheckNoError(err); rocksdb_close(db); size_t cflen; char** column_fams = rocksdb_list_column_families(db_options, dbname, &cflen, &err); CheckNoError(err); CheckEqual("default", column_fams[0], 7); CheckEqual("cf1", column_fams[1], 3); CheckCondition(cflen == 2); rocksdb_list_column_families_destroy(column_fams, cflen); rocksdb_options_t* cf_options = rocksdb_options_create(); const char* cf_names[2] = {"default", "cf1"}; const rocksdb_options_t* cf_opts[2] = {cf_options, cf_options}; rocksdb_column_family_handle_t* handles[2]; db = rocksdb_open_column_families(db_options, dbname, 2, cf_names, cf_opts, handles, &err); CheckNoError(err); rocksdb_put_cf(db, woptions, handles[1], "foo", 3, "hello", 5, &err); CheckNoError(err); CheckGetCF(db, roptions, handles[1], "foo", "hello"); rocksdb_delete_cf(db, woptions, handles[1], "foo", 3, &err); CheckNoError(err); CheckGetCF(db, roptions, handles[1], "foo", NULL); rocksdb_writebatch_t* wb = rocksdb_writebatch_create(); rocksdb_writebatch_put_cf(wb, handles[1], "baz", 3, "a", 1); rocksdb_writebatch_clear(wb); rocksdb_writebatch_put_cf(wb, handles[1], "bar", 3, "b", 1); rocksdb_writebatch_put_cf(wb, handles[1], "box", 3, "c", 1); rocksdb_writebatch_delete_cf(wb, handles[1], "bar", 3); rocksdb_write(db, woptions, wb, &err); CheckNoError(err); CheckGetCF(db, roptions, handles[1], "baz", NULL); CheckGetCF(db, roptions, handles[1], "bar", NULL); CheckGetCF(db, roptions, handles[1], "box", "c"); rocksdb_writebatch_destroy(wb); const char* keys[3] = { "box", "box", "barfooxx" }; const rocksdb_column_family_handle_t* get_handles[3] = { handles[0], handles[1], handles[1] }; const size_t keys_sizes[3] = { 3, 3, 8 }; char* vals[3]; size_t vals_sizes[3]; char* errs[3]; rocksdb_multi_get_cf(db, roptions, get_handles, 3, keys, keys_sizes, vals, vals_sizes, errs); int i; for (i = 0; i < 3; i++) { CheckEqual(NULL, errs[i], 0); switch (i) { case 0: CheckEqual(NULL, vals[i], vals_sizes[i]); // wrong cf break; case 1: CheckEqual("c", vals[i], vals_sizes[i]); // bingo break; case 2: CheckEqual(NULL, vals[i], vals_sizes[i]); // normal not found break; } Free(&vals[i]); } rocksdb_iterator_t* iter = rocksdb_create_iterator_cf(db, roptions, handles[1]); CheckCondition(!rocksdb_iter_valid(iter)); rocksdb_iter_seek_to_first(iter); CheckCondition(rocksdb_iter_valid(iter)); for (i = 0; rocksdb_iter_valid(iter) != 0; rocksdb_iter_next(iter)) { i++; } CheckCondition(i == 1); rocksdb_iter_get_error(iter, &err); CheckNoError(err); rocksdb_iter_destroy(iter); rocksdb_drop_column_family(db, handles[1], &err); CheckNoError(err); for (i = 0; i < 2; i++) { rocksdb_column_family_handle_destroy(handles[i]); } rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); rocksdb_options_destroy(db_options); rocksdb_options_destroy(cf_options); } StartPhase("prefix"); { // Create new database rocksdb_options_set_allow_mmap_reads(options, 1); rocksdb_options_set_prefix_extractor(options, rocksdb_slicetransform_create_fixed_prefix(3)); rocksdb_options_set_hash_skip_list_rep(options, 5000, 4, 4); rocksdb_options_set_plain_table_factory(options, 4, 10, 0.75, 16); db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo1", 4, "foo", 3, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo2", 4, "foo", 3, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo3", 4, "foo", 3, &err); CheckNoError(err); rocksdb_put(db, woptions, "bar1", 4, "bar", 3, &err); CheckNoError(err); rocksdb_put(db, woptions, "bar2", 4, "bar", 3, &err); CheckNoError(err); rocksdb_put(db, woptions, "bar3", 4, "bar", 3, &err); CheckNoError(err); rocksdb_iterator_t* iter = rocksdb_create_iterator(db, roptions); CheckCondition(!rocksdb_iter_valid(iter)); rocksdb_iter_seek(iter, "bar", 3); rocksdb_iter_get_error(iter, &err); CheckNoError(err); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "bar1", "bar"); rocksdb_iter_next(iter); CheckIter(iter, "bar2", "bar"); rocksdb_iter_next(iter); CheckIter(iter, "bar3", "bar"); rocksdb_iter_get_error(iter, &err); CheckNoError(err); rocksdb_iter_destroy(iter); rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); } StartPhase("cuckoo_options"); { rocksdb_cuckoo_table_options_t* cuckoo_options; cuckoo_options = rocksdb_cuckoo_options_create(); rocksdb_cuckoo_options_set_hash_ratio(cuckoo_options, 0.5); rocksdb_cuckoo_options_set_max_search_depth(cuckoo_options, 200); rocksdb_cuckoo_options_set_cuckoo_block_size(cuckoo_options, 10); rocksdb_cuckoo_options_set_identity_as_first_hash(cuckoo_options, 1); rocksdb_cuckoo_options_set_use_module_hash(cuckoo_options, 0); rocksdb_options_set_cuckoo_table_factory(options, cuckoo_options); db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_cuckoo_options_destroy(cuckoo_options); } StartPhase("iterate_upper_bound"); { // Create new empty database rocksdb_close(db); rocksdb_destroy_db(options, dbname, &err); CheckNoError(err); rocksdb_options_set_prefix_extractor(options, NULL); db = rocksdb_open(options, dbname, &err); CheckNoError(err); rocksdb_put(db, woptions, "a", 1, "0", 1, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo", 3, "bar", 3, &err); CheckNoError(err); rocksdb_put(db, woptions, "foo1", 4, "bar1", 4, &err); CheckNoError(err); rocksdb_put(db, woptions, "g1", 2, "0", 1, &err); CheckNoError(err); // testing basic case with no iterate_upper_bound and no prefix_extractor { rocksdb_readoptions_set_iterate_upper_bound(roptions, NULL, 0); rocksdb_iterator_t* iter = rocksdb_create_iterator(db, roptions); rocksdb_iter_seek(iter, "foo", 3); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "foo", "bar"); rocksdb_iter_next(iter); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "foo1", "bar1"); rocksdb_iter_next(iter); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "g1", "0"); rocksdb_iter_destroy(iter); } // testing iterate_upper_bound and forward iterator // to make sure it stops at bound { // iterate_upper_bound points beyond the last expected entry rocksdb_readoptions_set_iterate_upper_bound(roptions, "foo2", 4); rocksdb_iterator_t* iter = rocksdb_create_iterator(db, roptions); rocksdb_iter_seek(iter, "foo", 3); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "foo", "bar"); rocksdb_iter_next(iter); CheckCondition(rocksdb_iter_valid(iter)); CheckIter(iter, "foo1", "bar1"); rocksdb_iter_next(iter); // should stop here... CheckCondition(!rocksdb_iter_valid(iter)); rocksdb_iter_destroy(iter); } } StartPhase("cleanup"); rocksdb_close(db); rocksdb_options_destroy(options); rocksdb_block_based_options_destroy(table_options); rocksdb_readoptions_destroy(roptions); rocksdb_writeoptions_destroy(woptions); rocksdb_cache_destroy(cache); rocksdb_comparator_destroy(cmp); rocksdb_env_destroy(env); fprintf(stderr, "PASS\n"); return 0; } #else #include int main() { fprintf(stderr, "SKIPPED\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/column_family.cc000066400000000000000000001132011267530634000167060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/column_family.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include "db/compaction_picker.h" #include "db/db_impl.h" #include "db/internal_stats.h" #include "db/job_context.h" #include "db/table_properties_collector.h" #include "db/version_set.h" #include "db/write_controller.h" #include "db/writebuffer.h" #include "memtable/hash_skiplist_rep.h" #include "util/autovector.h" #include "util/compression.h" #include "util/options_helper.h" #include "util/thread_status_util.h" #include "util/xfunc.h" namespace rocksdb { ColumnFamilyHandleImpl::ColumnFamilyHandleImpl( ColumnFamilyData* column_family_data, DBImpl* db, InstrumentedMutex* mutex) : cfd_(column_family_data), db_(db), mutex_(mutex) { if (cfd_ != nullptr) { cfd_->Ref(); } } ColumnFamilyHandleImpl::~ColumnFamilyHandleImpl() { if (cfd_ != nullptr) { // Job id == 0 means that this is not our background process, but rather // user thread JobContext job_context(0); mutex_->Lock(); if (cfd_->Unref()) { delete cfd_; } db_->FindObsoleteFiles(&job_context, false, true); mutex_->Unlock(); if (job_context.HaveSomethingToDelete()) { db_->PurgeObsoleteFiles(job_context); } job_context.Clean(); } } uint32_t ColumnFamilyHandleImpl::GetID() const { return cfd()->GetID(); } const std::string& ColumnFamilyHandleImpl::GetName() const { return cfd()->GetName(); } Status ColumnFamilyHandleImpl::GetDescriptor(ColumnFamilyDescriptor* desc) { #ifndef ROCKSDB_LITE // accessing mutable cf-options requires db mutex. InstrumentedMutexLock l(mutex_); *desc = ColumnFamilyDescriptor( cfd()->GetName(), BuildColumnFamilyOptions(*cfd()->options(), *cfd()->GetLatestMutableCFOptions())); return Status::OK(); #else return Status::NotSupported(); #endif // !ROCKSDB_LITE } const Comparator* ColumnFamilyHandleImpl::user_comparator() const { return cfd()->user_comparator(); } void GetIntTblPropCollectorFactory( const ColumnFamilyOptions& cf_options, std::vector>* int_tbl_prop_collector_factories) { auto& collector_factories = cf_options.table_properties_collector_factories; for (size_t i = 0; i < cf_options.table_properties_collector_factories.size(); ++i) { assert(collector_factories[i]); int_tbl_prop_collector_factories->emplace_back( new UserKeyTablePropertiesCollectorFactory(collector_factories[i])); } // Add collector to collect internal key statistics int_tbl_prop_collector_factories->emplace_back( new InternalKeyPropertiesCollectorFactory); } Status CheckCompressionSupported(const ColumnFamilyOptions& cf_options) { if (!cf_options.compression_per_level.empty()) { for (size_t level = 0; level < cf_options.compression_per_level.size(); ++level) { if (!CompressionTypeSupported(cf_options.compression_per_level[level])) { return Status::InvalidArgument( "Compression type " + CompressionTypeToString(cf_options.compression_per_level[level]) + " is not linked with the binary."); } } } else { if (!CompressionTypeSupported(cf_options.compression)) { return Status::InvalidArgument( "Compression type " + CompressionTypeToString(cf_options.compression) + " is not linked with the binary."); } } return Status::OK(); } Status CheckConcurrentWritesSupported(const ColumnFamilyOptions& cf_options) { if (cf_options.inplace_update_support) { return Status::InvalidArgument( "In-place memtable updates (inplace_update_support) is not compatible " "with concurrent writes (allow_concurrent_memtable_write)"); } if (cf_options.filter_deletes) { return Status::InvalidArgument( "Delete filtering (filter_deletes) is not compatible with concurrent " "memtable writes (allow_concurrent_memtable_writes)"); } if (!cf_options.memtable_factory->IsInsertConcurrentlySupported()) { return Status::InvalidArgument( "Memtable doesn't concurrent writes (allow_concurrent_memtable_write)"); } return Status::OK(); } ColumnFamilyOptions SanitizeOptions(const DBOptions& db_options, const InternalKeyComparator* icmp, const ColumnFamilyOptions& src) { ColumnFamilyOptions result = src; result.comparator = icmp; #ifdef OS_MACOSX // TODO(icanadi) make write_buffer_size uint64_t instead of size_t ClipToRange(&result.write_buffer_size, ((size_t)64) << 10, ((size_t)1) << 30); #else ClipToRange(&result.write_buffer_size, ((size_t)64) << 10, ((size_t)64) << 30); #endif // if user sets arena_block_size, we trust user to use this value. Otherwise, // calculate a proper value from writer_buffer_size; if (result.arena_block_size <= 0) { result.arena_block_size = result.write_buffer_size / 8; // Align up to 4k const size_t align = 4 * 1024; result.arena_block_size = ((result.arena_block_size + align - 1) / align) * align; } result.min_write_buffer_number_to_merge = std::min(result.min_write_buffer_number_to_merge, result.max_write_buffer_number - 1); if (result.num_levels < 1) { result.num_levels = 1; } if (result.compaction_style == kCompactionStyleLevel && result.num_levels < 2) { result.num_levels = 2; } if (result.max_write_buffer_number < 2) { result.max_write_buffer_number = 2; } if (result.max_write_buffer_number_to_maintain < 0) { result.max_write_buffer_number_to_maintain = result.max_write_buffer_number; } XFUNC_TEST("memtablelist_history", "transaction_xftest_SanitizeOptions", xf_transaction_set_memtable_history1, xf_transaction_set_memtable_history, &result.max_write_buffer_number_to_maintain); XFUNC_TEST("memtablelist_history_clear", "transaction_xftest_SanitizeOptions", xf_transaction_clear_memtable_history1, xf_transaction_clear_memtable_history, &result.max_write_buffer_number_to_maintain); if (!result.prefix_extractor) { assert(result.memtable_factory); Slice name = result.memtable_factory->Name(); if (name.compare("HashSkipListRepFactory") == 0 || name.compare("HashLinkListRepFactory") == 0) { result.memtable_factory = std::make_shared(); } } if (result.compaction_style == kCompactionStyleFIFO) { result.num_levels = 1; // since we delete level0 files in FIFO compaction when there are too many // of them, these options don't really mean anything result.level0_file_num_compaction_trigger = std::numeric_limits::max(); result.level0_slowdown_writes_trigger = std::numeric_limits::max(); result.level0_stop_writes_trigger = std::numeric_limits::max(); } if (result.level0_file_num_compaction_trigger == 0) { Warn(db_options.info_log.get(), "level0_file_num_compaction_trigger cannot be 0"); result.level0_file_num_compaction_trigger = 1; } if (result.level0_stop_writes_trigger < result.level0_slowdown_writes_trigger || result.level0_slowdown_writes_trigger < result.level0_file_num_compaction_trigger) { Warn(db_options.info_log.get(), "This condition must be satisfied: " "level0_stop_writes_trigger(%d) >= " "level0_slowdown_writes_trigger(%d) >= " "level0_file_num_compaction_trigger(%d)", result.level0_stop_writes_trigger, result.level0_slowdown_writes_trigger, result.level0_file_num_compaction_trigger); if (result.level0_slowdown_writes_trigger < result.level0_file_num_compaction_trigger) { result.level0_slowdown_writes_trigger = result.level0_file_num_compaction_trigger; } if (result.level0_stop_writes_trigger < result.level0_slowdown_writes_trigger) { result.level0_stop_writes_trigger = result.level0_slowdown_writes_trigger; } Warn(db_options.info_log.get(), "Adjust the value to " "level0_stop_writes_trigger(%d)" "level0_slowdown_writes_trigger(%d)" "level0_file_num_compaction_trigger(%d)", result.level0_stop_writes_trigger, result.level0_slowdown_writes_trigger, result.level0_file_num_compaction_trigger); } if (result.soft_pending_compaction_bytes_limit == 0) { result.soft_pending_compaction_bytes_limit = result.hard_pending_compaction_bytes_limit; } else if (result.hard_pending_compaction_bytes_limit > 0 && result.soft_pending_compaction_bytes_limit > result.hard_pending_compaction_bytes_limit) { result.soft_pending_compaction_bytes_limit = result.hard_pending_compaction_bytes_limit; } if (result.level_compaction_dynamic_level_bytes) { if (result.compaction_style != kCompactionStyleLevel || db_options.db_paths.size() > 1U) { // 1. level_compaction_dynamic_level_bytes only makes sense for // level-based compaction. // 2. we don't yet know how to make both of this feature and multiple // DB path work. result.level_compaction_dynamic_level_bytes = false; } } return result; } int SuperVersion::dummy = 0; void* const SuperVersion::kSVInUse = &SuperVersion::dummy; void* const SuperVersion::kSVObsolete = nullptr; SuperVersion::~SuperVersion() { for (auto td : to_delete) { delete td; } } SuperVersion* SuperVersion::Ref() { refs.fetch_add(1, std::memory_order_relaxed); return this; } bool SuperVersion::Unref() { // fetch_sub returns the previous value of ref uint32_t previous_refs = refs.fetch_sub(1); assert(previous_refs > 0); return previous_refs == 1; } void SuperVersion::Cleanup() { assert(refs.load(std::memory_order_relaxed) == 0); imm->Unref(&to_delete); MemTable* m = mem->Unref(); if (m != nullptr) { auto* memory_usage = current->cfd()->imm()->current_memory_usage(); assert(*memory_usage >= m->ApproximateMemoryUsage()); *memory_usage -= m->ApproximateMemoryUsage(); to_delete.push_back(m); } current->Unref(); } void SuperVersion::Init(MemTable* new_mem, MemTableListVersion* new_imm, Version* new_current) { mem = new_mem; imm = new_imm; current = new_current; mem->Ref(); imm->Ref(); current->Ref(); refs.store(1, std::memory_order_relaxed); } namespace { void SuperVersionUnrefHandle(void* ptr) { // UnrefHandle is called when a thread exists or a ThreadLocalPtr gets // destroyed. When former happens, the thread shouldn't see kSVInUse. // When latter happens, we are in ~ColumnFamilyData(), no get should happen as // well. SuperVersion* sv = static_cast(ptr); if (sv->Unref()) { sv->db_mutex->Lock(); sv->Cleanup(); sv->db_mutex->Unlock(); delete sv; } } } // anonymous namespace ColumnFamilyData::ColumnFamilyData( uint32_t id, const std::string& name, Version* _dummy_versions, Cache* _table_cache, WriteBuffer* write_buffer, const ColumnFamilyOptions& cf_options, const DBOptions* db_options, const EnvOptions& env_options, ColumnFamilySet* column_family_set) : id_(id), name_(name), dummy_versions_(_dummy_versions), current_(nullptr), refs_(0), dropped_(false), internal_comparator_(cf_options.comparator), options_(*db_options, SanitizeOptions(*db_options, &internal_comparator_, cf_options)), ioptions_(options_), mutable_cf_options_(options_, ioptions_), write_buffer_(write_buffer), mem_(nullptr), imm_(options_.min_write_buffer_number_to_merge, options_.max_write_buffer_number_to_maintain), super_version_(nullptr), super_version_number_(0), local_sv_(new ThreadLocalPtr(&SuperVersionUnrefHandle)), next_(nullptr), prev_(nullptr), log_number_(0), column_family_set_(column_family_set), pending_flush_(false), pending_compaction_(false), prev_compaction_needed_bytes_(0) { Ref(); // Convert user defined table properties collector factories to internal ones. GetIntTblPropCollectorFactory(options_, &int_tbl_prop_collector_factories_); // if _dummy_versions is nullptr, then this is a dummy column family. if (_dummy_versions != nullptr) { internal_stats_.reset( new InternalStats(ioptions_.num_levels, db_options->env, this)); table_cache_.reset(new TableCache(ioptions_, env_options, _table_cache)); if (ioptions_.compaction_style == kCompactionStyleLevel) { compaction_picker_.reset( new LevelCompactionPicker(ioptions_, &internal_comparator_)); #ifndef ROCKSDB_LITE } else if (ioptions_.compaction_style == kCompactionStyleUniversal) { compaction_picker_.reset( new UniversalCompactionPicker(ioptions_, &internal_comparator_)); } else if (ioptions_.compaction_style == kCompactionStyleFIFO) { compaction_picker_.reset( new FIFOCompactionPicker(ioptions_, &internal_comparator_)); } else if (ioptions_.compaction_style == kCompactionStyleNone) { compaction_picker_.reset(new NullCompactionPicker( ioptions_, &internal_comparator_)); Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "Column family %s does not use any background compaction. " "Compactions can only be done via CompactFiles\n", GetName().c_str()); #endif // !ROCKSDB_LITE } else { Log(InfoLogLevel::ERROR_LEVEL, ioptions_.info_log, "Unable to recognize the specified compaction style %d. " "Column family %s will use kCompactionStyleLevel.\n", ioptions_.compaction_style, GetName().c_str()); compaction_picker_.reset( new LevelCompactionPicker(ioptions_, &internal_comparator_)); } if (column_family_set_->NumberOfColumnFamilies() < 10) { Log(InfoLogLevel::INFO_LEVEL, ioptions_.info_log, "--------------- Options for column family [%s]:\n", name.c_str()); options_.DumpCFOptions(ioptions_.info_log); } else { Log(InfoLogLevel::INFO_LEVEL, ioptions_.info_log, "\t(skipping printing options)\n"); } } RecalculateWriteStallConditions(mutable_cf_options_); } // DB mutex held ColumnFamilyData::~ColumnFamilyData() { assert(refs_.load(std::memory_order_relaxed) == 0); // remove from linked list auto prev = prev_; auto next = next_; prev->next_ = next; next->prev_ = prev; if (!dropped_ && column_family_set_ != nullptr) { // If it's dropped, it's already removed from column family set // If column_family_set_ == nullptr, this is dummy CFD and not in // ColumnFamilySet column_family_set_->RemoveColumnFamily(this); } if (current_ != nullptr) { current_->Unref(); } // It would be wrong if this ColumnFamilyData is in flush_queue_ or // compaction_queue_ and we destroyed it assert(!pending_flush_); assert(!pending_compaction_); if (super_version_ != nullptr) { // Release SuperVersion reference kept in ThreadLocalPtr. // This must be done outside of mutex_ since unref handler can lock mutex. super_version_->db_mutex->Unlock(); local_sv_.reset(); super_version_->db_mutex->Lock(); bool is_last_reference __attribute__((unused)); is_last_reference = super_version_->Unref(); assert(is_last_reference); super_version_->Cleanup(); delete super_version_; super_version_ = nullptr; } if (dummy_versions_ != nullptr) { // List must be empty assert(dummy_versions_->TEST_Next() == dummy_versions_); bool deleted __attribute__((unused)) = dummy_versions_->Unref(); assert(deleted); } if (mem_ != nullptr) { delete mem_->Unref(); } autovector to_delete; imm_.current()->Unref(&to_delete); for (MemTable* m : to_delete) { delete m; } } void ColumnFamilyData::SetDropped() { // can't drop default CF assert(id_ != 0); dropped_ = true; write_controller_token_.reset(); // remove from column_family_set column_family_set_->RemoveColumnFamily(this); } const double kSlowdownRatio = 1.2; namespace { std::unique_ptr SetupDelay( uint64_t max_write_rate, WriteController* write_controller, uint64_t compaction_needed_bytes, uint64_t prev_compaction_neeed_bytes, bool auto_comapctions_disabled) { const uint64_t kMinWriteRate = 1024u; // Minimum write rate 1KB/s. uint64_t write_rate = write_controller->delayed_write_rate(); if (auto_comapctions_disabled) { // When auto compaction is disabled, always use the value user gave. write_rate = max_write_rate; } else if (write_controller->NeedsDelay() && max_write_rate > kMinWriteRate) { // If user gives rate less than kMinWriteRate, don't adjust it. // // If already delayed, need to adjust based on previous compaction debt. // When there are two or more column families require delay, we always // increase or reduce write rate based on information for one single // column family. It is likely to be OK but we can improve if there is a // problem. // Ignore compaction_needed_bytes = 0 case because compaction_needed_bytes // is only available in level-based compaction // // If the compaction debt stays the same as previously, we also further slow // down. It usually means a mem table is full. It's mainly for the case // where both of flush and compaction are much slower than the speed we // insert to mem tables, so we need to actively slow down before we get // feedback signal from compaction and flushes to avoid the full stop // because of hitting the max write buffer number. if (prev_compaction_neeed_bytes > 0 && prev_compaction_neeed_bytes <= compaction_needed_bytes) { write_rate = static_cast(static_cast(write_rate) / kSlowdownRatio); if (write_rate < kMinWriteRate) { write_rate = kMinWriteRate; } } else if (prev_compaction_neeed_bytes > compaction_needed_bytes) { // We are speeding up by ratio of kSlowdownRatio when we have paid // compaction debt. But we'll never speed up to faster than the write rate // given by users. write_rate = static_cast(static_cast(write_rate) * kSlowdownRatio); if (write_rate > max_write_rate) { write_rate = max_write_rate; } } } return write_controller->GetDelayToken(write_rate); } int GetL0ThresholdSpeedupCompaction(int level0_file_num_compaction_trigger, int level0_slowdown_writes_trigger) { // SanitizeOptions() ensures it. assert(level0_file_num_compaction_trigger <= level0_slowdown_writes_trigger); // 1/4 of the way between L0 compaction trigger threshold and slowdown // condition. // Or twice as compaction trigger, if it is smaller. return std::min(level0_file_num_compaction_trigger * 2, level0_file_num_compaction_trigger + (level0_slowdown_writes_trigger - level0_file_num_compaction_trigger) / 4); } } // namespace void ColumnFamilyData::RecalculateWriteStallConditions( const MutableCFOptions& mutable_cf_options) { if (current_ != nullptr) { auto* vstorage = current_->storage_info(); auto write_controller = column_family_set_->write_controller_; uint64_t compaction_needed_bytes = vstorage->estimated_compaction_needed_bytes(); if (imm()->NumNotFlushed() >= mutable_cf_options.max_write_buffer_number) { write_controller_token_ = write_controller->GetStopToken(); internal_stats_->AddCFStats(InternalStats::MEMTABLE_COMPACTION, 1); Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Stopping writes because we have %d immutable memtables " "(waiting for flush), max_write_buffer_number is set to %d", name_.c_str(), imm()->NumNotFlushed(), mutable_cf_options.max_write_buffer_number); } else if (vstorage->l0_delay_trigger_count() >= mutable_cf_options.level0_stop_writes_trigger) { write_controller_token_ = write_controller->GetStopToken(); internal_stats_->AddCFStats(InternalStats::LEVEL0_NUM_FILES_TOTAL, 1); if (compaction_picker_->IsLevel0CompactionInProgress()) { internal_stats_->AddCFStats( InternalStats::LEVEL0_NUM_FILES_WITH_COMPACTION, 1); } Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Stopping writes because we have %d level-0 files", name_.c_str(), vstorage->l0_delay_trigger_count()); } else if (mutable_cf_options.hard_pending_compaction_bytes_limit > 0 && compaction_needed_bytes >= mutable_cf_options.hard_pending_compaction_bytes_limit) { write_controller_token_ = write_controller->GetStopToken(); internal_stats_->AddCFStats( InternalStats::HARD_PENDING_COMPACTION_BYTES_LIMIT, 1); Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Stopping writes because of estimated pending compaction " "bytes %" PRIu64, name_.c_str(), compaction_needed_bytes); } else if (mutable_cf_options.max_write_buffer_number > 3 && imm()->NumNotFlushed() >= mutable_cf_options.max_write_buffer_number - 1) { write_controller_token_ = SetupDelay(ioptions_.delayed_write_rate, write_controller, compaction_needed_bytes, prev_compaction_needed_bytes_, mutable_cf_options.disable_auto_compactions); internal_stats_->AddCFStats(InternalStats::MEMTABLE_SLOWDOWN, 1); Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Stalling writes because we have %d immutable memtables " "(waiting for flush), max_write_buffer_number is set to %d " "rate %" PRIu64, name_.c_str(), imm()->NumNotFlushed(), mutable_cf_options.max_write_buffer_number, write_controller->delayed_write_rate()); } else if (mutable_cf_options.level0_slowdown_writes_trigger >= 0 && vstorage->l0_delay_trigger_count() >= mutable_cf_options.level0_slowdown_writes_trigger) { write_controller_token_ = SetupDelay(ioptions_.delayed_write_rate, write_controller, compaction_needed_bytes, prev_compaction_needed_bytes_, mutable_cf_options.disable_auto_compactions); internal_stats_->AddCFStats(InternalStats::LEVEL0_SLOWDOWN_TOTAL, 1); if (compaction_picker_->IsLevel0CompactionInProgress()) { internal_stats_->AddCFStats( InternalStats::LEVEL0_SLOWDOWN_WITH_COMPACTION, 1); } Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Stalling writes because we have %d level-0 files " "rate %" PRIu64, name_.c_str(), vstorage->l0_delay_trigger_count(), write_controller->delayed_write_rate()); } else if (mutable_cf_options.soft_pending_compaction_bytes_limit > 0 && vstorage->estimated_compaction_needed_bytes() >= mutable_cf_options.soft_pending_compaction_bytes_limit) { write_controller_token_ = SetupDelay(ioptions_.delayed_write_rate, write_controller, compaction_needed_bytes, prev_compaction_needed_bytes_, mutable_cf_options.disable_auto_compactions); internal_stats_->AddCFStats( InternalStats::SOFT_PENDING_COMPACTION_BYTES_LIMIT, 1); Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Stalling writes because of estimated pending compaction " "bytes %" PRIu64 " rate %" PRIu64, name_.c_str(), vstorage->estimated_compaction_needed_bytes(), write_controller->delayed_write_rate()); } else if (vstorage->l0_delay_trigger_count() >= GetL0ThresholdSpeedupCompaction( mutable_cf_options.level0_file_num_compaction_trigger, mutable_cf_options.level0_slowdown_writes_trigger)) { write_controller_token_ = write_controller->GetCompactionPressureToken(); Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Increasing compaction threads because we have %d level-0 " "files ", name_.c_str(), vstorage->l0_delay_trigger_count()); } else if (vstorage->estimated_compaction_needed_bytes() >= mutable_cf_options.soft_pending_compaction_bytes_limit / 4) { // Increase compaction threads if bytes needed for compaction exceeds // 1/4 of threshold for slowing down. // If soft pending compaction byte limit is not set, always speed up // compaction. write_controller_token_ = write_controller->GetCompactionPressureToken(); if (mutable_cf_options.soft_pending_compaction_bytes_limit > 0) { Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] Increasing compaction threads because of estimated pending " "compaction " "bytes %" PRIu64, name_.c_str(), vstorage->estimated_compaction_needed_bytes()); } } else { write_controller_token_.reset(); } prev_compaction_needed_bytes_ = compaction_needed_bytes; } } const EnvOptions* ColumnFamilyData::soptions() const { return &(column_family_set_->env_options_); } void ColumnFamilyData::SetCurrent(Version* current_version) { current_ = current_version; } uint64_t ColumnFamilyData::GetNumLiveVersions() const { return VersionSet::GetNumLiveVersions(dummy_versions_); } uint64_t ColumnFamilyData::GetTotalSstFilesSize() const { return VersionSet::GetTotalSstFilesSize(dummy_versions_); } MemTable* ColumnFamilyData::ConstructNewMemtable( const MutableCFOptions& mutable_cf_options, SequenceNumber earliest_seq) { assert(current_ != nullptr); return new MemTable(internal_comparator_, ioptions_, mutable_cf_options, write_buffer_, earliest_seq); } void ColumnFamilyData::CreateNewMemtable( const MutableCFOptions& mutable_cf_options, SequenceNumber earliest_seq) { if (mem_ != nullptr) { delete mem_->Unref(); } SetMemtable(ConstructNewMemtable(mutable_cf_options, earliest_seq)); mem_->Ref(); } bool ColumnFamilyData::NeedsCompaction() const { return compaction_picker_->NeedsCompaction(current_->storage_info()); } Compaction* ColumnFamilyData::PickCompaction( const MutableCFOptions& mutable_options, LogBuffer* log_buffer) { auto* result = compaction_picker_->PickCompaction( GetName(), mutable_options, current_->storage_info(), log_buffer); if (result != nullptr) { result->SetInputVersion(current_); } return result; } const int ColumnFamilyData::kCompactAllLevels = -1; const int ColumnFamilyData::kCompactToBaseLevel = -2; Compaction* ColumnFamilyData::CompactRange( const MutableCFOptions& mutable_cf_options, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* conflict) { auto* result = compaction_picker_->CompactRange( GetName(), mutable_cf_options, current_->storage_info(), input_level, output_level, output_path_id, begin, end, compaction_end, conflict); if (result != nullptr) { result->SetInputVersion(current_); } return result; } SuperVersion* ColumnFamilyData::GetReferencedSuperVersion( InstrumentedMutex* db_mutex) { SuperVersion* sv = nullptr; sv = GetThreadLocalSuperVersion(db_mutex); sv->Ref(); if (!ReturnThreadLocalSuperVersion(sv)) { sv->Unref(); } return sv; } SuperVersion* ColumnFamilyData::GetThreadLocalSuperVersion( InstrumentedMutex* db_mutex) { SuperVersion* sv = nullptr; // The SuperVersion is cached in thread local storage to avoid acquiring // mutex when SuperVersion does not change since the last use. When a new // SuperVersion is installed, the compaction or flush thread cleans up // cached SuperVersion in all existing thread local storage. To avoid // acquiring mutex for this operation, we use atomic Swap() on the thread // local pointer to guarantee exclusive access. If the thread local pointer // is being used while a new SuperVersion is installed, the cached // SuperVersion can become stale. In that case, the background thread would // have swapped in kSVObsolete. We re-check the value at when returning // SuperVersion back to thread local, with an atomic compare and swap. // The superversion will need to be released if detected to be stale. void* ptr = local_sv_->Swap(SuperVersion::kSVInUse); // Invariant: // (1) Scrape (always) installs kSVObsolete in ThreadLocal storage // (2) the Swap above (always) installs kSVInUse, ThreadLocal storage // should only keep kSVInUse before ReturnThreadLocalSuperVersion call // (if no Scrape happens). assert(ptr != SuperVersion::kSVInUse); sv = static_cast(ptr); if (sv == SuperVersion::kSVObsolete || sv->version_number != super_version_number_.load()) { RecordTick(ioptions_.statistics, NUMBER_SUPERVERSION_ACQUIRES); SuperVersion* sv_to_delete = nullptr; if (sv && sv->Unref()) { RecordTick(ioptions_.statistics, NUMBER_SUPERVERSION_CLEANUPS); db_mutex->Lock(); // NOTE: underlying resources held by superversion (sst files) might // not be released until the next background job. sv->Cleanup(); sv_to_delete = sv; } else { db_mutex->Lock(); } sv = super_version_->Ref(); db_mutex->Unlock(); delete sv_to_delete; } assert(sv != nullptr); return sv; } bool ColumnFamilyData::ReturnThreadLocalSuperVersion(SuperVersion* sv) { assert(sv != nullptr); // Put the SuperVersion back void* expected = SuperVersion::kSVInUse; if (local_sv_->CompareAndSwap(static_cast(sv), expected)) { // When we see kSVInUse in the ThreadLocal, we are sure ThreadLocal // storage has not been altered and no Scrape has happened. The // SuperVersion is still current. return true; } else { // ThreadLocal scrape happened in the process of this GetImpl call (after // thread local Swap() at the beginning and before CompareAndSwap()). // This means the SuperVersion it holds is obsolete. assert(expected == SuperVersion::kSVObsolete); } return false; } SuperVersion* ColumnFamilyData::InstallSuperVersion( SuperVersion* new_superversion, InstrumentedMutex* db_mutex) { db_mutex->AssertHeld(); return InstallSuperVersion(new_superversion, db_mutex, mutable_cf_options_); } SuperVersion* ColumnFamilyData::InstallSuperVersion( SuperVersion* new_superversion, InstrumentedMutex* db_mutex, const MutableCFOptions& mutable_cf_options) { new_superversion->db_mutex = db_mutex; new_superversion->mutable_cf_options = mutable_cf_options; new_superversion->Init(mem_, imm_.current(), current_); SuperVersion* old_superversion = super_version_; super_version_ = new_superversion; ++super_version_number_; super_version_->version_number = super_version_number_; // Reset SuperVersions cached in thread local storage ResetThreadLocalSuperVersions(); RecalculateWriteStallConditions(mutable_cf_options); if (old_superversion != nullptr && old_superversion->Unref()) { old_superversion->Cleanup(); return old_superversion; // will let caller delete outside of mutex } return nullptr; } void ColumnFamilyData::ResetThreadLocalSuperVersions() { autovector sv_ptrs; local_sv_->Scrape(&sv_ptrs, SuperVersion::kSVObsolete); for (auto ptr : sv_ptrs) { assert(ptr); if (ptr == SuperVersion::kSVInUse) { continue; } auto sv = static_cast(ptr); if (sv->Unref()) { sv->Cleanup(); delete sv; } } } #ifndef ROCKSDB_LITE Status ColumnFamilyData::SetOptions( const std::unordered_map& options_map) { MutableCFOptions new_mutable_cf_options; Status s = GetMutableOptionsFromStrings(mutable_cf_options_, options_map, &new_mutable_cf_options); if (s.ok()) { mutable_cf_options_ = new_mutable_cf_options; mutable_cf_options_.RefreshDerivedOptions(ioptions_); } return s; } #endif // ROCKSDB_LITE ColumnFamilySet::ColumnFamilySet(const std::string& dbname, const DBOptions* db_options, const EnvOptions& env_options, Cache* table_cache, WriteBuffer* write_buffer, WriteController* write_controller) : max_column_family_(0), dummy_cfd_(new ColumnFamilyData(0, "", nullptr, nullptr, nullptr, ColumnFamilyOptions(), db_options, env_options, nullptr)), default_cfd_cache_(nullptr), db_name_(dbname), db_options_(db_options), env_options_(env_options), table_cache_(table_cache), write_buffer_(write_buffer), write_controller_(write_controller) { // initialize linked list dummy_cfd_->prev_ = dummy_cfd_; dummy_cfd_->next_ = dummy_cfd_; } ColumnFamilySet::~ColumnFamilySet() { while (column_family_data_.size() > 0) { // cfd destructor will delete itself from column_family_data_ auto cfd = column_family_data_.begin()->second; cfd->Unref(); delete cfd; } dummy_cfd_->Unref(); delete dummy_cfd_; } ColumnFamilyData* ColumnFamilySet::GetDefault() const { assert(default_cfd_cache_ != nullptr); return default_cfd_cache_; } ColumnFamilyData* ColumnFamilySet::GetColumnFamily(uint32_t id) const { auto cfd_iter = column_family_data_.find(id); if (cfd_iter != column_family_data_.end()) { return cfd_iter->second; } else { return nullptr; } } ColumnFamilyData* ColumnFamilySet::GetColumnFamily(const std::string& name) const { auto cfd_iter = column_families_.find(name); if (cfd_iter != column_families_.end()) { auto cfd = GetColumnFamily(cfd_iter->second); assert(cfd != nullptr); return cfd; } else { return nullptr; } } uint32_t ColumnFamilySet::GetNextColumnFamilyID() { return ++max_column_family_; } uint32_t ColumnFamilySet::GetMaxColumnFamily() { return max_column_family_; } void ColumnFamilySet::UpdateMaxColumnFamily(uint32_t new_max_column_family) { max_column_family_ = std::max(new_max_column_family, max_column_family_); } size_t ColumnFamilySet::NumberOfColumnFamilies() const { return column_families_.size(); } // under a DB mutex AND write thread ColumnFamilyData* ColumnFamilySet::CreateColumnFamily( const std::string& name, uint32_t id, Version* dummy_versions, const ColumnFamilyOptions& options) { assert(column_families_.find(name) == column_families_.end()); ColumnFamilyData* new_cfd = new ColumnFamilyData(id, name, dummy_versions, table_cache_, write_buffer_, options, db_options_, env_options_, this); column_families_.insert({name, id}); column_family_data_.insert({id, new_cfd}); max_column_family_ = std::max(max_column_family_, id); // add to linked list new_cfd->next_ = dummy_cfd_; auto prev = dummy_cfd_->prev_; new_cfd->prev_ = prev; prev->next_ = new_cfd; dummy_cfd_->prev_ = new_cfd; if (id == 0) { default_cfd_cache_ = new_cfd; } return new_cfd; } // REQUIRES: DB mutex held void ColumnFamilySet::FreeDeadColumnFamilies() { autovector to_delete; for (auto cfd = dummy_cfd_->next_; cfd != dummy_cfd_; cfd = cfd->next_) { if (cfd->refs_.load(std::memory_order_relaxed) == 0) { to_delete.push_back(cfd); } } for (auto cfd : to_delete) { // this is very rare, so it's not a problem that we do it under a mutex delete cfd; } } // under a DB mutex AND from a write thread void ColumnFamilySet::RemoveColumnFamily(ColumnFamilyData* cfd) { auto cfd_iter = column_family_data_.find(cfd->GetID()); assert(cfd_iter != column_family_data_.end()); column_family_data_.erase(cfd_iter); column_families_.erase(cfd->GetName()); } // under a DB mutex OR from a write thread bool ColumnFamilyMemTablesImpl::Seek(uint32_t column_family_id) { if (column_family_id == 0) { // optimization for common case current_ = column_family_set_->GetDefault(); } else { current_ = column_family_set_->GetColumnFamily(column_family_id); } handle_.SetCFD(current_); return current_ != nullptr; } uint64_t ColumnFamilyMemTablesImpl::GetLogNumber() const { assert(current_ != nullptr); return current_->GetLogNumber(); } MemTable* ColumnFamilyMemTablesImpl::GetMemTable() const { assert(current_ != nullptr); return current_->mem(); } ColumnFamilyHandle* ColumnFamilyMemTablesImpl::GetColumnFamilyHandle() { assert(current_ != nullptr); return &handle_; } uint32_t GetColumnFamilyID(ColumnFamilyHandle* column_family) { uint32_t column_family_id = 0; if (column_family != nullptr) { auto cfh = reinterpret_cast(column_family); column_family_id = cfh->GetID(); } return column_family_id; } const Comparator* GetColumnFamilyUserComparator( ColumnFamilyHandle* column_family) { if (column_family != nullptr) { auto cfh = reinterpret_cast(column_family); return cfh->user_comparator(); } return nullptr; } } // namespace rocksdb rocksdb-4.5.1/db/column_family.h000066400000000000000000000531571267530634000165650ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include "db/memtable_list.h" #include "db/write_batch_internal.h" #include "db/write_controller.h" #include "db/table_cache.h" #include "db/table_properties_collector.h" #include "rocksdb/compaction_job_stats.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "util/mutable_cf_options.h" #include "util/thread_local.h" namespace rocksdb { class Version; class VersionSet; class MemTable; class MemTableListVersion; class CompactionPicker; class Compaction; class InternalKey; class InternalStats; class ColumnFamilyData; class DBImpl; class LogBuffer; class InstrumentedMutex; class InstrumentedMutexLock; extern const double kSlowdownRatio; // ColumnFamilyHandleImpl is the class that clients use to access different // column families. It has non-trivial destructor, which gets called when client // is done using the column family class ColumnFamilyHandleImpl : public ColumnFamilyHandle { public: // create while holding the mutex ColumnFamilyHandleImpl( ColumnFamilyData* cfd, DBImpl* db, InstrumentedMutex* mutex); // destroy without mutex virtual ~ColumnFamilyHandleImpl(); virtual ColumnFamilyData* cfd() const { return cfd_; } virtual const Comparator* user_comparator() const; virtual uint32_t GetID() const override; virtual const std::string& GetName() const override; virtual Status GetDescriptor(ColumnFamilyDescriptor* desc) override; private: ColumnFamilyData* cfd_; DBImpl* db_; InstrumentedMutex* mutex_; }; // Does not ref-count ColumnFamilyData // We use this dummy ColumnFamilyHandleImpl because sometimes MemTableInserter // calls DBImpl methods. When this happens, MemTableInserter need access to // ColumnFamilyHandle (same as the client would need). In that case, we feed // MemTableInserter dummy ColumnFamilyHandle and enable it to call DBImpl // methods class ColumnFamilyHandleInternal : public ColumnFamilyHandleImpl { public: ColumnFamilyHandleInternal() : ColumnFamilyHandleImpl(nullptr, nullptr, nullptr) {} void SetCFD(ColumnFamilyData* _cfd) { internal_cfd_ = _cfd; } virtual ColumnFamilyData* cfd() const override { return internal_cfd_; } private: ColumnFamilyData* internal_cfd_; }; // holds references to memtable, all immutable memtables and version struct SuperVersion { // Accessing members of this class is not thread-safe and requires external // synchronization (ie db mutex held or on write thread). MemTable* mem; MemTableListVersion* imm; Version* current; MutableCFOptions mutable_cf_options; // Version number of the current SuperVersion uint64_t version_number; InstrumentedMutex* db_mutex; // should be called outside the mutex SuperVersion() = default; ~SuperVersion(); SuperVersion* Ref(); // If Unref() returns true, Cleanup() should be called with mutex held // before deleting this SuperVersion. bool Unref(); // call these two methods with db mutex held // Cleanup unrefs mem, imm and current. Also, it stores all memtables // that needs to be deleted in to_delete vector. Unrefing those // objects needs to be done in the mutex void Cleanup(); void Init(MemTable* new_mem, MemTableListVersion* new_imm, Version* new_current); // The value of dummy is not actually used. kSVInUse takes its address as a // mark in the thread local storage to indicate the SuperVersion is in use // by thread. This way, the value of kSVInUse is guaranteed to have no // conflict with SuperVersion object address and portable on different // platform. static int dummy; static void* const kSVInUse; static void* const kSVObsolete; private: std::atomic refs; // We need to_delete because during Cleanup(), imm->Unref() returns // all memtables that we need to free through this vector. We then // delete all those memtables outside of mutex, during destruction autovector to_delete; }; extern Status CheckCompressionSupported(const ColumnFamilyOptions& cf_options); extern Status CheckConcurrentWritesSupported( const ColumnFamilyOptions& cf_options); extern ColumnFamilyOptions SanitizeOptions(const DBOptions& db_options, const InternalKeyComparator* icmp, const ColumnFamilyOptions& src); // Wrap user defined table proproties collector factories `from cf_options` // into internal ones in int_tbl_prop_collector_factories. Add a system internal // one too. extern void GetIntTblPropCollectorFactory( const ColumnFamilyOptions& cf_options, std::vector>* int_tbl_prop_collector_factories); class ColumnFamilySet; // This class keeps all the data that a column family needs. // Most methods require DB mutex held, unless otherwise noted class ColumnFamilyData { public: ~ColumnFamilyData(); // thread-safe uint32_t GetID() const { return id_; } // thread-safe const std::string& GetName() const { return name_; } // Ref() can only be called from a context where the caller can guarantee // that ColumnFamilyData is alive (while holding a non-zero ref already, // holding a DB mutex, or as the leader in a write batch group). void Ref() { refs_.fetch_add(1, std::memory_order_relaxed); } // Unref decreases the reference count, but does not handle deletion // when the count goes to 0. If this method returns true then the // caller should delete the instance immediately, or later, by calling // FreeDeadColumnFamilies(). Unref() can only be called while holding // a DB mutex, or during single-threaded recovery. bool Unref() { int old_refs = refs_.fetch_sub(1, std::memory_order_relaxed); assert(old_refs > 0); return old_refs == 1; } // SetDropped() can only be called under following conditions: // 1) Holding a DB mutex, // 2) from single-threaded write thread, AND // 3) from single-threaded VersionSet::LogAndApply() // After dropping column family no other operation on that column family // will be executed. All the files and memory will be, however, kept around // until client drops the column family handle. That way, client can still // access data from dropped column family. // Column family can be dropped and still alive. In that state: // *) Compaction and flush is not executed on the dropped column family. // *) Client can continue reading from column family. Writes will fail unless // WriteOptions::ignore_missing_column_families is true // When the dropped column family is unreferenced, then we: // *) Remove column family from the linked list maintained by ColumnFamilySet // *) delete all memory associated with that column family // *) delete all the files associated with that column family void SetDropped(); bool IsDropped() const { return dropped_; } // thread-safe int NumberLevels() const { return ioptions_.num_levels; } void SetLogNumber(uint64_t log_number) { log_number_ = log_number; } uint64_t GetLogNumber() const { return log_number_; } // !!! To be deprecated! Please don't not use this function anymore! const Options* options() const { return &options_; } // thread-safe const EnvOptions* soptions() const; const ImmutableCFOptions* ioptions() const { return &ioptions_; } // REQUIRES: DB mutex held // This returns the MutableCFOptions used by current SuperVersion // You shoul use this API to reference MutableCFOptions most of the time. const MutableCFOptions* GetCurrentMutableCFOptions() const { return &(super_version_->mutable_cf_options); } // REQUIRES: DB mutex held // This returns the latest MutableCFOptions, which may be not in effect yet. const MutableCFOptions* GetLatestMutableCFOptions() const { return &mutable_cf_options_; } #ifndef ROCKSDB_LITE // REQUIRES: DB mutex held Status SetOptions( const std::unordered_map& options_map); #endif // ROCKSDB_LITE InternalStats* internal_stats() { return internal_stats_.get(); } MemTableList* imm() { return &imm_; } MemTable* mem() { return mem_; } Version* current() { return current_; } Version* dummy_versions() { return dummy_versions_; } void SetCurrent(Version* _current); uint64_t GetNumLiveVersions() const; // REQUIRE: DB mutex held uint64_t GetTotalSstFilesSize() const; // REQUIRE: DB mutex held void SetMemtable(MemTable* new_mem) { mem_ = new_mem; } // See Memtable constructor for explanation of earliest_seq param. MemTable* ConstructNewMemtable(const MutableCFOptions& mutable_cf_options, SequenceNumber earliest_seq); void CreateNewMemtable(const MutableCFOptions& mutable_cf_options, SequenceNumber earliest_seq); TableCache* table_cache() const { return table_cache_.get(); } // See documentation in compaction_picker.h // REQUIRES: DB mutex held bool NeedsCompaction() const; // REQUIRES: DB mutex held Compaction* PickCompaction(const MutableCFOptions& mutable_options, LogBuffer* log_buffer); // A flag to tell a manual compaction is to compact all levels together // instad of for specific level. static const int kCompactAllLevels; // A flag to tell a manual compaction's output is base level. static const int kCompactToBaseLevel; // REQUIRES: DB mutex held Compaction* CompactRange(const MutableCFOptions& mutable_cf_options, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict); CompactionPicker* compaction_picker() { return compaction_picker_.get(); } // thread-safe const Comparator* user_comparator() const { return internal_comparator_.user_comparator(); } // thread-safe const InternalKeyComparator& internal_comparator() const { return internal_comparator_; } const std::vector>* int_tbl_prop_collector_factories() const { return &int_tbl_prop_collector_factories_; } SuperVersion* GetSuperVersion() { return super_version_; } // thread-safe // Return a already referenced SuperVersion to be used safely. SuperVersion* GetReferencedSuperVersion(InstrumentedMutex* db_mutex); // thread-safe // Get SuperVersion stored in thread local storage. If it does not exist, // get a reference from a current SuperVersion. SuperVersion* GetThreadLocalSuperVersion(InstrumentedMutex* db_mutex); // Try to return SuperVersion back to thread local storage. Retrun true on // success and false on failure. It fails when the thread local storage // contains anything other than SuperVersion::kSVInUse flag. bool ReturnThreadLocalSuperVersion(SuperVersion* sv); // thread-safe uint64_t GetSuperVersionNumber() const { return super_version_number_.load(); } // will return a pointer to SuperVersion* if previous SuperVersion // if its reference count is zero and needs deletion or nullptr if not // As argument takes a pointer to allocated SuperVersion to enable // the clients to allocate SuperVersion outside of mutex. // IMPORTANT: Only call this from DBImpl::InstallSuperVersion() SuperVersion* InstallSuperVersion(SuperVersion* new_superversion, InstrumentedMutex* db_mutex, const MutableCFOptions& mutable_cf_options); SuperVersion* InstallSuperVersion(SuperVersion* new_superversion, InstrumentedMutex* db_mutex); void ResetThreadLocalSuperVersions(); // Protected by DB mutex void set_pending_flush(bool value) { pending_flush_ = value; } void set_pending_compaction(bool value) { pending_compaction_ = value; } bool pending_flush() { return pending_flush_; } bool pending_compaction() { return pending_compaction_; } // Recalculate some small conditions, which are changed only during // compaction, adding new memtable and/or // recalculation of compaction score. These values are used in // DBImpl::MakeRoomForWrite function to decide, if it need to make // a write stall void RecalculateWriteStallConditions( const MutableCFOptions& mutable_cf_options); private: friend class ColumnFamilySet; ColumnFamilyData(uint32_t id, const std::string& name, Version* dummy_versions, Cache* table_cache, WriteBuffer* write_buffer, const ColumnFamilyOptions& options, const DBOptions* db_options, const EnvOptions& env_options, ColumnFamilySet* column_family_set); uint32_t id_; const std::string name_; Version* dummy_versions_; // Head of circular doubly-linked list of versions. Version* current_; // == dummy_versions->prev_ std::atomic refs_; // outstanding references to ColumnFamilyData bool dropped_; // true if client dropped it const InternalKeyComparator internal_comparator_; std::vector> int_tbl_prop_collector_factories_; const Options options_; const ImmutableCFOptions ioptions_; MutableCFOptions mutable_cf_options_; std::unique_ptr table_cache_; std::unique_ptr internal_stats_; WriteBuffer* write_buffer_; MemTable* mem_; MemTableList imm_; SuperVersion* super_version_; // An ordinal representing the current SuperVersion. Updated by // InstallSuperVersion(), i.e. incremented every time super_version_ // changes. std::atomic super_version_number_; // Thread's local copy of SuperVersion pointer // This needs to be destructed before mutex_ std::unique_ptr local_sv_; // pointers for a circular linked list. we use it to support iterations over // all column families that are alive (note: dropped column families can also // be alive as long as client holds a reference) ColumnFamilyData* next_; ColumnFamilyData* prev_; // This is the earliest log file number that contains data from this // Column Family. All earlier log files must be ignored and not // recovered from uint64_t log_number_; // An object that keeps all the compaction stats // and picks the next compaction std::unique_ptr compaction_picker_; ColumnFamilySet* column_family_set_; std::unique_ptr write_controller_token_; // If true --> this ColumnFamily is currently present in DBImpl::flush_queue_ bool pending_flush_; // If true --> this ColumnFamily is currently present in // DBImpl::compaction_queue_ bool pending_compaction_; uint64_t prev_compaction_needed_bytes_; }; // ColumnFamilySet has interesting thread-safety requirements // * CreateColumnFamily() or RemoveColumnFamily() -- need to be protected by DB // mutex AND executed in the write thread. // CreateColumnFamily() should ONLY be called from VersionSet::LogAndApply() AND // single-threaded write thread. It is also called during Recovery and in // DumpManifest(). // RemoveColumnFamily() is only called from SetDropped(). DB mutex needs to be // held and it needs to be executed from the write thread. SetDropped() also // guarantees that it will be called only from single-threaded LogAndApply(), // but this condition is not that important. // * Iteration -- hold DB mutex, but you can release it in the body of // iteration. If you release DB mutex in body, reference the column // family before the mutex and unreference after you unlock, since the column // family might get dropped when the DB mutex is released // * GetDefault() -- thread safe // * GetColumnFamily() -- either inside of DB mutex or from a write thread // * GetNextColumnFamilyID(), GetMaxColumnFamily(), UpdateMaxColumnFamily(), // NumberOfColumnFamilies -- inside of DB mutex class ColumnFamilySet { public: // ColumnFamilySet supports iteration class iterator { public: explicit iterator(ColumnFamilyData* cfd) : current_(cfd) {} iterator& operator++() { // dropped column families might still be included in this iteration // (we're only removing them when client drops the last reference to the // column family). // dummy is never dead, so this will never be infinite do { current_ = current_->next_; } while (current_->refs_.load(std::memory_order_relaxed) == 0); return *this; } bool operator!=(const iterator& other) { return this->current_ != other.current_; } ColumnFamilyData* operator*() { return current_; } private: ColumnFamilyData* current_; }; ColumnFamilySet(const std::string& dbname, const DBOptions* db_options, const EnvOptions& env_options, Cache* table_cache, WriteBuffer* write_buffer, WriteController* write_controller); ~ColumnFamilySet(); ColumnFamilyData* GetDefault() const; // GetColumnFamily() calls return nullptr if column family is not found ColumnFamilyData* GetColumnFamily(uint32_t id) const; ColumnFamilyData* GetColumnFamily(const std::string& name) const; // this call will return the next available column family ID. it guarantees // that there is no column family with id greater than or equal to the // returned value in the current running instance or anytime in RocksDB // instance history. uint32_t GetNextColumnFamilyID(); uint32_t GetMaxColumnFamily(); void UpdateMaxColumnFamily(uint32_t new_max_column_family); size_t NumberOfColumnFamilies() const; ColumnFamilyData* CreateColumnFamily(const std::string& name, uint32_t id, Version* dummy_version, const ColumnFamilyOptions& options); iterator begin() { return iterator(dummy_cfd_->next_); } iterator end() { return iterator(dummy_cfd_); } // REQUIRES: DB mutex held // Don't call while iterating over ColumnFamilySet void FreeDeadColumnFamilies(); private: friend class ColumnFamilyData; // helper function that gets called from cfd destructor // REQUIRES: DB mutex held void RemoveColumnFamily(ColumnFamilyData* cfd); // column_families_ and column_family_data_ need to be protected: // * when mutating both conditions have to be satisfied: // 1. DB mutex locked // 2. thread currently in single-threaded write thread // * when reading, at least one condition needs to be satisfied: // 1. DB mutex locked // 2. accessed from a single-threaded write thread std::unordered_map column_families_; std::unordered_map column_family_data_; uint32_t max_column_family_; ColumnFamilyData* dummy_cfd_; // We don't hold the refcount here, since default column family always exists // We are also not responsible for cleaning up default_cfd_cache_. This is // just a cache that makes common case (accessing default column family) // faster ColumnFamilyData* default_cfd_cache_; const std::string db_name_; const DBOptions* const db_options_; const EnvOptions env_options_; Cache* table_cache_; WriteBuffer* write_buffer_; WriteController* write_controller_; }; // We use ColumnFamilyMemTablesImpl to provide WriteBatch a way to access // memtables of different column families (specified by ID in the write batch) class ColumnFamilyMemTablesImpl : public ColumnFamilyMemTables { public: explicit ColumnFamilyMemTablesImpl(ColumnFamilySet* column_family_set) : column_family_set_(column_family_set), current_(nullptr) {} // Constructs a ColumnFamilyMemTablesImpl equivalent to one constructed // with the arguments used to construct *orig. explicit ColumnFamilyMemTablesImpl(ColumnFamilyMemTablesImpl* orig) : column_family_set_(orig->column_family_set_), current_(nullptr) {} // sets current_ to ColumnFamilyData with column_family_id // returns false if column family doesn't exist // REQUIRES: use this function of DBImpl::column_family_memtables_ should be // under a DB mutex OR from a write thread bool Seek(uint32_t column_family_id) override; // Returns log number of the selected column family // REQUIRES: under a DB mutex OR from a write thread uint64_t GetLogNumber() const override; // REQUIRES: Seek() called first // REQUIRES: use this function of DBImpl::column_family_memtables_ should be // under a DB mutex OR from a write thread virtual MemTable* GetMemTable() const override; // Returns column family handle for the selected column family // REQUIRES: use this function of DBImpl::column_family_memtables_ should be // under a DB mutex OR from a write thread virtual ColumnFamilyHandle* GetColumnFamilyHandle() override; // Cannot be called while another thread is calling Seek(). // REQUIRES: use this function of DBImpl::column_family_memtables_ should be // under a DB mutex OR from a write thread virtual ColumnFamilyData* current() override { return current_; } private: ColumnFamilySet* column_family_set_; ColumnFamilyData* current_; ColumnFamilyHandleInternal handle_; }; extern uint32_t GetColumnFamilyID(ColumnFamilyHandle* column_family); extern const Comparator* GetColumnFamilyUserComparator( ColumnFamilyHandle* column_family); } // namespace rocksdb rocksdb-4.5.1/db/column_family_test.cc000066400000000000000000002562751267530634000177700ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include #include #include #include #include "db/db_impl.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "util/coding.h" #include "util/options_parser.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/testharness.h" #include "util/testutil.h" #include "utilities/merge_operators.h" namespace rocksdb { namespace { std::string RandomString(Random* rnd, int len) { std::string r; test::RandomString(rnd, len, &r); return r; } } // anonymous namespace // counts how many operations were performed class EnvCounter : public EnvWrapper { public: explicit EnvCounter(Env* base) : EnvWrapper(base), num_new_writable_file_(0) {} int GetNumberOfNewWritableFileCalls() { return num_new_writable_file_; } Status NewWritableFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { ++num_new_writable_file_; return EnvWrapper::NewWritableFile(f, r, soptions); } private: int num_new_writable_file_; }; class ColumnFamilyTest : public testing::Test { public: ColumnFamilyTest() : rnd_(139) { env_ = new EnvCounter(Env::Default()); dbname_ = test::TmpDir() + "/column_family_test"; db_options_.create_if_missing = true; db_options_.fail_if_options_file_error = true; db_options_.env = env_; DestroyDB(dbname_, Options(db_options_, column_family_options_)); } ~ColumnFamilyTest() { Close(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); Destroy(); delete env_; } void Close() { for (auto h : handles_) { if (h) { delete h; } } handles_.clear(); names_.clear(); delete db_; db_ = nullptr; } Status TryOpen(std::vector cf, std::vector options = {}) { std::vector column_families; names_.clear(); for (size_t i = 0; i < cf.size(); ++i) { column_families.push_back(ColumnFamilyDescriptor( cf[i], options.size() == 0 ? column_family_options_ : options[i])); names_.push_back(cf[i]); } return DB::Open(db_options_, dbname_, column_families, &handles_, &db_); } Status OpenReadOnly(std::vector cf, std::vector options = {}) { std::vector column_families; names_.clear(); for (size_t i = 0; i < cf.size(); ++i) { column_families.push_back(ColumnFamilyDescriptor( cf[i], options.size() == 0 ? column_family_options_ : options[i])); names_.push_back(cf[i]); } return DB::OpenForReadOnly(db_options_, dbname_, column_families, &handles_, &db_); } #ifndef ROCKSDB_LITE // ReadOnlyDB is not supported void AssertOpenReadOnly(std::vector cf, std::vector options = {}) { ASSERT_OK(OpenReadOnly(cf, options)); } #endif // !ROCKSDB_LITE void Open(std::vector cf, std::vector options = {}) { ASSERT_OK(TryOpen(cf, options)); } void Open() { Open({"default"}); } DBImpl* dbfull() { return reinterpret_cast(db_); } int GetProperty(int cf, std::string property) { std::string value; EXPECT_TRUE(dbfull()->GetProperty(handles_[cf], property, &value)); #ifndef CYGWIN return std::stoi(value); #else return std::strtol(value.c_str(), 0 /* off */, 10 /* base */); #endif } void Destroy() { Close(); ASSERT_OK(DestroyDB(dbname_, Options(db_options_, column_family_options_))); } void CreateColumnFamilies( const std::vector& cfs, const std::vector options = {}) { int cfi = static_cast(handles_.size()); handles_.resize(cfi + cfs.size()); names_.resize(cfi + cfs.size()); for (size_t i = 0; i < cfs.size(); ++i) { const auto& current_cf_opt = options.size() == 0 ? column_family_options_ : options[i]; ASSERT_OK( db_->CreateColumnFamily(current_cf_opt, cfs[i], &handles_[cfi])); names_[cfi] = cfs[i]; #ifndef ROCKSDB_LITE // RocksDBLite does not support GetDescriptor // Verify the CF options of the returned CF handle. ColumnFamilyDescriptor desc; ASSERT_OK(handles_[cfi]->GetDescriptor(&desc)); RocksDBOptionsParser::VerifyCFOptions(desc.options, current_cf_opt); #endif // !ROCKSDB_LITE cfi++; } } void Reopen(const std::vector options = {}) { std::vector names; for (auto name : names_) { if (name != "") { names.push_back(name); } } Close(); assert(options.size() == 0 || names.size() == options.size()); Open(names, options); } void CreateColumnFamiliesAndReopen(const std::vector& cfs) { CreateColumnFamilies(cfs); Reopen(); } void DropColumnFamilies(const std::vector& cfs) { for (auto cf : cfs) { ASSERT_OK(db_->DropColumnFamily(handles_[cf])); delete handles_[cf]; handles_[cf] = nullptr; names_[cf] = ""; } } void PutRandomData(int cf, int num, int key_value_size, bool save = false) { for (int i = 0; i < num; ++i) { // 10 bytes for key, rest is value if (!save) { ASSERT_OK(Put(cf, test::RandomKey(&rnd_, 11), RandomString(&rnd_, key_value_size - 10))); } else { std::string key = test::RandomKey(&rnd_, 11); keys_.insert(key); ASSERT_OK(Put(cf, key, RandomString(&rnd_, key_value_size - 10))); } } } void WaitForFlush(int cf) { #ifndef ROCKSDB_LITE // TEST functions are not supported in lite ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable(handles_[cf])); #endif // !ROCKSDB_LITE } void WaitForCompaction() { #ifndef ROCKSDB_LITE // TEST functions are not supported in lite ASSERT_OK(dbfull()->TEST_WaitForCompact()); #endif // !ROCKSDB_LITE } uint64_t MaxTotalInMemoryState() { #ifndef ROCKSDB_LITE return dbfull()->TEST_MaxTotalInMemoryState(); #else return 0; #endif // !ROCKSDB_LITE } void AssertMaxTotalInMemoryState(uint64_t value) { ASSERT_EQ(value, MaxTotalInMemoryState()); } Status Put(int cf, const std::string& key, const std::string& value) { return db_->Put(WriteOptions(), handles_[cf], Slice(key), Slice(value)); } Status Merge(int cf, const std::string& key, const std::string& value) { return db_->Merge(WriteOptions(), handles_[cf], Slice(key), Slice(value)); } Status Flush(int cf) { return db_->Flush(FlushOptions(), handles_[cf]); } std::string Get(int cf, const std::string& key) { ReadOptions options; options.verify_checksums = true; std::string result; Status s = db_->Get(options, handles_[cf], Slice(key), &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } void CompactAll(int cf) { ASSERT_OK(db_->CompactRange(CompactRangeOptions(), handles_[cf], nullptr, nullptr)); } void Compact(int cf, const Slice& start, const Slice& limit) { ASSERT_OK( db_->CompactRange(CompactRangeOptions(), handles_[cf], &start, &limit)); } int NumTableFilesAtLevel(int level, int cf) { return GetProperty(cf, "rocksdb.num-files-at-level" + ToString(level)); } #ifndef ROCKSDB_LITE // Return spread of files per level std::string FilesPerLevel(int cf) { std::string result; int last_non_zero_offset = 0; for (int level = 0; level < dbfull()->NumberLevels(handles_[cf]); level++) { int f = NumTableFilesAtLevel(level, cf); char buf[100]; snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); result += buf; if (f > 0) { last_non_zero_offset = static_cast(result.size()); } } result.resize(last_non_zero_offset); return result; } #endif void AssertFilesPerLevel(const std::string& value, int cf) { #ifndef ROCKSDB_LITE ASSERT_EQ(value, FilesPerLevel(cf)); #endif } #ifndef ROCKSDB_LITE // GetLiveFilesMetaData is not supported int CountLiveFiles() { std::vector metadata; db_->GetLiveFilesMetaData(&metadata); return static_cast(metadata.size()); } #endif // !ROCKSDB_LITE void AssertCountLiveFiles(int expected_value) { #ifndef ROCKSDB_LITE ASSERT_EQ(expected_value, CountLiveFiles()); #endif } // Do n memtable flushes, each of which produces an sstable // covering the range [small,large]. void MakeTables(int cf, int n, const std::string& small, const std::string& large) { for (int i = 0; i < n; i++) { ASSERT_OK(Put(cf, small, "begin")); ASSERT_OK(Put(cf, large, "end")); ASSERT_OK(db_->Flush(FlushOptions(), handles_[cf])); } } #ifndef ROCKSDB_LITE // GetSortedWalFiles is not supported int CountLiveLogFiles() { int micros_wait_for_log_deletion = 20000; env_->SleepForMicroseconds(micros_wait_for_log_deletion); int ret = 0; VectorLogPtr wal_files; Status s; // GetSortedWalFiles is a flakey function -- it gets all the wal_dir // children files and then later checks for their existence. if some of the // log files doesn't exist anymore, it reports an error. it does all of this // without DB mutex held, so if a background process deletes the log file // while the function is being executed, it returns an error. We retry the // function 10 times to avoid the error failing the test for (int retries = 0; retries < 10; ++retries) { wal_files.clear(); s = db_->GetSortedWalFiles(wal_files); if (s.ok()) { break; } } EXPECT_OK(s); for (const auto& wal : wal_files) { if (wal->Type() == kAliveLogFile) { ++ret; } } return ret; return 0; } #endif // !ROCKSDB_LITE void AssertCountLiveLogFiles(int value) { #ifndef ROCKSDB_LITE // GetSortedWalFiles is not supported ASSERT_EQ(value, CountLiveLogFiles()); #endif // !ROCKSDB_LITE } void AssertNumberOfImmutableMemtables(std::vector num_per_cf) { assert(num_per_cf.size() == handles_.size()); #ifndef ROCKSDB_LITE // GetProperty is not supported in lite for (size_t i = 0; i < num_per_cf.size(); ++i) { ASSERT_EQ(num_per_cf[i], GetProperty(static_cast(i), "rocksdb.num-immutable-mem-table")); } #endif // !ROCKSDB_LITE } void CopyFile(const std::string& source, const std::string& destination, uint64_t size = 0) { const EnvOptions soptions; unique_ptr srcfile; ASSERT_OK(env_->NewSequentialFile(source, &srcfile, soptions)); unique_ptr destfile; ASSERT_OK(env_->NewWritableFile(destination, &destfile, soptions)); if (size == 0) { // default argument means copy everything ASSERT_OK(env_->GetFileSize(source, &size)); } char buffer[4096]; Slice slice; while (size > 0) { uint64_t one = std::min(uint64_t(sizeof(buffer)), size); ASSERT_OK(srcfile->Read(one, &slice, buffer)); ASSERT_OK(destfile->Append(slice)); size -= slice.size(); } ASSERT_OK(destfile->Close()); } std::vector handles_; std::vector names_; std::set keys_; ColumnFamilyOptions column_family_options_; DBOptions db_options_; std::string dbname_; DB* db_ = nullptr; EnvCounter* env_; Random rnd_; }; TEST_F(ColumnFamilyTest, DontReuseColumnFamilyID) { for (int iter = 0; iter < 3; ++iter) { Open(); CreateColumnFamilies({"one", "two", "three"}); for (size_t i = 0; i < handles_.size(); ++i) { auto cfh = reinterpret_cast(handles_[i]); ASSERT_EQ(i, cfh->GetID()); } if (iter == 1) { Reopen(); } DropColumnFamilies({3}); Reopen(); if (iter == 2) { // this tests if max_column_family is correctly persisted with // WriteSnapshot() Reopen(); } CreateColumnFamilies({"three2"}); // ID 3 that was used for dropped column family "three" should not be reused auto cfh3 = reinterpret_cast(handles_[3]); ASSERT_EQ(4U, cfh3->GetID()); Close(); Destroy(); } } TEST_F(ColumnFamilyTest, AddDrop) { Open(); CreateColumnFamilies({"one", "two", "three"}); ASSERT_EQ("NOT_FOUND", Get(1, "fodor")); ASSERT_EQ("NOT_FOUND", Get(2, "fodor")); DropColumnFamilies({2}); ASSERT_EQ("NOT_FOUND", Get(1, "fodor")); CreateColumnFamilies({"four"}); ASSERT_EQ("NOT_FOUND", Get(3, "fodor")); ASSERT_OK(Put(1, "fodor", "mirko")); ASSERT_EQ("mirko", Get(1, "fodor")); ASSERT_EQ("NOT_FOUND", Get(3, "fodor")); Close(); ASSERT_TRUE(TryOpen({"default"}).IsInvalidArgument()); Open({"default", "one", "three", "four"}); DropColumnFamilies({1}); Reopen(); Close(); std::vector families; ASSERT_OK(DB::ListColumnFamilies(db_options_, dbname_, &families)); sort(families.begin(), families.end()); ASSERT_TRUE(families == std::vector({"default", "four", "three"})); } TEST_F(ColumnFamilyTest, DropTest) { // first iteration - dont reopen DB before dropping // second iteration - reopen DB before dropping for (int iter = 0; iter < 2; ++iter) { Open({"default"}); CreateColumnFamiliesAndReopen({"pikachu"}); for (int i = 0; i < 100; ++i) { ASSERT_OK(Put(1, ToString(i), "bar" + ToString(i))); } ASSERT_OK(Flush(1)); if (iter == 1) { Reopen(); } ASSERT_EQ("bar1", Get(1, "1")); AssertCountLiveFiles(1); DropColumnFamilies({1}); // make sure that all files are deleted when we drop the column family AssertCountLiveFiles(0); Destroy(); } } TEST_F(ColumnFamilyTest, WriteBatchFailure) { Open(); CreateColumnFamiliesAndReopen({"one", "two"}); WriteBatch batch; batch.Put(handles_[0], Slice("existing"), Slice("column-family")); batch.Put(handles_[1], Slice("non-existing"), Slice("column-family")); ASSERT_OK(db_->Write(WriteOptions(), &batch)); DropColumnFamilies({1}); WriteOptions woptions_ignore_missing_cf; woptions_ignore_missing_cf.ignore_missing_column_families = true; batch.Put(handles_[0], Slice("still here"), Slice("column-family")); ASSERT_OK(db_->Write(woptions_ignore_missing_cf, &batch)); ASSERT_EQ("column-family", Get(0, "still here")); Status s = db_->Write(WriteOptions(), &batch); ASSERT_TRUE(s.IsInvalidArgument()); Close(); } TEST_F(ColumnFamilyTest, ReadWrite) { Open(); CreateColumnFamiliesAndReopen({"one", "two"}); ASSERT_OK(Put(0, "foo", "v1")); ASSERT_OK(Put(0, "bar", "v2")); ASSERT_OK(Put(1, "mirko", "v3")); ASSERT_OK(Put(0, "foo", "v2")); ASSERT_OK(Put(2, "fodor", "v5")); for (int iter = 0; iter <= 3; ++iter) { ASSERT_EQ("v2", Get(0, "foo")); ASSERT_EQ("v2", Get(0, "bar")); ASSERT_EQ("v3", Get(1, "mirko")); ASSERT_EQ("v5", Get(2, "fodor")); ASSERT_EQ("NOT_FOUND", Get(0, "fodor")); ASSERT_EQ("NOT_FOUND", Get(1, "fodor")); ASSERT_EQ("NOT_FOUND", Get(2, "foo")); if (iter <= 1) { Reopen(); } } Close(); } TEST_F(ColumnFamilyTest, IgnoreRecoveredLog) { std::string backup_logs = dbname_ + "/backup_logs"; // delete old files in backup_logs directory ASSERT_OK(env_->CreateDirIfMissing(dbname_)); ASSERT_OK(env_->CreateDirIfMissing(backup_logs)); std::vector old_files; env_->GetChildren(backup_logs, &old_files); for (auto& file : old_files) { if (file != "." && file != "..") { env_->DeleteFile(backup_logs + "/" + file); } } column_family_options_.merge_operator = MergeOperators::CreateUInt64AddOperator(); db_options_.wal_dir = dbname_ + "/logs"; Destroy(); Open(); CreateColumnFamilies({"cf1", "cf2"}); // fill up the DB std::string one, two, three; PutFixed64(&one, 1); PutFixed64(&two, 2); PutFixed64(&three, 3); ASSERT_OK(Merge(0, "foo", one)); ASSERT_OK(Merge(1, "mirko", one)); ASSERT_OK(Merge(0, "foo", one)); ASSERT_OK(Merge(2, "bla", one)); ASSERT_OK(Merge(2, "fodor", one)); ASSERT_OK(Merge(0, "bar", one)); ASSERT_OK(Merge(2, "bla", one)); ASSERT_OK(Merge(1, "mirko", two)); ASSERT_OK(Merge(1, "franjo", one)); // copy the logs to backup std::vector logs; env_->GetChildren(db_options_.wal_dir, &logs); for (auto& log : logs) { if (log != ".." && log != ".") { CopyFile(db_options_.wal_dir + "/" + log, backup_logs + "/" + log); } } // recover the DB Close(); // 1. check consistency // 2. copy the logs from backup back to WAL dir. if the recovery happens // again on the same log files, this should lead to incorrect results // due to applying merge operator twice // 3. check consistency for (int iter = 0; iter < 2; ++iter) { // assert consistency Open({"default", "cf1", "cf2"}); ASSERT_EQ(two, Get(0, "foo")); ASSERT_EQ(one, Get(0, "bar")); ASSERT_EQ(three, Get(1, "mirko")); ASSERT_EQ(one, Get(1, "franjo")); ASSERT_EQ(one, Get(2, "fodor")); ASSERT_EQ(two, Get(2, "bla")); Close(); if (iter == 0) { // copy the logs from backup back to wal dir for (auto& log : logs) { if (log != ".." && log != ".") { CopyFile(backup_logs + "/" + log, db_options_.wal_dir + "/" + log); } } } } } TEST_F(ColumnFamilyTest, FlushTest) { Open(); CreateColumnFamiliesAndReopen({"one", "two"}); ASSERT_OK(Put(0, "foo", "v1")); ASSERT_OK(Put(0, "bar", "v2")); ASSERT_OK(Put(1, "mirko", "v3")); ASSERT_OK(Put(0, "foo", "v2")); ASSERT_OK(Put(2, "fodor", "v5")); for (int j = 0; j < 2; j++) { ReadOptions ro; std::vector iterators; // Hold super version. if (j == 0) { ASSERT_OK(db_->NewIterators(ro, handles_, &iterators)); } for (int i = 0; i < 3; ++i) { uint64_t max_total_in_memory_state = MaxTotalInMemoryState(); Flush(i); AssertMaxTotalInMemoryState(max_total_in_memory_state); } ASSERT_OK(Put(1, "foofoo", "bar")); ASSERT_OK(Put(0, "foofoo", "bar")); for (auto* it : iterators) { delete it; } } Reopen(); for (int iter = 0; iter <= 2; ++iter) { ASSERT_EQ("v2", Get(0, "foo")); ASSERT_EQ("v2", Get(0, "bar")); ASSERT_EQ("v3", Get(1, "mirko")); ASSERT_EQ("v5", Get(2, "fodor")); ASSERT_EQ("NOT_FOUND", Get(0, "fodor")); ASSERT_EQ("NOT_FOUND", Get(1, "fodor")); ASSERT_EQ("NOT_FOUND", Get(2, "foo")); if (iter <= 1) { Reopen(); } } Close(); } // Makes sure that obsolete log files get deleted TEST_F(ColumnFamilyTest, LogDeletionTest) { db_options_.max_total_wal_size = std::numeric_limits::max(); column_family_options_.arena_block_size = 4 * 1024; column_family_options_.write_buffer_size = 100000; // 100KB Open(); CreateColumnFamilies({"one", "two", "three", "four"}); // Each bracket is one log file. if number is in (), it means // we don't need it anymore (it's been flushed) // [] AssertCountLiveLogFiles(0); PutRandomData(0, 1, 100); // [0] PutRandomData(1, 1, 100); // [0, 1] PutRandomData(1, 1000, 100); WaitForFlush(1); // [0, (1)] [1] AssertCountLiveLogFiles(2); PutRandomData(0, 1, 100); // [0, (1)] [0, 1] AssertCountLiveLogFiles(2); PutRandomData(2, 1, 100); // [0, (1)] [0, 1, 2] PutRandomData(2, 1000, 100); WaitForFlush(2); // [0, (1)] [0, 1, (2)] [2] AssertCountLiveLogFiles(3); PutRandomData(2, 1000, 100); WaitForFlush(2); // [0, (1)] [0, 1, (2)] [(2)] [2] AssertCountLiveLogFiles(4); PutRandomData(3, 1, 100); // [0, (1)] [0, 1, (2)] [(2)] [2, 3] PutRandomData(1, 1, 100); // [0, (1)] [0, 1, (2)] [(2)] [1, 2, 3] AssertCountLiveLogFiles(4); PutRandomData(1, 1000, 100); WaitForFlush(1); // [0, (1)] [0, (1), (2)] [(2)] [(1), 2, 3] [1] AssertCountLiveLogFiles(5); PutRandomData(0, 1000, 100); WaitForFlush(0); // [(0), (1)] [(0), (1), (2)] [(2)] [(1), 2, 3] [1, (0)] [0] // delete obsolete logs --> // [(1), 2, 3] [1, (0)] [0] AssertCountLiveLogFiles(3); PutRandomData(0, 1000, 100); WaitForFlush(0); // [(1), 2, 3] [1, (0)], [(0)] [0] AssertCountLiveLogFiles(4); PutRandomData(1, 1000, 100); WaitForFlush(1); // [(1), 2, 3] [(1), (0)] [(0)] [0, (1)] [1] AssertCountLiveLogFiles(5); PutRandomData(2, 1000, 100); WaitForFlush(2); // [(1), (2), 3] [(1), (0)] [(0)] [0, (1)] [1, (2)], [2] AssertCountLiveLogFiles(6); PutRandomData(3, 1000, 100); WaitForFlush(3); // [(1), (2), (3)] [(1), (0)] [(0)] [0, (1)] [1, (2)], [2, (3)] [3] // delete obsolete logs --> // [0, (1)] [1, (2)], [2, (3)] [3] AssertCountLiveLogFiles(4); Close(); } // Makes sure that obsolete log files get deleted TEST_F(ColumnFamilyTest, DifferentWriteBufferSizes) { // disable flushing stale column families db_options_.max_total_wal_size = std::numeric_limits::max(); Open(); CreateColumnFamilies({"one", "two", "three"}); ColumnFamilyOptions default_cf, one, two, three; // setup options. all column families have max_write_buffer_number setup to 10 // "default" -> 100KB memtable, start flushing immediatelly // "one" -> 200KB memtable, start flushing with two immutable memtables // "two" -> 1MB memtable, start flushing with three immutable memtables // "three" -> 90KB memtable, start flushing with four immutable memtables default_cf.write_buffer_size = 100000; default_cf.arena_block_size = 4 * 4096; default_cf.max_write_buffer_number = 10; default_cf.min_write_buffer_number_to_merge = 1; default_cf.max_write_buffer_number_to_maintain = 0; one.write_buffer_size = 200000; one.arena_block_size = 4 * 4096; one.max_write_buffer_number = 10; one.min_write_buffer_number_to_merge = 2; one.max_write_buffer_number_to_maintain = 1; two.write_buffer_size = 1000000; two.arena_block_size = 4 * 4096; two.max_write_buffer_number = 10; two.min_write_buffer_number_to_merge = 3; two.max_write_buffer_number_to_maintain = 2; three.write_buffer_size = 4096 * 22; three.arena_block_size = 4096; three.max_write_buffer_number = 10; three.min_write_buffer_number_to_merge = 4; three.max_write_buffer_number_to_maintain = -1; Reopen({default_cf, one, two, three}); int micros_wait_for_flush = 10000; PutRandomData(0, 100, 1000); WaitForFlush(0); AssertNumberOfImmutableMemtables({0, 0, 0, 0}); AssertCountLiveLogFiles(1); PutRandomData(1, 200, 1000); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 0, 0}); AssertCountLiveLogFiles(2); PutRandomData(2, 1000, 1000); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 1, 0}); AssertCountLiveLogFiles(3); PutRandomData(2, 1000, 1000); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 2, 0}); AssertCountLiveLogFiles(4); PutRandomData(3, 93, 990); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 2, 1}); AssertCountLiveLogFiles(5); PutRandomData(3, 88, 990); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 2, 2}); AssertCountLiveLogFiles(6); PutRandomData(3, 88, 990); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 2, 3}); AssertCountLiveLogFiles(7); PutRandomData(0, 100, 1000); WaitForFlush(0); AssertNumberOfImmutableMemtables({0, 1, 2, 3}); AssertCountLiveLogFiles(8); PutRandomData(2, 100, 10000); WaitForFlush(2); AssertNumberOfImmutableMemtables({0, 1, 0, 3}); AssertCountLiveLogFiles(9); PutRandomData(3, 88, 990); WaitForFlush(3); AssertNumberOfImmutableMemtables({0, 1, 0, 0}); AssertCountLiveLogFiles(10); PutRandomData(3, 88, 990); env_->SleepForMicroseconds(micros_wait_for_flush); AssertNumberOfImmutableMemtables({0, 1, 0, 1}); AssertCountLiveLogFiles(11); PutRandomData(1, 200, 1000); WaitForFlush(1); AssertNumberOfImmutableMemtables({0, 0, 0, 1}); AssertCountLiveLogFiles(5); PutRandomData(3, 88 * 3, 990); WaitForFlush(3); PutRandomData(3, 88 * 4, 990); WaitForFlush(3); AssertNumberOfImmutableMemtables({0, 0, 0, 0}); AssertCountLiveLogFiles(12); PutRandomData(0, 100, 1000); WaitForFlush(0); AssertNumberOfImmutableMemtables({0, 0, 0, 0}); AssertCountLiveLogFiles(12); PutRandomData(2, 3 * 1000, 1000); WaitForFlush(2); AssertNumberOfImmutableMemtables({0, 0, 0, 0}); AssertCountLiveLogFiles(12); PutRandomData(1, 2*200, 1000); WaitForFlush(1); AssertNumberOfImmutableMemtables({0, 0, 0, 0}); AssertCountLiveLogFiles(7); Close(); } #ifndef ROCKSDB_LITE // Cuckoo is not supported in lite TEST_F(ColumnFamilyTest, MemtableNotSupportSnapshot) { Open(); auto* s1 = dbfull()->GetSnapshot(); ASSERT_TRUE(s1 != nullptr); dbfull()->ReleaseSnapshot(s1); // Add a column family that doesn't support snapshot ColumnFamilyOptions first; first.memtable_factory.reset(NewHashCuckooRepFactory(1024 * 1024)); CreateColumnFamilies({"first"}, {first}); auto* s2 = dbfull()->GetSnapshot(); ASSERT_TRUE(s2 == nullptr); // Add a column family that supports snapshot. Snapshot stays not supported. ColumnFamilyOptions second; CreateColumnFamilies({"second"}, {second}); auto* s3 = dbfull()->GetSnapshot(); ASSERT_TRUE(s3 == nullptr); Close(); } #endif // !ROCKSDB_LITE TEST_F(ColumnFamilyTest, DifferentMergeOperators) { Open(); CreateColumnFamilies({"first", "second"}); ColumnFamilyOptions default_cf, first, second; first.merge_operator = MergeOperators::CreateUInt64AddOperator(); second.merge_operator = MergeOperators::CreateStringAppendOperator(); Reopen({default_cf, first, second}); std::string one, two, three; PutFixed64(&one, 1); PutFixed64(&two, 2); PutFixed64(&three, 3); ASSERT_OK(Put(0, "foo", two)); ASSERT_OK(Put(0, "foo", one)); ASSERT_TRUE(Merge(0, "foo", two).IsNotSupported()); ASSERT_EQ(Get(0, "foo"), one); ASSERT_OK(Put(1, "foo", two)); ASSERT_OK(Put(1, "foo", one)); ASSERT_OK(Merge(1, "foo", two)); ASSERT_EQ(Get(1, "foo"), three); ASSERT_OK(Put(2, "foo", two)); ASSERT_OK(Put(2, "foo", one)); ASSERT_OK(Merge(2, "foo", two)); ASSERT_EQ(Get(2, "foo"), one + "," + two); Close(); } TEST_F(ColumnFamilyTest, DifferentCompactionStyles) { Open(); CreateColumnFamilies({"one", "two"}); ColumnFamilyOptions default_cf, one, two; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; two.compaction_style = kCompactionStyleLevel; two.num_levels = 4; two.level0_file_num_compaction_trigger = 3; two.write_buffer_size = 100000; Reopen({default_cf, one, two}); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger - 1; ++i) { PutRandomData(1, 10, 12000); PutRandomData(1, 1, 10); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } // SETUP column family "two" -- level style with 4 levels for (int i = 0; i < two.level0_file_num_compaction_trigger - 1; ++i) { PutRandomData(2, 10, 12000); PutRandomData(2, 1, 10); WaitForFlush(2); AssertFilesPerLevel(ToString(i + 1), 2); } // TRIGGER compaction "one" PutRandomData(1, 10, 12000); PutRandomData(1, 1, 10); // TRIGGER compaction "two" PutRandomData(2, 10, 12000); PutRandomData(2, 1, 10); // WAIT for compactions WaitForCompaction(); // VERIFY compaction "one" AssertFilesPerLevel("1", 1); // VERIFY compaction "two" AssertFilesPerLevel("0,1", 2); CompactAll(2); AssertFilesPerLevel("0,1", 2); Close(); } #ifndef ROCKSDB_LITE // Sync points not supported in RocksDB Lite TEST_F(ColumnFamilyTest, MultipleManualCompactions) { Open(); CreateColumnFamilies({"one", "two"}); ColumnFamilyOptions default_cf, one, two; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; two.compaction_style = kCompactionStyleLevel; two.num_levels = 4; two.level0_file_num_compaction_trigger = 3; two.write_buffer_size = 100000; Reopen({default_cf, one, two}); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } bool cf_1_1 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::MultiManual:4", "ColumnFamilyTest::MultiManual:1"}, {"ColumnFamilyTest::MultiManual:2", "ColumnFamilyTest::MultiManual:5"}, {"ColumnFamilyTest::MultiManual:2", "ColumnFamilyTest::MultiManual:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::MultiManual:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::MultiManual:3"); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::vector threads; threads.emplace_back([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); }); // SETUP column family "two" -- level style with 4 levels for (int i = 0; i < two.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(2, 10, 12000); PutRandomData(2, 1, 10); WaitForFlush(2); AssertFilesPerLevel(ToString(i + 1), 2); } threads.emplace_back([&] { TEST_SYNC_POINT("ColumnFamilyTest::MultiManual:1"); CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[2], nullptr, nullptr)); TEST_SYNC_POINT("ColumnFamilyTest::MultiManual:2"); }); TEST_SYNC_POINT("ColumnFamilyTest::MultiManual:5"); for (auto& t : threads) { t.join(); } // VERIFY compaction "one" AssertFilesPerLevel("1", 1); // VERIFY compaction "two" AssertFilesPerLevel("0,1", 2); CompactAll(2); AssertFilesPerLevel("0,1", 2); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } TEST_F(ColumnFamilyTest, AutomaticAndManualCompactions) { Open(); CreateColumnFamilies({"one", "two"}); ColumnFamilyOptions default_cf, one, two; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; two.compaction_style = kCompactionStyleLevel; two.num_levels = 4; two.level0_file_num_compaction_trigger = 3; two.write_buffer_size = 100000; Reopen({default_cf, one, two}); bool cf_1_1 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::AutoManual:4", "ColumnFamilyTest::AutoManual:1"}, {"ColumnFamilyTest::AutoManual:2", "ColumnFamilyTest::AutoManual:5"}, {"ColumnFamilyTest::AutoManual:2", "ColumnFamilyTest::AutoManual:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:4"); TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:3"); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:1"); // SETUP column family "two" -- level style with 4 levels for (int i = 0; i < two.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(2, 10, 12000); PutRandomData(2, 1, 10); WaitForFlush(2); AssertFilesPerLevel(ToString(i + 1), 2); } std::thread threads([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[2], nullptr, nullptr)); TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:2"); }); TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:5"); threads.join(); // WAIT for compactions WaitForCompaction(); // VERIFY compaction "one" AssertFilesPerLevel("1", 1); // VERIFY compaction "two" AssertFilesPerLevel("0,1", 2); CompactAll(2); AssertFilesPerLevel("0,1", 2); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } TEST_F(ColumnFamilyTest, ManualAndAutomaticCompactions) { Open(); CreateColumnFamilies({"one", "two"}); ColumnFamilyOptions default_cf, one, two; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; two.compaction_style = kCompactionStyleLevel; two.num_levels = 4; two.level0_file_num_compaction_trigger = 3; two.write_buffer_size = 100000; Reopen({default_cf, one, two}); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } bool cf_1_1 = true; bool cf_1_2 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:1"}, {"ColumnFamilyTest::ManualAuto:5", "ColumnFamilyTest::ManualAuto:2"}, {"ColumnFamilyTest::ManualAuto:2", "ColumnFamilyTest::ManualAuto:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:3"); } else if (cf_1_2) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:2"); cf_1_2 = false; } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread threads([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); }); TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:1"); // SETUP column family "two" -- level style with 4 levels for (int i = 0; i < two.level0_file_num_compaction_trigger; ++i) { PutRandomData(2, 10, 12000); PutRandomData(2, 1, 10); WaitForFlush(2); AssertFilesPerLevel(ToString(i + 1), 2); } TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:5"); threads.join(); // WAIT for compactions WaitForCompaction(); // VERIFY compaction "one" AssertFilesPerLevel("1", 1); // VERIFY compaction "two" AssertFilesPerLevel("0,1", 2); CompactAll(2); AssertFilesPerLevel("0,1", 2); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } TEST_F(ColumnFamilyTest, SameCFManualManualCompactions) { Open(); CreateColumnFamilies({"one"}); ColumnFamilyOptions default_cf, one; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; Reopen({default_cf, one}); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } bool cf_1_1 = true; bool cf_1_2 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::ManualManual:4", "ColumnFamilyTest::ManualManual:2"}, {"ColumnFamilyTest::ManualManual:4", "ColumnFamilyTest::ManualManual:5"}, {"ColumnFamilyTest::ManualManual:1", "ColumnFamilyTest::ManualManual:2"}, {"ColumnFamilyTest::ManualManual:1", "ColumnFamilyTest::ManualManual:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::ManualManual:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::ManualManual:3"); } else if (cf_1_2) { TEST_SYNC_POINT("ColumnFamilyTest::ManualManual:2"); cf_1_2 = false; } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread threads([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = true; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); }); TEST_SYNC_POINT("ColumnFamilyTest::ManualManual:5"); WaitForFlush(1); // Add more L0 files and force another manual compaction for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(one.level0_file_num_compaction_trigger + i), 1); } std::thread threads1([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); }); TEST_SYNC_POINT("ColumnFamilyTest::ManualManual:1"); threads.join(); threads1.join(); WaitForCompaction(); // VERIFY compaction "one" ASSERT_LE(NumTableFilesAtLevel(0, 1), 2); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactions) { Open(); CreateColumnFamilies({"one"}); ColumnFamilyOptions default_cf, one; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; Reopen({default_cf, one}); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } bool cf_1_1 = true; bool cf_1_2 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:2"}, {"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:5"}, {"ColumnFamilyTest::ManualAuto:1", "ColumnFamilyTest::ManualAuto:2"}, {"ColumnFamilyTest::ManualAuto:1", "ColumnFamilyTest::ManualAuto:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:3"); } else if (cf_1_2) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:2"); cf_1_2 = false; } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread threads([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); }); TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:5"); WaitForFlush(1); // Add more L0 files and force automatic compaction for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(one.level0_file_num_compaction_trigger + i), 1); } TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:1"); threads.join(); WaitForCompaction(); // VERIFY compaction "one" ASSERT_LE(NumTableFilesAtLevel(0, 1), 2); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) { Open(); CreateColumnFamilies({"one"}); ColumnFamilyOptions default_cf, one; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleLevel; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; Reopen({default_cf, one}); // SETUP column family "one" -- level style for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } bool cf_1_1 = true; bool cf_1_2 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:2"}, {"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:5"}, {"ColumnFamilyTest::ManualAuto:3", "ColumnFamilyTest::ManualAuto:2"}, {"LevelCompactionPicker::PickCompactionBySize:0", "ColumnFamilyTest::ManualAuto:3"}, {"ColumnFamilyTest::ManualAuto:1", "ColumnFamilyTest::ManualAuto:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:3"); } else if (cf_1_2) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:2"); cf_1_2 = false; } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread threads([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); }); TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:5"); // Add more L0 files and force automatic compaction for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(one.level0_file_num_compaction_trigger + i), 1); } TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:1"); threads.join(); WaitForCompaction(); // VERIFY compaction "one" AssertFilesPerLevel("0,1", 1); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } // This test checks for automatic getting a conflict if there is a // manual which has not yet been scheduled. // The manual compaction waits in NotScheduled // We generate more files and then trigger an automatic compaction // This will wait because there is an unscheduled manual compaction. // Once the conflict is hit, the manual compaction starts and ends // Then another automatic will start and end. TEST_F(ColumnFamilyTest, SameCFManualAutomaticConflict) { Open(); CreateColumnFamilies({"one"}); ColumnFamilyOptions default_cf, one; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; Reopen({default_cf, one}); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } bool cf_1_1 = true; bool cf_1_2 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"DBImpl::BackgroundCompaction()::Conflict", "ColumnFamilyTest::ManualAutoCon:7"}, {"ColumnFamilyTest::ManualAutoCon:9", "ColumnFamilyTest::ManualAutoCon:8"}, {"ColumnFamilyTest::ManualAutoCon:2", "ColumnFamilyTest::ManualAutoCon:6"}, {"ColumnFamilyTest::ManualAutoCon:4", "ColumnFamilyTest::ManualAutoCon:5"}, {"ColumnFamilyTest::ManualAutoCon:1", "ColumnFamilyTest::ManualAutoCon:2"}, {"ColumnFamilyTest::ManualAutoCon:1", "ColumnFamilyTest::ManualAutoCon:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:3"); } else if (cf_1_2) { cf_1_2 = false; TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:2"); } }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::RunManualCompaction:NotScheduled", [&](void* arg) { InstrumentedMutex* mutex = static_cast(arg); mutex->Unlock(); TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:9"); TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:7"); mutex->Lock(); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread threads([&] { CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK( db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:6"); }); TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:8"); WaitForFlush(1); // Add more L0 files and force automatic compaction for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(one.level0_file_num_compaction_trigger + i), 1); } TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:5"); // Add more L0 files and force automatic compaction for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); } TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:1"); threads.join(); WaitForCompaction(); // VERIFY compaction "one" ASSERT_LE(NumTableFilesAtLevel(0, 1), 3); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } // In this test, we generate enough files to trigger automatic compactions. // The automatic compaction waits in NonTrivial:AfterRun // We generate more files and then trigger an automatic compaction // This will wait because the automatic compaction has files it needs. // Once the conflict is hit, the automatic compaction starts and ends // Then the manual will run and end. TEST_F(ColumnFamilyTest, SameCFAutomaticManualCompactions) { Open(); CreateColumnFamilies({"one"}); ColumnFamilyOptions default_cf, one; db_options_.max_open_files = 20; // only 10 files in file cache db_options_.disableDataSync = true; db_options_.max_background_compactions = 3; default_cf.compaction_style = kCompactionStyleLevel; default_cf.num_levels = 3; default_cf.write_buffer_size = 64 << 10; // 64KB default_cf.target_file_size_base = 30 << 10; default_cf.source_compaction_factor = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options)); one.compaction_style = kCompactionStyleUniversal; one.num_levels = 1; // trigger compaction if there are >= 4 files one.level0_file_num_compaction_trigger = 4; one.write_buffer_size = 120000; Reopen({default_cf, one}); bool cf_1_1 = true; bool cf_1_2 = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"ColumnFamilyTest::AutoManual:4", "ColumnFamilyTest::AutoManual:2"}, {"ColumnFamilyTest::AutoManual:4", "ColumnFamilyTest::AutoManual:5"}, {"CompactionPicker::CompactRange:Conflict", "ColumnFamilyTest::AutoManual:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (cf_1_1) { TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:4"); cf_1_1 = false; TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:3"); } else if (cf_1_2) { TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:2"); cf_1_2 = false; } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // SETUP column family "one" -- universal style for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); AssertFilesPerLevel(ToString(i + 1), 1); } TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:5"); // Add another L0 file and force automatic compaction for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) { PutRandomData(1, 10, 12000, true); PutRandomData(1, 1, 10, true); WaitForFlush(1); } CompactRangeOptions compact_options; compact_options.exclusive_manual_compaction = false; ASSERT_OK(db_->CompactRange(compact_options, handles_[1], nullptr, nullptr)); TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:1"); WaitForCompaction(); // VERIFY compaction "one" AssertFilesPerLevel("1", 1); // Compare against saved keys std::set::iterator key_iter = keys_.begin(); while (key_iter != keys_.end()) { ASSERT_NE("NOT_FOUND", Get(1, *key_iter)); key_iter++; } Close(); } #endif // !ROCKSDB_LITE #ifndef ROCKSDB_LITE // Tailing interator not supported namespace { std::string IterStatus(Iterator* iter) { std::string result; if (iter->Valid()) { result = iter->key().ToString() + "->" + iter->value().ToString(); } else { result = "(invalid)"; } return result; } } // anonymous namespace TEST_F(ColumnFamilyTest, NewIteratorsTest) { // iter == 0 -- no tailing // iter == 2 -- tailing for (int iter = 0; iter < 2; ++iter) { Open(); CreateColumnFamiliesAndReopen({"one", "two"}); ASSERT_OK(Put(0, "a", "b")); ASSERT_OK(Put(1, "b", "a")); ASSERT_OK(Put(2, "c", "m")); ASSERT_OK(Put(2, "v", "t")); std::vector iterators; ReadOptions options; options.tailing = (iter == 1); ASSERT_OK(db_->NewIterators(options, handles_, &iterators)); for (auto it : iterators) { it->SeekToFirst(); } ASSERT_EQ(IterStatus(iterators[0]), "a->b"); ASSERT_EQ(IterStatus(iterators[1]), "b->a"); ASSERT_EQ(IterStatus(iterators[2]), "c->m"); ASSERT_OK(Put(1, "x", "x")); for (auto it : iterators) { it->Next(); } ASSERT_EQ(IterStatus(iterators[0]), "(invalid)"); if (iter == 0) { // no tailing ASSERT_EQ(IterStatus(iterators[1]), "(invalid)"); } else { // tailing ASSERT_EQ(IterStatus(iterators[1]), "x->x"); } ASSERT_EQ(IterStatus(iterators[2]), "v->t"); for (auto it : iterators) { delete it; } Destroy(); } } #endif // !ROCKSDB_LITE #ifndef ROCKSDB_LITE // ReadOnlyDB is not supported TEST_F(ColumnFamilyTest, ReadOnlyDBTest) { Open(); CreateColumnFamiliesAndReopen({"one", "two", "three", "four"}); ASSERT_OK(Put(0, "a", "b")); ASSERT_OK(Put(1, "foo", "bla")); ASSERT_OK(Put(2, "foo", "blabla")); ASSERT_OK(Put(3, "foo", "blablabla")); ASSERT_OK(Put(4, "foo", "blablablabla")); DropColumnFamilies({2}); Close(); // open only a subset of column families AssertOpenReadOnly({"default", "one", "four"}); ASSERT_EQ("NOT_FOUND", Get(0, "foo")); ASSERT_EQ("bla", Get(1, "foo")); ASSERT_EQ("blablablabla", Get(2, "foo")); // test newiterators { std::vector iterators; ASSERT_OK(db_->NewIterators(ReadOptions(), handles_, &iterators)); for (auto it : iterators) { it->SeekToFirst(); } ASSERT_EQ(IterStatus(iterators[0]), "a->b"); ASSERT_EQ(IterStatus(iterators[1]), "foo->bla"); ASSERT_EQ(IterStatus(iterators[2]), "foo->blablablabla"); for (auto it : iterators) { it->Next(); } ASSERT_EQ(IterStatus(iterators[0]), "(invalid)"); ASSERT_EQ(IterStatus(iterators[1]), "(invalid)"); ASSERT_EQ(IterStatus(iterators[2]), "(invalid)"); for (auto it : iterators) { delete it; } } Close(); // can't open dropped column family Status s = OpenReadOnly({"default", "one", "two"}); ASSERT_TRUE(!s.ok()); // Can't open without specifying default column family s = OpenReadOnly({"one", "four"}); ASSERT_TRUE(!s.ok()); } #endif // !ROCKSDB_LITE TEST_F(ColumnFamilyTest, DontRollEmptyLogs) { Open(); CreateColumnFamiliesAndReopen({"one", "two", "three", "four"}); for (size_t i = 0; i < handles_.size(); ++i) { PutRandomData(static_cast(i), 10, 100); } int num_writable_file_start = env_->GetNumberOfNewWritableFileCalls(); // this will trigger the flushes for (int i = 0; i <= 4; ++i) { ASSERT_OK(Flush(i)); } for (int i = 0; i < 4; ++i) { WaitForFlush(i); } int total_new_writable_files = env_->GetNumberOfNewWritableFileCalls() - num_writable_file_start; ASSERT_EQ(static_cast(total_new_writable_files), handles_.size() + 1); Close(); } TEST_F(ColumnFamilyTest, FlushStaleColumnFamilies) { Open(); CreateColumnFamilies({"one", "two"}); ColumnFamilyOptions default_cf, one, two; default_cf.write_buffer_size = 100000; // small write buffer size default_cf.arena_block_size = 4096; default_cf.disable_auto_compactions = true; one.disable_auto_compactions = true; two.disable_auto_compactions = true; db_options_.max_total_wal_size = 210000; Reopen({default_cf, one, two}); PutRandomData(2, 1, 10); // 10 bytes for (int i = 0; i < 2; ++i) { PutRandomData(0, 100, 1000); // flush WaitForFlush(0); AssertCountLiveFiles(i + 1); } // third flush. now, CF [two] should be detected as stale and flushed // column family 1 should not be flushed since it's empty PutRandomData(0, 100, 1000); // flush WaitForFlush(0); WaitForFlush(2); // 3 files for default column families, 1 file for column family [two], zero // files for column family [one], because it's empty AssertCountLiveFiles(4); Close(); } TEST_F(ColumnFamilyTest, CreateMissingColumnFamilies) { Status s = TryOpen({"one", "two"}); ASSERT_TRUE(!s.ok()); db_options_.create_missing_column_families = true; s = TryOpen({"default", "one", "two"}); ASSERT_TRUE(s.ok()); Close(); } TEST_F(ColumnFamilyTest, SanitizeOptions) { DBOptions db_options; for (int s = kCompactionStyleLevel; s <= kCompactionStyleUniversal; ++s) { for (int l = 0; l <= 2; l++) { for (int i = 1; i <= 3; i++) { for (int j = 1; j <= 3; j++) { for (int k = 1; k <= 3; k++) { ColumnFamilyOptions original; original.compaction_style = static_cast(s); original.num_levels = l; original.level0_stop_writes_trigger = i; original.level0_slowdown_writes_trigger = j; original.level0_file_num_compaction_trigger = k; original.write_buffer_size = l * 4 * 1024 * 1024 + i * 1024 * 1024 + j * 1024 + k; ColumnFamilyOptions result = SanitizeOptions(db_options, nullptr, original); ASSERT_TRUE(result.level0_stop_writes_trigger >= result.level0_slowdown_writes_trigger); ASSERT_TRUE(result.level0_slowdown_writes_trigger >= result.level0_file_num_compaction_trigger); ASSERT_TRUE(result.level0_file_num_compaction_trigger == original.level0_file_num_compaction_trigger); if (s == kCompactionStyleLevel) { ASSERT_GE(result.num_levels, 2); } else { ASSERT_GE(result.num_levels, 1); if (original.num_levels >= 1) { ASSERT_EQ(result.num_levels, original.num_levels); } } // Make sure Sanitize options sets arena_block_size to 1/8 of // the write_buffer_size, rounded up to a multiple of 4k. size_t expected_arena_block_size = l * 4 * 1024 * 1024 / 8 + i * 1024 * 1024 / 8; if (j + k != 0) { // not a multiple of 4k, round up 4k expected_arena_block_size += 4 * 1024; } ASSERT_EQ(expected_arena_block_size, result.arena_block_size); } } } } } } TEST_F(ColumnFamilyTest, ReadDroppedColumnFamily) { // iter 0 -- drop CF, don't reopen // iter 1 -- delete CF, reopen for (int iter = 0; iter < 2; ++iter) { db_options_.create_missing_column_families = true; db_options_.max_open_files = 20; // delete obsolete files always db_options_.delete_obsolete_files_period_micros = 0; Open({"default", "one", "two"}); ColumnFamilyOptions options; options.level0_file_num_compaction_trigger = 100; options.level0_slowdown_writes_trigger = 200; options.level0_stop_writes_trigger = 200; options.write_buffer_size = 100000; // small write buffer size Reopen({options, options, options}); // 1MB should create ~10 files for each CF int kKeysNum = 10000; PutRandomData(0, kKeysNum, 100); PutRandomData(1, kKeysNum, 100); PutRandomData(2, kKeysNum, 100); if (iter == 0) { // Drop CF two ASSERT_OK(db_->DropColumnFamily(handles_[2])); } else { // delete CF two delete handles_[2]; handles_[2] = nullptr; } // Add bunch more data to other CFs PutRandomData(0, kKeysNum, 100); PutRandomData(1, kKeysNum, 100); if (iter == 1) { Reopen(); } // Since we didn't delete CF handle, RocksDB's contract guarantees that // we're still able to read dropped CF for (int i = 0; i < 3; ++i) { std::unique_ptr iterator( db_->NewIterator(ReadOptions(), handles_[i])); int count = 0; for (iterator->SeekToFirst(); iterator->Valid(); iterator->Next()) { ASSERT_OK(iterator->status()); ++count; } ASSERT_OK(iterator->status()); ASSERT_EQ(count, kKeysNum * ((i == 2) ? 1 : 2)); } Close(); Destroy(); } } TEST_F(ColumnFamilyTest, FlushAndDropRaceCondition) { db_options_.create_missing_column_families = true; Open({"default", "one"}); ColumnFamilyOptions options; options.level0_file_num_compaction_trigger = 100; options.level0_slowdown_writes_trigger = 200; options.level0_stop_writes_trigger = 200; options.max_write_buffer_number = 20; options.write_buffer_size = 100000; // small write buffer size Reopen({options, options}); rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"VersionSet::LogAndApply::ColumnFamilyDrop:0", "FlushJob::WriteLevel0Table"}, {"VersionSet::LogAndApply::ColumnFamilyDrop:1", "FlushJob::InstallResults"}, {"FlushJob::InstallResults", "VersionSet::LogAndApply::ColumnFamilyDrop:2"}}); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); test::SleepingBackgroundTask sleeping_task; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task, Env::Priority::HIGH); // 1MB should create ~10 files for each CF int kKeysNum = 10000; PutRandomData(1, kKeysNum, 100); std::vector threads; threads.emplace_back([&] { ASSERT_OK(db_->DropColumnFamily(handles_[1])); }); sleeping_task.WakeUp(); sleeping_task.WaitUntilDone(); sleeping_task.Reset(); // now we sleep again. this is just so we're certain that flush job finished env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task, Env::Priority::HIGH); sleeping_task.WakeUp(); sleeping_task.WaitUntilDone(); { // Since we didn't delete CF handle, RocksDB's contract guarantees that // we're still able to read dropped CF std::unique_ptr iterator( db_->NewIterator(ReadOptions(), handles_[1])); int count = 0; for (iterator->SeekToFirst(); iterator->Valid(); iterator->Next()) { ASSERT_OK(iterator->status()); ++count; } ASSERT_OK(iterator->status()); ASSERT_EQ(count, kKeysNum); } for (auto& t : threads) { t.join(); } Close(); Destroy(); } #ifndef ROCKSDB_LITE // skipped as persisting options is not supported in ROCKSDB_LITE namespace { std::atomic test_stage(0); const int kMainThreadStartPersistingOptionsFile = 1; const int kChildThreadFinishDroppingColumnFamily = 2; const int kChildThreadWaitingMainThreadPersistOptions = 3; void DropSingleColumnFamily(ColumnFamilyTest* cf_test, int cf_id, std::vector* comparators) { while (test_stage < kMainThreadStartPersistingOptionsFile) { Env::Default()->SleepForMicroseconds(100); } cf_test->DropColumnFamilies({cf_id}); if ((*comparators)[cf_id]) { delete (*comparators)[cf_id]; (*comparators)[cf_id] = nullptr; } test_stage = kChildThreadFinishDroppingColumnFamily; } } // namespace TEST_F(ColumnFamilyTest, CreateAndDropRace) { const int kCfCount = 5; std::vector cf_opts; std::vector comparators; for (int i = 0; i < kCfCount; ++i) { cf_opts.emplace_back(); comparators.push_back(new test::SimpleSuffixReverseComparator()); cf_opts.back().comparator = comparators.back(); } db_options_.create_if_missing = true; db_options_.create_missing_column_families = true; auto main_thread_id = std::this_thread::get_id(); rocksdb::SyncPoint::GetInstance()->SetCallBack("PersistRocksDBOptions:start", [&](void* arg) { auto current_thread_id = std::this_thread::get_id(); // If it's the main thread hitting this sync-point, then it // will be blocked until some other thread update the test_stage. if (main_thread_id == current_thread_id) { test_stage = kMainThreadStartPersistingOptionsFile; while (test_stage < kChildThreadFinishDroppingColumnFamily) { Env::Default()->SleepForMicroseconds(100); } } }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "WriteThread::EnterUnbatched:Wait", [&](void* arg) { // This means a thread doing DropColumnFamily() is waiting for // other thread to finish persisting options. // In such case, we update the test_stage to unblock the main thread. test_stage = kChildThreadWaitingMainThreadPersistOptions; // Note that based on the test setting, this must not be the // main thread. ASSERT_NE(main_thread_id, std::this_thread::get_id()); }); // Create a database with four column families Open({"default", "one", "two", "three"}, {cf_opts[0], cf_opts[1], cf_opts[2], cf_opts[3]}); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // Start a thread that will drop the first column family // and its comparator std::thread drop_cf_thread(DropSingleColumnFamily, this, 1, &comparators); DropColumnFamilies({2}); drop_cf_thread.join(); Close(); Destroy(); for (auto* comparator : comparators) { if (comparator) { delete comparator; } } } #endif // !ROCKSDB_LITE TEST_F(ColumnFamilyTest, WriteStallSingleColumnFamily) { const uint64_t kBaseRate = 810000u; db_options_.delayed_write_rate = kBaseRate; db_options_.base_background_compactions = 2; db_options_.max_background_compactions = 6; Open({"default"}); ColumnFamilyData* cfd = static_cast(db_->DefaultColumnFamily())->cfd(); VersionStorageInfo* vstorage = cfd->current()->storage_info(); MutableCFOptions mutable_cf_options( Options(db_options_, column_family_options_), ImmutableCFOptions(Options(db_options_, column_family_options_))); mutable_cf_options.level0_slowdown_writes_trigger = 20; mutable_cf_options.level0_stop_writes_trigger = 10000; mutable_cf_options.soft_pending_compaction_bytes_limit = 200; mutable_cf_options.hard_pending_compaction_bytes_limit = 2000; vstorage->TEST_set_estimated_compaction_needed_bytes(50); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage->TEST_set_estimated_compaction_needed_bytes(201); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(400); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(500); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(450); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(205); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(202); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(201); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(198); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage->TEST_set_estimated_compaction_needed_bytes(399); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(599); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(2001); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(3001); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage->TEST_set_estimated_compaction_needed_bytes(390); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(100); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage->set_l0_delay_trigger_count(100); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(101); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->set_l0_delay_trigger_count(0); vstorage->TEST_set_estimated_compaction_needed_bytes(300); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->set_l0_delay_trigger_count(101); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2 / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(200); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->set_l0_delay_trigger_count(0); vstorage->TEST_set_estimated_compaction_needed_bytes(0); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); mutable_cf_options.disable_auto_compactions = true; dbfull()->TEST_write_controler().set_delayed_write_rate(kBaseRate); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage->set_l0_delay_trigger_count(50); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->set_l0_delay_trigger_count(60); vstorage->TEST_set_estimated_compaction_needed_bytes(300); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->set_l0_delay_trigger_count(70); vstorage->TEST_set_estimated_compaction_needed_bytes(500); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); mutable_cf_options.disable_auto_compactions = false; vstorage->set_l0_delay_trigger_count(71); vstorage->TEST_set_estimated_compaction_needed_bytes(501); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); } TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) { db_options_.base_background_compactions = 2; db_options_.max_background_compactions = 6; Open({"default"}); ColumnFamilyData* cfd = static_cast(db_->DefaultColumnFamily())->cfd(); VersionStorageInfo* vstorage = cfd->current()->storage_info(); MutableCFOptions mutable_cf_options( Options(db_options_, column_family_options_), ImmutableCFOptions(Options(db_options_, column_family_options_))); // Speed up threshold = min(4 * 2, 4 + (36 - 4)/4) = 8 mutable_cf_options.level0_file_num_compaction_trigger = 4; mutable_cf_options.level0_slowdown_writes_trigger = 36; mutable_cf_options.level0_stop_writes_trigger = 50; // Speedup threshold = 200 / 4 = 50 mutable_cf_options.soft_pending_compaction_bytes_limit = 200; mutable_cf_options.hard_pending_compaction_bytes_limit = 2000; vstorage->TEST_set_estimated_compaction_needed_bytes(40); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(50); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(300); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(45); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(7); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(9); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(6); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); // Speed up threshold = min(4 * 2, 4 + (12 - 4)/4) = 6 mutable_cf_options.level0_file_num_compaction_trigger = 4; mutable_cf_options.level0_slowdown_writes_trigger = 16; mutable_cf_options.level0_stop_writes_trigger = 30; vstorage->set_l0_delay_trigger_count(5); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(7); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(3); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); } TEST_F(ColumnFamilyTest, WriteStallTwoColumnFamilies) { const uint64_t kBaseRate = 810000u; db_options_.delayed_write_rate = kBaseRate; Open(); CreateColumnFamilies({"one"}); ColumnFamilyData* cfd = static_cast(db_->DefaultColumnFamily())->cfd(); VersionStorageInfo* vstorage = cfd->current()->storage_info(); ColumnFamilyData* cfd1 = static_cast(handles_[1])->cfd(); VersionStorageInfo* vstorage1 = cfd1->current()->storage_info(); MutableCFOptions mutable_cf_options( Options(db_options_, column_family_options_), ImmutableCFOptions(Options(db_options_, column_family_options_))); mutable_cf_options.level0_slowdown_writes_trigger = 20; mutable_cf_options.level0_stop_writes_trigger = 10000; mutable_cf_options.soft_pending_compaction_bytes_limit = 200; mutable_cf_options.hard_pending_compaction_bytes_limit = 2000; MutableCFOptions mutable_cf_options1 = mutable_cf_options; mutable_cf_options1.soft_pending_compaction_bytes_limit = 500; vstorage->TEST_set_estimated_compaction_needed_bytes(50); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage1->TEST_set_estimated_compaction_needed_bytes(201); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); vstorage1->TEST_set_estimated_compaction_needed_bytes(600); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(70); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage1->TEST_set_estimated_compaction_needed_bytes(800); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(300); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage1->TEST_set_estimated_compaction_needed_bytes(700); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage->TEST_set_estimated_compaction_needed_bytes(500); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2 / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); vstorage1->TEST_set_estimated_compaction_needed_bytes(600); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_TRUE(!dbfull()->TEST_write_controler().IsStopped()); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); ASSERT_EQ(kBaseRate / 1.2, dbfull()->TEST_write_controler().delayed_write_rate()); } TEST_F(ColumnFamilyTest, CompactionSpeedupTwoColumnFamilies) { db_options_.base_background_compactions = 2; db_options_.max_background_compactions = 6; column_family_options_.soft_pending_compaction_bytes_limit = 200; column_family_options_.hard_pending_compaction_bytes_limit = 2000; Open(); CreateColumnFamilies({"one"}); ColumnFamilyData* cfd = static_cast(db_->DefaultColumnFamily())->cfd(); VersionStorageInfo* vstorage = cfd->current()->storage_info(); ColumnFamilyData* cfd1 = static_cast(handles_[1])->cfd(); VersionStorageInfo* vstorage1 = cfd1->current()->storage_info(); MutableCFOptions mutable_cf_options( Options(db_options_, column_family_options_), ImmutableCFOptions(Options(db_options_, column_family_options_))); // Speed up threshold = min(4 * 2, 4 + (36 - 4)/4) = 8 mutable_cf_options.level0_file_num_compaction_trigger = 4; mutable_cf_options.level0_slowdown_writes_trigger = 36; mutable_cf_options.level0_stop_writes_trigger = 30; // Speedup threshold = 200 / 4 = 50 mutable_cf_options.soft_pending_compaction_bytes_limit = 200; mutable_cf_options.hard_pending_compaction_bytes_limit = 2000; MutableCFOptions mutable_cf_options1 = mutable_cf_options; mutable_cf_options1.level0_slowdown_writes_trigger = 16; vstorage->TEST_set_estimated_compaction_needed_bytes(40); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(60); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage1->TEST_set_estimated_compaction_needed_bytes(30); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage1->TEST_set_estimated_compaction_needed_bytes(70); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->TEST_set_estimated_compaction_needed_bytes(20); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage1->TEST_set_estimated_compaction_needed_bytes(3); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(9); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage1->set_l0_delay_trigger_count(2); cfd1->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(6, dbfull()->BGCompactionsAllowed()); vstorage->set_l0_delay_trigger_count(0); cfd->RecalculateWriteStallConditions(mutable_cf_options); ASSERT_EQ(2, dbfull()->BGCompactionsAllowed()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/compact_files_test.cc000066400000000000000000000152011267530634000177200ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include #include #include #include #include "rocksdb/db.h" #include "rocksdb/env.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/testharness.h" namespace rocksdb { class CompactFilesTest : public testing::Test { public: CompactFilesTest() { env_ = Env::Default(); db_name_ = test::TmpDir(env_) + "/compact_files_test"; } std::string db_name_; Env* env_; }; // A class which remembers the name of each flushed file. class FlushedFileCollector : public EventListener { public: FlushedFileCollector() {} ~FlushedFileCollector() {} virtual void OnFlushCompleted( DB* db, const FlushJobInfo& info) override { std::lock_guard lock(mutex_); flushed_files_.push_back(info.file_path); } std::vector GetFlushedFiles() { std::lock_guard lock(mutex_); std::vector result; for (auto fname : flushed_files_) { result.push_back(fname); } return result; } private: std::vector flushed_files_; std::mutex mutex_; }; TEST_F(CompactFilesTest, L0ConflictsFiles) { Options options; // to trigger compaction more easily const int kWriteBufferSize = 10000; const int kLevel0Trigger = 2; options.create_if_missing = true; options.compaction_style = kCompactionStyleLevel; // Small slowdown and stop trigger for experimental purpose. options.level0_slowdown_writes_trigger = 20; options.level0_stop_writes_trigger = 20; options.level0_stop_writes_trigger = 20; options.write_buffer_size = kWriteBufferSize; options.level0_file_num_compaction_trigger = kLevel0Trigger; options.compression = kNoCompression; DB* db = nullptr; DestroyDB(db_name_, options); Status s = DB::Open(options, db_name_, &db); assert(s.ok()); assert(db); rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"CompactFilesImpl:0", "BackgroundCallCompaction:0"}, {"BackgroundCallCompaction:1", "CompactFilesImpl:1"}, }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // create couple files // Background compaction starts and waits in BackgroundCallCompaction:0 for (int i = 0; i < kLevel0Trigger * 4; ++i) { db->Put(WriteOptions(), ToString(i), ""); db->Put(WriteOptions(), ToString(100 - i), ""); db->Flush(FlushOptions()); } rocksdb::ColumnFamilyMetaData meta; db->GetColumnFamilyMetaData(&meta); std::string file1; for (auto& file : meta.levels[0].files) { ASSERT_EQ(0, meta.levels[0].level); if (file1 == "") { file1 = file.db_path + "/" + file.name; } else { std::string file2 = file.db_path + "/" + file.name; // Another thread starts a compact files and creates an L0 compaction // The background compaction then notices that there is an L0 compaction // already in progress and doesn't do an L0 compaction // Once the background compaction finishes, the compact files finishes ASSERT_OK( db->CompactFiles(rocksdb::CompactionOptions(), {file1, file2}, 0)); break; } } rocksdb::SyncPoint::GetInstance()->DisableProcessing(); delete db; } TEST_F(CompactFilesTest, ObsoleteFiles) { Options options; // to trigger compaction more easily const int kWriteBufferSize = 10000; options.create_if_missing = true; // Disable RocksDB background compaction. options.compaction_style = kCompactionStyleNone; // Small slowdown and stop trigger for experimental purpose. options.level0_slowdown_writes_trigger = 20; options.level0_stop_writes_trigger = 20; options.write_buffer_size = kWriteBufferSize; options.max_write_buffer_number = 2; options.compression = kNoCompression; // Add listener FlushedFileCollector* collector = new FlushedFileCollector(); options.listeners.emplace_back(collector); DB* db = nullptr; DestroyDB(db_name_, options); Status s = DB::Open(options, db_name_, &db); assert(s.ok()); assert(db); // create couple files for (int i = 1000; i < 2000; ++i) { db->Put(WriteOptions(), ToString(i), std::string(kWriteBufferSize / 10, 'a' + (i % 26))); } auto l0_files = collector->GetFlushedFiles(); ASSERT_OK(db->CompactFiles(CompactionOptions(), l0_files, 1)); // verify all compaction input files are deleted for (auto fname : l0_files) { ASSERT_EQ(Status::NotFound(), env_->FileExists(fname)); } delete db; } TEST_F(CompactFilesTest, CapturingPendingFiles) { Options options; options.create_if_missing = true; // Disable RocksDB background compaction. options.compaction_style = kCompactionStyleNone; // Always do full scans for obsolete files (needed to reproduce the issue). options.delete_obsolete_files_period_micros = 0; // Add listener. FlushedFileCollector* collector = new FlushedFileCollector(); options.listeners.emplace_back(collector); DB* db = nullptr; DestroyDB(db_name_, options); Status s = DB::Open(options, db_name_, &db); assert(s.ok()); assert(db); // Create 5 files. for (int i = 0; i < 5; ++i) { db->Put(WriteOptions(), "key" + ToString(i), "value"); db->Flush(FlushOptions()); } auto l0_files = collector->GetFlushedFiles(); EXPECT_EQ(5, l0_files.size()); rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"CompactFilesImpl:2", "CompactFilesTest.CapturingPendingFiles:0"}, {"CompactFilesTest.CapturingPendingFiles:1", "CompactFilesImpl:3"}, }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // Start compacting files. std::thread compaction_thread( [&] { EXPECT_OK(db->CompactFiles(CompactionOptions(), l0_files, 1)); }); // In the meantime flush another file. TEST_SYNC_POINT("CompactFilesTest.CapturingPendingFiles:0"); db->Put(WriteOptions(), "key5", "value"); db->Flush(FlushOptions()); TEST_SYNC_POINT("CompactFilesTest.CapturingPendingFiles:1"); compaction_thread.join(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); delete db; // Make sure we can reopen the DB. s = DB::Open(options, db_name_, &db); ASSERT_TRUE(s.ok()); assert(db); delete db; } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as DBImpl::CompactFiles is not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/compacted_db_impl.cc000066400000000000000000000126601267530634000175040ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "db/compacted_db_impl.h" #include "db/db_impl.h" #include "db/version_set.h" #include "table/get_context.h" namespace rocksdb { extern void MarkKeyMayExist(void* arg); extern bool SaveValue(void* arg, const ParsedInternalKey& parsed_key, const Slice& v, bool hit_and_return); CompactedDBImpl::CompactedDBImpl( const DBOptions& options, const std::string& dbname) : DBImpl(options, dbname) { } CompactedDBImpl::~CompactedDBImpl() { } size_t CompactedDBImpl::FindFile(const Slice& key) { size_t left = 0; size_t right = files_.num_files - 1; while (left < right) { size_t mid = (left + right) >> 1; const FdWithKeyRange& f = files_.files[mid]; if (user_comparator_->Compare(ExtractUserKey(f.largest_key), key) < 0) { // Key at "mid.largest" is < "target". Therefore all // files at or before "mid" are uninteresting. left = mid + 1; } else { // Key at "mid.largest" is >= "target". Therefore all files // after "mid" are uninteresting. right = mid; } } return right; } Status CompactedDBImpl::Get(const ReadOptions& options, ColumnFamilyHandle*, const Slice& key, std::string* value) { GetContext get_context(user_comparator_, nullptr, nullptr, nullptr, GetContext::kNotFound, key, value, nullptr, nullptr, nullptr); LookupKey lkey(key, kMaxSequenceNumber); files_.files[FindFile(key)].fd.table_reader->Get( options, lkey.internal_key(), &get_context); if (get_context.State() == GetContext::kFound) { return Status::OK(); } return Status::NotFound(); } std::vector CompactedDBImpl::MultiGet(const ReadOptions& options, const std::vector&, const std::vector& keys, std::vector* values) { autovector reader_list; for (const auto& key : keys) { const FdWithKeyRange& f = files_.files[FindFile(key)]; if (user_comparator_->Compare(key, ExtractUserKey(f.smallest_key)) < 0) { reader_list.push_back(nullptr); } else { LookupKey lkey(key, kMaxSequenceNumber); f.fd.table_reader->Prepare(lkey.internal_key()); reader_list.push_back(f.fd.table_reader); } } std::vector statuses(keys.size(), Status::NotFound()); values->resize(keys.size()); int idx = 0; for (auto* r : reader_list) { if (r != nullptr) { GetContext get_context(user_comparator_, nullptr, nullptr, nullptr, GetContext::kNotFound, keys[idx], &(*values)[idx], nullptr, nullptr, nullptr); LookupKey lkey(keys[idx], kMaxSequenceNumber); r->Get(options, lkey.internal_key(), &get_context); if (get_context.State() == GetContext::kFound) { statuses[idx] = Status::OK(); } } ++idx; } return statuses; } Status CompactedDBImpl::Init(const Options& options) { mutex_.Lock(); ColumnFamilyDescriptor cf(kDefaultColumnFamilyName, ColumnFamilyOptions(options)); Status s = Recover({ cf }, true /* read only */, false); if (s.ok()) { cfd_ = reinterpret_cast( DefaultColumnFamily())->cfd(); delete cfd_->InstallSuperVersion(new SuperVersion(), &mutex_); } mutex_.Unlock(); if (!s.ok()) { return s; } NewThreadStatusCfInfo(cfd_); version_ = cfd_->GetSuperVersion()->current; user_comparator_ = cfd_->user_comparator(); auto* vstorage = version_->storage_info(); if (vstorage->num_non_empty_levels() == 0) { return Status::NotSupported("no file exists"); } const LevelFilesBrief& l0 = vstorage->LevelFilesBrief(0); // L0 should not have files if (l0.num_files > 1) { return Status::NotSupported("L0 contain more than 1 file"); } if (l0.num_files == 1) { if (vstorage->num_non_empty_levels() > 1) { return Status::NotSupported("Both L0 and other level contain files"); } files_ = l0; return Status::OK(); } for (int i = 1; i < vstorage->num_non_empty_levels() - 1; ++i) { if (vstorage->LevelFilesBrief(i).num_files > 0) { return Status::NotSupported("Other levels also contain files"); } } int level = vstorage->num_non_empty_levels() - 1; if (vstorage->LevelFilesBrief(level).num_files > 0) { files_ = vstorage->LevelFilesBrief(level); return Status::OK(); } return Status::NotSupported("no file exists"); } Status CompactedDBImpl::Open(const Options& options, const std::string& dbname, DB** dbptr) { *dbptr = nullptr; if (options.max_open_files != -1) { return Status::InvalidArgument("require max_open_files = -1"); } if (options.merge_operator.get() != nullptr) { return Status::InvalidArgument("merge operator is not supported"); } DBOptions db_options(options); std::unique_ptr db(new CompactedDBImpl(db_options, dbname)); Status s = db->Init(options); if (s.ok()) { Log(INFO_LEVEL, db->db_options_.info_log, "Opened the db as fully compacted mode"); LogFlush(db->db_options_.info_log); *dbptr = db.release(); } return s; } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/compacted_db_impl.h000066400000000000000000000067461267530634000173560ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include "db/db_impl.h" #include #include namespace rocksdb { class CompactedDBImpl : public DBImpl { public: CompactedDBImpl(const DBOptions& options, const std::string& dbname); virtual ~CompactedDBImpl(); static Status Open(const Options& options, const std::string& dbname, DB** dbptr); // Implementations of the DB interface using DB::Get; virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) override; using DB::MultiGet; virtual std::vector MultiGet( const ReadOptions& options, const std::vector&, const std::vector& keys, std::vector* values) override; using DBImpl::Put; virtual Status Put(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override { return Status::NotSupported("Not supported in compacted db mode."); } using DBImpl::Merge; virtual Status Merge(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override { return Status::NotSupported("Not supported in compacted db mode."); } using DBImpl::Delete; virtual Status Delete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) override { return Status::NotSupported("Not supported in compacted db mode."); } virtual Status Write(const WriteOptions& options, WriteBatch* updates) override { return Status::NotSupported("Not supported in compacted db mode."); } using DBImpl::CompactRange; virtual Status CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) override { return Status::NotSupported("Not supported in compacted db mode."); } virtual Status DisableFileDeletions() override { return Status::NotSupported("Not supported in compacted db mode."); } virtual Status EnableFileDeletions(bool force) override { return Status::NotSupported("Not supported in compacted db mode."); } virtual Status GetLiveFiles(std::vector&, uint64_t* manifest_file_size, bool flush_memtable = true) override { return Status::NotSupported("Not supported in compacted db mode."); } using DBImpl::Flush; virtual Status Flush(const FlushOptions& options, ColumnFamilyHandle* column_family) override { return Status::NotSupported("Not supported in compacted db mode."); } private: friend class DB; inline size_t FindFile(const Slice& key); Status Init(const Options& options); ColumnFamilyData* cfd_; Version* version_; const Comparator* user_comparator_; LevelFilesBrief files_; // No copying allowed CompactedDBImpl(const CompactedDBImpl&); void operator=(const CompactedDBImpl&); }; } #endif // ROCKSDB_LITE rocksdb-4.5.1/db/compaction.cc000066400000000000000000000401311267530634000162050ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/compaction.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include "rocksdb/compaction_filter.h" #include "db/column_family.h" #include "util/logging.h" #include "util/sync_point.h" namespace rocksdb { uint64_t TotalFileSize(const std::vector& files) { uint64_t sum = 0; for (size_t i = 0; i < files.size() && files[i]; i++) { sum += files[i]->fd.GetFileSize(); } return sum; } void Compaction::SetInputVersion(Version* _input_version) { input_version_ = _input_version; cfd_ = input_version_->cfd(); cfd_->Ref(); input_version_->Ref(); edit_.SetColumnFamily(cfd_->GetID()); } void Compaction::GetBoundaryKeys( VersionStorageInfo* vstorage, const std::vector& inputs, Slice* smallest_user_key, Slice* largest_user_key) { bool initialized = false; const Comparator* ucmp = vstorage->InternalComparator()->user_comparator(); for (size_t i = 0; i < inputs.size(); ++i) { if (inputs[i].files.empty()) { continue; } if (inputs[i].level == 0) { // we need to consider all files on level 0 for (const auto* f : inputs[i].files) { const Slice& start_user_key = f->smallest.user_key(); if (!initialized || ucmp->Compare(start_user_key, *smallest_user_key) < 0) { *smallest_user_key = start_user_key; } const Slice& end_user_key = f->largest.user_key(); if (!initialized || ucmp->Compare(end_user_key, *largest_user_key) > 0) { *largest_user_key = end_user_key; } initialized = true; } } else { // we only need to consider the first and last file const Slice& start_user_key = inputs[i].files[0]->smallest.user_key(); if (!initialized || ucmp->Compare(start_user_key, *smallest_user_key) < 0) { *smallest_user_key = start_user_key; } const Slice& end_user_key = inputs[i].files.back()->largest.user_key(); if (!initialized || ucmp->Compare(end_user_key, *largest_user_key) > 0) { *largest_user_key = end_user_key; } initialized = true; } } } // helper function to determine if compaction is creating files at the // bottommost level bool Compaction::IsBottommostLevel( int output_level, VersionStorageInfo* vstorage, const std::vector& inputs) { if (inputs[0].level == 0 && inputs[0].files.back() != vstorage->LevelFiles(0).back()) { return false; } Slice smallest_key, largest_key; GetBoundaryKeys(vstorage, inputs, &smallest_key, &largest_key); // Checks whether there are files living beyond the output_level. // If lower levels have files, it checks for overlap between files // if the compaction process and those files. // Bottomlevel optimizations can be made if there are no files in // lower levels or if there is no overlap with the files in // the lower levels. for (int i = output_level + 1; i < vstorage->num_levels(); i++) { // It is not the bottommost level if there are files in higher // levels when the output level is 0 or if there are files in // higher levels which overlap with files to be compacted. // output_level == 0 means that we want it to be considered // s the bottommost level only if the last file on the level // is a part of the files to be compacted - this is verified by // the first if condition in this function if (vstorage->NumLevelFiles(i) > 0 && (output_level == 0 || vstorage->OverlapInLevel(i, &smallest_key, &largest_key))) { return false; } } return true; } // test function to validate the functionality of IsBottommostLevel() // function -- determines if compaction with inputs and storage is bottommost bool Compaction::TEST_IsBottommostLevel( int output_level, VersionStorageInfo* vstorage, const std::vector& inputs) { return IsBottommostLevel(output_level, vstorage, inputs); } bool Compaction::IsFullCompaction( VersionStorageInfo* vstorage, const std::vector& inputs) { size_t num_files_in_compaction = 0; size_t total_num_files = 0; for (int l = 0; l < vstorage->num_levels(); l++) { total_num_files += vstorage->NumLevelFiles(l); } for (size_t i = 0; i < inputs.size(); i++) { num_files_in_compaction += inputs[i].size(); } return num_files_in_compaction == total_num_files; } Compaction::Compaction(VersionStorageInfo* vstorage, const MutableCFOptions& _mutable_cf_options, std::vector _inputs, int _output_level, uint64_t _target_file_size, uint64_t _max_grandparent_overlap_bytes, uint32_t _output_path_id, CompressionType _compression, std::vector _grandparents, bool _manual_compaction, double _score, bool _deletion_compaction, CompactionReason _compaction_reason) : start_level_(_inputs[0].level), output_level_(_output_level), max_output_file_size_(_target_file_size), max_grandparent_overlap_bytes_(_max_grandparent_overlap_bytes), mutable_cf_options_(_mutable_cf_options), input_version_(nullptr), number_levels_(vstorage->num_levels()), cfd_(nullptr), output_path_id_(_output_path_id), output_compression_(_compression), deletion_compaction_(_deletion_compaction), inputs_(std::move(_inputs)), grandparents_(std::move(_grandparents)), grandparent_index_(0), seen_key_(false), overlapped_bytes_(0), score_(_score), bottommost_level_(IsBottommostLevel(output_level_, vstorage, inputs_)), is_full_compaction_(IsFullCompaction(vstorage, inputs_)), is_manual_compaction_(_manual_compaction), compaction_reason_(_compaction_reason) { MarkFilesBeingCompacted(true); if (is_manual_compaction_) { compaction_reason_ = CompactionReason::kManualCompaction; } #ifndef NDEBUG for (size_t i = 1; i < inputs_.size(); ++i) { assert(inputs_[i].level > inputs_[i - 1].level); } #endif // setup input_levels_ { input_levels_.resize(num_input_levels()); for (size_t which = 0; which < num_input_levels(); which++) { DoGenerateLevelFilesBrief(&input_levels_[which], inputs_[which].files, &arena_); } } Slice smallest_user_key; GetBoundaryKeys(vstorage, inputs_, &smallest_user_key, &largest_user_key_); } Compaction::~Compaction() { if (input_version_ != nullptr) { input_version_->Unref(); } if (cfd_ != nullptr) { if (cfd_->Unref()) { delete cfd_; } } } bool Compaction::InputCompressionMatchesOutput() const { int base_level = input_version_->storage_info()->base_level(); bool matches = (GetCompressionType(*cfd_->ioptions(), start_level_, base_level) == output_compression_); if (matches) { TEST_SYNC_POINT("Compaction::InputCompressionMatchesOutput:Matches"); return true; } TEST_SYNC_POINT("Compaction::InputCompressionMatchesOutput:DidntMatch"); return matches; } bool Compaction::IsTrivialMove() const { // Avoid a move if there is lots of overlapping grandparent data. // Otherwise, the move could create a parent file that will require // a very expensive merge later on. // If start_level_== output_level_, the purpose is to force compaction // filter to be applied to that level, and thus cannot be a trivial move. // Check if start level have files with overlapping ranges if (start_level_ == 0 && input_version_->storage_info()->level0_non_overlapping() == false) { // We cannot move files from L0 to L1 if the files are overlapping return false; } if (is_manual_compaction_ && (cfd_->ioptions()->compaction_filter != nullptr || cfd_->ioptions()->compaction_filter_factory != nullptr)) { // This is a manual compaction and we have a compaction filter that should // be executed, we cannot do a trivial move return false; } // Used in universal compaction, where trivial move can be done if the // input files are non overlapping if ((cfd_->ioptions()->compaction_options_universal.allow_trivial_move) && (output_level_ != 0)) { return is_trivial_move_; } return (start_level_ != output_level_ && num_input_levels() == 1 && input(0, 0)->fd.GetPathId() == output_path_id() && InputCompressionMatchesOutput() && TotalFileSize(grandparents_) <= max_grandparent_overlap_bytes_); } void Compaction::AddInputDeletions(VersionEdit* out_edit) { for (size_t which = 0; which < num_input_levels(); which++) { for (size_t i = 0; i < inputs_[which].size(); i++) { out_edit->DeleteFile(level(which), inputs_[which][i]->fd.GetNumber()); } } } bool Compaction::KeyNotExistsBeyondOutputLevel( const Slice& user_key, std::vector* level_ptrs) const { assert(input_version_ != nullptr); assert(level_ptrs != nullptr); assert(level_ptrs->size() == static_cast(number_levels_)); assert(cfd_->ioptions()->compaction_style != kCompactionStyleFIFO); if (cfd_->ioptions()->compaction_style == kCompactionStyleUniversal) { return bottommost_level_; } // Maybe use binary search to find right entry instead of linear search? const Comparator* user_cmp = cfd_->user_comparator(); for (int lvl = output_level_ + 1; lvl < number_levels_; lvl++) { const std::vector& files = input_version_->storage_info()->LevelFiles(lvl); for (; level_ptrs->at(lvl) < files.size(); level_ptrs->at(lvl)++) { auto* f = files[level_ptrs->at(lvl)]; if (user_cmp->Compare(user_key, f->largest.user_key()) <= 0) { // We've advanced far enough if (user_cmp->Compare(user_key, f->smallest.user_key()) >= 0) { // Key falls in this file's range, so definitely // exists beyond output level return false; } break; } } } return true; } bool Compaction::ShouldStopBefore(const Slice& internal_key) { // Scan to find earliest grandparent file that contains key. const InternalKeyComparator* icmp = &cfd_->internal_comparator(); while (grandparent_index_ < grandparents_.size() && icmp->Compare(internal_key, grandparents_[grandparent_index_]->largest.Encode()) > 0) { if (seen_key_) { overlapped_bytes_ += grandparents_[grandparent_index_]->fd.GetFileSize(); } assert(grandparent_index_ + 1 >= grandparents_.size() || icmp->Compare(grandparents_[grandparent_index_]->largest.Encode(), grandparents_[grandparent_index_+1]->smallest.Encode()) < 0); grandparent_index_++; } seen_key_ = true; if (overlapped_bytes_ > max_grandparent_overlap_bytes_) { // Too much overlap for current output; start new output overlapped_bytes_ = 0; return true; } else { return false; } } // Mark (or clear) each file that is being compacted void Compaction::MarkFilesBeingCompacted(bool mark_as_compacted) { for (size_t i = 0; i < num_input_levels(); i++) { for (size_t j = 0; j < inputs_[i].size(); j++) { assert(mark_as_compacted ? !inputs_[i][j]->being_compacted : inputs_[i][j]->being_compacted); inputs_[i][j]->being_compacted = mark_as_compacted; } } } // Sample output: // If compacting 3 L0 files, 2 L3 files and 1 L4 file, and outputting to L5, // print: "3@0 + 2@3 + 1@4 files to L5" const char* Compaction::InputLevelSummary( InputLevelSummaryBuffer* scratch) const { int len = 0; bool is_first = true; for (auto& input_level : inputs_) { if (input_level.empty()) { continue; } if (!is_first) { len += snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, " + "); } else { is_first = false; } len += snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, "%" ROCKSDB_PRIszt "@%d", input_level.size(), input_level.level); } snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, " files to L%d", output_level()); return scratch->buffer; } uint64_t Compaction::CalculateTotalInputSize() const { uint64_t size = 0; for (auto& input_level : inputs_) { for (auto f : input_level.files) { size += f->fd.GetFileSize(); } } return size; } void Compaction::ReleaseCompactionFiles(Status status) { MarkFilesBeingCompacted(false); cfd_->compaction_picker()->ReleaseCompactionFiles(this, status); } void Compaction::ResetNextCompactionIndex() { assert(input_version_ != nullptr); input_version_->storage_info()->ResetNextCompactionIndex(start_level_); } namespace { int InputSummary(const std::vector& files, char* output, int len) { *output = '\0'; int write = 0; for (size_t i = 0; i < files.size(); i++) { int sz = len - write; int ret; char sztxt[16]; AppendHumanBytes(files.at(i)->fd.GetFileSize(), sztxt, 16); ret = snprintf(output + write, sz, "%" PRIu64 "(%s) ", files.at(i)->fd.GetNumber(), sztxt); if (ret < 0 || ret >= sz) break; write += ret; } // if files.size() is non-zero, overwrite the last space return write - !!files.size(); } } // namespace void Compaction::Summary(char* output, int len) { int write = snprintf(output, len, "Base version %" PRIu64 " Base level %d, inputs: [", input_version_->GetVersionNumber(), start_level_); if (write < 0 || write >= len) { return; } for (size_t level_iter = 0; level_iter < num_input_levels(); ++level_iter) { if (level_iter > 0) { write += snprintf(output + write, len - write, "], ["); if (write < 0 || write >= len) { return; } } write += InputSummary(inputs_[level_iter].files, output + write, len - write); if (write < 0 || write >= len) { return; } } snprintf(output + write, len - write, "]"); } uint64_t Compaction::OutputFilePreallocationSize() { uint64_t preallocation_size = 0; if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel || output_level() > 0) { preallocation_size = max_output_file_size_; } else { // output_level() == 0 assert(num_input_levels() > 0); for (const auto& f : inputs_[0].files) { preallocation_size += f->fd.GetFileSize(); } } // Over-estimate slightly so we don't end up just barely crossing // the threshold return preallocation_size + (preallocation_size / 10); } std::unique_ptr Compaction::CreateCompactionFilter() const { if (!cfd_->ioptions()->compaction_filter_factory) { return nullptr; } CompactionFilter::Context context; context.is_full_compaction = is_full_compaction_; context.is_manual_compaction = is_manual_compaction_; context.column_family_id = cfd_->GetID(); return cfd_->ioptions()->compaction_filter_factory->CreateCompactionFilter( context); } bool Compaction::IsOutputLevelEmpty() const { return inputs_.back().level != output_level_ || inputs_.back().empty(); } bool Compaction::ShouldFormSubcompactions() const { if (mutable_cf_options_.max_subcompactions <= 1 || cfd_ == nullptr) { return false; } if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel) { return start_level_ == 0 && !IsOutputLevelEmpty(); } else if (cfd_->ioptions()->compaction_style == kCompactionStyleUniversal) { return number_levels_ > 1 && output_level_ > 0; } else { return false; } } } // namespace rocksdb rocksdb-4.5.1/db/compaction.h000066400000000000000000000266111267530634000160560ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include "util/arena.h" #include "util/autovector.h" #include "util/mutable_cf_options.h" #include "db/version_set.h" namespace rocksdb { // The structure that manages compaction input files associated // with the same physical level. struct CompactionInputFiles { int level; std::vector files; inline bool empty() const { return files.empty(); } inline size_t size() const { return files.size(); } inline void clear() { files.clear(); } inline FileMetaData* operator[](size_t i) const { return files[i]; } }; class Version; class ColumnFamilyData; class VersionStorageInfo; class CompactionFilter; // A Compaction encapsulates information about a compaction. class Compaction { public: Compaction(VersionStorageInfo* input_version, const MutableCFOptions& mutable_cf_options, std::vector inputs, int output_level, uint64_t target_file_size, uint64_t max_grandparent_overlap_bytes, uint32_t output_path_id, CompressionType compression, std::vector grandparents, bool manual_compaction = false, double score = -1, bool deletion_compaction = false, CompactionReason compaction_reason = CompactionReason::kUnknown); // No copying allowed Compaction(const Compaction&) = delete; void operator=(const Compaction&) = delete; ~Compaction(); // Returns the level associated to the specified compaction input level. // If compaction_input_level is not specified, then input_level is set to 0. int level(size_t compaction_input_level = 0) const { return inputs_[compaction_input_level].level; } int start_level() const { return start_level_; } // Outputs will go to this level int output_level() const { return output_level_; } // Returns the number of input levels in this compaction. size_t num_input_levels() const { return inputs_.size(); } // Return the object that holds the edits to the descriptor done // by this compaction. VersionEdit* edit() { return &edit_; } // Returns the number of input files associated to the specified // compaction input level. // The function will return 0 if when "compaction_input_level" < 0 // or "compaction_input_level" >= "num_input_levels()". size_t num_input_files(size_t compaction_input_level) const { if (compaction_input_level < inputs_.size()) { return inputs_[compaction_input_level].size(); } return 0; } // Returns input version of the compaction Version* input_version() const { return input_version_; } // Returns the ColumnFamilyData associated with the compaction. ColumnFamilyData* column_family_data() const { return cfd_; } // Returns the file meta data of the 'i'th input file at the // specified compaction input level. // REQUIREMENT: "compaction_input_level" must be >= 0 and // < "input_levels()" FileMetaData* input(size_t compaction_input_level, size_t i) const { assert(compaction_input_level < inputs_.size()); return inputs_[compaction_input_level][i]; } // Returns the list of file meta data of the specified compaction // input level. // REQUIREMENT: "compaction_input_level" must be >= 0 and // < "input_levels()" const std::vector* inputs(size_t compaction_input_level) { assert(compaction_input_level < inputs_.size()); return &inputs_[compaction_input_level].files; } // Returns the LevelFilesBrief of the specified compaction input level. LevelFilesBrief* input_levels(size_t compaction_input_level) { return &input_levels_[compaction_input_level]; } // Maximum size of files to build during this compaction. uint64_t max_output_file_size() const { return max_output_file_size_; } // What compression for output CompressionType output_compression() const { return output_compression_; } // Whether need to write output file to second DB path. uint32_t output_path_id() const { return output_path_id_; } // Is this a trivial compaction that can be implemented by just // moving a single input file to the next level (no merging or splitting) bool IsTrivialMove() const; // If true, then the compaction can be done by simply deleting input files. bool deletion_compaction() const { return deletion_compaction_; } // Add all inputs to this compaction as delete operations to *edit. void AddInputDeletions(VersionEdit* edit); // Returns true if the available information we have guarantees that // the input "user_key" does not exist in any level beyond "output_level()". bool KeyNotExistsBeyondOutputLevel(const Slice& user_key, std::vector* level_ptrs) const; // Returns true iff we should stop building the current output // before processing "internal_key". bool ShouldStopBefore(const Slice& internal_key); // Clear all files to indicate that they are not being compacted // Delete this compaction from the list of running compactions. void ReleaseCompactionFiles(Status status); // Returns the summary of the compaction in "output" with maximum "len" // in bytes. The caller is responsible for the memory management of // "output". void Summary(char* output, int len); // Return the score that was used to pick this compaction run. double score() const { return score_; } // Is this compaction creating a file in the bottom most level? bool bottommost_level() { return bottommost_level_; } // Does this compaction include all sst files? bool is_full_compaction() { return is_full_compaction_; } // Was this compaction triggered manually by the client? bool is_manual_compaction() { return is_manual_compaction_; } // Used when allow_trivial_move option is set in // Universal compaction. If all the input files are // non overlapping, then is_trivial_move_ variable // will be set true, else false void set_is_trivial_move(bool trivial_move) { is_trivial_move_ = trivial_move; } // Used when allow_trivial_move option is set in // Universal compaction. Returns true, if the input files // are non-overlapping and can be trivially moved. bool is_trivial_move() { return is_trivial_move_; } // How many total levels are there? int number_levels() const { return number_levels_; } // Return the MutableCFOptions that should be used throughout the compaction // procedure const MutableCFOptions* mutable_cf_options() { return &mutable_cf_options_; } // Returns the size in bytes that the output file should be preallocated to. // In level compaction, that is max_file_size_. In universal compaction, that // is the sum of all input file sizes. uint64_t OutputFilePreallocationSize(); void SetInputVersion(Version* input_version); struct InputLevelSummaryBuffer { char buffer[128]; }; const char* InputLevelSummary(InputLevelSummaryBuffer* scratch) const; uint64_t CalculateTotalInputSize() const; // In case of compaction error, reset the nextIndex that is used // to pick up the next file to be compacted from files_by_size_ void ResetNextCompactionIndex(); // Create a CompactionFilter from compaction_filter_factory std::unique_ptr CreateCompactionFilter() const; // Is the input level corresponding to output_level_ empty? bool IsOutputLevelEmpty() const; // Should this compaction be broken up into smaller ones run in parallel? bool ShouldFormSubcompactions() const; // test function to validate the functionality of IsBottommostLevel() // function -- determines if compaction with inputs and storage is bottommost static bool TEST_IsBottommostLevel( int output_level, VersionStorageInfo* vstorage, const std::vector& inputs); TablePropertiesCollection GetOutputTableProperties() const { return output_table_properties_; } void SetOutputTableProperties(TablePropertiesCollection tp) { output_table_properties_ = std::move(tp); } Slice GetLargestUserKey() const { return largest_user_key_; } CompactionReason compaction_reason() { return compaction_reason_; } private: // mark (or clear) all files that are being compacted void MarkFilesBeingCompacted(bool mark_as_compacted); // get the smallest and largest key present in files to be compacted static void GetBoundaryKeys(VersionStorageInfo* vstorage, const std::vector& inputs, Slice* smallest_key, Slice* largest_key); // helper function to determine if compaction with inputs and storage is // bottommost static bool IsBottommostLevel( int output_level, VersionStorageInfo* vstorage, const std::vector& inputs); static bool IsFullCompaction(VersionStorageInfo* vstorage, const std::vector& inputs); const int start_level_; // the lowest level to be compacted const int output_level_; // levels to which output files are stored uint64_t max_output_file_size_; uint64_t max_grandparent_overlap_bytes_; MutableCFOptions mutable_cf_options_; Version* input_version_; VersionEdit edit_; const int number_levels_; ColumnFamilyData* cfd_; Arena arena_; // Arena used to allocate space for file_levels_ const uint32_t output_path_id_; CompressionType output_compression_; // If true, then the comaction can be done by simply deleting input files. const bool deletion_compaction_; // Compaction input files organized by level. Constant after construction const std::vector inputs_; // A copy of inputs_, organized more closely in memory autovector input_levels_; // State used to check for number of of overlapping grandparent files // (grandparent == "output_level_ + 1") std::vector grandparents_; size_t grandparent_index_; // Index in grandparent_starts_ bool seen_key_; // Some output key has been seen uint64_t overlapped_bytes_; // Bytes of overlap between current output // and grandparent files const double score_; // score that was used to pick this compaction. // Is this compaction creating a file in the bottom most level? const bool bottommost_level_; // Does this compaction include all sst files? const bool is_full_compaction_; // Is this compaction requested by the client? const bool is_manual_compaction_; // True if we can do trivial move in Universal multi level // compaction bool is_trivial_move_; // Does input compression match the output compression? bool InputCompressionMatchesOutput() const; // table properties of output files TablePropertiesCollection output_table_properties_; // largest user keys in compaction Slice largest_user_key_; // Reason for compaction CompactionReason compaction_reason_; }; // Utility function extern uint64_t TotalFileSize(const std::vector& files); } // namespace rocksdb rocksdb-4.5.1/db/compaction_iterator.cc000066400000000000000000000447561267530634000201370ustar00rootroot00000000000000// Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/compaction_iterator.h" #include "table/internal_iterator.h" namespace rocksdb { CompactionIterator::CompactionIterator( InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper, SequenceNumber last_sequence, std::vector* snapshots, SequenceNumber earliest_write_conflict_snapshot, Env* env, bool expect_valid_internal_key, Compaction* compaction, const CompactionFilter* compaction_filter, LogBuffer* log_buffer) : input_(input), cmp_(cmp), merge_helper_(merge_helper), snapshots_(snapshots), earliest_write_conflict_snapshot_(earliest_write_conflict_snapshot), env_(env), expect_valid_internal_key_(expect_valid_internal_key), compaction_(compaction), compaction_filter_(compaction_filter), log_buffer_(log_buffer), merge_out_iter_(merge_helper_) { assert(compaction_filter_ == nullptr || compaction_ != nullptr); bottommost_level_ = compaction_ == nullptr ? false : compaction_->bottommost_level(); if (compaction_ != nullptr) { level_ptrs_ = std::vector(compaction_->number_levels(), 0); } if (snapshots_->size() == 0) { // optimize for fast path if there are no snapshots visible_at_tip_ = last_sequence; earliest_snapshot_ = visible_at_tip_; latest_snapshot_ = 0; } else { visible_at_tip_ = 0; earliest_snapshot_ = snapshots_->at(0); latest_snapshot_ = snapshots_->back(); } if (compaction_filter_ != nullptr && compaction_filter_->IgnoreSnapshots()) { ignore_snapshots_ = true; } else { ignore_snapshots_ = false; } } void CompactionIterator::ResetRecordCounts() { iter_stats_.num_record_drop_user = 0; iter_stats_.num_record_drop_hidden = 0; iter_stats_.num_record_drop_obsolete = 0; } void CompactionIterator::SeekToFirst() { NextFromInput(); PrepareOutput(); } void CompactionIterator::Next() { // If there is a merge output, return it before continuing to process the // input. if (merge_out_iter_.Valid()) { merge_out_iter_.Next(); // Check if we returned all records of the merge output. if (merge_out_iter_.Valid()) { key_ = merge_out_iter_.key(); value_ = merge_out_iter_.value(); bool valid_key __attribute__((__unused__)) = ParseInternalKey(key_, &ikey_); // MergeUntil stops when it encounters a corrupt key and does not // include them in the result, so we expect the keys here to be valid. assert(valid_key); // Keep current_key_ in sync. current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type); key_ = current_key_.GetKey(); ikey_.user_key = current_key_.GetUserKey(); valid_ = true; } else { // MergeHelper moves the iterator to the first record after the merged // records, so even though we reached the end of the merge output, we do // not want to advance the iterator. NextFromInput(); } } else { // Only advance the input iterator if there is no merge output and the // iterator is not already at the next record. if (!at_next_) { input_->Next(); } NextFromInput(); } if (valid_) { // Record that we've ouputted a record for the current key. has_outputted_key_ = true; } PrepareOutput(); } void CompactionIterator::NextFromInput() { at_next_ = false; valid_ = false; while (!valid_ && input_->Valid()) { key_ = input_->key(); value_ = input_->value(); iter_stats_.num_input_records++; if (!ParseInternalKey(key_, &ikey_)) { // If `expect_valid_internal_key_` is false, return the corrupted key // and let the caller decide what to do with it. // TODO(noetzli): We should have a more elegant solution for this. if (expect_valid_internal_key_) { assert(!"Corrupted internal key not expected."); status_ = Status::Corruption("Corrupted internal key not expected."); break; } key_ = current_key_.SetKey(key_); has_current_user_key_ = false; current_user_key_sequence_ = kMaxSequenceNumber; current_user_key_snapshot_ = 0; iter_stats_.num_input_corrupt_records++; valid_ = true; break; } // Update input statistics if (ikey_.type == kTypeDeletion || ikey_.type == kTypeSingleDeletion) { iter_stats_.num_input_deletion_records++; } iter_stats_.total_input_raw_key_bytes += key_.size(); iter_stats_.total_input_raw_value_bytes += value_.size(); // Check whether the user key changed. After this if statement current_key_ // is a copy of the current input key (maybe converted to a delete by the // compaction filter). ikey_.user_key is pointing to the copy. if (!has_current_user_key_ || !cmp_->Equal(ikey_.user_key, current_user_key_)) { // First occurrence of this user key key_ = current_key_.SetKey(key_, &ikey_); current_user_key_ = ikey_.user_key; has_current_user_key_ = true; has_outputted_key_ = false; current_user_key_sequence_ = kMaxSequenceNumber; current_user_key_snapshot_ = 0; // apply the compaction filter to the first occurrence of the user key if (compaction_filter_ != nullptr && ikey_.type == kTypeValue && (visible_at_tip_ || ikey_.sequence > latest_snapshot_ || ignore_snapshots_)) { // If the user has specified a compaction filter and the sequence // number is greater than any external snapshot, then invoke the // filter. If the return value of the compaction filter is true, // replace the entry with a deletion marker. bool value_changed = false; bool to_delete = false; compaction_filter_value_.clear(); { StopWatchNano timer(env_, true); to_delete = compaction_filter_->Filter( compaction_->level(), ikey_.user_key, value_, &compaction_filter_value_, &value_changed); iter_stats_.total_filter_time += env_ != nullptr ? timer.ElapsedNanos() : 0; } if (to_delete) { // convert the current key to a delete ikey_.type = kTypeDeletion; current_key_.UpdateInternalKey(ikey_.sequence, kTypeDeletion); // no value associated with delete value_.clear(); iter_stats_.num_record_drop_user++; } else if (value_changed) { value_ = compaction_filter_value_; } } } else { // Update the current key to reflect the new sequence number/type without // copying the user key. // TODO(rven): Compaction filter does not process keys in this path // Need to have the compaction filter process multiple versions // if we have versions on both sides of a snapshot current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type); key_ = current_key_.GetKey(); ikey_.user_key = current_key_.GetUserKey(); } // If there are no snapshots, then this kv affect visibility at tip. // Otherwise, search though all existing snapshots to find the earliest // snapshot that is affected by this kv. SequenceNumber last_sequence __attribute__((__unused__)) = current_user_key_sequence_; current_user_key_sequence_ = ikey_.sequence; SequenceNumber last_snapshot = current_user_key_snapshot_; SequenceNumber prev_snapshot = 0; // 0 means no previous snapshot current_user_key_snapshot_ = visible_at_tip_ ? visible_at_tip_ : findEarliestVisibleSnapshot( ikey_.sequence, &prev_snapshot); if (clear_and_output_next_key_) { // In the previous iteration we encountered a single delete that we could // not compact out. We will keep this Put, but can drop it's data. // (See Optimization 3, below.) assert(ikey_.type == kTypeValue); assert(current_user_key_snapshot_ == last_snapshot); value_.clear(); valid_ = true; clear_and_output_next_key_ = false; } else if (ikey_.type == kTypeSingleDeletion) { // We can compact out a SingleDelete if: // 1) We encounter the corresponding PUT -OR- we know that this key // doesn't appear past this output level // =AND= // 2) We've already returned a record in this snapshot -OR- // there are no earlier earliest_write_conflict_snapshot. // // Rule 1 is needed for SingleDelete correctness. Rule 2 is needed to // allow Transactions to do write-conflict checking (if we compacted away // all keys, then we wouldn't know that a write happened in this // snapshot). If there is no earlier snapshot, then we know that there // are no active transactions that need to know about any writes. // // Optimization 3: // If we encounter a SingleDelete followed by a PUT and Rule 2 is NOT // true, then we must output a SingleDelete. In this case, we will decide // to also output the PUT. While we are compacting less by outputting the // PUT now, hopefully this will lead to better compaction in the future // when Rule 2 is later true (Ie, We are hoping we can later compact out // both the SingleDelete and the Put, while we couldn't if we only // outputted the SingleDelete now). // In this case, we can save space by removing the PUT's value as it will // never be read. // // Deletes and Merges are not supported on the same key that has a // SingleDelete as it is not possible to correctly do any partial // compaction of such a combination of operations. The result of mixing // those operations for a given key is documented as being undefined. So // we can choose how to handle such a combinations of operations. We will // try to compact out as much as we can in these cases. // The easiest way to process a SingleDelete during iteration is to peek // ahead at the next key. ParsedInternalKey next_ikey; input_->Next(); // Check whether the next key exists, is not corrupt, and is the same key // as the single delete. if (input_->Valid() && ParseInternalKey(input_->key(), &next_ikey) && cmp_->Equal(ikey_.user_key, next_ikey.user_key)) { // Check whether the next key belongs to the same snapshot as the // SingleDelete. if (prev_snapshot == 0 || next_ikey.sequence > prev_snapshot) { if (next_ikey.type == kTypeSingleDeletion) { // We encountered two SingleDeletes in a row. This could be due to // unexpected user input. // Skip the first SingleDelete and let the next iteration decide how // to handle the second SingleDelete // First SingleDelete has been skipped since we already called // input_->Next(). ++iter_stats_.num_record_drop_obsolete; } else if ((ikey_.sequence <= earliest_write_conflict_snapshot_) || has_outputted_key_) { // Found a matching value, we can drop the single delete and the // value. It is safe to drop both records since we've already // outputted a key in this snapshot, or there is no earlier // snapshot (Rule 2 above). // Note: it doesn't matter whether the second key is a Put or if it // is an unexpected Merge or Delete. We will compact it out // either way. ++iter_stats_.num_record_drop_hidden; ++iter_stats_.num_record_drop_obsolete; // Already called input_->Next() once. Call it a second time to // skip past the second key. input_->Next(); } else { // Found a matching value, but we cannot drop both keys since // there is an earlier snapshot and we need to leave behind a record // to know that a write happened in this snapshot (Rule 2 above). // Clear the value and output the SingleDelete. (The value will be // outputted on the next iteration.) ++iter_stats_.num_record_drop_hidden; // Setting valid_ to true will output the current SingleDelete valid_ = true; // Set up the Put to be outputted in the next iteration. // (Optimization 3). clear_and_output_next_key_ = true; } } else { // We hit the next snapshot without hitting a put, so the iterator // returns the single delete. valid_ = true; } } else { // We are at the end of the input, could not parse the next key, or hit // the next key. The iterator returns the single delete if the key // possibly exists beyond the current output level. We set // has_current_user_key to false so that if the iterator is at the next // key, we do not compare it again against the previous key at the next // iteration. If the next key is corrupt, we return before the // comparison, so the value of has_current_user_key does not matter. has_current_user_key_ = false; if (compaction_ != nullptr && ikey_.sequence <= earliest_snapshot_ && compaction_->KeyNotExistsBeyondOutputLevel(ikey_.user_key, &level_ptrs_)) { // Key doesn't exist outside of this range. // Can compact out this SingleDelete. ++iter_stats_.num_record_drop_obsolete; } else { // Output SingleDelete valid_ = true; } } if (valid_) { at_next_ = true; } } else if (last_snapshot == current_user_key_snapshot_) { // If the earliest snapshot is which this key is visible in // is the same as the visibility of a previous instance of the // same key, then this kv is not visible in any snapshot. // Hidden by an newer entry for same user key // TODO: why not > ? // // Note: Dropping this key will not affect TransactionDB write-conflict // checking since there has already been a record returned for this key // in this snapshot. assert(last_sequence >= current_user_key_sequence_); ++iter_stats_.num_record_drop_hidden; // (A) input_->Next(); } else if (compaction_ != nullptr && ikey_.type == kTypeDeletion && ikey_.sequence <= earliest_snapshot_ && compaction_->KeyNotExistsBeyondOutputLevel(ikey_.user_key, &level_ptrs_)) { // TODO(noetzli): This is the only place where we use compaction_ // (besides the constructor). We should probably get rid of this // dependency and find a way to do similar filtering during flushes. // // For this user key: // (1) there is no data in higher levels // (2) data in lower levels will have larger sequence numbers // (3) data in layers that are being compacted here and have // smaller sequence numbers will be dropped in the next // few iterations of this loop (by rule (A) above). // Therefore this deletion marker is obsolete and can be dropped. // // Note: Dropping this Delete will not affect TransactionDB // write-conflict checking since it is earlier than any snapshot. ++iter_stats_.num_record_drop_obsolete; input_->Next(); } else if (ikey_.type == kTypeMerge) { if (!merge_helper_->HasOperator()) { LogToBuffer(log_buffer_, "Options::merge_operator is null."); status_ = Status::InvalidArgument( "merge_operator is not properly initialized."); return; } // We know the merge type entry is not hidden, otherwise we would // have hit (A) // We encapsulate the merge related state machine in a different // object to minimize change to the existing flow. merge_helper_->MergeUntil(input_, prev_snapshot, bottommost_level_); merge_out_iter_.SeekToFirst(); if (merge_out_iter_.Valid()) { // NOTE: key, value, and ikey_ refer to old entries. // These will be correctly set below. key_ = merge_out_iter_.key(); value_ = merge_out_iter_.value(); bool valid_key __attribute__((__unused__)) = ParseInternalKey(key_, &ikey_); // MergeUntil stops when it encounters a corrupt key and does not // include them in the result, so we expect the keys here to valid. assert(valid_key); // Keep current_key_ in sync. current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type); key_ = current_key_.GetKey(); ikey_.user_key = current_key_.GetUserKey(); valid_ = true; } else { // all merge operands were filtered out. reset the user key, since the // batch consumed by the merge operator should not shadow any keys // coming after the merges has_current_user_key_ = false; } } else { valid_ = true; } } } void CompactionIterator::PrepareOutput() { // Zeroing out the sequence number leads to better compression. // If this is the bottommost level (no files in lower levels) // and the earliest snapshot is larger than this seqno // and the userkey differs from the last userkey in compaction // then we can squash the seqno to zero. // This is safe for TransactionDB write-conflict checking since transactions // only care about sequence number larger than any active snapshots. if (bottommost_level_ && valid_ && ikey_.sequence < earliest_snapshot_ && ikey_.type != kTypeMerge && !cmp_->Equal(compaction_->GetLargestUserKey(), ikey_.user_key)) { assert(ikey_.type != kTypeDeletion && ikey_.type != kTypeSingleDeletion); ikey_.sequence = 0; current_key_.UpdateInternalKey(0, ikey_.type); } } inline SequenceNumber CompactionIterator::findEarliestVisibleSnapshot( SequenceNumber in, SequenceNumber* prev_snapshot) { assert(snapshots_->size()); SequenceNumber prev __attribute__((unused)) = 0; for (const auto cur : *snapshots_) { assert(prev <= cur); if (cur >= in) { *prev_snapshot = prev; return cur; } prev = cur; assert(prev); } *prev_snapshot = prev; return kMaxSequenceNumber; } } // namespace rocksdb rocksdb-4.5.1/db/compaction_iterator.h000066400000000000000000000131101267530634000177550ustar00rootroot00000000000000// Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include #include "db/compaction.h" #include "db/merge_helper.h" #include "rocksdb/compaction_filter.h" #include "util/log_buffer.h" namespace rocksdb { struct CompactionIteratorStats { // Compaction statistics int64_t num_record_drop_user = 0; int64_t num_record_drop_hidden = 0; int64_t num_record_drop_obsolete = 0; uint64_t total_filter_time = 0; // Input statistics // TODO(noetzli): The stats are incomplete. They are lacking everything // consumed by MergeHelper. uint64_t num_input_records = 0; uint64_t num_input_deletion_records = 0; uint64_t num_input_corrupt_records = 0; uint64_t total_input_raw_key_bytes = 0; uint64_t total_input_raw_value_bytes = 0; }; class CompactionIterator { public: CompactionIterator(InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper, SequenceNumber last_sequence, std::vector* snapshots, SequenceNumber earliest_write_conflict_snapshot, Env* env, bool expect_valid_internal_key, Compaction* compaction = nullptr, const CompactionFilter* compaction_filter = nullptr, LogBuffer* log_buffer = nullptr); void ResetRecordCounts(); // Seek to the beginning of the compaction iterator output. // // REQUIRED: Call only once. void SeekToFirst(); // Produces the next record in the compaction. // // REQUIRED: SeekToFirst() has been called. void Next(); // Getters const Slice& key() const { return key_; } const Slice& value() const { return value_; } const Status& status() const { return status_; } const ParsedInternalKey& ikey() const { return ikey_; } bool Valid() const { return valid_; } const Slice& user_key() const { return current_user_key_; } const CompactionIteratorStats& iter_stats() const { return iter_stats_; } private: // Processes the input stream to find the next output void NextFromInput(); // Do last preparations before presenting the output to the callee. At this // point this only zeroes out the sequence number if possible for better // compression. void PrepareOutput(); // Given a sequence number, return the sequence number of the // earliest snapshot that this sequence number is visible in. // The snapshots themselves are arranged in ascending order of // sequence numbers. // Employ a sequential search because the total number of // snapshots are typically small. inline SequenceNumber findEarliestVisibleSnapshot( SequenceNumber in, SequenceNumber* prev_snapshot); InternalIterator* input_; const Comparator* cmp_; MergeHelper* merge_helper_; const std::vector* snapshots_; const SequenceNumber earliest_write_conflict_snapshot_; Env* env_; bool expect_valid_internal_key_; Compaction* compaction_; const CompactionFilter* compaction_filter_; LogBuffer* log_buffer_; bool bottommost_level_; bool valid_ = false; SequenceNumber visible_at_tip_; SequenceNumber earliest_snapshot_; SequenceNumber latest_snapshot_; bool ignore_snapshots_; // State // // Points to a copy of the current compaction iterator output (current_key_) // if valid_. Slice key_; // Points to the value in the underlying iterator that corresponds to the // current output. Slice value_; // The status is OK unless compaction iterator encounters a merge operand // while not having a merge operator defined. Status status_; // Stores the user key, sequence number and type of the current compaction // iterator output (or current key in the underlying iterator during // NextFromInput()). ParsedInternalKey ikey_; // Stores whether ikey_.user_key is valid. If set to false, the user key is // not compared against the current key in the underlying iterator. bool has_current_user_key_ = false; bool at_next_ = false; // If false, the iterator // Holds a copy of the current compaction iterator output (or current key in // the underlying iterator during NextFromInput()). IterKey current_key_; Slice current_user_key_; SequenceNumber current_user_key_sequence_; SequenceNumber current_user_key_snapshot_; // True if the iterator has already returned a record for the current key. bool has_outputted_key_ = false; // truncated the value of the next key and output it without applying any // compaction rules. This is used for outputting a put after a single delete. bool clear_and_output_next_key_ = false; MergeOutputIterator merge_out_iter_; std::string compaction_filter_value_; // "level_ptrs" holds indices that remember which file of an associated // level we were last checking during the last call to compaction-> // KeyNotExistsBeyondOutputLevel(). This allows future calls to the function // to pick off where it left off since each subcompaction's key range is // increasing so a later call to the function must be looking for a key that // is in or beyond the last file checked during the previous call std::vector level_ptrs_; CompactionIteratorStats iter_stats_; }; } // namespace rocksdb rocksdb-4.5.1/db/compaction_iterator_test.cc000066400000000000000000000051141267530634000211570ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/compaction_iterator.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class CompactionIteratorTest : public testing::Test { public: CompactionIteratorTest() : cmp_(BytewiseComparator()), snapshots_({}) {} void InitIterator(const std::vector& ks, const std::vector& vs, SequenceNumber last_sequence) { merge_helper_.reset(new MergeHelper(Env::Default(), cmp_, nullptr, nullptr, nullptr, 0U, false, 0)); iter_.reset(new test::VectorIterator(ks, vs)); iter_->SeekToFirst(); c_iter_.reset(new CompactionIterator( iter_.get(), cmp_, merge_helper_.get(), last_sequence, &snapshots_, kMaxSequenceNumber, Env::Default(), false)); } const Comparator* cmp_; std::vector snapshots_; std::unique_ptr merge_helper_; std::unique_ptr iter_; std::unique_ptr c_iter_; }; // It is possible that the output of the compaction iterator is empty even if // the input is not. TEST_F(CompactionIteratorTest, EmptyResult) { InitIterator({test::KeyStr("a", 5, kTypeSingleDeletion), test::KeyStr("a", 3, kTypeValue)}, {"", "val"}, 5); c_iter_->SeekToFirst(); ASSERT_FALSE(c_iter_->Valid()); } // If there is a corruption after a single deletion, the corrupted key should // be preserved. TEST_F(CompactionIteratorTest, CorruptionAfterSingleDeletion) { InitIterator({test::KeyStr("a", 5, kTypeSingleDeletion), test::KeyStr("a", 3, kTypeValue, true), test::KeyStr("b", 10, kTypeValue)}, {"", "val", "val2"}, 10); c_iter_->SeekToFirst(); ASSERT_TRUE(c_iter_->Valid()); ASSERT_EQ(test::KeyStr("a", 5, kTypeSingleDeletion), c_iter_->key().ToString()); c_iter_->Next(); ASSERT_TRUE(c_iter_->Valid()); ASSERT_EQ(test::KeyStr("a", 3, kTypeValue, true), c_iter_->key().ToString()); c_iter_->Next(); ASSERT_TRUE(c_iter_->Valid()); ASSERT_EQ(test::KeyStr("b", 10, kTypeValue), c_iter_->key().ToString()); c_iter_->Next(); ASSERT_FALSE(c_iter_->Valid()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/compaction_job.cc000066400000000000000000001216061267530634000170460ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/compaction_job.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include #include #include #include #include "db/builder.h" #include "db/db_iter.h" #include "db/dbformat.h" #include "db/event_helpers.h" #include "db/filename.h" #include "db/log_reader.h" #include "db/log_writer.h" #include "db/memtable.h" #include "db/memtable_list.h" #include "db/merge_context.h" #include "db/merge_helper.h" #include "db/version_set.h" #include "port/likely.h" #include "port/port.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/statistics.h" #include "rocksdb/status.h" #include "rocksdb/table.h" #include "table/block.h" #include "table/block_based_table_factory.h" #include "table/merger.h" #include "table/table_builder.h" #include "util/coding.h" #include "util/file_reader_writer.h" #include "util/iostats_context_imp.h" #include "util/log_buffer.h" #include "util/logging.h" #include "util/sst_file_manager_impl.h" #include "util/mutexlock.h" #include "util/perf_context_imp.h" #include "util/stop_watch.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/thread_status_util.h" namespace rocksdb { // Maintains state for each sub-compaction struct CompactionJob::SubcompactionState { Compaction* compaction; std::unique_ptr c_iter; // The boundaries of the key-range this compaction is interested in. No two // subcompactions may have overlapping key-ranges. // 'start' is inclusive, 'end' is exclusive, and nullptr means unbounded Slice *start, *end; // The return status of this subcompaction Status status; // Files produced by this subcompaction struct Output { FileMetaData meta; bool finished; std::shared_ptr table_properties; }; // State kept for output being generated std::vector outputs; std::unique_ptr outfile; std::unique_ptr builder; Output* current_output() { if (outputs.empty()) { // This subcompaction's outptut could be empty if compaction was aborted // before this subcompaction had a chance to generate any output files. // When subcompactions are executed sequentially this is more likely and // will be particulalry likely for the later subcompactions to be empty. // Once they are run in parallel however it should be much rarer. return nullptr; } else { return &outputs.back(); } } // State during the subcompaction uint64_t total_bytes; uint64_t num_input_records; uint64_t num_output_records; CompactionJobStats compaction_job_stats; uint64_t approx_size; SubcompactionState(Compaction* c, Slice* _start, Slice* _end, uint64_t size = 0) : compaction(c), start(_start), end(_end), outfile(nullptr), builder(nullptr), total_bytes(0), num_input_records(0), num_output_records(0), approx_size(size) { assert(compaction != nullptr); } SubcompactionState(SubcompactionState&& o) { *this = std::move(o); } SubcompactionState& operator=(SubcompactionState&& o) { compaction = std::move(o.compaction); start = std::move(o.start); end = std::move(o.end); status = std::move(o.status); outputs = std::move(o.outputs); outfile = std::move(o.outfile); builder = std::move(o.builder); total_bytes = std::move(o.total_bytes); num_input_records = std::move(o.num_input_records); num_output_records = std::move(o.num_output_records); compaction_job_stats = std::move(o.compaction_job_stats); approx_size = std::move(o.approx_size); return *this; } // Because member unique_ptrs do not have these. SubcompactionState(const SubcompactionState&) = delete; SubcompactionState& operator=(const SubcompactionState&) = delete; }; // Maintains state for the entire compaction struct CompactionJob::CompactionState { Compaction* const compaction; // REQUIRED: subcompaction states are stored in order of increasing // key-range std::vector sub_compact_states; Status status; uint64_t total_bytes; uint64_t num_input_records; uint64_t num_output_records; explicit CompactionState(Compaction* c) : compaction(c), total_bytes(0), num_input_records(0), num_output_records(0) {} size_t NumOutputFiles() { size_t total = 0; for (auto& s : sub_compact_states) { total += s.outputs.size(); } return total; } Slice SmallestUserKey() { for (const auto& sub_compact_state : sub_compact_states) { if (!sub_compact_state.outputs.empty() && sub_compact_state.outputs[0].finished) { return sub_compact_state.outputs[0].meta.smallest.user_key(); } } // If there is no finished output, return an empty slice. return Slice(nullptr, 0); } Slice LargestUserKey() { for (auto it = sub_compact_states.rbegin(); it < sub_compact_states.rend(); ++it) { if (!it->outputs.empty() && it->current_output()->finished) { assert(it->current_output() != nullptr); return it->current_output()->meta.largest.user_key(); } } // If there is no finished output, return an empty slice. return Slice(nullptr, 0); } }; void CompactionJob::AggregateStatistics() { for (SubcompactionState& sc : compact_->sub_compact_states) { compact_->total_bytes += sc.total_bytes; compact_->num_input_records += sc.num_input_records; compact_->num_output_records += sc.num_output_records; } if (compaction_job_stats_) { for (SubcompactionState& sc : compact_->sub_compact_states) { compaction_job_stats_->Add(sc.compaction_job_stats); } } } CompactionJob::CompactionJob( int job_id, Compaction* compaction, const DBOptions& db_options, const EnvOptions& env_options, VersionSet* versions, std::atomic* shutting_down, LogBuffer* log_buffer, Directory* db_directory, Directory* output_directory, Statistics* stats, std::vector existing_snapshots, SequenceNumber earliest_write_conflict_snapshot, std::shared_ptr table_cache, EventLogger* event_logger, bool paranoid_file_checks, bool measure_io_stats, const std::string& dbname, CompactionJobStats* compaction_job_stats) : job_id_(job_id), compact_(new CompactionState(compaction)), compaction_job_stats_(compaction_job_stats), compaction_stats_(1), dbname_(dbname), db_options_(db_options), env_options_(env_options), env_(db_options.env), versions_(versions), shutting_down_(shutting_down), log_buffer_(log_buffer), db_directory_(db_directory), output_directory_(output_directory), stats_(stats), existing_snapshots_(std::move(existing_snapshots)), earliest_write_conflict_snapshot_(earliest_write_conflict_snapshot), table_cache_(std::move(table_cache)), event_logger_(event_logger), paranoid_file_checks_(paranoid_file_checks), measure_io_stats_(measure_io_stats) { assert(log_buffer_ != nullptr); const auto* cfd = compact_->compaction->column_family_data(); ThreadStatusUtil::SetColumnFamily(cfd, cfd->ioptions()->env, cfd->options()->enable_thread_tracking); ThreadStatusUtil::SetThreadOperation(ThreadStatus::OP_COMPACTION); ReportStartedCompaction(compaction); } CompactionJob::~CompactionJob() { assert(compact_ == nullptr); ThreadStatusUtil::ResetThreadStatus(); } void CompactionJob::ReportStartedCompaction( Compaction* compaction) { const auto* cfd = compact_->compaction->column_family_data(); ThreadStatusUtil::SetColumnFamily(cfd, cfd->ioptions()->env, cfd->options()->enable_thread_tracking); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_JOB_ID, job_id_); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_INPUT_OUTPUT_LEVEL, (static_cast(compact_->compaction->start_level()) << 32) + compact_->compaction->output_level()); // In the current design, a CompactionJob is always created // for non-trivial compaction. assert(compaction->IsTrivialMove() == false || compaction->is_manual_compaction() == true); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_PROP_FLAGS, compaction->is_manual_compaction() + (compaction->deletion_compaction() << 1)); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_TOTAL_INPUT_BYTES, compaction->CalculateTotalInputSize()); IOSTATS_RESET(bytes_written); IOSTATS_RESET(bytes_read); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_BYTES_WRITTEN, 0); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_BYTES_READ, 0); // Set the thread operation after operation properties // to ensure GetThreadList() can always show them all together. ThreadStatusUtil::SetThreadOperation( ThreadStatus::OP_COMPACTION); if (compaction_job_stats_) { compaction_job_stats_->is_manual_compaction = compaction->is_manual_compaction(); } } void CompactionJob::Prepare() { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_COMPACTION_PREPARE); // Generate file_levels_ for compaction berfore making Iterator auto* c = compact_->compaction; assert(c->column_family_data() != nullptr); assert(c->column_family_data()->current()->storage_info() ->NumLevelFiles(compact_->compaction->level()) > 0); // Is this compaction producing files at the bottommost level? bottommost_level_ = c->bottommost_level(); if (c->ShouldFormSubcompactions()) { const uint64_t start_micros = env_->NowMicros(); GenSubcompactionBoundaries(); MeasureTime(stats_, SUBCOMPACTION_SETUP_TIME, env_->NowMicros() - start_micros); assert(sizes_.size() == boundaries_.size() + 1); for (size_t i = 0; i <= boundaries_.size(); i++) { Slice* start = i == 0 ? nullptr : &boundaries_[i - 1]; Slice* end = i == boundaries_.size() ? nullptr : &boundaries_[i]; compact_->sub_compact_states.emplace_back(c, start, end, sizes_[i]); } MeasureTime(stats_, NUM_SUBCOMPACTIONS_SCHEDULED, compact_->sub_compact_states.size()); } else { compact_->sub_compact_states.emplace_back(c, nullptr, nullptr); } } struct RangeWithSize { Range range; uint64_t size; RangeWithSize(const Slice& a, const Slice& b, uint64_t s = 0) : range(a, b), size(s) {} }; // Generates a histogram representing potential divisions of key ranges from // the input. It adds the starting and/or ending keys of certain input files // to the working set and then finds the approximate size of data in between // each consecutive pair of slices. Then it divides these ranges into // consecutive groups such that each group has a similar size. void CompactionJob::GenSubcompactionBoundaries() { auto* c = compact_->compaction; auto* cfd = c->column_family_data(); const Comparator* cfd_comparator = cfd->user_comparator(); std::vector bounds; int start_lvl = c->start_level(); int out_lvl = c->output_level(); // Add the starting and/or ending key of certain input files as a potential // boundary for (size_t lvl_idx = 0; lvl_idx < c->num_input_levels(); lvl_idx++) { int lvl = c->level(lvl_idx); if (lvl >= start_lvl && lvl <= out_lvl) { const LevelFilesBrief* flevel = c->input_levels(lvl_idx); size_t num_files = flevel->num_files; if (num_files == 0) { break; } if (lvl == 0) { // For level 0 add the starting and ending key of each file since the // files may have greatly differing key ranges (not range-partitioned) for (size_t i = 0; i < num_files; i++) { bounds.emplace_back(flevel->files[i].smallest_key); bounds.emplace_back(flevel->files[i].largest_key); } } else { // For all other levels add the smallest/largest key in the level to // encompass the range covered by that level bounds.emplace_back(flevel->files[0].smallest_key); bounds.emplace_back(flevel->files[num_files - 1].largest_key); if (lvl == out_lvl) { // For the last level include the starting keys of all files since // the last level is the largest and probably has the widest key // range. Since it's range partitioned, the ending key of one file // and the starting key of the next are very close (or identical). for (size_t i = 1; i < num_files; i++) { bounds.emplace_back(flevel->files[i].smallest_key); } } } } } std::sort(bounds.begin(), bounds.end(), [cfd_comparator] (const Slice& a, const Slice& b) -> bool { return cfd_comparator->Compare(ExtractUserKey(a), ExtractUserKey(b)) < 0; }); // Remove duplicated entries from bounds bounds.erase(std::unique(bounds.begin(), bounds.end(), [cfd_comparator] (const Slice& a, const Slice& b) -> bool { return cfd_comparator->Compare(ExtractUserKey(a), ExtractUserKey(b)) == 0; }), bounds.end()); // Combine consecutive pairs of boundaries into ranges with an approximate // size of data covered by keys in that range uint64_t sum = 0; std::vector ranges; auto* v = cfd->current(); for (auto it = bounds.begin();;) { const Slice a = *it; it++; if (it == bounds.end()) { break; } const Slice b = *it; uint64_t size = versions_->ApproximateSize(v, a, b, start_lvl, out_lvl + 1); ranges.emplace_back(a, b, size); sum += size; } // Group the ranges into subcompactions const double min_file_fill_percent = 4.0 / 5; uint64_t max_output_files = static_cast(std::ceil( sum / min_file_fill_percent / cfd->GetCurrentMutableCFOptions()->MaxFileSizeForLevel(out_lvl))); uint64_t subcompactions = std::min({static_cast(ranges.size()), static_cast(db_options_.max_subcompactions), max_output_files}); double mean = sum * 1.0 / subcompactions; if (subcompactions > 1) { // Greedily add ranges to the subcompaction until the sum of the ranges' // sizes becomes >= the expected mean size of a subcompaction sum = 0; for (size_t i = 0; i < ranges.size() - 1; i++) { sum += ranges[i].size; if (subcompactions == 1) { // If there's only one left to schedule then it goes to the end so no // need to put an end boundary continue; } if (sum >= mean) { boundaries_.emplace_back(ExtractUserKey(ranges[i].range.limit)); sizes_.emplace_back(sum); subcompactions--; sum = 0; } } sizes_.emplace_back(sum + ranges.back().size); } else { // Only one range so its size is the total sum of sizes computed above sizes_.emplace_back(sum); } } Status CompactionJob::Run() { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_COMPACTION_RUN); TEST_SYNC_POINT("CompactionJob::Run():Start"); log_buffer_->FlushBufferToLog(); LogCompaction(); const size_t num_threads = compact_->sub_compact_states.size(); assert(num_threads > 0); const uint64_t start_micros = env_->NowMicros(); // Launch a thread for each of subcompactions 1...num_threads-1 std::vector thread_pool; thread_pool.reserve(num_threads - 1); for (size_t i = 1; i < compact_->sub_compact_states.size(); i++) { thread_pool.emplace_back(&CompactionJob::ProcessKeyValueCompaction, this, &compact_->sub_compact_states[i]); } // Always schedule the first subcompaction (whether or not there are also // others) in the current thread to be efficient with resources ProcessKeyValueCompaction(&compact_->sub_compact_states[0]); // Wait for all other threads (if there are any) to finish execution for (auto& thread : thread_pool) { thread.join(); } if (output_directory_ && !db_options_.disableDataSync) { output_directory_->Fsync(); } compaction_stats_.micros = env_->NowMicros() - start_micros; MeasureTime(stats_, COMPACTION_TIME, compaction_stats_.micros); // Check if any thread encountered an error during execution Status status; for (const auto& state : compact_->sub_compact_states) { if (!state.status.ok()) { status = state.status; break; } } TablePropertiesCollection tp; auto sfm = static_cast(db_options_.sst_file_manager.get()); for (const auto& state : compact_->sub_compact_states) { for (const auto& output : state.outputs) { auto fn = TableFileName(db_options_.db_paths, output.meta.fd.GetNumber(), output.meta.fd.GetPathId()); tp[fn] = output.table_properties; if (sfm && output.meta.fd.GetPathId() == 0) { sfm->OnAddFile(fn); } } } compact_->compaction->SetOutputTableProperties(std::move(tp)); // Finish up all book-keeping to unify the subcompaction results AggregateStatistics(); UpdateCompactionStats(); RecordCompactionIOStats(); LogFlush(db_options_.info_log); TEST_SYNC_POINT("CompactionJob::Run():End"); compact_->status = status; return status; } Status CompactionJob::Install(const MutableCFOptions& mutable_cf_options, InstrumentedMutex* db_mutex) { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_COMPACTION_INSTALL); db_mutex->AssertHeld(); Status status = compact_->status; ColumnFamilyData* cfd = compact_->compaction->column_family_data(); cfd->internal_stats()->AddCompactionStats( compact_->compaction->output_level(), compaction_stats_); if (status.ok()) { status = InstallCompactionResults(mutable_cf_options, db_mutex); } VersionStorageInfo::LevelSummaryStorage tmp; auto vstorage = cfd->current()->storage_info(); const auto& stats = compaction_stats_; LogToBuffer( log_buffer_, "[%s] compacted to: %s, MB/sec: %.1f rd, %.1f wr, level %d, " "files in(%d, %d) out(%d) " "MB in(%.1f, %.1f) out(%.1f), read-write-amplify(%.1f) " "write-amplify(%.1f) %s, records in: %d, records dropped: %d\n", cfd->GetName().c_str(), vstorage->LevelSummary(&tmp), (stats.bytes_read_non_output_levels + stats.bytes_read_output_level) / static_cast(stats.micros), stats.bytes_written / static_cast(stats.micros), compact_->compaction->output_level(), stats.num_input_files_in_non_output_levels, stats.num_input_files_in_output_level, stats.num_output_files, stats.bytes_read_non_output_levels / 1048576.0, stats.bytes_read_output_level / 1048576.0, stats.bytes_written / 1048576.0, (stats.bytes_written + stats.bytes_read_output_level + stats.bytes_read_non_output_levels) / static_cast(stats.bytes_read_non_output_levels), stats.bytes_written / static_cast(stats.bytes_read_non_output_levels), status.ToString().c_str(), stats.num_input_records, stats.num_dropped_records); UpdateCompactionJobStats(stats); auto stream = event_logger_->LogToBuffer(log_buffer_); stream << "job" << job_id_ << "event" << "compaction_finished" << "compaction_time_micros" << compaction_stats_.micros << "output_level" << compact_->compaction->output_level() << "num_output_files" << compact_->NumOutputFiles() << "total_output_size" << compact_->total_bytes << "num_input_records" << compact_->num_input_records << "num_output_records" << compact_->num_output_records << "num_subcompactions" << compact_->sub_compact_states.size(); if (measure_io_stats_ && compaction_job_stats_ != nullptr) { stream << "file_write_nanos" << compaction_job_stats_->file_write_nanos; stream << "file_range_sync_nanos" << compaction_job_stats_->file_range_sync_nanos; stream << "file_fsync_nanos" << compaction_job_stats_->file_fsync_nanos; stream << "file_prepare_write_nanos" << compaction_job_stats_->file_prepare_write_nanos; } stream << "lsm_state"; stream.StartArray(); for (int level = 0; level < vstorage->num_levels(); ++level) { stream << vstorage->NumLevelFiles(level); } stream.EndArray(); CleanupCompaction(); return status; } void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) { assert(sub_compact != nullptr); std::unique_ptr input( versions_->MakeInputIterator(sub_compact->compaction)); AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_COMPACTION_PROCESS_KV); // I/O measurement variables PerfLevel prev_perf_level = PerfLevel::kEnableTime; const uint64_t kRecordStatsEvery = 1000; uint64_t prev_write_nanos = 0; uint64_t prev_fsync_nanos = 0; uint64_t prev_range_sync_nanos = 0; uint64_t prev_prepare_write_nanos = 0; if (measure_io_stats_) { prev_perf_level = GetPerfLevel(); SetPerfLevel(PerfLevel::kEnableTime); prev_write_nanos = IOSTATS(write_nanos); prev_fsync_nanos = IOSTATS(fsync_nanos); prev_range_sync_nanos = IOSTATS(range_sync_nanos); prev_prepare_write_nanos = IOSTATS(prepare_write_nanos); } ColumnFamilyData* cfd = sub_compact->compaction->column_family_data(); auto compaction_filter = cfd->ioptions()->compaction_filter; std::unique_ptr compaction_filter_from_factory = nullptr; if (compaction_filter == nullptr) { compaction_filter_from_factory = sub_compact->compaction->CreateCompactionFilter(); compaction_filter = compaction_filter_from_factory.get(); } MergeHelper merge( env_, cfd->user_comparator(), cfd->ioptions()->merge_operator, compaction_filter, db_options_.info_log.get(), cfd->ioptions()->min_partial_merge_operands, false /* internal key corruption is expected */, existing_snapshots_.empty() ? 0 : existing_snapshots_.back(), compact_->compaction->level(), db_options_.statistics.get()); TEST_SYNC_POINT("CompactionJob::Run():Inprogress"); Slice* start = sub_compact->start; Slice* end = sub_compact->end; if (start != nullptr) { IterKey start_iter; start_iter.SetInternalKey(*start, kMaxSequenceNumber, kValueTypeForSeek); input->Seek(start_iter.GetKey()); } else { input->SeekToFirst(); } Status status; sub_compact->c_iter.reset(new CompactionIterator( input.get(), cfd->user_comparator(), &merge, versions_->LastSequence(), &existing_snapshots_, earliest_write_conflict_snapshot_, env_, false, sub_compact->compaction, compaction_filter)); auto c_iter = sub_compact->c_iter.get(); c_iter->SeekToFirst(); const auto& c_iter_stats = c_iter->iter_stats(); // TODO(noetzli): check whether we could check !shutting_down_->... only // only occasionally (see diff D42687) while (status.ok() && !shutting_down_->load(std::memory_order_acquire) && !cfd->IsDropped() && c_iter->Valid()) { // Invariant: c_iter.status() is guaranteed to be OK if c_iter->Valid() // returns true. const Slice& key = c_iter->key(); const Slice& value = c_iter->value(); // If an end key (exclusive) is specified, check if the current key is // >= than it and exit if it is because the iterator is out of its range if (end != nullptr && cfd->user_comparator()->Compare(c_iter->user_key(), *end) >= 0) { break; } else if (sub_compact->compaction->ShouldStopBefore(key) && sub_compact->builder != nullptr) { status = FinishCompactionOutputFile(input->status(), sub_compact); if (!status.ok()) { break; } } if (c_iter_stats.num_input_records % kRecordStatsEvery == kRecordStatsEvery - 1) { RecordDroppedKeys(c_iter_stats, &sub_compact->compaction_job_stats); c_iter->ResetRecordCounts(); RecordCompactionIOStats(); } // Open output file if necessary if (sub_compact->builder == nullptr) { status = OpenCompactionOutputFile(sub_compact); if (!status.ok()) { break; } } assert(sub_compact->builder != nullptr); assert(sub_compact->current_output() != nullptr); sub_compact->builder->Add(key, value); sub_compact->current_output()->meta.UpdateBoundaries( key, c_iter->ikey().sequence); sub_compact->num_output_records++; // Close output file if it is big enough // TODO(aekmekji): determine if file should be closed earlier than this // during subcompactions (i.e. if output size, estimated by input size, is // going to be 1.2MB and max_output_file_size = 1MB, prefer to have 0.6MB // and 0.6MB instead of 1MB and 0.2MB) if (sub_compact->builder->FileSize() >= sub_compact->compaction->max_output_file_size()) { status = FinishCompactionOutputFile(input->status(), sub_compact); } c_iter->Next(); } sub_compact->num_input_records = c_iter_stats.num_input_records; sub_compact->compaction_job_stats.num_input_deletion_records = c_iter_stats.num_input_deletion_records; sub_compact->compaction_job_stats.num_corrupt_keys = c_iter_stats.num_input_corrupt_records; sub_compact->compaction_job_stats.total_input_raw_key_bytes += c_iter_stats.total_input_raw_key_bytes; sub_compact->compaction_job_stats.total_input_raw_value_bytes += c_iter_stats.total_input_raw_value_bytes; RecordTick(stats_, FILTER_OPERATION_TOTAL_TIME, c_iter_stats.total_filter_time); RecordDroppedKeys(c_iter_stats, &sub_compact->compaction_job_stats); RecordCompactionIOStats(); if (status.ok() && (shutting_down_->load(std::memory_order_acquire) || cfd->IsDropped())) { status = Status::ShutdownInProgress( "Database shutdown or Column family drop during compaction"); } if (status.ok() && sub_compact->builder != nullptr) { status = FinishCompactionOutputFile(input->status(), sub_compact); } if (status.ok()) { status = input->status(); } if (measure_io_stats_) { sub_compact->compaction_job_stats.file_write_nanos += IOSTATS(write_nanos) - prev_write_nanos; sub_compact->compaction_job_stats.file_fsync_nanos += IOSTATS(fsync_nanos) - prev_fsync_nanos; sub_compact->compaction_job_stats.file_range_sync_nanos += IOSTATS(range_sync_nanos) - prev_range_sync_nanos; sub_compact->compaction_job_stats.file_prepare_write_nanos += IOSTATS(prepare_write_nanos) - prev_prepare_write_nanos; if (prev_perf_level != PerfLevel::kEnableTime) { SetPerfLevel(prev_perf_level); } } sub_compact->c_iter.reset(); input.reset(); sub_compact->status = status; } void CompactionJob::RecordDroppedKeys( const CompactionIteratorStats& c_iter_stats, CompactionJobStats* compaction_job_stats) { if (c_iter_stats.num_record_drop_user > 0) { RecordTick(stats_, COMPACTION_KEY_DROP_USER, c_iter_stats.num_record_drop_user); } if (c_iter_stats.num_record_drop_hidden > 0) { RecordTick(stats_, COMPACTION_KEY_DROP_NEWER_ENTRY, c_iter_stats.num_record_drop_hidden); if (compaction_job_stats) { compaction_job_stats->num_records_replaced += c_iter_stats.num_record_drop_hidden; } } if (c_iter_stats.num_record_drop_obsolete > 0) { RecordTick(stats_, COMPACTION_KEY_DROP_OBSOLETE, c_iter_stats.num_record_drop_obsolete); if (compaction_job_stats) { compaction_job_stats->num_expired_deletion_records += c_iter_stats.num_record_drop_obsolete; } } } Status CompactionJob::FinishCompactionOutputFile( const Status& input_status, SubcompactionState* sub_compact) { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_COMPACTION_SYNC_FILE); assert(sub_compact != nullptr); assert(sub_compact->outfile); assert(sub_compact->builder != nullptr); assert(sub_compact->current_output() != nullptr); uint64_t output_number = sub_compact->current_output()->meta.fd.GetNumber(); assert(output_number != 0); TableProperties table_properties; // Check for iterator errors Status s = input_status; auto meta = &sub_compact->current_output()->meta; const uint64_t current_entries = sub_compact->builder->NumEntries(); meta->marked_for_compaction = sub_compact->builder->NeedCompact(); if (s.ok()) { s = sub_compact->builder->Finish(); } else { sub_compact->builder->Abandon(); } const uint64_t current_bytes = sub_compact->builder->FileSize(); meta->fd.file_size = current_bytes; sub_compact->current_output()->finished = true; sub_compact->total_bytes += current_bytes; // Finish and check for file errors if (s.ok() && !db_options_.disableDataSync) { StopWatch sw(env_, stats_, COMPACTION_OUTFILE_SYNC_MICROS); s = sub_compact->outfile->Sync(db_options_.use_fsync); } if (s.ok()) { s = sub_compact->outfile->Close(); } sub_compact->outfile.reset(); if (s.ok() && current_entries > 0) { // Verify that the table is usable ColumnFamilyData* cfd = sub_compact->compaction->column_family_data(); InternalIterator* iter = cfd->table_cache()->NewIterator( ReadOptions(), env_options_, cfd->internal_comparator(), meta->fd, nullptr, cfd->internal_stats()->GetFileReadHist( compact_->compaction->output_level()), false); s = iter->status(); if (s.ok() && paranoid_file_checks_) { for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {} s = iter->status(); } delete iter; if (s.ok()) { auto tp = sub_compact->builder->GetTableProperties(); sub_compact->current_output()->table_properties = std::make_shared(tp); TableFileCreationInfo info(std::move(tp)); info.db_name = dbname_; info.cf_name = cfd->GetName(); info.file_path = TableFileName(cfd->ioptions()->db_paths, meta->fd.GetNumber(), meta->fd.GetPathId()); info.file_size = meta->fd.GetFileSize(); info.job_id = job_id_; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] [JOB %d] Generated table #%" PRIu64 ": %" PRIu64 " keys, %" PRIu64 " bytes%s", cfd->GetName().c_str(), job_id_, output_number, current_entries, current_bytes, meta->marked_for_compaction ? " (need compaction)" : ""); EventHelpers::LogAndNotifyTableFileCreation( event_logger_, cfd->ioptions()->listeners, meta->fd, info); } } sub_compact->builder.reset(); return s; } Status CompactionJob::InstallCompactionResults( const MutableCFOptions& mutable_cf_options, InstrumentedMutex* db_mutex) { db_mutex->AssertHeld(); auto* compaction = compact_->compaction; // paranoia: verify that the files that we started with // still exist in the current version and in the same original level. // This ensures that a concurrent compaction did not erroneously // pick the same files to compact_. if (!versions_->VerifyCompactionFileConsistency(compaction)) { Compaction::InputLevelSummaryBuffer inputs_summary; Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "[%s] [JOB %d] Compaction %s aborted", compaction->column_family_data()->GetName().c_str(), job_id_, compaction->InputLevelSummary(&inputs_summary)); return Status::Corruption("Compaction input files inconsistent"); } { Compaction::InputLevelSummaryBuffer inputs_summary; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] [JOB %d] Compacted %s => %" PRIu64 " bytes", compaction->column_family_data()->GetName().c_str(), job_id_, compaction->InputLevelSummary(&inputs_summary), compact_->total_bytes); } // Add compaction outputs compaction->AddInputDeletions(compact_->compaction->edit()); for (const auto& sub_compact : compact_->sub_compact_states) { for (const auto& out : sub_compact.outputs) { compaction->edit()->AddFile(compaction->output_level(), out.meta); } } return versions_->LogAndApply(compaction->column_family_data(), mutable_cf_options, compaction->edit(), db_mutex, db_directory_); } void CompactionJob::RecordCompactionIOStats() { RecordTick(stats_, COMPACT_READ_BYTES, IOSTATS(bytes_read)); ThreadStatusUtil::IncreaseThreadOperationProperty( ThreadStatus::COMPACTION_BYTES_READ, IOSTATS(bytes_read)); IOSTATS_RESET(bytes_read); RecordTick(stats_, COMPACT_WRITE_BYTES, IOSTATS(bytes_written)); ThreadStatusUtil::IncreaseThreadOperationProperty( ThreadStatus::COMPACTION_BYTES_WRITTEN, IOSTATS(bytes_written)); IOSTATS_RESET(bytes_written); } Status CompactionJob::OpenCompactionOutputFile( SubcompactionState* sub_compact) { assert(sub_compact != nullptr); assert(sub_compact->builder == nullptr); // no need to lock because VersionSet::next_file_number_ is atomic uint64_t file_number = versions_->NewFileNumber(); // Make the output file unique_ptr writable_file; std::string fname = TableFileName(db_options_.db_paths, file_number, sub_compact->compaction->output_path_id()); Status s = NewWritableFile(env_, fname, &writable_file, env_options_); if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "[%s] [JOB %d] OpenCompactionOutputFiles for table #%" PRIu64 " fails at NewWritableFile with status %s", sub_compact->compaction->column_family_data()->GetName().c_str(), job_id_, file_number, s.ToString().c_str()); LogFlush(db_options_.info_log); return s; } SubcompactionState::Output out; out.meta.fd = FileDescriptor(file_number, sub_compact->compaction->output_path_id(), 0); out.finished = false; sub_compact->outputs.push_back(out); writable_file->SetIOPriority(Env::IO_LOW); writable_file->SetPreallocationBlockSize(static_cast( sub_compact->compaction->OutputFilePreallocationSize())); sub_compact->outfile.reset( new WritableFileWriter(std::move(writable_file), env_options_)); ColumnFamilyData* cfd = sub_compact->compaction->column_family_data(); // If the Column family flag is to only optimize filters for hits, // we can skip creating filters if this is the bottommost_level where // data is going to be found bool skip_filters = cfd->ioptions()->optimize_filters_for_hits && bottommost_level_; sub_compact->builder.reset(NewTableBuilder( *cfd->ioptions(), cfd->internal_comparator(), cfd->int_tbl_prop_collector_factories(), cfd->GetID(), sub_compact->outfile.get(), sub_compact->compaction->output_compression(), cfd->ioptions()->compression_opts, skip_filters)); LogFlush(db_options_.info_log); return s; } void CompactionJob::CleanupCompaction() { for (SubcompactionState& sub_compact : compact_->sub_compact_states) { const auto& sub_status = sub_compact.status; if (sub_compact.builder != nullptr) { // May happen if we get a shutdown call in the middle of compaction sub_compact.builder->Abandon(); sub_compact.builder.reset(); } else { assert(!sub_status.ok() || sub_compact.outfile == nullptr); } for (const auto& out : sub_compact.outputs) { // If this file was inserted into the table cache then remove // them here because this compaction was not committed. if (!sub_status.ok()) { TableCache::Evict(table_cache_.get(), out.meta.fd.GetNumber()); } } } delete compact_; compact_ = nullptr; } #ifndef ROCKSDB_LITE namespace { void CopyPrefix( const Slice& src, size_t prefix_length, std::string* dst) { assert(prefix_length > 0); size_t length = src.size() > prefix_length ? prefix_length : src.size(); dst->assign(src.data(), length); } } // namespace #endif // !ROCKSDB_LITE void CompactionJob::UpdateCompactionStats() { Compaction* compaction = compact_->compaction; compaction_stats_.num_input_files_in_non_output_levels = 0; compaction_stats_.num_input_files_in_output_level = 0; for (int input_level = 0; input_level < static_cast(compaction->num_input_levels()); ++input_level) { if (compaction->start_level() + input_level != compaction->output_level()) { UpdateCompactionInputStatsHelper( &compaction_stats_.num_input_files_in_non_output_levels, &compaction_stats_.bytes_read_non_output_levels, input_level); } else { UpdateCompactionInputStatsHelper( &compaction_stats_.num_input_files_in_output_level, &compaction_stats_.bytes_read_output_level, input_level); } } for (const auto& sub_compact : compact_->sub_compact_states) { size_t num_output_files = sub_compact.outputs.size(); if (sub_compact.builder != nullptr) { // An error occurred so ignore the last output. assert(num_output_files > 0); --num_output_files; } compaction_stats_.num_output_files += static_cast(num_output_files); for (const auto& out : sub_compact.outputs) { compaction_stats_.bytes_written += out.meta.fd.file_size; } if (sub_compact.num_input_records > sub_compact.num_output_records) { compaction_stats_.num_dropped_records += sub_compact.num_input_records - sub_compact.num_output_records; } } } void CompactionJob::UpdateCompactionInputStatsHelper( int* num_files, uint64_t* bytes_read, int input_level) { const Compaction* compaction = compact_->compaction; auto num_input_files = compaction->num_input_files(input_level); *num_files += static_cast(num_input_files); for (size_t i = 0; i < num_input_files; ++i) { const auto* file_meta = compaction->input(input_level, i); *bytes_read += file_meta->fd.GetFileSize(); compaction_stats_.num_input_records += static_cast(file_meta->num_entries); } } void CompactionJob::UpdateCompactionJobStats( const InternalStats::CompactionStats& stats) const { #ifndef ROCKSDB_LITE if (compaction_job_stats_) { compaction_job_stats_->elapsed_micros = stats.micros; // input information compaction_job_stats_->total_input_bytes = stats.bytes_read_non_output_levels + stats.bytes_read_output_level; compaction_job_stats_->num_input_records = compact_->num_input_records; compaction_job_stats_->num_input_files = stats.num_input_files_in_non_output_levels + stats.num_input_files_in_output_level; compaction_job_stats_->num_input_files_at_output_level = stats.num_input_files_in_output_level; // output information compaction_job_stats_->total_output_bytes = stats.bytes_written; compaction_job_stats_->num_output_records = compact_->num_output_records; compaction_job_stats_->num_output_files = stats.num_output_files; if (compact_->NumOutputFiles() > 0U) { CopyPrefix( compact_->SmallestUserKey(), CompactionJobStats::kMaxPrefixLength, &compaction_job_stats_->smallest_output_key_prefix); CopyPrefix( compact_->LargestUserKey(), CompactionJobStats::kMaxPrefixLength, &compaction_job_stats_->largest_output_key_prefix); } } #endif // !ROCKSDB_LITE } void CompactionJob::LogCompaction() { Compaction* compaction = compact_->compaction; ColumnFamilyData* cfd = compaction->column_family_data(); // Let's check if anything will get logged. Don't prepare all the info if // we're not logging if (db_options_.info_log_level <= InfoLogLevel::INFO_LEVEL) { Compaction::InputLevelSummaryBuffer inputs_summary; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] [JOB %d] Compacting %s, score %.2f", cfd->GetName().c_str(), job_id_, compaction->InputLevelSummary(&inputs_summary), compaction->score()); char scratch[2345]; compaction->Summary(scratch, sizeof(scratch)); Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] Compaction start summary: %s\n", cfd->GetName().c_str(), scratch); // build event logger report auto stream = event_logger_->Log(); stream << "job" << job_id_ << "event" << "compaction_started"; for (size_t i = 0; i < compaction->num_input_levels(); ++i) { stream << ("files_L" + ToString(compaction->level(i))); stream.StartArray(); for (auto f : *compaction->inputs(i)) { stream << f->fd.GetNumber(); } stream.EndArray(); } stream << "score" << compaction->score() << "input_data_size" << compaction->CalculateTotalInputSize(); } } } // namespace rocksdb rocksdb-4.5.1/db/compaction_job.h000066400000000000000000000124121267530634000167020ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include #include #include #include "db/column_family.h" #include "db/compaction_iterator.h" #include "db/dbformat.h" #include "db/flush_scheduler.h" #include "db/internal_stats.h" #include "db/job_context.h" #include "db/log_writer.h" #include "db/memtable_list.h" #include "db/version_edit.h" #include "db/write_controller.h" #include "db/write_thread.h" #include "port/port.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/compaction_job_stats.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "rocksdb/transaction_log.h" #include "table/scoped_arena_iterator.h" #include "util/autovector.h" #include "util/event_logger.h" #include "util/stop_watch.h" #include "util/thread_local.h" namespace rocksdb { class MemTable; class TableCache; class Version; class VersionEdit; class VersionSet; class Arena; class CompactionJob { public: CompactionJob(int job_id, Compaction* compaction, const DBOptions& db_options, const EnvOptions& env_options, VersionSet* versions, std::atomic* shutting_down, LogBuffer* log_buffer, Directory* db_directory, Directory* output_directory, Statistics* stats, std::vector existing_snapshots, SequenceNumber earliest_write_conflict_snapshot, std::shared_ptr table_cache, EventLogger* event_logger, bool paranoid_file_checks, bool measure_io_stats, const std::string& dbname, CompactionJobStats* compaction_job_stats); ~CompactionJob(); // no copy/move CompactionJob(CompactionJob&& job) = delete; CompactionJob(const CompactionJob& job) = delete; CompactionJob& operator=(const CompactionJob& job) = delete; // REQUIRED: mutex held void Prepare(); // REQUIRED mutex not held Status Run(); // REQUIRED: mutex held Status Install(const MutableCFOptions& mutable_cf_options, InstrumentedMutex* db_mutex); private: struct SubcompactionState; void AggregateStatistics(); void GenSubcompactionBoundaries(); // update the thread status for starting a compaction. void ReportStartedCompaction(Compaction* compaction); void AllocateCompactionOutputFileNumbers(); // Call compaction filter. Then iterate through input and compact the // kv-pairs void ProcessKeyValueCompaction(SubcompactionState* sub_compact); Status FinishCompactionOutputFile(const Status& input_status, SubcompactionState* sub_compact); Status InstallCompactionResults(const MutableCFOptions& mutable_cf_options, InstrumentedMutex* db_mutex); void RecordCompactionIOStats(); Status OpenCompactionOutputFile(SubcompactionState* sub_compact); void CleanupCompaction(); void UpdateCompactionJobStats( const InternalStats::CompactionStats& stats) const; void RecordDroppedKeys(const CompactionIteratorStats& c_iter_stats, CompactionJobStats* compaction_job_stats = nullptr); void UpdateCompactionStats(); void UpdateCompactionInputStatsHelper( int* num_files, uint64_t* bytes_read, int input_level); void LogCompaction(); int job_id_; // CompactionJob state struct CompactionState; CompactionState* compact_; CompactionJobStats* compaction_job_stats_; InternalStats::CompactionStats compaction_stats_; // DBImpl state const std::string& dbname_; const DBOptions& db_options_; const EnvOptions& env_options_; Env* env_; VersionSet* versions_; std::atomic* shutting_down_; LogBuffer* log_buffer_; Directory* db_directory_; Directory* output_directory_; Statistics* stats_; // If there were two snapshots with seq numbers s1 and // s2 and s1 < s2, and if we find two instances of a key k1 then lies // entirely within s1 and s2, then the earlier version of k1 can be safely // deleted because that version is not visible in any snapshot. std::vector existing_snapshots_; // This is the earliest snapshot that could be used for write-conflict // checking by a transaction. For any user-key newer than this snapshot, we // should make sure not to remove evidence that a write occurred. SequenceNumber earliest_write_conflict_snapshot_; std::shared_ptr table_cache_; EventLogger* event_logger_; bool bottommost_level_; bool paranoid_file_checks_; bool measure_io_stats_; // Stores the Slices that designate the boundaries for each subcompaction std::vector boundaries_; // Stores the approx size of keys covered in the range of each subcompaction std::vector sizes_; }; } // namespace rocksdb rocksdb-4.5.1/db/compaction_job_stats_test.cc000066400000000000000000001070041267530634000213170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include #include #include #include #include "db/db_impl.h" #include "db/dbformat.h" #include "db/filename.h" #include "db/job_context.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "memtable/hash_linklist_rep.h" #include "port/stack_trace.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/convenience.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/experimental.h" #include "rocksdb/filter_policy.h" #include "rocksdb/options.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "rocksdb/thread_status.h" #include "rocksdb/utilities/checkpoint.h" #include "rocksdb/utilities/write_batch_with_index.h" #include "table/block_based_table_factory.h" #include "table/mock_table.h" #include "table/plain_table_factory.h" #include "table/scoped_arena_iterator.h" #include "util/compression.h" #include "util/hash.h" #include "util/logging.h" #include "util/mock_env.h" #include "util/mutexlock.h" #include "util/rate_limiter.h" #include "util/statistics.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/testharness.h" #include "util/testutil.h" #include "util/thread_status_util.h" #include "util/xfunc.h" #include "utilities/merge_operators.h" #if !defined(IOS_CROSS_COMPILE) #ifndef ROCKSDB_LITE namespace rocksdb { static std::string RandomString(Random* rnd, int len, double ratio) { std::string r; test::CompressibleString(rnd, ratio, len, &r); return r; } std::string Key(uint64_t key, int length) { const int kBufSize = 1000; char buf[kBufSize]; if (length > kBufSize) { length = kBufSize; } snprintf(buf, kBufSize, "%0*" PRIu64, length, key); return std::string(buf); } class CompactionJobStatsTest : public testing::Test, public testing::WithParamInterface { public: std::string dbname_; std::string alternative_wal_dir_; Env* env_; DB* db_; std::vector handles_; uint32_t max_subcompactions_; Options last_options_; CompactionJobStatsTest() : env_(Env::Default()) { env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); dbname_ = test::TmpDir(env_) + "/compaction_job_stats_test"; alternative_wal_dir_ = dbname_ + "/wal"; Options options; options.create_if_missing = true; max_subcompactions_ = GetParam(); options.max_subcompactions = max_subcompactions_; auto delete_options = options; delete_options.wal_dir = alternative_wal_dir_; EXPECT_OK(DestroyDB(dbname_, delete_options)); // Destroy it for not alternative WAL dir is used. EXPECT_OK(DestroyDB(dbname_, options)); db_ = nullptr; Reopen(options); } ~CompactionJobStatsTest() { rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->LoadDependency({}); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); Close(); Options options; options.db_paths.emplace_back(dbname_, 0); options.db_paths.emplace_back(dbname_ + "_2", 0); options.db_paths.emplace_back(dbname_ + "_3", 0); options.db_paths.emplace_back(dbname_ + "_4", 0); EXPECT_OK(DestroyDB(dbname_, options)); } // Required if inheriting from testing::WithParamInterface<> static void SetUpTestCase() {} static void TearDownTestCase() {} DBImpl* dbfull() { return reinterpret_cast(db_); } void CreateColumnFamilies(const std::vector& cfs, const Options& options) { ColumnFamilyOptions cf_opts(options); size_t cfi = handles_.size(); handles_.resize(cfi + cfs.size()); for (auto cf : cfs) { ASSERT_OK(db_->CreateColumnFamily(cf_opts, cf, &handles_[cfi++])); } } void CreateAndReopenWithCF(const std::vector& cfs, const Options& options) { CreateColumnFamilies(cfs, options); std::vector cfs_plus_default = cfs; cfs_plus_default.insert(cfs_plus_default.begin(), kDefaultColumnFamilyName); ReopenWithColumnFamilies(cfs_plus_default, options); } void ReopenWithColumnFamilies(const std::vector& cfs, const std::vector& options) { ASSERT_OK(TryReopenWithColumnFamilies(cfs, options)); } void ReopenWithColumnFamilies(const std::vector& cfs, const Options& options) { ASSERT_OK(TryReopenWithColumnFamilies(cfs, options)); } Status TryReopenWithColumnFamilies( const std::vector& cfs, const std::vector& options) { Close(); EXPECT_EQ(cfs.size(), options.size()); std::vector column_families; for (size_t i = 0; i < cfs.size(); ++i) { column_families.push_back(ColumnFamilyDescriptor(cfs[i], options[i])); } DBOptions db_opts = DBOptions(options[0]); return DB::Open(db_opts, dbname_, column_families, &handles_, &db_); } Status TryReopenWithColumnFamilies(const std::vector& cfs, const Options& options) { Close(); std::vector v_opts(cfs.size(), options); return TryReopenWithColumnFamilies(cfs, v_opts); } void Reopen(const Options& options) { ASSERT_OK(TryReopen(options)); } void Close() { for (auto h : handles_) { delete h; } handles_.clear(); delete db_; db_ = nullptr; } void DestroyAndReopen(const Options& options) { // Destroy using last options Destroy(last_options_); ASSERT_OK(TryReopen(options)); } void Destroy(const Options& options) { Close(); ASSERT_OK(DestroyDB(dbname_, options)); } Status ReadOnlyReopen(const Options& options) { return DB::OpenForReadOnly(options, dbname_, &db_); } Status TryReopen(const Options& options) { Close(); last_options_ = options; return DB::Open(options, dbname_, &db_); } Status Flush(int cf = 0) { if (cf == 0) { return db_->Flush(FlushOptions()); } else { return db_->Flush(FlushOptions(), handles_[cf]); } } Status Put(const Slice& k, const Slice& v, WriteOptions wo = WriteOptions()) { return db_->Put(wo, k, v); } Status Put(int cf, const Slice& k, const Slice& v, WriteOptions wo = WriteOptions()) { return db_->Put(wo, handles_[cf], k, v); } Status Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); } Status Delete(int cf, const std::string& k) { return db_->Delete(WriteOptions(), handles_[cf], k); } std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) { ReadOptions options; options.verify_checksums = true; options.snapshot = snapshot; std::string result; Status s = db_->Get(options, k, &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } std::string Get(int cf, const std::string& k, const Snapshot* snapshot = nullptr) { ReadOptions options; options.verify_checksums = true; options.snapshot = snapshot; std::string result; Status s = db_->Get(options, handles_[cf], k, &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } int NumTableFilesAtLevel(int level, int cf = 0) { std::string property; if (cf == 0) { // default cfd EXPECT_TRUE(db_->GetProperty( "rocksdb.num-files-at-level" + NumberToString(level), &property)); } else { EXPECT_TRUE(db_->GetProperty( handles_[cf], "rocksdb.num-files-at-level" + NumberToString(level), &property)); } return atoi(property.c_str()); } // Return spread of files per level std::string FilesPerLevel(int cf = 0) { int num_levels = (cf == 0) ? db_->NumberLevels() : db_->NumberLevels(handles_[1]); std::string result; size_t last_non_zero_offset = 0; for (int level = 0; level < num_levels; level++) { int f = NumTableFilesAtLevel(level, cf); char buf[100]; snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); result += buf; if (f > 0) { last_non_zero_offset = result.size(); } } result.resize(last_non_zero_offset); return result; } uint64_t Size(const Slice& start, const Slice& limit, int cf = 0) { Range r(start, limit); uint64_t size; if (cf == 0) { db_->GetApproximateSizes(&r, 1, &size); } else { db_->GetApproximateSizes(handles_[1], &r, 1, &size); } return size; } void Compact(int cf, const Slice& start, const Slice& limit, uint32_t target_path_id) { CompactRangeOptions compact_options; compact_options.target_path_id = target_path_id; ASSERT_OK(db_->CompactRange(compact_options, handles_[cf], &start, &limit)); } void Compact(int cf, const Slice& start, const Slice& limit) { ASSERT_OK( db_->CompactRange(CompactRangeOptions(), handles_[cf], &start, &limit)); } void Compact(const Slice& start, const Slice& limit) { ASSERT_OK(db_->CompactRange(CompactRangeOptions(), &start, &limit)); } void TEST_Compact(int level, int cf, const Slice& start, const Slice& limit) { ASSERT_OK(dbfull()->TEST_CompactRange(level, &start, &limit, handles_[cf], true /* disallow trivial move */)); } // Do n memtable compactions, each of which produces an sstable // covering the range [small,large]. void MakeTables(int n, const std::string& small, const std::string& large, int cf = 0) { for (int i = 0; i < n; i++) { ASSERT_OK(Put(cf, small, "begin")); ASSERT_OK(Put(cf, large, "end")); ASSERT_OK(Flush(cf)); } } static void SetDeletionCompactionStats( CompactionJobStats *stats, uint64_t input_deletions, uint64_t expired_deletions, uint64_t records_replaced) { stats->num_input_deletion_records = input_deletions; stats->num_expired_deletion_records = expired_deletions; stats->num_records_replaced = records_replaced; } void MakeTableWithKeyValues( Random* rnd, uint64_t smallest, uint64_t largest, int key_size, int value_size, uint64_t interval, double ratio, int cf = 0) { for (auto key = smallest; key < largest; key += interval) { ASSERT_OK(Put(cf, Slice(Key(key, key_size)), Slice(RandomString(rnd, value_size, ratio)))); } ASSERT_OK(Flush(cf)); } // This function behaves with the implicit understanding that two // rounds of keys are inserted into the database, as per the behavior // of the DeletionStatsTest. void SelectivelyDeleteKeys(uint64_t smallest, uint64_t largest, uint64_t interval, int deletion_interval, int key_size, uint64_t cutoff_key_num, CompactionJobStats* stats, int cf = 0) { // interval needs to be >= 2 so that deletion entries can be inserted // that are intended to not result in an actual key deletion by using // an offset of 1 from another existing key ASSERT_GE(interval, 2); uint64_t ctr = 1; uint32_t deletions_made = 0; uint32_t num_deleted = 0; uint32_t num_expired = 0; for (auto key = smallest; key <= largest; key += interval, ctr++) { if (ctr % deletion_interval == 0) { ASSERT_OK(Delete(cf, Key(key, key_size))); deletions_made++; num_deleted++; if (key > cutoff_key_num) { num_expired++; } } } // Insert some deletions for keys that don't exist that // are both in and out of the key range ASSERT_OK(Delete(cf, Key(smallest+1, key_size))); deletions_made++; ASSERT_OK(Delete(cf, Key(smallest-1, key_size))); deletions_made++; num_expired++; ASSERT_OK(Delete(cf, Key(smallest-9, key_size))); deletions_made++; num_expired++; ASSERT_OK(Flush(cf)); SetDeletionCompactionStats(stats, deletions_made, num_expired, num_deleted); } }; // An EventListener which helps verify the compaction results in // test CompactionJobStatsTest. class CompactionJobStatsChecker : public EventListener { public: CompactionJobStatsChecker() : compression_enabled_(false), verify_next_comp_io_stats_(false) {} size_t NumberOfUnverifiedStats() { return expected_stats_.size(); } void set_verify_next_comp_io_stats(bool v) { verify_next_comp_io_stats_ = v; } // Once a compaction completed, this function will verify the returned // CompactionJobInfo with the oldest CompactionJobInfo added earlier // in "expected_stats_" which has not yet being used for verification. virtual void OnCompactionCompleted(DB *db, const CompactionJobInfo& ci) { if (verify_next_comp_io_stats_) { ASSERT_GT(ci.stats.file_write_nanos, 0); ASSERT_GT(ci.stats.file_range_sync_nanos, 0); ASSERT_GT(ci.stats.file_fsync_nanos, 0); ASSERT_GT(ci.stats.file_prepare_write_nanos, 0); verify_next_comp_io_stats_ = false; } std::lock_guard lock(mutex_); if (expected_stats_.size()) { Verify(ci.stats, expected_stats_.front()); expected_stats_.pop(); } } // A helper function which verifies whether two CompactionJobStats // match. The verification of all compaction stats are done by // ASSERT_EQ except for the total input / output bytes, which we // use ASSERT_GE and ASSERT_LE with a reasonable bias --- // 10% in uncompressed case and 20% when compression is used. virtual void Verify(const CompactionJobStats& current_stats, const CompactionJobStats& stats) { // time ASSERT_GT(current_stats.elapsed_micros, 0U); ASSERT_EQ(current_stats.num_input_records, stats.num_input_records); ASSERT_EQ(current_stats.num_input_files, stats.num_input_files); ASSERT_EQ(current_stats.num_input_files_at_output_level, stats.num_input_files_at_output_level); ASSERT_EQ(current_stats.num_output_records, stats.num_output_records); ASSERT_EQ(current_stats.num_output_files, stats.num_output_files); ASSERT_EQ(current_stats.is_manual_compaction, stats.is_manual_compaction); // file size double kFileSizeBias = compression_enabled_ ? 0.20 : 0.10; ASSERT_GE(current_stats.total_input_bytes * (1.00 + kFileSizeBias), stats.total_input_bytes); ASSERT_LE(current_stats.total_input_bytes, stats.total_input_bytes * (1.00 + kFileSizeBias)); ASSERT_GE(current_stats.total_output_bytes * (1.00 + kFileSizeBias), stats.total_output_bytes); ASSERT_LE(current_stats.total_output_bytes, stats.total_output_bytes * (1.00 + kFileSizeBias)); ASSERT_EQ(current_stats.total_input_raw_key_bytes, stats.total_input_raw_key_bytes); ASSERT_EQ(current_stats.total_input_raw_value_bytes, stats.total_input_raw_value_bytes); ASSERT_EQ(current_stats.num_records_replaced, stats.num_records_replaced); ASSERT_EQ(current_stats.num_corrupt_keys, stats.num_corrupt_keys); ASSERT_EQ( std::string(current_stats.smallest_output_key_prefix), std::string(stats.smallest_output_key_prefix)); ASSERT_EQ( std::string(current_stats.largest_output_key_prefix), std::string(stats.largest_output_key_prefix)); } // Add an expected compaction stats, which will be used to // verify the CompactionJobStats returned by the OnCompactionCompleted() // callback. void AddExpectedStats(const CompactionJobStats& stats) { std::lock_guard lock(mutex_); expected_stats_.push(stats); } void EnableCompression(bool flag) { compression_enabled_ = flag; } bool verify_next_comp_io_stats() const { return verify_next_comp_io_stats_; } private: std::mutex mutex_; std::queue expected_stats_; bool compression_enabled_; bool verify_next_comp_io_stats_; }; // An EventListener which helps verify the compaction statistics in // the test DeletionStatsTest. class CompactionJobDeletionStatsChecker : public CompactionJobStatsChecker { public: // Verifies whether two CompactionJobStats match. void Verify(const CompactionJobStats& current_stats, const CompactionJobStats& stats) { ASSERT_EQ( current_stats.num_input_deletion_records, stats.num_input_deletion_records); ASSERT_EQ( current_stats.num_expired_deletion_records, stats.num_expired_deletion_records); ASSERT_EQ( current_stats.num_records_replaced, stats.num_records_replaced); ASSERT_EQ(current_stats.num_corrupt_keys, stats.num_corrupt_keys); } }; namespace { uint64_t EstimatedFileSize( uint64_t num_records, size_t key_size, size_t value_size, double compression_ratio = 1.0, size_t block_size = 4096, int bloom_bits_per_key = 10) { const size_t kPerKeyOverhead = 8; const size_t kFooterSize = 512; uint64_t data_size = static_cast( num_records * (key_size + value_size * compression_ratio + kPerKeyOverhead)); return data_size + kFooterSize + num_records * bloom_bits_per_key / 8 // filter block + data_size * (key_size + 8) / block_size; // index block } namespace { void CopyPrefix( const Slice& src, size_t prefix_length, std::string* dst) { assert(prefix_length > 0); size_t length = src.size() > prefix_length ? prefix_length : src.size(); dst->assign(src.data(), length); } } // namespace CompactionJobStats NewManualCompactionJobStats( const std::string& smallest_key, const std::string& largest_key, size_t num_input_files, size_t num_input_files_at_output_level, uint64_t num_input_records, size_t key_size, size_t value_size, size_t num_output_files, uint64_t num_output_records, double compression_ratio, uint64_t num_records_replaced, bool is_manual = true) { CompactionJobStats stats; stats.Reset(); stats.num_input_records = num_input_records; stats.num_input_files = num_input_files; stats.num_input_files_at_output_level = num_input_files_at_output_level; stats.num_output_records = num_output_records; stats.num_output_files = num_output_files; stats.total_input_bytes = EstimatedFileSize( num_input_records / num_input_files, key_size, value_size, compression_ratio) * num_input_files; stats.total_output_bytes = EstimatedFileSize( num_output_records / num_output_files, key_size, value_size, compression_ratio) * num_output_files; stats.total_input_raw_key_bytes = num_input_records * (key_size + 8); stats.total_input_raw_value_bytes = num_input_records * value_size; stats.is_manual_compaction = is_manual; stats.num_records_replaced = num_records_replaced; CopyPrefix(smallest_key, CompactionJobStats::kMaxPrefixLength, &stats.smallest_output_key_prefix); CopyPrefix(largest_key, CompactionJobStats::kMaxPrefixLength, &stats.largest_output_key_prefix); return stats; } CompressionType GetAnyCompression() { if (Snappy_Supported()) { return kSnappyCompression; } else if (Zlib_Supported()) { return kZlibCompression; } else if (BZip2_Supported()) { return kBZip2Compression; } else if (LZ4_Supported()) { return kLZ4Compression; } return kNoCompression; } } // namespace TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) { Random rnd(301); const int kBufSize = 100; char buf[kBufSize]; uint64_t key_base = 100000000l; // Note: key_base must be multiple of num_keys_per_L0_file int num_keys_per_L0_file = 100; const int kTestScale = 8; const int kKeySize = 10; const int kValueSize = 1000; const double kCompressionRatio = 0.5; double compression_ratio = 1.0; uint64_t key_interval = key_base / num_keys_per_L0_file; // Whenever a compaction completes, this listener will try to // verify whether the returned CompactionJobStats matches // what we expect. The expected CompactionJobStats is added // via AddExpectedStats(). auto* stats_checker = new CompactionJobStatsChecker(); Options options; options.listeners.emplace_back(stats_checker); options.create_if_missing = true; options.max_background_flushes = 0; // just enough setting to hold off auto-compaction. options.level0_file_num_compaction_trigger = kTestScale + 1; options.num_levels = 3; options.compression = kNoCompression; options.max_subcompactions = max_subcompactions_; options.bytes_per_sync = 512 * 1024; options.compaction_measure_io_stats = true; for (int test = 0; test < 2; ++test) { DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // 1st Phase: generate "num_L0_files" L0 files. int num_L0_files = 0; for (uint64_t start_key = key_base; start_key <= key_base * kTestScale; start_key += key_base) { MakeTableWithKeyValues( &rnd, start_key, start_key + key_base - 1, kKeySize, kValueSize, key_interval, compression_ratio, 1); snprintf(buf, kBufSize, "%d", ++num_L0_files); ASSERT_EQ(std::string(buf), FilesPerLevel(1)); } ASSERT_EQ(ToString(num_L0_files), FilesPerLevel(1)); // 2nd Phase: perform L0 -> L1 compaction. int L0_compaction_count = 6; int count = 1; std::string smallest_key; std::string largest_key; for (uint64_t start_key = key_base; start_key <= key_base * L0_compaction_count; start_key += key_base, count++) { smallest_key = Key(start_key, 10); largest_key = Key(start_key + key_base - key_interval, 10); stats_checker->AddExpectedStats( NewManualCompactionJobStats( smallest_key, largest_key, 1, 0, num_keys_per_L0_file, kKeySize, kValueSize, 1, num_keys_per_L0_file, compression_ratio, 0)); ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 1U); TEST_Compact(0, 1, smallest_key, largest_key); snprintf(buf, kBufSize, "%d,%d", num_L0_files - count, count); ASSERT_EQ(std::string(buf), FilesPerLevel(1)); } // compact two files into one in the last L0 -> L1 compaction int num_remaining_L0 = num_L0_files - L0_compaction_count; smallest_key = Key(key_base * (L0_compaction_count + 1), 10); largest_key = Key(key_base * (kTestScale + 1) - key_interval, 10); stats_checker->AddExpectedStats( NewManualCompactionJobStats( smallest_key, largest_key, num_remaining_L0, 0, num_keys_per_L0_file * num_remaining_L0, kKeySize, kValueSize, 1, num_keys_per_L0_file * num_remaining_L0, compression_ratio, 0)); ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 1U); TEST_Compact(0, 1, smallest_key, largest_key); int num_L1_files = num_L0_files - num_remaining_L0 + 1; num_L0_files = 0; snprintf(buf, kBufSize, "%d,%d", num_L0_files, num_L1_files); ASSERT_EQ(std::string(buf), FilesPerLevel(1)); // 3rd Phase: generate sparse L0 files (wider key-range, same num of keys) int sparseness = 2; for (uint64_t start_key = key_base; start_key <= key_base * kTestScale; start_key += key_base * sparseness) { MakeTableWithKeyValues( &rnd, start_key, start_key + key_base * sparseness - 1, kKeySize, kValueSize, key_base * sparseness / num_keys_per_L0_file, compression_ratio, 1); snprintf(buf, kBufSize, "%d,%d", ++num_L0_files, num_L1_files); ASSERT_EQ(std::string(buf), FilesPerLevel(1)); } // 4th Phase: perform L0 -> L1 compaction again, expect higher write amp // When subcompactions are enabled, the number of output files increases // by 1 because multiple threads are consuming the input and generating // output files without coordinating to see if the output could fit into // a smaller number of files like it does when it runs sequentially int num_output_files = options.max_subcompactions > 1 ? 2 : 1; for (uint64_t start_key = key_base; num_L0_files > 1; start_key += key_base * sparseness) { smallest_key = Key(start_key, 10); largest_key = Key(start_key + key_base * sparseness - key_interval, 10); stats_checker->AddExpectedStats( NewManualCompactionJobStats( smallest_key, largest_key, 3, 2, num_keys_per_L0_file * 3, kKeySize, kValueSize, num_output_files, num_keys_per_L0_file * 2, // 1/3 of the data will be updated. compression_ratio, num_keys_per_L0_file)); ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 1U); Compact(1, smallest_key, largest_key); if (options.max_subcompactions == 1) { --num_L1_files; } snprintf(buf, kBufSize, "%d,%d", --num_L0_files, num_L1_files); ASSERT_EQ(std::string(buf), FilesPerLevel(1)); } // 5th Phase: Do a full compaction, which involves in two sub-compactions. // Here we expect to have 1 L0 files and 4 L1 files // In the first sub-compaction, we expect L0 compaction. smallest_key = Key(key_base, 10); largest_key = Key(key_base * (kTestScale + 1) - key_interval, 10); stats_checker->AddExpectedStats( NewManualCompactionJobStats( Key(key_base * (kTestScale + 1 - sparseness), 10), largest_key, 2, 1, num_keys_per_L0_file * 3, kKeySize, kValueSize, 1, num_keys_per_L0_file * 2, compression_ratio, num_keys_per_L0_file)); ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 1U); Compact(1, smallest_key, largest_key); num_L1_files = options.max_subcompactions > 1 ? 7 : 4; char L1_buf[4]; snprintf(L1_buf, sizeof(L1_buf), "0,%d", num_L1_files); std::string L1_files(L1_buf); ASSERT_EQ(L1_files, FilesPerLevel(1)); options.compression = GetAnyCompression(); if (options.compression == kNoCompression) { break; } stats_checker->EnableCompression(true); compression_ratio = kCompressionRatio; for (int i = 0; i < 5; i++) { ASSERT_OK(Put(1, Slice(Key(key_base + i, 10)), Slice(RandomString(&rnd, 512 * 1024, 1)))); } ASSERT_OK(Flush(1)); reinterpret_cast(db_)->TEST_WaitForCompact(); stats_checker->set_verify_next_comp_io_stats(true); std::atomic first_prepare_write(true); rocksdb::SyncPoint::GetInstance()->SetCallBack( "WritableFileWriter::Append:BeforePrepareWrite", [&](void* arg) { if (first_prepare_write.load()) { options.env->SleepForMicroseconds(3); first_prepare_write.store(false); } }); std::atomic first_flush(true); rocksdb::SyncPoint::GetInstance()->SetCallBack( "WritableFileWriter::Flush:BeforeAppend", [&](void* arg) { if (first_flush.load()) { options.env->SleepForMicroseconds(3); first_flush.store(false); } }); std::atomic first_sync(true); rocksdb::SyncPoint::GetInstance()->SetCallBack( "WritableFileWriter::SyncInternal:0", [&](void* arg) { if (first_sync.load()) { options.env->SleepForMicroseconds(3); first_sync.store(false); } }); std::atomic first_range_sync(true); rocksdb::SyncPoint::GetInstance()->SetCallBack( "WritableFileWriter::RangeSync:0", [&](void* arg) { if (first_range_sync.load()) { options.env->SleepForMicroseconds(3); first_range_sync.store(false); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Compact(1, smallest_key, largest_key); ASSERT_TRUE(!stats_checker->verify_next_comp_io_stats()); ASSERT_TRUE(!first_prepare_write.load()); ASSERT_TRUE(!first_flush.load()); ASSERT_TRUE(!first_sync.load()); ASSERT_TRUE(!first_range_sync.load()); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 0U); } TEST_P(CompactionJobStatsTest, DeletionStatsTest) { Random rnd(301); uint64_t key_base = 100000l; // Note: key_base must be multiple of num_keys_per_L0_file int num_keys_per_L0_file = 20; const int kTestScale = 8; // make sure this is even const int kKeySize = 10; const int kValueSize = 100; double compression_ratio = 1.0; uint64_t key_interval = key_base / num_keys_per_L0_file; uint64_t largest_key_num = key_base * (kTestScale + 1) - key_interval; uint64_t cutoff_key_num = key_base * (kTestScale / 2 + 1) - key_interval; const std::string smallest_key = Key(key_base - 10, kKeySize); const std::string largest_key = Key(largest_key_num + 10, kKeySize); // Whenever a compaction completes, this listener will try to // verify whether the returned CompactionJobStats matches // what we expect. auto* stats_checker = new CompactionJobDeletionStatsChecker(); Options options; options.listeners.emplace_back(stats_checker); options.create_if_missing = true; options.max_background_flushes = 0; options.level0_file_num_compaction_trigger = kTestScale+1; options.num_levels = 3; options.compression = kNoCompression; options.max_bytes_for_level_multiplier = 2; options.max_subcompactions = max_subcompactions_; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Stage 1: Generate several L0 files and then send them to L2 by // using CompactRangeOptions and CompactRange(). These files will // have a strict subset of the keys from the full key-range for (uint64_t start_key = key_base; start_key <= key_base * kTestScale / 2; start_key += key_base) { MakeTableWithKeyValues( &rnd, start_key, start_key + key_base - 1, kKeySize, kValueSize, key_interval, compression_ratio, 1); } CompactRangeOptions cr_options; cr_options.change_level = true; cr_options.target_level = 2; db_->CompactRange(cr_options, handles_[1], nullptr, nullptr); ASSERT_GT(NumTableFilesAtLevel(2, 1), 0); // Stage 2: Generate files including keys from the entire key range for (uint64_t start_key = key_base; start_key <= key_base * kTestScale; start_key += key_base) { MakeTableWithKeyValues( &rnd, start_key, start_key + key_base - 1, kKeySize, kValueSize, key_interval, compression_ratio, 1); } // Send these L0 files to L1 TEST_Compact(0, 1, smallest_key, largest_key); ASSERT_GT(NumTableFilesAtLevel(1, 1), 0); // Add a new record and flush so now there is a L0 file // with a value too (not just deletions from the next step) ASSERT_OK(Put(1, Key(key_base-6, kKeySize), "test")); ASSERT_OK(Flush(1)); // Stage 3: Generate L0 files with some deletions so now // there are files with the same key range in L0, L1, and L2 int deletion_interval = 3; CompactionJobStats first_compaction_stats; SelectivelyDeleteKeys(key_base, largest_key_num, key_interval, deletion_interval, kKeySize, cutoff_key_num, &first_compaction_stats, 1); stats_checker->AddExpectedStats(first_compaction_stats); // Stage 4: Trigger compaction and verify the stats TEST_Compact(0, 1, smallest_key, largest_key); } namespace { int GetUniversalCompactionInputUnits(uint32_t num_flushes) { uint32_t compaction_input_units; for (compaction_input_units = 1; num_flushes >= compaction_input_units; compaction_input_units *= 2) { if ((num_flushes & compaction_input_units) != 0) { return compaction_input_units > 1 ? compaction_input_units : 0; } } return 0; } } // namespace TEST_P(CompactionJobStatsTest, UniversalCompactionTest) { Random rnd(301); uint64_t key_base = 100000000l; // Note: key_base must be multiple of num_keys_per_L0_file int num_keys_per_table = 100; const uint32_t kTestScale = 8; const int kKeySize = 10; const int kValueSize = 900; double compression_ratio = 1.0; uint64_t key_interval = key_base / num_keys_per_table; auto* stats_checker = new CompactionJobStatsChecker(); Options options; options.listeners.emplace_back(stats_checker); options.create_if_missing = true; options.num_levels = 3; options.compression = kNoCompression; options.level0_file_num_compaction_trigger = 2; options.target_file_size_base = num_keys_per_table * 1000; options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.size_ratio = 1; options.compaction_options_universal.max_size_amplification_percent = 1000; options.max_subcompactions = max_subcompactions_; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Generates the expected CompactionJobStats for each compaction for (uint32_t num_flushes = 2; num_flushes <= kTestScale; num_flushes++) { // Here we treat one newly flushed file as an unit. // // For example, if a newly flushed file is 100k, and a compaction has // 4 input units, then this compaction inputs 400k. uint32_t num_input_units = GetUniversalCompactionInputUnits(num_flushes); if (num_input_units == 0) { continue; } // The following statement determines the expected smallest key // based on whether it is a full compaction. A full compaction only // happens when the number of flushes equals to the number of compaction // input runs. uint64_t smallest_key = (num_flushes == num_input_units) ? key_base : key_base * (num_flushes - 1); stats_checker->AddExpectedStats( NewManualCompactionJobStats( Key(smallest_key, 10), Key(smallest_key + key_base * num_input_units - key_interval, 10), num_input_units, num_input_units > 2 ? num_input_units / 2 : 0, num_keys_per_table * num_input_units, kKeySize, kValueSize, num_input_units, num_keys_per_table * num_input_units, 1.0, 0, false)); } ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 4U); for (uint64_t start_key = key_base; start_key <= key_base * kTestScale; start_key += key_base) { MakeTableWithKeyValues( &rnd, start_key, start_key + key_base - 1, kKeySize, kValueSize, key_interval, compression_ratio, 1); reinterpret_cast(db_)->TEST_WaitForCompact(); } ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 0U); } INSTANTIATE_TEST_CASE_P(CompactionJobStatsTest, CompactionJobStatsTest, ::testing::Values(1, 4)); } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED, not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE #else int main(int argc, char** argv) { return 0; } #endif // !defined(IOS_CROSS_COMPILE) rocksdb-4.5.1/db/compaction_job_test.cc000066400000000000000000001045221267530634000201030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include #include #include #include #include "db/compaction_job.h" #include "db/column_family.h" #include "db/version_set.h" #include "db/writebuffer.h" #include "rocksdb/cache.h" #include "rocksdb/db.h" #include "rocksdb/options.h" #include "table/mock_table.h" #include "util/file_reader_writer.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" #include "utilities/merge_operators.h" namespace rocksdb { namespace { void VerifyInitializationOfCompactionJobStats( const CompactionJobStats& compaction_job_stats) { #if !defined(IOS_CROSS_COMPILE) ASSERT_EQ(compaction_job_stats.elapsed_micros, 0U); ASSERT_EQ(compaction_job_stats.num_input_records, 0U); ASSERT_EQ(compaction_job_stats.num_input_files, 0U); ASSERT_EQ(compaction_job_stats.num_input_files_at_output_level, 0U); ASSERT_EQ(compaction_job_stats.num_output_records, 0U); ASSERT_EQ(compaction_job_stats.num_output_files, 0U); ASSERT_EQ(compaction_job_stats.is_manual_compaction, true); ASSERT_EQ(compaction_job_stats.total_input_bytes, 0U); ASSERT_EQ(compaction_job_stats.total_output_bytes, 0U); ASSERT_EQ(compaction_job_stats.total_input_raw_key_bytes, 0U); ASSERT_EQ(compaction_job_stats.total_input_raw_value_bytes, 0U); ASSERT_EQ(compaction_job_stats.smallest_output_key_prefix[0], 0); ASSERT_EQ(compaction_job_stats.largest_output_key_prefix[0], 0); ASSERT_EQ(compaction_job_stats.num_records_replaced, 0U); ASSERT_EQ(compaction_job_stats.num_input_deletion_records, 0U); ASSERT_EQ(compaction_job_stats.num_expired_deletion_records, 0U); ASSERT_EQ(compaction_job_stats.num_corrupt_keys, 0U); #endif // !defined(IOS_CROSS_COMPILE) } } // namespace // TODO(icanadi) Make it simpler once we mock out VersionSet class CompactionJobTest : public testing::Test { public: CompactionJobTest() : env_(Env::Default()), dbname_(test::TmpDir() + "/compaction_job_test"), mutable_cf_options_(Options(), ImmutableCFOptions(Options())), table_cache_(NewLRUCache(50000, 16)), write_buffer_(db_options_.db_write_buffer_size), versions_(new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(), &write_buffer_, &write_controller_)), shutting_down_(false), mock_table_factory_(new mock::MockTableFactory()) { EXPECT_OK(env_->CreateDirIfMissing(dbname_)); db_options_.db_paths.emplace_back(dbname_, std::numeric_limits::max()); } std::string GenerateFileName(uint64_t file_number) { FileMetaData meta; std::vector db_paths; db_paths.emplace_back(dbname_, std::numeric_limits::max()); meta.fd = FileDescriptor(file_number, 0, 0); return TableFileName(db_paths, meta.fd.GetNumber(), meta.fd.GetPathId()); } std::string KeyStr(const std::string& user_key, const SequenceNumber seq_num, const ValueType t) { return InternalKey(user_key, seq_num, t).Encode().ToString(); } void AddMockFile(const stl_wrappers::KVMap& contents, int level = 0) { assert(contents.size() > 0); bool first_key = true; std::string smallest, largest; InternalKey smallest_key, largest_key; SequenceNumber smallest_seqno = kMaxSequenceNumber; SequenceNumber largest_seqno = 0; for (auto kv : contents) { ParsedInternalKey key; std::string skey; std::string value; std::tie(skey, value) = kv; ParseInternalKey(skey, &key); smallest_seqno = std::min(smallest_seqno, key.sequence); largest_seqno = std::max(largest_seqno, key.sequence); if (first_key || cfd_->user_comparator()->Compare(key.user_key, smallest) < 0) { smallest.assign(key.user_key.data(), key.user_key.size()); smallest_key.DecodeFrom(skey); } if (first_key || cfd_->user_comparator()->Compare(key.user_key, largest) > 0) { largest.assign(key.user_key.data(), key.user_key.size()); largest_key.DecodeFrom(skey); } first_key = false; } uint64_t file_number = versions_->NewFileNumber(); EXPECT_OK(mock_table_factory_->CreateMockTable( env_, GenerateFileName(file_number), std::move(contents))); VersionEdit edit; edit.AddFile(level, file_number, 0, 10, smallest_key, largest_key, smallest_seqno, largest_seqno, false); mutex_.Lock(); versions_->LogAndApply(versions_->GetColumnFamilySet()->GetDefault(), mutable_cf_options_, &edit, &mutex_); mutex_.Unlock(); } void SetLastSequence(const SequenceNumber sequence_number) { versions_->SetLastSequence(sequence_number + 1); } // returns expected result after compaction stl_wrappers::KVMap CreateTwoFiles(bool gen_corrupted_keys) { auto expected_results = mock::MakeMockFile(); const int kKeysPerFile = 10000; const int kCorruptKeysPerFile = 200; const int kMatchingKeys = kKeysPerFile / 2; SequenceNumber sequence_number = 0; auto corrupt_id = [&](int id) { return gen_corrupted_keys && id > 0 && id <= kCorruptKeysPerFile; }; for (int i = 0; i < 2; ++i) { auto contents = mock::MakeMockFile(); for (int k = 0; k < kKeysPerFile; ++k) { auto key = ToString(i * kMatchingKeys + k); auto value = ToString(i * kKeysPerFile + k); InternalKey internal_key(key, ++sequence_number, kTypeValue); // This is how the key will look like once it's written in bottommost // file InternalKey bottommost_internal_key( key, (key == "9999") ? sequence_number : 0, kTypeValue); if (corrupt_id(k)) { test::CorruptKeyType(&internal_key); test::CorruptKeyType(&bottommost_internal_key); } contents.insert({ internal_key.Encode().ToString(), value }); if (i == 1 || k < kMatchingKeys || corrupt_id(k - kMatchingKeys)) { expected_results.insert( { bottommost_internal_key.Encode().ToString(), value }); } } AddMockFile(contents); } SetLastSequence(sequence_number); return expected_results; } void NewDB() { VersionEdit new_db; new_db.SetLogNumber(0); new_db.SetNextFile(2); new_db.SetLastSequence(0); const std::string manifest = DescriptorFileName(dbname_, 1); unique_ptr file; Status s = env_->NewWritableFile( manifest, &file, env_->OptimizeForManifestWrite(env_options_)); ASSERT_OK(s); unique_ptr file_writer( new WritableFileWriter(std::move(file), env_options_)); { log::Writer log(std::move(file_writer), 0, false); std::string record; new_db.EncodeTo(&record); s = log.AddRecord(record); } ASSERT_OK(s); // Make "CURRENT" file that points to the new manifest file. s = SetCurrentFile(env_, dbname_, 1, nullptr); std::vector column_families; cf_options_.table_factory = mock_table_factory_; cf_options_.merge_operator = merge_op_; cf_options_.compaction_filter = compaction_filter_.get(); column_families.emplace_back(kDefaultColumnFamilyName, cf_options_); EXPECT_OK(versions_->Recover(column_families, false)); cfd_ = versions_->GetColumnFamilySet()->GetDefault(); } void RunCompaction( const std::vector>& input_files, const stl_wrappers::KVMap& expected_results, const std::vector& snapshots = {}, SequenceNumber earliest_write_conflict_snapshot = kMaxSequenceNumber) { auto cfd = versions_->GetColumnFamilySet()->GetDefault(); size_t num_input_files = 0; std::vector compaction_input_files; for (size_t level = 0; level < input_files.size(); level++) { auto level_files = input_files[level]; CompactionInputFiles compaction_level; compaction_level.level = static_cast(level); compaction_level.files.insert(compaction_level.files.end(), level_files.begin(), level_files.end()); compaction_input_files.push_back(compaction_level); num_input_files += level_files.size(); } Compaction compaction(cfd->current()->storage_info(), *cfd->GetLatestMutableCFOptions(), compaction_input_files, 1, 1024 * 1024, 10, 0, kNoCompression, {}, true); compaction.SetInputVersion(cfd->current()); LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, db_options_.info_log.get()); mutex_.Lock(); EventLogger event_logger(db_options_.info_log.get()); CompactionJob compaction_job( 0, &compaction, db_options_, env_options_, versions_.get(), &shutting_down_, &log_buffer, nullptr, nullptr, nullptr, snapshots, earliest_write_conflict_snapshot, table_cache_, &event_logger, false, false, dbname_, &compaction_job_stats_); VerifyInitializationOfCompactionJobStats(compaction_job_stats_); compaction_job.Prepare(); mutex_.Unlock(); Status s; s = compaction_job.Run(); ASSERT_OK(s); mutex_.Lock(); ASSERT_OK(compaction_job.Install(*cfd->GetLatestMutableCFOptions(), &mutex_)); mutex_.Unlock(); if (expected_results.size() == 0) { ASSERT_GE(compaction_job_stats_.elapsed_micros, 0U); ASSERT_EQ(compaction_job_stats_.num_input_files, num_input_files); ASSERT_EQ(compaction_job_stats_.num_output_files, 0U); } else { ASSERT_GE(compaction_job_stats_.elapsed_micros, 0U); ASSERT_EQ(compaction_job_stats_.num_input_files, num_input_files); ASSERT_EQ(compaction_job_stats_.num_output_files, 1U); mock_table_factory_->AssertLatestFile(expected_results); } } Env* env_; std::string dbname_; EnvOptions env_options_; MutableCFOptions mutable_cf_options_; std::shared_ptr table_cache_; WriteController write_controller_; DBOptions db_options_; ColumnFamilyOptions cf_options_; WriteBuffer write_buffer_; std::unique_ptr versions_; InstrumentedMutex mutex_; std::atomic shutting_down_; std::shared_ptr mock_table_factory_; CompactionJobStats compaction_job_stats_; ColumnFamilyData* cfd_; std::unique_ptr compaction_filter_; std::shared_ptr merge_op_; }; TEST_F(CompactionJobTest, Simple) { NewDB(); auto expected_results = CreateTwoFiles(false); auto cfd = versions_->GetColumnFamilySet()->GetDefault(); auto files = cfd->current()->storage_info()->LevelFiles(0); ASSERT_EQ(2U, files.size()); RunCompaction({ files }, expected_results); } TEST_F(CompactionJobTest, SimpleCorrupted) { NewDB(); auto expected_results = CreateTwoFiles(true); auto cfd = versions_->GetColumnFamilySet()->GetDefault(); auto files = cfd->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); ASSERT_EQ(compaction_job_stats_.num_corrupt_keys, 400U); } TEST_F(CompactionJobTest, SimpleDeletion) { NewDB(); auto file1 = mock::MakeMockFile({{KeyStr("c", 4U, kTypeDeletion), ""}, {KeyStr("c", 3U, kTypeValue), "val"}}); AddMockFile(file1); auto file2 = mock::MakeMockFile({{KeyStr("b", 2U, kTypeValue), "val"}, {KeyStr("b", 1U, kTypeValue), "val"}}); AddMockFile(file2); auto expected_results = mock::MakeMockFile({{KeyStr("b", 0U, kTypeValue), "val"}}); SetLastSequence(4U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } TEST_F(CompactionJobTest, SimpleOverwrite) { NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("a", 3U, kTypeValue), "val2"}, {KeyStr("b", 4U, kTypeValue), "val3"}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({{KeyStr("a", 1U, kTypeValue), "val"}, {KeyStr("b", 2U, kTypeValue), "val"}}); AddMockFile(file2); auto expected_results = mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), "val2"}, {KeyStr("b", 4U, kTypeValue), "val3"}}); SetLastSequence(4U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } TEST_F(CompactionJobTest, SimpleNonLastLevel) { NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("a", 5U, kTypeValue), "val2"}, {KeyStr("b", 6U, kTypeValue), "val3"}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({{KeyStr("a", 3U, kTypeValue), "val"}, {KeyStr("b", 4U, kTypeValue), "val"}}); AddMockFile(file2, 1); auto file3 = mock::MakeMockFile({{KeyStr("a", 1U, kTypeValue), "val"}, {KeyStr("b", 2U, kTypeValue), "val"}}); AddMockFile(file3, 2); // Because level 1 is not the last level, the sequence numbers of a and b // cannot be set to 0 auto expected_results = mock::MakeMockFile({{KeyStr("a", 5U, kTypeValue), "val2"}, {KeyStr("b", 6U, kTypeValue), "val3"}}); SetLastSequence(6U); auto lvl0_files = cfd_->current()->storage_info()->LevelFiles(0); auto lvl1_files = cfd_->current()->storage_info()->LevelFiles(1); RunCompaction({lvl0_files, lvl1_files}, expected_results); } TEST_F(CompactionJobTest, SimpleMerge) { merge_op_ = MergeOperators::CreateStringAppendOperator(); NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("a", 5U, kTypeMerge), "5"}, {KeyStr("a", 4U, kTypeMerge), "4"}, {KeyStr("a", 3U, kTypeValue), "3"}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile( {{KeyStr("b", 2U, kTypeMerge), "2"}, {KeyStr("b", 1U, kTypeValue), "1"}}); AddMockFile(file2); auto expected_results = mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), "3,4,5"}, {KeyStr("b", 2U, kTypeValue), "1,2"}}); SetLastSequence(5U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } TEST_F(CompactionJobTest, NonAssocMerge) { merge_op_ = MergeOperators::CreateStringAppendTESTOperator(); NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("a", 5U, kTypeMerge), "5"}, {KeyStr("a", 4U, kTypeMerge), "4"}, {KeyStr("a", 3U, kTypeMerge), "3"}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile( {{KeyStr("b", 2U, kTypeMerge), "2"}, {KeyStr("b", 1U, kTypeMerge), "1"}}); AddMockFile(file2); auto expected_results = mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), "3,4,5"}, {KeyStr("b", 2U, kTypeMerge), "2"}, {KeyStr("b", 1U, kTypeMerge), "1"}}); SetLastSequence(5U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } // Filters merge operands with value 10. TEST_F(CompactionJobTest, MergeOperandFilter) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); compaction_filter_.reset(new test::FilterNumber(10U)); NewDB(); auto file1 = mock::MakeMockFile( {{KeyStr("a", 5U, kTypeMerge), test::EncodeInt(5U)}, {KeyStr("a", 4U, kTypeMerge), test::EncodeInt(10U)}, // Filtered {KeyStr("a", 3U, kTypeMerge), test::EncodeInt(3U)}}); AddMockFile(file1); auto file2 = mock::MakeMockFile({ {KeyStr("b", 2U, kTypeMerge), test::EncodeInt(2U)}, {KeyStr("b", 1U, kTypeMerge), test::EncodeInt(10U)} // Filtered }); AddMockFile(file2); auto expected_results = mock::MakeMockFile({{KeyStr("a", 0U, kTypeValue), test::EncodeInt(8U)}, {KeyStr("b", 2U, kTypeMerge), test::EncodeInt(2U)}}); SetLastSequence(5U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } TEST_F(CompactionJobTest, FilterSomeMergeOperands) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); compaction_filter_.reset(new test::FilterNumber(10U)); NewDB(); auto file1 = mock::MakeMockFile( {{KeyStr("a", 5U, kTypeMerge), test::EncodeInt(5U)}, {KeyStr("a", 4U, kTypeMerge), test::EncodeInt(10U)}, // Filtered {KeyStr("a", 3U, kTypeValue), test::EncodeInt(5U)}, {KeyStr("d", 8U, kTypeMerge), test::EncodeInt(10U)}}); AddMockFile(file1); auto file2 = mock::MakeMockFile({{KeyStr("b", 2U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 1U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("c", 2U, kTypeMerge), test::EncodeInt(3U)}, {KeyStr("c", 1U, kTypeValue), test::EncodeInt(7U)}, {KeyStr("d", 1U, kTypeValue), test::EncodeInt(6U)}}); AddMockFile(file2); auto file3 = mock::MakeMockFile({{KeyStr("a", 1U, kTypeMerge), test::EncodeInt(3U)}}); AddMockFile(file3, 2); auto expected_results = mock::MakeMockFile({ {KeyStr("a", 5U, kTypeValue), test::EncodeInt(10U)}, {KeyStr("c", 2U, kTypeValue), test::EncodeInt(10U)}, {KeyStr("d", 1U, kTypeValue), test::EncodeInt(6U)} // b does not appear because the operands are filtered }); SetLastSequence(5U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } // Test where all operands/merge results are filtered out. TEST_F(CompactionJobTest, FilterAllMergeOperands) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); compaction_filter_.reset(new test::FilterNumber(10U)); NewDB(); auto file1 = mock::MakeMockFile({{KeyStr("a", 11U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("a", 10U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("a", 9U, kTypeMerge), test::EncodeInt(10U)}}); AddMockFile(file1); auto file2 = mock::MakeMockFile({{KeyStr("b", 8U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 7U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 6U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 5U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 4U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 3U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 2U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("c", 2U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("c", 1U, kTypeMerge), test::EncodeInt(10U)}}); AddMockFile(file2); auto file3 = mock::MakeMockFile({{KeyStr("a", 2U, kTypeMerge), test::EncodeInt(10U)}, {KeyStr("b", 1U, kTypeMerge), test::EncodeInt(10U)}}); AddMockFile(file3, 2); SetLastSequence(11U); auto files = cfd_->current()->storage_info()->LevelFiles(0); stl_wrappers::KVMap empty_map; RunCompaction({files}, empty_map); } TEST_F(CompactionJobTest, SimpleSingleDelete) { NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("a", 5U, kTypeDeletion), ""}, {KeyStr("b", 6U, kTypeSingleDeletion), ""}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({{KeyStr("a", 3U, kTypeValue), "val"}, {KeyStr("b", 4U, kTypeValue), "val"}}); AddMockFile(file2); auto file3 = mock::MakeMockFile({ {KeyStr("a", 1U, kTypeValue), "val"}, }); AddMockFile(file3, 2); auto expected_results = mock::MakeMockFile({{KeyStr("a", 5U, kTypeDeletion), ""}}); SetLastSequence(6U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } TEST_F(CompactionJobTest, SingleDeleteSnapshots) { NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("A", 12U, kTypeSingleDeletion), ""}, {KeyStr("a", 12U, kTypeSingleDeletion), ""}, {KeyStr("b", 21U, kTypeSingleDeletion), ""}, {KeyStr("c", 22U, kTypeSingleDeletion), ""}, {KeyStr("d", 9U, kTypeSingleDeletion), ""}, {KeyStr("f", 21U, kTypeSingleDeletion), ""}, {KeyStr("j", 11U, kTypeSingleDeletion), ""}, {KeyStr("j", 9U, kTypeSingleDeletion), ""}, {KeyStr("k", 12U, kTypeSingleDeletion), ""}, {KeyStr("k", 11U, kTypeSingleDeletion), ""}, {KeyStr("l", 3U, kTypeSingleDeletion), ""}, {KeyStr("l", 2U, kTypeSingleDeletion), ""}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({ {KeyStr("0", 2U, kTypeSingleDeletion), ""}, {KeyStr("a", 11U, kTypeValue), "val1"}, {KeyStr("b", 11U, kTypeValue), "val2"}, {KeyStr("c", 21U, kTypeValue), "val3"}, {KeyStr("d", 8U, kTypeValue), "val4"}, {KeyStr("e", 2U, kTypeSingleDeletion), ""}, {KeyStr("f", 1U, kTypeValue), "val1"}, {KeyStr("g", 11U, kTypeSingleDeletion), ""}, {KeyStr("h", 2U, kTypeSingleDeletion), ""}, {KeyStr("m", 12U, kTypeValue), "val1"}, {KeyStr("m", 11U, kTypeSingleDeletion), ""}, {KeyStr("m", 8U, kTypeValue), "val2"}, }); AddMockFile(file2); auto file3 = mock::MakeMockFile({ {KeyStr("A", 1U, kTypeValue), "val"}, {KeyStr("e", 1U, kTypeValue), "val"}, }); AddMockFile(file3, 2); auto expected_results = mock::MakeMockFile({ {KeyStr("A", 12U, kTypeSingleDeletion), ""}, {KeyStr("a", 12U, kTypeSingleDeletion), ""}, {KeyStr("a", 11U, kTypeValue), ""}, {KeyStr("b", 21U, kTypeSingleDeletion), ""}, {KeyStr("b", 11U, kTypeValue), "val2"}, {KeyStr("c", 22U, kTypeSingleDeletion), ""}, {KeyStr("c", 21U, kTypeValue), ""}, {KeyStr("e", 2U, kTypeSingleDeletion), ""}, {KeyStr("f", 21U, kTypeSingleDeletion), ""}, {KeyStr("f", 1U, kTypeValue), "val1"}, {KeyStr("g", 11U, kTypeSingleDeletion), ""}, {KeyStr("j", 11U, kTypeSingleDeletion), ""}, {KeyStr("k", 11U, kTypeSingleDeletion), ""}, {KeyStr("m", 12U, kTypeValue), "val1"}, {KeyStr("m", 11U, kTypeSingleDeletion), ""}, {KeyStr("m", 8U, kTypeValue), "val2"}, }); SetLastSequence(22U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results, {10U, 20U}, 10U); } TEST_F(CompactionJobTest, EarliestWriteConflictSnapshot) { NewDB(); // Test multiple snapshots where the earliest snapshot is not a // write-conflic-snapshot. auto file1 = mock::MakeMockFile({ {KeyStr("A", 24U, kTypeSingleDeletion), ""}, {KeyStr("A", 23U, kTypeValue), "val"}, {KeyStr("B", 24U, kTypeSingleDeletion), ""}, {KeyStr("B", 23U, kTypeValue), "val"}, {KeyStr("D", 24U, kTypeSingleDeletion), ""}, {KeyStr("G", 32U, kTypeSingleDeletion), ""}, {KeyStr("G", 31U, kTypeValue), "val"}, {KeyStr("G", 24U, kTypeSingleDeletion), ""}, {KeyStr("G", 23U, kTypeValue), "val2"}, {KeyStr("H", 31U, kTypeValue), "val"}, {KeyStr("H", 24U, kTypeSingleDeletion), ""}, {KeyStr("H", 23U, kTypeValue), "val"}, {KeyStr("I", 35U, kTypeSingleDeletion), ""}, {KeyStr("I", 34U, kTypeValue), "val2"}, {KeyStr("I", 33U, kTypeSingleDeletion), ""}, {KeyStr("I", 32U, kTypeValue), "val3"}, {KeyStr("I", 31U, kTypeSingleDeletion), ""}, {KeyStr("J", 34U, kTypeValue), "val"}, {KeyStr("J", 33U, kTypeSingleDeletion), ""}, {KeyStr("J", 25U, kTypeValue), "val2"}, {KeyStr("J", 24U, kTypeSingleDeletion), ""}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({ {KeyStr("A", 14U, kTypeSingleDeletion), ""}, {KeyStr("A", 13U, kTypeValue), "val2"}, {KeyStr("C", 14U, kTypeSingleDeletion), ""}, {KeyStr("C", 13U, kTypeValue), "val"}, {KeyStr("E", 12U, kTypeSingleDeletion), ""}, {KeyStr("F", 4U, kTypeSingleDeletion), ""}, {KeyStr("F", 3U, kTypeValue), "val"}, {KeyStr("G", 14U, kTypeSingleDeletion), ""}, {KeyStr("G", 13U, kTypeValue), "val3"}, {KeyStr("H", 14U, kTypeSingleDeletion), ""}, {KeyStr("H", 13U, kTypeValue), "val2"}, {KeyStr("I", 13U, kTypeValue), "val4"}, {KeyStr("I", 12U, kTypeSingleDeletion), ""}, {KeyStr("I", 11U, kTypeValue), "val5"}, {KeyStr("J", 15U, kTypeValue), "val3"}, {KeyStr("J", 14U, kTypeSingleDeletion), ""}, }); AddMockFile(file2); auto expected_results = mock::MakeMockFile({ {KeyStr("A", 24U, kTypeSingleDeletion), ""}, {KeyStr("A", 23U, kTypeValue), ""}, {KeyStr("B", 24U, kTypeSingleDeletion), ""}, {KeyStr("B", 23U, kTypeValue), ""}, {KeyStr("D", 24U, kTypeSingleDeletion), ""}, {KeyStr("E", 12U, kTypeSingleDeletion), ""}, {KeyStr("G", 32U, kTypeSingleDeletion), ""}, {KeyStr("G", 31U, kTypeValue), ""}, {KeyStr("H", 31U, kTypeValue), "val"}, {KeyStr("I", 35U, kTypeSingleDeletion), ""}, {KeyStr("I", 34U, kTypeValue), ""}, {KeyStr("I", 31U, kTypeSingleDeletion), ""}, {KeyStr("I", 13U, kTypeValue), "val4"}, {KeyStr("J", 34U, kTypeValue), "val"}, {KeyStr("J", 33U, kTypeSingleDeletion), ""}, {KeyStr("J", 25U, kTypeValue), "val2"}, {KeyStr("J", 24U, kTypeSingleDeletion), ""}, {KeyStr("J", 15U, kTypeValue), "val3"}, {KeyStr("J", 14U, kTypeSingleDeletion), ""}, }); SetLastSequence(24U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results, {10U, 20U, 30U}, 20U); } TEST_F(CompactionJobTest, SingleDeleteZeroSeq) { NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("A", 10U, kTypeSingleDeletion), ""}, {KeyStr("dummy", 5U, kTypeValue), "val2"}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({ {KeyStr("A", 0U, kTypeValue), "val"}, }); AddMockFile(file2); auto expected_results = mock::MakeMockFile({ {KeyStr("dummy", 5U, kTypeValue), "val2"}, }); SetLastSequence(22U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results, {}); } TEST_F(CompactionJobTest, MultiSingleDelete) { // Tests three scenarios involving multiple single delete/put pairs: // // A: Put Snapshot SDel Put SDel -> Put Snapshot SDel // B: Snapshot Put SDel Put SDel Snapshot -> Snapshot SDel Snapshot // C: SDel Put SDel Snapshot Put -> Snapshot Put // D: (Put) SDel Snapshot Put SDel -> (Put) SDel Snapshot SDel // E: Put SDel Snapshot Put SDel -> Snapshot SDel // F: Put SDel Put Sdel Snapshot -> removed // G: Snapshot SDel Put SDel Put -> Snapshot Put SDel // H: (Put) Put SDel Put Sdel Snapshot -> Removed // I: (Put) Snapshot Put SDel Put SDel -> SDel // J: Put Put SDel Put SDel SDel Snapshot Put Put SDel SDel Put // -> Snapshot Put // K: SDel SDel Put SDel Put Put Snapshot SDel Put SDel SDel Put SDel // -> Snapshot Put Snapshot SDel // L: SDel Put Del Put SDel Snapshot Del Put Del SDel Put SDel // -> Snapshot SDel // M: (Put) SDel Put Del Put SDel Snapshot Put Del SDel Put SDel Del // -> SDel Snapshot Del NewDB(); auto file1 = mock::MakeMockFile({ {KeyStr("A", 14U, kTypeSingleDeletion), ""}, {KeyStr("A", 13U, kTypeValue), "val5"}, {KeyStr("A", 12U, kTypeSingleDeletion), ""}, {KeyStr("B", 14U, kTypeSingleDeletion), ""}, {KeyStr("B", 13U, kTypeValue), "val2"}, {KeyStr("C", 14U, kTypeValue), "val3"}, {KeyStr("D", 12U, kTypeSingleDeletion), ""}, {KeyStr("D", 11U, kTypeValue), "val4"}, {KeyStr("G", 15U, kTypeValue), "val"}, {KeyStr("G", 14U, kTypeSingleDeletion), ""}, {KeyStr("G", 13U, kTypeValue), "val"}, {KeyStr("I", 14U, kTypeSingleDeletion), ""}, {KeyStr("I", 13U, kTypeValue), "val"}, {KeyStr("J", 15U, kTypeValue), "val"}, {KeyStr("J", 14U, kTypeSingleDeletion), ""}, {KeyStr("J", 13U, kTypeSingleDeletion), ""}, {KeyStr("J", 12U, kTypeValue), "val"}, {KeyStr("J", 11U, kTypeValue), "val"}, {KeyStr("K", 16U, kTypeSingleDeletion), ""}, {KeyStr("K", 15U, kTypeValue), "val1"}, {KeyStr("K", 14U, kTypeSingleDeletion), ""}, {KeyStr("K", 13U, kTypeSingleDeletion), ""}, {KeyStr("K", 12U, kTypeValue), "val2"}, {KeyStr("K", 11U, kTypeSingleDeletion), ""}, {KeyStr("L", 16U, kTypeSingleDeletion), ""}, {KeyStr("L", 15U, kTypeValue), "val"}, {KeyStr("L", 14U, kTypeSingleDeletion), ""}, {KeyStr("L", 13U, kTypeDeletion), ""}, {KeyStr("L", 12U, kTypeValue), "val"}, {KeyStr("L", 11U, kTypeDeletion), ""}, {KeyStr("M", 16U, kTypeDeletion), ""}, {KeyStr("M", 15U, kTypeSingleDeletion), ""}, {KeyStr("M", 14U, kTypeValue), "val"}, {KeyStr("M", 13U, kTypeSingleDeletion), ""}, {KeyStr("M", 12U, kTypeDeletion), ""}, {KeyStr("M", 11U, kTypeValue), "val"}, }); AddMockFile(file1); auto file2 = mock::MakeMockFile({ {KeyStr("A", 10U, kTypeValue), "val"}, {KeyStr("B", 12U, kTypeSingleDeletion), ""}, {KeyStr("B", 11U, kTypeValue), "val2"}, {KeyStr("C", 10U, kTypeSingleDeletion), ""}, {KeyStr("C", 9U, kTypeValue), "val6"}, {KeyStr("C", 8U, kTypeSingleDeletion), ""}, {KeyStr("D", 10U, kTypeSingleDeletion), ""}, {KeyStr("E", 12U, kTypeSingleDeletion), ""}, {KeyStr("E", 11U, kTypeValue), "val"}, {KeyStr("E", 5U, kTypeSingleDeletion), ""}, {KeyStr("E", 4U, kTypeValue), "val"}, {KeyStr("F", 6U, kTypeSingleDeletion), ""}, {KeyStr("F", 5U, kTypeValue), "val"}, {KeyStr("F", 4U, kTypeSingleDeletion), ""}, {KeyStr("F", 3U, kTypeValue), "val"}, {KeyStr("G", 12U, kTypeSingleDeletion), ""}, {KeyStr("H", 6U, kTypeSingleDeletion), ""}, {KeyStr("H", 5U, kTypeValue), "val"}, {KeyStr("H", 4U, kTypeSingleDeletion), ""}, {KeyStr("H", 3U, kTypeValue), "val"}, {KeyStr("I", 12U, kTypeSingleDeletion), ""}, {KeyStr("I", 11U, kTypeValue), "val"}, {KeyStr("J", 6U, kTypeSingleDeletion), ""}, {KeyStr("J", 5U, kTypeSingleDeletion), ""}, {KeyStr("J", 4U, kTypeValue), "val"}, {KeyStr("J", 3U, kTypeSingleDeletion), ""}, {KeyStr("J", 2U, kTypeValue), "val"}, {KeyStr("K", 8U, kTypeValue), "val3"}, {KeyStr("K", 7U, kTypeValue), "val4"}, {KeyStr("K", 6U, kTypeSingleDeletion), ""}, {KeyStr("K", 5U, kTypeValue), "val5"}, {KeyStr("K", 2U, kTypeSingleDeletion), ""}, {KeyStr("K", 1U, kTypeSingleDeletion), ""}, {KeyStr("L", 5U, kTypeSingleDeletion), ""}, {KeyStr("L", 4U, kTypeValue), "val"}, {KeyStr("L", 3U, kTypeDeletion), ""}, {KeyStr("L", 2U, kTypeValue), "val"}, {KeyStr("L", 1U, kTypeSingleDeletion), ""}, {KeyStr("M", 10U, kTypeSingleDeletion), ""}, {KeyStr("M", 7U, kTypeValue), "val"}, {KeyStr("M", 5U, kTypeDeletion), ""}, {KeyStr("M", 4U, kTypeValue), "val"}, {KeyStr("M", 3U, kTypeSingleDeletion), ""}, }); AddMockFile(file2); auto file3 = mock::MakeMockFile({ {KeyStr("D", 1U, kTypeValue), "val"}, {KeyStr("H", 1U, kTypeValue), "val"}, {KeyStr("I", 2U, kTypeValue), "val"}, }); AddMockFile(file3, 2); auto file4 = mock::MakeMockFile({ {KeyStr("M", 1U, kTypeValue), "val"}, }); AddMockFile(file4, 2); auto expected_results = mock::MakeMockFile({{KeyStr("A", 14U, kTypeSingleDeletion), ""}, {KeyStr("A", 13U, kTypeValue), ""}, {KeyStr("A", 12U, kTypeSingleDeletion), ""}, {KeyStr("A", 10U, kTypeValue), "val"}, {KeyStr("B", 14U, kTypeSingleDeletion), ""}, {KeyStr("B", 13U, kTypeValue), ""}, {KeyStr("C", 14U, kTypeValue), "val3"}, {KeyStr("D", 12U, kTypeSingleDeletion), ""}, {KeyStr("D", 11U, kTypeValue), ""}, {KeyStr("D", 10U, kTypeSingleDeletion), ""}, {KeyStr("E", 12U, kTypeSingleDeletion), ""}, {KeyStr("E", 11U, kTypeValue), ""}, {KeyStr("G", 15U, kTypeValue), "val"}, {KeyStr("G", 12U, kTypeSingleDeletion), ""}, {KeyStr("I", 14U, kTypeSingleDeletion), ""}, {KeyStr("I", 13U, kTypeValue), ""}, {KeyStr("J", 15U, kTypeValue), "val"}, {KeyStr("K", 16U, kTypeSingleDeletion), ""}, {KeyStr("K", 15U, kTypeValue), ""}, {KeyStr("K", 11U, kTypeSingleDeletion), ""}, {KeyStr("K", 8U, kTypeValue), "val3"}, {KeyStr("L", 16U, kTypeSingleDeletion), ""}, {KeyStr("L", 15U, kTypeValue), ""}, {KeyStr("M", 16U, kTypeDeletion), ""}, {KeyStr("M", 3U, kTypeSingleDeletion), ""}}); SetLastSequence(22U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results, {10U}, 10U); } // This test documents the behavior where a corrupt key follows a deletion or a // single deletion and the (single) deletion gets removed while the corrupt key // gets written out. TODO(noetzli): We probably want a better way to treat // corrupt keys. TEST_F(CompactionJobTest, CorruptionAfterDeletion) { NewDB(); auto file1 = mock::MakeMockFile({{test::KeyStr("A", 6U, kTypeValue), "val3"}, {test::KeyStr("a", 5U, kTypeDeletion), ""}, {test::KeyStr("a", 4U, kTypeValue, true), "val"}}); AddMockFile(file1); auto file2 = mock::MakeMockFile({{test::KeyStr("b", 3U, kTypeSingleDeletion), ""}, {test::KeyStr("b", 2U, kTypeValue, true), "val"}, {test::KeyStr("c", 1U, kTypeValue), "val2"}}); AddMockFile(file2); auto expected_results = mock::MakeMockFile({{test::KeyStr("A", 0U, kTypeValue), "val3"}, {test::KeyStr("a", 0U, kTypeValue, true), "val"}, {test::KeyStr("b", 0U, kTypeValue, true), "val"}, {test::KeyStr("c", 1U, kTypeValue), "val2"}}); SetLastSequence(6U); auto files = cfd_->current()->storage_info()->LevelFiles(0); RunCompaction({files}, expected_results); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as CompactionJobStats is not supported in ROCKSDB_LITE\n"); return 0; } #endif // ROCKSDB_LITE rocksdb-4.5.1/db/compaction_picker.cc000066400000000000000000002041721267530634000175510ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/compaction_picker.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include "db/column_family.h" #include "db/filename.h" #include "util/log_buffer.h" #include "util/random.h" #include "util/statistics.h" #include "util/string_util.h" #include "util/sync_point.h" namespace rocksdb { namespace { uint64_t TotalCompensatedFileSize(const std::vector& files) { uint64_t sum = 0; for (size_t i = 0; i < files.size() && files[i]; i++) { sum += files[i]->compensated_file_size; } return sum; } // Universal compaction is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE // Used in universal compaction when trivial move is enabled. // This structure is used for the construction of min heap // that contains the file meta data, the level of the file // and the index of the file in that level struct InputFileInfo { InputFileInfo() : f(nullptr) {} FileMetaData* f; size_t level; size_t index; }; // Used in universal compaction when trivial move is enabled. // This comparator is used for the construction of min heap // based on the smallest key of the file. struct UserKeyComparator { explicit UserKeyComparator(const Comparator* ucmp) { ucmp_ = ucmp; } bool operator()(InputFileInfo i1, InputFileInfo i2) const { return (ucmp_->Compare(i1.f->smallest.user_key(), i2.f->smallest.user_key()) > 0); } private: const Comparator* ucmp_; }; typedef std::priority_queue, UserKeyComparator> SmallestKeyHeap; // This function creates the heap that is used to find if the files are // overlapping during universal compaction when the allow_trivial_move // is set. SmallestKeyHeap create_level_heap(Compaction* c, const Comparator* ucmp) { SmallestKeyHeap smallest_key_priority_q = SmallestKeyHeap(UserKeyComparator(ucmp)); InputFileInfo input_file; for (size_t l = 0; l < c->num_input_levels(); l++) { if (c->num_input_files(l) != 0) { if (l == 0 && c->start_level() == 0) { for (size_t i = 0; i < c->num_input_files(0); i++) { input_file.f = c->input(0, i); input_file.level = 0; input_file.index = i; smallest_key_priority_q.push(std::move(input_file)); } } else { input_file.f = c->input(l, 0); input_file.level = l; input_file.index = 0; smallest_key_priority_q.push(std::move(input_file)); } } } return smallest_key_priority_q; } #endif // !ROCKSDB_LITE } // anonymous namespace // Determine compression type, based on user options, level of the output // file and whether compression is disabled. // If enable_compression is false, then compression is always disabled no // matter what the values of the other two parameters are. // Otherwise, the compression type is determined based on options and level. CompressionType GetCompressionType(const ImmutableCFOptions& ioptions, int level, int base_level, const bool enable_compression) { if (!enable_compression) { // disable compression return kNoCompression; } // If the use has specified a different compression level for each level, // then pick the compression for that level. if (!ioptions.compression_per_level.empty()) { assert(level == 0 || level >= base_level); int idx = (level == 0) ? 0 : level - base_level + 1; const int n = static_cast(ioptions.compression_per_level.size()) - 1; // It is possible for level_ to be -1; in that case, we use level // 0's compression. This occurs mostly in backwards compatibility // situations when the builder doesn't know what level the file // belongs to. Likewise, if level is beyond the end of the // specified compression levels, use the last value. return ioptions.compression_per_level[std::max(0, std::min(idx, n))]; } else { return ioptions.compression; } } CompactionPicker::CompactionPicker(const ImmutableCFOptions& ioptions, const InternalKeyComparator* icmp) : ioptions_(ioptions), icmp_(icmp) {} CompactionPicker::~CompactionPicker() {} // Delete this compaction from the list of running compactions. void CompactionPicker::ReleaseCompactionFiles(Compaction* c, Status status) { if (c->start_level() == 0) { level0_compactions_in_progress_.erase(c); } if (!status.ok()) { c->ResetNextCompactionIndex(); } } void CompactionPicker::GetRange(const CompactionInputFiles& inputs, InternalKey* smallest, InternalKey* largest) { const int level = inputs.level; assert(!inputs.empty()); smallest->Clear(); largest->Clear(); if (level == 0) { for (size_t i = 0; i < inputs.size(); i++) { FileMetaData* f = inputs[i]; if (i == 0) { *smallest = f->smallest; *largest = f->largest; } else { if (icmp_->Compare(f->smallest, *smallest) < 0) { *smallest = f->smallest; } if (icmp_->Compare(f->largest, *largest) > 0) { *largest = f->largest; } } } } else { *smallest = inputs[0]->smallest; *largest = inputs[inputs.size() - 1]->largest; } } void CompactionPicker::GetRange(const CompactionInputFiles& inputs1, const CompactionInputFiles& inputs2, InternalKey* smallest, InternalKey* largest) { assert(!inputs1.empty() || !inputs2.empty()); if (inputs1.empty()) { GetRange(inputs2, smallest, largest); } else if (inputs2.empty()) { GetRange(inputs1, smallest, largest); } else { InternalKey smallest1, smallest2, largest1, largest2; GetRange(inputs1, &smallest1, &largest1); GetRange(inputs2, &smallest2, &largest2); *smallest = icmp_->Compare(smallest1, smallest2) < 0 ? smallest1 : smallest2; *largest = icmp_->Compare(largest1, largest2) < 0 ? largest2 : largest1; } } bool CompactionPicker::ExpandWhileOverlapping(const std::string& cf_name, VersionStorageInfo* vstorage, CompactionInputFiles* inputs) { // This isn't good compaction assert(!inputs->empty()); const int level = inputs->level; // GetOverlappingInputs will always do the right thing for level-0. // So we don't need to do any expansion if level == 0. if (level == 0) { return true; } InternalKey smallest, largest; // Keep expanding inputs until we are sure that there is a "clean cut" // boundary between the files in input and the surrounding files. // This will ensure that no parts of a key are lost during compaction. int hint_index = -1; size_t old_size; do { old_size = inputs->size(); GetRange(*inputs, &smallest, &largest); inputs->clear(); vstorage->GetOverlappingInputs(level, &smallest, &largest, &inputs->files, hint_index, &hint_index); } while (inputs->size() > old_size); // we started off with inputs non-empty and the previous loop only grew // inputs. thus, inputs should be non-empty here assert(!inputs->empty()); // If, after the expansion, there are files that are already under // compaction, then we must drop/cancel this compaction. if (FilesInCompaction(inputs->files)) { Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log, "[%s] ExpandWhileOverlapping() failure because some of the necessary" " compaction input files are currently being compacted.", cf_name.c_str()); return false; } return true; } // Returns true if any one of specified files are being compacted bool CompactionPicker::FilesInCompaction( const std::vector& files) { for (size_t i = 0; i < files.size(); i++) { if (files[i]->being_compacted) { return true; } } return false; } Compaction* CompactionPicker::FormCompaction( const CompactionOptions& compact_options, const std::vector& input_files, int output_level, VersionStorageInfo* vstorage, const MutableCFOptions& mutable_cf_options, uint32_t output_path_id) { uint64_t max_grandparent_overlap_bytes = output_level + 1 < vstorage->num_levels() ? mutable_cf_options.MaxGrandParentOverlapBytes(output_level + 1) : std::numeric_limits::max(); assert(input_files.size()); // TODO(rven ): we might be able to run concurrent level 0 compaction // if the key ranges of the two compactions do not overlap, but for now // we do not allow it. if ((input_files[0].level == 0) && !level0_compactions_in_progress_.empty()) { return nullptr; } auto c = new Compaction( vstorage, mutable_cf_options, input_files, output_level, compact_options.output_file_size_limit, max_grandparent_overlap_bytes, output_path_id, compact_options.compression, /* grandparents */ {}, true); // If it's level 0 compaction, make sure we don't execute any other level 0 // compactions in parallel if ((c != nullptr) && (input_files[0].level == 0)) { level0_compactions_in_progress_.insert(c); } return c; } Status CompactionPicker::GetCompactionInputsFromFileNumbers( std::vector* input_files, std::unordered_set* input_set, const VersionStorageInfo* vstorage, const CompactionOptions& compact_options) const { if (input_set->size() == 0U) { return Status::InvalidArgument( "Compaction must include at least one file."); } assert(input_files); std::vector matched_input_files; matched_input_files.resize(vstorage->num_levels()); int first_non_empty_level = -1; int last_non_empty_level = -1; // TODO(yhchiang): use a lazy-initialized mapping from // file_number to FileMetaData in Version. for (int level = 0; level < vstorage->num_levels(); ++level) { for (auto file : vstorage->LevelFiles(level)) { auto iter = input_set->find(file->fd.GetNumber()); if (iter != input_set->end()) { matched_input_files[level].files.push_back(file); input_set->erase(iter); last_non_empty_level = level; if (first_non_empty_level == -1) { first_non_empty_level = level; } } } } if (!input_set->empty()) { std::string message( "Cannot find matched SST files for the following file numbers:"); for (auto fn : *input_set) { message += " "; message += ToString(fn); } return Status::InvalidArgument(message); } for (int level = first_non_empty_level; level <= last_non_empty_level; ++level) { matched_input_files[level].level = level; input_files->emplace_back(std::move(matched_input_files[level])); } return Status::OK(); } // Returns true if any one of the parent files are being compacted bool CompactionPicker::RangeInCompaction(VersionStorageInfo* vstorage, const InternalKey* smallest, const InternalKey* largest, int level, int* level_index) { std::vector inputs; assert(level < NumberLevels()); vstorage->GetOverlappingInputs(level, smallest, largest, &inputs, *level_index, level_index); return FilesInCompaction(inputs); } // Populates the set of inputs of all other levels that overlap with the // start level. // Now we assume all levels except start level and output level are empty. // Will also attempt to expand "start level" if that doesn't expand // "output level" or cause "level" to include a file for compaction that has an // overlapping user-key with another file. // REQUIRES: input_level and output_level are different // REQUIRES: inputs->empty() == false // Returns false if files on parent level are currently in compaction, which // means that we can't compact them bool CompactionPicker::SetupOtherInputs( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, CompactionInputFiles* inputs, CompactionInputFiles* output_level_inputs, int* parent_index, int base_index) { assert(!inputs->empty()); assert(output_level_inputs->empty()); const int input_level = inputs->level; const int output_level = output_level_inputs->level; assert(input_level != output_level); // For now, we only support merging two levels, start level and output level. // We need to assert other levels are empty. for (int l = input_level + 1; l < output_level; l++) { assert(vstorage->NumLevelFiles(l) == 0); } InternalKey smallest, largest; // Get the range one last time. GetRange(*inputs, &smallest, &largest); // Populate the set of next-level files (inputs_GetOutputLevelInputs()) to // include in compaction vstorage->GetOverlappingInputs(output_level, &smallest, &largest, &output_level_inputs->files, *parent_index, parent_index); if (FilesInCompaction(output_level_inputs->files)) { return false; } // See if we can further grow the number of inputs in "level" without // changing the number of "level+1" files we pick up. We also choose NOT // to expand if this would cause "level" to include some entries for some // user key, while excluding other entries for the same user key. This // can happen when one user key spans multiple files. if (!output_level_inputs->empty()) { CompactionInputFiles expanded0; expanded0.level = input_level; // Get entire range covered by compaction InternalKey all_start, all_limit; GetRange(*inputs, *output_level_inputs, &all_start, &all_limit); vstorage->GetOverlappingInputs(input_level, &all_start, &all_limit, &expanded0.files, base_index, nullptr); const uint64_t inputs0_size = TotalCompensatedFileSize(inputs->files); const uint64_t inputs1_size = TotalCompensatedFileSize(output_level_inputs->files); const uint64_t expanded0_size = TotalCompensatedFileSize(expanded0.files); uint64_t limit = mutable_cf_options.ExpandedCompactionByteSizeLimit(input_level); if (expanded0.size() > inputs->size() && inputs1_size + expanded0_size < limit && !FilesInCompaction(expanded0.files) && !vstorage->HasOverlappingUserKey(&expanded0.files, input_level)) { InternalKey new_start, new_limit; GetRange(expanded0, &new_start, &new_limit); std::vector expanded1; vstorage->GetOverlappingInputs(output_level, &new_start, &new_limit, &expanded1, *parent_index, parent_index); if (expanded1.size() == output_level_inputs->size() && !FilesInCompaction(expanded1)) { Log(InfoLogLevel::INFO_LEVEL, ioptions_.info_log, "[%s] Expanding@%d %" ROCKSDB_PRIszt "+%" ROCKSDB_PRIszt "(%" PRIu64 "+%" PRIu64 " bytes) to %" ROCKSDB_PRIszt "+%" ROCKSDB_PRIszt " (%" PRIu64 "+%" PRIu64 "bytes)\n", cf_name.c_str(), input_level, inputs->size(), output_level_inputs->size(), inputs0_size, inputs1_size, expanded0.size(), expanded1.size(), expanded0_size, inputs1_size); smallest = new_start; largest = new_limit; inputs->files = expanded0.files; output_level_inputs->files = expanded1; } } } return true; } void CompactionPicker::GetGrandparents( VersionStorageInfo* vstorage, const CompactionInputFiles& inputs, const CompactionInputFiles& output_level_inputs, std::vector* grandparents) { InternalKey start, limit; GetRange(inputs, output_level_inputs, &start, &limit); // Compute the set of grandparent files that overlap this compaction // (parent == level+1; grandparent == level+2) if (output_level_inputs.level + 1 < NumberLevels()) { vstorage->GetOverlappingInputs(output_level_inputs.level + 1, &start, &limit, grandparents); } } Compaction* CompactionPicker::CompactRange( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict) { // CompactionPickerFIFO has its own implementation of compact range assert(ioptions_.compaction_style != kCompactionStyleFIFO); if (input_level == ColumnFamilyData::kCompactAllLevels) { assert(ioptions_.compaction_style == kCompactionStyleUniversal); // Universal compaction with more than one level always compacts all the // files together to the last level. assert(vstorage->num_levels() > 1); // DBImpl::CompactRange() set output level to be the last level assert(output_level == vstorage->num_levels() - 1); // DBImpl::RunManualCompaction will make full range for universal compaction assert(begin == nullptr); assert(end == nullptr); *compaction_end = nullptr; int start_level = 0; for (; start_level < vstorage->num_levels() && vstorage->NumLevelFiles(start_level) == 0; start_level++) { } if (start_level == vstorage->num_levels()) { return nullptr; } if ((start_level == 0) && (!level0_compactions_in_progress_.empty())) { *manual_conflict = true; // Only one level 0 compaction allowed return nullptr; } std::vector inputs(vstorage->num_levels() - start_level); for (int level = start_level; level < vstorage->num_levels(); level++) { inputs[level - start_level].level = level; auto& files = inputs[level - start_level].files; for (FileMetaData* f : vstorage->LevelFiles(level)) { files.push_back(f); } if (FilesInCompaction(files)) { *manual_conflict = true; return nullptr; } } Compaction* c = new Compaction( vstorage, mutable_cf_options, std::move(inputs), output_level, mutable_cf_options.MaxFileSizeForLevel(output_level), /* max_grandparent_overlap_bytes */ LLONG_MAX, output_path_id, GetCompressionType(ioptions_, output_level, 1), /* grandparents */ {}, /* is manual */ true); if (start_level == 0) { level0_compactions_in_progress_.insert(c); } return c; } CompactionInputFiles inputs; inputs.level = input_level; bool covering_the_whole_range = true; // All files are 'overlapping' in universal style compaction. // We have to compact the entire range in one shot. if (ioptions_.compaction_style == kCompactionStyleUniversal) { begin = nullptr; end = nullptr; } vstorage->GetOverlappingInputs(input_level, begin, end, &inputs.files); if (inputs.empty()) { return nullptr; } if ((input_level == 0) && (!level0_compactions_in_progress_.empty())) { // Only one level 0 compaction allowed TEST_SYNC_POINT("CompactionPicker::CompactRange:Conflict"); *manual_conflict = true; return nullptr; } // Avoid compacting too much in one shot in case the range is large. // But we cannot do this for level-0 since level-0 files can overlap // and we must not pick one file and drop another older file if the // two files overlap. if (input_level > 0) { const uint64_t limit = mutable_cf_options.MaxFileSizeForLevel(input_level) * mutable_cf_options.source_compaction_factor; uint64_t total = 0; for (size_t i = 0; i + 1 < inputs.size(); ++i) { uint64_t s = inputs[i]->compensated_file_size; total += s; if (total >= limit) { **compaction_end = inputs[i + 1]->smallest; covering_the_whole_range = false; inputs.files.resize(i + 1); break; } } } assert(output_path_id < static_cast(ioptions_.db_paths.size())); if (ExpandWhileOverlapping(cf_name, vstorage, &inputs) == false) { // manual compaction is now multi-threaded, so it can // happen that ExpandWhileOverlapping fails // we handle it higher in RunManualCompaction *manual_conflict = true; return nullptr; } if (covering_the_whole_range) { *compaction_end = nullptr; } CompactionInputFiles output_level_inputs; if (output_level == ColumnFamilyData::kCompactToBaseLevel) { assert(input_level == 0); output_level = vstorage->base_level(); assert(output_level > 0); } output_level_inputs.level = output_level; if (input_level != output_level) { int parent_index = -1; if (!SetupOtherInputs(cf_name, mutable_cf_options, vstorage, &inputs, &output_level_inputs, &parent_index, -1)) { // manual compaction is now multi-threaded, so it can // happen that SetupOtherInputs fails // we handle it higher in RunManualCompaction *manual_conflict = true; return nullptr; } } std::vector compaction_inputs({inputs}); if (!output_level_inputs.empty()) { compaction_inputs.push_back(output_level_inputs); } for (size_t i = 0; i < compaction_inputs.size(); i++) { if (FilesInCompaction(compaction_inputs[i].files)) { *manual_conflict = true; return nullptr; } } std::vector grandparents; GetGrandparents(vstorage, inputs, output_level_inputs, &grandparents); Compaction* compaction = new Compaction( vstorage, mutable_cf_options, std::move(compaction_inputs), output_level, mutable_cf_options.MaxFileSizeForLevel(output_level), mutable_cf_options.MaxGrandParentOverlapBytes(input_level), output_path_id, GetCompressionType(ioptions_, output_level, vstorage->base_level()), std::move(grandparents), /* is manual compaction */ true); TEST_SYNC_POINT_CALLBACK("CompactionPicker::CompactRange:Return", compaction); if (input_level == 0) { level0_compactions_in_progress_.insert(compaction); } return compaction; } #ifndef ROCKSDB_LITE namespace { // Test whether two files have overlapping key-ranges. bool HaveOverlappingKeyRanges( const Comparator* c, const SstFileMetaData& a, const SstFileMetaData& b) { if (c->Compare(a.smallestkey, b.smallestkey) >= 0) { if (c->Compare(a.smallestkey, b.largestkey) <= 0) { // b.smallestkey <= a.smallestkey <= b.largestkey return true; } } else if (c->Compare(a.largestkey, b.smallestkey) >= 0) { // a.smallestkey < b.smallestkey <= a.largestkey return true; } if (c->Compare(a.largestkey, b.largestkey) <= 0) { if (c->Compare(a.largestkey, b.smallestkey) >= 0) { // b.smallestkey <= a.largestkey <= b.largestkey return true; } } else if (c->Compare(a.smallestkey, b.largestkey) <= 0) { // a.smallestkey <= b.largestkey < a.largestkey return true; } return false; } } // namespace Status CompactionPicker::SanitizeCompactionInputFilesForAllLevels( std::unordered_set* input_files, const ColumnFamilyMetaData& cf_meta, const int output_level) const { auto& levels = cf_meta.levels; auto comparator = icmp_->user_comparator(); // TODO(yhchiang): If there is any input files of L1 or up and there // is at least one L0 files. All L0 files older than the L0 file needs // to be included. Otherwise, it is a false conditoin // TODO(yhchiang): add is_adjustable to CompactionOptions // the smallest and largest key of the current compaction input std::string smallestkey; std::string largestkey; // a flag for initializing smallest and largest key bool is_first = false; const int kNotFound = -1; // For each level, it does the following things: // 1. Find the first and the last compaction input files // in the current level. // 2. Include all files between the first and the last // compaction input files. // 3. Update the compaction key-range. // 4. For all remaining levels, include files that have // overlapping key-range with the compaction key-range. for (int l = 0; l <= output_level; ++l) { auto& current_files = levels[l].files; int first_included = static_cast(current_files.size()); int last_included = kNotFound; // identify the first and the last compaction input files // in the current level. for (size_t f = 0; f < current_files.size(); ++f) { if (input_files->find(TableFileNameToNumber(current_files[f].name)) != input_files->end()) { first_included = std::min(first_included, static_cast(f)); last_included = std::max(last_included, static_cast(f)); if (is_first == false) { smallestkey = current_files[f].smallestkey; largestkey = current_files[f].largestkey; is_first = true; } } } if (last_included == kNotFound) { continue; } if (l != 0) { // expend the compaction input of the current level if it // has overlapping key-range with other non-compaction input // files in the same level. while (first_included > 0) { if (comparator->Compare( current_files[first_included - 1].largestkey, current_files[first_included].smallestkey) < 0) { break; } first_included--; } while (last_included < static_cast(current_files.size()) - 1) { if (comparator->Compare( current_files[last_included + 1].smallestkey, current_files[last_included].largestkey) > 0) { break; } last_included++; } } // include all files between the first and the last compaction input files. for (int f = first_included; f <= last_included; ++f) { if (current_files[f].being_compacted) { return Status::Aborted( "Necessary compaction input file " + current_files[f].name + " is currently being compacted."); } input_files->insert( TableFileNameToNumber(current_files[f].name)); } // update smallest and largest key if (l == 0) { for (int f = first_included; f <= last_included; ++f) { if (comparator->Compare( smallestkey, current_files[f].smallestkey) > 0) { smallestkey = current_files[f].smallestkey; } if (comparator->Compare( largestkey, current_files[f].largestkey) < 0) { largestkey = current_files[f].largestkey; } } } else { if (comparator->Compare( smallestkey, current_files[first_included].smallestkey) > 0) { smallestkey = current_files[first_included].smallestkey; } if (comparator->Compare( largestkey, current_files[last_included].largestkey) < 0) { largestkey = current_files[last_included].largestkey; } } SstFileMetaData aggregated_file_meta; aggregated_file_meta.smallestkey = smallestkey; aggregated_file_meta.largestkey = largestkey; // For all lower levels, include all overlapping files. // We need to add overlapping files from the current level too because even // if there no input_files in level l, we would still need to add files // which overlap with the range containing the input_files in levels 0 to l // Level 0 doesn't need to be handled this way because files are sorted by // time and not by key for (int m = std::max(l, 1); m <= output_level; ++m) { for (auto& next_lv_file : levels[m].files) { if (HaveOverlappingKeyRanges( comparator, aggregated_file_meta, next_lv_file)) { if (next_lv_file.being_compacted) { return Status::Aborted( "File " + next_lv_file.name + " that has overlapping key range with one of the compaction " " input file is currently being compacted."); } input_files->insert( TableFileNameToNumber(next_lv_file.name)); } } } } return Status::OK(); } Status CompactionPicker::SanitizeCompactionInputFiles( std::unordered_set* input_files, const ColumnFamilyMetaData& cf_meta, const int output_level) const { assert(static_cast(cf_meta.levels.size()) - 1 == cf_meta.levels[cf_meta.levels.size() - 1].level); if (output_level >= static_cast(cf_meta.levels.size())) { return Status::InvalidArgument( "Output level for column family " + cf_meta.name + " must between [0, " + ToString(cf_meta.levels[cf_meta.levels.size() - 1].level) + "]."); } if (output_level > MaxOutputLevel()) { return Status::InvalidArgument( "Exceed the maximum output level defined by " "the current compaction algorithm --- " + ToString(MaxOutputLevel())); } if (output_level < 0) { return Status::InvalidArgument( "Output level cannot be negative."); } if (input_files->size() == 0) { return Status::InvalidArgument( "A compaction must contain at least one file."); } Status s = SanitizeCompactionInputFilesForAllLevels( input_files, cf_meta, output_level); if (!s.ok()) { return s; } // for all input files, check whether the file number matches // any currently-existing files. for (auto file_num : *input_files) { bool found = false; for (auto level_meta : cf_meta.levels) { for (auto file_meta : level_meta.files) { if (file_num == TableFileNameToNumber(file_meta.name)) { if (file_meta.being_compacted) { return Status::Aborted( "Specified compaction input file " + MakeTableFileName("", file_num) + " is already being compacted."); } found = true; break; } } if (found) { break; } } if (!found) { return Status::InvalidArgument( "Specified compaction input file " + MakeTableFileName("", file_num) + " does not exist in column family " + cf_meta.name + "."); } } return Status::OK(); } #endif // !ROCKSDB_LITE bool LevelCompactionPicker::NeedsCompaction(const VersionStorageInfo* vstorage) const { if (!vstorage->FilesMarkedForCompaction().empty()) { return true; } for (int i = 0; i <= vstorage->MaxInputLevel(); i++) { if (vstorage->CompactionScore(i) >= 1) { return true; } } return false; } void LevelCompactionPicker::PickFilesMarkedForCompactionExperimental( const std::string& cf_name, VersionStorageInfo* vstorage, CompactionInputFiles* inputs, int* level, int* output_level) { if (vstorage->FilesMarkedForCompaction().empty()) { return; } auto continuation = [&](std::pair level_file) { // If it's being compacted it has nothing to do here. // If this assert() fails that means that some function marked some // files as being_compacted, but didn't call ComputeCompactionScore() assert(!level_file.second->being_compacted); *level = level_file.first; *output_level = (*level == 0) ? vstorage->base_level() : *level + 1; if (*level == 0 && !level0_compactions_in_progress_.empty()) { return false; } inputs->files = {level_file.second}; inputs->level = *level; return ExpandWhileOverlapping(cf_name, vstorage, inputs); }; // take a chance on a random file first Random64 rnd(/* seed */ reinterpret_cast(vstorage)); size_t random_file_index = static_cast(rnd.Uniform( static_cast(vstorage->FilesMarkedForCompaction().size()))); if (continuation(vstorage->FilesMarkedForCompaction()[random_file_index])) { // found the compaction! return; } for (auto& level_file : vstorage->FilesMarkedForCompaction()) { if (continuation(level_file)) { // found the compaction! return; } } inputs->files.clear(); } Compaction* LevelCompactionPicker::PickCompaction( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) { int level = -1; int output_level = -1; int parent_index = -1; int base_index = -1; CompactionInputFiles inputs; double score = 0; CompactionReason compaction_reason = CompactionReason::kUnknown; // Find the compactions by size on all levels. bool skipped_l0 = false; for (int i = 0; i < NumberLevels() - 1; i++) { score = vstorage->CompactionScore(i); level = vstorage->CompactionScoreLevel(i); assert(i == 0 || score <= vstorage->CompactionScore(i - 1)); if (score >= 1) { if (skipped_l0 && level == vstorage->base_level()) { // If L0->base_level compaction is pending, don't schedule further // compaction from base level. Otherwise L0->base_level compaction // may starve. continue; } output_level = (level == 0) ? vstorage->base_level() : level + 1; if (PickCompactionBySize(vstorage, level, output_level, &inputs, &parent_index, &base_index) && ExpandWhileOverlapping(cf_name, vstorage, &inputs)) { // found the compaction! if (level == 0) { // L0 score = `num L0 files` / `level0_file_num_compaction_trigger` compaction_reason = CompactionReason::kLevelL0FilesNum; } else { // L1+ score = `Level files size` / `MaxBytesForLevel` compaction_reason = CompactionReason::kLevelMaxLevelSize; } break; } else { // didn't find the compaction, clear the inputs inputs.clear(); if (level == 0) { skipped_l0 = true; } } } } bool is_manual = false; // if we didn't find a compaction, check if there are any files marked for // compaction if (inputs.empty()) { is_manual = true; parent_index = base_index = -1; PickFilesMarkedForCompactionExperimental(cf_name, vstorage, &inputs, &level, &output_level); if (!inputs.empty()) { compaction_reason = CompactionReason::kFilesMarkedForCompaction; } } if (inputs.empty()) { return nullptr; } assert(level >= 0 && output_level >= 0); // Two level 0 compaction won't run at the same time, so don't need to worry // about files on level 0 being compacted. if (level == 0) { assert(level0_compactions_in_progress_.empty()); InternalKey smallest, largest; GetRange(inputs, &smallest, &largest); // Note that the next call will discard the file we placed in // c->inputs_[0] earlier and replace it with an overlapping set // which will include the picked file. inputs.files.clear(); vstorage->GetOverlappingInputs(0, &smallest, &largest, &inputs.files); // If we include more L0 files in the same compaction run it can // cause the 'smallest' and 'largest' key to get extended to a // larger range. So, re-invoke GetRange to get the new key range GetRange(inputs, &smallest, &largest); if (RangeInCompaction(vstorage, &smallest, &largest, output_level, &parent_index)) { return nullptr; } assert(!inputs.files.empty()); } // Setup input files from output level CompactionInputFiles output_level_inputs; output_level_inputs.level = output_level; if (!SetupOtherInputs(cf_name, mutable_cf_options, vstorage, &inputs, &output_level_inputs, &parent_index, base_index)) { return nullptr; } std::vector compaction_inputs({inputs}); if (!output_level_inputs.empty()) { compaction_inputs.push_back(output_level_inputs); } std::vector grandparents; GetGrandparents(vstorage, inputs, output_level_inputs, &grandparents); auto c = new Compaction( vstorage, mutable_cf_options, std::move(compaction_inputs), output_level, mutable_cf_options.MaxFileSizeForLevel(output_level), mutable_cf_options.MaxGrandParentOverlapBytes(level), GetPathId(ioptions_, mutable_cf_options, output_level), GetCompressionType(ioptions_, output_level, vstorage->base_level()), std::move(grandparents), is_manual, score, false /* deletion_compaction */, compaction_reason); // If it's level 0 compaction, make sure we don't execute any other level 0 // compactions in parallel if (level == 0) { level0_compactions_in_progress_.insert(c); } // Creating a compaction influences the compaction score because the score // takes running compactions into account (by skipping files that are already // being compacted). Since we just changed compaction score, we recalculate it // here { // this piece of code recomputes compaction score CompactionOptionsFIFO dummy_compaction_options_fifo; vstorage->ComputeCompactionScore(mutable_cf_options, dummy_compaction_options_fifo); } TEST_SYNC_POINT_CALLBACK("LevelCompactionPicker::PickCompaction:Return", c); return c; } /* * Find the optimal path to place a file * Given a level, finds the path where levels up to it will fit in levels * up to and including this path */ uint32_t LevelCompactionPicker::GetPathId( const ImmutableCFOptions& ioptions, const MutableCFOptions& mutable_cf_options, int level) { uint32_t p = 0; assert(!ioptions.db_paths.empty()); // size remaining in the most recent path uint64_t current_path_size = ioptions.db_paths[0].target_size; uint64_t level_size; int cur_level = 0; level_size = mutable_cf_options.max_bytes_for_level_base; // Last path is the fallback while (p < ioptions.db_paths.size() - 1) { if (level_size <= current_path_size) { if (cur_level == level) { // Does desired level fit in this path? return p; } else { current_path_size -= level_size; level_size *= mutable_cf_options.max_bytes_for_level_multiplier; cur_level++; continue; } } p++; current_path_size = ioptions.db_paths[p].target_size; } return p; } bool LevelCompactionPicker::PickCompactionBySize(VersionStorageInfo* vstorage, int level, int output_level, CompactionInputFiles* inputs, int* parent_index, int* base_index) { // level 0 files are overlapping. So we cannot pick more // than one concurrent compactions at this level. This // could be made better by looking at key-ranges that are // being compacted at level 0. if (level == 0 && !level0_compactions_in_progress_.empty()) { TEST_SYNC_POINT("LevelCompactionPicker::PickCompactionBySize:0"); return false; } inputs->clear(); assert(level >= 0); // Pick the largest file in this level that is not already // being compacted const std::vector& file_size = vstorage->FilesByCompactionPri(level); const std::vector& level_files = vstorage->LevelFiles(level); // record the first file that is not yet compacted int nextIndex = -1; for (unsigned int i = vstorage->NextCompactionIndex(level); i < file_size.size(); i++) { int index = file_size[i]; auto* f = level_files[index]; // do not pick a file to compact if it is being compacted // from n-1 level. if (f->being_compacted) { continue; } // remember the startIndex for the next call to PickCompaction if (nextIndex == -1) { nextIndex = i; } // Do not pick this file if its parents at level+1 are being compacted. // Maybe we can avoid redoing this work in SetupOtherInputs *parent_index = -1; if (RangeInCompaction(vstorage, &f->smallest, &f->largest, output_level, parent_index)) { continue; } inputs->files.push_back(f); inputs->level = level; *base_index = index; break; } // store where to start the iteration in the next call to PickCompaction vstorage->SetNextCompactionIndex(level, nextIndex); return inputs->size() > 0; } #ifndef ROCKSDB_LITE bool UniversalCompactionPicker::NeedsCompaction( const VersionStorageInfo* vstorage) const { const int kLevel0 = 0; return vstorage->CompactionScore(kLevel0) >= 1; } void UniversalCompactionPicker::SortedRun::Dump(char* out_buf, size_t out_buf_size, bool print_path) const { if (level == 0) { assert(file != nullptr); if (file->fd.GetPathId() == 0 || !print_path) { snprintf(out_buf, out_buf_size, "file %" PRIu64, file->fd.GetNumber()); } else { snprintf(out_buf, out_buf_size, "file %" PRIu64 "(path " "%" PRIu32 ")", file->fd.GetNumber(), file->fd.GetPathId()); } } else { snprintf(out_buf, out_buf_size, "level %d", level); } } void UniversalCompactionPicker::SortedRun::DumpSizeInfo( char* out_buf, size_t out_buf_size, size_t sorted_run_count) const { if (level == 0) { assert(file != nullptr); snprintf(out_buf, out_buf_size, "file %" PRIu64 "[%" ROCKSDB_PRIszt "] " "with size %" PRIu64 " (compensated size %" PRIu64 ")", file->fd.GetNumber(), sorted_run_count, file->fd.GetFileSize(), file->compensated_file_size); } else { snprintf(out_buf, out_buf_size, "level %d[%" ROCKSDB_PRIszt "] " "with size %" PRIu64 " (compensated size %" PRIu64 ")", level, sorted_run_count, size, compensated_file_size); } } std::vector UniversalCompactionPicker::CalculateSortedRuns( const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions) { std::vector ret; for (FileMetaData* f : vstorage.LevelFiles(0)) { ret.emplace_back(0, f, f->fd.GetFileSize(), f->compensated_file_size, f->being_compacted); } for (int level = 1; level < vstorage.num_levels(); level++) { uint64_t total_compensated_size = 0U; uint64_t total_size = 0U; bool being_compacted = false; bool is_first = true; for (FileMetaData* f : vstorage.LevelFiles(level)) { total_compensated_size += f->compensated_file_size; total_size += f->fd.GetFileSize(); if (ioptions.compaction_options_universal.allow_trivial_move == true) { if (f->being_compacted) { being_compacted = f->being_compacted; } } else { // Compaction always includes all files for a non-zero level, so for a // non-zero level, all the files should share the same being_compacted // value. // This assumption is only valid when // ioptions.compaction_options_universal.allow_trivial_move is false assert(is_first || f->being_compacted == being_compacted); } if (is_first) { being_compacted = f->being_compacted; is_first = false; } } if (total_compensated_size > 0) { ret.emplace_back(level, nullptr, total_size, total_compensated_size, being_compacted); } } return ret; } #ifndef NDEBUG namespace { // smallest_seqno and largest_seqno are set iff. `files` is not empty. void GetSmallestLargestSeqno(const std::vector& files, SequenceNumber* smallest_seqno, SequenceNumber* largest_seqno) { bool is_first = true; for (FileMetaData* f : files) { assert(f->smallest_seqno <= f->largest_seqno); if (is_first) { is_first = false; *smallest_seqno = f->smallest_seqno; *largest_seqno = f->largest_seqno; } else { if (f->smallest_seqno < *smallest_seqno) { *smallest_seqno = f->smallest_seqno; } if (f->largest_seqno > *largest_seqno) { *largest_seqno = f->largest_seqno; } } } } } // namespace #endif // Algorithm that checks to see if there are any overlapping // files in the input bool CompactionPicker::IsInputNonOverlapping(Compaction* c) { auto comparator = icmp_->user_comparator(); int first_iter = 1; InputFileInfo prev, curr, next; SmallestKeyHeap smallest_key_priority_q = create_level_heap(c, icmp_->user_comparator()); while (!smallest_key_priority_q.empty()) { curr = smallest_key_priority_q.top(); smallest_key_priority_q.pop(); if (first_iter) { prev = curr; first_iter = 0; } else { if (comparator->Compare(prev.f->largest.user_key(), curr.f->smallest.user_key()) >= 0) { // found overlapping files, return false return false; } assert(comparator->Compare(curr.f->largest.user_key(), prev.f->largest.user_key()) > 0); prev = curr; } next.f = nullptr; if (curr.level != 0 && curr.index < c->num_input_files(curr.level) - 1) { next.f = c->input(curr.level, curr.index + 1); next.level = curr.level; next.index = curr.index + 1; } if (next.f) { smallest_key_priority_q.push(std::move(next)); } } return true; } // Universal style of compaction. Pick files that are contiguous in // time-range to compact. // Compaction* UniversalCompactionPicker::PickCompaction( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) { const int kLevel0 = 0; double score = vstorage->CompactionScore(kLevel0); std::vector sorted_runs = CalculateSortedRuns(*vstorage, ioptions_); if (sorted_runs.size() == 0 || sorted_runs.size() < (unsigned int)mutable_cf_options.level0_file_num_compaction_trigger) { LogToBuffer(log_buffer, "[%s] Universal: nothing to do\n", cf_name.c_str()); return nullptr; } VersionStorageInfo::LevelSummaryStorage tmp; LogToBuffer(log_buffer, 3072, "[%s] Universal: sorted runs files(%" ROCKSDB_PRIszt "): %s\n", cf_name.c_str(), sorted_runs.size(), vstorage->LevelSummary(&tmp)); // Check for size amplification first. Compaction* c; if ((c = PickCompactionUniversalSizeAmp(cf_name, mutable_cf_options, vstorage, score, sorted_runs, log_buffer)) != nullptr) { LogToBuffer(log_buffer, "[%s] Universal: compacting for size amp\n", cf_name.c_str()); } else { // Size amplification is within limits. Try reducing read // amplification while maintaining file size ratios. unsigned int ratio = ioptions_.compaction_options_universal.size_ratio; if ((c = PickCompactionUniversalReadAmp( cf_name, mutable_cf_options, vstorage, score, ratio, UINT_MAX, sorted_runs, log_buffer)) != nullptr) { LogToBuffer(log_buffer, "[%s] Universal: compacting for size ratio\n", cf_name.c_str()); } else { // Size amplification and file size ratios are within configured limits. // If max read amplification is exceeding configured limits, then force // compaction without looking at filesize ratios and try to reduce // the number of files to fewer than level0_file_num_compaction_trigger. // This is guaranteed by NeedsCompaction() assert(sorted_runs.size() >= static_cast( mutable_cf_options.level0_file_num_compaction_trigger)); unsigned int num_files = static_cast(sorted_runs.size()) - mutable_cf_options.level0_file_num_compaction_trigger; if ((c = PickCompactionUniversalReadAmp( cf_name, mutable_cf_options, vstorage, score, UINT_MAX, num_files, sorted_runs, log_buffer)) != nullptr) { LogToBuffer(log_buffer, "[%s] Universal: compacting for file num -- %u\n", cf_name.c_str(), num_files); } } } if (c == nullptr) { return nullptr; } if (ioptions_.compaction_options_universal.allow_trivial_move == true) { c->set_is_trivial_move(IsInputNonOverlapping(c)); } // validate that all the chosen files of L0 are non overlapping in time #ifndef NDEBUG SequenceNumber prev_smallest_seqno = 0U; bool is_first = true; size_t level_index = 0U; if (c->start_level() == 0) { for (auto f : *c->inputs(0)) { assert(f->smallest_seqno <= f->largest_seqno); if (is_first) { is_first = false; } else { assert(prev_smallest_seqno > f->largest_seqno); } prev_smallest_seqno = f->smallest_seqno; } level_index = 1U; } for (; level_index < c->num_input_levels(); level_index++) { if (c->num_input_files(level_index) != 0) { SequenceNumber smallest_seqno = 0U; SequenceNumber largest_seqno = 0U; GetSmallestLargestSeqno(*(c->inputs(level_index)), &smallest_seqno, &largest_seqno); if (is_first) { is_first = false; } else if (prev_smallest_seqno > 0) { // A level is considered as the bottommost level if there are // no files in higher levels or if files in higher levels do // not overlap with the files being compacted. Sequence numbers // of files in bottommost level can be set to 0 to help // compression. As a result, the following assert may not hold // if the prev_smallest_seqno is 0. assert(prev_smallest_seqno > largest_seqno); } prev_smallest_seqno = smallest_seqno; } } #endif // update statistics MeasureTime(ioptions_.statistics, NUM_FILES_IN_SINGLE_COMPACTION, c->inputs(0)->size()); level0_compactions_in_progress_.insert(c); return c; } uint32_t UniversalCompactionPicker::GetPathId( const ImmutableCFOptions& ioptions, uint64_t file_size) { // Two conditions need to be satisfied: // (1) the target path needs to be able to hold the file's size // (2) Total size left in this and previous paths need to be not // smaller than expected future file size before this new file is // compacted, which is estimated based on size_ratio. // For example, if now we are compacting files of size (1, 1, 2, 4, 8), // we will make sure the target file, probably with size of 16, will be // placed in a path so that eventually when new files are generated and // compacted to (1, 1, 2, 4, 8, 16), all those files can be stored in or // before the path we chose. // // TODO(sdong): now the case of multiple column families is not // considered in this algorithm. So the target size can be violated in // that case. We need to improve it. uint64_t accumulated_size = 0; uint64_t future_size = file_size * (100 - ioptions.compaction_options_universal.size_ratio) / 100; uint32_t p = 0; assert(!ioptions.db_paths.empty()); for (; p < ioptions.db_paths.size() - 1; p++) { uint64_t target_size = ioptions.db_paths[p].target_size; if (target_size > file_size && accumulated_size + (target_size - file_size) > future_size) { return p; } accumulated_size += target_size; } return p; } // // Consider compaction files based on their size differences with // the next file in time order. // Compaction* UniversalCompactionPicker::PickCompactionUniversalReadAmp( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, double score, unsigned int ratio, unsigned int max_number_of_files_to_compact, const std::vector& sorted_runs, LogBuffer* log_buffer) { unsigned int min_merge_width = ioptions_.compaction_options_universal.min_merge_width; unsigned int max_merge_width = ioptions_.compaction_options_universal.max_merge_width; const SortedRun* sr = nullptr; bool done = false; size_t start_index = 0; unsigned int candidate_count = 0; unsigned int max_files_to_compact = std::min(max_merge_width, max_number_of_files_to_compact); min_merge_width = std::max(min_merge_width, 2U); // Caller checks the size before executing this function. This invariant is // important because otherwise we may have a possible integer underflow when // dealing with unsigned types. assert(sorted_runs.size() > 0); // Considers a candidate file only if it is smaller than the // total size accumulated so far. for (size_t loop = 0; loop < sorted_runs.size(); loop++) { candidate_count = 0; // Skip files that are already being compacted for (sr = nullptr; loop < sorted_runs.size(); loop++) { sr = &sorted_runs[loop]; if (!sr->being_compacted) { candidate_count = 1; break; } char file_num_buf[kFormatFileNumberBufSize]; sr->Dump(file_num_buf, sizeof(file_num_buf)); LogToBuffer(log_buffer, "[%s] Universal: %s" "[%d] being compacted, skipping", cf_name.c_str(), file_num_buf, loop); sr = nullptr; } // This file is not being compacted. Consider it as the // first candidate to be compacted. uint64_t candidate_size = sr != nullptr ? sr->compensated_file_size : 0; if (sr != nullptr) { char file_num_buf[kFormatFileNumberBufSize]; sr->Dump(file_num_buf, sizeof(file_num_buf), true); LogToBuffer(log_buffer, "[%s] Universal: Possible candidate %s[%d].", cf_name.c_str(), file_num_buf, loop); } // Check if the succeeding files need compaction. for (size_t i = loop + 1; candidate_count < max_files_to_compact && i < sorted_runs.size(); i++) { const SortedRun* succeeding_sr = &sorted_runs[i]; if (succeeding_sr->being_compacted) { break; } // Pick files if the total/last candidate file size (increased by the // specified ratio) is still larger than the next candidate file. // candidate_size is the total size of files picked so far with the // default kCompactionStopStyleTotalSize; with // kCompactionStopStyleSimilarSize, it's simply the size of the last // picked file. double sz = candidate_size * (100.0 + ratio) / 100.0; if (sz < static_cast(succeeding_sr->size)) { break; } if (ioptions_.compaction_options_universal.stop_style == kCompactionStopStyleSimilarSize) { // Similar-size stopping rule: also check the last picked file isn't // far larger than the next candidate file. sz = (succeeding_sr->size * (100.0 + ratio)) / 100.0; if (sz < static_cast(candidate_size)) { // If the small file we've encountered begins a run of similar-size // files, we'll pick them up on a future iteration of the outer // loop. If it's some lonely straggler, it'll eventually get picked // by the last-resort read amp strategy which disregards size ratios. break; } candidate_size = succeeding_sr->compensated_file_size; } else { // default kCompactionStopStyleTotalSize candidate_size += succeeding_sr->compensated_file_size; } candidate_count++; } // Found a series of consecutive files that need compaction. if (candidate_count >= (unsigned int)min_merge_width) { start_index = loop; done = true; break; } else { for (size_t i = loop; i < loop + candidate_count && i < sorted_runs.size(); i++) { const SortedRun* skipping_sr = &sorted_runs[i]; char file_num_buf[256]; skipping_sr->DumpSizeInfo(file_num_buf, sizeof(file_num_buf), loop); LogToBuffer(log_buffer, "[%s] Universal: Skipping %s", cf_name.c_str(), file_num_buf); } } } if (!done || candidate_count <= 1) { return nullptr; } size_t first_index_after = start_index + candidate_count; // Compression is enabled if files compacted earlier already reached // size ratio of compression. bool enable_compression = true; int ratio_to_compress = ioptions_.compaction_options_universal.compression_size_percent; if (ratio_to_compress >= 0) { uint64_t total_size = 0; for (auto& sorted_run : sorted_runs) { total_size += sorted_run.compensated_file_size; } uint64_t older_file_size = 0; for (size_t i = sorted_runs.size() - 1; i >= first_index_after; i--) { older_file_size += sorted_runs[i].size; if (older_file_size * 100L >= total_size * (long) ratio_to_compress) { enable_compression = false; break; } } } uint64_t estimated_total_size = 0; for (unsigned int i = 0; i < first_index_after; i++) { estimated_total_size += sorted_runs[i].size; } uint32_t path_id = GetPathId(ioptions_, estimated_total_size); int start_level = sorted_runs[start_index].level; int output_level; if (first_index_after == sorted_runs.size()) { output_level = vstorage->num_levels() - 1; } else if (sorted_runs[first_index_after].level == 0) { output_level = 0; } else { output_level = sorted_runs[first_index_after].level - 1; } std::vector inputs(vstorage->num_levels()); for (size_t i = 0; i < inputs.size(); ++i) { inputs[i].level = start_level + static_cast(i); } for (size_t i = start_index; i < first_index_after; i++) { auto& picking_sr = sorted_runs[i]; if (picking_sr.level == 0) { FileMetaData* picking_file = picking_sr.file; inputs[0].files.push_back(picking_file); } else { auto& files = inputs[picking_sr.level - start_level].files; for (auto* f : vstorage->LevelFiles(picking_sr.level)) { files.push_back(f); } } char file_num_buf[256]; picking_sr.DumpSizeInfo(file_num_buf, sizeof(file_num_buf), i); LogToBuffer(log_buffer, "[%s] Universal: Picking %s", cf_name.c_str(), file_num_buf); } CompactionReason compaction_reason; if (max_number_of_files_to_compact == UINT_MAX) { compaction_reason = CompactionReason::kUniversalSortedRunNum; } else { compaction_reason = CompactionReason::kUniversalSizeRatio; } return new Compaction( vstorage, mutable_cf_options, std::move(inputs), output_level, mutable_cf_options.MaxFileSizeForLevel(output_level), LLONG_MAX, path_id, GetCompressionType(ioptions_, start_level, 1, enable_compression), /* grandparents */ {}, /* is manual */ false, score, false /* deletion_compaction */, compaction_reason); } // Look at overall size amplification. If size amplification // exceeeds the configured value, then do a compaction // of the candidate files all the way upto the earliest // base file (overrides configured values of file-size ratios, // min_merge_width and max_merge_width). // Compaction* UniversalCompactionPicker::PickCompactionUniversalSizeAmp( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, double score, const std::vector& sorted_runs, LogBuffer* log_buffer) { // percentage flexibilty while reducing size amplification uint64_t ratio = ioptions_.compaction_options_universal. max_size_amplification_percent; unsigned int candidate_count = 0; uint64_t candidate_size = 0; size_t start_index = 0; const SortedRun* sr = nullptr; // Skip files that are already being compacted for (size_t loop = 0; loop < sorted_runs.size() - 1; loop++) { sr = &sorted_runs[loop]; if (!sr->being_compacted) { start_index = loop; // Consider this as the first candidate. break; } char file_num_buf[kFormatFileNumberBufSize]; sr->Dump(file_num_buf, sizeof(file_num_buf), true); LogToBuffer(log_buffer, "[%s] Universal: skipping %s[%d] compacted %s", cf_name.c_str(), file_num_buf, loop, " cannot be a candidate to reduce size amp.\n"); sr = nullptr; } if (sr == nullptr) { return nullptr; // no candidate files } { char file_num_buf[kFormatFileNumberBufSize]; sr->Dump(file_num_buf, sizeof(file_num_buf), true); LogToBuffer(log_buffer, "[%s] Universal: First candidate %s[%" ROCKSDB_PRIszt "] %s", cf_name.c_str(), file_num_buf, start_index, " to reduce size amp.\n"); } // keep adding up all the remaining files for (size_t loop = start_index; loop < sorted_runs.size() - 1; loop++) { sr = &sorted_runs[loop]; if (sr->being_compacted) { char file_num_buf[kFormatFileNumberBufSize]; sr->Dump(file_num_buf, sizeof(file_num_buf), true); LogToBuffer( log_buffer, "[%s] Universal: Possible candidate %s[%d] %s", cf_name.c_str(), file_num_buf, start_index, " is already being compacted. No size amp reduction possible.\n"); return nullptr; } candidate_size += sr->compensated_file_size; candidate_count++; } if (candidate_count == 0) { return nullptr; } // size of earliest file uint64_t earliest_file_size = sorted_runs.back().size; // size amplification = percentage of additional size if (candidate_size * 100 < ratio * earliest_file_size) { LogToBuffer( log_buffer, "[%s] Universal: size amp not needed. newer-files-total-size %" PRIu64 "earliest-file-size %" PRIu64, cf_name.c_str(), candidate_size, earliest_file_size); return nullptr; } else { LogToBuffer( log_buffer, "[%s] Universal: size amp needed. newer-files-total-size %" PRIu64 "earliest-file-size %" PRIu64, cf_name.c_str(), candidate_size, earliest_file_size); } assert(start_index < sorted_runs.size() - 1); // Estimate total file size uint64_t estimated_total_size = 0; for (size_t loop = start_index; loop < sorted_runs.size(); loop++) { estimated_total_size += sorted_runs[loop].size; } uint32_t path_id = GetPathId(ioptions_, estimated_total_size); int start_level = sorted_runs[start_index].level; std::vector inputs(vstorage->num_levels()); for (size_t i = 0; i < inputs.size(); ++i) { inputs[i].level = start_level + static_cast(i); } // We always compact all the files, so always compress. for (size_t loop = start_index; loop < sorted_runs.size(); loop++) { auto& picking_sr = sorted_runs[loop]; if (picking_sr.level == 0) { FileMetaData* f = picking_sr.file; inputs[0].files.push_back(f); } else { auto& files = inputs[picking_sr.level - start_level].files; for (auto* f : vstorage->LevelFiles(picking_sr.level)) { files.push_back(f); } } char file_num_buf[256]; picking_sr.DumpSizeInfo(file_num_buf, sizeof(file_num_buf), loop); LogToBuffer(log_buffer, "[%s] Universal: size amp picking %s", cf_name.c_str(), file_num_buf); } return new Compaction( vstorage, mutable_cf_options, std::move(inputs), vstorage->num_levels() - 1, mutable_cf_options.MaxFileSizeForLevel(vstorage->num_levels() - 1), /* max_grandparent_overlap_bytes */ LLONG_MAX, path_id, GetCompressionType(ioptions_, vstorage->num_levels() - 1, 1), /* grandparents */ {}, /* is manual */ false, score, false /* deletion_compaction */, CompactionReason::kUniversalSizeAmplification); } bool FIFOCompactionPicker::NeedsCompaction(const VersionStorageInfo* vstorage) const { const int kLevel0 = 0; return vstorage->CompactionScore(kLevel0) >= 1; } Compaction* FIFOCompactionPicker::PickCompaction( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) { assert(vstorage->num_levels() == 1); const int kLevel0 = 0; const std::vector& level_files = vstorage->LevelFiles(kLevel0); uint64_t total_size = 0; for (const auto& file : level_files) { total_size += file->fd.file_size; } if (total_size <= ioptions_.compaction_options_fifo.max_table_files_size || level_files.size() == 0) { // total size not exceeded LogToBuffer(log_buffer, "[%s] FIFO compaction: nothing to do. Total size %" PRIu64 ", max size %" PRIu64 "\n", cf_name.c_str(), total_size, ioptions_.compaction_options_fifo.max_table_files_size); return nullptr; } if (!level0_compactions_in_progress_.empty()) { LogToBuffer(log_buffer, "[%s] FIFO compaction: Already executing compaction. No need " "to run parallel compactions since compactions are very fast", cf_name.c_str()); return nullptr; } std::vector inputs; inputs.emplace_back(); inputs[0].level = 0; // delete old files (FIFO) for (auto ritr = level_files.rbegin(); ritr != level_files.rend(); ++ritr) { auto f = *ritr; total_size -= f->compensated_file_size; inputs[0].files.push_back(f); char tmp_fsize[16]; AppendHumanBytes(f->fd.GetFileSize(), tmp_fsize, sizeof(tmp_fsize)); LogToBuffer(log_buffer, "[%s] FIFO compaction: picking file %" PRIu64 " with size %s for deletion", cf_name.c_str(), f->fd.GetNumber(), tmp_fsize); if (total_size <= ioptions_.compaction_options_fifo.max_table_files_size) { break; } } Compaction* c = new Compaction( vstorage, mutable_cf_options, std::move(inputs), 0, 0, 0, 0, kNoCompression, {}, /* is manual */ false, vstorage->CompactionScore(0), /* is deletion compaction */ true, CompactionReason::kFIFOMaxSize); level0_compactions_in_progress_.insert(c); return c; } Compaction* FIFOCompactionPicker::CompactRange( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict) { assert(input_level == 0); assert(output_level == 0); *compaction_end = nullptr; LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, ioptions_.info_log); Compaction* c = PickCompaction(cf_name, mutable_cf_options, vstorage, &log_buffer); log_buffer.FlushBufferToLog(); return c; } #endif // !ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/compaction_picker.h000066400000000000000000000350521267530634000174120ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include "db/compaction.h" #include "db/version_set.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/status.h" #include "util/mutable_cf_options.h" namespace rocksdb { class LogBuffer; class Compaction; class VersionStorageInfo; struct CompactionInputFiles; class CompactionPicker { public: CompactionPicker(const ImmutableCFOptions& ioptions, const InternalKeyComparator* icmp); virtual ~CompactionPicker(); // Pick level and inputs for a new compaction. // Returns nullptr if there is no compaction to be done. // Otherwise returns a pointer to a heap-allocated object that // describes the compaction. Caller should delete the result. virtual Compaction* PickCompaction(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) = 0; // Return a compaction object for compacting the range [begin,end] in // the specified level. Returns nullptr if there is nothing in that // level that overlaps the specified range. Caller should delete // the result. // // The returned Compaction might not include the whole requested range. // In that case, compaction_end will be set to the next key that needs // compacting. In case the compaction will compact the whole range, // compaction_end will be set to nullptr. // Client is responsible for compaction_end storage -- when called, // *compaction_end should point to valid InternalKey! virtual Compaction* CompactRange( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict); // The maximum allowed output level. Default value is NumberLevels() - 1. virtual int MaxOutputLevel() const { return NumberLevels() - 1; } virtual bool NeedsCompaction(const VersionStorageInfo* vstorage) const = 0; // Sanitize the input set of compaction input files. // When the input parameters do not describe a valid compaction, the // function will try to fix the input_files by adding necessary // files. If it's not possible to conver an invalid input_files // into a valid one by adding more files, the function will return a // non-ok status with specific reason. #ifndef ROCKSDB_LITE Status SanitizeCompactionInputFiles( std::unordered_set* input_files, const ColumnFamilyMetaData& cf_meta, const int output_level) const; #endif // ROCKSDB_LITE // Free up the files that participated in a compaction void ReleaseCompactionFiles(Compaction* c, Status status); // Returns true if any one of the specified files are being compacted bool FilesInCompaction(const std::vector& files); // Takes a list of CompactionInputFiles and returns a (manual) Compaction // object. Compaction* FormCompaction( const CompactionOptions& compact_options, const std::vector& input_files, int output_level, VersionStorageInfo* vstorage, const MutableCFOptions& mutable_cf_options, uint32_t output_path_id); // Converts a set of compaction input file numbers into // a list of CompactionInputFiles. Status GetCompactionInputsFromFileNumbers( std::vector* input_files, std::unordered_set* input_set, const VersionStorageInfo* vstorage, const CompactionOptions& compact_options) const; // Used in universal compaction when the enabled_trivial_move // option is set. Checks whether there are any overlapping files // in the input. Returns true if the input files are non // overlapping. bool IsInputNonOverlapping(Compaction* c); // Is there currently a compaction involving level 0 taking place bool IsLevel0CompactionInProgress() const { return !level0_compactions_in_progress_.empty(); } protected: int NumberLevels() const { return ioptions_.num_levels; } // Stores the minimal range that covers all entries in inputs in // *smallest, *largest. // REQUIRES: inputs is not empty void GetRange(const CompactionInputFiles& inputs, InternalKey* smallest, InternalKey* largest); // Stores the minimal range that covers all entries in inputs1 and inputs2 // in *smallest, *largest. // REQUIRES: inputs is not empty void GetRange(const CompactionInputFiles& inputs1, const CompactionInputFiles& inputs2, InternalKey* smallest, InternalKey* largest); // Add more files to the inputs on "level" to make sure that // no newer version of a key is compacted to "level+1" while leaving an older // version in a "level". Otherwise, any Get() will search "level" first, // and will likely return an old/stale value for the key, since it always // searches in increasing order of level to find the value. This could // also scramble the order of merge operands. This function should be // called any time a new Compaction is created, and its inputs_[0] are // populated. // // Will return false if it is impossible to apply this compaction. bool ExpandWhileOverlapping(const std::string& cf_name, VersionStorageInfo* vstorage, CompactionInputFiles* inputs); // Returns true if any one of the parent files are being compacted bool RangeInCompaction(VersionStorageInfo* vstorage, const InternalKey* smallest, const InternalKey* largest, int level, int* index); bool SetupOtherInputs(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, CompactionInputFiles* inputs, CompactionInputFiles* output_level_inputs, int* parent_index, int base_index); void GetGrandparents(VersionStorageInfo* vstorage, const CompactionInputFiles& inputs, const CompactionInputFiles& output_level_inputs, std::vector* grandparents); const ImmutableCFOptions& ioptions_; // A helper function to SanitizeCompactionInputFiles() that // sanitizes "input_files" by adding necessary files. #ifndef ROCKSDB_LITE virtual Status SanitizeCompactionInputFilesForAllLevels( std::unordered_set* input_files, const ColumnFamilyMetaData& cf_meta, const int output_level) const; #endif // ROCKSDB_LITE // Keeps track of all compactions that are running on Level0. // It is protected by DB mutex std::set level0_compactions_in_progress_; const InternalKeyComparator* const icmp_; }; class LevelCompactionPicker : public CompactionPicker { public: LevelCompactionPicker(const ImmutableCFOptions& ioptions, const InternalKeyComparator* icmp) : CompactionPicker(ioptions, icmp) {} virtual Compaction* PickCompaction(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) override; virtual bool NeedsCompaction(const VersionStorageInfo* vstorage) const override; // Pick a path ID to place a newly generated file, with its level static uint32_t GetPathId(const ImmutableCFOptions& ioptions, const MutableCFOptions& mutable_cf_options, int level); private: // For the specfied level, pick a file that we want to compact. // Returns false if there is no file to compact. // If it returns true, inputs->files.size() will be exactly one. // If level is 0 and there is already a compaction on that level, this // function will return false. bool PickCompactionBySize(VersionStorageInfo* vstorage, int level, int output_level, CompactionInputFiles* inputs, int* parent_index, int* base_index); // If there is any file marked for compaction, put put it into inputs. // This is still experimental. It will return meaningful results only if // clients call experimental feature SuggestCompactRange() void PickFilesMarkedForCompactionExperimental(const std::string& cf_name, VersionStorageInfo* vstorage, CompactionInputFiles* inputs, int* level, int* output_level); }; #ifndef ROCKSDB_LITE class UniversalCompactionPicker : public CompactionPicker { public: UniversalCompactionPicker(const ImmutableCFOptions& ioptions, const InternalKeyComparator* icmp) : CompactionPicker(ioptions, icmp) {} virtual Compaction* PickCompaction(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) override; virtual int MaxOutputLevel() const override { return NumberLevels() - 1; } virtual bool NeedsCompaction(const VersionStorageInfo* vstorage) const override; private: struct SortedRun { SortedRun(int _level, FileMetaData* _file, uint64_t _size, uint64_t _compensated_file_size, bool _being_compacted) : level(_level), file(_file), size(_size), compensated_file_size(_compensated_file_size), being_compacted(_being_compacted) { assert(compensated_file_size > 0); assert(level != 0 || file != nullptr); } void Dump(char* out_buf, size_t out_buf_size, bool print_path = false) const; // sorted_run_count is added into the string to print void DumpSizeInfo(char* out_buf, size_t out_buf_size, size_t sorted_run_count) const; int level; // `file` Will be null for level > 0. For level = 0, the sorted run is // for this file. FileMetaData* file; // For level > 0, `size` and `compensated_file_size` are sum of sizes all // files in the level. `being_compacted` should be the same for all files // in a non-zero level. Use the value here. uint64_t size; uint64_t compensated_file_size; bool being_compacted; }; // Pick Universal compaction to limit read amplification Compaction* PickCompactionUniversalReadAmp( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, double score, unsigned int ratio, unsigned int num_files, const std::vector& sorted_runs, LogBuffer* log_buffer); // Pick Universal compaction to limit space amplification. Compaction* PickCompactionUniversalSizeAmp( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, double score, const std::vector& sorted_runs, LogBuffer* log_buffer); static std::vector CalculateSortedRuns( const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions); // Pick a path ID to place a newly generated file, with its estimated file // size. static uint32_t GetPathId(const ImmutableCFOptions& ioptions, uint64_t file_size); }; class FIFOCompactionPicker : public CompactionPicker { public: FIFOCompactionPicker(const ImmutableCFOptions& ioptions, const InternalKeyComparator* icmp) : CompactionPicker(ioptions, icmp) {} virtual Compaction* PickCompaction(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* version, LogBuffer* log_buffer) override; virtual Compaction* CompactRange( const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict) override; // The maximum allowed output level. Always returns 0. virtual int MaxOutputLevel() const override { return 0; } virtual bool NeedsCompaction(const VersionStorageInfo* vstorage) const override; }; class NullCompactionPicker : public CompactionPicker { public: NullCompactionPicker(const ImmutableCFOptions& ioptions, const InternalKeyComparator* icmp) : CompactionPicker(ioptions, icmp) {} virtual ~NullCompactionPicker() {} // Always return "nullptr" Compaction* PickCompaction(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, LogBuffer* log_buffer) override { return nullptr; } // Always return "nullptr" Compaction* CompactRange(const std::string& cf_name, const MutableCFOptions& mutable_cf_options, VersionStorageInfo* vstorage, int input_level, int output_level, uint32_t output_path_id, const InternalKey* begin, const InternalKey* end, InternalKey** compaction_end, bool* manual_conflict) override { return nullptr; } // Always returns false. virtual bool NeedsCompaction(const VersionStorageInfo* vstorage) const override { return false; } }; #endif // !ROCKSDB_LITE CompressionType GetCompressionType(const ImmutableCFOptions& ioptions, int level, int base_level, const bool enable_compression = true); } // namespace rocksdb rocksdb-4.5.1/db/compaction_picker_test.cc000066400000000000000000001045631267530634000206130ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/compaction.h" #include "db/compaction_picker.h" #include #include #include #include "util/logging.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class CountingLogger : public Logger { public: using Logger::Logv; virtual void Logv(const char* format, va_list ap) override { log_count++; } size_t log_count; }; class CompactionPickerTest : public testing::Test { public: const Comparator* ucmp_; InternalKeyComparator icmp_; Options options_; ImmutableCFOptions ioptions_; MutableCFOptions mutable_cf_options_; LevelCompactionPicker level_compaction_picker; std::string cf_name_; CountingLogger logger_; LogBuffer log_buffer_; uint32_t file_num_; CompactionOptionsFIFO fifo_options_; std::unique_ptr vstorage_; std::vector> files_; // does not own FileMetaData std::unordered_map> file_map_; // input files to compaction process. std::vector input_files_; int compaction_level_start_; CompactionPickerTest() : ucmp_(BytewiseComparator()), icmp_(ucmp_), ioptions_(options_), mutable_cf_options_(options_, ioptions_), level_compaction_picker(ioptions_, &icmp_), cf_name_("dummy"), log_buffer_(InfoLogLevel::INFO_LEVEL, &logger_), file_num_(1), vstorage_(nullptr) { fifo_options_.max_table_files_size = 1; mutable_cf_options_.RefreshDerivedOptions(ioptions_); ioptions_.db_paths.emplace_back("dummy", std::numeric_limits::max()); } ~CompactionPickerTest() { } void NewVersionStorage(int num_levels, CompactionStyle style) { DeleteVersionStorage(); options_.num_levels = num_levels; vstorage_.reset(new VersionStorageInfo( &icmp_, ucmp_, options_.num_levels, style, nullptr)); vstorage_->CalculateBaseBytes(ioptions_, mutable_cf_options_); } void DeleteVersionStorage() { vstorage_.reset(); files_.clear(); file_map_.clear(); input_files_.clear(); } void Add(int level, uint32_t file_number, const char* smallest, const char* largest, uint64_t file_size = 0, uint32_t path_id = 0, SequenceNumber smallest_seq = 100, SequenceNumber largest_seq = 100) { assert(level < vstorage_->num_levels()); FileMetaData* f = new FileMetaData; f->fd = FileDescriptor(file_number, path_id, file_size); f->smallest = InternalKey(smallest, smallest_seq, kTypeValue); f->largest = InternalKey(largest, largest_seq, kTypeValue); f->smallest_seqno = smallest_seq; f->largest_seqno = largest_seq; f->compensated_file_size = file_size; f->refs = 0; vstorage_->AddFile(level, f); files_.emplace_back(f); file_map_.insert({file_number, {f, level}}); } void SetCompactionInputFilesLevels(int level_count, int start_level) { input_files_.resize(level_count); for (int i = 0; i < level_count; ++i) { input_files_[i].level = start_level + i; } compaction_level_start_ = start_level; } void AddToCompactionFiles(uint32_t file_number) { auto iter = file_map_.find(file_number); assert(iter != file_map_.end()); int level = iter->second.second; assert(level < vstorage_->num_levels()); input_files_[level - compaction_level_start_].files.emplace_back( iter->second.first); } void UpdateVersionStorageInfo() { vstorage_->CalculateBaseBytes(ioptions_, mutable_cf_options_); vstorage_->UpdateFilesByCompactionPri(mutable_cf_options_); vstorage_->UpdateNumNonEmptyLevels(); vstorage_->GenerateFileIndexer(); vstorage_->GenerateLevelFilesBrief(); vstorage_->ComputeCompactionScore(mutable_cf_options_, fifo_options_); vstorage_->GenerateLevel0NonOverlapping(); vstorage_->SetFinalized(); } }; TEST_F(CompactionPickerTest, Empty) { NewVersionStorage(6, kCompactionStyleLevel); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() == nullptr); } TEST_F(CompactionPickerTest, Single) { NewVersionStorage(6, kCompactionStyleLevel); mutable_cf_options_.level0_file_num_compaction_trigger = 2; Add(0, 1U, "p", "q"); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() == nullptr); } TEST_F(CompactionPickerTest, Level0Trigger) { NewVersionStorage(6, kCompactionStyleLevel); mutable_cf_options_.level0_file_num_compaction_trigger = 2; Add(0, 1U, "150", "200"); Add(0, 2U, "200", "250"); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); } TEST_F(CompactionPickerTest, Level1Trigger) { NewVersionStorage(6, kCompactionStyleLevel); Add(1, 66U, "150", "200", 1000000000U); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(1U, compaction->num_input_files(0)); ASSERT_EQ(66U, compaction->input(0, 0)->fd.GetNumber()); } TEST_F(CompactionPickerTest, Level1Trigger2) { NewVersionStorage(6, kCompactionStyleLevel); Add(1, 66U, "150", "200", 1000000001U); Add(1, 88U, "201", "300", 1000000000U); Add(2, 6U, "150", "179", 1000000000U); Add(2, 7U, "180", "220", 1000000000U); Add(2, 8U, "221", "300", 1000000000U); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(1U, compaction->num_input_files(0)); ASSERT_EQ(2U, compaction->num_input_files(1)); ASSERT_EQ(66U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(6U, compaction->input(1, 0)->fd.GetNumber()); ASSERT_EQ(7U, compaction->input(1, 1)->fd.GetNumber()); } TEST_F(CompactionPickerTest, LevelMaxScore) { NewVersionStorage(6, kCompactionStyleLevel); mutable_cf_options_.target_file_size_base = 10000000; mutable_cf_options_.target_file_size_multiplier = 10; Add(0, 1U, "150", "200", 1000000000U); // Level 1 score 1.2 Add(1, 66U, "150", "200", 6000000U); Add(1, 88U, "201", "300", 6000000U); // Level 2 score 1.8. File 7 is the largest. Should be picked Add(2, 6U, "150", "179", 60000000U); Add(2, 7U, "180", "220", 60000001U); Add(2, 8U, "221", "300", 60000000U); // Level 3 score slightly larger than 1 Add(3, 26U, "150", "170", 260000000U); Add(3, 27U, "171", "179", 260000000U); Add(3, 28U, "191", "220", 260000000U); Add(3, 29U, "221", "300", 260000000U); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(1U, compaction->num_input_files(0)); ASSERT_EQ(7U, compaction->input(0, 0)->fd.GetNumber()); } TEST_F(CompactionPickerTest, NeedsCompactionLevel) { const int kLevels = 6; const int kFileCount = 20; for (int level = 0; level < kLevels - 1; ++level) { NewVersionStorage(kLevels, kCompactionStyleLevel); uint64_t file_size = vstorage_->MaxBytesForLevel(level) * 2 / kFileCount; for (int file_count = 1; file_count <= kFileCount; ++file_count) { // start a brand new version in each test. NewVersionStorage(kLevels, kCompactionStyleLevel); for (int i = 0; i < file_count; ++i) { Add(level, i, ToString((i + 100) * 1000).c_str(), ToString((i + 100) * 1000 + 999).c_str(), file_size, 0, i * 100, i * 100 + 99); } UpdateVersionStorageInfo(); ASSERT_EQ(vstorage_->CompactionScoreLevel(0), level); ASSERT_EQ(level_compaction_picker.NeedsCompaction(vstorage_.get()), vstorage_->CompactionScore(0) >= 1); // release the version storage DeleteVersionStorage(); } } } TEST_F(CompactionPickerTest, Level0TriggerDynamic) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 200; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200"); Add(0, 2U, "200", "250"); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); ASSERT_EQ(1, static_cast(compaction->num_input_levels())); ASSERT_EQ(num_levels - 1, compaction->output_level()); } TEST_F(CompactionPickerTest, Level0TriggerDynamic2) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 200; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200"); Add(0, 2U, "200", "250"); Add(num_levels - 1, 3U, "200", "250", 300U); UpdateVersionStorageInfo(); ASSERT_EQ(vstorage_->base_level(), num_levels - 2); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); ASSERT_EQ(1, static_cast(compaction->num_input_levels())); ASSERT_EQ(num_levels - 2, compaction->output_level()); } TEST_F(CompactionPickerTest, Level0TriggerDynamic3) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 200; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200"); Add(0, 2U, "200", "250"); Add(num_levels - 1, 3U, "200", "250", 300U); Add(num_levels - 1, 4U, "300", "350", 3000U); UpdateVersionStorageInfo(); ASSERT_EQ(vstorage_->base_level(), num_levels - 3); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); ASSERT_EQ(1, static_cast(compaction->num_input_levels())); ASSERT_EQ(num_levels - 3, compaction->output_level()); } TEST_F(CompactionPickerTest, Level0TriggerDynamic4) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 200; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200"); Add(0, 2U, "200", "250"); Add(num_levels - 1, 3U, "200", "250", 300U); Add(num_levels - 1, 4U, "300", "350", 3000U); Add(num_levels - 3, 5U, "150", "180", 3U); Add(num_levels - 3, 6U, "181", "300", 3U); Add(num_levels - 3, 7U, "400", "450", 3U); UpdateVersionStorageInfo(); ASSERT_EQ(vstorage_->base_level(), num_levels - 3); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); ASSERT_EQ(2U, compaction->num_input_files(1)); ASSERT_EQ(num_levels - 3, compaction->level(1)); ASSERT_EQ(5U, compaction->input(1, 0)->fd.GetNumber()); ASSERT_EQ(6U, compaction->input(1, 1)->fd.GetNumber()); ASSERT_EQ(2, static_cast(compaction->num_input_levels())); ASSERT_EQ(num_levels - 3, compaction->output_level()); } TEST_F(CompactionPickerTest, LevelTriggerDynamic4) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 200; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200"); Add(num_levels - 1, 3U, "200", "250", 300U); Add(num_levels - 1, 4U, "300", "350", 3000U); Add(num_levels - 1, 4U, "400", "450", 3U); Add(num_levels - 2, 5U, "150", "180", 300U); Add(num_levels - 2, 6U, "181", "350", 500U); Add(num_levels - 2, 7U, "400", "450", 200U); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(1U, compaction->num_input_files(0)); ASSERT_EQ(6U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->num_input_files(1)); ASSERT_EQ(3U, compaction->input(1, 0)->fd.GetNumber()); ASSERT_EQ(4U, compaction->input(1, 1)->fd.GetNumber()); ASSERT_EQ(2U, compaction->num_input_levels()); ASSERT_EQ(num_levels - 1, compaction->output_level()); } // Universal and FIFO Compactions are not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE TEST_F(CompactionPickerTest, NeedsCompactionUniversal) { NewVersionStorage(1, kCompactionStyleUniversal); UniversalCompactionPicker universal_compaction_picker( ioptions_, &icmp_); // must return false when there's no files. ASSERT_EQ(universal_compaction_picker.NeedsCompaction(vstorage_.get()), false); UpdateVersionStorageInfo(); // verify the trigger given different number of L0 files. for (int i = 1; i <= mutable_cf_options_.level0_file_num_compaction_trigger * 2; ++i) { NewVersionStorage(1, kCompactionStyleUniversal); Add(0, i, ToString((i + 100) * 1000).c_str(), ToString((i + 100) * 1000 + 999).c_str(), 1000000, 0, i * 100, i * 100 + 99); UpdateVersionStorageInfo(); ASSERT_EQ(level_compaction_picker.NeedsCompaction(vstorage_.get()), vstorage_->CompactionScore(0) >= 1); } } // Tests if the files can be trivially moved in multi level // universal compaction when allow_trivial_move option is set // In this test as the input files overlaps, they cannot // be trivially moved. TEST_F(CompactionPickerTest, CannotTrivialMoveUniversal) { const uint64_t kFileSize = 100000; ioptions_.compaction_options_universal.allow_trivial_move = true; NewVersionStorage(1, kCompactionStyleUniversal); UniversalCompactionPicker universal_compaction_picker(ioptions_, &icmp_); // must return false when there's no files. ASSERT_EQ(universal_compaction_picker.NeedsCompaction(vstorage_.get()), false); NewVersionStorage(3, kCompactionStyleUniversal); Add(0, 1U, "150", "200", kFileSize, 0, 500, 550); Add(0, 2U, "201", "250", kFileSize, 0, 401, 450); Add(0, 4U, "260", "300", kFileSize, 0, 260, 300); Add(1, 5U, "100", "151", kFileSize, 0, 200, 251); Add(1, 3U, "301", "350", kFileSize, 0, 101, 150); Add(2, 6U, "120", "200", kFileSize, 0, 20, 100); UpdateVersionStorageInfo(); std::unique_ptr compaction( universal_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(!compaction->is_trivial_move()); } // Tests if the files can be trivially moved in multi level // universal compaction when allow_trivial_move option is set // In this test as the input files doesn't overlaps, they should // be trivially moved. TEST_F(CompactionPickerTest, AllowsTrivialMoveUniversal) { const uint64_t kFileSize = 100000; ioptions_.compaction_options_universal.allow_trivial_move = true; UniversalCompactionPicker universal_compaction_picker(ioptions_, &icmp_); NewVersionStorage(3, kCompactionStyleUniversal); Add(0, 1U, "150", "200", kFileSize, 0, 500, 550); Add(0, 2U, "201", "250", kFileSize, 0, 401, 450); Add(0, 4U, "260", "300", kFileSize, 0, 260, 300); Add(1, 5U, "010", "080", kFileSize, 0, 200, 251); Add(2, 3U, "301", "350", kFileSize, 0, 101, 150); UpdateVersionStorageInfo(); std::unique_ptr compaction( universal_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction->is_trivial_move()); } TEST_F(CompactionPickerTest, NeedsCompactionFIFO) { NewVersionStorage(1, kCompactionStyleFIFO); const int kFileCount = mutable_cf_options_.level0_file_num_compaction_trigger * 3; const uint64_t kFileSize = 100000; const uint64_t kMaxSize = kFileSize * kFileCount / 2; fifo_options_.max_table_files_size = kMaxSize; ioptions_.compaction_options_fifo = fifo_options_; FIFOCompactionPicker fifo_compaction_picker(ioptions_, &icmp_); UpdateVersionStorageInfo(); // must return false when there's no files. ASSERT_EQ(fifo_compaction_picker.NeedsCompaction(vstorage_.get()), false); // verify whether compaction is needed based on the current // size of L0 files. uint64_t current_size = 0; for (int i = 1; i <= kFileCount; ++i) { NewVersionStorage(1, kCompactionStyleFIFO); Add(0, i, ToString((i + 100) * 1000).c_str(), ToString((i + 100) * 1000 + 999).c_str(), kFileSize, 0, i * 100, i * 100 + 99); current_size += kFileSize; UpdateVersionStorageInfo(); ASSERT_EQ(level_compaction_picker.NeedsCompaction(vstorage_.get()), vstorage_->CompactionScore(0) >= 1); } } #endif // ROCKSDB_LITE // This test exhibits the bug where we don't properly reset parent_index in // PickCompaction() TEST_F(CompactionPickerTest, ParentIndexResetBug) { int num_levels = ioptions_.num_levels; mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 200; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200"); // <- marked for compaction Add(1, 3U, "400", "500", 600); // <- this one needs compacting Add(2, 4U, "150", "200"); Add(2, 5U, "201", "210"); Add(2, 6U, "300", "310"); Add(2, 7U, "400", "500"); // <- being compacted vstorage_->LevelFiles(2)[3]->being_compacted = true; vstorage_->LevelFiles(0)[0]->marked_for_compaction = true; UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); } // This test checks ExpandWhileOverlapping() by having overlapping user keys // ranges (with different sequence numbers) in the input files. TEST_F(CompactionPickerTest, OverlappingUserKeys) { NewVersionStorage(6, kCompactionStyleLevel); Add(1, 1U, "100", "150", 1U); // Overlapping user keys Add(1, 2U, "200", "400", 1U); Add(1, 3U, "400", "500", 1000000000U, 0, 0); Add(2, 4U, "600", "700", 1U); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(1U, compaction->num_input_levels()); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(2U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(3U, compaction->input(0, 1)->fd.GetNumber()); } TEST_F(CompactionPickerTest, OverlappingUserKeys2) { NewVersionStorage(6, kCompactionStyleLevel); // Overlapping user keys on same level and output level Add(1, 1U, "200", "400", 1000000000U); Add(1, 2U, "400", "500", 1U, 0, 0); Add(2, 3U, "400", "600", 1U); // The following file is not in the compaction despite overlapping user keys Add(2, 4U, "600", "700", 1U, 0, 0); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_levels()); ASSERT_EQ(2U, compaction->num_input_files(0)); ASSERT_EQ(1U, compaction->num_input_files(1)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); ASSERT_EQ(3U, compaction->input(1, 0)->fd.GetNumber()); } TEST_F(CompactionPickerTest, OverlappingUserKeys3) { NewVersionStorage(6, kCompactionStyleLevel); // Chain of overlapping user key ranges (forces ExpandWhileOverlapping() to // expand multiple times) Add(1, 1U, "100", "150", 1U); Add(1, 2U, "150", "200", 1U, 0, 0); Add(1, 3U, "200", "250", 1000000000U, 0, 0); Add(1, 4U, "250", "300", 1U, 0, 0); Add(1, 5U, "300", "350", 1U, 0, 0); // Output level overlaps with the beginning and the end of the chain Add(2, 6U, "050", "100", 1U); Add(2, 7U, "350", "400", 1U); UpdateVersionStorageInfo(); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); ASSERT_EQ(2U, compaction->num_input_levels()); ASSERT_EQ(5U, compaction->num_input_files(0)); ASSERT_EQ(2U, compaction->num_input_files(1)); ASSERT_EQ(1U, compaction->input(0, 0)->fd.GetNumber()); ASSERT_EQ(2U, compaction->input(0, 1)->fd.GetNumber()); ASSERT_EQ(3U, compaction->input(0, 2)->fd.GetNumber()); ASSERT_EQ(4U, compaction->input(0, 3)->fd.GetNumber()); ASSERT_EQ(5U, compaction->input(0, 4)->fd.GetNumber()); ASSERT_EQ(6U, compaction->input(1, 0)->fd.GetNumber()); ASSERT_EQ(7U, compaction->input(1, 1)->fd.GetNumber()); } TEST_F(CompactionPickerTest, NotScheduleL1IfL0WithHigherPri1) { NewVersionStorage(6, kCompactionStyleLevel); mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 900000000U; // 6 L0 files, score 3. Add(0, 1U, "000", "400", 1U); Add(0, 2U, "001", "400", 1U, 0, 0); Add(0, 3U, "001", "400", 1000000000U, 0, 0); Add(0, 31U, "001", "400", 1000000000U, 0, 0); Add(0, 32U, "001", "400", 1000000000U, 0, 0); Add(0, 33U, "001", "400", 1000000000U, 0, 0); // L1 total size 2GB, score 2.2. If one file being comapcted, score 1.1. Add(1, 4U, "050", "300", 1000000000U, 0, 0); file_map_[4u].first->being_compacted = true; Add(1, 5U, "301", "350", 1000000000U, 0, 0); // Output level overlaps with the beginning and the end of the chain Add(2, 6U, "050", "100", 1U); Add(2, 7U, "300", "400", 1U); // No compaction should be scheduled, if L0 has higher priority than L1 // but L0->L1 compaction is blocked by a file in L1 being compacted. UpdateVersionStorageInfo(); ASSERT_EQ(0, vstorage_->CompactionScoreLevel(0)); ASSERT_EQ(1, vstorage_->CompactionScoreLevel(1)); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() == nullptr); } TEST_F(CompactionPickerTest, NotScheduleL1IfL0WithHigherPri2) { NewVersionStorage(6, kCompactionStyleLevel); mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 900000000U; // 6 L0 files, score 3. Add(0, 1U, "000", "400", 1U); Add(0, 2U, "001", "400", 1U, 0, 0); Add(0, 3U, "001", "400", 1000000000U, 0, 0); Add(0, 31U, "001", "400", 1000000000U, 0, 0); Add(0, 32U, "001", "400", 1000000000U, 0, 0); Add(0, 33U, "001", "400", 1000000000U, 0, 0); // L1 total size 2GB, score 2.2. If one file being comapcted, score 1.1. Add(1, 4U, "050", "300", 1000000000U, 0, 0); Add(1, 5U, "301", "350", 1000000000U, 0, 0); // Output level overlaps with the beginning and the end of the chain Add(2, 6U, "050", "100", 1U); Add(2, 7U, "300", "400", 1U); // If no file in L1 being compacted, L0->L1 compaction will be scheduled. UpdateVersionStorageInfo(); // being_compacted flag is cleared here. ASSERT_EQ(0, vstorage_->CompactionScoreLevel(0)); ASSERT_EQ(1, vstorage_->CompactionScoreLevel(1)); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); } TEST_F(CompactionPickerTest, NotScheduleL1IfL0WithHigherPri3) { NewVersionStorage(6, kCompactionStyleLevel); mutable_cf_options_.level0_file_num_compaction_trigger = 2; mutable_cf_options_.max_bytes_for_level_base = 900000000U; // 6 L0 files, score 3. Add(0, 1U, "000", "400", 1U); Add(0, 2U, "001", "400", 1U, 0, 0); Add(0, 3U, "001", "400", 1000000000U, 0, 0); Add(0, 31U, "001", "400", 1000000000U, 0, 0); Add(0, 32U, "001", "400", 1000000000U, 0, 0); Add(0, 33U, "001", "400", 1000000000U, 0, 0); // L1 score more than 6. Add(1, 4U, "050", "300", 1000000000U, 0, 0); file_map_[4u].first->being_compacted = true; Add(1, 5U, "301", "350", 1000000000U, 0, 0); Add(1, 51U, "351", "400", 6000000000U, 0, 0); // Output level overlaps with the beginning and the end of the chain Add(2, 6U, "050", "100", 1U); Add(2, 7U, "300", "400", 1U); // If score in L1 is larger than L0, L1 compaction goes through despite // there is pending L0 compaction. UpdateVersionStorageInfo(); ASSERT_EQ(1, vstorage_->CompactionScoreLevel(0)); ASSERT_EQ(0, vstorage_->CompactionScoreLevel(1)); std::unique_ptr compaction(level_compaction_picker.PickCompaction( cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_)); ASSERT_TRUE(compaction.get() != nullptr); } TEST_F(CompactionPickerTest, EstimateCompactionBytesNeeded1) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = false; mutable_cf_options_.level0_file_num_compaction_trigger = 3; mutable_cf_options_.max_bytes_for_level_base = 1000; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200", 200); Add(0, 2U, "150", "200", 200); Add(0, 3U, "150", "200", 200); // Level 1 is over target by 200 Add(1, 4U, "400", "500", 600); Add(1, 5U, "600", "700", 600); // Level 2 is less than target 10000 even added size of level 1 Add(2, 6U, "150", "200", 2500); Add(2, 7U, "201", "210", 2000); Add(2, 8U, "300", "310", 2500); Add(2, 9U, "400", "500", 2500); // Level 3 exceeds target 100,000 of 1000 Add(3, 10U, "400", "500", 101000); // Level 4 exceeds target 1,000,000 of 500 after adding size from level 3 Add(4, 11U, "400", "500", 999500); Add(5, 11U, "400", "500", 8000000); UpdateVersionStorageInfo(); ASSERT_EQ(2200u + 11000u + 5500u, vstorage_->estimated_compaction_needed_bytes()); } TEST_F(CompactionPickerTest, EstimateCompactionBytesNeeded2) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = false; mutable_cf_options_.level0_file_num_compaction_trigger = 3; mutable_cf_options_.max_bytes_for_level_base = 1000; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); Add(0, 1U, "150", "200", 200); Add(0, 2U, "150", "200", 200); Add(0, 4U, "150", "200", 200); Add(0, 5U, "150", "200", 200); Add(0, 6U, "150", "200", 200); // Level 1 is over target by Add(1, 7U, "400", "500", 200); Add(1, 8U, "600", "700", 200); // Level 2 is less than target 10000 even added size of level 1 Add(2, 9U, "150", "200", 9500); Add(3, 10U, "400", "500", 101000); UpdateVersionStorageInfo(); ASSERT_EQ(1400u + 4400u + 11000u, vstorage_->estimated_compaction_needed_bytes()); } TEST_F(CompactionPickerTest, EstimateCompactionBytesNeededDynamicLevel) { int num_levels = ioptions_.num_levels; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.level0_file_num_compaction_trigger = 3; mutable_cf_options_.max_bytes_for_level_base = 1000; mutable_cf_options_.max_bytes_for_level_multiplier = 10; NewVersionStorage(num_levels, kCompactionStyleLevel); // Set Last level size 50000 // num_levels - 1 target 5000 // num_levels - 2 is base level with taret 500 Add(num_levels - 1, 10U, "400", "500", 50000); Add(0, 1U, "150", "200", 200); Add(0, 2U, "150", "200", 200); Add(0, 4U, "150", "200", 200); Add(0, 5U, "150", "200", 200); Add(0, 6U, "150", "200", 200); // num_levels - 3 is over target by 100 + 1000 Add(num_levels - 3, 7U, "400", "500", 300); Add(num_levels - 3, 8U, "600", "700", 300); // Level 2 is over target by 1100 + 100 Add(num_levels - 2, 9U, "150", "200", 5100); UpdateVersionStorageInfo(); ASSERT_EQ(1600u + 12100u + 13200u, vstorage_->estimated_compaction_needed_bytes()); } TEST_F(CompactionPickerTest, IsBottommostLevelTest) { // case 1: Higher levels are empty NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "a", "m"); Add(0, 2U, "c", "z"); Add(1, 3U, "d", "e"); Add(1, 4U, "l", "p"); Add(2, 5U, "g", "i"); Add(2, 6U, "x", "z"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(2, 1); AddToCompactionFiles(3U); AddToCompactionFiles(5U); bool result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_TRUE(result); // case 2: Higher levels have no overlap NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "a", "m"); Add(0, 2U, "c", "z"); Add(1, 3U, "d", "e"); Add(1, 4U, "l", "p"); Add(2, 5U, "g", "i"); Add(2, 6U, "x", "z"); Add(3, 7U, "k", "p"); Add(3, 8U, "t", "w"); Add(4, 9U, "a", "b"); Add(5, 10U, "c", "cc"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(2, 1); AddToCompactionFiles(3U); AddToCompactionFiles(5U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_TRUE(result); // case 3.1: Higher levels (level 3) have overlap NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "a", "m"); Add(0, 2U, "c", "z"); Add(1, 3U, "d", "e"); Add(1, 4U, "l", "p"); Add(2, 5U, "g", "i"); Add(2, 6U, "x", "z"); Add(3, 7U, "e", "g"); Add(3, 8U, "h", "k"); Add(4, 9U, "a", "b"); Add(5, 10U, "c", "cc"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(2, 1); AddToCompactionFiles(3U); AddToCompactionFiles(5U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_FALSE(result); // case 3.2: Higher levels (level 5) have overlap DeleteVersionStorage(); NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "a", "m"); Add(0, 2U, "c", "z"); Add(1, 3U, "d", "e"); Add(1, 4U, "l", "p"); Add(2, 5U, "g", "i"); Add(2, 6U, "x", "z"); Add(3, 7U, "j", "k"); Add(3, 8U, "l", "m"); Add(4, 9U, "a", "b"); Add(5, 10U, "c", "cc"); Add(5, 11U, "h", "k"); Add(5, 12U, "y", "yy"); Add(5, 13U, "z", "zz"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(2, 1); AddToCompactionFiles(3U); AddToCompactionFiles(5U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_FALSE(result); // case 3.3: Higher levels (level 5) have overlap, but it's only overlapping // one key ("d") NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "a", "m"); Add(0, 2U, "c", "z"); Add(1, 3U, "d", "e"); Add(1, 4U, "l", "p"); Add(2, 5U, "g", "i"); Add(2, 6U, "x", "z"); Add(3, 7U, "j", "k"); Add(3, 8U, "l", "m"); Add(4, 9U, "a", "b"); Add(5, 10U, "c", "cc"); Add(5, 11U, "ccc", "d"); Add(5, 12U, "y", "yy"); Add(5, 13U, "z", "zz"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(2, 1); AddToCompactionFiles(3U); AddToCompactionFiles(5U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_FALSE(result); // Level 0 files overlap NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "s", "t"); Add(0, 2U, "a", "m"); Add(0, 3U, "b", "z"); Add(0, 4U, "e", "f"); Add(5, 10U, "y", "z"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(1, 0); AddToCompactionFiles(1U); AddToCompactionFiles(2U); AddToCompactionFiles(3U); AddToCompactionFiles(4U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_FALSE(result); // Level 0 files don't overlap NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "s", "t"); Add(0, 2U, "a", "m"); Add(0, 3U, "b", "k"); Add(0, 4U, "e", "f"); Add(5, 10U, "y", "z"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(1, 0); AddToCompactionFiles(1U); AddToCompactionFiles(2U); AddToCompactionFiles(3U); AddToCompactionFiles(4U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_TRUE(result); // Level 1 files overlap NewVersionStorage(6, kCompactionStyleLevel); Add(0, 1U, "s", "t"); Add(0, 2U, "a", "m"); Add(0, 3U, "b", "k"); Add(0, 4U, "e", "f"); Add(1, 5U, "a", "m"); Add(1, 6U, "n", "o"); Add(1, 7U, "w", "y"); Add(5, 10U, "y", "z"); UpdateVersionStorageInfo(); SetCompactionInputFilesLevels(2, 0); AddToCompactionFiles(1U); AddToCompactionFiles(2U); AddToCompactionFiles(3U); AddToCompactionFiles(4U); AddToCompactionFiles(5U); AddToCompactionFiles(6U); AddToCompactionFiles(7U); result = Compaction::TEST_IsBottommostLevel(2, vstorage_.get(), input_files_); ASSERT_FALSE(result); DeleteVersionStorage(); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/comparator_db_test.cc000066400000000000000000000305531267530634000177330ustar00rootroot00000000000000// Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include "memtable/stl_wrappers.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "util/hash.h" #include "util/kv_map.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" #include "utilities/merge_operators.h" using std::unique_ptr; namespace rocksdb { namespace { static const Comparator* comparator; class KVIter : public Iterator { public: explicit KVIter(const stl_wrappers::KVMap* map) : map_(map), iter_(map_->end()) {} virtual bool Valid() const override { return iter_ != map_->end(); } virtual void SeekToFirst() override { iter_ = map_->begin(); } virtual void SeekToLast() override { if (map_->empty()) { iter_ = map_->end(); } else { iter_ = map_->find(map_->rbegin()->first); } } virtual void Seek(const Slice& k) override { iter_ = map_->lower_bound(k.ToString()); } virtual void Next() override { ++iter_; } virtual void Prev() override { if (iter_ == map_->begin()) { iter_ = map_->end(); return; } --iter_; } virtual Slice key() const override { return iter_->first; } virtual Slice value() const override { return iter_->second; } virtual Status status() const override { return Status::OK(); } private: const stl_wrappers::KVMap* const map_; stl_wrappers::KVMap::const_iterator iter_; }; void AssertItersEqual(Iterator* iter1, Iterator* iter2) { ASSERT_EQ(iter1->Valid(), iter2->Valid()); if (iter1->Valid()) { ASSERT_EQ(iter1->key().ToString(), iter2->key().ToString()); ASSERT_EQ(iter1->value().ToString(), iter2->value().ToString()); } } // Measuring operations on DB (expect to be empty). // source_strings are candidate keys void DoRandomIteraratorTest(DB* db, std::vector source_strings, Random* rnd, int num_writes, int num_iter_ops, int num_trigger_flush) { stl_wrappers::KVMap map((stl_wrappers::LessOfComparator(comparator))); for (int i = 0; i < num_writes; i++) { if (num_trigger_flush > 0 && i != 0 && i % num_trigger_flush == 0) { db->Flush(FlushOptions()); } int type = rnd->Uniform(2); int index = rnd->Uniform(static_cast(source_strings.size())); auto& key = source_strings[index]; switch (type) { case 0: // put map[key] = key; ASSERT_OK(db->Put(WriteOptions(), key, key)); break; case 1: // delete if (map.find(key) != map.end()) { map.erase(key); } ASSERT_OK(db->Delete(WriteOptions(), key)); break; default: assert(false); } } std::unique_ptr iter(db->NewIterator(ReadOptions())); std::unique_ptr result_iter(new KVIter(&map)); bool is_valid = false; for (int i = 0; i < num_iter_ops; i++) { // Random walk and make sure iter and result_iter returns the // same key and value int type = rnd->Uniform(6); ASSERT_OK(iter->status()); switch (type) { case 0: // Seek to First iter->SeekToFirst(); result_iter->SeekToFirst(); break; case 1: // Seek to last iter->SeekToLast(); result_iter->SeekToLast(); break; case 2: { // Seek to random key auto key_idx = rnd->Uniform(static_cast(source_strings.size())); auto key = source_strings[key_idx]; iter->Seek(key); result_iter->Seek(key); break; } case 3: // Next if (is_valid) { iter->Next(); result_iter->Next(); } else { continue; } break; case 4: // Prev if (is_valid) { iter->Prev(); result_iter->Prev(); } else { continue; } break; default: { assert(type == 5); auto key_idx = rnd->Uniform(static_cast(source_strings.size())); auto key = source_strings[key_idx]; std::string result; auto status = db->Get(ReadOptions(), key, &result); if (map.find(key) == map.end()) { ASSERT_TRUE(status.IsNotFound()); } else { ASSERT_EQ(map[key], result); } break; } } AssertItersEqual(iter.get(), result_iter.get()); is_valid = iter->Valid(); } } class DoubleComparator : public Comparator { public: DoubleComparator() {} virtual const char* Name() const override { return "DoubleComparator"; } virtual int Compare(const Slice& a, const Slice& b) const override { #ifndef CYGWIN double da = std::stod(a.ToString()); double db = std::stod(b.ToString()); #else double da = std::strtod(a.ToString().c_str(), 0 /* endptr */); double db = std::strtod(a.ToString().c_str(), 0 /* endptr */); #endif if (da == db) { return a.compare(b); } else if (da > db) { return 1; } else { return -1; } } virtual void FindShortestSeparator(std::string* start, const Slice& limit) const override {} virtual void FindShortSuccessor(std::string* key) const override {} }; class HashComparator : public Comparator { public: HashComparator() {} virtual const char* Name() const override { return "HashComparator"; } virtual int Compare(const Slice& a, const Slice& b) const override { uint32_t ha = Hash(a.data(), a.size(), 66); uint32_t hb = Hash(b.data(), b.size(), 66); if (ha == hb) { return a.compare(b); } else if (ha > hb) { return 1; } else { return -1; } } virtual void FindShortestSeparator(std::string* start, const Slice& limit) const override {} virtual void FindShortSuccessor(std::string* key) const override {} }; class TwoStrComparator : public Comparator { public: TwoStrComparator() {} virtual const char* Name() const override { return "TwoStrComparator"; } virtual int Compare(const Slice& a, const Slice& b) const override { assert(a.size() >= 2); assert(b.size() >= 2); size_t size_a1 = static_cast(a[0]); size_t size_b1 = static_cast(b[0]); size_t size_a2 = static_cast(a[1]); size_t size_b2 = static_cast(b[1]); assert(size_a1 + size_a2 + 2 == a.size()); assert(size_b1 + size_b2 + 2 == b.size()); Slice a1 = Slice(a.data() + 2, size_a1); Slice b1 = Slice(b.data() + 2, size_b1); Slice a2 = Slice(a.data() + 2 + size_a1, size_a2); Slice b2 = Slice(b.data() + 2 + size_b1, size_b2); if (a1 != b1) { return a1.compare(b1); } return a2.compare(b2); } virtual void FindShortestSeparator(std::string* start, const Slice& limit) const override {} virtual void FindShortSuccessor(std::string* key) const override {} }; } // namespace class ComparatorDBTest : public testing::Test { private: std::string dbname_; Env* env_; DB* db_; Options last_options_; std::unique_ptr comparator_guard; public: ComparatorDBTest() : env_(Env::Default()), db_(nullptr) { comparator = BytewiseComparator(); dbname_ = test::TmpDir() + "/comparator_db_test"; EXPECT_OK(DestroyDB(dbname_, last_options_)); } ~ComparatorDBTest() { delete db_; EXPECT_OK(DestroyDB(dbname_, last_options_)); comparator = BytewiseComparator(); } DB* GetDB() { return db_; } void SetOwnedComparator(const Comparator* cmp) { comparator_guard.reset(cmp); comparator = cmp; last_options_.comparator = cmp; } // Return the current option configuration. Options* GetOptions() { return &last_options_; } void DestroyAndReopen() { // Destroy using last options Destroy(); ASSERT_OK(TryReopen()); } void Destroy() { delete db_; db_ = nullptr; ASSERT_OK(DestroyDB(dbname_, last_options_)); } Status TryReopen() { delete db_; db_ = nullptr; last_options_.create_if_missing = true; return DB::Open(last_options_, dbname_, &db_); } }; TEST_F(ComparatorDBTest, Bytewise) { for (int rand_seed = 301; rand_seed < 306; rand_seed++) { DestroyAndReopen(); Random rnd(rand_seed); DoRandomIteraratorTest(GetDB(), {"a", "b", "c", "d", "e", "f", "g", "h", "i"}, &rnd, 8, 100, 3); } } TEST_F(ComparatorDBTest, SimpleSuffixReverseComparator) { SetOwnedComparator(new test::SimpleSuffixReverseComparator()); for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) { Options* opt = GetOptions(); opt->comparator = comparator; DestroyAndReopen(); Random rnd(rnd_seed); std::vector source_strings; std::vector source_prefixes; // Randomly generate 5 prefixes for (int i = 0; i < 5; i++) { source_prefixes.push_back(test::RandomHumanReadableString(&rnd, 8)); } for (int j = 0; j < 20; j++) { int prefix_index = rnd.Uniform(static_cast(source_prefixes.size())); std::string key = source_prefixes[prefix_index] + test::RandomHumanReadableString(&rnd, rnd.Uniform(8)); source_strings.push_back(key); } DoRandomIteraratorTest(GetDB(), source_strings, &rnd, 30, 600, 66); } } TEST_F(ComparatorDBTest, Uint64Comparator) { SetOwnedComparator(test::Uint64Comparator()); for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) { Options* opt = GetOptions(); opt->comparator = comparator; DestroyAndReopen(); Random rnd(rnd_seed); Random64 rnd64(rnd_seed); std::vector source_strings; // Randomly generate source keys for (int i = 0; i < 100; i++) { uint64_t r = rnd64.Next(); std::string str; str.resize(8); memcpy(&str[0], static_cast(&r), 8); source_strings.push_back(str); } DoRandomIteraratorTest(GetDB(), source_strings, &rnd, 200, 1000, 66); } } TEST_F(ComparatorDBTest, DoubleComparator) { SetOwnedComparator(new DoubleComparator()); for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) { Options* opt = GetOptions(); opt->comparator = comparator; DestroyAndReopen(); Random rnd(rnd_seed); std::vector source_strings; // Randomly generate source keys for (int i = 0; i < 100; i++) { uint32_t r = rnd.Next(); uint32_t divide_order = rnd.Uniform(8); double to_divide = 1.0; for (uint32_t j = 0; j < divide_order; j++) { to_divide *= 10.0; } source_strings.push_back(ToString(r / to_divide)); } DoRandomIteraratorTest(GetDB(), source_strings, &rnd, 200, 1000, 66); } } TEST_F(ComparatorDBTest, HashComparator) { SetOwnedComparator(new HashComparator()); for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) { Options* opt = GetOptions(); opt->comparator = comparator; DestroyAndReopen(); Random rnd(rnd_seed); std::vector source_strings; // Randomly generate source keys for (int i = 0; i < 100; i++) { source_strings.push_back(test::RandomKey(&rnd, 8)); } DoRandomIteraratorTest(GetDB(), source_strings, &rnd, 200, 1000, 66); } } TEST_F(ComparatorDBTest, TwoStrComparator) { SetOwnedComparator(new TwoStrComparator()); for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) { Options* opt = GetOptions(); opt->comparator = comparator; DestroyAndReopen(); Random rnd(rnd_seed); std::vector source_strings; // Randomly generate source keys for (int i = 0; i < 100; i++) { std::string str; uint32_t size1 = rnd.Uniform(8); uint32_t size2 = rnd.Uniform(8); str.append(1, static_cast(size1)); str.append(1, static_cast(size2)); str.append(test::RandomKey(&rnd, size1)); str.append(test::RandomKey(&rnd, size2)); source_strings.push_back(str); } DoRandomIteraratorTest(GetDB(), source_strings, &rnd, 200, 1000, 66); } } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/convenience.cc000066400000000000000000000016401267530634000163470ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2012 Facebook. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. #ifndef ROCKSDB_LITE #include "rocksdb/convenience.h" #include "db/db_impl.h" namespace rocksdb { void CancelAllBackgroundWork(DB* db, bool wait) { (dynamic_cast(db))->CancelAllBackgroundWork(wait); } Status DeleteFilesInRange(DB* db, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) { return (dynamic_cast(db)) ->DeleteFilesInRange(column_family, begin, end); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/corruption_test.cc000066400000000000000000000350041267530634000173170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "rocksdb/db.h" #include #include #include #include #include #include "rocksdb/cache.h" #include "rocksdb/env.h" #include "rocksdb/table.h" #include "rocksdb/write_batch.h" #include "db/db_impl.h" #include "db/filename.h" #include "db/log_format.h" #include "db/version_set.h" #include "util/logging.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { static const int kValueSize = 1000; class CorruptionTest : public testing::Test { public: test::ErrorEnv env_; std::string dbname_; shared_ptr tiny_cache_; Options options_; DB* db_; CorruptionTest() { tiny_cache_ = NewLRUCache(100); options_.env = &env_; dbname_ = test::TmpDir() + "/corruption_test"; DestroyDB(dbname_, options_); db_ = nullptr; options_.create_if_missing = true; BlockBasedTableOptions table_options; table_options.block_size_deviation = 0; // make unit test pass for now options_.table_factory.reset(NewBlockBasedTableFactory(table_options)); Reopen(); options_.create_if_missing = false; } ~CorruptionTest() { delete db_; DestroyDB(dbname_, Options()); } void CloseDb() { delete db_; db_ = nullptr; } Status TryReopen(Options* options = nullptr) { delete db_; db_ = nullptr; Options opt = (options ? *options : options_); opt.env = &env_; opt.arena_block_size = 4096; BlockBasedTableOptions table_options; table_options.block_cache = tiny_cache_; table_options.block_size_deviation = 0; opt.table_factory.reset(NewBlockBasedTableFactory(table_options)); return DB::Open(opt, dbname_, &db_); } void Reopen(Options* options = nullptr) { ASSERT_OK(TryReopen(options)); } void RepairDB() { delete db_; db_ = nullptr; ASSERT_OK(::rocksdb::RepairDB(dbname_, options_)); } void Build(int n, int flush_every = 0) { std::string key_space, value_space; WriteBatch batch; for (int i = 0; i < n; i++) { if (flush_every != 0 && i != 0 && i % flush_every == 0) { DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_FlushMemTable(); } //if ((i % 100) == 0) fprintf(stderr, "@ %d of %d\n", i, n); Slice key = Key(i, &key_space); batch.Clear(); batch.Put(key, Value(i, &value_space)); ASSERT_OK(db_->Write(WriteOptions(), &batch)); } } void Check(int min_expected, int max_expected) { uint64_t next_expected = 0; uint64_t missed = 0; int bad_keys = 0; int bad_values = 0; int correct = 0; std::string value_space; // Do not verify checksums. If we verify checksums then the // db itself will raise errors because data is corrupted. // Instead, we want the reads to be successful and this test // will detect whether the appropriate corruptions have // occurred. Iterator* iter = db_->NewIterator(ReadOptions(false, true)); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { uint64_t key; Slice in(iter->key()); if (!ConsumeDecimalNumber(&in, &key) || !in.empty() || key < next_expected) { bad_keys++; continue; } missed += (key - next_expected); next_expected = key + 1; if (iter->value() != Value(static_cast(key), &value_space)) { bad_values++; } else { correct++; } } delete iter; fprintf(stderr, "expected=%d..%d; got=%d; bad_keys=%d; bad_values=%d; missed=%llu\n", min_expected, max_expected, correct, bad_keys, bad_values, static_cast(missed)); ASSERT_LE(min_expected, correct); ASSERT_GE(max_expected, correct); } void CorruptFile(const std::string& fname, int offset, int bytes_to_corrupt) { struct stat sbuf; if (stat(fname.c_str(), &sbuf) != 0) { const char* msg = strerror(errno); ASSERT_TRUE(false) << fname << ": " << msg; } if (offset < 0) { // Relative to end of file; make it absolute if (-offset > sbuf.st_size) { offset = 0; } else { offset = static_cast(sbuf.st_size + offset); } } if (offset > sbuf.st_size) { offset = static_cast(sbuf.st_size); } if (offset + bytes_to_corrupt > sbuf.st_size) { bytes_to_corrupt = static_cast(sbuf.st_size - offset); } // Do it std::string contents; Status s = ReadFileToString(Env::Default(), fname, &contents); ASSERT_TRUE(s.ok()) << s.ToString(); for (int i = 0; i < bytes_to_corrupt; i++) { contents[i + offset] ^= 0x80; } s = WriteStringToFile(Env::Default(), contents, fname); ASSERT_TRUE(s.ok()) << s.ToString(); } void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) { // Pick file to corrupt std::vector filenames; ASSERT_OK(env_.GetChildren(dbname_, &filenames)); uint64_t number; FileType type; std::string fname; int picked_number = -1; for (size_t i = 0; i < filenames.size(); i++) { if (ParseFileName(filenames[i], &number, &type) && type == filetype && static_cast(number) > picked_number) { // Pick latest file fname = dbname_ + "/" + filenames[i]; picked_number = static_cast(number); } } ASSERT_TRUE(!fname.empty()) << filetype; CorruptFile(fname, offset, bytes_to_corrupt); } // corrupts exactly one file at level `level`. if no file found at level, // asserts void CorruptTableFileAtLevel(int level, int offset, int bytes_to_corrupt) { std::vector metadata; db_->GetLiveFilesMetaData(&metadata); for (const auto& m : metadata) { if (m.level == level) { CorruptFile(dbname_ + "/" + m.name, offset, bytes_to_corrupt); return; } } ASSERT_TRUE(false) << "no file found at level"; } int Property(const std::string& name) { std::string property; int result; if (db_->GetProperty(name, &property) && sscanf(property.c_str(), "%d", &result) == 1) { return result; } else { return -1; } } // Return the ith key Slice Key(int i, std::string* storage) { char buf[100]; snprintf(buf, sizeof(buf), "%016d", i); storage->assign(buf, strlen(buf)); return Slice(*storage); } // Return the value to associate with the specified key Slice Value(int k, std::string* storage) { if (k == 0) { // Ugh. Random seed of 0 used to produce no entropy. This code // preserves the implementation that was in place when all of the // magic values in this file were picked. *storage = std::string(kValueSize, ' '); return Slice(*storage); } else { Random r(k); return test::RandomString(&r, kValueSize, storage); } } }; TEST_F(CorruptionTest, Recovery) { Build(100); Check(100, 100); #ifdef OS_WIN // On Wndows OS Disk cache does not behave properly // We do not call FlushBuffers on every Flush. If we do not close // the log file prior to the corruption we end up with the first // block not corrupted but only the second. However, under the debugger // things work just fine but never pass when running normally // For that reason people may want to run with unbuffered I/O. That option // is not available for WAL though. CloseDb(); #endif Corrupt(kLogFile, 19, 1); // WriteBatch tag for first record Corrupt(kLogFile, log::kBlockSize + 1000, 1); // Somewhere in second block ASSERT_TRUE(!TryReopen().ok()); options_.paranoid_checks = false; Reopen(&options_); // The 64 records in the first two log blocks are completely lost. Check(36, 36); } TEST_F(CorruptionTest, RecoverWriteError) { env_.writable_file_error_ = true; Status s = TryReopen(); ASSERT_TRUE(!s.ok()); } TEST_F(CorruptionTest, NewFileErrorDuringWrite) { // Do enough writing to force minor compaction env_.writable_file_error_ = true; const int num = static_cast(3 + (Options().write_buffer_size / kValueSize)); std::string value_storage; Status s; bool failed = false; for (int i = 0; i < num; i++) { WriteBatch batch; batch.Put("a", Value(100, &value_storage)); s = db_->Write(WriteOptions(), &batch); if (!s.ok()) { failed = true; } ASSERT_TRUE(!failed || !s.ok()); } ASSERT_TRUE(!s.ok()); ASSERT_GE(env_.num_writable_file_errors_, 1); env_.writable_file_error_ = false; Reopen(); } TEST_F(CorruptionTest, TableFile) { Build(100); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_FlushMemTable(); dbi->TEST_CompactRange(0, nullptr, nullptr); dbi->TEST_CompactRange(1, nullptr, nullptr); Corrupt(kTableFile, 100, 1); Check(99, 99); } TEST_F(CorruptionTest, TableFileIndexData) { Options options; // very big, we'll trigger flushes manually options.write_buffer_size = 100 * 1024 * 1024; Reopen(&options); // build 2 tables, flush at 5000 Build(10000, 5000); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_FlushMemTable(); // corrupt an index block of an entire file Corrupt(kTableFile, -2000, 500); Reopen(); // one full file should be readable, since only one was corrupted // the other file should be fully non-readable, since index was corrupted Check(5000, 5000); } TEST_F(CorruptionTest, MissingDescriptor) { Build(1000); RepairDB(); Reopen(); Check(1000, 1000); } TEST_F(CorruptionTest, SequenceNumberRecovery) { ASSERT_OK(db_->Put(WriteOptions(), "foo", "v1")); ASSERT_OK(db_->Put(WriteOptions(), "foo", "v2")); ASSERT_OK(db_->Put(WriteOptions(), "foo", "v3")); ASSERT_OK(db_->Put(WriteOptions(), "foo", "v4")); ASSERT_OK(db_->Put(WriteOptions(), "foo", "v5")); RepairDB(); Reopen(); std::string v; ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); ASSERT_EQ("v5", v); // Write something. If sequence number was not recovered properly, // it will be hidden by an earlier write. ASSERT_OK(db_->Put(WriteOptions(), "foo", "v6")); ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); ASSERT_EQ("v6", v); Reopen(); ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); ASSERT_EQ("v6", v); } TEST_F(CorruptionTest, CorruptedDescriptor) { ASSERT_OK(db_->Put(WriteOptions(), "foo", "hello")); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_FlushMemTable(); dbi->TEST_CompactRange(0, nullptr, nullptr); Corrupt(kDescriptorFile, 0, 1000); Status s = TryReopen(); ASSERT_TRUE(!s.ok()); RepairDB(); Reopen(); std::string v; ASSERT_OK(db_->Get(ReadOptions(), "foo", &v)); ASSERT_EQ("hello", v); } TEST_F(CorruptionTest, CompactionInputError) { Options options; Reopen(&options); Build(10); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_FlushMemTable(); dbi->TEST_CompactRange(0, nullptr, nullptr); dbi->TEST_CompactRange(1, nullptr, nullptr); ASSERT_EQ(1, Property("rocksdb.num-files-at-level2")); Corrupt(kTableFile, 100, 1); Check(9, 9); // Force compactions by writing lots of values Build(10000); Check(10000, 10000); } TEST_F(CorruptionTest, CompactionInputErrorParanoid) { Options options; options.paranoid_checks = true; options.write_buffer_size = 131072; options.max_write_buffer_number = 2; Reopen(&options); DBImpl* dbi = reinterpret_cast(db_); // Fill levels >= 1 for (int level = 1; level < dbi->NumberLevels(); level++) { dbi->Put(WriteOptions(), "", "begin"); dbi->Put(WriteOptions(), "~", "end"); dbi->TEST_FlushMemTable(); for (int comp_level = 0; comp_level < dbi->NumberLevels() - level; ++comp_level) { dbi->TEST_CompactRange(comp_level, nullptr, nullptr); } } Reopen(&options); dbi = reinterpret_cast(db_); Build(10); dbi->TEST_FlushMemTable(); dbi->TEST_WaitForCompact(); ASSERT_EQ(1, Property("rocksdb.num-files-at-level0")); CorruptTableFileAtLevel(0, 100, 1); Check(9, 9); // Write must eventually fail because of corrupted table Status s; std::string tmp1, tmp2; bool failed = false; for (int i = 0; i < 10000; i++) { s = db_->Put(WriteOptions(), Key(i, &tmp1), Value(i, &tmp2)); if (!s.ok()) { failed = true; } // if one write failed, every subsequent write must fail, too ASSERT_TRUE(!failed || !s.ok()) << "write did not fail in a corrupted db"; } ASSERT_TRUE(!s.ok()) << "write did not fail in corrupted paranoid db"; } TEST_F(CorruptionTest, UnrelatedKeys) { Build(10); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_FlushMemTable(); Corrupt(kTableFile, 100, 1); std::string tmp1, tmp2; ASSERT_OK(db_->Put(WriteOptions(), Key(1000, &tmp1), Value(1000, &tmp2))); std::string v; ASSERT_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v)); ASSERT_EQ(Value(1000, &tmp2).ToString(), v); dbi->TEST_FlushMemTable(); ASSERT_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v)); ASSERT_EQ(Value(1000, &tmp2).ToString(), v); } TEST_F(CorruptionTest, FileSystemStateCorrupted) { for (int iter = 0; iter < 2; ++iter) { Options options; options.paranoid_checks = true; options.create_if_missing = true; Reopen(&options); Build(10); ASSERT_OK(db_->Flush(FlushOptions())); DBImpl* dbi = reinterpret_cast(db_); std::vector metadata; dbi->GetLiveFilesMetaData(&metadata); ASSERT_GT(metadata.size(), size_t(0)); std::string filename = dbname_ + metadata[0].name; delete db_; db_ = nullptr; if (iter == 0) { // corrupt file size unique_ptr file; env_.NewWritableFile(filename, &file, EnvOptions()); file->Append(Slice("corrupted sst")); file.reset(); } else { // delete the file env_.DeleteFile(filename); } Status x = TryReopen(&options); ASSERT_TRUE(x.IsCorruption()); DestroyDB(dbname_, options_); Reopen(&options); } } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as RepairDB() is not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/cuckoo_table_db_test.cc000066400000000000000000000230571267530634000202170ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "db/db_impl.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "table/meta_blocks.h" #include "table/cuckoo_table_factory.h" #include "table/cuckoo_table_reader.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class CuckooTableDBTest : public testing::Test { private: std::string dbname_; Env* env_; DB* db_; public: CuckooTableDBTest() : env_(Env::Default()) { dbname_ = test::TmpDir() + "/cuckoo_table_db_test"; EXPECT_OK(DestroyDB(dbname_, Options())); db_ = nullptr; Reopen(); } ~CuckooTableDBTest() { delete db_; EXPECT_OK(DestroyDB(dbname_, Options())); } Options CurrentOptions() { Options options; options.table_factory.reset(NewCuckooTableFactory()); options.memtable_factory.reset(NewHashLinkListRepFactory(4, 0, 3, true)); options.allow_mmap_reads = true; options.create_if_missing = true; return options; } DBImpl* dbfull() { return reinterpret_cast(db_); } // The following util methods are copied from plain_table_db_test. void Reopen(Options* options = nullptr) { delete db_; db_ = nullptr; Options opts; if (options != nullptr) { opts = *options; } else { opts = CurrentOptions(); opts.create_if_missing = true; } ASSERT_OK(DB::Open(opts, dbname_, &db_)); } Status Put(const Slice& k, const Slice& v) { return db_->Put(WriteOptions(), k, v); } Status Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); } std::string Get(const std::string& k) { ReadOptions options; std::string result; Status s = db_->Get(options, k, &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } int NumTableFilesAtLevel(int level) { std::string property; EXPECT_TRUE(db_->GetProperty( "rocksdb.num-files-at-level" + NumberToString(level), &property)); return atoi(property.c_str()); } // Return spread of files per level std::string FilesPerLevel() { std::string result; size_t last_non_zero_offset = 0; for (int level = 0; level < db_->NumberLevels(); level++) { int f = NumTableFilesAtLevel(level); char buf[100]; snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); result += buf; if (f > 0) { last_non_zero_offset = result.size(); } } result.resize(last_non_zero_offset); return result; } }; TEST_F(CuckooTableDBTest, Flush) { // Try with empty DB first. ASSERT_TRUE(dbfull() != nullptr); ASSERT_EQ("NOT_FOUND", Get("key2")); // Add some values to db. Options options = CurrentOptions(); Reopen(&options); ASSERT_OK(Put("key1", "v1")); ASSERT_OK(Put("key2", "v2")); ASSERT_OK(Put("key3", "v3")); dbfull()->TEST_FlushMemTable(); TablePropertiesCollection ptc; reinterpret_cast(dbfull())->GetPropertiesOfAllTables(&ptc); ASSERT_EQ(1U, ptc.size()); ASSERT_EQ(3U, ptc.begin()->second->num_entries); ASSERT_EQ("1", FilesPerLevel()); ASSERT_EQ("v1", Get("key1")); ASSERT_EQ("v2", Get("key2")); ASSERT_EQ("v3", Get("key3")); ASSERT_EQ("NOT_FOUND", Get("key4")); // Now add more keys and flush. ASSERT_OK(Put("key4", "v4")); ASSERT_OK(Put("key5", "v5")); ASSERT_OK(Put("key6", "v6")); dbfull()->TEST_FlushMemTable(); reinterpret_cast(dbfull())->GetPropertiesOfAllTables(&ptc); ASSERT_EQ(2U, ptc.size()); auto row = ptc.begin(); ASSERT_EQ(3U, row->second->num_entries); ASSERT_EQ(3U, (++row)->second->num_entries); ASSERT_EQ("2", FilesPerLevel()); ASSERT_EQ("v1", Get("key1")); ASSERT_EQ("v2", Get("key2")); ASSERT_EQ("v3", Get("key3")); ASSERT_EQ("v4", Get("key4")); ASSERT_EQ("v5", Get("key5")); ASSERT_EQ("v6", Get("key6")); ASSERT_OK(Delete("key6")); ASSERT_OK(Delete("key5")); ASSERT_OK(Delete("key4")); dbfull()->TEST_FlushMemTable(); reinterpret_cast(dbfull())->GetPropertiesOfAllTables(&ptc); ASSERT_EQ(3U, ptc.size()); row = ptc.begin(); ASSERT_EQ(3U, row->second->num_entries); ASSERT_EQ(3U, (++row)->second->num_entries); ASSERT_EQ(3U, (++row)->second->num_entries); ASSERT_EQ("3", FilesPerLevel()); ASSERT_EQ("v1", Get("key1")); ASSERT_EQ("v2", Get("key2")); ASSERT_EQ("v3", Get("key3")); ASSERT_EQ("NOT_FOUND", Get("key4")); ASSERT_EQ("NOT_FOUND", Get("key5")); ASSERT_EQ("NOT_FOUND", Get("key6")); } TEST_F(CuckooTableDBTest, FlushWithDuplicateKeys) { Options options = CurrentOptions(); Reopen(&options); ASSERT_OK(Put("key1", "v1")); ASSERT_OK(Put("key2", "v2")); ASSERT_OK(Put("key1", "v3")); // Duplicate dbfull()->TEST_FlushMemTable(); TablePropertiesCollection ptc; reinterpret_cast(dbfull())->GetPropertiesOfAllTables(&ptc); ASSERT_EQ(1U, ptc.size()); ASSERT_EQ(2U, ptc.begin()->second->num_entries); ASSERT_EQ("1", FilesPerLevel()); ASSERT_EQ("v3", Get("key1")); ASSERT_EQ("v2", Get("key2")); } namespace { static std::string Key(int i) { char buf[100]; snprintf(buf, sizeof(buf), "key_______%06d", i); return std::string(buf); } static std::string Uint64Key(uint64_t i) { std::string str; str.resize(8); memcpy(&str[0], static_cast(&i), 8); return str; } } // namespace. TEST_F(CuckooTableDBTest, Uint64Comparator) { Options options = CurrentOptions(); options.comparator = test::Uint64Comparator(); Reopen(&options); ASSERT_OK(Put(Uint64Key(1), "v1")); ASSERT_OK(Put(Uint64Key(2), "v2")); ASSERT_OK(Put(Uint64Key(3), "v3")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get(Uint64Key(1))); ASSERT_EQ("v2", Get(Uint64Key(2))); ASSERT_EQ("v3", Get(Uint64Key(3))); ASSERT_EQ("NOT_FOUND", Get(Uint64Key(4))); // Add more keys. ASSERT_OK(Delete(Uint64Key(2))); // Delete. dbfull()->TEST_FlushMemTable(); ASSERT_OK(Put(Uint64Key(3), "v0")); // Update. ASSERT_OK(Put(Uint64Key(4), "v4")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get(Uint64Key(1))); ASSERT_EQ("NOT_FOUND", Get(Uint64Key(2))); ASSERT_EQ("v0", Get(Uint64Key(3))); ASSERT_EQ("v4", Get(Uint64Key(4))); } TEST_F(CuckooTableDBTest, CompactionIntoMultipleFiles) { // Create a big L0 file and check it compacts into multiple files in L1. Options options = CurrentOptions(); options.write_buffer_size = 270 << 10; // Two SST files should be created, each containing 14 keys. // Number of buckets will be 16. Total size ~156 KB. options.target_file_size_base = 160 << 10; Reopen(&options); // Write 28 values, each 10016 B ~ 10KB for (int idx = 0; idx < 28; ++idx) { ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + idx))); } dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ("1", FilesPerLevel()); dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, true /* disallow trivial move */); ASSERT_EQ("0,2", FilesPerLevel()); for (int idx = 0; idx < 28; ++idx) { ASSERT_EQ(std::string(10000, 'a' + idx), Get(Key(idx))); } } TEST_F(CuckooTableDBTest, SameKeyInsertedInTwoDifferentFilesAndCompacted) { // Insert same key twice so that they go to different SST files. Then wait for // compaction and check if the latest value is stored and old value removed. Options options = CurrentOptions(); options.write_buffer_size = 100 << 10; // 100KB options.level0_file_num_compaction_trigger = 2; Reopen(&options); // Write 11 values, each 10016 B for (int idx = 0; idx < 11; ++idx) { ASSERT_OK(Put(Key(idx), std::string(10000, 'a'))); } dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ("1", FilesPerLevel()); // Generate one more file in level-0, and should trigger level-0 compaction for (int idx = 0; idx < 11; ++idx) { ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + idx))); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_CompactRange(0, nullptr, nullptr); ASSERT_EQ("0,1", FilesPerLevel()); for (int idx = 0; idx < 11; ++idx) { ASSERT_EQ(std::string(10000, 'a' + idx), Get(Key(idx))); } } TEST_F(CuckooTableDBTest, AdaptiveTable) { Options options = CurrentOptions(); // Write some keys using cuckoo table. options.table_factory.reset(NewCuckooTableFactory()); Reopen(&options); ASSERT_OK(Put("key1", "v1")); ASSERT_OK(Put("key2", "v2")); ASSERT_OK(Put("key3", "v3")); dbfull()->TEST_FlushMemTable(); // Write some keys using plain table. options.create_if_missing = false; options.table_factory.reset(NewPlainTableFactory()); Reopen(&options); ASSERT_OK(Put("key4", "v4")); ASSERT_OK(Put("key1", "v5")); dbfull()->TEST_FlushMemTable(); // Write some keys using block based table. std::shared_ptr block_based_factory( NewBlockBasedTableFactory()); options.table_factory.reset(NewAdaptiveTableFactory(block_based_factory)); Reopen(&options); ASSERT_OK(Put("key5", "v6")); ASSERT_OK(Put("key2", "v7")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v5", Get("key1")); ASSERT_EQ("v7", Get("key2")); ASSERT_EQ("v3", Get("key3")); ASSERT_EQ("v4", Get("key4")); ASSERT_EQ("v6", Get("key5")); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as Cuckoo table is not supported in ROCKSDB_LITE\n"); return 0; } #endif // ROCKSDB_LITE rocksdb-4.5.1/db/db_bench.cc000066400000000000000000004366261267530634000156170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #ifndef GFLAGS #include int main() { fprintf(stderr, "Please install gflags to run rocksdb tools\n"); return 1; } #else #ifdef NUMA #include #include #endif #ifndef OS_WIN #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include "db/db_impl.h" #include "db/version_set.h" #include "rocksdb/options.h" #include "rocksdb/cache.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "rocksdb/write_batch.h" #include "rocksdb/slice.h" #include "rocksdb/filter_policy.h" #include "rocksdb/rate_limiter.h" #include "rocksdb/slice_transform.h" #include "rocksdb/perf_context.h" #include "rocksdb/utilities/flashcache.h" #include "rocksdb/utilities/transaction.h" #include "rocksdb/utilities/transaction_db.h" #include "rocksdb/utilities/optimistic_transaction_db.h" #include "port/port.h" #include "port/stack_trace.h" #include "util/crc32c.h" #include "util/compression.h" #include "util/histogram.h" #include "util/mutexlock.h" #include "util/random.h" #include "util/string_util.h" #include "util/statistics.h" #include "util/testutil.h" #include "util/xxhash.h" #include "hdfs/env_hdfs.h" #include "utilities/merge_operators.h" #ifdef OS_WIN #include // open/close #endif using GFLAGS::ParseCommandLineFlags; using GFLAGS::RegisterFlagValidator; using GFLAGS::SetUsageMessage; DEFINE_string(benchmarks, "fillseq," "fillsync," "fillrandom," "overwrite," "readrandom," "newiterator," "newiteratorwhilewriting," "seekrandom," "seekrandomwhilewriting," "seekrandomwhilemerging," "readseq," "readreverse," "compact," "readrandom," "multireadrandom," "readseq," "readtocache," "readreverse," "readwhilewriting," "readwhilemerging," "readrandomwriterandom," "updaterandom," "randomwithverify," "fill100K," "crc32c," "xxhash," "compress," "uncompress," "acquireload," "fillseekseq," "randomtransaction," "randomreplacekeys", "Comma-separated list of operations to run in the specified" " order. Available benchmarks:\n" "\tfillseq -- write N values in sequential key" " order in async mode\n" "\tfillrandom -- write N values in random key order in async" " mode\n" "\toverwrite -- overwrite N values in random key order in" " async mode\n" "\tfillsync -- write N/100 values in random key order in " "sync mode\n" "\tfill100K -- write N/1000 100K values in random order in" " async mode\n" "\tdeleteseq -- delete N keys in sequential order\n" "\tdeleterandom -- delete N keys in random order\n" "\treadseq -- read N times sequentially\n" "\treadtocache -- 1 thread reading database sequentially\n" "\treadreverse -- read N times in reverse order\n" "\treadrandom -- read N times in random order\n" "\treadmissing -- read N missing keys in random order\n" "\treadwhilewriting -- 1 writer, N threads doing random " "reads\n" "\treadwhilemerging -- 1 merger, N threads doing random " "reads\n" "\treadrandomwriterandom -- N threads doing random-read, " "random-write\n" "\tprefixscanrandom -- prefix scan N times in random order\n" "\tupdaterandom -- N threads doing read-modify-write for random " "keys\n" "\tappendrandom -- N threads doing read-modify-write with " "growing values\n" "\tmergerandom -- same as updaterandom/appendrandom using merge" " operator. " "Must be used with merge_operator\n" "\treadrandommergerandom -- perform N random read-or-merge " "operations. Must be used with merge_operator\n" "\tnewiterator -- repeated iterator creation\n" "\tseekrandom -- N random seeks, call Next seek_nexts times " "per seek\n" "\tseekrandomwhilewriting -- seekrandom and 1 thread doing " "overwrite\n" "\tseekrandomwhilemerging -- seekrandom and 1 thread doing " "merge\n" "\tcrc32c -- repeated crc32c of 4K of data\n" "\txxhash -- repeated xxHash of 4K of data\n" "\tacquireload -- load N*1000 times\n" "\tfillseekseq -- write N values in sequential key, then read " "them by seeking to each key\n" "\trandomtransaction -- execute N random transactions and " "verify correctness\n" "\trandomreplacekeys -- randomly replaces N keys by deleting " "the old version and putting the new version\n\n" "Meta operations:\n" "\tcompact -- Compact the entire DB\n" "\tstats -- Print DB stats\n" "\tlevelstats -- Print the number of files and bytes per level\n" "\tsstables -- Print sstable info\n" "\theapprofile -- Dump a heap profile (if supported by this" " port)\n"); DEFINE_int64(num, 1000000, "Number of key/values to place in database"); DEFINE_int64(numdistinct, 1000, "Number of distinct keys to use. Used in RandomWithVerify to " "read/write on fewer keys so that gets are more likely to find the" " key and puts are more likely to update the same key"); DEFINE_int64(merge_keys, -1, "Number of distinct keys to use for MergeRandom and " "ReadRandomMergeRandom. " "If negative, there will be FLAGS_num keys."); DEFINE_int32(num_column_families, 1, "Number of Column Families to use."); DEFINE_int32( num_hot_column_families, 0, "Number of Hot Column Families. If more than 0, only write to this " "number of column families. After finishing all the writes to them, " "create new set of column families and insert to them. Only used " "when num_column_families > 1."); DEFINE_int64(reads, -1, "Number of read operations to do. " "If negative, do FLAGS_num reads."); DEFINE_int32(bloom_locality, 0, "Control bloom filter probes locality"); DEFINE_int64(seed, 0, "Seed base for random number generators. " "When 0 it is deterministic."); DEFINE_int32(threads, 1, "Number of concurrent threads to run."); DEFINE_int32(duration, 0, "Time in seconds for the random-ops tests to run." " When 0 then num & reads determine the test duration"); DEFINE_int32(value_size, 100, "Size of each value"); DEFINE_int32(seek_nexts, 0, "How many times to call Next() after Seek() in " "fillseekseq, seekrandom, seekrandomwhilewriting and " "seekrandomwhilemerging"); DEFINE_bool(reverse_iterator, false, "When true use Prev rather than Next for iterators that do " "Seek and then Next"); DEFINE_bool(use_uint64_comparator, false, "use Uint64 user comparator"); DEFINE_int64(batch_size, 1, "Batch size"); static bool ValidateKeySize(const char* flagname, int32_t value) { return true; } static bool ValidateUint32Range(const char* flagname, uint64_t value) { if (value > std::numeric_limits::max()) { fprintf(stderr, "Invalid value for --%s: %lu, overflow\n", flagname, (unsigned long)value); return false; } return true; } DEFINE_int32(key_size, 16, "size of each key"); DEFINE_int32(num_multi_db, 0, "Number of DBs used in the benchmark. 0 means single DB."); DEFINE_double(compression_ratio, 0.5, "Arrange to generate values that shrink" " to this fraction of their original size after compression"); DEFINE_double(read_random_exp_range, 0.0, "Read random's key will be generated using distribution of " "num * exp(-r) where r is uniform number from 0 to this value. " "The larger the number is, the more skewed the reads are. " "Only used in readrandom and multireadrandom benchmarks."); DEFINE_bool(histogram, false, "Print histogram of operation timings"); DEFINE_bool(enable_numa, false, "Make operations aware of NUMA architecture and bind memory " "and cpus corresponding to nodes together. In NUMA, memory " "in same node as CPUs are closer when compared to memory in " "other nodes. Reads can be faster when the process is bound to " "CPU and memory of same node. Use \"$numactl --hardware\" command " "to see NUMA memory architecture."); DEFINE_int64(db_write_buffer_size, rocksdb::Options().db_write_buffer_size, "Number of bytes to buffer in all memtables before compacting"); DEFINE_int64(write_buffer_size, rocksdb::Options().write_buffer_size, "Number of bytes to buffer in memtable before compacting"); DEFINE_int32(max_write_buffer_number, rocksdb::Options().max_write_buffer_number, "The number of in-memory memtables. Each memtable is of size" "write_buffer_size."); DEFINE_int32(min_write_buffer_number_to_merge, rocksdb::Options().min_write_buffer_number_to_merge, "The minimum number of write buffers that will be merged together" "before writing to storage. This is cheap because it is an" "in-memory merge. If this feature is not enabled, then all these" "write buffers are flushed to L0 as separate files and this " "increases read amplification because a get request has to check" " in all of these files. Also, an in-memory merge may result in" " writing less data to storage if there are duplicate records " " in each of these individual write buffers."); DEFINE_int32(max_write_buffer_number_to_maintain, rocksdb::Options().max_write_buffer_number_to_maintain, "The total maximum number of write buffers to maintain in memory " "including copies of buffers that have already been flushed. " "Unlike max_write_buffer_number, this parameter does not affect " "flushing. This controls the minimum amount of write history " "that will be available in memory for conflict checking when " "Transactions are used. If this value is too low, some " "transactions may fail at commit time due to not being able to " "determine whether there were any write conflicts. Setting this " "value to 0 will cause write buffers to be freed immediately " "after they are flushed. If this value is set to -1, " "'max_write_buffer_number' will be used."); DEFINE_int32(max_background_compactions, rocksdb::Options().max_background_compactions, "The maximum number of concurrent background compactions" " that can occur in parallel."); DEFINE_uint64(subcompactions, 1, "Maximum number of subcompactions to divide L0-L1 compactions " "into."); static const bool FLAGS_subcompactions_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_subcompactions, &ValidateUint32Range); DEFINE_int32(max_background_flushes, rocksdb::Options().max_background_flushes, "The maximum number of concurrent background flushes" " that can occur in parallel."); static rocksdb::CompactionStyle FLAGS_compaction_style_e; DEFINE_int32(compaction_style, (int32_t) rocksdb::Options().compaction_style, "style of compaction: level-based vs universal"); static rocksdb::CompactionPri FLAGS_compaction_pri_e; DEFINE_int32(compaction_pri, (int32_t)rocksdb::Options().compaction_pri, "priority of files to compaction: by size or by data age"); DEFINE_int32(universal_size_ratio, 0, "Percentage flexibility while comparing file size" " (for universal compaction only)."); DEFINE_int32(universal_min_merge_width, 0, "The minimum number of files in a" " single compaction run (for universal compaction only)."); DEFINE_int32(universal_max_merge_width, 0, "The max number of files to compact" " in universal style compaction"); DEFINE_int32(universal_max_size_amplification_percent, 0, "The max size amplification for universal style compaction"); DEFINE_int32(universal_compression_size_percent, -1, "The percentage of the database to compress for universal " "compaction. -1 means compress everything."); DEFINE_bool(universal_allow_trivial_move, false, "Allow trivial move in universal compaction."); DEFINE_int64(cache_size, -1, "Number of bytes to use as a cache of uncompressed" "data. Negative means use default settings."); DEFINE_bool(cache_index_and_filter_blocks, false, "Cache index/filter blocks in block cache."); DEFINE_int32(block_size, static_cast(rocksdb::BlockBasedTableOptions().block_size), "Number of bytes in a block."); DEFINE_int32(block_restart_interval, rocksdb::BlockBasedTableOptions().block_restart_interval, "Number of keys between restart points " "for delta encoding of keys."); DEFINE_int64(compressed_cache_size, -1, "Number of bytes to use as a cache of compressed data."); DEFINE_int64(row_cache_size, 0, "Number of bytes to use as a cache of individual rows" " (0 = disabled)."); DEFINE_int32(open_files, rocksdb::Options().max_open_files, "Maximum number of files to keep open at the same time" " (use default if == 0)"); DEFINE_int32(file_opening_threads, rocksdb::Options().max_file_opening_threads, "If open_files is set to -1, this option set the number of " "threads that will be used to open files during DB::Open()"); DEFINE_int32(new_table_reader_for_compaction_inputs, true, "If true, uses a separate file handle for compaction inputs"); DEFINE_int32(compaction_readahead_size, 0, "Compaction readahead size"); DEFINE_int32(random_access_max_buffer_size, 1024 * 1024, "Maximum windows randomaccess buffer size"); DEFINE_int32(writable_file_max_buffer_size, 1024 * 1024, "Maximum write buffer for Writable File"); DEFINE_int32(skip_table_builder_flush, false, "Skip flushing block in " "table builder "); DEFINE_int32(bloom_bits, -1, "Bloom filter bits per key. Negative means" " use default settings."); DEFINE_int32(memtable_bloom_bits, 0, "Bloom filter bits per key for memtable. " "Negative means no bloom filter."); DEFINE_bool(use_existing_db, false, "If true, do not destroy the existing" " database. If you set this flag and also specify a benchmark that" " wants a fresh database, that benchmark will fail."); DEFINE_bool(show_table_properties, false, "If true, then per-level table" " properties will be printed on every stats-interval when" " stats_interval is set and stats_per_interval is on."); DEFINE_string(db, "", "Use the db with the following name."); static bool ValidateCacheNumshardbits(const char* flagname, int32_t value) { if (value >= 20) { fprintf(stderr, "Invalid value for --%s: %d, must be < 20\n", flagname, value); return false; } return true; } DEFINE_int32(cache_numshardbits, -1, "Number of shards for the block cache" " is 2 ** cache_numshardbits. Negative means use default settings." " This is applied only if FLAGS_cache_size is non-negative."); DEFINE_bool(verify_checksum, false, "Verify checksum for every block read" " from storage"); DEFINE_bool(statistics, false, "Database statistics"); static class std::shared_ptr dbstats; DEFINE_int64(writes, -1, "Number of write operations to do. If negative, do" " --num reads."); DEFINE_bool(sync, false, "Sync all writes to disk"); DEFINE_bool(disable_data_sync, false, "If true, do not wait until data is" " synced to disk."); DEFINE_bool(use_fsync, false, "If true, issue fsync instead of fdatasync"); DEFINE_bool(disable_wal, false, "If true, do not write WAL for write."); DEFINE_string(wal_dir, "", "If not empty, use the given dir for WAL"); DEFINE_int32(num_levels, 7, "The total number of levels"); DEFINE_int64(target_file_size_base, 2 * 1048576, "Target file size at level-1"); DEFINE_int32(target_file_size_multiplier, 1, "A multiplier to compute target level-N file size (N >= 2)"); DEFINE_uint64(max_bytes_for_level_base, 10 * 1048576, "Max bytes for level-1"); DEFINE_bool(level_compaction_dynamic_level_bytes, false, "Whether level size base is dynamic"); DEFINE_int32(max_bytes_for_level_multiplier, 10, "A multiplier to compute max bytes for level-N (N >= 2)"); static std::vector FLAGS_max_bytes_for_level_multiplier_additional_v; DEFINE_string(max_bytes_for_level_multiplier_additional, "", "A vector that specifies additional fanout per level"); DEFINE_int32(level0_stop_writes_trigger, rocksdb::Options().level0_stop_writes_trigger, "Number of files in level-0" " that will trigger put stop."); DEFINE_int32(level0_slowdown_writes_trigger, rocksdb::Options().level0_slowdown_writes_trigger, "Number of files in level-0" " that will slow down writes."); DEFINE_int32(level0_file_num_compaction_trigger, rocksdb::Options().level0_file_num_compaction_trigger, "Number of files in level-0" " when compactions start"); static bool ValidateInt32Percent(const char* flagname, int32_t value) { if (value <= 0 || value>=100) { fprintf(stderr, "Invalid value for --%s: %d, 0< pct <100 \n", flagname, value); return false; } return true; } DEFINE_int32(readwritepercent, 90, "Ratio of reads to reads/writes (expressed" " as percentage) for the ReadRandomWriteRandom workload. The " "default value 90 means 90% operations out of all reads and writes" " operations are reads. In other words, 9 gets for every 1 put."); DEFINE_int32(mergereadpercent, 70, "Ratio of merges to merges&reads (expressed" " as percentage) for the ReadRandomMergeRandom workload. The" " default value 70 means 70% out of all read and merge operations" " are merges. In other words, 7 merges for every 3 gets."); DEFINE_int32(deletepercent, 2, "Percentage of deletes out of reads/writes/" "deletes (used in RandomWithVerify only). RandomWithVerify " "calculates writepercent as (100 - FLAGS_readwritepercent - " "deletepercent), so deletepercent must be smaller than (100 - " "FLAGS_readwritepercent)"); DEFINE_uint64(delete_obsolete_files_period_micros, 0, "Ignored. Left here for backward compatibility"); #ifndef ROCKSDB_LITE DEFINE_bool(optimistic_transaction_db, false, "Open a OptimisticTransactionDB instance. " "Required for randomtransaction benchmark."); DEFINE_bool(transaction_db, false, "Open a TransactionDB instance. " "Required for randomtransaction benchmark."); DEFINE_uint64(transaction_sets, 2, "Number of keys each transaction will " "modify (use in RandomTransaction only). Max: 9999"); DEFINE_bool(transaction_set_snapshot, false, "Setting to true will have each transaction call SetSnapshot()" " upon creation."); DEFINE_int32(transaction_sleep, 0, "Max microseconds to sleep in between " "reading and writing a value (used in RandomTransaction only). "); DEFINE_uint64(transaction_lock_timeout, 100, "If using a transaction_db, specifies the lock wait timeout in" " milliseconds before failing a transaction waiting on a lock"); #endif // ROCKSDB_LITE DEFINE_bool(compaction_measure_io_stats, false, "Measure times spents on I/Os while in compactions. "); namespace { enum rocksdb::CompressionType StringToCompressionType(const char* ctype) { assert(ctype); if (!strcasecmp(ctype, "none")) return rocksdb::kNoCompression; else if (!strcasecmp(ctype, "snappy")) return rocksdb::kSnappyCompression; else if (!strcasecmp(ctype, "zlib")) return rocksdb::kZlibCompression; else if (!strcasecmp(ctype, "bzip2")) return rocksdb::kBZip2Compression; else if (!strcasecmp(ctype, "lz4")) return rocksdb::kLZ4Compression; else if (!strcasecmp(ctype, "lz4hc")) return rocksdb::kLZ4HCCompression; else if (!strcasecmp(ctype, "zstd")) return rocksdb::kZSTDNotFinalCompression; fprintf(stdout, "Cannot parse compression type '%s'\n", ctype); return rocksdb::kSnappyCompression; //default value } std::string ColumnFamilyName(size_t i) { if (i == 0) { return rocksdb::kDefaultColumnFamilyName; } else { char name[100]; snprintf(name, sizeof(name), "column_family_name_%06zu", i); return std::string(name); } } } // namespace DEFINE_string(compression_type, "snappy", "Algorithm to use to compress the database"); static enum rocksdb::CompressionType FLAGS_compression_type_e = rocksdb::kSnappyCompression; DEFINE_int32(compression_level, -1, "Compression level. For zlib this should be -1 for the " "default level, or between 0 and 9."); static bool ValidateCompressionLevel(const char* flagname, int32_t value) { if (value < -1 || value > 9) { fprintf(stderr, "Invalid value for --%s: %d, must be between -1 and 9\n", flagname, value); return false; } return true; } static const bool FLAGS_compression_level_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_compression_level, &ValidateCompressionLevel); DEFINE_int32(min_level_to_compress, -1, "If non-negative, compression starts" " from this level. Levels with number < min_level_to_compress are" " not compressed. Otherwise, apply compression_type to " "all levels."); static bool ValidateTableCacheNumshardbits(const char* flagname, int32_t value) { if (0 >= value || value > 20) { fprintf(stderr, "Invalid value for --%s: %d, must be 0 < val <= 20\n", flagname, value); return false; } return true; } DEFINE_int32(table_cache_numshardbits, 4, ""); DEFINE_string(hdfs, "", "Name of hdfs environment"); // posix or hdfs environment static rocksdb::Env* FLAGS_env = rocksdb::Env::Default(); DEFINE_int64(stats_interval, 0, "Stats are reported every N operations when " "this is greater than zero. When 0 the interval grows over time."); DEFINE_int64(stats_interval_seconds, 0, "Report stats every N seconds. This " "overrides stats_interval when both are > 0."); DEFINE_int32(stats_per_interval, 0, "Reports additional stats per interval when" " this is greater than 0."); DEFINE_int64(report_interval_seconds, 0, "If greater than zero, it will write simple stats in CVS format " "to --report_file every N seconds"); DEFINE_string(report_file, "report.csv", "Filename where some simple stats are reported to (if " "--report_interval_seconds is bigger than 0)"); DEFINE_int32(thread_status_per_interval, 0, "Takes and report a snapshot of the current status of each thread" " when this is greater than 0."); DEFINE_int32(perf_level, 0, "Level of perf collection"); static bool ValidateRateLimit(const char* flagname, double value) { const double EPSILON = 1e-10; if ( value < -EPSILON ) { fprintf(stderr, "Invalid value for --%s: %12.6f, must be >= 0.0\n", flagname, value); return false; } return true; } DEFINE_double(soft_rate_limit, 0.0, "DEPRECATED"); DEFINE_double(hard_rate_limit, 0.0, "DEPRECATED"); DEFINE_uint64(soft_pending_compaction_bytes_limit, 64ull * 1024 * 1024 * 1024, "Slowdown writes if pending compaction bytes exceed this number"); DEFINE_uint64(hard_pending_compaction_bytes_limit, 128ull * 1024 * 1024 * 1024, "Stop writes if pending compaction bytes exceed this number"); DEFINE_uint64(delayed_write_rate, 8388608u, "Limited bytes allowed to DB when soft_rate_limit or " "level0_slowdown_writes_trigger triggers"); DEFINE_bool(allow_concurrent_memtable_write, false, "Allow multi-writers to update mem tables in parallel."); DEFINE_bool(enable_write_thread_adaptive_yield, false, "Use a yielding spin loop for brief writer thread waits."); DEFINE_uint64( write_thread_max_yield_usec, 100, "Maximum microseconds for enable_write_thread_adaptive_yield operation."); DEFINE_uint64(write_thread_slow_yield_usec, 3, "The threshold at which a slow yield is considered a signal that " "other processes or threads want the core."); DEFINE_int32(rate_limit_delay_max_milliseconds, 1000, "When hard_rate_limit is set then this is the max time a put will" " be stalled."); DEFINE_uint64(rate_limiter_bytes_per_sec, 0, "Set options.rate_limiter value."); DEFINE_uint64( benchmark_write_rate_limit, 0, "If non-zero, db_bench will rate-limit the writes going into RocksDB. This " "is the global rate in bytes/second."); DEFINE_int32(max_grandparent_overlap_factor, 10, "Control maximum bytes of " "overlaps in grandparent (i.e., level+2) before we stop building a" " single file in a level->level+1 compaction."); #ifndef ROCKSDB_LITE DEFINE_bool(readonly, false, "Run read only benchmarks."); #endif // ROCKSDB_LITE DEFINE_bool(disable_auto_compactions, false, "Do not auto trigger compactions"); DEFINE_int32(source_compaction_factor, 1, "Cap the size of data in level-K for" " a compaction run that compacts Level-K with Level-(K+1) (for" " K >= 1)"); DEFINE_uint64(wal_ttl_seconds, 0, "Set the TTL for the WAL Files in seconds."); DEFINE_uint64(wal_size_limit_MB, 0, "Set the size limit for the WAL Files" " in MB."); DEFINE_uint64(max_total_wal_size, 0, "Set total max WAL size"); DEFINE_bool(bufferedio, rocksdb::EnvOptions().use_os_buffer, "Allow buffered io using OS buffers"); DEFINE_bool(mmap_read, rocksdb::EnvOptions().use_mmap_reads, "Allow reads to occur via mmap-ing files"); DEFINE_bool(mmap_write, rocksdb::EnvOptions().use_mmap_writes, "Allow writes to occur via mmap-ing files"); DEFINE_bool(advise_random_on_open, rocksdb::Options().advise_random_on_open, "Advise random access on table file open"); DEFINE_string(compaction_fadvice, "NORMAL", "Access pattern advice when a file is compacted"); static auto FLAGS_compaction_fadvice_e = rocksdb::Options().access_hint_on_compaction_start; DEFINE_bool(disable_flashcache_for_background_threads, false, "Disable flashcache for background threads"); DEFINE_string(flashcache_dev, "", "Path to flashcache device"); DEFINE_bool(use_tailing_iterator, false, "Use tailing iterator to access a series of keys instead of get"); DEFINE_bool(use_adaptive_mutex, rocksdb::Options().use_adaptive_mutex, "Use adaptive mutex"); DEFINE_uint64(bytes_per_sync, rocksdb::Options().bytes_per_sync, "Allows OS to incrementally sync SST files to disk while they are" " being written, in the background. Issue one request for every" " bytes_per_sync written. 0 turns it off."); DEFINE_uint64(wal_bytes_per_sync, rocksdb::Options().wal_bytes_per_sync, "Allows OS to incrementally sync WAL files to disk while they are" " being written, in the background. Issue one request for every" " wal_bytes_per_sync written. 0 turns it off."); DEFINE_bool(filter_deletes, false, " On true, deletes use bloom-filter and drop" " the delete if key not present"); DEFINE_bool(use_single_deletes, true, "Use single deletes (used in RandomReplaceKeys only)."); DEFINE_double(stddev, 2000.0, "Standard deviation of normal distribution used for picking keys" " (used in RandomReplaceKeys only)."); DEFINE_int32(max_successive_merges, 0, "Maximum number of successive merge" " operations on a key in the memtable"); static bool ValidatePrefixSize(const char* flagname, int32_t value) { if (value < 0 || value>=2000000000) { fprintf(stderr, "Invalid value for --%s: %d. 0<= PrefixSize <=2000000000\n", flagname, value); return false; } return true; } DEFINE_int32(prefix_size, 0, "control the prefix size for HashSkipList and " "plain table"); DEFINE_int64(keys_per_prefix, 0, "control average number of keys generated " "per prefix, 0 means no special handling of the prefix, " "i.e. use the prefix comes with the generated random number."); DEFINE_bool(enable_io_prio, false, "Lower the background flush/compaction " "threads' IO priority"); DEFINE_bool(identity_as_first_hash, false, "the first hash function of cuckoo " "table becomes an identity function. This is only valid when key " "is 8 bytes"); enum RepFactory { kSkipList, kPrefixHash, kVectorRep, kHashLinkedList, kCuckoo }; namespace { enum RepFactory StringToRepFactory(const char* ctype) { assert(ctype); if (!strcasecmp(ctype, "skip_list")) return kSkipList; else if (!strcasecmp(ctype, "prefix_hash")) return kPrefixHash; else if (!strcasecmp(ctype, "vector")) return kVectorRep; else if (!strcasecmp(ctype, "hash_linkedlist")) return kHashLinkedList; else if (!strcasecmp(ctype, "cuckoo")) return kCuckoo; fprintf(stdout, "Cannot parse memreptable %s\n", ctype); return kSkipList; } } // namespace static enum RepFactory FLAGS_rep_factory; DEFINE_string(memtablerep, "skip_list", ""); DEFINE_int64(hash_bucket_count, 1024 * 1024, "hash bucket count"); DEFINE_bool(use_plain_table, false, "if use plain table " "instead of block-based table format"); DEFINE_bool(use_cuckoo_table, false, "if use cuckoo table format"); DEFINE_double(cuckoo_hash_ratio, 0.9, "Hash ratio for Cuckoo SST table."); DEFINE_bool(use_hash_search, false, "if use kHashSearch " "instead of kBinarySearch. " "This is valid if only we use BlockTable"); DEFINE_bool(use_block_based_filter, false, "if use kBlockBasedFilter " "instead of kFullFilter for filter block. " "This is valid if only we use BlockTable"); DEFINE_string(merge_operator, "", "The merge operator to use with the database." "If a new merge operator is specified, be sure to use fresh" " database The possible merge operators are defined in" " utilities/merge_operators.h"); DEFINE_int32(skip_list_lookahead, 0, "Used with skip_list memtablerep; try " "linear search first for this many steps from the previous " "position"); DEFINE_bool(report_file_operations, false, "if report number of file " "operations"); static const bool FLAGS_soft_rate_limit_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_soft_rate_limit, &ValidateRateLimit); static const bool FLAGS_hard_rate_limit_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_hard_rate_limit, &ValidateRateLimit); static const bool FLAGS_prefix_size_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_prefix_size, &ValidatePrefixSize); static const bool FLAGS_key_size_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_key_size, &ValidateKeySize); static const bool FLAGS_cache_numshardbits_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_cache_numshardbits, &ValidateCacheNumshardbits); static const bool FLAGS_readwritepercent_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_readwritepercent, &ValidateInt32Percent); DEFINE_int32(disable_seek_compaction, false, "Not used, left here for backwards compatibility"); static const bool FLAGS_deletepercent_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_deletepercent, &ValidateInt32Percent); static const bool FLAGS_table_cache_numshardbits_dummy __attribute__((unused)) = RegisterFlagValidator(&FLAGS_table_cache_numshardbits, &ValidateTableCacheNumshardbits); namespace rocksdb { namespace { struct ReportFileOpCounters { std::atomic open_counter_; std::atomic read_counter_; std::atomic append_counter_; std::atomic bytes_read_; std::atomic bytes_written_; }; // A special Env to records and report file operations in db_bench class ReportFileOpEnv : public EnvWrapper { public: explicit ReportFileOpEnv(Env* base) : EnvWrapper(base) { reset(); } void reset() { counters_.open_counter_ = 0; counters_.read_counter_ = 0; counters_.append_counter_ = 0; counters_.bytes_read_ = 0; counters_.bytes_written_ = 0; } Status NewSequentialFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { class CountingFile : public SequentialFile { private: unique_ptr target_; ReportFileOpCounters* counters_; public: CountingFile(unique_ptr&& target, ReportFileOpCounters* counters) : target_(std::move(target)), counters_(counters) {} virtual Status Read(size_t n, Slice* result, char* scratch) override { counters_->read_counter_.fetch_add(1, std::memory_order_relaxed); Status rv = target_->Read(n, result, scratch); counters_->bytes_read_.fetch_add(result->size(), std::memory_order_relaxed); return rv; } virtual Status Skip(uint64_t n) override { return target_->Skip(n); } }; Status s = target()->NewSequentialFile(f, r, soptions); if (s.ok()) { counters()->open_counter_.fetch_add(1, std::memory_order_relaxed); r->reset(new CountingFile(std::move(*r), counters())); } return s; } Status NewRandomAccessFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { class CountingFile : public RandomAccessFile { private: unique_ptr target_; ReportFileOpCounters* counters_; public: CountingFile(unique_ptr&& target, ReportFileOpCounters* counters) : target_(std::move(target)), counters_(counters) {} virtual Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const override { counters_->read_counter_.fetch_add(1, std::memory_order_relaxed); Status rv = target_->Read(offset, n, result, scratch); counters_->bytes_read_.fetch_add(result->size(), std::memory_order_relaxed); return rv; } }; Status s = target()->NewRandomAccessFile(f, r, soptions); if (s.ok()) { counters()->open_counter_.fetch_add(1, std::memory_order_relaxed); r->reset(new CountingFile(std::move(*r), counters())); } return s; } Status NewWritableFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { class CountingFile : public WritableFile { private: unique_ptr target_; ReportFileOpCounters* counters_; public: CountingFile(unique_ptr&& target, ReportFileOpCounters* counters) : target_(std::move(target)), counters_(counters) {} Status Append(const Slice& data) override { counters_->append_counter_.fetch_add(1, std::memory_order_relaxed); Status rv = target_->Append(data); counters_->bytes_written_.fetch_add(data.size(), std::memory_order_relaxed); return rv; } Status Truncate(uint64_t size) override { return target_->Truncate(size); } Status Close() override { return target_->Close(); } Status Flush() override { return target_->Flush(); } Status Sync() override { return target_->Sync(); } }; Status s = target()->NewWritableFile(f, r, soptions); if (s.ok()) { counters()->open_counter_.fetch_add(1, std::memory_order_relaxed); r->reset(new CountingFile(std::move(*r), counters())); } return s; } // getter ReportFileOpCounters* counters() { return &counters_; } private: ReportFileOpCounters counters_; }; } // namespace // Helper for quickly generating random data. class RandomGenerator { private: std::string data_; unsigned int pos_; public: RandomGenerator() { // We use a limited amount of data over and over again and ensure // that it is larger than the compression window (32KB), and also // large enough to serve all typical value sizes we want to write. Random rnd(301); std::string piece; while (data_.size() < (unsigned)std::max(1048576, FLAGS_value_size)) { // Add a short fragment that is as compressible as specified // by FLAGS_compression_ratio. test::CompressibleString(&rnd, FLAGS_compression_ratio, 100, &piece); data_.append(piece); } pos_ = 0; } Slice Generate(unsigned int len) { assert(len <= data_.size()); if (pos_ + len > data_.size()) { pos_ = 0; } pos_ += len; return Slice(data_.data() + pos_ - len, len); } }; static void AppendWithSpace(std::string* str, Slice msg) { if (msg.empty()) return; if (!str->empty()) { str->push_back(' '); } str->append(msg.data(), msg.size()); } struct DBWithColumnFamilies { std::vector cfh; DB* db; #ifndef ROCKSDB_LITE OptimisticTransactionDB* opt_txn_db; #endif // ROCKSDB_LITE std::atomic num_created; // Need to be updated after all the // new entries in cfh are set. size_t num_hot; // Number of column families to be queried at each moment. // After each CreateNewCf(), another num_hot number of new // Column families will be created and used to be queried. port::Mutex create_cf_mutex; // Only one thread can execute CreateNewCf() DBWithColumnFamilies() : db(nullptr) #ifndef ROCKSDB_LITE , opt_txn_db(nullptr) #endif // ROCKSDB_LITE { cfh.clear(); num_created = 0; num_hot = 0; } DBWithColumnFamilies(const DBWithColumnFamilies& other) : cfh(other.cfh), db(other.db), #ifndef ROCKSDB_LITE opt_txn_db(other.opt_txn_db), #endif // ROCKSDB_LITE num_created(other.num_created.load()), num_hot(other.num_hot) {} void DeleteDBs() { std::for_each(cfh.begin(), cfh.end(), [](ColumnFamilyHandle* cfhi) { delete cfhi; }); cfh.clear(); #ifndef ROCKSDB_LITE if (opt_txn_db) { delete opt_txn_db; opt_txn_db = nullptr; } else { delete db; db = nullptr; } #else delete db; db = nullptr; #endif // ROCKSDB_LITE } ColumnFamilyHandle* GetCfh(int64_t rand_num) { assert(num_hot > 0); return cfh[num_created.load(std::memory_order_acquire) - num_hot + rand_num % num_hot]; } // stage: assume CF from 0 to stage * num_hot has be created. Need to create // stage * num_hot + 1 to stage * (num_hot + 1). void CreateNewCf(ColumnFamilyOptions options, int64_t stage) { MutexLock l(&create_cf_mutex); if ((stage + 1) * num_hot <= num_created) { // Already created. return; } auto new_num_created = num_created + num_hot; assert(new_num_created <= cfh.size()); for (size_t i = num_created; i < new_num_created; i++) { Status s = db->CreateColumnFamily(options, ColumnFamilyName(i), &(cfh[i])); if (!s.ok()) { fprintf(stderr, "create column family error: %s\n", s.ToString().c_str()); abort(); } } num_created.store(new_num_created, std::memory_order_release); } }; // a class that reports stats to CSV file class ReporterAgent { public: ReporterAgent(Env* env, const std::string& fname, uint64_t report_interval_secs) : env_(env), total_ops_done_(0), last_report_(0), report_interval_secs_(report_interval_secs), stop_(false) { auto s = env_->NewWritableFile(fname, &report_file_, EnvOptions()); if (s.ok()) { s = report_file_->Append(Header() + "\n"); } if (s.ok()) { s = report_file_->Flush(); } if (!s.ok()) { fprintf(stderr, "Can't open %s: %s\n", fname.c_str(), s.ToString().c_str()); abort(); } reporting_thread_ = std::thread([&]() { SleepAndReport(); }); } ~ReporterAgent() { { std::unique_lock lk(mutex_); stop_ = true; stop_cv_.notify_all(); } reporting_thread_.join(); } // thread safe void ReportFinishedOps(int64_t num_ops) { total_ops_done_.fetch_add(num_ops); } private: std::string Header() const { return "secs_elapsed,interval_qps"; } void SleepAndReport() { uint64_t kMicrosInSecond = 1000 * 1000; auto time_started = env_->NowMicros(); while (true) { { std::unique_lock lk(mutex_); if (stop_ || stop_cv_.wait_for(lk, std::chrono::seconds(report_interval_secs_), [&]() { return stop_; })) { // stopping break; } // else -> timeout, which means time for a report! } auto total_ops_done_snapshot = total_ops_done_.load(); // round the seconds elapsed auto secs_elapsed = (env_->NowMicros() - time_started + kMicrosInSecond / 2) / kMicrosInSecond; std::string report = ToString(secs_elapsed) + "," + ToString(total_ops_done_snapshot - last_report_) + "\n"; auto s = report_file_->Append(report); if (s.ok()) { s = report_file_->Flush(); } if (!s.ok()) { fprintf(stderr, "Can't write to report file (%s), stopping the reporting\n", s.ToString().c_str()); break; } last_report_ = total_ops_done_snapshot; } } Env* env_; std::unique_ptr report_file_; std::atomic total_ops_done_; int64_t last_report_; const uint64_t report_interval_secs_; std::thread reporting_thread_; std::mutex mutex_; // will notify on stop std::condition_variable stop_cv_; bool stop_; }; enum OperationType : unsigned char { kRead = 0, kWrite, kDelete, kSeek, kMerge, kUpdate, kCompress, kUncompress, kCrc, kHash, kOthers }; static std::unordered_map> OperationTypeString = { {kRead, "read"}, {kWrite, "write"}, {kDelete, "delete"}, {kSeek, "seek"}, {kMerge, "merge"}, {kUpdate, "update"}, {kCompress, "compress"}, {kCompress, "uncompress"}, {kCrc, "crc"}, {kHash, "hash"}, {kOthers, "op"} }; class Stats { private: int id_; uint64_t start_; uint64_t finish_; double seconds_; uint64_t done_; uint64_t last_report_done_; uint64_t next_report_; uint64_t bytes_; uint64_t last_op_finish_; uint64_t last_report_finish_; std::unordered_map> hist_; std::string message_; bool exclude_from_merge_; ReporterAgent* reporter_agent_; // does not own public: Stats() { Start(-1); } void SetReporterAgent(ReporterAgent* reporter_agent) { reporter_agent_ = reporter_agent; } void Start(int id) { id_ = id; next_report_ = FLAGS_stats_interval ? FLAGS_stats_interval : 100; last_op_finish_ = start_; hist_.clear(); done_ = 0; last_report_done_ = 0; bytes_ = 0; seconds_ = 0; start_ = FLAGS_env->NowMicros(); finish_ = start_; last_report_finish_ = start_; message_.clear(); // When set, stats from this thread won't be merged with others. exclude_from_merge_ = false; } void Merge(const Stats& other) { if (other.exclude_from_merge_) return; for (auto it = other.hist_.begin(); it != other.hist_.end(); ++it) { auto this_it = hist_.find(it->first); if (this_it != hist_.end()) { this_it->second.Merge(other.hist_.at(it->first)); } else { hist_.insert({ it->first, it->second }); } } done_ += other.done_; bytes_ += other.bytes_; seconds_ += other.seconds_; if (other.start_ < start_) start_ = other.start_; if (other.finish_ > finish_) finish_ = other.finish_; // Just keep the messages from one thread if (message_.empty()) message_ = other.message_; } void Stop() { finish_ = FLAGS_env->NowMicros(); seconds_ = (finish_ - start_) * 1e-6; } void AddMessage(Slice msg) { AppendWithSpace(&message_, msg); } void SetId(int id) { id_ = id; } void SetExcludeFromMerge() { exclude_from_merge_ = true; } void PrintThreadStatus() { std::vector thread_list; FLAGS_env->GetThreadList(&thread_list); fprintf(stderr, "\n%18s %10s %12s %20s %13s %45s %12s %s\n", "ThreadID", "ThreadType", "cfName", "Operation", "ElapsedTime", "Stage", "State", "OperationProperties"); int64_t current_time = 0; Env::Default()->GetCurrentTime(¤t_time); for (auto ts : thread_list) { fprintf(stderr, "%18" PRIu64 " %10s %12s %20s %13s %45s %12s", ts.thread_id, ThreadStatus::GetThreadTypeName(ts.thread_type).c_str(), ts.cf_name.c_str(), ThreadStatus::GetOperationName(ts.operation_type).c_str(), ThreadStatus::MicrosToString(ts.op_elapsed_micros).c_str(), ThreadStatus::GetOperationStageName(ts.operation_stage).c_str(), ThreadStatus::GetStateName(ts.state_type).c_str()); auto op_properties = ThreadStatus::InterpretOperationProperties( ts.operation_type, ts.op_properties); for (const auto& op_prop : op_properties) { fprintf(stderr, " %s %" PRIu64" |", op_prop.first.c_str(), op_prop.second); } fprintf(stderr, "\n"); } } void ResetLastOpTime() { // Set to now to avoid latency from calls to SleepForMicroseconds last_op_finish_ = FLAGS_env->NowMicros(); } void FinishedOps(DBWithColumnFamilies* db_with_cfh, DB* db, int64_t num_ops, enum OperationType op_type = kOthers) { if (reporter_agent_) { reporter_agent_->ReportFinishedOps(num_ops); } if (FLAGS_histogram) { uint64_t now = FLAGS_env->NowMicros(); uint64_t micros = now - last_op_finish_; if (hist_.find(op_type) == hist_.end()) { HistogramImpl hist_temp; hist_.insert({op_type, hist_temp}); } hist_[op_type].Add(micros); if (micros > 20000 && !FLAGS_stats_interval) { fprintf(stderr, "long op: %" PRIu64 " micros%30s\r", micros, ""); fflush(stderr); } last_op_finish_ = now; } done_ += num_ops; if (done_ >= next_report_) { if (!FLAGS_stats_interval) { if (next_report_ < 1000) next_report_ += 100; else if (next_report_ < 5000) next_report_ += 500; else if (next_report_ < 10000) next_report_ += 1000; else if (next_report_ < 50000) next_report_ += 5000; else if (next_report_ < 100000) next_report_ += 10000; else if (next_report_ < 500000) next_report_ += 50000; else next_report_ += 100000; fprintf(stderr, "... finished %" PRIu64 " ops%30s\r", done_, ""); } else { uint64_t now = FLAGS_env->NowMicros(); int64_t usecs_since_last = now - last_report_finish_; // Determine whether to print status where interval is either // each N operations or each N seconds. if (FLAGS_stats_interval_seconds && usecs_since_last < (FLAGS_stats_interval_seconds * 1000000)) { // Don't check again for this many operations next_report_ += FLAGS_stats_interval; } else { fprintf(stderr, "%s ... thread %d: (%" PRIu64 ",%" PRIu64 ") ops and " "(%.1f,%.1f) ops/second in (%.6f,%.6f) seconds\n", FLAGS_env->TimeToString(now/1000000).c_str(), id_, done_ - last_report_done_, done_, (done_ - last_report_done_) / (usecs_since_last / 1000000.0), done_ / ((now - start_) / 1000000.0), (now - last_report_finish_) / 1000000.0, (now - start_) / 1000000.0); if (FLAGS_stats_per_interval) { std::string stats; if (db_with_cfh && db_with_cfh->num_created.load()) { for (size_t i = 0; i < db_with_cfh->num_created.load(); ++i) { if (db->GetProperty(db_with_cfh->cfh[i], "rocksdb.cfstats", &stats)) fprintf(stderr, "%s\n", stats.c_str()); if (FLAGS_show_table_properties) { for (int level = 0; level < FLAGS_num_levels; ++level) { if (db->GetProperty( db_with_cfh->cfh[i], "rocksdb.aggregated-table-properties-at-level" + ToString(level), &stats)) { if (stats.find("# entries=0") == std::string::npos) { fprintf(stderr, "Level[%d]: %s\n", level, stats.c_str()); } } } } } } else if (db) { if (db->GetProperty("rocksdb.stats", &stats)) { fprintf(stderr, "%s\n", stats.c_str()); } if (FLAGS_show_table_properties) { for (int level = 0; level < FLAGS_num_levels; ++level) { if (db->GetProperty( "rocksdb.aggregated-table-properties-at-level" + ToString(level), &stats)) { if (stats.find("# entries=0") == std::string::npos) { fprintf(stderr, "Level[%d]: %s\n", level, stats.c_str()); } } } } } } next_report_ += FLAGS_stats_interval; last_report_finish_ = now; last_report_done_ = done_; } } if (id_ == 0 && FLAGS_thread_status_per_interval) { PrintThreadStatus(); } fflush(stderr); } } void AddBytes(int64_t n) { bytes_ += n; } void Report(const Slice& name) { // Pretend at least one op was done in case we are running a benchmark // that does not call FinishedOps(). if (done_ < 1) done_ = 1; std::string extra; if (bytes_ > 0) { // Rate is computed on actual elapsed time, not the sum of per-thread // elapsed times. double elapsed = (finish_ - start_) * 1e-6; char rate[100]; snprintf(rate, sizeof(rate), "%6.1f MB/s", (bytes_ / 1048576.0) / elapsed); extra = rate; } AppendWithSpace(&extra, message_); double elapsed = (finish_ - start_) * 1e-6; double throughput = (double)done_/elapsed; fprintf(stdout, "%-12s : %11.3f micros/op %ld ops/sec;%s%s\n", name.ToString().c_str(), elapsed * 1e6 / done_, (long)throughput, (extra.empty() ? "" : " "), extra.c_str()); if (FLAGS_histogram) { for (auto it = hist_.begin(); it != hist_.end(); ++it) { fprintf(stdout, "Microseconds per %s:\n%s\n", OperationTypeString[it->first].c_str(), it->second.ToString().c_str()); } } if (FLAGS_report_file_operations) { ReportFileOpEnv* env = static_cast(FLAGS_env); ReportFileOpCounters* counters = env->counters(); fprintf(stdout, "Num files opened: %d\n", counters->open_counter_.load(std::memory_order_relaxed)); fprintf(stdout, "Num Read(): %d\n", counters->read_counter_.load(std::memory_order_relaxed)); fprintf(stdout, "Num Append(): %d\n", counters->append_counter_.load(std::memory_order_relaxed)); fprintf(stdout, "Num bytes read: %" PRIu64 "\n", counters->bytes_read_.load(std::memory_order_relaxed)); fprintf(stdout, "Num bytes written: %" PRIu64 "\n", counters->bytes_written_.load(std::memory_order_relaxed)); env->reset(); } fflush(stdout); } }; // State shared by all concurrent executions of the same benchmark. struct SharedState { port::Mutex mu; port::CondVar cv; int total; int perf_level; std::shared_ptr write_rate_limiter; // Each thread goes through the following states: // (1) initializing // (2) waiting for others to be initialized // (3) running // (4) done long num_initialized; long num_done; bool start; SharedState() : cv(&mu), perf_level(FLAGS_perf_level) { } }; // Per-thread state for concurrent executions of the same benchmark. struct ThreadState { int tid; // 0..n-1 when running in n threads Random64 rand; // Has different seeds for different threads Stats stats; SharedState* shared; /* implicit */ ThreadState(int index) : tid(index), rand((FLAGS_seed ? FLAGS_seed : 1000) + index) { } }; class Duration { public: Duration(uint64_t max_seconds, int64_t max_ops, int64_t ops_per_stage = 0) { max_seconds_ = max_seconds; max_ops_= max_ops; ops_per_stage_ = (ops_per_stage > 0) ? ops_per_stage : max_ops; ops_ = 0; start_at_ = FLAGS_env->NowMicros(); } int64_t GetStage() { return std::min(ops_, max_ops_ - 1) / ops_per_stage_; } bool Done(int64_t increment) { if (increment <= 0) increment = 1; // avoid Done(0) and infinite loops ops_ += increment; if (max_seconds_) { // Recheck every appx 1000 ops (exact iff increment is factor of 1000) if ((ops_/1000) != ((ops_-increment)/1000)) { uint64_t now = FLAGS_env->NowMicros(); return ((now - start_at_) / 1000000) >= max_seconds_; } else { return false; } } else { return ops_ > max_ops_; } } private: uint64_t max_seconds_; int64_t max_ops_; int64_t ops_per_stage_; int64_t ops_; uint64_t start_at_; }; class Benchmark { private: std::shared_ptr cache_; std::shared_ptr compressed_cache_; std::shared_ptr filter_policy_; const SliceTransform* prefix_extractor_; DBWithColumnFamilies db_; std::vector multi_dbs_; int64_t num_; int value_size_; int key_size_; int prefix_size_; int64_t keys_per_prefix_; int64_t entries_per_batch_; WriteOptions write_options_; Options open_options_; // keep options around to properly destroy db later int64_t reads_; double read_random_exp_range_; int64_t writes_; int64_t readwrites_; int64_t merge_keys_; bool report_file_operations_; int cachedev_fd_; bool SanityCheck() { if (FLAGS_compression_ratio > 1) { fprintf(stderr, "compression_ratio should be between 0 and 1\n"); return false; } return true; } inline bool CompressSlice(const Slice& input, std::string* compressed) { bool ok = true; switch (FLAGS_compression_type_e) { case rocksdb::kSnappyCompression: ok = Snappy_Compress(Options().compression_opts, input.data(), input.size(), compressed); break; case rocksdb::kZlibCompression: ok = Zlib_Compress(Options().compression_opts, 2, input.data(), input.size(), compressed); break; case rocksdb::kBZip2Compression: ok = BZip2_Compress(Options().compression_opts, 2, input.data(), input.size(), compressed); break; case rocksdb::kLZ4Compression: ok = LZ4_Compress(Options().compression_opts, 2, input.data(), input.size(), compressed); break; case rocksdb::kLZ4HCCompression: ok = LZ4HC_Compress(Options().compression_opts, 2, input.data(), input.size(), compressed); break; case rocksdb::kZSTDNotFinalCompression: ok = ZSTD_Compress(Options().compression_opts, input.data(), input.size(), compressed); break; default: ok = false; } return ok; } void PrintHeader() { PrintEnvironment(); fprintf(stdout, "Keys: %d bytes each\n", FLAGS_key_size); fprintf(stdout, "Values: %d bytes each (%d bytes after compression)\n", FLAGS_value_size, static_cast(FLAGS_value_size * FLAGS_compression_ratio + 0.5)); fprintf(stdout, "Entries: %" PRIu64 "\n", num_); fprintf(stdout, "Prefix: %d bytes\n", FLAGS_prefix_size); fprintf(stdout, "Keys per prefix: %" PRIu64 "\n", keys_per_prefix_); fprintf(stdout, "RawSize: %.1f MB (estimated)\n", ((static_cast(FLAGS_key_size + FLAGS_value_size) * num_) / 1048576.0)); fprintf(stdout, "FileSize: %.1f MB (estimated)\n", (((FLAGS_key_size + FLAGS_value_size * FLAGS_compression_ratio) * num_) / 1048576.0)); fprintf(stdout, "Write rate: %" PRIu64 " bytes/second\n", FLAGS_benchmark_write_rate_limit); if (FLAGS_enable_numa) { fprintf(stderr, "Running in NUMA enabled mode.\n"); #ifndef NUMA fprintf(stderr, "NUMA is not defined in the system.\n"); exit(1); #else if (numa_available() == -1) { fprintf(stderr, "NUMA is not supported by the system.\n"); exit(1); } #endif } auto compression = CompressionTypeToString(FLAGS_compression_type_e); fprintf(stdout, "Compression: %s\n", compression.c_str()); switch (FLAGS_rep_factory) { case kPrefixHash: fprintf(stdout, "Memtablerep: prefix_hash\n"); break; case kSkipList: fprintf(stdout, "Memtablerep: skip_list\n"); break; case kVectorRep: fprintf(stdout, "Memtablerep: vector\n"); break; case kHashLinkedList: fprintf(stdout, "Memtablerep: hash_linkedlist\n"); break; case kCuckoo: fprintf(stdout, "Memtablerep: cuckoo\n"); break; } fprintf(stdout, "Perf Level: %d\n", FLAGS_perf_level); PrintWarnings(compression.c_str()); fprintf(stdout, "------------------------------------------------\n"); } void PrintWarnings(const char* compression) { #if defined(__GNUC__) && !defined(__OPTIMIZE__) fprintf(stdout, "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n" ); #endif #ifndef NDEBUG fprintf(stdout, "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); #endif if (FLAGS_compression_type_e != rocksdb::kNoCompression) { // The test string should not be too small. const int len = FLAGS_block_size; std::string input_str(len, 'y'); std::string compressed; bool result = CompressSlice(Slice(input_str), &compressed); if (!result) { fprintf(stdout, "WARNING: %s compression is not enabled\n", compression); } else if (compressed.size() >= input_str.size()) { fprintf(stdout, "WARNING: %s compression is not effective\n", compression); } } } // Current the following isn't equivalent to OS_LINUX. #if defined(__linux) static Slice TrimSpace(Slice s) { unsigned int start = 0; while (start < s.size() && isspace(s[start])) { start++; } unsigned int limit = static_cast(s.size()); while (limit > start && isspace(s[limit-1])) { limit--; } return Slice(s.data() + start, limit - start); } #endif void PrintEnvironment() { fprintf(stderr, "LevelDB: version %d.%d\n", kMajorVersion, kMinorVersion); #if defined(__linux) time_t now = time(nullptr); char buf[52]; // Lint complains about ctime() usage, so replace it with ctime_r(). The // requirement is to provide a buffer which is at least 26 bytes. fprintf(stderr, "Date: %s", ctime_r(&now, buf)); // ctime_r() adds newline FILE* cpuinfo = fopen("/proc/cpuinfo", "r"); if (cpuinfo != nullptr) { char line[1000]; int num_cpus = 0; std::string cpu_type; std::string cache_size; while (fgets(line, sizeof(line), cpuinfo) != nullptr) { const char* sep = strchr(line, ':'); if (sep == nullptr) { continue; } Slice key = TrimSpace(Slice(line, sep - 1 - line)); Slice val = TrimSpace(Slice(sep + 1)); if (key == "model name") { ++num_cpus; cpu_type = val.ToString(); } else if (key == "cache size") { cache_size = val.ToString(); } } fclose(cpuinfo); fprintf(stderr, "CPU: %d * %s\n", num_cpus, cpu_type.c_str()); fprintf(stderr, "CPUCache: %s\n", cache_size.c_str()); } #endif } public: Benchmark() : cache_( FLAGS_cache_size >= 0 ? (FLAGS_cache_numshardbits >= 1 ? NewLRUCache(FLAGS_cache_size, FLAGS_cache_numshardbits) : NewLRUCache(FLAGS_cache_size)) : nullptr), compressed_cache_(FLAGS_compressed_cache_size >= 0 ? (FLAGS_cache_numshardbits >= 1 ? NewLRUCache(FLAGS_compressed_cache_size, FLAGS_cache_numshardbits) : NewLRUCache(FLAGS_compressed_cache_size)) : nullptr), filter_policy_(FLAGS_bloom_bits >= 0 ? NewBloomFilterPolicy(FLAGS_bloom_bits, FLAGS_use_block_based_filter) : nullptr), prefix_extractor_(NewFixedPrefixTransform(FLAGS_prefix_size)), num_(FLAGS_num), value_size_(FLAGS_value_size), key_size_(FLAGS_key_size), prefix_size_(FLAGS_prefix_size), keys_per_prefix_(FLAGS_keys_per_prefix), entries_per_batch_(1), reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), read_random_exp_range_(0.0), writes_(FLAGS_writes < 0 ? FLAGS_num : FLAGS_writes), readwrites_( (FLAGS_writes < 0 && FLAGS_reads < 0) ? FLAGS_num : ((FLAGS_writes > FLAGS_reads) ? FLAGS_writes : FLAGS_reads)), merge_keys_(FLAGS_merge_keys < 0 ? FLAGS_num : FLAGS_merge_keys), report_file_operations_(FLAGS_report_file_operations), cachedev_fd_(-1) { if (report_file_operations_) { if (!FLAGS_hdfs.empty()) { fprintf(stderr, "--hdfs and --report_file_operations cannot be enabled " "at the same time"); exit(1); } FLAGS_env = new ReportFileOpEnv(rocksdb::Env::Default()); } if (FLAGS_prefix_size > FLAGS_key_size) { fprintf(stderr, "prefix size is larger than key size"); exit(1); } std::vector files; FLAGS_env->GetChildren(FLAGS_db, &files); for (size_t i = 0; i < files.size(); i++) { if (Slice(files[i]).starts_with("heap-")) { FLAGS_env->DeleteFile(FLAGS_db + "/" + files[i]); } } if (!FLAGS_use_existing_db) { Options options; if (!FLAGS_wal_dir.empty()) { options.wal_dir = FLAGS_wal_dir; } DestroyDB(FLAGS_db, options); } } ~Benchmark() { db_.DeleteDBs(); delete prefix_extractor_; if (cache_.get() != nullptr) { // this will leak, but we're shutting down so nobody cares cache_->DisownData(); } if (FLAGS_disable_flashcache_for_background_threads && cachedev_fd_ != -1) { // Dtor for this env should run before cachedev_fd_ is closed flashcache_aware_env_ = nullptr; close(cachedev_fd_); } } Slice AllocateKey(std::unique_ptr* key_guard) { char* data = new char[key_size_]; const char* const_data = data; key_guard->reset(const_data); return Slice(key_guard->get(), key_size_); } // Generate key according to the given specification and random number. // The resulting key will have the following format (if keys_per_prefix_ // is positive), extra trailing bytes are either cut off or paddd with '0'. // The prefix value is derived from key value. // ---------------------------- // | prefix 00000 | key 00000 | // ---------------------------- // If keys_per_prefix_ is 0, the key is simply a binary representation of // random number followed by trailing '0's // ---------------------------- // | key 00000 | // ---------------------------- void GenerateKeyFromInt(uint64_t v, int64_t num_keys, Slice* key) { char* start = const_cast(key->data()); char* pos = start; if (keys_per_prefix_ > 0) { int64_t num_prefix = num_keys / keys_per_prefix_; int64_t prefix = v % num_prefix; int bytes_to_fill = std::min(prefix_size_, 8); if (port::kLittleEndian) { for (int i = 0; i < bytes_to_fill; ++i) { pos[i] = (prefix >> ((bytes_to_fill - i - 1) << 3)) & 0xFF; } } else { memcpy(pos, static_cast(&prefix), bytes_to_fill); } if (prefix_size_ > 8) { // fill the rest with 0s memset(pos + 8, '0', prefix_size_ - 8); } pos += prefix_size_; } int bytes_to_fill = std::min(key_size_ - static_cast(pos - start), 8); if (port::kLittleEndian) { for (int i = 0; i < bytes_to_fill; ++i) { pos[i] = (v >> ((bytes_to_fill - i - 1) << 3)) & 0xFF; } } else { memcpy(pos, static_cast(&v), bytes_to_fill); } pos += bytes_to_fill; if (key_size_ > pos - start) { memset(pos, '0', key_size_ - (pos - start)); } } std::string GetDbNameForMultiple(std::string base_name, size_t id) { return base_name + ToString(id); } void Run() { if (!SanityCheck()) { exit(1); } PrintHeader(); Open(&open_options_); std::stringstream benchmark_stream(FLAGS_benchmarks); std::string name; while (std::getline(benchmark_stream, name, ',')) { // Sanitize parameters num_ = FLAGS_num; reads_ = (FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads); writes_ = (FLAGS_writes < 0 ? FLAGS_num : FLAGS_writes); value_size_ = FLAGS_value_size; key_size_ = FLAGS_key_size; entries_per_batch_ = FLAGS_batch_size; write_options_ = WriteOptions(); read_random_exp_range_ = FLAGS_read_random_exp_range; if (FLAGS_sync) { write_options_.sync = true; } write_options_.disableWAL = FLAGS_disable_wal; void (Benchmark::*method)(ThreadState*) = nullptr; void (Benchmark::*post_process_method)() = nullptr; bool fresh_db = false; int num_threads = FLAGS_threads; if (name == "fillseq") { fresh_db = true; method = &Benchmark::WriteSeq; } else if (name == "fillbatch") { fresh_db = true; entries_per_batch_ = 1000; method = &Benchmark::WriteSeq; } else if (name == "fillrandom") { fresh_db = true; method = &Benchmark::WriteRandom; } else if (name == "filluniquerandom") { fresh_db = true; if (num_threads > 1) { fprintf(stderr, "filluniquerandom multithreaded not supported" ", use 1 thread"); num_threads = 1; } method = &Benchmark::WriteUniqueRandom; } else if (name == "overwrite") { method = &Benchmark::WriteRandom; } else if (name == "fillsync") { fresh_db = true; num_ /= 1000; write_options_.sync = true; method = &Benchmark::WriteRandom; } else if (name == "fill100K") { fresh_db = true; num_ /= 1000; value_size_ = 100 * 1000; method = &Benchmark::WriteRandom; } else if (name == "readseq") { method = &Benchmark::ReadSequential; } else if (name == "readtocache") { method = &Benchmark::ReadSequential; num_threads = 1; reads_ = num_; } else if (name == "readreverse") { method = &Benchmark::ReadReverse; } else if (name == "readrandom") { method = &Benchmark::ReadRandom; } else if (name == "readrandomfast") { method = &Benchmark::ReadRandomFast; } else if (name == "multireadrandom") { fprintf(stderr, "entries_per_batch = %" PRIi64 "\n", entries_per_batch_); method = &Benchmark::MultiReadRandom; } else if (name == "readmissing") { ++key_size_; method = &Benchmark::ReadRandom; } else if (name == "newiterator") { method = &Benchmark::IteratorCreation; } else if (name == "newiteratorwhilewriting") { num_threads++; // Add extra thread for writing method = &Benchmark::IteratorCreationWhileWriting; } else if (name == "seekrandom") { method = &Benchmark::SeekRandom; } else if (name == "seekrandomwhilewriting") { num_threads++; // Add extra thread for writing method = &Benchmark::SeekRandomWhileWriting; } else if (name == "seekrandomwhilemerging") { num_threads++; // Add extra thread for merging method = &Benchmark::SeekRandomWhileMerging; } else if (name == "readrandomsmall") { reads_ /= 1000; method = &Benchmark::ReadRandom; } else if (name == "deleteseq") { method = &Benchmark::DeleteSeq; } else if (name == "deleterandom") { method = &Benchmark::DeleteRandom; } else if (name == "readwhilewriting") { num_threads++; // Add extra thread for writing method = &Benchmark::ReadWhileWriting; } else if (name == "readwhilemerging") { num_threads++; // Add extra thread for writing method = &Benchmark::ReadWhileMerging; } else if (name == "readrandomwriterandom") { method = &Benchmark::ReadRandomWriteRandom; } else if (name == "readrandommergerandom") { if (FLAGS_merge_operator.empty()) { fprintf(stdout, "%-12s : skipped (--merge_operator is unknown)\n", name.c_str()); exit(1); } method = &Benchmark::ReadRandomMergeRandom; } else if (name == "updaterandom") { method = &Benchmark::UpdateRandom; } else if (name == "appendrandom") { method = &Benchmark::AppendRandom; } else if (name == "mergerandom") { if (FLAGS_merge_operator.empty()) { fprintf(stdout, "%-12s : skipped (--merge_operator is unknown)\n", name.c_str()); exit(1); } method = &Benchmark::MergeRandom; } else if (name == "randomwithverify") { method = &Benchmark::RandomWithVerify; } else if (name == "fillseekseq") { method = &Benchmark::WriteSeqSeekSeq; } else if (name == "compact") { method = &Benchmark::Compact; } else if (name == "crc32c") { method = &Benchmark::Crc32c; } else if (name == "xxhash") { method = &Benchmark::xxHash; } else if (name == "acquireload") { method = &Benchmark::AcquireLoad; } else if (name == "compress") { method = &Benchmark::Compress; } else if (name == "uncompress") { method = &Benchmark::Uncompress; #ifndef ROCKSDB_LITE } else if (name == "randomtransaction") { method = &Benchmark::RandomTransaction; post_process_method = &Benchmark::RandomTransactionVerify; #endif // ROCKSDB_LITE } else if (name == "randomreplacekeys") { fresh_db = true; method = &Benchmark::RandomReplaceKeys; } else if (name == "stats") { PrintStats("rocksdb.stats"); } else if (name == "levelstats") { PrintStats("rocksdb.levelstats"); } else if (name == "sstables") { PrintStats("rocksdb.sstables"); } else if (!name.empty()) { // No error message for empty name fprintf(stderr, "unknown benchmark '%s'\n", name.c_str()); exit(1); } if (fresh_db) { if (FLAGS_use_existing_db) { fprintf(stdout, "%-12s : skipped (--use_existing_db is true)\n", name.c_str()); method = nullptr; } else { if (db_.db != nullptr) { db_.DeleteDBs(); DestroyDB(FLAGS_db, open_options_); } for (size_t i = 0; i < multi_dbs_.size(); i++) { delete multi_dbs_[i].db; DestroyDB(GetDbNameForMultiple(FLAGS_db, i), open_options_); } multi_dbs_.clear(); } Open(&open_options_); // use open_options for the last accessed } if (method != nullptr) { fprintf(stdout, "DB path: [%s]\n", FLAGS_db.c_str()); RunBenchmark(num_threads, name, method); } if (post_process_method != nullptr) { (this->*post_process_method)(); } } if (FLAGS_statistics) { fprintf(stdout, "STATISTICS:\n%s\n", dbstats->ToString().c_str()); } } private: std::unique_ptr flashcache_aware_env_; struct ThreadArg { Benchmark* bm; SharedState* shared; ThreadState* thread; void (Benchmark::*method)(ThreadState*); }; static void ThreadBody(void* v) { ThreadArg* arg = reinterpret_cast(v); SharedState* shared = arg->shared; ThreadState* thread = arg->thread; { MutexLock l(&shared->mu); shared->num_initialized++; if (shared->num_initialized >= shared->total) { shared->cv.SignalAll(); } while (!shared->start) { shared->cv.Wait(); } } SetPerfLevel(static_cast (shared->perf_level)); thread->stats.Start(thread->tid); (arg->bm->*(arg->method))(thread); thread->stats.Stop(); { MutexLock l(&shared->mu); shared->num_done++; if (shared->num_done >= shared->total) { shared->cv.SignalAll(); } } } void RunBenchmark(int n, Slice name, void (Benchmark::*method)(ThreadState*)) { SharedState shared; shared.total = n; shared.num_initialized = 0; shared.num_done = 0; shared.start = false; if (FLAGS_benchmark_write_rate_limit > 0) { shared.write_rate_limiter.reset( NewGenericRateLimiter(FLAGS_benchmark_write_rate_limit)); } std::unique_ptr reporter_agent; if (FLAGS_report_interval_seconds > 0) { reporter_agent.reset(new ReporterAgent(FLAGS_env, FLAGS_report_file, FLAGS_report_interval_seconds)); } ThreadArg* arg = new ThreadArg[n]; for (int i = 0; i < n; i++) { #ifdef NUMA if (FLAGS_enable_numa) { // Performs a local allocation of memory to threads in numa node. int n_nodes = numa_num_task_nodes(); // Number of nodes in NUMA. numa_exit_on_error = 1; int numa_node = i % n_nodes; bitmask* nodes = numa_allocate_nodemask(); numa_bitmask_clearall(nodes); numa_bitmask_setbit(nodes, numa_node); // numa_bind() call binds the process to the node and these // properties are passed on to the thread that is created in // StartThread method called later in the loop. numa_bind(nodes); numa_set_strict(1); numa_free_nodemask(nodes); } #endif arg[i].bm = this; arg[i].method = method; arg[i].shared = &shared; arg[i].thread = new ThreadState(i); arg[i].thread->stats.SetReporterAgent(reporter_agent.get()); arg[i].thread->shared = &shared; FLAGS_env->StartThread(ThreadBody, &arg[i]); } shared.mu.Lock(); while (shared.num_initialized < n) { shared.cv.Wait(); } shared.start = true; shared.cv.SignalAll(); while (shared.num_done < n) { shared.cv.Wait(); } shared.mu.Unlock(); // Stats for some threads can be excluded. Stats merge_stats; for (int i = 0; i < n; i++) { merge_stats.Merge(arg[i].thread->stats); } merge_stats.Report(name); for (int i = 0; i < n; i++) { delete arg[i].thread; } delete[] arg; } void Crc32c(ThreadState* thread) { // Checksum about 500MB of data total const int size = 4096; const char* label = "(4K per op)"; std::string data(size, 'x'); int64_t bytes = 0; uint32_t crc = 0; while (bytes < 500 * 1048576) { crc = crc32c::Value(data.data(), size); thread->stats.FinishedOps(nullptr, nullptr, 1, kCrc); bytes += size; } // Print so result is not dead fprintf(stderr, "... crc=0x%x\r", static_cast(crc)); thread->stats.AddBytes(bytes); thread->stats.AddMessage(label); } void xxHash(ThreadState* thread) { // Checksum about 500MB of data total const int size = 4096; const char* label = "(4K per op)"; std::string data(size, 'x'); int64_t bytes = 0; unsigned int xxh32 = 0; while (bytes < 500 * 1048576) { xxh32 = XXH32(data.data(), size, 0); thread->stats.FinishedOps(nullptr, nullptr, 1, kHash); bytes += size; } // Print so result is not dead fprintf(stderr, "... xxh32=0x%x\r", static_cast(xxh32)); thread->stats.AddBytes(bytes); thread->stats.AddMessage(label); } void AcquireLoad(ThreadState* thread) { int dummy; std::atomic ap(&dummy); int count = 0; void *ptr = nullptr; thread->stats.AddMessage("(each op is 1000 loads)"); while (count < 100000) { for (int i = 0; i < 1000; i++) { ptr = ap.load(std::memory_order_acquire); } count++; thread->stats.FinishedOps(nullptr, nullptr, 1, kOthers); } if (ptr == nullptr) exit(1); // Disable unused variable warning. } void Compress(ThreadState *thread) { RandomGenerator gen; Slice input = gen.Generate(FLAGS_block_size); int64_t bytes = 0; int64_t produced = 0; bool ok = true; std::string compressed; // Compress 1G while (ok && bytes < int64_t(1) << 30) { compressed.clear(); ok = CompressSlice(input, &compressed); produced += compressed.size(); bytes += input.size(); thread->stats.FinishedOps(nullptr, nullptr, 1, kCompress); } if (!ok) { thread->stats.AddMessage("(compression failure)"); } else { char buf[100]; snprintf(buf, sizeof(buf), "(output: %.1f%%)", (produced * 100.0) / bytes); thread->stats.AddMessage(buf); thread->stats.AddBytes(bytes); } } void Uncompress(ThreadState *thread) { RandomGenerator gen; Slice input = gen.Generate(FLAGS_block_size); std::string compressed; bool ok = CompressSlice(input, &compressed); int64_t bytes = 0; int decompress_size; while (ok && bytes < 1024 * 1048576) { char *uncompressed = nullptr; switch (FLAGS_compression_type_e) { case rocksdb::kSnappyCompression: { // get size and allocate here to make comparison fair size_t ulength = 0; if (!Snappy_GetUncompressedLength(compressed.data(), compressed.size(), &ulength)) { ok = false; break; } uncompressed = new char[ulength]; ok = Snappy_Uncompress(compressed.data(), compressed.size(), uncompressed); break; } case rocksdb::kZlibCompression: uncompressed = Zlib_Uncompress(compressed.data(), compressed.size(), &decompress_size, 2); ok = uncompressed != nullptr; break; case rocksdb::kBZip2Compression: uncompressed = BZip2_Uncompress(compressed.data(), compressed.size(), &decompress_size, 2); ok = uncompressed != nullptr; break; case rocksdb::kLZ4Compression: uncompressed = LZ4_Uncompress(compressed.data(), compressed.size(), &decompress_size, 2); ok = uncompressed != nullptr; break; case rocksdb::kLZ4HCCompression: uncompressed = LZ4_Uncompress(compressed.data(), compressed.size(), &decompress_size, 2); ok = uncompressed != nullptr; break; case rocksdb::kZSTDNotFinalCompression: uncompressed = ZSTD_Uncompress(compressed.data(), compressed.size(), &decompress_size); ok = uncompressed != nullptr; break; default: ok = false; } delete[] uncompressed; bytes += input.size(); thread->stats.FinishedOps(nullptr, nullptr, 1, kUncompress); } if (!ok) { thread->stats.AddMessage("(compression failure)"); } else { thread->stats.AddBytes(bytes); } } void Open(Options* opts) { Options& options = *opts; assert(db_.db == nullptr); options.create_if_missing = !FLAGS_use_existing_db; options.create_missing_column_families = FLAGS_num_column_families > 1; options.db_write_buffer_size = FLAGS_db_write_buffer_size; options.write_buffer_size = FLAGS_write_buffer_size; options.max_write_buffer_number = FLAGS_max_write_buffer_number; options.min_write_buffer_number_to_merge = FLAGS_min_write_buffer_number_to_merge; options.max_write_buffer_number_to_maintain = FLAGS_max_write_buffer_number_to_maintain; options.max_background_compactions = FLAGS_max_background_compactions; options.max_subcompactions = static_cast(FLAGS_subcompactions); options.max_background_flushes = FLAGS_max_background_flushes; options.compaction_style = FLAGS_compaction_style_e; options.compaction_pri = FLAGS_compaction_pri_e; if (FLAGS_prefix_size != 0) { options.prefix_extractor.reset( NewFixedPrefixTransform(FLAGS_prefix_size)); } if (FLAGS_use_uint64_comparator) { options.comparator = test::Uint64Comparator(); if (FLAGS_key_size != 8) { fprintf(stderr, "Using Uint64 comparator but key size is not 8.\n"); exit(1); } } options.memtable_prefix_bloom_bits = FLAGS_memtable_bloom_bits; options.bloom_locality = FLAGS_bloom_locality; options.max_open_files = FLAGS_open_files; options.max_file_opening_threads = FLAGS_file_opening_threads; options.new_table_reader_for_compaction_inputs = FLAGS_new_table_reader_for_compaction_inputs; options.compaction_readahead_size = FLAGS_compaction_readahead_size; options.random_access_max_buffer_size = FLAGS_random_access_max_buffer_size; options.writable_file_max_buffer_size = FLAGS_writable_file_max_buffer_size; options.statistics = dbstats; if (FLAGS_enable_io_prio) { FLAGS_env->LowerThreadPoolIOPriority(Env::LOW); FLAGS_env->LowerThreadPoolIOPriority(Env::HIGH); } if (FLAGS_disable_flashcache_for_background_threads && cachedev_fd_ == -1) { // Avoid creating the env twice when an use_existing_db is true cachedev_fd_ = open(FLAGS_flashcache_dev.c_str(), O_RDONLY); if (cachedev_fd_ < 0) { fprintf(stderr, "Open flash device failed\n"); exit(1); } flashcache_aware_env_ = NewFlashcacheAwareEnv(FLAGS_env, cachedev_fd_); if (flashcache_aware_env_.get() == nullptr) { fprintf(stderr, "Failed to open flashcache device at %s\n", FLAGS_flashcache_dev.c_str()); std::abort(); } options.env = flashcache_aware_env_.get(); } else { options.env = FLAGS_env; } options.disableDataSync = FLAGS_disable_data_sync; options.use_fsync = FLAGS_use_fsync; options.wal_dir = FLAGS_wal_dir; options.num_levels = FLAGS_num_levels; options.target_file_size_base = FLAGS_target_file_size_base; options.target_file_size_multiplier = FLAGS_target_file_size_multiplier; options.max_bytes_for_level_base = FLAGS_max_bytes_for_level_base; options.level_compaction_dynamic_level_bytes = FLAGS_level_compaction_dynamic_level_bytes; options.max_bytes_for_level_multiplier = FLAGS_max_bytes_for_level_multiplier; options.filter_deletes = FLAGS_filter_deletes; if (FLAGS_row_cache_size) { if (FLAGS_cache_numshardbits >= 1) { options.row_cache = NewLRUCache(FLAGS_row_cache_size, FLAGS_cache_numshardbits); } else { options.row_cache = NewLRUCache(FLAGS_row_cache_size); } } if ((FLAGS_prefix_size == 0) && (FLAGS_rep_factory == kPrefixHash || FLAGS_rep_factory == kHashLinkedList)) { fprintf(stderr, "prefix_size should be non-zero if PrefixHash or " "HashLinkedList memtablerep is used\n"); exit(1); } switch (FLAGS_rep_factory) { case kSkipList: options.memtable_factory.reset(new SkipListFactory( FLAGS_skip_list_lookahead)); break; #ifndef ROCKSDB_LITE case kPrefixHash: options.memtable_factory.reset( NewHashSkipListRepFactory(FLAGS_hash_bucket_count)); break; case kHashLinkedList: options.memtable_factory.reset(NewHashLinkListRepFactory( FLAGS_hash_bucket_count)); break; case kVectorRep: options.memtable_factory.reset( new VectorRepFactory ); break; case kCuckoo: options.memtable_factory.reset(NewHashCuckooRepFactory( options.write_buffer_size, FLAGS_key_size + FLAGS_value_size)); break; #else default: fprintf(stderr, "Only skip list is supported in lite mode\n"); exit(1); #endif // ROCKSDB_LITE } if (FLAGS_use_plain_table) { #ifndef ROCKSDB_LITE if (FLAGS_rep_factory != kPrefixHash && FLAGS_rep_factory != kHashLinkedList) { fprintf(stderr, "Waring: plain table is used with skipList\n"); } if (!FLAGS_mmap_read && !FLAGS_mmap_write) { fprintf(stderr, "plain table format requires mmap to operate\n"); exit(1); } int bloom_bits_per_key = FLAGS_bloom_bits; if (bloom_bits_per_key < 0) { bloom_bits_per_key = 0; } PlainTableOptions plain_table_options; plain_table_options.user_key_len = FLAGS_key_size; plain_table_options.bloom_bits_per_key = bloom_bits_per_key; plain_table_options.hash_table_ratio = 0.75; options.table_factory = std::shared_ptr( NewPlainTableFactory(plain_table_options)); #else fprintf(stderr, "Plain table is not supported in lite mode\n"); exit(1); #endif // ROCKSDB_LITE } else if (FLAGS_use_cuckoo_table) { #ifndef ROCKSDB_LITE if (FLAGS_cuckoo_hash_ratio > 1 || FLAGS_cuckoo_hash_ratio < 0) { fprintf(stderr, "Invalid cuckoo_hash_ratio\n"); exit(1); } rocksdb::CuckooTableOptions table_options; table_options.hash_table_ratio = FLAGS_cuckoo_hash_ratio; table_options.identity_as_first_hash = FLAGS_identity_as_first_hash; options.table_factory = std::shared_ptr( NewCuckooTableFactory(table_options)); #else fprintf(stderr, "Cuckoo table is not supported in lite mode\n"); exit(1); #endif // ROCKSDB_LITE } else { BlockBasedTableOptions block_based_options; if (FLAGS_use_hash_search) { if (FLAGS_prefix_size == 0) { fprintf(stderr, "prefix_size not assigned when enable use_hash_search \n"); exit(1); } block_based_options.index_type = BlockBasedTableOptions::kHashSearch; } else { block_based_options.index_type = BlockBasedTableOptions::kBinarySearch; } if (cache_ == nullptr) { block_based_options.no_block_cache = true; } block_based_options.cache_index_and_filter_blocks = FLAGS_cache_index_and_filter_blocks; block_based_options.block_cache = cache_; block_based_options.block_cache_compressed = compressed_cache_; block_based_options.block_size = FLAGS_block_size; block_based_options.block_restart_interval = FLAGS_block_restart_interval; block_based_options.filter_policy = filter_policy_; block_based_options.skip_table_builder_flush = FLAGS_skip_table_builder_flush; block_based_options.format_version = 2; options.table_factory.reset( NewBlockBasedTableFactory(block_based_options)); } if (FLAGS_max_bytes_for_level_multiplier_additional_v.size() > 0) { if (FLAGS_max_bytes_for_level_multiplier_additional_v.size() != (unsigned int)FLAGS_num_levels) { fprintf(stderr, "Insufficient number of fanouts specified %d\n", (int)FLAGS_max_bytes_for_level_multiplier_additional_v.size()); exit(1); } options.max_bytes_for_level_multiplier_additional = FLAGS_max_bytes_for_level_multiplier_additional_v; } options.level0_stop_writes_trigger = FLAGS_level0_stop_writes_trigger; options.level0_file_num_compaction_trigger = FLAGS_level0_file_num_compaction_trigger; options.level0_slowdown_writes_trigger = FLAGS_level0_slowdown_writes_trigger; options.compression = FLAGS_compression_type_e; options.compression_opts.level = FLAGS_compression_level; options.WAL_ttl_seconds = FLAGS_wal_ttl_seconds; options.WAL_size_limit_MB = FLAGS_wal_size_limit_MB; options.max_total_wal_size = FLAGS_max_total_wal_size; if (FLAGS_min_level_to_compress >= 0) { assert(FLAGS_min_level_to_compress <= FLAGS_num_levels); options.compression_per_level.resize(FLAGS_num_levels); for (int i = 0; i < FLAGS_min_level_to_compress; i++) { options.compression_per_level[i] = kNoCompression; } for (int i = FLAGS_min_level_to_compress; i < FLAGS_num_levels; i++) { options.compression_per_level[i] = FLAGS_compression_type_e; } } options.soft_rate_limit = FLAGS_soft_rate_limit; options.hard_rate_limit = FLAGS_hard_rate_limit; options.soft_pending_compaction_bytes_limit = FLAGS_soft_pending_compaction_bytes_limit; options.hard_pending_compaction_bytes_limit = FLAGS_hard_pending_compaction_bytes_limit; options.delayed_write_rate = FLAGS_delayed_write_rate; options.allow_concurrent_memtable_write = FLAGS_allow_concurrent_memtable_write; options.enable_write_thread_adaptive_yield = FLAGS_enable_write_thread_adaptive_yield; options.write_thread_max_yield_usec = FLAGS_write_thread_max_yield_usec; options.write_thread_slow_yield_usec = FLAGS_write_thread_slow_yield_usec; options.rate_limit_delay_max_milliseconds = FLAGS_rate_limit_delay_max_milliseconds; options.table_cache_numshardbits = FLAGS_table_cache_numshardbits; options.max_grandparent_overlap_factor = FLAGS_max_grandparent_overlap_factor; options.disable_auto_compactions = FLAGS_disable_auto_compactions; options.source_compaction_factor = FLAGS_source_compaction_factor; // fill storage options options.allow_os_buffer = FLAGS_bufferedio; options.allow_mmap_reads = FLAGS_mmap_read; options.allow_mmap_writes = FLAGS_mmap_write; options.advise_random_on_open = FLAGS_advise_random_on_open; options.access_hint_on_compaction_start = FLAGS_compaction_fadvice_e; options.use_adaptive_mutex = FLAGS_use_adaptive_mutex; options.bytes_per_sync = FLAGS_bytes_per_sync; options.wal_bytes_per_sync = FLAGS_wal_bytes_per_sync; // merge operator options options.merge_operator = MergeOperators::CreateFromStringId( FLAGS_merge_operator); if (options.merge_operator == nullptr && !FLAGS_merge_operator.empty()) { fprintf(stderr, "invalid merge operator: %s\n", FLAGS_merge_operator.c_str()); exit(1); } options.max_successive_merges = FLAGS_max_successive_merges; options.compaction_measure_io_stats = FLAGS_compaction_measure_io_stats; // set universal style compaction configurations, if applicable if (FLAGS_universal_size_ratio != 0) { options.compaction_options_universal.size_ratio = FLAGS_universal_size_ratio; } if (FLAGS_universal_min_merge_width != 0) { options.compaction_options_universal.min_merge_width = FLAGS_universal_min_merge_width; } if (FLAGS_universal_max_merge_width != 0) { options.compaction_options_universal.max_merge_width = FLAGS_universal_max_merge_width; } if (FLAGS_universal_max_size_amplification_percent != 0) { options.compaction_options_universal.max_size_amplification_percent = FLAGS_universal_max_size_amplification_percent; } if (FLAGS_universal_compression_size_percent != -1) { options.compaction_options_universal.compression_size_percent = FLAGS_universal_compression_size_percent; } options.compaction_options_universal.allow_trivial_move = FLAGS_universal_allow_trivial_move; if (FLAGS_thread_status_per_interval > 0) { options.enable_thread_tracking = true; } if (FLAGS_rate_limiter_bytes_per_sec > 0) { options.rate_limiter.reset( NewGenericRateLimiter(FLAGS_rate_limiter_bytes_per_sec)); } #ifndef ROCKSDB_LITE if (FLAGS_readonly && FLAGS_transaction_db) { fprintf(stderr, "Cannot use readonly flag with transaction_db\n"); exit(1); } #endif // ROCKSDB_LITE if (FLAGS_num_multi_db <= 1) { OpenDb(options, FLAGS_db, &db_); } else { multi_dbs_.clear(); multi_dbs_.resize(FLAGS_num_multi_db); for (int i = 0; i < FLAGS_num_multi_db; i++) { OpenDb(options, GetDbNameForMultiple(FLAGS_db, i), &multi_dbs_[i]); } } if (FLAGS_min_level_to_compress >= 0) { options.compression_per_level.clear(); } } void OpenDb(const Options& options, const std::string& db_name, DBWithColumnFamilies* db) { Status s; // Open with column families if necessary. if (FLAGS_num_column_families > 1) { size_t num_hot = FLAGS_num_column_families; if (FLAGS_num_hot_column_families > 0 && FLAGS_num_hot_column_families < FLAGS_num_column_families) { num_hot = FLAGS_num_hot_column_families; } else { FLAGS_num_hot_column_families = FLAGS_num_column_families; } std::vector column_families; for (size_t i = 0; i < num_hot; i++) { column_families.push_back(ColumnFamilyDescriptor( ColumnFamilyName(i), ColumnFamilyOptions(options))); } #ifndef ROCKSDB_LITE if (FLAGS_readonly) { s = DB::OpenForReadOnly(options, db_name, column_families, &db->cfh, &db->db); } else if (FLAGS_optimistic_transaction_db) { s = OptimisticTransactionDB::Open(options, db_name, column_families, &db->cfh, &db->opt_txn_db); if (s.ok()) { db->db = db->opt_txn_db->GetBaseDB(); } } else if (FLAGS_transaction_db) { TransactionDB* ptr; TransactionDBOptions txn_db_options; s = TransactionDB::Open(options, txn_db_options, db_name, column_families, &db->cfh, &ptr); if (s.ok()) { db->db = ptr; } } else { s = DB::Open(options, db_name, column_families, &db->cfh, &db->db); } #else s = DB::Open(options, db_name, column_families, &db->cfh, &db->db); #endif // ROCKSDB_LITE db->cfh.resize(FLAGS_num_column_families); db->num_created = num_hot; db->num_hot = num_hot; #ifndef ROCKSDB_LITE } else if (FLAGS_readonly) { s = DB::OpenForReadOnly(options, db_name, &db->db); } else if (FLAGS_optimistic_transaction_db) { s = OptimisticTransactionDB::Open(options, db_name, &db->opt_txn_db); if (s.ok()) { db->db = db->opt_txn_db->GetBaseDB(); } } else if (FLAGS_transaction_db) { TransactionDB* ptr; TransactionDBOptions txn_db_options; s = TransactionDB::Open(options, txn_db_options, db_name, &ptr); if (s.ok()) { db->db = ptr; } #endif // ROCKSDB_LITE } else { s = DB::Open(options, db_name, &db->db); } if (!s.ok()) { fprintf(stderr, "open error: %s\n", s.ToString().c_str()); exit(1); } } enum WriteMode { RANDOM, SEQUENTIAL, UNIQUE_RANDOM }; void WriteSeq(ThreadState* thread) { DoWrite(thread, SEQUENTIAL); } void WriteRandom(ThreadState* thread) { DoWrite(thread, RANDOM); } void WriteUniqueRandom(ThreadState* thread) { DoWrite(thread, UNIQUE_RANDOM); } class KeyGenerator { public: KeyGenerator(Random64* rand, WriteMode mode, uint64_t num, uint64_t num_per_set = 64 * 1024) : rand_(rand), mode_(mode), num_(num), next_(0) { if (mode_ == UNIQUE_RANDOM) { // NOTE: if memory consumption of this approach becomes a concern, // we can either break it into pieces and only random shuffle a section // each time. Alternatively, use a bit map implementation // (https://reviews.facebook.net/differential/diff/54627/) values_.resize(num_); for (uint64_t i = 0; i < num_; ++i) { values_[i] = i; } std::shuffle( values_.begin(), values_.end(), std::default_random_engine(static_cast(FLAGS_seed))); } } uint64_t Next() { switch (mode_) { case SEQUENTIAL: return next_++; case RANDOM: return rand_->Next() % num_; case UNIQUE_RANDOM: return values_[next_++]; } assert(false); return std::numeric_limits::max(); } private: Random64* rand_; WriteMode mode_; const uint64_t num_; uint64_t next_; std::vector values_; }; DB* SelectDB(ThreadState* thread) { return SelectDBWithCfh(thread)->db; } DBWithColumnFamilies* SelectDBWithCfh(ThreadState* thread) { return SelectDBWithCfh(thread->rand.Next()); } DBWithColumnFamilies* SelectDBWithCfh(uint64_t rand_int) { if (db_.db != nullptr) { return &db_; } else { return &multi_dbs_[rand_int % multi_dbs_.size()]; } } void DoWrite(ThreadState* thread, WriteMode write_mode) { const int test_duration = write_mode == RANDOM ? FLAGS_duration : 0; const int64_t num_ops = writes_ == 0 ? num_ : writes_; size_t num_key_gens = 1; if (db_.db == nullptr) { num_key_gens = multi_dbs_.size(); } std::vector> key_gens(num_key_gens); int64_t max_ops = num_ops * num_key_gens; int64_t ops_per_stage = max_ops; if (FLAGS_num_column_families > 1 && FLAGS_num_hot_column_families > 0) { ops_per_stage = (max_ops - 1) / (FLAGS_num_column_families / FLAGS_num_hot_column_families) + 1; } Duration duration(test_duration, max_ops, ops_per_stage); for (size_t i = 0; i < num_key_gens; i++) { key_gens[i].reset(new KeyGenerator(&(thread->rand), write_mode, num_, ops_per_stage)); } if (num_ != FLAGS_num) { char msg[100]; snprintf(msg, sizeof(msg), "(%" PRIu64 " ops)", num_); thread->stats.AddMessage(msg); } RandomGenerator gen; WriteBatch batch; Status s; int64_t bytes = 0; std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); int64_t stage = 0; while (!duration.Done(entries_per_batch_)) { if (duration.GetStage() != stage) { stage = duration.GetStage(); if (db_.db != nullptr) { db_.CreateNewCf(open_options_, stage); } else { for (auto& db : multi_dbs_) { db.CreateNewCf(open_options_, stage); } } } size_t id = thread->rand.Next() % num_key_gens; DBWithColumnFamilies* db_with_cfh = SelectDBWithCfh(id); batch.Clear(); if (thread->shared->write_rate_limiter.get() != nullptr) { thread->shared->write_rate_limiter->Request( entries_per_batch_ * (value_size_ + key_size_), Env::IO_HIGH); // Set time at which last op finished to Now() to hide latency and // sleep from rate limiter. Also, do the check once per batch, not // once per write. thread->stats.ResetLastOpTime(); } for (int64_t j = 0; j < entries_per_batch_; j++) { int64_t rand_num = key_gens[id]->Next(); GenerateKeyFromInt(rand_num, FLAGS_num, &key); if (FLAGS_num_column_families <= 1) { batch.Put(key, gen.Generate(value_size_)); } else { // We use same rand_num as seed for key and column family so that we // can deterministically find the cfh corresponding to a particular // key while reading the key. batch.Put(db_with_cfh->GetCfh(rand_num), key, gen.Generate(value_size_)); } bytes += value_size_ + key_size_; } s = db_with_cfh->db->Write(write_options_, &batch); thread->stats.FinishedOps(db_with_cfh, db_with_cfh->db, entries_per_batch_, kWrite); if (!s.ok()) { fprintf(stderr, "put error: %s\n", s.ToString().c_str()); exit(1); } } thread->stats.AddBytes(bytes); } void ReadSequential(ThreadState* thread) { if (db_.db != nullptr) { ReadSequential(thread, db_.db); } else { for (const auto& db_with_cfh : multi_dbs_) { ReadSequential(thread, db_with_cfh.db); } } } void ReadSequential(ThreadState* thread, DB* db) { ReadOptions options(FLAGS_verify_checksum, true); options.tailing = FLAGS_use_tailing_iterator; Iterator* iter = db->NewIterator(options); int64_t i = 0; int64_t bytes = 0; for (iter->SeekToFirst(); i < reads_ && iter->Valid(); iter->Next()) { bytes += iter->key().size() + iter->value().size(); thread->stats.FinishedOps(nullptr, db, 1, kRead); ++i; } delete iter; thread->stats.AddBytes(bytes); } void ReadReverse(ThreadState* thread) { if (db_.db != nullptr) { ReadReverse(thread, db_.db); } else { for (const auto& db_with_cfh : multi_dbs_) { ReadReverse(thread, db_with_cfh.db); } } } void ReadReverse(ThreadState* thread, DB* db) { Iterator* iter = db->NewIterator(ReadOptions(FLAGS_verify_checksum, true)); int64_t i = 0; int64_t bytes = 0; for (iter->SeekToLast(); i < reads_ && iter->Valid(); iter->Prev()) { bytes += iter->key().size() + iter->value().size(); thread->stats.FinishedOps(nullptr, db, 1, kRead); ++i; } delete iter; thread->stats.AddBytes(bytes); } void ReadRandomFast(ThreadState* thread) { int64_t read = 0; int64_t found = 0; int64_t nonexist = 0; ReadOptions options(FLAGS_verify_checksum, true); std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); std::string value; DB* db = SelectDBWithCfh(thread)->db; int64_t pot = 1; while (pot < FLAGS_num) { pot <<= 1; } Duration duration(FLAGS_duration, reads_); do { for (int i = 0; i < 100; ++i) { int64_t key_rand = thread->rand.Next() & (pot - 1); GenerateKeyFromInt(key_rand, FLAGS_num, &key); ++read; auto status = db->Get(options, key, &value); if (status.ok()) { ++found; } else if (!status.IsNotFound()) { fprintf(stderr, "Get returned an error: %s\n", status.ToString().c_str()); abort(); } if (key_rand >= FLAGS_num) { ++nonexist; } } thread->stats.FinishedOps(nullptr, db, 100, kRead); } while (!duration.Done(100)); char msg[100]; snprintf(msg, sizeof(msg), "(%" PRIu64 " of %" PRIu64 " found, " "issued %" PRIu64 " non-exist keys)\n", found, read, nonexist); thread->stats.AddMessage(msg); if (FLAGS_perf_level > 0) { thread->stats.AddMessage(perf_context.ToString()); } } int64_t GetRandomKey(Random64* rand) { uint64_t rand_int = rand->Next(); int64_t key_rand; if (read_random_exp_range_ == 0) { key_rand = rand_int % FLAGS_num; } else { const uint64_t kBigInt = static_cast(1U) << 62; long double order = -static_cast(rand_int % kBigInt) / static_cast(kBigInt) * read_random_exp_range_; long double exp_ran = std::exp(order); uint64_t rand_num = static_cast(exp_ran * static_cast(FLAGS_num)); // Map to a different number to avoid locality. const uint64_t kBigPrime = 0x5bd1e995; // Overflow is like %(2^64). Will have little impact of results. key_rand = static_cast((rand_num * kBigPrime) % FLAGS_num); } return key_rand; } void ReadRandom(ThreadState* thread) { int64_t read = 0; int64_t found = 0; int64_t bytes = 0; ReadOptions options(FLAGS_verify_checksum, true); std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); std::string value; Duration duration(FLAGS_duration, reads_); while (!duration.Done(1)) { DBWithColumnFamilies* db_with_cfh = SelectDBWithCfh(thread); // We use same key_rand as seed for key and column family so that we can // deterministically find the cfh corresponding to a particular key, as it // is done in DoWrite method. int64_t key_rand = GetRandomKey(&thread->rand); GenerateKeyFromInt(key_rand, FLAGS_num, &key); read++; Status s; if (FLAGS_num_column_families > 1) { s = db_with_cfh->db->Get(options, db_with_cfh->GetCfh(key_rand), key, &value); } else { s = db_with_cfh->db->Get(options, key, &value); } if (s.ok()) { found++; bytes += key.size() + value.size(); } else if (!s.IsNotFound()) { fprintf(stderr, "Get returned an error: %s\n", s.ToString().c_str()); abort(); } thread->stats.FinishedOps(db_with_cfh, db_with_cfh->db, 1, kRead); } char msg[100]; snprintf(msg, sizeof(msg), "(%" PRIu64 " of %" PRIu64 " found)\n", found, read); thread->stats.AddBytes(bytes); thread->stats.AddMessage(msg); if (FLAGS_perf_level > 0) { thread->stats.AddMessage(perf_context.ToString()); } } // Calls MultiGet over a list of keys from a random distribution. // Returns the total number of keys found. void MultiReadRandom(ThreadState* thread) { int64_t read = 0; int64_t found = 0; ReadOptions options(FLAGS_verify_checksum, true); std::vector keys; std::vector > key_guards; std::vector values(entries_per_batch_); while (static_cast(keys.size()) < entries_per_batch_) { key_guards.push_back(std::unique_ptr()); keys.push_back(AllocateKey(&key_guards.back())); } Duration duration(FLAGS_duration, reads_); while (!duration.Done(1)) { DB* db = SelectDB(thread); for (int64_t i = 0; i < entries_per_batch_; ++i) { GenerateKeyFromInt(GetRandomKey(&thread->rand), FLAGS_num, &keys[i]); } std::vector statuses = db->MultiGet(options, keys, &values); assert(static_cast(statuses.size()) == entries_per_batch_); read += entries_per_batch_; for (int64_t i = 0; i < entries_per_batch_; ++i) { if (statuses[i].ok()) { ++found; } else if (!statuses[i].IsNotFound()) { fprintf(stderr, "MultiGet returned an error: %s\n", statuses[i].ToString().c_str()); abort(); } } thread->stats.FinishedOps(nullptr, db, entries_per_batch_, kRead); } char msg[100]; snprintf(msg, sizeof(msg), "(%" PRIu64 " of %" PRIu64 " found)", found, read); thread->stats.AddMessage(msg); } void IteratorCreation(ThreadState* thread) { Duration duration(FLAGS_duration, reads_); ReadOptions options(FLAGS_verify_checksum, true); while (!duration.Done(1)) { DB* db = SelectDB(thread); Iterator* iter = db->NewIterator(options); delete iter; thread->stats.FinishedOps(nullptr, db, 1, kOthers); } } void IteratorCreationWhileWriting(ThreadState* thread) { if (thread->tid > 0) { IteratorCreation(thread); } else { BGWriter(thread, kWrite); } } void SeekRandom(ThreadState* thread) { int64_t read = 0; int64_t found = 0; int64_t bytes = 0; ReadOptions options(FLAGS_verify_checksum, true); options.tailing = FLAGS_use_tailing_iterator; Iterator* single_iter = nullptr; std::vector multi_iters; if (db_.db != nullptr) { single_iter = db_.db->NewIterator(options); } else { for (const auto& db_with_cfh : multi_dbs_) { multi_iters.push_back(db_with_cfh.db->NewIterator(options)); } } std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); Duration duration(FLAGS_duration, reads_); char value_buffer[256]; while (!duration.Done(1)) { if (!FLAGS_use_tailing_iterator) { if (db_.db != nullptr) { delete single_iter; single_iter = db_.db->NewIterator(options); } else { for (auto iter : multi_iters) { delete iter; } multi_iters.clear(); for (const auto& db_with_cfh : multi_dbs_) { multi_iters.push_back(db_with_cfh.db->NewIterator(options)); } } } // Pick a Iterator to use Iterator* iter_to_use = single_iter; if (single_iter == nullptr) { iter_to_use = multi_iters[thread->rand.Next() % multi_iters.size()]; } GenerateKeyFromInt(thread->rand.Next() % FLAGS_num, FLAGS_num, &key); iter_to_use->Seek(key); read++; if (iter_to_use->Valid() && iter_to_use->key().compare(key) == 0) { found++; } for (int j = 0; j < FLAGS_seek_nexts && iter_to_use->Valid(); ++j) { // Copy out iterator's value to make sure we read them. Slice value = iter_to_use->value(); memcpy(value_buffer, value.data(), std::min(value.size(), sizeof(value_buffer))); bytes += iter_to_use->key().size() + iter_to_use->value().size(); if (!FLAGS_reverse_iterator) { iter_to_use->Next(); } else { iter_to_use->Prev(); } assert(iter_to_use->status().ok()); } thread->stats.FinishedOps(&db_, db_.db, 1, kSeek); } delete single_iter; for (auto iter : multi_iters) { delete iter; } char msg[100]; snprintf(msg, sizeof(msg), "(%" PRIu64 " of %" PRIu64 " found)\n", found, read); thread->stats.AddBytes(bytes); thread->stats.AddMessage(msg); if (FLAGS_perf_level > 0) { thread->stats.AddMessage(perf_context.ToString()); } } void SeekRandomWhileWriting(ThreadState* thread) { if (thread->tid > 0) { SeekRandom(thread); } else { BGWriter(thread, kWrite); } } void SeekRandomWhileMerging(ThreadState* thread) { if (thread->tid > 0) { SeekRandom(thread); } else { BGWriter(thread, kMerge); } } void DoDelete(ThreadState* thread, bool seq) { WriteBatch batch; Duration duration(seq ? 0 : FLAGS_duration, num_); int64_t i = 0; std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); while (!duration.Done(entries_per_batch_)) { DB* db = SelectDB(thread); batch.Clear(); for (int64_t j = 0; j < entries_per_batch_; ++j) { const int64_t k = seq ? i + j : (thread->rand.Next() % FLAGS_num); GenerateKeyFromInt(k, FLAGS_num, &key); batch.Delete(key); } auto s = db->Write(write_options_, &batch); thread->stats.FinishedOps(nullptr, db, entries_per_batch_, kDelete); if (!s.ok()) { fprintf(stderr, "del error: %s\n", s.ToString().c_str()); exit(1); } i += entries_per_batch_; } } void DeleteSeq(ThreadState* thread) { DoDelete(thread, true); } void DeleteRandom(ThreadState* thread) { DoDelete(thread, false); } void ReadWhileWriting(ThreadState* thread) { if (thread->tid > 0) { ReadRandom(thread); } else { BGWriter(thread, kWrite); } } void ReadWhileMerging(ThreadState* thread) { if (thread->tid > 0) { ReadRandom(thread); } else { BGWriter(thread, kMerge); } } void BGWriter(ThreadState* thread, enum OperationType write_merge) { // Special thread that keeps writing until other threads are done. RandomGenerator gen; int64_t bytes = 0; std::unique_ptr write_rate_limiter; if (FLAGS_benchmark_write_rate_limit > 0) { write_rate_limiter.reset( NewGenericRateLimiter(FLAGS_benchmark_write_rate_limit)); } // Don't merge stats from this thread with the readers. thread->stats.SetExcludeFromMerge(); std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); while (true) { DB* db = SelectDB(thread); { MutexLock l(&thread->shared->mu); if (thread->shared->num_done + 1 >= thread->shared->num_initialized) { // Other threads have finished break; } } GenerateKeyFromInt(thread->rand.Next() % FLAGS_num, FLAGS_num, &key); Status s; if (write_merge == kWrite) { s = db->Put(write_options_, key, gen.Generate(value_size_)); } else { s = db->Merge(write_options_, key, gen.Generate(value_size_)); } if (!s.ok()) { fprintf(stderr, "put or merge error: %s\n", s.ToString().c_str()); exit(1); } bytes += key.size() + value_size_; thread->stats.FinishedOps(&db_, db_.db, 1, kWrite); if (FLAGS_benchmark_write_rate_limit > 0) { write_rate_limiter->Request( entries_per_batch_ * (value_size_ + key_size_), Env::IO_HIGH); } } thread->stats.AddBytes(bytes); } // Given a key K and value V, this puts (K+"0", V), (K+"1", V), (K+"2", V) // in DB atomically i.e in a single batch. Also refer GetMany. Status PutMany(DB* db, const WriteOptions& writeoptions, const Slice& key, const Slice& value) { std::string suffixes[3] = {"2", "1", "0"}; std::string keys[3]; WriteBatch batch; Status s; for (int i = 0; i < 3; i++) { keys[i] = key.ToString() + suffixes[i]; batch.Put(keys[i], value); } s = db->Write(writeoptions, &batch); return s; } // Given a key K, this deletes (K+"0", V), (K+"1", V), (K+"2", V) // in DB atomically i.e in a single batch. Also refer GetMany. Status DeleteMany(DB* db, const WriteOptions& writeoptions, const Slice& key) { std::string suffixes[3] = {"1", "2", "0"}; std::string keys[3]; WriteBatch batch; Status s; for (int i = 0; i < 3; i++) { keys[i] = key.ToString() + suffixes[i]; batch.Delete(keys[i]); } s = db->Write(writeoptions, &batch); return s; } // Given a key K and value V, this gets values for K+"0", K+"1" and K+"2" // in the same snapshot, and verifies that all the values are identical. // ASSUMES that PutMany was used to put (K, V) into the DB. Status GetMany(DB* db, const ReadOptions& readoptions, const Slice& key, std::string* value) { std::string suffixes[3] = {"0", "1", "2"}; std::string keys[3]; Slice key_slices[3]; std::string values[3]; ReadOptions readoptionscopy = readoptions; readoptionscopy.snapshot = db->GetSnapshot(); Status s; for (int i = 0; i < 3; i++) { keys[i] = key.ToString() + suffixes[i]; key_slices[i] = keys[i]; s = db->Get(readoptionscopy, key_slices[i], value); if (!s.ok() && !s.IsNotFound()) { fprintf(stderr, "get error: %s\n", s.ToString().c_str()); values[i] = ""; // we continue after error rather than exiting so that we can // find more errors if any } else if (s.IsNotFound()) { values[i] = ""; } else { values[i] = *value; } } db->ReleaseSnapshot(readoptionscopy.snapshot); if ((values[0] != values[1]) || (values[1] != values[2])) { fprintf(stderr, "inconsistent values for key %s: %s, %s, %s\n", key.ToString().c_str(), values[0].c_str(), values[1].c_str(), values[2].c_str()); // we continue after error rather than exiting so that we can // find more errors if any } return s; } // Differs from readrandomwriterandom in the following ways: // (a) Uses GetMany/PutMany to read/write key values. Refer to those funcs. // (b) Does deletes as well (per FLAGS_deletepercent) // (c) In order to achieve high % of 'found' during lookups, and to do // multiple writes (including puts and deletes) it uses upto // FLAGS_numdistinct distinct keys instead of FLAGS_num distinct keys. // (d) Does not have a MultiGet option. void RandomWithVerify(ThreadState* thread) { ReadOptions options(FLAGS_verify_checksum, true); RandomGenerator gen; std::string value; int64_t found = 0; int get_weight = 0; int put_weight = 0; int delete_weight = 0; int64_t gets_done = 0; int64_t puts_done = 0; int64_t deletes_done = 0; std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); // the number of iterations is the larger of read_ or write_ for (int64_t i = 0; i < readwrites_; i++) { DB* db = SelectDB(thread); if (get_weight == 0 && put_weight == 0 && delete_weight == 0) { // one batch completed, reinitialize for next batch get_weight = FLAGS_readwritepercent; delete_weight = FLAGS_deletepercent; put_weight = 100 - get_weight - delete_weight; } GenerateKeyFromInt(thread->rand.Next() % FLAGS_numdistinct, FLAGS_numdistinct, &key); if (get_weight > 0) { // do all the gets first Status s = GetMany(db, options, key, &value); if (!s.ok() && !s.IsNotFound()) { fprintf(stderr, "getmany error: %s\n", s.ToString().c_str()); // we continue after error rather than exiting so that we can // find more errors if any } else if (!s.IsNotFound()) { found++; } get_weight--; gets_done++; thread->stats.FinishedOps(&db_, db_.db, 1, kRead); } else if (put_weight > 0) { // then do all the corresponding number of puts // for all the gets we have done earlier Status s = PutMany(db, write_options_, key, gen.Generate(value_size_)); if (!s.ok()) { fprintf(stderr, "putmany error: %s\n", s.ToString().c_str()); exit(1); } put_weight--; puts_done++; thread->stats.FinishedOps(&db_, db_.db, 1, kWrite); } else if (delete_weight > 0) { Status s = DeleteMany(db, write_options_, key); if (!s.ok()) { fprintf(stderr, "deletemany error: %s\n", s.ToString().c_str()); exit(1); } delete_weight--; deletes_done++; thread->stats.FinishedOps(&db_, db_.db, 1, kDelete); } } char msg[100]; snprintf(msg, sizeof(msg), "( get:%" PRIu64 " put:%" PRIu64 " del:%" PRIu64 " total:%" \ PRIu64 " found:%" PRIu64 ")", gets_done, puts_done, deletes_done, readwrites_, found); thread->stats.AddMessage(msg); } // This is different from ReadWhileWriting because it does not use // an extra thread. void ReadRandomWriteRandom(ThreadState* thread) { ReadOptions options(FLAGS_verify_checksum, true); RandomGenerator gen; std::string value; int64_t found = 0; int get_weight = 0; int put_weight = 0; int64_t reads_done = 0; int64_t writes_done = 0; Duration duration(FLAGS_duration, readwrites_); std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); // the number of iterations is the larger of read_ or write_ while (!duration.Done(1)) { DB* db = SelectDB(thread); GenerateKeyFromInt(thread->rand.Next() % FLAGS_num, FLAGS_num, &key); if (get_weight == 0 && put_weight == 0) { // one batch completed, reinitialize for next batch get_weight = FLAGS_readwritepercent; put_weight = 100 - get_weight; } if (get_weight > 0) { // do all the gets first Status s = db->Get(options, key, &value); if (!s.ok() && !s.IsNotFound()) { fprintf(stderr, "get error: %s\n", s.ToString().c_str()); // we continue after error rather than exiting so that we can // find more errors if any } else if (!s.IsNotFound()) { found++; } get_weight--; reads_done++; thread->stats.FinishedOps(nullptr, db, 1, kRead); } else if (put_weight > 0) { // then do all the corresponding number of puts // for all the gets we have done earlier Status s = db->Put(write_options_, key, gen.Generate(value_size_)); if (!s.ok()) { fprintf(stderr, "put error: %s\n", s.ToString().c_str()); exit(1); } put_weight--; writes_done++; thread->stats.FinishedOps(nullptr, db, 1, kWrite); } } char msg[100]; snprintf(msg, sizeof(msg), "( reads:%" PRIu64 " writes:%" PRIu64 \ " total:%" PRIu64 " found:%" PRIu64 ")", reads_done, writes_done, readwrites_, found); thread->stats.AddMessage(msg); } // // Read-modify-write for random keys void UpdateRandom(ThreadState* thread) { ReadOptions options(FLAGS_verify_checksum, true); RandomGenerator gen; std::string value; int64_t found = 0; int64_t bytes = 0; Duration duration(FLAGS_duration, readwrites_); std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); // the number of iterations is the larger of read_ or write_ while (!duration.Done(1)) { DB* db = SelectDB(thread); GenerateKeyFromInt(thread->rand.Next() % FLAGS_num, FLAGS_num, &key); auto status = db->Get(options, key, &value); if (status.ok()) { ++found; bytes += key.size() + value.size(); } else if (!status.IsNotFound()) { fprintf(stderr, "Get returned an error: %s\n", status.ToString().c_str()); abort(); } Status s = db->Put(write_options_, key, gen.Generate(value_size_)); if (!s.ok()) { fprintf(stderr, "put error: %s\n", s.ToString().c_str()); exit(1); } bytes += key.size() + value_size_; thread->stats.FinishedOps(nullptr, db, 1, kUpdate); } char msg[100]; snprintf(msg, sizeof(msg), "( updates:%" PRIu64 " found:%" PRIu64 ")", readwrites_, found); thread->stats.AddBytes(bytes); thread->stats.AddMessage(msg); } // Read-modify-write for random keys. // Each operation causes the key grow by value_size (simulating an append). // Generally used for benchmarking against merges of similar type void AppendRandom(ThreadState* thread) { ReadOptions options(FLAGS_verify_checksum, true); RandomGenerator gen; std::string value; int64_t found = 0; int64_t bytes = 0; std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); // The number of iterations is the larger of read_ or write_ Duration duration(FLAGS_duration, readwrites_); while (!duration.Done(1)) { DB* db = SelectDB(thread); GenerateKeyFromInt(thread->rand.Next() % FLAGS_num, FLAGS_num, &key); auto status = db->Get(options, key, &value); if (status.ok()) { ++found; bytes += key.size() + value.size(); } else if (!status.IsNotFound()) { fprintf(stderr, "Get returned an error: %s\n", status.ToString().c_str()); abort(); } else { // If not existing, then just assume an empty string of data value.clear(); } // Update the value (by appending data) Slice operand = gen.Generate(value_size_); if (value.size() > 0) { // Use a delimiter to match the semantics for StringAppendOperator value.append(1,','); } value.append(operand.data(), operand.size()); // Write back to the database Status s = db->Put(write_options_, key, value); if (!s.ok()) { fprintf(stderr, "put error: %s\n", s.ToString().c_str()); exit(1); } bytes += key.size() + value.size(); thread->stats.FinishedOps(nullptr, db, 1, kUpdate); } char msg[100]; snprintf(msg, sizeof(msg), "( updates:%" PRIu64 " found:%" PRIu64 ")", readwrites_, found); thread->stats.AddBytes(bytes); thread->stats.AddMessage(msg); } // Read-modify-write for random keys (using MergeOperator) // The merge operator to use should be defined by FLAGS_merge_operator // Adjust FLAGS_value_size so that the keys are reasonable for this operator // Assumes that the merge operator is non-null (i.e.: is well-defined) // // For example, use FLAGS_merge_operator="uint64add" and FLAGS_value_size=8 // to simulate random additions over 64-bit integers using merge. // // The number of merges on the same key can be controlled by adjusting // FLAGS_merge_keys. void MergeRandom(ThreadState* thread) { RandomGenerator gen; int64_t bytes = 0; std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); // The number of iterations is the larger of read_ or write_ Duration duration(FLAGS_duration, readwrites_); while (!duration.Done(1)) { DB* db = SelectDB(thread); GenerateKeyFromInt(thread->rand.Next() % merge_keys_, merge_keys_, &key); Status s = db->Merge(write_options_, key, gen.Generate(value_size_)); if (!s.ok()) { fprintf(stderr, "merge error: %s\n", s.ToString().c_str()); exit(1); } bytes += key.size() + value_size_; thread->stats.FinishedOps(nullptr, db, 1, kMerge); } // Print some statistics char msg[100]; snprintf(msg, sizeof(msg), "( updates:%" PRIu64 ")", readwrites_); thread->stats.AddBytes(bytes); thread->stats.AddMessage(msg); } // Read and merge random keys. The amount of reads and merges are controlled // by adjusting FLAGS_num and FLAGS_mergereadpercent. The number of distinct // keys (and thus also the number of reads and merges on the same key) can be // adjusted with FLAGS_merge_keys. // // As with MergeRandom, the merge operator to use should be defined by // FLAGS_merge_operator. void ReadRandomMergeRandom(ThreadState* thread) { ReadOptions options(FLAGS_verify_checksum, true); RandomGenerator gen; std::string value; int64_t num_hits = 0; int64_t num_gets = 0; int64_t num_merges = 0; size_t max_length = 0; std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); // the number of iterations is the larger of read_ or write_ Duration duration(FLAGS_duration, readwrites_); while (!duration.Done(1)) { DB* db = SelectDB(thread); GenerateKeyFromInt(thread->rand.Next() % merge_keys_, merge_keys_, &key); bool do_merge = int(thread->rand.Next() % 100) < FLAGS_mergereadpercent; if (do_merge) { Status s = db->Merge(write_options_, key, gen.Generate(value_size_)); if (!s.ok()) { fprintf(stderr, "merge error: %s\n", s.ToString().c_str()); exit(1); } num_merges++; thread->stats.FinishedOps(nullptr, db, 1, kMerge); } else { Status s = db->Get(options, key, &value); if (value.length() > max_length) max_length = value.length(); if (!s.ok() && !s.IsNotFound()) { fprintf(stderr, "get error: %s\n", s.ToString().c_str()); // we continue after error rather than exiting so that we can // find more errors if any } else if (!s.IsNotFound()) { num_hits++; } num_gets++; thread->stats.FinishedOps(nullptr, db, 1, kRead); } } char msg[100]; snprintf(msg, sizeof(msg), "(reads:%" PRIu64 " merges:%" PRIu64 " total:%" PRIu64 " hits:%" PRIu64 " maxlength:%" ROCKSDB_PRIszt ")", num_gets, num_merges, readwrites_, num_hits, max_length); thread->stats.AddMessage(msg); } void WriteSeqSeekSeq(ThreadState* thread) { writes_ = FLAGS_num; DoWrite(thread, SEQUENTIAL); // exclude writes from the ops/sec calculation thread->stats.Start(thread->tid); DB* db = SelectDB(thread); std::unique_ptr iter( db->NewIterator(ReadOptions(FLAGS_verify_checksum, true))); std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); for (int64_t i = 0; i < FLAGS_num; ++i) { GenerateKeyFromInt(i, FLAGS_num, &key); iter->Seek(key); assert(iter->Valid() && iter->key() == key); thread->stats.FinishedOps(nullptr, db, 1, kSeek); for (int j = 0; j < FLAGS_seek_nexts && i + 1 < FLAGS_num; ++j) { if (!FLAGS_reverse_iterator) { iter->Next(); } else { iter->Prev(); } GenerateKeyFromInt(++i, FLAGS_num, &key); assert(iter->Valid() && iter->key() == key); thread->stats.FinishedOps(nullptr, db, 1, kSeek); } iter->Seek(key); assert(iter->Valid() && iter->key() == key); thread->stats.FinishedOps(nullptr, db, 1, kSeek); } } #ifndef ROCKSDB_LITE // This benchmark stress tests Transactions. For a given --duration (or // total number of --writes, a Transaction will perform a read-modify-write // to increment the value of a key in each of N(--transaction-sets) sets of // keys (where each set has --num keys). If --threads is set, this will be // done in parallel. // // To test transactions, use --transaction_db=true. Not setting this // parameter // will run the same benchmark without transactions. // // RandomTransactionVerify() will then validate the correctness of the results // by checking if the sum of all keys in each set is the same. void RandomTransaction(ThreadState* thread) { ReadOptions options(FLAGS_verify_checksum, true); Duration duration(FLAGS_duration, readwrites_); ReadOptions read_options(FLAGS_verify_checksum, true); std::string value; DB* db = db_.db; uint64_t transactions_done = 0; uint64_t transactions_aborted = 0; Status s; uint64_t num_prefix_ranges = FLAGS_transaction_sets; if (num_prefix_ranges == 0 || num_prefix_ranges > 9999) { fprintf(stderr, "invalid value for transaction_sets\n"); abort(); } if (FLAGS_num_multi_db > 1) { fprintf(stderr, "Cannot run RandomTransaction benchmark with " "FLAGS_multi_db > 1."); abort(); } while (!duration.Done(1)) { Transaction* txn = nullptr; WriteBatch* batch = nullptr; if (FLAGS_optimistic_transaction_db) { txn = db_.opt_txn_db->BeginTransaction(write_options_); assert(txn); } else if (FLAGS_transaction_db) { TransactionDB* txn_db = reinterpret_cast(db_.db); TransactionOptions txn_options; txn_options.lock_timeout = FLAGS_transaction_lock_timeout; txn = txn_db->BeginTransaction(write_options_, txn_options); assert(txn); } else { batch = new WriteBatch(); } if (txn && FLAGS_transaction_set_snapshot) { txn->SetSnapshot(); } // pick a random number to use to increment a key in each set uint64_t incr = (thread->rand.Next() % 100) + 1; bool failed = false; // For each set, pick a key at random and increment it for (uint8_t i = 0; i < num_prefix_ranges; i++) { uint64_t int_value; char prefix_buf[5]; // key format: [SET#][random#] std::string rand_key = ToString(thread->rand.Next() % FLAGS_num); Slice base_key(rand_key); // Pad prefix appropriately so we can iterate over each set snprintf(prefix_buf, sizeof(prefix_buf), "%04d", i + 1); std::string full_key = std::string(prefix_buf) + base_key.ToString(); Slice key(full_key); if (txn) { s = txn->GetForUpdate(read_options, key, &value); } else { s = db->Get(read_options, key, &value); } if (s.ok()) { int_value = std::stoull(value); if (int_value == 0 || int_value == ULONG_MAX) { fprintf(stderr, "Get returned unexpected value: %s\n", value.c_str()); abort(); } } else if (s.IsNotFound()) { int_value = 0; } else if (!(s.IsBusy() || s.IsTimedOut() || s.IsTryAgain())) { fprintf(stderr, "Get returned an unexpected error: %s\n", s.ToString().c_str()); abort(); } else { failed = true; break; } if (FLAGS_transaction_sleep > 0) { FLAGS_env->SleepForMicroseconds(thread->rand.Next() % FLAGS_transaction_sleep); } std::string sum = ToString(int_value + incr); if (txn) { s = txn->Put(key, sum); if (!s.ok()) { // Since we did a GetForUpdate, Put should not fail. fprintf(stderr, "Put returned an unexpected error: %s\n", s.ToString().c_str()); abort(); } } else { batch->Put(key, sum); } } if (txn) { if (failed) { transactions_aborted++; txn->Rollback(); s = Status::OK(); } else { s = txn->Commit(); } } else { s = db->Write(write_options_, batch); } if (!s.ok()) { failed = true; // Ideally, we'd want to run this stress test with enough concurrency // on a small enough set of keys that we get some failed transactions // due to conflicts. if (FLAGS_optimistic_transaction_db && (s.IsBusy() || s.IsTimedOut() || s.IsTryAgain())) { transactions_aborted++; } else if (FLAGS_transaction_db && s.IsExpired()) { transactions_aborted++; } else { fprintf(stderr, "Unexpected write error: %s\n", s.ToString().c_str()); abort(); } } delete txn; delete batch; if (!failed) { thread->stats.FinishedOps(nullptr, db, 1, kOthers); } transactions_done++; } char msg[100]; if (FLAGS_optimistic_transaction_db || FLAGS_transaction_db) { snprintf(msg, sizeof(msg), "( transactions:%" PRIu64 " aborts:%" PRIu64 ")", transactions_done, transactions_aborted); } else { snprintf(msg, sizeof(msg), "( batches:%" PRIu64 " )", transactions_done); } thread->stats.AddMessage(msg); if (FLAGS_perf_level > 0) { thread->stats.AddMessage(perf_context.ToString()); } } // Verifies consistency of data after RandomTransaction() has been run. // Since each iteration of RandomTransaction() incremented a key in each set // by the same value, the sum of the keys in each set should be the same. void RandomTransactionVerify() { if (!FLAGS_transaction_db && !FLAGS_optimistic_transaction_db) { // transactions not used, nothing to verify. return; } uint64_t prev_total = 0; // For each set of keys with the same prefix, sum all the values for (uint32_t i = 0; i < FLAGS_transaction_sets; i++) { char prefix_buf[5]; snprintf(prefix_buf, sizeof(prefix_buf), "%04u", i + 1); uint64_t total = 0; Iterator* iter = db_.db->NewIterator(ReadOptions()); for (iter->Seek(Slice(prefix_buf, 4)); iter->Valid(); iter->Next()) { Slice key = iter->key(); // stop when we reach a different prefix if (key.ToString().compare(0, 4, prefix_buf) != 0) { break; } Slice value = iter->value(); uint64_t int_value = std::stoull(value.ToString()); if (int_value == 0 || int_value == ULONG_MAX) { fprintf(stderr, "Iter returned unexpected value: %s\n", value.ToString().c_str()); abort(); } total += int_value; } delete iter; if (i > 0) { if (total != prev_total) { fprintf(stderr, "RandomTransactionVerify found inconsistent totals. " "Set[%" PRIu32 "]: %" PRIu64 ", Set[%" PRIu32 "]: %" PRIu64 " \n", i - 1, prev_total, i, total); abort(); } } prev_total = total; } fprintf(stdout, "RandomTransactionVerify Success!\n"); } #endif // ROCKSDB_LITE // Writes and deletes random keys without overwriting keys. // // This benchmark is intended to partially replicate the behavior of MyRocks // secondary indices: All data is stored in keys and updates happen by // deleting the old version of the key and inserting the new version. void RandomReplaceKeys(ThreadState* thread) { std::unique_ptr key_guard; Slice key = AllocateKey(&key_guard); std::vector counters(FLAGS_numdistinct, 0); size_t max_counter = 50; RandomGenerator gen; Status s; DB* db = SelectDB(thread); for (int64_t i = 0; i < FLAGS_numdistinct; i++) { GenerateKeyFromInt(i * max_counter, FLAGS_num, &key); s = db->Put(write_options_, key, gen.Generate(value_size_)); if (!s.ok()) { fprintf(stderr, "Operation failed: %s\n", s.ToString().c_str()); exit(1); } } db->GetSnapshot(); std::default_random_engine generator; std::normal_distribution distribution(FLAGS_numdistinct / 2.0, FLAGS_stddev); Duration duration(FLAGS_duration, FLAGS_num); while (!duration.Done(1)) { int64_t rnd_id = static_cast(distribution(generator)); int64_t key_id = std::max(std::min(FLAGS_numdistinct - 1, rnd_id), static_cast(0)); GenerateKeyFromInt(key_id * max_counter + counters[key_id], FLAGS_num, &key); s = FLAGS_use_single_deletes ? db->SingleDelete(write_options_, key) : db->Delete(write_options_, key); if (s.ok()) { counters[key_id] = (counters[key_id] + 1) % max_counter; GenerateKeyFromInt(key_id * max_counter + counters[key_id], FLAGS_num, &key); s = db->Put(write_options_, key, Slice()); } if (!s.ok()) { fprintf(stderr, "Operation failed: %s\n", s.ToString().c_str()); exit(1); } thread->stats.FinishedOps(nullptr, db, 1, kOthers); } char msg[200]; snprintf(msg, sizeof(msg), "use single deletes: %d, " "standard deviation: %lf\n", FLAGS_use_single_deletes, FLAGS_stddev); thread->stats.AddMessage(msg); } void Compact(ThreadState* thread) { DB* db = SelectDB(thread); db->CompactRange(CompactRangeOptions(), nullptr, nullptr); } void PrintStats(const char* key) { if (db_.db != nullptr) { PrintStats(db_.db, key, false); } for (const auto& db_with_cfh : multi_dbs_) { PrintStats(db_with_cfh.db, key, true); } } void PrintStats(DB* db, const char* key, bool print_header = false) { if (print_header) { fprintf(stdout, "\n==== DB: %s ===\n", db->GetName().c_str()); } std::string stats; if (!db->GetProperty(key, &stats)) { stats = "(failed)"; } fprintf(stdout, "\n%s\n", stats.c_str()); } }; } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); SetUsageMessage(std::string("\nUSAGE:\n") + std::string(argv[0]) + " [OPTIONS]..."); ParseCommandLineFlags(&argc, &argv, true); FLAGS_compaction_style_e = (rocksdb::CompactionStyle) FLAGS_compaction_style; if (FLAGS_statistics) { dbstats = rocksdb::CreateDBStatistics(); } FLAGS_compaction_pri_e = (rocksdb::CompactionPri)FLAGS_compaction_pri; std::vector fanout = rocksdb::StringSplit( FLAGS_max_bytes_for_level_multiplier_additional, ','); for (size_t j = 0; j < fanout.size(); j++) { FLAGS_max_bytes_for_level_multiplier_additional_v.push_back( #ifndef CYGWIN std::stoi(fanout[j])); #else stoi(fanout[j])); #endif } FLAGS_compression_type_e = StringToCompressionType(FLAGS_compression_type.c_str()); if (!FLAGS_hdfs.empty()) { FLAGS_env = new rocksdb::HdfsEnv(FLAGS_hdfs); } if (!strcasecmp(FLAGS_compaction_fadvice.c_str(), "NONE")) FLAGS_compaction_fadvice_e = rocksdb::Options::NONE; else if (!strcasecmp(FLAGS_compaction_fadvice.c_str(), "NORMAL")) FLAGS_compaction_fadvice_e = rocksdb::Options::NORMAL; else if (!strcasecmp(FLAGS_compaction_fadvice.c_str(), "SEQUENTIAL")) FLAGS_compaction_fadvice_e = rocksdb::Options::SEQUENTIAL; else if (!strcasecmp(FLAGS_compaction_fadvice.c_str(), "WILLNEED")) FLAGS_compaction_fadvice_e = rocksdb::Options::WILLNEED; else { fprintf(stdout, "Unknown compaction fadvice:%s\n", FLAGS_compaction_fadvice.c_str()); } FLAGS_rep_factory = StringToRepFactory(FLAGS_memtablerep.c_str()); // The number of background threads should be at least as much the // max number of concurrent compactions. FLAGS_env->SetBackgroundThreads(FLAGS_max_background_compactions); FLAGS_env->SetBackgroundThreads(FLAGS_max_background_flushes, rocksdb::Env::Priority::HIGH); // Choose a location for the test database if none given with --db= if (FLAGS_db.empty()) { std::string default_db_path; rocksdb::Env::Default()->GetTestDirectory(&default_db_path); default_db_path += "/dbbench"; FLAGS_db = default_db_path; } if (FLAGS_stats_interval_seconds > 0) { // When both are set then FLAGS_stats_interval determines the frequency // at which the timer is checked for FLAGS_stats_interval_seconds FLAGS_stats_interval = 1000; } rocksdb::Benchmark benchmark; benchmark.Run(); return 0; } #endif // GFLAGS rocksdb-4.5.1/db/db_compaction_filter_test.cc000066400000000000000000000557021267530634000212700ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_test_util.h" #include "port/stack_trace.h" namespace rocksdb { static int cfilter_count = 0; // This is a static filter used for filtering // kvs during the compaction process. static std::string NEW_VALUE = "NewValue"; class DBTestCompactionFilter : public DBTestBase { public: DBTestCompactionFilter() : DBTestBase("/db_compaction_filter_test") {} }; class KeepFilter : public CompactionFilter { public: virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { cfilter_count++; return false; } virtual const char* Name() const override { return "KeepFilter"; } }; class DeleteFilter : public CompactionFilter { public: virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { cfilter_count++; return true; } virtual const char* Name() const override { return "DeleteFilter"; } }; class DeleteISFilter : public CompactionFilter { public: virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { cfilter_count++; int i = std::stoi(key.ToString()); if (i > 5 && i <= 105) { return true; } return false; } virtual bool IgnoreSnapshots() const override { return true; } virtual const char* Name() const override { return "DeleteFilter"; } }; class DelayFilter : public CompactionFilter { public: explicit DelayFilter(DBTestBase* d) : db_test(d) {} virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { db_test->env_->addon_time_.fetch_add(1000); return true; } virtual const char* Name() const override { return "DelayFilter"; } private: DBTestBase* db_test; }; class ConditionalFilter : public CompactionFilter { public: explicit ConditionalFilter(const std::string* filtered_value) : filtered_value_(filtered_value) {} virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { return value.ToString() == *filtered_value_; } virtual const char* Name() const override { return "ConditionalFilter"; } private: const std::string* filtered_value_; }; class ChangeFilter : public CompactionFilter { public: explicit ChangeFilter() {} virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { assert(new_value != nullptr); *new_value = NEW_VALUE; *value_changed = true; return false; } virtual const char* Name() const override { return "ChangeFilter"; } }; class KeepFilterFactory : public CompactionFilterFactory { public: explicit KeepFilterFactory(bool check_context = false, bool check_context_cf_id = false) : check_context_(check_context), check_context_cf_id_(check_context_cf_id), compaction_filter_created_(false) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { if (check_context_) { EXPECT_EQ(expect_full_compaction_.load(), context.is_full_compaction); EXPECT_EQ(expect_manual_compaction_.load(), context.is_manual_compaction); } if (check_context_cf_id_) { EXPECT_EQ(expect_cf_id_.load(), context.column_family_id); } compaction_filter_created_ = true; return std::unique_ptr(new KeepFilter()); } bool compaction_filter_created() const { return compaction_filter_created_; } virtual const char* Name() const override { return "KeepFilterFactory"; } bool check_context_; bool check_context_cf_id_; std::atomic_bool expect_full_compaction_; std::atomic_bool expect_manual_compaction_; std::atomic expect_cf_id_; bool compaction_filter_created_; }; class DeleteFilterFactory : public CompactionFilterFactory { public: virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { if (context.is_manual_compaction) { return std::unique_ptr(new DeleteFilter()); } else { return std::unique_ptr(nullptr); } } virtual const char* Name() const override { return "DeleteFilterFactory"; } }; // Delete Filter Factory which ignores snapshots class DeleteISFilterFactory : public CompactionFilterFactory { public: virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { if (context.is_manual_compaction) { return std::unique_ptr(new DeleteISFilter()); } else { return std::unique_ptr(nullptr); } } virtual const char* Name() const override { return "DeleteFilterFactory"; } }; class DelayFilterFactory : public CompactionFilterFactory { public: explicit DelayFilterFactory(DBTestBase* d) : db_test(d) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { return std::unique_ptr(new DelayFilter(db_test)); } virtual const char* Name() const override { return "DelayFilterFactory"; } private: DBTestBase* db_test; }; class ConditionalFilterFactory : public CompactionFilterFactory { public: explicit ConditionalFilterFactory(const Slice& filtered_value) : filtered_value_(filtered_value.ToString()) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { return std::unique_ptr( new ConditionalFilter(&filtered_value_)); } virtual const char* Name() const override { return "ConditionalFilterFactory"; } private: std::string filtered_value_; }; class ChangeFilterFactory : public CompactionFilterFactory { public: explicit ChangeFilterFactory() {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { return std::unique_ptr(new ChangeFilter()); } virtual const char* Name() const override { return "ChangeFilterFactory"; } }; #ifndef ROCKSDB_LITE TEST_F(DBTestCompactionFilter, CompactionFilter) { Options options = CurrentOptions(); options.max_open_files = -1; options.num_levels = 3; options.compaction_filter_factory = std::make_shared(); options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Write 100K keys, these are written to a few files in L0. const std::string value(10, 'x'); for (int i = 0; i < 100000; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); Put(1, key, value); } ASSERT_OK(Flush(1)); // Push all files to the highest level L2. Verify that // the compaction is each level invokes the filter for // all the keys in that level. cfilter_count = 0; dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); ASSERT_EQ(cfilter_count, 100000); cfilter_count = 0; dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); ASSERT_EQ(cfilter_count, 100000); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0); ASSERT_NE(NumTableFilesAtLevel(2, 1), 0); cfilter_count = 0; // All the files are in the lowest level. // Verify that all but the 100001st record // has sequence number zero. The 100001st record // is at the tip of this snapshot and cannot // be zeroed out. int count = 0; int total = 0; Arena arena; { ScopedArenaIterator iter( dbfull()->NewInternalIterator(&arena, handles_[1])); iter->SeekToFirst(); ASSERT_OK(iter->status()); while (iter->Valid()) { ParsedInternalKey ikey(Slice(), 0, kTypeValue); ikey.sequence = -1; ASSERT_EQ(ParseInternalKey(iter->key(), &ikey), true); total++; if (ikey.sequence != 0) { count++; } iter->Next(); } } ASSERT_EQ(total, 100000); ASSERT_EQ(count, 1); // overwrite all the 100K keys once again. for (int i = 0; i < 100000; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); ASSERT_OK(Put(1, key, value)); } ASSERT_OK(Flush(1)); // push all files to the highest level L2. This // means that all keys should pass at least once // via the compaction filter cfilter_count = 0; dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); ASSERT_EQ(cfilter_count, 100000); cfilter_count = 0; dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); ASSERT_EQ(cfilter_count, 100000); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0); ASSERT_NE(NumTableFilesAtLevel(2, 1), 0); // create a new database with the compaction // filter in such a way that it deletes all keys options.compaction_filter_factory = std::make_shared(); options.create_if_missing = true; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // write all the keys once again. for (int i = 0; i < 100000; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); ASSERT_OK(Put(1, key, value)); } ASSERT_OK(Flush(1)); ASSERT_NE(NumTableFilesAtLevel(0, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(2, 1), 0); // Push all files to the highest level L2. This // triggers the compaction filter to delete all keys, // verify that at the end of the compaction process, // nothing is left. cfilter_count = 0; dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); ASSERT_EQ(cfilter_count, 100000); cfilter_count = 0; dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); ASSERT_EQ(cfilter_count, 0); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0); { // Scan the entire database to ensure that nothing is left std::unique_ptr iter( db_->NewIterator(ReadOptions(), handles_[1])); iter->SeekToFirst(); count = 0; while (iter->Valid()) { count++; iter->Next(); } ASSERT_EQ(count, 0); } // The sequence number of the remaining record // is not zeroed out even though it is at the // level Lmax because this record is at the tip count = 0; { ScopedArenaIterator iter( dbfull()->NewInternalIterator(&arena, handles_[1])); iter->SeekToFirst(); ASSERT_OK(iter->status()); while (iter->Valid()) { ParsedInternalKey ikey(Slice(), 0, kTypeValue); ASSERT_EQ(ParseInternalKey(iter->key(), &ikey), true); ASSERT_NE(ikey.sequence, (unsigned)0); count++; iter->Next(); } ASSERT_EQ(count, 0); } } // Tests the edge case where compaction does not produce any output -- all // entries are deleted. The compaction should create bunch of 'DeleteFile' // entries in VersionEdit, but none of the 'AddFile's. TEST_F(DBTestCompactionFilter, CompactionFilterDeletesAll) { Options options; options.compaction_filter_factory = std::make_shared(); options.disable_auto_compactions = true; options.create_if_missing = true; options = CurrentOptions(options); DestroyAndReopen(options); // put some data for (int table = 0; table < 4; ++table) { for (int i = 0; i < 10 + table; ++i) { Put(ToString(table * 100 + i), "val"); } Flush(); } // this will produce empty file (delete compaction filter) ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ(0U, CountLiveFiles()); Reopen(options); Iterator* itr = db_->NewIterator(ReadOptions()); itr->SeekToFirst(); // empty db ASSERT_TRUE(!itr->Valid()); delete itr; } #endif // ROCKSDB_LITE TEST_F(DBTestCompactionFilter, CompactionFilterWithValueChange) { do { Options options; options.num_levels = 3; options.compaction_filter_factory = std::make_shared(); options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Write 100K+1 keys, these are written to a few files // in L0. We do this so that the current snapshot points // to the 100001 key.The compaction filter is not invoked // on keys that are visible via a snapshot because we // anyways cannot delete it. const std::string value(10, 'x'); for (int i = 0; i < 100001; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); Put(1, key, value); } // push all files to lower levels ASSERT_OK(Flush(1)); if (option_config_ != kUniversalCompactionMultiLevel && option_config_ != kUniversalSubcompactions) { dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); } else { dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); } // re-write all data again for (int i = 0; i < 100001; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); Put(1, key, value); } // push all files to lower levels. This should // invoke the compaction filter for all 100000 keys. ASSERT_OK(Flush(1)); if (option_config_ != kUniversalCompactionMultiLevel && option_config_ != kUniversalSubcompactions) { dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); } else { dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); } // verify that all keys now have the new value that // was set by the compaction process. for (int i = 0; i < 100001; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); std::string newvalue = Get(1, key); ASSERT_EQ(newvalue.compare(NEW_VALUE), 0); } } while (ChangeCompactOptions()); } TEST_F(DBTestCompactionFilter, CompactionFilterWithMergeOperator) { std::string one, two, three, four; PutFixed64(&one, 1); PutFixed64(&two, 2); PutFixed64(&three, 3); PutFixed64(&four, 4); Options options; options = CurrentOptions(options); options.create_if_missing = true; options.merge_operator = MergeOperators::CreateUInt64AddOperator(); options.num_levels = 3; // Filter out keys with value is 2. options.compaction_filter_factory = std::make_shared(two); DestroyAndReopen(options); // In the same compaction, a value type needs to be deleted based on // compaction filter, and there is a merge type for the key. compaction // filter result is ignored. ASSERT_OK(db_->Put(WriteOptions(), "foo", two)); ASSERT_OK(Flush()); ASSERT_OK(db_->Merge(WriteOptions(), "foo", one)); ASSERT_OK(Flush()); std::string newvalue = Get("foo"); ASSERT_EQ(newvalue, three); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); newvalue = Get("foo"); ASSERT_EQ(newvalue, three); // value key can be deleted based on compaction filter, leaving only // merge keys. ASSERT_OK(db_->Put(WriteOptions(), "bar", two)); ASSERT_OK(Flush()); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); newvalue = Get("bar"); ASSERT_EQ("NOT_FOUND", newvalue); ASSERT_OK(db_->Merge(WriteOptions(), "bar", two)); ASSERT_OK(Flush()); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); newvalue = Get("bar"); ASSERT_EQ(two, two); // Compaction filter never applies to merge keys. ASSERT_OK(db_->Put(WriteOptions(), "foobar", one)); ASSERT_OK(Flush()); ASSERT_OK(db_->Merge(WriteOptions(), "foobar", two)); ASSERT_OK(Flush()); newvalue = Get("foobar"); ASSERT_EQ(newvalue, three); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); newvalue = Get("foobar"); ASSERT_EQ(newvalue, three); // In the same compaction, both of value type and merge type keys need to be // deleted based on compaction filter, and there is a merge type for the key. // For both keys, compaction filter results are ignored. ASSERT_OK(db_->Put(WriteOptions(), "barfoo", two)); ASSERT_OK(Flush()); ASSERT_OK(db_->Merge(WriteOptions(), "barfoo", two)); ASSERT_OK(Flush()); newvalue = Get("barfoo"); ASSERT_EQ(newvalue, four); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); newvalue = Get("barfoo"); ASSERT_EQ(newvalue, four); } #ifndef ROCKSDB_LITE TEST_F(DBTestCompactionFilter, CompactionFilterContextManual) { KeepFilterFactory* filter = new KeepFilterFactory(true, true); Options options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.compaction_filter_factory.reset(filter); options.compression = kNoCompression; options.level0_file_num_compaction_trigger = 8; Reopen(options); int num_keys_per_file = 400; for (int j = 0; j < 3; j++) { // Write several keys. const std::string value(10, 'x'); for (int i = 0; i < num_keys_per_file; i++) { char key[100]; snprintf(key, sizeof(key), "B%08d%02d", i, j); Put(key, value); } dbfull()->TEST_FlushMemTable(); // Make sure next file is much smaller so automatic compaction will not // be triggered. num_keys_per_file /= 2; } dbfull()->TEST_WaitForCompact(); // Force a manual compaction cfilter_count = 0; filter->expect_manual_compaction_.store(true); filter->expect_full_compaction_.store(true); filter->expect_cf_id_.store(0); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(cfilter_count, 700); ASSERT_EQ(NumSortedRuns(0), 1); ASSERT_TRUE(filter->compaction_filter_created()); // Verify total number of keys is correct after manual compaction. { int count = 0; int total = 0; Arena arena; ScopedArenaIterator iter(dbfull()->NewInternalIterator(&arena)); iter->SeekToFirst(); ASSERT_OK(iter->status()); while (iter->Valid()) { ParsedInternalKey ikey(Slice(), 0, kTypeValue); ikey.sequence = -1; ASSERT_EQ(ParseInternalKey(iter->key(), &ikey), true); total++; if (ikey.sequence != 0) { count++; } iter->Next(); } ASSERT_EQ(total, 700); ASSERT_EQ(count, 2); } } #endif // ROCKSDB_LITE TEST_F(DBTestCompactionFilter, CompactionFilterContextCfId) { KeepFilterFactory* filter = new KeepFilterFactory(false, true); filter->expect_cf_id_.store(1); Options options = CurrentOptions(); options.compaction_filter_factory.reset(filter); options.compression = kNoCompression; options.level0_file_num_compaction_trigger = 2; CreateAndReopenWithCF({"pikachu"}, options); int num_keys_per_file = 400; for (int j = 0; j < 3; j++) { // Write several keys. const std::string value(10, 'x'); for (int i = 0; i < num_keys_per_file; i++) { char key[100]; snprintf(key, sizeof(key), "B%08d%02d", i, j); Put(1, key, value); } Flush(1); // Make sure next file is much smaller so automatic compaction will not // be triggered. num_keys_per_file /= 2; } dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(filter->compaction_filter_created()); } #ifndef ROCKSDB_LITE // Compaction filters should only be applied to records that are newer than the // latest snapshot. This test inserts records and applies a delete filter. TEST_F(DBTestCompactionFilter, CompactionFilterSnapshot) { Options options; options.compaction_filter_factory = std::make_shared(); options.disable_auto_compactions = true; options.create_if_missing = true; options = CurrentOptions(options); DestroyAndReopen(options); // Put some data. const Snapshot* snapshot = nullptr; for (int table = 0; table < 4; ++table) { for (int i = 0; i < 10; ++i) { Put(ToString(table * 100 + i), "val"); } Flush(); if (table == 0) { snapshot = db_->GetSnapshot(); } } assert(snapshot != nullptr); cfilter_count = 0; ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); // The filter should delete 10 records. ASSERT_EQ(30U, cfilter_count); // Release the snapshot and compact again -> now all records should be // removed. db_->ReleaseSnapshot(snapshot); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ(0U, CountLiveFiles()); } // Compaction filters should only be applied to records that are newer than the // latest snapshot. However, if the compaction filter asks to ignore snapshots // records newer than the snapshot will also be processed TEST_F(DBTestCompactionFilter, CompactionFilterIgnoreSnapshot) { std::string five = ToString(5); Options options; options.compaction_filter_factory = std::make_shared(); options.disable_auto_compactions = true; options.create_if_missing = true; options = CurrentOptions(options); DestroyAndReopen(options); // Put some data. const Snapshot* snapshot = nullptr; for (int table = 0; table < 4; ++table) { for (int i = 0; i < 10; ++i) { Put(ToString(table * 100 + i), "val"); } Flush(); if (table == 0) { snapshot = db_->GetSnapshot(); } } assert(snapshot != nullptr); cfilter_count = 0; ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); // The filter should delete 40 records. ASSERT_EQ(40U, cfilter_count); { // Scan the entire database as of the snapshot to ensure // that nothing is left ReadOptions read_options; read_options.snapshot = snapshot; std::unique_ptr iter(db_->NewIterator(read_options)); iter->SeekToFirst(); int count = 0; while (iter->Valid()) { count++; iter->Next(); } ASSERT_EQ(count, 6); read_options.snapshot = 0; std::unique_ptr iter1(db_->NewIterator(read_options)); iter1->SeekToFirst(); count = 0; while (iter1->Valid()) { count++; iter1->Next(); } // We have deleted 10 keys from 40 using the compaction filter // Keys 6-9 before the snapshot and 100-105 after the snapshot ASSERT_EQ(count, 30); } // Release the snapshot and compact again -> now all records should be // removed. db_->ReleaseSnapshot(snapshot); } #endif // ROCKSDB_LITE } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/db_compaction_test.cc000066400000000000000000002441371267530634000177250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_test_util.h" #include "port/stack_trace.h" #include "rocksdb/experimental.h" #include "rocksdb/utilities/convenience.h" #include "util/sync_point.h" namespace rocksdb { // SYNC_POINT is not supported in released Windows mode. #if !defined(ROCKSDB_LITE) class DBCompactionTest : public DBTestBase { public: DBCompactionTest() : DBTestBase("/db_compaction_test") {} }; class DBCompactionTestWithParam : public DBTestBase, public testing::WithParamInterface> { public: DBCompactionTestWithParam() : DBTestBase("/db_compaction_test") { max_subcompactions_ = std::get<0>(GetParam()); exclusive_manual_compaction_ = std::get<1>(GetParam()); } // Required if inheriting from testing::WithParamInterface<> static void SetUpTestCase() {} static void TearDownTestCase() {} uint32_t max_subcompactions_; bool exclusive_manual_compaction_; }; namespace { class OnFileDeletionListener : public EventListener { public: OnFileDeletionListener() : matched_count_(0), expected_file_name_("") {} void SetExpectedFileName( const std::string file_name) { expected_file_name_ = file_name; } void VerifyMatchedCount(size_t expected_value) { ASSERT_EQ(matched_count_, expected_value); } void OnTableFileDeleted( const TableFileDeletionInfo& info) override { if (expected_file_name_ != "") { ASSERT_EQ(expected_file_name_, info.file_path); expected_file_name_ = ""; matched_count_++; } } private: size_t matched_count_; std::string expected_file_name_; }; class FlushedFileCollector : public EventListener { public: FlushedFileCollector() {} ~FlushedFileCollector() {} virtual void OnFlushCompleted(DB* db, const FlushJobInfo& info) override { std::lock_guard lock(mutex_); flushed_files_.push_back(info.file_path); } std::vector GetFlushedFiles() { std::lock_guard lock(mutex_); std::vector result; for (auto fname : flushed_files_) { result.push_back(fname); } return result; } void ClearFlushedFiles() { flushed_files_.clear(); } private: std::vector flushed_files_; std::mutex mutex_; }; static const int kCDTValueSize = 1000; static const int kCDTKeysPerBuffer = 4; static const int kCDTNumLevels = 8; Options DeletionTriggerOptions() { Options options; options.compression = kNoCompression; options.write_buffer_size = kCDTKeysPerBuffer * (kCDTValueSize + 24); options.min_write_buffer_number_to_merge = 1; options.max_write_buffer_number_to_maintain = 0; options.num_levels = kCDTNumLevels; options.level0_file_num_compaction_trigger = 1; options.target_file_size_base = options.write_buffer_size * 2; options.target_file_size_multiplier = 2; options.max_bytes_for_level_base = options.target_file_size_base * options.target_file_size_multiplier; options.max_bytes_for_level_multiplier = 2; options.disable_auto_compactions = false; return options; } bool HaveOverlappingKeyRanges( const Comparator* c, const SstFileMetaData& a, const SstFileMetaData& b) { if (c->Compare(a.smallestkey, b.smallestkey) >= 0) { if (c->Compare(a.smallestkey, b.largestkey) <= 0) { // b.smallestkey <= a.smallestkey <= b.largestkey return true; } } else if (c->Compare(a.largestkey, b.smallestkey) >= 0) { // a.smallestkey < b.smallestkey <= a.largestkey return true; } if (c->Compare(a.largestkey, b.largestkey) <= 0) { if (c->Compare(a.largestkey, b.smallestkey) >= 0) { // b.smallestkey <= a.largestkey <= b.largestkey return true; } } else if (c->Compare(a.smallestkey, b.largestkey) <= 0) { // a.smallestkey <= b.largestkey < a.largestkey return true; } return false; } // Identifies all files between level "min_level" and "max_level" // which has overlapping key range with "input_file_meta". void GetOverlappingFileNumbersForLevelCompaction( const ColumnFamilyMetaData& cf_meta, const Comparator* comparator, int min_level, int max_level, const SstFileMetaData* input_file_meta, std::set* overlapping_file_names) { std::set overlapping_files; overlapping_files.insert(input_file_meta); for (int m = min_level; m <= max_level; ++m) { for (auto& file : cf_meta.levels[m].files) { for (auto* included_file : overlapping_files) { if (HaveOverlappingKeyRanges( comparator, *included_file, file)) { overlapping_files.insert(&file); overlapping_file_names->insert(file.name); break; } } } } } void VerifyCompactionResult( const ColumnFamilyMetaData& cf_meta, const std::set& overlapping_file_numbers) { #ifndef NDEBUG for (auto& level : cf_meta.levels) { for (auto& file : level.files) { assert(overlapping_file_numbers.find(file.name) == overlapping_file_numbers.end()); } } #endif } const SstFileMetaData* PickFileRandomly( const ColumnFamilyMetaData& cf_meta, Random* rand, int* level = nullptr) { auto file_id = rand->Uniform(static_cast( cf_meta.file_count)) + 1; for (auto& level_meta : cf_meta.levels) { if (file_id <= level_meta.files.size()) { if (level != nullptr) { *level = level_meta.level; } auto result = rand->Uniform(file_id); return &(level_meta.files[result]); } file_id -= static_cast(level_meta.files.size()); } assert(false); return nullptr; } } // anonymous namespace // All the TEST_P tests run once with sub_compactions disabled (i.e. // options.max_subcompactions = 1) and once with it enabled TEST_P(DBCompactionTestWithParam, CompactionDeletionTrigger) { for (int tid = 0; tid < 3; ++tid) { uint64_t db_size[2]; Options options = CurrentOptions(DeletionTriggerOptions()); options.max_subcompactions = max_subcompactions_; if (tid == 1) { // the following only disable stats update in DB::Open() // and should not affect the result of this test. options.skip_stats_update_on_db_open = true; } else if (tid == 2) { // third pass with universal compaction options.compaction_style = kCompactionStyleUniversal; options.num_levels = 1; } DestroyAndReopen(options); Random rnd(301); const int kTestSize = kCDTKeysPerBuffer * 1024; std::vector values; for (int k = 0; k < kTestSize; ++k) { values.push_back(RandomString(&rnd, kCDTValueSize)); ASSERT_OK(Put(Key(k), values[k])); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); db_size[0] = Size(Key(0), Key(kTestSize - 1)); for (int k = 0; k < kTestSize; ++k) { ASSERT_OK(Delete(Key(k))); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); db_size[1] = Size(Key(0), Key(kTestSize - 1)); // must have much smaller db size. ASSERT_GT(db_size[0] / 3, db_size[1]); } } TEST_F(DBCompactionTest, SkipStatsUpdateTest) { // This test verify UpdateAccumulatedStats is not on by observing // the compaction behavior when there are many of deletion entries. // The test will need to be updated if the internal behavior changes. Options options = DeletionTriggerOptions(); options = CurrentOptions(options); options.env = env_; DestroyAndReopen(options); Random rnd(301); const int kTestSize = kCDTKeysPerBuffer * 512; std::vector values; for (int k = 0; k < kTestSize; ++k) { values.push_back(RandomString(&rnd, kCDTValueSize)); ASSERT_OK(Put(Key(k), values[k])); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); for (int k = 0; k < kTestSize; ++k) { ASSERT_OK(Delete(Key(k))); } // Reopen the DB with stats-update disabled options.skip_stats_update_on_db_open = true; env_->random_file_open_counter_.store(0); Reopen(options); // As stats-update is disabled, we expect a very low number of // random file open. // Note that this number must be changed accordingly if we change // the number of files needed to be opened in the DB::Open process. const int kMaxFileOpenCount = 10; ASSERT_LT(env_->random_file_open_counter_.load(), kMaxFileOpenCount); // Repeat the reopen process, but this time we enable // stats-update. options.skip_stats_update_on_db_open = false; env_->random_file_open_counter_.store(0); Reopen(options); // Since we do a normal stats update on db-open, there // will be more random open files. ASSERT_GT(env_->random_file_open_counter_.load(), kMaxFileOpenCount); } TEST_F(DBCompactionTest, TestTableReaderForCompaction) { Options options; options = CurrentOptions(options); options.env = env_; options.new_table_reader_for_compaction_inputs = true; options.max_open_files = 100; options.level0_file_num_compaction_trigger = 3; DestroyAndReopen(options); Random rnd(301); int num_table_cache_lookup = 0; int num_new_table_reader = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "TableCache::FindTable:0", [&](void* arg) { assert(arg != nullptr); bool no_io = *(reinterpret_cast(arg)); if (!no_io) { // filter out cases for table properties queries. num_table_cache_lookup++; } }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "TableCache::GetTableReader:0", [&](void* arg) { num_new_table_reader++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); for (int k = 0; k < options.level0_file_num_compaction_trigger; ++k) { ASSERT_OK(Put(Key(k), Key(k))); ASSERT_OK(Put(Key(10 - k), "bar")); if (k < options.level0_file_num_compaction_trigger - 1) { num_table_cache_lookup = 0; Flush(); dbfull()->TEST_WaitForCompact(); // preloading iterator issues one table cache lookup and create // a new table reader. ASSERT_EQ(num_table_cache_lookup, 1); ASSERT_EQ(num_new_table_reader, 1); num_table_cache_lookup = 0; num_new_table_reader = 0; ASSERT_EQ(Key(k), Get(Key(k))); // lookup iterator from table cache and no need to create a new one. ASSERT_EQ(num_table_cache_lookup, 1); ASSERT_EQ(num_new_table_reader, 0); } } num_table_cache_lookup = 0; num_new_table_reader = 0; Flush(); dbfull()->TEST_WaitForCompact(); // Preloading iterator issues one table cache lookup and creates // a new table reader. One file is created for flush and one for compaction. // Compaction inputs make no table cache look-up. ASSERT_EQ(num_table_cache_lookup, 2); // Create new iterator for: // (1) 1 for verifying flush results // (2) 3 for compaction input files // (3) 1 for verifying compaction results. ASSERT_EQ(num_new_table_reader, 5); num_table_cache_lookup = 0; num_new_table_reader = 0; ASSERT_EQ(Key(1), Get(Key(1))); ASSERT_EQ(num_table_cache_lookup, 1); ASSERT_EQ(num_new_table_reader, 0); num_table_cache_lookup = 0; num_new_table_reader = 0; CompactRangeOptions cro; cro.change_level = true; cro.target_level = 2; cro.bottommost_level_compaction = BottommostLevelCompaction::kForce; db_->CompactRange(cro, nullptr, nullptr); // Only verifying compaction outputs issues one table cache lookup. ASSERT_EQ(num_table_cache_lookup, 1); // One for compaction input, one for verifying compaction results. ASSERT_EQ(num_new_table_reader, 2); num_table_cache_lookup = 0; num_new_table_reader = 0; ASSERT_EQ(Key(1), Get(Key(1))); ASSERT_EQ(num_table_cache_lookup, 1); ASSERT_EQ(num_new_table_reader, 0); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); } TEST_P(DBCompactionTestWithParam, CompactionDeletionTriggerReopen) { for (int tid = 0; tid < 2; ++tid) { uint64_t db_size[3]; Options options = CurrentOptions(DeletionTriggerOptions()); options.max_subcompactions = max_subcompactions_; if (tid == 1) { // second pass with universal compaction options.compaction_style = kCompactionStyleUniversal; options.num_levels = 1; } DestroyAndReopen(options); Random rnd(301); // round 1 --- insert key/value pairs. const int kTestSize = kCDTKeysPerBuffer * 512; std::vector values; for (int k = 0; k < kTestSize; ++k) { values.push_back(RandomString(&rnd, kCDTValueSize)); ASSERT_OK(Put(Key(k), values[k])); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); db_size[0] = Size(Key(0), Key(kTestSize - 1)); Close(); // round 2 --- disable auto-compactions and issue deletions. options.create_if_missing = false; options.disable_auto_compactions = true; Reopen(options); for (int k = 0; k < kTestSize; ++k) { ASSERT_OK(Delete(Key(k))); } db_size[1] = Size(Key(0), Key(kTestSize - 1)); Close(); // as auto_compaction is off, we shouldn't see too much reduce // in db size. ASSERT_LT(db_size[0] / 3, db_size[1]); // round 3 --- reopen db with auto_compaction on and see if // deletion compensation still work. options.disable_auto_compactions = false; Reopen(options); // insert relatively small amount of data to trigger auto compaction. for (int k = 0; k < kTestSize / 10; ++k) { ASSERT_OK(Put(Key(k), values[k])); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); db_size[2] = Size(Key(0), Key(kTestSize - 1)); // this time we're expecting significant drop in size. ASSERT_GT(db_size[0] / 3, db_size[2]); } } TEST_F(DBCompactionTest, DisableStatsUpdateReopen) { uint64_t db_size[3]; for (int test = 0; test < 2; ++test) { Options options = CurrentOptions(DeletionTriggerOptions()); options.skip_stats_update_on_db_open = (test == 0); env_->random_read_counter_.Reset(); DestroyAndReopen(options); Random rnd(301); // round 1 --- insert key/value pairs. const int kTestSize = kCDTKeysPerBuffer * 512; std::vector values; for (int k = 0; k < kTestSize; ++k) { values.push_back(RandomString(&rnd, kCDTValueSize)); ASSERT_OK(Put(Key(k), values[k])); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); db_size[0] = Size(Key(0), Key(kTestSize - 1)); Close(); // round 2 --- disable auto-compactions and issue deletions. options.create_if_missing = false; options.disable_auto_compactions = true; env_->random_read_counter_.Reset(); Reopen(options); for (int k = 0; k < kTestSize; ++k) { ASSERT_OK(Delete(Key(k))); } db_size[1] = Size(Key(0), Key(kTestSize - 1)); Close(); // as auto_compaction is off, we shouldn't see too much reduce // in db size. ASSERT_LT(db_size[0] / 3, db_size[1]); // round 3 --- reopen db with auto_compaction on and see if // deletion compensation still work. options.disable_auto_compactions = false; Reopen(options); dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); db_size[2] = Size(Key(0), Key(kTestSize - 1)); if (options.skip_stats_update_on_db_open) { // If update stats on DB::Open is disable, we don't expect // deletion entries taking effect. ASSERT_LT(db_size[0] / 3, db_size[2]); } else { // Otherwise, we should see a significant drop in db size. ASSERT_GT(db_size[0] / 3, db_size[2]); } } } TEST_P(DBCompactionTestWithParam, CompactionTrigger) { const int kNumKeysPerFile = 100; Options options; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.num_levels = 3; options.level0_file_num_compaction_trigger = 3; options.max_subcompactions = max_subcompactions_; options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile)); options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); for (int num = 0; num < options.level0_file_num_compaction_trigger - 1; num++) { std::vector values; // Write 100KB (100 values, each 1K) for (int i = 0; i < kNumKeysPerFile; i++) { values.push_back(RandomString(&rnd, 990)); ASSERT_OK(Put(1, Key(i), values[i])); } // put extra key to trigger flush ASSERT_OK(Put(1, "", "")); dbfull()->TEST_WaitForFlushMemTable(handles_[1]); ASSERT_EQ(NumTableFilesAtLevel(0, 1), num + 1); } // generate one more file in level-0, and should trigger level-0 compaction std::vector values; for (int i = 0; i < kNumKeysPerFile; i++) { values.push_back(RandomString(&rnd, 990)); ASSERT_OK(Put(1, Key(i), values[i])); } // put extra key to trigger flush ASSERT_OK(Put(1, "", "")); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(1, 1), 1); } TEST_F(DBCompactionTest, BGCompactionsAllowed) { // Create several column families. Make compaction triggers in all of them // and see number of compactions scheduled to be less than allowed. const int kNumKeysPerFile = 100; Options options; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.num_levels = 3; // Should speed up compaction when there are 4 files. options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 20; options.soft_pending_compaction_bytes_limit = 1 << 30; // Infinitely large options.base_background_compactions = 1; options.max_background_compactions = 3; options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile)); options = CurrentOptions(options); // Block all threads in thread pool. const size_t kTotalTasks = 4; env_->SetBackgroundThreads(4, Env::LOW); test::SleepingBackgroundTask sleeping_tasks[kTotalTasks]; for (size_t i = 0; i < kTotalTasks; i++) { env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_tasks[i], Env::Priority::LOW); sleeping_tasks[i].WaitUntilSleeping(); } CreateAndReopenWithCF({"one", "two", "three"}, options); Random rnd(301); for (int cf = 0; cf < 4; cf++) { for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) { for (int i = 0; i < kNumKeysPerFile; i++) { ASSERT_OK(Put(cf, Key(i), "")); } // put extra key to trigger flush ASSERT_OK(Put(cf, "", "")); dbfull()->TEST_WaitForFlushMemTable(handles_[cf]); ASSERT_EQ(NumTableFilesAtLevel(0, cf), num + 1); } } // Now all column families qualify compaction but only one should be // scheduled, because no column family hits speed up condition. ASSERT_EQ(1, env_->GetThreadPoolQueueLen(Env::Priority::LOW)); // Create two more files for one column family, which triggers speed up // condition, three compactions will be scheduled. for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) { for (int i = 0; i < kNumKeysPerFile; i++) { ASSERT_OK(Put(2, Key(i), "")); } // put extra key to trigger flush ASSERT_OK(Put(2, "", "")); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); ASSERT_EQ(options.level0_file_num_compaction_trigger + num + 1, NumTableFilesAtLevel(0, 2)); } ASSERT_EQ(3, env_->GetThreadPoolQueueLen(Env::Priority::LOW)); // Unblock all threads to unblock all compactions. for (size_t i = 0; i < kTotalTasks; i++) { sleeping_tasks[i].WakeUp(); sleeping_tasks[i].WaitUntilDone(); } dbfull()->TEST_WaitForCompact(); // Verify number of compactions allowed will come back to 1. for (size_t i = 0; i < kTotalTasks; i++) { sleeping_tasks[i].Reset(); env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_tasks[i], Env::Priority::LOW); sleeping_tasks[i].WaitUntilSleeping(); } for (int cf = 0; cf < 4; cf++) { for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) { for (int i = 0; i < kNumKeysPerFile; i++) { ASSERT_OK(Put(cf, Key(i), "")); } // put extra key to trigger flush ASSERT_OK(Put(cf, "", "")); dbfull()->TEST_WaitForFlushMemTable(handles_[cf]); ASSERT_EQ(NumTableFilesAtLevel(0, cf), num + 1); } } // Now all column families qualify compaction but only one should be // scheduled, because no column family hits speed up condition. ASSERT_EQ(1, env_->GetThreadPoolQueueLen(Env::Priority::LOW)); for (size_t i = 0; i < kTotalTasks; i++) { sleeping_tasks[i].WakeUp(); sleeping_tasks[i].WaitUntilDone(); } } TEST_P(DBCompactionTestWithParam, CompactionsGenerateMultipleFiles) { Options options; options.write_buffer_size = 100000000; // Large write buffer options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); // Write 8MB (80 values, each 100K) ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); std::vector values; for (int i = 0; i < 80; i++) { values.push_back(RandomString(&rnd, 100000)); ASSERT_OK(Put(1, Key(i), values[i])); } // Reopening moves updates to level-0 ReopenWithColumnFamilies({"default", "pikachu"}, options); dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1], true /* disallow trivial move */); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_GT(NumTableFilesAtLevel(1, 1), 1); for (int i = 0; i < 80; i++) { ASSERT_EQ(Get(1, Key(i)), values[i]); } } TEST_F(DBCompactionTest, MinorCompactionsHappen) { do { Options options; options.write_buffer_size = 10000; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); const int N = 500; int starting_num_tables = TotalTableFiles(1); for (int i = 0; i < N; i++) { ASSERT_OK(Put(1, Key(i), Key(i) + std::string(1000, 'v'))); } int ending_num_tables = TotalTableFiles(1); ASSERT_GT(ending_num_tables, starting_num_tables); for (int i = 0; i < N; i++) { ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(1, Key(i))); } ReopenWithColumnFamilies({"default", "pikachu"}, options); for (int i = 0; i < N; i++) { ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(1, Key(i))); } } while (ChangeCompactOptions()); } TEST_F(DBCompactionTest, ZeroSeqIdCompaction) { Options options; options.compaction_style = kCompactionStyleLevel; options.level0_file_num_compaction_trigger = 3; FlushedFileCollector* collector = new FlushedFileCollector(); options.listeners.emplace_back(collector); // compaction options CompactionOptions compact_opt; compact_opt.compression = kNoCompression; compact_opt.output_file_size_limit = 4096; const size_t key_len = static_cast(compact_opt.output_file_size_limit) / 5; options = CurrentOptions(options); DestroyAndReopen(options); std::vector snaps; // create first file and flush to l0 for (auto& key : {"1", "2", "3", "3", "3", "3"}) { Put(key, std::string(key_len, 'A')); snaps.push_back(dbfull()->GetSnapshot()); } Flush(); dbfull()->TEST_WaitForFlushMemTable(); // create second file and flush to l0 for (auto& key : {"3", "4", "5", "6", "7", "8"}) { Put(key, std::string(key_len, 'A')); snaps.push_back(dbfull()->GetSnapshot()); } Flush(); dbfull()->TEST_WaitForFlushMemTable(); // move both files down to l1 dbfull()->CompactFiles(compact_opt, collector->GetFlushedFiles(), 1); // release snap so that first instance of key(3) can have seqId=0 for (auto snap : snaps) { dbfull()->ReleaseSnapshot(snap); } // create 3 files in l0 so to trigger compaction for (int i = 0; i < options.level0_file_num_compaction_trigger; i++) { Put("2", std::string(1, 'A')); Flush(); dbfull()->TEST_WaitForFlushMemTable(); } dbfull()->TEST_WaitForCompact(); ASSERT_OK(Put("", "")); } // Check that writes done during a memtable compaction are recovered // if the database is shutdown during the memtable compaction. TEST_F(DBCompactionTest, RecoverDuringMemtableCompaction) { do { Options options; options.env = env_; options.write_buffer_size = 1000000; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Trigger a long memtable compaction and reopen the database during it ASSERT_OK(Put(1, "foo", "v1")); // Goes to 1st log file ASSERT_OK(Put(1, "big1", std::string(10000000, 'x'))); // Fills memtable ASSERT_OK(Put(1, "big2", std::string(1000, 'y'))); // Triggers compaction ASSERT_OK(Put(1, "bar", "v2")); // Goes to new log file ReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("v2", Get(1, "bar")); ASSERT_EQ(std::string(10000000, 'x'), Get(1, "big1")); ASSERT_EQ(std::string(1000, 'y'), Get(1, "big2")); } while (ChangeOptions()); } TEST_P(DBCompactionTestWithParam, TrivialMoveOneFile) { int32_t trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options; options.write_buffer_size = 100000000; options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); DestroyAndReopen(options); int32_t num_keys = 80; int32_t value_size = 100 * 1024; // 100 KB Random rnd(301); std::vector values; for (int i = 0; i < num_keys; i++) { values.push_back(RandomString(&rnd, value_size)); ASSERT_OK(Put(Key(i), values[i])); } // Reopening moves updates to L0 Reopen(options); ASSERT_EQ(NumTableFilesAtLevel(0, 0), 1); // 1 file in L0 ASSERT_EQ(NumTableFilesAtLevel(1, 0), 0); // 0 files in L1 std::vector metadata; db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(metadata.size(), 1U); LiveFileMetaData level0_file = metadata[0]; // L0 file meta CompactRangeOptions cro; cro.exclusive_manual_compaction = exclusive_manual_compaction_; // Compaction will initiate a trivial move from L0 to L1 dbfull()->CompactRange(cro, nullptr, nullptr); // File moved From L0 to L1 ASSERT_EQ(NumTableFilesAtLevel(0, 0), 0); // 0 files in L0 ASSERT_EQ(NumTableFilesAtLevel(1, 0), 1); // 1 file in L1 metadata.clear(); db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(metadata.size(), 1U); ASSERT_EQ(metadata[0].name /* level1_file.name */, level0_file.name); ASSERT_EQ(metadata[0].size /* level1_file.size */, level0_file.size); for (int i = 0; i < num_keys; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } ASSERT_EQ(trivial_move, 1); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_P(DBCompactionTestWithParam, TrivialMoveNonOverlappingFiles) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.disable_auto_compactions = true; options.write_buffer_size = 10 * 1024 * 1024; options.max_subcompactions = max_subcompactions_; DestroyAndReopen(options); // non overlapping ranges std::vector> ranges = { {100, 199}, {300, 399}, {0, 99}, {200, 299}, {600, 699}, {400, 499}, {500, 550}, {551, 599}, }; int32_t value_size = 10 * 1024; // 10 KB Random rnd(301); std::map values; for (size_t i = 0; i < ranges.size(); i++) { for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) { values[j] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(j), values[j])); } ASSERT_OK(Flush()); } int32_t level0_files = NumTableFilesAtLevel(0, 0); ASSERT_EQ(level0_files, ranges.size()); // Multiple files in L0 ASSERT_EQ(NumTableFilesAtLevel(1, 0), 0); // No files in L1 CompactRangeOptions cro; cro.exclusive_manual_compaction = exclusive_manual_compaction_; // Since data is non-overlapping we expect compaction to initiate // a trivial move db_->CompactRange(cro, nullptr, nullptr); // We expect that all the files were trivially moved from L0 to L1 ASSERT_EQ(NumTableFilesAtLevel(0, 0), 0); ASSERT_EQ(NumTableFilesAtLevel(1, 0) /* level1_files */, level0_files); for (size_t i = 0; i < ranges.size(); i++) { for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) { ASSERT_EQ(Get(Key(j)), values[j]); } } ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); trivial_move = 0; non_trivial_move = 0; values.clear(); DestroyAndReopen(options); // Same ranges as above but overlapping ranges = { {100, 199}, {300, 399}, {0, 99}, {200, 299}, {600, 699}, {400, 499}, {500, 560}, // this range overlap with the next one {551, 599}, }; for (size_t i = 0; i < ranges.size(); i++) { for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) { values[j] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(j), values[j])); } ASSERT_OK(Flush()); } db_->CompactRange(cro, nullptr, nullptr); for (size_t i = 0; i < ranges.size(); i++) { for (int32_t j = ranges[i].first; j <= ranges[i].second; j++) { ASSERT_EQ(Get(Key(j)), values[j]); } } ASSERT_EQ(trivial_move, 0); ASSERT_EQ(non_trivial_move, 1); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_P(DBCompactionTestWithParam, TrivialMoveTargetLevel) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.disable_auto_compactions = true; options.write_buffer_size = 10 * 1024 * 1024; options.num_levels = 7; options.max_subcompactions = max_subcompactions_; DestroyAndReopen(options); int32_t value_size = 10 * 1024; // 10 KB // Add 2 non-overlapping files Random rnd(301); std::map values; // file 1 [0 => 300] for (int32_t i = 0; i <= 300; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // file 2 [600 => 700] for (int32_t i = 600; i <= 700; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 2 files in L0 ASSERT_EQ("2", FilesPerLevel(0)); CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 6; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); // 2 files in L6 ASSERT_EQ("0,0,0,0,0,0,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); for (int32_t i = 0; i <= 300; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } for (int32_t i = 600; i <= 700; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } } TEST_P(DBCompactionTestWithParam, ManualCompactionPartial) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); bool first = true; bool second = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"DBCompaction::ManualPartial:4", "DBCompaction::ManualPartial:1"}, {"DBCompaction::ManualPartial:2", "DBCompaction::ManualPartial:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (first) { TEST_SYNC_POINT("DBCompaction::ManualPartial:4"); first = false; TEST_SYNC_POINT("DBCompaction::ManualPartial:3"); } else if (second) { TEST_SYNC_POINT("DBCompaction::ManualPartial:2"); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.write_buffer_size = 10 * 1024 * 1024; options.num_levels = 7; options.max_subcompactions = max_subcompactions_; options.level0_file_num_compaction_trigger = 3; options.max_background_compactions = 3; options.target_file_size_base = 1 << 23; // 8 MB DestroyAndReopen(options); int32_t value_size = 10 * 1024; // 10 KB // Add 2 non-overlapping files Random rnd(301); std::map values; // file 1 [0 => 100] for (int32_t i = 0; i < 100; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // file 2 [100 => 300] for (int32_t i = 100; i < 300; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 2 files in L0 ASSERT_EQ("2", FilesPerLevel(0)); CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 6; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; // Trivial move the two non-overlapping files to level 6 ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); // 2 files in L6 ASSERT_EQ("0,0,0,0,0,0,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); // file 3 [ 0 => 200] for (int32_t i = 0; i < 200; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 1 files in L0 ASSERT_EQ("1,0,0,0,0,0,2", FilesPerLevel(0)); ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, false)); ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, nullptr, false)); ASSERT_OK(dbfull()->TEST_CompactRange(2, nullptr, nullptr, nullptr, false)); ASSERT_OK(dbfull()->TEST_CompactRange(3, nullptr, nullptr, nullptr, false)); ASSERT_OK(dbfull()->TEST_CompactRange(4, nullptr, nullptr, nullptr, false)); // 2 files in L6, 1 file in L5 ASSERT_EQ("0,0,0,0,0,1,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 6); ASSERT_EQ(non_trivial_move, 0); std::thread threads([&] { compact_options.change_level = false; compact_options.exclusive_manual_compaction = false; std::string begin_string = Key(0); std::string end_string = Key(199); Slice begin(begin_string); Slice end(end_string); ASSERT_OK(db_->CompactRange(compact_options, &begin, &end)); }); TEST_SYNC_POINT("DBCompaction::ManualPartial:1"); // file 4 [300 => 400) for (int32_t i = 300; i <= 400; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // file 5 [400 => 500) for (int32_t i = 400; i <= 500; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // file 6 [500 => 600) for (int32_t i = 500; i <= 600; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 3 files in L0 ASSERT_EQ("3,0,0,0,0,1,2", FilesPerLevel(0)); // 1 file in L6, 1 file in L1 dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,1,0,0,0,0,1", FilesPerLevel(0)); threads.join(); for (int32_t i = 0; i < 600; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } } TEST_F(DBCompactionTest, ManualPartialFill) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); bool first = true; bool second = true; rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"DBCompaction::PartialFill:4", "DBCompaction::PartialFill:1"}, {"DBCompaction::PartialFill:2", "DBCompaction::PartialFill:3"}}); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) { if (first) { TEST_SYNC_POINT("DBCompaction::PartialFill:4"); first = false; TEST_SYNC_POINT("DBCompaction::PartialFill:3"); } else if (second) { } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.write_buffer_size = 10 * 1024 * 1024; options.max_bytes_for_level_multiplier = 2; options.num_levels = 4; options.level0_file_num_compaction_trigger = 3; options.max_background_compactions = 3; DestroyAndReopen(options); int32_t value_size = 10 * 1024; // 10 KB // Add 2 non-overlapping files Random rnd(301); std::map values; // file 1 [0 => 100] for (int32_t i = 0; i < 100; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // file 2 [100 => 300] for (int32_t i = 100; i < 300; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 2 files in L0 ASSERT_EQ("2", FilesPerLevel(0)); CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 2; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); // 2 files in L2 ASSERT_EQ("0,0,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); // file 3 [ 0 => 200] for (int32_t i = 0; i < 200; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 2 files in L2, 1 in L0 ASSERT_EQ("1,0,2", FilesPerLevel(0)); ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, false)); // 2 files in L2, 1 in L1 ASSERT_EQ("0,1,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 2); ASSERT_EQ(non_trivial_move, 0); std::thread threads([&] { compact_options.change_level = false; compact_options.exclusive_manual_compaction = false; std::string begin_string = Key(0); std::string end_string = Key(199); Slice begin(begin_string); Slice end(end_string); ASSERT_OK(db_->CompactRange(compact_options, &begin, &end)); }); TEST_SYNC_POINT("DBCompaction::PartialFill:1"); // Many files 4 [300 => 4300) for (int32_t i = 0; i <= 5; i++) { for (int32_t j = 300; j < 4300; j++) { if (j == 2300) { ASSERT_OK(Flush()); dbfull()->TEST_WaitForFlushMemTable(); } values[j] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(j), values[j])); } } // Verify level sizes uint64_t target_size = 4 * options.max_bytes_for_level_base; for (int32_t i = 1; i < options.num_levels; i++) { ASSERT_LE(SizeAtLevel(i), target_size); target_size *= options.max_bytes_for_level_multiplier; } TEST_SYNC_POINT("DBCompaction::PartialFill:2"); dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); threads.join(); for (int32_t i = 0; i < 4300; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } } TEST_F(DBCompactionTest, DeleteFileRange) { Options options = CurrentOptions(); options.write_buffer_size = 10 * 1024 * 1024; options.max_bytes_for_level_multiplier = 2; options.num_levels = 4; options.level0_file_num_compaction_trigger = 3; options.max_background_compactions = 3; DestroyAndReopen(options); int32_t value_size = 10 * 1024; // 10 KB // Add 2 non-overlapping files Random rnd(301); std::map values; // file 1 [0 => 100] for (int32_t i = 0; i < 100; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // file 2 [100 => 300] for (int32_t i = 100; i < 300; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // 2 files in L0 ASSERT_EQ("2", FilesPerLevel(0)); CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 2; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); // 2 files in L2 ASSERT_EQ("0,0,2", FilesPerLevel(0)); // file 3 [ 0 => 200] for (int32_t i = 0; i < 200; i++) { values[i] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); // Many files 4 [300 => 4300) for (int32_t i = 0; i <= 5; i++) { for (int32_t j = 300; j < 4300; j++) { if (j == 2300) { ASSERT_OK(Flush()); dbfull()->TEST_WaitForFlushMemTable(); } values[j] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(j), values[j])); } } ASSERT_OK(Flush()); dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); // Verify level sizes uint64_t target_size = 4 * options.max_bytes_for_level_base; for (int32_t i = 1; i < options.num_levels; i++) { ASSERT_LE(SizeAtLevel(i), target_size); target_size *= options.max_bytes_for_level_multiplier; } size_t old_num_files = CountFiles(); std::string begin_string = Key(1000); std::string end_string = Key(2000); Slice begin(begin_string); Slice end(end_string); ASSERT_OK(DeleteFilesInRange(db_, db_->DefaultColumnFamily(), &begin, &end)); int32_t deleted_count = 0; for (int32_t i = 0; i < 4300; i++) { if (i < 1000 || i > 2000) { ASSERT_EQ(Get(Key(i)), values[i]); } else { ReadOptions roptions; std::string result; Status s = db_->Get(roptions, Key(i), &result); ASSERT_TRUE(s.IsNotFound() || s.ok()); if (s.IsNotFound()) { deleted_count++; } } } ASSERT_GT(deleted_count, 0); begin_string = Key(5000); end_string = Key(6000); Slice begin1(begin_string); Slice end1(end_string); // Try deleting files in range which contain no keys ASSERT_OK( DeleteFilesInRange(db_, db_->DefaultColumnFamily(), &begin1, &end1)); // Push data from level 0 to level 1 to force all data to be deleted // Note that we don't delete level 0 files compact_options.change_level = true; compact_options.target_level = 1; ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr)); ASSERT_OK( DeleteFilesInRange(db_, db_->DefaultColumnFamily(), nullptr, nullptr)); int32_t deleted_count2 = 0; for (int32_t i = 0; i < 4300; i++) { ReadOptions roptions; std::string result; Status s = db_->Get(roptions, Key(i), &result); ASSERT_TRUE(s.IsNotFound()); deleted_count2++; } ASSERT_GT(deleted_count2, deleted_count); size_t new_num_files = CountFiles(); ASSERT_GT(old_num_files, new_num_files); } TEST_P(DBCompactionTestWithParam, TrivialMoveToLastLevelWithFiles) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options; options.write_buffer_size = 100000000; options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); DestroyAndReopen(options); int32_t value_size = 10 * 1024; // 10 KB Random rnd(301); std::vector values; // File with keys [ 0 => 99 ] for (int i = 0; i < 100; i++) { values.push_back(RandomString(&rnd, value_size)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); ASSERT_EQ("1", FilesPerLevel(0)); // Compaction will do L0=>L1 (trivial move) then move L1 files to L3 CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 3; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); ASSERT_EQ("0,0,0,1", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); // File with keys [ 100 => 199 ] for (int i = 100; i < 200; i++) { values.push_back(RandomString(&rnd, value_size)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); ASSERT_EQ("1,0,0,1", FilesPerLevel(0)); CompactRangeOptions cro; cro.exclusive_manual_compaction = exclusive_manual_compaction_; // Compaction will do L0=>L1 L1=>L2 L2=>L3 (3 trivial moves) ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr)); ASSERT_EQ("0,0,0,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 4); ASSERT_EQ(non_trivial_move, 0); for (int i = 0; i < 200; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_P(DBCompactionTestWithParam, LevelCompactionThirdPath) { Options options = CurrentOptions(); options.db_paths.emplace_back(dbname_, 500 * 1024); options.db_paths.emplace_back(dbname_ + "_2", 4 * 1024 * 1024); options.db_paths.emplace_back(dbname_ + "_3", 1024 * 1024 * 1024); options.memtable_factory.reset( new SpecialSkipListFactory(KNumKeysByGenerateNewFile - 1)); options.compaction_style = kCompactionStyleLevel; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 2; options.num_levels = 4; options.max_bytes_for_level_base = 400 * 1024; options.max_subcompactions = max_subcompactions_; // options = CurrentOptions(options); std::vector filenames; env_->GetChildren(options.db_paths[1].path, &filenames); // Delete archival files. for (size_t i = 0; i < filenames.size(); ++i) { env_->DeleteFile(options.db_paths[1].path + "/" + filenames[i]); } env_->DeleteDir(options.db_paths[1].path); Reopen(options); Random rnd(301); int key_idx = 0; // First three 110KB files are not going to second path. // After that, (100K, 200K) for (int num = 0; num < 3; num++) { GenerateNewFile(&rnd, &key_idx); } // Another 110KB triggers a compaction to 400K file to fill up first path GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(3, GetSstFileCount(options.db_paths[1].path)); // (1, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4", FilesPerLevel(0)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 1) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,1", FilesPerLevel(0)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 2) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,2", FilesPerLevel(0)); ASSERT_EQ(2, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 3) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,3", FilesPerLevel(0)); ASSERT_EQ(3, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,4", FilesPerLevel(0)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 5) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,5", FilesPerLevel(0)); ASSERT_EQ(5, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 6) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,6", FilesPerLevel(0)); ASSERT_EQ(6, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 7) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,7", FilesPerLevel(0)); ASSERT_EQ(7, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 4, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,8", FilesPerLevel(0)); ASSERT_EQ(8, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(4, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Reopen(options); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Destroy(options); } TEST_P(DBCompactionTestWithParam, LevelCompactionPathUse) { Options options = CurrentOptions(); options.db_paths.emplace_back(dbname_, 500 * 1024); options.db_paths.emplace_back(dbname_ + "_2", 4 * 1024 * 1024); options.db_paths.emplace_back(dbname_ + "_3", 1024 * 1024 * 1024); options.memtable_factory.reset( new SpecialSkipListFactory(KNumKeysByGenerateNewFile - 1)); options.compaction_style = kCompactionStyleLevel; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 2; options.num_levels = 4; options.max_bytes_for_level_base = 400 * 1024; options.max_subcompactions = max_subcompactions_; // options = CurrentOptions(options); std::vector filenames; env_->GetChildren(options.db_paths[1].path, &filenames); // Delete archival files. for (size_t i = 0; i < filenames.size(); ++i) { env_->DeleteFile(options.db_paths[1].path + "/" + filenames[i]); } env_->DeleteDir(options.db_paths[1].path); Reopen(options); Random rnd(301); int key_idx = 0; // Always gets compacted into 1 Level1 file, // 0/1 Level 0 file for (int num = 0; num < 3; num++) { key_idx = 0; GenerateNewFile(&rnd, &key_idx); } key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,1", FilesPerLevel(0)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("0,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("0,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("0,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("0,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); key_idx = 0; GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,1", FilesPerLevel(0)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Reopen(options); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Destroy(options); } TEST_P(DBCompactionTestWithParam, ConvertCompactionStyle) { Random rnd(301); int max_key_level_insert = 200; int max_key_universal_insert = 600; // Stage 1: generate a db with level compaction Options options; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.num_levels = 4; options.level0_file_num_compaction_trigger = 3; options.max_bytes_for_level_base = 500 << 10; // 500KB options.max_bytes_for_level_multiplier = 1; options.target_file_size_base = 200 << 10; // 200KB options.target_file_size_multiplier = 1; options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); for (int i = 0; i <= max_key_level_insert; i++) { // each value is 10K ASSERT_OK(Put(1, Key(i), RandomString(&rnd, 10000))); } ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); ASSERT_GT(TotalTableFiles(1, 4), 1); int non_level0_num_files = 0; for (int i = 1; i < options.num_levels; i++) { non_level0_num_files += NumTableFilesAtLevel(i, 1); } ASSERT_GT(non_level0_num_files, 0); // Stage 2: reopen with universal compaction - should fail options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.num_levels = 1; options = CurrentOptions(options); Status s = TryReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_TRUE(s.IsInvalidArgument()); // Stage 3: compact into a single file and move the file to level 0 options = CurrentOptions(); options.disable_auto_compactions = true; options.target_file_size_base = INT_MAX; options.target_file_size_multiplier = 1; options.max_bytes_for_level_base = INT_MAX; options.max_bytes_for_level_multiplier = 1; options.num_levels = 4; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 0; compact_options.bottommost_level_compaction = BottommostLevelCompaction::kForce; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; dbfull()->CompactRange(compact_options, handles_[1], nullptr, nullptr); // Only 1 file in L0 ASSERT_EQ("1", FilesPerLevel(1)); // Stage 4: re-open in universal compaction style and do some db operations options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.num_levels = 4; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 3; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); options.num_levels = 1; ReopenWithColumnFamilies({"default", "pikachu"}, options); for (int i = max_key_level_insert / 2; i <= max_key_universal_insert; i++) { ASSERT_OK(Put(1, Key(i), RandomString(&rnd, 10000))); } dbfull()->Flush(FlushOptions()); ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); for (int i = 1; i < options.num_levels; i++) { ASSERT_EQ(NumTableFilesAtLevel(i, 1), 0); } // verify keys inserted in both level compaction style and universal // compaction style std::string keys_in_db; Iterator* iter = dbfull()->NewIterator(ReadOptions(), handles_[1]); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { keys_in_db.append(iter->key().ToString()); keys_in_db.push_back(','); } delete iter; std::string expected_keys; for (int i = 0; i <= max_key_universal_insert; i++) { expected_keys.append(Key(i)); expected_keys.push_back(','); } ASSERT_EQ(keys_in_db, expected_keys); } TEST_F(DBCompactionTest, L0_CompactionBug_Issue44_a) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "b", "v")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_OK(Delete(1, "b")); ASSERT_OK(Delete(1, "a")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_OK(Delete(1, "a")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "a", "v")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("(a->v)", Contents(1)); env_->SleepForMicroseconds(1000000); // Wait for compaction to finish ASSERT_EQ("(a->v)", Contents(1)); } while (ChangeCompactOptions()); } TEST_F(DBCompactionTest, L0_CompactionBug_Issue44_b) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); Put(1, "", ""); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Delete(1, "e"); Put(1, "", ""); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Put(1, "c", "cv"); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Put(1, "", ""); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Put(1, "", ""); env_->SleepForMicroseconds(1000000); // Wait for compaction to finish ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Put(1, "d", "dv"); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Put(1, "", ""); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); Delete(1, "d"); Delete(1, "b"); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("(->)(c->cv)", Contents(1)); env_->SleepForMicroseconds(1000000); // Wait for compaction to finish ASSERT_EQ("(->)(c->cv)", Contents(1)); } while (ChangeCompactOptions()); } TEST_P(DBCompactionTestWithParam, ManualCompaction) { Options options = CurrentOptions(); options.max_subcompactions = max_subcompactions_; options.statistics = rocksdb::CreateDBStatistics(); CreateAndReopenWithCF({"pikachu"}, options); // iter - 0 with 7 levels // iter - 1 with 3 levels for (int iter = 0; iter < 2; ++iter) { MakeTables(3, "p", "q", 1); ASSERT_EQ("1,1,1", FilesPerLevel(1)); // Compaction range falls before files Compact(1, "", "c"); ASSERT_EQ("1,1,1", FilesPerLevel(1)); // Compaction range falls after files Compact(1, "r", "z"); ASSERT_EQ("1,1,1", FilesPerLevel(1)); // Compaction range overlaps files Compact(1, "p1", "p9"); ASSERT_EQ("0,0,1", FilesPerLevel(1)); // Populate a different range MakeTables(3, "c", "e", 1); ASSERT_EQ("1,1,2", FilesPerLevel(1)); // Compact just the new range Compact(1, "b", "f"); ASSERT_EQ("0,0,2", FilesPerLevel(1)); // Compact all MakeTables(1, "a", "z", 1); ASSERT_EQ("1,0,2", FilesPerLevel(1)); uint64_t prev_block_cache_add = options.statistics->getTickerCount(BLOCK_CACHE_ADD); CompactRangeOptions cro; cro.exclusive_manual_compaction = exclusive_manual_compaction_; db_->CompactRange(cro, handles_[1], nullptr, nullptr); // Verify manual compaction doesn't fill block cache ASSERT_EQ(prev_block_cache_add, options.statistics->getTickerCount(BLOCK_CACHE_ADD)); ASSERT_EQ("0,0,1", FilesPerLevel(1)); if (iter == 0) { options = CurrentOptions(); options.max_background_flushes = 0; options.num_levels = 3; options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); } } } TEST_P(DBCompactionTestWithParam, ManualLevelCompactionOutputPathId) { Options options = CurrentOptions(); options.db_paths.emplace_back(dbname_ + "_2", 2 * 10485760); options.db_paths.emplace_back(dbname_ + "_3", 100 * 10485760); options.db_paths.emplace_back(dbname_ + "_4", 120 * 10485760); options.max_subcompactions = max_subcompactions_; CreateAndReopenWithCF({"pikachu"}, options); // iter - 0 with 7 levels // iter - 1 with 3 levels for (int iter = 0; iter < 2; ++iter) { for (int i = 0; i < 3; ++i) { ASSERT_OK(Put(1, "p", "begin")); ASSERT_OK(Put(1, "q", "end")); ASSERT_OK(Flush(1)); } ASSERT_EQ("3", FilesPerLevel(1)); ASSERT_EQ(3, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); // Compaction range falls before files Compact(1, "", "c"); ASSERT_EQ("3", FilesPerLevel(1)); // Compaction range falls after files Compact(1, "r", "z"); ASSERT_EQ("3", FilesPerLevel(1)); // Compaction range overlaps files Compact(1, "p1", "p9", 1); ASSERT_EQ("0,1", FilesPerLevel(1)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); // Populate a different range for (int i = 0; i < 3; ++i) { ASSERT_OK(Put(1, "c", "begin")); ASSERT_OK(Put(1, "e", "end")); ASSERT_OK(Flush(1)); } ASSERT_EQ("3,1", FilesPerLevel(1)); // Compact just the new range Compact(1, "b", "f", 1); ASSERT_EQ("0,2", FilesPerLevel(1)); ASSERT_EQ(2, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); // Compact all ASSERT_OK(Put(1, "a", "begin")); ASSERT_OK(Put(1, "z", "end")); ASSERT_OK(Flush(1)); ASSERT_EQ("1,2", FilesPerLevel(1)); ASSERT_EQ(2, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[0].path)); CompactRangeOptions compact_options; compact_options.target_path_id = 1; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; db_->CompactRange(compact_options, handles_[1], nullptr, nullptr); ASSERT_EQ("0,1", FilesPerLevel(1)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); if (iter == 0) { DestroyAndReopen(options); options = CurrentOptions(); options.db_paths.emplace_back(dbname_ + "_2", 2 * 10485760); options.db_paths.emplace_back(dbname_ + "_3", 100 * 10485760); options.db_paths.emplace_back(dbname_ + "_4", 120 * 10485760); options.max_background_flushes = 1; options.num_levels = 3; options.create_if_missing = true; CreateAndReopenWithCF({"pikachu"}, options); } } } TEST_F(DBCompactionTest, FilesDeletedAfterCompaction) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v2")); Compact(1, "a", "z"); const size_t num_files = CountLiveFiles(); for (int i = 0; i < 10; i++) { ASSERT_OK(Put(1, "foo", "v2")); Compact(1, "a", "z"); } ASSERT_EQ(CountLiveFiles(), num_files); } while (ChangeCompactOptions()); } // Check level comapction with compact files TEST_P(DBCompactionTestWithParam, DISABLED_CompactFilesOnLevelCompaction) { const int kTestKeySize = 16; const int kTestValueSize = 984; const int kEntrySize = kTestKeySize + kTestValueSize; const int kEntriesPerBuffer = 100; Options options; options.create_if_missing = true; options.write_buffer_size = kEntrySize * kEntriesPerBuffer; options.compaction_style = kCompactionStyleLevel; options.target_file_size_base = options.write_buffer_size; options.max_bytes_for_level_base = options.target_file_size_base * 2; options.level0_stop_writes_trigger = 2; options.max_bytes_for_level_multiplier = 2; options.compression = kNoCompression; options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); for (int key = 64 * kEntriesPerBuffer; key >= 0; --key) { ASSERT_OK(Put(1, ToString(key), RandomString(&rnd, kTestValueSize))); } dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForCompact(); ColumnFamilyMetaData cf_meta; dbfull()->GetColumnFamilyMetaData(handles_[1], &cf_meta); int output_level = static_cast(cf_meta.levels.size()) - 1; for (int file_picked = 5; file_picked > 0; --file_picked) { std::set overlapping_file_names; std::vector compaction_input_file_names; for (int f = 0; f < file_picked; ++f) { int level = 0; auto file_meta = PickFileRandomly(cf_meta, &rnd, &level); compaction_input_file_names.push_back(file_meta->name); GetOverlappingFileNumbersForLevelCompaction( cf_meta, options.comparator, level, output_level, file_meta, &overlapping_file_names); } ASSERT_OK(dbfull()->CompactFiles( CompactionOptions(), handles_[1], compaction_input_file_names, output_level)); // Make sure all overlapping files do not exist after compaction dbfull()->GetColumnFamilyMetaData(handles_[1], &cf_meta); VerifyCompactionResult(cf_meta, overlapping_file_names); } // make sure all key-values are still there. for (int key = 64 * kEntriesPerBuffer; key >= 0; --key) { ASSERT_NE(Get(1, ToString(key)), "NOT_FOUND"); } } TEST_P(DBCompactionTestWithParam, PartialCompactionFailure) { Options options; const int kKeySize = 16; const int kKvSize = 1000; const int kKeysPerBuffer = 100; const int kNumL1Files = 5; options.create_if_missing = true; options.write_buffer_size = kKeysPerBuffer * kKvSize; options.max_write_buffer_number = 2; options.target_file_size_base = options.write_buffer_size * (options.max_write_buffer_number - 1); options.level0_file_num_compaction_trigger = kNumL1Files; options.max_bytes_for_level_base = options.level0_file_num_compaction_trigger * options.target_file_size_base; options.max_bytes_for_level_multiplier = 2; options.compression = kNoCompression; options.max_subcompactions = max_subcompactions_; env_->SetBackgroundThreads(1, Env::HIGH); env_->SetBackgroundThreads(1, Env::LOW); // stop the compaction thread until we simulate the file creation failure. test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); options.env = env_; DestroyAndReopen(options); const int kNumInsertedKeys = options.level0_file_num_compaction_trigger * (options.max_write_buffer_number - 1) * kKeysPerBuffer; Random rnd(301); std::vector keys; std::vector values; for (int k = 0; k < kNumInsertedKeys; ++k) { keys.emplace_back(RandomString(&rnd, kKeySize)); values.emplace_back(RandomString(&rnd, kKvSize - kKeySize)); ASSERT_OK(Put(Slice(keys[k]), Slice(values[k]))); dbfull()->TEST_WaitForFlushMemTable(); } dbfull()->TEST_FlushMemTable(true); // Make sure the number of L0 files can trigger compaction. ASSERT_GE(NumTableFilesAtLevel(0), options.level0_file_num_compaction_trigger); auto previous_num_level0_files = NumTableFilesAtLevel(0); // Fail the first file creation. env_->non_writable_count_ = 1; sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); // Expect compaction to fail here as one file will fail its // creation. ASSERT_TRUE(!dbfull()->TEST_WaitForCompact().ok()); // Verify L0 -> L1 compaction does fail. ASSERT_EQ(NumTableFilesAtLevel(1), 0); // Verify all L0 files are still there. ASSERT_EQ(NumTableFilesAtLevel(0), previous_num_level0_files); // All key-values must exist after compaction fails. for (int k = 0; k < kNumInsertedKeys; ++k) { ASSERT_EQ(values[k], Get(keys[k])); } env_->non_writable_count_ = 0; // Make sure RocksDB will not get into corrupted state. Reopen(options); // Verify again after reopen. for (int k = 0; k < kNumInsertedKeys; ++k) { ASSERT_EQ(values[k], Get(keys[k])); } } TEST_P(DBCompactionTestWithParam, DeleteMovedFileAfterCompaction) { // iter 1 -- delete_obsolete_files_period_micros == 0 for (int iter = 0; iter < 2; ++iter) { // This test triggers move compaction and verifies that the file is not // deleted when it's part of move compaction Options options = CurrentOptions(); options.env = env_; if (iter == 1) { options.delete_obsolete_files_period_micros = 0; } options.create_if_missing = true; options.level0_file_num_compaction_trigger = 2; // trigger compaction when we have 2 files OnFileDeletionListener* listener = new OnFileDeletionListener(); options.listeners.emplace_back(listener); options.max_subcompactions = max_subcompactions_; DestroyAndReopen(options); Random rnd(301); // Create two 1MB sst files for (int i = 0; i < 2; ++i) { // Create 1MB sst file for (int j = 0; j < 100; ++j) { ASSERT_OK(Put(Key(i * 50 + j), RandomString(&rnd, 10 * 1024))); } ASSERT_OK(Flush()); } // this should execute L0->L1 dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,1", FilesPerLevel(0)); // block compactions test::SleepingBackgroundTask sleeping_task; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task, Env::Priority::LOW); options.max_bytes_for_level_base = 1024 * 1024; // 1 MB Reopen(options); std::unique_ptr iterator(db_->NewIterator(ReadOptions())); ASSERT_EQ("0,1", FilesPerLevel(0)); // let compactions go sleeping_task.WakeUp(); sleeping_task.WaitUntilDone(); // this should execute L1->L2 (move) dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,0,1", FilesPerLevel(0)); std::vector metadata; db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(metadata.size(), 1U); auto moved_file_name = metadata[0].name; // Create two more 1MB sst files for (int i = 0; i < 2; ++i) { // Create 1MB sst file for (int j = 0; j < 100; ++j) { ASSERT_OK(Put(Key(i * 50 + j + 100), RandomString(&rnd, 10 * 1024))); } ASSERT_OK(Flush()); } // this should execute both L0->L1 and L1->L2 (merge with previous file) dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,0,2", FilesPerLevel(0)); // iterator is holding the file ASSERT_OK(env_->FileExists(dbname_ + moved_file_name)); listener->SetExpectedFileName(dbname_ + moved_file_name); iterator.reset(); // this file should have been compacted away ASSERT_NOK(env_->FileExists(dbname_ + moved_file_name)); listener->VerifyMatchedCount(1); } } TEST_P(DBCompactionTestWithParam, CompressLevelCompaction) { if (!Zlib_Supported()) { return; } Options options = CurrentOptions(); options.memtable_factory.reset( new SpecialSkipListFactory(KNumKeysByGenerateNewFile - 1)); options.compaction_style = kCompactionStyleLevel; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 2; options.num_levels = 4; options.max_bytes_for_level_base = 400 * 1024; options.max_subcompactions = max_subcompactions_; // First two levels have no compression, so that a trivial move between // them will be allowed. Level 2 has Zlib compression so that a trivial // move to level 3 will not be allowed options.compression_per_level = {kNoCompression, kNoCompression, kZlibCompression}; int matches = 0, didnt_match = 0, trivial_move = 0, non_trivial = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "Compaction::InputCompressionMatchesOutput:Matches", [&](void* arg) { matches++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "Compaction::InputCompressionMatchesOutput:DidntMatch", [&](void* arg) { didnt_match++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Reopen(options); Random rnd(301); int key_idx = 0; // First three 110KB files are going to level 0 // After that, (100K, 200K) for (int num = 0; num < 3; num++) { GenerateNewFile(&rnd, &key_idx); } // Another 110KB triggers a compaction to 400K file to fill up level 0 GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(4, GetSstFileCount(dbname_)); // (1, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4", FilesPerLevel(0)); // (1, 4, 1) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,1", FilesPerLevel(0)); // (1, 4, 2) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,2", FilesPerLevel(0)); // (1, 4, 3) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,3", FilesPerLevel(0)); // (1, 4, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,4", FilesPerLevel(0)); // (1, 4, 5) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,5", FilesPerLevel(0)); // (1, 4, 6) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,6", FilesPerLevel(0)); // (1, 4, 7) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,7", FilesPerLevel(0)); // (1, 4, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ("1,4,8", FilesPerLevel(0)); ASSERT_EQ(matches, 12); // Currently, the test relies on the number of calls to // InputCompressionMatchesOutput() per compaction. const int kCallsToInputCompressionMatch = 2; ASSERT_EQ(didnt_match, 8 * kCallsToInputCompressionMatch); ASSERT_EQ(trivial_move, 12); ASSERT_EQ(non_trivial, 8); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Reopen(options); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Destroy(options); } TEST_F(DBCompactionTest, SanitizeCompactionOptionsTest) { Options options = CurrentOptions(); options.max_background_compactions = 5; options.soft_pending_compaction_bytes_limit = 0; options.hard_pending_compaction_bytes_limit = 100; options.create_if_missing = true; DestroyAndReopen(options); ASSERT_EQ(5, db_->GetOptions().base_background_compactions); ASSERT_EQ(100, db_->GetOptions().soft_pending_compaction_bytes_limit); options.base_background_compactions = 4; options.max_background_compactions = 3; options.soft_pending_compaction_bytes_limit = 200; options.hard_pending_compaction_bytes_limit = 150; DestroyAndReopen(options); ASSERT_EQ(3, db_->GetOptions().base_background_compactions); ASSERT_EQ(150, db_->GetOptions().soft_pending_compaction_bytes_limit); } // This tests for a bug that could cause two level0 compactions running // concurrently // TODO(aekmekji): Make sure that the reason this fails when run with // max_subcompactions > 1 is not a correctness issue but just inherent to // running parallel L0-L1 compactions TEST_F(DBCompactionTest, SuggestCompactRangeNoTwoLevel0Compactions) { Options options = CurrentOptions(); options.compaction_style = kCompactionStyleLevel; options.write_buffer_size = 110 << 10; options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 4; options.num_levels = 4; options.compression = kNoCompression; options.max_bytes_for_level_base = 450 << 10; options.target_file_size_base = 98 << 10; options.max_write_buffer_number = 2; options.max_background_compactions = 2; DestroyAndReopen(options); // fill up the DB Random rnd(301); for (int num = 0; num < 10; num++) { GenerateNewRandomFile(&rnd); } db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"CompactionJob::Run():Start", "DBCompactionTest::SuggestCompactRangeNoTwoLevel0Compactions:1"}, {"DBCompactionTest::SuggestCompactRangeNoTwoLevel0Compactions:2", "CompactionJob::Run():End"}}); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // trigger L0 compaction for (int num = 0; num < options.level0_file_num_compaction_trigger + 1; num++) { GenerateNewRandomFile(&rnd, /* nowait */ true); ASSERT_OK(Flush()); } TEST_SYNC_POINT( "DBCompactionTest::SuggestCompactRangeNoTwoLevel0Compactions:1"); GenerateNewRandomFile(&rnd, /* nowait */ true); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_OK(experimental::SuggestCompactRange(db_, nullptr, nullptr)); for (int num = 0; num < options.level0_file_num_compaction_trigger + 1; num++) { GenerateNewRandomFile(&rnd, /* nowait */ true); ASSERT_OK(Flush()); } TEST_SYNC_POINT( "DBCompactionTest::SuggestCompactRangeNoTwoLevel0Compactions:2"); dbfull()->TEST_WaitForCompact(); } TEST_P(DBCompactionTestWithParam, ForceBottommostLevelCompaction) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options; options.write_buffer_size = 100000000; options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); DestroyAndReopen(options); int32_t value_size = 10 * 1024; // 10 KB Random rnd(301); std::vector values; // File with keys [ 0 => 99 ] for (int i = 0; i < 100; i++) { values.push_back(RandomString(&rnd, value_size)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); ASSERT_EQ("1", FilesPerLevel(0)); // Compaction will do L0=>L1 (trivial move) then move L1 files to L3 CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 3; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); ASSERT_EQ("0,0,0,1", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); // File with keys [ 100 => 199 ] for (int i = 100; i < 200; i++) { values.push_back(RandomString(&rnd, value_size)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); ASSERT_EQ("1,0,0,1", FilesPerLevel(0)); // Compaction will do L0=>L1 L1=>L2 L2=>L3 (3 trivial moves) // then compacte the bottommost level L3=>L3 (non trivial move) compact_options = CompactRangeOptions(); compact_options.bottommost_level_compaction = BottommostLevelCompaction::kForce; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); ASSERT_EQ("0,0,0,1", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 4); ASSERT_EQ(non_trivial_move, 1); // File with keys [ 200 => 299 ] for (int i = 200; i < 300; i++) { values.push_back(RandomString(&rnd, value_size)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Flush()); ASSERT_EQ("1,0,0,1", FilesPerLevel(0)); trivial_move = 0; non_trivial_move = 0; compact_options = CompactRangeOptions(); compact_options.bottommost_level_compaction = BottommostLevelCompaction::kSkip; // Compaction will do L0=>L1 L1=>L2 L2=>L3 (3 trivial moves) // and will skip bottommost level compaction ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); ASSERT_EQ("0,0,0,2", FilesPerLevel(0)); ASSERT_EQ(trivial_move, 3); ASSERT_EQ(non_trivial_move, 0); for (int i = 0; i < 300; i++) { ASSERT_EQ(Get(Key(i)), values[i]); } rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } INSTANTIATE_TEST_CASE_P(DBCompactionTestWithParam, DBCompactionTestWithParam, ::testing::Values(std::make_tuple(1, true), std::make_tuple(1, false), std::make_tuple(4, true), std::make_tuple(4, false))); class CompactionPriTest : public DBTestBase, public testing::WithParamInterface { public: CompactionPriTest() : DBTestBase("/compaction_pri_test") { compaction_pri_ = GetParam(); } // Required if inheriting from testing::WithParamInterface<> static void SetUpTestCase() {} static void TearDownTestCase() {} uint32_t compaction_pri_; }; TEST_P(CompactionPriTest, Test) { Options options; options.write_buffer_size = 16 * 1024; options.compaction_pri = static_cast(compaction_pri_); options.hard_pending_compaction_bytes_limit = 256 * 1024; options.max_bytes_for_level_base = 64 * 1024; options.max_bytes_for_level_multiplier = 4; options.compression = kNoCompression; options = CurrentOptions(options); DestroyAndReopen(options); Random rnd(301); const int kNKeys = 5000; int keys[kNKeys]; for (int i = 0; i < kNKeys; i++) { keys[i] = i; } std::random_shuffle(std::begin(keys), std::end(keys)); for (int i = 0; i < kNKeys; i++) { ASSERT_OK(Put(Key(keys[i]), RandomString(&rnd, 102))); } dbfull()->TEST_WaitForCompact(); for (int i = 0; i < kNKeys; i++) { ASSERT_NE("NOT_FOUND", Get(Key(i))); } } INSTANTIATE_TEST_CASE_P(CompactionPriTest, CompactionPriTest, ::testing::Values(0, 1, 2)); #endif // !defined(ROCKSDB_LITE) } // namespace rocksdb int main(int argc, char** argv) { #if !defined(ROCKSDB_LITE) rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); #else return 0; #endif } rocksdb-4.5.1/db/db_dynamic_level_test.cc000066400000000000000000000410071267530634000203730ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Introduction of SyncPoint effectively disabled building and running this test // in Release build. // which is a pity, it is a good test #if !defined(ROCKSDB_LITE) #include "db/db_test_util.h" #include "port/stack_trace.h" namespace rocksdb { class DBTestDynamicLevel : public DBTestBase { public: DBTestDynamicLevel() : DBTestBase("/db_dynamic_level_test") {} }; TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase) { if (!Snappy_Supported() || !LZ4_Supported()) { return; } // Use InMemoryEnv, or it would be too slow. unique_ptr env(new MockEnv(env_)); const int kNKeys = 1000; int keys[kNKeys]; auto verify_func = [&]() { for (int i = 0; i < kNKeys; i++) { ASSERT_NE("NOT_FOUND", Get(Key(i))); ASSERT_NE("NOT_FOUND", Get(Key(kNKeys * 2 + i))); if (i < kNKeys / 10) { ASSERT_EQ("NOT_FOUND", Get(Key(kNKeys + keys[i]))); } else { ASSERT_NE("NOT_FOUND", Get(Key(kNKeys + keys[i]))); } } }; Random rnd(301); for (int ordered_insert = 0; ordered_insert <= 1; ordered_insert++) { for (int i = 0; i < kNKeys; i++) { keys[i] = i; } if (ordered_insert == 0) { std::random_shuffle(std::begin(keys), std::end(keys)); } for (int max_background_compactions = 1; max_background_compactions < 4; max_background_compactions += 2) { Options options; options.env = env.get(); options.create_if_missing = true; options.db_write_buffer_size = 2048; options.write_buffer_size = 2048; options.max_write_buffer_number = 2; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 2; options.target_file_size_base = 2048; options.level_compaction_dynamic_level_bytes = true; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.soft_rate_limit = 1.1; options.max_background_compactions = max_background_compactions; options.num_levels = 5; options.compression_per_level.resize(3); options.compression_per_level[0] = kNoCompression; options.compression_per_level[1] = kLZ4Compression; options.compression_per_level[2] = kSnappyCompression; DestroyAndReopen(options); for (int i = 0; i < kNKeys; i++) { int key = keys[i]; ASSERT_OK(Put(Key(kNKeys + key), RandomString(&rnd, 102))); ASSERT_OK(Put(Key(key), RandomString(&rnd, 102))); ASSERT_OK(Put(Key(kNKeys * 2 + key), RandomString(&rnd, 102))); ASSERT_OK(Delete(Key(kNKeys + keys[i / 10]))); env_->SleepForMicroseconds(5000); } uint64_t int_prop; ASSERT_TRUE(db_->GetIntProperty("rocksdb.background-errors", &int_prop)); ASSERT_EQ(0U, int_prop); // Verify DB for (int j = 0; j < 2; j++) { verify_func(); if (j == 0) { Reopen(options); } } // Test compact range works dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); // All data should be in the last level. ColumnFamilyMetaData cf_meta; db_->GetColumnFamilyMetaData(&cf_meta); ASSERT_EQ(5U, cf_meta.levels.size()); for (int i = 0; i < 4; i++) { ASSERT_EQ(0U, cf_meta.levels[i].files.size()); } ASSERT_GT(cf_meta.levels[4U].files.size(), 0U); verify_func(); Close(); } } env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); } // Test specific cases in dynamic max bytes TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) { Random rnd(301); int kMaxKey = 1000000; Options options = CurrentOptions(); options.create_if_missing = true; options.db_write_buffer_size = 2048; options.write_buffer_size = 2048; options.max_write_buffer_number = 2; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 9999; options.level0_stop_writes_trigger = 9999; options.target_file_size_base = 2048; options.level_compaction_dynamic_level_bytes = true; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.max_background_compactions = 2; options.num_levels = 5; options.expanded_compaction_factor = 0; // Force not expanding in compactions BlockBasedTableOptions table_options; table_options.block_size = 1024; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); DestroyAndReopen(options); ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "true"}, })); uint64_t int_prop; std::string str_prop; // Initial base level is the last level ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(4U, int_prop); // Put about 7K to L0 for (int i = 0; i < 70; i++) { ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); } ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "false"}, })); Flush(); dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(4U, int_prop); // Insert extra about 3.5K to L0. After they are compacted to L4, base level // should be changed to L3. ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "true"}, })); for (int i = 0; i < 70; i++) { ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); } ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "false"}, })); Flush(); dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(3U, int_prop); ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level1", &str_prop)); ASSERT_EQ("0", str_prop); ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level2", &str_prop)); ASSERT_EQ("0", str_prop); // Trigger parallel compaction, and the first one would change the base // level. // Hold compaction jobs to make sure rocksdb::SyncPoint::GetInstance()->SetCallBack( "CompactionJob::Run():Start", [&](void* arg) { env_->SleepForMicroseconds(100000); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "true"}, })); // Write about 10K more for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); } ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "false"}, })); Flush(); // Wait for 200 milliseconds before proceeding compactions to make sure two // parallel ones are executed. env_->SleepForMicroseconds(200000); dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(3U, int_prop); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); // Trigger a condition that the compaction changes base level and L0->Lbase // happens at the same time. // We try to make last levels' targets to be 10K, 40K, 160K, add triggers // another compaction from 40K->160K. ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "true"}, })); // Write about 150K more for (int i = 0; i < 1350; i++) { ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); } ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "false"}, })); Flush(); dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(2U, int_prop); // Keep Writing data until base level changed 2->1. There will be L0->L2 // compaction going on at the same time. rocksdb::SyncPoint::GetInstance()->EnableProcessing(); for (int attempt = 0; attempt <= 20; attempt++) { // Write about 5K more data with two flushes. It should be flush to level 2 // but when it is applied, base level is already 1. for (int i = 0; i < 50; i++) { ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); } Flush(); ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); if (int_prop == 2U) { env_->SleepForMicroseconds(50000); } else { break; } } rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); env_->SleepForMicroseconds(200000); ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(1U, int_prop); } // Test specific cases in dynamic max bytes TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) { Random rnd(301); int kMaxKey = 1000000; Options options = CurrentOptions(); options.create_if_missing = true; options.db_write_buffer_size = 2048; options.write_buffer_size = 2048; options.max_write_buffer_number = 2; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 9999; options.level0_stop_writes_trigger = 9999; options.target_file_size_base = 2; options.level_compaction_dynamic_level_bytes = true; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.max_background_compactions = 1; const int kNumLevels = 5; options.num_levels = kNumLevels; options.expanded_compaction_factor = 0; // Force not expanding in compactions BlockBasedTableOptions table_options; table_options.block_size = 1024; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); DestroyAndReopen(options); // Compact against empty DB dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); uint64_t int_prop; std::string str_prop; // Initial base level is the last level ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(4U, int_prop); // Put about 7K to L0 for (int i = 0; i < 140; i++) { ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); } Flush(); dbfull()->TEST_WaitForCompact(); if (NumTableFilesAtLevel(0) == 0) { // Make sure level 0 is not empty ASSERT_OK(Put(Key(static_cast(rnd.Uniform(kMaxKey))), RandomString(&rnd, 80))); Flush(); } ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(3U, int_prop); ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level1", &str_prop)); ASSERT_EQ("0", str_prop); ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level2", &str_prop)); ASSERT_EQ("0", str_prop); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); std::set output_levels; rocksdb::SyncPoint::GetInstance()->SetCallBack( "CompactionPicker::CompactRange:Return", [&](void* arg) { Compaction* compaction = reinterpret_cast(arg); output_levels.insert(compaction->output_level()); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(output_levels.size(), 2); ASSERT_TRUE(output_levels.find(3) != output_levels.end()); ASSERT_TRUE(output_levels.find(4) != output_levels.end()); ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level0", &str_prop)); ASSERT_EQ("0", str_prop); ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level3", &str_prop)); ASSERT_EQ("0", str_prop); // Base level is still level 3. ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop)); ASSERT_EQ(3U, int_prop); } TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBaseInc) { Options options = CurrentOptions(); options.create_if_missing = true; options.db_write_buffer_size = 2048; options.write_buffer_size = 2048; options.max_write_buffer_number = 2; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 2; options.target_file_size_base = 2048; options.level_compaction_dynamic_level_bytes = true; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.soft_rate_limit = 1.1; options.max_background_compactions = 2; options.num_levels = 5; DestroyAndReopen(options); int non_trivial = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Random rnd(301); const int total_keys = 3000; const int random_part_size = 100; for (int i = 0; i < total_keys; i++) { std::string value = RandomString(&rnd, random_part_size); PutFixed32(&value, static_cast(i)); ASSERT_OK(Put(Key(i), value)); } Flush(); dbfull()->TEST_WaitForCompact(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); ASSERT_EQ(non_trivial, 0); for (int i = 0; i < total_keys; i++) { std::string value = Get(Key(i)); ASSERT_EQ(DecodeFixed32(value.c_str() + random_part_size), static_cast(i)); } env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); } TEST_F(DBTestDynamicLevel, MigrateToDynamicLevelMaxBytesBase) { Random rnd(301); const int kMaxKey = 2000; Options options; options.create_if_missing = true; options.db_write_buffer_size = 2048; options.write_buffer_size = 2048; options.max_write_buffer_number = 8; options.level0_file_num_compaction_trigger = 4; options.level0_slowdown_writes_trigger = 4; options.level0_stop_writes_trigger = 8; options.target_file_size_base = 2048; options.level_compaction_dynamic_level_bytes = false; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.soft_rate_limit = 1.1; options.num_levels = 8; DestroyAndReopen(options); auto verify_func = [&](int num_keys, bool if_sleep) { for (int i = 0; i < num_keys; i++) { ASSERT_NE("NOT_FOUND", Get(Key(kMaxKey + i))); if (i < num_keys / 10) { ASSERT_EQ("NOT_FOUND", Get(Key(i))); } else { ASSERT_NE("NOT_FOUND", Get(Key(i))); } if (if_sleep && i % 1000 == 0) { // Without it, valgrind may choose not to give another // thread a chance to run before finishing the function, // causing the test to be extremely slow. env_->SleepForMicroseconds(1); } } }; int total_keys = 1000; for (int i = 0; i < total_keys; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 102))); ASSERT_OK(Put(Key(kMaxKey + i), RandomString(&rnd, 102))); ASSERT_OK(Delete(Key(i / 10))); } verify_func(total_keys, false); dbfull()->TEST_WaitForCompact(); options.level_compaction_dynamic_level_bytes = true; options.disable_auto_compactions = true; Reopen(options); verify_func(total_keys, false); std::atomic_bool compaction_finished; compaction_finished = false; // Issue manual compaction in one thread and still verify DB state // in main thread. std::thread t([&]() { CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = options.num_levels - 1; dbfull()->CompactRange(compact_options, nullptr, nullptr); compaction_finished.store(true); }); do { verify_func(total_keys, true); } while (!compaction_finished.load()); t.join(); ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "false"}, })); int total_keys2 = 2000; for (int i = total_keys; i < total_keys2; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 102))); ASSERT_OK(Put(Key(kMaxKey + i), RandomString(&rnd, 102))); ASSERT_OK(Delete(Key(i / 10))); } verify_func(total_keys2, false); dbfull()->TEST_WaitForCompact(); verify_func(total_keys2, false); // Base level is not level 1 ASSERT_EQ(NumTableFilesAtLevel(1), 0); ASSERT_EQ(NumTableFilesAtLevel(2), 0); } } // namespace rocksdb #endif // !defined(ROCKSDB_LITE) int main(int argc, char** argv) { #if !defined(ROCKSDB_LITE) rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); #else return 0; #endif } rocksdb-4.5.1/db/db_filesnapshot.cc000066400000000000000000000104211267530634000172140ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2012 Facebook. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. #ifndef ROCKSDB_LITE #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include "db/db_impl.h" #include "db/filename.h" #include "db/job_context.h" #include "db/version_set.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "port/port.h" #include "util/mutexlock.h" #include "util/sync_point.h" #include "util/file_util.h" namespace rocksdb { Status DBImpl::DisableFileDeletions() { InstrumentedMutexLock l(&mutex_); ++disable_delete_obsolete_files_; if (disable_delete_obsolete_files_ == 1) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "File Deletions Disabled"); } else { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "File Deletions Disabled, but already disabled. Counter: %d", disable_delete_obsolete_files_); } return Status::OK(); } Status DBImpl::EnableFileDeletions(bool force) { // Job id == 0 means that this is not our background process, but rather // user thread JobContext job_context(0); bool should_purge_files = false; { InstrumentedMutexLock l(&mutex_); if (force) { // if force, we need to enable file deletions right away disable_delete_obsolete_files_ = 0; } else if (disable_delete_obsolete_files_ > 0) { --disable_delete_obsolete_files_; } if (disable_delete_obsolete_files_ == 0) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "File Deletions Enabled"); should_purge_files = true; FindObsoleteFiles(&job_context, true); } else { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "File Deletions Enable, but not really enabled. Counter: %d", disable_delete_obsolete_files_); } } if (should_purge_files) { PurgeObsoleteFiles(job_context); } job_context.Clean(); LogFlush(db_options_.info_log); return Status::OK(); } int DBImpl::IsFileDeletionsEnabled() const { return disable_delete_obsolete_files_; } Status DBImpl::GetLiveFiles(std::vector& ret, uint64_t* manifest_file_size, bool flush_memtable) { *manifest_file_size = 0; mutex_.Lock(); if (flush_memtable) { // flush all dirty data to disk. Status status; for (auto cfd : *versions_->GetColumnFamilySet()) { if (cfd->IsDropped()) { continue; } cfd->Ref(); mutex_.Unlock(); status = FlushMemTable(cfd, FlushOptions()); TEST_SYNC_POINT("DBImpl::GetLiveFiles:1"); TEST_SYNC_POINT("DBImpl::GetLiveFiles:2"); mutex_.Lock(); cfd->Unref(); if (!status.ok()) { break; } } versions_->GetColumnFamilySet()->FreeDeadColumnFamilies(); if (!status.ok()) { mutex_.Unlock(); Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Cannot Flush data %s\n", status.ToString().c_str()); return status; } } // Make a set of all of the live *.sst files std::vector live; for (auto cfd : *versions_->GetColumnFamilySet()) { if (cfd->IsDropped()) { continue; } cfd->current()->AddLiveFiles(&live); } ret.clear(); ret.reserve(live.size() + 2); //*.sst + CURRENT + MANIFEST // create names of the live files. The names are not absolute // paths, instead they are relative to dbname_; for (auto live_file : live) { ret.push_back(MakeTableFileName("", live_file.GetNumber())); } ret.push_back(CurrentFileName("")); ret.push_back(DescriptorFileName("", versions_->manifest_file_number())); // find length of manifest file while holding the mutex lock *manifest_file_size = versions_->manifest_file_size(); mutex_.Unlock(); return Status::OK(); } Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) { return wal_manager_.GetSortedWalFiles(files); } } #endif // ROCKSDB_LITE rocksdb-4.5.1/db/db_impl.cc000066400000000000000000006252201267530634000154670ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_impl.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #ifdef OS_SOLARIS #include #endif #include #include #include #include #include #include #include #include #include #include #include #include "db/auto_roll_logger.h" #include "db/builder.h" #include "db/compaction_job.h" #include "db/db_info_dumper.h" #include "db/db_iter.h" #include "db/dbformat.h" #include "db/event_helpers.h" #include "db/filename.h" #include "db/flush_job.h" #include "db/forward_iterator.h" #include "db/job_context.h" #include "db/log_reader.h" #include "db/log_writer.h" #include "db/managed_iterator.h" #include "db/memtable.h" #include "db/memtable_list.h" #include "db/merge_context.h" #include "db/merge_helper.h" #include "db/table_cache.h" #include "db/table_properties_collector.h" #include "db/transaction_log_impl.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "db/write_callback.h" #include "db/writebuffer.h" #include "db/xfunc_test_points.h" #include "memtable/hash_linklist_rep.h" #include "memtable/hash_skiplist_rep.h" #include "port/likely.h" #include "port/port.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/merge_operator.h" #include "rocksdb/sst_file_writer.h" #include "rocksdb/statistics.h" #include "rocksdb/status.h" #include "rocksdb/table.h" #include "rocksdb/version.h" #include "rocksdb/wal_filter.h" #include "table/block.h" #include "table/block_based_table_factory.h" #include "table/merger.h" #include "table/table_builder.h" #include "table/two_level_iterator.h" #include "util/autovector.h" #include "util/build_version.h" #include "util/coding.h" #include "util/compression.h" #include "util/crc32c.h" #include "util/file_reader_writer.h" #include "util/file_util.h" #include "util/iostats_context_imp.h" #include "util/log_buffer.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/sst_file_manager_impl.h" #include "util/options_helper.h" #include "util/options_parser.h" #include "util/perf_context_imp.h" #include "util/stop_watch.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/thread_status_updater.h" #include "util/thread_status_util.h" #include "util/xfunc.h" namespace rocksdb { const std::string kDefaultColumnFamilyName("default"); void DumpRocksDBBuildVersion(Logger * log); struct DBImpl::WriteContext { autovector superversions_to_free_; autovector memtables_to_free_; ~WriteContext() { for (auto& sv : superversions_to_free_) { delete sv; } for (auto& m : memtables_to_free_) { delete m; } } }; Options SanitizeOptions(const std::string& dbname, const InternalKeyComparator* icmp, const Options& src) { auto db_options = SanitizeOptions(dbname, DBOptions(src)); auto cf_options = SanitizeOptions(db_options, icmp, ColumnFamilyOptions(src)); return Options(db_options, cf_options); } DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) { DBOptions result = src; // result.max_open_files means an "infinite" open files. if (result.max_open_files != -1) { int max_max_open_files = port::GetMaxOpenFiles(); if (max_max_open_files == -1) { max_max_open_files = 1000000; } ClipToRange(&result.max_open_files, 20, max_max_open_files); } if (result.info_log == nullptr) { Status s = CreateLoggerFromOptions(dbname, result, &result.info_log); if (!s.ok()) { // No place suitable for logging result.info_log = nullptr; } } if (result.base_background_compactions == -1) { result.base_background_compactions = result.max_background_compactions; } if (result.base_background_compactions > result.max_background_compactions) { result.base_background_compactions = result.max_background_compactions; } result.env->IncBackgroundThreadsIfNeeded(src.max_background_compactions, Env::Priority::LOW); result.env->IncBackgroundThreadsIfNeeded(src.max_background_flushes, Env::Priority::HIGH); if (result.rate_limiter.get() != nullptr) { if (result.bytes_per_sync == 0) { result.bytes_per_sync = 1024 * 1024; } } if (result.WAL_ttl_seconds > 0 || result.WAL_size_limit_MB > 0) { result.recycle_log_file_num = false; } if (result.wal_dir.empty()) { // Use dbname as default result.wal_dir = dbname; } if (result.wal_dir.back() == '/') { result.wal_dir = result.wal_dir.substr(0, result.wal_dir.size() - 1); } if (result.db_paths.size() == 0) { result.db_paths.emplace_back(dbname, std::numeric_limits::max()); } if (result.compaction_readahead_size > 0) { result.new_table_reader_for_compaction_inputs = true; } return result; } namespace { Status SanitizeOptionsByTable( const DBOptions& db_opts, const std::vector& column_families) { Status s; for (auto cf : column_families) { s = cf.options.table_factory->SanitizeOptions(db_opts, cf.options); if (!s.ok()) { return s; } } return Status::OK(); } CompressionType GetCompressionFlush(const ImmutableCFOptions& ioptions) { // Compressing memtable flushes might not help unless the sequential load // optimization is used for leveled compaction. Otherwise the CPU and // latency overhead is not offset by saving much space. bool can_compress; if (ioptions.compaction_style == kCompactionStyleUniversal) { can_compress = (ioptions.compaction_options_universal.compression_size_percent < 0); } else { // For leveled compress when min_level_to_compress == 0. can_compress = ioptions.compression_per_level.empty() || ioptions.compression_per_level[0] != kNoCompression; } if (can_compress) { return ioptions.compression; } else { return kNoCompression; } } void DumpSupportInfo(Logger* logger) { Log(InfoLogLevel::INFO_LEVEL, logger, "Compression algorithms supported:"); Log(InfoLogLevel::INFO_LEVEL, logger, "\tSnappy supported: %d", Snappy_Supported()); Log(InfoLogLevel::INFO_LEVEL, logger, "\tZlib supported: %d", Zlib_Supported()); Log(InfoLogLevel::INFO_LEVEL, logger, "\tBzip supported: %d", BZip2_Supported()); Log(InfoLogLevel::INFO_LEVEL, logger, "\tLZ4 supported: %d", LZ4_Supported()); Log(InfoLogLevel::INFO_LEVEL, logger, "Fast CRC32 supported: %d", crc32c::IsFastCrc32Supported()); } } // namespace DBImpl::DBImpl(const DBOptions& options, const std::string& dbname) : env_(options.env), dbname_(dbname), db_options_(SanitizeOptions(dbname, options)), stats_(db_options_.statistics.get()), db_lock_(nullptr), mutex_(stats_, env_, DB_MUTEX_WAIT_MICROS, options.use_adaptive_mutex), shutting_down_(false), bg_cv_(&mutex_), logfile_number_(0), log_dir_synced_(false), log_empty_(true), default_cf_handle_(nullptr), log_sync_cv_(&mutex_), total_log_size_(0), max_total_in_memory_state_(0), is_snapshot_supported_(true), write_buffer_(options.db_write_buffer_size), write_thread_(options.enable_write_thread_adaptive_yield ? options.write_thread_max_yield_usec : 0, options.write_thread_slow_yield_usec), write_controller_(options.delayed_write_rate), last_batch_group_size_(0), unscheduled_flushes_(0), unscheduled_compactions_(0), bg_compaction_scheduled_(0), num_running_compactions_(0), bg_flush_scheduled_(0), num_running_flushes_(0), disable_delete_obsolete_files_(0), delete_obsolete_files_next_run_( options.env->NowMicros() + db_options_.delete_obsolete_files_period_micros), last_stats_dump_time_microsec_(0), next_job_id_(1), has_unpersisted_data_(false), env_options_(db_options_), #ifndef ROCKSDB_LITE wal_manager_(db_options_, env_options_), #endif // ROCKSDB_LITE event_logger_(db_options_.info_log.get()), bg_work_paused_(0), refitting_level_(false), opened_successfully_(false) { env_->GetAbsolutePath(dbname, &db_absolute_path_); // Reserve ten files or so for other uses and give the rest to TableCache. // Give a large number for setting of "infinite" open files. const int table_cache_size = (db_options_.max_open_files == -1) ? 4194304 : db_options_.max_open_files - 10; table_cache_ = NewLRUCache(table_cache_size, db_options_.table_cache_numshardbits); versions_.reset(new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(), &write_buffer_, &write_controller_)); column_family_memtables_.reset( new ColumnFamilyMemTablesImpl(versions_->GetColumnFamilySet())); DumpRocksDBBuildVersion(db_options_.info_log.get()); DumpDBFileSummary(db_options_, dbname_); db_options_.Dump(db_options_.info_log.get()); DumpSupportInfo(db_options_.info_log.get()); } // Will lock the mutex_, will wait for completion if wait is true void DBImpl::CancelAllBackgroundWork(bool wait) { InstrumentedMutexLock l(&mutex_); shutting_down_.store(true, std::memory_order_release); bg_cv_.SignalAll(); if (!wait) { return; } // Wait for background work to finish while (bg_compaction_scheduled_ || bg_flush_scheduled_) { bg_cv_.Wait(); } } DBImpl::~DBImpl() { mutex_.Lock(); if (!shutting_down_.load(std::memory_order_acquire) && has_unpersisted_data_) { for (auto cfd : *versions_->GetColumnFamilySet()) { if (!cfd->IsDropped() && !cfd->mem()->IsEmpty()) { cfd->Ref(); mutex_.Unlock(); FlushMemTable(cfd, FlushOptions()); mutex_.Lock(); cfd->Unref(); } } versions_->GetColumnFamilySet()->FreeDeadColumnFamilies(); } mutex_.Unlock(); // CancelAllBackgroundWork called with false means we just set the shutdown // marker. After this we do a variant of the waiting and unschedule work // (to consider: moving all the waiting into CancelAllBackgroundWork(true)) CancelAllBackgroundWork(false); int compactions_unscheduled = env_->UnSchedule(this, Env::Priority::LOW); int flushes_unscheduled = env_->UnSchedule(this, Env::Priority::HIGH); mutex_.Lock(); bg_compaction_scheduled_ -= compactions_unscheduled; bg_flush_scheduled_ -= flushes_unscheduled; // Wait for background work to finish while (bg_compaction_scheduled_ || bg_flush_scheduled_) { bg_cv_.Wait(); } EraseThreadStatusDbInfo(); flush_scheduler_.Clear(); while (!flush_queue_.empty()) { auto cfd = PopFirstFromFlushQueue(); if (cfd->Unref()) { delete cfd; } } while (!compaction_queue_.empty()) { auto cfd = PopFirstFromCompactionQueue(); if (cfd->Unref()) { delete cfd; } } if (default_cf_handle_ != nullptr) { // we need to delete handle outside of lock because it does its own locking mutex_.Unlock(); delete default_cf_handle_; mutex_.Lock(); } // Clean up obsolete files due to SuperVersion release. // (1) Need to delete to obsolete files before closing because RepairDB() // scans all existing files in the file system and builds manifest file. // Keeping obsolete files confuses the repair process. // (2) Need to check if we Open()/Recover() the DB successfully before // deleting because if VersionSet recover fails (may be due to corrupted // manifest file), it is not able to identify live files correctly. As a // result, all "live" files can get deleted by accident. However, corrupted // manifest is recoverable by RepairDB(). if (opened_successfully_) { JobContext job_context(next_job_id_.fetch_add(1)); FindObsoleteFiles(&job_context, true); mutex_.Unlock(); // manifest number starting from 2 job_context.manifest_file_number = 1; if (job_context.HaveSomethingToDelete()) { PurgeObsoleteFiles(job_context); } job_context.Clean(); mutex_.Lock(); } for (auto l : logs_to_free_) { delete l; } for (auto& log : logs_) { log.ClearWriter(); } logs_.clear(); // versions need to be destroyed before table_cache since it can hold // references to table_cache. versions_.reset(); mutex_.Unlock(); if (db_lock_ != nullptr) { env_->UnlockFile(db_lock_); } LogFlush(db_options_.info_log); } Status DBImpl::NewDB() { VersionEdit new_db; new_db.SetLogNumber(0); new_db.SetNextFile(2); new_db.SetLastSequence(0); Status s; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Creating manifest 1 \n"); const std::string manifest = DescriptorFileName(dbname_, 1); { unique_ptr file; EnvOptions env_options = env_->OptimizeForManifestWrite(env_options_); s = NewWritableFile(env_, manifest, &file, env_options); if (!s.ok()) { return s; } file->SetPreallocationBlockSize(db_options_.manifest_preallocation_size); unique_ptr file_writer( new WritableFileWriter(std::move(file), env_options)); log::Writer log(std::move(file_writer), 0, false); std::string record; new_db.EncodeTo(&record); s = log.AddRecord(record); if (s.ok()) { s = SyncManifest(env_, &db_options_, log.file()); } } if (s.ok()) { // Make "CURRENT" file that points to the new manifest file. s = SetCurrentFile(env_, dbname_, 1, directories_.GetDbDir()); } else { env_->DeleteFile(manifest); } return s; } void DBImpl::MaybeIgnoreError(Status* s) const { if (s->ok() || db_options_.paranoid_checks) { // No change needed } else { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Ignoring error %s", s->ToString().c_str()); *s = Status::OK(); } } const Status DBImpl::CreateArchivalDirectory() { if (db_options_.WAL_ttl_seconds > 0 || db_options_.WAL_size_limit_MB > 0) { std::string archivalPath = ArchivalDirectory(db_options_.wal_dir); return env_->CreateDirIfMissing(archivalPath); } return Status::OK(); } void DBImpl::PrintStatistics() { auto dbstats = db_options_.statistics.get(); if (dbstats) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "STATISTICS:\n %s", dbstats->ToString().c_str()); } } void DBImpl::MaybeDumpStats() { if (db_options_.stats_dump_period_sec == 0) return; const uint64_t now_micros = env_->NowMicros(); if (last_stats_dump_time_microsec_ + db_options_.stats_dump_period_sec * 1000000 <= now_micros) { // Multiple threads could race in here simultaneously. // However, the last one will update last_stats_dump_time_microsec_ // atomically. We could see more than one dump during one dump // period in rare cases. last_stats_dump_time_microsec_ = now_micros; #ifndef ROCKSDB_LITE const DBPropertyInfo* cf_property_info = GetPropertyInfo(DB::Properties::kCFStats); assert(cf_property_info != nullptr); const DBPropertyInfo* db_property_info = GetPropertyInfo(DB::Properties::kDBStats); assert(db_property_info != nullptr); std::string stats; { InstrumentedMutexLock l(&mutex_); for (auto cfd : *versions_->GetColumnFamilySet()) { cfd->internal_stats()->GetStringProperty( *cf_property_info, DB::Properties::kCFStats, &stats); } default_cf_internal_stats_->GetStringProperty( *db_property_info, DB::Properties::kDBStats, &stats); } Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "------- DUMPING STATS -------"); Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "%s", stats.c_str()); #endif // !ROCKSDB_LITE PrintStatistics(); } } // * Returns the list of live files in 'sst_live' // If it's doing full scan: // * Returns the list of all files in the filesystem in // 'full_scan_candidate_files'. // Otherwise, gets obsolete files from VersionSet. // no_full_scan = true -- never do the full scan using GetChildren() // force = false -- don't force the full scan, except every // db_options_.delete_obsolete_files_period_micros // force = true -- force the full scan void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force, bool no_full_scan) { mutex_.AssertHeld(); // if deletion is disabled, do nothing if (disable_delete_obsolete_files_ > 0) { return; } bool doing_the_full_scan = false; // logic for figurint out if we're doing the full scan if (no_full_scan) { doing_the_full_scan = false; } else if (force || db_options_.delete_obsolete_files_period_micros == 0) { doing_the_full_scan = true; } else { const uint64_t now_micros = env_->NowMicros(); if (delete_obsolete_files_next_run_ < now_micros) { doing_the_full_scan = true; delete_obsolete_files_next_run_ = now_micros + db_options_.delete_obsolete_files_period_micros; } } // don't delete files that might be currently written to from compaction // threads // Since job_context->min_pending_output is set, until file scan finishes, // mutex_ cannot be released. Otherwise, we might see no min_pending_output // here but later find newer generated unfinalized files while scannint. if (!pending_outputs_.empty()) { job_context->min_pending_output = *pending_outputs_.begin(); } else { // delete all of them job_context->min_pending_output = std::numeric_limits::max(); } // Get obsolete files. This function will also update the list of // pending files in VersionSet(). versions_->GetObsoleteFiles(&job_context->sst_delete_files, job_context->min_pending_output); // store the current filenum, lognum, etc job_context->manifest_file_number = versions_->manifest_file_number(); job_context->pending_manifest_file_number = versions_->pending_manifest_file_number(); job_context->log_number = versions_->MinLogNumber(); job_context->prev_log_number = versions_->prev_log_number(); versions_->AddLiveFiles(&job_context->sst_live); if (doing_the_full_scan) { for (size_t path_id = 0; path_id < db_options_.db_paths.size(); path_id++) { // set of all files in the directory. We'll exclude files that are still // alive in the subsequent processings. std::vector files; env_->GetChildren(db_options_.db_paths[path_id].path, &files); // Ignore errors for (std::string file : files) { // TODO(icanadi) clean up this mess to avoid having one-off "/" prefixes job_context->full_scan_candidate_files.emplace_back( "/" + file, static_cast(path_id)); } } //Add log files in wal_dir if (db_options_.wal_dir != dbname_) { std::vector log_files; env_->GetChildren(db_options_.wal_dir, &log_files); // Ignore errors for (std::string log_file : log_files) { job_context->full_scan_candidate_files.emplace_back(log_file, 0); } } // Add info log files in db_log_dir if (!db_options_.db_log_dir.empty() && db_options_.db_log_dir != dbname_) { std::vector info_log_files; // Ignore errors env_->GetChildren(db_options_.db_log_dir, &info_log_files); for (std::string log_file : info_log_files) { job_context->full_scan_candidate_files.emplace_back(log_file, 0); } } } if (!alive_log_files_.empty()) { uint64_t min_log_number = versions_->MinLogNumber(); // find newly obsoleted log files while (alive_log_files_.begin()->number < min_log_number) { auto& earliest = *alive_log_files_.begin(); if (db_options_.recycle_log_file_num > log_recycle_files.size()) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "adding log %" PRIu64 " to recycle list\n", earliest.number); log_recycle_files.push_back(earliest.number); } else { job_context->log_delete_files.push_back(earliest.number); } total_log_size_ -= earliest.size; alive_log_files_.pop_front(); // Current log should always stay alive since it can't have // number < MinLogNumber(). assert(alive_log_files_.size()); } while (!logs_.empty() && logs_.front().number < min_log_number) { auto& log = logs_.front(); if (log.getting_synced) { log_sync_cv_.Wait(); // logs_ could have changed while we were waiting. continue; } logs_to_free_.push_back(log.ReleaseWriter()); logs_.pop_front(); } // Current log cannot be obsolete. assert(!logs_.empty()); } // We're just cleaning up for DB::Write(). assert(job_context->logs_to_free.empty()); job_context->logs_to_free = logs_to_free_; logs_to_free_.clear(); } namespace { bool CompareCandidateFile(const JobContext::CandidateFileInfo& first, const JobContext::CandidateFileInfo& second) { if (first.file_name > second.file_name) { return true; } else if (first.file_name < second.file_name) { return false; } else { return (first.path_id > second.path_id); } } }; // namespace // Diffs the files listed in filenames and those that do not // belong to live files are posibly removed. Also, removes all the // files in sst_delete_files and log_delete_files. // It is not necessary to hold the mutex when invoking this method. void DBImpl::PurgeObsoleteFiles(const JobContext& state) { // we'd better have sth to delete assert(state.HaveSomethingToDelete()); // this checks if FindObsoleteFiles() was run before. If not, don't do // PurgeObsoleteFiles(). If FindObsoleteFiles() was run, we need to also // run PurgeObsoleteFiles(), even if disable_delete_obsolete_files_ is true if (state.manifest_file_number == 0) { return; } // Now, convert live list to an unordered map, WITHOUT mutex held; // set is slow. std::unordered_map sst_live_map; for (const FileDescriptor& fd : state.sst_live) { sst_live_map[fd.GetNumber()] = &fd; } auto candidate_files = state.full_scan_candidate_files; candidate_files.reserve(candidate_files.size() + state.sst_delete_files.size() + state.log_delete_files.size()); // We may ignore the dbname when generating the file names. const char* kDumbDbName = ""; for (auto file : state.sst_delete_files) { candidate_files.emplace_back( MakeTableFileName(kDumbDbName, file->fd.GetNumber()), file->fd.GetPathId()); delete file; } for (auto file_num : state.log_delete_files) { if (file_num > 0) { candidate_files.emplace_back(LogFileName(kDumbDbName, file_num).substr(1), 0); } } // dedup state.candidate_files so we don't try to delete the same // file twice sort(candidate_files.begin(), candidate_files.end(), CompareCandidateFile); candidate_files.erase(unique(candidate_files.begin(), candidate_files.end()), candidate_files.end()); std::vector old_info_log_files; InfoLogPrefix info_log_prefix(!db_options_.db_log_dir.empty(), dbname_); for (const auto& candidate_file : candidate_files) { std::string to_delete = candidate_file.file_name; uint32_t path_id = candidate_file.path_id; uint64_t number; FileType type; // Ignore file if we cannot recognize it. if (!ParseFileName(to_delete, &number, info_log_prefix.prefix, &type)) { continue; } bool keep = true; switch (type) { case kLogFile: keep = ((number >= state.log_number) || (number == state.prev_log_number)); break; case kDescriptorFile: // Keep my manifest file, and any newer incarnations' // (can happen during manifest roll) keep = (number >= state.manifest_file_number); break; case kTableFile: // If the second condition is not there, this makes // DontDeletePendingOutputs fail keep = (sst_live_map.find(number) != sst_live_map.end()) || number >= state.min_pending_output; break; case kTempFile: // Any temp files that are currently being written to must // be recorded in pending_outputs_, which is inserted into "live". // Also, SetCurrentFile creates a temp file when writing out new // manifest, which is equal to state.pending_manifest_file_number. We // should not delete that file // // TODO(yhchiang): carefully modify the third condition to safely // remove the temp options files. keep = (sst_live_map.find(number) != sst_live_map.end()) || (number == state.pending_manifest_file_number) || (to_delete.find(kOptionsFileNamePrefix) != std::string::npos); break; case kInfoLogFile: keep = true; if (number != 0) { old_info_log_files.push_back(to_delete); } break; case kCurrentFile: case kDBLockFile: case kIdentityFile: case kMetaDatabase: case kOptionsFile: keep = true; break; } if (keep) { continue; } std::string fname; if (type == kTableFile) { // evict from cache TableCache::Evict(table_cache_.get(), number); fname = TableFileName(db_options_.db_paths, number, path_id); } else { fname = ((type == kLogFile) ? db_options_.wal_dir : dbname_) + "/" + to_delete; } #ifndef ROCKSDB_LITE if (type == kLogFile && (db_options_.WAL_ttl_seconds > 0 || db_options_.WAL_size_limit_MB > 0)) { wal_manager_.ArchiveWALFile(fname, number); continue; } #endif // !ROCKSDB_LITE Status file_deletion_status; if (type == kTableFile) { file_deletion_status = DeleteSSTFile(&db_options_, fname, path_id); } else { file_deletion_status = env_->DeleteFile(fname); } if (file_deletion_status.ok()) { Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[JOB %d] Delete %s type=%d #%" PRIu64 " -- %s\n", state.job_id, fname.c_str(), type, number, file_deletion_status.ToString().c_str()); } else if (env_->FileExists(fname).IsNotFound()) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[JOB %d] Tried to delete a non-existing file %s type=%d #%" PRIu64 " -- %s\n", state.job_id, fname.c_str(), type, number, file_deletion_status.ToString().c_str()); } else { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "[JOB %d] Failed to delete %s type=%d #%" PRIu64 " -- %s\n", state.job_id, fname.c_str(), type, number, file_deletion_status.ToString().c_str()); } if (type == kTableFile) { EventHelpers::LogAndNotifyTableFileDeletion( &event_logger_, state.job_id, number, fname, file_deletion_status, GetName(), db_options_.listeners); } } // Delete old info log files. size_t old_info_log_file_count = old_info_log_files.size(); if (old_info_log_file_count >= db_options_.keep_log_file_num) { std::sort(old_info_log_files.begin(), old_info_log_files.end()); size_t end = old_info_log_file_count - db_options_.keep_log_file_num; for (unsigned int i = 0; i <= end; i++) { std::string& to_delete = old_info_log_files.at(i); std::string full_path_to_delete = (db_options_.db_log_dir.empty() ? dbname_ : db_options_.db_log_dir) + "/" + to_delete; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[JOB %d] Delete info log file %s\n", state.job_id, full_path_to_delete.c_str()); Status s = env_->DeleteFile(full_path_to_delete); if (!s.ok()) { if (env_->FileExists(full_path_to_delete).IsNotFound()) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[JOB %d] Tried to delete non-existing info log file %s FAILED " "-- %s\n", state.job_id, to_delete.c_str(), s.ToString().c_str()); } else { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "[JOB %d] Delete info log file %s FAILED -- %s\n", state.job_id, to_delete.c_str(), s.ToString().c_str()); } } } } #ifndef ROCKSDB_LITE wal_manager_.PurgeObsoleteWALFiles(); #endif // ROCKSDB_LITE LogFlush(db_options_.info_log); } void DBImpl::DeleteObsoleteFiles() { mutex_.AssertHeld(); JobContext job_context(next_job_id_.fetch_add(1)); FindObsoleteFiles(&job_context, true); mutex_.Unlock(); if (job_context.HaveSomethingToDelete()) { PurgeObsoleteFiles(job_context); } job_context.Clean(); mutex_.Lock(); } Status DBImpl::Directories::CreateAndNewDirectory( Env* env, const std::string& dirname, std::unique_ptr* directory) const { // We call CreateDirIfMissing() as the directory may already exist (if we // are reopening a DB), when this happens we don't want creating the // directory to cause an error. However, we need to check if creating the // directory fails or else we may get an obscure message about the lock // file not existing. One real-world example of this occurring is if // env->CreateDirIfMissing() doesn't create intermediate directories, e.g. // when dbname_ is "dir/db" but when "dir" doesn't exist. Status s = env->CreateDirIfMissing(dirname); if (!s.ok()) { return s; } return env->NewDirectory(dirname, directory); } Status DBImpl::Directories::SetDirectories( Env* env, const std::string& dbname, const std::string& wal_dir, const std::vector& data_paths) { Status s = CreateAndNewDirectory(env, dbname, &db_dir_); if (!s.ok()) { return s; } if (!wal_dir.empty() && dbname != wal_dir) { s = CreateAndNewDirectory(env, wal_dir, &wal_dir_); if (!s.ok()) { return s; } } data_dirs_.clear(); for (auto& p : data_paths) { const std::string db_path = p.path; if (db_path == dbname) { data_dirs_.emplace_back(nullptr); } else { std::unique_ptr path_directory; s = CreateAndNewDirectory(env, db_path, &path_directory); if (!s.ok()) { return s; } data_dirs_.emplace_back(path_directory.release()); } } assert(data_dirs_.size() == data_paths.size()); return Status::OK(); } Directory* DBImpl::Directories::GetDataDir(size_t path_id) { assert(path_id < data_dirs_.size()); Directory* ret_dir = data_dirs_[path_id].get(); if (ret_dir == nullptr) { // Should use db_dir_ return db_dir_.get(); } return ret_dir; } Status DBImpl::Recover( const std::vector& column_families, bool read_only, bool error_if_log_file_exist) { mutex_.AssertHeld(); bool is_new_db = false; assert(db_lock_ == nullptr); if (!read_only) { Status s = directories_.SetDirectories(env_, dbname_, db_options_.wal_dir, db_options_.db_paths); if (!s.ok()) { return s; } s = env_->LockFile(LockFileName(dbname_), &db_lock_); if (!s.ok()) { return s; } s = env_->FileExists(CurrentFileName(dbname_)); if (s.IsNotFound()) { if (db_options_.create_if_missing) { s = NewDB(); is_new_db = true; if (!s.ok()) { return s; } } else { return Status::InvalidArgument( dbname_, "does not exist (create_if_missing is false)"); } } else if (s.ok()) { if (db_options_.error_if_exists) { return Status::InvalidArgument( dbname_, "exists (error_if_exists is true)"); } } else { // Unexpected error reading file assert(s.IsIOError()); return s; } // Check for the IDENTITY file and create it if not there s = env_->FileExists(IdentityFileName(dbname_)); if (s.IsNotFound()) { s = SetIdentityFile(env_, dbname_); if (!s.ok()) { return s; } } else if (!s.ok()) { assert(s.IsIOError()); return s; } } Status s = versions_->Recover(column_families, read_only); if (db_options_.paranoid_checks && s.ok()) { s = CheckConsistency(); } if (s.ok()) { SequenceNumber max_sequence(kMaxSequenceNumber); default_cf_handle_ = new ColumnFamilyHandleImpl( versions_->GetColumnFamilySet()->GetDefault(), this, &mutex_); default_cf_internal_stats_ = default_cf_handle_->cfd()->internal_stats(); single_column_family_mode_ = versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1; // Recover from all newer log files than the ones named in the // descriptor (new log files may have been added by the previous // incarnation without registering them in the descriptor). // // Note that prev_log_number() is no longer used, but we pay // attention to it in case we are recovering a database // produced by an older version of rocksdb. const uint64_t min_log = versions_->MinLogNumber(); const uint64_t prev_log = versions_->prev_log_number(); std::vector filenames; s = env_->GetChildren(db_options_.wal_dir, &filenames); if (!s.ok()) { return s; } std::vector logs; for (size_t i = 0; i < filenames.size(); i++) { uint64_t number; FileType type; if (ParseFileName(filenames[i], &number, &type) && type == kLogFile) { if (is_new_db) { return Status::Corruption( "While creating a new Db, wal_dir contains " "existing log file: ", filenames[i]); } else if ((number >= min_log) || (number == prev_log)) { logs.push_back(number); } } } if (logs.size() > 0 && error_if_log_file_exist) { return Status::Corruption("" "The db was opened in readonly mode with error_if_log_file_exist" "flag but a log file already exists"); } if (!logs.empty()) { // Recover in the order in which the logs were generated std::sort(logs.begin(), logs.end()); s = RecoverLogFiles(logs, &max_sequence, read_only); if (!s.ok()) { // Clear memtables if recovery failed for (auto cfd : *versions_->GetColumnFamilySet()) { cfd->CreateNewMemtable(*cfd->GetLatestMutableCFOptions(), kMaxSequenceNumber); } } } SetTickerCount(stats_, SEQUENCE_NUMBER, versions_->LastSequence()); } // Initial value max_total_in_memory_state_ = 0; for (auto cfd : *versions_->GetColumnFamilySet()) { auto* mutable_cf_options = cfd->GetLatestMutableCFOptions(); max_total_in_memory_state_ += mutable_cf_options->write_buffer_size * mutable_cf_options->max_write_buffer_number; } return s; } // REQUIRES: log_numbers are sorted in ascending order Status DBImpl::RecoverLogFiles(const std::vector& log_numbers, SequenceNumber* max_sequence, bool read_only) { struct LogReporter : public log::Reader::Reporter { Env* env; Logger* info_log; const char* fname; Status* status; // nullptr if db_options_.paranoid_checks==false virtual void Corruption(size_t bytes, const Status& s) override { Log(InfoLogLevel::WARN_LEVEL, info_log, "%s%s: dropping %d bytes; %s", (this->status == nullptr ? "(ignoring error) " : ""), fname, static_cast(bytes), s.ToString().c_str()); if (this->status != nullptr && this->status->ok()) { *this->status = s; } } }; mutex_.AssertHeld(); Status status; std::unordered_map version_edits; // no need to refcount because iteration is under mutex for (auto cfd : *versions_->GetColumnFamilySet()) { VersionEdit edit; edit.SetColumnFamily(cfd->GetID()); version_edits.insert({cfd->GetID(), edit}); } int job_id = next_job_id_.fetch_add(1); { auto stream = event_logger_.Log(); stream << "job" << job_id << "event" << "recovery_started"; stream << "log_files"; stream.StartArray(); for (auto log_number : log_numbers) { stream << log_number; } stream.EndArray(); } bool continue_replay_log = true; for (auto log_number : log_numbers) { // The previous incarnation may not have written any MANIFEST // records after allocating this log number. So we manually // update the file number allocation counter in VersionSet. versions_->MarkFileNumberUsedDuringRecovery(log_number); // Open the log file std::string fname = LogFileName(db_options_.wal_dir, log_number); unique_ptr file_reader; { unique_ptr file; status = env_->NewSequentialFile(fname, &file, env_options_); if (!status.ok()) { MaybeIgnoreError(&status); if (!status.ok()) { return status; } else { // Fail with one log file, but that's ok. // Try next one. continue; } } file_reader.reset(new SequentialFileReader(std::move(file))); } // Create the log reader. LogReporter reporter; reporter.env = env_; reporter.info_log = db_options_.info_log.get(); reporter.fname = fname.c_str(); if (!db_options_.paranoid_checks || db_options_.wal_recovery_mode == WALRecoveryMode::kSkipAnyCorruptedRecords) { reporter.status = nullptr; } else { reporter.status = &status; } // We intentially make log::Reader do checksumming even if // paranoid_checks==false so that corruptions cause entire commits // to be skipped instead of propagating bad information (like overly // large sequence numbers). log::Reader reader(db_options_.info_log, std::move(file_reader), &reporter, true /*checksum*/, 0 /*initial_offset*/, log_number); Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Recovering log #%" PRIu64 " mode %d skip-recovery %d", log_number, db_options_.wal_recovery_mode, !continue_replay_log); // Determine if we should tolerate incomplete records at the tail end of the // Read all the records and add to a memtable std::string scratch; Slice record; WriteBatch batch; if (!continue_replay_log) { uint64_t bytes; if (env_->GetFileSize(fname, &bytes).ok()) { auto info_log = db_options_.info_log.get(); Log(InfoLogLevel::WARN_LEVEL, info_log, "%s: dropping %d bytes", fname.c_str(), static_cast(bytes)); } } while ( continue_replay_log && reader.ReadRecord(&record, &scratch, db_options_.wal_recovery_mode) && status.ok()) { if (record.size() < 12) { reporter.Corruption(record.size(), Status::Corruption("log record too small")); continue; } WriteBatchInternal::SetContents(&batch, record); #ifndef ROCKSDB_LITE if (db_options_.wal_filter != nullptr) { WriteBatch new_batch; bool batch_changed = false; WalFilter::WalProcessingOption wal_processing_option = db_options_.wal_filter->LogRecord(batch, &new_batch, &batch_changed); switch (wal_processing_option) { case WalFilter::WalProcessingOption::kContinueProcessing: // do nothing, proceeed normally break; case WalFilter::WalProcessingOption::kIgnoreCurrentRecord: // skip current record continue; case WalFilter::WalProcessingOption::kStopReplay: // skip current record and stop replay continue_replay_log = false; continue; case WalFilter::WalProcessingOption::kCorruptedRecord: { status = Status::Corruption("Corruption reported by Wal Filter ", db_options_.wal_filter->Name()); MaybeIgnoreError(&status); if (!status.ok()) { reporter.Corruption(record.size(), status); continue; } break; } default: { assert(false); // unhandled case status = Status::NotSupported( "Unknown WalProcessingOption returned" " by Wal Filter ", db_options_.wal_filter->Name()); MaybeIgnoreError(&status); if (!status.ok()) { return status; } else { // Ignore the error with current record processing. continue; } } } if (batch_changed) { // Make sure that the count in the new batch is // within the orignal count. int new_count = WriteBatchInternal::Count(&new_batch); int original_count = WriteBatchInternal::Count(&batch); if (new_count > original_count) { Log(InfoLogLevel::FATAL_LEVEL, db_options_.info_log, "Recovering log #%" PRIu64 " mode %d log filter %s returned " "more records (%d) than original (%d) which is not allowed. " "Aborting recovery.", log_number, db_options_.wal_recovery_mode, db_options_.wal_filter->Name(), new_count, original_count); status = Status::NotSupported( "More than original # of records " "returned by Wal Filter ", db_options_.wal_filter->Name()); return status; } // Set the same sequence number in the new_batch // as the original batch. WriteBatchInternal::SetSequence(&new_batch, WriteBatchInternal::Sequence(&batch)); batch = new_batch; } } #endif // ROCKSDB_LITE // If column family was not found, it might mean that the WAL write // batch references to the column family that was dropped after the // insert. We don't want to fail the whole write batch in that case -- // we just ignore the update. // That's why we set ignore missing column families to true status = WriteBatchInternal::InsertInto(&batch, column_family_memtables_.get(), &flush_scheduler_, true, log_number); MaybeIgnoreError(&status); if (!status.ok()) { // We are treating this as a failure while reading since we read valid // blocks that do not form coherent data reporter.Corruption(record.size(), status); continue; } const SequenceNumber last_seq = WriteBatchInternal::Sequence(&batch) + WriteBatchInternal::Count(&batch) - 1; if ((*max_sequence == kMaxSequenceNumber) || (last_seq > *max_sequence)) { *max_sequence = last_seq; } if (!read_only) { // we can do this because this is called before client has access to the // DB and there is only a single thread operating on DB ColumnFamilyData* cfd; while ((cfd = flush_scheduler_.TakeNextColumnFamily()) != nullptr) { cfd->Unref(); // If this asserts, it means that InsertInto failed in // filtering updates to already-flushed column families assert(cfd->GetLogNumber() <= log_number); auto iter = version_edits.find(cfd->GetID()); assert(iter != version_edits.end()); VersionEdit* edit = &iter->second; status = WriteLevel0TableForRecovery(job_id, cfd, cfd->mem(), edit); if (!status.ok()) { // Reflect errors immediately so that conditions like full // file-systems cause the DB::Open() to fail. return status; } cfd->CreateNewMemtable(*cfd->GetLatestMutableCFOptions(), *max_sequence); } } } if (!status.ok()) { if (db_options_.wal_recovery_mode == WALRecoveryMode::kSkipAnyCorruptedRecords) { // We should ignore all errors unconditionally status = Status::OK(); } else if (db_options_.wal_recovery_mode == WALRecoveryMode::kPointInTimeRecovery) { // We should ignore the error but not continue replaying status = Status::OK(); continue_replay_log = false; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Point in time recovered to log #%" PRIu64 " seq #%" PRIu64, log_number, *max_sequence); } else { assert(db_options_.wal_recovery_mode == WALRecoveryMode::kTolerateCorruptedTailRecords || db_options_.wal_recovery_mode == WALRecoveryMode::kAbsoluteConsistency); return status; } } flush_scheduler_.Clear(); if ((*max_sequence != kMaxSequenceNumber) && (versions_->LastSequence() < *max_sequence)) { versions_->SetLastSequence(*max_sequence); } } if (!read_only) { // no need to refcount since client still doesn't have access // to the DB and can not drop column families while we iterate auto max_log_number = log_numbers.back(); for (auto cfd : *versions_->GetColumnFamilySet()) { auto iter = version_edits.find(cfd->GetID()); assert(iter != version_edits.end()); VersionEdit* edit = &iter->second; if (cfd->GetLogNumber() > max_log_number) { // Column family cfd has already flushed the data // from all logs. Memtable has to be empty because // we filter the updates based on log_number // (in WriteBatch::InsertInto) assert(cfd->mem()->GetFirstSequenceNumber() == 0); assert(edit->NumEntries() == 0); continue; } // flush the final memtable (if non-empty) if (cfd->mem()->GetFirstSequenceNumber() != 0) { status = WriteLevel0TableForRecovery(job_id, cfd, cfd->mem(), edit); if (!status.ok()) { // Recovery failed break; } cfd->CreateNewMemtable(*cfd->GetLatestMutableCFOptions(), *max_sequence); } // write MANIFEST with update // writing log_number in the manifest means that any log file // with number strongly less than (log_number + 1) is already // recovered and should be ignored on next reincarnation. // Since we already recovered max_log_number, we want all logs // with numbers `<= max_log_number` (includes this one) to be ignored edit->SetLogNumber(max_log_number + 1); // we must mark the next log number as used, even though it's // not actually used. that is because VersionSet assumes // VersionSet::next_file_number_ always to be strictly greater than any // log number versions_->MarkFileNumberUsedDuringRecovery(max_log_number + 1); status = versions_->LogAndApply( cfd, *cfd->GetLatestMutableCFOptions(), edit, &mutex_); if (!status.ok()) { // Recovery failed break; } } } event_logger_.Log() << "job" << job_id << "event" << "recovery_finished"; return status; } Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd, MemTable* mem, VersionEdit* edit) { mutex_.AssertHeld(); const uint64_t start_micros = env_->NowMicros(); FileMetaData meta; meta.fd = FileDescriptor(versions_->NewFileNumber(), 0, 0); auto pending_outputs_inserted_elem = CaptureCurrentFileNumberInPendingOutputs(); ReadOptions ro; ro.total_order_seek = true; Arena arena; Status s; TableProperties table_properties; { ScopedArenaIterator iter(mem->NewIterator(ro, &arena)); Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] [WriteLevel0TableForRecovery]" " Level-0 table #%" PRIu64 ": started", cfd->GetName().c_str(), meta.fd.GetNumber()); bool paranoid_file_checks = cfd->GetLatestMutableCFOptions()->paranoid_file_checks; { mutex_.Unlock(); TableFileCreationInfo info; SequenceNumber earliest_write_conflict_snapshot; std::vector snapshot_seqs = snapshots_.GetAll(&earliest_write_conflict_snapshot); s = BuildTable( dbname_, env_, *cfd->ioptions(), env_options_, cfd->table_cache(), iter.get(), &meta, cfd->internal_comparator(), cfd->int_tbl_prop_collector_factories(), cfd->GetID(), snapshot_seqs, earliest_write_conflict_snapshot, GetCompressionFlush(*cfd->ioptions()), cfd->ioptions()->compression_opts, paranoid_file_checks, cfd->internal_stats(), Env::IO_HIGH, &info.table_properties); LogFlush(db_options_.info_log); Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] [WriteLevel0TableForRecovery]" " Level-0 table #%" PRIu64 ": %" PRIu64 " bytes %s", cfd->GetName().c_str(), meta.fd.GetNumber(), meta.fd.GetFileSize(), s.ToString().c_str()); // output to event logger if (s.ok()) { info.db_name = dbname_; info.cf_name = cfd->GetName(); info.file_path = TableFileName(db_options_.db_paths, meta.fd.GetNumber(), meta.fd.GetPathId()); info.file_size = meta.fd.GetFileSize(); info.job_id = job_id; EventHelpers::LogAndNotifyTableFileCreation( &event_logger_, db_options_.listeners, meta.fd, info); } mutex_.Lock(); } } ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem); // Note that if file_size is zero, the file has been deleted and // should not be added to the manifest. int level = 0; if (s.ok() && meta.fd.GetFileSize() > 0) { edit->AddFile(level, meta.fd.GetNumber(), meta.fd.GetPathId(), meta.fd.GetFileSize(), meta.smallest, meta.largest, meta.smallest_seqno, meta.largest_seqno, meta.marked_for_compaction); } InternalStats::CompactionStats stats(1); stats.micros = env_->NowMicros() - start_micros; stats.bytes_written = meta.fd.GetFileSize(); stats.num_output_files = 1; cfd->internal_stats()->AddCompactionStats(level, stats); cfd->internal_stats()->AddCFStats( InternalStats::BYTES_FLUSHED, meta.fd.GetFileSize()); RecordTick(stats_, COMPACT_WRITE_BYTES, meta.fd.GetFileSize()); return s; } Status DBImpl::FlushMemTableToOutputFile( ColumnFamilyData* cfd, const MutableCFOptions& mutable_cf_options, bool* made_progress, JobContext* job_context, LogBuffer* log_buffer) { mutex_.AssertHeld(); assert(cfd->imm()->NumNotFlushed() != 0); assert(cfd->imm()->IsFlushPending()); SequenceNumber earliest_write_conflict_snapshot; std::vector snapshot_seqs = snapshots_.GetAll(&earliest_write_conflict_snapshot); FlushJob flush_job( dbname_, cfd, db_options_, mutable_cf_options, env_options_, versions_.get(), &mutex_, &shutting_down_, snapshot_seqs, earliest_write_conflict_snapshot, job_context, log_buffer, directories_.GetDbDir(), directories_.GetDataDir(0U), GetCompressionFlush(*cfd->ioptions()), stats_, &event_logger_); FileMetaData file_meta; // Within flush_job.Run, rocksdb may call event listener to notify // file creation and deletion. // // Note that flush_job.Run will unlock and lock the db_mutex, // and EventListener callback will be called when the db_mutex // is unlocked by the current thread. Status s = flush_job.Run(&file_meta); if (s.ok()) { InstallSuperVersionAndScheduleWorkWrapper(cfd, job_context, mutable_cf_options); if (made_progress) { *made_progress = 1; } VersionStorageInfo::LevelSummaryStorage tmp; LogToBuffer(log_buffer, "[%s] Level summary: %s\n", cfd->GetName().c_str(), cfd->current()->storage_info()->LevelSummary(&tmp)); } if (!s.ok() && !s.IsShutdownInProgress() && db_options_.paranoid_checks && bg_error_.ok()) { // if a bad error happened (not ShutdownInProgress) and paranoid_checks is // true, mark DB read-only bg_error_ = s; } RecordFlushIOStats(); #ifndef ROCKSDB_LITE if (s.ok()) { // may temporarily unlock and lock the mutex. NotifyOnFlushCompleted(cfd, &file_meta, mutable_cf_options, job_context->job_id, flush_job.GetTableProperties()); auto sfm = static_cast(db_options_.sst_file_manager.get()); if (sfm) { // Notify sst_file_manager that a new file was added std::string file_path = MakeTableFileName(db_options_.db_paths[0].path, file_meta.fd.GetNumber()); sfm->OnAddFile(file_path); } } #endif // ROCKSDB_LITE return s; } void DBImpl::NotifyOnFlushCompleted(ColumnFamilyData* cfd, FileMetaData* file_meta, const MutableCFOptions& mutable_cf_options, int job_id, TableProperties prop) { #ifndef ROCKSDB_LITE if (db_options_.listeners.size() == 0U) { return; } mutex_.AssertHeld(); if (shutting_down_.load(std::memory_order_acquire)) { return; } bool triggered_writes_slowdown = (cfd->current()->storage_info()->NumLevelFiles(0) >= mutable_cf_options.level0_slowdown_writes_trigger); bool triggered_writes_stop = (cfd->current()->storage_info()->NumLevelFiles(0) >= mutable_cf_options.level0_stop_writes_trigger); // release lock while notifying events mutex_.Unlock(); { FlushJobInfo info; info.cf_name = cfd->GetName(); // TODO(yhchiang): make db_paths dynamic in case flush does not // go to L0 in the future. info.file_path = MakeTableFileName(db_options_.db_paths[0].path, file_meta->fd.GetNumber()); info.thread_id = env_->GetThreadID(); info.job_id = job_id; info.triggered_writes_slowdown = triggered_writes_slowdown; info.triggered_writes_stop = triggered_writes_stop; info.smallest_seqno = file_meta->smallest_seqno; info.largest_seqno = file_meta->largest_seqno; info.table_properties = prop; for (auto listener : db_options_.listeners) { listener->OnFlushCompleted(this, info); } } mutex_.Lock(); // no need to signal bg_cv_ as it will be signaled at the end of the // flush process. #endif // ROCKSDB_LITE } Status DBImpl::CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) { if (options.target_path_id >= db_options_.db_paths.size()) { return Status::InvalidArgument("Invalid target path ID"); } auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); bool exclusive = options.exclusive_manual_compaction; Status s = FlushMemTable(cfd, FlushOptions()); if (!s.ok()) { LogFlush(db_options_.info_log); return s; } int max_level_with_files = 0; { InstrumentedMutexLock l(&mutex_); Version* base = cfd->current(); for (int level = 1; level < base->storage_info()->num_non_empty_levels(); level++) { if (base->storage_info()->OverlapInLevel(level, begin, end)) { max_level_with_files = level; } } } int final_output_level = 0; if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal && cfd->NumberLevels() > 1) { // Always compact all files together. s = RunManualCompaction(cfd, ColumnFamilyData::kCompactAllLevels, cfd->NumberLevels() - 1, options.target_path_id, begin, end, exclusive); final_output_level = cfd->NumberLevels() - 1; } else { for (int level = 0; level <= max_level_with_files; level++) { int output_level; // in case the compaction is universal or if we're compacting the // bottom-most level, the output level will be the same as input one. // level 0 can never be the bottommost level (i.e. if all files are in // level 0, we will compact to level 1) if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal || cfd->ioptions()->compaction_style == kCompactionStyleFIFO) { output_level = level; } else if (level == max_level_with_files && level > 0) { if (options.bottommost_level_compaction == BottommostLevelCompaction::kSkip) { // Skip bottommost level compaction continue; } else if (options.bottommost_level_compaction == BottommostLevelCompaction::kIfHaveCompactionFilter && cfd->ioptions()->compaction_filter == nullptr && cfd->ioptions()->compaction_filter_factory == nullptr) { // Skip bottommost level compaction since we don't have a compaction // filter continue; } output_level = level; } else { output_level = level + 1; if (cfd->ioptions()->compaction_style == kCompactionStyleLevel && cfd->ioptions()->level_compaction_dynamic_level_bytes && level == 0) { output_level = ColumnFamilyData::kCompactToBaseLevel; } } s = RunManualCompaction(cfd, level, output_level, options.target_path_id, begin, end, exclusive); if (!s.ok()) { break; } if (output_level == ColumnFamilyData::kCompactToBaseLevel) { final_output_level = cfd->NumberLevels() - 1; } else if (output_level > final_output_level) { final_output_level = output_level; } TEST_SYNC_POINT("DBImpl::RunManualCompaction()::1"); TEST_SYNC_POINT("DBImpl::RunManualCompaction()::2"); } } if (!s.ok()) { LogFlush(db_options_.info_log); return s; } if (options.change_level) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[RefitLevel] waiting for background threads to stop"); s = PauseBackgroundWork(); if (s.ok()) { s = ReFitLevel(cfd, final_output_level, options.target_level); } ContinueBackgroundWork(); } LogFlush(db_options_.info_log); { InstrumentedMutexLock l(&mutex_); // an automatic compaction that has been scheduled might have been // preempted by the manual compactions. Need to schedule it back. MaybeScheduleFlushOrCompaction(); } return s; } Status DBImpl::CompactFiles( const CompactionOptions& compact_options, ColumnFamilyHandle* column_family, const std::vector& input_file_names, const int output_level, const int output_path_id) { #ifdef ROCKSDB_LITE // not supported in lite version return Status::NotSupported("Not supported in ROCKSDB LITE"); #else if (column_family == nullptr) { return Status::InvalidArgument("ColumnFamilyHandle must be non-null."); } auto cfd = reinterpret_cast(column_family)->cfd(); assert(cfd); Status s; JobContext job_context(0, true); LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, db_options_.info_log.get()); // Perform CompactFiles SuperVersion* sv = GetAndRefSuperVersion(cfd); { InstrumentedMutexLock l(&mutex_); s = CompactFilesImpl(compact_options, cfd, sv->current, input_file_names, output_level, output_path_id, &job_context, &log_buffer); } ReturnAndCleanupSuperVersion(cfd, sv); // Find and delete obsolete files { InstrumentedMutexLock l(&mutex_); // If !s.ok(), this means that Compaction failed. In that case, we want // to delete all obsolete files we might have created and we force // FindObsoleteFiles(). This is because job_context does not // catch all created files if compaction failed. FindObsoleteFiles(&job_context, !s.ok()); } // release the mutex // delete unnecessary files if any, this is done outside the mutex if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) { // Have to flush the info logs before bg_compaction_scheduled_-- // because if bg_flush_scheduled_ becomes 0 and the lock is // released, the deconstructor of DB can kick in and destroy all the // states of DB so info_log might not be available after that point. // It also applies to access other states that DB owns. log_buffer.FlushBufferToLog(); if (job_context.HaveSomethingToDelete()) { // no mutex is locked here. No need to Unlock() and Lock() here. PurgeObsoleteFiles(job_context); } job_context.Clean(); } return s; #endif // ROCKSDB_LITE } #ifndef ROCKSDB_LITE Status DBImpl::CompactFilesImpl( const CompactionOptions& compact_options, ColumnFamilyData* cfd, Version* version, const std::vector& input_file_names, const int output_level, int output_path_id, JobContext* job_context, LogBuffer* log_buffer) { mutex_.AssertHeld(); if (shutting_down_.load(std::memory_order_acquire)) { return Status::ShutdownInProgress(); } std::unordered_set input_set; for (auto file_name : input_file_names) { input_set.insert(TableFileNameToNumber(file_name)); } ColumnFamilyMetaData cf_meta; // TODO(yhchiang): can directly use version here if none of the // following functions call is pluggable to external developers. version->GetColumnFamilyMetaData(&cf_meta); if (output_path_id < 0) { if (db_options_.db_paths.size() == 1U) { output_path_id = 0; } else { return Status::NotSupported( "Automatic output path selection is not " "yet supported in CompactFiles()"); } } Status s = cfd->compaction_picker()->SanitizeCompactionInputFiles( &input_set, cf_meta, output_level); if (!s.ok()) { return s; } std::vector input_files; s = cfd->compaction_picker()->GetCompactionInputsFromFileNumbers( &input_files, &input_set, version->storage_info(), compact_options); if (!s.ok()) { return s; } for (auto inputs : input_files) { if (cfd->compaction_picker()->FilesInCompaction(inputs.files)) { return Status::Aborted( "Some of the necessary compaction input " "files are already being compacted"); } } // At this point, CompactFiles will be run. bg_compaction_scheduled_++; unique_ptr c; assert(cfd->compaction_picker()); c.reset(cfd->compaction_picker()->FormCompaction( compact_options, input_files, output_level, version->storage_info(), *cfd->GetLatestMutableCFOptions(), output_path_id)); if (!c) { return Status::Aborted("Another Level 0 compaction is running"); } c->SetInputVersion(version); // deletion compaction currently not allowed in CompactFiles. assert(!c->deletion_compaction()); SequenceNumber earliest_write_conflict_snapshot; std::vector snapshot_seqs = snapshots_.GetAll(&earliest_write_conflict_snapshot); auto pending_outputs_inserted_elem = CaptureCurrentFileNumberInPendingOutputs(); assert(is_snapshot_supported_ || snapshots_.empty()); CompactionJob compaction_job( job_context->job_id, c.get(), db_options_, env_options_, versions_.get(), &shutting_down_, log_buffer, directories_.GetDbDir(), directories_.GetDataDir(c->output_path_id()), stats_, snapshot_seqs, earliest_write_conflict_snapshot, table_cache_, &event_logger_, c->mutable_cf_options()->paranoid_file_checks, c->mutable_cf_options()->compaction_measure_io_stats, dbname_, nullptr); // Here we pass a nullptr for CompactionJobStats because // CompactFiles does not trigger OnCompactionCompleted(), // which is the only place where CompactionJobStats is // returned. The idea of not triggering OnCompationCompleted() // is that CompactFiles runs in the caller thread, so the user // should always know when it completes. As a result, it makes // less sense to notify the users something they should already // know. // // In the future, if we would like to add CompactionJobStats // support for CompactFiles, we should have CompactFiles API // pass a pointer of CompactionJobStats as the out-value // instead of using EventListener. compaction_job.Prepare(); mutex_.Unlock(); TEST_SYNC_POINT("CompactFilesImpl:0"); TEST_SYNC_POINT("CompactFilesImpl:1"); compaction_job.Run(); TEST_SYNC_POINT("CompactFilesImpl:2"); TEST_SYNC_POINT("CompactFilesImpl:3"); mutex_.Lock(); Status status = compaction_job.Install(*c->mutable_cf_options(), &mutex_); if (status.ok()) { InstallSuperVersionAndScheduleWorkWrapper( c->column_family_data(), job_context, *c->mutable_cf_options()); } c->ReleaseCompactionFiles(s); ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem); if (status.ok()) { // Done } else if (status.IsShutdownInProgress()) { // Ignore compaction errors found during shutting down } else { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "[%s] [JOB %d] Compaction error: %s", c->column_family_data()->GetName().c_str(), job_context->job_id, status.ToString().c_str()); if (db_options_.paranoid_checks && bg_error_.ok()) { bg_error_ = status; } } c.reset(); bg_compaction_scheduled_--; if (bg_compaction_scheduled_ == 0) { bg_cv_.SignalAll(); } return status; } #endif // ROCKSDB_LITE Status DBImpl::PauseBackgroundWork() { InstrumentedMutexLock guard_lock(&mutex_); bg_work_paused_++; while (bg_compaction_scheduled_ > 0 || bg_flush_scheduled_ > 0) { bg_cv_.Wait(); } return Status::OK(); } Status DBImpl::ContinueBackgroundWork() { InstrumentedMutexLock guard_lock(&mutex_); if (bg_work_paused_ == 0) { return Status::InvalidArgument(); } assert(bg_work_paused_ > 0); bg_work_paused_--; if (bg_work_paused_ == 0) { MaybeScheduleFlushOrCompaction(); } return Status::OK(); } void DBImpl::NotifyOnCompactionCompleted( ColumnFamilyData* cfd, Compaction *c, const Status &st, const CompactionJobStats& compaction_job_stats, const int job_id) { #ifndef ROCKSDB_LITE if (db_options_.listeners.size() == 0U) { return; } mutex_.AssertHeld(); if (shutting_down_.load(std::memory_order_acquire)) { return; } // release lock while notifying events mutex_.Unlock(); { CompactionJobInfo info; info.cf_name = cfd->GetName(); info.status = st; info.thread_id = env_->GetThreadID(); info.job_id = job_id; info.base_input_level = c->start_level(); info.output_level = c->output_level(); info.stats = compaction_job_stats; info.table_properties = c->GetOutputTableProperties(); info.compaction_reason = c->compaction_reason(); for (size_t i = 0; i < c->num_input_levels(); ++i) { for (const auto fmd : *c->inputs(i)) { auto fn = TableFileName(db_options_.db_paths, fmd->fd.GetNumber(), fmd->fd.GetPathId()); info.input_files.push_back(fn); if (info.table_properties.count(fn) == 0) { std::shared_ptr tp; auto s = cfd->current()->GetTableProperties(&tp, fmd, &fn); if (s.ok()) { info.table_properties[fn] = tp; } } } } for (const auto newf : c->edit()->GetNewFiles()) { info.output_files.push_back( TableFileName(db_options_.db_paths, newf.second.fd.GetNumber(), newf.second.fd.GetPathId())); } for (auto listener : db_options_.listeners) { listener->OnCompactionCompleted(this, info); } } mutex_.Lock(); // no need to signal bg_cv_ as it will be signaled at the end of the // flush process. #endif // ROCKSDB_LITE } Status DBImpl::SetOptions(ColumnFamilyHandle* column_family, const std::unordered_map& options_map) { #ifdef ROCKSDB_LITE return Status::NotSupported("Not supported in ROCKSDB LITE"); #else auto* cfd = reinterpret_cast(column_family)->cfd(); if (options_map.empty()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "SetOptions() on column family [%s], empty input", cfd->GetName().c_str()); return Status::InvalidArgument("empty input"); } MutableCFOptions new_options; Status s; Status persist_options_status; { InstrumentedMutexLock l(&mutex_); s = cfd->SetOptions(options_map); if (s.ok()) { new_options = *cfd->GetLatestMutableCFOptions(); } if (s.ok()) { // Persist RocksDB options under the single write thread WriteThread::Writer w; write_thread_.EnterUnbatched(&w, &mutex_); persist_options_status = WriteOptionsFile(); write_thread_.ExitUnbatched(&w); } } Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "SetOptions() on column family [%s], inputs:", cfd->GetName().c_str()); for (const auto& o : options_map) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "%s: %s\n", o.first.c_str(), o.second.c_str()); } if (s.ok()) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] SetOptions succeeded", cfd->GetName().c_str()); new_options.Dump(db_options_.info_log.get()); if (!persist_options_status.ok()) { if (db_options_.fail_if_options_file_error) { s = Status::IOError( "SetOptions succeeded, but unable to persist options", persist_options_status.ToString()); } Warn(db_options_.info_log, "Unable to persist options in SetOptions() -- %s", persist_options_status.ToString().c_str()); } } else { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "[%s] SetOptions failed", cfd->GetName().c_str()); } LogFlush(db_options_.info_log); return s; #endif // ROCKSDB_LITE } // return the same level if it cannot be moved int DBImpl::FindMinimumEmptyLevelFitting(ColumnFamilyData* cfd, const MutableCFOptions& mutable_cf_options, int level) { mutex_.AssertHeld(); const auto* vstorage = cfd->current()->storage_info(); int minimum_level = level; for (int i = level - 1; i > 0; --i) { // stop if level i is not empty if (vstorage->NumLevelFiles(i) > 0) break; // stop if level i is too small (cannot fit the level files) if (vstorage->MaxBytesForLevel(i) < vstorage->NumLevelBytes(level)) { break; } minimum_level = i; } return minimum_level; } // REQUIREMENT: block all background work by calling PauseBackgroundWork() // before calling this function Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) { assert(level < cfd->NumberLevels()); if (target_level >= cfd->NumberLevels()) { return Status::InvalidArgument("Target level exceeds number of levels"); } std::unique_ptr superversion_to_free; std::unique_ptr new_superversion(new SuperVersion()); Status status; InstrumentedMutexLock guard_lock(&mutex_); // only allow one thread refitting if (refitting_level_) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[ReFitLevel] another thread is refitting"); return Status::NotSupported("another thread is refitting"); } refitting_level_ = true; const MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions(); // move to a smaller level int to_level = target_level; if (target_level < 0) { to_level = FindMinimumEmptyLevelFitting(cfd, mutable_cf_options, level); } auto* vstorage = cfd->current()->storage_info(); if (to_level > level) { if (level == 0) { return Status::NotSupported( "Cannot change from level 0 to other levels."); } // Check levels are empty for a trivial move for (int l = level + 1; l <= to_level; l++) { if (vstorage->NumLevelFiles(l) > 0) { return Status::NotSupported( "Levels between source and target are not empty for a move."); } } } if (to_level != level) { Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] Before refitting:\n%s", cfd->GetName().c_str(), cfd->current()->DebugString().data()); VersionEdit edit; edit.SetColumnFamily(cfd->GetID()); for (const auto& f : vstorage->LevelFiles(level)) { edit.DeleteFile(level, f->fd.GetNumber()); edit.AddFile(to_level, f->fd.GetNumber(), f->fd.GetPathId(), f->fd.GetFileSize(), f->smallest, f->largest, f->smallest_seqno, f->largest_seqno, f->marked_for_compaction); } Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] Apply version edit:\n%s", cfd->GetName().c_str(), edit.DebugString().data()); status = versions_->LogAndApply(cfd, mutable_cf_options, &edit, &mutex_, directories_.GetDbDir()); superversion_to_free.reset(InstallSuperVersionAndScheduleWork( cfd, new_superversion.release(), mutable_cf_options)); Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] LogAndApply: %s\n", cfd->GetName().c_str(), status.ToString().data()); if (status.ok()) { Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] After refitting:\n%s", cfd->GetName().c_str(), cfd->current()->DebugString().data()); } } refitting_level_ = false; return status; } int DBImpl::NumberLevels(ColumnFamilyHandle* column_family) { auto cfh = reinterpret_cast(column_family); return cfh->cfd()->NumberLevels(); } int DBImpl::MaxMemCompactionLevel(ColumnFamilyHandle* column_family) { return 0; } int DBImpl::Level0StopWriteTrigger(ColumnFamilyHandle* column_family) { auto cfh = reinterpret_cast(column_family); InstrumentedMutexLock l(&mutex_); return cfh->cfd()->GetSuperVersion()-> mutable_cf_options.level0_stop_writes_trigger; } Status DBImpl::Flush(const FlushOptions& flush_options, ColumnFamilyHandle* column_family) { auto cfh = reinterpret_cast(column_family); return FlushMemTable(cfh->cfd(), flush_options); } Status DBImpl::SyncWAL() { autovector logs_to_sync; bool need_log_dir_sync; uint64_t current_log_number; { InstrumentedMutexLock l(&mutex_); assert(!logs_.empty()); // This SyncWAL() call only cares about logs up to this number. current_log_number = logfile_number_; while (logs_.front().number <= current_log_number && logs_.front().getting_synced) { log_sync_cv_.Wait(); } // First check that logs are safe to sync in background. for (auto it = logs_.begin(); it != logs_.end() && it->number <= current_log_number; ++it) { if (!it->writer->file()->writable_file()->IsSyncThreadSafe()) { return Status::NotSupported( "SyncWAL() is not supported for this implementation of WAL file", db_options_.allow_mmap_writes ? "try setting Options::allow_mmap_writes to false" : Slice()); } } for (auto it = logs_.begin(); it != logs_.end() && it->number <= current_log_number; ++it) { auto& log = *it; assert(!log.getting_synced); log.getting_synced = true; logs_to_sync.push_back(log.writer); } need_log_dir_sync = !log_dir_synced_; } RecordTick(stats_, WAL_FILE_SYNCED); Status status; for (log::Writer* log : logs_to_sync) { status = log->file()->SyncWithoutFlush(db_options_.use_fsync); if (!status.ok()) { break; } } if (status.ok() && need_log_dir_sync) { status = directories_.GetWalDir()->Fsync(); } { InstrumentedMutexLock l(&mutex_); MarkLogsSynced(current_log_number, need_log_dir_sync, status); } return status; } void DBImpl::MarkLogsSynced( uint64_t up_to, bool synced_dir, const Status& status) { mutex_.AssertHeld(); if (synced_dir && logfile_number_ == up_to && status.ok()) { log_dir_synced_ = true; } for (auto it = logs_.begin(); it != logs_.end() && it->number <= up_to;) { auto& log = *it; assert(log.getting_synced); if (status.ok() && logs_.size() > 1) { logs_to_free_.push_back(log.ReleaseWriter()); it = logs_.erase(it); } else { log.getting_synced = false; ++it; } } assert(logs_.empty() || (logs_.size() == 1 && !logs_[0].getting_synced)); log_sync_cv_.SignalAll(); } SequenceNumber DBImpl::GetLatestSequenceNumber() const { return versions_->LastSequence(); } Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level, int output_level, uint32_t output_path_id, const Slice* begin, const Slice* end, bool exclusive, bool disallow_trivial_move) { assert(input_level == ColumnFamilyData::kCompactAllLevels || input_level >= 0); InternalKey begin_storage, end_storage; CompactionArg* ca; bool scheduled = false; bool manual_conflict = false; ManualCompaction manual; manual.cfd = cfd; manual.input_level = input_level; manual.output_level = output_level; manual.output_path_id = output_path_id; manual.done = false; manual.in_progress = false; manual.incomplete = false; manual.exclusive = exclusive; manual.disallow_trivial_move = disallow_trivial_move; // For universal compaction, we enforce every manual compaction to compact // all files. if (begin == nullptr || cfd->ioptions()->compaction_style == kCompactionStyleUniversal || cfd->ioptions()->compaction_style == kCompactionStyleFIFO) { manual.begin = nullptr; } else { begin_storage.SetMaxPossibleForUserKey(*begin); manual.begin = &begin_storage; } if (end == nullptr || cfd->ioptions()->compaction_style == kCompactionStyleUniversal || cfd->ioptions()->compaction_style == kCompactionStyleFIFO) { manual.end = nullptr; } else { end_storage.SetMinPossibleForUserKey(*end); manual.end = &end_storage; } InstrumentedMutexLock l(&mutex_); // When a manual compaction arrives, temporarily disable scheduling of // non-manual compactions and wait until the number of scheduled compaction // jobs drops to zero. This is needed to ensure that this manual compaction // can compact any range of keys/files. // // HasPendingManualCompaction() is true when at least one thread is inside // RunManualCompaction(), i.e. during that time no other compaction will // get scheduled (see MaybeScheduleFlushOrCompaction). // // Note that the following loop doesn't stop more that one thread calling // RunManualCompaction() from getting to the second while loop below. // However, only one of them will actually schedule compaction, while // others will wait on a condition variable until it completes. AddManualCompaction(&manual); TEST_SYNC_POINT_CALLBACK("DBImpl::RunManualCompaction:NotScheduled", &mutex_); if (exclusive) { while (bg_compaction_scheduled_ > 0) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] Manual compaction waiting for all other scheduled background " "compactions to finish", cfd->GetName().c_str()); bg_cv_.Wait(); } } Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] Manual compaction starting", cfd->GetName().c_str()); // We don't check bg_error_ here, because if we get the error in compaction, // the compaction will set manual.status to bg_error_ and set manual.done to // true. while (!manual.done) { assert(HasPendingManualCompaction()); manual_conflict = false; if (ShouldntRunManualCompaction(&manual) || (manual.in_progress == true) || scheduled || ((manual.manual_end = &manual.tmp_storage1)&&( (manual.compaction = manual.cfd->CompactRange( *manual.cfd->GetLatestMutableCFOptions(), manual.input_level, manual.output_level, manual.output_path_id, manual.begin, manual.end, &manual.manual_end, &manual_conflict)) == nullptr) && manual_conflict)) { // exclusive manual compactions should not see a conflict during // CompactRange assert(!exclusive || !manual_conflict); // Running either this or some other manual compaction bg_cv_.Wait(); if (scheduled && manual.incomplete == true) { assert(!manual.in_progress); scheduled = false; manual.incomplete = false; } } else if (!scheduled) { if (manual.compaction == nullptr) { manual.done = true; bg_cv_.SignalAll(); continue; } ca = new CompactionArg; ca->db = this; ca->m = &manual; manual.incomplete = false; bg_compaction_scheduled_++; env_->Schedule(&DBImpl::BGWorkCompaction, ca, Env::Priority::LOW, this, &DBImpl::UnscheduleCallback); scheduled = true; } } assert(!manual.in_progress); assert(HasPendingManualCompaction()); RemoveManualCompaction(&manual); bg_cv_.SignalAll(); return manual.status; } InternalIterator* DBImpl::NewInternalIterator( Arena* arena, ColumnFamilyHandle* column_family) { ColumnFamilyData* cfd; if (column_family == nullptr) { cfd = default_cf_handle_->cfd(); } else { auto cfh = reinterpret_cast(column_family); cfd = cfh->cfd(); } mutex_.Lock(); SuperVersion* super_version = cfd->GetSuperVersion()->Ref(); mutex_.Unlock(); ReadOptions roptions; return NewInternalIterator(roptions, cfd, super_version, arena); } Status DBImpl::FlushMemTable(ColumnFamilyData* cfd, const FlushOptions& flush_options) { Status s; { WriteContext context; InstrumentedMutexLock guard_lock(&mutex_); if (cfd->imm()->NumNotFlushed() == 0 && cfd->mem()->IsEmpty()) { // Nothing to flush return Status::OK(); } WriteThread::Writer w; write_thread_.EnterUnbatched(&w, &mutex_); // SwitchMemtable() will release and reacquire mutex // during execution s = SwitchMemtable(cfd, &context); write_thread_.ExitUnbatched(&w); cfd->imm()->FlushRequested(); // schedule flush SchedulePendingFlush(cfd); MaybeScheduleFlushOrCompaction(); } if (s.ok() && flush_options.wait) { // Wait until the compaction completes s = WaitForFlushMemTable(cfd); } return s; } Status DBImpl::WaitForFlushMemTable(ColumnFamilyData* cfd) { Status s; // Wait until the compaction completes InstrumentedMutexLock l(&mutex_); while (cfd->imm()->NumNotFlushed() > 0 && bg_error_.ok()) { if (shutting_down_.load(std::memory_order_acquire)) { return Status::ShutdownInProgress(); } bg_cv_.Wait(); } if (!bg_error_.ok()) { s = bg_error_; } return s; } Status DBImpl::EnableAutoCompaction( const std::vector& column_family_handles) { Status s; for (auto cf_ptr : column_family_handles) { Status status = this->SetOptions(cf_ptr, {{"disable_auto_compactions", "false"}}); if (status.ok()) { ColumnFamilyData* cfd = reinterpret_cast(cf_ptr)->cfd(); InstrumentedMutexLock guard_lock(&mutex_); delete this->InstallSuperVersionAndScheduleWork( cfd, nullptr, *cfd->GetLatestMutableCFOptions()); } else { s = status; } } return s; } void DBImpl::MaybeScheduleFlushOrCompaction() { mutex_.AssertHeld(); if (!opened_successfully_) { // Compaction may introduce data race to DB open return; } if (bg_work_paused_ > 0) { // we paused the background work return; } else if (shutting_down_.load(std::memory_order_acquire)) { // DB is being deleted; no more background compactions return; } while (unscheduled_flushes_ > 0 && bg_flush_scheduled_ < db_options_.max_background_flushes) { unscheduled_flushes_--; bg_flush_scheduled_++; env_->Schedule(&DBImpl::BGWorkFlush, this, Env::Priority::HIGH, this); } auto bg_compactions_allowed = BGCompactionsAllowed(); // special case -- if max_background_flushes == 0, then schedule flush on a // compaction thread if (db_options_.max_background_flushes == 0) { while (unscheduled_flushes_ > 0 && bg_flush_scheduled_ + bg_compaction_scheduled_ < bg_compactions_allowed) { unscheduled_flushes_--; bg_flush_scheduled_++; env_->Schedule(&DBImpl::BGWorkFlush, this, Env::Priority::LOW, this); } } if (HasExclusiveManualCompaction()) { // only manual compactions are allowed to run. don't schedule automatic // compactions return; } while (bg_compaction_scheduled_ < bg_compactions_allowed && unscheduled_compactions_ > 0) { CompactionArg* ca = new CompactionArg; ca->db = this; ca->m = nullptr; bg_compaction_scheduled_++; unscheduled_compactions_--; env_->Schedule(&DBImpl::BGWorkCompaction, ca, Env::Priority::LOW, this, &DBImpl::UnscheduleCallback); } } int DBImpl::BGCompactionsAllowed() const { if (write_controller_.NeedSpeedupCompaction()) { return db_options_.max_background_compactions; } else { return db_options_.base_background_compactions; } } void DBImpl::AddToCompactionQueue(ColumnFamilyData* cfd) { assert(!cfd->pending_compaction()); cfd->Ref(); compaction_queue_.push_back(cfd); cfd->set_pending_compaction(true); } ColumnFamilyData* DBImpl::PopFirstFromCompactionQueue() { assert(!compaction_queue_.empty()); auto cfd = *compaction_queue_.begin(); compaction_queue_.pop_front(); assert(cfd->pending_compaction()); cfd->set_pending_compaction(false); return cfd; } void DBImpl::AddToFlushQueue(ColumnFamilyData* cfd) { assert(!cfd->pending_flush()); cfd->Ref(); flush_queue_.push_back(cfd); cfd->set_pending_flush(true); } ColumnFamilyData* DBImpl::PopFirstFromFlushQueue() { assert(!flush_queue_.empty()); auto cfd = *flush_queue_.begin(); flush_queue_.pop_front(); assert(cfd->pending_flush()); cfd->set_pending_flush(false); return cfd; } void DBImpl::SchedulePendingFlush(ColumnFamilyData* cfd) { if (!cfd->pending_flush() && cfd->imm()->IsFlushPending()) { AddToFlushQueue(cfd); ++unscheduled_flushes_; } } void DBImpl::SchedulePendingCompaction(ColumnFamilyData* cfd) { if (!cfd->pending_compaction() && cfd->NeedsCompaction()) { AddToCompactionQueue(cfd); ++unscheduled_compactions_; } } void DBImpl::RecordFlushIOStats() { RecordTick(stats_, FLUSH_WRITE_BYTES, IOSTATS(bytes_written)); IOSTATS_RESET(bytes_written); } void DBImpl::BGWorkFlush(void* db) { IOSTATS_SET_THREAD_POOL_ID(Env::Priority::HIGH); TEST_SYNC_POINT("DBImpl::BGWorkFlush"); reinterpret_cast(db)->BackgroundCallFlush(); TEST_SYNC_POINT("DBImpl::BGWorkFlush:done"); } void DBImpl::BGWorkCompaction(void* arg) { CompactionArg ca = *(reinterpret_cast(arg)); delete reinterpret_cast(arg); IOSTATS_SET_THREAD_POOL_ID(Env::Priority::LOW); TEST_SYNC_POINT("DBImpl::BGWorkCompaction"); reinterpret_cast(ca.db)->BackgroundCallCompaction(ca.m); } void DBImpl::UnscheduleCallback(void* arg) { CompactionArg ca = *(reinterpret_cast(arg)); delete reinterpret_cast(arg); if ((ca.m != nullptr) && (ca.m->compaction != nullptr)) { delete ca.m->compaction; } TEST_SYNC_POINT("DBImpl::UnscheduleCallback"); } Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context, LogBuffer* log_buffer) { mutex_.AssertHeld(); Status status = bg_error_; if (status.ok() && shutting_down_.load(std::memory_order_acquire)) { status = Status::ShutdownInProgress(); } if (!status.ok()) { return status; } ColumnFamilyData* cfd = nullptr; while (!flush_queue_.empty()) { // This cfd is already referenced auto first_cfd = PopFirstFromFlushQueue(); if (first_cfd->IsDropped() || !first_cfd->imm()->IsFlushPending()) { // can't flush this CF, try next one if (first_cfd->Unref()) { delete first_cfd; } continue; } // found a flush! cfd = first_cfd; break; } if (cfd != nullptr) { const MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions(); LogToBuffer( log_buffer, "Calling FlushMemTableToOutputFile with column " "family [%s], flush slots available %d, compaction slots allowed %d, " "compaction slots scheduled %d", cfd->GetName().c_str(), db_options_.max_background_flushes, bg_flush_scheduled_, BGCompactionsAllowed() - bg_compaction_scheduled_); status = FlushMemTableToOutputFile(cfd, mutable_cf_options, made_progress, job_context, log_buffer); if (cfd->Unref()) { delete cfd; } } return status; } void DBImpl::BackgroundCallFlush() { bool made_progress = false; JobContext job_context(next_job_id_.fetch_add(1), true); assert(bg_flush_scheduled_); LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, db_options_.info_log.get()); { InstrumentedMutexLock l(&mutex_); num_running_flushes_++; auto pending_outputs_inserted_elem = CaptureCurrentFileNumberInPendingOutputs(); Status s = BackgroundFlush(&made_progress, &job_context, &log_buffer); if (!s.ok() && !s.IsShutdownInProgress()) { // Wait a little bit before retrying background flush in // case this is an environmental problem and we do not want to // chew up resources for failed flushes for the duration of // the problem. uint64_t error_cnt = default_cf_internal_stats_->BumpAndGetBackgroundErrorCount(); bg_cv_.SignalAll(); // In case a waiter can proceed despite the error mutex_.Unlock(); Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Waiting after background flush error: %s" "Accumulated background error counts: %" PRIu64, s.ToString().c_str(), error_cnt); log_buffer.FlushBufferToLog(); LogFlush(db_options_.info_log); env_->SleepForMicroseconds(1000000); mutex_.Lock(); } ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem); // If flush failed, we want to delete all temporary files that we might have // created. Thus, we force full scan in FindObsoleteFiles() FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress()); // delete unnecessary files if any, this is done outside the mutex if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) { mutex_.Unlock(); // Have to flush the info logs before bg_flush_scheduled_-- // because if bg_flush_scheduled_ becomes 0 and the lock is // released, the deconstructor of DB can kick in and destroy all the // states of DB so info_log might not be available after that point. // It also applies to access other states that DB owns. log_buffer.FlushBufferToLog(); if (job_context.HaveSomethingToDelete()) { PurgeObsoleteFiles(job_context); } job_context.Clean(); mutex_.Lock(); } assert(num_running_flushes_ > 0); num_running_flushes_--; bg_flush_scheduled_--; // See if there's more work to be done MaybeScheduleFlushOrCompaction(); RecordFlushIOStats(); bg_cv_.SignalAll(); // IMPORTANT: there should be no code after calling SignalAll. This call may // signal the DB destructor that it's OK to proceed with destruction. In // that case, all DB variables will be dealloacated and referencing them // will cause trouble. } } void DBImpl::BackgroundCallCompaction(void* arg) { bool made_progress = false; ManualCompaction* m = reinterpret_cast(arg); JobContext job_context(next_job_id_.fetch_add(1), true); TEST_SYNC_POINT("BackgroundCallCompaction:0"); MaybeDumpStats(); LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, db_options_.info_log.get()); { InstrumentedMutexLock l(&mutex_); num_running_compactions_++; auto pending_outputs_inserted_elem = CaptureCurrentFileNumberInPendingOutputs(); assert(bg_compaction_scheduled_); Status s = BackgroundCompaction(&made_progress, &job_context, &log_buffer, m); TEST_SYNC_POINT("BackgroundCallCompaction:1"); if (!s.ok() && !s.IsShutdownInProgress()) { // Wait a little bit before retrying background compaction in // case this is an environmental problem and we do not want to // chew up resources for failed compactions for the duration of // the problem. uint64_t error_cnt = default_cf_internal_stats_->BumpAndGetBackgroundErrorCount(); bg_cv_.SignalAll(); // In case a waiter can proceed despite the error mutex_.Unlock(); log_buffer.FlushBufferToLog(); Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Waiting after background compaction error: %s, " "Accumulated background error counts: %" PRIu64, s.ToString().c_str(), error_cnt); LogFlush(db_options_.info_log); env_->SleepForMicroseconds(1000000); mutex_.Lock(); } ReleaseFileNumberFromPendingOutputs(pending_outputs_inserted_elem); // If compaction failed, we want to delete all temporary files that we might // have created (they might not be all recorded in job_context in case of a // failure). Thus, we force full scan in FindObsoleteFiles() FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress()); // delete unnecessary files if any, this is done outside the mutex if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) { mutex_.Unlock(); // Have to flush the info logs before bg_compaction_scheduled_-- // because if bg_flush_scheduled_ becomes 0 and the lock is // released, the deconstructor of DB can kick in and destroy all the // states of DB so info_log might not be available after that point. // It also applies to access other states that DB owns. log_buffer.FlushBufferToLog(); if (job_context.HaveSomethingToDelete()) { PurgeObsoleteFiles(job_context); } job_context.Clean(); mutex_.Lock(); } assert(num_running_compactions_ > 0); num_running_compactions_--; bg_compaction_scheduled_--; versions_->GetColumnFamilySet()->FreeDeadColumnFamilies(); // See if there's more work to be done MaybeScheduleFlushOrCompaction(); if (made_progress || bg_compaction_scheduled_ == 0 || HasPendingManualCompaction()) { // signal if // * made_progress -- need to wakeup DelayWrite // * bg_compaction_scheduled_ == 0 -- need to wakeup ~DBImpl // * HasPendingManualCompaction -- need to wakeup RunManualCompaction // If none of this is true, there is no need to signal since nobody is // waiting for it bg_cv_.SignalAll(); } // IMPORTANT: there should be no code after calling SignalAll. This call may // signal the DB destructor that it's OK to proceed with destruction. In // that case, all DB variables will be dealloacated and referencing them // will cause trouble. } } Status DBImpl::BackgroundCompaction(bool* made_progress, JobContext* job_context, LogBuffer* log_buffer, void* arg) { ManualCompaction* manual_compaction = reinterpret_cast(arg); *made_progress = false; mutex_.AssertHeld(); bool is_manual = (manual_compaction != nullptr); // (manual_compaction->in_progress == false); bool trivial_move_disallowed = is_manual && manual_compaction->disallow_trivial_move; CompactionJobStats compaction_job_stats; Status status = bg_error_; if (status.ok() && shutting_down_.load(std::memory_order_acquire)) { status = Status::ShutdownInProgress(); } if (!status.ok()) { if (is_manual) { manual_compaction->status = status; manual_compaction->done = true; manual_compaction->in_progress = false; delete manual_compaction->compaction; manual_compaction = nullptr; } return status; } if (is_manual) { // another thread cannot pick up the same work manual_compaction->in_progress = true; } unique_ptr c; // InternalKey manual_end_storage; // InternalKey* manual_end = &manual_end_storage; if (is_manual) { ManualCompaction* m = manual_compaction; assert(m->in_progress); c.reset(std::move(m->compaction)); if (!c) { m->done = true; m->manual_end = nullptr; LogToBuffer(log_buffer, "[%s] Manual compaction from level-%d from %s .. " "%s; nothing to do\n", m->cfd->GetName().c_str(), m->input_level, (m->begin ? m->begin->DebugString().c_str() : "(begin)"), (m->end ? m->end->DebugString().c_str() : "(end)")); } else { LogToBuffer(log_buffer, "[%s] Manual compaction from level-%d to level-%d from %s .. " "%s; will stop at %s\n", m->cfd->GetName().c_str(), m->input_level, c->output_level(), (m->begin ? m->begin->DebugString().c_str() : "(begin)"), (m->end ? m->end->DebugString().c_str() : "(end)"), ((m->done || m->manual_end == nullptr) ? "(end)" : m->manual_end->DebugString().c_str())); } } else if (!compaction_queue_.empty()) { // cfd is referenced here auto cfd = PopFirstFromCompactionQueue(); // We unreference here because the following code will take a Ref() on // this cfd if it is going to use it (Compaction class holds a // reference). // This will all happen under a mutex so we don't have to be afraid of // somebody else deleting it. if (cfd->Unref()) { delete cfd; // This was the last reference of the column family, so no need to // compact. return Status::OK(); } if (HaveManualCompaction(cfd)) { // Can't compact right now, but try again later TEST_SYNC_POINT("DBImpl::BackgroundCompaction()::Conflict"); return Status::OK(); } // Pick up latest mutable CF Options and use it throughout the // compaction job // Compaction makes a copy of the latest MutableCFOptions. It should be used // throughout the compaction procedure to make sure consistency. It will // eventually be installed into SuperVersion auto* mutable_cf_options = cfd->GetLatestMutableCFOptions(); if (!mutable_cf_options->disable_auto_compactions && !cfd->IsDropped()) { // NOTE: try to avoid unnecessary copy of MutableCFOptions if // compaction is not necessary. Need to make sure mutex is held // until we make a copy in the following code c.reset(cfd->PickCompaction(*mutable_cf_options, log_buffer)); if (c != nullptr) { // update statistics MeasureTime(stats_, NUM_FILES_IN_SINGLE_COMPACTION, c->inputs(0)->size()); // There are three things that can change compaction score: // 1) When flush or compaction finish. This case is covered by // InstallSuperVersionAndScheduleWork // 2) When MutableCFOptions changes. This case is also covered by // InstallSuperVersionAndScheduleWork, because this is when the new // options take effect. // 3) When we Pick a new compaction, we "remove" those files being // compacted from the calculation, which then influences compaction // score. Here we check if we need the new compaction even without the // files that are currently being compacted. If we need another // compaction, we might be able to execute it in parallel, so we add it // to the queue and schedule a new thread. if (cfd->NeedsCompaction()) { // Yes, we need more compactions! AddToCompactionQueue(cfd); ++unscheduled_compactions_; MaybeScheduleFlushOrCompaction(); } } } } if (!c) { // Nothing to do LogToBuffer(log_buffer, "Compaction nothing to do"); } else if (c->deletion_compaction()) { // TODO(icanadi) Do we want to honor snapshots here? i.e. not delete old // file if there is alive snapshot pointing to it assert(c->num_input_files(1) == 0); assert(c->level() == 0); assert(c->column_family_data()->ioptions()->compaction_style == kCompactionStyleFIFO); compaction_job_stats.num_input_files = c->num_input_files(0); for (const auto& f : *c->inputs(0)) { c->edit()->DeleteFile(c->level(), f->fd.GetNumber()); } status = versions_->LogAndApply(c->column_family_data(), *c->mutable_cf_options(), c->edit(), &mutex_, directories_.GetDbDir()); InstallSuperVersionAndScheduleWorkWrapper( c->column_family_data(), job_context, *c->mutable_cf_options()); LogToBuffer(log_buffer, "[%s] Deleted %d files\n", c->column_family_data()->GetName().c_str(), c->num_input_files(0)); *made_progress = true; } else if (!trivial_move_disallowed && c->IsTrivialMove()) { TEST_SYNC_POINT("DBImpl::BackgroundCompaction:TrivialMove"); // Instrument for event update // TODO(yhchiang): add op details for showing trivial-move. ThreadStatusUtil::SetColumnFamily( c->column_family_data(), c->column_family_data()->ioptions()->env, c->column_family_data()->options()->enable_thread_tracking); ThreadStatusUtil::SetThreadOperation(ThreadStatus::OP_COMPACTION); compaction_job_stats.num_input_files = c->num_input_files(0); // Move files to next level int32_t moved_files = 0; int64_t moved_bytes = 0; for (unsigned int l = 0; l < c->num_input_levels(); l++) { if (c->level(l) == c->output_level()) { continue; } for (size_t i = 0; i < c->num_input_files(l); i++) { FileMetaData* f = c->input(l, i); c->edit()->DeleteFile(c->level(l), f->fd.GetNumber()); c->edit()->AddFile(c->output_level(), f->fd.GetNumber(), f->fd.GetPathId(), f->fd.GetFileSize(), f->smallest, f->largest, f->smallest_seqno, f->largest_seqno, f->marked_for_compaction); LogToBuffer(log_buffer, "[%s] Moving #%" PRIu64 " to level-%d %" PRIu64 " bytes\n", c->column_family_data()->GetName().c_str(), f->fd.GetNumber(), c->output_level(), f->fd.GetFileSize()); ++moved_files; moved_bytes += f->fd.GetFileSize(); } } status = versions_->LogAndApply(c->column_family_data(), *c->mutable_cf_options(), c->edit(), &mutex_, directories_.GetDbDir()); // Use latest MutableCFOptions InstallSuperVersionAndScheduleWorkWrapper( c->column_family_data(), job_context, *c->mutable_cf_options()); VersionStorageInfo::LevelSummaryStorage tmp; c->column_family_data()->internal_stats()->IncBytesMoved(c->output_level(), moved_bytes); { event_logger_.LogToBuffer(log_buffer) << "job" << job_context->job_id << "event" << "trivial_move" << "destination_level" << c->output_level() << "files" << moved_files << "total_files_size" << moved_bytes; } LogToBuffer( log_buffer, "[%s] Moved #%d files to level-%d %" PRIu64 " bytes %s: %s\n", c->column_family_data()->GetName().c_str(), moved_files, c->output_level(), moved_bytes, status.ToString().c_str(), c->column_family_data()->current()->storage_info()->LevelSummary(&tmp)); *made_progress = true; // Clear Instrument ThreadStatusUtil::ResetThreadStatus(); } else { int output_level __attribute__((unused)) = c->output_level(); TEST_SYNC_POINT_CALLBACK("DBImpl::BackgroundCompaction:NonTrivial", &output_level); SequenceNumber earliest_write_conflict_snapshot; std::vector snapshot_seqs = snapshots_.GetAll(&earliest_write_conflict_snapshot); assert(is_snapshot_supported_ || snapshots_.empty()); CompactionJob compaction_job( job_context->job_id, c.get(), db_options_, env_options_, versions_.get(), &shutting_down_, log_buffer, directories_.GetDbDir(), directories_.GetDataDir(c->output_path_id()), stats_, snapshot_seqs, earliest_write_conflict_snapshot, table_cache_, &event_logger_, c->mutable_cf_options()->paranoid_file_checks, c->mutable_cf_options()->compaction_measure_io_stats, dbname_, &compaction_job_stats); compaction_job.Prepare(); mutex_.Unlock(); compaction_job.Run(); TEST_SYNC_POINT("DBImpl::BackgroundCompaction:NonTrivial:AfterRun"); mutex_.Lock(); status = compaction_job.Install(*c->mutable_cf_options(), &mutex_); if (status.ok()) { InstallSuperVersionAndScheduleWorkWrapper( c->column_family_data(), job_context, *c->mutable_cf_options()); } *made_progress = true; } if (c != nullptr) { NotifyOnCompactionCompleted( c->column_family_data(), c.get(), status, compaction_job_stats, job_context->job_id); c->ReleaseCompactionFiles(status); *made_progress = true; } // this will unref its input_version and column_family_data c.reset(); if (status.ok()) { // Done } else if (status.IsShutdownInProgress()) { // Ignore compaction errors found during shutting down } else { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Compaction error: %s", status.ToString().c_str()); if (db_options_.paranoid_checks && bg_error_.ok()) { bg_error_ = status; } } if (is_manual) { ManualCompaction* m = manual_compaction; if (!status.ok()) { m->status = status; m->done = true; } // For universal compaction: // Because universal compaction always happens at level 0, so one // compaction will pick up all overlapped files. No files will be // filtered out due to size limit and left for a successive compaction. // So we can safely conclude the current compaction. // // Also note that, if we don't stop here, then the current compaction // writes a new file back to level 0, which will be used in successive // compaction. Hence the manual compaction will never finish. // // Stop the compaction if manual_end points to nullptr -- this means // that we compacted the whole range. manual_end should always point // to nullptr in case of universal compaction if (m->manual_end == nullptr) { m->done = true; } if (!m->done) { // We only compacted part of the requested range. Update *m // to the range that is left to be compacted. // Universal and FIFO compactions should always compact the whole range assert(m->cfd->ioptions()->compaction_style != kCompactionStyleUniversal || m->cfd->ioptions()->num_levels > 1); assert(m->cfd->ioptions()->compaction_style != kCompactionStyleFIFO); m->tmp_storage = *m->manual_end; m->begin = &m->tmp_storage; m->incomplete = true; } m->in_progress = false; // not being processed anymore } return status; } bool DBImpl::HasPendingManualCompaction() { return (!manual_compaction_dequeue_.empty()); } void DBImpl::AddManualCompaction(DBImpl::ManualCompaction* m) { manual_compaction_dequeue_.push_back(m); } void DBImpl::RemoveManualCompaction(DBImpl::ManualCompaction* m) { // Remove from queue std::deque::iterator it = manual_compaction_dequeue_.begin(); while (it != manual_compaction_dequeue_.end()) { if (m == (*it)) { it = manual_compaction_dequeue_.erase(it); return; } it++; } assert(false); return; } bool DBImpl::ShouldntRunManualCompaction(ManualCompaction* m) { if (m->exclusive) { return (bg_compaction_scheduled_ > 0); } std::deque::iterator it = manual_compaction_dequeue_.begin(); bool seen = false; while (it != manual_compaction_dequeue_.end()) { if (m == (*it)) { it++; seen = true; continue; } else if (MCOverlap(m, (*it)) && (!seen && !(*it)->in_progress)) { // Consider the other manual compaction *it, conflicts if: // overlaps with m // and (*it) is ahead in the queue and is not yet in progress return true; } it++; } return false; } bool DBImpl::HaveManualCompaction(ColumnFamilyData* cfd) { // Remove from priority queue std::deque::iterator it = manual_compaction_dequeue_.begin(); while (it != manual_compaction_dequeue_.end()) { if ((*it)->exclusive) { return true; } if ((cfd == (*it)->cfd) && (!((*it)->in_progress || (*it)->done))) { // Allow automatic compaction if manual compaction is // is in progress return true; } it++; } return false; } bool DBImpl::HasExclusiveManualCompaction() { // Remove from priority queue std::deque::iterator it = manual_compaction_dequeue_.begin(); while (it != manual_compaction_dequeue_.end()) { if ((*it)->exclusive) { return true; } it++; } return false; } bool DBImpl::MCOverlap(ManualCompaction* m, ManualCompaction* m1) { if ((m->exclusive) || (m1->exclusive)) { return true; } if (m->cfd != m1->cfd) { return false; } return true; } namespace { struct IterState { IterState(DBImpl* _db, InstrumentedMutex* _mu, SuperVersion* _super_version) : db(_db), mu(_mu), super_version(_super_version) {} DBImpl* db; InstrumentedMutex* mu; SuperVersion* super_version; }; static void CleanupIteratorState(void* arg1, void* arg2) { IterState* state = reinterpret_cast(arg1); if (state->super_version->Unref()) { // Job id == 0 means that this is not our background process, but rather // user thread JobContext job_context(0); state->mu->Lock(); state->super_version->Cleanup(); state->db->FindObsoleteFiles(&job_context, false, true); state->mu->Unlock(); delete state->super_version; if (job_context.HaveSomethingToDelete()) { state->db->PurgeObsoleteFiles(job_context); } job_context.Clean(); } delete state; } } // namespace InternalIterator* DBImpl::NewInternalIterator(const ReadOptions& read_options, ColumnFamilyData* cfd, SuperVersion* super_version, Arena* arena) { InternalIterator* internal_iter; assert(arena != nullptr); // Need to create internal iterator from the arena. MergeIteratorBuilder merge_iter_builder(&cfd->internal_comparator(), arena); // Collect iterator for mutable mem merge_iter_builder.AddIterator( super_version->mem->NewIterator(read_options, arena)); // Collect all needed child iterators for immutable memtables super_version->imm->AddIterators(read_options, &merge_iter_builder); // Collect iterators for files in L0 - Ln super_version->current->AddIterators(read_options, env_options_, &merge_iter_builder); internal_iter = merge_iter_builder.Finish(); IterState* cleanup = new IterState(this, &mutex_, super_version); internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, nullptr); return internal_iter; } ColumnFamilyHandle* DBImpl::DefaultColumnFamily() const { return default_cf_handle_; } Status DBImpl::Get(const ReadOptions& read_options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) { return GetImpl(read_options, column_family, key, value); } // JobContext gets created and destructed outside of the lock -- // we // use this convinently to: // * malloc one SuperVersion() outside of the lock -- new_superversion // * delete SuperVersion()s outside of the lock -- superversions_to_free // // However, if InstallSuperVersionAndScheduleWork() gets called twice with the // same job_context, we can't reuse the SuperVersion() that got // malloced because // first call already used it. In that rare case, we take a hit and create a // new SuperVersion() inside of the mutex. We do similar thing // for superversion_to_free void DBImpl::InstallSuperVersionAndScheduleWorkWrapper( ColumnFamilyData* cfd, JobContext* job_context, const MutableCFOptions& mutable_cf_options) { mutex_.AssertHeld(); SuperVersion* old_superversion = InstallSuperVersionAndScheduleWork( cfd, job_context->new_superversion, mutable_cf_options); job_context->new_superversion = nullptr; job_context->superversions_to_free.push_back(old_superversion); } SuperVersion* DBImpl::InstallSuperVersionAndScheduleWork( ColumnFamilyData* cfd, SuperVersion* new_sv, const MutableCFOptions& mutable_cf_options) { mutex_.AssertHeld(); // Update max_total_in_memory_state_ size_t old_memtable_size = 0; auto* old_sv = cfd->GetSuperVersion(); if (old_sv) { old_memtable_size = old_sv->mutable_cf_options.write_buffer_size * old_sv->mutable_cf_options.max_write_buffer_number; } auto* old = cfd->InstallSuperVersion( new_sv ? new_sv : new SuperVersion(), &mutex_, mutable_cf_options); // Whenever we install new SuperVersion, we might need to issue new flushes or // compactions. SchedulePendingFlush(cfd); SchedulePendingCompaction(cfd); MaybeScheduleFlushOrCompaction(); // Update max_total_in_memory_state_ max_total_in_memory_state_ = max_total_in_memory_state_ - old_memtable_size + mutable_cf_options.write_buffer_size * mutable_cf_options.max_write_buffer_number; return old; } Status DBImpl::GetImpl(const ReadOptions& read_options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value, bool* value_found) { StopWatch sw(env_, stats_, DB_GET); PERF_TIMER_GUARD(get_snapshot_time); auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); SequenceNumber snapshot; if (read_options.snapshot != nullptr) { snapshot = reinterpret_cast( read_options.snapshot)->number_; } else { snapshot = versions_->LastSequence(); } // Acquire SuperVersion SuperVersion* sv = GetAndRefSuperVersion(cfd); // Prepare to store a list of merge operations if merge occurs. MergeContext merge_context; Status s; // First look in the memtable, then in the immutable memtable (if any). // s is both in/out. When in, s could either be OK or MergeInProgress. // merge_operands will contain the sequence of merges in the latter case. LookupKey lkey(key, snapshot); PERF_TIMER_STOP(get_snapshot_time); bool skip_memtable = (read_options.read_tier == kPersistedTier && has_unpersisted_data_); bool done = false; if (!skip_memtable) { if (sv->mem->Get(lkey, value, &s, &merge_context)) { done = true; RecordTick(stats_, MEMTABLE_HIT); } else if (sv->imm->Get(lkey, value, &s, &merge_context)) { done = true; RecordTick(stats_, MEMTABLE_HIT); } } if (!done) { PERF_TIMER_GUARD(get_from_output_files_time); sv->current->Get(read_options, lkey, value, &s, &merge_context, value_found); RecordTick(stats_, MEMTABLE_MISS); } { PERF_TIMER_GUARD(get_post_process_time); ReturnAndCleanupSuperVersion(cfd, sv); RecordTick(stats_, NUMBER_KEYS_READ); RecordTick(stats_, BYTES_READ, value->size()); MeasureTime(stats_, BYTES_PER_READ, value->size()); } return s; } std::vector DBImpl::MultiGet( const ReadOptions& read_options, const std::vector& column_family, const std::vector& keys, std::vector* values) { StopWatch sw(env_, stats_, DB_MULTIGET); PERF_TIMER_GUARD(get_snapshot_time); SequenceNumber snapshot; struct MultiGetColumnFamilyData { ColumnFamilyData* cfd; SuperVersion* super_version; }; std::unordered_map multiget_cf_data; // fill up and allocate outside of mutex for (auto cf : column_family) { auto cfh = reinterpret_cast(cf); auto cfd = cfh->cfd(); if (multiget_cf_data.find(cfd->GetID()) == multiget_cf_data.end()) { auto mgcfd = new MultiGetColumnFamilyData(); mgcfd->cfd = cfd; multiget_cf_data.insert({cfd->GetID(), mgcfd}); } } mutex_.Lock(); if (read_options.snapshot != nullptr) { snapshot = reinterpret_cast( read_options.snapshot)->number_; } else { snapshot = versions_->LastSequence(); } for (auto mgd_iter : multiget_cf_data) { mgd_iter.second->super_version = mgd_iter.second->cfd->GetSuperVersion()->Ref(); } mutex_.Unlock(); // Contain a list of merge operations if merge occurs. MergeContext merge_context; // Note: this always resizes the values array size_t num_keys = keys.size(); std::vector stat_list(num_keys); values->resize(num_keys); // Keep track of bytes that we read for statistics-recording later uint64_t bytes_read = 0; PERF_TIMER_STOP(get_snapshot_time); // For each of the given keys, apply the entire "get" process as follows: // First look in the memtable, then in the immutable memtable (if any). // s is both in/out. When in, s could either be OK or MergeInProgress. // merge_operands will contain the sequence of merges in the latter case. for (size_t i = 0; i < num_keys; ++i) { merge_context.Clear(); Status& s = stat_list[i]; std::string* value = &(*values)[i]; LookupKey lkey(keys[i], snapshot); auto cfh = reinterpret_cast(column_family[i]); auto mgd_iter = multiget_cf_data.find(cfh->cfd()->GetID()); assert(mgd_iter != multiget_cf_data.end()); auto mgd = mgd_iter->second; auto super_version = mgd->super_version; bool skip_memtable = (read_options.read_tier == kPersistedTier && has_unpersisted_data_); bool done = false; if (!skip_memtable) { if (super_version->mem->Get(lkey, value, &s, &merge_context)) { done = true; // TODO(?): RecordTick(stats_, MEMTABLE_HIT)? } else if (super_version->imm->Get(lkey, value, &s, &merge_context)) { done = true; // TODO(?): RecordTick(stats_, MEMTABLE_HIT)? } } if (!done) { PERF_TIMER_GUARD(get_from_output_files_time); super_version->current->Get(read_options, lkey, value, &s, &merge_context); // TODO(?): RecordTick(stats_, MEMTABLE_MISS)? } if (s.ok()) { bytes_read += value->size(); } } // Post processing (decrement reference counts and record statistics) PERF_TIMER_GUARD(get_post_process_time); autovector superversions_to_delete; // TODO(icanadi) do we need lock here or just around Cleanup()? mutex_.Lock(); for (auto mgd_iter : multiget_cf_data) { auto mgd = mgd_iter.second; if (mgd->super_version->Unref()) { mgd->super_version->Cleanup(); superversions_to_delete.push_back(mgd->super_version); } } mutex_.Unlock(); for (auto td : superversions_to_delete) { delete td; } for (auto mgd : multiget_cf_data) { delete mgd.second; } RecordTick(stats_, NUMBER_MULTIGET_CALLS); RecordTick(stats_, NUMBER_MULTIGET_KEYS_READ, num_keys); RecordTick(stats_, NUMBER_MULTIGET_BYTES_READ, bytes_read); MeasureTime(stats_, BYTES_PER_MULTIGET, bytes_read); PERF_TIMER_STOP(get_post_process_time); return stat_list; } #ifndef ROCKSDB_LITE Status DBImpl::AddFile(ColumnFamilyHandle* column_family, const std::string& file_path, bool move_file) { Status status; auto cfh = reinterpret_cast(column_family); ColumnFamilyData* cfd = cfh->cfd(); ExternalSstFileInfo file_info; file_info.file_path = file_path; status = env_->GetFileSize(file_path, &file_info.file_size); if (!status.ok()) { return status; } // Access the file using TableReader to extract // version, number of entries, smallest user key, largest user key std::unique_ptr sst_file; status = env_->NewRandomAccessFile(file_path, &sst_file, env_options_); if (!status.ok()) { return status; } std::unique_ptr sst_file_reader; sst_file_reader.reset(new RandomAccessFileReader(std::move(sst_file))); std::unique_ptr table_reader; status = cfd->ioptions()->table_factory->NewTableReader( TableReaderOptions(*cfd->ioptions(), env_options_, cfd->internal_comparator()), std::move(sst_file_reader), file_info.file_size, &table_reader); if (!status.ok()) { return status; } // Get the external sst file version from table properties const UserCollectedProperties& user_collected_properties = table_reader->GetTableProperties()->user_collected_properties; UserCollectedProperties::const_iterator external_sst_file_version_iter = user_collected_properties.find(ExternalSstFilePropertyNames::kVersion); if (external_sst_file_version_iter == user_collected_properties.end()) { return Status::InvalidArgument("Generated table version not found"); } file_info.version = DecodeFixed32(external_sst_file_version_iter->second.c_str()); if (file_info.version == 1) { // version 1 imply that all sequence numbers in table equal 0 file_info.sequence_number = 0; } else { return Status::InvalidArgument("Generated table version is not supported"); } // Get number of entries in table file_info.num_entries = table_reader->GetTableProperties()->num_entries; ParsedInternalKey key; std::unique_ptr iter( table_reader->NewIterator(ReadOptions())); // Get first (smallest) key from file iter->SeekToFirst(); if (!ParseInternalKey(iter->key(), &key)) { return Status::Corruption("Generated table have corrupted keys"); } if (key.sequence != 0) { return Status::Corruption("Generated table have non zero sequence number"); } file_info.smallest_key = key.user_key.ToString(); // Get last (largest) key from file iter->SeekToLast(); if (!ParseInternalKey(iter->key(), &key)) { return Status::Corruption("Generated table have corrupted keys"); } if (key.sequence != 0) { return Status::Corruption("Generated table have non zero sequence number"); } file_info.largest_key = key.user_key.ToString(); return AddFile(column_family, &file_info, move_file); } Status DBImpl::AddFile(ColumnFamilyHandle* column_family, const ExternalSstFileInfo* file_info, bool move_file) { Status status; auto cfh = reinterpret_cast(column_family); ColumnFamilyData* cfd = cfh->cfd(); if (file_info->num_entries == 0) { return Status::InvalidArgument("File contain no entries"); } if (file_info->version != 1) { return Status::InvalidArgument("Generated table version is not supported"); } // version 1 imply that file have only Put Operations with Sequence Number = 0 FileMetaData meta; meta.smallest = InternalKey(file_info->smallest_key, file_info->sequence_number, ValueType::kTypeValue); meta.largest = InternalKey(file_info->largest_key, file_info->sequence_number, ValueType::kTypeValue); if (!meta.smallest.Valid() || !meta.largest.Valid()) { return Status::Corruption("Generated table have corrupted keys"); } meta.smallest_seqno = file_info->sequence_number; meta.largest_seqno = file_info->sequence_number; if (meta.smallest_seqno != 0 || meta.largest_seqno != 0) { return Status::InvalidArgument( "Non zero sequence numbers are not supported"); } // Generate a location for the new table meta.fd = FileDescriptor(versions_->NewFileNumber(), 0, file_info->file_size); std::string db_fname = TableFileName( db_options_.db_paths, meta.fd.GetNumber(), meta.fd.GetPathId()); if (move_file) { status = env_->LinkFile(file_info->file_path, db_fname); if (status.IsNotSupported()) { // Original file is on a different FS, use copy instead of hard linking status = CopyFile(env_, file_info->file_path, db_fname, 0); } } else { status = CopyFile(env_, file_info->file_path, db_fname, 0); } if (!status.ok()) { return status; } { InstrumentedMutexLock l(&mutex_); const MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions(); WriteThread::Writer w; write_thread_.EnterUnbatched(&w, &mutex_); if (!snapshots_.empty()) { // Check that no snapshots are being held status = Status::NotSupported("Cannot add a file while holding snapshots"); } if (status.ok()) { // Verify that added file key range dont overlap with any keys in DB SuperVersion* sv = cfd->GetSuperVersion()->Ref(); Arena arena; ReadOptions ro; ro.total_order_seek = true; ScopedArenaIterator iter(NewInternalIterator(ro, cfd, sv, &arena)); InternalKey range_start(file_info->smallest_key, kMaxSequenceNumber, kTypeValue); iter->Seek(range_start.Encode()); status = iter->status(); if (status.ok() && iter->Valid()) { ParsedInternalKey seek_result; if (ParseInternalKey(iter->key(), &seek_result)) { auto* vstorage = cfd->current()->storage_info(); if (vstorage->InternalComparator()->user_comparator()->Compare( seek_result.user_key, file_info->largest_key) <= 0) { status = Status::NotSupported("Cannot add overlapping range"); } } else { status = Status::Corruption("DB have corrupted keys"); } } } if (status.ok()) { // Add file to L0 VersionEdit edit; edit.SetColumnFamily(cfd->GetID()); edit.AddFile(0, meta.fd.GetNumber(), meta.fd.GetPathId(), meta.fd.GetFileSize(), meta.smallest, meta.largest, meta.smallest_seqno, meta.largest_seqno, meta.marked_for_compaction); status = versions_->LogAndApply(cfd, mutable_cf_options, &edit, &mutex_, directories_.GetDbDir()); } write_thread_.ExitUnbatched(&w); if (status.ok()) { delete InstallSuperVersionAndScheduleWork(cfd, nullptr, mutable_cf_options); } } if (!status.ok()) { // We failed to add the file to the database Status s = env_->DeleteFile(db_fname); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "AddFile() clean up for file %s failed : %s", db_fname.c_str(), s.ToString().c_str()); } } else if (status.ok() && move_file) { // The file was moved and added successfully, remove original file link Status s = env_->DeleteFile(file_info->file_path); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "%s was added to DB successfully but failed to remove original file " "link : %s", file_info->file_path.c_str(), s.ToString().c_str()); } } return status; } #endif // ROCKSDB_LITE Status DBImpl::CreateColumnFamily(const ColumnFamilyOptions& cf_options, const std::string& column_family_name, ColumnFamilyHandle** handle) { Status s; Status persist_options_status; *handle = nullptr; s = CheckCompressionSupported(cf_options); if (s.ok() && db_options_.allow_concurrent_memtable_write) { s = CheckConcurrentWritesSupported(cf_options); } if (!s.ok()) { return s; } { InstrumentedMutexLock l(&mutex_); if (versions_->GetColumnFamilySet()->GetColumnFamily(column_family_name) != nullptr) { return Status::InvalidArgument("Column family already exists"); } VersionEdit edit; edit.AddColumnFamily(column_family_name); uint32_t new_id = versions_->GetColumnFamilySet()->GetNextColumnFamilyID(); edit.SetColumnFamily(new_id); edit.SetLogNumber(logfile_number_); edit.SetComparatorName(cf_options.comparator->Name()); // LogAndApply will both write the creation in MANIFEST and create // ColumnFamilyData object Options opt(db_options_, cf_options); { // write thread WriteThread::Writer w; write_thread_.EnterUnbatched(&w, &mutex_); // LogAndApply will both write the creation in MANIFEST and create // ColumnFamilyData object s = versions_->LogAndApply( nullptr, MutableCFOptions(opt, ImmutableCFOptions(opt)), &edit, &mutex_, directories_.GetDbDir(), false, &cf_options); if (s.ok()) { // If the column family was created successfully, we then persist // the updated RocksDB options under the same single write thread persist_options_status = WriteOptionsFile(); } write_thread_.ExitUnbatched(&w); } if (s.ok()) { single_column_family_mode_ = false; auto* cfd = versions_->GetColumnFamilySet()->GetColumnFamily(column_family_name); assert(cfd != nullptr); delete InstallSuperVersionAndScheduleWork( cfd, nullptr, *cfd->GetLatestMutableCFOptions()); if (!cfd->mem()->IsSnapshotSupported()) { is_snapshot_supported_ = false; } *handle = new ColumnFamilyHandleImpl(cfd, this, &mutex_); Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Created column family [%s] (ID %u)", column_family_name.c_str(), (unsigned)cfd->GetID()); } else { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Creating column family [%s] FAILED -- %s", column_family_name.c_str(), s.ToString().c_str()); } } // InstrumentedMutexLock l(&mutex_) // this is outside the mutex if (s.ok()) { NewThreadStatusCfInfo( reinterpret_cast(*handle)->cfd()); if (!persist_options_status.ok()) { if (db_options_.fail_if_options_file_error) { s = Status::IOError( "ColumnFamily has been created, but unable to persist" "options in CreateColumnFamily()", persist_options_status.ToString().c_str()); } Warn(db_options_.info_log, "Unable to persist options in CreateColumnFamily() -- %s", persist_options_status.ToString().c_str()); } } return s; } Status DBImpl::DropColumnFamily(ColumnFamilyHandle* column_family) { auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); if (cfd->GetID() == 0) { return Status::InvalidArgument("Can't drop default column family"); } bool cf_support_snapshot = cfd->mem()->IsSnapshotSupported(); VersionEdit edit; edit.DropColumnFamily(); edit.SetColumnFamily(cfd->GetID()); Status s; Status options_persist_status; { InstrumentedMutexLock l(&mutex_); if (cfd->IsDropped()) { s = Status::InvalidArgument("Column family already dropped!\n"); } if (s.ok()) { // we drop column family from a single write thread WriteThread::Writer w; write_thread_.EnterUnbatched(&w, &mutex_); s = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(), &edit, &mutex_); if (s.ok()) { // If the column family was dropped successfully, we then persist // the updated RocksDB options under the same single write thread options_persist_status = WriteOptionsFile(); } write_thread_.ExitUnbatched(&w); } if (!cf_support_snapshot) { // Dropped Column Family doesn't support snapshot. Need to recalculate // is_snapshot_supported_. bool new_is_snapshot_supported = true; for (auto c : *versions_->GetColumnFamilySet()) { if (!c->IsDropped() && !c->mem()->IsSnapshotSupported()) { new_is_snapshot_supported = false; break; } } is_snapshot_supported_ = new_is_snapshot_supported; } } if (s.ok()) { // Note that here we erase the associated cf_info of the to-be-dropped // cfd before its ref-count goes to zero to avoid having to erase cf_info // later inside db_mutex. EraseThreadStatusCfInfo(cfd); assert(cfd->IsDropped()); auto* mutable_cf_options = cfd->GetLatestMutableCFOptions(); max_total_in_memory_state_ -= mutable_cf_options->write_buffer_size * mutable_cf_options->max_write_buffer_number; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Dropped column family with id %u\n", cfd->GetID()); if (!options_persist_status.ok()) { if (db_options_.fail_if_options_file_error) { s = Status::IOError( "ColumnFamily has been dropped, but unable to persist " "options in DropColumnFamily()", options_persist_status.ToString().c_str()); } Warn(db_options_.info_log, "Unable to persist options in DropColumnFamily() -- %s", options_persist_status.ToString().c_str()); } } else { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Dropping column family with id %u FAILED -- %s\n", cfd->GetID(), s.ToString().c_str()); } return s; } bool DBImpl::KeyMayExist(const ReadOptions& read_options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value, bool* value_found) { if (value_found != nullptr) { // falsify later if key-may-exist but can't fetch value *value_found = true; } ReadOptions roptions = read_options; roptions.read_tier = kBlockCacheTier; // read from block cache only auto s = GetImpl(roptions, column_family, key, value, value_found); // If block_cache is enabled and the index block of the table didn't // not present in block_cache, the return value will be Status::Incomplete. // In this case, key may still exist in the table. return s.ok() || s.IsIncomplete(); } Iterator* DBImpl::NewIterator(const ReadOptions& read_options, ColumnFamilyHandle* column_family) { if (read_options.read_tier == kPersistedTier) { return NewErrorIterator(Status::NotSupported( "ReadTier::kPersistedData is not yet supported in iterators.")); } auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); XFUNC_TEST("", "managed_new", managed_new1, xf_manage_new, reinterpret_cast(this), const_cast(&read_options), is_snapshot_supported_); if (read_options.managed) { #ifdef ROCKSDB_LITE // not supported in lite version return NewErrorIterator(Status::InvalidArgument( "Managed Iterators not supported in RocksDBLite.")); #else if ((read_options.tailing) || (read_options.snapshot != nullptr) || (is_snapshot_supported_)) { return new ManagedIterator(this, read_options, cfd); } // Managed iter not supported return NewErrorIterator(Status::InvalidArgument( "Managed Iterators not supported without snapshots.")); #endif } else if (read_options.tailing) { #ifdef ROCKSDB_LITE // not supported in lite version return nullptr; #else SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_); auto iter = new ForwardIterator(this, read_options, cfd, sv); return NewDBIterator( env_, *cfd->ioptions(), cfd->user_comparator(), iter, kMaxSequenceNumber, sv->mutable_cf_options.max_sequential_skip_in_iterations, read_options.iterate_upper_bound, read_options.prefix_same_as_start, read_options.pin_data); #endif } else { SequenceNumber latest_snapshot = versions_->LastSequence(); SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_); auto snapshot = read_options.snapshot != nullptr ? reinterpret_cast( read_options.snapshot)->number_ : latest_snapshot; // Try to generate a DB iterator tree in continuous memory area to be // cache friendly. Here is an example of result: // +-------------------------------+ // | | // | ArenaWrappedDBIter | // | + | // | +---> Inner Iterator ------------+ // | | | | // | | +-- -- -- -- -- -- -- --+ | // | +--- | Arena | | // | | | | // | Allocated Memory: | | // | | +-------------------+ | // | | | DBIter | <---+ // | | + | // | | | +-> iter_ ------------+ // | | | | | // | | +-------------------+ | // | | | MergingIterator | <---+ // | | + | // | | | +->child iter1 ------------+ // | | | | | | // | | +->child iter2 ----------+ | // | | | | | | | // | | | +->child iter3 --------+ | | // | | | | | | // | | +-------------------+ | | | // | | | Iterator1 | <--------+ // | | +-------------------+ | | // | | | Iterator2 | <------+ // | | +-------------------+ | // | | | Iterator3 | <----+ // | | +-------------------+ // | | | // +-------+-----------------------+ // // ArenaWrappedDBIter inlines an arena area where all the iterators in // the iterator tree are allocated in the order of being accessed when // querying. // Laying out the iterators in the order of being accessed makes it more // likely that any iterator pointer is close to the iterator it points to so // that they are likely to be in the same cache line and/or page. ArenaWrappedDBIter* db_iter = NewArenaWrappedDbIterator( env_, *cfd->ioptions(), cfd->user_comparator(), snapshot, sv->mutable_cf_options.max_sequential_skip_in_iterations, read_options.iterate_upper_bound, read_options.prefix_same_as_start, read_options.pin_data); InternalIterator* internal_iter = NewInternalIterator(read_options, cfd, sv, db_iter->GetArena()); db_iter->SetIterUnderDBIter(internal_iter); return db_iter; } // To stop compiler from complaining return nullptr; } Status DBImpl::NewIterators( const ReadOptions& read_options, const std::vector& column_families, std::vector* iterators) { if (read_options.read_tier == kPersistedTier) { return Status::NotSupported( "ReadTier::kPersistedData is not yet supported in iterators."); } iterators->clear(); iterators->reserve(column_families.size()); XFUNC_TEST("", "managed_new", managed_new1, xf_manage_new, reinterpret_cast(this), const_cast(&read_options), is_snapshot_supported_); if (read_options.managed) { #ifdef ROCKSDB_LITE return Status::InvalidArgument( "Managed interator not supported in RocksDB lite"); #else if ((!read_options.tailing) && (read_options.snapshot == nullptr) && (!is_snapshot_supported_)) { return Status::InvalidArgument( "Managed interator not supported without snapshots"); } for (auto cfh : column_families) { auto cfd = reinterpret_cast(cfh)->cfd(); auto iter = new ManagedIterator(this, read_options, cfd); iterators->push_back(iter); } #endif } else if (read_options.tailing) { #ifdef ROCKSDB_LITE return Status::InvalidArgument( "Tailing interator not supported in RocksDB lite"); #else for (auto cfh : column_families) { auto cfd = reinterpret_cast(cfh)->cfd(); SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_); auto iter = new ForwardIterator(this, read_options, cfd, sv); iterators->push_back(NewDBIterator( env_, *cfd->ioptions(), cfd->user_comparator(), iter, kMaxSequenceNumber, sv->mutable_cf_options.max_sequential_skip_in_iterations, nullptr, false, read_options.pin_data)); } #endif } else { SequenceNumber latest_snapshot = versions_->LastSequence(); for (size_t i = 0; i < column_families.size(); ++i) { auto* cfd = reinterpret_cast( column_families[i])->cfd(); SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_); auto snapshot = read_options.snapshot != nullptr ? reinterpret_cast( read_options.snapshot)->number_ : latest_snapshot; ArenaWrappedDBIter* db_iter = NewArenaWrappedDbIterator( env_, *cfd->ioptions(), cfd->user_comparator(), snapshot, sv->mutable_cf_options.max_sequential_skip_in_iterations, nullptr, false, read_options.pin_data); InternalIterator* internal_iter = NewInternalIterator(read_options, cfd, sv, db_iter->GetArena()); db_iter->SetIterUnderDBIter(internal_iter); iterators->push_back(db_iter); } } return Status::OK(); } const Snapshot* DBImpl::GetSnapshot() { return GetSnapshotImpl(false); } #ifndef ROCKSDB_LITE const Snapshot* DBImpl::GetSnapshotForWriteConflictBoundary() { return GetSnapshotImpl(true); } #endif // ROCKSDB_LITE const Snapshot* DBImpl::GetSnapshotImpl(bool is_write_conflict_boundary) { int64_t unix_time = 0; env_->GetCurrentTime(&unix_time); // Ignore error SnapshotImpl* s = new SnapshotImpl; InstrumentedMutexLock l(&mutex_); // returns null if the underlying memtable does not support snapshot. if (!is_snapshot_supported_) { delete s; return nullptr; } return snapshots_.New(s, versions_->LastSequence(), unix_time, is_write_conflict_boundary); } void DBImpl::ReleaseSnapshot(const Snapshot* s) { const SnapshotImpl* casted_s = reinterpret_cast(s); { InstrumentedMutexLock l(&mutex_); snapshots_.Delete(casted_s); } delete casted_s; } // Convenience methods Status DBImpl::Put(const WriteOptions& o, ColumnFamilyHandle* column_family, const Slice& key, const Slice& val) { return DB::Put(o, column_family, key, val); } Status DBImpl::Merge(const WriteOptions& o, ColumnFamilyHandle* column_family, const Slice& key, const Slice& val) { auto cfh = reinterpret_cast(column_family); if (!cfh->cfd()->ioptions()->merge_operator) { return Status::NotSupported("Provide a merge_operator when opening DB"); } else { return DB::Merge(o, column_family, key, val); } } Status DBImpl::Delete(const WriteOptions& write_options, ColumnFamilyHandle* column_family, const Slice& key) { return DB::Delete(write_options, column_family, key); } Status DBImpl::SingleDelete(const WriteOptions& write_options, ColumnFamilyHandle* column_family, const Slice& key) { return DB::SingleDelete(write_options, column_family, key); } Status DBImpl::Write(const WriteOptions& write_options, WriteBatch* my_batch) { return WriteImpl(write_options, my_batch, nullptr); } #ifndef ROCKSDB_LITE Status DBImpl::WriteWithCallback(const WriteOptions& write_options, WriteBatch* my_batch, WriteCallback* callback) { return WriteImpl(write_options, my_batch, callback); } #endif // ROCKSDB_LITE Status DBImpl::WriteImpl(const WriteOptions& write_options, WriteBatch* my_batch, WriteCallback* callback) { if (my_batch == nullptr) { return Status::Corruption("Batch is nullptr!"); } if (write_options.timeout_hint_us != 0) { return Status::InvalidArgument("timeout_hint_us is deprecated"); } Status status; bool callback_failed = false; bool xfunc_attempted_write = false; XFUNC_TEST("transaction", "transaction_xftest_write_impl", xf_transaction_write1, xf_transaction_write, write_options, db_options_, my_batch, callback, this, &status, &xfunc_attempted_write); if (xfunc_attempted_write) { // Test already did the write return status; } PERF_TIMER_GUARD(write_pre_and_post_process_time); WriteThread::Writer w; w.batch = my_batch; w.sync = write_options.sync; w.disableWAL = write_options.disableWAL; w.in_batch_group = false; w.has_callback = (callback != nullptr) ? true : false; if (!write_options.disableWAL) { RecordTick(stats_, WRITE_WITH_WAL); } StopWatch write_sw(env_, db_options_.statistics.get(), DB_WRITE); write_thread_.JoinBatchGroup(&w); if (w.state == WriteThread::STATE_PARALLEL_FOLLOWER) { // we are a non-leader in a parallel group PERF_TIMER_GUARD(write_memtable_time); ColumnFamilyMemTablesImpl column_family_memtables( versions_->GetColumnFamilySet()); WriteBatchInternal::SetSequence(w.batch, w.sequence); w.status = WriteBatchInternal::InsertInto( w.batch, &column_family_memtables, &flush_scheduler_, write_options.ignore_missing_column_families, 0 /*log_number*/, this, true /*dont_filter_deletes*/, true /*concurrent_memtable_writes*/); if (write_thread_.CompleteParallelWorker(&w)) { // we're responsible for early exit auto last_sequence = w.parallel_group->last_writer->sequence + WriteBatchInternal::Count(w.parallel_group->last_writer->batch) - 1; SetTickerCount(stats_, SEQUENCE_NUMBER, last_sequence); versions_->SetLastSequence(last_sequence); write_thread_.EarlyExitParallelGroup(&w); } assert(w.state == WriteThread::STATE_COMPLETED); // STATE_COMPLETED conditional below handles exit } if (w.state == WriteThread::STATE_COMPLETED) { // write is complete and leader has updated sequence RecordTick(stats_, WRITE_DONE_BY_OTHER); return w.status; } // else we are the leader of the write batch group assert(w.state == WriteThread::STATE_GROUP_LEADER); WriteContext context; mutex_.Lock(); if (!write_options.disableWAL) { default_cf_internal_stats_->AddDBStats(InternalStats::WRITE_WITH_WAL, 1); } RecordTick(stats_, WRITE_DONE_BY_SELF); default_cf_internal_stats_->AddDBStats(InternalStats::WRITE_DONE_BY_SELF, 1); // Once reaches this point, the current writer "w" will try to do its write // job. It may also pick up some of the remaining writers in the "writers_" // when it finds suitable, and finish them in the same write batch. // This is how a write job could be done by the other writer. assert(!single_column_family_mode_ || versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1); uint64_t max_total_wal_size = (db_options_.max_total_wal_size == 0) ? 4 * max_total_in_memory_state_ : db_options_.max_total_wal_size; if (UNLIKELY(!single_column_family_mode_ && alive_log_files_.begin()->getting_flushed == false && total_log_size_ > max_total_wal_size)) { uint64_t flush_column_family_if_log_file = alive_log_files_.begin()->number; alive_log_files_.begin()->getting_flushed = true; Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Flushing all column families with data in WAL number %" PRIu64 ". Total log size is %" PRIu64 " while max_total_wal_size is %" PRIu64, flush_column_family_if_log_file, total_log_size_, max_total_wal_size); // no need to refcount because drop is happening in write thread, so can't // happen while we're in the write thread for (auto cfd : *versions_->GetColumnFamilySet()) { if (cfd->IsDropped()) { continue; } if (cfd->GetLogNumber() <= flush_column_family_if_log_file) { status = SwitchMemtable(cfd, &context); if (!status.ok()) { break; } cfd->imm()->FlushRequested(); SchedulePendingFlush(cfd); } } MaybeScheduleFlushOrCompaction(); } else if (UNLIKELY(write_buffer_.ShouldFlush())) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Flushing column family with largest mem table size. Write buffer is " "using %" PRIu64 " bytes out of a total of %" PRIu64 ".", write_buffer_.memory_usage(), write_buffer_.buffer_size()); // no need to refcount because drop is happening in write thread, so can't // happen while we're in the write thread ColumnFamilyData* largest_cfd = nullptr; size_t largest_cfd_size = 0; for (auto cfd : *versions_->GetColumnFamilySet()) { if (cfd->IsDropped()) { continue; } if (!cfd->mem()->IsEmpty()) { // We only consider active mem table, hoping immutable memtable is // already in the process of flushing. size_t cfd_size = cfd->mem()->ApproximateMemoryUsage(); if (largest_cfd == nullptr || cfd_size > largest_cfd_size) { largest_cfd = cfd; largest_cfd_size = cfd_size; } } } if (largest_cfd != nullptr) { status = SwitchMemtable(largest_cfd, &context); if (status.ok()) { largest_cfd->imm()->FlushRequested(); SchedulePendingFlush(largest_cfd); MaybeScheduleFlushOrCompaction(); } } } if (UNLIKELY(status.ok() && !bg_error_.ok())) { status = bg_error_; } if (UNLIKELY(status.ok() && !flush_scheduler_.Empty())) { status = ScheduleFlushes(&context); } if (UNLIKELY(status.ok() && (write_controller_.IsStopped() || write_controller_.NeedsDelay()))) { PERF_TIMER_STOP(write_pre_and_post_process_time); PERF_TIMER_GUARD(write_delay_time); // We don't know size of curent batch so that we always use the size // for previous one. It might create a fairness issue that expiration // might happen for smaller writes but larger writes can go through. // Can optimize it if it is an issue. status = DelayWrite(last_batch_group_size_); PERF_TIMER_START(write_pre_and_post_process_time); } uint64_t last_sequence = versions_->LastSequence(); WriteThread::Writer* last_writer = &w; autovector write_batch_group; bool need_log_sync = !write_options.disableWAL && write_options.sync; bool need_log_dir_sync = need_log_sync && !log_dir_synced_; if (status.ok()) { if (need_log_sync) { while (logs_.front().getting_synced) { log_sync_cv_.Wait(); } for (auto& log : logs_) { assert(!log.getting_synced); log.getting_synced = true; } } // Add to log and apply to memtable. We can release the lock // during this phase since &w is currently responsible for logging // and protects against concurrent loggers and concurrent writes // into memtables mutex_.Unlock(); if (callback != nullptr) { // If this write has a validation callback, check to see if this write // is able to be written. Must be called on the write thread. status = callback->Callback(this); callback_failed = true; } } else { mutex_.Unlock(); } // At this point the mutex is unlocked bool exit_completed_early = false; last_batch_group_size_ = write_thread_.EnterAsBatchGroupLeader( &w, &last_writer, &write_batch_group); if (status.ok()) { // Rules for when we can update the memtable concurrently // 1. supported by memtable // 2. Puts are not okay if inplace_update_support // 3. Deletes or SingleDeletes are not okay if filtering deletes // (controlled by both batch and memtable setting) // 4. Merges are not okay // // Rules 1..3 are enforced by checking the options // during startup (CheckConcurrentWritesSupported), so if // options.allow_concurrent_memtable_write is true then they can be // assumed to be true. Rule 4 is checked for each batch. We could // relax rules 2 and 3 if we could prevent write batches from referring // more than once to a particular key. bool parallel = db_options_.allow_concurrent_memtable_write && write_batch_group.size() > 1; int total_count = 0; uint64_t total_byte_size = 0; for (auto b : write_batch_group) { total_count += WriteBatchInternal::Count(b); total_byte_size = WriteBatchInternal::AppendedByteSize( total_byte_size, WriteBatchInternal::ByteSize(b)); parallel = parallel && !b->HasMerge(); } const SequenceNumber current_sequence = last_sequence + 1; last_sequence += total_count; // Record statistics RecordTick(stats_, NUMBER_KEYS_WRITTEN, total_count); RecordTick(stats_, BYTES_WRITTEN, total_byte_size); MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size); PERF_TIMER_STOP(write_pre_and_post_process_time); if (write_options.disableWAL) { has_unpersisted_data_ = true; } uint64_t log_size = 0; if (!write_options.disableWAL) { PERF_TIMER_GUARD(write_wal_time); WriteBatch* merged_batch = nullptr; if (write_batch_group.size() == 1) { merged_batch = write_batch_group[0]; } else { // WAL needs all of the batches flattened into a single batch. // We could avoid copying here with an iov-like AddRecord // interface merged_batch = &tmp_batch_; for (auto b : write_batch_group) { WriteBatchInternal::Append(merged_batch, b); } } WriteBatchInternal::SetSequence(merged_batch, current_sequence); assert(WriteBatchInternal::Count(merged_batch) == total_count); assert(WriteBatchInternal::ByteSize(merged_batch) == total_byte_size); Slice log_entry = WriteBatchInternal::Contents(merged_batch); status = logs_.back().writer->AddRecord(log_entry); total_log_size_ += log_entry.size(); alive_log_files_.back().AddSize(log_entry.size()); log_empty_ = false; log_size = log_entry.size(); RecordTick(stats_, WAL_FILE_BYTES, log_size); if (status.ok() && need_log_sync) { RecordTick(stats_, WAL_FILE_SYNCED); StopWatch sw(env_, stats_, WAL_FILE_SYNC_MICROS); // It's safe to access logs_ with unlocked mutex_ here because: // - we've set getting_synced=true for all logs, // so other threads won't pop from logs_ while we're here, // - only writer thread can push to logs_, and we're in // writer thread, so no one will push to logs_, // - as long as other threads don't modify it, it's safe to read // from std::deque from multiple threads concurrently. for (auto& log : logs_) { status = log.writer->file()->Sync(db_options_.use_fsync); if (!status.ok()) { break; } } if (status.ok() && need_log_dir_sync) { // We only sync WAL directory the first time WAL syncing is // requested, so that in case users never turn on WAL sync, // we can avoid the disk I/O in the write code path. status = directories_.GetWalDir()->Fsync(); } } if (merged_batch == &tmp_batch_) { tmp_batch_.Clear(); } } if (status.ok()) { PERF_TIMER_GUARD(write_memtable_time); { // Update stats while we are an exclusive group leader, so we know // that nobody else can be writing to these particular stats. // We're optimistic, updating the stats before we successfully // commit. That lets us release our leader status early in // some cases. auto stats = default_cf_internal_stats_; stats->AddDBStats(InternalStats::BYTES_WRITTEN, total_byte_size); stats->AddDBStats(InternalStats::NUMBER_KEYS_WRITTEN, total_count); if (!write_options.disableWAL) { if (write_options.sync) { stats->AddDBStats(InternalStats::WAL_FILE_SYNCED, 1); } stats->AddDBStats(InternalStats::WAL_FILE_BYTES, log_size); } uint64_t for_other = write_batch_group.size() - 1; if (for_other > 0) { stats->AddDBStats(InternalStats::WRITE_DONE_BY_OTHER, for_other); if (!write_options.disableWAL) { stats->AddDBStats(InternalStats::WRITE_WITH_WAL, for_other); } } } if (!parallel) { status = WriteBatchInternal::InsertInto( write_batch_group, current_sequence, column_family_memtables_.get(), &flush_scheduler_, write_options.ignore_missing_column_families, 0 /*log_number*/, this, false /*dont_filter_deletes*/); } else { WriteThread::ParallelGroup pg; pg.leader = &w; pg.last_writer = last_writer; pg.early_exit_allowed = !need_log_sync; pg.running.store(static_cast(write_batch_group.size()), std::memory_order_relaxed); write_thread_.LaunchParallelFollowers(&pg, current_sequence); ColumnFamilyMemTablesImpl column_family_memtables( versions_->GetColumnFamilySet()); assert(w.sequence == current_sequence); WriteBatchInternal::SetSequence(w.batch, w.sequence); w.status = WriteBatchInternal::InsertInto( w.batch, &column_family_memtables, &flush_scheduler_, write_options.ignore_missing_column_families, 0 /*log_number*/, this, true /*dont_filter_deletes*/, true /*concurrent_memtable_writes*/); assert(last_writer->sequence + WriteBatchInternal::Count(last_writer->batch) - 1 == last_sequence); // CompleteParallelWorker returns true if this thread should // handle exit, false means somebody else did exit_completed_early = !write_thread_.CompleteParallelWorker(&w); status = w.status; assert(status.ok() || !exit_completed_early); } if (status.ok() && !exit_completed_early) { SetTickerCount(stats_, SEQUENCE_NUMBER, last_sequence); versions_->SetLastSequence(last_sequence); if (!need_log_sync) { write_thread_.ExitAsBatchGroupLeader(&w, last_writer, status); exit_completed_early = true; } } // A non-OK status here indicates that the state implied by the // WAL has diverged from the in-memory state. This could be // because of a corrupt write_batch (very bad), or because the // client specified an invalid column family and didn't specify // ignore_missing_column_families. // // Is setting bg_error_ enough here? This will at least stop // compaction and fail any further writes. if (!status.ok() && bg_error_.ok()) { bg_error_ = status; } } } PERF_TIMER_START(write_pre_and_post_process_time); if (db_options_.paranoid_checks && !status.ok() && !callback_failed && !status.IsBusy()) { mutex_.Lock(); if (bg_error_.ok()) { bg_error_ = status; // stop compaction & fail any further writes } mutex_.Unlock(); } if (need_log_sync) { mutex_.Lock(); MarkLogsSynced(logfile_number_, need_log_dir_sync, status); mutex_.Unlock(); } if (!exit_completed_early) { write_thread_.ExitAsBatchGroupLeader(&w, last_writer, status); } return status; } // REQUIRES: mutex_ is held // REQUIRES: this thread is currently at the front of the writer queue Status DBImpl::DelayWrite(uint64_t num_bytes) { uint64_t time_delayed = 0; bool delayed = false; { StopWatch sw(env_, stats_, WRITE_STALL, &time_delayed); auto delay = write_controller_.GetDelay(env_, num_bytes); if (delay > 0) { mutex_.Unlock(); delayed = true; TEST_SYNC_POINT("DBImpl::DelayWrite:Sleep"); // hopefully we don't have to sleep more than 2 billion microseconds env_->SleepForMicroseconds(static_cast(delay)); mutex_.Lock(); } while (bg_error_.ok() && write_controller_.IsStopped()) { delayed = true; TEST_SYNC_POINT("DBImpl::DelayWrite:Wait"); bg_cv_.Wait(); } } if (delayed) { default_cf_internal_stats_->AddDBStats(InternalStats::WRITE_STALL_MICROS, time_delayed); RecordTick(stats_, STALL_MICROS, time_delayed); } return bg_error_; } Status DBImpl::ScheduleFlushes(WriteContext* context) { ColumnFamilyData* cfd; while ((cfd = flush_scheduler_.TakeNextColumnFamily()) != nullptr) { auto status = SwitchMemtable(cfd, context); if (cfd->Unref()) { delete cfd; } if (!status.ok()) { return status; } } return Status::OK(); } // REQUIRES: mutex_ is held // REQUIRES: this thread is currently at the front of the writer queue Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) { mutex_.AssertHeld(); unique_ptr lfile; log::Writer* new_log = nullptr; MemTable* new_mem = nullptr; // Attempt to switch to a new memtable and trigger flush of old. // Do this without holding the dbmutex lock. assert(versions_->prev_log_number() == 0); bool creating_new_log = !log_empty_; uint64_t recycle_log_number = 0; if (creating_new_log && db_options_.recycle_log_file_num && !log_recycle_files.empty()) { recycle_log_number = log_recycle_files.front(); log_recycle_files.pop_front(); } uint64_t new_log_number = creating_new_log ? versions_->NewFileNumber() : logfile_number_; SuperVersion* new_superversion = nullptr; const MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions(); mutex_.Unlock(); Status s; { if (creating_new_log) { EnvOptions opt_env_opt = env_->OptimizeForLogWrite(env_options_, db_options_); if (recycle_log_number) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "reusing log %" PRIu64 " from recycle list\n", recycle_log_number); s = env_->ReuseWritableFile( LogFileName(db_options_.wal_dir, new_log_number), LogFileName(db_options_.wal_dir, recycle_log_number), &lfile, opt_env_opt); } else { s = NewWritableFile(env_, LogFileName(db_options_.wal_dir, new_log_number), &lfile, opt_env_opt); } if (s.ok()) { // Our final size should be less than write_buffer_size // (compression, etc) but err on the side of caution. lfile->SetPreallocationBlockSize( mutable_cf_options.write_buffer_size / 10 + mutable_cf_options.write_buffer_size); unique_ptr file_writer( new WritableFileWriter(std::move(lfile), opt_env_opt)); new_log = new log::Writer(std::move(file_writer), new_log_number, db_options_.recycle_log_file_num > 0); } } if (s.ok()) { SequenceNumber seq = versions_->LastSequence(); new_mem = cfd->ConstructNewMemtable(mutable_cf_options, seq); new_superversion = new SuperVersion(); } } Log(InfoLogLevel::DEBUG_LEVEL, db_options_.info_log, "[%s] New memtable created with log file: #%" PRIu64 "\n", cfd->GetName().c_str(), new_log_number); mutex_.Lock(); if (!s.ok()) { // how do we fail if we're not creating new log? assert(creating_new_log); assert(!new_mem); assert(!new_log); return s; } if (creating_new_log) { logfile_number_ = new_log_number; assert(new_log != nullptr); log_empty_ = true; log_dir_synced_ = false; logs_.emplace_back(logfile_number_, new_log); alive_log_files_.push_back(LogFileNumberSize(logfile_number_)); for (auto loop_cfd : *versions_->GetColumnFamilySet()) { // all this is just optimization to delete logs that // are no longer needed -- if CF is empty, that means it // doesn't need that particular log to stay alive, so we just // advance the log number. no need to persist this in the manifest if (loop_cfd->mem()->GetFirstSequenceNumber() == 0 && loop_cfd->imm()->NumNotFlushed() == 0) { loop_cfd->SetLogNumber(logfile_number_); } } } cfd->mem()->SetNextLogNumber(logfile_number_); cfd->imm()->Add(cfd->mem(), &context->memtables_to_free_); new_mem->Ref(); cfd->SetMemtable(new_mem); context->superversions_to_free_.push_back(InstallSuperVersionAndScheduleWork( cfd, new_superversion, mutable_cf_options)); return s; } #ifndef ROCKSDB_LITE Status DBImpl::GetPropertiesOfAllTables(ColumnFamilyHandle* column_family, TablePropertiesCollection* props) { auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); // Increment the ref count mutex_.Lock(); auto version = cfd->current(); version->Ref(); mutex_.Unlock(); auto s = version->GetPropertiesOfAllTables(props); // Decrement the ref count mutex_.Lock(); version->Unref(); mutex_.Unlock(); return s; } Status DBImpl::GetPropertiesOfTablesInRange(ColumnFamilyHandle* column_family, const Range* range, std::size_t n, TablePropertiesCollection* props) { auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); // Increment the ref count mutex_.Lock(); auto version = cfd->current(); version->Ref(); mutex_.Unlock(); auto s = version->GetPropertiesOfTablesInRange(range, n, props); // Decrement the ref count mutex_.Lock(); version->Unref(); mutex_.Unlock(); return s; } #endif // ROCKSDB_LITE const std::string& DBImpl::GetName() const { return dbname_; } Env* DBImpl::GetEnv() const { return env_; } const Options& DBImpl::GetOptions(ColumnFamilyHandle* column_family) const { auto cfh = reinterpret_cast(column_family); return *cfh->cfd()->options(); } const DBOptions& DBImpl::GetDBOptions() const { return db_options_; } bool DBImpl::GetProperty(ColumnFamilyHandle* column_family, const Slice& property, std::string* value) { const DBPropertyInfo* property_info = GetPropertyInfo(property); value->clear(); auto cfd = reinterpret_cast(column_family)->cfd(); if (property_info == nullptr) { return false; } else if (property_info->handle_int) { uint64_t int_value; bool ret_value = GetIntPropertyInternal(cfd, *property_info, false, &int_value); if (ret_value) { *value = ToString(int_value); } return ret_value; } else if (property_info->handle_string) { InstrumentedMutexLock l(&mutex_); return cfd->internal_stats()->GetStringProperty(*property_info, property, value); } // Shouldn't reach here since exactly one of handle_string and handle_int // should be non-nullptr. assert(false); return false; } bool DBImpl::GetIntProperty(ColumnFamilyHandle* column_family, const Slice& property, uint64_t* value) { const DBPropertyInfo* property_info = GetPropertyInfo(property); if (property_info == nullptr || property_info->handle_int == nullptr) { return false; } auto cfd = reinterpret_cast(column_family)->cfd(); return GetIntPropertyInternal(cfd, *property_info, false, value); } bool DBImpl::GetIntPropertyInternal(ColumnFamilyData* cfd, const DBPropertyInfo& property_info, bool is_locked, uint64_t* value) { assert(property_info.handle_int != nullptr); if (!property_info.need_out_of_mutex) { if (is_locked) { mutex_.AssertHeld(); return cfd->internal_stats()->GetIntProperty(property_info, value, this); } else { InstrumentedMutexLock l(&mutex_); return cfd->internal_stats()->GetIntProperty(property_info, value, this); } } else { SuperVersion* sv = nullptr; if (!is_locked) { sv = GetAndRefSuperVersion(cfd); } else { sv = cfd->GetSuperVersion(); } bool ret = cfd->internal_stats()->GetIntPropertyOutOfMutex( property_info, sv->current, value); if (!is_locked) { ReturnAndCleanupSuperVersion(cfd, sv); } return ret; } } bool DBImpl::GetAggregatedIntProperty(const Slice& property, uint64_t* aggregated_value) { const DBPropertyInfo* property_info = GetPropertyInfo(property); if (property_info == nullptr || property_info->handle_int == nullptr) { return false; } uint64_t sum = 0; { // Needs mutex to protect the list of column families. InstrumentedMutexLock l(&mutex_); uint64_t value; for (auto* cfd : *versions_->GetColumnFamilySet()) { if (GetIntPropertyInternal(cfd, *property_info, true, &value)) { sum += value; } else { return false; } } } *aggregated_value = sum; return true; } SuperVersion* DBImpl::GetAndRefSuperVersion(ColumnFamilyData* cfd) { // TODO(ljin): consider using GetReferencedSuperVersion() directly return cfd->GetThreadLocalSuperVersion(&mutex_); } // REQUIRED: this function should only be called on the write thread or if the // mutex is held. SuperVersion* DBImpl::GetAndRefSuperVersion(uint32_t column_family_id) { auto column_family_set = versions_->GetColumnFamilySet(); auto cfd = column_family_set->GetColumnFamily(column_family_id); if (!cfd) { return nullptr; } return GetAndRefSuperVersion(cfd); } // REQUIRED: mutex is NOT held SuperVersion* DBImpl::GetAndRefSuperVersionUnlocked(uint32_t column_family_id) { ColumnFamilyData* cfd; { InstrumentedMutexLock l(&mutex_); auto column_family_set = versions_->GetColumnFamilySet(); cfd = column_family_set->GetColumnFamily(column_family_id); } if (!cfd) { return nullptr; } return GetAndRefSuperVersion(cfd); } void DBImpl::ReturnAndCleanupSuperVersion(ColumnFamilyData* cfd, SuperVersion* sv) { bool unref_sv = !cfd->ReturnThreadLocalSuperVersion(sv); if (unref_sv) { // Release SuperVersion if (sv->Unref()) { { InstrumentedMutexLock l(&mutex_); sv->Cleanup(); } delete sv; RecordTick(stats_, NUMBER_SUPERVERSION_CLEANUPS); } RecordTick(stats_, NUMBER_SUPERVERSION_RELEASES); } } // REQUIRED: this function should only be called on the write thread. void DBImpl::ReturnAndCleanupSuperVersion(uint32_t column_family_id, SuperVersion* sv) { auto column_family_set = versions_->GetColumnFamilySet(); auto cfd = column_family_set->GetColumnFamily(column_family_id); // If SuperVersion is held, and we successfully fetched a cfd using // GetAndRefSuperVersion(), it must still exist. assert(cfd != nullptr); ReturnAndCleanupSuperVersion(cfd, sv); } // REQUIRED: Mutex should NOT be held. void DBImpl::ReturnAndCleanupSuperVersionUnlocked(uint32_t column_family_id, SuperVersion* sv) { ColumnFamilyData* cfd; { InstrumentedMutexLock l(&mutex_); auto column_family_set = versions_->GetColumnFamilySet(); cfd = column_family_set->GetColumnFamily(column_family_id); } // If SuperVersion is held, and we successfully fetched a cfd using // GetAndRefSuperVersion(), it must still exist. assert(cfd != nullptr); ReturnAndCleanupSuperVersion(cfd, sv); } // REQUIRED: this function should only be called on the write thread or if the // mutex is held. ColumnFamilyHandle* DBImpl::GetColumnFamilyHandle(uint32_t column_family_id) { ColumnFamilyMemTables* cf_memtables = column_family_memtables_.get(); if (!cf_memtables->Seek(column_family_id)) { return nullptr; } return cf_memtables->GetColumnFamilyHandle(); } // REQUIRED: mutex is NOT held. ColumnFamilyHandle* DBImpl::GetColumnFamilyHandleUnlocked( uint32_t column_family_id) { ColumnFamilyMemTables* cf_memtables = column_family_memtables_.get(); InstrumentedMutexLock l(&mutex_); if (!cf_memtables->Seek(column_family_id)) { return nullptr; } return cf_memtables->GetColumnFamilyHandle(); } void DBImpl::GetApproximateSizes(ColumnFamilyHandle* column_family, const Range* range, int n, uint64_t* sizes, bool include_memtable) { Version* v; auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); SuperVersion* sv = GetAndRefSuperVersion(cfd); v = sv->current; for (int i = 0; i < n; i++) { // Convert user_key into a corresponding internal key. InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek); InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek); sizes[i] = versions_->ApproximateSize(v, k1.Encode(), k2.Encode()); if (include_memtable) { sizes[i] += sv->mem->ApproximateSize(k1.Encode(), k2.Encode()); sizes[i] += sv->imm->ApproximateSize(k1.Encode(), k2.Encode()); } } ReturnAndCleanupSuperVersion(cfd, sv); } std::list::iterator DBImpl::CaptureCurrentFileNumberInPendingOutputs() { // We need to remember the iterator of our insert, because after the // background job is done, we need to remove that element from // pending_outputs_. pending_outputs_.push_back(versions_->current_next_file_number()); auto pending_outputs_inserted_elem = pending_outputs_.end(); --pending_outputs_inserted_elem; return pending_outputs_inserted_elem; } void DBImpl::ReleaseFileNumberFromPendingOutputs( std::list::iterator v) { pending_outputs_.erase(v); } #ifndef ROCKSDB_LITE Status DBImpl::GetUpdatesSince( SequenceNumber seq, unique_ptr* iter, const TransactionLogIterator::ReadOptions& read_options) { RecordTick(stats_, GET_UPDATES_SINCE_CALLS); if (seq > versions_->LastSequence()) { return Status::NotFound("Requested sequence not yet written in the db"); } return wal_manager_.GetUpdatesSince(seq, iter, read_options, versions_.get()); } Status DBImpl::DeleteFile(std::string name) { uint64_t number; FileType type; WalFileType log_type; if (!ParseFileName(name, &number, &type, &log_type) || (type != kTableFile && type != kLogFile)) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "DeleteFile %s failed.\n", name.c_str()); return Status::InvalidArgument("Invalid file name"); } Status status; if (type == kLogFile) { // Only allow deleting archived log files if (log_type != kArchivedLogFile) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "DeleteFile %s failed - not archived log.\n", name.c_str()); return Status::NotSupported("Delete only supported for archived logs"); } status = env_->DeleteFile(db_options_.wal_dir + "/" + name.c_str()); if (!status.ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "DeleteFile %s failed -- %s.\n", name.c_str(), status.ToString().c_str()); } return status; } int level; FileMetaData* metadata; ColumnFamilyData* cfd; VersionEdit edit; JobContext job_context(next_job_id_.fetch_add(1), true); { InstrumentedMutexLock l(&mutex_); status = versions_->GetMetadataForFile(number, &level, &metadata, &cfd); if (!status.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "DeleteFile %s failed. File not found\n", name.c_str()); job_context.Clean(); return Status::InvalidArgument("File not found"); } assert(level < cfd->NumberLevels()); // If the file is being compacted no need to delete. if (metadata->being_compacted) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "DeleteFile %s Skipped. File about to be compacted\n", name.c_str()); job_context.Clean(); return Status::OK(); } // Only the files in the last level can be deleted externally. // This is to make sure that any deletion tombstones are not // lost. Check that the level passed is the last level. auto* vstoreage = cfd->current()->storage_info(); for (int i = level + 1; i < cfd->NumberLevels(); i++) { if (vstoreage->NumLevelFiles(i) != 0) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "DeleteFile %s FAILED. File not in last level\n", name.c_str()); job_context.Clean(); return Status::InvalidArgument("File not in last level"); } } // if level == 0, it has to be the oldest file if (level == 0 && vstoreage->LevelFiles(0).back()->fd.GetNumber() != number) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "DeleteFile %s failed ---" " target file in level 0 must be the oldest.", name.c_str()); job_context.Clean(); return Status::InvalidArgument("File in level 0, but not oldest"); } edit.SetColumnFamily(cfd->GetID()); edit.DeleteFile(level, number); status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(), &edit, &mutex_, directories_.GetDbDir()); if (status.ok()) { InstallSuperVersionAndScheduleWorkWrapper( cfd, &job_context, *cfd->GetLatestMutableCFOptions()); } FindObsoleteFiles(&job_context, false); } // lock released here LogFlush(db_options_.info_log); // remove files outside the db-lock if (job_context.HaveSomethingToDelete()) { // Call PurgeObsoleteFiles() without holding mutex. PurgeObsoleteFiles(job_context); } job_context.Clean(); return status; } Status DBImpl::DeleteFilesInRange(ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) { Status status; auto cfh = reinterpret_cast(column_family); ColumnFamilyData* cfd = cfh->cfd(); VersionEdit edit; std::vector deleted_files; JobContext job_context(next_job_id_.fetch_add(1), true); { InstrumentedMutexLock l(&mutex_); Version* input_version = cfd->current(); auto* vstorage = input_version->storage_info(); for (int i = 1; i < cfd->NumberLevels(); i++) { if (vstorage->LevelFiles(i).empty() || !vstorage->OverlapInLevel(i, begin, end)) { continue; } std::vector level_files; InternalKey begin_storage, end_storage, *begin_key, *end_key; if (begin == nullptr) { begin_key = nullptr; } else { begin_storage.SetMaxPossibleForUserKey(*begin); begin_key = &begin_storage; } if (end == nullptr) { end_key = nullptr; } else { end_storage.SetMinPossibleForUserKey(*end); end_key = &end_storage; } vstorage->GetOverlappingInputs(i, begin_key, end_key, &level_files, -1, nullptr, false); FileMetaData* level_file; for (uint32_t j = 0; j < level_files.size(); j++) { level_file = level_files[j]; if (((begin == nullptr) || (cfd->internal_comparator().user_comparator()->Compare( level_file->smallest.user_key(), *begin) >= 0)) && ((end == nullptr) || (cfd->internal_comparator().user_comparator()->Compare( level_file->largest.user_key(), *end) <= 0))) { if (level_file->being_compacted) { continue; } edit.SetColumnFamily(cfd->GetID()); edit.DeleteFile(i, level_file->fd.GetNumber()); deleted_files.push_back(level_file); level_file->being_compacted = true; } } } if (edit.GetDeletedFiles().empty()) { job_context.Clean(); return Status::OK(); } input_version->Ref(); status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(), &edit, &mutex_, directories_.GetDbDir()); if (status.ok()) { InstallSuperVersionAndScheduleWorkWrapper( cfd, &job_context, *cfd->GetLatestMutableCFOptions()); } for (auto* deleted_file : deleted_files) { deleted_file->being_compacted = false; } input_version->Unref(); FindObsoleteFiles(&job_context, false); } // lock released here LogFlush(db_options_.info_log); // remove files outside the db-lock if (job_context.HaveSomethingToDelete()) { // Call PurgeObsoleteFiles() without holding mutex. PurgeObsoleteFiles(job_context); } job_context.Clean(); return status; } void DBImpl::GetLiveFilesMetaData(std::vector* metadata) { InstrumentedMutexLock l(&mutex_); versions_->GetLiveFilesMetaData(metadata); } void DBImpl::GetColumnFamilyMetaData( ColumnFamilyHandle* column_family, ColumnFamilyMetaData* cf_meta) { assert(column_family); auto* cfd = reinterpret_cast(column_family)->cfd(); auto* sv = GetAndRefSuperVersion(cfd); sv->current->GetColumnFamilyMetaData(cf_meta); ReturnAndCleanupSuperVersion(cfd, sv); } #endif // ROCKSDB_LITE Status DBImpl::CheckConsistency() { mutex_.AssertHeld(); std::vector metadata; versions_->GetLiveFilesMetaData(&metadata); std::string corruption_messages; for (const auto& md : metadata) { // md.name has a leading "/". std::string file_path = md.db_path + md.name; uint64_t fsize = 0; Status s = env_->GetFileSize(file_path, &fsize); if (!s.ok() && env_->GetFileSize(Rocks2LevelTableFileName(file_path), &fsize).ok()) { s = Status::OK(); } if (!s.ok()) { corruption_messages += "Can't access " + md.name + ": " + s.ToString() + "\n"; } else if (fsize != md.size) { corruption_messages += "Sst file size mismatch: " + file_path + ". Size recorded in manifest " + ToString(md.size) + ", actual size " + ToString(fsize) + "\n"; } } if (corruption_messages.size() == 0) { return Status::OK(); } else { return Status::Corruption(corruption_messages); } } Status DBImpl::GetDbIdentity(std::string& identity) const { std::string idfilename = IdentityFileName(dbname_); const EnvOptions soptions; unique_ptr id_file_reader; Status s; { unique_ptr idfile; s = env_->NewSequentialFile(idfilename, &idfile, soptions); if (!s.ok()) { return s; } id_file_reader.reset(new SequentialFileReader(std::move(idfile))); } uint64_t file_size; s = env_->GetFileSize(idfilename, &file_size); if (!s.ok()) { return s; } char* buffer = reinterpret_cast(alloca(file_size)); Slice id; s = id_file_reader->Read(static_cast(file_size), &id, buffer); if (!s.ok()) { return s; } identity.assign(id.ToString()); // If last character is '\n' remove it from identity if (identity.size() > 0 && identity.back() == '\n') { identity.pop_back(); } return s; } // Default implementations of convenience methods that subclasses of DB // can call if they wish Status DB::Put(const WriteOptions& opt, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) { // Pre-allocate size of write batch conservatively. // 8 bytes are taken by header, 4 bytes for count, 1 byte for type, // and we allocate 11 extra bytes for key length, as well as value length. WriteBatch batch(key.size() + value.size() + 24); batch.Put(column_family, key, value); return Write(opt, &batch); } Status DB::Delete(const WriteOptions& opt, ColumnFamilyHandle* column_family, const Slice& key) { WriteBatch batch; batch.Delete(column_family, key); return Write(opt, &batch); } Status DB::SingleDelete(const WriteOptions& opt, ColumnFamilyHandle* column_family, const Slice& key) { WriteBatch batch; batch.SingleDelete(column_family, key); return Write(opt, &batch); } Status DB::Merge(const WriteOptions& opt, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) { WriteBatch batch; batch.Merge(column_family, key, value); return Write(opt, &batch); } // Default implementation -- returns not supported status Status DB::CreateColumnFamily(const ColumnFamilyOptions& cf_options, const std::string& column_family_name, ColumnFamilyHandle** handle) { return Status::NotSupported(""); } Status DB::DropColumnFamily(ColumnFamilyHandle* column_family) { return Status::NotSupported(""); } DB::~DB() { } Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) { DBOptions db_options(options); ColumnFamilyOptions cf_options(options); std::vector column_families; column_families.push_back( ColumnFamilyDescriptor(kDefaultColumnFamilyName, cf_options)); std::vector handles; Status s = DB::Open(db_options, dbname, column_families, &handles, dbptr); if (s.ok()) { assert(handles.size() == 1); // i can delete the handle since DBImpl is always holding a reference to // default column family delete handles[0]; } return s; } Status DB::Open(const DBOptions& db_options, const std::string& dbname, const std::vector& column_families, std::vector* handles, DB** dbptr) { Status s = SanitizeOptionsByTable(db_options, column_families); if (!s.ok()) { return s; } for (auto& cfd : column_families) { s = CheckCompressionSupported(cfd.options); if (s.ok() && db_options.allow_concurrent_memtable_write) { s = CheckConcurrentWritesSupported(cfd.options); } if (!s.ok()) { return s; } if (db_options.db_paths.size() > 1) { if ((cfd.options.compaction_style != kCompactionStyleUniversal) && (cfd.options.compaction_style != kCompactionStyleLevel)) { return Status::NotSupported( "More than one DB paths are only supported in " "universal and level compaction styles. "); } } } if (db_options.db_paths.size() > 4) { return Status::NotSupported( "More than four DB paths are not supported yet. "); } *dbptr = nullptr; handles->clear(); size_t max_write_buffer_size = 0; for (auto cf : column_families) { max_write_buffer_size = std::max(max_write_buffer_size, cf.options.write_buffer_size); } DBImpl* impl = new DBImpl(db_options, dbname); s = impl->env_->CreateDirIfMissing(impl->db_options_.wal_dir); if (s.ok()) { for (auto db_path : impl->db_options_.db_paths) { s = impl->env_->CreateDirIfMissing(db_path.path); if (!s.ok()) { break; } } } if (!s.ok()) { delete impl; return s; } s = impl->CreateArchivalDirectory(); if (!s.ok()) { delete impl; return s; } impl->mutex_.Lock(); // Handles create_if_missing, error_if_exists s = impl->Recover(column_families); if (s.ok()) { uint64_t new_log_number = impl->versions_->NewFileNumber(); unique_ptr lfile; EnvOptions soptions(db_options); EnvOptions opt_env_options = impl->db_options_.env->OptimizeForLogWrite(soptions, impl->db_options_); s = NewWritableFile(impl->db_options_.env, LogFileName(impl->db_options_.wal_dir, new_log_number), &lfile, opt_env_options); if (s.ok()) { lfile->SetPreallocationBlockSize((max_write_buffer_size / 10) + max_write_buffer_size); impl->logfile_number_ = new_log_number; unique_ptr file_writer( new WritableFileWriter(std::move(lfile), opt_env_options)); impl->logs_.emplace_back( new_log_number, new log::Writer(std::move(file_writer), new_log_number, impl->db_options_.recycle_log_file_num > 0)); // set column family handles for (auto cf : column_families) { auto cfd = impl->versions_->GetColumnFamilySet()->GetColumnFamily(cf.name); if (cfd != nullptr) { handles->push_back( new ColumnFamilyHandleImpl(cfd, impl, &impl->mutex_)); impl->NewThreadStatusCfInfo(cfd); } else { if (db_options.create_missing_column_families) { // missing column family, create it ColumnFamilyHandle* handle; impl->mutex_.Unlock(); s = impl->CreateColumnFamily(cf.options, cf.name, &handle); impl->mutex_.Lock(); if (s.ok()) { handles->push_back(handle); } else { break; } } else { s = Status::InvalidArgument("Column family not found: ", cf.name); break; } } } } if (s.ok()) { for (auto cfd : *impl->versions_->GetColumnFamilySet()) { delete impl->InstallSuperVersionAndScheduleWork( cfd, nullptr, *cfd->GetLatestMutableCFOptions()); } impl->alive_log_files_.push_back( DBImpl::LogFileNumberSize(impl->logfile_number_)); impl->DeleteObsoleteFiles(); s = impl->directories_.GetDbDir()->Fsync(); } } if (s.ok()) { for (auto cfd : *impl->versions_->GetColumnFamilySet()) { if (cfd->ioptions()->compaction_style == kCompactionStyleFIFO) { auto* vstorage = cfd->current()->storage_info(); for (int i = 1; i < vstorage->num_levels(); ++i) { int num_files = vstorage->NumLevelFiles(i); if (num_files > 0) { s = Status::InvalidArgument( "Not all files are at level 0. Cannot " "open with FIFO compaction style."); break; } } } if (!cfd->mem()->IsSnapshotSupported()) { impl->is_snapshot_supported_ = false; } if (cfd->ioptions()->merge_operator != nullptr && !cfd->mem()->IsMergeOperatorSupported()) { s = Status::InvalidArgument( "The memtable of column family %s does not support merge operator " "its options.merge_operator is non-null", cfd->GetName().c_str()); } if (!s.ok()) { break; } } } TEST_SYNC_POINT("DBImpl::Open:Opened"); Status persist_options_status; if (s.ok()) { // Persist RocksDB Options before scheduling the compaction. // The WriteOptionsFile() will release and lock the mutex internally. persist_options_status = impl->WriteOptionsFile(); *dbptr = impl; impl->opened_successfully_ = true; impl->MaybeScheduleFlushOrCompaction(); } impl->mutex_.Unlock(); auto sfm = static_cast( impl->db_options_.sst_file_manager.get()); if (s.ok() && sfm) { // Notify SstFileManager about all sst files that already exist in // db_paths[0] when the DB is opened. auto& db_path = impl->db_options_.db_paths[0]; std::vector existing_files; impl->db_options_.env->GetChildren(db_path.path, &existing_files); for (auto& file_name : existing_files) { uint64_t file_number; FileType file_type; std::string file_path = db_path.path + "/" + file_name; if (ParseFileName(file_name, &file_number, &file_type) && file_type == kTableFile) { sfm->OnAddFile(file_path); } } } if (s.ok()) { Log(InfoLogLevel::INFO_LEVEL, impl->db_options_.info_log, "DB pointer %p", impl); LogFlush(impl->db_options_.info_log); if (!persist_options_status.ok()) { if (db_options.fail_if_options_file_error) { s = Status::IOError( "DB::Open() failed --- Unable to persist Options file", persist_options_status.ToString()); } Warn(impl->db_options_.info_log, "Unable to persist options in DB::Open() -- %s", persist_options_status.ToString().c_str()); } } if (!s.ok()) { for (auto* h : *handles) { delete h; } handles->clear(); delete impl; *dbptr = nullptr; } return s; } Status DB::ListColumnFamilies(const DBOptions& db_options, const std::string& name, std::vector* column_families) { return VersionSet::ListColumnFamilies(column_families, name, db_options.env); } Snapshot::~Snapshot() { } Status DestroyDB(const std::string& dbname, const Options& options) { const InternalKeyComparator comparator(options.comparator); const Options& soptions(SanitizeOptions(dbname, &comparator, options)); Env* env = soptions.env; std::vector filenames; // Ignore error in case directory does not exist env->GetChildren(dbname, &filenames); FileLock* lock; const std::string lockname = LockFileName(dbname); Status result = env->LockFile(lockname, &lock); if (result.ok()) { uint64_t number; FileType type; InfoLogPrefix info_log_prefix(!options.db_log_dir.empty(), dbname); for (size_t i = 0; i < filenames.size(); i++) { if (ParseFileName(filenames[i], &number, info_log_prefix.prefix, &type) && type != kDBLockFile) { // Lock file will be deleted at end Status del; std::string path_to_delete = dbname + "/" + filenames[i]; if (type == kMetaDatabase) { del = DestroyDB(path_to_delete, options); } else if (type == kTableFile) { del = DeleteSSTFile(&options, path_to_delete, 0); } else { del = env->DeleteFile(path_to_delete); } if (result.ok() && !del.ok()) { result = del; } } } for (size_t path_id = 0; path_id < options.db_paths.size(); path_id++) { const auto& db_path = options.db_paths[path_id]; env->GetChildren(db_path.path, &filenames); for (size_t i = 0; i < filenames.size(); i++) { if (ParseFileName(filenames[i], &number, &type) && type == kTableFile) { // Lock file will be deleted at end std::string table_path = db_path.path + "/" + filenames[i]; Status del = DeleteSSTFile(&options, table_path, static_cast(path_id)); if (result.ok() && !del.ok()) { result = del; } } } } std::vector walDirFiles; std::string archivedir = ArchivalDirectory(dbname); if (dbname != soptions.wal_dir) { env->GetChildren(soptions.wal_dir, &walDirFiles); archivedir = ArchivalDirectory(soptions.wal_dir); } // Delete log files in the WAL dir for (const auto& file : walDirFiles) { if (ParseFileName(file, &number, &type) && type == kLogFile) { Status del = env->DeleteFile(soptions.wal_dir + "/" + file); if (result.ok() && !del.ok()) { result = del; } } } std::vector archiveFiles; env->GetChildren(archivedir, &archiveFiles); // Delete archival files. for (size_t i = 0; i < archiveFiles.size(); ++i) { if (ParseFileName(archiveFiles[i], &number, &type) && type == kLogFile) { Status del = env->DeleteFile(archivedir + "/" + archiveFiles[i]); if (result.ok() && !del.ok()) { result = del; } } } // ignore case where no archival directory is present. env->DeleteDir(archivedir); env->UnlockFile(lock); // Ignore error since state is already gone env->DeleteFile(lockname); env->DeleteDir(dbname); // Ignore error in case dir contains other files env->DeleteDir(soptions.wal_dir); } return result; } Status DBImpl::WriteOptionsFile() { #ifndef ROCKSDB_LITE mutex_.AssertHeld(); std::vector cf_names; std::vector cf_opts; // This part requires mutex to protect the column family options for (auto cfd : *versions_->GetColumnFamilySet()) { if (cfd->IsDropped()) { continue; } cf_names.push_back(cfd->GetName()); cf_opts.push_back(BuildColumnFamilyOptions( *cfd->options(), *cfd->GetLatestMutableCFOptions())); } // Unlock during expensive operations. New writes cannot get here // because the single write thread ensures all new writes get queued. mutex_.Unlock(); std::string file_name = TempOptionsFileName(GetName(), versions_->NewFileNumber()); Status s = PersistRocksDBOptions(GetDBOptions(), cf_names, cf_opts, file_name, GetEnv()); if (s.ok()) { s = RenameTempFileToOptionsFile(file_name); } mutex_.Lock(); return s; #else return Status::OK(); #endif // !ROCKSDB_LITE } #ifndef ROCKSDB_LITE namespace { void DeleteOptionsFilesHelper(const std::map& filenames, const size_t num_files_to_keep, const std::shared_ptr& info_log, Env* env) { if (filenames.size() <= num_files_to_keep) { return; } for (auto iter = std::next(filenames.begin(), num_files_to_keep); iter != filenames.end(); ++iter) { if (!env->DeleteFile(iter->second).ok()) { Warn(info_log, "Unable to delete options file %s", iter->second.c_str()); } } } } // namespace #endif // !ROCKSDB_LITE Status DBImpl::DeleteObsoleteOptionsFiles() { #ifndef ROCKSDB_LITE std::vector filenames; // use ordered map to store keep the filenames sorted from the newest // to the oldest. std::map options_filenames; Status s; s = GetEnv()->GetChildren(GetName(), &filenames); if (!s.ok()) { return s; } for (auto& filename : filenames) { uint64_t file_number; FileType type; if (ParseFileName(filename, &file_number, &type) && type == kOptionsFile) { options_filenames.insert( {std::numeric_limits::max() - file_number, GetName() + "/" + filename}); } } // Keeps the latest 2 Options file const size_t kNumOptionsFilesKept = 2; DeleteOptionsFilesHelper(options_filenames, kNumOptionsFilesKept, db_options_.info_log, GetEnv()); return Status::OK(); #else return Status::OK(); #endif // !ROCKSDB_LITE } Status DBImpl::RenameTempFileToOptionsFile(const std::string& file_name) { #ifndef ROCKSDB_LITE Status s; std::string options_file_name = OptionsFileName(GetName(), versions_->NewFileNumber()); // Retry if the file name happen to conflict with an existing one. s = GetEnv()->RenameFile(file_name, options_file_name); DeleteObsoleteOptionsFiles(); return s; #else return Status::OK(); #endif // !ROCKSDB_LITE } #if ROCKSDB_USING_THREAD_STATUS void DBImpl::NewThreadStatusCfInfo( ColumnFamilyData* cfd) const { if (db_options_.enable_thread_tracking) { ThreadStatusUtil::NewColumnFamilyInfo(this, cfd, cfd->GetName(), cfd->ioptions()->env); } } void DBImpl::EraseThreadStatusCfInfo( ColumnFamilyData* cfd) const { if (db_options_.enable_thread_tracking) { ThreadStatusUtil::EraseColumnFamilyInfo(cfd); } } void DBImpl::EraseThreadStatusDbInfo() const { if (db_options_.enable_thread_tracking) { ThreadStatusUtil::EraseDatabaseInfo(this); } } #else void DBImpl::NewThreadStatusCfInfo( ColumnFamilyData* cfd) const { } void DBImpl::EraseThreadStatusCfInfo( ColumnFamilyData* cfd) const { } void DBImpl::EraseThreadStatusDbInfo() const { } #endif // ROCKSDB_USING_THREAD_STATUS // // A global method that can dump out the build version void DumpRocksDBBuildVersion(Logger * log) { #if !defined(IOS_CROSS_COMPILE) // if we compile with Xcode, we don't run build_detect_vesion, so we don't // generate util/build_version.cc Header(log, "RocksDB version: %d.%d.%d\n", ROCKSDB_MAJOR, ROCKSDB_MINOR, ROCKSDB_PATCH); Header(log, "Git sha %s", rocksdb_build_git_sha); Header(log, "Compile date %s", rocksdb_build_compile_date); #endif } #ifndef ROCKSDB_LITE SequenceNumber DBImpl::GetEarliestMemTableSequenceNumber(SuperVersion* sv, bool include_history) { // Find the earliest sequence number that we know we can rely on reading // from the memtable without needing to check sst files. SequenceNumber earliest_seq = sv->imm->GetEarliestSequenceNumber(include_history); if (earliest_seq == kMaxSequenceNumber) { earliest_seq = sv->mem->GetEarliestSequenceNumber(); } assert(sv->mem->GetEarliestSequenceNumber() >= earliest_seq); return earliest_seq; } #endif // ROCKSDB_LITE #ifndef ROCKSDB_LITE Status DBImpl::GetLatestSequenceForKey(SuperVersion* sv, const Slice& key, bool cache_only, SequenceNumber* seq, bool* found_record_for_key) { Status s; MergeContext merge_context; SequenceNumber current_seq = versions_->LastSequence(); LookupKey lkey(key, current_seq); *seq = kMaxSequenceNumber; *found_record_for_key = false; // Check if there is a record for this key in the latest memtable sv->mem->Get(lkey, nullptr, &s, &merge_context, seq); if (!(s.ok() || s.IsNotFound() || s.IsMergeInProgress())) { // unexpected error reading memtable. Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Unexpected status returned from MemTable::Get: %s\n", s.ToString().c_str()); return s; } if (*seq != kMaxSequenceNumber) { // Found a sequence number, no need to check immutable memtables *found_record_for_key = true; return Status::OK(); } // Check if there is a record for this key in the immutable memtables sv->imm->Get(lkey, nullptr, &s, &merge_context, seq); if (!(s.ok() || s.IsNotFound() || s.IsMergeInProgress())) { // unexpected error reading memtable. Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Unexpected status returned from MemTableList::Get: %s\n", s.ToString().c_str()); return s; } if (*seq != kMaxSequenceNumber) { // Found a sequence number, no need to check memtable history *found_record_for_key = true; return Status::OK(); } // Check if there is a record for this key in the immutable memtables sv->imm->GetFromHistory(lkey, nullptr, &s, &merge_context, seq); if (!(s.ok() || s.IsNotFound() || s.IsMergeInProgress())) { // unexpected error reading memtable. Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Unexpected status returned from MemTableList::GetFromHistory: %s\n", s.ToString().c_str()); return s; } if (*seq != kMaxSequenceNumber) { // Found a sequence number, no need to check SST files *found_record_for_key = true; return Status::OK(); } // TODO(agiardullo): possible optimization: consider checking cached // SST files if cache_only=true? if (!cache_only) { // Check tables ReadOptions read_options; sv->current->Get(read_options, lkey, nullptr, &s, &merge_context, nullptr /* value_found */, found_record_for_key, seq); if (!(s.ok() || s.IsNotFound() || s.IsMergeInProgress())) { // unexpected error reading SST files Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Unexpected status returned from Version::Get: %s\n", s.ToString().c_str()); return s; } } return Status::OK(); } #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/db_impl.h000066400000000000000000001114641267530634000153310ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include #include #include #include "db/column_family.h" #include "db/compaction_job.h" #include "db/dbformat.h" #include "db/flush_job.h" #include "db/flush_scheduler.h" #include "db/internal_stats.h" #include "db/log_writer.h" #include "db/snapshot_impl.h" #include "db/version_edit.h" #include "db/wal_manager.h" #include "db/write_controller.h" #include "db/write_thread.h" #include "db/writebuffer.h" #include "memtable_list.h" #include "port/port.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "rocksdb/transaction_log.h" #include "table/scoped_arena_iterator.h" #include "util/autovector.h" #include "util/event_logger.h" #include "util/hash.h" #include "util/instrumented_mutex.h" #include "util/stop_watch.h" #include "util/thread_local.h" namespace rocksdb { class MemTable; class TableCache; class Version; class VersionEdit; class VersionSet; class Arena; class WriteCallback; struct JobContext; struct ExternalSstFileInfo; class DBImpl : public DB { public: DBImpl(const DBOptions& options, const std::string& dbname); virtual ~DBImpl(); // Implementations of the DB interface using DB::Put; virtual Status Put(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override; using DB::Merge; virtual Status Merge(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override; using DB::Delete; virtual Status Delete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) override; using DB::SingleDelete; virtual Status SingleDelete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) override; using DB::Write; virtual Status Write(const WriteOptions& options, WriteBatch* updates) override; using DB::Get; virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) override; using DB::MultiGet; virtual std::vector MultiGet( const ReadOptions& options, const std::vector& column_family, const std::vector& keys, std::vector* values) override; virtual Status CreateColumnFamily(const ColumnFamilyOptions& options, const std::string& column_family, ColumnFamilyHandle** handle) override; virtual Status DropColumnFamily(ColumnFamilyHandle* column_family) override; // Returns false if key doesn't exist in the database and true if it may. // If value_found is not passed in as null, then return the value if found in // memory. On return, if value was found, then value_found will be set to true // , otherwise false. using DB::KeyMayExist; virtual bool KeyMayExist(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value, bool* value_found = nullptr) override; using DB::NewIterator; virtual Iterator* NewIterator(const ReadOptions& options, ColumnFamilyHandle* column_family) override; virtual Status NewIterators( const ReadOptions& options, const std::vector& column_families, std::vector* iterators) override; virtual const Snapshot* GetSnapshot() override; virtual void ReleaseSnapshot(const Snapshot* snapshot) override; using DB::GetProperty; virtual bool GetProperty(ColumnFamilyHandle* column_family, const Slice& property, std::string* value) override; using DB::GetIntProperty; virtual bool GetIntProperty(ColumnFamilyHandle* column_family, const Slice& property, uint64_t* value) override; using DB::GetAggregatedIntProperty; virtual bool GetAggregatedIntProperty(const Slice& property, uint64_t* aggregated_value) override; using DB::GetApproximateSizes; virtual void GetApproximateSizes(ColumnFamilyHandle* column_family, const Range* range, int n, uint64_t* sizes, bool include_memtable = false) override; using DB::CompactRange; virtual Status CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) override; using DB::CompactFiles; virtual Status CompactFiles(const CompactionOptions& compact_options, ColumnFamilyHandle* column_family, const std::vector& input_file_names, const int output_level, const int output_path_id = -1) override; virtual Status PauseBackgroundWork() override; virtual Status ContinueBackgroundWork() override; virtual Status EnableAutoCompaction( const std::vector& column_family_handles) override; using DB::SetOptions; Status SetOptions( ColumnFamilyHandle* column_family, const std::unordered_map& options_map) override; using DB::NumberLevels; virtual int NumberLevels(ColumnFamilyHandle* column_family) override; using DB::MaxMemCompactionLevel; virtual int MaxMemCompactionLevel(ColumnFamilyHandle* column_family) override; using DB::Level0StopWriteTrigger; virtual int Level0StopWriteTrigger( ColumnFamilyHandle* column_family) override; virtual const std::string& GetName() const override; virtual Env* GetEnv() const override; using DB::GetOptions; virtual const Options& GetOptions( ColumnFamilyHandle* column_family) const override; using DB::GetDBOptions; virtual const DBOptions& GetDBOptions() const override; using DB::Flush; virtual Status Flush(const FlushOptions& options, ColumnFamilyHandle* column_family) override; virtual Status SyncWAL() override; virtual SequenceNumber GetLatestSequenceNumber() const override; #ifndef ROCKSDB_LITE virtual Status DisableFileDeletions() override; virtual Status EnableFileDeletions(bool force) override; virtual int IsFileDeletionsEnabled() const; // All the returned filenames start with "/" virtual Status GetLiveFiles(std::vector&, uint64_t* manifest_file_size, bool flush_memtable = true) override; virtual Status GetSortedWalFiles(VectorLogPtr& files) override; virtual Status GetUpdatesSince( SequenceNumber seq_number, unique_ptr* iter, const TransactionLogIterator::ReadOptions& read_options = TransactionLogIterator::ReadOptions()) override; virtual Status DeleteFile(std::string name) override; Status DeleteFilesInRange(ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end); virtual void GetLiveFilesMetaData( std::vector* metadata) override; // Obtains the meta data of the specified column family of the DB. // Status::NotFound() will be returned if the current DB does not have // any column family match the specified name. // TODO(yhchiang): output parameter is placed in the end in this codebase. virtual void GetColumnFamilyMetaData( ColumnFamilyHandle* column_family, ColumnFamilyMetaData* metadata) override; // experimental API Status SuggestCompactRange(ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end); Status PromoteL0(ColumnFamilyHandle* column_family, int target_level); // Similar to Write() but will call the callback once on the single write // thread to determine whether it is safe to perform the write. virtual Status WriteWithCallback(const WriteOptions& write_options, WriteBatch* my_batch, WriteCallback* callback); // Returns the sequence number that is guaranteed to be smaller than or equal // to the sequence number of any key that could be inserted into the current // memtables. It can then be assumed that any write with a larger(or equal) // sequence number will be present in this memtable or a later memtable. // // If the earliest sequence number could not be determined, // kMaxSequenceNumber will be returned. // // If include_history=true, will also search Memtables in MemTableList // History. SequenceNumber GetEarliestMemTableSequenceNumber(SuperVersion* sv, bool include_history); // For a given key, check to see if there are any records for this key // in the memtables, including memtable history. If cache_only is false, // SST files will also be checked. // // If a key is found, *found_record_for_key will be set to true and // *seq will will be set to the stored sequence number for the latest // operation on this key or kMaxSequenceNumber if unknown. // If no key is found, *found_record_for_key will be set to false. // // Note: If cache_only=false, it is possible for *seq to be set to 0 if // the sequence number has been cleared from the record. If the caller is // holding an active db snapshot, we know the missing sequence must be less // than the snapshot's sequence number (sequence numbers are only cleared // when there are no earlier active snapshots). // // If NotFound is returned and found_record_for_key is set to false, then no // record for this key was found. If the caller is holding an active db // snapshot, we know that no key could have existing after this snapshot // (since we do not compact keys that have an earlier snapshot). // // Returns OK or NotFound on success, // other status on unexpected error. Status GetLatestSequenceForKey(SuperVersion* sv, const Slice& key, bool cache_only, SequenceNumber* seq, bool* found_record_for_key); using DB::AddFile; virtual Status AddFile(ColumnFamilyHandle* column_family, const ExternalSstFileInfo* file_info, bool move_file) override; virtual Status AddFile(ColumnFamilyHandle* column_family, const std::string& file_path, bool move_file) override; #endif // ROCKSDB_LITE // Similar to GetSnapshot(), but also lets the db know that this snapshot // will be used for transaction write-conflict checking. The DB can then // make sure not to compact any keys that would prevent a write-conflict from // being detected. const Snapshot* GetSnapshotForWriteConflictBoundary(); // checks if all live files exist on file system and that their file sizes // match to our in-memory records virtual Status CheckConsistency(); virtual Status GetDbIdentity(std::string& identity) const override; Status RunManualCompaction(ColumnFamilyData* cfd, int input_level, int output_level, uint32_t output_path_id, const Slice* begin, const Slice* end, bool exclusive, bool disallow_trivial_move = false); // Return an internal iterator over the current state of the database. // The keys of this iterator are internal keys (see format.h). // The returned iterator should be deleted when no longer needed. InternalIterator* NewInternalIterator( Arena* arena, ColumnFamilyHandle* column_family = nullptr); #ifndef NDEBUG // Extra methods (for testing) that are not in the public DB interface // Implemented in db_impl_debug.cc // Compact any files in the named level that overlap [*begin, *end] Status TEST_CompactRange(int level, const Slice* begin, const Slice* end, ColumnFamilyHandle* column_family = nullptr, bool disallow_trivial_move = false); // Force current memtable contents to be flushed. Status TEST_FlushMemTable(bool wait = true); // Wait for memtable compaction Status TEST_WaitForFlushMemTable(ColumnFamilyHandle* column_family = nullptr); // Wait for any compaction Status TEST_WaitForCompact(); // Return the maximum overlapping data (in bytes) at next level for any // file at a level >= 1. int64_t TEST_MaxNextLevelOverlappingBytes(ColumnFamilyHandle* column_family = nullptr); // Return the current manifest file no. uint64_t TEST_Current_Manifest_FileNo(); // get total level0 file size. Only for testing. uint64_t TEST_GetLevel0TotalSize(); void TEST_GetFilesMetaData(ColumnFamilyHandle* column_family, std::vector>* metadata); void TEST_LockMutex(); void TEST_UnlockMutex(); // REQUIRES: mutex locked void* TEST_BeginWrite(); // REQUIRES: mutex locked // pass the pointer that you got from TEST_BeginWrite() void TEST_EndWrite(void* w); uint64_t TEST_MaxTotalInMemoryState() const { return max_total_in_memory_state_; } size_t TEST_LogsToFreeSize(); uint64_t TEST_LogfileNumber(); // Returns column family name to ImmutableCFOptions map. Status TEST_GetAllImmutableCFOptions( std::unordered_map* iopts_map); Cache* TEST_table_cache() { return table_cache_.get(); } WriteController& TEST_write_controler() { return write_controller_; } #endif // NDEBUG // Return maximum background compaction alowed to be scheduled based on // compaction status. int BGCompactionsAllowed() const; // Returns the list of live files in 'live' and the list // of all files in the filesystem in 'candidate_files'. // If force == false and the last call was less than // db_options_.delete_obsolete_files_period_micros microseconds ago, // it will not fill up the job_context void FindObsoleteFiles(JobContext* job_context, bool force, bool no_full_scan = false); // Diffs the files listed in filenames and those that do not // belong to live files are posibly removed. Also, removes all the // files in sst_delete_files and log_delete_files. // It is not necessary to hold the mutex when invoking this method. void PurgeObsoleteFiles(const JobContext& background_contet); ColumnFamilyHandle* DefaultColumnFamily() const override; const SnapshotList& snapshots() const { return snapshots_; } void CancelAllBackgroundWork(bool wait); // Find Super version and reference it. Based on options, it might return // the thread local cached one. // Call ReturnAndCleanupSuperVersion() when it is no longer needed. SuperVersion* GetAndRefSuperVersion(ColumnFamilyData* cfd); // Similar to the previous function but looks up based on a column family id. // nullptr will be returned if this column family no longer exists. // REQUIRED: this function should only be called on the write thread or if the // mutex is held. SuperVersion* GetAndRefSuperVersion(uint32_t column_family_id); // Same as above, should called without mutex held and not on write thread. SuperVersion* GetAndRefSuperVersionUnlocked(uint32_t column_family_id); // Un-reference the super version and return it to thread local cache if // needed. If it is the last reference of the super version. Clean it up // after un-referencing it. void ReturnAndCleanupSuperVersion(ColumnFamilyData* cfd, SuperVersion* sv); // Similar to the previous function but looks up based on a column family id. // nullptr will be returned if this column family no longer exists. // REQUIRED: this function should only be called on the write thread. void ReturnAndCleanupSuperVersion(uint32_t colun_family_id, SuperVersion* sv); // Same as above, should called without mutex held and not on write thread. void ReturnAndCleanupSuperVersionUnlocked(uint32_t colun_family_id, SuperVersion* sv); // REQUIRED: this function should only be called on the write thread or if the // mutex is held. Return value only valid until next call to this function or // mutex is released. ColumnFamilyHandle* GetColumnFamilyHandle(uint32_t column_family_id); // Same as above, should called without mutex held and not on write thread. ColumnFamilyHandle* GetColumnFamilyHandleUnlocked(uint32_t column_family_id); // Returns the number of currently running flushes. // REQUIREMENT: mutex_ must be held when calling this function. int num_running_flushes() { mutex_.AssertHeld(); return num_running_flushes_; } // Returns the number of currently running compactions. // REQUIREMENT: mutex_ must be held when calling this function. int num_running_compactions() { mutex_.AssertHeld(); return num_running_compactions_; } protected: Env* const env_; const std::string dbname_; unique_ptr versions_; const DBOptions db_options_; Statistics* stats_; InternalIterator* NewInternalIterator(const ReadOptions&, ColumnFamilyData* cfd, SuperVersion* super_version, Arena* arena); // Except in DB::Open(), WriteOptionsFile can only be called when: // 1. WriteThread::Writer::EnterUnbatched() is used. // 2. db_mutex is held Status WriteOptionsFile(); // The following two functions can only be called when: // 1. WriteThread::Writer::EnterUnbatched() is used. // 2. db_mutex is NOT held Status RenameTempFileToOptionsFile(const std::string& file_name); Status DeleteObsoleteOptionsFiles(); void NotifyOnFlushCompleted(ColumnFamilyData* cfd, FileMetaData* file_meta, const MutableCFOptions& mutable_cf_options, int job_id, TableProperties prop); void NotifyOnCompactionCompleted(ColumnFamilyData* cfd, Compaction *c, const Status &st, const CompactionJobStats& job_stats, int job_id); void NewThreadStatusCfInfo(ColumnFamilyData* cfd) const; void EraseThreadStatusCfInfo(ColumnFamilyData* cfd) const; void EraseThreadStatusDbInfo() const; Status WriteImpl(const WriteOptions& options, WriteBatch* updates, WriteCallback* callback); private: friend class DB; friend class InternalStats; #ifndef ROCKSDB_LITE friend class ForwardIterator; #endif friend struct SuperVersion; friend class CompactedDBImpl; #ifndef NDEBUG friend class XFTransactionWriteHandler; #endif struct CompactionState; struct WriteContext; Status NewDB(); // Recover the descriptor from persistent storage. May do a significant // amount of work to recover recently logged updates. Any changes to // be made to the descriptor are added to *edit. Status Recover(const std::vector& column_families, bool read_only = false, bool error_if_log_file_exist = false); void MaybeIgnoreError(Status* s) const; const Status CreateArchivalDirectory(); // Delete any unneeded files and stale in-memory entries. void DeleteObsoleteFiles(); // Background process needs to call // auto x = CaptureCurrentFileNumberInPendingOutputs() // // ReleaseFileNumberFromPendingOutputs(x) // This will protect any temporary files created while is // executing from being deleted. // ----------- // This function will capture current file number and append it to // pending_outputs_. This will prevent any background process to delete any // file created after this point. std::list::iterator CaptureCurrentFileNumberInPendingOutputs(); // This function should be called with the result of // CaptureCurrentFileNumberInPendingOutputs(). It then marks that any file // created between the calls CaptureCurrentFileNumberInPendingOutputs() and // ReleaseFileNumberFromPendingOutputs() can now be deleted (if it's not live // and blocked by any other pending_outputs_ calls) void ReleaseFileNumberFromPendingOutputs(std::list::iterator v); // Flush the in-memory write buffer to storage. Switches to a new // log-file/memtable and writes a new descriptor iff successful. Status FlushMemTableToOutputFile(ColumnFamilyData* cfd, const MutableCFOptions& mutable_cf_options, bool* madeProgress, JobContext* job_context, LogBuffer* log_buffer); // REQUIRES: log_numbers are sorted in ascending order Status RecoverLogFiles(const std::vector& log_numbers, SequenceNumber* max_sequence, bool read_only); // The following two methods are used to flush a memtable to // storage. The first one is used atdatabase RecoveryTime (when the // database is opened) and is heavyweight because it holds the mutex // for the entire period. The second method WriteLevel0Table supports // concurrent flush memtables to storage. Status WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd, MemTable* mem, VersionEdit* edit); // num_bytes: for slowdown case, delay time is calculated based on // `num_bytes` going through. Status DelayWrite(uint64_t num_bytes); Status ScheduleFlushes(WriteContext* context); Status SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context); // Force current memtable contents to be flushed. Status FlushMemTable(ColumnFamilyData* cfd, const FlushOptions& options); // Wait for memtable flushed Status WaitForFlushMemTable(ColumnFamilyData* cfd); void RecordFlushIOStats(); void RecordCompactionIOStats(); #ifndef ROCKSDB_LITE Status CompactFilesImpl( const CompactionOptions& compact_options, ColumnFamilyData* cfd, Version* version, const std::vector& input_file_names, const int output_level, int output_path_id, JobContext* job_context, LogBuffer* log_buffer); #endif // ROCKSDB_LITE ColumnFamilyData* GetColumnFamilyDataByName(const std::string& cf_name); void MaybeScheduleFlushOrCompaction(); void SchedulePendingFlush(ColumnFamilyData* cfd); void SchedulePendingCompaction(ColumnFamilyData* cfd); static void BGWorkCompaction(void* arg); static void BGWorkFlush(void* db); static void UnscheduleCallback(void* arg); void BackgroundCallCompaction(void* arg); void BackgroundCallFlush(); Status BackgroundCompaction(bool* madeProgress, JobContext* job_context, LogBuffer* log_buffer, void* m = 0); Status BackgroundFlush(bool* madeProgress, JobContext* job_context, LogBuffer* log_buffer); void PrintStatistics(); // dump rocksdb.stats to LOG void MaybeDumpStats(); // Return the minimum empty level that could hold the total data in the // input level. Return the input level, if such level could not be found. int FindMinimumEmptyLevelFitting(ColumnFamilyData* cfd, const MutableCFOptions& mutable_cf_options, int level); // Move the files in the input level to the target level. // If target_level < 0, automatically calculate the minimum level that could // hold the data set. Status ReFitLevel(ColumnFamilyData* cfd, int level, int target_level = -1); // helper functions for adding and removing from flush & compaction queues void AddToCompactionQueue(ColumnFamilyData* cfd); ColumnFamilyData* PopFirstFromCompactionQueue(); void AddToFlushQueue(ColumnFamilyData* cfd); ColumnFamilyData* PopFirstFromFlushQueue(); // helper function to call after some of the logs_ were synced void MarkLogsSynced(uint64_t up_to, bool synced_dir, const Status& status); const Snapshot* GetSnapshotImpl(bool is_write_conflict_boundary); // table_cache_ provides its own synchronization std::shared_ptr table_cache_; // Lock over the persistent DB state. Non-nullptr iff successfully acquired. FileLock* db_lock_; // The mutex for options file related operations. // NOTE: should never acquire options_file_mutex_ and mutex_ at the // same time. InstrumentedMutex options_files_mutex_; // State below is protected by mutex_ InstrumentedMutex mutex_; std::atomic shutting_down_; // This condition variable is signaled on these conditions: // * whenever bg_compaction_scheduled_ goes down to 0 // * if AnyManualCompaction, whenever a compaction finishes, even if it hasn't // made any progress // * whenever a compaction made any progress // * whenever bg_flush_scheduled_ value decreases (i.e. whenever a flush is // done, even if it didn't make any progress) // * whenever there is an error in background flush or compaction InstrumentedCondVar bg_cv_; uint64_t logfile_number_; std::deque log_recycle_files; // a list of log files that we can recycle bool log_dir_synced_; bool log_empty_; ColumnFamilyHandleImpl* default_cf_handle_; InternalStats* default_cf_internal_stats_; unique_ptr column_family_memtables_; struct LogFileNumberSize { explicit LogFileNumberSize(uint64_t _number) : number(_number) {} void AddSize(uint64_t new_size) { size += new_size; } uint64_t number; uint64_t size = 0; bool getting_flushed = false; }; struct LogWriterNumber { // pass ownership of _writer LogWriterNumber(uint64_t _number, log::Writer* _writer) : number(_number), writer(_writer) {} log::Writer* ReleaseWriter() { auto* w = writer; writer = nullptr; return w; } void ClearWriter() { delete writer; writer = nullptr; } uint64_t number; // Visual Studio doesn't support deque's member to be noncopyable because // of a unique_ptr as a member. log::Writer* writer; // own // true for some prefix of logs_ bool getting_synced = false; }; std::deque alive_log_files_; // Log files that aren't fully synced, and the current log file. // Synchronization: // - push_back() is done from write thread with locked mutex_, // - pop_front() is done from any thread with locked mutex_, // - back() and items with getting_synced=true are not popped, // - it follows that write thread with unlocked mutex_ can safely access // back() and items with getting_synced=true. std::deque logs_; // Signaled when getting_synced becomes false for some of the logs_. InstrumentedCondVar log_sync_cv_; uint64_t total_log_size_; // only used for dynamically adjusting max_total_wal_size. it is a sum of // [write_buffer_size * max_write_buffer_number] over all column families uint64_t max_total_in_memory_state_; // If true, we have only one (default) column family. We use this to optimize // some code-paths bool single_column_family_mode_; // If this is non-empty, we need to delete these log files in background // threads. Protected by db mutex. autovector logs_to_free_; bool is_snapshot_supported_; // Class to maintain directories for all database paths other than main one. class Directories { public: Status SetDirectories(Env* env, const std::string& dbname, const std::string& wal_dir, const std::vector& data_paths); Directory* GetDataDir(size_t path_id); Directory* GetWalDir() { if (wal_dir_) { return wal_dir_.get(); } return db_dir_.get(); } Directory* GetDbDir() { return db_dir_.get(); } private: std::unique_ptr db_dir_; std::vector> data_dirs_; std::unique_ptr wal_dir_; Status CreateAndNewDirectory(Env* env, const std::string& dirname, std::unique_ptr* directory) const; }; Directories directories_; WriteBuffer write_buffer_; WriteThread write_thread_; WriteBatch tmp_batch_; WriteController write_controller_; // Size of the last batch group. In slowdown mode, next write needs to // sleep if it uses up the quota. uint64_t last_batch_group_size_; FlushScheduler flush_scheduler_; SnapshotList snapshots_; // For each background job, pending_outputs_ keeps the current file number at // the time that background job started. // FindObsoleteFiles()/PurgeObsoleteFiles() never deletes any file that has // number bigger than any of the file number in pending_outputs_. Since file // numbers grow monotonically, this also means that pending_outputs_ is always // sorted. After a background job is done executing, its file number is // deleted from pending_outputs_, which allows PurgeObsoleteFiles() to clean // it up. // State is protected with db mutex. std::list pending_outputs_; // flush_queue_ and compaction_queue_ hold column families that we need to // flush and compact, respectively. // A column family is inserted into flush_queue_ when it satisfies condition // cfd->imm()->IsFlushPending() // A column family is inserted into compaction_queue_ when it satisfied // condition cfd->NeedsCompaction() // Column families in this list are all Ref()-erenced // TODO(icanadi) Provide some kind of ReferencedColumnFamily class that will // do RAII on ColumnFamilyData // Column families are in this queue when they need to be flushed or // compacted. Consumers of these queues are flush and compaction threads. When // column family is put on this queue, we increase unscheduled_flushes_ and // unscheduled_compactions_. When these variables are bigger than zero, that // means we need to schedule background threads for compaction and thread. // Once the background threads are scheduled, we decrease unscheduled_flushes_ // and unscheduled_compactions_. That way we keep track of number of // compaction and flush threads we need to schedule. This scheduling is done // in MaybeScheduleFlushOrCompaction() // invariant(column family present in flush_queue_ <==> // ColumnFamilyData::pending_flush_ == true) std::deque flush_queue_; // invariant(column family present in compaction_queue_ <==> // ColumnFamilyData::pending_compaction_ == true) std::deque compaction_queue_; int unscheduled_flushes_; int unscheduled_compactions_; // count how many background compactions are running or have been scheduled int bg_compaction_scheduled_; // stores the number of compactions are currently running int num_running_compactions_; // number of background memtable flush jobs, submitted to the HIGH pool int bg_flush_scheduled_; // stores the number of flushes are currently running int num_running_flushes_; // Information for a manual compaction struct ManualCompaction { ColumnFamilyData* cfd; int input_level; int output_level; uint32_t output_path_id; Status status; bool done; bool in_progress; // compaction request being processed? bool incomplete; // only part of requested range compacted bool exclusive; // current behavior of only one manual bool disallow_trivial_move; // Force actual compaction to run const InternalKey* begin; // nullptr means beginning of key range const InternalKey* end; // nullptr means end of key range InternalKey* manual_end; // how far we are compacting InternalKey tmp_storage; // Used to keep track of compaction progress InternalKey tmp_storage1; // Used to keep track of compaction progress Compaction* compaction; }; std::deque manual_compaction_dequeue_; struct CompactionArg { DBImpl* db; ManualCompaction* m; }; // Have we encountered a background error in paranoid mode? Status bg_error_; // shall we disable deletion of obsolete files // if 0 the deletion is enabled. // if non-zero, files will not be getting deleted // This enables two different threads to call // EnableFileDeletions() and DisableFileDeletions() // without any synchronization int disable_delete_obsolete_files_; // next time when we should run DeleteObsoleteFiles with full scan uint64_t delete_obsolete_files_next_run_; // last time stats were dumped to LOG std::atomic last_stats_dump_time_microsec_; // Each flush or compaction gets its own job id. this counter makes sure // they're unique std::atomic next_job_id_; // A flag indicating whether the current rocksdb database has any // data that is not yet persisted into either WAL or SST file. // Used when disableWAL is true. bool has_unpersisted_data_; static const int KEEP_LOG_FILE_NUM = 1000; // MSVC version 1800 still does not have constexpr for ::max() static const uint64_t kNoTimeOut = port::kMaxUint64; std::string db_absolute_path_; // The options to access storage files const EnvOptions env_options_; #ifndef ROCKSDB_LITE WalManager wal_manager_; #endif // ROCKSDB_LITE // Unified interface for logging events EventLogger event_logger_; // A value of >0 temporarily disables scheduling of background work int bg_work_paused_; // Guard against multiple concurrent refitting bool refitting_level_; // Indicate DB was opened successfully bool opened_successfully_; // No copying allowed DBImpl(const DBImpl&); void operator=(const DBImpl&); // Return the earliest snapshot where seqno is visible. // Store the snapshot right before that, if any, in prev_snapshot inline SequenceNumber findEarliestVisibleSnapshot( SequenceNumber in, std::vector& snapshots, SequenceNumber* prev_snapshot); // Background threads call this function, which is just a wrapper around // the InstallSuperVersion() function. Background threads carry // job_context which can have new_superversion already // allocated. void InstallSuperVersionAndScheduleWorkWrapper( ColumnFamilyData* cfd, JobContext* job_context, const MutableCFOptions& mutable_cf_options); // All ColumnFamily state changes go through this function. Here we analyze // the new state and we schedule background work if we detect that the new // state needs flush or compaction. SuperVersion* InstallSuperVersionAndScheduleWork( ColumnFamilyData* cfd, SuperVersion* new_sv, const MutableCFOptions& mutable_cf_options); #ifndef ROCKSDB_LITE using DB::GetPropertiesOfAllTables; virtual Status GetPropertiesOfAllTables(ColumnFamilyHandle* column_family, TablePropertiesCollection* props) override; virtual Status GetPropertiesOfTablesInRange( ColumnFamilyHandle* column_family, const Range* range, std::size_t n, TablePropertiesCollection* props) override; #endif // ROCKSDB_LITE // Function that Get and KeyMayExist call with no_io true or false // Note: 'value_found' from KeyMayExist propagates here Status GetImpl(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value, bool* value_found = nullptr); bool GetIntPropertyInternal(ColumnFamilyData* cfd, const DBPropertyInfo& property_info, bool is_locked, uint64_t* value); bool HasPendingManualCompaction(); bool HasExclusiveManualCompaction(); void AddManualCompaction(ManualCompaction* m); void RemoveManualCompaction(ManualCompaction* m); bool ShouldntRunManualCompaction(ManualCompaction* m); bool HaveManualCompaction(ColumnFamilyData* cfd); bool MCOverlap(ManualCompaction* m, ManualCompaction* m1); }; // Sanitize db options. The caller should delete result.info_log if // it is not equal to src.info_log. extern Options SanitizeOptions(const std::string& db, const InternalKeyComparator* icmp, const Options& src); extern DBOptions SanitizeOptions(const std::string& db, const DBOptions& src); // Fix user-supplied options to be reasonable template static void ClipToRange(T* ptr, V minvalue, V maxvalue) { if (static_cast(*ptr) > maxvalue) *ptr = maxvalue; if (static_cast(*ptr) < minvalue) *ptr = minvalue; } } // namespace rocksdb rocksdb-4.5.1/db/db_impl_debug.cc000066400000000000000000000112071267530634000166270ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef NDEBUG #include "db/db_impl.h" #include "util/thread_status_updater.h" namespace rocksdb { uint64_t DBImpl::TEST_GetLevel0TotalSize() { InstrumentedMutexLock l(&mutex_); return default_cf_handle_->cfd()->current()->storage_info()->NumLevelBytes(0); } int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes( ColumnFamilyHandle* column_family) { ColumnFamilyData* cfd; if (column_family == nullptr) { cfd = default_cf_handle_->cfd(); } else { auto cfh = reinterpret_cast(column_family); cfd = cfh->cfd(); } InstrumentedMutexLock l(&mutex_); return cfd->current()->storage_info()->MaxNextLevelOverlappingBytes(); } void DBImpl::TEST_GetFilesMetaData( ColumnFamilyHandle* column_family, std::vector>* metadata) { auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); InstrumentedMutexLock l(&mutex_); metadata->resize(NumberLevels()); for (int level = 0; level < NumberLevels(); level++) { const std::vector& files = cfd->current()->storage_info()->LevelFiles(level); (*metadata)[level].clear(); for (const auto& f : files) { (*metadata)[level].push_back(*f); } } } uint64_t DBImpl::TEST_Current_Manifest_FileNo() { return versions_->manifest_file_number(); } Status DBImpl::TEST_CompactRange(int level, const Slice* begin, const Slice* end, ColumnFamilyHandle* column_family, bool disallow_trivial_move) { ColumnFamilyData* cfd; if (column_family == nullptr) { cfd = default_cf_handle_->cfd(); } else { auto cfh = reinterpret_cast(column_family); cfd = cfh->cfd(); } int output_level = (cfd->ioptions()->compaction_style == kCompactionStyleUniversal || cfd->ioptions()->compaction_style == kCompactionStyleFIFO) ? level : level + 1; return RunManualCompaction(cfd, level, output_level, 0, begin, end, true, disallow_trivial_move); } Status DBImpl::TEST_FlushMemTable(bool wait) { FlushOptions fo; fo.wait = wait; return FlushMemTable(default_cf_handle_->cfd(), fo); } Status DBImpl::TEST_WaitForFlushMemTable(ColumnFamilyHandle* column_family) { ColumnFamilyData* cfd; if (column_family == nullptr) { cfd = default_cf_handle_->cfd(); } else { auto cfh = reinterpret_cast(column_family); cfd = cfh->cfd(); } return WaitForFlushMemTable(cfd); } Status DBImpl::TEST_WaitForCompact() { // Wait until the compaction completes // TODO: a bug here. This function actually does not necessarily // wait for compact. It actually waits for scheduled compaction // OR flush to finish. InstrumentedMutexLock l(&mutex_); while ((bg_compaction_scheduled_ || bg_flush_scheduled_) && bg_error_.ok()) { bg_cv_.Wait(); } return bg_error_; } void DBImpl::TEST_LockMutex() { mutex_.Lock(); } void DBImpl::TEST_UnlockMutex() { mutex_.Unlock(); } void* DBImpl::TEST_BeginWrite() { auto w = new WriteThread::Writer(); write_thread_.EnterUnbatched(w, &mutex_); return reinterpret_cast(w); } void DBImpl::TEST_EndWrite(void* w) { auto writer = reinterpret_cast(w); write_thread_.ExitUnbatched(writer); delete writer; } size_t DBImpl::TEST_LogsToFreeSize() { InstrumentedMutexLock l(&mutex_); return logs_to_free_.size(); } uint64_t DBImpl::TEST_LogfileNumber() { InstrumentedMutexLock l(&mutex_); return logfile_number_; } Status DBImpl::TEST_GetAllImmutableCFOptions( std::unordered_map* iopts_map) { std::vector cf_names; std::vector iopts; { InstrumentedMutexLock l(&mutex_); for (auto cfd : *versions_->GetColumnFamilySet()) { cf_names.push_back(cfd->GetName()); iopts.push_back(cfd->ioptions()); } } iopts_map->clear(); for (size_t i = 0; i < cf_names.size(); ++i) { iopts_map->insert({cf_names[i], iopts[i]}); } return Status::OK(); } } // namespace rocksdb #endif // NDEBUG rocksdb-4.5.1/db/db_impl_experimental.cc000066400000000000000000000125161267530634000202420ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_impl.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include "db/column_family.h" #include "db/job_context.h" #include "db/version_set.h" #include "rocksdb/status.h" namespace rocksdb { #ifndef ROCKSDB_LITE Status DBImpl::SuggestCompactRange(ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) { auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); InternalKey start_key, end_key; if (begin != nullptr) { start_key.SetMaxPossibleForUserKey(*begin); } if (end != nullptr) { end_key.SetMinPossibleForUserKey(*end); } { InstrumentedMutexLock l(&mutex_); auto vstorage = cfd->current()->storage_info(); for (int level = 0; level < vstorage->num_non_empty_levels() - 1; ++level) { std::vector inputs; vstorage->GetOverlappingInputs( level, begin == nullptr ? nullptr : &start_key, end == nullptr ? nullptr : &end_key, &inputs); for (auto f : inputs) { f->marked_for_compaction = true; } } // Since we have some more files to compact, we should also recompute // compaction score vstorage->ComputeCompactionScore(*cfd->GetLatestMutableCFOptions(), CompactionOptionsFIFO()); SchedulePendingCompaction(cfd); MaybeScheduleFlushOrCompaction(); } return Status::OK(); } Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) { assert(column_family); if (target_level < 1) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "PromoteL0 FAILED. Invalid target level %d\n", target_level); return Status::InvalidArgument("Invalid target level"); } Status status; VersionEdit edit; JobContext job_context(next_job_id_.fetch_add(1), true); { InstrumentedMutexLock l(&mutex_); auto* cfd = static_cast(column_family)->cfd(); const auto* vstorage = cfd->current()->storage_info(); if (target_level >= vstorage->num_levels()) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "PromoteL0 FAILED. Target level %d does not exist\n", target_level); job_context.Clean(); return Status::InvalidArgument("Target level does not exist"); } // Sort L0 files by range. const InternalKeyComparator* icmp = &cfd->internal_comparator(); auto l0_files = vstorage->LevelFiles(0); std::sort(l0_files.begin(), l0_files.end(), [icmp](FileMetaData* f1, FileMetaData* f2) { return icmp->Compare(f1->largest, f2->largest) < 0; }); // Check that no L0 file is being compacted and that they have // non-overlapping ranges. for (size_t i = 0; i < l0_files.size(); ++i) { auto f = l0_files[i]; if (f->being_compacted) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "PromoteL0 FAILED. File %" PRIu64 " being compacted\n", f->fd.GetNumber()); job_context.Clean(); return Status::InvalidArgument("PromoteL0 called during L0 compaction"); } if (i == 0) continue; auto prev_f = l0_files[i - 1]; if (icmp->Compare(prev_f->largest, f->smallest) >= 0) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "PromoteL0 FAILED. Files %" PRIu64 " and %" PRIu64 " have overlapping ranges\n", prev_f->fd.GetNumber(), f->fd.GetNumber()); job_context.Clean(); return Status::InvalidArgument("L0 has overlapping files"); } } // Check that all levels up to target_level are empty. for (int level = 1; level <= target_level; ++level) { if (vstorage->NumLevelFiles(level) > 0) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "PromoteL0 FAILED. Level %d not empty\n", level); job_context.Clean(); return Status::InvalidArgument( "All levels up to target_level " "must be empty"); } } edit.SetColumnFamily(cfd->GetID()); for (const auto& f : l0_files) { edit.DeleteFile(0, f->fd.GetNumber()); edit.AddFile(target_level, f->fd.GetNumber(), f->fd.GetPathId(), f->fd.GetFileSize(), f->smallest, f->largest, f->smallest_seqno, f->largest_seqno, f->marked_for_compaction); } status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(), &edit, &mutex_, directories_.GetDbDir()); if (status.ok()) { InstallSuperVersionAndScheduleWorkWrapper( cfd, &job_context, *cfd->GetLatestMutableCFOptions()); } } // lock released here LogFlush(db_options_.info_log); job_context.Clean(); return status; } #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/db_impl_readonly.cc000066400000000000000000000144271267530634000173650ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/db_impl_readonly.h" #include "db/compacted_db_impl.h" #include "db/db_impl.h" #include "db/merge_context.h" #include "db/db_iter.h" #include "util/perf_context_imp.h" namespace rocksdb { #ifndef ROCKSDB_LITE DBImplReadOnly::DBImplReadOnly(const DBOptions& db_options, const std::string& dbname) : DBImpl(db_options, dbname) { Log(INFO_LEVEL, db_options_.info_log, "Opening the db in read only mode"); LogFlush(db_options_.info_log); } DBImplReadOnly::~DBImplReadOnly() { } // Implementations of the DB interface Status DBImplReadOnly::Get(const ReadOptions& read_options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) { Status s; SequenceNumber snapshot = versions_->LastSequence(); auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); SuperVersion* super_version = cfd->GetSuperVersion(); MergeContext merge_context; LookupKey lkey(key, snapshot); if (super_version->mem->Get(lkey, value, &s, &merge_context)) { } else { PERF_TIMER_GUARD(get_from_output_files_time); super_version->current->Get(read_options, lkey, value, &s, &merge_context); } return s; } Iterator* DBImplReadOnly::NewIterator(const ReadOptions& read_options, ColumnFamilyHandle* column_family) { auto cfh = reinterpret_cast(column_family); auto cfd = cfh->cfd(); SuperVersion* super_version = cfd->GetSuperVersion()->Ref(); SequenceNumber latest_snapshot = versions_->LastSequence(); auto db_iter = NewArenaWrappedDbIterator( env_, *cfd->ioptions(), cfd->user_comparator(), (read_options.snapshot != nullptr ? reinterpret_cast( read_options.snapshot)->number_ : latest_snapshot), super_version->mutable_cf_options.max_sequential_skip_in_iterations); auto internal_iter = NewInternalIterator( read_options, cfd, super_version, db_iter->GetArena()); db_iter->SetIterUnderDBIter(internal_iter); return db_iter; } Status DBImplReadOnly::NewIterators( const ReadOptions& read_options, const std::vector& column_families, std::vector* iterators) { if (iterators == nullptr) { return Status::InvalidArgument("iterators not allowed to be nullptr"); } iterators->clear(); iterators->reserve(column_families.size()); SequenceNumber latest_snapshot = versions_->LastSequence(); for (auto cfh : column_families) { auto* cfd = reinterpret_cast(cfh)->cfd(); auto* sv = cfd->GetSuperVersion()->Ref(); auto* db_iter = NewArenaWrappedDbIterator( env_, *cfd->ioptions(), cfd->user_comparator(), (read_options.snapshot != nullptr ? reinterpret_cast( read_options.snapshot)->number_ : latest_snapshot), sv->mutable_cf_options.max_sequential_skip_in_iterations); auto* internal_iter = NewInternalIterator( read_options, cfd, sv, db_iter->GetArena()); db_iter->SetIterUnderDBIter(internal_iter); iterators->push_back(db_iter); } return Status::OK(); } Status DB::OpenForReadOnly(const Options& options, const std::string& dbname, DB** dbptr, bool error_if_log_file_exist) { *dbptr = nullptr; // Try to first open DB as fully compacted DB Status s; s = CompactedDBImpl::Open(options, dbname, dbptr); if (s.ok()) { return s; } DBOptions db_options(options); ColumnFamilyOptions cf_options(options); std::vector column_families; column_families.push_back( ColumnFamilyDescriptor(kDefaultColumnFamilyName, cf_options)); std::vector handles; s = DB::OpenForReadOnly(db_options, dbname, column_families, &handles, dbptr); if (s.ok()) { assert(handles.size() == 1); // i can delete the handle since DBImpl is always holding a // reference to default column family delete handles[0]; } return s; } Status DB::OpenForReadOnly( const DBOptions& db_options, const std::string& dbname, const std::vector& column_families, std::vector* handles, DB** dbptr, bool error_if_log_file_exist) { *dbptr = nullptr; handles->clear(); DBImplReadOnly* impl = new DBImplReadOnly(db_options, dbname); impl->mutex_.Lock(); Status s = impl->Recover(column_families, true /* read only */, error_if_log_file_exist); if (s.ok()) { // set column family handles for (auto cf : column_families) { auto cfd = impl->versions_->GetColumnFamilySet()->GetColumnFamily(cf.name); if (cfd == nullptr) { s = Status::InvalidArgument("Column family not found: ", cf.name); break; } handles->push_back(new ColumnFamilyHandleImpl(cfd, impl, &impl->mutex_)); } } if (s.ok()) { for (auto cfd : *impl->versions_->GetColumnFamilySet()) { delete cfd->InstallSuperVersion(new SuperVersion(), &impl->mutex_); } } impl->mutex_.Unlock(); if (s.ok()) { *dbptr = impl; for (auto* h : *handles) { impl->NewThreadStatusCfInfo( reinterpret_cast(h)->cfd()); } } else { for (auto h : *handles) { delete h; } handles->clear(); delete impl; } return s; } #else // !ROCKSDB_LITE Status DB::OpenForReadOnly(const Options& options, const std::string& dbname, DB** dbptr, bool error_if_log_file_exist) { return Status::NotSupported("Not supported in ROCKSDB_LITE."); } Status DB::OpenForReadOnly( const DBOptions& db_options, const std::string& dbname, const std::vector& column_families, std::vector* handles, DB** dbptr, bool error_if_log_file_exist) { return Status::NotSupported("Not supported in ROCKSDB_LITE."); } #endif // !ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/db_impl_readonly.h000066400000000000000000000103401267530634000172150ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include "db/db_impl.h" #include #include namespace rocksdb { class DBImplReadOnly : public DBImpl { public: DBImplReadOnly(const DBOptions& options, const std::string& dbname); virtual ~DBImplReadOnly(); // Implementations of the DB interface using DB::Get; virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) override; // TODO: Implement ReadOnly MultiGet? using DBImpl::NewIterator; virtual Iterator* NewIterator(const ReadOptions&, ColumnFamilyHandle* column_family) override; virtual Status NewIterators( const ReadOptions& options, const std::vector& column_families, std::vector* iterators) override; using DBImpl::Put; virtual Status Put(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override { return Status::NotSupported("Not supported operation in read only mode."); } using DBImpl::Merge; virtual Status Merge(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override { return Status::NotSupported("Not supported operation in read only mode."); } using DBImpl::Delete; virtual Status Delete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) override { return Status::NotSupported("Not supported operation in read only mode."); } using DBImpl::SingleDelete; virtual Status SingleDelete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) override { return Status::NotSupported("Not supported operation in read only mode."); } virtual Status Write(const WriteOptions& options, WriteBatch* updates) override { return Status::NotSupported("Not supported operation in read only mode."); } using DBImpl::CompactRange; virtual Status CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) override { return Status::NotSupported("Not supported operation in read only mode."); } using DBImpl::CompactFiles; virtual Status CompactFiles( const CompactionOptions& compact_options, ColumnFamilyHandle* column_family, const std::vector& input_file_names, const int output_level, const int output_path_id = -1) override { return Status::NotSupported("Not supported operation in read only mode."); } #ifndef ROCKSDB_LITE virtual Status DisableFileDeletions() override { return Status::NotSupported("Not supported operation in read only mode."); } virtual Status EnableFileDeletions(bool force) override { return Status::NotSupported("Not supported operation in read only mode."); } virtual Status GetLiveFiles(std::vector&, uint64_t* manifest_file_size, bool flush_memtable = true) override { return Status::NotSupported("Not supported operation in read only mode."); } #endif // ROCKSDB_LITE using DBImpl::Flush; virtual Status Flush(const FlushOptions& options, ColumnFamilyHandle* column_family) override { return Status::NotSupported("Not supported operation in read only mode."); } using DBImpl::SyncWAL; virtual Status SyncWAL() override { return Status::NotSupported("Not supported operation in read only mode."); } private: friend class DB; // No copying allowed DBImplReadOnly(const DBImplReadOnly&); void operator=(const DBImplReadOnly&); }; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/db_info_dumper.cc000066400000000000000000000073231267530634000170330ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include "db/db_info_dumper.h" #include #include #include #include #include #include "db/filename.h" #include "rocksdb/options.h" #include "rocksdb/env.h" namespace rocksdb { void DumpDBFileSummary(const DBOptions& options, const std::string& dbname) { if (options.info_log == nullptr) { return; } auto* env = options.env; uint64_t number = 0; FileType type = kInfoLogFile; std::vector files; uint64_t file_num = 0; uint64_t file_size; std::string file_info, wal_info; Header(options.info_log, "DB SUMMARY\n"); // Get files in dbname dir if (!env->GetChildren(dbname, &files).ok()) { Error(options.info_log, "Error when reading %s dir\n", dbname.c_str()); } std::sort(files.begin(), files.end()); for (std::string file : files) { if (!ParseFileName(file, &number, &type)) { continue; } switch (type) { case kCurrentFile: Header(options.info_log, "CURRENT file: %s\n", file.c_str()); break; case kIdentityFile: Header(options.info_log, "IDENTITY file: %s\n", file.c_str()); break; case kDescriptorFile: env->GetFileSize(dbname + "/" + file, &file_size); Header(options.info_log, "MANIFEST file: %s size: %" PRIu64 " Bytes\n", file.c_str(), file_size); break; case kLogFile: env->GetFileSize(dbname + "/" + file, &file_size); char str[16]; snprintf(str, sizeof(str), "%" PRIu64, file_size); wal_info.append(file).append(" size: "). append(str).append(" ; "); break; case kTableFile: if (++file_num < 10) { file_info.append(file).append(" "); } break; default: break; } } // Get sst files in db_path dir for (auto& db_path : options.db_paths) { if (dbname.compare(db_path.path) != 0) { if (!env->GetChildren(db_path.path, &files).ok()) { Error(options.info_log, "Error when reading %s dir\n", db_path.path.c_str()); continue; } std::sort(files.begin(), files.end()); for (std::string file : files) { if (ParseFileName(file, &number, &type)) { if (type == kTableFile && ++file_num < 10) { file_info.append(file).append(" "); } } } } Header(options.info_log, "SST files in %s dir, Total Num: %" PRIu64 ", files: %s\n", db_path.path.c_str(), file_num, file_info.c_str()); file_num = 0; file_info.clear(); } // Get wal file in wal_dir if (dbname.compare(options.wal_dir) != 0) { if (!env->GetChildren(options.wal_dir, &files).ok()) { Error(options.info_log, "Error when reading %s dir\n", options.wal_dir.c_str()); return; } wal_info.clear(); for (std::string file : files) { if (ParseFileName(file, &number, &type)) { if (type == kLogFile) { env->GetFileSize(options.wal_dir + "/" + file, &file_size); char str[16]; snprintf(str, sizeof(str), "%" PRIu64, file_size); wal_info.append(file).append(" size: "). append(str).append(" ; "); } } } } Header(options.info_log, "Write Ahead Log file in %s: %s\n", options.wal_dir.c_str(), wal_info.c_str()); } } // namespace rocksdb rocksdb-4.5.1/db/db_info_dumper.h000066400000000000000000000007341267530634000166740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include "rocksdb/options.h" namespace rocksdb { void DumpDBFileSummary(const DBOptions& options, const std::string& dbname); } // namespace rocksdb rocksdb-4.5.1/db/db_inplace_update_test.cc000066400000000000000000000125431267530634000205400ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_test_util.h" #include "port/stack_trace.h" namespace rocksdb { class DBTestInPlaceUpdate : public DBTestBase { public: DBTestInPlaceUpdate() : DBTestBase("/db_inplace_update_test") {} }; TEST_F(DBTestInPlaceUpdate, InPlaceUpdate) { do { Options options; options.create_if_missing = true; options.inplace_update_support = true; options.env = env_; options.write_buffer_size = 100000; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Update key with values of smaller size int numValues = 10; for (int i = numValues; i > 0; i--) { std::string value = DummyString(i, 'a'); ASSERT_OK(Put(1, "key", value)); ASSERT_EQ(value, Get(1, "key")); } // Only 1 instance for that key. validateNumberOfEntries(1, 1); } while (ChangeCompactOptions()); } TEST_F(DBTestInPlaceUpdate, InPlaceUpdateLargeNewValue) { do { Options options; options.create_if_missing = true; options.inplace_update_support = true; options.env = env_; options.write_buffer_size = 100000; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Update key with values of larger size int numValues = 10; for (int i = 0; i < numValues; i++) { std::string value = DummyString(i, 'a'); ASSERT_OK(Put(1, "key", value)); ASSERT_EQ(value, Get(1, "key")); } // All 10 updates exist in the internal iterator validateNumberOfEntries(numValues, 1); } while (ChangeCompactOptions()); } TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackSmallerSize) { do { Options options; options.create_if_missing = true; options.inplace_update_support = true; options.env = env_; options.write_buffer_size = 100000; options.inplace_callback = rocksdb::DBTestInPlaceUpdate::updateInPlaceSmallerSize; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Update key with values of smaller size int numValues = 10; ASSERT_OK(Put(1, "key", DummyString(numValues, 'a'))); ASSERT_EQ(DummyString(numValues, 'c'), Get(1, "key")); for (int i = numValues; i > 0; i--) { ASSERT_OK(Put(1, "key", DummyString(i, 'a'))); ASSERT_EQ(DummyString(i - 1, 'b'), Get(1, "key")); } // Only 1 instance for that key. validateNumberOfEntries(1, 1); } while (ChangeCompactOptions()); } TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackSmallerVarintSize) { do { Options options; options.create_if_missing = true; options.inplace_update_support = true; options.env = env_; options.write_buffer_size = 100000; options.inplace_callback = rocksdb::DBTestInPlaceUpdate::updateInPlaceSmallerVarintSize; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Update key with values of smaller varint size int numValues = 265; ASSERT_OK(Put(1, "key", DummyString(numValues, 'a'))); ASSERT_EQ(DummyString(numValues, 'c'), Get(1, "key")); for (int i = numValues; i > 0; i--) { ASSERT_OK(Put(1, "key", DummyString(i, 'a'))); ASSERT_EQ(DummyString(1, 'b'), Get(1, "key")); } // Only 1 instance for that key. validateNumberOfEntries(1, 1); } while (ChangeCompactOptions()); } TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackLargeNewValue) { do { Options options; options.create_if_missing = true; options.inplace_update_support = true; options.env = env_; options.write_buffer_size = 100000; options.inplace_callback = rocksdb::DBTestInPlaceUpdate::updateInPlaceLargerSize; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Update key with values of larger size int numValues = 10; for (int i = 0; i < numValues; i++) { ASSERT_OK(Put(1, "key", DummyString(i, 'a'))); ASSERT_EQ(DummyString(i, 'c'), Get(1, "key")); } // No inplace updates. All updates are puts with new seq number // All 10 updates exist in the internal iterator validateNumberOfEntries(numValues, 1); } while (ChangeCompactOptions()); } TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackNoAction) { do { Options options; options.create_if_missing = true; options.inplace_update_support = true; options.env = env_; options.write_buffer_size = 100000; options.inplace_callback = rocksdb::DBTestInPlaceUpdate::updateInPlaceNoAction; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Callback function requests no actions from db ASSERT_OK(Put(1, "key", DummyString(1, 'a'))); ASSERT_EQ(Get(1, "key"), "NOT_FOUND"); } while (ChangeCompactOptions()); } } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/db_iter.cc000066400000000000000000000701551267530634000154720ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_iter.h" #include #include #include #include #include "db/filename.h" #include "db/dbformat.h" #include "port/port.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/iterator.h" #include "rocksdb/merge_operator.h" #include "table/internal_iterator.h" #include "util/arena.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/perf_context_imp.h" namespace rocksdb { #if 0 static void DumpInternalIter(Iterator* iter) { for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ParsedInternalKey k; if (!ParseInternalKey(iter->key(), &k)) { fprintf(stderr, "Corrupt '%s'\n", EscapeString(iter->key()).c_str()); } else { fprintf(stderr, "@ '%s'\n", k.DebugString().c_str()); } } } #endif // Memtables and sstables that make the DB representation contain // (userkey,seq,type) => uservalue entries. DBIter // combines multiple entries for the same userkey found in the DB // representation into a single entry while accounting for sequence // numbers, deletion markers, overwrites, etc. class DBIter: public Iterator { public: // The following is grossly complicated. TODO: clean it up // Which direction is the iterator currently moving? // (1) When moving forward, the internal iterator is positioned at // the exact entry that yields this->key(), this->value() // (2) When moving backwards, the internal iterator is positioned // just before all entries whose user key == this->key(). enum Direction { kForward, kReverse }; DBIter(Env* env, const ImmutableCFOptions& ioptions, const Comparator* cmp, InternalIterator* iter, SequenceNumber s, bool arena_mode, uint64_t max_sequential_skip_in_iterations, const Slice* iterate_upper_bound = nullptr, bool prefix_same_as_start = false) : arena_mode_(arena_mode), env_(env), logger_(ioptions.info_log), user_comparator_(cmp), user_merge_operator_(ioptions.merge_operator), iter_(iter), sequence_(s), direction_(kForward), valid_(false), current_entry_is_merged_(false), statistics_(ioptions.statistics), iterate_upper_bound_(iterate_upper_bound), prefix_same_as_start_(prefix_same_as_start), iter_pinned_(false) { RecordTick(statistics_, NO_ITERATORS); prefix_extractor_ = ioptions.prefix_extractor; max_skip_ = max_sequential_skip_in_iterations; } virtual ~DBIter() { RecordTick(statistics_, NO_ITERATORS, -1); if (!arena_mode_) { delete iter_; } else { iter_->~InternalIterator(); } } virtual void SetIter(InternalIterator* iter) { assert(iter_ == nullptr); iter_ = iter; if (iter_ && iter_pinned_) { iter_->PinData(); } } virtual bool Valid() const override { return valid_; } virtual Slice key() const override { assert(valid_); return saved_key_.GetKey(); } virtual Slice value() const override { assert(valid_); return (direction_ == kForward && !current_entry_is_merged_) ? iter_->value() : saved_value_; } virtual Status status() const override { if (status_.ok()) { return iter_->status(); } else { return status_; } } virtual Status PinData() { Status s; if (iter_) { s = iter_->PinData(); } if (s.ok()) { // Even if iter_ is nullptr, we set iter_pinned_ to true so that when // iter_ is updated using SetIter, we Pin it. iter_pinned_ = true; } return s; } virtual Status ReleasePinnedData() { Status s; if (iter_) { s = iter_->ReleasePinnedData(); } if (s.ok()) { iter_pinned_ = false; } return s; } virtual bool IsKeyPinned() const override { assert(valid_); return iter_pinned_ && saved_key_.IsKeyPinned(); } virtual void Next() override; virtual void Prev() override; virtual void Seek(const Slice& target) override; virtual void SeekToFirst() override; virtual void SeekToLast() override; private: void ReverseToBackward(); void PrevInternal(); void FindParseableKey(ParsedInternalKey* ikey, Direction direction); bool FindValueForCurrentKey(); bool FindValueForCurrentKeyUsingSeek(); void FindPrevUserKey(); void FindNextUserKey(); inline void FindNextUserEntry(bool skipping); void FindNextUserEntryInternal(bool skipping); bool ParseKey(ParsedInternalKey* key); void MergeValuesNewToOld(); inline void ClearSavedValue() { if (saved_value_.capacity() > 1048576) { std::string empty; swap(empty, saved_value_); } else { saved_value_.clear(); } } const SliceTransform* prefix_extractor_; bool arena_mode_; Env* const env_; Logger* logger_; const Comparator* const user_comparator_; const MergeOperator* const user_merge_operator_; InternalIterator* iter_; SequenceNumber const sequence_; Status status_; IterKey saved_key_; std::string saved_value_; Direction direction_; bool valid_; bool current_entry_is_merged_; Statistics* statistics_; uint64_t max_skip_; const Slice* iterate_upper_bound_; IterKey prefix_start_; bool prefix_same_as_start_; bool iter_pinned_; // List of operands for merge operator. std::deque merge_operands_; // No copying allowed DBIter(const DBIter&); void operator=(const DBIter&); }; inline bool DBIter::ParseKey(ParsedInternalKey* ikey) { if (!ParseInternalKey(iter_->key(), ikey)) { status_ = Status::Corruption("corrupted internal key in DBIter"); Log(InfoLogLevel::ERROR_LEVEL, logger_, "corrupted internal key in DBIter: %s", iter_->key().ToString(true).c_str()); return false; } else { return true; } } void DBIter::Next() { assert(valid_); if (direction_ == kReverse) { FindNextUserKey(); direction_ = kForward; if (!iter_->Valid()) { iter_->SeekToFirst(); } } else if (iter_->Valid() && !current_entry_is_merged_) { // If the current value is not a merge, the iter position is the // current key, which is already returned. We can safely issue a // Next() without checking the current key. // If the current key is a merge, very likely iter already points // to the next internal position. iter_->Next(); PERF_COUNTER_ADD(internal_key_skipped_count, 1); } // Now we point to the next internal position, for both of merge and // not merge cases. if (!iter_->Valid()) { valid_ = false; return; } FindNextUserEntry(true /* skipping the current user key */); if (statistics_ != nullptr) { RecordTick(statistics_, NUMBER_DB_NEXT); if (valid_) { RecordTick(statistics_, NUMBER_DB_NEXT_FOUND); RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size()); } } if (valid_ && prefix_extractor_ && prefix_same_as_start_ && prefix_extractor_->Transform(saved_key_.GetKey()) .compare(prefix_start_.GetKey()) != 0) { valid_ = false; } } // PRE: saved_key_ has the current user key if skipping // POST: saved_key_ should have the next user key if valid_, // if the current entry is a result of merge // current_entry_is_merged_ => true // saved_value_ => the merged value // // NOTE: In between, saved_key_ can point to a user key that has // a delete marker inline void DBIter::FindNextUserEntry(bool skipping) { PERF_TIMER_GUARD(find_next_user_entry_time); FindNextUserEntryInternal(skipping); } // Actual implementation of DBIter::FindNextUserEntry() void DBIter::FindNextUserEntryInternal(bool skipping) { // Loop until we hit an acceptable entry to yield assert(iter_->Valid()); assert(direction_ == kForward); current_entry_is_merged_ = false; uint64_t num_skipped = 0; do { ParsedInternalKey ikey; if (ParseKey(&ikey)) { if (iterate_upper_bound_ != nullptr && ikey.user_key.compare(*iterate_upper_bound_) >= 0) { break; } if (ikey.sequence <= sequence_) { if (skipping && user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0) { num_skipped++; // skip this entry PERF_COUNTER_ADD(internal_key_skipped_count, 1); } else { switch (ikey.type) { case kTypeDeletion: case kTypeSingleDeletion: // Arrange to skip all upcoming entries for this key since // they are hidden by this deletion. saved_key_.SetKey(ikey.user_key, !iter_->IsKeyPinned() /* copy */); skipping = true; num_skipped = 0; PERF_COUNTER_ADD(internal_delete_skipped_count, 1); break; case kTypeValue: valid_ = true; saved_key_.SetKey(ikey.user_key, !iter_->IsKeyPinned() /* copy */); return; case kTypeMerge: // By now, we are sure the current ikey is going to yield a value saved_key_.SetKey(ikey.user_key, !iter_->IsKeyPinned() /* copy */); current_entry_is_merged_ = true; valid_ = true; MergeValuesNewToOld(); // Go to a different state machine return; default: assert(false); break; } } } } // If we have sequentially iterated via numerous keys and still not // found the next user-key, then it is better to seek so that we can // avoid too many key comparisons. We seek to the last occurrence of // our current key by looking for sequence number 0 and type deletion // (the smallest type). if (skipping && num_skipped > max_skip_) { num_skipped = 0; std::string last_key; AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), 0, kTypeDeletion)); iter_->Seek(last_key); RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION); } else { iter_->Next(); } } while (iter_->Valid()); valid_ = false; } // Merge values of the same user key starting from the current iter_ position // Scan from the newer entries to older entries. // PRE: iter_->key() points to the first merge type entry // saved_key_ stores the user key // POST: saved_value_ has the merged value for the user key // iter_ points to the next entry (or invalid) void DBIter::MergeValuesNewToOld() { if (!user_merge_operator_) { Log(InfoLogLevel::ERROR_LEVEL, logger_, "Options::merge_operator is null."); status_ = Status::InvalidArgument("user_merge_operator_ must be set."); valid_ = false; return; } // Start the merge process by pushing the first operand std::deque operands; operands.push_front(iter_->value().ToString()); ParsedInternalKey ikey; for (iter_->Next(); iter_->Valid(); iter_->Next()) { if (!ParseKey(&ikey)) { // skip corrupted key continue; } if (!user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) { // hit the next user key, stop right here break; } else if (kTypeDeletion == ikey.type || kTypeSingleDeletion == ikey.type) { // hit a delete with the same user key, stop right here // iter_ is positioned after delete iter_->Next(); break; } else if (kTypeValue == ikey.type) { // hit a put, merge the put value with operands and store the // final result in saved_value_. We are done! // ignore corruption if there is any. const Slice val = iter_->value(); { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); user_merge_operator_->FullMerge(ikey.user_key, &val, operands, &saved_value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } // iter_ is positioned after put iter_->Next(); return; } else if (kTypeMerge == ikey.type) { // hit a merge, add the value as an operand and run associative merge. // when complete, add result to operands and continue. const Slice& val = iter_->value(); operands.push_front(val.ToString()); } else { assert(false); } } { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); // we either exhausted all internal keys under this user key, or hit // a deletion marker. // feed null as the existing value to the merge operator, such that // client can differentiate this scenario and do things accordingly. user_merge_operator_->FullMerge(saved_key_.GetKey(), nullptr, operands, &saved_value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } } void DBIter::Prev() { assert(valid_); if (direction_ == kForward) { ReverseToBackward(); } PrevInternal(); if (statistics_ != nullptr) { RecordTick(statistics_, NUMBER_DB_PREV); if (valid_) { RecordTick(statistics_, NUMBER_DB_PREV_FOUND); RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size()); } } if (valid_ && prefix_extractor_ && prefix_same_as_start_ && prefix_extractor_->Transform(saved_key_.GetKey()) .compare(prefix_start_.GetKey()) != 0) { valid_ = false; } } void DBIter::ReverseToBackward() { if (current_entry_is_merged_) { // Not placed in the same key. Need to call Prev() until finding the // previous key. if (!iter_->Valid()) { iter_->SeekToLast(); } ParsedInternalKey ikey; FindParseableKey(&ikey, kReverse); while (iter_->Valid() && user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) > 0) { iter_->Prev(); FindParseableKey(&ikey, kReverse); } } #ifndef NDEBUG if (iter_->Valid()) { ParsedInternalKey ikey; assert(ParseKey(&ikey)); assert(user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0); } #endif FindPrevUserKey(); direction_ = kReverse; } void DBIter::PrevInternal() { if (!iter_->Valid()) { valid_ = false; return; } ParsedInternalKey ikey; while (iter_->Valid()) { saved_key_.SetKey(ExtractUserKey(iter_->key()), !iter_->IsKeyPinned() /* copy */); if (FindValueForCurrentKey()) { valid_ = true; if (!iter_->Valid()) { return; } FindParseableKey(&ikey, kReverse); if (user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) { FindPrevUserKey(); } return; } if (!iter_->Valid()) { break; } FindParseableKey(&ikey, kReverse); if (user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) { FindPrevUserKey(); } } // We haven't found any key - iterator is not valid assert(!iter_->Valid()); valid_ = false; } // This function checks, if the entry with biggest sequence_number <= sequence_ // is non kTypeDeletion or kTypeSingleDeletion. If it's not, we save value in // saved_value_ bool DBIter::FindValueForCurrentKey() { assert(iter_->Valid()); merge_operands_.clear(); // last entry before merge (could be kTypeDeletion, kTypeSingleDeletion or // kTypeValue) ValueType last_not_merge_type = kTypeDeletion; ValueType last_key_entry_type = kTypeDeletion; ParsedInternalKey ikey; FindParseableKey(&ikey, kReverse); size_t num_skipped = 0; while (iter_->Valid() && ikey.sequence <= sequence_ && user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) { // We iterate too much: let's use Seek() to avoid too much key comparisons if (num_skipped >= max_skip_) { return FindValueForCurrentKeyUsingSeek(); } last_key_entry_type = ikey.type; switch (last_key_entry_type) { case kTypeValue: merge_operands_.clear(); saved_value_ = iter_->value().ToString(); last_not_merge_type = kTypeValue; break; case kTypeDeletion: case kTypeSingleDeletion: merge_operands_.clear(); last_not_merge_type = last_key_entry_type; PERF_COUNTER_ADD(internal_delete_skipped_count, 1); break; case kTypeMerge: assert(user_merge_operator_ != nullptr); merge_operands_.push_back(iter_->value().ToString()); break; default: assert(false); } PERF_COUNTER_ADD(internal_key_skipped_count, 1); assert(user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())); iter_->Prev(); ++num_skipped; FindParseableKey(&ikey, kReverse); } switch (last_key_entry_type) { case kTypeDeletion: case kTypeSingleDeletion: valid_ = false; return false; case kTypeMerge: if (last_not_merge_type == kTypeDeletion) { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); user_merge_operator_->FullMerge(saved_key_.GetKey(), nullptr, merge_operands_, &saved_value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } else { assert(last_not_merge_type == kTypeValue); std::string last_put_value = saved_value_; Slice temp_slice(last_put_value); { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); user_merge_operator_->FullMerge(saved_key_.GetKey(), &temp_slice, merge_operands_, &saved_value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } } break; case kTypeValue: // do nothing - we've already has value in saved_value_ break; default: assert(false); break; } valid_ = true; return true; } // This function is used in FindValueForCurrentKey. // We use Seek() function instead of Prev() to find necessary value bool DBIter::FindValueForCurrentKeyUsingSeek() { std::string last_key; AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), sequence_, kValueTypeForSeek)); iter_->Seek(last_key); RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION); // assume there is at least one parseable key for this user key ParsedInternalKey ikey; FindParseableKey(&ikey, kForward); if (ikey.type == kTypeValue || ikey.type == kTypeDeletion || ikey.type == kTypeSingleDeletion) { if (ikey.type == kTypeValue) { saved_value_ = iter_->value().ToString(); valid_ = true; return true; } valid_ = false; return false; } // kTypeMerge. We need to collect all kTypeMerge values and save them // in operands std::deque operands; while (iter_->Valid() && user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()) && ikey.type == kTypeMerge) { operands.push_front(iter_->value().ToString()); iter_->Next(); FindParseableKey(&ikey, kForward); } if (!iter_->Valid() || !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()) || ikey.type == kTypeDeletion || ikey.type == kTypeSingleDeletion) { { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); user_merge_operator_->FullMerge(saved_key_.GetKey(), nullptr, operands, &saved_value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } // Make iter_ valid and point to saved_key_ if (!iter_->Valid() || !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) { iter_->Seek(last_key); RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION); } valid_ = true; return true; } const Slice& val = iter_->value(); { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); user_merge_operator_->FullMerge(saved_key_.GetKey(), &val, operands, &saved_value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } valid_ = true; return true; } // Used in Next to change directions // Go to next user key // Don't use Seek(), // because next user key will be very close void DBIter::FindNextUserKey() { if (!iter_->Valid()) { return; } ParsedInternalKey ikey; FindParseableKey(&ikey, kForward); while (iter_->Valid() && !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) { iter_->Next(); FindParseableKey(&ikey, kForward); } } // Go to previous user_key void DBIter::FindPrevUserKey() { if (!iter_->Valid()) { return; } size_t num_skipped = 0; ParsedInternalKey ikey; FindParseableKey(&ikey, kReverse); int cmp; while (iter_->Valid() && ((cmp = user_comparator_->Compare( ikey.user_key, saved_key_.GetKey())) == 0 || (cmp > 0 && ikey.sequence > sequence_))) { if (cmp == 0) { if (num_skipped >= max_skip_) { num_skipped = 0; IterKey last_key; last_key.SetInternalKey(ParsedInternalKey( saved_key_.GetKey(), kMaxSequenceNumber, kValueTypeForSeek)); iter_->Seek(last_key.GetKey()); RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION); } else { ++num_skipped; } } iter_->Prev(); FindParseableKey(&ikey, kReverse); } } // Skip all unparseable keys void DBIter::FindParseableKey(ParsedInternalKey* ikey, Direction direction) { while (iter_->Valid() && !ParseKey(ikey)) { if (direction == kReverse) { iter_->Prev(); } else { iter_->Next(); } } } void DBIter::Seek(const Slice& target) { StopWatch sw(env_, statistics_, DB_SEEK); saved_key_.Clear(); // now savved_key is used to store internal key. saved_key_.SetInternalKey(target, sequence_); { PERF_TIMER_GUARD(seek_internal_seek_time); iter_->Seek(saved_key_.GetKey()); } RecordTick(statistics_, NUMBER_DB_SEEK); if (iter_->Valid()) { direction_ = kForward; ClearSavedValue(); FindNextUserEntry(false /* not skipping */); if (statistics_ != nullptr) { if (valid_) { RecordTick(statistics_, NUMBER_DB_SEEK_FOUND); RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size()); } } } else { valid_ = false; } if (valid_ && prefix_extractor_ && prefix_same_as_start_) { prefix_start_.SetKey(prefix_extractor_->Transform(target)); } } void DBIter::SeekToFirst() { // Don't use iter_::Seek() if we set a prefix extractor // because prefix seek will be used. if (prefix_extractor_ != nullptr) { max_skip_ = std::numeric_limits::max(); } direction_ = kForward; ClearSavedValue(); { PERF_TIMER_GUARD(seek_internal_seek_time); iter_->SeekToFirst(); } RecordTick(statistics_, NUMBER_DB_SEEK); if (iter_->Valid()) { FindNextUserEntry(false /* not skipping */); if (statistics_ != nullptr) { if (valid_) { RecordTick(statistics_, NUMBER_DB_SEEK_FOUND); RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size()); } } } else { valid_ = false; } if (valid_ && prefix_extractor_ && prefix_same_as_start_) { prefix_start_.SetKey(prefix_extractor_->Transform(saved_key_.GetKey())); } } void DBIter::SeekToLast() { // Don't use iter_::Seek() if we set a prefix extractor // because prefix seek will be used. if (prefix_extractor_ != nullptr) { max_skip_ = std::numeric_limits::max(); } direction_ = kReverse; ClearSavedValue(); { PERF_TIMER_GUARD(seek_internal_seek_time); iter_->SeekToLast(); } // When the iterate_upper_bound is set to a value, // it will seek to the last key before the // ReadOptions.iterate_upper_bound if (iter_->Valid() && iterate_upper_bound_ != nullptr) { saved_key_.SetKey(*iterate_upper_bound_, false /* copy */); std::string last_key; AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), kMaxSequenceNumber, kValueTypeForSeek)); iter_->Seek(last_key); if (!iter_->Valid()) { iter_->SeekToLast(); } else { iter_->Prev(); if (!iter_->Valid()) { valid_ = false; return; } } } PrevInternal(); if (statistics_ != nullptr) { RecordTick(statistics_, NUMBER_DB_SEEK); if (valid_) { RecordTick(statistics_, NUMBER_DB_SEEK_FOUND); RecordTick(statistics_, ITER_BYTES_READ, key().size() + value().size()); } } if (valid_ && prefix_extractor_ && prefix_same_as_start_) { prefix_start_.SetKey(prefix_extractor_->Transform(saved_key_.GetKey())); } } Iterator* NewDBIterator(Env* env, const ImmutableCFOptions& ioptions, const Comparator* user_key_comparator, InternalIterator* internal_iter, const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations, const Slice* iterate_upper_bound, bool prefix_same_as_start, bool pin_data) { DBIter* db_iter = new DBIter(env, ioptions, user_key_comparator, internal_iter, sequence, false, max_sequential_skip_in_iterations, iterate_upper_bound, prefix_same_as_start); if (pin_data) { db_iter->PinData(); } return db_iter; } ArenaWrappedDBIter::~ArenaWrappedDBIter() { db_iter_->~DBIter(); } void ArenaWrappedDBIter::SetDBIter(DBIter* iter) { db_iter_ = iter; } void ArenaWrappedDBIter::SetIterUnderDBIter(InternalIterator* iter) { static_cast(db_iter_)->SetIter(iter); } inline bool ArenaWrappedDBIter::Valid() const { return db_iter_->Valid(); } inline void ArenaWrappedDBIter::SeekToFirst() { db_iter_->SeekToFirst(); } inline void ArenaWrappedDBIter::SeekToLast() { db_iter_->SeekToLast(); } inline void ArenaWrappedDBIter::Seek(const Slice& target) { db_iter_->Seek(target); } inline void ArenaWrappedDBIter::Next() { db_iter_->Next(); } inline void ArenaWrappedDBIter::Prev() { db_iter_->Prev(); } inline Slice ArenaWrappedDBIter::key() const { return db_iter_->key(); } inline Slice ArenaWrappedDBIter::value() const { return db_iter_->value(); } inline Status ArenaWrappedDBIter::status() const { return db_iter_->status(); } inline Status ArenaWrappedDBIter::PinData() { return db_iter_->PinData(); } inline Status ArenaWrappedDBIter::ReleasePinnedData() { return db_iter_->ReleasePinnedData(); } inline bool ArenaWrappedDBIter::IsKeyPinned() const { return db_iter_->IsKeyPinned(); } void ArenaWrappedDBIter::RegisterCleanup(CleanupFunction function, void* arg1, void* arg2) { db_iter_->RegisterCleanup(function, arg1, arg2); } ArenaWrappedDBIter* NewArenaWrappedDbIterator( Env* env, const ImmutableCFOptions& ioptions, const Comparator* user_key_comparator, const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations, const Slice* iterate_upper_bound, bool prefix_same_as_start, bool pin_data) { ArenaWrappedDBIter* iter = new ArenaWrappedDBIter(); Arena* arena = iter->GetArena(); auto mem = arena->AllocateAligned(sizeof(DBIter)); DBIter* db_iter = new (mem) DBIter(env, ioptions, user_key_comparator, nullptr, sequence, true, max_sequential_skip_in_iterations, iterate_upper_bound, prefix_same_as_start); iter->SetDBIter(db_iter); if (pin_data) { iter->PinData(); } return iter; } } // namespace rocksdb rocksdb-4.5.1/db/db_iter.h000066400000000000000000000063151267530634000153310ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include "rocksdb/db.h" #include "rocksdb/iterator.h" #include "db/dbformat.h" #include "util/arena.h" #include "util/autovector.h" namespace rocksdb { class Arena; class DBIter; class InternalIterator; // Return a new iterator that converts internal keys (yielded by // "*internal_iter") that were live at the specified "sequence" number // into appropriate user keys. extern Iterator* NewDBIterator(Env* env, const ImmutableCFOptions& options, const Comparator* user_key_comparator, InternalIterator* internal_iter, const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations, const Slice* iterate_upper_bound = nullptr, bool prefix_same_as_start = false, bool pin_data = false); // A wrapper iterator which wraps DB Iterator and the arena, with which the DB // iterator is supposed be allocated. This class is used as an entry point of // a iterator hierarchy whose memory can be allocated inline. In that way, // accessing the iterator tree can be more cache friendly. It is also faster // to allocate. class ArenaWrappedDBIter : public Iterator { public: virtual ~ArenaWrappedDBIter(); // Get the arena to be used to allocate memory for DBIter to be wrapped, // as well as child iterators in it. virtual Arena* GetArena() { return &arena_; } // Set the DB Iterator to be wrapped virtual void SetDBIter(DBIter* iter); // Set the internal iterator wrapped inside the DB Iterator. Usually it is // a merging iterator. virtual void SetIterUnderDBIter(InternalIterator* iter); virtual bool Valid() const override; virtual void SeekToFirst() override; virtual void SeekToLast() override; virtual void Seek(const Slice& target) override; virtual void Next() override; virtual void Prev() override; virtual Slice key() const override; virtual Slice value() const override; virtual Status status() const override; void RegisterCleanup(CleanupFunction function, void* arg1, void* arg2); virtual Status PinData(); virtual Status ReleasePinnedData(); virtual bool IsKeyPinned() const override; private: DBIter* db_iter_; Arena arena_; }; // Generate the arena wrapped iterator class. extern ArenaWrappedDBIter* NewArenaWrappedDbIterator( Env* env, const ImmutableCFOptions& options, const Comparator* user_key_comparator, const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations, const Slice* iterate_upper_bound = nullptr, bool prefix_same_as_start = false, bool pin_data = false); } // namespace rocksdb rocksdb-4.5.1/db/db_iter_test.cc000066400000000000000000002326621267530634000165340ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include #include "db/db_iter.h" #include "db/dbformat.h" #include "rocksdb/comparator.h" #include "rocksdb/options.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice.h" #include "rocksdb/statistics.h" #include "table/iterator_wrapper.h" #include "table/merger.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/testharness.h" #include "utilities/merge_operators.h" namespace rocksdb { static uint64_t TestGetTickerCount(const Options& options, Tickers ticker_type) { return options.statistics->getTickerCount(ticker_type); } class TestIterator : public InternalIterator { public: explicit TestIterator(const Comparator* comparator) : initialized_(false), valid_(false), sequence_number_(0), iter_(0), cmp(comparator) {} void AddPut(std::string argkey, std::string argvalue) { Add(argkey, kTypeValue, argvalue); } void AddDeletion(std::string argkey) { Add(argkey, kTypeDeletion, std::string()); } void AddSingleDeletion(std::string argkey) { Add(argkey, kTypeSingleDeletion, std::string()); } void AddMerge(std::string argkey, std::string argvalue) { Add(argkey, kTypeMerge, argvalue); } void Add(std::string argkey, ValueType type, std::string argvalue) { Add(argkey, type, argvalue, sequence_number_++); } void Add(std::string argkey, ValueType type, std::string argvalue, size_t seq_num, bool update_iter = false) { valid_ = true; ParsedInternalKey internal_key(argkey, seq_num, type); data_.push_back( std::pair(std::string(), argvalue)); AppendInternalKey(&data_.back().first, internal_key); if (update_iter && valid_ && cmp.Compare(data_.back().first, key()) < 0) { // insert a key smaller than current key Finish(); // data_[iter_] is not anymore the current element of the iterator. // Increment it to reposition it to the right position. iter_++; } } // should be called before operations with iterator void Finish() { initialized_ = true; std::sort(data_.begin(), data_.end(), [this](std::pair a, std::pair b) { return (cmp.Compare(a.first, b.first) < 0); }); } virtual bool Valid() const override { assert(initialized_); return valid_; } virtual void SeekToFirst() override { assert(initialized_); valid_ = (data_.size() > 0); iter_ = 0; } virtual void SeekToLast() override { assert(initialized_); valid_ = (data_.size() > 0); iter_ = data_.size() - 1; } virtual void Seek(const Slice& target) override { assert(initialized_); SeekToFirst(); if (!valid_) { return; } while (iter_ < data_.size() && (cmp.Compare(data_[iter_].first, target) < 0)) { ++iter_; } if (iter_ == data_.size()) { valid_ = false; } } virtual void Next() override { assert(initialized_); if (data_.empty() || (iter_ == data_.size() - 1)) { valid_ = false; } else { ++iter_; } } virtual void Prev() override { assert(initialized_); if (iter_ == 0) { valid_ = false; } else { --iter_; } } virtual Slice key() const override { assert(initialized_); return data_[iter_].first; } virtual Slice value() const override { assert(initialized_); return data_[iter_].second; } virtual Status status() const override { assert(initialized_); return Status::OK(); } private: bool initialized_; bool valid_; size_t sequence_number_; size_t iter_; InternalKeyComparator cmp; std::vector> data_; }; class DBIteratorTest : public testing::Test { public: Env* env_; DBIteratorTest() : env_(Env::Default()) {} }; TEST_F(DBIteratorTest, DBIteratorPrevNext) { Options options; { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddDeletion("a"); internal_iter->AddDeletion("a"); internal_iter->AddDeletion("a"); internal_iter->AddDeletion("a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "val_a"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); } // Test to check the SeekToLast() with iterate_upper_bound not set { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("c", "val_c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); } // Test to check the SeekToLast() with iterate_upper_bound set { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("d", "val_d"); internal_iter->AddPut("e", "val_e"); internal_iter->AddPut("f", "val_f"); internal_iter->Finish(); Slice prefix("d"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); } // Test to check the SeekToLast() iterate_upper_bound set to a key that // is not Put yet { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("d", "val_d"); internal_iter->Finish(); Slice prefix("z"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "d"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "d"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); } // Test to check the SeekToLast() with iterate_upper_bound set to the // first key { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("b", "val_b"); internal_iter->Finish(); Slice prefix("a"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); db_iter->SeekToLast(); ASSERT_TRUE(!db_iter->Valid()); } // Test case to check SeekToLast with iterate_upper_bound set // (same key put may times - SeekToLast should start with the // maximum sequence id of the upper bound) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("c", "val_c"); internal_iter->AddPut("c", "val_c"); internal_iter->Finish(); Slice prefix("c"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 7, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); SetPerfLevel(kEnableCount); ASSERT_TRUE(GetPerfLevel() == kEnableCount); perf_context.Reset(); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(static_cast(perf_context.internal_key_skipped_count), 1); ASSERT_EQ(db_iter->key().ToString(), "b"); SetPerfLevel(kDisable); } // Test to check the SeekToLast() with the iterate_upper_bound set // (Checking the value of the key which has sequence ids greater than // and less that the iterator's sequence id) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a1"); internal_iter->AddPut("a", "val_a2"); internal_iter->AddPut("b", "val_b1"); internal_iter->AddPut("c", "val_c1"); internal_iter->AddPut("c", "val_c2"); internal_iter->AddPut("c", "val_c3"); internal_iter->AddPut("b", "val_b2"); internal_iter->AddPut("d", "val_d1"); internal_iter->Finish(); Slice prefix("c"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 4, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b1"); } // Test to check the SeekToLast() with the iterate_upper_bound set to the // key that is deleted { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddDeletion("a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("c", "val_c"); internal_iter->Finish(); Slice prefix("a"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); db_iter->SeekToLast(); ASSERT_TRUE(!db_iter->Valid()); } // Test to check the SeekToLast() with the iterate_upper_bound set // (Deletion cases) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddDeletion("b"); internal_iter->AddPut("c", "val_c"); internal_iter->Finish(); Slice prefix("c"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); } // Test to check the SeekToLast() with iterate_upper_bound set // (Deletion cases - Lot of internal keys after the upper_bound // is deleted) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddDeletion("c"); internal_iter->AddDeletion("d"); internal_iter->AddDeletion("e"); internal_iter->AddDeletion("f"); internal_iter->AddDeletion("g"); internal_iter->AddDeletion("h"); internal_iter->Finish(); Slice prefix("c"); ReadOptions ro; ro.iterate_upper_bound = &prefix; std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 7, options.max_sequential_skip_in_iterations, ro.iterate_upper_bound)); SetPerfLevel(kEnableCount); ASSERT_TRUE(GetPerfLevel() == kEnableCount); perf_context.Reset(); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(static_cast(perf_context.internal_delete_skipped_count), 0); ASSERT_EQ(db_iter->key().ToString(), "b"); SetPerfLevel(kDisable); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddDeletion("a"); internal_iter->AddDeletion("a"); internal_iter->AddDeletion("a"); internal_iter->AddDeletion("a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "val_a"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "val_a"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b"); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("a", "val_a"); internal_iter->AddPut("b", "val_b"); internal_iter->AddPut("c", "val_c"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "val_c"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val_b"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "val_c"); } } TEST_F(DBIteratorTest, DBIteratorEmpty) { Options options; { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 0, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 0, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(!db_iter->Valid()); } } TEST_F(DBIteratorTest, DBIteratorUseSkipCountSkips) { Options options; options.statistics = rocksdb::CreateDBStatistics(); options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); for (size_t i = 0; i < 200; ++i) { internal_iter->AddPut("a", "a"); internal_iter->AddPut("b", "b"); internal_iter->AddPut("c", "c"); } internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "c"); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1u); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "b"); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 2u); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "a"); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 3u); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 3u); } TEST_F(DBIteratorTest, DBIteratorUseSkip) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); { for (size_t i = 0; i < 200; ++i) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("b", "merge_1"); internal_iter->AddMerge("a", "merge_2"); for (size_t k = 0; k < 200; ++k) { internal_iter->AddPut("c", ToString(k)); } internal_iter->Finish(); options.statistics = rocksdb::CreateDBStatistics(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, i + 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), ToString(i)); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_2"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } { for (size_t i = 0; i < 200; ++i) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("b", "merge_1"); internal_iter->AddMerge("a", "merge_2"); for (size_t k = 0; k < 200; ++k) { internal_iter->AddDeletion("c"); } internal_iter->AddPut("c", "200"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, i + 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_2"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("b", "merge_1"); internal_iter->AddMerge("a", "merge_2"); for (size_t i = 0; i < 200; ++i) { internal_iter->AddDeletion("c"); } internal_iter->AddPut("c", "200"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 202, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "200"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_2"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } { for (size_t i = 0; i < 200; ++i) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); for (size_t k = 0; k < 200; ++k) { internal_iter->AddDeletion("c"); } internal_iter->AddPut("c", "200"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, i, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(!db_iter->Valid()); db_iter->SeekToFirst(); ASSERT_TRUE(!db_iter->Valid()); } TestIterator* internal_iter = new TestIterator(BytewiseComparator()); for (size_t i = 0; i < 200; ++i) { internal_iter->AddDeletion("c"); } internal_iter->AddPut("c", "200"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 200, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "200"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "200"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); } { for (size_t i = 0; i < 200; ++i) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("b", "merge_1"); internal_iter->AddMerge("a", "merge_2"); for (size_t k = 0; k < 200; ++k) { internal_iter->AddPut("d", ToString(k)); } for (size_t k = 0; k < 200; ++k) { internal_iter->AddPut("c", ToString(k)); } internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, i + 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "d"); ASSERT_EQ(db_iter->value().ToString(), ToString(i)); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_2"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } { for (size_t i = 0; i < 200; ++i) { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("b", "b"); internal_iter->AddMerge("a", "a"); for (size_t k = 0; k < 200; ++k) { internal_iter->AddMerge("c", ToString(k)); } internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, i + 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); std::string merge_result = "0"; for (size_t j = 1; j <= i; ++j) { merge_result += "," + ToString(j); } ASSERT_EQ(db_iter->value().ToString(), merge_result); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "b"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "a"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } } TEST_F(DBIteratorTest, DBIterator1) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "0"); internal_iter->AddPut("b", "0"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("a", "1"); internal_iter->AddMerge("b", "2"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 1, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "0"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); db_iter->Next(); ASSERT_FALSE(db_iter->Valid()); } TEST_F(DBIteratorTest, DBIterator2) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "0"); internal_iter->AddPut("b", "0"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("a", "1"); internal_iter->AddMerge("b", "2"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 0, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "0"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); } TEST_F(DBIteratorTest, DBIterator3) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "0"); internal_iter->AddPut("b", "0"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("a", "1"); internal_iter->AddMerge("b", "2"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "0"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); } TEST_F(DBIteratorTest, DBIterator4) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "0"); internal_iter->AddPut("b", "0"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("a", "1"); internal_iter->AddMerge("b", "2"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 4, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "0,1"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "2"); db_iter->Next(); ASSERT_TRUE(!db_iter->Valid()); } TEST_F(DBIteratorTest, DBIterator5) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 0, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 1, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1,merge_2"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1,merge_2,merge_3"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 3, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "put_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 4, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "put_1,merge_4"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 5, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "put_1,merge_4,merge_5"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddPut("a", "put_1"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 6, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "put_1,merge_4,merge_5,merge_6"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } TEST_F(DBIteratorTest, DBIterator6) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 0, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 1, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1,merge_2"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1,merge_2,merge_3"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 3, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 4, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_4"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 5, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_4,merge_5"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("a", "merge_3"); internal_iter->AddDeletion("a"); internal_iter->AddMerge("a", "merge_4"); internal_iter->AddMerge("a", "merge_5"); internal_iter->AddMerge("a", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 6, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_4,merge_5,merge_6"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } TEST_F(DBIteratorTest, DBIterator7) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 0, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 2, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "val,merge_2"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 4, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_3"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 5, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "merge_4"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_3"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 6, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "merge_4,merge_5"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_3"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 7, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "merge_4,merge_5"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 9, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "merge_4,merge_5"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_6,merge_7"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 13, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "merge_4,merge_5"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_6,merge_7,merge_8,merge_9,merge_10,merge_11"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddPut("b", "val"); internal_iter->AddMerge("b", "merge_2"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("c", "merge_4"); internal_iter->AddMerge("c", "merge_5"); internal_iter->AddDeletion("b"); internal_iter->AddMerge("b", "merge_6"); internal_iter->AddMerge("b", "merge_7"); internal_iter->AddMerge("b", "merge_8"); internal_iter->AddMerge("b", "merge_9"); internal_iter->AddMerge("b", "merge_10"); internal_iter->AddMerge("b", "merge_11"); internal_iter->AddDeletion("c"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 14, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_6,merge_7,merge_8,merge_9,merge_10,merge_11"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1"); db_iter->Prev(); ASSERT_TRUE(!db_iter->Valid()); } } TEST_F(DBIteratorTest, DBIterator8) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddDeletion("a"); internal_iter->AddPut("a", "0"); internal_iter->AddPut("b", "0"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "0"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "0"); } // TODO(3.13): fix the issue of Seek() then Prev() which might not necessary // return the biggest element smaller than the seek key. TEST_F(DBIteratorTest, DBIterator9) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); { TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddMerge("a", "merge_1"); internal_iter->AddMerge("a", "merge_2"); internal_iter->AddMerge("b", "merge_3"); internal_iter->AddMerge("b", "merge_4"); internal_iter->AddMerge("d", "merge_5"); internal_iter->AddMerge("d", "merge_6"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_3,merge_4"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "d"); ASSERT_EQ(db_iter->value().ToString(), "merge_5,merge_6"); db_iter->Seek("b"); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_3,merge_4"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "merge_1,merge_2"); db_iter->Seek("c"); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "d"); ASSERT_EQ(db_iter->value().ToString(), "merge_5,merge_6"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "merge_3,merge_4"); } } // TODO(3.13): fix the issue of Seek() then Prev() which might not necessary // return the biggest element smaller than the seek key. TEST_F(DBIteratorTest, DBIterator10) { Options options; TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "1"); internal_iter->AddPut("b", "2"); internal_iter->AddPut("c", "3"); internal_iter->AddPut("d", "4"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, options.max_sequential_skip_in_iterations)); db_iter->Seek("c"); ASSERT_TRUE(db_iter->Valid()); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "2"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "3"); } TEST_F(DBIteratorTest, SeekToLastOccurrenceSeq0) { Options options; options.merge_operator = nullptr; TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "1"); internal_iter->AddPut("b", "2"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, 0 /* force seek */)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "1"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); ASSERT_EQ(db_iter->value().ToString(), "2"); db_iter->Next(); ASSERT_FALSE(db_iter->Valid()); } TEST_F(DBIteratorTest, DBIterator11) { Options options; options.merge_operator = MergeOperators::CreateFromStringId("stringappend"); TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "0"); internal_iter->AddPut("b", "0"); internal_iter->AddSingleDeletion("b"); internal_iter->AddMerge("a", "1"); internal_iter->AddMerge("b", "2"); internal_iter->Finish(); std::unique_ptr db_iter(NewDBIterator( env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 1, options.max_sequential_skip_in_iterations)); db_iter->SeekToFirst(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "0"); db_iter->Next(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "b"); db_iter->Next(); ASSERT_FALSE(db_iter->Valid()); } TEST_F(DBIteratorTest, DBIterator12) { Options options; options.merge_operator = nullptr; TestIterator* internal_iter = new TestIterator(BytewiseComparator()); internal_iter->AddPut("a", "1"); internal_iter->AddPut("b", "2"); internal_iter->AddPut("c", "3"); internal_iter->AddSingleDeletion("b"); internal_iter->Finish(); std::unique_ptr db_iter( NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(), internal_iter, 10, 0)); db_iter->SeekToLast(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "c"); ASSERT_EQ(db_iter->value().ToString(), "3"); db_iter->Prev(); ASSERT_TRUE(db_iter->Valid()); ASSERT_EQ(db_iter->key().ToString(), "a"); ASSERT_EQ(db_iter->value().ToString(), "1"); db_iter->Prev(); ASSERT_FALSE(db_iter->Valid()); } class DBIterWithMergeIterTest : public testing::Test { public: DBIterWithMergeIterTest() : env_(Env::Default()), icomp_(BytewiseComparator()) { options_.merge_operator = nullptr; internal_iter1_ = new TestIterator(BytewiseComparator()); internal_iter1_->Add("a", kTypeValue, "1", 3u); internal_iter1_->Add("f", kTypeValue, "2", 5u); internal_iter1_->Add("g", kTypeValue, "3", 7u); internal_iter1_->Finish(); internal_iter2_ = new TestIterator(BytewiseComparator()); internal_iter2_->Add("a", kTypeValue, "4", 6u); internal_iter2_->Add("b", kTypeValue, "5", 1u); internal_iter2_->Add("c", kTypeValue, "6", 2u); internal_iter2_->Add("d", kTypeValue, "7", 3u); internal_iter2_->Finish(); std::vector child_iters; child_iters.push_back(internal_iter1_); child_iters.push_back(internal_iter2_); InternalKeyComparator icomp(BytewiseComparator()); InternalIterator* merge_iter = NewMergingIterator(&icomp_, &child_iters[0], 2u); db_iter_.reset(NewDBIterator(env_, ImmutableCFOptions(options_), BytewiseComparator(), merge_iter, 8 /* read data earlier than seqId 8 */, 3 /* max iterators before reseek */)); } Env* env_; Options options_; TestIterator* internal_iter1_; TestIterator* internal_iter2_; InternalKeyComparator icomp_; Iterator* merge_iter_; std::unique_ptr db_iter_; }; TEST_F(DBIterWithMergeIterTest, InnerMergeIterator1) { db_iter_->SeekToFirst(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); db_iter_->Next(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Next(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Next(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Next(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Next(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); db_iter_->Next(); ASSERT_FALSE(db_iter_->Valid()); } TEST_F(DBIterWithMergeIterTest, InnerMergeIterator2) { // Test Prev() when one child iterator is at its end. db_iter_->Seek("g"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace1) { // Test Prev() when one child iterator is at its end but more rows // are added. db_iter_->Seek("f"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); // Test call back inserts a key in the end of the mem table after // MergeIterator::Prev() realized the mem table iterator is at its end // and before an SeekToLast() is called. rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforeSeekToLast", [&](void* arg) { internal_iter2_->Add("z", kTypeValue, "7", 12u); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace2) { // Test Prev() when one child iterator is at its end but more rows // are added. db_iter_->Seek("f"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); // Test call back inserts entries for update a key in the end of the // mem table after MergeIterator::Prev() realized the mem tableiterator is at // its end and before an SeekToLast() is called. rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforeSeekToLast", [&](void* arg) { internal_iter2_->Add("z", kTypeValue, "7", 12u); internal_iter2_->Add("z", kTypeValue, "7", 11u); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace3) { // Test Prev() when one child iterator is at its end but more rows // are added and max_skipped is triggered. db_iter_->Seek("f"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); // Test call back inserts entries for update a key in the end of the // mem table after MergeIterator::Prev() realized the mem table iterator is at // its end and before an SeekToLast() is called. rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforeSeekToLast", [&](void* arg) { internal_iter2_->Add("z", kTypeValue, "7", 16u, true); internal_iter2_->Add("z", kTypeValue, "7", 15u, true); internal_iter2_->Add("z", kTypeValue, "7", 14u, true); internal_iter2_->Add("z", kTypeValue, "7", 13u, true); internal_iter2_->Add("z", kTypeValue, "7", 12u, true); internal_iter2_->Add("z", kTypeValue, "7", 11u, true); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace4) { // Test Prev() when one child iterator has more rows inserted // between Seek() and Prev() when changing directions. internal_iter2_->Add("z", kTypeValue, "9", 4u); db_iter_->Seek("g"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); // Test call back inserts entries for update a key before "z" in // mem table after MergeIterator::Prev() calls mem table iterator's // Seek() and before calling Prev() rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforePrev", [&](void* arg) { IteratorWrapper* it = reinterpret_cast(arg); if (it->key().starts_with("z")) { internal_iter2_->Add("x", kTypeValue, "7", 16u, true); internal_iter2_->Add("x", kTypeValue, "7", 15u, true); internal_iter2_->Add("x", kTypeValue, "7", 14u, true); internal_iter2_->Add("x", kTypeValue, "7", 13u, true); internal_iter2_->Add("x", kTypeValue, "7", 12u, true); internal_iter2_->Add("x", kTypeValue, "7", 11u, true); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace5) { internal_iter2_->Add("z", kTypeValue, "9", 4u); // Test Prev() when one child iterator has more rows inserted // between Seek() and Prev() when changing directions. db_iter_->Seek("g"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); // Test call back inserts entries for update a key before "z" in // mem table after MergeIterator::Prev() calls mem table iterator's // Seek() and before calling Prev() rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforePrev", [&](void* arg) { IteratorWrapper* it = reinterpret_cast(arg); if (it->key().starts_with("z")) { internal_iter2_->Add("x", kTypeValue, "7", 16u, true); internal_iter2_->Add("x", kTypeValue, "7", 15u, true); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace6) { internal_iter2_->Add("z", kTypeValue, "9", 4u); // Test Prev() when one child iterator has more rows inserted // between Seek() and Prev() when changing directions. db_iter_->Seek("g"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); // Test call back inserts an entry for update a key before "z" in // mem table after MergeIterator::Prev() calls mem table iterator's // Seek() and before calling Prev() rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforePrev", [&](void* arg) { IteratorWrapper* it = reinterpret_cast(arg); if (it->key().starts_with("z")) { internal_iter2_->Add("x", kTypeValue, "7", 16u, true); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace7) { internal_iter1_->Add("u", kTypeValue, "10", 4u); internal_iter1_->Add("v", kTypeValue, "11", 4u); internal_iter1_->Add("w", kTypeValue, "12", 4u); internal_iter2_->Add("z", kTypeValue, "9", 4u); // Test Prev() when one child iterator has more rows inserted // between Seek() and Prev() when changing directions. db_iter_->Seek("g"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); // Test call back inserts entries for update a key before "z" in // mem table after MergeIterator::Prev() calls mem table iterator's // Seek() and before calling Prev() rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforePrev", [&](void* arg) { IteratorWrapper* it = reinterpret_cast(arg); if (it->key().starts_with("z")) { internal_iter2_->Add("x", kTypeValue, "7", 16u, true); internal_iter2_->Add("x", kTypeValue, "7", 15u, true); internal_iter2_->Add("x", kTypeValue, "7", 14u, true); internal_iter2_->Add("x", kTypeValue, "7", 13u, true); internal_iter2_->Add("x", kTypeValue, "7", 12u, true); internal_iter2_->Add("x", kTypeValue, "7", 11u, true); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "c"); ASSERT_EQ(db_iter_->value().ToString(), "6"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "b"); ASSERT_EQ(db_iter_->value().ToString(), "5"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "a"); ASSERT_EQ(db_iter_->value().ToString(), "4"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBIterWithMergeIterTest, InnerMergeIteratorDataRace8) { // internal_iter1_: a, f, g // internal_iter2_: a, b, c, d, adding (z) internal_iter2_->Add("z", kTypeValue, "9", 4u); // Test Prev() when one child iterator has more rows inserted // between Seek() and Prev() when changing directions. db_iter_->Seek("g"); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "g"); ASSERT_EQ(db_iter_->value().ToString(), "3"); // Test call back inserts two keys before "z" in mem table after // MergeIterator::Prev() calls mem table iterator's Seek() and // before calling Prev() rocksdb::SyncPoint::GetInstance()->SetCallBack( "MergeIterator::Prev:BeforePrev", [&](void* arg) { IteratorWrapper* it = reinterpret_cast(arg); if (it->key().starts_with("z")) { internal_iter2_->Add("x", kTypeValue, "7", 16u, true); internal_iter2_->Add("y", kTypeValue, "7", 17u, true); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "f"); ASSERT_EQ(db_iter_->value().ToString(), "2"); db_iter_->Prev(); ASSERT_TRUE(db_iter_->Valid()); ASSERT_EQ(db_iter_->key().ToString(), "d"); ASSERT_EQ(db_iter_->value().ToString(), "7"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/db_log_iter_test.cc000066400000000000000000000231011267530634000173570ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Introduction of SyncPoint effectively disabled building and running this test // in Release build. // which is a pity, it is a good test #if !defined(ROCKSDB_LITE) #include "db/db_test_util.h" #include "port/stack_trace.h" namespace rocksdb { class DBTestXactLogIterator : public DBTestBase { public: DBTestXactLogIterator() : DBTestBase("/db_log_iter_test") {} std::unique_ptr OpenTransactionLogIter( const SequenceNumber seq) { unique_ptr iter; Status status = dbfull()->GetUpdatesSince(seq, &iter); EXPECT_OK(status); EXPECT_TRUE(iter->Valid()); return iter; } }; namespace { SequenceNumber ReadRecords( std::unique_ptr& iter, int& count) { count = 0; SequenceNumber lastSequence = 0; BatchResult res; while (iter->Valid()) { res = iter->GetBatch(); EXPECT_TRUE(res.sequence > lastSequence); ++count; lastSequence = res.sequence; EXPECT_OK(iter->status()); iter->Next(); } return res.sequence; } void ExpectRecords( const int expected_no_records, std::unique_ptr& iter) { int num_records; ReadRecords(iter, num_records); ASSERT_EQ(num_records, expected_no_records); } } // namespace TEST_F(DBTestXactLogIterator, TransactionLogIterator) { do { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); Put(0, "key1", DummyString(1024)); Put(1, "key2", DummyString(1024)); Put(1, "key2", DummyString(1024)); ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 3U); { auto iter = OpenTransactionLogIter(0); ExpectRecords(3, iter); } ReopenWithColumnFamilies({"default", "pikachu"}, options); env_->SleepForMicroseconds(2 * 1000 * 1000); { Put(0, "key4", DummyString(1024)); Put(1, "key5", DummyString(1024)); Put(0, "key6", DummyString(1024)); } { auto iter = OpenTransactionLogIter(0); ExpectRecords(6, iter); } } while (ChangeCompactOptions()); } #ifndef NDEBUG // sync point is not included with DNDEBUG build TEST_F(DBTestXactLogIterator, TransactionLogIteratorRace) { static const int LOG_ITERATOR_RACE_TEST_COUNT = 2; static const char* sync_points[LOG_ITERATOR_RACE_TEST_COUNT][4] = { {"WalManager::GetSortedWalFiles:1", "WalManager::PurgeObsoleteFiles:1", "WalManager::PurgeObsoleteFiles:2", "WalManager::GetSortedWalFiles:2"}, {"WalManager::GetSortedWalsOfType:1", "WalManager::PurgeObsoleteFiles:1", "WalManager::PurgeObsoleteFiles:2", "WalManager::GetSortedWalsOfType:2"}}; for (int test = 0; test < LOG_ITERATOR_RACE_TEST_COUNT; ++test) { // Setup sync point dependency to reproduce the race condition of // a log file moved to archived dir, in the middle of GetSortedWalFiles rocksdb::SyncPoint::GetInstance()->LoadDependency( { { sync_points[test][0], sync_points[test][1] }, { sync_points[test][2], sync_points[test][3] }, }); do { rocksdb::SyncPoint::GetInstance()->ClearTrace(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); Options options = OptionsForLogIterTest(); DestroyAndReopen(options); Put("key1", DummyString(1024)); dbfull()->Flush(FlushOptions()); Put("key2", DummyString(1024)); dbfull()->Flush(FlushOptions()); Put("key3", DummyString(1024)); dbfull()->Flush(FlushOptions()); Put("key4", DummyString(1024)); ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 4U); { auto iter = OpenTransactionLogIter(0); ExpectRecords(4, iter); } rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // trigger async flush, and log move. Well, log move will // wait until the GetSortedWalFiles:1 to reproduce the race // condition FlushOptions flush_options; flush_options.wait = false; dbfull()->Flush(flush_options); // "key5" would be written in a new memtable and log Put("key5", DummyString(1024)); { // this iter would miss "key4" if not fixed auto iter = OpenTransactionLogIter(0); ExpectRecords(5, iter); } } while (ChangeCompactOptions()); } } #endif TEST_F(DBTestXactLogIterator, TransactionLogIteratorStallAtLastRecord) { do { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); Put("key1", DummyString(1024)); auto iter = OpenTransactionLogIter(0); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); iter->Next(); ASSERT_TRUE(!iter->Valid()); ASSERT_OK(iter->status()); Put("key2", DummyString(1024)); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); } while (ChangeCompactOptions()); } TEST_F(DBTestXactLogIterator, TransactionLogIteratorCheckAfterRestart) { do { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); Put("key1", DummyString(1024)); Put("key2", DummyString(1023)); dbfull()->Flush(FlushOptions()); Reopen(options); auto iter = OpenTransactionLogIter(0); ExpectRecords(2, iter); } while (ChangeCompactOptions()); } TEST_F(DBTestXactLogIterator, TransactionLogIteratorCorruptedLog) { do { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); for (int i = 0; i < 1024; i++) { Put("key"+ToString(i), DummyString(10)); } dbfull()->Flush(FlushOptions()); // Corrupt this log to create a gap rocksdb::VectorLogPtr wal_files; ASSERT_OK(dbfull()->GetSortedWalFiles(wal_files)); const auto logfile_path = dbname_ + "/" + wal_files.front()->PathName(); if (mem_env_) { mem_env_->Truncate(logfile_path, wal_files.front()->SizeFileBytes() / 2); } else { ASSERT_EQ(0, truncate(logfile_path.c_str(), wal_files.front()->SizeFileBytes() / 2)); } // Insert a new entry to a new log file Put("key1025", DummyString(10)); // Try to read from the beginning. Should stop before the gap and read less // than 1025 entries auto iter = OpenTransactionLogIter(0); int count; SequenceNumber last_sequence_read = ReadRecords(iter, count); ASSERT_LT(last_sequence_read, 1025U); // Try to read past the gap, should be able to seek to key1025 auto iter2 = OpenTransactionLogIter(last_sequence_read + 1); ExpectRecords(1, iter2); } while (ChangeCompactOptions()); } TEST_F(DBTestXactLogIterator, TransactionLogIteratorBatchOperations) { do { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); WriteBatch batch; batch.Put(handles_[1], "key1", DummyString(1024)); batch.Put(handles_[0], "key2", DummyString(1024)); batch.Put(handles_[1], "key3", DummyString(1024)); batch.Delete(handles_[0], "key2"); dbfull()->Write(WriteOptions(), &batch); Flush(1); Flush(0); ReopenWithColumnFamilies({"default", "pikachu"}, options); Put(1, "key4", DummyString(1024)); auto iter = OpenTransactionLogIter(3); ExpectRecords(2, iter); } while (ChangeCompactOptions()); } TEST_F(DBTestXactLogIterator, TransactionLogIteratorBlobs) { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); { WriteBatch batch; batch.Put(handles_[1], "key1", DummyString(1024)); batch.Put(handles_[0], "key2", DummyString(1024)); batch.PutLogData(Slice("blob1")); batch.Put(handles_[1], "key3", DummyString(1024)); batch.PutLogData(Slice("blob2")); batch.Delete(handles_[0], "key2"); dbfull()->Write(WriteOptions(), &batch); ReopenWithColumnFamilies({"default", "pikachu"}, options); } auto res = OpenTransactionLogIter(0)->GetBatch(); struct Handler : public WriteBatch::Handler { std::string seen; virtual Status PutCF(uint32_t cf, const Slice& key, const Slice& value) override { seen += "Put(" + ToString(cf) + ", " + key.ToString() + ", " + ToString(value.size()) + ")"; return Status::OK(); } virtual Status MergeCF(uint32_t cf, const Slice& key, const Slice& value) override { seen += "Merge(" + ToString(cf) + ", " + key.ToString() + ", " + ToString(value.size()) + ")"; return Status::OK(); } virtual void LogData(const Slice& blob) override { seen += "LogData(" + blob.ToString() + ")"; } virtual Status DeleteCF(uint32_t cf, const Slice& key) override { seen += "Delete(" + ToString(cf) + ", " + key.ToString() + ")"; return Status::OK(); } } handler; res.writeBatchPtr->Iterate(&handler); ASSERT_EQ( "Put(1, key1, 1024)" "Put(0, key2, 1024)" "LogData(blob1)" "Put(1, key3, 1024)" "LogData(blob2)" "Delete(0, key2)", handler.seen); } } // namespace rocksdb #endif // !defined(ROCKSDB_LITE) int main(int argc, char** argv) { #if !defined(ROCKSDB_LITE) rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); #else return 0; #endif } rocksdb-4.5.1/db/db_properties_test.cc000066400000000000000000001254421267530634000177620ustar00rootroot00000000000000// Copyright (c) 2016, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include #include #include #include "db/db_test_util.h" #include "port/stack_trace.h" #include "rocksdb/options.h" #include "rocksdb/perf_context.h" #include "rocksdb/perf_level.h" #include "rocksdb/table.h" #include "util/random.h" namespace rocksdb { class DBPropertiesTest : public DBTestBase { public: DBPropertiesTest() : DBTestBase("/db_properties_test") {} }; #ifndef ROCKSDB_LITE TEST_F(DBPropertiesTest, Empty) { do { Options options; options.env = env_; options.write_buffer_size = 100000; // Small write buffer options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); std::string num; ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ("0", num); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ("1", num); // Block sync calls env_->delay_sstable_sync_.store(true, std::memory_order_release); Put(1, "k1", std::string(100000, 'x')); // Fill memtable ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ("2", num); Put(1, "k2", std::string(100000, 'y')); // Trigger compaction ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ("1", num); ASSERT_EQ("v1", Get(1, "foo")); // Release sync calls env_->delay_sstable_sync_.store(false, std::memory_order_release); ASSERT_OK(db_->DisableFileDeletions()); ASSERT_TRUE( dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num)); ASSERT_EQ("1", num); ASSERT_OK(db_->DisableFileDeletions()); ASSERT_TRUE( dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num)); ASSERT_EQ("2", num); ASSERT_OK(db_->DisableFileDeletions()); ASSERT_TRUE( dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num)); ASSERT_EQ("3", num); ASSERT_OK(db_->EnableFileDeletions(false)); ASSERT_TRUE( dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num)); ASSERT_EQ("2", num); ASSERT_OK(db_->EnableFileDeletions()); ASSERT_TRUE( dbfull()->GetProperty("rocksdb.is-file-deletions-enabled", &num)); ASSERT_EQ("0", num); } while (ChangeOptions()); } TEST_F(DBPropertiesTest, GetAggregatedIntPropertyTest) { const int kKeySize = 100; const int kValueSize = 500; const int kKeyNum = 100; Options options; options.env = env_; options.create_if_missing = true; options.write_buffer_size = (kKeySize + kValueSize) * kKeyNum / 10; // Make them never flush options.min_write_buffer_number_to_merge = 1000; options.max_write_buffer_number = 1000; options = CurrentOptions(options); CreateAndReopenWithCF({"one", "two", "three", "four"}, options); Random rnd(301); for (auto* handle : handles_) { for (int i = 0; i < kKeyNum; ++i) { db_->Put(WriteOptions(), handle, RandomString(&rnd, kKeySize), RandomString(&rnd, kValueSize)); } } uint64_t manual_sum = 0; uint64_t api_sum = 0; uint64_t value = 0; for (auto* handle : handles_) { ASSERT_TRUE( db_->GetIntProperty(handle, DB::Properties::kSizeAllMemTables, &value)); manual_sum += value; } ASSERT_TRUE(db_->GetAggregatedIntProperty(DB::Properties::kSizeAllMemTables, &api_sum)); ASSERT_GT(manual_sum, 0); ASSERT_EQ(manual_sum, api_sum); ASSERT_FALSE(db_->GetAggregatedIntProperty(DB::Properties::kDBStats, &value)); uint64_t before_flush_trm; uint64_t after_flush_trm; for (auto* handle : handles_) { ASSERT_TRUE(db_->GetAggregatedIntProperty( DB::Properties::kEstimateTableReadersMem, &before_flush_trm)); // Issue flush and expect larger memory usage of table readers. db_->Flush(FlushOptions(), handle); ASSERT_TRUE(db_->GetAggregatedIntProperty( DB::Properties::kEstimateTableReadersMem, &after_flush_trm)); ASSERT_GT(after_flush_trm, before_flush_trm); } } namespace { void ResetTableProperties(TableProperties* tp) { tp->data_size = 0; tp->index_size = 0; tp->filter_size = 0; tp->raw_key_size = 0; tp->raw_value_size = 0; tp->num_data_blocks = 0; tp->num_entries = 0; } void ParseTablePropertiesString(std::string tp_string, TableProperties* tp) { double dummy_double; std::replace(tp_string.begin(), tp_string.end(), ';', ' '); std::replace(tp_string.begin(), tp_string.end(), '=', ' '); ResetTableProperties(tp); sscanf(tp_string.c_str(), "# data blocks %" SCNu64 " # entries %" SCNu64 " raw key size %" SCNu64 " raw average key size %lf " " raw value size %" SCNu64 " raw average value size %lf " " data block size %" SCNu64 " index block size %" SCNu64 " filter block size %" SCNu64, &tp->num_data_blocks, &tp->num_entries, &tp->raw_key_size, &dummy_double, &tp->raw_value_size, &dummy_double, &tp->data_size, &tp->index_size, &tp->filter_size); } void VerifySimilar(uint64_t a, uint64_t b, double bias) { ASSERT_EQ(a == 0U, b == 0U); if (a == 0) { return; } double dbl_a = static_cast(a); double dbl_b = static_cast(b); if (dbl_a > dbl_b) { ASSERT_LT(static_cast(dbl_a - dbl_b) / (dbl_a + dbl_b), bias); } else { ASSERT_LT(static_cast(dbl_b - dbl_a) / (dbl_a + dbl_b), bias); } } void VerifyTableProperties(const TableProperties& base_tp, const TableProperties& new_tp, double filter_size_bias = 0.1, double index_size_bias = 0.1, double data_size_bias = 0.1, double num_data_blocks_bias = 0.05) { VerifySimilar(base_tp.data_size, new_tp.data_size, data_size_bias); VerifySimilar(base_tp.index_size, new_tp.index_size, index_size_bias); VerifySimilar(base_tp.filter_size, new_tp.filter_size, filter_size_bias); VerifySimilar(base_tp.num_data_blocks, new_tp.num_data_blocks, num_data_blocks_bias); ASSERT_EQ(base_tp.raw_key_size, new_tp.raw_key_size); ASSERT_EQ(base_tp.raw_value_size, new_tp.raw_value_size); ASSERT_EQ(base_tp.num_entries, new_tp.num_entries); } void GetExpectedTableProperties(TableProperties* expected_tp, const int kKeySize, const int kValueSize, const int kKeysPerTable, const int kTableCount, const int kBloomBitsPerKey, const size_t kBlockSize) { const int kKeyCount = kTableCount * kKeysPerTable; const int kAvgSuccessorSize = kKeySize / 2; const int kEncodingSavePerKey = kKeySize / 4; expected_tp->raw_key_size = kKeyCount * (kKeySize + 8); expected_tp->raw_value_size = kKeyCount * kValueSize; expected_tp->num_entries = kKeyCount; expected_tp->num_data_blocks = kTableCount * (kKeysPerTable * (kKeySize - kEncodingSavePerKey + kValueSize)) / kBlockSize; expected_tp->data_size = kTableCount * (kKeysPerTable * (kKeySize + 8 + kValueSize)); expected_tp->index_size = expected_tp->num_data_blocks * (kAvgSuccessorSize + 12); expected_tp->filter_size = kTableCount * (kKeysPerTable * kBloomBitsPerKey / 8); } } // anonymous namespace TEST_F(DBPropertiesTest, ValidatePropertyInfo) { for (const auto& ppt_name_and_info : InternalStats::ppt_name_to_info) { // If C++ gets a std::string_literal, this would be better to check at // compile-time using static_assert. ASSERT_TRUE(ppt_name_and_info.first.empty() || !isdigit(ppt_name_and_info.first.back())); ASSERT_TRUE((ppt_name_and_info.second.handle_string == nullptr) != (ppt_name_and_info.second.handle_int == nullptr)); } } TEST_F(DBPropertiesTest, AggregatedTableProperties) { for (int kTableCount = 40; kTableCount <= 100; kTableCount += 30) { const int kKeysPerTable = 100; const int kKeySize = 80; const int kValueSize = 200; const int kBloomBitsPerKey = 20; Options options = CurrentOptions(); options.level0_file_num_compaction_trigger = 8; options.compression = kNoCompression; options.create_if_missing = true; BlockBasedTableOptions table_options; table_options.filter_policy.reset( NewBloomFilterPolicy(kBloomBitsPerKey, false)); table_options.block_size = 1024; options.table_factory.reset(new BlockBasedTableFactory(table_options)); DestroyAndReopen(options); Random rnd(5632); for (int table = 1; table <= kTableCount; ++table) { for (int i = 0; i < kKeysPerTable; ++i) { db_->Put(WriteOptions(), RandomString(&rnd, kKeySize), RandomString(&rnd, kValueSize)); } db_->Flush(FlushOptions()); } std::string property; db_->GetProperty(DB::Properties::kAggregatedTableProperties, &property); TableProperties expected_tp; GetExpectedTableProperties(&expected_tp, kKeySize, kValueSize, kKeysPerTable, kTableCount, kBloomBitsPerKey, table_options.block_size); TableProperties output_tp; ParseTablePropertiesString(property, &output_tp); VerifyTableProperties(expected_tp, output_tp); } } TEST_F(DBPropertiesTest, ReadLatencyHistogramByLevel) { Options options = CurrentOptions(); options.write_buffer_size = 110 << 10; options.level0_file_num_compaction_trigger = 6; options.num_levels = 4; options.compression = kNoCompression; options.max_bytes_for_level_base = 4500 << 10; options.target_file_size_base = 98 << 10; options.max_write_buffer_number = 2; options.statistics = rocksdb::CreateDBStatistics(); options.max_open_files = 100; BlockBasedTableOptions table_options; table_options.no_block_cache = true; DestroyAndReopen(options); int key_index = 0; Random rnd(301); for (int num = 0; num < 8; num++) { Put("foo", "bar"); GenerateNewFile(&rnd, &key_index); dbfull()->TEST_WaitForCompact(); } dbfull()->TEST_WaitForCompact(); std::string prop; ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); // Get() after flushes, See latency histogram tracked. for (int key = 0; key < key_index; key++) { Get(Key(key)); } ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); ASSERT_NE(std::string::npos, prop.find("** Level 0 read latency histogram")); ASSERT_NE(std::string::npos, prop.find("** Level 1 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram")); // Reopen and issue Get(). See thee latency tracked Reopen(options); dbfull()->TEST_WaitForCompact(); for (int key = 0; key < key_index; key++) { Get(Key(key)); } ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); ASSERT_NE(std::string::npos, prop.find("** Level 0 read latency histogram")); ASSERT_NE(std::string::npos, prop.find("** Level 1 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram")); // Reopen and issue iterating. See thee latency tracked Reopen(options); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); ASSERT_EQ(std::string::npos, prop.find("** Level 0 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 1 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram")); { unique_ptr iter(db_->NewIterator(ReadOptions())); for (iter->Seek(Key(0)); iter->Valid(); iter->Next()) { } } ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); ASSERT_NE(std::string::npos, prop.find("** Level 0 read latency histogram")); ASSERT_NE(std::string::npos, prop.find("** Level 1 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram")); // options.max_open_files preloads table readers. options.max_open_files = -1; Reopen(options); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); ASSERT_NE(std::string::npos, prop.find("** Level 0 read latency histogram")); ASSERT_NE(std::string::npos, prop.find("** Level 1 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram")); for (int key = 0; key < key_index; key++) { Get(Key(key)); } ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop)); ASSERT_NE(std::string::npos, prop.find("** Level 0 read latency histogram")); ASSERT_NE(std::string::npos, prop.find("** Level 1 read latency histogram")); ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram")); } TEST_F(DBPropertiesTest, AggregatedTablePropertiesAtLevel) { const int kTableCount = 100; const int kKeysPerTable = 10; const int kKeySize = 50; const int kValueSize = 400; const int kMaxLevel = 7; const int kBloomBitsPerKey = 20; Random rnd(301); Options options = CurrentOptions(); options.level0_file_num_compaction_trigger = 8; options.compression = kNoCompression; options.create_if_missing = true; options.level0_file_num_compaction_trigger = 2; options.target_file_size_base = 8192; options.max_bytes_for_level_base = 10000; options.max_bytes_for_level_multiplier = 2; // This ensures there no compaction happening when we call GetProperty(). options.disable_auto_compactions = true; BlockBasedTableOptions table_options; table_options.filter_policy.reset( NewBloomFilterPolicy(kBloomBitsPerKey, false)); table_options.block_size = 1024; options.table_factory.reset(new BlockBasedTableFactory(table_options)); DestroyAndReopen(options); std::string level_tp_strings[kMaxLevel]; std::string tp_string; TableProperties level_tps[kMaxLevel]; TableProperties tp, sum_tp, expected_tp; for (int table = 1; table <= kTableCount; ++table) { for (int i = 0; i < kKeysPerTable; ++i) { db_->Put(WriteOptions(), RandomString(&rnd, kKeySize), RandomString(&rnd, kValueSize)); } db_->Flush(FlushOptions()); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); ResetTableProperties(&sum_tp); for (int level = 0; level < kMaxLevel; ++level) { db_->GetProperty( DB::Properties::kAggregatedTablePropertiesAtLevel + ToString(level), &level_tp_strings[level]); ParseTablePropertiesString(level_tp_strings[level], &level_tps[level]); sum_tp.data_size += level_tps[level].data_size; sum_tp.index_size += level_tps[level].index_size; sum_tp.filter_size += level_tps[level].filter_size; sum_tp.raw_key_size += level_tps[level].raw_key_size; sum_tp.raw_value_size += level_tps[level].raw_value_size; sum_tp.num_data_blocks += level_tps[level].num_data_blocks; sum_tp.num_entries += level_tps[level].num_entries; } db_->GetProperty(DB::Properties::kAggregatedTableProperties, &tp_string); ParseTablePropertiesString(tp_string, &tp); ASSERT_EQ(sum_tp.data_size, tp.data_size); ASSERT_EQ(sum_tp.index_size, tp.index_size); ASSERT_EQ(sum_tp.filter_size, tp.filter_size); ASSERT_EQ(sum_tp.raw_key_size, tp.raw_key_size); ASSERT_EQ(sum_tp.raw_value_size, tp.raw_value_size); ASSERT_EQ(sum_tp.num_data_blocks, tp.num_data_blocks); ASSERT_EQ(sum_tp.num_entries, tp.num_entries); if (table > 3) { GetExpectedTableProperties(&expected_tp, kKeySize, kValueSize, kKeysPerTable, table, kBloomBitsPerKey, table_options.block_size); // Gives larger bias here as index block size, filter block size, // and data block size become much harder to estimate in this test. VerifyTableProperties(tp, expected_tp, 0.5, 0.4, 0.4, 0.25); } } } TEST_F(DBPropertiesTest, NumImmutableMemTable) { do { Options options = CurrentOptions(); WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; options.max_write_buffer_number = 4; options.min_write_buffer_number_to_merge = 3; options.max_write_buffer_number_to_maintain = 4; options.write_buffer_size = 1000000; CreateAndReopenWithCF({"pikachu"}, options); std::string big_value(1000000 * 2, 'x'); std::string num; SetPerfLevel(kEnableTime); ASSERT_TRUE(GetPerfLevel() == kEnableTime); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k1", big_value)); ASSERT_TRUE(dbfull()->GetProperty(handles_[1], "rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], DB::Properties::kNumImmutableMemTableFlushed, &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ(num, "1"); perf_context.Reset(); Get(1, "k1"); ASSERT_EQ(1, static_cast(perf_context.get_from_memtable_count)); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k2", big_value)); ASSERT_TRUE(dbfull()->GetProperty(handles_[1], "rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "1"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ(num, "1"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-imm-mem-tables", &num)); ASSERT_EQ(num, "1"); perf_context.Reset(); Get(1, "k1"); ASSERT_EQ(2, static_cast(perf_context.get_from_memtable_count)); perf_context.Reset(); Get(1, "k2"); ASSERT_EQ(1, static_cast(perf_context.get_from_memtable_count)); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k3", big_value)); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.cur-size-active-mem-table", &num)); ASSERT_TRUE(dbfull()->GetProperty(handles_[1], "rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "2"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &num)); ASSERT_EQ(num, "1"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.num-entries-imm-mem-tables", &num)); ASSERT_EQ(num, "2"); perf_context.Reset(); Get(1, "k2"); ASSERT_EQ(2, static_cast(perf_context.get_from_memtable_count)); perf_context.Reset(); Get(1, "k3"); ASSERT_EQ(1, static_cast(perf_context.get_from_memtable_count)); perf_context.Reset(); Get(1, "k1"); ASSERT_EQ(3, static_cast(perf_context.get_from_memtable_count)); ASSERT_OK(Flush(1)); ASSERT_TRUE(dbfull()->GetProperty(handles_[1], "rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], DB::Properties::kNumImmutableMemTableFlushed, &num)); ASSERT_EQ(num, "3"); ASSERT_TRUE(dbfull()->GetProperty( handles_[1], "rocksdb.cur-size-active-mem-table", &num)); // "192" is the size of the metadata of an empty skiplist, this would // break if we change the default skiplist implementation ASSERT_EQ(num, "192"); uint64_t int_num; uint64_t base_total_size; ASSERT_TRUE(dbfull()->GetIntProperty( handles_[1], "rocksdb.estimate-num-keys", &base_total_size)); ASSERT_OK(dbfull()->Delete(writeOpt, handles_[1], "k2")); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k3", "")); ASSERT_OK(dbfull()->Delete(writeOpt, handles_[1], "k3")); ASSERT_TRUE(dbfull()->GetIntProperty( handles_[1], "rocksdb.num-deletes-active-mem-table", &int_num)); ASSERT_EQ(int_num, 2U); ASSERT_TRUE(dbfull()->GetIntProperty( handles_[1], "rocksdb.num-entries-active-mem-table", &int_num)); ASSERT_EQ(int_num, 3U); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k2", big_value)); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k2", big_value)); ASSERT_TRUE(dbfull()->GetIntProperty( handles_[1], "rocksdb.num-entries-imm-mem-tables", &int_num)); ASSERT_EQ(int_num, 4U); ASSERT_TRUE(dbfull()->GetIntProperty( handles_[1], "rocksdb.num-deletes-imm-mem-tables", &int_num)); ASSERT_EQ(int_num, 2U); ASSERT_TRUE(dbfull()->GetIntProperty( handles_[1], "rocksdb.estimate-num-keys", &int_num)); ASSERT_EQ(int_num, base_total_size + 1); SetPerfLevel(kDisable); ASSERT_TRUE(GetPerfLevel() == kDisable); } while (ChangeCompactOptions()); } TEST_F(DBPropertiesTest, GetProperty) { // Set sizes to both background thread pool to be 1 and block them. env_->SetBackgroundThreads(1, Env::HIGH); env_->SetBackgroundThreads(1, Env::LOW); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); test::SleepingBackgroundTask sleeping_task_high; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_high, Env::Priority::HIGH); Options options = CurrentOptions(); WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; options.compaction_style = kCompactionStyleUniversal; options.level0_file_num_compaction_trigger = 1; options.compaction_options_universal.size_ratio = 50; options.max_background_compactions = 1; options.max_background_flushes = 1; options.max_write_buffer_number = 10; options.min_write_buffer_number_to_merge = 1; options.max_write_buffer_number_to_maintain = 0; options.write_buffer_size = 1000000; Reopen(options); std::string big_value(1000000 * 2, 'x'); std::string num; uint64_t int_num; SetPerfLevel(kEnableTime); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_EQ(int_num, 0U); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-live-data-size", &int_num)); ASSERT_EQ(int_num, 0U); ASSERT_OK(dbfull()->Put(writeOpt, "k1", big_value)); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.mem-table-flush-pending", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.compaction-pending", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.estimate-num-keys", &num)); ASSERT_EQ(num, "1"); perf_context.Reset(); ASSERT_OK(dbfull()->Put(writeOpt, "k2", big_value)); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "1"); ASSERT_OK(dbfull()->Delete(writeOpt, "k-non-existing")); ASSERT_OK(dbfull()->Put(writeOpt, "k3", big_value)); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.num-immutable-mem-table", &num)); ASSERT_EQ(num, "2"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.mem-table-flush-pending", &num)); ASSERT_EQ(num, "1"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.compaction-pending", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.estimate-num-keys", &num)); ASSERT_EQ(num, "2"); // Verify the same set of properties through GetIntProperty ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.num-immutable-mem-table", &int_num)); ASSERT_EQ(int_num, 2U); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.mem-table-flush-pending", &int_num)); ASSERT_EQ(int_num, 1U); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.compaction-pending", &int_num)); ASSERT_EQ(int_num, 0U); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.estimate-num-keys", &int_num)); ASSERT_EQ(int_num, 2U); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_EQ(int_num, 0U); sleeping_task_high.WakeUp(); sleeping_task_high.WaitUntilDone(); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_OK(dbfull()->Put(writeOpt, "k4", big_value)); ASSERT_OK(dbfull()->Put(writeOpt, "k5", big_value)); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.mem-table-flush-pending", &num)); ASSERT_EQ(num, "0"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.compaction-pending", &num)); ASSERT_EQ(num, "1"); ASSERT_TRUE(dbfull()->GetProperty("rocksdb.estimate-num-keys", &num)); ASSERT_EQ(num, "4"); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_GT(int_num, 0U); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); // Wait for compaction to be done. This is important because otherwise RocksDB // might schedule a compaction when reopening the database, failing assertion // (A) as a result. dbfull()->TEST_WaitForCompact(); options.max_open_files = 10; Reopen(options); // After reopening, no table reader is loaded, so no memory for table readers ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_EQ(int_num, 0U); // (A) ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.estimate-num-keys", &int_num)); ASSERT_GT(int_num, 0U); // After reading a key, at least one table reader is loaded. Get("k5"); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_GT(int_num, 0U); // Test rocksdb.num-live-versions { options.level0_file_num_compaction_trigger = 20; Reopen(options); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.num-live-versions", &int_num)); ASSERT_EQ(int_num, 1U); // Use an iterator to hold current version std::unique_ptr iter1(dbfull()->NewIterator(ReadOptions())); ASSERT_OK(dbfull()->Put(writeOpt, "k6", big_value)); Flush(); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.num-live-versions", &int_num)); ASSERT_EQ(int_num, 2U); // Use an iterator to hold current version std::unique_ptr iter2(dbfull()->NewIterator(ReadOptions())); ASSERT_OK(dbfull()->Put(writeOpt, "k7", big_value)); Flush(); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.num-live-versions", &int_num)); ASSERT_EQ(int_num, 3U); iter2.reset(); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.num-live-versions", &int_num)); ASSERT_EQ(int_num, 2U); iter1.reset(); ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.num-live-versions", &int_num)); ASSERT_EQ(int_num, 1U); } } TEST_F(DBPropertiesTest, ApproximateMemoryUsage) { const int kNumRounds = 10; // TODO(noetzli) kFlushesPerRound does not really correlate with how many // flushes happen. const int kFlushesPerRound = 10; const int kWritesPerFlush = 10; const int kKeySize = 100; const int kValueSize = 1000; Options options; options.write_buffer_size = 1000; // small write buffer options.min_write_buffer_number_to_merge = 4; options.compression = kNoCompression; options.create_if_missing = true; options = CurrentOptions(options); DestroyAndReopen(options); Random rnd(301); std::vector iters; uint64_t active_mem; uint64_t unflushed_mem; uint64_t all_mem; uint64_t prev_all_mem; // Phase 0. The verify the initial value of all these properties are the same // as we have no mem-tables. dbfull()->GetIntProperty("rocksdb.cur-size-active-mem-table", &active_mem); dbfull()->GetIntProperty("rocksdb.cur-size-all-mem-tables", &unflushed_mem); dbfull()->GetIntProperty("rocksdb.size-all-mem-tables", &all_mem); ASSERT_EQ(all_mem, active_mem); ASSERT_EQ(all_mem, unflushed_mem); // Phase 1. Simply issue Put() and expect "cur-size-all-mem-tables" equals to // "size-all-mem-tables" for (int r = 0; r < kNumRounds; ++r) { for (int f = 0; f < kFlushesPerRound; ++f) { for (int w = 0; w < kWritesPerFlush; ++w) { Put(RandomString(&rnd, kKeySize), RandomString(&rnd, kValueSize)); } } // Make sure that there is no flush between getting the two properties. dbfull()->TEST_WaitForFlushMemTable(); dbfull()->GetIntProperty("rocksdb.cur-size-all-mem-tables", &unflushed_mem); dbfull()->GetIntProperty("rocksdb.size-all-mem-tables", &all_mem); // in no iterator case, these two number should be the same. ASSERT_EQ(unflushed_mem, all_mem); } prev_all_mem = all_mem; // Phase 2. Keep issuing Put() but also create new iterators. This time we // expect "size-all-mem-tables" > "cur-size-all-mem-tables". for (int r = 0; r < kNumRounds; ++r) { iters.push_back(db_->NewIterator(ReadOptions())); for (int f = 0; f < kFlushesPerRound; ++f) { for (int w = 0; w < kWritesPerFlush; ++w) { Put(RandomString(&rnd, kKeySize), RandomString(&rnd, kValueSize)); } } // Force flush to prevent flush from happening between getting the // properties or after getting the properties and before the new round. Flush(); // In the second round, add iterators. dbfull()->GetIntProperty("rocksdb.cur-size-active-mem-table", &active_mem); dbfull()->GetIntProperty("rocksdb.cur-size-all-mem-tables", &unflushed_mem); dbfull()->GetIntProperty("rocksdb.size-all-mem-tables", &all_mem); ASSERT_GT(all_mem, active_mem); ASSERT_GT(all_mem, unflushed_mem); ASSERT_GT(all_mem, prev_all_mem); prev_all_mem = all_mem; } // Phase 3. Delete iterators and expect "size-all-mem-tables" shrinks // whenever we release an iterator. for (auto* iter : iters) { delete iter; dbfull()->GetIntProperty("rocksdb.size-all-mem-tables", &all_mem); // Expect the size shrinking ASSERT_LT(all_mem, prev_all_mem); prev_all_mem = all_mem; } // Expect all these three counters to be the same. dbfull()->GetIntProperty("rocksdb.cur-size-active-mem-table", &active_mem); dbfull()->GetIntProperty("rocksdb.cur-size-all-mem-tables", &unflushed_mem); dbfull()->GetIntProperty("rocksdb.size-all-mem-tables", &all_mem); ASSERT_EQ(active_mem, unflushed_mem); ASSERT_EQ(unflushed_mem, all_mem); // Phase 5. Reopen, and expect all these three counters to be the same again. Reopen(options); dbfull()->GetIntProperty("rocksdb.cur-size-active-mem-table", &active_mem); dbfull()->GetIntProperty("rocksdb.cur-size-all-mem-tables", &unflushed_mem); dbfull()->GetIntProperty("rocksdb.size-all-mem-tables", &all_mem); ASSERT_EQ(active_mem, unflushed_mem); ASSERT_EQ(unflushed_mem, all_mem); } TEST_F(DBPropertiesTest, EstimatePendingCompBytes) { // Set sizes to both background thread pool to be 1 and block them. env_->SetBackgroundThreads(1, Env::HIGH); env_->SetBackgroundThreads(1, Env::LOW); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); Options options = CurrentOptions(); WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; options.compaction_style = kCompactionStyleLevel; options.level0_file_num_compaction_trigger = 2; options.max_background_compactions = 1; options.max_background_flushes = 1; options.max_write_buffer_number = 10; options.min_write_buffer_number_to_merge = 1; options.max_write_buffer_number_to_maintain = 0; options.write_buffer_size = 1000000; Reopen(options); std::string big_value(1000000 * 2, 'x'); std::string num; uint64_t int_num; ASSERT_OK(dbfull()->Put(writeOpt, "k1", big_value)); Flush(); ASSERT_TRUE(dbfull()->GetIntProperty( "rocksdb.estimate-pending-compaction-bytes", &int_num)); ASSERT_EQ(int_num, 0U); ASSERT_OK(dbfull()->Put(writeOpt, "k2", big_value)); Flush(); ASSERT_TRUE(dbfull()->GetIntProperty( "rocksdb.estimate-pending-compaction-bytes", &int_num)); ASSERT_EQ(int_num, 0U); ASSERT_OK(dbfull()->Put(writeOpt, "k3", big_value)); Flush(); ASSERT_TRUE(dbfull()->GetIntProperty( "rocksdb.estimate-pending-compaction-bytes", &int_num)); ASSERT_GT(int_num, 0U); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(dbfull()->GetIntProperty( "rocksdb.estimate-pending-compaction-bytes", &int_num)); ASSERT_EQ(int_num, 0U); } #endif // ROCKSDB_LITE class CountingUserTblPropCollector : public TablePropertiesCollector { public: const char* Name() const override { return "CountingUserTblPropCollector"; } Status Finish(UserCollectedProperties* properties) override { std::string encoded; PutVarint32(&encoded, count_); *properties = UserCollectedProperties{ {"CountingUserTblPropCollector", message_}, {"Count", encoded}, }; return Status::OK(); } Status AddUserKey(const Slice& user_key, const Slice& value, EntryType type, SequenceNumber seq, uint64_t file_size) override { ++count_; return Status::OK(); } virtual UserCollectedProperties GetReadableProperties() const override { return UserCollectedProperties{}; } private: std::string message_ = "Rocksdb"; uint32_t count_ = 0; }; class CountingUserTblPropCollectorFactory : public TablePropertiesCollectorFactory { public: explicit CountingUserTblPropCollectorFactory( uint32_t expected_column_family_id) : expected_column_family_id_(expected_column_family_id), num_created_(0) {} virtual TablePropertiesCollector* CreateTablePropertiesCollector( TablePropertiesCollectorFactory::Context context) override { EXPECT_EQ(expected_column_family_id_, context.column_family_id); num_created_++; return new CountingUserTblPropCollector(); } const char* Name() const override { return "CountingUserTblPropCollectorFactory"; } void set_expected_column_family_id(uint32_t v) { expected_column_family_id_ = v; } uint32_t expected_column_family_id_; uint32_t num_created_; }; class CountingDeleteTabPropCollector : public TablePropertiesCollector { public: const char* Name() const override { return "CountingDeleteTabPropCollector"; } Status AddUserKey(const Slice& user_key, const Slice& value, EntryType type, SequenceNumber seq, uint64_t file_size) override { if (type == kEntryDelete) { num_deletes_++; } return Status::OK(); } bool NeedCompact() const override { return num_deletes_ > 10; } UserCollectedProperties GetReadableProperties() const override { return UserCollectedProperties{}; } Status Finish(UserCollectedProperties* properties) override { *properties = UserCollectedProperties{{"num_delete", ToString(num_deletes_)}}; return Status::OK(); } private: uint32_t num_deletes_ = 0; }; class CountingDeleteTabPropCollectorFactory : public TablePropertiesCollectorFactory { public: virtual TablePropertiesCollector* CreateTablePropertiesCollector( TablePropertiesCollectorFactory::Context context) override { return new CountingDeleteTabPropCollector(); } const char* Name() const override { return "CountingDeleteTabPropCollectorFactory"; } }; #ifndef ROCKSDB_LITE TEST_F(DBPropertiesTest, GetUserDefinedTableProperties) { Options options = CurrentOptions(); options.level0_file_num_compaction_trigger = (1 << 30); options.max_background_flushes = 0; options.table_properties_collector_factories.resize(1); std::shared_ptr collector_factory = std::make_shared(0); options.table_properties_collector_factories[0] = collector_factory; Reopen(options); // Create 4 tables for (int table = 0; table < 4; ++table) { for (int i = 0; i < 10 + table; ++i) { db_->Put(WriteOptions(), ToString(table * 100 + i), "val"); } db_->Flush(FlushOptions()); } TablePropertiesCollection props; ASSERT_OK(db_->GetPropertiesOfAllTables(&props)); ASSERT_EQ(4U, props.size()); uint32_t sum = 0; for (const auto& item : props) { auto& user_collected = item.second->user_collected_properties; ASSERT_TRUE(user_collected.find("CountingUserTblPropCollector") != user_collected.end()); ASSERT_EQ(user_collected.at("CountingUserTblPropCollector"), "Rocksdb"); ASSERT_TRUE(user_collected.find("Count") != user_collected.end()); Slice key(user_collected.at("Count")); uint32_t count; ASSERT_TRUE(GetVarint32(&key, &count)); sum += count; } ASSERT_EQ(10u + 11u + 12u + 13u, sum); ASSERT_GT(collector_factory->num_created_, 0U); collector_factory->num_created_ = 0; dbfull()->TEST_CompactRange(0, nullptr, nullptr); ASSERT_GT(collector_factory->num_created_, 0U); } #endif // ROCKSDB_LITE TEST_F(DBPropertiesTest, UserDefinedTablePropertiesContext) { Options options = CurrentOptions(); options.level0_file_num_compaction_trigger = 3; options.max_background_flushes = 0; options.table_properties_collector_factories.resize(1); std::shared_ptr collector_factory = std::make_shared(1); options.table_properties_collector_factories[0] = collector_factory, CreateAndReopenWithCF({"pikachu"}, options); // Create 2 files for (int table = 0; table < 2; ++table) { for (int i = 0; i < 10 + table; ++i) { Put(1, ToString(table * 100 + i), "val"); } Flush(1); } ASSERT_GT(collector_factory->num_created_, 0U); collector_factory->num_created_ = 0; // Trigger automatic compactions. for (int table = 0; table < 3; ++table) { for (int i = 0; i < 10 + table; ++i) { Put(1, ToString(table * 100 + i), "val"); } Flush(1); dbfull()->TEST_WaitForCompact(); } ASSERT_GT(collector_factory->num_created_, 0U); collector_factory->num_created_ = 0; dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); ASSERT_GT(collector_factory->num_created_, 0U); // Come back to write to default column family collector_factory->num_created_ = 0; collector_factory->set_expected_column_family_id(0); // default CF // Create 4 tables in default column family for (int table = 0; table < 2; ++table) { for (int i = 0; i < 10 + table; ++i) { Put(ToString(table * 100 + i), "val"); } Flush(); } ASSERT_GT(collector_factory->num_created_, 0U); collector_factory->num_created_ = 0; // Trigger automatic compactions. for (int table = 0; table < 3; ++table) { for (int i = 0; i < 10 + table; ++i) { Put(ToString(table * 100 + i), "val"); } Flush(); dbfull()->TEST_WaitForCompact(); } ASSERT_GT(collector_factory->num_created_, 0U); collector_factory->num_created_ = 0; dbfull()->TEST_CompactRange(0, nullptr, nullptr); ASSERT_GT(collector_factory->num_created_, 0U); } #ifndef ROCKSDB_LITE TEST_F(DBPropertiesTest, TablePropertiesNeedCompactTest) { Random rnd(301); Options options; options.create_if_missing = true; options.write_buffer_size = 4096; options.max_write_buffer_number = 8; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 4; options.target_file_size_base = 2048; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.soft_pending_compaction_bytes_limit = 1024 * 1024; options.num_levels = 8; std::shared_ptr collector_factory = std::make_shared(); options.table_properties_collector_factories.resize(1); options.table_properties_collector_factories[0] = collector_factory; DestroyAndReopen(options); const int kMaxKey = 1000; for (int i = 0; i < kMaxKey; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 102))); ASSERT_OK(Put(Key(kMaxKey + i), RandomString(&rnd, 102))); } Flush(); dbfull()->TEST_WaitForCompact(); if (NumTableFilesAtLevel(0) == 1) { // Clear Level 0 so that when later flush a file with deletions, // we don't trigger an organic compaction. ASSERT_OK(Put(Key(0), "")); ASSERT_OK(Put(Key(kMaxKey * 2), "")); Flush(); dbfull()->TEST_WaitForCompact(); } ASSERT_EQ(NumTableFilesAtLevel(0), 0); { int c = 0; std::unique_ptr iter(db_->NewIterator(ReadOptions())); iter->Seek(Key(kMaxKey - 100)); while (iter->Valid() && iter->key().compare(Key(kMaxKey + 100)) < 0) { iter->Next(); ++c; } ASSERT_EQ(c, 200); } Delete(Key(0)); for (int i = kMaxKey - 100; i < kMaxKey + 100; i++) { Delete(Key(i)); } Delete(Key(kMaxKey * 2)); Flush(); dbfull()->TEST_WaitForCompact(); { SetPerfLevel(kEnableCount); perf_context.Reset(); int c = 0; std::unique_ptr iter(db_->NewIterator(ReadOptions())); iter->Seek(Key(kMaxKey - 100)); while (iter->Valid() && iter->key().compare(Key(kMaxKey + 100)) < 0) { iter->Next(); } ASSERT_EQ(c, 0); ASSERT_LT(perf_context.internal_delete_skipped_count, 30u); ASSERT_LT(perf_context.internal_key_skipped_count, 30u); SetPerfLevel(kDisable); } } TEST_F(DBPropertiesTest, NeedCompactHintPersistentTest) { Random rnd(301); Options options; options.create_if_missing = true; options.max_write_buffer_number = 8; options.level0_file_num_compaction_trigger = 10; options.level0_slowdown_writes_trigger = 10; options.level0_stop_writes_trigger = 10; options.disable_auto_compactions = true; std::shared_ptr collector_factory = std::make_shared(); options.table_properties_collector_factories.resize(1); options.table_properties_collector_factories[0] = collector_factory; DestroyAndReopen(options); const int kMaxKey = 100; for (int i = 0; i < kMaxKey; i++) { ASSERT_OK(Put(Key(i), "")); } Flush(); dbfull()->TEST_WaitForFlushMemTable(); for (int i = 1; i < kMaxKey - 1; i++) { Delete(Key(i)); } Flush(); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(NumTableFilesAtLevel(0), 2); // Restart the DB. Although number of files didn't reach // options.level0_file_num_compaction_trigger, compaction should // still be triggered because of the need-compaction hint. options.disable_auto_compactions = false; Reopen(options); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(0), 0); { SetPerfLevel(kEnableCount); perf_context.Reset(); int c = 0; std::unique_ptr iter(db_->NewIterator(ReadOptions())); for (iter->Seek(Key(0)); iter->Valid(); iter->Next()) { c++; } ASSERT_EQ(c, 2); ASSERT_EQ(perf_context.internal_delete_skipped_count, 0); // We iterate every key twice. Is it a bug? ASSERT_LE(perf_context.internal_key_skipped_count, 2); SetPerfLevel(kDisable); } } #endif // ROCKSDB_LITE } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/db_table_properties_test.cc000066400000000000000000000156511267530634000211310ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include #include #include "db/db_test_util.h" #include "port/stack_trace.h" #include "rocksdb/db.h" #include "util/testharness.h" #include "util/testutil.h" #ifndef ROCKSDB_LITE namespace rocksdb { // A helper function that ensures the table properties returned in // `GetPropertiesOfAllTablesTest` is correct. // This test assumes entries size is different for each of the tables. namespace { void VerifyTableProperties(DB* db, uint64_t expected_entries_size) { TablePropertiesCollection props; ASSERT_OK(db->GetPropertiesOfAllTables(&props)); ASSERT_EQ(4U, props.size()); std::unordered_set unique_entries; // Indirect test uint64_t sum = 0; for (const auto& item : props) { unique_entries.insert(item.second->num_entries); sum += item.second->num_entries; } ASSERT_EQ(props.size(), unique_entries.size()); ASSERT_EQ(expected_entries_size, sum); } } // namespace class DBTablePropertiesTest : public DBTestBase { public: DBTablePropertiesTest() : DBTestBase("/db_table_properties_test") {} TablePropertiesCollection TestGetPropertiesOfTablesInRange( std::vector ranges, std::size_t* num_properties = nullptr, std::size_t* num_files = nullptr); }; TEST_F(DBTablePropertiesTest, GetPropertiesOfAllTablesTest) { Options options = CurrentOptions(); options.level0_file_num_compaction_trigger = 8; Reopen(options); // Create 4 tables for (int table = 0; table < 4; ++table) { for (int i = 0; i < 10 + table; ++i) { db_->Put(WriteOptions(), ToString(table * 100 + i), "val"); } db_->Flush(FlushOptions()); } // 1. Read table properties directly from file Reopen(options); VerifyTableProperties(db_, 10 + 11 + 12 + 13); // 2. Put two tables to table cache and Reopen(options); // fetch key from 1st and 2nd table, which will internally place that table to // the table cache. for (int i = 0; i < 2; ++i) { Get(ToString(i * 100 + 0)); } VerifyTableProperties(db_, 10 + 11 + 12 + 13); // 3. Put all tables to table cache Reopen(options); // fetch key from 1st and 2nd table, which will internally place that table to // the table cache. for (int i = 0; i < 4; ++i) { Get(ToString(i * 100 + 0)); } VerifyTableProperties(db_, 10 + 11 + 12 + 13); } TablePropertiesCollection DBTablePropertiesTest::TestGetPropertiesOfTablesInRange( std::vector ranges, std::size_t* num_properties, std::size_t* num_files) { // Since we deref zero element in the vector it can not be empty // otherwise we pass an address to some random memory EXPECT_GT(ranges.size(), 0U); // run the query TablePropertiesCollection props; EXPECT_OK(db_->GetPropertiesOfTablesInRange( db_->DefaultColumnFamily(), &ranges[0], ranges.size(), &props)); // Make sure that we've received properties for those and for those files // only which fall within requested ranges std::vector vmd; db_->GetLiveFilesMetaData(&vmd); for (auto& md : vmd) { std::string fn = md.db_path + md.name; bool in_range = false; for (auto& r : ranges) { // smallestkey < limit && largestkey >= start if (r.limit.compare(md.smallestkey) >= 0 && r.start.compare(md.largestkey) <= 0) { in_range = true; EXPECT_GT(props.count(fn), 0); } } if (!in_range) { EXPECT_EQ(props.count(fn), 0); } } if (num_properties) { *num_properties = props.size(); } if (num_files) { *num_files = vmd.size(); } return props; } TEST_F(DBTablePropertiesTest, GetPropertiesOfTablesInRange) { // Fixed random sead Random rnd(301); Options options; options.create_if_missing = true; options.write_buffer_size = 4096; options.max_write_buffer_number = 8; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 4; options.target_file_size_base = 2048; options.max_bytes_for_level_base = 10240; options.max_bytes_for_level_multiplier = 4; options.soft_rate_limit = 1.1; options.num_levels = 8; DestroyAndReopen(options); // build a decent LSM for (int i = 0; i < 10000; i++) { ASSERT_OK(Put(test::RandomKey(&rnd, 5), RandomString(&rnd, 102))); } Flush(); db_->PauseBackgroundWork(); // Ensure that we have at least L0, L1 and L2 ASSERT_GT(NumTableFilesAtLevel(0), 0); ASSERT_GT(NumTableFilesAtLevel(1), 0); ASSERT_GT(NumTableFilesAtLevel(2), 0); // Query the largest range std::size_t num_properties, num_files; TestGetPropertiesOfTablesInRange( {Range(test::RandomKey(&rnd, 5, test::RandomKeyType::SMALLEST), test::RandomKey(&rnd, 5, test::RandomKeyType::LARGEST))}, &num_properties, &num_files); ASSERT_EQ(num_properties, num_files); // Query the empty range TestGetPropertiesOfTablesInRange( {Range(test::RandomKey(&rnd, 5, test::RandomKeyType::LARGEST), test::RandomKey(&rnd, 5, test::RandomKeyType::SMALLEST))}, &num_properties, &num_files); ASSERT_GT(num_files, 0); ASSERT_EQ(num_properties, 0); // Query the middle rangee TestGetPropertiesOfTablesInRange( {Range(test::RandomKey(&rnd, 5, test::RandomKeyType::MIDDLE), test::RandomKey(&rnd, 5, test::RandomKeyType::LARGEST))}, &num_properties, &num_files); ASSERT_GT(num_files, 0); ASSERT_GT(num_files, num_properties); ASSERT_GT(num_properties, 0); // Query a bunch of random ranges for (int j = 0; j < 100; j++) { // create a bunch of ranges std::vector random_keys; // Random returns numbers with zero included // when we pass empty ranges TestGetPropertiesOfTablesInRange() // derefs random memory in the empty ranges[0] // so want to be greater than zero and even since // the below loop requires that random_keys.size() to be even. auto n = 2 * (rnd.Uniform(50) + 1); for (uint32_t i = 0; i < n; ++i) { random_keys.push_back(test::RandomKey(&rnd, 5)); } ASSERT_GT(random_keys.size(), 0U); ASSERT_EQ((random_keys.size() % 2), 0U); std::vector ranges; auto it = random_keys.begin(); while (it != random_keys.end()) { ranges.push_back(Range(*it, *(it + 1))); it += 2; } TestGetPropertiesOfTablesInRange(std::move(ranges)); } } } // namespace rocksdb #endif // ROCKSDB_LITE int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/db_tailing_iter_test.cc000066400000000000000000000477371267530634000202520ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Introduction of SyncPoint effectively disabled building and running this test // in Release build. // which is a pity, it is a good test #if !defined(ROCKSDB_LITE) #include "db/db_test_util.h" #include "db/forward_iterator.h" #include "port/stack_trace.h" namespace rocksdb { class DBTestTailingIterator : public DBTestBase { public: DBTestTailingIterator() : DBTestBase("/db_tailing_iterator_test") {} }; TEST_F(DBTestTailingIterator, TailingIteratorSingle) { ReadOptions read_options; read_options.tailing = true; std::unique_ptr iter(db_->NewIterator(read_options)); iter->SeekToFirst(); ASSERT_TRUE(!iter->Valid()); // add a record and check that iter can see it ASSERT_OK(db_->Put(WriteOptions(), "mirko", "fodor")); iter->SeekToFirst(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().ToString(), "mirko"); iter->Next(); ASSERT_TRUE(!iter->Valid()); } TEST_F(DBTestTailingIterator, TailingIteratorKeepAdding) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); std::string value(1024, 'a'); const int num_records = 10000; for (int i = 0; i < num_records; ++i) { char buf[32]; snprintf(buf, sizeof(buf), "%016d", i); Slice key(buf, 16); ASSERT_OK(Put(1, key, value)); iter->Seek(key); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); } } TEST_F(DBTestTailingIterator, TailingIteratorSeekToNext) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); std::unique_ptr itern(db_->NewIterator(read_options, handles_[1])); std::string value(1024, 'a'); const int num_records = 1000; for (int i = 1; i < num_records; ++i) { char buf1[32]; char buf2[32]; snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5); Slice key(buf1, 20); ASSERT_OK(Put(1, key, value)); if (i % 100 == 99) { ASSERT_OK(Flush(1)); } snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2); Slice target(buf2, 20); iter->Seek(target); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); if (i == 1) { itern->SeekToFirst(); } else { itern->Next(); } ASSERT_TRUE(itern->Valid()); ASSERT_EQ(itern->key().compare(key), 0); } rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); for (int i = 2 * num_records; i > 0; --i) { char buf1[32]; char buf2[32]; snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5); Slice key(buf1, 20); ASSERT_OK(Put(1, key, value)); if (i % 100 == 99) { ASSERT_OK(Flush(1)); } snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2); Slice target(buf2, 20); iter->Seek(target); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); } } TEST_F(DBTestTailingIterator, TailingIteratorTrimSeekToNext) { const uint64_t k150KB = 150 * 1024; Options options; options.write_buffer_size = k150KB; options.max_write_buffer_number = 3; options.min_write_buffer_number_to_merge = 2; CreateAndReopenWithCF({"pikachu"}, options); ReadOptions read_options; read_options.tailing = true; int num_iters, deleted_iters; char bufe[32]; snprintf(bufe, sizeof(bufe), "00b0%016d", 0); Slice keyu(bufe, 20); read_options.iterate_upper_bound = &keyu; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); std::unique_ptr itern(db_->NewIterator(read_options, handles_[1])); std::unique_ptr iterh(db_->NewIterator(read_options, handles_[1])); std::string value(1024, 'a'); bool file_iters_deleted = false; bool file_iters_renewed_null = false; bool file_iters_renewed_copy = false; rocksdb::SyncPoint::GetInstance()->SetCallBack( "ForwardIterator::SeekInternal:Return", [&](void* arg) { ForwardIterator* fiter = reinterpret_cast(arg); ASSERT_TRUE(!file_iters_deleted || fiter->TEST_CheckDeletedIters(&deleted_iters, &num_iters)); }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "ForwardIterator::Next:Return", [&](void* arg) { ForwardIterator* fiter = reinterpret_cast(arg); ASSERT_TRUE(!file_iters_deleted || fiter->TEST_CheckDeletedIters(&deleted_iters, &num_iters)); }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "ForwardIterator::RenewIterators:Null", [&](void* arg) { file_iters_renewed_null = true; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "ForwardIterator::RenewIterators:Copy", [&](void* arg) { file_iters_renewed_copy = true; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); const int num_records = 1000; for (int i = 1; i < num_records; ++i) { char buf1[32]; char buf2[32]; char buf3[32]; char buf4[32]; snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5); snprintf(buf3, sizeof(buf3), "00b0%016d", i * 5); Slice key(buf1, 20); ASSERT_OK(Put(1, key, value)); Slice keyn(buf3, 20); ASSERT_OK(Put(1, keyn, value)); if (i % 100 == 99) { ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); if (i == 299) { file_iters_deleted = true; } snprintf(buf4, sizeof(buf4), "00a0%016d", i * 5 / 2); Slice target(buf4, 20); iterh->Seek(target); ASSERT_TRUE(iter->Valid()); for (int j = (i + 1) * 5 / 2; j < i * 5; j += 5) { iterh->Next(); ASSERT_TRUE(iterh->Valid()); } if (i == 299) { file_iters_deleted = false; } } file_iters_deleted = true; snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2); Slice target(buf2, 20); iter->Seek(target); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); ASSERT_LE(num_iters, 1); if (i == 1) { itern->SeekToFirst(); } else { itern->Next(); } ASSERT_TRUE(itern->Valid()); ASSERT_EQ(itern->key().compare(key), 0); ASSERT_LE(num_iters, 1); file_iters_deleted = false; } ASSERT_TRUE(file_iters_renewed_null); ASSERT_TRUE(file_iters_renewed_copy); iter = 0; itern = 0; iterh = 0; BlockBasedTableOptions table_options; table_options.no_block_cache = true; table_options.block_cache_compressed = nullptr; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); ReopenWithColumnFamilies({"default", "pikachu"}, options); read_options.read_tier = kBlockCacheTier; std::unique_ptr iteri(db_->NewIterator(read_options, handles_[1])); char buf5[32]; snprintf(buf5, sizeof(buf5), "00a0%016d", (num_records / 2) * 5 - 2); Slice target1(buf5, 20); iteri->Seek(target1); ASSERT_TRUE(iteri->status().IsIncomplete()); iteri = 0; read_options.read_tier = kReadAllTier; options.table_factory.reset(NewBlockBasedTableFactory()); ReopenWithColumnFamilies({"default", "pikachu"}, options); iter.reset(db_->NewIterator(read_options, handles_[1])); for (int i = 2 * num_records; i > 0; --i) { char buf1[32]; char buf2[32]; snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5); Slice key(buf1, 20); ASSERT_OK(Put(1, key, value)); if (i % 100 == 99) { ASSERT_OK(Flush(1)); } snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2); Slice target(buf2, 20); iter->Seek(target); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); } } TEST_F(DBTestTailingIterator, TailingIteratorDeletes) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); // write a single record, read it using the iterator, then delete it ASSERT_OK(Put(1, "0test", "test")); iter->SeekToFirst(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().ToString(), "0test"); ASSERT_OK(Delete(1, "0test")); // write many more records const int num_records = 10000; std::string value(1024, 'A'); for (int i = 0; i < num_records; ++i) { char buf[32]; snprintf(buf, sizeof(buf), "1%015d", i); Slice key(buf, 16); ASSERT_OK(Put(1, key, value)); } // force a flush to make sure that no records are read from memtable ASSERT_OK(Flush(1)); // skip "0test" iter->Next(); // make sure we can read all new records using the existing iterator int count = 0; for (; iter->Valid(); iter->Next(), ++count) ; ASSERT_EQ(count, num_records); } TEST_F(DBTestTailingIterator, TailingIteratorPrefixSeek) { XFUNC_TEST("", "dbtest_prefix", prefix_skip1, XFuncPoint::SetSkip, kSkipNoPrefix); ReadOptions read_options; read_options.tailing = true; Options options = CurrentOptions(); options.env = env_; options.create_if_missing = true; options.disable_auto_compactions = true; options.prefix_extractor.reset(NewFixedPrefixTransform(2)); options.memtable_factory.reset(NewHashSkipListRepFactory(16)); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); ASSERT_OK(Put(1, "0101", "test")); ASSERT_OK(Flush(1)); ASSERT_OK(Put(1, "0202", "test")); // Seek(0102) shouldn't find any records since 0202 has a different prefix iter->Seek("0102"); ASSERT_TRUE(!iter->Valid()); iter->Seek("0202"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().ToString(), "0202"); iter->Next(); ASSERT_TRUE(!iter->Valid()); XFUNC_TEST("", "dbtest_prefix", prefix_skip1, XFuncPoint::SetSkip, 0); } TEST_F(DBTestTailingIterator, TailingIteratorIncomplete) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; read_options.read_tier = kBlockCacheTier; std::string key("key"); std::string value("value"); ASSERT_OK(db_->Put(WriteOptions(), key, value)); std::unique_ptr iter(db_->NewIterator(read_options)); iter->SeekToFirst(); // we either see the entry or it's not in cache ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete()); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); iter->SeekToFirst(); // should still be true after compaction ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete()); } TEST_F(DBTestTailingIterator, TailingIteratorSeekToSame) { Options options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 1000; CreateAndReopenWithCF({"pikachu"}, options); ReadOptions read_options; read_options.tailing = true; const int NROWS = 10000; // Write rows with keys 00000, 00002, 00004 etc. for (int i = 0; i < NROWS; ++i) { char buf[100]; snprintf(buf, sizeof(buf), "%05d", 2*i); std::string key(buf); std::string value("value"); ASSERT_OK(db_->Put(WriteOptions(), key, value)); } std::unique_ptr iter(db_->NewIterator(read_options)); // Seek to 00001. We expect to find 00002. std::string start_key = "00001"; iter->Seek(start_key); ASSERT_TRUE(iter->Valid()); std::string found = iter->key().ToString(); ASSERT_EQ("00002", found); // Now seek to the same key. The iterator should remain in the same // position. iter->Seek(found); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(found, iter->key().ToString()); } // Sets iterate_upper_bound and verifies that ForwardIterator doesn't call // Seek() on immutable iterators when target key is >= prev_key and all // iterators, including the memtable iterator, are over the upper bound. TEST_F(DBTestTailingIterator, TailingIteratorUpperBound) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); const Slice upper_bound("20", 3); ReadOptions read_options; read_options.tailing = true; read_options.iterate_upper_bound = &upper_bound; ASSERT_OK(Put(1, "11", "11")); ASSERT_OK(Put(1, "12", "12")); ASSERT_OK(Put(1, "22", "22")); ASSERT_OK(Flush(1)); // flush all those keys to an immutable SST file // Add another key to the memtable. ASSERT_OK(Put(1, "21", "21")); std::unique_ptr it(db_->NewIterator(read_options, handles_[1])); it->Seek("12"); ASSERT_TRUE(it->Valid()); ASSERT_EQ("12", it->key().ToString()); it->Next(); // Not valid since "21" is over the upper bound. ASSERT_FALSE(it->Valid()); // This keeps track of the number of times NeedToSeekImmutable() was true. int immutable_seeks = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "ForwardIterator::SeekInternal:Immutable", [&](void* arg) { ++immutable_seeks; }); // Seek to 13. This should not require any immutable seeks. rocksdb::SyncPoint::GetInstance()->EnableProcessing(); it->Seek("13"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); ASSERT_FALSE(it->Valid()); ASSERT_EQ(0, immutable_seeks); } TEST_F(DBTestTailingIterator, ManagedTailingIteratorSingle) { ReadOptions read_options; read_options.tailing = true; read_options.managed = true; std::unique_ptr iter(db_->NewIterator(read_options)); iter->SeekToFirst(); ASSERT_TRUE(!iter->Valid()); // add a record and check that iter can see it ASSERT_OK(db_->Put(WriteOptions(), "mirko", "fodor")); iter->SeekToFirst(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().ToString(), "mirko"); iter->Next(); ASSERT_TRUE(!iter->Valid()); } TEST_F(DBTestTailingIterator, ManagedTailingIteratorKeepAdding) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; read_options.managed = true; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); std::string value(1024, 'a'); const int num_records = 10000; for (int i = 0; i < num_records; ++i) { char buf[32]; snprintf(buf, sizeof(buf), "%016d", i); Slice key(buf, 16); ASSERT_OK(Put(1, key, value)); iter->Seek(key); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); } } TEST_F(DBTestTailingIterator, ManagedTailingIteratorSeekToNext) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; read_options.managed = true; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); std::string value(1024, 'a'); const int num_records = 1000; for (int i = 1; i < num_records; ++i) { char buf1[32]; char buf2[32]; snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5); Slice key(buf1, 20); ASSERT_OK(Put(1, key, value)); if (i % 100 == 99) { ASSERT_OK(Flush(1)); } snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2); Slice target(buf2, 20); iter->Seek(target); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); } for (int i = 2 * num_records; i > 0; --i) { char buf1[32]; char buf2[32]; snprintf(buf1, sizeof(buf1), "00a0%016d", i * 5); Slice key(buf1, 20); ASSERT_OK(Put(1, key, value)); if (i % 100 == 99) { ASSERT_OK(Flush(1)); } snprintf(buf2, sizeof(buf2), "00a0%016d", i * 5 - 2); Slice target(buf2, 20); iter->Seek(target); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(key), 0); } } TEST_F(DBTestTailingIterator, ManagedTailingIteratorDeletes) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; read_options.managed = true; std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); // write a single record, read it using the iterator, then delete it ASSERT_OK(Put(1, "0test", "test")); iter->SeekToFirst(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().ToString(), "0test"); ASSERT_OK(Delete(1, "0test")); // write many more records const int num_records = 10000; std::string value(1024, 'A'); for (int i = 0; i < num_records; ++i) { char buf[32]; snprintf(buf, sizeof(buf), "1%015d", i); Slice key(buf, 16); ASSERT_OK(Put(1, key, value)); } // force a flush to make sure that no records are read from memtable ASSERT_OK(Flush(1)); // skip "0test" iter->Next(); // make sure we can read all new records using the existing iterator int count = 0; for (; iter->Valid(); iter->Next(), ++count) { } ASSERT_EQ(count, num_records); } TEST_F(DBTestTailingIterator, ManagedTailingIteratorPrefixSeek) { XFUNC_TEST("", "dbtest_prefix", prefix_skip1, XFuncPoint::SetSkip, kSkipNoPrefix); ReadOptions read_options; read_options.tailing = true; read_options.managed = true; Options options = CurrentOptions(); options.env = env_; options.create_if_missing = true; options.disable_auto_compactions = true; options.prefix_extractor.reset(NewFixedPrefixTransform(2)); options.memtable_factory.reset(NewHashSkipListRepFactory(16)); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); std::unique_ptr iter(db_->NewIterator(read_options, handles_[1])); ASSERT_OK(Put(1, "0101", "test")); ASSERT_OK(Flush(1)); ASSERT_OK(Put(1, "0202", "test")); // Seek(0102) shouldn't find any records since 0202 has a different prefix iter->Seek("0102"); ASSERT_TRUE(!iter->Valid()); iter->Seek("0202"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().ToString(), "0202"); iter->Next(); ASSERT_TRUE(!iter->Valid()); XFUNC_TEST("", "dbtest_prefix", prefix_skip1, XFuncPoint::SetSkip, 0); } TEST_F(DBTestTailingIterator, ManagedTailingIteratorIncomplete) { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ReadOptions read_options; read_options.tailing = true; read_options.managed = true; read_options.read_tier = kBlockCacheTier; std::string key = "key"; std::string value = "value"; ASSERT_OK(db_->Put(WriteOptions(), key, value)); std::unique_ptr iter(db_->NewIterator(read_options)); iter->SeekToFirst(); // we either see the entry or it's not in cache ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete()); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); iter->SeekToFirst(); // should still be true after compaction ASSERT_TRUE(iter->Valid() || iter->status().IsIncomplete()); } TEST_F(DBTestTailingIterator, ManagedTailingIteratorSeekToSame) { Options options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 1000; CreateAndReopenWithCF({"pikachu"}, options); ReadOptions read_options; read_options.tailing = true; read_options.managed = true; const int NROWS = 10000; // Write rows with keys 00000, 00002, 00004 etc. for (int i = 0; i < NROWS; ++i) { char buf[100]; snprintf(buf, sizeof(buf), "%05d", 2 * i); std::string key(buf); std::string value("value"); ASSERT_OK(db_->Put(WriteOptions(), key, value)); } std::unique_ptr iter(db_->NewIterator(read_options)); // Seek to 00001. We expect to find 00002. std::string start_key = "00001"; iter->Seek(start_key); ASSERT_TRUE(iter->Valid()); std::string found = iter->key().ToString(); ASSERT_EQ("00002", found); // Now seek to the same key. The iterator should remain in the same // position. iter->Seek(found); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(found, iter->key().ToString()); } } // namespace rocksdb #endif // !defined(ROCKSDB_LITE) int main(int argc, char** argv) { #if !defined(ROCKSDB_LITE) rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); #else return 0; #endif } rocksdb-4.5.1/db/db_test.cc000066400000000000000000012762521267530634000155150ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Introduction of SyncPoint effectively disabled building and running this test // in Release build. // which is a pity, it is a good test #include #include #include #include #include #include #include #ifndef OS_WIN #include #endif #ifdef OS_SOLARIS #include #endif #include "db/filename.h" #include "db/dbformat.h" #include "db/db_impl.h" #include "db/db_test_util.h" #include "db/job_context.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "memtable/hash_linklist_rep.h" #include "port/stack_trace.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/convenience.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/experimental.h" #include "rocksdb/sst_file_manager.h" #include "rocksdb/filter_policy.h" #include "rocksdb/options.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "rocksdb/snapshot.h" #include "rocksdb/sst_file_writer.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "rocksdb/thread_status.h" #include "rocksdb/wal_filter.h" #include "rocksdb/utilities/write_batch_with_index.h" #include "rocksdb/utilities/checkpoint.h" #include "rocksdb/utilities/optimistic_transaction_db.h" #include "table/block_based_table_factory.h" #include "table/mock_table.h" #include "table/plain_table_factory.h" #include "table/scoped_arena_iterator.h" #include "util/file_reader_writer.h" #include "util/hash.h" #include "utilities/merge_operators.h" #include "util/logging.h" #include "util/compression.h" #include "util/mutexlock.h" #include "util/rate_limiter.h" #include "util/sst_file_manager_impl.h" #include "util/statistics.h" #include "util/sync_point.h" #include "util/testharness.h" #include "util/testutil.h" #include "util/mock_env.h" #include "util/string_util.h" #include "util/thread_status_util.h" #include "util/xfunc.h" namespace rocksdb { static uint64_t TestGetTickerCount(const Options& options, Tickers ticker_type) { return options.statistics->getTickerCount(ticker_type); } #ifndef ROCKSDB_LITE // A helper function that ensures the table properties returned in // `GetPropertiesOfAllTablesTest` is correct. // This test assumes entries size is different for each of the tables. namespace { uint64_t GetNumberOfSstFilesForColumnFamily(DB* db, std::string column_family_name) { std::vector metadata; db->GetLiveFilesMetaData(&metadata); uint64_t result = 0; for (auto& fileMetadata : metadata) { result += (fileMetadata.column_family_name == column_family_name); } return result; } } // namespace #endif // ROCKSDB_LITE class DBTest : public DBTestBase { public: DBTest() : DBTestBase("/db_test") {} }; class DBTestWithParam : public DBTest, public testing::WithParamInterface> { public: DBTestWithParam() { max_subcompactions_ = std::get<0>(GetParam()); exclusive_manual_compaction_ = std::get<1>(GetParam()); } // Required if inheriting from testing::WithParamInterface<> static void SetUpTestCase() {} static void TearDownTestCase() {} uint32_t max_subcompactions_; bool exclusive_manual_compaction_; }; TEST_F(DBTest, MockEnvTest) { unique_ptr env{new MockEnv(Env::Default())}; Options options; options.create_if_missing = true; options.env = env.get(); DB* db; const Slice keys[] = {Slice("aaa"), Slice("bbb"), Slice("ccc")}; const Slice vals[] = {Slice("foo"), Slice("bar"), Slice("baz")}; ASSERT_OK(DB::Open(options, "/dir/db", &db)); for (size_t i = 0; i < 3; ++i) { ASSERT_OK(db->Put(WriteOptions(), keys[i], vals[i])); } for (size_t i = 0; i < 3; ++i) { std::string res; ASSERT_OK(db->Get(ReadOptions(), keys[i], &res)); ASSERT_TRUE(res == vals[i]); } Iterator* iterator = db->NewIterator(ReadOptions()); iterator->SeekToFirst(); for (size_t i = 0; i < 3; ++i) { ASSERT_TRUE(iterator->Valid()); ASSERT_TRUE(keys[i] == iterator->key()); ASSERT_TRUE(vals[i] == iterator->value()); iterator->Next(); } ASSERT_TRUE(!iterator->Valid()); delete iterator; // TEST_FlushMemTable() is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE DBImpl* dbi = reinterpret_cast(db); ASSERT_OK(dbi->TEST_FlushMemTable()); for (size_t i = 0; i < 3; ++i) { std::string res; ASSERT_OK(db->Get(ReadOptions(), keys[i], &res)); ASSERT_TRUE(res == vals[i]); } #endif // ROCKSDB_LITE delete db; } // NewMemEnv returns nullptr in ROCKSDB_LITE since class InMemoryEnv isn't // defined. #ifndef ROCKSDB_LITE TEST_F(DBTest, MemEnvTest) { unique_ptr env{NewMemEnv(Env::Default())}; Options options; options.create_if_missing = true; options.env = env.get(); DB* db; const Slice keys[] = {Slice("aaa"), Slice("bbb"), Slice("ccc")}; const Slice vals[] = {Slice("foo"), Slice("bar"), Slice("baz")}; ASSERT_OK(DB::Open(options, "/dir/db", &db)); for (size_t i = 0; i < 3; ++i) { ASSERT_OK(db->Put(WriteOptions(), keys[i], vals[i])); } for (size_t i = 0; i < 3; ++i) { std::string res; ASSERT_OK(db->Get(ReadOptions(), keys[i], &res)); ASSERT_TRUE(res == vals[i]); } Iterator* iterator = db->NewIterator(ReadOptions()); iterator->SeekToFirst(); for (size_t i = 0; i < 3; ++i) { ASSERT_TRUE(iterator->Valid()); ASSERT_TRUE(keys[i] == iterator->key()); ASSERT_TRUE(vals[i] == iterator->value()); iterator->Next(); } ASSERT_TRUE(!iterator->Valid()); delete iterator; DBImpl* dbi = reinterpret_cast(db); ASSERT_OK(dbi->TEST_FlushMemTable()); for (size_t i = 0; i < 3; ++i) { std::string res; ASSERT_OK(db->Get(ReadOptions(), keys[i], &res)); ASSERT_TRUE(res == vals[i]); } delete db; options.create_if_missing = false; ASSERT_OK(DB::Open(options, "/dir/db", &db)); for (size_t i = 0; i < 3; ++i) { std::string res; ASSERT_OK(db->Get(ReadOptions(), keys[i], &res)); ASSERT_TRUE(res == vals[i]); } delete db; } #endif // ROCKSDB_LITE TEST_F(DBTest, WriteEmptyBatch) { Options options; options.env = env_; options.write_buffer_size = 100000; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "foo", "bar")); env_->sync_counter_.store(0); WriteOptions wo; wo.sync = true; wo.disableWAL = false; WriteBatch empty_batch; ASSERT_OK(dbfull()->Write(wo, &empty_batch)); ASSERT_GE(env_->sync_counter_.load(), 1); // make sure we can re-open it. ASSERT_OK(TryReopenWithColumnFamilies({"default", "pikachu"}, options)); ASSERT_EQ("bar", Get(1, "foo")); } #ifndef ROCKSDB_LITE TEST_F(DBTest, ReadOnlyDB) { ASSERT_OK(Put("foo", "v1")); ASSERT_OK(Put("bar", "v2")); ASSERT_OK(Put("foo", "v3")); Close(); auto options = CurrentOptions(); assert(options.env = env_); ASSERT_OK(ReadOnlyReopen(options)); ASSERT_EQ("v3", Get("foo")); ASSERT_EQ("v2", Get("bar")); Iterator* iter = db_->NewIterator(ReadOptions()); int count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(iter->status()); ++count; } ASSERT_EQ(count, 2); delete iter; Close(); // Reopen and flush memtable. Reopen(options); Flush(); Close(); // Now check keys in read only mode. ASSERT_OK(ReadOnlyReopen(options)); ASSERT_EQ("v3", Get("foo")); ASSERT_EQ("v2", Get("bar")); ASSERT_TRUE(db_->SyncWAL().IsNotSupported()); } TEST_F(DBTest, CompactedDB) { const uint64_t kFileSize = 1 << 20; Options options; options.disable_auto_compactions = true; options.write_buffer_size = kFileSize; options.target_file_size_base = kFileSize; options.max_bytes_for_level_base = 1 << 30; options.compression = kNoCompression; options = CurrentOptions(options); Reopen(options); // 1 L0 file, use CompactedDB if max_open_files = -1 ASSERT_OK(Put("aaa", DummyString(kFileSize / 2, '1'))); Flush(); Close(); ASSERT_OK(ReadOnlyReopen(options)); Status s = Put("new", "value"); ASSERT_EQ(s.ToString(), "Not implemented: Not supported operation in read only mode."); ASSERT_EQ(DummyString(kFileSize / 2, '1'), Get("aaa")); Close(); options.max_open_files = -1; ASSERT_OK(ReadOnlyReopen(options)); s = Put("new", "value"); ASSERT_EQ(s.ToString(), "Not implemented: Not supported in compacted db mode."); ASSERT_EQ(DummyString(kFileSize / 2, '1'), Get("aaa")); Close(); Reopen(options); // Add more L0 files ASSERT_OK(Put("bbb", DummyString(kFileSize / 2, '2'))); Flush(); ASSERT_OK(Put("aaa", DummyString(kFileSize / 2, 'a'))); Flush(); ASSERT_OK(Put("bbb", DummyString(kFileSize / 2, 'b'))); ASSERT_OK(Put("eee", DummyString(kFileSize / 2, 'e'))); Flush(); Close(); ASSERT_OK(ReadOnlyReopen(options)); // Fallback to read-only DB s = Put("new", "value"); ASSERT_EQ(s.ToString(), "Not implemented: Not supported operation in read only mode."); Close(); // Full compaction Reopen(options); // Add more keys ASSERT_OK(Put("fff", DummyString(kFileSize / 2, 'f'))); ASSERT_OK(Put("hhh", DummyString(kFileSize / 2, 'h'))); ASSERT_OK(Put("iii", DummyString(kFileSize / 2, 'i'))); ASSERT_OK(Put("jjj", DummyString(kFileSize / 2, 'j'))); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(3, NumTableFilesAtLevel(1)); Close(); // CompactedDB ASSERT_OK(ReadOnlyReopen(options)); s = Put("new", "value"); ASSERT_EQ(s.ToString(), "Not implemented: Not supported in compacted db mode."); ASSERT_EQ("NOT_FOUND", Get("abc")); ASSERT_EQ(DummyString(kFileSize / 2, 'a'), Get("aaa")); ASSERT_EQ(DummyString(kFileSize / 2, 'b'), Get("bbb")); ASSERT_EQ("NOT_FOUND", Get("ccc")); ASSERT_EQ(DummyString(kFileSize / 2, 'e'), Get("eee")); ASSERT_EQ(DummyString(kFileSize / 2, 'f'), Get("fff")); ASSERT_EQ("NOT_FOUND", Get("ggg")); ASSERT_EQ(DummyString(kFileSize / 2, 'h'), Get("hhh")); ASSERT_EQ(DummyString(kFileSize / 2, 'i'), Get("iii")); ASSERT_EQ(DummyString(kFileSize / 2, 'j'), Get("jjj")); ASSERT_EQ("NOT_FOUND", Get("kkk")); // MultiGet std::vector values; std::vector status_list = dbfull()->MultiGet(ReadOptions(), std::vector({Slice("aaa"), Slice("ccc"), Slice("eee"), Slice("ggg"), Slice("iii"), Slice("kkk")}), &values); ASSERT_EQ(status_list.size(), static_cast(6)); ASSERT_EQ(values.size(), static_cast(6)); ASSERT_OK(status_list[0]); ASSERT_EQ(DummyString(kFileSize / 2, 'a'), values[0]); ASSERT_TRUE(status_list[1].IsNotFound()); ASSERT_OK(status_list[2]); ASSERT_EQ(DummyString(kFileSize / 2, 'e'), values[2]); ASSERT_TRUE(status_list[3].IsNotFound()); ASSERT_OK(status_list[4]); ASSERT_EQ(DummyString(kFileSize / 2, 'i'), values[4]); ASSERT_TRUE(status_list[5].IsNotFound()); } // Make sure that when options.block_cache is set, after a new table is // created its index/filter blocks are added to block cache. TEST_F(DBTest, IndexAndFilterBlocksOfNewTableAddedToCache) { Options options = CurrentOptions(); options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); BlockBasedTableOptions table_options; table_options.cache_index_and_filter_blocks = true; table_options.filter_policy.reset(NewBloomFilterPolicy(20)); options.table_factory.reset(new BlockBasedTableFactory(table_options)); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "key", "val")); // Create a new table. ASSERT_OK(Flush(1)); // index/filter blocks added to block cache right after table creation. ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_INDEX_MISS)); ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(2, /* only index/filter were added */ TestGetTickerCount(options, BLOCK_CACHE_ADD)); ASSERT_EQ(0, TestGetTickerCount(options, BLOCK_CACHE_DATA_MISS)); uint64_t int_num; ASSERT_TRUE( dbfull()->GetIntProperty("rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_EQ(int_num, 0U); // Make sure filter block is in cache. std::string value; ReadOptions ropt; db_->KeyMayExist(ReadOptions(), handles_[1], "key", &value); // Miss count should remain the same. ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); db_->KeyMayExist(ReadOptions(), handles_[1], "key", &value); ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(2, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); // Make sure index block is in cache. auto index_block_hit = TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT); value = Get(1, "key"); ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(index_block_hit + 1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); value = Get(1, "key"); ASSERT_EQ(1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(index_block_hit + 2, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); } TEST_F(DBTest, ParanoidFileChecks) { Options options = CurrentOptions(); options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); options.level0_file_num_compaction_trigger = 2; options.paranoid_file_checks = true; BlockBasedTableOptions table_options; table_options.cache_index_and_filter_blocks = false; table_options.filter_policy.reset(NewBloomFilterPolicy(20)); options.table_factory.reset(new BlockBasedTableFactory(table_options)); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "1_key", "val")); ASSERT_OK(Put(1, "9_key", "val")); // Create a new table. ASSERT_OK(Flush(1)); ASSERT_EQ(1, /* read and cache data block */ TestGetTickerCount(options, BLOCK_CACHE_ADD)); ASSERT_OK(Put(1, "1_key2", "val2")); ASSERT_OK(Put(1, "9_key2", "val2")); // Create a new SST file. This will further trigger a compaction // and generate another file. ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(3, /* Totally 3 files created up to now */ TestGetTickerCount(options, BLOCK_CACHE_ADD)); // After disabling options.paranoid_file_checks. NO further block // is added after generating a new file. ASSERT_OK( dbfull()->SetOptions(handles_[1], {{"paranoid_file_checks", "false"}})); ASSERT_OK(Put(1, "1_key3", "val3")); ASSERT_OK(Put(1, "9_key3", "val3")); ASSERT_OK(Flush(1)); ASSERT_OK(Put(1, "1_key4", "val4")); ASSERT_OK(Put(1, "9_key4", "val4")); ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(3, /* Totally 3 files created up to now */ TestGetTickerCount(options, BLOCK_CACHE_ADD)); } TEST_F(DBTest, LevelLimitReopen) { Options options = CurrentOptions(); CreateAndReopenWithCF({"pikachu"}, options); const std::string value(1024 * 1024, ' '); int i = 0; while (NumTableFilesAtLevel(2, 1) == 0) { ASSERT_OK(Put(1, Key(i++), value)); } options.num_levels = 1; options.max_bytes_for_level_multiplier_additional.resize(1, 1); Status s = TryReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_EQ(s.IsInvalidArgument(), true); ASSERT_EQ(s.ToString(), "Invalid argument: db has more levels than options.num_levels"); options.num_levels = 10; options.max_bytes_for_level_multiplier_additional.resize(10, 1); ASSERT_OK(TryReopenWithColumnFamilies({"default", "pikachu"}, options)); } #endif // ROCKSDB_LITE TEST_F(DBTest, PutDeleteGet) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_OK(Put(1, "foo", "v2")); ASSERT_EQ("v2", Get(1, "foo")); ASSERT_OK(Delete(1, "foo")); ASSERT_EQ("NOT_FOUND", Get(1, "foo")); } while (ChangeOptions()); } TEST_F(DBTest, PutSingleDeleteGet) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_OK(Put(1, "foo2", "v2")); ASSERT_EQ("v2", Get(1, "foo2")); ASSERT_OK(SingleDelete(1, "foo")); ASSERT_EQ("NOT_FOUND", Get(1, "foo")); // Skip HashCuckooRep as it does not support single delete. FIFO and // universal compaction do not apply to the test case. Skip MergePut // because single delete does not get removed when it encounters a merge. } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction | kSkipUniversalCompaction | kSkipMergePut)); } TEST_F(DBTest, ReadFromPersistedTier) { do { Random rnd(301); Options options = CurrentOptions(); for (int disableWAL = 0; disableWAL <= 1; ++disableWAL) { CreateAndReopenWithCF({"pikachu"}, options); WriteOptions wopt; wopt.disableWAL = (disableWAL == 1); // 1st round: put but not flush ASSERT_OK(db_->Put(wopt, handles_[1], "foo", "first")); ASSERT_OK(db_->Put(wopt, handles_[1], "bar", "one")); ASSERT_EQ("first", Get(1, "foo")); ASSERT_EQ("one", Get(1, "bar")); // Read directly from persited data. ReadOptions ropt; ropt.read_tier = kPersistedTier; std::string value; if (wopt.disableWAL) { // as data has not yet being flushed, we expect not found. ASSERT_TRUE(db_->Get(ropt, handles_[1], "foo", &value).IsNotFound()); ASSERT_TRUE(db_->Get(ropt, handles_[1], "bar", &value).IsNotFound()); } else { ASSERT_OK(db_->Get(ropt, handles_[1], "foo", &value)); ASSERT_OK(db_->Get(ropt, handles_[1], "bar", &value)); } // Multiget std::vector multiget_cfs; multiget_cfs.push_back(handles_[1]); multiget_cfs.push_back(handles_[1]); std::vector multiget_keys; multiget_keys.push_back("foo"); multiget_keys.push_back("bar"); std::vector multiget_values; auto statuses = db_->MultiGet(ropt, multiget_cfs, multiget_keys, &multiget_values); if (wopt.disableWAL) { ASSERT_TRUE(statuses[0].IsNotFound()); ASSERT_TRUE(statuses[1].IsNotFound()); } else { ASSERT_OK(statuses[0]); ASSERT_OK(statuses[1]); } // 2nd round: flush and put a new value in memtable. ASSERT_OK(Flush(1)); ASSERT_OK(db_->Put(wopt, handles_[1], "rocksdb", "hello")); // once the data has been flushed, we are able to get the // data when kPersistedTier is used. ASSERT_TRUE(db_->Get(ropt, handles_[1], "foo", &value).ok()); ASSERT_EQ(value, "first"); ASSERT_TRUE(db_->Get(ropt, handles_[1], "bar", &value).ok()); ASSERT_EQ(value, "one"); if (wopt.disableWAL) { ASSERT_TRUE( db_->Get(ropt, handles_[1], "rocksdb", &value).IsNotFound()); } else { ASSERT_OK(db_->Get(ropt, handles_[1], "rocksdb", &value)); ASSERT_EQ(value, "hello"); } // Expect same result in multiget multiget_cfs.push_back(handles_[1]); multiget_keys.push_back("rocksdb"); statuses = db_->MultiGet(ropt, multiget_cfs, multiget_keys, &multiget_values); ASSERT_TRUE(statuses[0].ok()); ASSERT_EQ("first", multiget_values[0]); ASSERT_TRUE(statuses[1].ok()); ASSERT_EQ("one", multiget_values[1]); if (wopt.disableWAL) { ASSERT_TRUE(statuses[2].IsNotFound()); } else { ASSERT_OK(statuses[2]); } // 3rd round: delete and flush ASSERT_OK(db_->Delete(wopt, handles_[1], "foo")); Flush(1); ASSERT_OK(db_->Delete(wopt, handles_[1], "bar")); ASSERT_TRUE(db_->Get(ropt, handles_[1], "foo", &value).IsNotFound()); if (wopt.disableWAL) { // Still expect finding the value as its delete has not yet being // flushed. ASSERT_TRUE(db_->Get(ropt, handles_[1], "bar", &value).ok()); ASSERT_EQ(value, "one"); } else { ASSERT_TRUE(db_->Get(ropt, handles_[1], "bar", &value).IsNotFound()); } ASSERT_TRUE(db_->Get(ropt, handles_[1], "rocksdb", &value).ok()); ASSERT_EQ(value, "hello"); statuses = db_->MultiGet(ropt, multiget_cfs, multiget_keys, &multiget_values); ASSERT_TRUE(statuses[0].IsNotFound()); if (wopt.disableWAL) { ASSERT_TRUE(statuses[1].ok()); ASSERT_EQ("one", multiget_values[1]); } else { ASSERT_TRUE(statuses[1].IsNotFound()); } ASSERT_TRUE(statuses[2].ok()); ASSERT_EQ("hello", multiget_values[2]); if (wopt.disableWAL == 0) { DestroyAndReopen(options); } } } while (ChangeOptions(kSkipHashCuckoo)); } TEST_F(DBTest, PersistedTierOnIterator) { // The test needs to be changed if kPersistedTier is supported in iterator. Options options = CurrentOptions(); CreateAndReopenWithCF({"pikachu"}, options); ReadOptions ropt; ropt.read_tier = kPersistedTier; auto* iter = db_->NewIterator(ropt, handles_[1]); ASSERT_TRUE(iter->status().IsNotSupported()); delete iter; std::vector iters; ASSERT_TRUE(db_->NewIterators(ropt, {handles_[1]}, &iters).IsNotSupported()); Close(); } TEST_F(DBTest, SingleDeleteFlush) { // Test to check whether flushing preserves a single delete hidden // behind a put. do { Random rnd(301); Options options = CurrentOptions(); options.disable_auto_compactions = true; CreateAndReopenWithCF({"pikachu"}, options); // Put values on second level (so that they will not be in the same // compaction as the other operations. Put(1, "foo", "first"); Put(1, "bar", "one"); ASSERT_OK(Flush(1)); MoveFilesToLevel(2, 1); // (Single) delete hidden by a put SingleDelete(1, "foo"); Put(1, "foo", "second"); Delete(1, "bar"); Put(1, "bar", "two"); ASSERT_OK(Flush(1)); SingleDelete(1, "foo"); Delete(1, "bar"); ASSERT_OK(Flush(1)); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ("NOT_FOUND", Get(1, "bar")); ASSERT_EQ("NOT_FOUND", Get(1, "foo")); // Skip HashCuckooRep as it does not support single delete. FIFO and // universal compaction do not apply to the test case. Skip MergePut // because merges cannot be combined with single deletions. } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction | kSkipUniversalCompaction | kSkipMergePut)); } TEST_F(DBTest, SingleDeletePutFlush) { // Single deletes that encounter the matching put in a flush should get // removed. do { Random rnd(301); Options options = CurrentOptions(); options.disable_auto_compactions = true; CreateAndReopenWithCF({"pikachu"}, options); Put(1, "foo", Slice()); Put(1, "a", Slice()); SingleDelete(1, "a"); ASSERT_OK(Flush(1)); ASSERT_EQ("[ ]", AllEntriesFor("a", 1)); // Skip HashCuckooRep as it does not support single delete. FIFO and // universal compaction do not apply to the test case. Skip MergePut // because merges cannot be combined with single deletions. } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction | kSkipUniversalCompaction | kSkipMergePut)); } TEST_F(DBTest, EmptyFlush) { // It is possible to produce empty flushes when using single deletes. Tests // whether empty flushes cause issues. do { Random rnd(301); Options options = CurrentOptions(); options.disable_auto_compactions = true; CreateAndReopenWithCF({"pikachu"}, options); Put(1, "a", Slice()); SingleDelete(1, "a"); ASSERT_OK(Flush(1)); ASSERT_EQ("[ ]", AllEntriesFor("a", 1)); // Skip HashCuckooRep as it does not support single delete. FIFO and // universal compaction do not apply to the test case. Skip MergePut // because merges cannot be combined with single deletions. } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction | kSkipUniversalCompaction | kSkipMergePut)); } // Disable because not all platform can run it. // It requires more than 9GB memory to run it, With single allocation // of more than 3GB. TEST_F(DBTest, DISABLED_VeryLargeValue) { const size_t kValueSize = 3221225472u; // 3GB value const size_t kKeySize = 8388608u; // 8MB key std::string raw(kValueSize, 'v'); std::string key1(kKeySize, 'c'); std::string key2(kKeySize, 'd'); Options options; options.env = env_; options.write_buffer_size = 100000; // Small write buffer options.paranoid_checks = true; options = CurrentOptions(options); DestroyAndReopen(options); ASSERT_OK(Put("boo", "v1")); ASSERT_OK(Put("foo", "v1")); ASSERT_OK(Put(key1, raw)); raw[0] = 'w'; ASSERT_OK(Put(key2, raw)); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(1, NumTableFilesAtLevel(0)); std::string value; Status s = db_->Get(ReadOptions(), key1, &value); ASSERT_OK(s); ASSERT_EQ(kValueSize, value.size()); ASSERT_EQ('v', value[0]); s = db_->Get(ReadOptions(), key2, &value); ASSERT_OK(s); ASSERT_EQ(kValueSize, value.size()); ASSERT_EQ('w', value[0]); // Compact all files. Flush(); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); // Check DB is not in read-only state. ASSERT_OK(Put("boo", "v1")); s = db_->Get(ReadOptions(), key1, &value); ASSERT_OK(s); ASSERT_EQ(kValueSize, value.size()); ASSERT_EQ('v', value[0]); s = db_->Get(ReadOptions(), key2, &value); ASSERT_OK(s); ASSERT_EQ(kValueSize, value.size()); ASSERT_EQ('w', value[0]); } TEST_F(DBTest, GetFromImmutableLayer) { do { Options options; options.env = env_; options.write_buffer_size = 100000; // Small write buffer options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_EQ("v1", Get(1, "foo")); // Block sync calls env_->delay_sstable_sync_.store(true, std::memory_order_release); Put(1, "k1", std::string(100000, 'x')); // Fill memtable Put(1, "k2", std::string(100000, 'y')); // Trigger flush ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("NOT_FOUND", Get(0, "foo")); // Release sync calls env_->delay_sstable_sync_.store(false, std::memory_order_release); } while (ChangeOptions()); } TEST_F(DBTest, GetFromVersions) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_OK(Flush(1)); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("NOT_FOUND", Get(0, "foo")); } while (ChangeOptions()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, GetSnapshot) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions(options_override)); // Try with both a short key and a long key for (int i = 0; i < 2; i++) { std::string key = (i == 0) ? std::string("foo") : std::string(200, 'x'); ASSERT_OK(Put(1, key, "v1")); const Snapshot* s1 = db_->GetSnapshot(); if (option_config_ == kHashCuckoo) { // Unsupported case. ASSERT_TRUE(s1 == nullptr); break; } ASSERT_OK(Put(1, key, "v2")); ASSERT_EQ("v2", Get(1, key)); ASSERT_EQ("v1", Get(1, key, s1)); ASSERT_OK(Flush(1)); ASSERT_EQ("v2", Get(1, key)); ASSERT_EQ("v1", Get(1, key, s1)); db_->ReleaseSnapshot(s1); } } while (ChangeOptions()); } #endif // ROCKSDB_LITE TEST_F(DBTest, GetLevel0Ordering) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); // Check that we process level-0 files in correct order. The code // below generates two level-0 files where the earlier one comes // before the later one in the level-0 file list since the earlier // one has a smaller "smallest" key. ASSERT_OK(Put(1, "bar", "b")); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_OK(Flush(1)); ASSERT_OK(Put(1, "foo", "v2")); ASSERT_OK(Flush(1)); ASSERT_EQ("v2", Get(1, "foo")); } while (ChangeOptions()); } TEST_F(DBTest, WrongLevel0Config) { Options options = CurrentOptions(); Close(); ASSERT_OK(DestroyDB(dbname_, options)); options.level0_stop_writes_trigger = 1; options.level0_slowdown_writes_trigger = 2; options.level0_file_num_compaction_trigger = 3; ASSERT_OK(DB::Open(options, dbname_, &db_)); } #ifndef ROCKSDB_LITE TEST_F(DBTest, GetOrderedByLevels) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); Compact(1, "a", "z"); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_OK(Put(1, "foo", "v2")); ASSERT_EQ("v2", Get(1, "foo")); ASSERT_OK(Flush(1)); ASSERT_EQ("v2", Get(1, "foo")); } while (ChangeOptions()); } TEST_F(DBTest, GetPicksCorrectFile) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); // Arrange to have multiple files in a non-level-0 level. ASSERT_OK(Put(1, "a", "va")); Compact(1, "a", "b"); ASSERT_OK(Put(1, "x", "vx")); Compact(1, "x", "y"); ASSERT_OK(Put(1, "f", "vf")); Compact(1, "f", "g"); ASSERT_EQ("va", Get(1, "a")); ASSERT_EQ("vf", Get(1, "f")); ASSERT_EQ("vx", Get(1, "x")); } while (ChangeOptions()); } TEST_F(DBTest, GetEncountersEmptyLevel) { do { Options options = CurrentOptions(); options.disableDataSync = true; CreateAndReopenWithCF({"pikachu"}, options); // Arrange for the following to happen: // * sstable A in level 0 // * nothing in level 1 // * sstable B in level 2 // Then do enough Get() calls to arrange for an automatic compaction // of sstable A. A bug would cause the compaction to be marked as // occurring at level 1 (instead of the correct level 0). // Step 1: First place sstables in levels 0 and 2 Put(1, "a", "begin"); Put(1, "z", "end"); ASSERT_OK(Flush(1)); dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); Put(1, "a", "begin"); Put(1, "z", "end"); ASSERT_OK(Flush(1)); ASSERT_GT(NumTableFilesAtLevel(0, 1), 0); ASSERT_GT(NumTableFilesAtLevel(2, 1), 0); // Step 2: clear level 1 if necessary. dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 1); ASSERT_EQ(NumTableFilesAtLevel(1, 1), 0); ASSERT_EQ(NumTableFilesAtLevel(2, 1), 1); // Step 3: read a bunch of times for (int i = 0; i < 1000; i++) { ASSERT_EQ("NOT_FOUND", Get(1, "missing")); } // Step 4: Wait for compaction to finish dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 1); // XXX } while (ChangeOptions(kSkipUniversalCompaction | kSkipFIFOCompaction)); } #endif // ROCKSDB_LITE // KeyMayExist can lead to a few false positives, but not false negatives. // To make test deterministic, use a much larger number of bits per key-20 than // bits in the key, so that false positives are eliminated TEST_F(DBTest, KeyMayExist) { do { ReadOptions ropts; std::string value; anon::OptionsOverride options_override; options_override.filter_policy.reset(NewBloomFilterPolicy(20)); Options options = CurrentOptions(options_override); options.statistics = rocksdb::CreateDBStatistics(); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_TRUE(!db_->KeyMayExist(ropts, handles_[1], "a", &value)); ASSERT_OK(Put(1, "a", "b")); bool value_found = false; ASSERT_TRUE( db_->KeyMayExist(ropts, handles_[1], "a", &value, &value_found)); ASSERT_TRUE(value_found); ASSERT_EQ("b", value); ASSERT_OK(Flush(1)); value.clear(); uint64_t numopen = TestGetTickerCount(options, NO_FILE_OPENS); uint64_t cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); ASSERT_TRUE( db_->KeyMayExist(ropts, handles_[1], "a", &value, &value_found)); ASSERT_TRUE(!value_found); // assert that no new files were opened and no new blocks were // read into block cache. ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); ASSERT_OK(Delete(1, "a")); numopen = TestGetTickerCount(options, NO_FILE_OPENS); cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); ASSERT_TRUE(!db_->KeyMayExist(ropts, handles_[1], "a", &value)); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); ASSERT_OK(Flush(1)); dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1], true /* disallow trivial move */); numopen = TestGetTickerCount(options, NO_FILE_OPENS); cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); ASSERT_TRUE(!db_->KeyMayExist(ropts, handles_[1], "a", &value)); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); ASSERT_OK(Delete(1, "c")); numopen = TestGetTickerCount(options, NO_FILE_OPENS); cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); ASSERT_TRUE(!db_->KeyMayExist(ropts, handles_[1], "c", &value)); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); // KeyMayExist function only checks data in block caches, which is not used // by plain table format. } while ( ChangeOptions(kSkipPlainTable | kSkipHashIndex | kSkipFIFOCompaction)); } TEST_F(DBTest, NonBlockingIteration) { do { ReadOptions non_blocking_opts, regular_opts; Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); non_blocking_opts.read_tier = kBlockCacheTier; CreateAndReopenWithCF({"pikachu"}, options); // write one kv to the database. ASSERT_OK(Put(1, "a", "b")); // scan using non-blocking iterator. We should find it because // it is in memtable. Iterator* iter = db_->NewIterator(non_blocking_opts, handles_[1]); int count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(iter->status()); count++; } ASSERT_EQ(count, 1); delete iter; // flush memtable to storage. Now, the key should not be in the // memtable neither in the block cache. ASSERT_OK(Flush(1)); // verify that a non-blocking iterator does not find any // kvs. Neither does it do any IOs to storage. uint64_t numopen = TestGetTickerCount(options, NO_FILE_OPENS); uint64_t cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); iter = db_->NewIterator(non_blocking_opts, handles_[1]); count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { count++; } ASSERT_EQ(count, 0); ASSERT_TRUE(iter->status().IsIncomplete()); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); delete iter; // read in the specified block via a regular get ASSERT_EQ(Get(1, "a"), "b"); // verify that we can find it via a non-blocking scan numopen = TestGetTickerCount(options, NO_FILE_OPENS); cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); iter = db_->NewIterator(non_blocking_opts, handles_[1]); count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(iter->status()); count++; } ASSERT_EQ(count, 1); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); delete iter; // This test verifies block cache behaviors, which is not used by plain // table format. // Exclude kHashCuckoo as it does not support iteration currently } while (ChangeOptions(kSkipPlainTable | kSkipNoSeekToLast | kSkipHashCuckoo | kSkipMmapReads)); } #ifndef ROCKSDB_LITE TEST_F(DBTest, ManagedNonBlockingIteration) { do { ReadOptions non_blocking_opts, regular_opts; Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); non_blocking_opts.read_tier = kBlockCacheTier; non_blocking_opts.managed = true; CreateAndReopenWithCF({"pikachu"}, options); // write one kv to the database. ASSERT_OK(Put(1, "a", "b")); // scan using non-blocking iterator. We should find it because // it is in memtable. Iterator* iter = db_->NewIterator(non_blocking_opts, handles_[1]); int count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(iter->status()); count++; } ASSERT_EQ(count, 1); delete iter; // flush memtable to storage. Now, the key should not be in the // memtable neither in the block cache. ASSERT_OK(Flush(1)); // verify that a non-blocking iterator does not find any // kvs. Neither does it do any IOs to storage. int64_t numopen = TestGetTickerCount(options, NO_FILE_OPENS); int64_t cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); iter = db_->NewIterator(non_blocking_opts, handles_[1]); count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { count++; } ASSERT_EQ(count, 0); ASSERT_TRUE(iter->status().IsIncomplete()); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); delete iter; // read in the specified block via a regular get ASSERT_EQ(Get(1, "a"), "b"); // verify that we can find it via a non-blocking scan numopen = TestGetTickerCount(options, NO_FILE_OPENS); cache_added = TestGetTickerCount(options, BLOCK_CACHE_ADD); iter = db_->NewIterator(non_blocking_opts, handles_[1]); count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(iter->status()); count++; } ASSERT_EQ(count, 1); ASSERT_EQ(numopen, TestGetTickerCount(options, NO_FILE_OPENS)); ASSERT_EQ(cache_added, TestGetTickerCount(options, BLOCK_CACHE_ADD)); delete iter; // This test verifies block cache behaviors, which is not used by plain // table format. // Exclude kHashCuckoo as it does not support iteration currently } while (ChangeOptions(kSkipPlainTable | kSkipNoSeekToLast | kSkipHashCuckoo | kSkipMmapReads)); } #endif // ROCKSDB_LITE // A delete is skipped for key if KeyMayExist(key) returns False // Tests Writebatch consistency and proper delete behaviour TEST_F(DBTest, FilterDeletes) { do { anon::OptionsOverride options_override; options_override.filter_policy.reset(NewBloomFilterPolicy(20)); Options options = CurrentOptions(options_override); options.filter_deletes = true; CreateAndReopenWithCF({"pikachu"}, options); WriteBatch batch; batch.Delete(handles_[1], "a"); dbfull()->Write(WriteOptions(), &batch); ASSERT_EQ(AllEntriesFor("a", 1), "[ ]"); // Delete skipped batch.Clear(); batch.Put(handles_[1], "a", "b"); batch.Delete(handles_[1], "a"); dbfull()->Write(WriteOptions(), &batch); ASSERT_EQ(Get(1, "a"), "NOT_FOUND"); ASSERT_EQ(AllEntriesFor("a", 1), "[ DEL, b ]"); // Delete issued batch.Clear(); batch.Delete(handles_[1], "c"); batch.Put(handles_[1], "c", "d"); dbfull()->Write(WriteOptions(), &batch); ASSERT_EQ(Get(1, "c"), "d"); ASSERT_EQ(AllEntriesFor("c", 1), "[ d ]"); // Delete skipped batch.Clear(); ASSERT_OK(Flush(1)); // A stray Flush batch.Delete(handles_[1], "c"); dbfull()->Write(WriteOptions(), &batch); ASSERT_EQ(AllEntriesFor("c", 1), "[ DEL, d ]"); // Delete issued batch.Clear(); } while (ChangeCompactOptions()); } TEST_F(DBTest, GetFilterByPrefixBloom) { Options options = last_options_; options.prefix_extractor.reset(NewFixedPrefixTransform(8)); options.statistics = rocksdb::CreateDBStatistics(); BlockBasedTableOptions bbto; bbto.filter_policy.reset(NewBloomFilterPolicy(10, false)); bbto.whole_key_filtering = false; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); DestroyAndReopen(options); WriteOptions wo; ReadOptions ro; FlushOptions fo; fo.wait = true; std::string value; ASSERT_OK(dbfull()->Put(wo, "barbarbar", "foo")); ASSERT_OK(dbfull()->Put(wo, "barbarbar2", "foo2")); ASSERT_OK(dbfull()->Put(wo, "foofoofoo", "bar")); dbfull()->Flush(fo); ASSERT_EQ("foo", Get("barbarbar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); ASSERT_EQ("foo2", Get("barbarbar2")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); ASSERT_EQ("NOT_FOUND", Get("barbarbar3")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); ASSERT_EQ("NOT_FOUND", Get("barfoofoo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("NOT_FOUND", Get("foobarbar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2); } TEST_F(DBTest, WholeKeyFilterProp) { Options options = last_options_; options.prefix_extractor.reset(NewFixedPrefixTransform(3)); options.statistics = rocksdb::CreateDBStatistics(); BlockBasedTableOptions bbto; bbto.filter_policy.reset(NewBloomFilterPolicy(10, false)); bbto.whole_key_filtering = false; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); DestroyAndReopen(options); WriteOptions wo; ReadOptions ro; FlushOptions fo; fo.wait = true; std::string value; ASSERT_OK(dbfull()->Put(wo, "foobar", "foo")); // Needs insert some keys to make sure files are not filtered out by key // ranges. ASSERT_OK(dbfull()->Put(wo, "aaa", "")); ASSERT_OK(dbfull()->Put(wo, "zzz", "")); dbfull()->Flush(fo); Reopen(options); ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); // Reopen with whole key filtering enabled and prefix extractor // NULL. Bloom filter should be off for both of whole key and // prefix bloom. bbto.whole_key_filtering = true; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); options.prefix_extractor.reset(); Reopen(options); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); // Write DB with only full key filtering. ASSERT_OK(dbfull()->Put(wo, "foobar", "foo")); // Needs insert some keys to make sure files are not filtered out by key // ranges. ASSERT_OK(dbfull()->Put(wo, "aaa", "")); ASSERT_OK(dbfull()->Put(wo, "zzz", "")); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); // Reopen with both of whole key off and prefix extractor enabled. // Still no bloom filter should be used. options.prefix_extractor.reset(NewFixedPrefixTransform(3)); bbto.whole_key_filtering = false; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); Reopen(options); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); // Try to create a DB with mixed files: ASSERT_OK(dbfull()->Put(wo, "foobar", "foo")); // Needs insert some keys to make sure files are not filtered out by key // ranges. ASSERT_OK(dbfull()->Put(wo, "aaa", "")); ASSERT_OK(dbfull()->Put(wo, "zzz", "")); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); options.prefix_extractor.reset(); bbto.whole_key_filtering = true; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); Reopen(options); // Try to create a DB with mixed files. ASSERT_OK(dbfull()->Put(wo, "barfoo", "bar")); // In this case needs insert some keys to make sure files are // not filtered out by key ranges. ASSERT_OK(dbfull()->Put(wo, "aaa", "")); ASSERT_OK(dbfull()->Put(wo, "zzz", "")); Flush(); // Now we have two files: // File 1: An older file with prefix bloom. // File 2: A newer file with whole bloom filter. ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1); ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 3); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4); ASSERT_EQ("bar", Get("barfoo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4); // Reopen with the same setting: only whole key is used Reopen(options); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4); ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 5); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 6); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7); ASSERT_EQ("bar", Get("barfoo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7); // Restart with both filters are allowed options.prefix_extractor.reset(NewFixedPrefixTransform(3)); bbto.whole_key_filtering = true; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); Reopen(options); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7); // File 1 will has it filtered out. // File 2 will not, as prefix `foo` exists in the file. ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 8); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 10); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11); ASSERT_EQ("bar", Get("barfoo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11); // Restart with only prefix bloom is allowed. options.prefix_extractor.reset(NewFixedPrefixTransform(3)); bbto.whole_key_filtering = false; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); Reopen(options); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11); ASSERT_EQ("NOT_FOUND", Get("foo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11); ASSERT_EQ("NOT_FOUND", Get("bar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12); ASSERT_EQ("foo", Get("foobar")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12); ASSERT_EQ("bar", Get("barfoo")); ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12); } TEST_F(DBTest, IterSeekBeforePrev) { ASSERT_OK(Put("a", "b")); ASSERT_OK(Put("c", "d")); dbfull()->Flush(FlushOptions()); ASSERT_OK(Put("0", "f")); ASSERT_OK(Put("1", "h")); dbfull()->Flush(FlushOptions()); ASSERT_OK(Put("2", "j")); auto iter = db_->NewIterator(ReadOptions()); iter->Seek(Slice("c")); iter->Prev(); iter->Seek(Slice("a")); iter->Prev(); delete iter; } namespace { std::string MakeLongKey(size_t length, char c) { return std::string(length, c); } } // namespace TEST_F(DBTest, IterLongKeys) { ASSERT_OK(Put(MakeLongKey(20, 0), "0")); ASSERT_OK(Put(MakeLongKey(32, 2), "2")); ASSERT_OK(Put("a", "b")); dbfull()->Flush(FlushOptions()); ASSERT_OK(Put(MakeLongKey(50, 1), "1")); ASSERT_OK(Put(MakeLongKey(127, 3), "3")); ASSERT_OK(Put(MakeLongKey(64, 4), "4")); auto iter = db_->NewIterator(ReadOptions()); // Create a key that needs to be skipped for Seq too new iter->Seek(MakeLongKey(20, 0)); ASSERT_EQ(IterStatus(iter), MakeLongKey(20, 0) + "->0"); iter->Next(); ASSERT_EQ(IterStatus(iter), MakeLongKey(50, 1) + "->1"); iter->Next(); ASSERT_EQ(IterStatus(iter), MakeLongKey(32, 2) + "->2"); iter->Next(); ASSERT_EQ(IterStatus(iter), MakeLongKey(127, 3) + "->3"); iter->Next(); ASSERT_EQ(IterStatus(iter), MakeLongKey(64, 4) + "->4"); delete iter; iter = db_->NewIterator(ReadOptions()); iter->Seek(MakeLongKey(50, 1)); ASSERT_EQ(IterStatus(iter), MakeLongKey(50, 1) + "->1"); iter->Next(); ASSERT_EQ(IterStatus(iter), MakeLongKey(32, 2) + "->2"); iter->Next(); ASSERT_EQ(IterStatus(iter), MakeLongKey(127, 3) + "->3"); delete iter; } TEST_F(DBTest, IterNextWithNewerSeq) { ASSERT_OK(Put("0", "0")); dbfull()->Flush(FlushOptions()); ASSERT_OK(Put("a", "b")); ASSERT_OK(Put("c", "d")); ASSERT_OK(Put("d", "e")); auto iter = db_->NewIterator(ReadOptions()); // Create a key that needs to be skipped for Seq too new for (uint64_t i = 0; i < last_options_.max_sequential_skip_in_iterations + 1; i++) { ASSERT_OK(Put("b", "f")); } iter->Seek(Slice("a")); ASSERT_EQ(IterStatus(iter), "a->b"); iter->Next(); ASSERT_EQ(IterStatus(iter), "c->d"); delete iter; } TEST_F(DBTest, IterPrevWithNewerSeq) { ASSERT_OK(Put("0", "0")); dbfull()->Flush(FlushOptions()); ASSERT_OK(Put("a", "b")); ASSERT_OK(Put("c", "d")); ASSERT_OK(Put("d", "e")); auto iter = db_->NewIterator(ReadOptions()); // Create a key that needs to be skipped for Seq too new for (uint64_t i = 0; i < last_options_.max_sequential_skip_in_iterations + 1; i++) { ASSERT_OK(Put("b", "f")); } iter->Seek(Slice("d")); ASSERT_EQ(IterStatus(iter), "d->e"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "c->d"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "a->b"); iter->Prev(); delete iter; } TEST_F(DBTest, IterPrevWithNewerSeq2) { ASSERT_OK(Put("0", "0")); dbfull()->Flush(FlushOptions()); ASSERT_OK(Put("a", "b")); ASSERT_OK(Put("c", "d")); ASSERT_OK(Put("d", "e")); auto iter = db_->NewIterator(ReadOptions()); iter->Seek(Slice("c")); ASSERT_EQ(IterStatus(iter), "c->d"); // Create a key that needs to be skipped for Seq too new for (uint64_t i = 0; i < last_options_.max_sequential_skip_in_iterations + 1; i++) { ASSERT_OK(Put("b", "f")); } iter->Prev(); ASSERT_EQ(IterStatus(iter), "a->b"); iter->Prev(); delete iter; } TEST_F(DBTest, IterEmpty) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->Seek("foo"); ASSERT_EQ(IterStatus(iter), "(invalid)"); delete iter; } while (ChangeCompactOptions()); } TEST_F(DBTest, IterSingle) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "a", "va")); Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->Seek(""); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->Seek("a"); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->Seek("b"); ASSERT_EQ(IterStatus(iter), "(invalid)"); delete iter; } while (ChangeCompactOptions()); } TEST_F(DBTest, IterMulti) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "a", "va")); ASSERT_OK(Put(1, "b", "vb")); ASSERT_OK(Put(1, "c", "vc")); Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "b->vb"); iter->Next(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "b->vb"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->Seek(""); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Seek("a"); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Seek("ax"); ASSERT_EQ(IterStatus(iter), "b->vb"); iter->Seek("b"); ASSERT_EQ(IterStatus(iter), "b->vb"); iter->Seek("z"); ASSERT_EQ(IterStatus(iter), "(invalid)"); // Switch from reverse to forward iter->SeekToLast(); iter->Prev(); iter->Prev(); iter->Next(); ASSERT_EQ(IterStatus(iter), "b->vb"); // Switch from forward to reverse iter->SeekToFirst(); iter->Next(); iter->Next(); iter->Prev(); ASSERT_EQ(IterStatus(iter), "b->vb"); // Make sure iter stays at snapshot ASSERT_OK(Put(1, "a", "va2")); ASSERT_OK(Put(1, "a2", "va3")); ASSERT_OK(Put(1, "b", "vb2")); ASSERT_OK(Put(1, "c", "vc2")); ASSERT_OK(Delete(1, "b")); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "b->vb"); iter->Next(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "b->vb"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "(invalid)"); delete iter; } while (ChangeCompactOptions()); } // Check that we can skip over a run of user keys // by using reseek rather than sequential scan TEST_F(DBTest, IterReseek) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; Options options = CurrentOptions(options_override); options.max_sequential_skip_in_iterations = 3; options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // insert three keys with same userkey and verify that // reseek is not invoked. For each of these test cases, // verify that we can find the next key "b". ASSERT_OK(Put(1, "a", "zero")); ASSERT_OK(Put(1, "a", "one")); ASSERT_OK(Put(1, "a", "two")); ASSERT_OK(Put(1, "b", "bone")); Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 0); ASSERT_EQ(IterStatus(iter), "a->two"); iter->Next(); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 0); ASSERT_EQ(IterStatus(iter), "b->bone"); delete iter; // insert a total of three keys with same userkey and verify // that reseek is still not invoked. ASSERT_OK(Put(1, "a", "three")); iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->three"); iter->Next(); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 0); ASSERT_EQ(IterStatus(iter), "b->bone"); delete iter; // insert a total of four keys with same userkey and verify // that reseek is invoked. ASSERT_OK(Put(1, "a", "four")); iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->four"); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 0); iter->Next(); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), 1); ASSERT_EQ(IterStatus(iter), "b->bone"); delete iter; // Testing reverse iterator // At this point, we have three versions of "a" and one version of "b". // The reseek statistics is already at 1. int num_reseeks = (int)TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION); // Insert another version of b and assert that reseek is not invoked ASSERT_OK(Put(1, "b", "btwo")); iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "b->btwo"); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), num_reseeks); iter->Prev(); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), num_reseeks + 1); ASSERT_EQ(IterStatus(iter), "a->four"); delete iter; // insert two more versions of b. This makes a total of 4 versions // of b and 4 versions of a. ASSERT_OK(Put(1, "b", "bthree")); ASSERT_OK(Put(1, "b", "bfour")); iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "b->bfour"); ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), num_reseeks + 2); iter->Prev(); // the previous Prev call should have invoked reseek ASSERT_EQ(TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION), num_reseeks + 3); ASSERT_EQ(IterStatus(iter), "a->four"); delete iter; } TEST_F(DBTest, IterSmallAndLargeMix) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "a", "va")); ASSERT_OK(Put(1, "b", std::string(100000, 'b'))); ASSERT_OK(Put(1, "c", "vc")); ASSERT_OK(Put(1, "d", std::string(100000, 'd'))); ASSERT_OK(Put(1, "e", std::string(100000, 'e'))); Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Next(); ASSERT_EQ(IterStatus(iter), "b->" + std::string(100000, 'b')); iter->Next(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Next(); ASSERT_EQ(IterStatus(iter), "d->" + std::string(100000, 'd')); iter->Next(); ASSERT_EQ(IterStatus(iter), "e->" + std::string(100000, 'e')); iter->Next(); ASSERT_EQ(IterStatus(iter), "(invalid)"); iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), "e->" + std::string(100000, 'e')); iter->Prev(); ASSERT_EQ(IterStatus(iter), "d->" + std::string(100000, 'd')); iter->Prev(); ASSERT_EQ(IterStatus(iter), "c->vc"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "b->" + std::string(100000, 'b')); iter->Prev(); ASSERT_EQ(IterStatus(iter), "a->va"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "(invalid)"); delete iter; } while (ChangeCompactOptions()); } TEST_F(DBTest, IterMultiWithDelete) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "ka", "va")); ASSERT_OK(Put(1, "kb", "vb")); ASSERT_OK(Put(1, "kc", "vc")); ASSERT_OK(Delete(1, "kb")); ASSERT_EQ("NOT_FOUND", Get(1, "kb")); Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); iter->Seek("kc"); ASSERT_EQ(IterStatus(iter), "kc->vc"); if (!CurrentOptions().merge_operator) { // TODO: merge operator does not support backward iteration yet if (kPlainTableAllBytesPrefix != option_config_&& kBlockBasedTableWithWholeKeyHashIndex != option_config_ && kHashLinkList != option_config_) { iter->Prev(); ASSERT_EQ(IterStatus(iter), "ka->va"); } } delete iter; } while (ChangeOptions()); } TEST_F(DBTest, IterPrevMaxSkip) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); for (int i = 0; i < 2; i++) { ASSERT_OK(Put(1, "key1", "v1")); ASSERT_OK(Put(1, "key2", "v2")); ASSERT_OK(Put(1, "key3", "v3")); ASSERT_OK(Put(1, "key4", "v4")); ASSERT_OK(Put(1, "key5", "v5")); } VerifyIterLast("key5->v5", 1); ASSERT_OK(Delete(1, "key5")); VerifyIterLast("key4->v4", 1); ASSERT_OK(Delete(1, "key4")); VerifyIterLast("key3->v3", 1); ASSERT_OK(Delete(1, "key3")); VerifyIterLast("key2->v2", 1); ASSERT_OK(Delete(1, "key2")); VerifyIterLast("key1->v1", 1); ASSERT_OK(Delete(1, "key1")); VerifyIterLast("(invalid)", 1); } while (ChangeOptions(kSkipMergePut | kSkipNoSeekToLast)); } TEST_F(DBTest, IterWithSnapshot) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions(options_override)); ASSERT_OK(Put(1, "key1", "val1")); ASSERT_OK(Put(1, "key2", "val2")); ASSERT_OK(Put(1, "key3", "val3")); ASSERT_OK(Put(1, "key4", "val4")); ASSERT_OK(Put(1, "key5", "val5")); const Snapshot *snapshot = db_->GetSnapshot(); ReadOptions options; options.snapshot = snapshot; Iterator* iter = db_->NewIterator(options, handles_[1]); // Put more values after the snapshot ASSERT_OK(Put(1, "key100", "val100")); ASSERT_OK(Put(1, "key101", "val101")); iter->Seek("key5"); ASSERT_EQ(IterStatus(iter), "key5->val5"); if (!CurrentOptions().merge_operator) { // TODO: merge operator does not support backward iteration yet if (kPlainTableAllBytesPrefix != option_config_&& kBlockBasedTableWithWholeKeyHashIndex != option_config_ && kHashLinkList != option_config_) { iter->Prev(); ASSERT_EQ(IterStatus(iter), "key4->val4"); iter->Prev(); ASSERT_EQ(IterStatus(iter), "key3->val3"); iter->Next(); ASSERT_EQ(IterStatus(iter), "key4->val4"); iter->Next(); ASSERT_EQ(IterStatus(iter), "key5->val5"); } iter->Next(); ASSERT_TRUE(!iter->Valid()); } db_->ReleaseSnapshot(snapshot); delete iter; // skip as HashCuckooRep does not support snapshot } while (ChangeOptions(kSkipHashCuckoo)); } TEST_F(DBTest, Recover) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_OK(Put(1, "baz", "v5")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("v5", Get(1, "baz")); ASSERT_OK(Put(1, "bar", "v2")); ASSERT_OK(Put(1, "foo", "v3")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("v3", Get(1, "foo")); ASSERT_OK(Put(1, "foo", "v4")); ASSERT_EQ("v4", Get(1, "foo")); ASSERT_EQ("v2", Get(1, "bar")); ASSERT_EQ("v5", Get(1, "baz")); } while (ChangeOptions()); } TEST_F(DBTest, RecoverWithTableHandle) { do { Options options; options.create_if_missing = true; options.write_buffer_size = 100; options.disable_auto_compactions = true; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_OK(Put(1, "bar", "v2")); ASSERT_OK(Flush(1)); ASSERT_OK(Put(1, "foo", "v3")); ASSERT_OK(Put(1, "bar", "v4")); ASSERT_OK(Flush(1)); ASSERT_OK(Put(1, "big", std::string(100, 'a'))); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); std::vector> files; dbfull()->TEST_GetFilesMetaData(handles_[1], &files); size_t total_files = 0; for (const auto& level : files) { total_files += level.size(); } ASSERT_EQ(total_files, 3); for (const auto& level : files) { for (const auto& file : level) { if (kInfiniteMaxOpenFiles == option_config_) { ASSERT_TRUE(file.table_reader_handle != nullptr); } else { ASSERT_TRUE(file.table_reader_handle == nullptr); } } } } while (ChangeOptions()); } TEST_F(DBTest, IgnoreRecoveredLog) { std::string backup_logs = dbname_ + "/backup_logs"; // delete old files in backup_logs directory env_->CreateDirIfMissing(backup_logs); std::vector old_files; env_->GetChildren(backup_logs, &old_files); for (auto& file : old_files) { if (file != "." && file != "..") { env_->DeleteFile(backup_logs + "/" + file); } } do { Options options = CurrentOptions(); options.create_if_missing = true; options.merge_operator = MergeOperators::CreateUInt64AddOperator(); options.wal_dir = dbname_ + "/logs"; DestroyAndReopen(options); // fill up the DB std::string one, two; PutFixed64(&one, 1); PutFixed64(&two, 2); ASSERT_OK(db_->Merge(WriteOptions(), Slice("foo"), Slice(one))); ASSERT_OK(db_->Merge(WriteOptions(), Slice("foo"), Slice(one))); ASSERT_OK(db_->Merge(WriteOptions(), Slice("bar"), Slice(one))); // copy the logs to backup std::vector logs; env_->GetChildren(options.wal_dir, &logs); for (auto& log : logs) { if (log != ".." && log != ".") { CopyFile(options.wal_dir + "/" + log, backup_logs + "/" + log); } } // recover the DB Reopen(options); ASSERT_EQ(two, Get("foo")); ASSERT_EQ(one, Get("bar")); Close(); // copy the logs from backup back to wal dir for (auto& log : logs) { if (log != ".." && log != ".") { CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log); } } // this should ignore the log files, recovery should not happen again // if the recovery happens, the same merge operator would be called twice, // leading to incorrect results Reopen(options); ASSERT_EQ(two, Get("foo")); ASSERT_EQ(one, Get("bar")); Close(); Destroy(options); Reopen(options); Close(); // copy the logs from backup back to wal dir env_->CreateDirIfMissing(options.wal_dir); for (auto& log : logs) { if (log != ".." && log != ".") { CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log); } } // assert that we successfully recovered only from logs, even though we // destroyed the DB Reopen(options); ASSERT_EQ(two, Get("foo")); ASSERT_EQ(one, Get("bar")); // Recovery will fail if DB directory doesn't exist. Destroy(options); // copy the logs from backup back to wal dir env_->CreateDirIfMissing(options.wal_dir); for (auto& log : logs) { if (log != ".." && log != ".") { CopyFile(backup_logs + "/" + log, options.wal_dir + "/" + log); // we won't be needing this file no more env_->DeleteFile(backup_logs + "/" + log); } } Status s = TryReopen(options); ASSERT_TRUE(!s.ok()); } while (ChangeOptions(kSkipHashCuckoo)); } TEST_F(DBTest, CheckLock) { do { DB* localdb; Options options = CurrentOptions(); ASSERT_OK(TryReopen(options)); // second open should fail ASSERT_TRUE(!(DB::Open(options, dbname_, &localdb)).ok()); } while (ChangeCompactOptions()); } TEST_F(DBTest, FlushMultipleMemtable) { do { Options options = CurrentOptions(); WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; options.max_write_buffer_number = 4; options.min_write_buffer_number_to_merge = 3; options.max_write_buffer_number_to_maintain = -1; CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v1")); ASSERT_OK(Flush(1)); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v1")); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("v1", Get(1, "bar")); ASSERT_OK(Flush(1)); } while (ChangeCompactOptions()); } TEST_F(DBTest, FlushEmptyColumnFamily) { // Block flush thread and disable compaction thread env_->SetBackgroundThreads(1, Env::HIGH); env_->SetBackgroundThreads(1, Env::LOW); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); test::SleepingBackgroundTask sleeping_task_high; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_high, Env::Priority::HIGH); Options options = CurrentOptions(); // disable compaction options.disable_auto_compactions = true; WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; options.max_write_buffer_number = 2; options.min_write_buffer_number_to_merge = 1; options.max_write_buffer_number_to_maintain = 1; CreateAndReopenWithCF({"pikachu"}, options); // Compaction can still go through even if no thread can flush the // mem table. ASSERT_OK(Flush(0)); ASSERT_OK(Flush(1)); // Insert can go through ASSERT_OK(dbfull()->Put(writeOpt, handles_[0], "foo", "v1")); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v1")); ASSERT_EQ("v1", Get(0, "foo")); ASSERT_EQ("v1", Get(1, "bar")); sleeping_task_high.WakeUp(); sleeping_task_high.WaitUntilDone(); // Flush can still go through. ASSERT_OK(Flush(0)); ASSERT_OK(Flush(1)); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); } TEST_F(DBTest, FLUSH) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; SetPerfLevel(kEnableTime);; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v1")); // this will now also flush the last 2 writes ASSERT_OK(Flush(1)); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v1")); perf_context.Reset(); Get(1, "foo"); ASSERT_TRUE((int) perf_context.get_from_output_files_time > 0); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("v1", Get(1, "bar")); writeOpt.disableWAL = true; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v2")); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v2")); ASSERT_OK(Flush(1)); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("v2", Get(1, "bar")); perf_context.Reset(); ASSERT_EQ("v2", Get(1, "foo")); ASSERT_TRUE((int) perf_context.get_from_output_files_time > 0); writeOpt.disableWAL = false; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v3")); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v3")); ASSERT_OK(Flush(1)); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); // 'foo' should be there because its put // has WAL enabled. ASSERT_EQ("v3", Get(1, "foo")); ASSERT_EQ("v3", Get(1, "bar")); SetPerfLevel(kDisable); } while (ChangeCompactOptions()); } TEST_F(DBTest, RecoveryWithEmptyLog) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_OK(Put(1, "foo", "v2")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v3")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("v3", Get(1, "foo")); } while (ChangeOptions()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, FlushSchedule) { Options options = CurrentOptions(); options.disable_auto_compactions = true; options.level0_stop_writes_trigger = 1 << 10; options.level0_slowdown_writes_trigger = 1 << 10; options.min_write_buffer_number_to_merge = 1; options.max_write_buffer_number_to_maintain = 1; options.max_write_buffer_number = 2; options.write_buffer_size = 120 * 1024; CreateAndReopenWithCF({"pikachu"}, options); std::vector threads; std::atomic thread_num(0); // each column family will have 5 thread, each thread generating 2 memtables. // each column family should end up with 10 table files std::function fill_memtable_func = [&]() { int a = thread_num.fetch_add(1); Random rnd(a); WriteOptions wo; // this should fill up 2 memtables for (int k = 0; k < 5000; ++k) { ASSERT_OK(db_->Put(wo, handles_[a & 1], RandomString(&rnd, 13), "")); } }; for (int i = 0; i < 10; ++i) { threads.emplace_back(fill_memtable_func); } for (auto& t : threads) { t.join(); } auto default_tables = GetNumberOfSstFilesForColumnFamily(db_, "default"); auto pikachu_tables = GetNumberOfSstFilesForColumnFamily(db_, "pikachu"); ASSERT_LE(default_tables, static_cast(10)); ASSERT_GT(default_tables, static_cast(0)); ASSERT_LE(pikachu_tables, static_cast(10)); ASSERT_GT(pikachu_tables, static_cast(0)); } #endif // ROCKSDB_LITE TEST_F(DBTest, ManifestRollOver) { do { Options options; options.max_manifest_file_size = 10 ; // 10 bytes options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); { ASSERT_OK(Put(1, "manifest_key1", std::string(1000, '1'))); ASSERT_OK(Put(1, "manifest_key2", std::string(1000, '2'))); ASSERT_OK(Put(1, "manifest_key3", std::string(1000, '3'))); uint64_t manifest_before_flush = dbfull()->TEST_Current_Manifest_FileNo(); ASSERT_OK(Flush(1)); // This should trigger LogAndApply. uint64_t manifest_after_flush = dbfull()->TEST_Current_Manifest_FileNo(); ASSERT_GT(manifest_after_flush, manifest_before_flush); ReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_GT(dbfull()->TEST_Current_Manifest_FileNo(), manifest_after_flush); // check if a new manifest file got inserted or not. ASSERT_EQ(std::string(1000, '1'), Get(1, "manifest_key1")); ASSERT_EQ(std::string(1000, '2'), Get(1, "manifest_key2")); ASSERT_EQ(std::string(1000, '3'), Get(1, "manifest_key3")); } } while (ChangeCompactOptions()); } TEST_F(DBTest, IdentityAcrossRestarts) { do { std::string id1; ASSERT_OK(db_->GetDbIdentity(id1)); Options options = CurrentOptions(); Reopen(options); std::string id2; ASSERT_OK(db_->GetDbIdentity(id2)); // id1 should match id2 because identity was not regenerated ASSERT_EQ(id1.compare(id2), 0); std::string idfilename = IdentityFileName(dbname_); ASSERT_OK(env_->DeleteFile(idfilename)); Reopen(options); std::string id3; ASSERT_OK(db_->GetDbIdentity(id3)); // id1 should NOT match id3 because identity was regenerated ASSERT_NE(id1.compare(id3), 0); } while (ChangeCompactOptions()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, RecoverWithLargeLog) { do { { Options options = CurrentOptions(); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "big1", std::string(200000, '1'))); ASSERT_OK(Put(1, "big2", std::string(200000, '2'))); ASSERT_OK(Put(1, "small3", std::string(10, '3'))); ASSERT_OK(Put(1, "small4", std::string(10, '4'))); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); } // Make sure that if we re-open with a small write buffer size that // we flush table files in the middle of a large log file. Options options; options.write_buffer_size = 100000; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 3); ASSERT_EQ(std::string(200000, '1'), Get(1, "big1")); ASSERT_EQ(std::string(200000, '2'), Get(1, "big2")); ASSERT_EQ(std::string(10, '3'), Get(1, "small3")); ASSERT_EQ(std::string(10, '4'), Get(1, "small4")); ASSERT_GT(NumTableFilesAtLevel(0, 1), 1); } while (ChangeCompactOptions()); } #endif // ROCKSDB_LITE namespace { class KeepFilter : public CompactionFilter { public: virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { return false; } virtual const char* Name() const override { return "KeepFilter"; } }; class KeepFilterFactory : public CompactionFilterFactory { public: explicit KeepFilterFactory(bool check_context = false) : check_context_(check_context) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { if (check_context_) { EXPECT_EQ(expect_full_compaction_.load(), context.is_full_compaction); EXPECT_EQ(expect_manual_compaction_.load(), context.is_manual_compaction); } return std::unique_ptr(new KeepFilter()); } virtual const char* Name() const override { return "KeepFilterFactory"; } bool check_context_; std::atomic_bool expect_full_compaction_; std::atomic_bool expect_manual_compaction_; }; class DelayFilter : public CompactionFilter { public: explicit DelayFilter(DBTestBase* d) : db_test(d) {} virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { db_test->env_->addon_time_.fetch_add(1000); return true; } virtual const char* Name() const override { return "DelayFilter"; } private: DBTestBase* db_test; }; class DelayFilterFactory : public CompactionFilterFactory { public: explicit DelayFilterFactory(DBTestBase* d) : db_test(d) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { return std::unique_ptr(new DelayFilter(db_test)); } virtual const char* Name() const override { return "DelayFilterFactory"; } private: DBTestBase* db_test; }; } // namespace #ifndef ROCKSDB_LITE TEST_F(DBTest, CompressedCache) { if (!Snappy_Supported()) { return; } int num_iter = 80; // Run this test three iterations. // Iteration 1: only a uncompressed block cache // Iteration 2: only a compressed block cache // Iteration 3: both block cache and compressed cache // Iteration 4: both block cache and compressed cache, but DB is not // compressed for (int iter = 0; iter < 4; iter++) { Options options; options.write_buffer_size = 64*1024; // small write buffer options.statistics = rocksdb::CreateDBStatistics(); options = CurrentOptions(options); BlockBasedTableOptions table_options; switch (iter) { case 0: // only uncompressed block cache table_options.block_cache = NewLRUCache(8*1024); table_options.block_cache_compressed = nullptr; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); break; case 1: // no block cache, only compressed cache table_options.no_block_cache = true; table_options.block_cache = nullptr; table_options.block_cache_compressed = NewLRUCache(8*1024); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); break; case 2: // both compressed and uncompressed block cache table_options.block_cache = NewLRUCache(1024); table_options.block_cache_compressed = NewLRUCache(8*1024); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); break; case 3: // both block cache and compressed cache, but DB is not compressed // also, make block cache sizes bigger, to trigger block cache hits table_options.block_cache = NewLRUCache(1024 * 1024); table_options.block_cache_compressed = NewLRUCache(8 * 1024 * 1024); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.compression = kNoCompression; break; default: ASSERT_TRUE(false); } CreateAndReopenWithCF({"pikachu"}, options); // default column family doesn't have block cache Options no_block_cache_opts; no_block_cache_opts.statistics = options.statistics; no_block_cache_opts = CurrentOptions(no_block_cache_opts); BlockBasedTableOptions table_options_no_bc; table_options_no_bc.no_block_cache = true; no_block_cache_opts.table_factory.reset( NewBlockBasedTableFactory(table_options_no_bc)); ReopenWithColumnFamilies({"default", "pikachu"}, std::vector({no_block_cache_opts, options})); Random rnd(301); // Write 8MB (80 values, each 100K) ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); std::vector values; std::string str; for (int i = 0; i < num_iter; i++) { if (i % 4 == 0) { // high compression ratio str = RandomString(&rnd, 1000); } values.push_back(str); ASSERT_OK(Put(1, Key(i), values[i])); } // flush all data from memtable so that reads are from block cache ASSERT_OK(Flush(1)); for (int i = 0; i < num_iter; i++) { ASSERT_EQ(Get(1, Key(i)), values[i]); } // check that we triggered the appropriate code paths in the cache switch (iter) { case 0: // only uncompressed block cache ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_MISS), 0); ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_COMPRESSED_MISS), 0); break; case 1: // no block cache, only compressed cache ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_MISS), 0); ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_COMPRESSED_MISS), 0); break; case 2: // both compressed and uncompressed block cache ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_MISS), 0); ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_COMPRESSED_MISS), 0); break; case 3: // both compressed and uncompressed block cache ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_MISS), 0); ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_HIT), 0); ASSERT_GT(TestGetTickerCount(options, BLOCK_CACHE_COMPRESSED_MISS), 0); // compressed doesn't have any hits since blocks are not compressed on // storage ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_COMPRESSED_HIT), 0); break; default: ASSERT_TRUE(false); } options.create_if_missing = true; DestroyAndReopen(options); } } static std::string CompressibleString(Random* rnd, int len) { std::string r; test::CompressibleString(rnd, 0.8, len, &r); return r; } #endif // ROCKSDB_LITE TEST_F(DBTest, FailMoreDbPaths) { Options options = CurrentOptions(); options.db_paths.emplace_back(dbname_, 10000000); options.db_paths.emplace_back(dbname_ + "_2", 1000000); options.db_paths.emplace_back(dbname_ + "_3", 1000000); options.db_paths.emplace_back(dbname_ + "_4", 1000000); options.db_paths.emplace_back(dbname_ + "_5", 1000000); ASSERT_TRUE(TryReopen(options).IsNotSupported()); } void CheckColumnFamilyMeta(const ColumnFamilyMetaData& cf_meta) { uint64_t cf_size = 0; uint64_t cf_csize = 0; size_t file_count = 0; for (auto level_meta : cf_meta.levels) { uint64_t level_size = 0; uint64_t level_csize = 0; file_count += level_meta.files.size(); for (auto file_meta : level_meta.files) { level_size += file_meta.size; } ASSERT_EQ(level_meta.size, level_size); cf_size += level_size; cf_csize += level_csize; } ASSERT_EQ(cf_meta.file_count, file_count); ASSERT_EQ(cf_meta.size, cf_size); } #ifndef ROCKSDB_LITE TEST_F(DBTest, ColumnFamilyMetaDataTest) { Options options = CurrentOptions(); options.create_if_missing = true; DestroyAndReopen(options); Random rnd(301); int key_index = 0; ColumnFamilyMetaData cf_meta; for (int i = 0; i < 100; ++i) { GenerateNewFile(&rnd, &key_index); db_->GetColumnFamilyMetaData(&cf_meta); CheckColumnFamilyMeta(cf_meta); } } namespace { void MinLevelHelper(DBTest* self, Options& options) { Random rnd(301); for (int num = 0; num < options.level0_file_num_compaction_trigger - 1; num++) { std::vector values; // Write 120KB (12 values, each 10K) for (int i = 0; i < 12; i++) { values.push_back(DBTestBase::RandomString(&rnd, 10000)); ASSERT_OK(self->Put(DBTestBase::Key(i), values[i])); } self->dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(self->NumTableFilesAtLevel(0), num + 1); } //generate one more file in level-0, and should trigger level-0 compaction std::vector values; for (int i = 0; i < 12; i++) { values.push_back(DBTestBase::RandomString(&rnd, 10000)); ASSERT_OK(self->Put(DBTestBase::Key(i), values[i])); } self->dbfull()->TEST_WaitForCompact(); ASSERT_EQ(self->NumTableFilesAtLevel(0), 0); ASSERT_EQ(self->NumTableFilesAtLevel(1), 1); } // returns false if the calling-Test should be skipped bool MinLevelToCompress(CompressionType& type, Options& options, int wbits, int lev, int strategy) { fprintf(stderr, "Test with compression options : window_bits = %d, level = %d, strategy = %d}\n", wbits, lev, strategy); options.write_buffer_size = 100<<10; //100KB options.arena_block_size = 4096; options.num_levels = 3; options.level0_file_num_compaction_trigger = 3; options.create_if_missing = true; if (Snappy_Supported()) { type = kSnappyCompression; fprintf(stderr, "using snappy\n"); } else if (Zlib_Supported()) { type = kZlibCompression; fprintf(stderr, "using zlib\n"); } else if (BZip2_Supported()) { type = kBZip2Compression; fprintf(stderr, "using bzip2\n"); } else if (LZ4_Supported()) { type = kLZ4Compression; fprintf(stderr, "using lz4\n"); } else { fprintf(stderr, "skipping test, compression disabled\n"); return false; } options.compression_per_level.resize(options.num_levels); // do not compress L0 for (int i = 0; i < 1; i++) { options.compression_per_level[i] = kNoCompression; } for (int i = 1; i < options.num_levels; i++) { options.compression_per_level[i] = type; } return true; } } // namespace TEST_F(DBTest, MinLevelToCompress1) { Options options = CurrentOptions(); CompressionType type = kSnappyCompression; if (!MinLevelToCompress(type, options, -14, -1, 0)) { return; } Reopen(options); MinLevelHelper(this, options); // do not compress L0 and L1 for (int i = 0; i < 2; i++) { options.compression_per_level[i] = kNoCompression; } for (int i = 2; i < options.num_levels; i++) { options.compression_per_level[i] = type; } DestroyAndReopen(options); MinLevelHelper(this, options); } TEST_F(DBTest, MinLevelToCompress2) { Options options = CurrentOptions(); CompressionType type = kSnappyCompression; if (!MinLevelToCompress(type, options, 15, -1, 0)) { return; } Reopen(options); MinLevelHelper(this, options); // do not compress L0 and L1 for (int i = 0; i < 2; i++) { options.compression_per_level[i] = kNoCompression; } for (int i = 2; i < options.num_levels; i++) { options.compression_per_level[i] = type; } DestroyAndReopen(options); MinLevelHelper(this, options); } TEST_F(DBTest, RepeatedWritesToSameKey) { do { Options options; options.env = env_; options.write_buffer_size = 100000; // Small write buffer options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // We must have at most one file per level except for level-0, // which may have up to kL0_StopWritesTrigger files. const int kMaxFiles = options.num_levels + options.level0_stop_writes_trigger; Random rnd(301); std::string value = RandomString(&rnd, static_cast(2 * options.write_buffer_size)); for (int i = 0; i < 5 * kMaxFiles; i++) { ASSERT_OK(Put(1, "key", value)); ASSERT_LE(TotalTableFiles(1), kMaxFiles); } } while (ChangeCompactOptions()); } #endif // ROCKSDB_LITE TEST_F(DBTest, SparseMerge) { do { Options options = CurrentOptions(); options.compression = kNoCompression; CreateAndReopenWithCF({"pikachu"}, options); FillLevels("A", "Z", 1); // Suppose there is: // small amount of data with prefix A // large amount of data with prefix B // small amount of data with prefix C // and that recent updates have made small changes to all three prefixes. // Check that we do not do a compaction that merges all of B in one shot. const std::string value(1000, 'x'); Put(1, "A", "va"); // Write approximately 100MB of "B" values for (int i = 0; i < 100000; i++) { char key[100]; snprintf(key, sizeof(key), "B%010d", i); Put(1, key, value); } Put(1, "C", "vc"); ASSERT_OK(Flush(1)); dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); // Make sparse update Put(1, "A", "va2"); Put(1, "B100", "bvalue2"); Put(1, "C", "vc2"); ASSERT_OK(Flush(1)); // Compactions should not cause us to create a situation where // a file overlaps too much data at the next level. ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(handles_[1]), 20 * 1048576); dbfull()->TEST_CompactRange(0, nullptr, nullptr); ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(handles_[1]), 20 * 1048576); dbfull()->TEST_CompactRange(1, nullptr, nullptr); ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(handles_[1]), 20 * 1048576); } while (ChangeCompactOptions()); } #ifndef ROCKSDB_LITE static bool Between(uint64_t val, uint64_t low, uint64_t high) { bool result = (val >= low) && (val <= high); if (!result) { fprintf(stderr, "Value %llu is not in range [%llu, %llu]\n", (unsigned long long)(val), (unsigned long long)(low), (unsigned long long)(high)); } return result; } TEST_F(DBTest, ApproximateSizesMemTable) { Options options; options.write_buffer_size = 100000000; // Large write buffer options.compression = kNoCompression; options.create_if_missing = true; options = CurrentOptions(options); DestroyAndReopen(options); const int N = 128; Random rnd(301); for (int i = 0; i < N; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024))); } uint64_t size; std::string start = Key(50); std::string end = Key(60); Range r(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_GT(size, 6000); ASSERT_LT(size, 204800); // Zero if not including mem table db_->GetApproximateSizes(&r, 1, &size, false); ASSERT_EQ(size, 0); start = Key(500); end = Key(600); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_EQ(size, 0); for (int i = 0; i < N; i++) { ASSERT_OK(Put(Key(1000 + i), RandomString(&rnd, 1024))); } start = Key(500); end = Key(600); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_EQ(size, 0); start = Key(100); end = Key(1020); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_GT(size, 6000); options.max_write_buffer_number = 8; options.min_write_buffer_number_to_merge = 5; options.write_buffer_size = 1024 * N; // Not very large DestroyAndReopen(options); int keys[N * 3]; for (int i = 0; i < N; i++) { keys[i * 3] = i * 5; keys[i * 3 + 1] = i * 5 + 1; keys[i * 3 + 2] = i * 5 + 2; } std::random_shuffle(std::begin(keys), std::end(keys)); for (int i = 0; i < N * 3; i++) { ASSERT_OK(Put(Key(keys[i] + 1000), RandomString(&rnd, 1024))); } start = Key(100); end = Key(300); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_EQ(size, 0); start = Key(1050); end = Key(1080); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_GT(size, 6000); start = Key(2100); end = Key(2300); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size, true); ASSERT_EQ(size, 0); start = Key(1050); end = Key(1080); r = Range(start, end); uint64_t size_with_mt, size_without_mt; db_->GetApproximateSizes(&r, 1, &size_with_mt, true); ASSERT_GT(size_with_mt, 6000); db_->GetApproximateSizes(&r, 1, &size_without_mt, false); ASSERT_EQ(size_without_mt, 0); Flush(); for (int i = 0; i < N; i++) { ASSERT_OK(Put(Key(i + 1000), RandomString(&rnd, 1024))); } start = Key(1050); end = Key(1080); r = Range(start, end); db_->GetApproximateSizes(&r, 1, &size_with_mt, true); db_->GetApproximateSizes(&r, 1, &size_without_mt, false); ASSERT_GT(size_with_mt, size_without_mt); ASSERT_GT(size_without_mt, 6000); } TEST_F(DBTest, ApproximateSizes) { do { Options options; options.write_buffer_size = 100000000; // Large write buffer options.compression = kNoCompression; options.create_if_missing = true; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_TRUE(Between(Size("", "xyz", 1), 0, 0)); ReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_TRUE(Between(Size("", "xyz", 1), 0, 0)); // Write 8MB (80 values, each 100K) ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); const int N = 80; static const int S1 = 100000; static const int S2 = 105000; // Allow some expansion from metadata Random rnd(301); for (int i = 0; i < N; i++) { ASSERT_OK(Put(1, Key(i), RandomString(&rnd, S1))); } // 0 because GetApproximateSizes() does not account for memtable space ASSERT_TRUE(Between(Size("", Key(50), 1), 0, 0)); // Check sizes across recovery by reopening a few times for (int run = 0; run < 3; run++) { ReopenWithColumnFamilies({"default", "pikachu"}, options); for (int compact_start = 0; compact_start < N; compact_start += 10) { for (int i = 0; i < N; i += 10) { ASSERT_TRUE(Between(Size("", Key(i), 1), S1 * i, S2 * i)); ASSERT_TRUE(Between(Size("", Key(i) + ".suffix", 1), S1 * (i + 1), S2 * (i + 1))); ASSERT_TRUE(Between(Size(Key(i), Key(i + 10), 1), S1 * 10, S2 * 10)); } ASSERT_TRUE(Between(Size("", Key(50), 1), S1 * 50, S2 * 50)); ASSERT_TRUE( Between(Size("", Key(50) + ".suffix", 1), S1 * 50, S2 * 50)); std::string cstart_str = Key(compact_start); std::string cend_str = Key(compact_start + 9); Slice cstart = cstart_str; Slice cend = cend_str; dbfull()->TEST_CompactRange(0, &cstart, &cend, handles_[1]); } ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_GT(NumTableFilesAtLevel(1, 1), 0); } // ApproximateOffsetOf() is not yet implemented in plain table format. } while (ChangeOptions(kSkipUniversalCompaction | kSkipFIFOCompaction | kSkipPlainTable | kSkipHashIndex)); } TEST_F(DBTest, ApproximateSizes_MixOfSmallAndLarge) { do { Options options = CurrentOptions(); options.compression = kNoCompression; CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); std::string big1 = RandomString(&rnd, 100000); ASSERT_OK(Put(1, Key(0), RandomString(&rnd, 10000))); ASSERT_OK(Put(1, Key(1), RandomString(&rnd, 10000))); ASSERT_OK(Put(1, Key(2), big1)); ASSERT_OK(Put(1, Key(3), RandomString(&rnd, 10000))); ASSERT_OK(Put(1, Key(4), big1)); ASSERT_OK(Put(1, Key(5), RandomString(&rnd, 10000))); ASSERT_OK(Put(1, Key(6), RandomString(&rnd, 300000))); ASSERT_OK(Put(1, Key(7), RandomString(&rnd, 10000))); // Check sizes across recovery by reopening a few times for (int run = 0; run < 3; run++) { ReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_TRUE(Between(Size("", Key(0), 1), 0, 0)); ASSERT_TRUE(Between(Size("", Key(1), 1), 10000, 11000)); ASSERT_TRUE(Between(Size("", Key(2), 1), 20000, 21000)); ASSERT_TRUE(Between(Size("", Key(3), 1), 120000, 121000)); ASSERT_TRUE(Between(Size("", Key(4), 1), 130000, 131000)); ASSERT_TRUE(Between(Size("", Key(5), 1), 230000, 231000)); ASSERT_TRUE(Between(Size("", Key(6), 1), 240000, 241000)); ASSERT_TRUE(Between(Size("", Key(7), 1), 540000, 541000)); ASSERT_TRUE(Between(Size("", Key(8), 1), 550000, 560000)); ASSERT_TRUE(Between(Size(Key(3), Key(5), 1), 110000, 111000)); dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]); } // ApproximateOffsetOf() is not yet implemented in plain table format. } while (ChangeOptions(kSkipPlainTable)); } #endif // ROCKSDB_LITE TEST_F(DBTest, IteratorPinsRef) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); Put(1, "foo", "hello"); // Get iterator that will yield the current contents of the DB. Iterator* iter = db_->NewIterator(ReadOptions(), handles_[1]); // Write to force compactions Put(1, "foo", "newvalue1"); for (int i = 0; i < 100; i++) { // 100K values ASSERT_OK(Put(1, Key(i), Key(i) + std::string(100000, 'v'))); } Put(1, "foo", "newvalue2"); iter->SeekToFirst(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("foo", iter->key().ToString()); ASSERT_EQ("hello", iter->value().ToString()); iter->Next(); ASSERT_TRUE(!iter->Valid()); delete iter; } while (ChangeCompactOptions()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, Snapshot) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions(options_override)); Put(0, "foo", "0v1"); Put(1, "foo", "1v1"); const Snapshot* s1 = db_->GetSnapshot(); ASSERT_EQ(1U, GetNumSnapshots()); uint64_t time_snap1 = GetTimeOldestSnapshots(); ASSERT_GT(time_snap1, 0U); Put(0, "foo", "0v2"); Put(1, "foo", "1v2"); env_->addon_time_.fetch_add(1); const Snapshot* s2 = db_->GetSnapshot(); ASSERT_EQ(2U, GetNumSnapshots()); ASSERT_EQ(time_snap1, GetTimeOldestSnapshots()); Put(0, "foo", "0v3"); Put(1, "foo", "1v3"); { ManagedSnapshot s3(db_); ASSERT_EQ(3U, GetNumSnapshots()); ASSERT_EQ(time_snap1, GetTimeOldestSnapshots()); Put(0, "foo", "0v4"); Put(1, "foo", "1v4"); ASSERT_EQ("0v1", Get(0, "foo", s1)); ASSERT_EQ("1v1", Get(1, "foo", s1)); ASSERT_EQ("0v2", Get(0, "foo", s2)); ASSERT_EQ("1v2", Get(1, "foo", s2)); ASSERT_EQ("0v3", Get(0, "foo", s3.snapshot())); ASSERT_EQ("1v3", Get(1, "foo", s3.snapshot())); ASSERT_EQ("0v4", Get(0, "foo")); ASSERT_EQ("1v4", Get(1, "foo")); } ASSERT_EQ(2U, GetNumSnapshots()); ASSERT_EQ(time_snap1, GetTimeOldestSnapshots()); ASSERT_EQ("0v1", Get(0, "foo", s1)); ASSERT_EQ("1v1", Get(1, "foo", s1)); ASSERT_EQ("0v2", Get(0, "foo", s2)); ASSERT_EQ("1v2", Get(1, "foo", s2)); ASSERT_EQ("0v4", Get(0, "foo")); ASSERT_EQ("1v4", Get(1, "foo")); db_->ReleaseSnapshot(s1); ASSERT_EQ("0v2", Get(0, "foo", s2)); ASSERT_EQ("1v2", Get(1, "foo", s2)); ASSERT_EQ("0v4", Get(0, "foo")); ASSERT_EQ("1v4", Get(1, "foo")); ASSERT_EQ(1U, GetNumSnapshots()); ASSERT_LT(time_snap1, GetTimeOldestSnapshots()); db_->ReleaseSnapshot(s2); ASSERT_EQ(0U, GetNumSnapshots()); ASSERT_EQ("0v4", Get(0, "foo")); ASSERT_EQ("1v4", Get(1, "foo")); } while (ChangeOptions(kSkipHashCuckoo)); } TEST_F(DBTest, HiddenValuesAreRemoved) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { Options options = CurrentOptions(options_override); CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); FillLevels("a", "z", 1); std::string big = RandomString(&rnd, 50000); Put(1, "foo", big); Put(1, "pastfoo", "v"); const Snapshot* snapshot = db_->GetSnapshot(); Put(1, "foo", "tiny"); Put(1, "pastfoo2", "v2"); // Advance sequence number one more ASSERT_OK(Flush(1)); ASSERT_GT(NumTableFilesAtLevel(0, 1), 0); ASSERT_EQ(big, Get(1, "foo", snapshot)); ASSERT_TRUE(Between(Size("", "pastfoo", 1), 50000, 60000)); db_->ReleaseSnapshot(snapshot); ASSERT_EQ(AllEntriesFor("foo", 1), "[ tiny, " + big + " ]"); Slice x("x"); dbfull()->TEST_CompactRange(0, nullptr, &x, handles_[1]); ASSERT_EQ(AllEntriesFor("foo", 1), "[ tiny ]"); ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); ASSERT_GE(NumTableFilesAtLevel(1, 1), 1); dbfull()->TEST_CompactRange(1, nullptr, &x, handles_[1]); ASSERT_EQ(AllEntriesFor("foo", 1), "[ tiny ]"); ASSERT_TRUE(Between(Size("", "pastfoo", 1), 0, 1000)); // ApproximateOffsetOf() is not yet implemented in plain table format, // which is used by Size(). // skip HashCuckooRep as it does not support snapshot } while (ChangeOptions(kSkipUniversalCompaction | kSkipFIFOCompaction | kSkipPlainTable | kSkipHashCuckoo)); } #endif // ROCKSDB_LITE TEST_F(DBTest, CompactBetweenSnapshots) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { Options options = CurrentOptions(options_override); options.disable_auto_compactions = true; CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); FillLevels("a", "z", 1); Put(1, "foo", "first"); const Snapshot* snapshot1 = db_->GetSnapshot(); Put(1, "foo", "second"); Put(1, "foo", "third"); Put(1, "foo", "fourth"); const Snapshot* snapshot2 = db_->GetSnapshot(); Put(1, "foo", "fifth"); Put(1, "foo", "sixth"); // All entries (including duplicates) exist // before any compaction or flush is triggered. ASSERT_EQ(AllEntriesFor("foo", 1), "[ sixth, fifth, fourth, third, second, first ]"); ASSERT_EQ("sixth", Get(1, "foo")); ASSERT_EQ("fourth", Get(1, "foo", snapshot2)); ASSERT_EQ("first", Get(1, "foo", snapshot1)); // After a flush, "second", "third" and "fifth" should // be removed ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ sixth, fourth, first ]"); // after we release the snapshot1, only two values left db_->ReleaseSnapshot(snapshot1); FillLevels("a", "z", 1); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); // We have only one valid snapshot snapshot2. Since snapshot1 is // not valid anymore, "first" should be removed by a compaction. ASSERT_EQ("sixth", Get(1, "foo")); ASSERT_EQ("fourth", Get(1, "foo", snapshot2)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ sixth, fourth ]"); // after we release the snapshot2, only one value should be left db_->ReleaseSnapshot(snapshot2); FillLevels("a", "z", 1); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ("sixth", Get(1, "foo")); ASSERT_EQ(AllEntriesFor("foo", 1), "[ sixth ]"); // skip HashCuckooRep as it does not support snapshot } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction)); } TEST_F(DBTest, UnremovableSingleDelete) { // If we compact: // // Put(A, v1) Snapshot SingleDelete(A) Put(A, v2) // // We do not want to end up with: // // Put(A, v1) Snapshot Put(A, v2) // // Because a subsequent SingleDelete(A) would delete the Put(A, v2) // but not Put(A, v1), so Get(A) would return v1. anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { Options options = CurrentOptions(options_override); options.disable_auto_compactions = true; CreateAndReopenWithCF({"pikachu"}, options); Put(1, "foo", "first"); const Snapshot* snapshot = db_->GetSnapshot(); SingleDelete(1, "foo"); Put(1, "foo", "second"); ASSERT_OK(Flush(1)); ASSERT_EQ("first", Get(1, "foo", snapshot)); ASSERT_EQ("second", Get(1, "foo")); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ("[ second, SDEL, first ]", AllEntriesFor("foo", 1)); SingleDelete(1, "foo"); ASSERT_EQ("first", Get(1, "foo", snapshot)); ASSERT_EQ("NOT_FOUND", Get(1, "foo")); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ("first", Get(1, "foo", snapshot)); ASSERT_EQ("NOT_FOUND", Get(1, "foo")); db_->ReleaseSnapshot(snapshot); // Skip HashCuckooRep as it does not support single delete. FIFO and // universal compaction do not apply to the test case. Skip MergePut // because single delete does not get removed when it encounters a merge. } while (ChangeOptions(kSkipHashCuckoo | kSkipFIFOCompaction | kSkipUniversalCompaction | kSkipMergePut)); } #ifndef ROCKSDB_LITE TEST_F(DBTest, DeletionMarkers1) { Options options = CurrentOptions(); options.max_background_flushes = 0; CreateAndReopenWithCF({"pikachu"}, options); Put(1, "foo", "v1"); ASSERT_OK(Flush(1)); const int last = 2; MoveFilesToLevel(last, 1); // foo => v1 is now in last level ASSERT_EQ(NumTableFilesAtLevel(last, 1), 1); // Place a table at level last-1 to prevent merging with preceding mutation Put(1, "a", "begin"); Put(1, "z", "end"); Flush(1); MoveFilesToLevel(last - 1, 1); ASSERT_EQ(NumTableFilesAtLevel(last, 1), 1); ASSERT_EQ(NumTableFilesAtLevel(last - 1, 1), 1); Delete(1, "foo"); Put(1, "foo", "v2"); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2, DEL, v1 ]"); ASSERT_OK(Flush(1)); // Moves to level last-2 ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2, v1 ]"); Slice z("z"); dbfull()->TEST_CompactRange(last - 2, nullptr, &z, handles_[1]); // DEL eliminated, but v1 remains because we aren't compacting that level // (DEL can be eliminated because v2 hides v1). ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2, v1 ]"); dbfull()->TEST_CompactRange(last - 1, nullptr, nullptr, handles_[1]); // Merging last-1 w/ last, so we are the base level for "foo", so // DEL is removed. (as is v1). ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2 ]"); } TEST_F(DBTest, DeletionMarkers2) { Options options = CurrentOptions(); CreateAndReopenWithCF({"pikachu"}, options); Put(1, "foo", "v1"); ASSERT_OK(Flush(1)); const int last = 2; MoveFilesToLevel(last, 1); // foo => v1 is now in last level ASSERT_EQ(NumTableFilesAtLevel(last, 1), 1); // Place a table at level last-1 to prevent merging with preceding mutation Put(1, "a", "begin"); Put(1, "z", "end"); Flush(1); MoveFilesToLevel(last - 1, 1); ASSERT_EQ(NumTableFilesAtLevel(last, 1), 1); ASSERT_EQ(NumTableFilesAtLevel(last - 1, 1), 1); Delete(1, "foo"); ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL, v1 ]"); ASSERT_OK(Flush(1)); // Moves to level last-2 ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL, v1 ]"); dbfull()->TEST_CompactRange(last - 2, nullptr, nullptr, handles_[1]); // DEL kept: "last" file overlaps ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL, v1 ]"); dbfull()->TEST_CompactRange(last - 1, nullptr, nullptr, handles_[1]); // Merging last-1 w/ last, so we are the base level for "foo", so // DEL is removed. (as is v1). ASSERT_EQ(AllEntriesFor("foo", 1), "[ ]"); } TEST_F(DBTest, OverlapInLevel0) { do { Options options = CurrentOptions(); CreateAndReopenWithCF({"pikachu"}, options); //Fill levels 1 and 2 to disable the pushing of new memtables to levels > 0. ASSERT_OK(Put(1, "100", "v100")); ASSERT_OK(Put(1, "999", "v999")); Flush(1); MoveFilesToLevel(2, 1); ASSERT_OK(Delete(1, "100")); ASSERT_OK(Delete(1, "999")); Flush(1); MoveFilesToLevel(1, 1); ASSERT_EQ("0,1,1", FilesPerLevel(1)); // Make files spanning the following ranges in level-0: // files[0] 200 .. 900 // files[1] 300 .. 500 // Note that files are sorted by smallest key. ASSERT_OK(Put(1, "300", "v300")); ASSERT_OK(Put(1, "500", "v500")); Flush(1); ASSERT_OK(Put(1, "200", "v200")); ASSERT_OK(Put(1, "600", "v600")); ASSERT_OK(Put(1, "900", "v900")); Flush(1); ASSERT_EQ("2,1,1", FilesPerLevel(1)); // Compact away the placeholder files we created initially dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]); dbfull()->TEST_CompactRange(2, nullptr, nullptr, handles_[1]); ASSERT_EQ("2", FilesPerLevel(1)); // Do a memtable compaction. Before bug-fix, the compaction would // not detect the overlap with level-0 files and would incorrectly place // the deletion in a deeper level. ASSERT_OK(Delete(1, "600")); Flush(1); ASSERT_EQ("3", FilesPerLevel(1)); ASSERT_EQ("NOT_FOUND", Get(1, "600")); } while (ChangeOptions(kSkipUniversalCompaction | kSkipFIFOCompaction)); } #endif // ROCKSDB_LITE TEST_F(DBTest, ComparatorCheck) { class NewComparator : public Comparator { public: virtual const char* Name() const override { return "rocksdb.NewComparator"; } virtual int Compare(const Slice& a, const Slice& b) const override { return BytewiseComparator()->Compare(a, b); } virtual void FindShortestSeparator(std::string* s, const Slice& l) const override { BytewiseComparator()->FindShortestSeparator(s, l); } virtual void FindShortSuccessor(std::string* key) const override { BytewiseComparator()->FindShortSuccessor(key); } }; Options new_options, options; NewComparator cmp; do { options = CurrentOptions(); CreateAndReopenWithCF({"pikachu"}, options); new_options = CurrentOptions(); new_options.comparator = &cmp; // only the non-default column family has non-matching comparator Status s = TryReopenWithColumnFamilies({"default", "pikachu"}, std::vector({options, new_options})); ASSERT_TRUE(!s.ok()); ASSERT_TRUE(s.ToString().find("comparator") != std::string::npos) << s.ToString(); } while (ChangeCompactOptions()); } TEST_F(DBTest, CustomComparator) { class NumberComparator : public Comparator { public: virtual const char* Name() const override { return "test.NumberComparator"; } virtual int Compare(const Slice& a, const Slice& b) const override { return ToNumber(a) - ToNumber(b); } virtual void FindShortestSeparator(std::string* s, const Slice& l) const override { ToNumber(*s); // Check format ToNumber(l); // Check format } virtual void FindShortSuccessor(std::string* key) const override { ToNumber(*key); // Check format } private: static int ToNumber(const Slice& x) { // Check that there are no extra characters. EXPECT_TRUE(x.size() >= 2 && x[0] == '[' && x[x.size() - 1] == ']') << EscapeString(x); int val; char ignored; EXPECT_TRUE(sscanf(x.ToString().c_str(), "[%i]%c", &val, &ignored) == 1) << EscapeString(x); return val; } }; Options new_options; NumberComparator cmp; do { new_options = CurrentOptions(); new_options.create_if_missing = true; new_options.comparator = &cmp; new_options.write_buffer_size = 4096; // Compact more often new_options.arena_block_size = 4096; new_options = CurrentOptions(new_options); DestroyAndReopen(new_options); CreateAndReopenWithCF({"pikachu"}, new_options); ASSERT_OK(Put(1, "[10]", "ten")); ASSERT_OK(Put(1, "[0x14]", "twenty")); for (int i = 0; i < 2; i++) { ASSERT_EQ("ten", Get(1, "[10]")); ASSERT_EQ("ten", Get(1, "[0xa]")); ASSERT_EQ("twenty", Get(1, "[20]")); ASSERT_EQ("twenty", Get(1, "[0x14]")); ASSERT_EQ("NOT_FOUND", Get(1, "[15]")); ASSERT_EQ("NOT_FOUND", Get(1, "[0xf]")); Compact(1, "[0]", "[9999]"); } for (int run = 0; run < 2; run++) { for (int i = 0; i < 1000; i++) { char buf[100]; snprintf(buf, sizeof(buf), "[%d]", i*10); ASSERT_OK(Put(1, buf, buf)); } Compact(1, "[0]", "[1000000]"); } } while (ChangeCompactOptions()); } TEST_F(DBTest, DBOpen_Options) { Options options = CurrentOptions(); std::string dbname = test::TmpDir(env_) + "/db_options_test"; ASSERT_OK(DestroyDB(dbname, options)); // Does not exist, and create_if_missing == false: error DB* db = nullptr; options.create_if_missing = false; Status s = DB::Open(options, dbname, &db); ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != nullptr); ASSERT_TRUE(db == nullptr); // Does not exist, and create_if_missing == true: OK options.create_if_missing = true; s = DB::Open(options, dbname, &db); ASSERT_OK(s); ASSERT_TRUE(db != nullptr); delete db; db = nullptr; // Does exist, and error_if_exists == true: error options.create_if_missing = false; options.error_if_exists = true; s = DB::Open(options, dbname, &db); ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != nullptr); ASSERT_TRUE(db == nullptr); // Does exist, and error_if_exists == false: OK options.create_if_missing = true; options.error_if_exists = false; s = DB::Open(options, dbname, &db); ASSERT_OK(s); ASSERT_TRUE(db != nullptr); delete db; db = nullptr; } TEST_F(DBTest, DBOpen_Change_NumLevels) { Options options = CurrentOptions(); options.create_if_missing = true; DestroyAndReopen(options); ASSERT_TRUE(db_ != nullptr); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "a", "123")); ASSERT_OK(Put(1, "b", "234")); Flush(1); MoveFilesToLevel(3, 1); Close(); options.create_if_missing = false; options.num_levels = 2; Status s = TryReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_TRUE(strstr(s.ToString().c_str(), "Invalid argument") != nullptr); ASSERT_TRUE(db_ == nullptr); } TEST_F(DBTest, DestroyDBMetaDatabase) { std::string dbname = test::TmpDir(env_) + "/db_meta"; ASSERT_OK(env_->CreateDirIfMissing(dbname)); std::string metadbname = MetaDatabaseName(dbname, 0); ASSERT_OK(env_->CreateDirIfMissing(metadbname)); std::string metametadbname = MetaDatabaseName(metadbname, 0); ASSERT_OK(env_->CreateDirIfMissing(metametadbname)); // Destroy previous versions if they exist. Using the long way. Options options = CurrentOptions(); ASSERT_OK(DestroyDB(metametadbname, options)); ASSERT_OK(DestroyDB(metadbname, options)); ASSERT_OK(DestroyDB(dbname, options)); // Setup databases DB* db = nullptr; ASSERT_OK(DB::Open(options, dbname, &db)); delete db; db = nullptr; ASSERT_OK(DB::Open(options, metadbname, &db)); delete db; db = nullptr; ASSERT_OK(DB::Open(options, metametadbname, &db)); delete db; db = nullptr; // Delete databases ASSERT_OK(DestroyDB(dbname, options)); // Check if deletion worked. options.create_if_missing = false; ASSERT_TRUE(!(DB::Open(options, dbname, &db)).ok()); ASSERT_TRUE(!(DB::Open(options, metadbname, &db)).ok()); ASSERT_TRUE(!(DB::Open(options, metametadbname, &db)).ok()); } #ifndef ROCKSDB_LITE // Check that number of files does not grow when writes are dropped TEST_F(DBTest, DropWrites) { do { Options options = CurrentOptions(); options.env = env_; options.paranoid_checks = false; Reopen(options); ASSERT_OK(Put("foo", "v1")); ASSERT_EQ("v1", Get("foo")); Compact("a", "z"); const size_t num_files = CountFiles(); // Force out-of-space errors env_->drop_writes_.store(true, std::memory_order_release); env_->sleep_counter_.Reset(); env_->no_sleep_ = true; for (int i = 0; i < 5; i++) { if (option_config_ != kUniversalCompactionMultiLevel && option_config_ != kUniversalSubcompactions) { for (int level = 0; level < dbfull()->NumberLevels(); level++) { if (level > 0 && level == dbfull()->NumberLevels() - 1) { break; } dbfull()->TEST_CompactRange(level, nullptr, nullptr, nullptr, true /* disallow trivial move */); } } else { dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); } } std::string property_value; ASSERT_TRUE(db_->GetProperty("rocksdb.background-errors", &property_value)); ASSERT_EQ("5", property_value); env_->drop_writes_.store(false, std::memory_order_release); ASSERT_LT(CountFiles(), num_files + 3); // Check that compaction attempts slept after errors // TODO @krad: Figure out why ASSERT_EQ 5 keeps failing in certain compiler // versions ASSERT_GE(env_->sleep_counter_.Read(), 4); } while (ChangeCompactOptions()); } // Check background error counter bumped on flush failures. TEST_F(DBTest, DropWritesFlush) { do { Options options = CurrentOptions(); options.env = env_; options.max_background_flushes = 1; Reopen(options); ASSERT_OK(Put("foo", "v1")); // Force out-of-space errors env_->drop_writes_.store(true, std::memory_order_release); std::string property_value; // Background error count is 0 now. ASSERT_TRUE(db_->GetProperty("rocksdb.background-errors", &property_value)); ASSERT_EQ("0", property_value); dbfull()->TEST_FlushMemTable(true); ASSERT_TRUE(db_->GetProperty("rocksdb.background-errors", &property_value)); ASSERT_EQ("1", property_value); env_->drop_writes_.store(false, std::memory_order_release); } while (ChangeCompactOptions()); } #endif // ROCKSDB_LITE // Check that CompactRange() returns failure if there is not enough space left // on device TEST_F(DBTest, NoSpaceCompactRange) { do { Options options = CurrentOptions(); options.env = env_; options.disable_auto_compactions = true; Reopen(options); // generate 5 tables for (int i = 0; i < 5; ++i) { ASSERT_OK(Put(Key(i), Key(i) + "v")); ASSERT_OK(Flush()); } // Force out-of-space errors env_->no_space_.store(true, std::memory_order_release); Status s = dbfull()->TEST_CompactRange(0, nullptr, nullptr, nullptr, true /* disallow trivial move */); ASSERT_TRUE(s.IsIOError()); env_->no_space_.store(false, std::memory_order_release); } while (ChangeCompactOptions()); } TEST_F(DBTest, NonWritableFileSystem) { do { Options options = CurrentOptions(); options.write_buffer_size = 4096; options.arena_block_size = 4096; options.env = env_; Reopen(options); ASSERT_OK(Put("foo", "v1")); env_->non_writeable_rate_.store(100); std::string big(100000, 'x'); int errors = 0; for (int i = 0; i < 20; i++) { if (!Put("foo", big).ok()) { errors++; env_->SleepForMicroseconds(100000); } } ASSERT_GT(errors, 0); env_->non_writeable_rate_.store(0); } while (ChangeCompactOptions()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, ManifestWriteError) { // Test for the following problem: // (a) Compaction produces file F // (b) Log record containing F is written to MANIFEST file, but Sync() fails // (c) GC deletes F // (d) After reopening DB, reads fail since deleted F is named in log record // We iterate twice. In the second iteration, everything is the // same except the log record never makes it to the MANIFEST file. for (int iter = 0; iter < 2; iter++) { std::atomic* error_type = (iter == 0) ? &env_->manifest_sync_error_ : &env_->manifest_write_error_; // Insert foo=>bar mapping Options options = CurrentOptions(); options.env = env_; options.create_if_missing = true; options.error_if_exists = false; options.paranoid_checks = true; DestroyAndReopen(options); ASSERT_OK(Put("foo", "bar")); ASSERT_EQ("bar", Get("foo")); // Memtable compaction (will succeed) Flush(); ASSERT_EQ("bar", Get("foo")); const int last = 2; MoveFilesToLevel(2); ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo=>bar is now in last level // Merging compaction (will fail) error_type->store(true, std::memory_order_release); dbfull()->TEST_CompactRange(last, nullptr, nullptr); // Should fail ASSERT_EQ("bar", Get("foo")); error_type->store(false, std::memory_order_release); // Since paranoid_checks=true, writes should fail ASSERT_NOK(Put("foo2", "bar2")); // Recovery: should not lose data ASSERT_EQ("bar", Get("foo")); // Try again with paranoid_checks=false Close(); options.paranoid_checks = false; Reopen(options); // Merging compaction (will fail) error_type->store(true, std::memory_order_release); dbfull()->TEST_CompactRange(last, nullptr, nullptr); // Should fail ASSERT_EQ("bar", Get("foo")); // Recovery: should not lose data error_type->store(false, std::memory_order_release); Reopen(options); ASSERT_EQ("bar", Get("foo")); // Since paranoid_checks=false, writes should succeed ASSERT_OK(Put("foo2", "bar2")); ASSERT_EQ("bar", Get("foo")); ASSERT_EQ("bar2", Get("foo2")); } } #endif // ROCKSDB_LITE TEST_F(DBTest, PutFailsParanoid) { // Test the following: // (a) A random put fails in paranoid mode (simulate by sync fail) // (b) All other puts have to fail, even if writes would succeed // (c) All of that should happen ONLY if paranoid_checks = true Options options = CurrentOptions(); options.env = env_; options.create_if_missing = true; options.error_if_exists = false; options.paranoid_checks = true; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); Status s; ASSERT_OK(Put(1, "foo", "bar")); ASSERT_OK(Put(1, "foo1", "bar1")); // simulate error env_->log_write_error_.store(true, std::memory_order_release); s = Put(1, "foo2", "bar2"); ASSERT_TRUE(!s.ok()); env_->log_write_error_.store(false, std::memory_order_release); s = Put(1, "foo3", "bar3"); // the next put should fail, too ASSERT_TRUE(!s.ok()); // but we're still able to read ASSERT_EQ("bar", Get(1, "foo")); // do the same thing with paranoid checks off options.paranoid_checks = false; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "foo", "bar")); ASSERT_OK(Put(1, "foo1", "bar1")); // simulate error env_->log_write_error_.store(true, std::memory_order_release); s = Put(1, "foo2", "bar2"); ASSERT_TRUE(!s.ok()); env_->log_write_error_.store(false, std::memory_order_release); s = Put(1, "foo3", "bar3"); // the next put should NOT fail ASSERT_TRUE(s.ok()); } TEST_F(DBTest, BloomFilter) { do { Options options = CurrentOptions(); env_->count_random_reads_ = true; options.env = env_; // ChangeCompactOptions() only changes compaction style, which does not // trigger reset of table_factory BlockBasedTableOptions table_options; table_options.no_block_cache = true; table_options.filter_policy.reset(NewBloomFilterPolicy(10)); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); CreateAndReopenWithCF({"pikachu"}, options); // Populate multiple layers const int N = 10000; for (int i = 0; i < N; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } Compact(1, "a", "z"); for (int i = 0; i < N; i += 100) { ASSERT_OK(Put(1, Key(i), Key(i))); } Flush(1); // Prevent auto compactions triggered by seeks env_->delay_sstable_sync_.store(true, std::memory_order_release); // Lookup present keys. Should rarely read from small sstable. env_->random_read_counter_.Reset(); for (int i = 0; i < N; i++) { ASSERT_EQ(Key(i), Get(1, Key(i))); } int reads = env_->random_read_counter_.Read(); fprintf(stderr, "%d present => %d reads\n", N, reads); ASSERT_GE(reads, N); ASSERT_LE(reads, N + 2*N/100); // Lookup present keys. Should rarely read from either sstable. env_->random_read_counter_.Reset(); for (int i = 0; i < N; i++) { ASSERT_EQ("NOT_FOUND", Get(1, Key(i) + ".missing")); } reads = env_->random_read_counter_.Read(); fprintf(stderr, "%d missing => %d reads\n", N, reads); ASSERT_LE(reads, 3*N/100); env_->delay_sstable_sync_.store(false, std::memory_order_release); Close(); } while (ChangeCompactOptions()); } TEST_F(DBTest, BloomFilterRate) { while (ChangeFilterOptions()) { Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); CreateAndReopenWithCF({"pikachu"}, options); const int maxKey = 10000; for (int i = 0; i < maxKey; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } // Add a large key to make the file contain wide range ASSERT_OK(Put(1, Key(maxKey + 55555), Key(maxKey + 55555))); Flush(1); // Check if they can be found for (int i = 0; i < maxKey; i++) { ASSERT_EQ(Key(i), Get(1, Key(i))); } ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); // Check if filter is useful for (int i = 0; i < maxKey; i++) { ASSERT_EQ("NOT_FOUND", Get(1, Key(i+33333))); } ASSERT_GE(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), maxKey*0.98); } } TEST_F(DBTest, BloomFilterCompatibility) { Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); BlockBasedTableOptions table_options; table_options.filter_policy.reset(NewBloomFilterPolicy(10, true)); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); // Create with block based filter CreateAndReopenWithCF({"pikachu"}, options); const int maxKey = 10000; for (int i = 0; i < maxKey; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } ASSERT_OK(Put(1, Key(maxKey + 55555), Key(maxKey + 55555))); Flush(1); // Check db with full filter table_options.filter_policy.reset(NewBloomFilterPolicy(10, false)); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); ReopenWithColumnFamilies({"default", "pikachu"}, options); // Check if they can be found for (int i = 0; i < maxKey; i++) { ASSERT_EQ(Key(i), Get(1, Key(i))); } ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); } TEST_F(DBTest, BloomFilterReverseCompatibility) { Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); BlockBasedTableOptions table_options; table_options.filter_policy.reset(NewBloomFilterPolicy(10, false)); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); // Create with full filter CreateAndReopenWithCF({"pikachu"}, options); const int maxKey = 10000; for (int i = 0; i < maxKey; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } ASSERT_OK(Put(1, Key(maxKey + 55555), Key(maxKey + 55555))); Flush(1); // Check db with block_based filter table_options.filter_policy.reset(NewBloomFilterPolicy(10, true)); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); ReopenWithColumnFamilies({"default", "pikachu"}, options); // Check if they can be found for (int i = 0; i < maxKey; i++) { ASSERT_EQ(Key(i), Get(1, Key(i))); } ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); } namespace { // A wrapped bloom over default FilterPolicy class WrappedBloom : public FilterPolicy { public: explicit WrappedBloom(int bits_per_key) : filter_(NewBloomFilterPolicy(bits_per_key)), counter_(0) {} ~WrappedBloom() { delete filter_; } const char* Name() const override { return "WrappedRocksDbFilterPolicy"; } void CreateFilter(const rocksdb::Slice* keys, int n, std::string* dst) const override { std::unique_ptr user_keys(new rocksdb::Slice[n]); for (int i = 0; i < n; ++i) { user_keys[i] = convertKey(keys[i]); } return filter_->CreateFilter(user_keys.get(), n, dst); } bool KeyMayMatch(const rocksdb::Slice& key, const rocksdb::Slice& filter) const override { counter_++; return filter_->KeyMayMatch(convertKey(key), filter); } uint32_t GetCounter() { return counter_; } private: const FilterPolicy* filter_; mutable uint32_t counter_; rocksdb::Slice convertKey(const rocksdb::Slice& key) const { return key; } }; } // namespace TEST_F(DBTest, BloomFilterWrapper) { Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); BlockBasedTableOptions table_options; WrappedBloom* policy = new WrappedBloom(10); table_options.filter_policy.reset(policy); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); CreateAndReopenWithCF({"pikachu"}, options); const int maxKey = 10000; for (int i = 0; i < maxKey; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } // Add a large key to make the file contain wide range ASSERT_OK(Put(1, Key(maxKey + 55555), Key(maxKey + 55555))); ASSERT_EQ(0U, policy->GetCounter()); Flush(1); // Check if they can be found for (int i = 0; i < maxKey; i++) { ASSERT_EQ(Key(i), Get(1, Key(i))); } ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); ASSERT_EQ(1U * maxKey, policy->GetCounter()); // Check if filter is useful for (int i = 0; i < maxKey; i++) { ASSERT_EQ("NOT_FOUND", Get(1, Key(i+33333))); } ASSERT_GE(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), maxKey*0.98); ASSERT_EQ(2U * maxKey, policy->GetCounter()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, SnapshotFiles) { do { Options options = CurrentOptions(); options.write_buffer_size = 100000000; // Large write buffer CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); // Write 8MB (80 values, each 100K) ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); std::vector values; for (int i = 0; i < 80; i++) { values.push_back(RandomString(&rnd, 100000)); ASSERT_OK(Put((i < 40), Key(i), values[i])); } // assert that nothing makes it to disk yet. ASSERT_EQ(NumTableFilesAtLevel(0, 1), 0); // get a file snapshot uint64_t manifest_number = 0; uint64_t manifest_size = 0; std::vector files; dbfull()->DisableFileDeletions(); dbfull()->GetLiveFiles(files, &manifest_size); // CURRENT, MANIFEST, *.sst files (one for each CF) ASSERT_EQ(files.size(), 4U); uint64_t number = 0; FileType type; // copy these files to a new snapshot directory std::string snapdir = dbname_ + ".snapdir/"; ASSERT_OK(env_->CreateDirIfMissing(snapdir)); for (size_t i = 0; i < files.size(); i++) { // our clients require that GetLiveFiles returns // files with "/" as first character! ASSERT_EQ(files[i][0], '/'); std::string src = dbname_ + files[i]; std::string dest = snapdir + files[i]; uint64_t size; ASSERT_OK(env_->GetFileSize(src, &size)); // record the number and the size of the // latest manifest file if (ParseFileName(files[i].substr(1), &number, &type)) { if (type == kDescriptorFile) { if (number > manifest_number) { manifest_number = number; ASSERT_GE(size, manifest_size); size = manifest_size; // copy only valid MANIFEST data } } } CopyFile(src, dest, size); } // release file snapshot dbfull()->DisableFileDeletions(); // overwrite one key, this key should not appear in the snapshot std::vector extras; for (unsigned int i = 0; i < 1; i++) { extras.push_back(RandomString(&rnd, 100000)); ASSERT_OK(Put(0, Key(i), extras[i])); } // verify that data in the snapshot are correct std::vector column_families; column_families.emplace_back("default", ColumnFamilyOptions()); column_families.emplace_back("pikachu", ColumnFamilyOptions()); std::vector cf_handles; DB* snapdb; DBOptions opts; opts.env = env_; opts.create_if_missing = false; Status stat = DB::Open(opts, snapdir, column_families, &cf_handles, &snapdb); ASSERT_OK(stat); ReadOptions roptions; std::string val; for (unsigned int i = 0; i < 80; i++) { stat = snapdb->Get(roptions, cf_handles[i < 40], Key(i), &val); ASSERT_EQ(values[i].compare(val), 0); } for (auto cfh : cf_handles) { delete cfh; } delete snapdb; // look at the new live files after we added an 'extra' key // and after we took the first snapshot. uint64_t new_manifest_number = 0; uint64_t new_manifest_size = 0; std::vector newfiles; dbfull()->DisableFileDeletions(); dbfull()->GetLiveFiles(newfiles, &new_manifest_size); // find the new manifest file. assert that this manifest file is // the same one as in the previous snapshot. But its size should be // larger because we added an extra key after taking the // previous shapshot. for (size_t i = 0; i < newfiles.size(); i++) { std::string src = dbname_ + "/" + newfiles[i]; // record the lognumber and the size of the // latest manifest file if (ParseFileName(newfiles[i].substr(1), &number, &type)) { if (type == kDescriptorFile) { if (number > new_manifest_number) { uint64_t size; new_manifest_number = number; ASSERT_OK(env_->GetFileSize(src, &size)); ASSERT_GE(size, new_manifest_size); } } } } ASSERT_EQ(manifest_number, new_manifest_number); ASSERT_GT(new_manifest_size, manifest_size); // release file snapshot dbfull()->DisableFileDeletions(); } while (ChangeCompactOptions()); } #endif TEST_F(DBTest, CompactOnFlush) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; do { Options options = CurrentOptions(options_override); options.disable_auto_compactions = true; CreateAndReopenWithCF({"pikachu"}, options); Put(1, "foo", "v1"); ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v1 ]"); // Write two new keys Put(1, "a", "begin"); Put(1, "z", "end"); Flush(1); // Case1: Delete followed by a put Delete(1, "foo"); Put(1, "foo", "v2"); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2, DEL, v1 ]"); // After the current memtable is flushed, the DEL should // have been removed ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2, v1 ]"); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v2 ]"); // Case 2: Delete followed by another delete Delete(1, "foo"); Delete(1, "foo"); ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL, DEL, v2 ]"); ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL, v2 ]"); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ(AllEntriesFor("foo", 1), "[ ]"); // Case 3: Put followed by a delete Put(1, "foo", "v3"); Delete(1, "foo"); ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL, v3 ]"); ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ DEL ]"); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ(AllEntriesFor("foo", 1), "[ ]"); // Case 4: Put followed by another Put Put(1, "foo", "v4"); Put(1, "foo", "v5"); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v5, v4 ]"); ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v5 ]"); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v5 ]"); // clear database Delete(1, "foo"); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ(AllEntriesFor("foo", 1), "[ ]"); // Case 5: Put followed by snapshot followed by another Put // Both puts should remain. Put(1, "foo", "v6"); const Snapshot* snapshot = db_->GetSnapshot(); Put(1, "foo", "v7"); ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v7, v6 ]"); db_->ReleaseSnapshot(snapshot); // clear database Delete(1, "foo"); dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ(AllEntriesFor("foo", 1), "[ ]"); // Case 5: snapshot followed by a put followed by another Put // Only the last put should remain. const Snapshot* snapshot1 = db_->GetSnapshot(); Put(1, "foo", "v8"); Put(1, "foo", "v9"); ASSERT_OK(Flush(1)); ASSERT_EQ(AllEntriesFor("foo", 1), "[ v9 ]"); db_->ReleaseSnapshot(snapshot1); } while (ChangeCompactOptions()); } namespace { std::vector ListSpecificFiles( Env* env, const std::string& path, const FileType expected_file_type) { std::vector files; std::vector file_numbers; env->GetChildren(path, &files); uint64_t number; FileType type; for (size_t i = 0; i < files.size(); ++i) { if (ParseFileName(files[i], &number, &type)) { if (type == expected_file_type) { file_numbers.push_back(number); } } } return file_numbers; } std::vector ListTableFiles(Env* env, const std::string& path) { return ListSpecificFiles(env, path, kTableFile); } } // namespace TEST_F(DBTest, FlushOneColumnFamily) { Options options = CurrentOptions(); CreateAndReopenWithCF({"pikachu", "ilya", "muromec", "dobrynia", "nikitich", "alyosha", "popovich"}, options); ASSERT_OK(Put(0, "Default", "Default")); ASSERT_OK(Put(1, "pikachu", "pikachu")); ASSERT_OK(Put(2, "ilya", "ilya")); ASSERT_OK(Put(3, "muromec", "muromec")); ASSERT_OK(Put(4, "dobrynia", "dobrynia")); ASSERT_OK(Put(5, "nikitich", "nikitich")); ASSERT_OK(Put(6, "alyosha", "alyosha")); ASSERT_OK(Put(7, "popovich", "popovich")); for (int i = 0; i < 8; ++i) { Flush(i); auto tables = ListTableFiles(env_, dbname_); ASSERT_EQ(tables.size(), i + 1U); } } #ifndef ROCKSDB_LITE // In https://reviews.facebook.net/D20661 we change // recovery behavior: previously for each log file each column family // memtable was flushed, even it was empty. Now it's changed: // we try to create the smallest number of table files by merging // updates from multiple logs TEST_F(DBTest, RecoverCheckFileAmountWithSmallWriteBuffer) { Options options = CurrentOptions(); options.write_buffer_size = 5000000; CreateAndReopenWithCF({"pikachu", "dobrynia", "nikitich"}, options); // Since we will reopen DB with smaller write_buffer_size, // each key will go to new SST file ASSERT_OK(Put(1, Key(10), DummyString(1000000))); ASSERT_OK(Put(1, Key(10), DummyString(1000000))); ASSERT_OK(Put(1, Key(10), DummyString(1000000))); ASSERT_OK(Put(1, Key(10), DummyString(1000000))); ASSERT_OK(Put(3, Key(10), DummyString(1))); // Make 'dobrynia' to be flushed and new WAL file to be created ASSERT_OK(Put(2, Key(10), DummyString(7500000))); ASSERT_OK(Put(2, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); { auto tables = ListTableFiles(env_, dbname_); ASSERT_EQ(tables.size(), static_cast(1)); // Make sure 'dobrynia' was flushed: check sst files amount ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(1)); } // New WAL file ASSERT_OK(Put(1, Key(1), DummyString(1))); ASSERT_OK(Put(1, Key(1), DummyString(1))); ASSERT_OK(Put(3, Key(10), DummyString(1))); ASSERT_OK(Put(3, Key(10), DummyString(1))); ASSERT_OK(Put(3, Key(10), DummyString(1))); options.write_buffer_size = 4096; options.arena_block_size = 4096; ReopenWithColumnFamilies({"default", "pikachu", "dobrynia", "nikitich"}, options); { // No inserts => default is empty ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(0)); // First 4 keys goes to separate SSTs + 1 more SST for 2 smaller keys ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(5)); // 1 SST for big key + 1 SST for small one ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(2)); // 1 SST for all keys ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(1)); } } // In https://reviews.facebook.net/D20661 we change // recovery behavior: previously for each log file each column family // memtable was flushed, even it wasn't empty. Now it's changed: // we try to create the smallest number of table files by merging // updates from multiple logs TEST_F(DBTest, RecoverCheckFileAmount) { Options options = CurrentOptions(); options.write_buffer_size = 100000; options.arena_block_size = 4 * 1024; CreateAndReopenWithCF({"pikachu", "dobrynia", "nikitich"}, options); ASSERT_OK(Put(0, Key(1), DummyString(1))); ASSERT_OK(Put(1, Key(1), DummyString(1))); ASSERT_OK(Put(2, Key(1), DummyString(1))); // Make 'nikitich' memtable to be flushed ASSERT_OK(Put(3, Key(10), DummyString(1002400))); ASSERT_OK(Put(3, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); // 4 memtable are not flushed, 1 sst file { auto tables = ListTableFiles(env_, dbname_); ASSERT_EQ(tables.size(), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(1)); } // Memtable for 'nikitich' has flushed, new WAL file has opened // 4 memtable still not flushed // Write to new WAL file ASSERT_OK(Put(0, Key(1), DummyString(1))); ASSERT_OK(Put(1, Key(1), DummyString(1))); ASSERT_OK(Put(2, Key(1), DummyString(1))); // Fill up 'nikitich' one more time ASSERT_OK(Put(3, Key(10), DummyString(1002400))); // make it flush ASSERT_OK(Put(3, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); // There are still 4 memtable not flushed, and 2 sst tables ASSERT_OK(Put(0, Key(1), DummyString(1))); ASSERT_OK(Put(1, Key(1), DummyString(1))); ASSERT_OK(Put(2, Key(1), DummyString(1))); { auto tables = ListTableFiles(env_, dbname_); ASSERT_EQ(tables.size(), static_cast(2)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(2)); } ReopenWithColumnFamilies({"default", "pikachu", "dobrynia", "nikitich"}, options); { std::vector table_files = ListTableFiles(env_, dbname_); // Check, that records for 'default', 'dobrynia' and 'pikachu' from // first, second and third WALs went to the same SST. // So, there is 6 SSTs: three for 'nikitich', one for 'default', one for // 'dobrynia', one for 'pikachu' ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(3)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(1)); } } TEST_F(DBTest, SharedWriteBuffer) { Options options = CurrentOptions(); options.db_write_buffer_size = 100000; // this is the real limit options.write_buffer_size = 500000; // this is never hit CreateAndReopenWithCF({"pikachu", "dobrynia", "nikitich"}, options); // Trigger a flush on CF "nikitich" ASSERT_OK(Put(0, Key(1), DummyString(1))); ASSERT_OK(Put(1, Key(1), DummyString(1))); ASSERT_OK(Put(3, Key(1), DummyString(90000))); ASSERT_OK(Put(2, Key(2), DummyString(20000))); ASSERT_OK(Put(2, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[0]); dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); { ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(1)); } // "dobrynia": 20KB // Flush 'dobrynia' ASSERT_OK(Put(3, Key(2), DummyString(40000))); ASSERT_OK(Put(2, Key(2), DummyString(70000))); ASSERT_OK(Put(0, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); { ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(1)); } // "nikitich" still has has data of 80KB // Inserting Data in "dobrynia" triggers "nikitich" flushing. ASSERT_OK(Put(3, Key(2), DummyString(40000))); ASSERT_OK(Put(2, Key(2), DummyString(40000))); ASSERT_OK(Put(0, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); { ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(2)); } // "dobrynia" still has 40KB ASSERT_OK(Put(1, Key(2), DummyString(20000))); ASSERT_OK(Put(0, Key(1), DummyString(10000))); ASSERT_OK(Put(0, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[0]); dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); // This should triggers no flush { ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(2)); } // "default": 10KB, "pikachu": 20KB, "dobrynia": 40KB ASSERT_OK(Put(1, Key(2), DummyString(40000))); ASSERT_OK(Put(0, Key(1), DummyString(1))); dbfull()->TEST_WaitForFlushMemTable(handles_[0]); dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForFlushMemTable(handles_[2]); dbfull()->TEST_WaitForFlushMemTable(handles_[3]); // This should triggers flush of "pikachu" { ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(0)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(2)); } // "default": 10KB, "dobrynia": 40KB // Some remaining writes so 'default', 'dobrynia' and 'nikitich' flush on // closure. ASSERT_OK(Put(3, Key(1), DummyString(1))); ReopenWithColumnFamilies({"default", "pikachu", "dobrynia", "nikitich"}, options); { ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"), static_cast(1)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"), static_cast(2)); ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"), static_cast(3)); } } #endif // ROCKSDB_LITE TEST_F(DBTest, PurgeInfoLogs) { Options options = CurrentOptions(); options.keep_log_file_num = 5; options.create_if_missing = true; for (int mode = 0; mode <= 1; mode++) { if (mode == 1) { options.db_log_dir = dbname_ + "_logs"; env_->CreateDirIfMissing(options.db_log_dir); } else { options.db_log_dir = ""; } for (int i = 0; i < 8; i++) { Reopen(options); } std::vector files; env_->GetChildren(options.db_log_dir.empty() ? dbname_ : options.db_log_dir, &files); int info_log_count = 0; for (std::string file : files) { if (file.find("LOG") != std::string::npos) { info_log_count++; } } ASSERT_EQ(5, info_log_count); Destroy(options); // For mode (1), test DestroyDB() to delete all the logs under DB dir. // For mode (2), no info log file should have been put under DB dir. std::vector db_files; env_->GetChildren(dbname_, &db_files); for (std::string file : db_files) { ASSERT_TRUE(file.find("LOG") == std::string::npos); } if (mode == 1) { // Cleaning up env_->GetChildren(options.db_log_dir, &files); for (std::string file : files) { env_->DeleteFile(options.db_log_dir + "/" + file); } env_->DeleteDir(options.db_log_dir); } } } TEST_F(DBTest, SyncMultipleLogs) { const uint64_t kNumBatches = 2; const int kBatchSize = 1000; Options options = CurrentOptions(); options.create_if_missing = true; options.write_buffer_size = 4096; Reopen(options); WriteBatch batch; WriteOptions wo; wo.sync = true; for (uint64_t b = 0; b < kNumBatches; b++) { batch.Clear(); for (int i = 0; i < kBatchSize; i++) { batch.Put(Key(i), DummyString(128)); } dbfull()->Write(wo, &batch); } ASSERT_OK(dbfull()->SyncWAL()); } #ifndef ROCKSDB_LITE // // Test WAL recovery for the various modes available // class RecoveryTestHelper { public: // Number of WAL files to generate static const int kWALFilesCount = 10; // Starting number for the WAL file name like 00010.log static const int kWALFileOffset = 10; // Keys to be written per WAL file static const int kKeysPerWALFile = 1024; // Size of the value static const int kValueSize = 10; // Create WAL files with values filled in static void FillData(DBTest* test, Options& options, const size_t wal_count, size_t& count) { DBOptions& db_options = options; count = 0; shared_ptr table_cache = NewLRUCache(50000, 16); EnvOptions env_options; WriteBuffer write_buffer(db_options.db_write_buffer_size); unique_ptr versions; unique_ptr wal_manager; WriteController write_controller; versions.reset(new VersionSet(test->dbname_, &db_options, env_options, table_cache.get(), &write_buffer, &write_controller)); wal_manager.reset(new WalManager(db_options, env_options)); std::unique_ptr current_log_writer; for (size_t j = kWALFileOffset; j < wal_count + kWALFileOffset; j++) { uint64_t current_log_number = j; std::string fname = LogFileName(test->dbname_, current_log_number); unique_ptr file; ASSERT_OK(db_options.env->NewWritableFile(fname, &file, env_options)); unique_ptr file_writer( new WritableFileWriter(std::move(file), env_options)); current_log_writer.reset( new log::Writer(std::move(file_writer), current_log_number, db_options.recycle_log_file_num > 0)); for (int i = 0; i < kKeysPerWALFile; i++) { std::string key = "key" + ToString(count++); std::string value = test->DummyString(kValueSize); assert(current_log_writer.get() != nullptr); uint64_t seq = versions->LastSequence() + 1; WriteBatch batch; batch.Put(key, value); WriteBatchInternal::SetSequence(&batch, seq); current_log_writer->AddRecord(WriteBatchInternal::Contents(&batch)); versions->SetLastSequence(seq); } } } // Recreate and fill the store with some data static size_t FillData(DBTest* test, Options& options) { options.create_if_missing = true; test->DestroyAndReopen(options); test->Close(); size_t count = 0; FillData(test, options, kWALFilesCount, count); return count; } // Read back all the keys we wrote and return the number of keys found static size_t GetData(DBTest* test) { size_t count = 0; for (size_t i = 0; i < kWALFilesCount * kKeysPerWALFile; i++) { if (test->Get("key" + ToString(i)) != "NOT_FOUND") { ++count; } } return count; } // Manuall corrupt the specified WAL static void CorruptWAL(DBTest* test, Options& options, const double off, const double len, const int wal_file_id, const bool trunc = false) { Env* env = options.env; std::string fname = LogFileName(test->dbname_, wal_file_id); uint64_t size; ASSERT_OK(env->GetFileSize(fname, &size)); ASSERT_GT(size, 0); #ifdef OS_WIN // Windows disk cache behaves differently. When we truncate // the original content is still in the cache due to the original // handle is still open. Generally, in Windows, one prohibits // shared access to files and it is not needed for WAL but we allow // it to induce corruption at various tests. test->Close(); #endif if (trunc) { ASSERT_EQ(0, truncate(fname.c_str(), static_cast(size * off))); } else { InduceCorruption(fname, static_cast(size * off), static_cast(size * len)); } } // Overwrite data with 'a' from offset for length len static void InduceCorruption(const std::string& filename, size_t offset, size_t len) { ASSERT_GT(len, 0U); int fd = open(filename.c_str(), O_RDWR); ASSERT_GT(fd, 0); ASSERT_EQ(offset, lseek(fd, static_cast(offset), SEEK_SET)); void* buf = alloca(len); memset(buf, 'a', len); ASSERT_EQ(len, write(fd, buf, static_cast(len))); close(fd); } }; // Test scope: // - We expect to open the data store when there is incomplete trailing writes // at the end of any of the logs // - We do not expect to open the data store for corruption TEST_F(DBTest, kTolerateCorruptedTailRecords) { const int jstart = RecoveryTestHelper::kWALFileOffset; const int jend = jstart + RecoveryTestHelper::kWALFilesCount; for (auto trunc : {true, false}) { /* Corruption style */ for (int i = 0; i < 4; i++) { /* Corruption offset position */ for (int j = jstart; j < jend; j++) { /* WAL file */ // Fill data for testing Options options = CurrentOptions(); const size_t row_count = RecoveryTestHelper::FillData(this, options); // test checksum failure or parsing RecoveryTestHelper::CorruptWAL(this, options, /*off=*/i * .3, /*len%=*/.1, /*wal=*/j, trunc); if (trunc) { options.wal_recovery_mode = WALRecoveryMode::kTolerateCorruptedTailRecords; options.create_if_missing = false; ASSERT_OK(TryReopen(options)); const size_t recovered_row_count = RecoveryTestHelper::GetData(this); ASSERT_TRUE(i == 0 || recovered_row_count > 0); ASSERT_LT(recovered_row_count, row_count); } else { options.wal_recovery_mode = WALRecoveryMode::kTolerateCorruptedTailRecords; ASSERT_NOK(TryReopen(options)); } } } } } // Test scope: // We don't expect the data store to be opened if there is any corruption // (leading, middle or trailing -- incomplete writes or corruption) TEST_F(DBTest, kAbsoluteConsistency) { const int jstart = RecoveryTestHelper::kWALFileOffset; const int jend = jstart + RecoveryTestHelper::kWALFilesCount; // Verify clean slate behavior Options options = CurrentOptions(); const size_t row_count = RecoveryTestHelper::FillData(this, options); options.wal_recovery_mode = WALRecoveryMode::kAbsoluteConsistency; options.create_if_missing = false; ASSERT_OK(TryReopen(options)); ASSERT_EQ(RecoveryTestHelper::GetData(this), row_count); for (auto trunc : {true, false}) { /* Corruption style */ for (int i = 0; i < 4; i++) { /* Corruption offset position */ if (trunc && i == 0) { continue; } for (int j = jstart; j < jend; j++) { /* wal files */ // fill with new date RecoveryTestHelper::FillData(this, options); // corrupt the wal RecoveryTestHelper::CorruptWAL(this, options, /*off=*/i * .3, /*len%=*/.1, j, trunc); // verify options.wal_recovery_mode = WALRecoveryMode::kAbsoluteConsistency; options.create_if_missing = false; ASSERT_NOK(TryReopen(options)); } } } } // Test scope: // - We expect to open data store under all circumstances // - We expect only data upto the point where the first error was encountered TEST_F(DBTest, kPointInTimeRecovery) { const int jstart = RecoveryTestHelper::kWALFileOffset; const int jend = jstart + RecoveryTestHelper::kWALFilesCount; const int maxkeys = RecoveryTestHelper::kWALFilesCount * RecoveryTestHelper::kKeysPerWALFile; for (auto trunc : {true, false}) { /* Corruption style */ for (int i = 0; i < 4; i++) { /* Offset of corruption */ for (int j = jstart; j < jend; j++) { /* WAL file */ // Fill data for testing Options options = CurrentOptions(); const size_t row_count = RecoveryTestHelper::FillData(this, options); // Corrupt the wal RecoveryTestHelper::CorruptWAL(this, options, /*off=*/i * .3, /*len%=*/.1, j, trunc); // Verify options.wal_recovery_mode = WALRecoveryMode::kPointInTimeRecovery; options.create_if_missing = false; ASSERT_OK(TryReopen(options)); // Probe data for invariants size_t recovered_row_count = RecoveryTestHelper::GetData(this); ASSERT_LT(recovered_row_count, row_count); bool expect_data = true; for (size_t k = 0; k < maxkeys; ++k) { bool found = Get("key" + ToString(i)) != "NOT_FOUND"; if (expect_data && !found) { expect_data = false; } ASSERT_EQ(found, expect_data); } const size_t min = RecoveryTestHelper::kKeysPerWALFile * (j - RecoveryTestHelper::kWALFileOffset); ASSERT_GE(recovered_row_count, min); if (!trunc && i != 0) { const size_t max = RecoveryTestHelper::kKeysPerWALFile * (j - RecoveryTestHelper::kWALFileOffset + 1); ASSERT_LE(recovered_row_count, max); } } } } } // Test scope: // - We expect to open the data store under all scenarios // - We expect to have recovered records past the corruption zone TEST_F(DBTest, kSkipAnyCorruptedRecords) { const int jstart = RecoveryTestHelper::kWALFileOffset; const int jend = jstart + RecoveryTestHelper::kWALFilesCount; for (auto trunc : {true, false}) { /* Corruption style */ for (int i = 0; i < 4; i++) { /* Corruption offset */ for (int j = jstart; j < jend; j++) { /* wal files */ // Fill data for testing Options options = CurrentOptions(); const size_t row_count = RecoveryTestHelper::FillData(this, options); // Corrupt the WAL RecoveryTestHelper::CorruptWAL(this, options, /*off=*/i * .3, /*len%=*/.1, j, trunc); // Verify behavior options.wal_recovery_mode = WALRecoveryMode::kSkipAnyCorruptedRecords; options.create_if_missing = false; ASSERT_OK(TryReopen(options)); // Probe data for invariants size_t recovered_row_count = RecoveryTestHelper::GetData(this); ASSERT_LT(recovered_row_count, row_count); if (!trunc) { ASSERT_TRUE(i != 0 || recovered_row_count > 0); } } } } } // Multi-threaded test: namespace { static const int kColumnFamilies = 10; static const int kNumThreads = 10; static const int kTestSeconds = 10; static const int kNumKeys = 1000; struct MTState { DBTest* test; std::atomic stop; std::atomic counter[kNumThreads]; std::atomic thread_done[kNumThreads]; }; struct MTThread { MTState* state; int id; }; static void MTThreadBody(void* arg) { MTThread* t = reinterpret_cast(arg); int id = t->id; DB* db = t->state->test->db_; int counter = 0; fprintf(stderr, "... starting thread %d\n", id); Random rnd(1000 + id); char valbuf[1500]; while (t->state->stop.load(std::memory_order_acquire) == false) { t->state->counter[id].store(counter, std::memory_order_release); int key = rnd.Uniform(kNumKeys); char keybuf[20]; snprintf(keybuf, sizeof(keybuf), "%016d", key); if (rnd.OneIn(2)) { // Write values of the form . // into each of the CFs // We add some padding for force compactions. int unique_id = rnd.Uniform(1000000); // Half of the time directly use WriteBatch. Half of the time use // WriteBatchWithIndex. if (rnd.OneIn(2)) { WriteBatch batch; for (int cf = 0; cf < kColumnFamilies; ++cf) { snprintf(valbuf, sizeof(valbuf), "%d.%d.%d.%d.%-1000d", key, id, static_cast(counter), cf, unique_id); batch.Put(t->state->test->handles_[cf], Slice(keybuf), Slice(valbuf)); } ASSERT_OK(db->Write(WriteOptions(), &batch)); } else { WriteBatchWithIndex batch(db->GetOptions().comparator); for (int cf = 0; cf < kColumnFamilies; ++cf) { snprintf(valbuf, sizeof(valbuf), "%d.%d.%d.%d.%-1000d", key, id, static_cast(counter), cf, unique_id); batch.Put(t->state->test->handles_[cf], Slice(keybuf), Slice(valbuf)); } ASSERT_OK(db->Write(WriteOptions(), batch.GetWriteBatch())); } } else { // Read a value and verify that it matches the pattern written above // and that writes to all column families were atomic (unique_id is the // same) std::vector keys(kColumnFamilies, Slice(keybuf)); std::vector values; std::vector statuses = db->MultiGet(ReadOptions(), t->state->test->handles_, keys, &values); Status s = statuses[0]; // all statuses have to be the same for (size_t i = 1; i < statuses.size(); ++i) { // they are either both ok or both not-found ASSERT_TRUE((s.ok() && statuses[i].ok()) || (s.IsNotFound() && statuses[i].IsNotFound())); } if (s.IsNotFound()) { // Key has not yet been written } else { // Check that the writer thread counter is >= the counter in the value ASSERT_OK(s); int unique_id = -1; for (int i = 0; i < kColumnFamilies; ++i) { int k, w, c, cf, u; ASSERT_EQ(5, sscanf(values[i].c_str(), "%d.%d.%d.%d.%d", &k, &w, &c, &cf, &u)) << values[i]; ASSERT_EQ(k, key); ASSERT_GE(w, 0); ASSERT_LT(w, kNumThreads); ASSERT_LE(c, t->state->counter[w].load(std::memory_order_acquire)); ASSERT_EQ(cf, i); if (i == 0) { unique_id = u; } else { // this checks that updates across column families happened // atomically -- all unique ids are the same ASSERT_EQ(u, unique_id); } } } } counter++; } t->state->thread_done[id].store(true, std::memory_order_release); fprintf(stderr, "... stopping thread %d after %d ops\n", id, int(counter)); } } // namespace class MultiThreadedDBTest : public DBTest, public ::testing::WithParamInterface { public: virtual void SetUp() override { option_config_ = GetParam(); } static std::vector GenerateOptionConfigs() { std::vector optionConfigs; for (int optionConfig = kDefault; optionConfig < kEnd; ++optionConfig) { // skip as HashCuckooRep does not support snapshot if (optionConfig != kHashCuckoo) { optionConfigs.push_back(optionConfig); } } return optionConfigs; } }; TEST_P(MultiThreadedDBTest, MultiThreaded) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; std::vector cfs; for (int i = 1; i < kColumnFamilies; ++i) { cfs.push_back(ToString(i)); } CreateAndReopenWithCF(cfs, CurrentOptions(options_override)); // Initialize state MTState mt; mt.test = this; mt.stop.store(false, std::memory_order_release); for (int id = 0; id < kNumThreads; id++) { mt.counter[id].store(0, std::memory_order_release); mt.thread_done[id].store(false, std::memory_order_release); } // Start threads MTThread thread[kNumThreads]; for (int id = 0; id < kNumThreads; id++) { thread[id].state = &mt; thread[id].id = id; env_->StartThread(MTThreadBody, &thread[id]); } // Let them run for a while env_->SleepForMicroseconds(kTestSeconds * 1000000); // Stop the threads and wait for them to finish mt.stop.store(true, std::memory_order_release); for (int id = 0; id < kNumThreads; id++) { while (mt.thread_done[id].load(std::memory_order_acquire) == false) { env_->SleepForMicroseconds(100000); } } } INSTANTIATE_TEST_CASE_P( MultiThreaded, MultiThreadedDBTest, ::testing::ValuesIn(MultiThreadedDBTest::GenerateOptionConfigs())); #endif // ROCKSDB_LITE // Group commit test: namespace { static const int kGCNumThreads = 4; static const int kGCNumKeys = 1000; struct GCThread { DB* db; int id; std::atomic done; }; static void GCThreadBody(void* arg) { GCThread* t = reinterpret_cast(arg); int id = t->id; DB* db = t->db; WriteOptions wo; for (int i = 0; i < kGCNumKeys; ++i) { std::string kv(ToString(i + id * kGCNumKeys)); ASSERT_OK(db->Put(wo, kv, kv)); } t->done = true; } } // namespace TEST_F(DBTest, GroupCommitTest) { do { Options options = CurrentOptions(); options.env = env_; env_->log_write_slowdown_.store(100); options.statistics = rocksdb::CreateDBStatistics(); Reopen(options); // Start threads GCThread thread[kGCNumThreads]; for (int id = 0; id < kGCNumThreads; id++) { thread[id].id = id; thread[id].db = db_; thread[id].done = false; env_->StartThread(GCThreadBody, &thread[id]); } for (int id = 0; id < kGCNumThreads; id++) { while (thread[id].done == false) { env_->SleepForMicroseconds(100000); } } env_->log_write_slowdown_.store(0); ASSERT_GT(TestGetTickerCount(options, WRITE_DONE_BY_OTHER), 0); std::vector expected_db; for (int i = 0; i < kGCNumThreads * kGCNumKeys; ++i) { expected_db.push_back(ToString(i)); } sort(expected_db.begin(), expected_db.end()); Iterator* itr = db_->NewIterator(ReadOptions()); itr->SeekToFirst(); for (auto x : expected_db) { ASSERT_TRUE(itr->Valid()); ASSERT_EQ(itr->key().ToString(), x); ASSERT_EQ(itr->value().ToString(), x); itr->Next(); } ASSERT_TRUE(!itr->Valid()); delete itr; HistogramData hist_data = {0, 0, 0, 0, 0}; options.statistics->histogramData(DB_WRITE, &hist_data); ASSERT_GT(hist_data.average, 0.0); } while (ChangeOptions(kSkipNoSeekToLast)); } namespace { typedef std::map KVMap; } class ModelDB: public DB { public: class ModelSnapshot : public Snapshot { public: KVMap map_; virtual SequenceNumber GetSequenceNumber() const override { // no need to call this assert(false); return 0; } }; explicit ModelDB(const Options& options) : options_(options) {} using DB::Put; virtual Status Put(const WriteOptions& o, ColumnFamilyHandle* cf, const Slice& k, const Slice& v) override { WriteBatch batch; batch.Put(cf, k, v); return Write(o, &batch); } using DB::Delete; virtual Status Delete(const WriteOptions& o, ColumnFamilyHandle* cf, const Slice& key) override { WriteBatch batch; batch.Delete(cf, key); return Write(o, &batch); } using DB::SingleDelete; virtual Status SingleDelete(const WriteOptions& o, ColumnFamilyHandle* cf, const Slice& key) override { WriteBatch batch; batch.SingleDelete(cf, key); return Write(o, &batch); } using DB::Merge; virtual Status Merge(const WriteOptions& o, ColumnFamilyHandle* cf, const Slice& k, const Slice& v) override { WriteBatch batch; batch.Merge(cf, k, v); return Write(o, &batch); } using DB::Get; virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* cf, const Slice& key, std::string* value) override { return Status::NotSupported(key); } using DB::MultiGet; virtual std::vector MultiGet( const ReadOptions& options, const std::vector& column_family, const std::vector& keys, std::vector* values) override { std::vector s(keys.size(), Status::NotSupported("Not implemented.")); return s; } #ifndef ROCKSDB_LITE using DB::AddFile; virtual Status AddFile(ColumnFamilyHandle* column_family, const ExternalSstFileInfo* file_path, bool move_file) override { return Status::NotSupported("Not implemented."); } virtual Status AddFile(ColumnFamilyHandle* column_family, const std::string& file_path, bool move_file) override { return Status::NotSupported("Not implemented."); } using DB::GetPropertiesOfAllTables; virtual Status GetPropertiesOfAllTables( ColumnFamilyHandle* column_family, TablePropertiesCollection* props) override { return Status(); } virtual Status GetPropertiesOfTablesInRange( ColumnFamilyHandle* column_family, const Range* range, std::size_t n, TablePropertiesCollection* props) override { return Status(); } #endif // ROCKSDB_LITE using DB::KeyMayExist; virtual bool KeyMayExist(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value, bool* value_found = nullptr) override { if (value_found != nullptr) { *value_found = false; } return true; // Not Supported directly } using DB::NewIterator; virtual Iterator* NewIterator(const ReadOptions& options, ColumnFamilyHandle* column_family) override { if (options.snapshot == nullptr) { KVMap* saved = new KVMap; *saved = map_; return new ModelIter(saved, true); } else { const KVMap* snapshot_state = &(reinterpret_cast(options.snapshot)->map_); return new ModelIter(snapshot_state, false); } } virtual Status NewIterators( const ReadOptions& options, const std::vector& column_family, std::vector* iterators) override { return Status::NotSupported("Not supported yet"); } virtual const Snapshot* GetSnapshot() override { ModelSnapshot* snapshot = new ModelSnapshot; snapshot->map_ = map_; return snapshot; } virtual void ReleaseSnapshot(const Snapshot* snapshot) override { delete reinterpret_cast(snapshot); } virtual Status Write(const WriteOptions& options, WriteBatch* batch) override { class Handler : public WriteBatch::Handler { public: KVMap* map_; virtual void Put(const Slice& key, const Slice& value) override { (*map_)[key.ToString()] = value.ToString(); } virtual void Merge(const Slice& key, const Slice& value) override { // ignore merge for now //(*map_)[key.ToString()] = value.ToString(); } virtual void Delete(const Slice& key) override { map_->erase(key.ToString()); } }; Handler handler; handler.map_ = &map_; return batch->Iterate(&handler); } using DB::GetProperty; virtual bool GetProperty(ColumnFamilyHandle* column_family, const Slice& property, std::string* value) override { return false; } using DB::GetIntProperty; virtual bool GetIntProperty(ColumnFamilyHandle* column_family, const Slice& property, uint64_t* value) override { return false; } using DB::GetAggregatedIntProperty; virtual bool GetAggregatedIntProperty(const Slice& property, uint64_t* value) override { return false; } using DB::GetApproximateSizes; virtual void GetApproximateSizes(ColumnFamilyHandle* column_family, const Range* range, int n, uint64_t* sizes, bool include_memtable) override { for (int i = 0; i < n; i++) { sizes[i] = 0; } } using DB::CompactRange; virtual Status CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* start, const Slice* end) override { return Status::NotSupported("Not supported operation."); } using DB::CompactFiles; virtual Status CompactFiles( const CompactionOptions& compact_options, ColumnFamilyHandle* column_family, const std::vector& input_file_names, const int output_level, const int output_path_id = -1) override { return Status::NotSupported("Not supported operation."); } Status PauseBackgroundWork() override { return Status::NotSupported("Not supported operation."); } Status ContinueBackgroundWork() override { return Status::NotSupported("Not supported operation."); } Status EnableAutoCompaction( const std::vector& column_family_handles) override { return Status::NotSupported("Not supported operation."); } using DB::NumberLevels; virtual int NumberLevels(ColumnFamilyHandle* column_family) override { return 1; } using DB::MaxMemCompactionLevel; virtual int MaxMemCompactionLevel( ColumnFamilyHandle* column_family) override { return 1; } using DB::Level0StopWriteTrigger; virtual int Level0StopWriteTrigger( ColumnFamilyHandle* column_family) override { return -1; } virtual const std::string& GetName() const override { return name_; } virtual Env* GetEnv() const override { return nullptr; } using DB::GetOptions; virtual const Options& GetOptions( ColumnFamilyHandle* column_family) const override { return options_; } using DB::GetDBOptions; virtual const DBOptions& GetDBOptions() const override { return options_; } using DB::Flush; virtual Status Flush(const rocksdb::FlushOptions& options, ColumnFamilyHandle* column_family) override { Status ret; return ret; } virtual Status SyncWAL() override { return Status::OK(); } #ifndef ROCKSDB_LITE virtual Status DisableFileDeletions() override { return Status::OK(); } virtual Status EnableFileDeletions(bool force) override { return Status::OK(); } virtual Status GetLiveFiles(std::vector&, uint64_t* size, bool flush_memtable = true) override { return Status::OK(); } virtual Status GetSortedWalFiles(VectorLogPtr& files) override { return Status::OK(); } virtual Status DeleteFile(std::string name) override { return Status::OK(); } virtual Status GetUpdatesSince( rocksdb::SequenceNumber, unique_ptr*, const TransactionLogIterator::ReadOptions& read_options = TransactionLogIterator::ReadOptions()) override { return Status::NotSupported("Not supported in Model DB"); } virtual void GetColumnFamilyMetaData( ColumnFamilyHandle* column_family, ColumnFamilyMetaData* metadata) override {} #endif // ROCKSDB_LITE virtual Status GetDbIdentity(std::string& identity) const override { return Status::OK(); } virtual SequenceNumber GetLatestSequenceNumber() const override { return 0; } virtual ColumnFamilyHandle* DefaultColumnFamily() const override { return nullptr; } private: class ModelIter: public Iterator { public: ModelIter(const KVMap* map, bool owned) : map_(map), owned_(owned), iter_(map_->end()) { } ~ModelIter() { if (owned_) delete map_; } virtual bool Valid() const override { return iter_ != map_->end(); } virtual void SeekToFirst() override { iter_ = map_->begin(); } virtual void SeekToLast() override { if (map_->empty()) { iter_ = map_->end(); } else { iter_ = map_->find(map_->rbegin()->first); } } virtual void Seek(const Slice& k) override { iter_ = map_->lower_bound(k.ToString()); } virtual void Next() override { ++iter_; } virtual void Prev() override { if (iter_ == map_->begin()) { iter_ = map_->end(); return; } --iter_; } virtual Slice key() const override { return iter_->first; } virtual Slice value() const override { return iter_->second; } virtual Status status() const override { return Status::OK(); } private: const KVMap* const map_; const bool owned_; // Do we own map_ KVMap::const_iterator iter_; }; const Options options_; KVMap map_; std::string name_ = ""; }; static std::string RandomKey(Random* rnd, int minimum = 0) { int len; do { len = (rnd->OneIn(3) ? 1 // Short sometimes to encourage collisions : (rnd->OneIn(100) ? rnd->Skewed(10) : rnd->Uniform(10))); } while (len < minimum); return test::RandomKey(rnd, len); } static bool CompareIterators(int step, DB* model, DB* db, const Snapshot* model_snap, const Snapshot* db_snap) { ReadOptions options; options.snapshot = model_snap; Iterator* miter = model->NewIterator(options); options.snapshot = db_snap; Iterator* dbiter = db->NewIterator(options); bool ok = true; int count = 0; for (miter->SeekToFirst(), dbiter->SeekToFirst(); ok && miter->Valid() && dbiter->Valid(); miter->Next(), dbiter->Next()) { count++; if (miter->key().compare(dbiter->key()) != 0) { fprintf(stderr, "step %d: Key mismatch: '%s' vs. '%s'\n", step, EscapeString(miter->key()).c_str(), EscapeString(dbiter->key()).c_str()); ok = false; break; } if (miter->value().compare(dbiter->value()) != 0) { fprintf(stderr, "step %d: Value mismatch for key '%s': '%s' vs. '%s'\n", step, EscapeString(miter->key()).c_str(), EscapeString(miter->value()).c_str(), EscapeString(miter->value()).c_str()); ok = false; } } if (ok) { if (miter->Valid() != dbiter->Valid()) { fprintf(stderr, "step %d: Mismatch at end of iterators: %d vs. %d\n", step, miter->Valid(), dbiter->Valid()); ok = false; } } delete miter; delete dbiter; return ok; } class DBTestRandomized : public DBTest, public ::testing::WithParamInterface { public: virtual void SetUp() override { option_config_ = GetParam(); } static std::vector GenerateOptionConfigs() { std::vector option_configs; // skip cuckoo hash as it does not support snapshot. for (int option_config = kDefault; option_config < kEnd; ++option_config) { if (!ShouldSkipOptions(option_config, kSkipDeletesFilterFirst | kSkipNoSeekToLast | kSkipHashCuckoo)) { option_configs.push_back(option_config); } } option_configs.push_back(kBlockBasedTableWithIndexRestartInterval); return option_configs; } }; INSTANTIATE_TEST_CASE_P( DBTestRandomized, DBTestRandomized, ::testing::ValuesIn(DBTestRandomized::GenerateOptionConfigs())); TEST_P(DBTestRandomized, Randomized) { anon::OptionsOverride options_override; options_override.skip_policy = kSkipNoSnapshot; Options options = CurrentOptions(options_override); DestroyAndReopen(options); Random rnd(test::RandomSeed() + GetParam()); ModelDB model(options); const int N = 10000; const Snapshot* model_snap = nullptr; const Snapshot* db_snap = nullptr; std::string k, v; for (int step = 0; step < N; step++) { // TODO(sanjay): Test Get() works int p = rnd.Uniform(100); int minimum = 0; if (option_config_ == kHashSkipList || option_config_ == kHashLinkList || option_config_ == kHashCuckoo || option_config_ == kPlainTableFirstBytePrefix || option_config_ == kBlockBasedTableWithWholeKeyHashIndex || option_config_ == kBlockBasedTableWithPrefixHashIndex) { minimum = 1; } if (p < 45) { // Put k = RandomKey(&rnd, minimum); v = RandomString(&rnd, rnd.OneIn(20) ? 100 + rnd.Uniform(100) : rnd.Uniform(8)); ASSERT_OK(model.Put(WriteOptions(), k, v)); ASSERT_OK(db_->Put(WriteOptions(), k, v)); } else if (p < 90) { // Delete k = RandomKey(&rnd, minimum); ASSERT_OK(model.Delete(WriteOptions(), k)); ASSERT_OK(db_->Delete(WriteOptions(), k)); } else { // Multi-element batch WriteBatch b; const int num = rnd.Uniform(8); for (int i = 0; i < num; i++) { if (i == 0 || !rnd.OneIn(10)) { k = RandomKey(&rnd, minimum); } else { // Periodically re-use the same key from the previous iter, so // we have multiple entries in the write batch for the same key } if (rnd.OneIn(2)) { v = RandomString(&rnd, rnd.Uniform(10)); b.Put(k, v); } else { b.Delete(k); } } ASSERT_OK(model.Write(WriteOptions(), &b)); ASSERT_OK(db_->Write(WriteOptions(), &b)); } if ((step % 100) == 0) { // For DB instances that use the hash index + block-based table, the // iterator will be invalid right when seeking a non-existent key, right // than return a key that is close to it. if (option_config_ != kBlockBasedTableWithWholeKeyHashIndex && option_config_ != kBlockBasedTableWithPrefixHashIndex) { ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr)); ASSERT_TRUE(CompareIterators(step, &model, db_, model_snap, db_snap)); } // Save a snapshot from each DB this time that we'll use next // time we compare things, to make sure the current state is // preserved with the snapshot if (model_snap != nullptr) model.ReleaseSnapshot(model_snap); if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap); Reopen(options); ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr)); model_snap = model.GetSnapshot(); db_snap = db_->GetSnapshot(); } } if (model_snap != nullptr) model.ReleaseSnapshot(model_snap); if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap); } TEST_F(DBTest, MultiGetSimple) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "k1", "v1")); ASSERT_OK(Put(1, "k2", "v2")); ASSERT_OK(Put(1, "k3", "v3")); ASSERT_OK(Put(1, "k4", "v4")); ASSERT_OK(Delete(1, "k4")); ASSERT_OK(Put(1, "k5", "v5")); ASSERT_OK(Delete(1, "no_key")); std::vector keys({"k1", "k2", "k3", "k4", "k5", "no_key"}); std::vector values(20, "Temporary data to be overwritten"); std::vector cfs(keys.size(), handles_[1]); std::vector s = db_->MultiGet(ReadOptions(), cfs, keys, &values); ASSERT_EQ(values.size(), keys.size()); ASSERT_EQ(values[0], "v1"); ASSERT_EQ(values[1], "v2"); ASSERT_EQ(values[2], "v3"); ASSERT_EQ(values[4], "v5"); ASSERT_OK(s[0]); ASSERT_OK(s[1]); ASSERT_OK(s[2]); ASSERT_TRUE(s[3].IsNotFound()); ASSERT_OK(s[4]); ASSERT_TRUE(s[5].IsNotFound()); } while (ChangeCompactOptions()); } TEST_F(DBTest, MultiGetEmpty) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); // Empty Key Set std::vector keys; std::vector values; std::vector cfs; std::vector s = db_->MultiGet(ReadOptions(), cfs, keys, &values); ASSERT_EQ(s.size(), 0U); // Empty Database, Empty Key Set Options options = CurrentOptions(); options.create_if_missing = true; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); s = db_->MultiGet(ReadOptions(), cfs, keys, &values); ASSERT_EQ(s.size(), 0U); // Empty Database, Search for Keys keys.resize(2); keys[0] = "a"; keys[1] = "b"; cfs.push_back(handles_[0]); cfs.push_back(handles_[1]); s = db_->MultiGet(ReadOptions(), cfs, keys, &values); ASSERT_EQ((int)s.size(), 2); ASSERT_TRUE(s[0].IsNotFound() && s[1].IsNotFound()); } while (ChangeCompactOptions()); } #ifndef ROCKSDB_LITE namespace { void PrefixScanInit(DBTest *dbtest) { char buf[100]; std::string keystr; const int small_range_sstfiles = 5; const int big_range_sstfiles = 5; // Generate 11 sst files with the following prefix ranges. // GROUP 0: [0,10] (level 1) // GROUP 1: [1,2], [2,3], [3,4], [4,5], [5, 6] (level 0) // GROUP 2: [0,6], [0,7], [0,8], [0,9], [0,10] (level 0) // // A seek with the previous API would do 11 random I/Os (to all the // files). With the new API and a prefix filter enabled, we should // only do 2 random I/O, to the 2 files containing the key. // GROUP 0 snprintf(buf, sizeof(buf), "%02d______:start", 0); keystr = std::string(buf); ASSERT_OK(dbtest->Put(keystr, keystr)); snprintf(buf, sizeof(buf), "%02d______:end", 10); keystr = std::string(buf); ASSERT_OK(dbtest->Put(keystr, keystr)); dbtest->Flush(); dbtest->dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); // move to level 1 // GROUP 1 for (int i = 1; i <= small_range_sstfiles; i++) { snprintf(buf, sizeof(buf), "%02d______:start", i); keystr = std::string(buf); ASSERT_OK(dbtest->Put(keystr, keystr)); snprintf(buf, sizeof(buf), "%02d______:end", i+1); keystr = std::string(buf); ASSERT_OK(dbtest->Put(keystr, keystr)); dbtest->Flush(); } // GROUP 2 for (int i = 1; i <= big_range_sstfiles; i++) { snprintf(buf, sizeof(buf), "%02d______:start", 0); keystr = std::string(buf); ASSERT_OK(dbtest->Put(keystr, keystr)); snprintf(buf, sizeof(buf), "%02d______:end", small_range_sstfiles+i+1); keystr = std::string(buf); ASSERT_OK(dbtest->Put(keystr, keystr)); dbtest->Flush(); } } } // namespace TEST_F(DBTest, PrefixScan) { XFUNC_TEST("", "dbtest_prefix", prefix_skip1, XFuncPoint::SetSkip, kSkipNoPrefix); while (ChangeFilterOptions()) { int count; Slice prefix; Slice key; char buf[100]; Iterator* iter; snprintf(buf, sizeof(buf), "03______:"); prefix = Slice(buf, 8); key = Slice(buf, 9); ASSERT_EQ(key.difference_offset(prefix), 8); ASSERT_EQ(prefix.difference_offset(key), 8); // db configs env_->count_random_reads_ = true; Options options = CurrentOptions(); options.env = env_; options.prefix_extractor.reset(NewFixedPrefixTransform(8)); options.disable_auto_compactions = true; options.max_background_compactions = 2; options.create_if_missing = true; options.memtable_factory.reset(NewHashSkipListRepFactory(16)); BlockBasedTableOptions table_options; table_options.no_block_cache = true; table_options.filter_policy.reset(NewBloomFilterPolicy(10)); table_options.whole_key_filtering = false; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); // 11 RAND I/Os DestroyAndReopen(options); PrefixScanInit(this); count = 0; env_->random_read_counter_.Reset(); iter = db_->NewIterator(ReadOptions()); for (iter->Seek(prefix); iter->Valid(); iter->Next()) { if (! iter->key().starts_with(prefix)) { break; } count++; } ASSERT_OK(iter->status()); delete iter; ASSERT_EQ(count, 2); ASSERT_EQ(env_->random_read_counter_.Read(), 2); Close(); } // end of while XFUNC_TEST("", "dbtest_prefix", prefix_skip1, XFuncPoint::SetSkip, 0); } #endif // ROCKSDB_LITE TEST_F(DBTest, BlockBasedTablePrefixIndexTest) { // create a DB with block prefix index BlockBasedTableOptions table_options; Options options = CurrentOptions(); table_options.index_type = BlockBasedTableOptions::kHashSearch; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.prefix_extractor.reset(NewFixedPrefixTransform(1)); Reopen(options); ASSERT_OK(Put("k1", "v1")); Flush(); ASSERT_OK(Put("k2", "v2")); // Reopen it without prefix extractor, make sure everything still works. // RocksDB should just fall back to the binary index. table_options.index_type = BlockBasedTableOptions::kBinarySearch; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.prefix_extractor.reset(); Reopen(options); ASSERT_EQ("v1", Get("k1")); ASSERT_EQ("v2", Get("k2")); } TEST_F(DBTest, ChecksumTest) { BlockBasedTableOptions table_options; Options options = CurrentOptions(); table_options.checksum = kCRC32c; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); Reopen(options); ASSERT_OK(Put("a", "b")); ASSERT_OK(Put("c", "d")); ASSERT_OK(Flush()); // table with crc checksum table_options.checksum = kxxHash; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); Reopen(options); ASSERT_OK(Put("e", "f")); ASSERT_OK(Put("g", "h")); ASSERT_OK(Flush()); // table with xxhash checksum table_options.checksum = kCRC32c; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); Reopen(options); ASSERT_EQ("b", Get("a")); ASSERT_EQ("d", Get("c")); ASSERT_EQ("f", Get("e")); ASSERT_EQ("h", Get("g")); table_options.checksum = kCRC32c; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); Reopen(options); ASSERT_EQ("b", Get("a")); ASSERT_EQ("d", Get("c")); ASSERT_EQ("f", Get("e")); ASSERT_EQ("h", Get("g")); } #ifndef ROCKSDB_LITE TEST_P(DBTestWithParam, FIFOCompactionTest) { for (int iter = 0; iter < 2; ++iter) { // first iteration -- auto compaction // second iteration -- manual compaction Options options; options.compaction_style = kCompactionStyleFIFO; options.write_buffer_size = 100 << 10; // 100KB options.arena_block_size = 4096; options.compaction_options_fifo.max_table_files_size = 500 << 10; // 500KB options.compression = kNoCompression; options.create_if_missing = true; options.max_subcompactions = max_subcompactions_; if (iter == 1) { options.disable_auto_compactions = true; } options = CurrentOptions(options); DestroyAndReopen(options); Random rnd(301); for (int i = 0; i < 6; ++i) { for (int j = 0; j < 110; ++j) { ASSERT_OK(Put(ToString(i * 100 + j), RandomString(&rnd, 980))); } // flush should happen here ASSERT_OK(dbfull()->TEST_WaitForFlushMemTable()); } if (iter == 0) { ASSERT_OK(dbfull()->TEST_WaitForCompact()); } else { CompactRangeOptions cro; cro.exclusive_manual_compaction = exclusive_manual_compaction_; ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr)); } // only 5 files should survive ASSERT_EQ(NumTableFilesAtLevel(0), 5); for (int i = 0; i < 50; ++i) { // these keys should be deleted in previous compaction ASSERT_EQ("NOT_FOUND", Get(ToString(i))); } } } #endif // ROCKSDB_LITE // verify that we correctly deprecated timeout_hint_us TEST_F(DBTest, SimpleWriteTimeoutTest) { WriteOptions write_opt; write_opt.timeout_hint_us = 0; ASSERT_OK(Put(Key(1), Key(1) + std::string(100, 'v'), write_opt)); write_opt.timeout_hint_us = 10; ASSERT_NOK(Put(Key(1), Key(1) + std::string(100, 'v'), write_opt)); } #ifndef ROCKSDB_LITE /* * This test is not reliable enough as it heavily depends on disk behavior. */ TEST_F(DBTest, RateLimitingTest) { Options options = CurrentOptions(); options.write_buffer_size = 1 << 20; // 1MB options.level0_file_num_compaction_trigger = 2; options.target_file_size_base = 1 << 20; // 1MB options.max_bytes_for_level_base = 4 << 20; // 4MB options.max_bytes_for_level_multiplier = 4; options.compression = kNoCompression; options.create_if_missing = true; options.env = env_; options.IncreaseParallelism(4); DestroyAndReopen(options); WriteOptions wo; wo.disableWAL = true; // # no rate limiting Random rnd(301); uint64_t start = env_->NowMicros(); // Write ~96M data for (int64_t i = 0; i < (96 << 10); ++i) { ASSERT_OK(Put(RandomString(&rnd, 32), RandomString(&rnd, (1 << 10) + 1), wo)); } uint64_t elapsed = env_->NowMicros() - start; double raw_rate = env_->bytes_written_ * 1000000.0 / elapsed; Close(); // # rate limiting with 0.7 x threshold options.rate_limiter.reset( NewGenericRateLimiter(static_cast(0.7 * raw_rate))); env_->bytes_written_ = 0; DestroyAndReopen(options); start = env_->NowMicros(); // Write ~96M data for (int64_t i = 0; i < (96 << 10); ++i) { ASSERT_OK(Put(RandomString(&rnd, 32), RandomString(&rnd, (1 << 10) + 1), wo)); } elapsed = env_->NowMicros() - start; Close(); ASSERT_EQ(options.rate_limiter->GetTotalBytesThrough(), env_->bytes_written_); double ratio = env_->bytes_written_ * 1000000 / elapsed / raw_rate; fprintf(stderr, "write rate ratio = %.2lf, expected 0.7\n", ratio); ASSERT_TRUE(ratio < 0.8); // # rate limiting with half of the raw_rate options.rate_limiter.reset( NewGenericRateLimiter(static_cast(raw_rate / 2))); env_->bytes_written_ = 0; DestroyAndReopen(options); start = env_->NowMicros(); // Write ~96M data for (int64_t i = 0; i < (96 << 10); ++i) { ASSERT_OK(Put(RandomString(&rnd, 32), RandomString(&rnd, (1 << 10) + 1), wo)); } elapsed = env_->NowMicros() - start; Close(); ASSERT_EQ(options.rate_limiter->GetTotalBytesThrough(), env_->bytes_written_); ratio = env_->bytes_written_ * 1000000 / elapsed / raw_rate; fprintf(stderr, "write rate ratio = %.2lf, expected 0.5\n", ratio); ASSERT_LT(ratio, 0.6); } TEST_F(DBTest, TableOptionsSanitizeTest) { Options options = CurrentOptions(); options.create_if_missing = true; DestroyAndReopen(options); ASSERT_EQ(db_->GetOptions().allow_mmap_reads, false); options.table_factory.reset(new PlainTableFactory()); options.prefix_extractor.reset(NewNoopTransform()); Destroy(options); ASSERT_TRUE(!TryReopen(options).IsNotSupported()); // Test for check of prefix_extractor when hash index is used for // block-based table BlockBasedTableOptions to; to.index_type = BlockBasedTableOptions::kHashSearch; options = CurrentOptions(); options.create_if_missing = true; options.table_factory.reset(NewBlockBasedTableFactory(to)); ASSERT_TRUE(TryReopen(options).IsInvalidArgument()); options.prefix_extractor.reset(NewFixedPrefixTransform(1)); ASSERT_OK(TryReopen(options)); } TEST_F(DBTest, ConcurrentMemtableNotSupported) { Options options = CurrentOptions(); options.allow_concurrent_memtable_write = true; options.soft_pending_compaction_bytes_limit = 0; options.hard_pending_compaction_bytes_limit = 100; options.create_if_missing = true; DestroyDB(dbname_, options); options.memtable_factory.reset(NewHashLinkListRepFactory(4, 0, 3, true, 4)); ASSERT_NOK(TryReopen(options)); options.memtable_factory.reset(new SkipListFactory); ASSERT_OK(TryReopen(options)); ColumnFamilyOptions cf_options(options); cf_options.memtable_factory.reset( NewHashLinkListRepFactory(4, 0, 3, true, 4)); ColumnFamilyHandle* handle; ASSERT_NOK(db_->CreateColumnFamily(cf_options, "name", &handle)); } #endif // ROCKSDB_LITE TEST_F(DBTest, SanitizeNumThreads) { for (int attempt = 0; attempt < 2; attempt++) { const size_t kTotalTasks = 8; test::SleepingBackgroundTask sleeping_tasks[kTotalTasks]; Options options = CurrentOptions(); if (attempt == 0) { options.max_background_compactions = 3; options.max_background_flushes = 2; } options.create_if_missing = true; DestroyAndReopen(options); for (size_t i = 0; i < kTotalTasks; i++) { // Insert 5 tasks to low priority queue and 5 tasks to high priority queue env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_tasks[i], (i < 4) ? Env::Priority::LOW : Env::Priority::HIGH); } // Wait 100 milliseconds for they are scheduled. env_->SleepForMicroseconds(100000); // pool size 3, total task 4. Queue size should be 1. ASSERT_EQ(1U, options.env->GetThreadPoolQueueLen(Env::Priority::LOW)); // pool size 2, total task 4. Queue size should be 2. ASSERT_EQ(2U, options.env->GetThreadPoolQueueLen(Env::Priority::HIGH)); for (size_t i = 0; i < kTotalTasks; i++) { sleeping_tasks[i].WakeUp(); sleeping_tasks[i].WaitUntilDone(); } ASSERT_OK(Put("abc", "def")); ASSERT_EQ("def", Get("abc")); Flush(); ASSERT_EQ("def", Get("abc")); } } TEST_F(DBTest, DBIteratorBoundTest) { Options options = CurrentOptions(); options.env = env_; options.create_if_missing = true; options.prefix_extractor = nullptr; DestroyAndReopen(options); ASSERT_OK(Put("a", "0")); ASSERT_OK(Put("foo", "bar")); ASSERT_OK(Put("foo1", "bar1")); ASSERT_OK(Put("g1", "0")); // testing basic case with no iterate_upper_bound and no prefix_extractor { ReadOptions ro; ro.iterate_upper_bound = nullptr; std::unique_ptr iter(db_->NewIterator(ro)); iter->Seek("foo"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("foo")), 0); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("foo1")), 0); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("g1")), 0); } // testing iterate_upper_bound and forward iterator // to make sure it stops at bound { ReadOptions ro; // iterate_upper_bound points beyond the last expected entry Slice prefix("foo2"); ro.iterate_upper_bound = &prefix; std::unique_ptr iter(db_->NewIterator(ro)); iter->Seek("foo"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("foo")), 0); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(("foo1")), 0); iter->Next(); // should stop here... ASSERT_TRUE(!iter->Valid()); } // Testing SeekToLast with iterate_upper_bound set { ReadOptions ro; Slice prefix("foo"); ro.iterate_upper_bound = &prefix; std::unique_ptr iter(db_->NewIterator(ro)); iter->SeekToLast(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("a")), 0); } // prefix is the first letter of the key options.prefix_extractor.reset(NewFixedPrefixTransform(1)); DestroyAndReopen(options); ASSERT_OK(Put("a", "0")); ASSERT_OK(Put("foo", "bar")); ASSERT_OK(Put("foo1", "bar1")); ASSERT_OK(Put("g1", "0")); // testing with iterate_upper_bound and prefix_extractor // Seek target and iterate_upper_bound are not is same prefix // This should be an error { ReadOptions ro; Slice upper_bound("g"); ro.iterate_upper_bound = &upper_bound; std::unique_ptr iter(db_->NewIterator(ro)); iter->Seek("foo"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("foo", iter->key().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("foo1", iter->key().ToString()); iter->Next(); ASSERT_TRUE(!iter->Valid()); } // testing that iterate_upper_bound prevents iterating over deleted items // if the bound has already reached { options.prefix_extractor = nullptr; DestroyAndReopen(options); ASSERT_OK(Put("a", "0")); ASSERT_OK(Put("b", "0")); ASSERT_OK(Put("b1", "0")); ASSERT_OK(Put("c", "0")); ASSERT_OK(Put("d", "0")); ASSERT_OK(Put("e", "0")); ASSERT_OK(Delete("c")); ASSERT_OK(Delete("d")); // base case with no bound ReadOptions ro; ro.iterate_upper_bound = nullptr; std::unique_ptr iter(db_->NewIterator(ro)); iter->Seek("b"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("b")), 0); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(("b1")), 0); perf_context.Reset(); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(static_cast(perf_context.internal_delete_skipped_count), 2); // now testing with iterate_bound Slice prefix("c"); ro.iterate_upper_bound = &prefix; iter.reset(db_->NewIterator(ro)); perf_context.Reset(); iter->Seek("b"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Slice("b")), 0); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(("b1")), 0); iter->Next(); // the iteration should stop as soon as the the bound key is reached // even though the key is deleted // hence internal_delete_skipped_count should be 0 ASSERT_TRUE(!iter->Valid()); ASSERT_EQ(static_cast(perf_context.internal_delete_skipped_count), 0); } } TEST_F(DBTest, WriteSingleThreadEntry) { std::vector threads; dbfull()->TEST_LockMutex(); auto w = dbfull()->TEST_BeginWrite(); threads.emplace_back([&] { Put("a", "b"); }); env_->SleepForMicroseconds(10000); threads.emplace_back([&] { Flush(); }); env_->SleepForMicroseconds(10000); dbfull()->TEST_UnlockMutex(); dbfull()->TEST_LockMutex(); dbfull()->TEST_EndWrite(w); dbfull()->TEST_UnlockMutex(); for (auto& t : threads) { t.join(); } } TEST_F(DBTest, DisableDataSyncTest) { env_->sync_counter_.store(0); // iter 0 -- no sync // iter 1 -- sync for (int iter = 0; iter < 2; ++iter) { Options options = CurrentOptions(); options.disableDataSync = iter == 0; options.create_if_missing = true; options.num_levels = 10; options.env = env_; Reopen(options); CreateAndReopenWithCF({"pikachu"}, options); MakeTables(10, "a", "z"); Compact("a", "z"); if (iter == 0) { ASSERT_EQ(env_->sync_counter_.load(), 0); } else { ASSERT_GT(env_->sync_counter_.load(), 0); } Destroy(options); } } #ifndef ROCKSDB_LITE TEST_F(DBTest, DynamicMemtableOptions) { const uint64_t k64KB = 1 << 16; const uint64_t k128KB = 1 << 17; const uint64_t k5KB = 5 * 1024; const int kNumPutsBeforeWaitForFlush = 64; Options options; options.env = env_; options.create_if_missing = true; options.compression = kNoCompression; options.max_background_compactions = 1; options.write_buffer_size = k64KB; options.arena_block_size = 16 * 1024; options.max_write_buffer_number = 2; // Don't trigger compact/slowdown/stop options.level0_file_num_compaction_trigger = 1024; options.level0_slowdown_writes_trigger = 1024; options.level0_stop_writes_trigger = 1024; DestroyAndReopen(options); auto gen_l0_kb = [this, kNumPutsBeforeWaitForFlush](int size) { Random rnd(301); for (int i = 0; i < size; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024))); // The following condition prevents a race condition between flush jobs // acquiring work and this thread filling up multiple memtables. Without // this, the flush might produce less files than expected because // multiple memtables are flushed into a single L0 file. This race // condition affects assertion (A). if (i % kNumPutsBeforeWaitForFlush == kNumPutsBeforeWaitForFlush - 1) { dbfull()->TEST_WaitForFlushMemTable(); } } dbfull()->TEST_WaitForFlushMemTable(); }; // Test write_buffer_size gen_l0_kb(64); ASSERT_EQ(NumTableFilesAtLevel(0), 1); ASSERT_LT(SizeAtLevel(0), k64KB + k5KB); ASSERT_GT(SizeAtLevel(0), k64KB - k5KB * 2); // Clean up L0 dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(NumTableFilesAtLevel(0), 0); // Increase buffer size ASSERT_OK(dbfull()->SetOptions({ {"write_buffer_size", "131072"}, })); // The existing memtable is still 64KB in size, after it becomes immutable, // the next memtable will be 128KB in size. Write 256KB total, we should // have a 64KB L0 file, a 128KB L0 file, and a memtable with 64KB data gen_l0_kb(256); ASSERT_EQ(NumTableFilesAtLevel(0), 2); // (A) ASSERT_LT(SizeAtLevel(0), k128KB + k64KB + 2 * k5KB); ASSERT_GT(SizeAtLevel(0), k128KB + k64KB - 4 * k5KB); // Test max_write_buffer_number // Block compaction thread, which will also block the flushes because // max_background_flushes == 0, so flushes are getting executed by the // compaction thread env_->SetBackgroundThreads(1, Env::LOW); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); // Start from scratch and disable compaction/flush. Flush can only happen // during compaction but trigger is pretty high options.max_background_flushes = 0; options.disable_auto_compactions = true; DestroyAndReopen(options); // Put until writes are stopped, bounded by 256 puts. We should see stop at // ~128KB int count = 0; Random rnd(301); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::DelayWrite:Wait", [&](void* arg) { sleeping_task_low.WakeUp(); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); while (!sleeping_task_low.WokenUp() && count < 256) { ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), WriteOptions())); count++; } ASSERT_GT(static_cast(count), 128 * 0.8); ASSERT_LT(static_cast(count), 128 * 1.2); sleeping_task_low.WaitUntilDone(); // Increase ASSERT_OK(dbfull()->SetOptions({ {"max_write_buffer_number", "8"}, })); // Clean up memtable and L0 dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); sleeping_task_low.Reset(); env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); count = 0; while (!sleeping_task_low.WokenUp() && count < 1024) { ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), WriteOptions())); count++; } // Windows fails this test. Will tune in the future and figure out // approp number #ifndef OS_WIN ASSERT_GT(static_cast(count), 512 * 0.8); ASSERT_LT(static_cast(count), 512 * 1.2); #endif sleeping_task_low.WaitUntilDone(); // Decrease ASSERT_OK(dbfull()->SetOptions({ {"max_write_buffer_number", "4"}, })); // Clean up memtable and L0 dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); sleeping_task_low.Reset(); env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); count = 0; while (!sleeping_task_low.WokenUp() && count < 1024) { ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), WriteOptions())); count++; } // Windows fails this test. Will tune in the future and figure out // approp number #ifndef OS_WIN ASSERT_GT(static_cast(count), 256 * 0.8); ASSERT_LT(static_cast(count), 266 * 1.2); #endif sleeping_task_low.WaitUntilDone(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } #endif // ROCKSDB_LITE #if ROCKSDB_USING_THREAD_STATUS namespace { void VerifyOperationCount(Env* env, ThreadStatus::OperationType op_type, int expected_count) { int op_count = 0; std::vector thread_list; ASSERT_OK(env->GetThreadList(&thread_list)); for (auto thread : thread_list) { if (thread.operation_type == op_type) { op_count++; } } ASSERT_EQ(op_count, expected_count); } } // namespace TEST_F(DBTest, GetThreadStatus) { Options options; options.env = env_; options.enable_thread_tracking = true; TryReopen(options); std::vector thread_list; Status s = env_->GetThreadList(&thread_list); for (int i = 0; i < 2; ++i) { // repeat the test with differet number of high / low priority threads const int kTestCount = 3; const unsigned int kHighPriCounts[kTestCount] = {3, 2, 5}; const unsigned int kLowPriCounts[kTestCount] = {10, 15, 3}; for (int test = 0; test < kTestCount; ++test) { // Change the number of threads in high / low priority pool. env_->SetBackgroundThreads(kHighPriCounts[test], Env::HIGH); env_->SetBackgroundThreads(kLowPriCounts[test], Env::LOW); // Wait to ensure the all threads has been registered env_->SleepForMicroseconds(100000); s = env_->GetThreadList(&thread_list); ASSERT_OK(s); unsigned int thread_type_counts[ThreadStatus::NUM_THREAD_TYPES]; memset(thread_type_counts, 0, sizeof(thread_type_counts)); for (auto thread : thread_list) { ASSERT_LT(thread.thread_type, ThreadStatus::NUM_THREAD_TYPES); thread_type_counts[thread.thread_type]++; } // Verify the total number of threades ASSERT_EQ( thread_type_counts[ThreadStatus::HIGH_PRIORITY] + thread_type_counts[ThreadStatus::LOW_PRIORITY], kHighPriCounts[test] + kLowPriCounts[test]); // Verify the number of high-priority threads ASSERT_EQ( thread_type_counts[ThreadStatus::HIGH_PRIORITY], kHighPriCounts[test]); // Verify the number of low-priority threads ASSERT_EQ( thread_type_counts[ThreadStatus::LOW_PRIORITY], kLowPriCounts[test]); } if (i == 0) { // repeat the test with multiple column families CreateAndReopenWithCF({"pikachu", "about-to-remove"}, options); env_->GetThreadStatusUpdater()->TEST_VerifyColumnFamilyInfoMap( handles_, true); } } db_->DropColumnFamily(handles_[2]); delete handles_[2]; handles_.erase(handles_.begin() + 2); env_->GetThreadStatusUpdater()->TEST_VerifyColumnFamilyInfoMap( handles_, true); Close(); env_->GetThreadStatusUpdater()->TEST_VerifyColumnFamilyInfoMap( handles_, true); } TEST_F(DBTest, DisableThreadStatus) { Options options; options.env = env_; options.enable_thread_tracking = false; TryReopen(options); CreateAndReopenWithCF({"pikachu", "about-to-remove"}, options); // Verify non of the column family info exists env_->GetThreadStatusUpdater()->TEST_VerifyColumnFamilyInfoMap( handles_, false); } TEST_F(DBTest, ThreadStatusFlush) { Options options; options.env = env_; options.write_buffer_size = 100000; // Small write buffer options.enable_thread_tracking = true; options = CurrentOptions(options); rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"FlushJob::FlushJob()", "DBTest::ThreadStatusFlush:1"}, {"DBTest::ThreadStatusFlush:2", "FlushJob::LogAndNotifyTableFileCreation()"}, }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); CreateAndReopenWithCF({"pikachu"}, options); VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 0); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_EQ("v1", Get(1, "foo")); VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 0); uint64_t num_running_flushes = 0; db_->GetIntProperty(DB::Properties::kNumRunningFlushes, &num_running_flushes); ASSERT_EQ(num_running_flushes, 0); Put(1, "k1", std::string(100000, 'x')); // Fill memtable Put(1, "k2", std::string(100000, 'y')); // Trigger flush // The first sync point is to make sure there's one flush job // running when we perform VerifyOperationCount(). TEST_SYNC_POINT("DBTest::ThreadStatusFlush:1"); VerifyOperationCount(env_, ThreadStatus::OP_FLUSH, 1); db_->GetIntProperty(DB::Properties::kNumRunningFlushes, &num_running_flushes); ASSERT_EQ(num_running_flushes, 1); // This second sync point is to ensure the flush job will not // be completed until we already perform VerifyOperationCount(). TEST_SYNC_POINT("DBTest::ThreadStatusFlush:2"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_P(DBTestWithParam, ThreadStatusSingleCompaction) { const int kTestKeySize = 16; const int kTestValueSize = 984; const int kEntrySize = kTestKeySize + kTestValueSize; const int kEntriesPerBuffer = 100; Options options; options.create_if_missing = true; options.write_buffer_size = kEntrySize * kEntriesPerBuffer; options.compaction_style = kCompactionStyleLevel; options.target_file_size_base = options.write_buffer_size; options.max_bytes_for_level_base = options.target_file_size_base * 2; options.max_bytes_for_level_multiplier = 2; options.compression = kNoCompression; options = CurrentOptions(options); options.env = env_; options.enable_thread_tracking = true; const int kNumL0Files = 4; options.level0_file_num_compaction_trigger = kNumL0Files; options.max_subcompactions = max_subcompactions_; rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"DBTest::ThreadStatusSingleCompaction:0", "DBImpl::BGWorkCompaction"}, {"CompactionJob::Run():Start", "DBTest::ThreadStatusSingleCompaction:1"}, {"DBTest::ThreadStatusSingleCompaction:2", "CompactionJob::Run():End"}, }); for (int tests = 0; tests < 2; ++tests) { DestroyAndReopen(options); rocksdb::SyncPoint::GetInstance()->ClearTrace(); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Random rnd(301); // The Put Phase. for (int file = 0; file < kNumL0Files; ++file) { for (int key = 0; key < kEntriesPerBuffer; ++key) { ASSERT_OK(Put(ToString(key + file * kEntriesPerBuffer), RandomString(&rnd, kTestValueSize))); } Flush(); } // This makes sure a compaction won't be scheduled until // we have done with the above Put Phase. uint64_t num_running_compactions = 0; db_->GetIntProperty(DB::Properties::kNumRunningCompactions, &num_running_compactions); ASSERT_EQ(num_running_compactions, 0); TEST_SYNC_POINT("DBTest::ThreadStatusSingleCompaction:0"); ASSERT_GE(NumTableFilesAtLevel(0), options.level0_file_num_compaction_trigger); // This makes sure at least one compaction is running. TEST_SYNC_POINT("DBTest::ThreadStatusSingleCompaction:1"); if (options.enable_thread_tracking) { // expecting one single L0 to L1 compaction VerifyOperationCount(env_, ThreadStatus::OP_COMPACTION, 1); } else { // If thread tracking is not enabled, compaction count should be 0. VerifyOperationCount(env_, ThreadStatus::OP_COMPACTION, 0); } db_->GetIntProperty(DB::Properties::kNumRunningCompactions, &num_running_compactions); ASSERT_EQ(num_running_compactions, 1); // TODO(yhchiang): adding assert to verify each compaction stage. TEST_SYNC_POINT("DBTest::ThreadStatusSingleCompaction:2"); // repeat the test with disabling thread tracking. options.enable_thread_tracking = false; rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } } TEST_P(DBTestWithParam, PreShutdownManualCompaction) { Options options = CurrentOptions(); options.max_background_flushes = 0; options.max_subcompactions = max_subcompactions_; CreateAndReopenWithCF({"pikachu"}, options); // iter - 0 with 7 levels // iter - 1 with 3 levels for (int iter = 0; iter < 2; ++iter) { MakeTables(3, "p", "q", 1); ASSERT_EQ("1,1,1", FilesPerLevel(1)); // Compaction range falls before files Compact(1, "", "c"); ASSERT_EQ("1,1,1", FilesPerLevel(1)); // Compaction range falls after files Compact(1, "r", "z"); ASSERT_EQ("1,1,1", FilesPerLevel(1)); // Compaction range overlaps files Compact(1, "p1", "p9"); ASSERT_EQ("0,0,1", FilesPerLevel(1)); // Populate a different range MakeTables(3, "c", "e", 1); ASSERT_EQ("1,1,2", FilesPerLevel(1)); // Compact just the new range Compact(1, "b", "f"); ASSERT_EQ("0,0,2", FilesPerLevel(1)); // Compact all MakeTables(1, "a", "z", 1); ASSERT_EQ("1,0,2", FilesPerLevel(1)); CancelAllBackgroundWork(db_); db_->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_EQ("1,0,2", FilesPerLevel(1)); if (iter == 0) { options = CurrentOptions(); options.max_background_flushes = 0; options.num_levels = 3; options.create_if_missing = true; DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); } } } TEST_F(DBTest, PreShutdownFlush) { Options options = CurrentOptions(); options.max_background_flushes = 0; CreateAndReopenWithCF({"pikachu"}, options); ASSERT_OK(Put(1, "key", "value")); CancelAllBackgroundWork(db_); Status s = db_->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr); ASSERT_TRUE(s.IsShutdownInProgress()); } TEST_P(DBTestWithParam, PreShutdownMultipleCompaction) { const int kTestKeySize = 16; const int kTestValueSize = 984; const int kEntrySize = kTestKeySize + kTestValueSize; const int kEntriesPerBuffer = 40; const int kNumL0Files = 4; const int kHighPriCount = 3; const int kLowPriCount = 5; env_->SetBackgroundThreads(kHighPriCount, Env::HIGH); env_->SetBackgroundThreads(kLowPriCount, Env::LOW); Options options; options.create_if_missing = true; options.write_buffer_size = kEntrySize * kEntriesPerBuffer; options.compaction_style = kCompactionStyleLevel; options.target_file_size_base = options.write_buffer_size; options.max_bytes_for_level_base = options.target_file_size_base * kNumL0Files; options.compression = kNoCompression; options = CurrentOptions(options); options.env = env_; options.enable_thread_tracking = true; options.level0_file_num_compaction_trigger = kNumL0Files; options.max_bytes_for_level_multiplier = 2; options.max_background_compactions = kLowPriCount; options.level0_stop_writes_trigger = 1 << 10; options.level0_slowdown_writes_trigger = 1 << 10; options.max_subcompactions = max_subcompactions_; TryReopen(options); Random rnd(301); std::vector thread_list; // Delay both flush and compaction rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"FlushJob::FlushJob()", "CompactionJob::Run():Start"}, {"CompactionJob::Run():Start", "DBTest::PreShutdownMultipleCompaction:Preshutdown"}, {"CompactionJob::Run():Start", "DBTest::PreShutdownMultipleCompaction:VerifyCompaction"}, {"DBTest::PreShutdownMultipleCompaction:Preshutdown", "CompactionJob::Run():End"}, {"CompactionJob::Run():End", "DBTest::PreShutdownMultipleCompaction:VerifyPreshutdown"}}); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // Make rocksdb busy int key = 0; // check how many threads are doing compaction using GetThreadList int operation_count[ThreadStatus::NUM_OP_TYPES] = {0}; for (int file = 0; file < 16 * kNumL0Files; ++file) { for (int k = 0; k < kEntriesPerBuffer; ++k) { ASSERT_OK(Put(ToString(key++), RandomString(&rnd, kTestValueSize))); } Status s = env_->GetThreadList(&thread_list); for (auto thread : thread_list) { operation_count[thread.operation_type]++; } // Speed up the test if (operation_count[ThreadStatus::OP_FLUSH] > 1 && operation_count[ThreadStatus::OP_COMPACTION] > 0.6 * options.max_background_compactions) { break; } if (file == 15 * kNumL0Files) { TEST_SYNC_POINT("DBTest::PreShutdownMultipleCompaction:Preshutdown"); } } TEST_SYNC_POINT("DBTest::PreShutdownMultipleCompaction:Preshutdown"); ASSERT_GE(operation_count[ThreadStatus::OP_COMPACTION], 1); CancelAllBackgroundWork(db_); TEST_SYNC_POINT("DBTest::PreShutdownMultipleCompaction:VerifyPreshutdown"); dbfull()->TEST_WaitForCompact(); // Record the number of compactions at a time. for (int i = 0; i < ThreadStatus::NUM_OP_TYPES; ++i) { operation_count[i] = 0; } Status s = env_->GetThreadList(&thread_list); for (auto thread : thread_list) { operation_count[thread.operation_type]++; } ASSERT_EQ(operation_count[ThreadStatus::OP_COMPACTION], 0); } TEST_P(DBTestWithParam, PreShutdownCompactionMiddle) { const int kTestKeySize = 16; const int kTestValueSize = 984; const int kEntrySize = kTestKeySize + kTestValueSize; const int kEntriesPerBuffer = 40; const int kNumL0Files = 4; const int kHighPriCount = 3; const int kLowPriCount = 5; env_->SetBackgroundThreads(kHighPriCount, Env::HIGH); env_->SetBackgroundThreads(kLowPriCount, Env::LOW); Options options; options.create_if_missing = true; options.write_buffer_size = kEntrySize * kEntriesPerBuffer; options.compaction_style = kCompactionStyleLevel; options.target_file_size_base = options.write_buffer_size; options.max_bytes_for_level_base = options.target_file_size_base * kNumL0Files; options.compression = kNoCompression; options = CurrentOptions(options); options.env = env_; options.enable_thread_tracking = true; options.level0_file_num_compaction_trigger = kNumL0Files; options.max_bytes_for_level_multiplier = 2; options.max_background_compactions = kLowPriCount; options.level0_stop_writes_trigger = 1 << 10; options.level0_slowdown_writes_trigger = 1 << 10; options.max_subcompactions = max_subcompactions_; TryReopen(options); Random rnd(301); std::vector thread_list; // Delay both flush and compaction rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"DBTest::PreShutdownCompactionMiddle:Preshutdown", "CompactionJob::Run():Inprogress"}, {"CompactionJob::Run():Start", "DBTest::PreShutdownCompactionMiddle:VerifyCompaction"}, {"CompactionJob::Run():Inprogress", "CompactionJob::Run():End"}, {"CompactionJob::Run():End", "DBTest::PreShutdownCompactionMiddle:VerifyPreshutdown"}}); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); // Make rocksdb busy int key = 0; // check how many threads are doing compaction using GetThreadList int operation_count[ThreadStatus::NUM_OP_TYPES] = {0}; for (int file = 0; file < 16 * kNumL0Files; ++file) { for (int k = 0; k < kEntriesPerBuffer; ++k) { ASSERT_OK(Put(ToString(key++), RandomString(&rnd, kTestValueSize))); } Status s = env_->GetThreadList(&thread_list); for (auto thread : thread_list) { operation_count[thread.operation_type]++; } // Speed up the test if (operation_count[ThreadStatus::OP_FLUSH] > 1 && operation_count[ThreadStatus::OP_COMPACTION] > 0.6 * options.max_background_compactions) { break; } if (file == 15 * kNumL0Files) { TEST_SYNC_POINT("DBTest::PreShutdownCompactionMiddle:VerifyCompaction"); } } ASSERT_GE(operation_count[ThreadStatus::OP_COMPACTION], 1); CancelAllBackgroundWork(db_); TEST_SYNC_POINT("DBTest::PreShutdownCompactionMiddle:Preshutdown"); TEST_SYNC_POINT("DBTest::PreShutdownCompactionMiddle:VerifyPreshutdown"); dbfull()->TEST_WaitForCompact(); // Record the number of compactions at a time. for (int i = 0; i < ThreadStatus::NUM_OP_TYPES; ++i) { operation_count[i] = 0; } Status s = env_->GetThreadList(&thread_list); for (auto thread : thread_list) { operation_count[thread.operation_type]++; } ASSERT_EQ(operation_count[ThreadStatus::OP_COMPACTION], 0); } #endif // ROCKSDB_USING_THREAD_STATUS #ifndef ROCKSDB_LITE TEST_F(DBTest, FlushOnDestroy) { WriteOptions wo; wo.disableWAL = true; ASSERT_OK(Put("foo", "v1", wo)); CancelAllBackgroundWork(db_); } namespace { class OnFileDeletionListener : public EventListener { public: OnFileDeletionListener() : matched_count_(0), expected_file_name_("") {} void SetExpectedFileName( const std::string file_name) { expected_file_name_ = file_name; } void VerifyMatchedCount(size_t expected_value) { ASSERT_EQ(matched_count_, expected_value); } void OnTableFileDeleted( const TableFileDeletionInfo& info) override { if (expected_file_name_ != "") { ASSERT_EQ(expected_file_name_, info.file_path); expected_file_name_ = ""; matched_count_++; } } private: size_t matched_count_; std::string expected_file_name_; }; } // namespace TEST_F(DBTest, DynamicLevelCompressionPerLevel) { if (!Snappy_Supported()) { return; } const int kNKeys = 120; int keys[kNKeys]; for (int i = 0; i < kNKeys; i++) { keys[i] = i; } std::random_shuffle(std::begin(keys), std::end(keys)); Random rnd(301); Options options; options.create_if_missing = true; options.db_write_buffer_size = 20480; options.write_buffer_size = 20480; options.max_write_buffer_number = 2; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 2; options.target_file_size_base = 2048; options.level_compaction_dynamic_level_bytes = true; options.max_bytes_for_level_base = 102400; options.max_bytes_for_level_multiplier = 4; options.max_background_compactions = 1; options.num_levels = 5; options.compression_per_level.resize(3); options.compression_per_level[0] = kNoCompression; options.compression_per_level[1] = kNoCompression; options.compression_per_level[2] = kSnappyCompression; OnFileDeletionListener* listener = new OnFileDeletionListener(); options.listeners.emplace_back(listener); DestroyAndReopen(options); // Insert more than 80K. L4 should be base level. Neither L0 nor L4 should // be compressed, so total data size should be more than 80K. for (int i = 0; i < 20; i++) { ASSERT_OK(Put(Key(keys[i]), CompressibleString(&rnd, 4000))); } Flush(); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(1), 0); ASSERT_EQ(NumTableFilesAtLevel(2), 0); ASSERT_EQ(NumTableFilesAtLevel(3), 0); ASSERT_GT(SizeAtLevel(0) + SizeAtLevel(4), 20U * 4000U); // Insert 400KB. Some data will be compressed for (int i = 21; i < 120; i++) { ASSERT_OK(Put(Key(keys[i]), CompressibleString(&rnd, 4000))); } Flush(); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(1), 0); ASSERT_EQ(NumTableFilesAtLevel(2), 0); ASSERT_LT(SizeAtLevel(0) + SizeAtLevel(3) + SizeAtLevel(4), 120U * 4000U); // Make sure data in files in L3 is not compacted by removing all files // in L4 and calculate number of rows ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "true"}, })); ColumnFamilyMetaData cf_meta; db_->GetColumnFamilyMetaData(&cf_meta); for (auto file : cf_meta.levels[4].files) { listener->SetExpectedFileName(dbname_ + file.name); ASSERT_OK(dbfull()->DeleteFile(file.name)); } listener->VerifyMatchedCount(cf_meta.levels[4].files.size()); int num_keys = 0; std::unique_ptr iter(db_->NewIterator(ReadOptions())); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { num_keys++; } ASSERT_OK(iter->status()); ASSERT_GT(SizeAtLevel(0) + SizeAtLevel(3), num_keys * 4000U); } TEST_F(DBTest, DynamicLevelCompressionPerLevel2) { if (!Snappy_Supported() || !LZ4_Supported() || !Zlib_Supported()) { return; } const int kNKeys = 500; int keys[kNKeys]; for (int i = 0; i < kNKeys; i++) { keys[i] = i; } std::random_shuffle(std::begin(keys), std::end(keys)); Random rnd(301); Options options; options.create_if_missing = true; options.db_write_buffer_size = 6000; options.write_buffer_size = 6000; options.max_write_buffer_number = 2; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 2; options.soft_pending_compaction_bytes_limit = 1024 * 1024; // Use file size to distinguish levels // L1: 10, L2: 20, L3 40, L4 80 // L0 is less than 30 options.target_file_size_base = 10; options.target_file_size_multiplier = 2; options.level_compaction_dynamic_level_bytes = true; options.max_bytes_for_level_base = 200; options.max_bytes_for_level_multiplier = 8; options.max_background_compactions = 1; options.num_levels = 5; std::shared_ptr mtf(new mock::MockTableFactory); options.table_factory = mtf; options.compression_per_level.resize(3); options.compression_per_level[0] = kNoCompression; options.compression_per_level[1] = kLZ4Compression; options.compression_per_level[2] = kZlibCompression; DestroyAndReopen(options); // When base level is L4, L4 is LZ4. std::atomic num_zlib(0); std::atomic num_lz4(0); std::atomic num_no(0); rocksdb::SyncPoint::GetInstance()->SetCallBack( "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) { Compaction* compaction = reinterpret_cast(arg); if (compaction->output_level() == 4) { ASSERT_TRUE(compaction->output_compression() == kLZ4Compression); num_lz4.fetch_add(1); } }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "FlushJob::WriteLevel0Table:output_compression", [&](void* arg) { auto* compression = reinterpret_cast(arg); ASSERT_TRUE(*compression == kNoCompression); num_no.fetch_add(1); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(keys[i]), RandomString(&rnd, 200))); if (i % 25 == 0) { dbfull()->TEST_WaitForFlushMemTable(); } } Flush(); dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); ASSERT_EQ(NumTableFilesAtLevel(1), 0); ASSERT_EQ(NumTableFilesAtLevel(2), 0); ASSERT_EQ(NumTableFilesAtLevel(3), 0); ASSERT_GT(NumTableFilesAtLevel(4), 0); ASSERT_GT(num_no.load(), 2); ASSERT_GT(num_lz4.load(), 0); int prev_num_files_l4 = NumTableFilesAtLevel(4); // After base level turn L4->L3, L3 becomes LZ4 and L4 becomes Zlib num_lz4.store(0); num_no.store(0); rocksdb::SyncPoint::GetInstance()->SetCallBack( "LevelCompactionPicker::PickCompaction:Return", [&](void* arg) { Compaction* compaction = reinterpret_cast(arg); if (compaction->output_level() == 4 && compaction->start_level() == 3) { ASSERT_TRUE(compaction->output_compression() == kZlibCompression); num_zlib.fetch_add(1); } else { ASSERT_TRUE(compaction->output_compression() == kLZ4Compression); num_lz4.fetch_add(1); } }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "FlushJob::WriteLevel0Table:output_compression", [&](void* arg) { auto* compression = reinterpret_cast(arg); ASSERT_TRUE(*compression == kNoCompression); num_no.fetch_add(1); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); for (int i = 101; i < 500; i++) { ASSERT_OK(Put(Key(keys[i]), RandomString(&rnd, 200))); if (i % 100 == 99) { Flush(); dbfull()->TEST_WaitForCompact(); } } rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); ASSERT_EQ(NumTableFilesAtLevel(1), 0); ASSERT_EQ(NumTableFilesAtLevel(2), 0); ASSERT_GT(NumTableFilesAtLevel(3), 0); ASSERT_GT(NumTableFilesAtLevel(4), prev_num_files_l4); ASSERT_GT(num_no.load(), 2); ASSERT_GT(num_lz4.load(), 0); ASSERT_GT(num_zlib.load(), 0); } TEST_F(DBTest, DynamicCompactionOptions) { // minimum write buffer size is enforced at 64KB const uint64_t k32KB = 1 << 15; const uint64_t k64KB = 1 << 16; const uint64_t k128KB = 1 << 17; const uint64_t k1MB = 1 << 20; const uint64_t k4KB = 1 << 12; Options options; options.env = env_; options.create_if_missing = true; options.compression = kNoCompression; options.soft_pending_compaction_bytes_limit = 1024 * 1024; options.write_buffer_size = k64KB; options.arena_block_size = 4 * k4KB; options.max_write_buffer_number = 2; // Compaction related options options.level0_file_num_compaction_trigger = 3; options.level0_slowdown_writes_trigger = 4; options.level0_stop_writes_trigger = 8; options.max_grandparent_overlap_factor = 10; options.expanded_compaction_factor = 25; options.source_compaction_factor = 1; options.target_file_size_base = k64KB; options.target_file_size_multiplier = 1; options.max_bytes_for_level_base = k128KB; options.max_bytes_for_level_multiplier = 4; // Block flush thread and disable compaction thread env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); DestroyAndReopen(options); auto gen_l0_kb = [this](int start, int size, int stride) { Random rnd(301); for (int i = 0; i < size; i++) { ASSERT_OK(Put(Key(start + stride * i), RandomString(&rnd, 1024))); } dbfull()->TEST_WaitForFlushMemTable(); }; // Write 3 files that have the same key range. // Since level0_file_num_compaction_trigger is 3, compaction should be // triggered. The compaction should result in one L1 file gen_l0_kb(0, 64, 1); ASSERT_EQ(NumTableFilesAtLevel(0), 1); gen_l0_kb(0, 64, 1); ASSERT_EQ(NumTableFilesAtLevel(0), 2); gen_l0_kb(0, 64, 1); dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,1", FilesPerLevel()); std::vector metadata; db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(1U, metadata.size()); ASSERT_LE(metadata[0].size, k64KB + k4KB); ASSERT_GE(metadata[0].size, k64KB - k4KB); // Test compaction trigger and target_file_size_base // Reduce compaction trigger to 2, and reduce L1 file size to 32KB. // Writing to 64KB L0 files should trigger a compaction. Since these // 2 L0 files have the same key range, compaction merge them and should // result in 2 32KB L1 files. ASSERT_OK(dbfull()->SetOptions({ {"level0_file_num_compaction_trigger", "2"}, {"target_file_size_base", ToString(k32KB) } })); gen_l0_kb(0, 64, 1); ASSERT_EQ("1,1", FilesPerLevel()); gen_l0_kb(0, 64, 1); dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,2", FilesPerLevel()); metadata.clear(); db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(2U, metadata.size()); ASSERT_LE(metadata[0].size, k32KB + k4KB); ASSERT_GE(metadata[0].size, k32KB - k4KB); ASSERT_LE(metadata[1].size, k32KB + k4KB); ASSERT_GE(metadata[1].size, k32KB - k4KB); // Test max_bytes_for_level_base // Increase level base size to 256KB and write enough data that will // fill L1 and L2. L1 size should be around 256KB while L2 size should be // around 256KB x 4. ASSERT_OK(dbfull()->SetOptions({ {"max_bytes_for_level_base", ToString(k1MB) } })); // writing 96 x 64KB => 6 * 1024KB // (L1 + L2) = (1 + 4) * 1024KB for (int i = 0; i < 96; ++i) { gen_l0_kb(i, 64, 96); } dbfull()->TEST_WaitForCompact(); ASSERT_GT(SizeAtLevel(1), k1MB / 2); ASSERT_LT(SizeAtLevel(1), k1MB + k1MB / 2); // Within (0.5, 1.5) of 4MB. ASSERT_GT(SizeAtLevel(2), 2 * k1MB); ASSERT_LT(SizeAtLevel(2), 6 * k1MB); // Test max_bytes_for_level_multiplier and // max_bytes_for_level_base. Now, reduce both mulitplier and level base, // After filling enough data that can fit in L1 - L3, we should see L1 size // reduces to 128KB from 256KB which was asserted previously. Same for L2. ASSERT_OK(dbfull()->SetOptions({ {"max_bytes_for_level_multiplier", "2"}, {"max_bytes_for_level_base", ToString(k128KB) } })); // writing 20 x 64KB = 10 x 128KB // (L1 + L2 + L3) = (1 + 2 + 4) * 128KB for (int i = 0; i < 20; ++i) { gen_l0_kb(i, 64, 32); } dbfull()->TEST_WaitForCompact(); uint64_t total_size = SizeAtLevel(1) + SizeAtLevel(2) + SizeAtLevel(3); ASSERT_TRUE(total_size < k128KB * 7 * 1.5); // Test level0_stop_writes_trigger. // Clean up memtable and L0. Block compaction threads. If continue to write // and flush memtables. We should see put stop after 8 memtable flushes // since level0_stop_writes_trigger = 8 dbfull()->TEST_FlushMemTable(true); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); // Block compaction test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); sleeping_task_low.WaitUntilSleeping(); ASSERT_EQ(NumTableFilesAtLevel(0), 0); int count = 0; Random rnd(301); WriteOptions wo; while (count < 64) { ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), wo)); dbfull()->TEST_FlushMemTable(true); count++; if (dbfull()->TEST_write_controler().IsStopped()) { sleeping_task_low.WakeUp(); break; } } // Stop trigger = 8 ASSERT_EQ(count, 8); // Unblock sleeping_task_low.WaitUntilDone(); // Now reduce level0_stop_writes_trigger to 6. Clear up memtables and L0. // Block compaction thread again. Perform the put and memtable flushes // until we see the stop after 6 memtable flushes. ASSERT_OK(dbfull()->SetOptions({ {"level0_stop_writes_trigger", "6"} })); dbfull()->TEST_FlushMemTable(true); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(NumTableFilesAtLevel(0), 0); // Block compaction again sleeping_task_low.Reset(); env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); sleeping_task_low.WaitUntilSleeping(); count = 0; while (count < 64) { ASSERT_OK(Put(Key(count), RandomString(&rnd, 1024), wo)); dbfull()->TEST_FlushMemTable(true); count++; if (dbfull()->TEST_write_controler().IsStopped()) { sleeping_task_low.WakeUp(); break; } } ASSERT_EQ(count, 6); // Unblock sleeping_task_low.WaitUntilDone(); // Test disable_auto_compactions // Compaction thread is unblocked but auto compaction is disabled. Write // 4 L0 files and compaction should be triggered. If auto compaction is // disabled, then TEST_WaitForCompact will be waiting for nothing. Number of // L0 files do not change after the call. ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "true"} })); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(NumTableFilesAtLevel(0), 0); for (int i = 0; i < 4; ++i) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024))); // Wait for compaction so that put won't stop dbfull()->TEST_FlushMemTable(true); } dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(0), 4); // Enable auto compaction and perform the same test, # of L0 files should be // reduced after compaction. ASSERT_OK(dbfull()->SetOptions({ {"disable_auto_compactions", "false"} })); dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(NumTableFilesAtLevel(0), 0); for (int i = 0; i < 4; ++i) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024))); // Wait for compaction so that put won't stop dbfull()->TEST_FlushMemTable(true); } dbfull()->TEST_WaitForCompact(); ASSERT_LT(NumTableFilesAtLevel(0), 4); } #endif // ROCKSDB_LITE TEST_F(DBTest, FileCreationRandomFailure) { Options options; options.env = env_; options.create_if_missing = true; options.write_buffer_size = 100000; // Small write buffer options.target_file_size_base = 200000; options.max_bytes_for_level_base = 1000000; options.max_bytes_for_level_multiplier = 2; DestroyAndReopen(options); Random rnd(301); const int kCDTKeysPerBuffer = 4; const int kTestSize = kCDTKeysPerBuffer * 4096; const int kTotalIteration = 100; // the second half of the test involves in random failure // of file creation. const int kRandomFailureTest = kTotalIteration / 2; std::vector values; for (int i = 0; i < kTestSize; ++i) { values.push_back("NOT_FOUND"); } for (int j = 0; j < kTotalIteration; ++j) { if (j == kRandomFailureTest) { env_->non_writeable_rate_.store(90); } for (int k = 0; k < kTestSize; ++k) { // here we expect some of the Put fails. std::string value = RandomString(&rnd, 100); Status s = Put(Key(k), Slice(value)); if (s.ok()) { // update the latest successful put values[k] = value; } // But everything before we simulate the failure-test should succeed. if (j < kRandomFailureTest) { ASSERT_OK(s); } } } // If rocksdb does not do the correct job, internal assert will fail here. dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); // verify we have the latest successful update for (int k = 0; k < kTestSize; ++k) { auto v = Get(Key(k)); ASSERT_EQ(v, values[k]); } // reopen and reverify we have the latest successful update env_->non_writeable_rate_.store(0); Reopen(options); for (int k = 0; k < kTestSize; ++k) { auto v = Get(Key(k)); ASSERT_EQ(v, values[k]); } } #ifndef ROCKSDB_LITE TEST_F(DBTest, DynamicMiscOptions) { // Test max_sequential_skip_in_iterations Options options; options.env = env_; options.create_if_missing = true; options.max_sequential_skip_in_iterations = 16; options.compression = kNoCompression; options.statistics = rocksdb::CreateDBStatistics(); DestroyAndReopen(options); auto assert_reseek_count = [this, &options](int key_start, int num_reseek) { int key0 = key_start; int key1 = key_start + 1; int key2 = key_start + 2; Random rnd(301); ASSERT_OK(Put(Key(key0), RandomString(&rnd, 8))); for (int i = 0; i < 10; ++i) { ASSERT_OK(Put(Key(key1), RandomString(&rnd, 8))); } ASSERT_OK(Put(Key(key2), RandomString(&rnd, 8))); std::unique_ptr iter(db_->NewIterator(ReadOptions())); iter->Seek(Key(key1)); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Key(key1)), 0); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->key().compare(Key(key2)), 0); ASSERT_EQ(num_reseek, TestGetTickerCount(options, NUMBER_OF_RESEEKS_IN_ITERATION)); }; // No reseek assert_reseek_count(100, 0); ASSERT_OK(dbfull()->SetOptions({ {"max_sequential_skip_in_iterations", "4"} })); // Clear memtable and make new option effective dbfull()->TEST_FlushMemTable(true); // Trigger reseek assert_reseek_count(200, 1); ASSERT_OK(dbfull()->SetOptions({ {"max_sequential_skip_in_iterations", "16"} })); // Clear memtable and make new option effective dbfull()->TEST_FlushMemTable(true); // No reseek assert_reseek_count(300, 1); } #endif // ROCKSDB_LITE TEST_F(DBTest, DontDeletePendingOutputs) { Options options; options.env = env_; options.create_if_missing = true; DestroyAndReopen(options); // Every time we write to a table file, call FOF/POF with full DB scan. This // will make sure our pending_outputs_ protection work correctly std::function purge_obsolete_files_function = [&]() { JobContext job_context(0); dbfull()->TEST_LockMutex(); dbfull()->FindObsoleteFiles(&job_context, true /*force*/); dbfull()->TEST_UnlockMutex(); dbfull()->PurgeObsoleteFiles(job_context); job_context.Clean(); }; env_->table_write_callback_ = &purge_obsolete_files_function; for (int i = 0; i < 2; ++i) { ASSERT_OK(Put("a", "begin")); ASSERT_OK(Put("z", "end")); ASSERT_OK(Flush()); } // If pending output guard does not work correctly, PurgeObsoleteFiles() will // delete the file that Compaction is trying to create, causing this: error // db/db_test.cc:975: IO error: // /tmp/rocksdbtest-1552237650/db_test/000009.sst: No such file or directory Compact("a", "b"); } #ifndef ROCKSDB_LITE TEST_F(DBTest, DontDeleteMovedFile) { // This test triggers move compaction and verifies that the file is not // deleted when it's part of move compaction Options options = CurrentOptions(); options.env = env_; options.create_if_missing = true; options.max_bytes_for_level_base = 1024 * 1024; // 1 MB options.level0_file_num_compaction_trigger = 2; // trigger compaction when we have 2 files DestroyAndReopen(options); Random rnd(301); // Create two 1MB sst files for (int i = 0; i < 2; ++i) { // Create 1MB sst file for (int j = 0; j < 100; ++j) { ASSERT_OK(Put(Key(i * 50 + j), RandomString(&rnd, 10 * 1024))); } ASSERT_OK(Flush()); } // this should execute both L0->L1 and L1->(move)->L2 compactions dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,0,1", FilesPerLevel(0)); // If the moved file is actually deleted (the move-safeguard in // ~Version::Version() is not there), we get this failure: // Corruption: Can't access /000009.sst Reopen(options); } TEST_F(DBTest, OptimizeFiltersForHits) { Options options = CurrentOptions(); options.write_buffer_size = 64 * 1024; options.arena_block_size = 4 * 1024; options.target_file_size_base = 64 * 1024; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 4; options.max_bytes_for_level_base = 256 * 1024; options.max_write_buffer_number = 2; options.max_background_compactions = 8; options.max_background_flushes = 8; options.compression = kNoCompression; options.compaction_style = kCompactionStyleLevel; options.level_compaction_dynamic_level_bytes = true; BlockBasedTableOptions bbto; bbto.cache_index_and_filter_blocks = true; bbto.filter_policy.reset(NewBloomFilterPolicy(10, true)); bbto.whole_key_filtering = true; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); options.optimize_filters_for_hits = true; options.statistics = rocksdb::CreateDBStatistics(); CreateAndReopenWithCF({"mypikachu"}, options); int numkeys = 200000; // Generate randomly shuffled keys, so the updates are almost // random. std::vector keys; keys.reserve(numkeys); for (int i = 0; i < numkeys; i += 2) { keys.push_back(i); } std::random_shuffle(std::begin(keys), std::end(keys)); int num_inserted = 0; for (int key : keys) { ASSERT_OK(Put(1, Key(key), "val")); if (++num_inserted % 1000 == 0) { dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } } ASSERT_OK(Put(1, Key(0), "val")); ASSERT_OK(Put(1, Key(numkeys), "val")); ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); if (NumTableFilesAtLevel(0, 1) == 0) { // No Level 0 file. Create one. ASSERT_OK(Put(1, Key(0), "val")); ASSERT_OK(Put(1, Key(numkeys), "val")); ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); } for (int i = 1; i < numkeys; i += 2) { ASSERT_EQ(Get(1, Key(i)), "NOT_FOUND"); } ASSERT_EQ(0, TestGetTickerCount(options, GET_HIT_L0)); ASSERT_EQ(0, TestGetTickerCount(options, GET_HIT_L1)); ASSERT_EQ(0, TestGetTickerCount(options, GET_HIT_L2_AND_UP)); // Now we have three sorted run, L0, L5 and L6 with most files in L6 have // no bloom filter. Most keys be checked bloom filters twice. ASSERT_GT(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 65000 * 2); ASSERT_LT(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 120000 * 2); for (int i = 0; i < numkeys; i += 2) { ASSERT_EQ(Get(1, Key(i)), "val"); } // Part 2 (read path): rewrite last level with blooms, then verify they get // cached only if !optimize_filters_for_hits options.disable_auto_compactions = true; options.num_levels = 9; options.optimize_filters_for_hits = false; options.statistics = CreateDBStatistics(); bbto.block_cache.reset(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); ReopenWithColumnFamilies({"default", "mypikachu"}, options); MoveFilesToLevel(7 /* level */, 1 /* column family index */); std::string value = Get(1, Key(0)); uint64_t prev_cache_filter_hits = TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT); value = Get(1, Key(0)); ASSERT_EQ(prev_cache_filter_hits + 1, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); // Now that we know the filter blocks exist in the last level files, see if // filter caching is skipped for this optimization options.optimize_filters_for_hits = true; options.statistics = CreateDBStatistics(); bbto.block_cache.reset(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); ReopenWithColumnFamilies({"default", "mypikachu"}, options); value = Get(1, Key(0)); ASSERT_EQ(0, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(0, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); ASSERT_EQ(2 /* index and data block */, TestGetTickerCount(options, BLOCK_CACHE_ADD)); // Check filter block ignored for files preloaded during DB::Open() options.max_open_files = -1; options.statistics = CreateDBStatistics(); bbto.block_cache.reset(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); ReopenWithColumnFamilies({"default", "mypikachu"}, options); uint64_t prev_cache_filter_misses = TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS); prev_cache_filter_hits = TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT); Get(1, Key(0)); ASSERT_EQ(prev_cache_filter_misses, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(prev_cache_filter_hits, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); // Check filter block ignored for file trivially-moved to bottom level bbto.block_cache.reset(); options.max_open_files = 100; // setting > -1 makes it not preload all files options.statistics = CreateDBStatistics(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); ReopenWithColumnFamilies({"default", "mypikachu"}, options); ASSERT_OK(Put(1, Key(numkeys + 1), "val")); ASSERT_OK(Flush(1)); int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); CompactRangeOptions compact_options; compact_options.bottommost_level_compaction = BottommostLevelCompaction::kSkip; compact_options.change_level = true; compact_options.target_level = 7; db_->CompactRange(compact_options, handles_[1], nullptr, nullptr); ASSERT_EQ(trivial_move, 1); ASSERT_EQ(non_trivial_move, 0); prev_cache_filter_hits = TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT); prev_cache_filter_misses = TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS); value = Get(1, Key(numkeys + 1)); ASSERT_EQ(prev_cache_filter_hits, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); ASSERT_EQ(prev_cache_filter_misses, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); // Check filter block not cached for iterator bbto.block_cache.reset(); options.statistics = CreateDBStatistics(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); ReopenWithColumnFamilies({"default", "mypikachu"}, options); std::unique_ptr iter(db_->NewIterator(ReadOptions(), handles_[1])); iter->SeekToFirst(); ASSERT_EQ(0, TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS)); ASSERT_EQ(0, TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT)); ASSERT_EQ(2 /* index and data block */, TestGetTickerCount(options, BLOCK_CACHE_ADD)); } #endif // ROCKSDB_LITE TEST_F(DBTest, L0L1L2AndUpHitCounter) { Options options = CurrentOptions(); options.write_buffer_size = 32 * 1024; options.target_file_size_base = 32 * 1024; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 4; options.max_bytes_for_level_base = 64 * 1024; options.max_write_buffer_number = 2; options.max_background_compactions = 8; options.max_background_flushes = 8; options.statistics = rocksdb::CreateDBStatistics(); CreateAndReopenWithCF({"mypikachu"}, options); int numkeys = 20000; for (int i = 0; i < numkeys; i++) { ASSERT_OK(Put(1, Key(i), "val")); } ASSERT_EQ(0, TestGetTickerCount(options, GET_HIT_L0)); ASSERT_EQ(0, TestGetTickerCount(options, GET_HIT_L1)); ASSERT_EQ(0, TestGetTickerCount(options, GET_HIT_L2_AND_UP)); ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); for (int i = 0; i < numkeys; i++) { ASSERT_EQ(Get(1, Key(i)), "val"); } ASSERT_GT(TestGetTickerCount(options, GET_HIT_L0), 100); ASSERT_GT(TestGetTickerCount(options, GET_HIT_L1), 100); ASSERT_GT(TestGetTickerCount(options, GET_HIT_L2_AND_UP), 100); ASSERT_EQ(numkeys, TestGetTickerCount(options, GET_HIT_L0) + TestGetTickerCount(options, GET_HIT_L1) + TestGetTickerCount(options, GET_HIT_L2_AND_UP)); } TEST_F(DBTest, EncodeDecompressedBlockSizeTest) { // iter 0 -- zlib // iter 1 -- bzip2 // iter 2 -- lz4 // iter 3 -- lz4HC CompressionType compressions[] = {kZlibCompression, kBZip2Compression, kLZ4Compression, kLZ4HCCompression}; for (int iter = 0; iter < 4; ++iter) { if (!CompressionTypeSupported(compressions[iter])) { continue; } // first_table_version 1 -- generate with table_version == 1, read with // table_version == 2 // first_table_version 2 -- generate with table_version == 2, read with // table_version == 1 for (int first_table_version = 1; first_table_version <= 2; ++first_table_version) { BlockBasedTableOptions table_options; table_options.format_version = first_table_version; table_options.filter_policy.reset(NewBloomFilterPolicy(10)); Options options = CurrentOptions(); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.create_if_missing = true; options.compression = compressions[iter]; DestroyAndReopen(options); int kNumKeysWritten = 100000; Random rnd(301); for (int i = 0; i < kNumKeysWritten; ++i) { // compressible string ASSERT_OK(Put(Key(i), RandomString(&rnd, 128) + std::string(128, 'a'))); } table_options.format_version = first_table_version == 1 ? 2 : 1; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); Reopen(options); for (int i = 0; i < kNumKeysWritten; ++i) { auto r = Get(Key(i)); ASSERT_EQ(r.substr(128), std::string(128, 'a')); } } } } TEST_F(DBTest, MutexWaitStatsDisabledByDefault) { Options options = CurrentOptions(); options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); CreateAndReopenWithCF({"pikachu"}, options); const uint64_t kMutexWaitDelay = 100; ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, kMutexWaitDelay); ASSERT_OK(Put("hello", "rocksdb")); ASSERT_EQ(TestGetTickerCount(options, DB_MUTEX_WAIT_MICROS), 0); ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0); } TEST_F(DBTest, MutexWaitStats) { Options options = CurrentOptions(); options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); options.statistics->stats_level_ = StatsLevel::kAll; CreateAndReopenWithCF({"pikachu"}, options); const uint64_t kMutexWaitDelay = 100; ThreadStatusUtil::TEST_SetStateDelay( ThreadStatus::STATE_MUTEX_WAIT, kMutexWaitDelay); ASSERT_OK(Put("hello", "rocksdb")); ASSERT_GE(TestGetTickerCount( options, DB_MUTEX_WAIT_MICROS), kMutexWaitDelay); ThreadStatusUtil::TEST_SetStateDelay( ThreadStatus::STATE_MUTEX_WAIT, 0); } #ifndef ROCKSDB_LITE // This reproduces a bug where we don't delete a file because when it was // supposed to be deleted, it was blocked by pending_outputs // Consider: // 1. current file_number is 13 // 2. compaction (1) starts, blocks deletion of all files starting with 13 // (pending outputs) // 3. file 13 is created by compaction (2) // 4. file 13 is consumed by compaction (3) and file 15 was created. Since file // 13 has no references, it is put into VersionSet::obsolete_files_ // 5. FindObsoleteFiles() gets file 13 from VersionSet::obsolete_files_. File 13 // is deleted from obsolete_files_ set. // 6. PurgeObsoleteFiles() tries to delete file 13, but this file is blocked by // pending outputs since compaction (1) is still running. It is not deleted and // it is not present in obsolete_files_ anymore. Therefore, we never delete it. TEST_F(DBTest, DeleteObsoleteFilesPendingOutputs) { Options options = CurrentOptions(); options.env = env_; options.write_buffer_size = 2 * 1024 * 1024; // 2 MB options.max_bytes_for_level_base = 1024 * 1024; // 1 MB options.level0_file_num_compaction_trigger = 2; // trigger compaction when we have 2 files options.max_background_flushes = 2; options.max_background_compactions = 2; OnFileDeletionListener* listener = new OnFileDeletionListener(); options.listeners.emplace_back(listener); Reopen(options); Random rnd(301); // Create two 1MB sst files for (int i = 0; i < 2; ++i) { // Create 1MB sst file for (int j = 0; j < 100; ++j) { ASSERT_OK(Put(Key(i * 50 + j), RandomString(&rnd, 10 * 1024))); } ASSERT_OK(Flush()); } // this should execute both L0->L1 and L1->(move)->L2 compactions dbfull()->TEST_WaitForCompact(); ASSERT_EQ("0,0,1", FilesPerLevel(0)); test::SleepingBackgroundTask blocking_thread; port::Mutex mutex_; bool already_blocked(false); // block the flush std::function block_first_time = [&]() { bool blocking = false; { MutexLock l(&mutex_); if (!already_blocked) { blocking = true; already_blocked = true; } } if (blocking) { blocking_thread.DoSleep(); } }; env_->table_write_callback_ = &block_first_time; // Create 1MB sst file for (int j = 0; j < 256; ++j) { ASSERT_OK(Put(Key(j), RandomString(&rnd, 10 * 1024))); } // this should trigger a flush, which is blocked with block_first_time // pending_file is protecting all the files created after ASSERT_OK(dbfull()->TEST_CompactRange(2, nullptr, nullptr)); ASSERT_EQ("0,0,0,1", FilesPerLevel(0)); std::vector metadata; db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(metadata.size(), 1U); auto file_on_L2 = metadata[0].name; listener->SetExpectedFileName(dbname_ + file_on_L2); ASSERT_OK(dbfull()->TEST_CompactRange(3, nullptr, nullptr, nullptr, true /* disallow trivial move */)); ASSERT_EQ("0,0,0,0,1", FilesPerLevel(0)); // finish the flush! blocking_thread.WakeUp(); blocking_thread.WaitUntilDone(); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ("1,0,0,0,1", FilesPerLevel(0)); metadata.clear(); db_->GetLiveFilesMetaData(&metadata); ASSERT_EQ(metadata.size(), 2U); // This file should have been deleted during last compaction ASSERT_EQ(Status::NotFound(), env_->FileExists(dbname_ + file_on_L2)); listener->VerifyMatchedCount(1); } #endif // ROCKSDB_LITE TEST_F(DBTest, CloseSpeedup) { Options options = CurrentOptions(); options.compaction_style = kCompactionStyleLevel; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 2; options.num_levels = 4; options.max_bytes_for_level_base = 400 * 1024; options.max_write_buffer_number = 16; // Block background threads env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); test::SleepingBackgroundTask sleeping_task_high; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_high, Env::Priority::HIGH); std::vector filenames; env_->GetChildren(dbname_, &filenames); // Delete archival files. for (size_t i = 0; i < filenames.size(); ++i) { env_->DeleteFile(dbname_ + "/" + filenames[i]); } env_->DeleteDir(dbname_); DestroyAndReopen(options); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); Random rnd(301); int key_idx = 0; // First three 110KB files are not going to level 2 // After that, (100K, 200K) for (int num = 0; num < 5; num++) { GenerateNewFile(&rnd, &key_idx, true); } ASSERT_EQ(0, GetSstFileCount(dbname_)); Close(); ASSERT_EQ(0, GetSstFileCount(dbname_)); // Unblock background threads sleeping_task_high.WakeUp(); sleeping_task_high.WaitUntilDone(); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); Destroy(options); } class DelayedMergeOperator : public MergeOperator { private: DBTest* db_test_; public: explicit DelayedMergeOperator(DBTest* d) : db_test_(d) {} virtual bool FullMerge(const Slice& key, const Slice* existing_value, const std::deque& operand_list, std::string* new_value, Logger* logger) const override { db_test_->env_->addon_time_.fetch_add(1000); *new_value = ""; return true; } virtual const char* Name() const override { return "DelayedMergeOperator"; } }; TEST_F(DBTest, MergeTestTime) { std::string one, two, three; PutFixed64(&one, 1); PutFixed64(&two, 2); PutFixed64(&three, 3); // Enable time profiling SetPerfLevel(kEnableTime); this->env_->addon_time_.store(0); this->env_->time_elapse_only_sleep_ = true; this->env_->no_sleep_ = true; Options options; options = CurrentOptions(options); options.statistics = rocksdb::CreateDBStatistics(); options.merge_operator.reset(new DelayedMergeOperator(this)); DestroyAndReopen(options); ASSERT_EQ(TestGetTickerCount(options, MERGE_OPERATION_TOTAL_TIME), 0); db_->Put(WriteOptions(), "foo", one); ASSERT_OK(Flush()); ASSERT_OK(db_->Merge(WriteOptions(), "foo", two)); ASSERT_OK(Flush()); ASSERT_OK(db_->Merge(WriteOptions(), "foo", three)); ASSERT_OK(Flush()); ReadOptions opt; opt.verify_checksums = true; opt.snapshot = nullptr; std::string result; db_->Get(opt, "foo", &result); ASSERT_EQ(1000000, TestGetTickerCount(options, MERGE_OPERATION_TOTAL_TIME)); ReadOptions read_options; std::unique_ptr iter(db_->NewIterator(read_options)); int count = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(iter->status()); ++count; } ASSERT_EQ(1, count); ASSERT_EQ(2000000, TestGetTickerCount(options, MERGE_OPERATION_TOTAL_TIME)); #if ROCKSDB_USING_THREAD_STATUS ASSERT_GT(TestGetTickerCount(options, FLUSH_WRITE_BYTES), 0); #endif // ROCKSDB_USING_THREAD_STATUS this->env_->time_elapse_only_sleep_ = false; } #ifndef ROCKSDB_LITE TEST_P(DBTestWithParam, MergeCompactionTimeTest) { SetPerfLevel(kEnableTime); Options options; options = CurrentOptions(options); options.compaction_filter_factory = std::make_shared(); options.statistics = rocksdb::CreateDBStatistics(); options.merge_operator.reset(new DelayedMergeOperator(this)); options.compaction_style = kCompactionStyleUniversal; options.max_subcompactions = max_subcompactions_; DestroyAndReopen(options); for (int i = 0; i < 1000; i++) { ASSERT_OK(db_->Merge(WriteOptions(), "foo", "TEST")); ASSERT_OK(Flush()); } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); ASSERT_NE(TestGetTickerCount(options, MERGE_OPERATION_TOTAL_TIME), 0); } TEST_P(DBTestWithParam, FilterCompactionTimeTest) { Options options; options.compaction_filter_factory = std::make_shared(this); options.disable_auto_compactions = true; options.create_if_missing = true; options.statistics = rocksdb::CreateDBStatistics(); options.max_subcompactions = max_subcompactions_; options = CurrentOptions(options); DestroyAndReopen(options); // put some data for (int table = 0; table < 4; ++table) { for (int i = 0; i < 10 + table; ++i) { Put(ToString(table * 100 + i), "val"); } Flush(); } CompactRangeOptions cro; cro.exclusive_manual_compaction = exclusive_manual_compaction_; ASSERT_OK(db_->CompactRange(cro, nullptr, nullptr)); ASSERT_EQ(0U, CountLiveFiles()); Reopen(options); Iterator* itr = db_->NewIterator(ReadOptions()); itr->SeekToFirst(); ASSERT_NE(TestGetTickerCount(options, FILTER_OPERATION_TOTAL_TIME), 0); delete itr; } #endif // ROCKSDB_LITE TEST_F(DBTest, TestLogCleanup) { Options options = CurrentOptions(); options.write_buffer_size = 64 * 1024; // very small // only two memtables allowed ==> only two log files options.max_write_buffer_number = 2; Reopen(options); for (int i = 0; i < 100000; ++i) { Put(Key(i), "val"); // only 2 memtables will be alive, so logs_to_free needs to always be below // 2 ASSERT_LT(dbfull()->TEST_LogsToFreeSize(), static_cast(3)); } } #ifndef ROCKSDB_LITE TEST_F(DBTest, EmptyCompactedDB) { Options options; options.max_open_files = -1; options = CurrentOptions(options); Close(); ASSERT_OK(ReadOnlyReopen(options)); Status s = Put("new", "value"); ASSERT_TRUE(s.IsNotSupported()); Close(); } #endif // ROCKSDB_LITE #ifndef ROCKSDB_LITE TEST_F(DBTest, SuggestCompactRangeTest) { class CompactionFilterFactoryGetContext : public CompactionFilterFactory { public: virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { saved_context = context; std::unique_ptr empty_filter; return empty_filter; } const char* Name() const override { return "CompactionFilterFactoryGetContext"; } static bool IsManual(CompactionFilterFactory* compaction_filter_factory) { return reinterpret_cast( compaction_filter_factory)->saved_context.is_manual_compaction; } CompactionFilter::Context saved_context; }; Options options = CurrentOptions(); options.memtable_factory.reset( new SpecialSkipListFactory(DBTestBase::kNumKeysByGenerateNewRandomFile)); options.compaction_style = kCompactionStyleLevel; options.compaction_filter_factory.reset( new CompactionFilterFactoryGetContext()); options.write_buffer_size = 200 << 10; options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 4; options.num_levels = 4; options.compression = kNoCompression; options.max_bytes_for_level_base = 450 << 10; options.target_file_size_base = 98 << 10; options.max_grandparent_overlap_factor = 1 << 20; // inf Reopen(options); Random rnd(301); for (int num = 0; num < 3; num++) { GenerateNewRandomFile(&rnd); } GenerateNewRandomFile(&rnd); ASSERT_EQ("0,4", FilesPerLevel(0)); ASSERT_TRUE(!CompactionFilterFactoryGetContext::IsManual( options.compaction_filter_factory.get())); GenerateNewRandomFile(&rnd); ASSERT_EQ("1,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("2,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("3,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("0,4,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("1,4,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("2,4,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("3,4,4", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("0,4,8", FilesPerLevel(0)); GenerateNewRandomFile(&rnd); ASSERT_EQ("1,4,8", FilesPerLevel(0)); // compact it three times for (int i = 0; i < 3; ++i) { ASSERT_OK(experimental::SuggestCompactRange(db_, nullptr, nullptr)); dbfull()->TEST_WaitForCompact(); } // All files are compacted ASSERT_EQ(0, NumTableFilesAtLevel(0)); ASSERT_EQ(0, NumTableFilesAtLevel(1)); GenerateNewRandomFile(&rnd); ASSERT_EQ(1, NumTableFilesAtLevel(0)); // nonoverlapping with the file on level 0 Slice start("a"), end("b"); ASSERT_OK(experimental::SuggestCompactRange(db_, &start, &end)); dbfull()->TEST_WaitForCompact(); // should not compact the level 0 file ASSERT_EQ(1, NumTableFilesAtLevel(0)); start = Slice("j"); end = Slice("m"); ASSERT_OK(experimental::SuggestCompactRange(db_, &start, &end)); dbfull()->TEST_WaitForCompact(); ASSERT_TRUE(CompactionFilterFactoryGetContext::IsManual( options.compaction_filter_factory.get())); // now it should compact the level 0 file ASSERT_EQ(0, NumTableFilesAtLevel(0)); ASSERT_EQ(1, NumTableFilesAtLevel(1)); } TEST_F(DBTest, PromoteL0) { Options options = CurrentOptions(); options.disable_auto_compactions = true; options.write_buffer_size = 10 * 1024 * 1024; DestroyAndReopen(options); // non overlapping ranges std::vector> ranges = { {81, 160}, {0, 80}, {161, 240}, {241, 320}}; int32_t value_size = 10 * 1024; // 10 KB Random rnd(301); std::map values; for (const auto& range : ranges) { for (int32_t j = range.first; j < range.second; j++) { values[j] = RandomString(&rnd, value_size); ASSERT_OK(Put(Key(j), values[j])); } ASSERT_OK(Flush()); } int32_t level0_files = NumTableFilesAtLevel(0, 0); ASSERT_EQ(level0_files, ranges.size()); ASSERT_EQ(NumTableFilesAtLevel(1, 0), 0); // No files in L1 // Promote L0 level to L2. ASSERT_OK(experimental::PromoteL0(db_, db_->DefaultColumnFamily(), 2)); // We expect that all the files were trivially moved from L0 to L2 ASSERT_EQ(NumTableFilesAtLevel(0, 0), 0); ASSERT_EQ(NumTableFilesAtLevel(2, 0), level0_files); for (const auto& kv : values) { ASSERT_EQ(Get(Key(kv.first)), kv.second); } } TEST_F(DBTest, PromoteL0Failure) { Options options = CurrentOptions(); options.disable_auto_compactions = true; options.write_buffer_size = 10 * 1024 * 1024; DestroyAndReopen(options); // Produce two L0 files with overlapping ranges. ASSERT_OK(Put(Key(0), "")); ASSERT_OK(Put(Key(3), "")); ASSERT_OK(Flush()); ASSERT_OK(Put(Key(1), "")); ASSERT_OK(Flush()); Status status; // Fails because L0 has overlapping files. status = experimental::PromoteL0(db_, db_->DefaultColumnFamily()); ASSERT_TRUE(status.IsInvalidArgument()); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); // Now there is a file in L1. ASSERT_GE(NumTableFilesAtLevel(1, 0), 1); ASSERT_OK(Put(Key(5), "")); ASSERT_OK(Flush()); // Fails because L1 is non-empty. status = experimental::PromoteL0(db_, db_->DefaultColumnFamily()); ASSERT_TRUE(status.IsInvalidArgument()); } #endif // ROCKSDB_LITE // Github issue #596 TEST_F(DBTest, HugeNumberOfLevels) { Options options = CurrentOptions(); options.write_buffer_size = 2 * 1024 * 1024; // 2MB options.max_bytes_for_level_base = 2 * 1024 * 1024; // 2MB options.num_levels = 12; options.max_background_compactions = 10; options.max_bytes_for_level_multiplier = 2; options.level_compaction_dynamic_level_bytes = true; DestroyAndReopen(options); Random rnd(301); for (int i = 0; i < 300000; ++i) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024))); } ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); } TEST_F(DBTest, AutomaticConflictsWithManualCompaction) { Options options = CurrentOptions(); options.write_buffer_size = 2 * 1024 * 1024; // 2MB options.max_bytes_for_level_base = 2 * 1024 * 1024; // 2MB options.num_levels = 12; options.max_background_compactions = 10; options.max_bytes_for_level_multiplier = 2; options.level_compaction_dynamic_level_bytes = true; DestroyAndReopen(options); Random rnd(301); for (int i = 0; i < 300000; ++i) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024))); } std::atomic callback_count(0); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction()::Conflict", [&](void* arg) { callback_count.fetch_add(1); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); CompactRangeOptions croptions; croptions.exclusive_manual_compaction = false; ASSERT_OK(db_->CompactRange(croptions, nullptr, nullptr)); ASSERT_GE(callback_count.load(), 1); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); for (int i = 0; i < 300000; ++i) { ASSERT_NE("NOT_FOUND", Get(Key(i))); } } // Github issue #595 // Large write batch with column families TEST_F(DBTest, LargeBatchWithColumnFamilies) { Options options; options.env = env_; options = CurrentOptions(options); options.write_buffer_size = 100000; // Small write buffer CreateAndReopenWithCF({"pikachu"}, options); int64_t j = 0; for (int i = 0; i < 5; i++) { for (int pass = 1; pass <= 3; pass++) { WriteBatch batch; size_t write_size = 1024 * 1024 * (5 + i); fprintf(stderr, "prepare: %" ROCKSDB_PRIszt " MB, pass:%d\n", (write_size / 1024 / 1024), pass); for (;;) { std::string data(3000, j++ % 127 + 20); data += ToString(j); batch.Put(handles_[0], Slice(data), Slice(data)); if (batch.GetDataSize() > write_size) { break; } } fprintf(stderr, "write: %" ROCKSDB_PRIszt " MB\n", (batch.GetDataSize() / 1024 / 1024)); ASSERT_OK(dbfull()->Write(WriteOptions(), &batch)); fprintf(stderr, "done\n"); } } // make sure we can re-open it. ASSERT_OK(TryReopenWithColumnFamilies({"default", "pikachu"}, options)); } // Make sure that Flushes can proceed in parallel with CompactRange() TEST_F(DBTest, FlushesInParallelWithCompactRange) { // iter == 0 -- leveled // iter == 1 -- leveled, but throw in a flush between two levels compacting // iter == 2 -- universal for (int iter = 0; iter < 3; ++iter) { Options options = CurrentOptions(); if (iter < 2) { options.compaction_style = kCompactionStyleLevel; } else { options.compaction_style = kCompactionStyleUniversal; } options.write_buffer_size = 110 << 10; options.level0_file_num_compaction_trigger = 4; options.num_levels = 4; options.compression = kNoCompression; options.max_bytes_for_level_base = 450 << 10; options.target_file_size_base = 98 << 10; options.max_write_buffer_number = 2; DestroyAndReopen(options); Random rnd(301); for (int num = 0; num < 14; num++) { GenerateNewRandomFile(&rnd); } if (iter == 1) { rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"DBImpl::RunManualCompaction()::1", "DBTest::FlushesInParallelWithCompactRange:1"}, {"DBTest::FlushesInParallelWithCompactRange:2", "DBImpl::RunManualCompaction()::2"}}); } else { rocksdb::SyncPoint::GetInstance()->LoadDependency( {{"CompactionJob::Run():Start", "DBTest::FlushesInParallelWithCompactRange:1"}, {"DBTest::FlushesInParallelWithCompactRange:2", "CompactionJob::Run():End"}}); } rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::vector threads; threads.emplace_back([&]() { Compact("a", "z"); }); TEST_SYNC_POINT("DBTest::FlushesInParallelWithCompactRange:1"); // this has to start a flush. if flushes are blocked, this will try to // create // 3 memtables, and that will fail because max_write_buffer_number is 2 for (int num = 0; num < 3; num++) { GenerateNewRandomFile(&rnd, /* nowait */ true); } TEST_SYNC_POINT("DBTest::FlushesInParallelWithCompactRange:2"); for (auto& t : threads) { t.join(); } rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } } TEST_F(DBTest, DelayedWriteRate) { const int kEntriesPerMemTable = 100; const int kTotalFlushes = 20; Options options; env_->SetBackgroundThreads(1, Env::LOW); options.env = env_; env_->no_sleep_ = true; options = CurrentOptions(options); options.write_buffer_size = 100000000; options.max_write_buffer_number = 256; options.max_background_compactions = 1; options.level0_file_num_compaction_trigger = 3; options.level0_slowdown_writes_trigger = 3; options.level0_stop_writes_trigger = 999999; options.delayed_write_rate = 20000000; // Start with 200MB/s options.memtable_factory.reset( new SpecialSkipListFactory(kEntriesPerMemTable)); CreateAndReopenWithCF({"pikachu"}, options); // Block compactions test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); for (int i = 0; i < 3; i++) { Put(Key(i), std::string(10000, 'x')); Flush(); } // These writes will be slowed down to 1KB/s uint64_t estimated_sleep_time = 0; Random rnd(301); Put("", ""); uint64_t cur_rate = options.delayed_write_rate; for (int i = 0; i < kTotalFlushes; i++) { uint64_t size_memtable = 0; for (int j = 0; j < kEntriesPerMemTable; j++) { auto rand_num = rnd.Uniform(20); // Spread the size range to more. size_t entry_size = rand_num * rand_num * rand_num; WriteOptions wo; Put(Key(i), std::string(entry_size, 'x'), wo); size_memtable += entry_size + 18; // Occasionally sleep a while if (rnd.Uniform(20) == 6) { env_->SleepForMicroseconds(2666); } } dbfull()->TEST_WaitForFlushMemTable(); estimated_sleep_time += size_memtable * 1000000u / cur_rate; // Slow down twice. One for memtable switch and one for flush finishes. cur_rate = static_cast(static_cast(cur_rate) / kSlowdownRatio / kSlowdownRatio); } // Estimate the total sleep time fall into the rough range. ASSERT_GT(env_->addon_time_.load(), static_cast(estimated_sleep_time / 2)); ASSERT_LT(env_->addon_time_.load(), static_cast(estimated_sleep_time * 2)); env_->no_sleep_ = false; rocksdb::SyncPoint::GetInstance()->DisableProcessing(); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); } TEST_F(DBTest, HardLimit) { Options options; options.env = env_; env_->SetBackgroundThreads(1, Env::LOW); options = CurrentOptions(options); options.max_write_buffer_number = 256; options.write_buffer_size = 110 << 10; // 110KB options.arena_block_size = 4 * 1024; options.level0_file_num_compaction_trigger = 4; options.level0_slowdown_writes_trigger = 999999; options.level0_stop_writes_trigger = 999999; options.hard_pending_compaction_bytes_limit = 800 << 10; options.max_bytes_for_level_base = 10000000000u; options.max_background_compactions = 1; options.memtable_factory.reset( new SpecialSkipListFactory(KNumKeysByGenerateNewFile - 1)); env_->SetBackgroundThreads(1, Env::LOW); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); CreateAndReopenWithCF({"pikachu"}, options); std::atomic callback_count(0); rocksdb::SyncPoint::GetInstance()->SetCallBack("DBImpl::DelayWrite:Wait", [&](void* arg) { callback_count.fetch_add(1); sleeping_task_low.WakeUp(); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Random rnd(301); int key_idx = 0; for (int num = 0; num < 5; num++) { GenerateNewFile(&rnd, &key_idx, true); } ASSERT_EQ(0, callback_count.load()); for (int num = 0; num < 5; num++) { GenerateNewFile(&rnd, &key_idx, true); dbfull()->TEST_WaitForFlushMemTable(); } ASSERT_GE(callback_count.load(), 1); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); sleeping_task_low.WaitUntilDone(); } #ifndef ROCKSDB_LITE TEST_F(DBTest, SoftLimit) { Options options; options.env = env_; options = CurrentOptions(options); options.write_buffer_size = 100000; // Small write buffer options.max_write_buffer_number = 256; options.level0_file_num_compaction_trigger = 1; options.level0_slowdown_writes_trigger = 3; options.level0_stop_writes_trigger = 999999; options.delayed_write_rate = 20000; // About 200KB/s limited rate options.soft_pending_compaction_bytes_limit = 200000; options.target_file_size_base = 99999999; // All into one file options.max_bytes_for_level_base = 50000; options.max_bytes_for_level_multiplier = 10; options.max_background_compactions = 1; options.compression = kNoCompression; Reopen(options); Put(Key(0), ""); test::SleepingBackgroundTask sleeping_task_low; // Block compactions env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); sleeping_task_low.WaitUntilSleeping(); // Create 3 L0 files, making score of L0 to be 3. for (int i = 0; i < 3; i++) { Put(Key(i), std::string(5000, 'x')); Put(Key(100 - i), std::string(5000, 'x')); // Flush the file. File size is around 30KB. Flush(); } ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); sleeping_task_low.Reset(); dbfull()->TEST_WaitForCompact(); // Now there is one L1 file but doesn't trigger soft_rate_limit // The L1 file size is around 30KB. ASSERT_EQ(NumTableFilesAtLevel(1), 1); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); // Only allow one compactin going through. rocksdb::SyncPoint::GetInstance()->SetCallBack( "BackgroundCallCompaction:0", [&](void* arg) { // Schedule a sleeping task. sleeping_task_low.Reset(); env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); sleeping_task_low.WaitUntilSleeping(); // Create 3 L0 files, making score of L0 to be 3 for (int i = 0; i < 3; i++) { Put(Key(10 + i), std::string(5000, 'x')); Put(Key(90 - i), std::string(5000, 'x')); // Flush the file. File size is around 30KB. Flush(); } // Wake up sleep task to enable compaction to run and waits // for it to go to sleep state again to make sure one compaction // goes through. sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilSleeping(); // Now there is one L1 file (around 60KB) which exceeds 50KB base by 10KB // Given level multiplier 10, estimated pending compaction is around 100KB // doesn't trigger soft_pending_compaction_bytes_limit ASSERT_EQ(NumTableFilesAtLevel(1), 1); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); // Create 3 L0 files, making score of L0 to be 3, higher than L0. for (int i = 0; i < 3; i++) { Put(Key(20 + i), std::string(5000, 'x')); Put(Key(80 - i), std::string(5000, 'x')); // Flush the file. File size is around 30KB. Flush(); } // Wake up sleep task to enable compaction to run and waits // for it to go to sleep state again to make sure one compaction // goes through. sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilSleeping(); // Now there is one L1 file (around 90KB) which exceeds 50KB base by 40KB // Given level multiplier 10, estimated pending compaction is around 400KB // triggerring soft_pending_compaction_bytes_limit ASSERT_EQ(NumTableFilesAtLevel(1), 1); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilSleeping(); ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); // shrink level base so L2 will hit soft limit easier. ASSERT_OK(dbfull()->SetOptions({ {"max_bytes_for_level_base", "5000"}, })); Put("", ""); Flush(); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); sleeping_task_low.WaitUntilSleeping(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); sleeping_task_low.WakeUp(); sleeping_task_low.WaitUntilDone(); } TEST_F(DBTest, LastWriteBufferDelay) { Options options; options.env = env_; options = CurrentOptions(options); options.write_buffer_size = 100000; options.max_write_buffer_number = 4; options.delayed_write_rate = 20000; options.compression = kNoCompression; options.disable_auto_compactions = true; int kNumKeysPerMemtable = 3; options.memtable_factory.reset( new SpecialSkipListFactory(kNumKeysPerMemtable)); Reopen(options); test::SleepingBackgroundTask sleeping_task; // Block flushes env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task, Env::Priority::HIGH); sleeping_task.WaitUntilSleeping(); // Create 3 L0 files, making score of L0 to be 3. for (int i = 0; i < 3; i++) { // Fill one mem table for (int j = 0; j < kNumKeysPerMemtable; j++) { Put(Key(j), ""); } ASSERT_TRUE(!dbfull()->TEST_write_controler().NeedsDelay()); } // Inserting a new entry would create a new mem table, triggering slow down. Put(Key(0), ""); ASSERT_TRUE(dbfull()->TEST_write_controler().NeedsDelay()); sleeping_task.WakeUp(); sleeping_task.WaitUntilDone(); } #endif // ROCKSDB_LITE TEST_F(DBTest, FailWhenCompressionNotSupportedTest) { CompressionType compressions[] = {kZlibCompression, kBZip2Compression, kLZ4Compression, kLZ4HCCompression}; for (int iter = 0; iter < 4; ++iter) { if (!CompressionTypeSupported(compressions[iter])) { // not supported, we should fail the Open() Options options = CurrentOptions(); options.compression = compressions[iter]; ASSERT_TRUE(!TryReopen(options).ok()); // Try if CreateColumnFamily also fails options.compression = kNoCompression; ASSERT_OK(TryReopen(options)); ColumnFamilyOptions cf_options(options); cf_options.compression = compressions[iter]; ColumnFamilyHandle* handle; ASSERT_TRUE(!db_->CreateColumnFamily(cf_options, "name", &handle).ok()); } } } #ifndef ROCKSDB_LITE TEST_F(DBTest, RowCache) { Options options = CurrentOptions(); options.statistics = rocksdb::CreateDBStatistics(); options.row_cache = NewLRUCache(8192); DestroyAndReopen(options); ASSERT_OK(Put("foo", "bar")); ASSERT_OK(Flush()); ASSERT_EQ(TestGetTickerCount(options, ROW_CACHE_HIT), 0); ASSERT_EQ(TestGetTickerCount(options, ROW_CACHE_MISS), 0); ASSERT_EQ(Get("foo"), "bar"); ASSERT_EQ(TestGetTickerCount(options, ROW_CACHE_HIT), 0); ASSERT_EQ(TestGetTickerCount(options, ROW_CACHE_MISS), 1); ASSERT_EQ(Get("foo"), "bar"); ASSERT_EQ(TestGetTickerCount(options, ROW_CACHE_HIT), 1); ASSERT_EQ(TestGetTickerCount(options, ROW_CACHE_MISS), 1); } #endif // ROCKSDB_LITE // TODO(3.13): fix the issue of Seek() + Prev() which might not necessary // return the biggest key which is smaller than the seek key. TEST_F(DBTest, PrevAfterMerge) { Options options; options.create_if_missing = true; options.merge_operator = MergeOperators::CreatePutOperator(); DestroyAndReopen(options); // write three entries with different keys using Merge() WriteOptions wopts; db_->Merge(wopts, "1", "data1"); db_->Merge(wopts, "2", "data2"); db_->Merge(wopts, "3", "data3"); std::unique_ptr it(db_->NewIterator(ReadOptions())); it->Seek("2"); ASSERT_TRUE(it->Valid()); ASSERT_EQ("2", it->key().ToString()); it->Prev(); ASSERT_TRUE(it->Valid()); ASSERT_EQ("1", it->key().ToString()); } TEST_F(DBTest, DeletingOldWalAfterDrop) { rocksdb::SyncPoint::GetInstance()->LoadDependency( { { "Test:AllowFlushes", "DBImpl::BGWorkFlush" }, { "DBImpl::BGWorkFlush:done", "Test:WaitForFlush"} }); rocksdb::SyncPoint::GetInstance()->ClearTrace(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); Options options = CurrentOptions(); options.max_total_wal_size = 8192; options.compression = kNoCompression; options.write_buffer_size = 1 << 20; options.level0_file_num_compaction_trigger = (1<<30); options.level0_slowdown_writes_trigger = (1<<30); options.level0_stop_writes_trigger = (1<<30); options.disable_auto_compactions = true; DestroyAndReopen(options); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); CreateColumnFamilies({"cf1", "cf2"}, options); ASSERT_OK(Put(0, "key1", DummyString(8192))); ASSERT_OK(Put(0, "key2", DummyString(8192))); // the oldest wal should now be getting_flushed ASSERT_OK(db_->DropColumnFamily(handles_[0])); // all flushes should now do nothing because their CF is dropped TEST_SYNC_POINT("Test:AllowFlushes"); TEST_SYNC_POINT("Test:WaitForFlush"); uint64_t lognum1 = dbfull()->TEST_LogfileNumber(); ASSERT_OK(Put(1, "key3", DummyString(8192))); ASSERT_OK(Put(1, "key4", DummyString(8192))); // new wal should have been created uint64_t lognum2 = dbfull()->TEST_LogfileNumber(); EXPECT_GT(lognum2, lognum1); } #ifndef ROCKSDB_LITE TEST_F(DBTest, DBWithSstFileManager) { std::shared_ptr sst_file_manager(NewSstFileManager(env_)); auto sfm = static_cast(sst_file_manager.get()); int files_added = 0; int files_deleted = 0; int files_moved = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "SstFileManagerImpl::OnAddFile", [&](void* arg) { files_added++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "SstFileManagerImpl::OnDeleteFile", [&](void* arg) { files_deleted++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "SstFileManagerImpl::OnMoveFile", [&](void* arg) { files_moved++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.sst_file_manager = sst_file_manager; DestroyAndReopen(options); Random rnd(301); for (int i = 0; i < 25; i++) { GenerateNewRandomFile(&rnd); ASSERT_OK(Flush()); dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); // Verify that we are tracking all sst files in dbname_ ASSERT_EQ(sfm->GetTrackedFiles(), GetAllSSTFiles()); } ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); auto files_in_db = GetAllSSTFiles(); // Verify that we are tracking all sst files in dbname_ ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db); // Verify the total files size uint64_t total_files_size = 0; for (auto& file_to_size : files_in_db) { total_files_size += file_to_size.second; } ASSERT_EQ(sfm->GetTotalSize(), total_files_size); // We flushed at least 25 files ASSERT_GE(files_added, 25); // Compaction must have deleted some files ASSERT_GT(files_deleted, 0); // No files were moved ASSERT_EQ(files_moved, 0); Close(); Reopen(options); ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db); ASSERT_EQ(sfm->GetTotalSize(), total_files_size); // Verify that we track all the files again after the DB is closed and opened Close(); sst_file_manager.reset(NewSstFileManager(env_)); options.sst_file_manager = sst_file_manager; sfm = static_cast(sst_file_manager.get()); Reopen(options); ASSERT_EQ(sfm->GetTrackedFiles(), files_in_db); ASSERT_EQ(sfm->GetTotalSize(), total_files_size); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBTest, RateLimitedDelete) { rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"DBTest::RateLimitedDelete:1", "DeleteScheduler::BackgroundEmptyTrash"}, }); std::vector penalties; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DeleteScheduler::BackgroundEmptyTrash:Wait", [&](void* arg) { penalties.push_back(*(static_cast(arg))); }); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); Options options = CurrentOptions(); options.disable_auto_compactions = true; options.env = env_; std::string trash_dir = test::TmpDir(env_) + "/trash"; int64_t rate_bytes_per_sec = 1024 * 10; // 10 Kbs / Sec Status s; options.sst_file_manager.reset(NewSstFileManager( env_, nullptr, trash_dir, rate_bytes_per_sec, false, &s)); ASSERT_OK(s); auto sfm = static_cast(options.sst_file_manager.get()); Destroy(last_options_); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); ASSERT_OK(TryReopen(options)); // Create 4 files in L0 for (char v = 'a'; v <= 'd'; v++) { ASSERT_OK(Put("Key2", DummyString(1024, v))); ASSERT_OK(Put("Key3", DummyString(1024, v))); ASSERT_OK(Put("Key4", DummyString(1024, v))); ASSERT_OK(Put("Key1", DummyString(1024, v))); ASSERT_OK(Put("Key4", DummyString(1024, v))); ASSERT_OK(Flush()); } // We created 4 sst files in L0 ASSERT_EQ("4", FilesPerLevel(0)); std::vector metadata; db_->GetLiveFilesMetaData(&metadata); // Compaction will move the 4 files in L0 to trash and create 1 L1 file ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ("0,1", FilesPerLevel(0)); uint64_t delete_start_time = env_->NowMicros(); // Hold BackgroundEmptyTrash TEST_SYNC_POINT("DBTest::RateLimitedDelete:1"); sfm->WaitForEmptyTrash(); uint64_t time_spent_deleting = env_->NowMicros() - delete_start_time; uint64_t total_files_size = 0; uint64_t expected_penlty = 0; ASSERT_EQ(penalties.size(), metadata.size()); for (size_t i = 0; i < metadata.size(); i++) { total_files_size += metadata[i].size; expected_penlty = ((total_files_size * 1000000) / rate_bytes_per_sec); ASSERT_EQ(expected_penlty, penalties[i]); } ASSERT_GT(time_spent_deleting, expected_penlty * 0.9); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } // Create a DB with 2 db_paths, and generate multiple files in the 2 // db_paths using CompactRangeOptions, make sure that files that were // deleted from first db_path were deleted using DeleteScheduler and // files in the second path were not. TEST_F(DBTest, DeleteSchedulerMultipleDBPaths) { int bg_delete_file = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DeleteScheduler::DeleteTrashFile:DeleteFile", [&](void* arg) { bg_delete_file++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.disable_auto_compactions = true; options.db_paths.emplace_back(dbname_, 1024 * 100); options.db_paths.emplace_back(dbname_ + "_2", 1024 * 100); options.env = env_; std::string trash_dir = test::TmpDir(env_) + "/trash"; int64_t rate_bytes_per_sec = 1024 * 1024; // 1 Mb / Sec Status s; options.sst_file_manager.reset(NewSstFileManager( env_, nullptr, trash_dir, rate_bytes_per_sec, false, &s)); ASSERT_OK(s); auto sfm = static_cast(options.sst_file_manager.get()); DestroyAndReopen(options); // Create 4 files in L0 for (int i = 0; i < 4; i++) { ASSERT_OK(Put("Key" + ToString(i), DummyString(1024, 'A'))); ASSERT_OK(Flush()); } // We created 4 sst files in L0 ASSERT_EQ("4", FilesPerLevel(0)); // Compaction will delete files from L0 in first db path and generate a new // file in L1 in second db path CompactRangeOptions compact_options; compact_options.target_path_id = 1; Slice begin("Key0"); Slice end("Key3"); ASSERT_OK(db_->CompactRange(compact_options, &begin, &end)); ASSERT_EQ("0,1", FilesPerLevel(0)); // Create 4 files in L0 for (int i = 4; i < 8; i++) { ASSERT_OK(Put("Key" + ToString(i), DummyString(1024, 'B'))); ASSERT_OK(Flush()); } ASSERT_EQ("4,1", FilesPerLevel(0)); // Compaction will delete files from L0 in first db path and generate a new // file in L1 in second db path begin = "Key4"; end = "Key7"; ASSERT_OK(db_->CompactRange(compact_options, &begin, &end)); ASSERT_EQ("0,2", FilesPerLevel(0)); sfm->WaitForEmptyTrash(); ASSERT_EQ(bg_delete_file, 8); compact_options.bottommost_level_compaction = BottommostLevelCompaction::kForce; ASSERT_OK(db_->CompactRange(compact_options, nullptr, nullptr)); ASSERT_EQ("0,1", FilesPerLevel(0)); sfm->WaitForEmptyTrash(); ASSERT_EQ(bg_delete_file, 8); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBTest, DestroyDBWithRateLimitedDelete) { int bg_delete_file = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DeleteScheduler::DeleteTrashFile:DeleteFile", [&](void* arg) { bg_delete_file++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options = CurrentOptions(); options.disable_auto_compactions = true; options.env = env_; DestroyAndReopen(options); // Create 4 files in L0 for (int i = 0; i < 4; i++) { ASSERT_OK(Put("Key" + ToString(i), DummyString(1024, 'A'))); ASSERT_OK(Flush()); } // We created 4 sst files in L0 ASSERT_EQ("4", FilesPerLevel(0)); // Close DB and destroy it using DeleteScheduler Close(); std::string trash_dir = test::TmpDir(env_) + "/trash"; int64_t rate_bytes_per_sec = 1024 * 1024; // 1 Mb / Sec Status s; options.sst_file_manager.reset(NewSstFileManager( env_, nullptr, trash_dir, rate_bytes_per_sec, false, &s)); ASSERT_OK(s); ASSERT_OK(DestroyDB(dbname_, options)); auto sfm = static_cast(options.sst_file_manager.get()); sfm->WaitForEmptyTrash(); // We have deleted the 4 sst files in the delete_scheduler ASSERT_EQ(bg_delete_file, 4); } #endif // ROCKSDB_LITE TEST_F(DBTest, UnsupportedManualSync) { DestroyAndReopen(CurrentOptions()); env_->is_wal_sync_thread_safe_.store(false); Status s = db_->SyncWAL(); ASSERT_TRUE(s.IsNotSupported()); } #ifndef ROCKSDB_LITE TEST_F(DBTest, OpenDBWithInfiniteMaxOpenFiles) { // Open DB with infinite max open files // - First iteration use 1 thread to open files // - Second iteration use 5 threads to open files for (int iter = 0; iter < 2; iter++) { Options options; options.create_if_missing = true; options.write_buffer_size = 100000; options.disable_auto_compactions = true; options.max_open_files = -1; if (iter == 0) { options.max_file_opening_threads = 1; } else { options.max_file_opening_threads = 5; } options = CurrentOptions(options); DestroyAndReopen(options); // Create 12 Files in L0 (then move then to L2) for (int i = 0; i < 12; i++) { std::string k = "L2_" + Key(i); ASSERT_OK(Put(k, k + std::string(1000, 'a'))); ASSERT_OK(Flush()); } CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 2; db_->CompactRange(compact_options, nullptr, nullptr); // Create 12 Files in L0 for (int i = 0; i < 12; i++) { std::string k = "L0_" + Key(i); ASSERT_OK(Put(k, k + std::string(1000, 'a'))); ASSERT_OK(Flush()); } Close(); // Reopening the DB will load all exisitng files Reopen(options); ASSERT_EQ("12,0,12", FilesPerLevel(0)); std::vector> files; dbfull()->TEST_GetFilesMetaData(db_->DefaultColumnFamily(), &files); for (const auto& level : files) { for (const auto& file : level) { ASSERT_TRUE(file.table_reader_handle != nullptr); } } for (int i = 0; i < 12; i++) { ASSERT_EQ(Get("L0_" + Key(i)), "L0_" + Key(i) + std::string(1000, 'a')); ASSERT_EQ(Get("L2_" + Key(i)), "L2_" + Key(i) + std::string(1000, 'a')); } } } TEST_F(DBTest, GetTotalSstFilesSize) { Options options = CurrentOptions(); options.disable_auto_compactions = true; options.compression = kNoCompression; DestroyAndReopen(options); // Generate 5 files in L0 for (int i = 0; i < 5; i++) { for (int j = 0; j < 10; j++) { std::string val = "val_file_" + ToString(i); ASSERT_OK(Put(Key(j), val)); } Flush(); } ASSERT_EQ("5", FilesPerLevel(0)); std::vector live_files_meta; dbfull()->GetLiveFilesMetaData(&live_files_meta); ASSERT_EQ(live_files_meta.size(), 5); uint64_t single_file_size = live_files_meta[0].size; uint64_t live_sst_files_size = 0; uint64_t total_sst_files_size = 0; for (const auto& file_meta : live_files_meta) { live_sst_files_size += file_meta.size; } ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 5 // Total SST files = 5 ASSERT_EQ(live_sst_files_size, 5 * single_file_size); ASSERT_EQ(total_sst_files_size, 5 * single_file_size); // hold current version std::unique_ptr iter1(dbfull()->NewIterator(ReadOptions())); // Compact 5 files into 1 file in L0 ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ("0,1", FilesPerLevel(0)); live_files_meta.clear(); dbfull()->GetLiveFilesMetaData(&live_files_meta); ASSERT_EQ(live_files_meta.size(), 1); live_sst_files_size = 0; total_sst_files_size = 0; for (const auto& file_meta : live_files_meta) { live_sst_files_size += file_meta.size; } ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 1 (compacted file) // Total SST files = 6 (5 original files + compacted file) ASSERT_EQ(live_sst_files_size, 1 * single_file_size); ASSERT_EQ(total_sst_files_size, 6 * single_file_size); // hold current version std::unique_ptr iter2(dbfull()->NewIterator(ReadOptions())); // Delete all keys and compact, this will delete all live files for (int i = 0; i < 10; i++) { ASSERT_OK(Delete(Key(i))); } Flush(); ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ("", FilesPerLevel(0)); live_files_meta.clear(); dbfull()->GetLiveFilesMetaData(&live_files_meta); ASSERT_EQ(live_files_meta.size(), 0); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 0 // Total SST files = 6 (5 original files + compacted file) ASSERT_EQ(total_sst_files_size, 6 * single_file_size); iter1.reset(); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 0 // Total SST files = 1 (compacted file) ASSERT_EQ(total_sst_files_size, 1 * single_file_size); iter2.reset(); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 0 // Total SST files = 0 ASSERT_EQ(total_sst_files_size, 0); } TEST_F(DBTest, GetTotalSstFilesSizeVersionsFilesShared) { Options options = CurrentOptions(); options.disable_auto_compactions = true; options.compression = kNoCompression; DestroyAndReopen(options); // Generate 5 files in L0 for (int i = 0; i < 5; i++) { ASSERT_OK(Put(Key(i), "val")); Flush(); } ASSERT_EQ("5", FilesPerLevel(0)); std::vector live_files_meta; dbfull()->GetLiveFilesMetaData(&live_files_meta); ASSERT_EQ(live_files_meta.size(), 5); uint64_t single_file_size = live_files_meta[0].size; uint64_t live_sst_files_size = 0; uint64_t total_sst_files_size = 0; for (const auto& file_meta : live_files_meta) { live_sst_files_size += file_meta.size; } ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 5 // Total SST files = 5 ASSERT_EQ(live_sst_files_size, 5 * single_file_size); ASSERT_EQ(total_sst_files_size, 5 * single_file_size); // hold current version std::unique_ptr iter1(dbfull()->NewIterator(ReadOptions())); // Compaction will do trivial move from L0 to L1 ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ("0,5", FilesPerLevel(0)); live_files_meta.clear(); dbfull()->GetLiveFilesMetaData(&live_files_meta); ASSERT_EQ(live_files_meta.size(), 5); live_sst_files_size = 0; total_sst_files_size = 0; for (const auto& file_meta : live_files_meta) { live_sst_files_size += file_meta.size; } ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 5 // Total SST files = 5 (used in 2 version) ASSERT_EQ(live_sst_files_size, 5 * single_file_size); ASSERT_EQ(total_sst_files_size, 5 * single_file_size); // hold current version std::unique_ptr iter2(dbfull()->NewIterator(ReadOptions())); // Delete all keys and compact, this will delete all live files for (int i = 0; i < 5; i++) { ASSERT_OK(Delete(Key(i))); } Flush(); ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ("", FilesPerLevel(0)); live_files_meta.clear(); dbfull()->GetLiveFilesMetaData(&live_files_meta); ASSERT_EQ(live_files_meta.size(), 0); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 0 // Total SST files = 5 (used in 2 version) ASSERT_EQ(total_sst_files_size, 5 * single_file_size); iter1.reset(); iter2.reset(); ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.total-sst-files-size", &total_sst_files_size)); // Live SST files = 0 // Total SST files = 0 ASSERT_EQ(total_sst_files_size, 0); } TEST_F(DBTest, AddExternalSstFile) { do { std::string sst_files_folder = test::TmpDir(env_) + "/sst_files/"; env_->CreateDir(sst_files_folder); Options options = CurrentOptions(); options.env = env_; const ImmutableCFOptions ioptions(options); SstFileWriter sst_file_writer(EnvOptions(), ioptions, options.comparator); // file1.sst (0 => 99) std::string file1 = sst_files_folder + "file1.sst"; ASSERT_OK(sst_file_writer.Open(file1)); for (int k = 0; k < 100; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val")); } ExternalSstFileInfo file1_info; Status s = sst_file_writer.Finish(&file1_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file1_info.file_path, file1); ASSERT_EQ(file1_info.num_entries, 100); ASSERT_EQ(file1_info.smallest_key, Key(0)); ASSERT_EQ(file1_info.largest_key, Key(99)); // sst_file_writer already finished, cannot add this value s = sst_file_writer.Add(Key(100), "bad_val"); ASSERT_FALSE(s.ok()) << s.ToString(); // file2.sst (100 => 199) std::string file2 = sst_files_folder + "file2.sst"; ASSERT_OK(sst_file_writer.Open(file2)); for (int k = 100; k < 200; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val")); } // Cannot add this key because it's not after last added key s = sst_file_writer.Add(Key(99), "bad_val"); ASSERT_FALSE(s.ok()) << s.ToString(); ExternalSstFileInfo file2_info; s = sst_file_writer.Finish(&file2_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file2_info.file_path, file2); ASSERT_EQ(file2_info.num_entries, 100); ASSERT_EQ(file2_info.smallest_key, Key(100)); ASSERT_EQ(file2_info.largest_key, Key(199)); // file3.sst (195 => 299) // This file values overlap with file2 values std::string file3 = sst_files_folder + "file3.sst"; ASSERT_OK(sst_file_writer.Open(file3)); for (int k = 195; k < 300; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val_overlap")); } ExternalSstFileInfo file3_info; s = sst_file_writer.Finish(&file3_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file3_info.file_path, file3); ASSERT_EQ(file3_info.num_entries, 105); ASSERT_EQ(file3_info.smallest_key, Key(195)); ASSERT_EQ(file3_info.largest_key, Key(299)); // file4.sst (30 => 39) // This file values overlap with file1 values std::string file4 = sst_files_folder + "file4.sst"; ASSERT_OK(sst_file_writer.Open(file4)); for (int k = 30; k < 40; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val_overlap")); } ExternalSstFileInfo file4_info; s = sst_file_writer.Finish(&file4_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file4_info.file_path, file4); ASSERT_EQ(file4_info.num_entries, 10); ASSERT_EQ(file4_info.smallest_key, Key(30)); ASSERT_EQ(file4_info.largest_key, Key(39)); // file5.sst (400 => 499) std::string file5 = sst_files_folder + "file5.sst"; ASSERT_OK(sst_file_writer.Open(file5)); for (int k = 400; k < 500; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val")); } ExternalSstFileInfo file5_info; s = sst_file_writer.Finish(&file5_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file5_info.file_path, file5); ASSERT_EQ(file5_info.num_entries, 100); ASSERT_EQ(file5_info.smallest_key, Key(400)); ASSERT_EQ(file5_info.largest_key, Key(499)); // Cannot create an empty sst file std::string file_empty = sst_files_folder + "file_empty.sst"; ExternalSstFileInfo file_empty_info; s = sst_file_writer.Finish(&file_empty_info); ASSERT_NOK(s); DestroyAndReopen(options); // Add file using file path s = db_->AddFile(file1); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(db_->GetLatestSequenceNumber(), 0U); for (int k = 0; k < 100; k++) { ASSERT_EQ(Get(Key(k)), Key(k) + "_val"); } // Add file while holding a snapshot will fail const Snapshot* s1 = db_->GetSnapshot(); if (s1 != nullptr) { ASSERT_NOK(db_->AddFile(&file2_info)); db_->ReleaseSnapshot(s1); } // We can add the file after releaseing the snapshot ASSERT_OK(db_->AddFile(&file2_info)); ASSERT_EQ(db_->GetLatestSequenceNumber(), 0U); for (int k = 0; k < 200; k++) { ASSERT_EQ(Get(Key(k)), Key(k) + "_val"); } // This file have overlapping values with the exisitng data s = db_->AddFile(file3); ASSERT_FALSE(s.ok()) << s.ToString(); // This file have overlapping values with the exisitng data s = db_->AddFile(&file4_info); ASSERT_FALSE(s.ok()) << s.ToString(); // Overwrite values of keys divisible by 5 for (int k = 0; k < 200; k += 5) { ASSERT_OK(Put(Key(k), Key(k) + "_val_new")); } ASSERT_NE(db_->GetLatestSequenceNumber(), 0U); // Key range of file5 (400 => 499) dont overlap with any keys in DB ASSERT_OK(db_->AddFile(file5)); // Make sure values are correct before and after flush/compaction for (int i = 0; i < 2; i++) { for (int k = 0; k < 200; k++) { std::string value = Key(k) + "_val"; if (k % 5 == 0) { value += "_new"; } ASSERT_EQ(Get(Key(k)), value); } for (int k = 400; k < 500; k++) { std::string value = Key(k) + "_val"; ASSERT_EQ(Get(Key(k)), value); } ASSERT_OK(Flush()); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); } Close(); options.disable_auto_compactions = true; Reopen(options); // Delete keys in range (400 => 499) for (int k = 400; k < 500; k++) { ASSERT_OK(Delete(Key(k))); } // We deleted range (400 => 499) but cannot add file5 because // of the range tombstones ASSERT_NOK(db_->AddFile(file5)); // Compacting the DB will remove the tombstones ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); // Now we can add the file ASSERT_OK(db_->AddFile(file5)); // Verify values of file5 in DB for (int k = 400; k < 500; k++) { std::string value = Key(k) + "_val"; ASSERT_EQ(Get(Key(k)), value); } } while (ChangeOptions(kSkipPlainTable | kSkipUniversalCompaction | kSkipFIFOCompaction)); } TEST_F(DBTest, AddExternalSstFileNoCopy) { std::string sst_files_folder = test::TmpDir(env_) + "/sst_files/"; env_->CreateDir(sst_files_folder); Options options = CurrentOptions(); options.env = env_; const ImmutableCFOptions ioptions(options); SstFileWriter sst_file_writer(EnvOptions(), ioptions, options.comparator); // file1.sst (0 => 99) std::string file1 = sst_files_folder + "file1.sst"; ASSERT_OK(sst_file_writer.Open(file1)); for (int k = 0; k < 100; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val")); } ExternalSstFileInfo file1_info; Status s = sst_file_writer.Finish(&file1_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file1_info.file_path, file1); ASSERT_EQ(file1_info.num_entries, 100); ASSERT_EQ(file1_info.smallest_key, Key(0)); ASSERT_EQ(file1_info.largest_key, Key(99)); // file2.sst (100 => 299) std::string file2 = sst_files_folder + "file2.sst"; ASSERT_OK(sst_file_writer.Open(file2)); for (int k = 100; k < 300; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val")); } ExternalSstFileInfo file2_info; s = sst_file_writer.Finish(&file2_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file2_info.file_path, file2); ASSERT_EQ(file2_info.num_entries, 200); ASSERT_EQ(file2_info.smallest_key, Key(100)); ASSERT_EQ(file2_info.largest_key, Key(299)); // file3.sst (110 => 124) .. overlap with file2.sst std::string file3 = sst_files_folder + "file3.sst"; ASSERT_OK(sst_file_writer.Open(file3)); for (int k = 110; k < 125; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k) + "_val_overlap")); } ExternalSstFileInfo file3_info; s = sst_file_writer.Finish(&file3_info); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(file3_info.file_path, file3); ASSERT_EQ(file3_info.num_entries, 15); ASSERT_EQ(file3_info.smallest_key, Key(110)); ASSERT_EQ(file3_info.largest_key, Key(124)); s = db_->AddFile(&file1_info, true /* move file */); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_EQ(Status::NotFound(), env_->FileExists(file1)); s = db_->AddFile(&file2_info, false /* copy file */); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_OK(env_->FileExists(file2)); // This file have overlapping values with the exisitng data s = db_->AddFile(&file3_info, true /* move file */); ASSERT_FALSE(s.ok()) << s.ToString(); ASSERT_OK(env_->FileExists(file3)); for (int k = 0; k < 300; k++) { ASSERT_EQ(Get(Key(k)), Key(k) + "_val"); } } TEST_F(DBTest, AddExternalSstFileMultiThreaded) { std::string sst_files_folder = test::TmpDir(env_) + "/sst_files/"; // Bulk load 10 files every file contain 1000 keys int num_files = 10; int keys_per_file = 1000; // Generate file names std::vector file_names; for (int i = 0; i < num_files; i++) { std::string file_name = "file_" + ToString(i) + ".sst"; file_names.push_back(sst_files_folder + file_name); } do { env_->CreateDir(sst_files_folder); Options options = CurrentOptions(); const ImmutableCFOptions ioptions(options); std::atomic thread_num(0); std::function write_file_func = [&]() { int file_idx = thread_num.fetch_add(1); int range_start = file_idx * keys_per_file; int range_end = range_start + keys_per_file; SstFileWriter sst_file_writer(EnvOptions(), ioptions, options.comparator); ASSERT_OK(sst_file_writer.Open(file_names[file_idx])); for (int k = range_start; k < range_end; k++) { ASSERT_OK(sst_file_writer.Add(Key(k), Key(k))); } Status s = sst_file_writer.Finish(); ASSERT_TRUE(s.ok()) << s.ToString(); }; // Write num_files files in parallel std::vector sst_writer_threads; for (int i = 0; i < num_files; ++i) { sst_writer_threads.emplace_back(write_file_func); } for (auto& t : sst_writer_threads) { t.join(); } fprintf(stderr, "Wrote %d files (%d keys)\n", num_files, num_files * keys_per_file); thread_num.store(0); std::atomic files_added(0); std::function load_file_func = [&]() { // We intentionally add every file twice, and assert that it was added // only once and the other add failed int thread_id = thread_num.fetch_add(1); int file_idx = thread_id / 2; // sometimes we use copy, sometimes link .. the result should be the same bool move_file = (thread_id % 3 == 0); Status s = db_->AddFile(file_names[file_idx], move_file); if (s.ok()) { files_added++; } }; // Bulk load num_files files in parallel std::vector add_file_threads; DestroyAndReopen(options); for (int i = 0; i < num_files * 2; ++i) { add_file_threads.emplace_back(load_file_func); } for (auto& t : add_file_threads) { t.join(); } ASSERT_EQ(files_added.load(), num_files); fprintf(stderr, "Loaded %d files (%d keys)\n", num_files, num_files * keys_per_file); // Overwrite values of keys divisible by 100 for (int k = 0; k < num_files * keys_per_file; k += 100) { std::string key = Key(k); Status s = Put(key, key + "_new"); ASSERT_TRUE(s.ok()); } for (int i = 0; i < 2; i++) { // Make sure the values are correct before and after flush/compaction for (int k = 0; k < num_files * keys_per_file; ++k) { std::string key = Key(k); std::string value = (k % 100 == 0) ? (key + "_new") : key; ASSERT_EQ(Get(key), value); } ASSERT_OK(Flush()); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); } fprintf(stderr, "Verified %d values\n", num_files * keys_per_file); } while (ChangeOptions(kSkipPlainTable | kSkipUniversalCompaction | kSkipFIFOCompaction)); } TEST_F(DBTest, AddExternalSstFileOverlappingRanges) { std::string sst_files_folder = test::TmpDir(env_) + "/sst_files/"; Random rnd(301); do { env_->CreateDir(sst_files_folder); Options options = CurrentOptions(); DestroyAndReopen(options); const ImmutableCFOptions ioptions(options); SstFileWriter sst_file_writer(EnvOptions(), ioptions, options.comparator); printf("Option config = %d\n", option_config_); std::vector> key_ranges; for (int i = 0; i < 500; i++) { int range_start = rnd.Uniform(20000); int keys_per_range = 10 + rnd.Uniform(41); key_ranges.emplace_back(range_start, range_start + keys_per_range); } int memtable_add = 0; int success_add_file = 0; int failed_add_file = 0; std::map true_data; for (size_t i = 0; i < key_ranges.size(); i++) { int range_start = key_ranges[i].first; int range_end = key_ranges[i].second; Status s; std::string range_val = "range_" + ToString(i); // For 20% of ranges we use DB::Put, for 80% we use DB::AddFile if (i && i % 5 == 0) { // Use DB::Put to insert range (insert into memtable) range_val += "_put"; for (int k = range_start; k <= range_end; k++) { s = Put(Key(k), range_val); ASSERT_OK(s); } memtable_add++; } else { // Use DB::AddFile to insert range range_val += "_add_file"; // Generate the file containing the range std::string file_name = sst_files_folder + env_->GenerateUniqueId(); ASSERT_OK(sst_file_writer.Open(file_name)); for (int k = range_start; k <= range_end; k++) { s = sst_file_writer.Add(Key(k), range_val); ASSERT_OK(s); } ExternalSstFileInfo file_info; s = sst_file_writer.Finish(&file_info); ASSERT_OK(s); // Insert the generated file s = db_->AddFile(&file_info); auto it = true_data.lower_bound(Key(range_start)); if (it != true_data.end() && it->first <= Key(range_end)) { // This range overlap with data already exist in DB ASSERT_NOK(s); failed_add_file++; } else { ASSERT_OK(s); success_add_file++; } } if (s.ok()) { // Update true_data map to include the new inserted data for (int k = range_start; k <= range_end; k++) { true_data[Key(k)] = range_val; } } // Flush / Compact the DB if (i && i % 50 == 0) { Flush(); } if (i && i % 75 == 0) { db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); } } printf( "Total: %zu ranges\n" "AddFile()|Success: %d ranges\n" "AddFile()|RangeConflict: %d ranges\n" "Put(): %d ranges\n", key_ranges.size(), success_add_file, failed_add_file, memtable_add); // Verify the correctness of the data for (const auto& kv : true_data) { ASSERT_EQ(Get(kv.first), kv.second); } printf("keys/values verified\n"); } while (ChangeOptions(kSkipPlainTable | kSkipUniversalCompaction | kSkipFIFOCompaction)); } #endif // ROCKSDB_LITE // 1 Create some SST files by inserting K-V pairs into DB // 2 Close DB and change suffix from ".sst" to ".ldb" for every other SST file // 3 Open DB and check if all key can be read TEST_F(DBTest, SSTsWithLdbSuffixHandling) { Options options = CurrentOptions(); options.write_buffer_size = 110 << 10; // 110KB options.num_levels = 4; DestroyAndReopen(options); Random rnd(301); int key_id = 0; for (int i = 0; i < 10; ++i) { GenerateNewFile(&rnd, &key_id, false); } Flush(); Close(); int const num_files = GetSstFileCount(dbname_); ASSERT_GT(num_files, 0); std::vector filenames; GetSstFiles(dbname_, &filenames); int num_ldb_files = 0; for (size_t i = 0; i < filenames.size(); ++i) { if (i & 1) { continue; } std::string const rdb_name = dbname_ + "/" + filenames[i]; std::string const ldb_name = Rocks2LevelTableFileName(rdb_name); ASSERT_TRUE(env_->RenameFile(rdb_name, ldb_name).ok()); ++num_ldb_files; } ASSERT_GT(num_ldb_files, 0); ASSERT_EQ(num_files, GetSstFileCount(dbname_)); Reopen(options); for (int k = 0; k < key_id; ++k) { ASSERT_NE("NOT_FOUND", Get(Key(k))); } Destroy(options); } TEST_F(DBTest, PinnedDataIteratorRandomized) { enum TestConfig { NORMAL, CLOSE_AND_OPEN, COMPACT_BEFORE_READ, FLUSH_EVERY_1000, MAX }; // Generate Random data Random rnd(301); int puts = 100000; int key_pool = static_cast(puts * 0.7); int key_size = 100; int val_size = 1000; int seeks_percentage = 20; // 20% of keys will be used to test seek() int delete_percentage = 20; // 20% of keys will be deleted int merge_percentage = 20; // 20% of keys will be added using Merge() for (int run_config = 0; run_config < TestConfig::MAX; run_config++) { Options options = CurrentOptions(); BlockBasedTableOptions table_options; table_options.use_delta_encoding = false; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.merge_operator = MergeOperators::CreatePutOperator(); DestroyAndReopen(options); std::vector generated_keys(key_pool); for (int i = 0; i < key_pool; i++) { generated_keys[i] = RandomString(&rnd, key_size); } std::map true_data; std::vector random_keys; std::vector deleted_keys; for (int i = 0; i < puts; i++) { auto& k = generated_keys[rnd.Next() % key_pool]; auto v = RandomString(&rnd, val_size); // Insert data to true_data map and to DB true_data[k] = v; if (rnd.OneIn(static_cast(100.0 / merge_percentage))) { ASSERT_OK(db_->Merge(WriteOptions(), k, v)); } else { ASSERT_OK(Put(k, v)); } // Pick random keys to be used to test Seek() if (rnd.OneIn(static_cast(100.0 / seeks_percentage))) { random_keys.push_back(k); } // Delete some random keys if (rnd.OneIn(static_cast(100.0 / delete_percentage))) { deleted_keys.push_back(k); true_data.erase(k); ASSERT_OK(Delete(k)); } if (run_config == TestConfig::FLUSH_EVERY_1000) { if (i && i % 1000 == 0) { Flush(); } } } if (run_config == TestConfig::CLOSE_AND_OPEN) { Close(); Reopen(options); } else if (run_config == TestConfig::COMPACT_BEFORE_READ) { db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); } ReadOptions ro; ro.pin_data = true; auto iter = db_->NewIterator(ro); { // Test Seek to random keys printf("Testing seek on %zu keys\n", random_keys.size()); std::vector keys_slices; std::vector true_keys; for (auto& k : random_keys) { iter->Seek(k); if (!iter->Valid()) { ASSERT_EQ(true_data.lower_bound(k), true_data.end()); continue; } ASSERT_TRUE(iter->IsKeyPinned()); keys_slices.push_back(iter->key()); true_keys.push_back(true_data.lower_bound(k)->first); } for (size_t i = 0; i < keys_slices.size(); i++) { ASSERT_EQ(keys_slices[i].ToString(), true_keys[i]); } } { // Test iterating all data forward printf("Testing iterating forward on all keys\n"); std::vector all_keys; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_TRUE(iter->IsKeyPinned()); all_keys.push_back(iter->key()); } ASSERT_EQ(all_keys.size(), true_data.size()); // Verify that all keys slices are valid auto data_iter = true_data.begin(); for (size_t i = 0; i < all_keys.size(); i++) { ASSERT_EQ(all_keys[i].ToString(), data_iter->first); data_iter++; } } { // Test iterating all data backward printf("Testing iterating backward on all keys\n"); std::vector all_keys; for (iter->SeekToLast(); iter->Valid(); iter->Prev()) { ASSERT_TRUE(iter->IsKeyPinned()); all_keys.push_back(iter->key()); } ASSERT_EQ(all_keys.size(), true_data.size()); // Verify that all keys slices are valid (backward) auto data_iter = true_data.rbegin(); for (size_t i = 0; i < all_keys.size(); i++) { ASSERT_EQ(all_keys[i].ToString(), data_iter->first); data_iter++; } } delete iter; } } #ifndef ROCKSDB_LITE TEST_F(DBTest, PinnedDataIteratorMultipleFiles) { Options options = CurrentOptions(); BlockBasedTableOptions table_options; table_options.use_delta_encoding = false; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.disable_auto_compactions = true; options.write_buffer_size = 1024 * 1024 * 10; // 10 Mb DestroyAndReopen(options); std::map true_data; // Generate 4 sst files in L2 Random rnd(301); for (int i = 1; i <= 1000; i++) { std::string k = Key(i * 3); std::string v = RandomString(&rnd, 100); ASSERT_OK(Put(k, v)); true_data[k] = v; if (i % 250 == 0) { ASSERT_OK(Flush()); } } ASSERT_EQ(FilesPerLevel(0), "4"); ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr)); ASSERT_EQ(FilesPerLevel(0), "0,4"); // Generate 4 sst files in L0 for (int i = 1; i <= 1000; i++) { std::string k = Key(i * 2); std::string v = RandomString(&rnd, 100); ASSERT_OK(Put(k, v)); true_data[k] = v; if (i % 250 == 0) { ASSERT_OK(Flush()); } } ASSERT_EQ(FilesPerLevel(0), "4,4"); // Add some keys/values in memtables for (int i = 1; i <= 1000; i++) { std::string k = Key(i); std::string v = RandomString(&rnd, 100); ASSERT_OK(Put(k, v)); true_data[k] = v; } ASSERT_EQ(FilesPerLevel(0), "4,4"); ReadOptions ro; ro.pin_data = true; auto iter = db_->NewIterator(ro); std::vector> results; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_TRUE(iter->IsKeyPinned()); results.emplace_back(iter->key(), iter->value().ToString()); } ASSERT_EQ(results.size(), true_data.size()); auto data_iter = true_data.begin(); for (size_t i = 0; i < results.size(); i++, data_iter++) { auto& kv = results[i]; ASSERT_EQ(kv.first, data_iter->first); ASSERT_EQ(kv.second, data_iter->second); } delete iter; } #endif TEST_F(DBTest, PinnedDataIteratorMergeOperator) { Options options = CurrentOptions(); BlockBasedTableOptions table_options; table_options.use_delta_encoding = false; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.merge_operator = MergeOperators::CreateUInt64AddOperator(); DestroyAndReopen(options); std::string numbers[7]; for (int val = 0; val <= 6; val++) { PutFixed64(numbers + val, val); } // +1 all keys in range [ 0 => 999] for (int i = 0; i < 1000; i++) { WriteOptions wo; ASSERT_OK(db_->Merge(wo, Key(i), numbers[1])); } // +2 all keys divisible by 2 in range [ 0 => 999] for (int i = 0; i < 1000; i += 2) { WriteOptions wo; ASSERT_OK(db_->Merge(wo, Key(i), numbers[2])); } // +3 all keys divisible by 5 in range [ 0 => 999] for (int i = 0; i < 1000; i += 5) { WriteOptions wo; ASSERT_OK(db_->Merge(wo, Key(i), numbers[3])); } ReadOptions ro; ro.pin_data = true; auto iter = db_->NewIterator(ro); std::vector> results; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_TRUE(iter->IsKeyPinned()); results.emplace_back(iter->key(), iter->value().ToString()); } ASSERT_EQ(results.size(), 1000); for (size_t i = 0; i < results.size(); i++) { auto& kv = results[i]; ASSERT_EQ(kv.first, Key(static_cast(i))); int expected_val = 1; if (i % 2 == 0) { expected_val += 2; } if (i % 5 == 0) { expected_val += 3; } ASSERT_EQ(kv.second, numbers[expected_val]); } delete iter; } TEST_F(DBTest, PinnedDataIteratorReadAfterUpdate) { Options options = CurrentOptions(); BlockBasedTableOptions table_options; table_options.use_delta_encoding = false; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); options.write_buffer_size = 100000; DestroyAndReopen(options); Random rnd(301); std::map true_data; for (int i = 0; i < 1000; i++) { std::string k = RandomString(&rnd, 10); std::string v = RandomString(&rnd, 1000); ASSERT_OK(Put(k, v)); true_data[k] = v; } ReadOptions ro; ro.pin_data = true; auto iter = db_->NewIterator(ro); // Delete 50% of the keys and update the other 50% for (auto& kv : true_data) { if (rnd.OneIn(2)) { ASSERT_OK(Delete(kv.first)); } else { std::string new_val = RandomString(&rnd, 1000); ASSERT_OK(Put(kv.first, new_val)); } } std::vector> results; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_TRUE(iter->IsKeyPinned()); results.emplace_back(iter->key(), iter->value().ToString()); } auto data_iter = true_data.begin(); for (size_t i = 0; i < results.size(); i++, data_iter++) { auto& kv = results[i]; ASSERT_EQ(kv.first, data_iter->first); ASSERT_EQ(kv.second, data_iter->second); } delete iter; } INSTANTIATE_TEST_CASE_P(DBTestWithParam, DBTestWithParam, ::testing::Combine(::testing::Values(1, 4), ::testing::Bool())); TEST_F(DBTest, PauseBackgroundWorkTest) { Options options; options.write_buffer_size = 100000; // Small write buffer options = CurrentOptions(options); Reopen(options); std::vector threads; std::atomic done(false); db_->PauseBackgroundWork(); threads.emplace_back([&]() { Random rnd(301); for (int i = 0; i < 10000; ++i) { Put(RandomString(&rnd, 10), RandomString(&rnd, 10)); } done.store(true); }); env_->SleepForMicroseconds(200000); // make sure the thread is not done ASSERT_EQ(false, done.load()); db_->ContinueBackgroundWork(); for (auto& t : threads) { t.join(); } // now it's done ASSERT_EQ(true, done.load()); } #ifndef ROCKSDB_LITE namespace { void ValidateKeyExistence(DB* db, const std::vector& keys_must_exist, const std::vector& keys_must_not_exist) { // Ensure that expected keys exist std::vector values; if (keys_must_exist.size() > 0) { std::vector status_list = db->MultiGet(ReadOptions(), keys_must_exist, &values); for (size_t i = 0; i < keys_must_exist.size(); i++) { ASSERT_OK(status_list[i]); } } // Ensure that given keys don't exist if (keys_must_not_exist.size() > 0) { std::vector status_list = db->MultiGet(ReadOptions(), keys_must_not_exist, &values); for (size_t i = 0; i < keys_must_not_exist.size(); i++) { ASSERT_TRUE(status_list[i].IsNotFound()); } } } } // namespace TEST_F(DBTest, WalFilterTest) { class TestWalFilter : public WalFilter { private: // Processing option that is requested to be applied at the given index WalFilter::WalProcessingOption wal_processing_option_; // Index at which to apply wal_processing_option_ // At other indexes default wal_processing_option::kContinueProcessing is // returned. size_t apply_option_at_record_index_; // Current record index, incremented with each record encountered. size_t current_record_index_; public: TestWalFilter(WalFilter::WalProcessingOption wal_processing_option, size_t apply_option_for_record_index) : wal_processing_option_(wal_processing_option), apply_option_at_record_index_(apply_option_for_record_index), current_record_index_(0) {} virtual WalProcessingOption LogRecord(const WriteBatch& batch, WriteBatch* new_batch, bool* batch_changed) const override { WalFilter::WalProcessingOption option_to_return; if (current_record_index_ == apply_option_at_record_index_) { option_to_return = wal_processing_option_; } else { option_to_return = WalProcessingOption::kContinueProcessing; } // Filter is passed as a const object for RocksDB to not modify the // object, however we modify it for our own purpose here and hence // cast the constness away. (const_cast(this)->current_record_index_)++; return option_to_return; } virtual const char* Name() const override { return "TestWalFilter"; } }; // Create 3 batches with two keys each std::vector> batch_keys(3); batch_keys[0].push_back("key1"); batch_keys[0].push_back("key2"); batch_keys[1].push_back("key3"); batch_keys[1].push_back("key4"); batch_keys[2].push_back("key5"); batch_keys[2].push_back("key6"); // Test with all WAL processing options for (int option = 0; option < static_cast( WalFilter::WalProcessingOption::kWalProcessingOptionMax); option++) { Options options = OptionsForLogIterTest(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Write given keys in given batches for (size_t i = 0; i < batch_keys.size(); i++) { WriteBatch batch; for (size_t j = 0; j < batch_keys[i].size(); j++) { batch.Put(handles_[0], batch_keys[i][j], DummyString(1024)); } dbfull()->Write(WriteOptions(), &batch); } WalFilter::WalProcessingOption wal_processing_option = static_cast(option); // Create a test filter that would apply wal_processing_option at the first // record size_t apply_option_for_record_index = 1; TestWalFilter test_wal_filter(wal_processing_option, apply_option_for_record_index); // Reopen database with option to use WAL filter options = OptionsForLogIterTest(); options.wal_filter = &test_wal_filter; Status status = TryReopenWithColumnFamilies({"default", "pikachu"}, options); if (wal_processing_option == WalFilter::WalProcessingOption::kCorruptedRecord) { assert(!status.ok()); // In case of corruption we can turn off paranoid_checks to reopen // databse options.paranoid_checks = false; ReopenWithColumnFamilies({"default", "pikachu"}, options); } else { assert(status.ok()); } // Compute which keys we expect to be found // and which we expect not to be found after recovery. std::vector keys_must_exist; std::vector keys_must_not_exist; switch (wal_processing_option) { case WalFilter::WalProcessingOption::kCorruptedRecord: case WalFilter::WalProcessingOption::kContinueProcessing: { fprintf(stderr, "Testing with complete WAL processing\n"); // we expect all records to be processed for (size_t i = 0; i < batch_keys.size(); i++) { for (size_t j = 0; j < batch_keys[i].size(); j++) { keys_must_exist.push_back(Slice(batch_keys[i][j])); } } break; } case WalFilter::WalProcessingOption::kIgnoreCurrentRecord: { fprintf(stderr, "Testing with ignoring record %" ROCKSDB_PRIszt " only\n", apply_option_for_record_index); // We expect the record with apply_option_for_record_index to be not // found. for (size_t i = 0; i < batch_keys.size(); i++) { for (size_t j = 0; j < batch_keys[i].size(); j++) { if (i == apply_option_for_record_index) { keys_must_not_exist.push_back(Slice(batch_keys[i][j])); } else { keys_must_exist.push_back(Slice(batch_keys[i][j])); } } } break; } case WalFilter::WalProcessingOption::kStopReplay: { fprintf(stderr, "Testing with stopping replay from record %" ROCKSDB_PRIszt "\n", apply_option_for_record_index); // We expect records beyond apply_option_for_record_index to be not // found. for (size_t i = 0; i < batch_keys.size(); i++) { for (size_t j = 0; j < batch_keys[i].size(); j++) { if (i >= apply_option_for_record_index) { keys_must_not_exist.push_back(Slice(batch_keys[i][j])); } else { keys_must_exist.push_back(Slice(batch_keys[i][j])); } } } break; } default: assert(false); // unhandled case } bool checked_after_reopen = false; while (true) { // Ensure that expected keys exists // and not expected keys don't exist after recovery ValidateKeyExistence(db_, keys_must_exist, keys_must_not_exist); if (checked_after_reopen) { break; } // reopen database again to make sure previous log(s) are not used //(even if they were skipped) // reopn database with option to use WAL filter options = OptionsForLogIterTest(); ReopenWithColumnFamilies({"default", "pikachu"}, options); checked_after_reopen = true; } } } TEST_F(DBTest, WalFilterTestWithChangeBatch) { class ChangeBatchHandler : public WriteBatch::Handler { private: // Batch to insert keys in WriteBatch* new_write_batch_; // Number of keys to add in the new batch size_t num_keys_to_add_in_new_batch_; // Number of keys added to new batch size_t num_keys_added_; public: ChangeBatchHandler(WriteBatch* new_write_batch, size_t num_keys_to_add_in_new_batch) : new_write_batch_(new_write_batch), num_keys_to_add_in_new_batch_(num_keys_to_add_in_new_batch), num_keys_added_(0) {} virtual void Put(const Slice& key, const Slice& value) override { if (num_keys_added_ < num_keys_to_add_in_new_batch_) { new_write_batch_->Put(key, value); ++num_keys_added_; } } }; class TestWalFilterWithChangeBatch : public WalFilter { private: // Index at which to start changing records size_t change_records_from_index_; // Number of keys to add in the new batch size_t num_keys_to_add_in_new_batch_; // Current record index, incremented with each record encountered. size_t current_record_index_; public: TestWalFilterWithChangeBatch(size_t change_records_from_index, size_t num_keys_to_add_in_new_batch) : change_records_from_index_(change_records_from_index), num_keys_to_add_in_new_batch_(num_keys_to_add_in_new_batch), current_record_index_(0) {} virtual WalProcessingOption LogRecord(const WriteBatch& batch, WriteBatch* new_batch, bool* batch_changed) const override { if (current_record_index_ >= change_records_from_index_) { ChangeBatchHandler handler(new_batch, num_keys_to_add_in_new_batch_); batch.Iterate(&handler); *batch_changed = true; } // Filter is passed as a const object for RocksDB to not modify the // object, however we modify it for our own purpose here and hence // cast the constness away. (const_cast(this) ->current_record_index_)++; return WalProcessingOption::kContinueProcessing; } virtual const char* Name() const override { return "TestWalFilterWithChangeBatch"; } }; std::vector> batch_keys(3); batch_keys[0].push_back("key1"); batch_keys[0].push_back("key2"); batch_keys[1].push_back("key3"); batch_keys[1].push_back("key4"); batch_keys[2].push_back("key5"); batch_keys[2].push_back("key6"); Options options = OptionsForLogIterTest(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Write given keys in given batches for (size_t i = 0; i < batch_keys.size(); i++) { WriteBatch batch; for (size_t j = 0; j < batch_keys[i].size(); j++) { batch.Put(handles_[0], batch_keys[i][j], DummyString(1024)); } dbfull()->Write(WriteOptions(), &batch); } // Create a test filter that would apply wal_processing_option at the first // record size_t change_records_from_index = 1; size_t num_keys_to_add_in_new_batch = 1; TestWalFilterWithChangeBatch test_wal_filter_with_change_batch( change_records_from_index, num_keys_to_add_in_new_batch); // Reopen database with option to use WAL filter options = OptionsForLogIterTest(); options.wal_filter = &test_wal_filter_with_change_batch; ReopenWithColumnFamilies({"default", "pikachu"}, options); // Ensure that all keys exist before change_records_from_index_ // And after that index only single key exists // as our filter adds only single key for each batch std::vector keys_must_exist; std::vector keys_must_not_exist; for (size_t i = 0; i < batch_keys.size(); i++) { for (size_t j = 0; j < batch_keys[i].size(); j++) { if (i >= change_records_from_index && j >= num_keys_to_add_in_new_batch) { keys_must_not_exist.push_back(Slice(batch_keys[i][j])); } else { keys_must_exist.push_back(Slice(batch_keys[i][j])); } } } bool checked_after_reopen = false; while (true) { // Ensure that expected keys exists // and not expected keys don't exist after recovery ValidateKeyExistence(db_, keys_must_exist, keys_must_not_exist); if (checked_after_reopen) { break; } // reopen database again to make sure previous log(s) are not used //(even if they were skipped) // reopn database with option to use WAL filter options = OptionsForLogIterTest(); ReopenWithColumnFamilies({"default", "pikachu"}, options); checked_after_reopen = true; } } TEST_F(DBTest, WalFilterTestWithChangeBatchExtraKeys) { class TestWalFilterWithChangeBatchAddExtraKeys : public WalFilter { public: virtual WalProcessingOption LogRecord(const WriteBatch& batch, WriteBatch* new_batch, bool* batch_changed) const override { *new_batch = batch; new_batch->Put("key_extra", "value_extra"); *batch_changed = true; return WalProcessingOption::kContinueProcessing; } virtual const char* Name() const override { return "WalFilterTestWithChangeBatchExtraKeys"; } }; std::vector> batch_keys(3); batch_keys[0].push_back("key1"); batch_keys[0].push_back("key2"); batch_keys[1].push_back("key3"); batch_keys[1].push_back("key4"); batch_keys[2].push_back("key5"); batch_keys[2].push_back("key6"); Options options = OptionsForLogIterTest(); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Write given keys in given batches for (size_t i = 0; i < batch_keys.size(); i++) { WriteBatch batch; for (size_t j = 0; j < batch_keys[i].size(); j++) { batch.Put(handles_[0], batch_keys[i][j], DummyString(1024)); } dbfull()->Write(WriteOptions(), &batch); } // Create a test filter that would add extra keys TestWalFilterWithChangeBatchAddExtraKeys test_wal_filter_extra_keys; // Reopen database with option to use WAL filter options = OptionsForLogIterTest(); options.wal_filter = &test_wal_filter_extra_keys; Status status = TryReopenWithColumnFamilies({"default", "pikachu"}, options); ASSERT_TRUE(status.IsNotSupported()); // Reopen without filter, now reopen should succeed - previous // attempt to open must not have altered the db. options = OptionsForLogIterTest(); ReopenWithColumnFamilies({"default", "pikachu"}, options); std::vector keys_must_exist; std::vector keys_must_not_exist; // empty vector for (size_t i = 0; i < batch_keys.size(); i++) { for (size_t j = 0; j < batch_keys[i].size(); j++) { keys_must_exist.push_back(Slice(batch_keys[i][j])); } } ValidateKeyExistence(db_, keys_must_exist, keys_must_not_exist); } #endif // ROCKSDB_LITE class SliceTransformLimitedDomain : public SliceTransform { const char* Name() const override { return "SliceTransformLimitedDomain"; } Slice Transform(const Slice& src) const override { return Slice(src.data(), 5); } bool InDomain(const Slice& src) const override { // prefix will be x???? return src.size() >= 5 && src[0] == 'x'; } bool InRange(const Slice& dst) const override { // prefix will be x???? return dst.size() == 5 && dst[0] == 'x'; } }; TEST_F(DBTest, PrefixExtractorFullFilter) { BlockBasedTableOptions bbto; // Full Filter Block bbto.filter_policy.reset(rocksdb::NewBloomFilterPolicy(10, false)); bbto.whole_key_filtering = false; Options options = CurrentOptions(); options.prefix_extractor = std::make_shared(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); DestroyAndReopen(options); ASSERT_OK(Put("x1111_AAAA", "val1")); ASSERT_OK(Put("x1112_AAAA", "val2")); ASSERT_OK(Put("x1113_AAAA", "val3")); ASSERT_OK(Put("x1114_AAAA", "val4")); // Not in domain, wont be added to filter ASSERT_OK(Put("zzzzz_AAAA", "val5")); ASSERT_OK(Flush()); ASSERT_EQ(Get("x1111_AAAA"), "val1"); ASSERT_EQ(Get("x1112_AAAA"), "val2"); ASSERT_EQ(Get("x1113_AAAA"), "val3"); ASSERT_EQ(Get("x1114_AAAA"), "val4"); // Was not added to filter but rocksdb will try to read it from the filter ASSERT_EQ(Get("zzzzz_AAAA"), "val5"); } TEST_F(DBTest, PrefixExtractorBlockFilter) { BlockBasedTableOptions bbto; // Block Filter Block bbto.filter_policy.reset(rocksdb::NewBloomFilterPolicy(10, true)); Options options = CurrentOptions(); options.prefix_extractor = std::make_shared(); options.table_factory.reset(NewBlockBasedTableFactory(bbto)); DestroyAndReopen(options); ASSERT_OK(Put("x1113_AAAA", "val3")); ASSERT_OK(Put("x1114_AAAA", "val4")); // Not in domain, wont be added to filter ASSERT_OK(Put("zzzzz_AAAA", "val1")); ASSERT_OK(Put("zzzzz_AAAB", "val2")); ASSERT_OK(Put("zzzzz_AAAC", "val3")); ASSERT_OK(Put("zzzzz_AAAD", "val4")); ASSERT_OK(Flush()); std::vector iter_res; auto iter = db_->NewIterator(ReadOptions()); // Seek to a key that was not in Domain for (iter->Seek("zzzzz_AAAA"); iter->Valid(); iter->Next()) { iter_res.emplace_back(iter->value().ToString()); } std::vector expected_res = {"val1", "val2", "val3", "val4"}; ASSERT_EQ(iter_res, expected_res); delete iter; } #ifndef ROCKSDB_LITE class BloomStatsTestWithParam : public DBTest, public testing::WithParamInterface> { public: BloomStatsTestWithParam() { use_block_table_ = std::get<0>(GetParam()); use_block_based_builder_ = std::get<1>(GetParam()); options_.create_if_missing = true; options_.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(4)); options_.memtable_prefix_bloom_bits = 8 * 1024; if (use_block_table_) { BlockBasedTableOptions table_options; table_options.hash_index_allow_collision = false; table_options.filter_policy.reset( NewBloomFilterPolicy(10, use_block_based_builder_)); options_.table_factory.reset(NewBlockBasedTableFactory(table_options)); } else { PlainTableOptions table_options; options_.table_factory.reset(NewPlainTableFactory(table_options)); } perf_context.Reset(); DestroyAndReopen(options_); } ~BloomStatsTestWithParam() { perf_context.Reset(); Destroy(options_); } // Required if inheriting from testing::WithParamInterface<> static void SetUpTestCase() {} static void TearDownTestCase() {} bool use_block_table_; bool use_block_based_builder_; Options options_; }; // 1 Insert 2 K-V pairs into DB // 2 Call Get() for both keys - expext memtable bloom hit stat to be 2 // 3 Call Get() for nonexisting key - expect memtable bloom miss stat to be 1 // 4 Call Flush() to create SST // 5 Call Get() for both keys - expext SST bloom hit stat to be 2 // 6 Call Get() for nonexisting key - expect SST bloom miss stat to be 1 // Test both: block and plain SST TEST_P(BloomStatsTestWithParam, BloomStatsTest) { std::string key1("AAAA"); std::string key2("RXDB"); // not in DB std::string key3("ZBRA"); std::string value1("Value1"); std::string value3("Value3"); ASSERT_OK(Put(key1, value1, WriteOptions())); ASSERT_OK(Put(key3, value3, WriteOptions())); // check memtable bloom stats ASSERT_EQ(value1, Get(key1)); ASSERT_EQ(1, perf_context.bloom_memtable_hit_count); ASSERT_EQ(value3, Get(key3)); ASSERT_EQ(2, perf_context.bloom_memtable_hit_count); ASSERT_EQ(0, perf_context.bloom_memtable_miss_count); ASSERT_EQ("NOT_FOUND", Get(key2)); ASSERT_EQ(1, perf_context.bloom_memtable_miss_count); ASSERT_EQ(2, perf_context.bloom_memtable_hit_count); // sanity checks ASSERT_EQ(0, perf_context.bloom_sst_hit_count); ASSERT_EQ(0, perf_context.bloom_sst_miss_count); Flush(); // sanity checks ASSERT_EQ(0, perf_context.bloom_sst_hit_count); ASSERT_EQ(0, perf_context.bloom_sst_miss_count); // check SST bloom stats // NOTE: hits per get differs because of code paths differences // in BlockBasedTable::Get() int hits_per_get = use_block_table_ && !use_block_based_builder_ ? 2 : 1; ASSERT_EQ(value1, Get(key1)); ASSERT_EQ(hits_per_get, perf_context.bloom_sst_hit_count); ASSERT_EQ(value3, Get(key3)); ASSERT_EQ(2 * hits_per_get, perf_context.bloom_sst_hit_count); ASSERT_EQ("NOT_FOUND", Get(key2)); ASSERT_EQ(1, perf_context.bloom_sst_miss_count); } // Same scenario as in BloomStatsTest but using an iterator TEST_P(BloomStatsTestWithParam, BloomStatsTestWithIter) { std::string key1("AAAA"); std::string key2("RXDB"); // not in DB std::string key3("ZBRA"); std::string value1("Value1"); std::string value3("Value3"); ASSERT_OK(Put(key1, value1, WriteOptions())); ASSERT_OK(Put(key3, value3, WriteOptions())); unique_ptr iter(dbfull()->NewIterator(ReadOptions())); // check memtable bloom stats iter->Seek(key1); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(value1, iter->value().ToString()); ASSERT_EQ(1, perf_context.bloom_memtable_hit_count); ASSERT_EQ(0, perf_context.bloom_memtable_miss_count); iter->Seek(key3); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(value3, iter->value().ToString()); ASSERT_EQ(2, perf_context.bloom_memtable_hit_count); ASSERT_EQ(0, perf_context.bloom_memtable_miss_count); iter->Seek(key2); ASSERT_OK(iter->status()); ASSERT_TRUE(!iter->Valid()); ASSERT_EQ(1, perf_context.bloom_memtable_miss_count); ASSERT_EQ(2, perf_context.bloom_memtable_hit_count); Flush(); iter.reset(dbfull()->NewIterator(ReadOptions())); // Check SST bloom stats iter->Seek(key1); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(value1, iter->value().ToString()); ASSERT_EQ(1, perf_context.bloom_sst_hit_count); iter->Seek(key3); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(value3, iter->value().ToString()); ASSERT_EQ(2, perf_context.bloom_sst_hit_count); iter->Seek(key2); ASSERT_OK(iter->status()); ASSERT_TRUE(!iter->Valid()); ASSERT_EQ(1, perf_context.bloom_sst_miss_count); ASSERT_EQ(2, perf_context.bloom_sst_hit_count); } INSTANTIATE_TEST_CASE_P(BloomStatsTestWithParam, BloomStatsTestWithParam, ::testing::Values(std::make_tuple(true, true), std::make_tuple(true, false), std::make_tuple(false, false))); #endif // ROCKSDB_LITE } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/db_test_util.cc000066400000000000000000000776031267530634000165500ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_test_util.h" namespace rocksdb { // Special Env used to delay background operations SpecialEnv::SpecialEnv(Env* base) : EnvWrapper(base), rnd_(301), sleep_counter_(this), addon_time_(0), time_elapse_only_sleep_(false), no_sleep_(false) { delay_sstable_sync_.store(false, std::memory_order_release); drop_writes_.store(false, std::memory_order_release); no_space_.store(false, std::memory_order_release); non_writable_.store(false, std::memory_order_release); count_random_reads_ = false; count_sequential_reads_ = false; manifest_sync_error_.store(false, std::memory_order_release); manifest_write_error_.store(false, std::memory_order_release); log_write_error_.store(false, std::memory_order_release); random_file_open_counter_.store(0, std::memory_order_relaxed); log_write_slowdown_ = 0; bytes_written_ = 0; sync_counter_ = 0; non_writeable_rate_ = 0; new_writable_count_ = 0; non_writable_count_ = 0; table_write_callback_ = nullptr; } DBTestBase::DBTestBase(const std::string path) : option_config_(kDefault), mem_env_(!getenv("MEM_ENV") ? nullptr : new MockEnv(Env::Default())), env_(new SpecialEnv(mem_env_ ? mem_env_ : Env::Default())) { env_->SetBackgroundThreads(1, Env::LOW); env_->SetBackgroundThreads(1, Env::HIGH); dbname_ = test::TmpDir(env_) + path; alternative_wal_dir_ = dbname_ + "/wal"; alternative_db_log_dir_ = dbname_ + "/db_log_dir"; auto options = CurrentOptions(); auto delete_options = options; delete_options.wal_dir = alternative_wal_dir_; EXPECT_OK(DestroyDB(dbname_, delete_options)); // Destroy it for not alternative WAL dir is used. EXPECT_OK(DestroyDB(dbname_, options)); db_ = nullptr; Reopen(options); Random::GetTLSInstance()->Reset(0xdeadbeef); } DBTestBase::~DBTestBase() { rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->LoadDependency({}); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); Close(); Options options; options.db_paths.emplace_back(dbname_, 0); options.db_paths.emplace_back(dbname_ + "_2", 0); options.db_paths.emplace_back(dbname_ + "_3", 0); options.db_paths.emplace_back(dbname_ + "_4", 0); EXPECT_OK(DestroyDB(dbname_, options)); delete env_; } bool DBTestBase::ShouldSkipOptions(int option_config, int skip_mask) { #ifdef ROCKSDB_LITE // These options are not supported in ROCKSDB_LITE if (option_config == kHashSkipList || option_config == kPlainTableFirstBytePrefix || option_config == kPlainTableCappedPrefix || option_config == kPlainTableCappedPrefixNonMmap || option_config == kPlainTableAllBytesPrefix || option_config == kVectorRep || option_config == kHashLinkList || option_config == kHashCuckoo || option_config == kUniversalCompaction || option_config == kUniversalCompactionMultiLevel || option_config == kUniversalSubcompactions || option_config == kFIFOCompaction || option_config == kConcurrentSkipList) { return true; } #endif if ((skip_mask & kSkipDeletesFilterFirst) && option_config == kDeletesFilterFirst) { return true; } if ((skip_mask & kSkipUniversalCompaction) && (option_config == kUniversalCompaction || option_config == kUniversalCompactionMultiLevel)) { return true; } if ((skip_mask & kSkipMergePut) && option_config == kMergePut) { return true; } if ((skip_mask & kSkipNoSeekToLast) && (option_config == kHashLinkList || option_config == kHashSkipList)) { return true; } if ((skip_mask & kSkipPlainTable) && (option_config == kPlainTableAllBytesPrefix || option_config == kPlainTableFirstBytePrefix || option_config == kPlainTableCappedPrefix || option_config == kPlainTableCappedPrefixNonMmap)) { return true; } if ((skip_mask & kSkipHashIndex) && (option_config == kBlockBasedTableWithPrefixHashIndex || option_config == kBlockBasedTableWithWholeKeyHashIndex)) { return true; } if ((skip_mask & kSkipHashCuckoo) && (option_config == kHashCuckoo)) { return true; } if ((skip_mask & kSkipFIFOCompaction) && option_config == kFIFOCompaction) { return true; } if ((skip_mask & kSkipMmapReads) && option_config == kWalDirAndMmapReads) { return true; } return false; } // Switch to a fresh database with the next option configuration to // test. Return false if there are no more configurations to test. bool DBTestBase::ChangeOptions(int skip_mask) { for (option_config_++; option_config_ < kEnd; option_config_++) { if (ShouldSkipOptions(option_config_, skip_mask)) { continue; } break; } if (option_config_ >= kEnd) { Destroy(last_options_); return false; } else { auto options = CurrentOptions(); options.create_if_missing = true; DestroyAndReopen(options); return true; } } // Switch between different compaction styles. bool DBTestBase::ChangeCompactOptions() { if (option_config_ == kDefault) { option_config_ = kUniversalCompaction; Destroy(last_options_); auto options = CurrentOptions(); options.create_if_missing = true; TryReopen(options); return true; } else if (option_config_ == kUniversalCompaction) { option_config_ = kUniversalCompactionMultiLevel; Destroy(last_options_); auto options = CurrentOptions(); options.create_if_missing = true; TryReopen(options); return true; } else if (option_config_ == kUniversalCompactionMultiLevel) { option_config_ = kLevelSubcompactions; Destroy(last_options_); auto options = CurrentOptions(); assert(options.max_subcompactions > 1); TryReopen(options); return true; } else if (option_config_ == kLevelSubcompactions) { option_config_ = kUniversalSubcompactions; Destroy(last_options_); auto options = CurrentOptions(); assert(options.max_subcompactions > 1); TryReopen(options); return true; } else { return false; } } // Switch between different filter policy // Jump from kDefault to kFilter to kFullFilter bool DBTestBase::ChangeFilterOptions() { if (option_config_ == kDefault) { option_config_ = kFilter; } else if (option_config_ == kFilter) { option_config_ = kFullFilterWithNewTableReaderForCompactions; } else { return false; } Destroy(last_options_); auto options = CurrentOptions(); options.create_if_missing = true; TryReopen(options); return true; } // Return the current option configuration. Options DBTestBase::CurrentOptions( const anon::OptionsOverride& options_override) { Options options; options.write_buffer_size = 4090 * 4096; return CurrentOptions(options, options_override); } Options DBTestBase::CurrentOptions( const Options& defaultOptions, const anon::OptionsOverride& options_override) { // this redundant copy is to minimize code change w/o having lint error. Options options = defaultOptions; XFUNC_TEST("", "dbtest_options", inplace_options1, GetXFTestOptions, reinterpret_cast(&options), options_override.skip_policy); BlockBasedTableOptions table_options; bool set_block_based_table_factory = true; switch (option_config_) { #ifndef ROCKSDB_LITE case kHashSkipList: options.prefix_extractor.reset(NewFixedPrefixTransform(1)); options.memtable_factory.reset(NewHashSkipListRepFactory(16)); break; case kPlainTableFirstBytePrefix: options.table_factory.reset(new PlainTableFactory()); options.prefix_extractor.reset(NewFixedPrefixTransform(1)); options.allow_mmap_reads = true; options.max_sequential_skip_in_iterations = 999999; set_block_based_table_factory = false; break; case kPlainTableCappedPrefix: options.table_factory.reset(new PlainTableFactory()); options.prefix_extractor.reset(NewCappedPrefixTransform(8)); options.allow_mmap_reads = true; options.max_sequential_skip_in_iterations = 999999; set_block_based_table_factory = false; break; case kPlainTableCappedPrefixNonMmap: options.table_factory.reset(new PlainTableFactory()); options.prefix_extractor.reset(NewCappedPrefixTransform(8)); options.allow_mmap_reads = false; options.max_sequential_skip_in_iterations = 999999; set_block_based_table_factory = false; break; case kPlainTableAllBytesPrefix: options.table_factory.reset(new PlainTableFactory()); options.prefix_extractor.reset(NewNoopTransform()); options.allow_mmap_reads = true; options.max_sequential_skip_in_iterations = 999999; set_block_based_table_factory = false; break; case kVectorRep: options.memtable_factory.reset(new VectorRepFactory(100)); break; case kHashLinkList: options.prefix_extractor.reset(NewFixedPrefixTransform(1)); options.memtable_factory.reset( NewHashLinkListRepFactory(4, 0, 3, true, 4)); break; case kHashCuckoo: options.memtable_factory.reset( NewHashCuckooRepFactory(options.write_buffer_size)); break; #endif // ROCKSDB_LITE case kMergePut: options.merge_operator = MergeOperators::CreatePutOperator(); break; case kFilter: table_options.filter_policy.reset(NewBloomFilterPolicy(10, true)); break; case kFullFilterWithNewTableReaderForCompactions: table_options.filter_policy.reset(NewBloomFilterPolicy(10, false)); options.new_table_reader_for_compaction_inputs = true; options.compaction_readahead_size = 10 * 1024 * 1024; break; case kUncompressed: options.compression = kNoCompression; break; case kNumLevel_3: options.num_levels = 3; break; case kDBLogDir: options.db_log_dir = alternative_db_log_dir_; break; case kWalDirAndMmapReads: options.wal_dir = alternative_wal_dir_; // mmap reads should be orthogonal to WalDir setting, so we piggyback to // this option config to test mmap reads as well options.allow_mmap_reads = true; break; case kManifestFileSize: options.max_manifest_file_size = 50; // 50 bytes case kPerfOptions: options.soft_rate_limit = 2.0; options.delayed_write_rate = 8 * 1024 * 1024; // TODO(3.13) -- test more options break; case kDeletesFilterFirst: options.filter_deletes = true; break; case kUniversalCompaction: options.compaction_style = kCompactionStyleUniversal; options.num_levels = 1; break; case kUniversalCompactionMultiLevel: options.compaction_style = kCompactionStyleUniversal; options.num_levels = 8; break; case kCompressedBlockCache: options.allow_mmap_writes = true; table_options.block_cache_compressed = NewLRUCache(8 * 1024 * 1024); break; case kInfiniteMaxOpenFiles: options.max_open_files = -1; break; case kxxHashChecksum: { table_options.checksum = kxxHash; break; } case kFIFOCompaction: { options.compaction_style = kCompactionStyleFIFO; break; } case kBlockBasedTableWithPrefixHashIndex: { table_options.index_type = BlockBasedTableOptions::kHashSearch; options.prefix_extractor.reset(NewFixedPrefixTransform(1)); break; } case kBlockBasedTableWithWholeKeyHashIndex: { table_options.index_type = BlockBasedTableOptions::kHashSearch; options.prefix_extractor.reset(NewNoopTransform()); break; } case kBlockBasedTableWithIndexRestartInterval: { table_options.index_block_restart_interval = 8; break; } case kOptimizeFiltersForHits: { options.optimize_filters_for_hits = true; set_block_based_table_factory = true; break; } case kRowCache: { options.row_cache = NewLRUCache(1024 * 1024); break; } case kRecycleLogFiles: { options.recycle_log_file_num = 2; break; } case kLevelSubcompactions: { options.max_subcompactions = 4; break; } case kUniversalSubcompactions: { options.compaction_style = kCompactionStyleUniversal; options.num_levels = 8; options.max_subcompactions = 4; break; } case kConcurrentSkipList: { options.allow_concurrent_memtable_write = true; options.enable_write_thread_adaptive_yield = true; break; } default: break; } if (options_override.filter_policy) { table_options.filter_policy = options_override.filter_policy; } if (set_block_based_table_factory) { options.table_factory.reset(NewBlockBasedTableFactory(table_options)); } options.env = env_; options.create_if_missing = true; options.fail_if_options_file_error = true; return options; } void DBTestBase::CreateColumnFamilies(const std::vector& cfs, const Options& options) { ColumnFamilyOptions cf_opts(options); size_t cfi = handles_.size(); handles_.resize(cfi + cfs.size()); for (auto cf : cfs) { ASSERT_OK(db_->CreateColumnFamily(cf_opts, cf, &handles_[cfi++])); } } void DBTestBase::CreateAndReopenWithCF(const std::vector& cfs, const Options& options) { CreateColumnFamilies(cfs, options); std::vector cfs_plus_default = cfs; cfs_plus_default.insert(cfs_plus_default.begin(), kDefaultColumnFamilyName); ReopenWithColumnFamilies(cfs_plus_default, options); } void DBTestBase::ReopenWithColumnFamilies(const std::vector& cfs, const std::vector& options) { ASSERT_OK(TryReopenWithColumnFamilies(cfs, options)); } void DBTestBase::ReopenWithColumnFamilies(const std::vector& cfs, const Options& options) { ASSERT_OK(TryReopenWithColumnFamilies(cfs, options)); } Status DBTestBase::TryReopenWithColumnFamilies( const std::vector& cfs, const std::vector& options) { Close(); EXPECT_EQ(cfs.size(), options.size()); std::vector column_families; for (size_t i = 0; i < cfs.size(); ++i) { column_families.push_back(ColumnFamilyDescriptor(cfs[i], options[i])); } DBOptions db_opts = DBOptions(options[0]); return DB::Open(db_opts, dbname_, column_families, &handles_, &db_); } Status DBTestBase::TryReopenWithColumnFamilies( const std::vector& cfs, const Options& options) { Close(); std::vector v_opts(cfs.size(), options); return TryReopenWithColumnFamilies(cfs, v_opts); } void DBTestBase::Reopen(const Options& options) { ASSERT_OK(TryReopen(options)); } void DBTestBase::Close() { for (auto h : handles_) { delete h; } handles_.clear(); delete db_; db_ = nullptr; } void DBTestBase::DestroyAndReopen(const Options& options) { // Destroy using last options Destroy(last_options_); ASSERT_OK(TryReopen(options)); } void DBTestBase::Destroy(const Options& options) { Close(); ASSERT_OK(DestroyDB(dbname_, options)); } Status DBTestBase::ReadOnlyReopen(const Options& options) { return DB::OpenForReadOnly(options, dbname_, &db_); } Status DBTestBase::TryReopen(const Options& options) { Close(); last_options_ = options; return DB::Open(options, dbname_, &db_); } Status DBTestBase::Flush(int cf) { if (cf == 0) { return db_->Flush(FlushOptions()); } else { return db_->Flush(FlushOptions(), handles_[cf]); } } Status DBTestBase::Put(const Slice& k, const Slice& v, WriteOptions wo) { if (kMergePut == option_config_) { return db_->Merge(wo, k, v); } else { return db_->Put(wo, k, v); } } Status DBTestBase::Put(int cf, const Slice& k, const Slice& v, WriteOptions wo) { if (kMergePut == option_config_) { return db_->Merge(wo, handles_[cf], k, v); } else { return db_->Put(wo, handles_[cf], k, v); } } Status DBTestBase::Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); } Status DBTestBase::Delete(int cf, const std::string& k) { return db_->Delete(WriteOptions(), handles_[cf], k); } Status DBTestBase::SingleDelete(const std::string& k) { return db_->SingleDelete(WriteOptions(), k); } Status DBTestBase::SingleDelete(int cf, const std::string& k) { return db_->SingleDelete(WriteOptions(), handles_[cf], k); } std::string DBTestBase::Get(const std::string& k, const Snapshot* snapshot) { ReadOptions options; options.verify_checksums = true; options.snapshot = snapshot; std::string result; Status s = db_->Get(options, k, &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } std::string DBTestBase::Get(int cf, const std::string& k, const Snapshot* snapshot) { ReadOptions options; options.verify_checksums = true; options.snapshot = snapshot; std::string result; Status s = db_->Get(options, handles_[cf], k, &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } uint64_t DBTestBase::GetNumSnapshots() { uint64_t int_num; EXPECT_TRUE(dbfull()->GetIntProperty("rocksdb.num-snapshots", &int_num)); return int_num; } uint64_t DBTestBase::GetTimeOldestSnapshots() { uint64_t int_num; EXPECT_TRUE( dbfull()->GetIntProperty("rocksdb.oldest-snapshot-time", &int_num)); return int_num; } // Return a string that contains all key,value pairs in order, // formatted like "(k1->v1)(k2->v2)". std::string DBTestBase::Contents(int cf) { std::vector forward; std::string result; Iterator* iter = (cf == 0) ? db_->NewIterator(ReadOptions()) : db_->NewIterator(ReadOptions(), handles_[cf]); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { std::string s = IterStatus(iter); result.push_back('('); result.append(s); result.push_back(')'); forward.push_back(s); } // Check reverse iteration results are the reverse of forward results unsigned int matched = 0; for (iter->SeekToLast(); iter->Valid(); iter->Prev()) { EXPECT_LT(matched, forward.size()); EXPECT_EQ(IterStatus(iter), forward[forward.size() - matched - 1]); matched++; } EXPECT_EQ(matched, forward.size()); delete iter; return result; } std::string DBTestBase::AllEntriesFor(const Slice& user_key, int cf) { Arena arena; ScopedArenaIterator iter; if (cf == 0) { iter.set(dbfull()->NewInternalIterator(&arena)); } else { iter.set(dbfull()->NewInternalIterator(&arena, handles_[cf])); } InternalKey target(user_key, kMaxSequenceNumber, kTypeValue); iter->Seek(target.Encode()); std::string result; if (!iter->status().ok()) { result = iter->status().ToString(); } else { result = "[ "; bool first = true; while (iter->Valid()) { ParsedInternalKey ikey(Slice(), 0, kTypeValue); if (!ParseInternalKey(iter->key(), &ikey)) { result += "CORRUPTED"; } else { if (!last_options_.comparator->Equal(ikey.user_key, user_key)) { break; } if (!first) { result += ", "; } first = false; switch (ikey.type) { case kTypeValue: result += iter->value().ToString(); break; case kTypeMerge: // keep it the same as kTypeValue for testing kMergePut result += iter->value().ToString(); break; case kTypeDeletion: result += "DEL"; break; case kTypeSingleDeletion: result += "SDEL"; break; default: assert(false); break; } } iter->Next(); } if (!first) { result += " "; } result += "]"; } return result; } #ifndef ROCKSDB_LITE int DBTestBase::NumSortedRuns(int cf) { ColumnFamilyMetaData cf_meta; if (cf == 0) { db_->GetColumnFamilyMetaData(&cf_meta); } else { db_->GetColumnFamilyMetaData(handles_[cf], &cf_meta); } int num_sr = static_cast(cf_meta.levels[0].files.size()); for (size_t i = 1U; i < cf_meta.levels.size(); i++) { if (cf_meta.levels[i].files.size() > 0) { num_sr++; } } return num_sr; } uint64_t DBTestBase::TotalSize(int cf) { ColumnFamilyMetaData cf_meta; if (cf == 0) { db_->GetColumnFamilyMetaData(&cf_meta); } else { db_->GetColumnFamilyMetaData(handles_[cf], &cf_meta); } return cf_meta.size; } uint64_t DBTestBase::SizeAtLevel(int level) { std::vector metadata; db_->GetLiveFilesMetaData(&metadata); uint64_t sum = 0; for (const auto& m : metadata) { if (m.level == level) { sum += m.size; } } return sum; } size_t DBTestBase::TotalLiveFiles(int cf) { ColumnFamilyMetaData cf_meta; if (cf == 0) { db_->GetColumnFamilyMetaData(&cf_meta); } else { db_->GetColumnFamilyMetaData(handles_[cf], &cf_meta); } size_t num_files = 0; for (auto& level : cf_meta.levels) { num_files += level.files.size(); } return num_files; } size_t DBTestBase::CountLiveFiles() { std::vector metadata; db_->GetLiveFilesMetaData(&metadata); return metadata.size(); } #endif // ROCKSDB_LITE int DBTestBase::NumTableFilesAtLevel(int level, int cf) { std::string property; if (cf == 0) { // default cfd EXPECT_TRUE(db_->GetProperty( "rocksdb.num-files-at-level" + NumberToString(level), &property)); } else { EXPECT_TRUE(db_->GetProperty( handles_[cf], "rocksdb.num-files-at-level" + NumberToString(level), &property)); } return atoi(property.c_str()); } int DBTestBase::TotalTableFiles(int cf, int levels) { if (levels == -1) { levels = CurrentOptions().num_levels; } int result = 0; for (int level = 0; level < levels; level++) { result += NumTableFilesAtLevel(level, cf); } return result; } // Return spread of files per level std::string DBTestBase::FilesPerLevel(int cf) { int num_levels = (cf == 0) ? db_->NumberLevels() : db_->NumberLevels(handles_[1]); std::string result; size_t last_non_zero_offset = 0; for (int level = 0; level < num_levels; level++) { int f = NumTableFilesAtLevel(level, cf); char buf[100]; snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); result += buf; if (f > 0) { last_non_zero_offset = result.size(); } } result.resize(last_non_zero_offset); return result; } size_t DBTestBase::CountFiles() { std::vector files; env_->GetChildren(dbname_, &files); std::vector logfiles; if (dbname_ != last_options_.wal_dir) { env_->GetChildren(last_options_.wal_dir, &logfiles); } return files.size() + logfiles.size(); } uint64_t DBTestBase::Size(const Slice& start, const Slice& limit, int cf) { Range r(start, limit); uint64_t size; if (cf == 0) { db_->GetApproximateSizes(&r, 1, &size); } else { db_->GetApproximateSizes(handles_[1], &r, 1, &size); } return size; } void DBTestBase::Compact(int cf, const Slice& start, const Slice& limit, uint32_t target_path_id) { CompactRangeOptions compact_options; compact_options.target_path_id = target_path_id; ASSERT_OK(db_->CompactRange(compact_options, handles_[cf], &start, &limit)); } void DBTestBase::Compact(int cf, const Slice& start, const Slice& limit) { ASSERT_OK( db_->CompactRange(CompactRangeOptions(), handles_[cf], &start, &limit)); } void DBTestBase::Compact(const Slice& start, const Slice& limit) { ASSERT_OK(db_->CompactRange(CompactRangeOptions(), &start, &limit)); } // Do n memtable compactions, each of which produces an sstable // covering the range [small,large]. void DBTestBase::MakeTables(int n, const std::string& small, const std::string& large, int cf) { for (int i = 0; i < n; i++) { ASSERT_OK(Put(cf, small, "begin")); ASSERT_OK(Put(cf, large, "end")); ASSERT_OK(Flush(cf)); MoveFilesToLevel(n - i - 1, cf); } } // Prevent pushing of new sstables into deeper levels by adding // tables that cover a specified range to all levels. void DBTestBase::FillLevels(const std::string& smallest, const std::string& largest, int cf) { MakeTables(db_->NumberLevels(handles_[cf]), smallest, largest, cf); } void DBTestBase::MoveFilesToLevel(int level, int cf) { for (int l = 0; l < level; ++l) { if (cf > 0) { dbfull()->TEST_CompactRange(l, nullptr, nullptr, handles_[cf]); } else { dbfull()->TEST_CompactRange(l, nullptr, nullptr); } } } void DBTestBase::DumpFileCounts(const char* label) { fprintf(stderr, "---\n%s:\n", label); fprintf(stderr, "maxoverlap: %" PRIu64 "\n", dbfull()->TEST_MaxNextLevelOverlappingBytes()); for (int level = 0; level < db_->NumberLevels(); level++) { int num = NumTableFilesAtLevel(level); if (num > 0) { fprintf(stderr, " level %3d : %d files\n", level, num); } } } std::string DBTestBase::DumpSSTableList() { std::string property; db_->GetProperty("rocksdb.sstables", &property); return property; } void DBTestBase::GetSstFiles(std::string path, std::vector* files) { env_->GetChildren(path, files); files->erase( std::remove_if(files->begin(), files->end(), [](std::string name) { uint64_t number; FileType type; return !(ParseFileName(name, &number, &type) && type == kTableFile); }), files->end()); } int DBTestBase::GetSstFileCount(std::string path) { std::vector files; GetSstFiles(path, &files); return static_cast(files.size()); } // this will generate non-overlapping files since it keeps increasing key_idx void DBTestBase::GenerateNewFile(int cf, Random* rnd, int* key_idx, bool nowait) { for (int i = 0; i < KNumKeysByGenerateNewFile; i++) { ASSERT_OK(Put(cf, Key(*key_idx), RandomString(rnd, (i == 99) ? 1 : 990))); (*key_idx)++; } if (!nowait) { dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } } // this will generate non-overlapping files since it keeps increasing key_idx void DBTestBase::GenerateNewFile(Random* rnd, int* key_idx, bool nowait) { for (int i = 0; i < KNumKeysByGenerateNewFile; i++) { ASSERT_OK(Put(Key(*key_idx), RandomString(rnd, (i == 99) ? 1 : 990))); (*key_idx)++; } if (!nowait) { dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } } const int DBTestBase::kNumKeysByGenerateNewRandomFile = 51; void DBTestBase::GenerateNewRandomFile(Random* rnd, bool nowait) { for (int i = 0; i < kNumKeysByGenerateNewRandomFile; i++) { ASSERT_OK(Put("key" + RandomString(rnd, 7), RandomString(rnd, 2000))); } ASSERT_OK(Put("key" + RandomString(rnd, 7), RandomString(rnd, 200))); if (!nowait) { dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } } std::string DBTestBase::IterStatus(Iterator* iter) { std::string result; if (iter->Valid()) { result = iter->key().ToString() + "->" + iter->value().ToString(); } else { result = "(invalid)"; } return result; } Options DBTestBase::OptionsForLogIterTest() { Options options = CurrentOptions(); options.create_if_missing = true; options.WAL_ttl_seconds = 1000; return options; } std::string DBTestBase::DummyString(size_t len, char c) { return std::string(len, c); } void DBTestBase::VerifyIterLast(std::string expected_key, int cf) { Iterator* iter; ReadOptions ro; if (cf == 0) { iter = db_->NewIterator(ro); } else { iter = db_->NewIterator(ro, handles_[cf]); } iter->SeekToLast(); ASSERT_EQ(IterStatus(iter), expected_key); delete iter; } // Used to test InplaceUpdate // If previous value is nullptr or delta is > than previous value, // sets newValue with delta // If previous value is not empty, // updates previous value with 'b' string of previous value size - 1. UpdateStatus DBTestBase::updateInPlaceSmallerSize(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue) { if (prevValue == nullptr) { *newValue = std::string(delta.size(), 'c'); return UpdateStatus::UPDATED; } else { *prevSize = *prevSize - 1; std::string str_b = std::string(*prevSize, 'b'); memcpy(prevValue, str_b.c_str(), str_b.size()); return UpdateStatus::UPDATED_INPLACE; } } UpdateStatus DBTestBase::updateInPlaceSmallerVarintSize(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue) { if (prevValue == nullptr) { *newValue = std::string(delta.size(), 'c'); return UpdateStatus::UPDATED; } else { *prevSize = 1; std::string str_b = std::string(*prevSize, 'b'); memcpy(prevValue, str_b.c_str(), str_b.size()); return UpdateStatus::UPDATED_INPLACE; } } UpdateStatus DBTestBase::updateInPlaceLargerSize(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue) { *newValue = std::string(delta.size(), 'c'); return UpdateStatus::UPDATED; } UpdateStatus DBTestBase::updateInPlaceNoAction(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue) { return UpdateStatus::UPDATE_FAILED; } // Utility method to test InplaceUpdate void DBTestBase::validateNumberOfEntries(int numValues, int cf) { ScopedArenaIterator iter; Arena arena; if (cf != 0) { iter.set(dbfull()->NewInternalIterator(&arena, handles_[cf])); } else { iter.set(dbfull()->NewInternalIterator(&arena)); } iter->SeekToFirst(); ASSERT_EQ(iter->status().ok(), true); int seq = numValues; while (iter->Valid()) { ParsedInternalKey ikey; ikey.sequence = -1; ASSERT_EQ(ParseInternalKey(iter->key(), &ikey), true); // checks sequence number for updates ASSERT_EQ(ikey.sequence, (unsigned)seq--); iter->Next(); } ASSERT_EQ(0, seq); } void DBTestBase::CopyFile(const std::string& source, const std::string& destination, uint64_t size) { const EnvOptions soptions; unique_ptr srcfile; ASSERT_OK(env_->NewSequentialFile(source, &srcfile, soptions)); unique_ptr destfile; ASSERT_OK(env_->NewWritableFile(destination, &destfile, soptions)); if (size == 0) { // default argument means copy everything ASSERT_OK(env_->GetFileSize(source, &size)); } char buffer[4096]; Slice slice; while (size > 0) { uint64_t one = std::min(uint64_t(sizeof(buffer)), size); ASSERT_OK(srcfile->Read(one, &slice, buffer)); ASSERT_OK(destfile->Append(slice)); size -= slice.size(); } ASSERT_OK(destfile->Close()); } std::unordered_map DBTestBase::GetAllSSTFiles() { std::unordered_map res; std::vector files; env_->GetChildren(dbname_, &files); for (auto& file_name : files) { uint64_t number; FileType type; std::string file_path = dbname_ + "/" + file_name; if (ParseFileName(file_name, &number, &type) && type == kTableFile) { uint64_t file_size = 0; env_->GetFileSize(file_path, &file_size); res[file_path] = file_size; } } return res; } } // namespace rocksdb rocksdb-4.5.1/db/db_test_util.h000066400000000000000000000563441267530634000164110ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #ifndef OS_WIN #include #endif #include #include #include #include #include #include #include #include #include "db/db_impl.h" #include "db/dbformat.h" #include "db/filename.h" #include "memtable/hash_linklist_rep.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/convenience.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/table.h" #include "rocksdb/utilities/checkpoint.h" #include "table/block_based_table_factory.h" #include "table/mock_table.h" #include "table/plain_table_factory.h" #include "table/scoped_arena_iterator.h" #include "util/compression.h" #include "util/mock_env.h" #include "util/mutexlock.h" #include "util/string_util.h" // SyncPoint is not supported in Released Windows Mode. #if !(defined NDEBUG) || !defined(OS_WIN) #include "util/sync_point.h" #endif // !(defined NDEBUG) || !defined(OS_WIN) #include "util/testharness.h" #include "util/testutil.h" #include "util/xfunc.h" #include "utilities/merge_operators.h" namespace rocksdb { namespace anon { class AtomicCounter { public: explicit AtomicCounter(Env* env = NULL) : env_(env), cond_count_(&mu_), count_(0) {} void Increment() { MutexLock l(&mu_); count_++; cond_count_.SignalAll(); } int Read() { MutexLock l(&mu_); return count_; } bool WaitFor(int count) { MutexLock l(&mu_); uint64_t start = env_->NowMicros(); while (count_ < count) { uint64_t now = env_->NowMicros(); cond_count_.TimedWait(now + /*1s*/ 1 * 1000 * 1000); if (env_->NowMicros() - start > /*10s*/ 10 * 1000 * 1000) { return false; } if (count_ < count) { GTEST_LOG_(WARNING) << "WaitFor is taking more time than usual"; } } return true; } void Reset() { MutexLock l(&mu_); count_ = 0; cond_count_.SignalAll(); } private: Env* env_; port::Mutex mu_; port::CondVar cond_count_; int count_; }; struct OptionsOverride { std::shared_ptr filter_policy = nullptr; // Used as a bit mask of individual enums in which to skip an XF test point int skip_policy = 0; }; } // namespace anon // A hacky skip list mem table that triggers flush after number of entries. class SpecialMemTableRep : public MemTableRep { public: explicit SpecialMemTableRep(MemTableAllocator* allocator, MemTableRep* memtable, int num_entries_flush) : MemTableRep(allocator), memtable_(memtable), num_entries_flush_(num_entries_flush), num_entries_(0) {} virtual KeyHandle Allocate(const size_t len, char** buf) override { return memtable_->Allocate(len, buf); } // Insert key into the list. // REQUIRES: nothing that compares equal to key is currently in the list. virtual void Insert(KeyHandle handle) override { memtable_->Insert(handle); num_entries_++; } // Returns true iff an entry that compares equal to key is in the list. virtual bool Contains(const char* key) const override { return memtable_->Contains(key); } virtual size_t ApproximateMemoryUsage() override { // Return a high memory usage when number of entries exceeds the threshold // to trigger a flush. return (num_entries_ < num_entries_flush_) ? 0 : 1024 * 1024 * 1024; } virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) override { memtable_->Get(k, callback_args, callback_func); } uint64_t ApproximateNumEntries(const Slice& start_ikey, const Slice& end_ikey) override { return memtable_->ApproximateNumEntries(start_ikey, end_ikey); } virtual MemTableRep::Iterator* GetIterator(Arena* arena = nullptr) override { return memtable_->GetIterator(arena); } virtual ~SpecialMemTableRep() override {} private: unique_ptr memtable_; int num_entries_flush_; int num_entries_; }; // The factory for the hacky skip list mem table that triggers flush after // number of entries exceeds a threshold. class SpecialSkipListFactory : public MemTableRepFactory { public: // After number of inserts exceeds `num_entries_flush` in a mem table, trigger // flush. explicit SpecialSkipListFactory(int num_entries_flush) : num_entries_flush_(num_entries_flush) {} virtual MemTableRep* CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) override { return new SpecialMemTableRep( allocator, factory_.CreateMemTableRep(compare, allocator, transform, 0), num_entries_flush_); } virtual const char* Name() const override { return "SkipListFactory"; } private: SkipListFactory factory_; int num_entries_flush_; }; // Special Env used to delay background operations class SpecialEnv : public EnvWrapper { public: explicit SpecialEnv(Env* base); Status NewWritableFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { class SSTableFile : public WritableFile { private: SpecialEnv* env_; unique_ptr base_; public: SSTableFile(SpecialEnv* env, unique_ptr&& base) : env_(env), base_(std::move(base)) {} Status Append(const Slice& data) override { if (env_->table_write_callback_) { (*env_->table_write_callback_)(); } if (env_->drop_writes_.load(std::memory_order_acquire)) { // Drop writes on the floor return Status::OK(); } else if (env_->no_space_.load(std::memory_order_acquire)) { return Status::IOError("No space left on device"); } else { env_->bytes_written_ += data.size(); return base_->Append(data); } } Status Truncate(uint64_t size) override { return base_->Truncate(size); } Status Close() override { // SyncPoint is not supported in Released Windows Mode. #if !(defined NDEBUG) || !defined(OS_WIN) // Check preallocation size // preallocation size is never passed to base file. size_t preallocation_size = preallocation_block_size(); TEST_SYNC_POINT_CALLBACK("DBTestWritableFile.GetPreallocationStatus", &preallocation_size); #endif // !(defined NDEBUG) || !defined(OS_WIN) return base_->Close(); } Status Flush() override { return base_->Flush(); } Status Sync() override { ++env_->sync_counter_; while (env_->delay_sstable_sync_.load(std::memory_order_acquire)) { env_->SleepForMicroseconds(100000); } return base_->Sync(); } void SetIOPriority(Env::IOPriority pri) override { base_->SetIOPriority(pri); } Env::IOPriority GetIOPriority() override { return base_->GetIOPriority(); } }; class ManifestFile : public WritableFile { public: ManifestFile(SpecialEnv* env, unique_ptr&& b) : env_(env), base_(std::move(b)) {} Status Append(const Slice& data) override { if (env_->manifest_write_error_.load(std::memory_order_acquire)) { return Status::IOError("simulated writer error"); } else { return base_->Append(data); } } Status Truncate(uint64_t size) override { return base_->Truncate(size); } Status Close() override { return base_->Close(); } Status Flush() override { return base_->Flush(); } Status Sync() override { ++env_->sync_counter_; if (env_->manifest_sync_error_.load(std::memory_order_acquire)) { return Status::IOError("simulated sync error"); } else { return base_->Sync(); } } uint64_t GetFileSize() override { return base_->GetFileSize(); } private: SpecialEnv* env_; unique_ptr base_; }; class WalFile : public WritableFile { public: WalFile(SpecialEnv* env, unique_ptr&& b) : env_(env), base_(std::move(b)) {} Status Append(const Slice& data) override { #if !(defined NDEBUG) || !defined(OS_WIN) TEST_SYNC_POINT("SpecialEnv::WalFile::Append:1"); #endif Status s; if (env_->log_write_error_.load(std::memory_order_acquire)) { s = Status::IOError("simulated writer error"); } else { int slowdown = env_->log_write_slowdown_.load(std::memory_order_acquire); if (slowdown > 0) { env_->SleepForMicroseconds(slowdown); } s = base_->Append(data); } #if !(defined NDEBUG) || !defined(OS_WIN) TEST_SYNC_POINT("SpecialEnv::WalFile::Append:2"); #endif return s; } Status Truncate(uint64_t size) override { return base_->Truncate(size); } Status Close() override { return base_->Close(); } Status Flush() override { return base_->Flush(); } Status Sync() override { ++env_->sync_counter_; return base_->Sync(); } bool IsSyncThreadSafe() const override { return env_->is_wal_sync_thread_safe_.load(); } private: SpecialEnv* env_; unique_ptr base_; }; if (non_writeable_rate_.load(std::memory_order_acquire) > 0) { uint32_t random_number; { MutexLock l(&rnd_mutex_); random_number = rnd_.Uniform(100); } if (random_number < non_writeable_rate_.load()) { return Status::IOError("simulated random write error"); } } new_writable_count_++; if (non_writable_count_.load() > 0) { non_writable_count_--; return Status::IOError("simulated write error"); } Status s = target()->NewWritableFile(f, r, soptions); if (s.ok()) { if (strstr(f.c_str(), ".sst") != nullptr) { r->reset(new SSTableFile(this, std::move(*r))); } else if (strstr(f.c_str(), "MANIFEST") != nullptr) { r->reset(new ManifestFile(this, std::move(*r))); } else if (strstr(f.c_str(), "log") != nullptr) { r->reset(new WalFile(this, std::move(*r))); } } return s; } Status NewRandomAccessFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { class CountingFile : public RandomAccessFile { public: CountingFile(unique_ptr&& target, anon::AtomicCounter* counter) : target_(std::move(target)), counter_(counter) {} virtual Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const override { counter_->Increment(); return target_->Read(offset, n, result, scratch); } private: unique_ptr target_; anon::AtomicCounter* counter_; }; Status s = target()->NewRandomAccessFile(f, r, soptions); random_file_open_counter_++; if (s.ok() && count_random_reads_) { r->reset(new CountingFile(std::move(*r), &random_read_counter_)); } return s; } Status NewSequentialFile(const std::string& f, unique_ptr* r, const EnvOptions& soptions) override { class CountingFile : public SequentialFile { public: CountingFile(unique_ptr&& target, anon::AtomicCounter* counter) : target_(std::move(target)), counter_(counter) {} virtual Status Read(size_t n, Slice* result, char* scratch) override { counter_->Increment(); return target_->Read(n, result, scratch); } virtual Status Skip(uint64_t n) override { return target_->Skip(n); } private: unique_ptr target_; anon::AtomicCounter* counter_; }; Status s = target()->NewSequentialFile(f, r, soptions); if (s.ok() && count_sequential_reads_) { r->reset(new CountingFile(std::move(*r), &sequential_read_counter_)); } return s; } virtual void SleepForMicroseconds(int micros) override { sleep_counter_.Increment(); if (no_sleep_ || time_elapse_only_sleep_) { addon_time_.fetch_add(micros); } if (!no_sleep_) { target()->SleepForMicroseconds(micros); } } virtual Status GetCurrentTime(int64_t* unix_time) override { Status s; if (!time_elapse_only_sleep_) { s = target()->GetCurrentTime(unix_time); } if (s.ok()) { *unix_time += addon_time_.load(); } return s; } virtual uint64_t NowNanos() override { return (time_elapse_only_sleep_ ? 0 : target()->NowNanos()) + addon_time_.load() * 1000; } virtual uint64_t NowMicros() override { return (time_elapse_only_sleep_ ? 0 : target()->NowMicros()) + addon_time_.load(); } Random rnd_; port::Mutex rnd_mutex_; // Lock to pretect rnd_ // sstable Sync() calls are blocked while this pointer is non-nullptr. std::atomic delay_sstable_sync_; // Drop writes on the floor while this pointer is non-nullptr. std::atomic drop_writes_; // Simulate no-space errors while this pointer is non-nullptr. std::atomic no_space_; // Simulate non-writable file system while this pointer is non-nullptr std::atomic non_writable_; // Force sync of manifest files to fail while this pointer is non-nullptr std::atomic manifest_sync_error_; // Force write to manifest files to fail while this pointer is non-nullptr std::atomic manifest_write_error_; // Force write to log files to fail while this pointer is non-nullptr std::atomic log_write_error_; // Slow down every log write, in micro-seconds. std::atomic log_write_slowdown_; bool count_random_reads_; anon::AtomicCounter random_read_counter_; std::atomic random_file_open_counter_; bool count_sequential_reads_; anon::AtomicCounter sequential_read_counter_; anon::AtomicCounter sleep_counter_; std::atomic bytes_written_; std::atomic sync_counter_; std::atomic non_writeable_rate_; std::atomic new_writable_count_; std::atomic non_writable_count_; std::function* table_write_callback_; std::atomic addon_time_; bool time_elapse_only_sleep_; bool no_sleep_; std::atomic is_wal_sync_thread_safe_{true}; }; class DBTestBase : public testing::Test { protected: // Sequence of option configurations to try enum OptionConfig { kDefault = 0, kBlockBasedTableWithPrefixHashIndex = 1, kBlockBasedTableWithWholeKeyHashIndex = 2, kPlainTableFirstBytePrefix = 3, kPlainTableCappedPrefix = 4, kPlainTableCappedPrefixNonMmap = 5, kPlainTableAllBytesPrefix = 6, kVectorRep = 7, kHashLinkList = 8, kHashCuckoo = 9, kMergePut = 10, kFilter = 11, kFullFilterWithNewTableReaderForCompactions = 12, kUncompressed = 13, kNumLevel_3 = 14, kDBLogDir = 15, kWalDirAndMmapReads = 16, kManifestFileSize = 17, kPerfOptions = 18, kDeletesFilterFirst = 19, kHashSkipList = 20, kUniversalCompaction = 21, kUniversalCompactionMultiLevel = 22, kCompressedBlockCache = 23, kInfiniteMaxOpenFiles = 24, kxxHashChecksum = 25, kFIFOCompaction = 26, kOptimizeFiltersForHits = 27, kRowCache = 28, kRecycleLogFiles = 29, kConcurrentSkipList = 30, kEnd = 31, kLevelSubcompactions = 31, kUniversalSubcompactions = 32, kBlockBasedTableWithIndexRestartInterval = 33, }; int option_config_; public: std::string dbname_; std::string alternative_wal_dir_; std::string alternative_db_log_dir_; MockEnv* mem_env_; SpecialEnv* env_; DB* db_; std::vector handles_; Options last_options_; // Skip some options, as they may not be applicable to a specific test. // To add more skip constants, use values 4, 8, 16, etc. enum OptionSkip { kNoSkip = 0, kSkipDeletesFilterFirst = 1, kSkipUniversalCompaction = 2, kSkipMergePut = 4, kSkipPlainTable = 8, kSkipHashIndex = 16, kSkipNoSeekToLast = 32, kSkipHashCuckoo = 64, kSkipFIFOCompaction = 128, kSkipMmapReads = 256, }; explicit DBTestBase(const std::string path); ~DBTestBase(); static std::string RandomString(Random* rnd, int len) { std::string r; test::RandomString(rnd, len, &r); return r; } static std::string Key(int i) { char buf[100]; snprintf(buf, sizeof(buf), "key%06d", i); return std::string(buf); } static bool ShouldSkipOptions(int option_config, int skip_mask = kNoSkip); // Switch to a fresh database with the next option configuration to // test. Return false if there are no more configurations to test. bool ChangeOptions(int skip_mask = kNoSkip); // Switch between different compaction styles (we have only 2 now). bool ChangeCompactOptions(); // Switch between different filter policy // Jump from kDefault to kFilter to kFullFilter bool ChangeFilterOptions(); // Return the current option configuration. Options CurrentOptions( const anon::OptionsOverride& options_override = anon::OptionsOverride()); Options CurrentOptions( const Options& defaultOptions, const anon::OptionsOverride& options_override = anon::OptionsOverride()); DBImpl* dbfull() { return reinterpret_cast(db_); } void CreateColumnFamilies(const std::vector& cfs, const Options& options); void CreateAndReopenWithCF(const std::vector& cfs, const Options& options); void ReopenWithColumnFamilies(const std::vector& cfs, const std::vector& options); void ReopenWithColumnFamilies(const std::vector& cfs, const Options& options); Status TryReopenWithColumnFamilies(const std::vector& cfs, const std::vector& options); Status TryReopenWithColumnFamilies(const std::vector& cfs, const Options& options); void Reopen(const Options& options); void Close(); void DestroyAndReopen(const Options& options); void Destroy(const Options& options); Status ReadOnlyReopen(const Options& options); Status TryReopen(const Options& options); Status Flush(int cf = 0); Status Put(const Slice& k, const Slice& v, WriteOptions wo = WriteOptions()); Status Put(int cf, const Slice& k, const Slice& v, WriteOptions wo = WriteOptions()); Status Delete(const std::string& k); Status Delete(int cf, const std::string& k); Status SingleDelete(const std::string& k); Status SingleDelete(int cf, const std::string& k); std::string Get(const std::string& k, const Snapshot* snapshot = nullptr); std::string Get(int cf, const std::string& k, const Snapshot* snapshot = nullptr); uint64_t GetNumSnapshots(); uint64_t GetTimeOldestSnapshots(); // Return a string that contains all key,value pairs in order, // formatted like "(k1->v1)(k2->v2)". std::string Contents(int cf = 0); std::string AllEntriesFor(const Slice& user_key, int cf = 0); #ifndef ROCKSDB_LITE int NumSortedRuns(int cf = 0); uint64_t TotalSize(int cf = 0); uint64_t SizeAtLevel(int level); size_t TotalLiveFiles(int cf = 0); size_t CountLiveFiles(); #endif // ROCKSDB_LITE int NumTableFilesAtLevel(int level, int cf = 0); int TotalTableFiles(int cf = 0, int levels = -1); // Return spread of files per level std::string FilesPerLevel(int cf = 0); size_t CountFiles(); uint64_t Size(const Slice& start, const Slice& limit, int cf = 0); void Compact(int cf, const Slice& start, const Slice& limit, uint32_t target_path_id); void Compact(int cf, const Slice& start, const Slice& limit); void Compact(const Slice& start, const Slice& limit); // Do n memtable compactions, each of which produces an sstable // covering the range [small,large]. void MakeTables(int n, const std::string& small, const std::string& large, int cf = 0); // Prevent pushing of new sstables into deeper levels by adding // tables that cover a specified range to all levels. void FillLevels(const std::string& smallest, const std::string& largest, int cf); void MoveFilesToLevel(int level, int cf = 0); void DumpFileCounts(const char* label); std::string DumpSSTableList(); void GetSstFiles(std::string path, std::vector* files); int GetSstFileCount(std::string path); // this will generate non-overlapping files since it keeps increasing key_idx void GenerateNewFile(Random* rnd, int* key_idx, bool nowait = false); void GenerateNewFile(int fd, Random* rnd, int* key_idx, bool nowait = false); static const int kNumKeysByGenerateNewRandomFile; static const int KNumKeysByGenerateNewFile = 100; void GenerateNewRandomFile(Random* rnd, bool nowait = false); std::string IterStatus(Iterator* iter); Options OptionsForLogIterTest(); std::string DummyString(size_t len, char c = 'a'); void VerifyIterLast(std::string expected_key, int cf = 0); // Used to test InplaceUpdate // If previous value is nullptr or delta is > than previous value, // sets newValue with delta // If previous value is not empty, // updates previous value with 'b' string of previous value size - 1. static UpdateStatus updateInPlaceSmallerSize(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue); static UpdateStatus updateInPlaceSmallerVarintSize(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue); static UpdateStatus updateInPlaceLargerSize(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue); static UpdateStatus updateInPlaceNoAction(char* prevValue, uint32_t* prevSize, Slice delta, std::string* newValue); // Utility method to test InplaceUpdate void validateNumberOfEntries(int numValues, int cf = 0); void CopyFile(const std::string& source, const std::string& destination, uint64_t size = 0); std::unordered_map GetAllSSTFiles(); }; } // namespace rocksdb rocksdb-4.5.1/db/db_universal_compaction_test.cc000066400000000000000000001342501267530634000220070ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_test_util.h" #include "port/stack_trace.h" #if !defined(ROCKSDB_LITE) #include "util/sync_point.h" namespace rocksdb { static uint64_t TestGetTickerCount(const Options& options, Tickers ticker_type) { return options.statistics->getTickerCount(ticker_type); } static std::string CompressibleString(Random* rnd, int len) { std::string r; test::CompressibleString(rnd, 0.8, len, &r); return r; } class DBTestUniversalCompactionBase : public DBTestBase, public ::testing::WithParamInterface> { public: explicit DBTestUniversalCompactionBase( const std::string& path) : DBTestBase(path) {} virtual void SetUp() override { num_levels_ = std::get<0>(GetParam()); exclusive_manual_compaction_ = std::get<1>(GetParam()); } int num_levels_; bool exclusive_manual_compaction_; }; class DBTestUniversalCompaction : public DBTestUniversalCompactionBase { public: DBTestUniversalCompaction() : DBTestUniversalCompactionBase("/db_universal_compaction_test") {} }; namespace { void VerifyCompactionResult( const ColumnFamilyMetaData& cf_meta, const std::set& overlapping_file_numbers) { #ifndef NDEBUG for (auto& level : cf_meta.levels) { for (auto& file : level.files) { assert(overlapping_file_numbers.find(file.name) == overlapping_file_numbers.end()); } } #endif } class KeepFilter : public CompactionFilter { public: virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { return false; } virtual const char* Name() const override { return "KeepFilter"; } }; class KeepFilterFactory : public CompactionFilterFactory { public: explicit KeepFilterFactory(bool check_context = false) : check_context_(check_context) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { if (check_context_) { EXPECT_EQ(expect_full_compaction_.load(), context.is_full_compaction); EXPECT_EQ(expect_manual_compaction_.load(), context.is_manual_compaction); } return std::unique_ptr(new KeepFilter()); } virtual const char* Name() const override { return "KeepFilterFactory"; } bool check_context_; std::atomic_bool expect_full_compaction_; std::atomic_bool expect_manual_compaction_; }; class DelayFilter : public CompactionFilter { public: explicit DelayFilter(DBTestBase* d) : db_test(d) {} virtual bool Filter(int level, const Slice& key, const Slice& value, std::string* new_value, bool* value_changed) const override { db_test->env_->addon_time_.fetch_add(1000); return true; } virtual const char* Name() const override { return "DelayFilter"; } private: DBTestBase* db_test; }; class DelayFilterFactory : public CompactionFilterFactory { public: explicit DelayFilterFactory(DBTestBase* d) : db_test(d) {} virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) override { return std::unique_ptr(new DelayFilter(db_test)); } virtual const char* Name() const override { return "DelayFilterFactory"; } private: DBTestBase* db_test; }; } // namespace // Make sure we don't trigger a problem if the trigger conditon is given // to be 0, which is invalid. TEST_P(DBTestUniversalCompaction, UniversalCompactionSingleSortedRun) { Options options; options = CurrentOptions(options); options.compaction_style = kCompactionStyleUniversal; options.num_levels = num_levels_; // Config universal compaction to always compact to one single sorted run. options.level0_file_num_compaction_trigger = 0; options.compaction_options_universal.size_ratio = 10; options.compaction_options_universal.min_merge_width = 2; options.compaction_options_universal.max_size_amplification_percent = 1; options.write_buffer_size = 105 << 10; // 105KB options.arena_block_size = 4 << 10; options.target_file_size_base = 32 << 10; // 32KB // trigger compaction if there are >= 4 files KeepFilterFactory* filter = new KeepFilterFactory(true); filter->expect_manual_compaction_.store(false); options.compaction_filter_factory.reset(filter); DestroyAndReopen(options); ASSERT_EQ(1, db_->GetOptions().level0_file_num_compaction_trigger); Random rnd(301); int key_idx = 0; filter->expect_full_compaction_.store(true); for (int num = 0; num < 16; num++) { // Write 100KB file. And immediately it should be compacted to one file. GenerateNewFile(&rnd, &key_idx); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumSortedRuns(0), 1); } } TEST_P(DBTestUniversalCompaction, OptimizeFiltersForHits) { Options options; options = CurrentOptions(options); options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.size_ratio = 5; options.num_levels = num_levels_; options.write_buffer_size = 105 << 10; // 105KB options.arena_block_size = 4 << 10; options.target_file_size_base = 32 << 10; // 32KB // trigger compaction if there are >= 4 files options.level0_file_num_compaction_trigger = 4; BlockBasedTableOptions bbto; bbto.cache_index_and_filter_blocks = true; bbto.filter_policy.reset(NewBloomFilterPolicy(10, false)); bbto.whole_key_filtering = true; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); options.optimize_filters_for_hits = true; options.statistics = rocksdb::CreateDBStatistics(); options.memtable_factory.reset(new SpecialSkipListFactory(3)); DestroyAndReopen(options); // block compaction from happening env_->SetBackgroundThreads(1, Env::LOW); test::SleepingBackgroundTask sleeping_task_low; env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::LOW); for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) { Put(Key(num * 10), "val"); if (num) { dbfull()->TEST_WaitForFlushMemTable(); } Put(Key(30 + num * 10), "val"); Put(Key(60 + num * 10), "val"); } Put("", ""); dbfull()->TEST_WaitForFlushMemTable(); // Query set of non existing keys for (int i = 5; i < 90; i += 10) { ASSERT_EQ(Get(Key(i)), "NOT_FOUND"); } // Make sure bloom filter is used at least once. ASSERT_GT(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0); auto prev_counter = TestGetTickerCount(options, BLOOM_FILTER_USEFUL); // Make sure bloom filter is used for all but the last L0 file when looking // up a non-existent key that's in the range of all L0 files. ASSERT_EQ(Get(Key(35)), "NOT_FOUND"); ASSERT_EQ(prev_counter + NumTableFilesAtLevel(0) - 1, TestGetTickerCount(options, BLOOM_FILTER_USEFUL)); prev_counter = TestGetTickerCount(options, BLOOM_FILTER_USEFUL); // Unblock compaction and wait it for happening. sleeping_task_low.WakeUp(); dbfull()->TEST_WaitForCompact(); // The same queries will not trigger bloom filter for (int i = 5; i < 90; i += 10) { ASSERT_EQ(Get(Key(i)), "NOT_FOUND"); } ASSERT_EQ(prev_counter, TestGetTickerCount(options, BLOOM_FILTER_USEFUL)); } // TODO(kailiu) The tests on UniversalCompaction has some issues: // 1. A lot of magic numbers ("11" or "12"). // 2. Made assumption on the memtable flush conditions, which may change from // time to time. TEST_P(DBTestUniversalCompaction, UniversalCompactionTrigger) { Options options; options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.size_ratio = 5; options.num_levels = num_levels_; options.write_buffer_size = 105 << 10; // 105KB options.arena_block_size = 4 << 10; options.target_file_size_base = 32 << 10; // 32KB // trigger compaction if there are >= 4 files options.level0_file_num_compaction_trigger = 4; KeepFilterFactory* filter = new KeepFilterFactory(true); filter->expect_manual_compaction_.store(false); options.compaction_filter_factory.reset(filter); options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBTestWritableFile.GetPreallocationStatus", [&](void* arg) { ASSERT_TRUE(arg != nullptr); size_t preallocation_size = *(static_cast(arg)); if (num_levels_ > 3) { ASSERT_LE(preallocation_size, options.target_file_size_base * 1.1); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Random rnd(301); int key_idx = 0; filter->expect_full_compaction_.store(true); // Stage 1: // Generate a set of files at level 0, but don't trigger level-0 // compaction. for (int num = 0; num < options.level0_file_num_compaction_trigger - 1; num++) { // Write 100KB GenerateNewFile(1, &rnd, &key_idx); } // Generate one more file at level-0, which should trigger level-0 // compaction. GenerateNewFile(1, &rnd, &key_idx); // Suppose each file flushed from mem table has size 1. Now we compact // (level0_file_num_compaction_trigger+1)=4 files and should have a big // file of size 4. ASSERT_EQ(NumSortedRuns(1), 1); // Stage 2: // Now we have one file at level 0, with size 4. We also have some data in // mem table. Let's continue generating new files at level 0, but don't // trigger level-0 compaction. // First, clean up memtable before inserting new data. This will generate // a level-0 file, with size around 0.4 (according to previously written // data amount). filter->expect_full_compaction_.store(false); ASSERT_OK(Flush(1)); for (int num = 0; num < options.level0_file_num_compaction_trigger - 3; num++) { GenerateNewFile(1, &rnd, &key_idx); ASSERT_EQ(NumSortedRuns(1), num + 3); } // Generate one more file at level-0, which should trigger level-0 // compaction. GenerateNewFile(1, &rnd, &key_idx); // Before compaction, we have 4 files at level 0, with size 4, 0.4, 1, 1. // After compaction, we should have 2 files, with size 4, 2.4. ASSERT_EQ(NumSortedRuns(1), 2); // Stage 3: // Now we have 2 files at level 0, with size 4 and 2.4. Continue // generating new files at level 0. for (int num = 0; num < options.level0_file_num_compaction_trigger - 3; num++) { GenerateNewFile(1, &rnd, &key_idx); ASSERT_EQ(NumSortedRuns(1), num + 3); } // Generate one more file at level-0, which should trigger level-0 // compaction. GenerateNewFile(1, &rnd, &key_idx); // Before compaction, we have 4 files at level 0, with size 4, 2.4, 1, 1. // After compaction, we should have 3 files, with size 4, 2.4, 2. ASSERT_EQ(NumSortedRuns(1), 3); // Stage 4: // Now we have 3 files at level 0, with size 4, 2.4, 2. Let's generate a // new file of size 1. GenerateNewFile(1, &rnd, &key_idx); dbfull()->TEST_WaitForCompact(); // Level-0 compaction is triggered, but no file will be picked up. ASSERT_EQ(NumSortedRuns(1), 4); // Stage 5: // Now we have 4 files at level 0, with size 4, 2.4, 2, 1. Let's generate // a new file of size 1. filter->expect_full_compaction_.store(true); GenerateNewFile(1, &rnd, &key_idx); dbfull()->TEST_WaitForCompact(); // All files at level 0 will be compacted into a single one. ASSERT_EQ(NumSortedRuns(1), 1); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_P(DBTestUniversalCompaction, UniversalCompactionSizeAmplification) { Options options; options.compaction_style = kCompactionStyleUniversal; options.num_levels = num_levels_; options.write_buffer_size = 100 << 10; // 100KB options.target_file_size_base = 32 << 10; // 32KB options.level0_file_num_compaction_trigger = 3; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Trigger compaction if size amplification exceeds 110% options.compaction_options_universal.max_size_amplification_percent = 110; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); Random rnd(301); int key_idx = 0; // Generate two files in Level 0. Both files are approx the same size. for (int num = 0; num < options.level0_file_num_compaction_trigger - 1; num++) { // Write 110KB (11 values, each 10K) for (int i = 0; i < 11; i++) { ASSERT_OK(Put(1, Key(key_idx), RandomString(&rnd, 10000))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(handles_[1]); ASSERT_EQ(NumSortedRuns(1), num + 1); } ASSERT_EQ(NumSortedRuns(1), 2); // Flush whatever is remaining in memtable. This is typically // small, which should not trigger size ratio based compaction // but will instead trigger size amplification. ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); // Verify that size amplification did occur ASSERT_EQ(NumSortedRuns(1), 1); } TEST_P(DBTestUniversalCompaction, CompactFilesOnUniversalCompaction) { const int kTestKeySize = 16; const int kTestValueSize = 984; const int kEntrySize = kTestKeySize + kTestValueSize; const int kEntriesPerBuffer = 10; ChangeCompactOptions(); Options options; options.create_if_missing = true; options.write_buffer_size = kEntrySize * kEntriesPerBuffer; options.compaction_style = kCompactionStyleLevel; options.num_levels = 1; options.target_file_size_base = options.write_buffer_size; options.compression = kNoCompression; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); ASSERT_EQ(options.compaction_style, kCompactionStyleUniversal); Random rnd(301); for (int key = 1024 * kEntriesPerBuffer; key >= 0; --key) { ASSERT_OK(Put(1, ToString(key), RandomString(&rnd, kTestValueSize))); } dbfull()->TEST_WaitForFlushMemTable(handles_[1]); dbfull()->TEST_WaitForCompact(); ColumnFamilyMetaData cf_meta; dbfull()->GetColumnFamilyMetaData(handles_[1], &cf_meta); std::vector compaction_input_file_names; for (auto file : cf_meta.levels[0].files) { if (rnd.OneIn(2)) { compaction_input_file_names.push_back(file.name); } } if (compaction_input_file_names.size() == 0) { compaction_input_file_names.push_back( cf_meta.levels[0].files[0].name); } // expect fail since universal compaction only allow L0 output ASSERT_FALSE(dbfull() ->CompactFiles(CompactionOptions(), handles_[1], compaction_input_file_names, 1) .ok()); // expect ok and verify the compacted files no longer exist. ASSERT_OK(dbfull()->CompactFiles( CompactionOptions(), handles_[1], compaction_input_file_names, 0)); dbfull()->GetColumnFamilyMetaData(handles_[1], &cf_meta); VerifyCompactionResult( cf_meta, std::set(compaction_input_file_names.begin(), compaction_input_file_names.end())); compaction_input_file_names.clear(); // Pick the first and the last file, expect everything is // compacted into one single file. compaction_input_file_names.push_back( cf_meta.levels[0].files[0].name); compaction_input_file_names.push_back( cf_meta.levels[0].files[ cf_meta.levels[0].files.size() - 1].name); ASSERT_OK(dbfull()->CompactFiles( CompactionOptions(), handles_[1], compaction_input_file_names, 0)); dbfull()->GetColumnFamilyMetaData(handles_[1], &cf_meta); ASSERT_EQ(cf_meta.levels[0].files.size(), 1U); } TEST_P(DBTestUniversalCompaction, UniversalCompactionTargetLevel) { Options options; options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 100 << 10; // 100KB options.num_levels = 7; options.disable_auto_compactions = true; options = CurrentOptions(options); DestroyAndReopen(options); // Generate 3 overlapping files Random rnd(301); for (int i = 0; i < 210; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 100))); } ASSERT_OK(Flush()); for (int i = 200; i < 300; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 100))); } ASSERT_OK(Flush()); for (int i = 250; i < 260; i++) { ASSERT_OK(Put(Key(i), RandomString(&rnd, 100))); } ASSERT_OK(Flush()); ASSERT_EQ("3", FilesPerLevel(0)); // Compact all files into 1 file and put it in L4 CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 4; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; db_->CompactRange(compact_options, nullptr, nullptr); ASSERT_EQ("0,0,0,0,1", FilesPerLevel(0)); } class DBTestUniversalCompactionMultiLevels : public DBTestUniversalCompactionBase { public: DBTestUniversalCompactionMultiLevels() : DBTestUniversalCompactionBase( "/db_universal_compaction_multi_levels_test") {} }; TEST_P(DBTestUniversalCompactionMultiLevels, UniversalCompactionMultiLevels) { Options options; options.compaction_style = kCompactionStyleUniversal; options.num_levels = num_levels_; options.write_buffer_size = 100 << 10; // 100KB options.level0_file_num_compaction_trigger = 8; options.max_background_compactions = 3; options.target_file_size_base = 32 * 1024; options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); // Trigger compaction if size amplification exceeds 110% options.compaction_options_universal.max_size_amplification_percent = 110; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); Random rnd(301); int num_keys = 100000; for (int i = 0; i < num_keys * 2; i++) { ASSERT_OK(Put(1, Key(i % num_keys), Key(i))); } dbfull()->TEST_WaitForCompact(); for (int i = num_keys; i < num_keys * 2; i++) { ASSERT_EQ(Get(1, Key(i % num_keys)), Key(i)); } } // Tests universal compaction with trivial move enabled TEST_P(DBTestUniversalCompactionMultiLevels, UniversalCompactionTrivialMove) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; ASSERT_TRUE(arg != nullptr); int output_level = *(static_cast(arg)); ASSERT_EQ(output_level, 0); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options; options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.allow_trivial_move = true; options.num_levels = 3; options.write_buffer_size = 100 << 10; // 100KB options.level0_file_num_compaction_trigger = 3; options.max_background_compactions = 2; options.target_file_size_base = 32 * 1024; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Trigger compaction if size amplification exceeds 110% options.compaction_options_universal.max_size_amplification_percent = 110; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); Random rnd(301); int num_keys = 150000; for (int i = 0; i < num_keys; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } std::vector values; ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); ASSERT_GT(trivial_move, 0); ASSERT_GT(non_trivial_move, 0); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } INSTANTIATE_TEST_CASE_P(DBTestUniversalCompactionMultiLevels, DBTestUniversalCompactionMultiLevels, ::testing::Combine(::testing::Values(3, 20), ::testing::Bool())); class DBTestUniversalCompactionParallel : public DBTestUniversalCompactionBase { public: DBTestUniversalCompactionParallel() : DBTestUniversalCompactionBase( "/db_universal_compaction_prallel_test") {} }; TEST_P(DBTestUniversalCompactionParallel, UniversalCompactionParallel) { Options options; options.compaction_style = kCompactionStyleUniversal; options.num_levels = num_levels_; options.write_buffer_size = 1 << 10; // 1KB options.level0_file_num_compaction_trigger = 3; options.max_background_compactions = 3; options.max_background_flushes = 3; options.target_file_size_base = 1 * 1024; options.compaction_options_universal.max_size_amplification_percent = 110; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Delay every compaction so multiple compactions will happen. std::atomic num_compactions_running(0); std::atomic has_parallel(false); rocksdb::SyncPoint::GetInstance()->SetCallBack("CompactionJob::Run():Start", [&](void* arg) { if (num_compactions_running.fetch_add(1) > 0) { has_parallel.store(true); return; } for (int nwait = 0; nwait < 20000; nwait++) { if (has_parallel.load() || num_compactions_running.load() > 1) { has_parallel.store(true); break; } env_->SleepForMicroseconds(1000); } }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "CompactionJob::Run():End", [&](void* arg) { num_compactions_running.fetch_add(-1); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); Random rnd(301); int num_keys = 30000; for (int i = 0; i < num_keys * 2; i++) { ASSERT_OK(Put(1, Key(i % num_keys), Key(i))); } dbfull()->TEST_WaitForCompact(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); ASSERT_EQ(num_compactions_running.load(), 0); ASSERT_TRUE(has_parallel.load()); for (int i = num_keys; i < num_keys * 2; i++) { ASSERT_EQ(Get(1, Key(i % num_keys)), Key(i)); } // Reopen and check. ReopenWithColumnFamilies({"default", "pikachu"}, options); for (int i = num_keys; i < num_keys * 2; i++) { ASSERT_EQ(Get(1, Key(i % num_keys)), Key(i)); } } INSTANTIATE_TEST_CASE_P(DBTestUniversalCompactionParallel, DBTestUniversalCompactionParallel, ::testing::Combine(::testing::Values(1, 10), ::testing::Bool())); TEST_P(DBTestUniversalCompaction, UniversalCompactionOptions) { Options options; options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 105 << 10; // 105KB options.arena_block_size = 4 << 10; // 4KB options.target_file_size_base = 32 << 10; // 32KB options.level0_file_num_compaction_trigger = 4; options.num_levels = num_levels_; options.compaction_options_universal.compression_size_percent = -1; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); Random rnd(301); int key_idx = 0; for (int num = 0; num < options.level0_file_num_compaction_trigger; num++) { // Write 100KB (100 values, each 1K) for (int i = 0; i < 100; i++) { ASSERT_OK(Put(1, Key(key_idx), RandomString(&rnd, 990))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(handles_[1]); if (num < options.level0_file_num_compaction_trigger - 1) { ASSERT_EQ(NumSortedRuns(1), num + 1); } } dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumSortedRuns(1), 1); } TEST_P(DBTestUniversalCompaction, UniversalCompactionStopStyleSimilarSize) { Options options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 105 << 10; // 105KB options.arena_block_size = 4 << 10; // 4KB options.target_file_size_base = 32 << 10; // 32KB // trigger compaction if there are >= 4 files options.level0_file_num_compaction_trigger = 4; options.compaction_options_universal.size_ratio = 10; options.compaction_options_universal.stop_style = kCompactionStopStyleSimilarSize; options.num_levels = num_levels_; DestroyAndReopen(options); Random rnd(301); int key_idx = 0; // Stage 1: // Generate a set of files at level 0, but don't trigger level-0 // compaction. for (int num = 0; num < options.level0_file_num_compaction_trigger - 1; num++) { // Write 100KB (100 values, each 1K) for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(key_idx), RandomString(&rnd, 990))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(NumSortedRuns(), num + 1); } // Generate one more file at level-0, which should trigger level-0 // compaction. for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(key_idx), RandomString(&rnd, 990))); key_idx++; } dbfull()->TEST_WaitForCompact(); // Suppose each file flushed from mem table has size 1. Now we compact // (level0_file_num_compaction_trigger+1)=4 files and should have a big // file of size 4. ASSERT_EQ(NumSortedRuns(), 1); // Stage 2: // Now we have one file at level 0, with size 4. We also have some data in // mem table. Let's continue generating new files at level 0, but don't // trigger level-0 compaction. // First, clean up memtable before inserting new data. This will generate // a level-0 file, with size around 0.4 (according to previously written // data amount). dbfull()->Flush(FlushOptions()); for (int num = 0; num < options.level0_file_num_compaction_trigger - 3; num++) { // Write 110KB (11 values, each 10K) for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(key_idx), RandomString(&rnd, 990))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(NumSortedRuns(), num + 3); } // Generate one more file at level-0, which should trigger level-0 // compaction. for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(key_idx), RandomString(&rnd, 990))); key_idx++; } dbfull()->TEST_WaitForCompact(); // Before compaction, we have 4 files at level 0, with size 4, 0.4, 1, 1. // After compaction, we should have 3 files, with size 4, 0.4, 2. ASSERT_EQ(NumSortedRuns(), 3); // Stage 3: // Now we have 3 files at level 0, with size 4, 0.4, 2. Generate one // more file at level-0, which should trigger level-0 compaction. for (int i = 0; i < 100; i++) { ASSERT_OK(Put(Key(key_idx), RandomString(&rnd, 990))); key_idx++; } dbfull()->TEST_WaitForCompact(); // Level-0 compaction is triggered, but no file will be picked up. ASSERT_EQ(NumSortedRuns(), 4); } TEST_P(DBTestUniversalCompaction, UniversalCompactionCompressRatio1) { if (!Snappy_Supported()) { return; } Options options; options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 100 << 10; // 100KB options.target_file_size_base = 32 << 10; // 32KB options.level0_file_num_compaction_trigger = 2; options.num_levels = num_levels_; options.compaction_options_universal.compression_size_percent = 70; options = CurrentOptions(options); DestroyAndReopen(options); Random rnd(301); int key_idx = 0; // The first compaction (2) is compressed. for (int num = 0; num < 2; num++) { // Write 110KB (11 values, each 10K) for (int i = 0; i < 11; i++) { ASSERT_OK(Put(Key(key_idx), CompressibleString(&rnd, 10000))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } ASSERT_LT(TotalSize(), 110000U * 2 * 0.9); // The second compaction (4) is compressed for (int num = 0; num < 2; num++) { // Write 110KB (11 values, each 10K) for (int i = 0; i < 11; i++) { ASSERT_OK(Put(Key(key_idx), CompressibleString(&rnd, 10000))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } ASSERT_LT(TotalSize(), 110000 * 4 * 0.9); // The third compaction (2 4) is compressed since this time it is // (1 1 3.2) and 3.2/5.2 doesn't reach ratio. for (int num = 0; num < 2; num++) { // Write 110KB (11 values, each 10K) for (int i = 0; i < 11; i++) { ASSERT_OK(Put(Key(key_idx), CompressibleString(&rnd, 10000))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } ASSERT_LT(TotalSize(), 110000 * 6 * 0.9); // When we start for the compaction up to (2 4 8), the latest // compressed is not compressed. for (int num = 0; num < 8; num++) { // Write 110KB (11 values, each 10K) for (int i = 0; i < 11; i++) { ASSERT_OK(Put(Key(key_idx), CompressibleString(&rnd, 10000))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } ASSERT_GT(TotalSize(), 110000 * 11 * 0.8 + 110000 * 2); } TEST_P(DBTestUniversalCompaction, UniversalCompactionCompressRatio2) { if (!Snappy_Supported()) { return; } Options options; options.compaction_style = kCompactionStyleUniversal; options.write_buffer_size = 100 << 10; // 100KB options.target_file_size_base = 32 << 10; // 32KB options.level0_file_num_compaction_trigger = 2; options.num_levels = num_levels_; options.compaction_options_universal.compression_size_percent = 95; options = CurrentOptions(options); DestroyAndReopen(options); Random rnd(301); int key_idx = 0; // When we start for the compaction up to (2 4 8), the latest // compressed is compressed given the size ratio to compress. for (int num = 0; num < 14; num++) { // Write 120KB (12 values, each 10K) for (int i = 0; i < 12; i++) { ASSERT_OK(Put(Key(key_idx), CompressibleString(&rnd, 10000))); key_idx++; } dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } ASSERT_LT(TotalSize(), 120000U * 12 * 0.8 + 120000 * 2); } // Test that checks trivial move in universal compaction TEST_P(DBTestUniversalCompaction, UniversalCompactionTrivialMoveTest1) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; ASSERT_TRUE(arg != nullptr); int output_level = *(static_cast(arg)); ASSERT_EQ(output_level, 0); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options; options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.allow_trivial_move = true; options.num_levels = 2; options.write_buffer_size = 100 << 10; // 100KB options.level0_file_num_compaction_trigger = 3; options.max_background_compactions = 1; options.target_file_size_base = 32 * 1024; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Trigger compaction if size amplification exceeds 110% options.compaction_options_universal.max_size_amplification_percent = 110; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); Random rnd(301); int num_keys = 250000; for (int i = 0; i < num_keys; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } std::vector values; ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); ASSERT_GT(trivial_move, 0); ASSERT_GT(non_trivial_move, 0); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } // Test that checks trivial move in universal compaction TEST_P(DBTestUniversalCompaction, UniversalCompactionTrivialMoveTest2) { int32_t trivial_move = 0; int32_t non_trivial_move = 0; rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:TrivialMove", [&](void* arg) { trivial_move++; }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BackgroundCompaction:NonTrivial", [&](void* arg) { non_trivial_move++; }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); Options options; options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.allow_trivial_move = true; options.num_levels = 15; options.write_buffer_size = 100 << 10; // 100KB options.level0_file_num_compaction_trigger = 8; options.max_background_compactions = 4; options.target_file_size_base = 64 * 1024; options = CurrentOptions(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); // Trigger compaction if size amplification exceeds 110% options.compaction_options_universal.max_size_amplification_percent = 110; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); Random rnd(301); int num_keys = 500000; for (int i = 0; i < num_keys; i++) { ASSERT_OK(Put(1, Key(i), Key(i))); } std::vector values; ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); ASSERT_GT(trivial_move, 0); ASSERT_EQ(non_trivial_move, 0); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_P(DBTestUniversalCompaction, UniversalCompactionFourPaths) { Options options; options.db_paths.emplace_back(dbname_, 300 * 1024); options.db_paths.emplace_back(dbname_ + "_2", 300 * 1024); options.db_paths.emplace_back(dbname_ + "_3", 500 * 1024); options.db_paths.emplace_back(dbname_ + "_4", 1024 * 1024 * 1024); options.memtable_factory.reset( new SpecialSkipListFactory(KNumKeysByGenerateNewFile - 1)); options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.size_ratio = 5; options.write_buffer_size = 110 << 10; // 105KB options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 2; options.num_levels = 1; options = CurrentOptions(options); std::vector filenames; env_->GetChildren(options.db_paths[1].path, &filenames); // Delete archival files. for (size_t i = 0; i < filenames.size(); ++i) { env_->DeleteFile(options.db_paths[1].path + "/" + filenames[i]); } env_->DeleteDir(options.db_paths[1].path); Reopen(options); Random rnd(301); int key_idx = 0; // First three 110KB files are not going to second path. // After that, (100K, 200K) for (int num = 0; num < 3; num++) { GenerateNewFile(&rnd, &key_idx); } // Another 110KB triggers a compaction to 400K file to second path GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); // (1, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1,1,4) -> (2, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); // (1, 2, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 1, 2, 4) -> (8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[3].path)); // (1, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[3].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 1, 8) -> (2, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[3].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); // (1, 2, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[3].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 1, 2, 8) -> (4, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[3].path)); // (1, 4, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[3].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[2].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Reopen(options); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Destroy(options); } TEST_P(DBTestUniversalCompaction, IncreaseUniversalCompactionNumLevels) { std::function verify_func = [&](int num_keys_in_db) { std::string keys_in_db; Iterator* iter = dbfull()->NewIterator(ReadOptions(), handles_[1]); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { keys_in_db.append(iter->key().ToString()); keys_in_db.push_back(','); } delete iter; std::string expected_keys; for (int i = 0; i <= num_keys_in_db; i++) { expected_keys.append(Key(i)); expected_keys.push_back(','); } ASSERT_EQ(keys_in_db, expected_keys); }; Random rnd(301); int max_key1 = 200; int max_key2 = 600; int max_key3 = 800; const int KNumKeysPerFile = 10; // Stage 1: open a DB with universal compaction, num_levels=1 Options options = CurrentOptions(); options.compaction_style = kCompactionStyleUniversal; options.num_levels = 1; options.write_buffer_size = 200 << 10; // 200KB options.level0_file_num_compaction_trigger = 3; options.memtable_factory.reset(new SpecialSkipListFactory(KNumKeysPerFile)); options = CurrentOptions(options); CreateAndReopenWithCF({"pikachu"}, options); for (int i = 0; i <= max_key1; i++) { // each value is 10K ASSERT_OK(Put(1, Key(i), RandomString(&rnd, 10000))); } ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); int non_level0_num_files = 0; for (int i = 1; i < options.num_levels; i++) { non_level0_num_files += NumTableFilesAtLevel(i, 1); } ASSERT_EQ(non_level0_num_files, 0); // Stage 2: reopen with universal compaction, num_levels=4 options.compaction_style = kCompactionStyleUniversal; options.num_levels = 4; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); verify_func(max_key1); // Insert more keys for (int i = max_key1 + 1; i <= max_key2; i++) { // each value is 10K ASSERT_OK(Put(1, Key(i), RandomString(&rnd, 10000))); } ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); verify_func(max_key2); // Compaction to non-L0 has happened. ASSERT_GT(NumTableFilesAtLevel(options.num_levels - 1, 1), 0); // Stage 3: Revert it back to one level and revert to num_levels=1. options.num_levels = 4; options.target_file_size_base = INT_MAX; ReopenWithColumnFamilies({"default", "pikachu"}, options); // Compact all to level 0 CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 0; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; dbfull()->CompactRange(compact_options, handles_[1], nullptr, nullptr); // Need to restart it once to remove higher level records in manifest. ReopenWithColumnFamilies({"default", "pikachu"}, options); // Final reopen options.compaction_style = kCompactionStyleUniversal; options.num_levels = 1; options = CurrentOptions(options); ReopenWithColumnFamilies({"default", "pikachu"}, options); // Insert more keys for (int i = max_key2 + 1; i <= max_key3; i++) { // each value is 10K ASSERT_OK(Put(1, Key(i), RandomString(&rnd, 10000))); } ASSERT_OK(Flush(1)); dbfull()->TEST_WaitForCompact(); verify_func(max_key3); } TEST_P(DBTestUniversalCompaction, UniversalCompactionSecondPathRatio) { if (!Snappy_Supported()) { return; } Options options; options.db_paths.emplace_back(dbname_, 500 * 1024); options.db_paths.emplace_back(dbname_ + "_2", 1024 * 1024 * 1024); options.compaction_style = kCompactionStyleUniversal; options.compaction_options_universal.size_ratio = 5; options.write_buffer_size = 110 << 10; // 105KB options.arena_block_size = 4 * 1024; options.arena_block_size = 4 << 10; options.level0_file_num_compaction_trigger = 2; options.num_levels = 1; options.memtable_factory.reset( new SpecialSkipListFactory(KNumKeysByGenerateNewFile - 1)); options = CurrentOptions(options); std::vector filenames; env_->GetChildren(options.db_paths[1].path, &filenames); // Delete archival files. for (size_t i = 0; i < filenames.size(); ++i) { env_->DeleteFile(options.db_paths[1].path + "/" + filenames[i]); } env_->DeleteDir(options.db_paths[1].path); Reopen(options); Random rnd(301); int key_idx = 0; // First three 110KB files are not going to second path. // After that, (100K, 200K) for (int num = 0; num < 3; num++) { GenerateNewFile(&rnd, &key_idx); } // Another 110KB triggers a compaction to 400K file to second path GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); // (1, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1,1,4) -> (2, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 2, 4) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(2, GetSstFileCount(dbname_)); // (1, 1, 2, 4) -> (8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); // (1, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 1, 8) -> (2, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); // (1, 2, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(2, GetSstFileCount(dbname_)); // (1, 1, 2, 8) -> (4, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(2, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(0, GetSstFileCount(dbname_)); // (1, 4, 8) GenerateNewFile(&rnd, &key_idx); ASSERT_EQ(2, GetSstFileCount(options.db_paths[1].path)); ASSERT_EQ(1, GetSstFileCount(dbname_)); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Reopen(options); for (int i = 0; i < key_idx; i++) { auto v = Get(Key(i)); ASSERT_NE(v, "NOT_FOUND"); ASSERT_TRUE(v.size() == 1 || v.size() == 990); } Destroy(options); } INSTANTIATE_TEST_CASE_P(UniversalCompactionNumLevels, DBTestUniversalCompaction, ::testing::Combine(::testing::Values(1, 3, 5), ::testing::Bool())); class DBTestUniversalManualCompactionOutputPathId : public DBTestUniversalCompactionBase { public: DBTestUniversalManualCompactionOutputPathId() : DBTestUniversalCompactionBase( "/db_universal_compaction_manual_pid_test") {} }; TEST_P(DBTestUniversalManualCompactionOutputPathId, ManualCompactionOutputPathId) { Options options = CurrentOptions(); options.create_if_missing = true; options.db_paths.emplace_back(dbname_, 1000000000); options.db_paths.emplace_back(dbname_ + "_2", 1000000000); options.compaction_style = kCompactionStyleUniversal; options.num_levels = num_levels_; options.target_file_size_base = 1 << 30; // Big size options.level0_file_num_compaction_trigger = 10; Destroy(options); DestroyAndReopen(options); CreateAndReopenWithCF({"pikachu"}, options); MakeTables(3, "p", "q", 1); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(2, TotalLiveFiles(1)); ASSERT_EQ(2, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[1].path)); // Full compaction to DB path 0 CompactRangeOptions compact_options; compact_options.target_path_id = 1; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; db_->CompactRange(compact_options, handles_[1], nullptr, nullptr); ASSERT_EQ(1, TotalLiveFiles(1)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ReopenWithColumnFamilies({kDefaultColumnFamilyName, "pikachu"}, options); ASSERT_EQ(1, TotalLiveFiles(1)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); MakeTables(1, "p", "q", 1); ASSERT_EQ(2, TotalLiveFiles(1)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); ReopenWithColumnFamilies({kDefaultColumnFamilyName, "pikachu"}, options); ASSERT_EQ(2, TotalLiveFiles(1)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[1].path)); // Full compaction to DB path 0 compact_options.target_path_id = 0; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; db_->CompactRange(compact_options, handles_[1], nullptr, nullptr); ASSERT_EQ(1, TotalLiveFiles(1)); ASSERT_EQ(1, GetSstFileCount(options.db_paths[0].path)); ASSERT_EQ(0, GetSstFileCount(options.db_paths[1].path)); // Fail when compacting to an invalid path ID compact_options.target_path_id = 2; compact_options.exclusive_manual_compaction = exclusive_manual_compaction_; ASSERT_TRUE(db_->CompactRange(compact_options, handles_[1], nullptr, nullptr) .IsInvalidArgument()); } INSTANTIATE_TEST_CASE_P(DBTestUniversalManualCompactionOutputPathId, DBTestUniversalManualCompactionOutputPathId, ::testing::Combine(::testing::Values(1, 8), ::testing::Bool())); } // namespace rocksdb #endif // !defined(ROCKSDB_LITE) int main(int argc, char** argv) { #if !defined(ROCKSDB_LITE) rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); #else return 0; #endif } rocksdb-4.5.1/db/db_wal_test.cc000066400000000000000000000106621267530634000163460ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/db_test_util.h" #include "port/stack_trace.h" #include "util/sync_point.h" namespace rocksdb { class DBWALTest : public DBTestBase { public: DBWALTest() : DBTestBase("/db_wal_test") {} }; TEST_F(DBWALTest, WAL) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); WriteOptions writeOpt = WriteOptions(); writeOpt.disableWAL = true; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v1")); ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v1")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); ASSERT_EQ("v1", Get(1, "foo")); ASSERT_EQ("v1", Get(1, "bar")); writeOpt.disableWAL = false; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v2")); writeOpt.disableWAL = true; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v2")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); // Both value's should be present. ASSERT_EQ("v2", Get(1, "bar")); ASSERT_EQ("v2", Get(1, "foo")); writeOpt.disableWAL = true; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v3")); writeOpt.disableWAL = false; ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v3")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); // again both values should be present. ASSERT_EQ("v3", Get(1, "foo")); ASSERT_EQ("v3", Get(1, "bar")); } while (ChangeCompactOptions()); } TEST_F(DBWALTest, RollLog) { do { CreateAndReopenWithCF({"pikachu"}, CurrentOptions()); ASSERT_OK(Put(1, "foo", "v1")); ASSERT_OK(Put(1, "baz", "v5")); ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); for (int i = 0; i < 10; i++) { ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); } ASSERT_OK(Put(1, "foo", "v4")); for (int i = 0; i < 10; i++) { ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions()); } } while (ChangeOptions()); } TEST_F(DBWALTest, SyncWALNotBlockWrite) { Options options = CurrentOptions(); options.max_write_buffer_number = 4; DestroyAndReopen(options); ASSERT_OK(Put("foo1", "bar1")); ASSERT_OK(Put("foo5", "bar5")); rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"WritableFileWriter::SyncWithoutFlush:1", "DBWALTest::SyncWALNotBlockWrite:1"}, {"DBWALTest::SyncWALNotBlockWrite:2", "WritableFileWriter::SyncWithoutFlush:2"}, }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread thread([&]() { ASSERT_OK(db_->SyncWAL()); }); TEST_SYNC_POINT("DBWALTest::SyncWALNotBlockWrite:1"); ASSERT_OK(Put("foo2", "bar2")); ASSERT_OK(Put("foo3", "bar3")); FlushOptions fo; fo.wait = false; ASSERT_OK(db_->Flush(fo)); ASSERT_OK(Put("foo4", "bar4")); TEST_SYNC_POINT("DBWALTest::SyncWALNotBlockWrite:2"); thread.join(); ASSERT_EQ(Get("foo1"), "bar1"); ASSERT_EQ(Get("foo2"), "bar2"); ASSERT_EQ(Get("foo3"), "bar3"); ASSERT_EQ(Get("foo4"), "bar4"); ASSERT_EQ(Get("foo5"), "bar5"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } TEST_F(DBWALTest, SyncWALNotWaitWrite) { ASSERT_OK(Put("foo1", "bar1")); ASSERT_OK(Put("foo3", "bar3")); rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"SpecialEnv::WalFile::Append:1", "DBWALTest::SyncWALNotWaitWrite:1"}, {"DBWALTest::SyncWALNotWaitWrite:2", "SpecialEnv::WalFile::Append:2"}, }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); std::thread thread([&]() { ASSERT_OK(Put("foo2", "bar2")); }); TEST_SYNC_POINT("DBWALTest::SyncWALNotWaitWrite:1"); ASSERT_OK(db_->SyncWAL()); TEST_SYNC_POINT("DBWALTest::SyncWALNotWaitWrite:2"); thread.join(); ASSERT_EQ(Get("foo1"), "bar1"); ASSERT_EQ(Get("foo2"), "bar2"); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); } } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/dbformat.cc000066400000000000000000000122571267530634000156570ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/dbformat.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include "port/port.h" #include "util/coding.h" #include "util/perf_context_imp.h" namespace rocksdb { uint64_t PackSequenceAndType(uint64_t seq, ValueType t) { assert(seq <= kMaxSequenceNumber); assert(IsValueType(t)); return (seq << 8) | t; } void UnPackSequenceAndType(uint64_t packed, uint64_t* seq, ValueType* t) { *seq = packed >> 8; *t = static_cast(packed & 0xff); assert(*seq <= kMaxSequenceNumber); assert(IsValueType(*t)); } void AppendInternalKey(std::string* result, const ParsedInternalKey& key) { result->append(key.user_key.data(), key.user_key.size()); PutFixed64(result, PackSequenceAndType(key.sequence, key.type)); } std::string ParsedInternalKey::DebugString(bool hex) const { char buf[50]; snprintf(buf, sizeof(buf), "' @ %" PRIu64 ": %d", sequence, static_cast(type)); std::string result = "'"; result += user_key.ToString(hex); result += buf; return result; } std::string InternalKey::DebugString(bool hex) const { std::string result; ParsedInternalKey parsed; if (ParseInternalKey(rep_, &parsed)) { result = parsed.DebugString(hex); } else { result = "(bad)"; result.append(EscapeString(rep_)); } return result; } const char* InternalKeyComparator::Name() const { return name_.c_str(); } int InternalKeyComparator::Compare(const Slice& akey, const Slice& bkey) const { // Order by: // increasing user key (according to user-supplied comparator) // decreasing sequence number // decreasing type (though sequence# should be enough to disambiguate) int r = user_comparator_->Compare(ExtractUserKey(akey), ExtractUserKey(bkey)); PERF_COUNTER_ADD(user_key_comparison_count, 1); if (r == 0) { const uint64_t anum = DecodeFixed64(akey.data() + akey.size() - 8); const uint64_t bnum = DecodeFixed64(bkey.data() + bkey.size() - 8); if (anum > bnum) { r = -1; } else if (anum < bnum) { r = +1; } } return r; } int InternalKeyComparator::Compare(const ParsedInternalKey& a, const ParsedInternalKey& b) const { // Order by: // increasing user key (according to user-supplied comparator) // decreasing sequence number // decreasing type (though sequence# should be enough to disambiguate) int r = user_comparator_->Compare(a.user_key, b.user_key); PERF_COUNTER_ADD(user_key_comparison_count, 1); if (r == 0) { if (a.sequence > b.sequence) { r = -1; } else if (a.sequence < b.sequence) { r = +1; } else if (a.type > b.type) { r = -1; } else if (a.type < b.type) { r = +1; } } return r; } void InternalKeyComparator::FindShortestSeparator( std::string* start, const Slice& limit) const { // Attempt to shorten the user portion of the key Slice user_start = ExtractUserKey(*start); Slice user_limit = ExtractUserKey(limit); std::string tmp(user_start.data(), user_start.size()); user_comparator_->FindShortestSeparator(&tmp, user_limit); if (tmp.size() < user_start.size() && user_comparator_->Compare(user_start, tmp) < 0) { // User key has become shorter physically, but larger logically. // Tack on the earliest possible number to the shortened user key. PutFixed64(&tmp, PackSequenceAndType(kMaxSequenceNumber,kValueTypeForSeek)); assert(this->Compare(*start, tmp) < 0); assert(this->Compare(tmp, limit) < 0); start->swap(tmp); } } void InternalKeyComparator::FindShortSuccessor(std::string* key) const { Slice user_key = ExtractUserKey(*key); std::string tmp(user_key.data(), user_key.size()); user_comparator_->FindShortSuccessor(&tmp); if (tmp.size() < user_key.size() && user_comparator_->Compare(user_key, tmp) < 0) { // User key has become shorter physically, but larger logically. // Tack on the earliest possible number to the shortened user key. PutFixed64(&tmp, PackSequenceAndType(kMaxSequenceNumber,kValueTypeForSeek)); assert(this->Compare(*key, tmp) < 0); key->swap(tmp); } } LookupKey::LookupKey(const Slice& _user_key, SequenceNumber s) { size_t usize = _user_key.size(); size_t needed = usize + 13; // A conservative estimate char* dst; if (needed <= sizeof(space_)) { dst = space_; } else { dst = new char[needed]; } start_ = dst; // NOTE: We don't support users keys of more than 2GB :) dst = EncodeVarint32(dst, static_cast(usize + 8)); kstart_ = dst; memcpy(dst, _user_key.data(), usize); dst += usize; EncodeFixed64(dst, PackSequenceAndType(s, kValueTypeForSeek)); dst += 8; end_ = dst; } } // namespace rocksdb rocksdb-4.5.1/db/dbformat.h000066400000000000000000000371401267530634000155170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/filter_policy.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "rocksdb/types.h" #include "util/coding.h" #include "util/logging.h" namespace rocksdb { class InternalKey; // Value types encoded as the last component of internal keys. // DO NOT CHANGE THESE ENUM VALUES: they are embedded in the on-disk // data structures. // The highest bit of the value type needs to be reserved to SST tables // for them to do more flexible encoding. enum ValueType : unsigned char { kTypeDeletion = 0x0, kTypeValue = 0x1, kTypeMerge = 0x2, kTypeLogData = 0x3, // WAL only. kTypeColumnFamilyDeletion = 0x4, // WAL only. kTypeColumnFamilyValue = 0x5, // WAL only. kTypeColumnFamilyMerge = 0x6, // WAL only. kTypeSingleDeletion = 0x7, kTypeColumnFamilySingleDeletion = 0x8, // WAL only. kMaxValue = 0x7F // Not used for storing records. }; // kValueTypeForSeek defines the ValueType that should be passed when // constructing a ParsedInternalKey object for seeking to a particular // sequence number (since we sort sequence numbers in decreasing order // and the value type is embedded as the low 8 bits in the sequence // number in internal keys, we need to use the highest-numbered // ValueType, not the lowest). static const ValueType kValueTypeForSeek = kTypeSingleDeletion; // Checks whether a type is a value type (i.e. a type used in memtables and sst // files). inline bool IsValueType(ValueType t) { return t <= kTypeMerge || t == kTypeSingleDeletion; } // We leave eight bits empty at the bottom so a type and sequence# // can be packed together into 64-bits. static const SequenceNumber kMaxSequenceNumber = ((0x1ull << 56) - 1); struct ParsedInternalKey { Slice user_key; SequenceNumber sequence; ValueType type; ParsedInternalKey() { } // Intentionally left uninitialized (for speed) ParsedInternalKey(const Slice& u, const SequenceNumber& seq, ValueType t) : user_key(u), sequence(seq), type(t) { } std::string DebugString(bool hex = false) const; }; // Return the length of the encoding of "key". inline size_t InternalKeyEncodingLength(const ParsedInternalKey& key) { return key.user_key.size() + 8; } // Pack a sequence number and a ValueType into a uint64_t extern uint64_t PackSequenceAndType(uint64_t seq, ValueType t); // Given the result of PackSequenceAndType, store the sequence number in *seq // and the ValueType in *t. extern void UnPackSequenceAndType(uint64_t packed, uint64_t* seq, ValueType* t); // Append the serialization of "key" to *result. extern void AppendInternalKey(std::string* result, const ParsedInternalKey& key); // Attempt to parse an internal key from "internal_key". On success, // stores the parsed data in "*result", and returns true. // // On error, returns false, leaves "*result" in an undefined state. extern bool ParseInternalKey(const Slice& internal_key, ParsedInternalKey* result); // Returns the user key portion of an internal key. inline Slice ExtractUserKey(const Slice& internal_key) { assert(internal_key.size() >= 8); return Slice(internal_key.data(), internal_key.size() - 8); } inline ValueType ExtractValueType(const Slice& internal_key) { assert(internal_key.size() >= 8); const size_t n = internal_key.size(); uint64_t num = DecodeFixed64(internal_key.data() + n - 8); unsigned char c = num & 0xff; return static_cast(c); } // A comparator for internal keys that uses a specified comparator for // the user key portion and breaks ties by decreasing sequence number. class InternalKeyComparator : public Comparator { private: const Comparator* user_comparator_; std::string name_; public: explicit InternalKeyComparator(const Comparator* c) : user_comparator_(c), name_("rocksdb.InternalKeyComparator:" + std::string(user_comparator_->Name())) { } virtual ~InternalKeyComparator() {} virtual const char* Name() const override; virtual int Compare(const Slice& a, const Slice& b) const override; virtual void FindShortestSeparator(std::string* start, const Slice& limit) const override; virtual void FindShortSuccessor(std::string* key) const override; const Comparator* user_comparator() const { return user_comparator_; } int Compare(const InternalKey& a, const InternalKey& b) const; int Compare(const ParsedInternalKey& a, const ParsedInternalKey& b) const; }; // Modules in this directory should keep internal keys wrapped inside // the following class instead of plain strings so that we do not // incorrectly use string comparisons instead of an InternalKeyComparator. class InternalKey { private: std::string rep_; public: InternalKey() { } // Leave rep_ as empty to indicate it is invalid InternalKey(const Slice& _user_key, SequenceNumber s, ValueType t) { AppendInternalKey(&rep_, ParsedInternalKey(_user_key, s, t)); } // sets the internal key to be bigger or equal to all internal keys with this // user key void SetMaxPossibleForUserKey(const Slice& _user_key) { AppendInternalKey(&rep_, ParsedInternalKey(_user_key, kMaxSequenceNumber, kValueTypeForSeek)); } // sets the internal key to be smaller or equal to all internal keys with this // user key void SetMinPossibleForUserKey(const Slice& _user_key) { AppendInternalKey( &rep_, ParsedInternalKey(_user_key, 0, static_cast(0))); } bool Valid() const { ParsedInternalKey parsed; return ParseInternalKey(Slice(rep_), &parsed); } void DecodeFrom(const Slice& s) { rep_.assign(s.data(), s.size()); } Slice Encode() const { assert(!rep_.empty()); return rep_; } Slice user_key() const { return ExtractUserKey(rep_); } size_t size() { return rep_.size(); } void SetFrom(const ParsedInternalKey& p) { rep_.clear(); AppendInternalKey(&rep_, p); } void Clear() { rep_.clear(); } std::string DebugString(bool hex = false) const; }; inline int InternalKeyComparator::Compare( const InternalKey& a, const InternalKey& b) const { return Compare(a.Encode(), b.Encode()); } inline bool ParseInternalKey(const Slice& internal_key, ParsedInternalKey* result) { const size_t n = internal_key.size(); if (n < 8) return false; uint64_t num = DecodeFixed64(internal_key.data() + n - 8); unsigned char c = num & 0xff; result->sequence = num >> 8; result->type = static_cast(c); assert(result->type <= ValueType::kMaxValue); result->user_key = Slice(internal_key.data(), n - 8); return IsValueType(result->type); } // Update the sequence number in the internal key. // Guarantees not to invalidate ikey.data(). inline void UpdateInternalKey(std::string* ikey, uint64_t seq, ValueType t) { size_t ikey_sz = ikey->size(); assert(ikey_sz >= 8); uint64_t newval = (seq << 8) | t; // Note: Since C++11, strings are guaranteed to be stored contiguously and // string::operator[]() is guaranteed not to change ikey.data(). EncodeFixed64(&(*ikey)[ikey_sz - 8], newval); } // Get the sequence number from the internal key inline uint64_t GetInternalKeySeqno(const Slice& internal_key) { const size_t n = internal_key.size(); assert(n >= 8); uint64_t num = DecodeFixed64(internal_key.data() + n - 8); return num >> 8; } // A helper class useful for DBImpl::Get() class LookupKey { public: // Initialize *this for looking up user_key at a snapshot with // the specified sequence number. LookupKey(const Slice& _user_key, SequenceNumber sequence); ~LookupKey(); // Return a key suitable for lookup in a MemTable. Slice memtable_key() const { return Slice(start_, static_cast(end_ - start_)); } // Return an internal key (suitable for passing to an internal iterator) Slice internal_key() const { return Slice(kstart_, static_cast(end_ - kstart_)); } // Return the user key Slice user_key() const { return Slice(kstart_, static_cast(end_ - kstart_ - 8)); } private: // We construct a char array of the form: // klength varint32 <-- start_ // userkey char[klength] <-- kstart_ // tag uint64 // <-- end_ // The array is a suitable MemTable key. // The suffix starting with "userkey" can be used as an InternalKey. const char* start_; const char* kstart_; const char* end_; char space_[200]; // Avoid allocation for short keys // No copying allowed LookupKey(const LookupKey&); void operator=(const LookupKey&); }; inline LookupKey::~LookupKey() { if (start_ != space_) delete[] start_; } class IterKey { public: IterKey() : buf_(space_), buf_size_(sizeof(space_)), key_(buf_), key_size_(0) {} ~IterKey() { ResetBuffer(); } Slice GetKey() const { return Slice(key_, key_size_); } Slice GetUserKey() const { assert(key_size_ >= 8); return Slice(key_, key_size_ - 8); } size_t Size() const { return key_size_; } void Clear() { key_size_ = 0; } // Append "non_shared_data" to its back, from "shared_len" // This function is used in Block::Iter::ParseNextKey // shared_len: bytes in [0, shard_len-1] would be remained // non_shared_data: data to be append, its length must be >= non_shared_len void TrimAppend(const size_t shared_len, const char* non_shared_data, const size_t non_shared_len) { assert(shared_len <= key_size_); size_t total_size = shared_len + non_shared_len; if (IsKeyPinned() /* key is not in buf_ */) { // Copy the key from external memory to buf_ (copy shared_len bytes) EnlargeBufferIfNeeded(total_size); memcpy(buf_, key_, shared_len); } else if (total_size > buf_size_) { // Need to allocate space, delete previous space char* p = new char[total_size]; memcpy(p, key_, shared_len); if (buf_ != space_) { delete[] buf_; } buf_ = p; buf_size_ = total_size; } memcpy(buf_ + shared_len, non_shared_data, non_shared_len); key_ = buf_; key_size_ = total_size; } Slice SetKey(const Slice& key, bool copy = true) { size_t size = key.size(); if (copy) { // Copy key to buf_ EnlargeBufferIfNeeded(size); memcpy(buf_, key.data(), size); key_ = buf_; } else { // Update key_ to point to external memory key_ = key.data(); } key_size_ = size; return Slice(key_, key_size_); } // Copies the content of key, updates the reference to the user key in ikey // and returns a Slice referencing the new copy. Slice SetKey(const Slice& key, ParsedInternalKey* ikey) { size_t key_n = key.size(); assert(key_n >= 8); SetKey(key); ikey->user_key = Slice(key_, key_n - 8); return Slice(key_, key_n); } // Update the sequence number in the internal key. Guarantees not to // invalidate slices to the key (and the user key). void UpdateInternalKey(uint64_t seq, ValueType t) { assert(!IsKeyPinned()); assert(key_size_ >= 8); uint64_t newval = (seq << 8) | t; EncodeFixed64(&buf_[key_size_ - 8], newval); } bool IsKeyPinned() const { return (key_ != buf_); } void SetInternalKey(const Slice& key_prefix, const Slice& user_key, SequenceNumber s, ValueType value_type = kValueTypeForSeek) { size_t psize = key_prefix.size(); size_t usize = user_key.size(); EnlargeBufferIfNeeded(psize + usize + sizeof(uint64_t)); if (psize > 0) { memcpy(buf_, key_prefix.data(), psize); } memcpy(buf_ + psize, user_key.data(), usize); EncodeFixed64(buf_ + usize + psize, PackSequenceAndType(s, value_type)); key_ = buf_; key_size_ = psize + usize + sizeof(uint64_t); } void SetInternalKey(const Slice& user_key, SequenceNumber s, ValueType value_type = kValueTypeForSeek) { SetInternalKey(Slice(), user_key, s, value_type); } void Reserve(size_t size) { EnlargeBufferIfNeeded(size); key_size_ = size; } void SetInternalKey(const ParsedInternalKey& parsed_key) { SetInternalKey(Slice(), parsed_key); } void SetInternalKey(const Slice& key_prefix, const ParsedInternalKey& parsed_key_suffix) { SetInternalKey(key_prefix, parsed_key_suffix.user_key, parsed_key_suffix.sequence, parsed_key_suffix.type); } void EncodeLengthPrefixedKey(const Slice& key) { auto size = key.size(); EnlargeBufferIfNeeded(size + static_cast(VarintLength(size))); char* ptr = EncodeVarint32(buf_, static_cast(size)); memcpy(ptr, key.data(), size); key_ = buf_; } private: char* buf_; size_t buf_size_; const char* key_; size_t key_size_; char space_[32]; // Avoid allocation for short keys void ResetBuffer() { if (buf_ != space_) { delete[] buf_; buf_ = space_; } buf_size_ = sizeof(space_); key_size_ = 0; } // Enlarge the buffer size if needed based on key_size. // By default, static allocated buffer is used. Once there is a key // larger than the static allocated buffer, another buffer is dynamically // allocated, until a larger key buffer is requested. In that case, we // reallocate buffer and delete the old one. void EnlargeBufferIfNeeded(size_t key_size) { // If size is smaller than buffer size, continue using current buffer, // or the static allocated one, as default if (key_size > buf_size_) { // Need to enlarge the buffer. ResetBuffer(); buf_ = new char[key_size]; buf_size_ = key_size; } } // No copying allowed IterKey(const IterKey&) = delete; void operator=(const IterKey&) = delete; }; class InternalKeySliceTransform : public SliceTransform { public: explicit InternalKeySliceTransform(const SliceTransform* transform) : transform_(transform) {} virtual const char* Name() const override { return transform_->Name(); } virtual Slice Transform(const Slice& src) const override { auto user_key = ExtractUserKey(src); return transform_->Transform(user_key); } virtual bool InDomain(const Slice& src) const override { auto user_key = ExtractUserKey(src); return transform_->InDomain(user_key); } virtual bool InRange(const Slice& dst) const override { auto user_key = ExtractUserKey(dst); return transform_->InRange(user_key); } const SliceTransform* user_prefix_extractor() const { return transform_; } private: // Like comparator, InternalKeySliceTransform will not take care of the // deletion of transform_ const SliceTransform* const transform_; }; // Read record from a write batch piece from input. // tag, column_family, key, value and blob are return values. Callers own the // Slice they point to. // Tag is defined as ValueType. // input will be advanced to after the record. extern Status ReadRecordFromWriteBatch(Slice* input, char* tag, uint32_t* column_family, Slice* key, Slice* value, Slice* blob); } // namespace rocksdb rocksdb-4.5.1/db/dbformat_test.cc000066400000000000000000000137421267530634000167160ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/dbformat.h" #include "util/logging.h" #include "util/testharness.h" namespace rocksdb { static std::string IKey(const std::string& user_key, uint64_t seq, ValueType vt) { std::string encoded; AppendInternalKey(&encoded, ParsedInternalKey(user_key, seq, vt)); return encoded; } static std::string Shorten(const std::string& s, const std::string& l) { std::string result = s; InternalKeyComparator(BytewiseComparator()).FindShortestSeparator(&result, l); return result; } static std::string ShortSuccessor(const std::string& s) { std::string result = s; InternalKeyComparator(BytewiseComparator()).FindShortSuccessor(&result); return result; } static void TestKey(const std::string& key, uint64_t seq, ValueType vt) { std::string encoded = IKey(key, seq, vt); Slice in(encoded); ParsedInternalKey decoded("", 0, kTypeValue); ASSERT_TRUE(ParseInternalKey(in, &decoded)); ASSERT_EQ(key, decoded.user_key.ToString()); ASSERT_EQ(seq, decoded.sequence); ASSERT_EQ(vt, decoded.type); ASSERT_TRUE(!ParseInternalKey(Slice("bar"), &decoded)); } class FormatTest : public testing::Test {}; TEST_F(FormatTest, InternalKey_EncodeDecode) { const char* keys[] = { "", "k", "hello", "longggggggggggggggggggggg" }; const uint64_t seq[] = { 1, 2, 3, (1ull << 8) - 1, 1ull << 8, (1ull << 8) + 1, (1ull << 16) - 1, 1ull << 16, (1ull << 16) + 1, (1ull << 32) - 1, 1ull << 32, (1ull << 32) + 1 }; for (unsigned int k = 0; k < sizeof(keys) / sizeof(keys[0]); k++) { for (unsigned int s = 0; s < sizeof(seq) / sizeof(seq[0]); s++) { TestKey(keys[k], seq[s], kTypeValue); TestKey("hello", 1, kTypeDeletion); } } } TEST_F(FormatTest, InternalKeyShortSeparator) { // When user keys are same ASSERT_EQ(IKey("foo", 100, kTypeValue), Shorten(IKey("foo", 100, kTypeValue), IKey("foo", 99, kTypeValue))); ASSERT_EQ(IKey("foo", 100, kTypeValue), Shorten(IKey("foo", 100, kTypeValue), IKey("foo", 101, kTypeValue))); ASSERT_EQ(IKey("foo", 100, kTypeValue), Shorten(IKey("foo", 100, kTypeValue), IKey("foo", 100, kTypeValue))); ASSERT_EQ(IKey("foo", 100, kTypeValue), Shorten(IKey("foo", 100, kTypeValue), IKey("foo", 100, kTypeDeletion))); // When user keys are misordered ASSERT_EQ(IKey("foo", 100, kTypeValue), Shorten(IKey("foo", 100, kTypeValue), IKey("bar", 99, kTypeValue))); // When user keys are different, but correctly ordered ASSERT_EQ(IKey("g", kMaxSequenceNumber, kValueTypeForSeek), Shorten(IKey("foo", 100, kTypeValue), IKey("hello", 200, kTypeValue))); // When start user key is prefix of limit user key ASSERT_EQ(IKey("foo", 100, kTypeValue), Shorten(IKey("foo", 100, kTypeValue), IKey("foobar", 200, kTypeValue))); // When limit user key is prefix of start user key ASSERT_EQ(IKey("foobar", 100, kTypeValue), Shorten(IKey("foobar", 100, kTypeValue), IKey("foo", 200, kTypeValue))); } TEST_F(FormatTest, InternalKeyShortestSuccessor) { ASSERT_EQ(IKey("g", kMaxSequenceNumber, kValueTypeForSeek), ShortSuccessor(IKey("foo", 100, kTypeValue))); ASSERT_EQ(IKey("\xff\xff", 100, kTypeValue), ShortSuccessor(IKey("\xff\xff", 100, kTypeValue))); } TEST_F(FormatTest, IterKeyOperation) { IterKey k; const char p[] = "abcdefghijklmnopqrstuvwxyz"; const char q[] = "0123456789"; ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("")); k.TrimAppend(0, p, 3); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("abc")); k.TrimAppend(1, p, 3); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("aabc")); k.TrimAppend(0, p, 26); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("abcdefghijklmnopqrstuvwxyz")); k.TrimAppend(26, q, 10); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("abcdefghijklmnopqrstuvwxyz0123456789")); k.TrimAppend(36, q, 1); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("abcdefghijklmnopqrstuvwxyz01234567890")); k.TrimAppend(26, q, 1); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("abcdefghijklmnopqrstuvwxyz0")); // Size going up, memory allocation is triggered k.TrimAppend(27, p, 26); ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()), std::string("abcdefghijklmnopqrstuvwxyz0" "abcdefghijklmnopqrstuvwxyz")); } TEST_F(FormatTest, UpdateInternalKey) { std::string user_key("abcdefghijklmnopqrstuvwxyz"); uint64_t new_seq = 0x123456; ValueType new_val_type = kTypeDeletion; std::string ikey; AppendInternalKey(&ikey, ParsedInternalKey(user_key, 100U, kTypeValue)); size_t ikey_size = ikey.size(); UpdateInternalKey(&ikey, new_seq, new_val_type); ASSERT_EQ(ikey_size, ikey.size()); Slice in(ikey); ParsedInternalKey decoded; ASSERT_TRUE(ParseInternalKey(in, &decoded)); ASSERT_EQ(user_key, decoded.user_key.ToString()); ASSERT_EQ(new_seq, decoded.sequence); ASSERT_EQ(new_val_type, decoded.type); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/deletefile_test.cc000066400000000000000000000275161267530634000172260ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "rocksdb/db.h" #include "db/db_impl.h" #include "db/filename.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" #include "rocksdb/env.h" #include "rocksdb/transaction_log.h" #include #include #include #include namespace rocksdb { class DeleteFileTest : public testing::Test { public: std::string dbname_; Options options_; DB* db_; Env* env_; int numlevels_; DeleteFileTest() { db_ = nullptr; env_ = Env::Default(); options_.enable_thread_tracking = true; options_.write_buffer_size = 1024*1024*1000; options_.target_file_size_base = 1024*1024*1000; options_.max_bytes_for_level_base = 1024*1024*1000; options_.WAL_ttl_seconds = 300; // Used to test log files options_.WAL_size_limit_MB = 1024; // Used to test log files dbname_ = test::TmpDir() + "/deletefile_test"; options_.wal_dir = dbname_ + "/wal_files"; // clean up all the files that might have been there before std::vector old_files; env_->GetChildren(dbname_, &old_files); for (auto file : old_files) { env_->DeleteFile(dbname_ + "/" + file); } env_->GetChildren(options_.wal_dir, &old_files); for (auto file : old_files) { env_->DeleteFile(options_.wal_dir + "/" + file); } DestroyDB(dbname_, options_); numlevels_ = 7; EXPECT_OK(ReopenDB(true)); } Status ReopenDB(bool create) { delete db_; if (create) { DestroyDB(dbname_, options_); } db_ = nullptr; options_.create_if_missing = create; return DB::Open(options_, dbname_, &db_); } void CloseDB() { delete db_; db_ = nullptr; } void AddKeys(int numkeys, int startkey = 0) { WriteOptions options; options.sync = false; ReadOptions roptions; for (int i = startkey; i < (numkeys + startkey) ; i++) { std::string temp = ToString(i); Slice key(temp); Slice value(temp); ASSERT_OK(db_->Put(options, key, value)); } } int numKeysInLevels( std::vector &metadata, std::vector *keysperlevel = nullptr) { if (keysperlevel != nullptr) { keysperlevel->resize(numlevels_); } int numKeys = 0; for (size_t i = 0; i < metadata.size(); i++) { int startkey = atoi(metadata[i].smallestkey.c_str()); int endkey = atoi(metadata[i].largestkey.c_str()); int numkeysinfile = (endkey - startkey + 1); numKeys += numkeysinfile; if (keysperlevel != nullptr) { (*keysperlevel)[(int)metadata[i].level] += numkeysinfile; } fprintf(stderr, "level %d name %s smallest %s largest %s\n", metadata[i].level, metadata[i].name.c_str(), metadata[i].smallestkey.c_str(), metadata[i].largestkey.c_str()); } return numKeys; } void CreateTwoLevels() { AddKeys(50000, 10000); DBImpl* dbi = reinterpret_cast(db_); ASSERT_OK(dbi->TEST_FlushMemTable()); ASSERT_OK(dbi->TEST_WaitForFlushMemTable()); for (int i = 0; i < 2; ++i) { ASSERT_OK(dbi->TEST_CompactRange(i, nullptr, nullptr)); } AddKeys(50000, 10000); ASSERT_OK(dbi->TEST_FlushMemTable()); ASSERT_OK(dbi->TEST_WaitForFlushMemTable()); ASSERT_OK(dbi->TEST_CompactRange(0, nullptr, nullptr)); } void CheckFileTypeCounts(std::string& dir, int required_log, int required_sst, int required_manifest) { std::vector filenames; env_->GetChildren(dir, &filenames); int log_cnt = 0, sst_cnt = 0, manifest_cnt = 0; for (auto file : filenames) { uint64_t number; FileType type; if (ParseFileName(file, &number, &type)) { log_cnt += (type == kLogFile); sst_cnt += (type == kTableFile); manifest_cnt += (type == kDescriptorFile); } } ASSERT_EQ(required_log, log_cnt); ASSERT_EQ(required_sst, sst_cnt); ASSERT_EQ(required_manifest, manifest_cnt); } }; TEST_F(DeleteFileTest, AddKeysAndQueryLevels) { CreateTwoLevels(); std::vector metadata; db_->GetLiveFilesMetaData(&metadata); std::string level1file = ""; int level1keycount = 0; std::string level2file = ""; int level2keycount = 0; int level1index = 0; int level2index = 1; ASSERT_EQ((int)metadata.size(), 2); if (metadata[0].level == 2) { level1index = 1; level2index = 0; } level1file = metadata[level1index].name; int startkey = atoi(metadata[level1index].smallestkey.c_str()); int endkey = atoi(metadata[level1index].largestkey.c_str()); level1keycount = (endkey - startkey + 1); level2file = metadata[level2index].name; startkey = atoi(metadata[level2index].smallestkey.c_str()); endkey = atoi(metadata[level2index].largestkey.c_str()); level2keycount = (endkey - startkey + 1); // COntrolled setup. Levels 1 and 2 should both have 50K files. // This is a little fragile as it depends on the current // compaction heuristics. ASSERT_EQ(level1keycount, 50000); ASSERT_EQ(level2keycount, 50000); Status status = db_->DeleteFile("0.sst"); ASSERT_TRUE(status.IsInvalidArgument()); // intermediate level files cannot be deleted. status = db_->DeleteFile(level1file); ASSERT_TRUE(status.IsInvalidArgument()); // Lowest level file deletion should succeed. ASSERT_OK(db_->DeleteFile(level2file)); CloseDB(); } TEST_F(DeleteFileTest, PurgeObsoleteFilesTest) { CreateTwoLevels(); // there should be only one (empty) log file because CreateTwoLevels() // flushes the memtables to disk CheckFileTypeCounts(options_.wal_dir, 1, 0, 0); // 2 ssts, 1 manifest CheckFileTypeCounts(dbname_, 0, 2, 1); std::string first("0"), last("999999"); CompactRangeOptions compact_options; compact_options.change_level = true; compact_options.target_level = 2; Slice first_slice(first), last_slice(last); db_->CompactRange(compact_options, &first_slice, &last_slice); // 1 sst after compaction CheckFileTypeCounts(dbname_, 0, 1, 1); // this time, we keep an iterator alive ReopenDB(true); Iterator *itr = 0; CreateTwoLevels(); itr = db_->NewIterator(ReadOptions()); db_->CompactRange(compact_options, &first_slice, &last_slice); // 3 sst after compaction with live iterator CheckFileTypeCounts(dbname_, 0, 3, 1); delete itr; // 1 sst after iterator deletion CheckFileTypeCounts(dbname_, 0, 1, 1); CloseDB(); } TEST_F(DeleteFileTest, DeleteFileWithIterator) { CreateTwoLevels(); ReadOptions options; Iterator* it = db_->NewIterator(options); std::vector metadata; db_->GetLiveFilesMetaData(&metadata); std::string level2file = ""; ASSERT_EQ((int)metadata.size(), 2); if (metadata[0].level == 1) { level2file = metadata[1].name; } else { level2file = metadata[0].name; } Status status = db_->DeleteFile(level2file); fprintf(stdout, "Deletion status %s: %s\n", level2file.c_str(), status.ToString().c_str()); ASSERT_TRUE(status.ok()); it->SeekToFirst(); int numKeysIterated = 0; while(it->Valid()) { numKeysIterated++; it->Next(); } ASSERT_EQ(numKeysIterated, 50000); delete it; CloseDB(); } TEST_F(DeleteFileTest, DeleteLogFiles) { AddKeys(10, 0); VectorLogPtr logfiles; db_->GetSortedWalFiles(logfiles); ASSERT_GT(logfiles.size(), 0UL); // Take the last log file which is expected to be alive and try to delete it // Should not succeed because live logs are not allowed to be deleted std::unique_ptr alive_log = std::move(logfiles.back()); ASSERT_EQ(alive_log->Type(), kAliveLogFile); ASSERT_OK(env_->FileExists(options_.wal_dir + "/" + alive_log->PathName())); fprintf(stdout, "Deleting alive log file %s\n", alive_log->PathName().c_str()); ASSERT_TRUE(!db_->DeleteFile(alive_log->PathName()).ok()); ASSERT_OK(env_->FileExists(options_.wal_dir + "/" + alive_log->PathName())); logfiles.clear(); // Call Flush to bring about a new working log file and add more keys // Call Flush again to flush out memtable and move alive log to archived log // and try to delete the archived log file FlushOptions fopts; db_->Flush(fopts); AddKeys(10, 0); db_->Flush(fopts); db_->GetSortedWalFiles(logfiles); ASSERT_GT(logfiles.size(), 0UL); std::unique_ptr archived_log = std::move(logfiles.front()); ASSERT_EQ(archived_log->Type(), kArchivedLogFile); ASSERT_OK( env_->FileExists(options_.wal_dir + "/" + archived_log->PathName())); fprintf(stdout, "Deleting archived log file %s\n", archived_log->PathName().c_str()); ASSERT_OK(db_->DeleteFile(archived_log->PathName())); ASSERT_EQ(Status::NotFound(), env_->FileExists(options_.wal_dir + "/" + archived_log->PathName())); CloseDB(); } TEST_F(DeleteFileTest, DeleteNonDefaultColumnFamily) { CloseDB(); DBOptions db_options; db_options.create_if_missing = true; db_options.create_missing_column_families = true; std::vector column_families; column_families.emplace_back(); column_families.emplace_back("new_cf", ColumnFamilyOptions()); std::vector handles; rocksdb::DB* db; ASSERT_OK(DB::Open(db_options, dbname_, column_families, &handles, &db)); Random rnd(5); for (int i = 0; i < 1000; ++i) { ASSERT_OK(db->Put(WriteOptions(), handles[1], test::RandomKey(&rnd, 10), test::RandomKey(&rnd, 10))); } ASSERT_OK(db->Flush(FlushOptions(), handles[1])); for (int i = 0; i < 1000; ++i) { ASSERT_OK(db->Put(WriteOptions(), handles[1], test::RandomKey(&rnd, 10), test::RandomKey(&rnd, 10))); } ASSERT_OK(db->Flush(FlushOptions(), handles[1])); std::vector metadata; db->GetLiveFilesMetaData(&metadata); ASSERT_EQ(2U, metadata.size()); ASSERT_EQ("new_cf", metadata[0].column_family_name); ASSERT_EQ("new_cf", metadata[1].column_family_name); auto old_file = metadata[0].smallest_seqno < metadata[1].smallest_seqno ? metadata[0].name : metadata[1].name; auto new_file = metadata[0].smallest_seqno > metadata[1].smallest_seqno ? metadata[0].name : metadata[1].name; ASSERT_TRUE(db->DeleteFile(new_file).IsInvalidArgument()); ASSERT_OK(db->DeleteFile(old_file)); { std::unique_ptr itr(db->NewIterator(ReadOptions(), handles[1])); int count = 0; for (itr->SeekToFirst(); itr->Valid(); itr->Next()) { ASSERT_OK(itr->status()); ++count; } ASSERT_EQ(count, 1000); } delete handles[0]; delete handles[1]; delete db; ASSERT_OK(DB::Open(db_options, dbname_, column_families, &handles, &db)); { std::unique_ptr itr(db->NewIterator(ReadOptions(), handles[1])); int count = 0; for (itr->SeekToFirst(); itr->Valid(); itr->Next()) { ASSERT_OK(itr->status()); ++count; } ASSERT_EQ(count, 1000); } delete handles[0]; delete handles[1]; delete db; } } //namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as DBImpl::DeleteFile is not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/event_helpers.cc000066400000000000000000000066151267530634000167250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/event_helpers.h" namespace rocksdb { namespace { template inline T SafeDivide(T a, T b) { return b == 0 ? 0 : a / b; } } // namespace void EventHelpers::AppendCurrentTime(JSONWriter* jwriter) { *jwriter << "time_micros" << std::chrono::duration_cast( std::chrono::system_clock::now().time_since_epoch()).count(); } void EventHelpers::LogAndNotifyTableFileCreation( EventLogger* event_logger, const std::vector>& listeners, const FileDescriptor& fd, const TableFileCreationInfo& info) { assert(event_logger); JSONWriter jwriter; AppendCurrentTime(&jwriter); jwriter << "cf_name" << info.cf_name << "job" << info.job_id << "event" << "table_file_creation" << "file_number" << fd.GetNumber() << "file_size" << fd.GetFileSize(); // table_properties { jwriter << "table_properties"; jwriter.StartObject(); // basic properties: jwriter << "data_size" << info.table_properties.data_size << "index_size" << info.table_properties.index_size << "filter_size" << info.table_properties.filter_size << "raw_key_size" << info.table_properties.raw_key_size << "raw_average_key_size" << SafeDivide( info.table_properties.raw_key_size, info.table_properties.num_entries) << "raw_value_size" << info.table_properties.raw_value_size << "raw_average_value_size" << SafeDivide( info.table_properties.raw_value_size, info.table_properties.num_entries) << "num_data_blocks" << info.table_properties.num_data_blocks << "num_entries" << info.table_properties.num_entries << "filter_policy_name" << info.table_properties.filter_policy_name; // user collected properties for (const auto& prop : info.table_properties.readable_properties) { jwriter << prop.first << prop.second; } jwriter.EndObject(); } jwriter.EndObject(); event_logger->Log(jwriter); #ifndef ROCKSDB_LITE if (listeners.size() == 0) { return; } for (auto listener : listeners) { listener->OnTableFileCreated(info); } #endif // !ROCKSDB_LITE } void EventHelpers::LogAndNotifyTableFileDeletion( EventLogger* event_logger, int job_id, uint64_t file_number, const std::string& file_path, const Status& status, const std::string& dbname, const std::vector>& listeners) { JSONWriter jwriter; AppendCurrentTime(&jwriter); jwriter << "job" << job_id << "event" << "table_file_deletion" << "file_number" << file_number; if (!status.ok()) { jwriter << "status" << status.ToString(); } jwriter.EndObject(); event_logger->Log(jwriter); #ifndef ROCKSDB_LITE TableFileDeletionInfo info; info.db_name = dbname; info.job_id = job_id; info.file_path = file_path; info.status = status; for (auto listener : listeners) { listener->OnTableFileDeleted(info); } #endif // !ROCKSDB_LITE } } // namespace rocksdb rocksdb-4.5.1/db/event_helpers.h000066400000000000000000000021701267530634000165570ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include "db/column_family.h" #include "db/version_edit.h" #include "rocksdb/listener.h" #include "rocksdb/table_properties.h" #include "util/event_logger.h" namespace rocksdb { class EventHelpers { public: static void AppendCurrentTime(JSONWriter* json_writer); static void LogAndNotifyTableFileCreation( EventLogger* event_logger, const std::vector>& listeners, const FileDescriptor& fd, const TableFileCreationInfo& info); static void LogAndNotifyTableFileDeletion( EventLogger* event_logger, int job_id, uint64_t file_number, const std::string& file_path, const Status& status, const std::string& db_name, const std::vector>& listeners); }; } // namespace rocksdb rocksdb-4.5.1/db/experimental.cc000066400000000000000000000031631267530634000165520ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "rocksdb/experimental.h" #include "db/db_impl.h" namespace rocksdb { namespace experimental { #ifndef ROCKSDB_LITE Status SuggestCompactRange(DB* db, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) { auto dbimpl = dynamic_cast(db); if (dbimpl == nullptr) { return Status::InvalidArgument("Didn't recognize DB object"); } return dbimpl->SuggestCompactRange(column_family, begin, end); } Status PromoteL0(DB* db, ColumnFamilyHandle* column_family, int target_level) { auto dbimpl = dynamic_cast(db); if (dbimpl == nullptr) { return Status::InvalidArgument("Didn't recognize DB object"); } return dbimpl->PromoteL0(column_family, target_level); } #else // ROCKSDB_LITE Status SuggestCompactRange(DB* db, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) { return Status::NotSupported("Not supported in RocksDB LITE"); } Status PromoteL0(DB* db, ColumnFamilyHandle* column_family, int target_level) { return Status::NotSupported("Not supported in RocksDB LITE"); } #endif // ROCKSDB_LITE Status SuggestCompactRange(DB* db, const Slice* begin, const Slice* end) { return SuggestCompactRange(db, db->DefaultColumnFamily(), begin, end); } } // namespace experimental } // namespace rocksdb rocksdb-4.5.1/db/fault_injection_test.cc000066400000000000000000000655221267530634000203000ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright 2014 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // This test uses a custom Env to keep track of the state of a filesystem as of // the last "sync". It then checks for data loss errors by purposely dropping // file data (or entire files) not protected by a "sync". #include #include #include "db/db_impl.h" #include "db/filename.h" #include "db/log_format.h" #include "db/version_set.h" #include "rocksdb/cache.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/table.h" #include "rocksdb/write_batch.h" #include "util/logging.h" #include "util/mock_env.h" #include "util/mutexlock.h" #include "util/sync_point.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { static const int kValueSize = 1000; static const int kMaxNumValues = 2000; static const size_t kNumIterations = 3; class TestWritableFile; class FaultInjectionTestEnv; namespace { // Assume a filename, and not a directory name like "/foo/bar/" static std::string GetDirName(const std::string filename) { size_t found = filename.find_last_of("/\\"); if (found == std::string::npos) { return ""; } else { return filename.substr(0, found); } } // Trim the tailing "/" in the end of `str` static std::string TrimDirname(const std::string& str) { size_t found = str.find_last_not_of("/"); if (found == std::string::npos) { return str; } return str.substr(0, found + 1); } // Return pair of a full path. static std::pair GetDirAndName( const std::string& name) { std::string dirname = GetDirName(name); std::string fname = name.substr(dirname.size() + 1); return std::make_pair(dirname, fname); } // A basic file truncation function suitable for this test. Status Truncate(Env* env, const std::string& filename, uint64_t length) { unique_ptr orig_file; const EnvOptions options; Status s = env->NewSequentialFile(filename, &orig_file, options); if (!s.ok()) { fprintf(stderr, "Cannot truncate file %s: %s\n", filename.c_str(), s.ToString().c_str()); return s; } std::unique_ptr scratch(new char[length]); rocksdb::Slice result; s = orig_file->Read(length, &result, scratch.get()); #ifdef OS_WIN orig_file.reset(); #endif if (s.ok()) { std::string tmp_name = GetDirName(filename) + "/truncate.tmp"; unique_ptr tmp_file; s = env->NewWritableFile(tmp_name, &tmp_file, options); if (s.ok()) { s = tmp_file->Append(result); if (s.ok()) { s = env->RenameFile(tmp_name, filename); } else { fprintf(stderr, "Cannot rename file %s to %s: %s\n", tmp_name.c_str(), filename.c_str(), s.ToString().c_str()); env->DeleteFile(tmp_name); } } } if (!s.ok()) { fprintf(stderr, "Cannot truncate file %s: %s\n", filename.c_str(), s.ToString().c_str()); } return s; } struct FileState { std::string filename_; ssize_t pos_; ssize_t pos_at_last_sync_; ssize_t pos_at_last_flush_; explicit FileState(const std::string& filename) : filename_(filename), pos_(-1), pos_at_last_sync_(-1), pos_at_last_flush_(-1) { } FileState() : pos_(-1), pos_at_last_sync_(-1), pos_at_last_flush_(-1) {} bool IsFullySynced() const { return pos_ <= 0 || pos_ == pos_at_last_sync_; } Status DropUnsyncedData(Env* env) const; Status DropRandomUnsyncedData(Env* env, Random* rand) const; }; } // anonymous namespace // A wrapper around WritableFileWriter* file // is written to or sync'ed. class TestWritableFile : public WritableFile { public: explicit TestWritableFile(const std::string& fname, unique_ptr&& f, FaultInjectionTestEnv* env); virtual ~TestWritableFile(); virtual Status Append(const Slice& data) override; virtual Status Truncate(uint64_t size) override { return target_->Truncate(size); } virtual Status Close() override; virtual Status Flush() override; virtual Status Sync() override; virtual bool IsSyncThreadSafe() const override { return true; } private: FileState state_; unique_ptr target_; bool writable_file_opened_; FaultInjectionTestEnv* env_; }; class TestDirectory : public Directory { public: explicit TestDirectory(FaultInjectionTestEnv* env, std::string dirname, Directory* dir) : env_(env), dirname_(dirname), dir_(dir) {} ~TestDirectory() {} virtual Status Fsync() override; private: FaultInjectionTestEnv* env_; std::string dirname_; unique_ptr dir_; }; class FaultInjectionTestEnv : public EnvWrapper { public: explicit FaultInjectionTestEnv(Env* base) : EnvWrapper(base), filesystem_active_(true) {} virtual ~FaultInjectionTestEnv() { } Status NewDirectory(const std::string& name, unique_ptr* result) override { unique_ptr r; Status s = target()->NewDirectory(name, &r); EXPECT_OK(s); if (!s.ok()) { return s; } result->reset(new TestDirectory(this, TrimDirname(name), r.release())); return Status::OK(); } Status NewWritableFile(const std::string& fname, unique_ptr* result, const EnvOptions& soptions) override { if (!IsFilesystemActive()) { return Status::Corruption("Not Active"); } // Not allow overwriting files Status s = target()->FileExists(fname); if (s.ok()) { return Status::Corruption("File already exists."); } else if (!s.IsNotFound()) { assert(s.IsIOError()); return s; } s = target()->NewWritableFile(fname, result, soptions); if (s.ok()) { result->reset(new TestWritableFile(fname, std::move(*result), this)); // WritableFileWriter* file is opened // again then it will be truncated - so forget our saved state. UntrackFile(fname); MutexLock l(&mutex_); open_files_.insert(fname); auto dir_and_name = GetDirAndName(fname); auto& list = dir_to_new_files_since_last_sync_[dir_and_name.first]; list.insert(dir_and_name.second); } return s; } virtual Status DeleteFile(const std::string& f) override { if (!IsFilesystemActive()) { return Status::Corruption("Not Active"); } Status s = EnvWrapper::DeleteFile(f); if (!s.ok()) { fprintf(stderr, "Cannot delete file %s: %s\n", f.c_str(), s.ToString().c_str()); } EXPECT_OK(s); if (s.ok()) { UntrackFile(f); } return s; } virtual Status RenameFile(const std::string& s, const std::string& t) override { if (!IsFilesystemActive()) { return Status::Corruption("Not Active"); } Status ret = EnvWrapper::RenameFile(s, t); if (ret.ok()) { MutexLock l(&mutex_); if (db_file_state_.find(s) != db_file_state_.end()) { db_file_state_[t] = db_file_state_[s]; db_file_state_.erase(s); } auto sdn = GetDirAndName(s); auto tdn = GetDirAndName(t); if (dir_to_new_files_since_last_sync_[sdn.first].erase(sdn.second) != 0) { auto& tlist = dir_to_new_files_since_last_sync_[tdn.first]; assert(tlist.find(tdn.second) == tlist.end()); tlist.insert(tdn.second); } } return ret; } void WritableFileClosed(const FileState& state) { MutexLock l(&mutex_); if (open_files_.find(state.filename_) != open_files_.end()) { db_file_state_[state.filename_] = state; open_files_.erase(state.filename_); } } // For every file that is not fully synced, make a call to `func` with // FileState of the file as the parameter. Status DropFileData(std::function func) { Status s; MutexLock l(&mutex_); for (std::map::const_iterator it = db_file_state_.begin(); s.ok() && it != db_file_state_.end(); ++it) { const FileState& state = it->second; if (!state.IsFullySynced()) { s = func(target(), state); } } return s; } Status DropUnsyncedFileData() { return DropFileData([&](Env* env, const FileState& state) { return state.DropUnsyncedData(env); }); } Status DropRandomUnsyncedFileData(Random* rnd) { return DropFileData([&](Env* env, const FileState& state) { return state.DropRandomUnsyncedData(env, rnd); }); } Status DeleteFilesCreatedAfterLastDirSync() { // Because DeleteFile access this container make a copy to avoid deadlock std::map> map_copy; { MutexLock l(&mutex_); map_copy.insert(dir_to_new_files_since_last_sync_.begin(), dir_to_new_files_since_last_sync_.end()); } for (auto& pair : map_copy) { for (std::string name : pair.second) { Status s = DeleteFile(pair.first + "/" + name); if (!s.ok()) { return s; } } } return Status::OK(); } void ResetState() { MutexLock l(&mutex_); db_file_state_.clear(); dir_to_new_files_since_last_sync_.clear(); SetFilesystemActiveNoLock(true); } void UntrackFile(const std::string& f) { MutexLock l(&mutex_); auto dir_and_name = GetDirAndName(f); dir_to_new_files_since_last_sync_[dir_and_name.first].erase( dir_and_name.second); db_file_state_.erase(f); open_files_.erase(f); } void SyncDir(const std::string& dirname) { MutexLock l(&mutex_); dir_to_new_files_since_last_sync_.erase(dirname); } // Setting the filesystem to inactive is the test equivalent to simulating a // system reset. Setting to inactive will freeze our saved filesystem state so // that it will stop being recorded. It can then be reset back to the state at // the time of the reset. bool IsFilesystemActive() { MutexLock l(&mutex_); return filesystem_active_; } void SetFilesystemActiveNoLock(bool active) { filesystem_active_ = active; } void SetFilesystemActive(bool active) { MutexLock l(&mutex_); SetFilesystemActiveNoLock(active); } void AssertNoOpenFile() { ASSERT_TRUE(open_files_.empty()); } private: port::Mutex mutex_; std::map db_file_state_; std::set open_files_; std::unordered_map> dir_to_new_files_since_last_sync_; bool filesystem_active_; // Record flushes, syncs, writes }; Status FileState::DropUnsyncedData(Env* env) const { ssize_t sync_pos = pos_at_last_sync_ == -1 ? 0 : pos_at_last_sync_; return Truncate(env, filename_, sync_pos); } Status FileState::DropRandomUnsyncedData(Env* env, Random* rand) const { ssize_t sync_pos = pos_at_last_sync_ == -1 ? 0 : pos_at_last_sync_; assert(pos_ >= sync_pos); int range = static_cast(pos_ - sync_pos); uint64_t truncated_size = static_cast(sync_pos) + rand->Uniform(range); return Truncate(env, filename_, truncated_size); } Status TestDirectory::Fsync() { env_->SyncDir(dirname_); return dir_->Fsync(); } TestWritableFile::TestWritableFile(const std::string& fname, unique_ptr&& f, FaultInjectionTestEnv* env) : state_(fname), target_(std::move(f)), writable_file_opened_(true), env_(env) { assert(target_ != nullptr); state_.pos_ = 0; } TestWritableFile::~TestWritableFile() { if (writable_file_opened_) { Close(); } } Status TestWritableFile::Append(const Slice& data) { if (!env_->IsFilesystemActive()) { return Status::Corruption("Not Active"); } Status s = target_->Append(data); if (s.ok()) { state_.pos_ += data.size(); } return s; } Status TestWritableFile::Close() { writable_file_opened_ = false; Status s = target_->Close(); if (s.ok()) { env_->WritableFileClosed(state_); } return s; } Status TestWritableFile::Flush() { Status s = target_->Flush(); if (s.ok() && env_->IsFilesystemActive()) { state_.pos_at_last_flush_ = state_.pos_; } return s; } Status TestWritableFile::Sync() { if (!env_->IsFilesystemActive()) { return Status::OK(); } // No need to actual sync. state_.pos_at_last_sync_ = state_.pos_; return Status::OK(); } class FaultInjectionTest : public testing::Test, public testing::WithParamInterface { protected: enum OptionConfig { kDefault, kDifferentDataDir, kWalDir, kSyncWal, kWalDirSyncWal, kMultiLevels, kEnd, }; int option_config_; // When need to make sure data is persistent, sync WAL bool sync_use_wal_; // When need to make sure data is persistent, call DB::CompactRange() bool sync_use_compact_; bool sequential_order_; protected: public: enum ExpectedVerifResult { kValExpectFound, kValExpectNoError }; enum ResetMethod { kResetDropUnsyncedData, kResetDropRandomUnsyncedData, kResetDeleteUnsyncedFiles, kResetDropAndDeleteUnsynced }; std::unique_ptr base_env_; FaultInjectionTestEnv* env_; std::string dbname_; shared_ptr tiny_cache_; Options options_; DB* db_; FaultInjectionTest() : option_config_(kDefault), sync_use_wal_(false), sync_use_compact_(true), base_env_(nullptr), env_(NULL), db_(NULL) { } ~FaultInjectionTest() { rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); } bool ChangeOptions() { option_config_++; if (option_config_ >= kEnd) { return false; } else { if (option_config_ == kMultiLevels) { base_env_.reset(new MockEnv(Env::Default())); } return true; } } // Return the current option configuration. Options CurrentOptions() { sync_use_wal_ = false; sync_use_compact_ = true; Options options; switch (option_config_) { case kWalDir: options.wal_dir = test::TmpDir(env_) + "/fault_test_wal"; break; case kDifferentDataDir: options.db_paths.emplace_back(test::TmpDir(env_) + "/fault_test_data", 1000000U); break; case kSyncWal: sync_use_wal_ = true; sync_use_compact_ = false; break; case kWalDirSyncWal: options.wal_dir = test::TmpDir(env_) + "/fault_test_wal"; sync_use_wal_ = true; sync_use_compact_ = false; break; case kMultiLevels: options.write_buffer_size = 64 * 1024; options.target_file_size_base = 64 * 1024; options.level0_file_num_compaction_trigger = 2; options.level0_slowdown_writes_trigger = 2; options.level0_stop_writes_trigger = 4; options.max_bytes_for_level_base = 128 * 1024; options.max_write_buffer_number = 2; options.max_background_compactions = 8; options.max_background_flushes = 8; sync_use_wal_ = true; sync_use_compact_ = false; break; default: break; } return options; } Status NewDB() { assert(db_ == NULL); assert(tiny_cache_ == nullptr); assert(env_ == NULL); env_ = new FaultInjectionTestEnv(base_env_ ? base_env_.get() : Env::Default()); options_ = CurrentOptions(); options_.env = env_; options_.paranoid_checks = true; BlockBasedTableOptions table_options; tiny_cache_ = NewLRUCache(100); table_options.block_cache = tiny_cache_; options_.table_factory.reset(NewBlockBasedTableFactory(table_options)); dbname_ = test::TmpDir() + "/fault_test"; EXPECT_OK(DestroyDB(dbname_, options_)); options_.create_if_missing = true; Status s = OpenDB(); options_.create_if_missing = false; return s; } void SetUp() override { sequential_order_ = GetParam(); ASSERT_OK(NewDB()); } void TearDown() override { CloseDB(); Status s = DestroyDB(dbname_, options_); delete env_; env_ = NULL; tiny_cache_.reset(); ASSERT_OK(s); } void Build(const WriteOptions& write_options, int start_idx, int num_vals) { std::string key_space, value_space; WriteBatch batch; for (int i = start_idx; i < start_idx + num_vals; i++) { Slice key = Key(i, &key_space); batch.Clear(); batch.Put(key, Value(i, &value_space)); ASSERT_OK(db_->Write(write_options, &batch)); } } Status ReadValue(int i, std::string* val) const { std::string key_space, value_space; Slice key = Key(i, &key_space); Value(i, &value_space); ReadOptions options; return db_->Get(options, key, val); } Status Verify(int start_idx, int num_vals, ExpectedVerifResult expected) const { std::string val; std::string value_space; Status s; for (int i = start_idx; i < start_idx + num_vals && s.ok(); i++) { Value(i, &value_space); s = ReadValue(i, &val); if (s.ok()) { EXPECT_EQ(value_space, val); } if (expected == kValExpectFound) { if (!s.ok()) { fprintf(stderr, "Error when read %dth record (expect found): %s\n", i, s.ToString().c_str()); return s; } } else if (!s.ok() && !s.IsNotFound()) { fprintf(stderr, "Error when read %dth record: %s\n", i, s.ToString().c_str()); return s; } } return Status::OK(); } // Return the ith key Slice Key(int i, std::string* storage) const { int num = i; if (!sequential_order_) { // random transfer const int m = 0x5bd1e995; num *= m; num ^= num << 24; } char buf[100]; snprintf(buf, sizeof(buf), "%016d", num); storage->assign(buf, strlen(buf)); return Slice(*storage); } // Return the value to associate with the specified key Slice Value(int k, std::string* storage) const { Random r(k); return test::RandomString(&r, kValueSize, storage); } void CloseDB() { delete db_; db_ = NULL; } Status OpenDB() { CloseDB(); env_->ResetState(); return DB::Open(options_, dbname_, &db_); } void DeleteAllData() { Iterator* iter = db_->NewIterator(ReadOptions()); WriteOptions options; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(db_->Delete(WriteOptions(), iter->key())); } delete iter; FlushOptions flush_options; flush_options.wait = true; db_->Flush(flush_options); } // rnd cannot be null for kResetDropRandomUnsyncedData void ResetDBState(ResetMethod reset_method, Random* rnd = nullptr) { env_->AssertNoOpenFile(); switch (reset_method) { case kResetDropUnsyncedData: ASSERT_OK(env_->DropUnsyncedFileData()); break; case kResetDropRandomUnsyncedData: ASSERT_OK(env_->DropRandomUnsyncedFileData(rnd)); break; case kResetDeleteUnsyncedFiles: ASSERT_OK(env_->DeleteFilesCreatedAfterLastDirSync()); break; case kResetDropAndDeleteUnsynced: ASSERT_OK(env_->DropUnsyncedFileData()); ASSERT_OK(env_->DeleteFilesCreatedAfterLastDirSync()); break; default: assert(false); } } void PartialCompactTestPreFault(int num_pre_sync, int num_post_sync) { DeleteAllData(); WriteOptions write_options; write_options.sync = sync_use_wal_; Build(write_options, 0, num_pre_sync); if (sync_use_compact_) { db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); } write_options.sync = false; Build(write_options, num_pre_sync, num_post_sync); } void PartialCompactTestReopenWithFault(ResetMethod reset_method, int num_pre_sync, int num_post_sync, Random* rnd = nullptr) { env_->SetFilesystemActive(false); CloseDB(); ResetDBState(reset_method, rnd); ASSERT_OK(OpenDB()); ASSERT_OK(Verify(0, num_pre_sync, FaultInjectionTest::kValExpectFound)); ASSERT_OK(Verify(num_pre_sync, num_post_sync, FaultInjectionTest::kValExpectNoError)); WaitCompactionFinish(); ASSERT_OK(Verify(0, num_pre_sync, FaultInjectionTest::kValExpectFound)); ASSERT_OK(Verify(num_pre_sync, num_post_sync, FaultInjectionTest::kValExpectNoError)); } void NoWriteTestPreFault() { } void NoWriteTestReopenWithFault(ResetMethod reset_method) { CloseDB(); ResetDBState(reset_method); ASSERT_OK(OpenDB()); } void WaitCompactionFinish() { static_cast(db_)->TEST_WaitForCompact(); ASSERT_OK(db_->Put(WriteOptions(), "", "")); } }; TEST_P(FaultInjectionTest, FaultTest) { do { Random rnd(301); for (size_t idx = 0; idx < kNumIterations; idx++) { int num_pre_sync = rnd.Uniform(kMaxNumValues); int num_post_sync = rnd.Uniform(kMaxNumValues); PartialCompactTestPreFault(num_pre_sync, num_post_sync); PartialCompactTestReopenWithFault(kResetDropUnsyncedData, num_pre_sync, num_post_sync); NoWriteTestPreFault(); NoWriteTestReopenWithFault(kResetDropUnsyncedData); PartialCompactTestPreFault(num_pre_sync, num_post_sync); PartialCompactTestReopenWithFault(kResetDropRandomUnsyncedData, num_pre_sync, num_post_sync, &rnd); NoWriteTestPreFault(); NoWriteTestReopenWithFault(kResetDropUnsyncedData); // Setting a separate data path won't pass the test as we don't sync // it after creating new files, PartialCompactTestPreFault(num_pre_sync, num_post_sync); PartialCompactTestReopenWithFault(kResetDropAndDeleteUnsynced, num_pre_sync, num_post_sync); NoWriteTestPreFault(); NoWriteTestReopenWithFault(kResetDropAndDeleteUnsynced); PartialCompactTestPreFault(num_pre_sync, num_post_sync); // No new files created so we expect all values since no files will be // dropped. PartialCompactTestReopenWithFault(kResetDeleteUnsyncedFiles, num_pre_sync, num_post_sync); NoWriteTestPreFault(); NoWriteTestReopenWithFault(kResetDeleteUnsyncedFiles); } } while (ChangeOptions()); } // Previous log file is not fsynced if sync is forced after log rolling. TEST_P(FaultInjectionTest, WriteOptionSyncTest) { test::SleepingBackgroundTask sleeping_task_low; env_->SetBackgroundThreads(1, Env::HIGH); // Block the job queue to prevent flush job from running. env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::HIGH); sleeping_task_low.WaitUntilSleeping(); WriteOptions write_options; write_options.sync = false; std::string key_space, value_space; ASSERT_OK( db_->Put(write_options, Key(1, &key_space), Value(1, &value_space))); FlushOptions flush_options; flush_options.wait = false; ASSERT_OK(db_->Flush(flush_options)); write_options.sync = true; ASSERT_OK( db_->Put(write_options, Key(2, &key_space), Value(2, &value_space))); env_->SetFilesystemActive(false); NoWriteTestReopenWithFault(kResetDropAndDeleteUnsynced); sleeping_task_low.WakeUp(); ASSERT_OK(OpenDB()); std::string val; Value(2, &value_space); ASSERT_OK(ReadValue(2, &val)); ASSERT_EQ(value_space, val); Value(1, &value_space); ASSERT_OK(ReadValue(1, &val)); ASSERT_EQ(value_space, val); } TEST_P(FaultInjectionTest, UninstalledCompaction) { options_.target_file_size_base = 32 * 1024; options_.write_buffer_size = 100 << 10; // 100KB options_.level0_file_num_compaction_trigger = 6; options_.level0_stop_writes_trigger = 1 << 10; options_.level0_slowdown_writes_trigger = 1 << 10; options_.max_background_compactions = 1; OpenDB(); if (!sequential_order_) { rocksdb::SyncPoint::GetInstance()->LoadDependency({ {"FaultInjectionTest::FaultTest:0", "DBImpl::BGWorkCompaction"}, {"CompactionJob::Run():End", "FaultInjectionTest::FaultTest:1"}, {"FaultInjectionTest::FaultTest:2", "DBImpl::BackgroundCompaction:NonTrivial:AfterRun"}, }); } rocksdb::SyncPoint::GetInstance()->EnableProcessing(); int kNumKeys = 1000; Build(WriteOptions(), 0, kNumKeys); FlushOptions flush_options; flush_options.wait = true; db_->Flush(flush_options); ASSERT_OK(db_->Put(WriteOptions(), "", "")); TEST_SYNC_POINT("FaultInjectionTest::FaultTest:0"); TEST_SYNC_POINT("FaultInjectionTest::FaultTest:1"); env_->SetFilesystemActive(false); TEST_SYNC_POINT("FaultInjectionTest::FaultTest:2"); CloseDB(); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); ResetDBState(kResetDropUnsyncedData); std::atomic opened(false); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::Open:Opened", [&](void* arg) { opened.store(true); }); rocksdb::SyncPoint::GetInstance()->SetCallBack( "DBImpl::BGWorkCompaction", [&](void* arg) { ASSERT_TRUE(opened.load()); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); ASSERT_OK(OpenDB()); ASSERT_OK(Verify(0, kNumKeys, FaultInjectionTest::kValExpectFound)); WaitCompactionFinish(); ASSERT_OK(Verify(0, kNumKeys, FaultInjectionTest::kValExpectFound)); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks(); } TEST_P(FaultInjectionTest, ManualLogSyncTest) { test::SleepingBackgroundTask sleeping_task_low; env_->SetBackgroundThreads(1, Env::HIGH); // Block the job queue to prevent flush job from running. env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task_low, Env::Priority::HIGH); sleeping_task_low.WaitUntilSleeping(); WriteOptions write_options; write_options.sync = false; std::string key_space, value_space; ASSERT_OK( db_->Put(write_options, Key(1, &key_space), Value(1, &value_space))); FlushOptions flush_options; flush_options.wait = false; ASSERT_OK(db_->Flush(flush_options)); ASSERT_OK( db_->Put(write_options, Key(2, &key_space), Value(2, &value_space))); ASSERT_OK(db_->SyncWAL()); env_->SetFilesystemActive(false); NoWriteTestReopenWithFault(kResetDropAndDeleteUnsynced); sleeping_task_low.WakeUp(); ASSERT_OK(OpenDB()); std::string val; Value(2, &value_space); ASSERT_OK(ReadValue(2, &val)); ASSERT_EQ(value_space, val); Value(1, &value_space); ASSERT_OK(ReadValue(1, &val)); ASSERT_EQ(value_space, val); } INSTANTIATE_TEST_CASE_P(FaultTest, FaultInjectionTest, ::testing::Bool()); } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/file_indexer.cc000066400000000000000000000170531267530634000165150ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/file_indexer.h" #include #include "rocksdb/comparator.h" #include "db/version_edit.h" namespace rocksdb { FileIndexer::FileIndexer(const Comparator* ucmp) : num_levels_(0), ucmp_(ucmp), level_rb_(nullptr) {} size_t FileIndexer::NumLevelIndex() const { return next_level_index_.size(); } size_t FileIndexer::LevelIndexSize(size_t level) const { if (level >= next_level_index_.size()) { return 0; } return next_level_index_[level].num_index; } void FileIndexer::GetNextLevelIndex(const size_t level, const size_t file_index, const int cmp_smallest, const int cmp_largest, int32_t* left_bound, int32_t* right_bound) const { assert(level > 0); // Last level, no hint if (level == num_levels_ - 1) { *left_bound = 0; *right_bound = -1; return; } assert(level < num_levels_ - 1); assert(static_cast(file_index) <= level_rb_[level]); const IndexUnit* index_units = next_level_index_[level].index_units; const auto& index = index_units[file_index]; if (cmp_smallest < 0) { *left_bound = (level > 0 && file_index > 0) ? index_units[file_index - 1].largest_lb : 0; *right_bound = index.smallest_rb; } else if (cmp_smallest == 0) { *left_bound = index.smallest_lb; *right_bound = index.smallest_rb; } else if (cmp_smallest > 0 && cmp_largest < 0) { *left_bound = index.smallest_lb; *right_bound = index.largest_rb; } else if (cmp_largest == 0) { *left_bound = index.largest_lb; *right_bound = index.largest_rb; } else if (cmp_largest > 0) { *left_bound = index.largest_lb; *right_bound = level_rb_[level + 1]; } else { assert(false); } assert(*left_bound >= 0); assert(*left_bound <= *right_bound + 1); assert(*right_bound <= level_rb_[level + 1]); } void FileIndexer::UpdateIndex(Arena* arena, const size_t num_levels, std::vector* const files) { if (files == nullptr) { return; } if (num_levels == 0) { // uint_32 0-1 would cause bad behavior num_levels_ = num_levels; return; } assert(level_rb_ == nullptr); // level_rb_ should be init here num_levels_ = num_levels; next_level_index_.resize(num_levels); char* mem = arena->AllocateAligned(num_levels_ * sizeof(int32_t)); level_rb_ = new (mem) int32_t[num_levels_]; for (size_t i = 0; i < num_levels_; i++) { level_rb_[i] = -1; } // L1 - Ln-1 for (size_t level = 1; level < num_levels_ - 1; ++level) { const auto& upper_files = files[level]; const int32_t upper_size = static_cast(upper_files.size()); const auto& lower_files = files[level + 1]; level_rb_[level] = static_cast(upper_files.size()) - 1; if (upper_size == 0) { continue; } IndexLevel& index_level = next_level_index_[level]; index_level.num_index = upper_size; mem = arena->AllocateAligned(upper_size * sizeof(IndexUnit)); index_level.index_units = new (mem) IndexUnit[upper_size]; CalculateLB( upper_files, lower_files, &index_level, [this](const FileMetaData * a, const FileMetaData * b)->int { return ucmp_->Compare(a->smallest.user_key(), b->largest.user_key()); }, [](IndexUnit* index, int32_t f_idx) { index->smallest_lb = f_idx; }); CalculateLB( upper_files, lower_files, &index_level, [this](const FileMetaData * a, const FileMetaData * b)->int { return ucmp_->Compare(a->largest.user_key(), b->largest.user_key()); }, [](IndexUnit* index, int32_t f_idx) { index->largest_lb = f_idx; }); CalculateRB( upper_files, lower_files, &index_level, [this](const FileMetaData * a, const FileMetaData * b)->int { return ucmp_->Compare(a->smallest.user_key(), b->smallest.user_key()); }, [](IndexUnit* index, int32_t f_idx) { index->smallest_rb = f_idx; }); CalculateRB( upper_files, lower_files, &index_level, [this](const FileMetaData * a, const FileMetaData * b)->int { return ucmp_->Compare(a->largest.user_key(), b->smallest.user_key()); }, [](IndexUnit* index, int32_t f_idx) { index->largest_rb = f_idx; }); } level_rb_[num_levels_ - 1] = static_cast(files[num_levels_ - 1].size()) - 1; } void FileIndexer::CalculateLB( const std::vector& upper_files, const std::vector& lower_files, IndexLevel* index_level, std::function cmp_op, std::function set_index) { const int32_t upper_size = static_cast(upper_files.size()); const int32_t lower_size = static_cast(lower_files.size()); int32_t upper_idx = 0; int32_t lower_idx = 0; IndexUnit* index = index_level->index_units; while (upper_idx < upper_size && lower_idx < lower_size) { int cmp = cmp_op(upper_files[upper_idx], lower_files[lower_idx]); if (cmp == 0) { set_index(&index[upper_idx], lower_idx); ++upper_idx; ++lower_idx; } else if (cmp > 0) { // Lower level's file (largest) is smaller, a key won't hit in that // file. Move to next lower file ++lower_idx; } else { // Lower level's file becomes larger, update the index, and // move to the next upper file set_index(&index[upper_idx], lower_idx); ++upper_idx; } } while (upper_idx < upper_size) { // Lower files are exhausted, that means the remaining upper files are // greater than any lower files. Set the index to be the lower level size. set_index(&index[upper_idx], lower_size); ++upper_idx; } } void FileIndexer::CalculateRB( const std::vector& upper_files, const std::vector& lower_files, IndexLevel* index_level, std::function cmp_op, std::function set_index) { const int32_t upper_size = static_cast(upper_files.size()); const int32_t lower_size = static_cast(lower_files.size()); int32_t upper_idx = upper_size - 1; int32_t lower_idx = lower_size - 1; IndexUnit* index = index_level->index_units; while (upper_idx >= 0 && lower_idx >= 0) { int cmp = cmp_op(upper_files[upper_idx], lower_files[lower_idx]); if (cmp == 0) { set_index(&index[upper_idx], lower_idx); --upper_idx; --lower_idx; } else if (cmp < 0) { // Lower level's file (smallest) is larger, a key won't hit in that // file. Move to next lower file. --lower_idx; } else { // Lower level's file becomes smaller, update the index, and move to // the next the upper file set_index(&index[upper_idx], lower_idx); --upper_idx; } } while (upper_idx >= 0) { // Lower files are exhausted, that means the remaining upper files are // smaller than any lower files. Set it to -1. set_index(&index[upper_idx], -1); --upper_idx; } } } // namespace rocksdb rocksdb-4.5.1/db/file_indexer.h000066400000000000000000000141241267530634000163530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include "port/port.h" #include "util/arena.h" #include "util/autovector.h" namespace rocksdb { class Comparator; struct FileMetaData; struct FdWithKeyRange; struct FileLevel; // The file tree structure in Version is prebuilt and the range of each file // is known. On Version::Get(), it uses binary search to find a potential file // and then check if a target key can be found in the file by comparing the key // to each file's smallest and largest key. The results of these comparisons // can be reused beyond checking if a key falls into a file's range. // With some pre-calculated knowledge, each key comparison that has been done // can serve as a hint to narrow down further searches: if a key compared to // be smaller than a file's smallest or largest, that comparison can be used // to find out the right bound of next binary search. Similarly, if a key // compared to be larger than a file's smallest or largest, it can be utilized // to find out the left bound of next binary search. // With these hints: it can greatly reduce the range of binary search, // especially for bottom levels, given that one file most likely overlaps with // only N files from level below (where N is max_bytes_for_level_multiplier). // So on level L, we will only look at ~N files instead of N^L files on the // naive approach. class FileIndexer { public: explicit FileIndexer(const Comparator* ucmp); size_t NumLevelIndex() const; size_t LevelIndexSize(size_t level) const; // Return a file index range in the next level to search for a key based on // smallest and largest key comparison for the current file specified by // level and file_index. When *left_index < *right_index, both index should // be valid and fit in the vector size. void GetNextLevelIndex(const size_t level, const size_t file_index, const int cmp_smallest, const int cmp_largest, int32_t* left_bound, int32_t* right_bound) const; void UpdateIndex(Arena* arena, const size_t num_levels, std::vector* const files); enum { // MSVC version 1800 still does not have constexpr for ::max() kLevelMaxIndex = rocksdb::port::kMaxInt32 }; private: size_t num_levels_; const Comparator* ucmp_; struct IndexUnit { IndexUnit() : smallest_lb(0), largest_lb(0), smallest_rb(-1), largest_rb(-1) {} // During file search, a key is compared against smallest and largest // from a FileMetaData. It can have 3 possible outcomes: // (1) key is smaller than smallest, implying it is also smaller than // larger. Precalculated index based on "smallest < smallest" can // be used to provide right bound. // (2) key is in between smallest and largest. // Precalculated index based on "smallest > greatest" can be used to // provide left bound. // Precalculated index based on "largest < smallest" can be used to // provide right bound. // (3) key is larger than largest, implying it is also larger than smallest. // Precalculated index based on "largest > largest" can be used to // provide left bound. // // As a result, we will need to do: // Compare smallest (<=) and largest keys from upper level file with // smallest key from lower level to get a right bound. // Compare smallest (>=) and largest keys from upper level file with // largest key from lower level to get a left bound. // // Example: // level 1: [50 - 60] // level 2: [1 - 40], [45 - 55], [58 - 80] // A key 35, compared to be less than 50, 3rd file on level 2 can be // skipped according to rule (1). LB = 0, RB = 1. // A key 53, sits in the middle 50 and 60. 1st file on level 2 can be // skipped according to rule (2)-a, but the 3rd file cannot be skipped // because 60 is greater than 58. LB = 1, RB = 2. // A key 70, compared to be larger than 60. 1st and 2nd file can be skipped // according to rule (3). LB = 2, RB = 2. // // Point to a left most file in a lower level that may contain a key, // which compares greater than smallest of a FileMetaData (upper level) int32_t smallest_lb; // Point to a left most file in a lower level that may contain a key, // which compares greater than largest of a FileMetaData (upper level) int32_t largest_lb; // Point to a right most file in a lower level that may contain a key, // which compares smaller than smallest of a FileMetaData (upper level) int32_t smallest_rb; // Point to a right most file in a lower level that may contain a key, // which compares smaller than largest of a FileMetaData (upper level) int32_t largest_rb; }; // Data structure to store IndexUnits in a whole level struct IndexLevel { size_t num_index; IndexUnit* index_units; IndexLevel() : num_index(0), index_units(nullptr) {} }; void CalculateLB( const std::vector& upper_files, const std::vector& lower_files, IndexLevel* index_level, std::function cmp_op, std::function set_index); void CalculateRB( const std::vector& upper_files, const std::vector& lower_files, IndexLevel* index_level, std::function cmp_op, std::function set_index); autovector next_level_index_; int32_t* level_rb_; }; } // namespace rocksdb rocksdb-4.5.1/db/file_indexer_test.cc000066400000000000000000000235241267530634000175540ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include #include "db/file_indexer.h" #include "db/dbformat.h" #include "db/version_edit.h" #include "port/stack_trace.h" #include "rocksdb/comparator.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class IntComparator : public Comparator { public: int Compare(const Slice& a, const Slice& b) const override { assert(a.size() == 8); assert(b.size() == 8); int64_t diff = *reinterpret_cast(a.data()) - *reinterpret_cast(b.data()); if (diff < 0) { return -1; } else if (diff == 0) { return 0; } else { return 1; } } const char* Name() const override { return "IntComparator"; } void FindShortestSeparator(std::string* start, const Slice& limit) const override {} void FindShortSuccessor(std::string* key) const override {} }; class FileIndexerTest : public testing::Test { public: FileIndexerTest() : kNumLevels(4), files(new std::vector[kNumLevels]) {} ~FileIndexerTest() { ClearFiles(); delete[] files; } void AddFile(int level, int64_t smallest, int64_t largest) { auto* f = new FileMetaData(); f->smallest = IntKey(smallest); f->largest = IntKey(largest); files[level].push_back(f); } InternalKey IntKey(int64_t v) { return InternalKey(Slice(reinterpret_cast(&v), 8), 0, kTypeValue); } void ClearFiles() { for (uint32_t i = 0; i < kNumLevels; ++i) { for (auto* f : files[i]) { delete f; } files[i].clear(); } } void GetNextLevelIndex(const uint32_t level, const uint32_t file_index, const int cmp_smallest, const int cmp_largest, int32_t* left_index, int32_t* right_index) { *left_index = 100; *right_index = 100; indexer->GetNextLevelIndex(level, file_index, cmp_smallest, cmp_largest, left_index, right_index); } int32_t left = 100; int32_t right = 100; const uint32_t kNumLevels; IntComparator ucmp; FileIndexer* indexer; std::vector* files; }; // Case 0: Empty TEST_F(FileIndexerTest, Empty) { Arena arena; indexer = new FileIndexer(&ucmp); indexer->UpdateIndex(&arena, 0, files); delete indexer; } // Case 1: no overlap, files are on the left of next level files TEST_F(FileIndexerTest, no_overlap_left) { Arena arena; indexer = new FileIndexer(&ucmp); // level 1 AddFile(1, 100, 200); AddFile(1, 300, 400); AddFile(1, 500, 600); // level 2 AddFile(2, 1500, 1600); AddFile(2, 1601, 1699); AddFile(2, 1700, 1800); // level 3 AddFile(3, 2500, 2600); AddFile(3, 2601, 2699); AddFile(3, 2700, 2800); indexer->UpdateIndex(&arena, kNumLevels, files); for (uint32_t level = 1; level < 3; ++level) { for (uint32_t f = 0; f < 3; ++f) { GetNextLevelIndex(level, f, -1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(level, f, 0, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(level, f, 1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(level, f, 1, 0, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(level, f, 1, 1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(2, right); } } delete indexer; ClearFiles(); } // Case 2: no overlap, files are on the right of next level files TEST_F(FileIndexerTest, no_overlap_right) { Arena arena; indexer = new FileIndexer(&ucmp); // level 1 AddFile(1, 2100, 2200); AddFile(1, 2300, 2400); AddFile(1, 2500, 2600); // level 2 AddFile(2, 1500, 1600); AddFile(2, 1501, 1699); AddFile(2, 1700, 1800); // level 3 AddFile(3, 500, 600); AddFile(3, 501, 699); AddFile(3, 700, 800); indexer->UpdateIndex(&arena, kNumLevels, files); for (uint32_t level = 1; level < 3; ++level) { for (uint32_t f = 0; f < 3; ++f) { GetNextLevelIndex(level, f, -1, -1, &left, &right); ASSERT_EQ(f == 0 ? 0 : 3, left); ASSERT_EQ(2, right); GetNextLevelIndex(level, f, 0, -1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(level, f, 1, -1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(level, f, 1, -1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(level, f, 1, 0, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(level, f, 1, 1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); } } delete indexer; } // Case 3: empty L2 TEST_F(FileIndexerTest, empty_L2) { Arena arena; indexer = new FileIndexer(&ucmp); for (uint32_t i = 1; i < kNumLevels; ++i) { ASSERT_EQ(0U, indexer->LevelIndexSize(i)); } // level 1 AddFile(1, 2100, 2200); AddFile(1, 2300, 2400); AddFile(1, 2500, 2600); // level 3 AddFile(3, 500, 600); AddFile(3, 501, 699); AddFile(3, 700, 800); indexer->UpdateIndex(&arena, kNumLevels, files); for (uint32_t f = 0; f < 3; ++f) { GetNextLevelIndex(1, f, -1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(1, f, 0, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(1, f, 1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(1, f, 1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(1, f, 1, 0, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); GetNextLevelIndex(1, f, 1, 1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(-1, right); } delete indexer; ClearFiles(); } // Case 4: mixed TEST_F(FileIndexerTest, mixed) { Arena arena; indexer = new FileIndexer(&ucmp); // level 1 AddFile(1, 100, 200); AddFile(1, 250, 400); AddFile(1, 450, 500); // level 2 AddFile(2, 100, 150); // 0 AddFile(2, 200, 250); // 1 AddFile(2, 251, 300); // 2 AddFile(2, 301, 350); // 3 AddFile(2, 500, 600); // 4 // level 3 AddFile(3, 0, 50); AddFile(3, 100, 200); AddFile(3, 201, 250); indexer->UpdateIndex(&arena, kNumLevels, files); // level 1, 0 GetNextLevelIndex(1, 0, -1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(0, right); GetNextLevelIndex(1, 0, 0, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(0, right); GetNextLevelIndex(1, 0, 1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(1, right); GetNextLevelIndex(1, 0, 1, 0, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(1, 0, 1, 1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(4, right); // level 1, 1 GetNextLevelIndex(1, 1, -1, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(1, 1, 0, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(1, 1, 1, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(3, right); GetNextLevelIndex(1, 1, 1, 0, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(3, right); GetNextLevelIndex(1, 1, 1, 1, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(4, right); // level 1, 2 GetNextLevelIndex(1, 2, -1, -1, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(3, right); GetNextLevelIndex(1, 2, 0, -1, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(3, right); GetNextLevelIndex(1, 2, 1, -1, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(4, right); GetNextLevelIndex(1, 2, 1, 0, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(4, right); GetNextLevelIndex(1, 2, 1, 1, &left, &right); ASSERT_EQ(4, left); ASSERT_EQ(4, right); // level 2, 0 GetNextLevelIndex(2, 0, -1, -1, &left, &right); ASSERT_EQ(0, left); ASSERT_EQ(1, right); GetNextLevelIndex(2, 0, 0, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(2, 0, 1, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(2, 0, 1, 0, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(2, 0, 1, 1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(2, right); // level 2, 1 GetNextLevelIndex(2, 1, -1, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(2, 1, 0, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(1, right); GetNextLevelIndex(2, 1, 1, -1, &left, &right); ASSERT_EQ(1, left); ASSERT_EQ(2, right); GetNextLevelIndex(2, 1, 1, 0, &left, &right); ASSERT_EQ(2, left); ASSERT_EQ(2, right); GetNextLevelIndex(2, 1, 1, 1, &left, &right); ASSERT_EQ(2, left); ASSERT_EQ(2, right); // level 2, [2 - 4], no overlap for (uint32_t f = 2; f <= 4; ++f) { GetNextLevelIndex(2, f, -1, -1, &left, &right); ASSERT_EQ(f == 2 ? 2 : 3, left); ASSERT_EQ(2, right); GetNextLevelIndex(2, f, 0, -1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(2, f, 1, -1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(2, f, 1, 0, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); GetNextLevelIndex(2, f, 1, 1, &left, &right); ASSERT_EQ(3, left); ASSERT_EQ(2, right); } delete indexer; ClearFiles(); } } // namespace rocksdb int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/filename.cc000066400000000000000000000302561267530634000156400ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include "db/filename.h" #include #include #include #include #include "db/dbformat.h" #include "rocksdb/env.h" #include "util/file_reader_writer.h" #include "util/logging.h" #include "util/stop_watch.h" #include "util/string_util.h" #include "util/sync_point.h" namespace rocksdb { static const std::string kRocksDbTFileExt = "sst"; static const std::string kLevelDbTFileExt = "ldb"; // Given a path, flatten the path name by replacing all chars not in // {[0-9,a-z,A-Z,-,_,.]} with _. And append '_LOG\0' at the end. // Return the number of chars stored in dest not including the trailing '\0'. static size_t GetInfoLogPrefix(const std::string& path, char* dest, int len) { const char suffix[] = "_LOG"; size_t write_idx = 0; size_t i = 0; size_t src_len = path.size(); while (i < src_len && write_idx < len - sizeof(suffix)) { if ((path[i] >= 'a' && path[i] <= 'z') || (path[i] >= '0' && path[i] <= '9') || (path[i] >= 'A' && path[i] <= 'Z') || path[i] == '-' || path[i] == '.' || path[i] == '_'){ dest[write_idx++] = path[i]; } else { if (i > 0) { dest[write_idx++] = '_'; } } i++; } assert(sizeof(suffix) <= len - write_idx); // "\0" is automatically added by snprintf snprintf(dest + write_idx, len - write_idx, suffix); write_idx += sizeof(suffix) - 1; return write_idx; } static std::string MakeFileName(const std::string& name, uint64_t number, const char* suffix) { char buf[100]; snprintf(buf, sizeof(buf), "/%06llu.%s", static_cast(number), suffix); return name + buf; } std::string LogFileName(const std::string& name, uint64_t number) { assert(number > 0); return MakeFileName(name, number, "log"); } std::string ArchivalDirectory(const std::string& dir) { return dir + "/" + ARCHIVAL_DIR; } std::string ArchivedLogFileName(const std::string& name, uint64_t number) { assert(number > 0); return MakeFileName(name + "/" + ARCHIVAL_DIR, number, "log"); } std::string MakeTableFileName(const std::string& path, uint64_t number) { return MakeFileName(path, number, kRocksDbTFileExt.c_str()); } std::string Rocks2LevelTableFileName(const std::string& fullname) { assert(fullname.size() > kRocksDbTFileExt.size() + 1); if (fullname.size() <= kRocksDbTFileExt.size() + 1) { return ""; } return fullname.substr(0, fullname.size() - kRocksDbTFileExt.size()) + kLevelDbTFileExt; } uint64_t TableFileNameToNumber(const std::string& name) { uint64_t number = 0; uint64_t base = 1; int pos = static_cast(name.find_last_of('.')); while (--pos >= 0 && name[pos] >= '0' && name[pos] <= '9') { number += (name[pos] - '0') * base; base *= 10; } return number; } std::string TableFileName(const std::vector& db_paths, uint64_t number, uint32_t path_id) { assert(number > 0); std::string path; if (path_id >= db_paths.size()) { path = db_paths.back().path; } else { path = db_paths[path_id].path; } return MakeTableFileName(path, number); } void FormatFileNumber(uint64_t number, uint32_t path_id, char* out_buf, size_t out_buf_size) { if (path_id == 0) { snprintf(out_buf, out_buf_size, "%" PRIu64, number); } else { snprintf(out_buf, out_buf_size, "%" PRIu64 "(path " "%" PRIu32 ")", number, path_id); } } std::string DescriptorFileName(const std::string& dbname, uint64_t number) { assert(number > 0); char buf[100]; snprintf(buf, sizeof(buf), "/MANIFEST-%06llu", static_cast(number)); return dbname + buf; } std::string CurrentFileName(const std::string& dbname) { return dbname + "/CURRENT"; } std::string LockFileName(const std::string& dbname) { return dbname + "/LOCK"; } std::string TempFileName(const std::string& dbname, uint64_t number) { return MakeFileName(dbname, number, kTempFileNameSuffix.c_str()); } InfoLogPrefix::InfoLogPrefix(bool has_log_dir, const std::string& db_absolute_path) { if (!has_log_dir) { const char kInfoLogPrefix[] = "LOG"; // "\0" is automatically added to the end snprintf(buf, sizeof(buf), kInfoLogPrefix); prefix = Slice(buf, sizeof(kInfoLogPrefix) - 1); } else { size_t len = GetInfoLogPrefix(db_absolute_path, buf, sizeof(buf)); prefix = Slice(buf, len); } } std::string InfoLogFileName(const std::string& dbname, const std::string& db_path, const std::string& log_dir) { if (log_dir.empty()) { return dbname + "/LOG"; } InfoLogPrefix info_log_prefix(true, db_path); return log_dir + "/" + info_log_prefix.buf; } // Return the name of the old info log file for "dbname". std::string OldInfoLogFileName(const std::string& dbname, uint64_t ts, const std::string& db_path, const std::string& log_dir) { char buf[50]; snprintf(buf, sizeof(buf), "%llu", static_cast(ts)); if (log_dir.empty()) { return dbname + "/LOG.old." + buf; } InfoLogPrefix info_log_prefix(true, db_path); return log_dir + "/" + info_log_prefix.buf + ".old." + buf; } std::string OptionsFileName(const std::string& dbname, uint64_t file_num) { char buffer[256]; snprintf(buffer, sizeof(buffer), "%s%06" PRIu64, kOptionsFileNamePrefix.c_str(), file_num); return dbname + "/" + buffer; } std::string TempOptionsFileName(const std::string& dbname, uint64_t file_num) { char buffer[256]; snprintf(buffer, sizeof(buffer), "%s%06" PRIu64 ".%s", kOptionsFileNamePrefix.c_str(), file_num, kTempFileNameSuffix.c_str()); return dbname + "/" + buffer; } std::string MetaDatabaseName(const std::string& dbname, uint64_t number) { char buf[100]; snprintf(buf, sizeof(buf), "/METADB-%llu", static_cast(number)); return dbname + buf; } std::string IdentityFileName(const std::string& dbname) { return dbname + "/IDENTITY"; } // Owned filenames have the form: // dbname/IDENTITY // dbname/CURRENT // dbname/LOCK // dbname/ // dbname/.old.[0-9]+ // dbname/MANIFEST-[0-9]+ // dbname/[0-9]+.(log|sst) // dbname/METADB-[0-9]+ // dbname/OPTIONS-[0-9]+ // dbname/OPTIONS-[0-9]+.dbtmp // Disregards / at the beginning bool ParseFileName(const std::string& fname, uint64_t* number, FileType* type, WalFileType* log_type) { return ParseFileName(fname, number, "", type, log_type); } bool ParseFileName(const std::string& fname, uint64_t* number, const Slice& info_log_name_prefix, FileType* type, WalFileType* log_type) { Slice rest(fname); if (fname.length() > 1 && fname[0] == '/') { rest.remove_prefix(1); } if (rest == "IDENTITY") { *number = 0; *type = kIdentityFile; } else if (rest == "CURRENT") { *number = 0; *type = kCurrentFile; } else if (rest == "LOCK") { *number = 0; *type = kDBLockFile; } else if (info_log_name_prefix.size() > 0 && rest.starts_with(info_log_name_prefix)) { rest.remove_prefix(info_log_name_prefix.size()); if (rest == "" || rest == ".old") { *number = 0; *type = kInfoLogFile; } else if (rest.starts_with(".old.")) { uint64_t ts_suffix; // sizeof also counts the trailing '\0'. rest.remove_prefix(sizeof(".old.") - 1); if (!ConsumeDecimalNumber(&rest, &ts_suffix)) { return false; } *number = ts_suffix; *type = kInfoLogFile; } } else if (rest.starts_with("MANIFEST-")) { rest.remove_prefix(strlen("MANIFEST-")); uint64_t num; if (!ConsumeDecimalNumber(&rest, &num)) { return false; } if (!rest.empty()) { return false; } *type = kDescriptorFile; *number = num; } else if (rest.starts_with("METADB-")) { rest.remove_prefix(strlen("METADB-")); uint64_t num; if (!ConsumeDecimalNumber(&rest, &num)) { return false; } if (!rest.empty()) { return false; } *type = kMetaDatabase; *number = num; } else if (rest.starts_with(kOptionsFileNamePrefix)) { uint64_t ts_suffix; bool is_temp_file = false; rest.remove_prefix(kOptionsFileNamePrefix.size()); const std::string kTempFileNameSuffixWithDot = std::string(".") + kTempFileNameSuffix; if (rest.ends_with(kTempFileNameSuffixWithDot)) { rest.remove_suffix(kTempFileNameSuffixWithDot.size()); is_temp_file = true; } if (!ConsumeDecimalNumber(&rest, &ts_suffix)) { return false; } *number = ts_suffix; *type = is_temp_file ? kTempFile : kOptionsFile; } else { // Avoid strtoull() to keep filename format independent of the // current locale bool archive_dir_found = false; if (rest.starts_with(ARCHIVAL_DIR)) { if (rest.size() <= ARCHIVAL_DIR.size()) { return false; } rest.remove_prefix(ARCHIVAL_DIR.size() + 1); // Add 1 to remove / also if (log_type) { *log_type = kArchivedLogFile; } archive_dir_found = true; } uint64_t num; if (!ConsumeDecimalNumber(&rest, &num)) { return false; } if (rest.size() <= 1 || rest[0] != '.') { return false; } rest.remove_prefix(1); Slice suffix = rest; if (suffix == Slice("log")) { *type = kLogFile; if (log_type && !archive_dir_found) { *log_type = kAliveLogFile; } } else if (archive_dir_found) { return false; // Archive dir can contain only log files } else if (suffix == Slice(kRocksDbTFileExt) || suffix == Slice(kLevelDbTFileExt)) { *type = kTableFile; } else if (suffix == Slice(kTempFileNameSuffix)) { *type = kTempFile; } else { return false; } *number = num; } return true; } Status SetCurrentFile(Env* env, const std::string& dbname, uint64_t descriptor_number, Directory* directory_to_fsync) { // Remove leading "dbname/" and add newline to manifest file name std::string manifest = DescriptorFileName(dbname, descriptor_number); Slice contents = manifest; assert(contents.starts_with(dbname + "/")); contents.remove_prefix(dbname.size() + 1); std::string tmp = TempFileName(dbname, descriptor_number); Status s = WriteStringToFile(env, contents.ToString() + "\n", tmp, true); if (s.ok()) { TEST_KILL_RANDOM("SetCurrentFile:0", rocksdb_kill_odds * REDUCE_ODDS2); s = env->RenameFile(tmp, CurrentFileName(dbname)); TEST_KILL_RANDOM("SetCurrentFile:1", rocksdb_kill_odds * REDUCE_ODDS2); } if (s.ok()) { if (directory_to_fsync != nullptr) { directory_to_fsync->Fsync(); } } else { env->DeleteFile(tmp); } return s; } Status SetIdentityFile(Env* env, const std::string& dbname) { std::string id = env->GenerateUniqueId(); assert(!id.empty()); // Reserve the filename dbname/000000.dbtmp for the temporary identity file std::string tmp = TempFileName(dbname, 0); Status s = WriteStringToFile(env, id, tmp, true); if (s.ok()) { s = env->RenameFile(tmp, IdentityFileName(dbname)); } if (!s.ok()) { env->DeleteFile(tmp); } return s; } Status SyncManifest(Env* env, const DBOptions* db_options, WritableFileWriter* file) { TEST_KILL_RANDOM("SyncManifest:0", rocksdb_kill_odds * REDUCE_ODDS2); if (db_options->disableDataSync) { return Status::OK(); } else { StopWatch sw(env, db_options->statistics.get(), MANIFEST_FILE_SYNC_MICROS); return file->Sync(db_options->use_fsync); } } } // namespace rocksdb rocksdb-4.5.1/db/filename.h000066400000000000000000000146331267530634000155030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // File names used by DB code #pragma once #include #include #include #include #include "port/port.h" #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/status.h" #include "rocksdb/transaction_log.h" namespace rocksdb { class Env; class Directory; class WritableFileWriter; enum FileType { kLogFile, kDBLockFile, kTableFile, kDescriptorFile, kCurrentFile, kTempFile, kInfoLogFile, // Either the current one, or an old one kMetaDatabase, kIdentityFile, kOptionsFile }; // Return the name of the log file with the specified number // in the db named by "dbname". The result will be prefixed with // "dbname". extern std::string LogFileName(const std::string& dbname, uint64_t number); static const std::string ARCHIVAL_DIR = "archive"; extern std::string ArchivalDirectory(const std::string& dbname); // Return the name of the archived log file with the specified number // in the db named by "dbname". The result will be prefixed with "dbname". extern std::string ArchivedLogFileName(const std::string& dbname, uint64_t num); extern std::string MakeTableFileName(const std::string& name, uint64_t number); // Return the name of sstable with LevelDB suffix // created from RocksDB sstable suffixed name extern std::string Rocks2LevelTableFileName(const std::string& fullname); // the reverse function of MakeTableFileName // TODO(yhchiang): could merge this function with ParseFileName() extern uint64_t TableFileNameToNumber(const std::string& name); // Return the name of the sstable with the specified number // in the db named by "dbname". The result will be prefixed with // "dbname". extern std::string TableFileName(const std::vector& db_paths, uint64_t number, uint32_t path_id); // Sufficient buffer size for FormatFileNumber. const size_t kFormatFileNumberBufSize = 38; extern void FormatFileNumber(uint64_t number, uint32_t path_id, char* out_buf, size_t out_buf_size); // Return the name of the descriptor file for the db named by // "dbname" and the specified incarnation number. The result will be // prefixed with "dbname". extern std::string DescriptorFileName(const std::string& dbname, uint64_t number); // Return the name of the current file. This file contains the name // of the current manifest file. The result will be prefixed with // "dbname". extern std::string CurrentFileName(const std::string& dbname); // Return the name of the lock file for the db named by // "dbname". The result will be prefixed with "dbname". extern std::string LockFileName(const std::string& dbname); // Return the name of a temporary file owned by the db named "dbname". // The result will be prefixed with "dbname". extern std::string TempFileName(const std::string& dbname, uint64_t number); // A helper structure for prefix of info log names. struct InfoLogPrefix { char buf[260]; Slice prefix; // Prefix with DB absolute path encoded explicit InfoLogPrefix(bool has_log_dir, const std::string& db_absolute_path); // Default Prefix explicit InfoLogPrefix(); }; // Return the name of the info log file for "dbname". extern std::string InfoLogFileName(const std::string& dbname, const std::string& db_path = "", const std::string& log_dir = ""); // Return the name of the old info log file for "dbname". extern std::string OldInfoLogFileName(const std::string& dbname, uint64_t ts, const std::string& db_path = "", const std::string& log_dir = ""); static const std::string kOptionsFileNamePrefix = "OPTIONS-"; static const std::string kTempFileNameSuffix = "dbtmp"; // Return a options file name given the "dbname" and file number. // Format: OPTIONS-[number].dbtmp extern std::string OptionsFileName(const std::string& dbname, uint64_t file_num); // Return a temp options file name given the "dbname" and file number. // Format: OPTIONS-[number] extern std::string TempOptionsFileName(const std::string& dbname, uint64_t file_num); // Return the name to use for a metadatabase. The result will be prefixed with // "dbname". extern std::string MetaDatabaseName(const std::string& dbname, uint64_t number); // Return the name of the Identity file which stores a unique number for the db // that will get regenerated if the db loses all its data and is recreated fresh // either from a backup-image or empty extern std::string IdentityFileName(const std::string& dbname); // If filename is a rocksdb file, store the type of the file in *type. // The number encoded in the filename is stored in *number. If the // filename was successfully parsed, returns true. Else return false. // info_log_name_prefix is the path of info logs. extern bool ParseFileName(const std::string& filename, uint64_t* number, const Slice& info_log_name_prefix, FileType* type, WalFileType* log_type = nullptr); // Same as previous function, but skip info log files. extern bool ParseFileName(const std::string& filename, uint64_t* number, FileType* type, WalFileType* log_type = nullptr); // Make the CURRENT file point to the descriptor file with the // specified number. extern Status SetCurrentFile(Env* env, const std::string& dbname, uint64_t descriptor_number, Directory* directory_to_fsync); // Make the IDENTITY file for the db extern Status SetIdentityFile(Env* env, const std::string& dbname); // Sync manifest file `file`. extern Status SyncManifest(Env* env, const DBOptions* db_options, WritableFileWriter* file); } // namespace rocksdb rocksdb-4.5.1/db/filename_test.cc000066400000000000000000000132341267530634000166740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/filename.h" #include "db/dbformat.h" #include "port/port.h" #include "util/logging.h" #include "util/testharness.h" namespace rocksdb { class FileNameTest : public testing::Test {}; TEST_F(FileNameTest, Parse) { Slice db; FileType type; uint64_t number; char kDefautInfoLogDir = 1; char kDifferentInfoLogDir = 2; char kNoCheckLogDir = 4; char kAllMode = kDefautInfoLogDir | kDifferentInfoLogDir | kNoCheckLogDir; // Successful parses static struct { const char* fname; uint64_t number; FileType type; char mode; } cases[] = { {"100.log", 100, kLogFile, kAllMode}, {"0.log", 0, kLogFile, kAllMode}, {"0.sst", 0, kTableFile, kAllMode}, {"CURRENT", 0, kCurrentFile, kAllMode}, {"LOCK", 0, kDBLockFile, kAllMode}, {"MANIFEST-2", 2, kDescriptorFile, kAllMode}, {"MANIFEST-7", 7, kDescriptorFile, kAllMode}, {"METADB-2", 2, kMetaDatabase, kAllMode}, {"METADB-7", 7, kMetaDatabase, kAllMode}, {"LOG", 0, kInfoLogFile, kDefautInfoLogDir}, {"LOG.old", 0, kInfoLogFile, kDefautInfoLogDir}, {"LOG.old.6688", 6688, kInfoLogFile, kDefautInfoLogDir}, {"rocksdb_dir_LOG", 0, kInfoLogFile, kDifferentInfoLogDir}, {"rocksdb_dir_LOG.old", 0, kInfoLogFile, kDifferentInfoLogDir}, {"rocksdb_dir_LOG.old.6688", 6688, kInfoLogFile, kDifferentInfoLogDir}, {"18446744073709551615.log", 18446744073709551615ull, kLogFile, kAllMode}, }; for (char mode : {kDifferentInfoLogDir, kDefautInfoLogDir, kNoCheckLogDir}) { for (unsigned int i = 0; i < sizeof(cases) / sizeof(cases[0]); i++) { InfoLogPrefix info_log_prefix(mode != kDefautInfoLogDir, "/rocksdb/dir"); if (cases[i].mode & mode) { std::string f = cases[i].fname; if (mode == kNoCheckLogDir) { ASSERT_TRUE(ParseFileName(f, &number, &type)) << f; } else { ASSERT_TRUE(ParseFileName(f, &number, info_log_prefix.prefix, &type)) << f; } ASSERT_EQ(cases[i].type, type) << f; ASSERT_EQ(cases[i].number, number) << f; } } } // Errors static const char* errors[] = { "", "foo", "foo-dx-100.log", ".log", "", "manifest", "CURREN", "CURRENTX", "MANIFES", "MANIFEST", "MANIFEST-", "XMANIFEST-3", "MANIFEST-3x", "META", "METADB", "METADB-", "XMETADB-3", "METADB-3x", "LOC", "LOCKx", "LO", "LOGx", "18446744073709551616.log", "184467440737095516150.log", "100", "100.", "100.lop" }; for (unsigned int i = 0; i < sizeof(errors) / sizeof(errors[0]); i++) { std::string f = errors[i]; ASSERT_TRUE(!ParseFileName(f, &number, &type)) << f; }; } TEST_F(FileNameTest, InfoLogFileName) { std::string dbname = ("/data/rocksdb"); std::string db_absolute_path; Env::Default()->GetAbsolutePath(dbname, &db_absolute_path); ASSERT_EQ("/data/rocksdb/LOG", InfoLogFileName(dbname, db_absolute_path, "")); ASSERT_EQ("/data/rocksdb/LOG.old.666", OldInfoLogFileName(dbname, 666u, db_absolute_path, "")); ASSERT_EQ("/data/rocksdb_log/data_rocksdb_LOG", InfoLogFileName(dbname, db_absolute_path, "/data/rocksdb_log")); ASSERT_EQ( "/data/rocksdb_log/data_rocksdb_LOG.old.666", OldInfoLogFileName(dbname, 666u, db_absolute_path, "/data/rocksdb_log")); } TEST_F(FileNameTest, Construction) { uint64_t number; FileType type; std::string fname; fname = CurrentFileName("foo"); ASSERT_EQ("foo/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(0U, number); ASSERT_EQ(kCurrentFile, type); fname = LockFileName("foo"); ASSERT_EQ("foo/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(0U, number); ASSERT_EQ(kDBLockFile, type); fname = LogFileName("foo", 192); ASSERT_EQ("foo/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(192U, number); ASSERT_EQ(kLogFile, type); fname = TableFileName({DbPath("bar", 0)}, 200, 0); std::string fname1 = TableFileName({DbPath("foo", 0), DbPath("bar", 0)}, 200, 1); ASSERT_EQ(fname, fname1); ASSERT_EQ("bar/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(200U, number); ASSERT_EQ(kTableFile, type); fname = DescriptorFileName("bar", 100); ASSERT_EQ("bar/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(100U, number); ASSERT_EQ(kDescriptorFile, type); fname = TempFileName("tmp", 999); ASSERT_EQ("tmp/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(999U, number); ASSERT_EQ(kTempFile, type); fname = MetaDatabaseName("met", 100); ASSERT_EQ("met/", std::string(fname.data(), 4)); ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(100U, number); ASSERT_EQ(kMetaDatabase, type); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/flush_job.cc000066400000000000000000000263341267530634000160350ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/flush_job.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include "db/builder.h" #include "db/db_iter.h" #include "db/dbformat.h" #include "db/event_helpers.h" #include "db/filename.h" #include "db/log_reader.h" #include "db/log_writer.h" #include "db/memtable.h" #include "db/memtable_list.h" #include "db/merge_context.h" #include "db/version_set.h" #include "port/likely.h" #include "port/port.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/statistics.h" #include "rocksdb/status.h" #include "rocksdb/table.h" #include "table/block.h" #include "table/block_based_table_factory.h" #include "table/merger.h" #include "table/table_builder.h" #include "table/two_level_iterator.h" #include "util/coding.h" #include "util/event_logger.h" #include "util/file_util.h" #include "util/iostats_context_imp.h" #include "util/log_buffer.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/perf_context_imp.h" #include "util/stop_watch.h" #include "util/sync_point.h" #include "util/thread_status_util.h" namespace rocksdb { FlushJob::FlushJob(const std::string& dbname, ColumnFamilyData* cfd, const DBOptions& db_options, const MutableCFOptions& mutable_cf_options, const EnvOptions& env_options, VersionSet* versions, InstrumentedMutex* db_mutex, std::atomic* shutting_down, std::vector existing_snapshots, SequenceNumber earliest_write_conflict_snapshot, JobContext* job_context, LogBuffer* log_buffer, Directory* db_directory, Directory* output_file_directory, CompressionType output_compression, Statistics* stats, EventLogger* event_logger) : dbname_(dbname), cfd_(cfd), db_options_(db_options), mutable_cf_options_(mutable_cf_options), env_options_(env_options), versions_(versions), db_mutex_(db_mutex), shutting_down_(shutting_down), existing_snapshots_(std::move(existing_snapshots)), earliest_write_conflict_snapshot_(earliest_write_conflict_snapshot), job_context_(job_context), log_buffer_(log_buffer), db_directory_(db_directory), output_file_directory_(output_file_directory), output_compression_(output_compression), stats_(stats), event_logger_(event_logger) { // Update the thread status to indicate flush. ReportStartedFlush(); TEST_SYNC_POINT("FlushJob::FlushJob()"); } FlushJob::~FlushJob() { ThreadStatusUtil::ResetThreadStatus(); } void FlushJob::ReportStartedFlush() { ThreadStatusUtil::SetColumnFamily(cfd_, cfd_->ioptions()->env, cfd_->options()->enable_thread_tracking); ThreadStatusUtil::SetThreadOperation(ThreadStatus::OP_FLUSH); ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::COMPACTION_JOB_ID, job_context_->job_id); IOSTATS_RESET(bytes_written); } void FlushJob::ReportFlushInputSize(const autovector& mems) { uint64_t input_size = 0; for (auto* mem : mems) { input_size += mem->ApproximateMemoryUsage(); } ThreadStatusUtil::IncreaseThreadOperationProperty( ThreadStatus::FLUSH_BYTES_MEMTABLES, input_size); } void FlushJob::RecordFlushIOStats() { ThreadStatusUtil::SetThreadOperationProperty( ThreadStatus::FLUSH_BYTES_WRITTEN, IOSTATS(bytes_written)); } Status FlushJob::Run(FileMetaData* file_meta) { AutoThreadOperationStageUpdater stage_run( ThreadStatus::STAGE_FLUSH_RUN); // Save the contents of the earliest memtable as a new Table FileMetaData meta; autovector mems; cfd_->imm()->PickMemtablesToFlush(&mems); if (mems.empty()) { LogToBuffer(log_buffer_, "[%s] Nothing in memtable to flush", cfd_->GetName().c_str()); return Status::OK(); } ReportFlushInputSize(mems); // entries mems are (implicitly) sorted in ascending order by their created // time. We will use the first memtable's `edit` to keep the meta info for // this flush. MemTable* m = mems[0]; VersionEdit* edit = m->GetEdits(); edit->SetPrevLogNumber(0); // SetLogNumber(log_num) indicates logs with number smaller than log_num // will no longer be picked up for recovery. edit->SetLogNumber(mems.back()->GetNextLogNumber()); edit->SetColumnFamily(cfd_->GetID()); // This will release and re-acquire the mutex. Status s = WriteLevel0Table(mems, edit, &meta); if (s.ok() && (shutting_down_->load(std::memory_order_acquire) || cfd_->IsDropped())) { s = Status::ShutdownInProgress( "Database shutdown or Column family drop during flush"); } if (!s.ok()) { cfd_->imm()->RollbackMemtableFlush(mems, meta.fd.GetNumber()); } else { TEST_SYNC_POINT("FlushJob::InstallResults"); // Replace immutable memtable with the generated Table s = cfd_->imm()->InstallMemtableFlushResults( cfd_, mutable_cf_options_, mems, versions_, db_mutex_, meta.fd.GetNumber(), &job_context_->memtables_to_free, db_directory_, log_buffer_); } if (s.ok() && file_meta != nullptr) { *file_meta = meta; } RecordFlushIOStats(); auto stream = event_logger_->LogToBuffer(log_buffer_); stream << "job" << job_context_->job_id << "event" << "flush_finished"; stream << "lsm_state"; stream.StartArray(); auto vstorage = cfd_->current()->storage_info(); for (int level = 0; level < vstorage->num_levels(); ++level) { stream << vstorage->NumLevelFiles(level); } stream.EndArray(); return s; } Status FlushJob::WriteLevel0Table(const autovector& mems, VersionEdit* edit, FileMetaData* meta) { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_FLUSH_WRITE_L0); db_mutex_->AssertHeld(); const uint64_t start_micros = db_options_.env->NowMicros(); // path 0 for level 0 file. meta->fd = FileDescriptor(versions_->NewFileNumber(), 0, 0); Version* base = cfd_->current(); base->Ref(); // it is likely that we do not need this reference Status s; { db_mutex_->Unlock(); if (log_buffer_) { log_buffer_->FlushBufferToLog(); } std::vector memtables; ReadOptions ro; ro.total_order_seek = true; Arena arena; uint64_t total_num_entries = 0, total_num_deletes = 0; size_t total_memory_usage = 0; for (MemTable* m : mems) { Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] [JOB %d] Flushing memtable with next log file: %" PRIu64 "\n", cfd_->GetName().c_str(), job_context_->job_id, m->GetNextLogNumber()); memtables.push_back(m->NewIterator(ro, &arena)); total_num_entries += m->num_entries(); total_num_deletes += m->num_deletes(); total_memory_usage += m->ApproximateMemoryUsage(); } event_logger_->Log() << "job" << job_context_->job_id << "event" << "flush_started" << "num_memtables" << mems.size() << "num_entries" << total_num_entries << "num_deletes" << total_num_deletes << "memory_usage" << total_memory_usage; TableFileCreationInfo info; { ScopedArenaIterator iter( NewMergingIterator(&cfd_->internal_comparator(), &memtables[0], static_cast(memtables.size()), &arena)); Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] [JOB %d] Level-0 flush table #%" PRIu64 ": started", cfd_->GetName().c_str(), job_context_->job_id, meta->fd.GetNumber()); TEST_SYNC_POINT_CALLBACK("FlushJob::WriteLevel0Table:output_compression", &output_compression_); s = BuildTable(dbname_, db_options_.env, *cfd_->ioptions(), env_options_, cfd_->table_cache(), iter.get(), meta, cfd_->internal_comparator(), cfd_->int_tbl_prop_collector_factories(), cfd_->GetID(), existing_snapshots_, earliest_write_conflict_snapshot_, output_compression_, cfd_->ioptions()->compression_opts, mutable_cf_options_.paranoid_file_checks, cfd_->internal_stats(), Env::IO_HIGH, &table_properties_); info.table_properties = table_properties_; LogFlush(db_options_.info_log); } Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "[%s] [JOB %d] Level-0 flush table #%" PRIu64 ": %" PRIu64 " bytes %s" "%s", cfd_->GetName().c_str(), job_context_->job_id, meta->fd.GetNumber(), meta->fd.GetFileSize(), s.ToString().c_str(), meta->marked_for_compaction ? " (needs compaction)" : ""); // output to event logger if (s.ok()) { info.db_name = dbname_; info.cf_name = cfd_->GetName(); info.file_path = TableFileName(db_options_.db_paths, meta->fd.GetNumber(), meta->fd.GetPathId()); info.file_size = meta->fd.GetFileSize(); info.job_id = job_context_->job_id; EventHelpers::LogAndNotifyTableFileCreation( event_logger_, db_options_.listeners, meta->fd, info); TEST_SYNC_POINT("FlushJob::LogAndNotifyTableFileCreation()"); } if (!db_options_.disableDataSync && output_file_directory_ != nullptr) { output_file_directory_->Fsync(); } TEST_SYNC_POINT("FlushJob::WriteLevel0Table"); db_mutex_->Lock(); } base->Unref(); // re-acquire the most current version base = cfd_->current(); // Note that if file_size is zero, the file has been deleted and // should not be added to the manifest. if (s.ok() && meta->fd.GetFileSize() > 0) { // if we have more than 1 background thread, then we cannot // insert files directly into higher levels because some other // threads could be concurrently producing compacted files for // that key range. // Add file to L0 edit->AddFile(0 /* level */, meta->fd.GetNumber(), meta->fd.GetPathId(), meta->fd.GetFileSize(), meta->smallest, meta->largest, meta->smallest_seqno, meta->largest_seqno, meta->marked_for_compaction); } InternalStats::CompactionStats stats(1); stats.micros = db_options_.env->NowMicros() - start_micros; stats.bytes_written = meta->fd.GetFileSize(); cfd_->internal_stats()->AddCompactionStats(0 /* level */, stats); cfd_->internal_stats()->AddCFStats(InternalStats::BYTES_FLUSHED, meta->fd.GetFileSize()); RecordTick(stats_, COMPACT_WRITE_BYTES, meta->fd.GetFileSize()); return s; } } // namespace rocksdb rocksdb-4.5.1/db/flush_job.h000066400000000000000000000063361267530634000156770ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include #include #include "db/dbformat.h" #include "db/column_family.h" #include "db/log_writer.h" #include "db/memtable_list.h" #include "db/snapshot_impl.h" #include "db/version_edit.h" #include "port/port.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "rocksdb/transaction_log.h" #include "table/scoped_arena_iterator.h" #include "util/autovector.h" #include "util/event_logger.h" #include "util/instrumented_mutex.h" #include "util/stop_watch.h" #include "util/thread_local.h" #include "db/internal_stats.h" #include "db/write_controller.h" #include "db/flush_scheduler.h" #include "db/write_thread.h" #include "db/job_context.h" namespace rocksdb { class MemTable; class TableCache; class Version; class VersionEdit; class VersionSet; class Arena; class FlushJob { public: // TODO(icanadi) make effort to reduce number of parameters here // IMPORTANT: mutable_cf_options needs to be alive while FlushJob is alive FlushJob(const std::string& dbname, ColumnFamilyData* cfd, const DBOptions& db_options, const MutableCFOptions& mutable_cf_options, const EnvOptions& env_options, VersionSet* versions, InstrumentedMutex* db_mutex, std::atomic* shutting_down, std::vector existing_snapshots, SequenceNumber earliest_write_conflict_snapshot, JobContext* job_context, LogBuffer* log_buffer, Directory* db_directory, Directory* output_file_directory, CompressionType output_compression, Statistics* stats, EventLogger* event_logger); ~FlushJob(); Status Run(FileMetaData* file_meta = nullptr); TableProperties GetTableProperties() const { return table_properties_; } private: void ReportStartedFlush(); void ReportFlushInputSize(const autovector& mems); void RecordFlushIOStats(); Status WriteLevel0Table(const autovector& mems, VersionEdit* edit, FileMetaData* meta); const std::string& dbname_; ColumnFamilyData* cfd_; const DBOptions& db_options_; const MutableCFOptions& mutable_cf_options_; const EnvOptions& env_options_; VersionSet* versions_; InstrumentedMutex* db_mutex_; std::atomic* shutting_down_; std::vector existing_snapshots_; SequenceNumber earliest_write_conflict_snapshot_; JobContext* job_context_; LogBuffer* log_buffer_; Directory* db_directory_; Directory* output_file_directory_; CompressionType output_compression_; Statistics* stats_; EventLogger* event_logger_; TableProperties table_properties_; }; } // namespace rocksdb rocksdb-4.5.1/db/flush_job_test.cc000066400000000000000000000170751267530634000170760ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include "db/flush_job.h" #include "db/column_family.h" #include "db/version_set.h" #include "db/writebuffer.h" #include "rocksdb/cache.h" #include "util/file_reader_writer.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" #include "table/mock_table.h" namespace rocksdb { // TODO(icanadi) Mock out everything else: // 1. VersionSet // 2. Memtable class FlushJobTest : public testing::Test { public: FlushJobTest() : env_(Env::Default()), dbname_(test::TmpDir() + "/flush_job_test"), table_cache_(NewLRUCache(50000, 16)), write_buffer_(db_options_.db_write_buffer_size), versions_(new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(), &write_buffer_, &write_controller_)), shutting_down_(false), mock_table_factory_(new mock::MockTableFactory()) { EXPECT_OK(env_->CreateDirIfMissing(dbname_)); db_options_.db_paths.emplace_back(dbname_, std::numeric_limits::max()); // TODO(icanadi) Remove this once we mock out VersionSet NewDB(); std::vector column_families; cf_options_.table_factory = mock_table_factory_; column_families.emplace_back(kDefaultColumnFamilyName, cf_options_); EXPECT_OK(versions_->Recover(column_families, false)); } void NewDB() { VersionEdit new_db; new_db.SetLogNumber(0); new_db.SetNextFile(2); new_db.SetLastSequence(0); const std::string manifest = DescriptorFileName(dbname_, 1); unique_ptr file; Status s = env_->NewWritableFile( manifest, &file, env_->OptimizeForManifestWrite(env_options_)); ASSERT_OK(s); unique_ptr file_writer( new WritableFileWriter(std::move(file), EnvOptions())); { log::Writer log(std::move(file_writer), 0, false); std::string record; new_db.EncodeTo(&record); s = log.AddRecord(record); } ASSERT_OK(s); // Make "CURRENT" file that points to the new manifest file. s = SetCurrentFile(env_, dbname_, 1, nullptr); } Env* env_; std::string dbname_; EnvOptions env_options_; std::shared_ptr table_cache_; WriteController write_controller_; DBOptions db_options_; WriteBuffer write_buffer_; ColumnFamilyOptions cf_options_; std::unique_ptr versions_; InstrumentedMutex mutex_; std::atomic shutting_down_; std::shared_ptr mock_table_factory_; }; TEST_F(FlushJobTest, Empty) { JobContext job_context(0); auto cfd = versions_->GetColumnFamilySet()->GetDefault(); EventLogger event_logger(db_options_.info_log.get()); FlushJob flush_job(dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_, *cfd->GetLatestMutableCFOptions(), env_options_, versions_.get(), &mutex_, &shutting_down_, {}, kMaxSequenceNumber, &job_context, nullptr, nullptr, nullptr, kNoCompression, nullptr, &event_logger); ASSERT_OK(flush_job.Run()); job_context.Clean(); } TEST_F(FlushJobTest, NonEmpty) { JobContext job_context(0); auto cfd = versions_->GetColumnFamilySet()->GetDefault(); auto new_mem = cfd->ConstructNewMemtable(*cfd->GetLatestMutableCFOptions(), kMaxSequenceNumber); new_mem->Ref(); auto inserted_keys = mock::MakeMockFile(); // Test data: // seqno [ 1, 2 ... 8998, 8999, 9000, 9001, 9002 ... 9999 ] // key [ 1001, 1002 ... 9998, 9999, 0, 1, 2 ... 999 ] // Expected: // smallest_key = "0" // largest_key = "9999" // smallest_seqno = 1 // smallest_seqno = 9999 for (int i = 1; i < 10000; ++i) { std::string key(ToString((i + 1000) % 10000)); std::string value("value" + key); new_mem->Add(SequenceNumber(i), kTypeValue, key, value); InternalKey internal_key(key, SequenceNumber(i), kTypeValue); inserted_keys.insert({internal_key.Encode().ToString(), value}); } autovector to_delete; cfd->imm()->Add(new_mem, &to_delete); for (auto& m : to_delete) { delete m; } EventLogger event_logger(db_options_.info_log.get()); FlushJob flush_job(dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_, *cfd->GetLatestMutableCFOptions(), env_options_, versions_.get(), &mutex_, &shutting_down_, {}, kMaxSequenceNumber, &job_context, nullptr, nullptr, nullptr, kNoCompression, nullptr, &event_logger); FileMetaData fd; mutex_.Lock(); ASSERT_OK(flush_job.Run(&fd)); mutex_.Unlock(); ASSERT_EQ(ToString(0), fd.smallest.user_key().ToString()); ASSERT_EQ(ToString(9999), fd.largest.user_key().ToString()); ASSERT_EQ(1, fd.smallest_seqno); ASSERT_EQ(9999, fd.largest_seqno); mock_table_factory_->AssertSingleFile(inserted_keys); job_context.Clean(); } TEST_F(FlushJobTest, Snapshots) { JobContext job_context(0); auto cfd = versions_->GetColumnFamilySet()->GetDefault(); auto new_mem = cfd->ConstructNewMemtable(*cfd->GetLatestMutableCFOptions(), kMaxSequenceNumber); std::vector snapshots; std::set snapshots_set; int keys = 10000; int max_inserts_per_keys = 8; Random rnd(301); for (int i = 0; i < keys / 2; ++i) { snapshots.push_back(rnd.Uniform(keys * (max_inserts_per_keys / 2)) + 1); snapshots_set.insert(snapshots.back()); } std::sort(snapshots.begin(), snapshots.end()); new_mem->Ref(); SequenceNumber current_seqno = 0; auto inserted_keys = mock::MakeMockFile(); for (int i = 1; i < keys; ++i) { std::string key(ToString(i)); int insertions = rnd.Uniform(max_inserts_per_keys); for (int j = 0; j < insertions; ++j) { std::string value(test::RandomHumanReadableString(&rnd, 10)); auto seqno = ++current_seqno; new_mem->Add(SequenceNumber(seqno), kTypeValue, key, value); // a key is visible only if: // 1. it's the last one written (j == insertions - 1) // 2. there's a snapshot pointing at it bool visible = (j == insertions - 1) || (snapshots_set.find(seqno) != snapshots_set.end()); if (visible) { InternalKey internal_key(key, seqno, kTypeValue); inserted_keys.insert({internal_key.Encode().ToString(), value}); } } } autovector to_delete; cfd->imm()->Add(new_mem, &to_delete); for (auto& m : to_delete) { delete m; } EventLogger event_logger(db_options_.info_log.get()); FlushJob flush_job(dbname_, versions_->GetColumnFamilySet()->GetDefault(), db_options_, *cfd->GetLatestMutableCFOptions(), env_options_, versions_.get(), &mutex_, &shutting_down_, snapshots, kMaxSequenceNumber, &job_context, nullptr, nullptr, nullptr, kNoCompression, nullptr, &event_logger); mutex_.Lock(); ASSERT_OK(flush_job.Run()); mutex_.Unlock(); mock_table_factory_->AssertSingleFile(inserted_keys); job_context.Clean(); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/flush_scheduler.cc000066400000000000000000000040431267530634000172320ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/flush_scheduler.h" #include #include "db/column_family.h" namespace rocksdb { void FlushScheduler::ScheduleFlush(ColumnFamilyData* cfd) { #ifndef NDEBUG { std::lock_guard lock(checking_mutex_); assert(checking_set_.count(cfd) == 0); checking_set_.insert(cfd); } #endif // NDEBUG cfd->Ref(); Node* node = new Node{cfd, head_.load(std::memory_order_relaxed)}; while (!head_.compare_exchange_strong( node->next, node, std::memory_order_relaxed, std::memory_order_relaxed)) { // failing CAS updates the first param, so we are already set for // retry. TakeNextColumnFamily won't happen until after another // inter-thread synchronization, so we don't even need release // semantics for this CAS } } ColumnFamilyData* FlushScheduler::TakeNextColumnFamily() { while (true) { if (Empty()) { return nullptr; } // dequeue the head Node* node = head_.load(std::memory_order_relaxed); head_.store(node->next, std::memory_order_relaxed); ColumnFamilyData* cfd = node->column_family; delete node; #ifndef NDEBUG { auto iter = checking_set_.find(cfd); assert(iter != checking_set_.end()); checking_set_.erase(iter); } #endif // NDEBUG if (!cfd->IsDropped()) { // success return cfd; } // no longer relevant, retry if (cfd->Unref()) { delete cfd; } } } bool FlushScheduler::Empty() { auto rv = head_.load(std::memory_order_relaxed) == nullptr; assert(rv == checking_set_.empty()); return rv; } void FlushScheduler::Clear() { ColumnFamilyData* cfd; while ((cfd = TakeNextColumnFamily()) != nullptr) { if (cfd->Unref()) { delete cfd; } } assert(Empty()); } } // namespace rocksdb rocksdb-4.5.1/db/flush_scheduler.h000066400000000000000000000023611267530634000170750ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include namespace rocksdb { class ColumnFamilyData; // Unless otherwise noted, all methods on FlushScheduler should be called // only with the DB mutex held or from a single-threaded recovery context. class FlushScheduler { public: FlushScheduler() : head_(nullptr) {} // May be called from multiple threads at once, but not concurrent with // any other method calls on this instance void ScheduleFlush(ColumnFamilyData* cfd); // Removes and returns Ref()-ed column family. Client needs to Unref(). // Filters column families that have been dropped. ColumnFamilyData* TakeNextColumnFamily(); bool Empty(); void Clear(); private: struct Node { ColumnFamilyData* column_family; Node* next; }; std::atomic head_; #ifndef NDEBUG std::mutex checking_mutex_; std::set checking_set_; #endif // NDEBUG }; } // namespace rocksdb rocksdb-4.5.1/db/forward_iterator.cc000066400000000000000000000564661267530634000174500ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "db/forward_iterator.h" #include #include #include #include "db/job_context.h" #include "db/db_impl.h" #include "db/db_iter.h" #include "db/column_family.h" #include "rocksdb/env.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "table/merger.h" #include "db/dbformat.h" #include "util/sync_point.h" namespace rocksdb { // Usage: // LevelIterator iter; // iter.SetFileIndex(file_index); // iter.Seek(target); // iter.Next() class LevelIterator : public InternalIterator { public: LevelIterator(const ColumnFamilyData* const cfd, const ReadOptions& read_options, const std::vector& files) : cfd_(cfd), read_options_(read_options), files_(files), valid_(false), file_index_(std::numeric_limits::max()) {} void SetFileIndex(uint32_t file_index) { assert(file_index < files_.size()); if (file_index != file_index_) { file_index_ = file_index; Reset(); } valid_ = false; } void Reset() { assert(file_index_ < files_.size()); file_iter_.reset(cfd_->table_cache()->NewIterator( read_options_, *(cfd_->soptions()), cfd_->internal_comparator(), files_[file_index_]->fd, nullptr /* table_reader_ptr */, nullptr, false)); } void SeekToLast() override { status_ = Status::NotSupported("LevelIterator::SeekToLast()"); valid_ = false; } void Prev() override { status_ = Status::NotSupported("LevelIterator::Prev()"); valid_ = false; } bool Valid() const override { return valid_; } void SeekToFirst() override { SetFileIndex(0); file_iter_->SeekToFirst(); valid_ = file_iter_->Valid(); } void Seek(const Slice& internal_key) override { assert(file_iter_ != nullptr); file_iter_->Seek(internal_key); valid_ = file_iter_->Valid(); } void Next() override { assert(valid_); file_iter_->Next(); for (;;) { if (file_iter_->status().IsIncomplete() || file_iter_->Valid()) { valid_ = !file_iter_->status().IsIncomplete(); return; } if (file_index_ + 1 >= files_.size()) { valid_ = false; return; } SetFileIndex(file_index_ + 1); file_iter_->SeekToFirst(); } } Slice key() const override { assert(valid_); return file_iter_->key(); } Slice value() const override { assert(valid_); return file_iter_->value(); } Status status() const override { if (!status_.ok()) { return status_; } else if (file_iter_ && !file_iter_->status().ok()) { return file_iter_->status(); } return Status::OK(); } private: const ColumnFamilyData* const cfd_; const ReadOptions& read_options_; const std::vector& files_; bool valid_; uint32_t file_index_; Status status_; std::unique_ptr file_iter_; }; ForwardIterator::ForwardIterator(DBImpl* db, const ReadOptions& read_options, ColumnFamilyData* cfd, SuperVersion* current_sv) : db_(db), read_options_(read_options), cfd_(cfd), prefix_extractor_(cfd->ioptions()->prefix_extractor), user_comparator_(cfd->user_comparator()), immutable_min_heap_(MinIterComparator(&cfd_->internal_comparator())), sv_(current_sv), mutable_iter_(nullptr), current_(nullptr), valid_(false), status_(Status::OK()), immutable_status_(Status::OK()), has_iter_trimmed_for_upper_bound_(false), current_over_upper_bound_(false), is_prev_set_(false), is_prev_inclusive_(false) { if (sv_) { RebuildIterators(false); } } ForwardIterator::~ForwardIterator() { Cleanup(true); } void ForwardIterator::SVCleanup() { if (sv_ != nullptr && sv_->Unref()) { // Job id == 0 means that this is not our background process, but rather // user thread JobContext job_context(0); db_->mutex_.Lock(); sv_->Cleanup(); db_->FindObsoleteFiles(&job_context, false, true); db_->mutex_.Unlock(); delete sv_; if (job_context.HaveSomethingToDelete()) { db_->PurgeObsoleteFiles(job_context); } job_context.Clean(); } } void ForwardIterator::Cleanup(bool release_sv) { if (mutable_iter_ != nullptr) { mutable_iter_->~InternalIterator(); } for (auto* m : imm_iters_) { m->~InternalIterator(); } imm_iters_.clear(); for (auto* f : l0_iters_) { delete f; } l0_iters_.clear(); for (auto* l : level_iters_) { delete l; } level_iters_.clear(); if (release_sv) { SVCleanup(); } } bool ForwardIterator::Valid() const { // See UpdateCurrent(). return valid_ ? !current_over_upper_bound_ : false; } void ForwardIterator::SeekToFirst() { if (sv_ == nullptr) { RebuildIterators(true); } else if (sv_->version_number != cfd_->GetSuperVersionNumber()) { RenewIterators(); } else if (immutable_status_.IsIncomplete()) { ResetIncompleteIterators(); } SeekInternal(Slice(), true); } bool ForwardIterator::IsOverUpperBound(const Slice& internal_key) const { return !(read_options_.iterate_upper_bound == nullptr || cfd_->internal_comparator().user_comparator()->Compare( ExtractUserKey(internal_key), *read_options_.iterate_upper_bound) < 0); } void ForwardIterator::Seek(const Slice& internal_key) { if (IsOverUpperBound(internal_key)) { valid_ = false; } if (sv_ == nullptr) { RebuildIterators(true); } else if (sv_->version_number != cfd_->GetSuperVersionNumber()) { RenewIterators(); } else if (immutable_status_.IsIncomplete()) { ResetIncompleteIterators(); } SeekInternal(internal_key, false); } void ForwardIterator::SeekInternal(const Slice& internal_key, bool seek_to_first) { assert(mutable_iter_); // mutable seek_to_first ? mutable_iter_->SeekToFirst() : mutable_iter_->Seek(internal_key); // immutable // TODO(ljin): NeedToSeekImmutable has negative impact on performance // if it turns to need to seek immutable often. We probably want to have // an option to turn it off. if (seek_to_first || NeedToSeekImmutable(internal_key)) { immutable_status_ = Status::OK(); if ((has_iter_trimmed_for_upper_bound_) && (cfd_->internal_comparator().InternalKeyComparator::Compare( prev_key_.GetKey(), internal_key) > 0)) { // Some iterators are trimmed. Need to rebuild. RebuildIterators(true); // Already seeked mutable iter, so seek again seek_to_first ? mutable_iter_->SeekToFirst() : mutable_iter_->Seek(internal_key); } { auto tmp = MinIterHeap(MinIterComparator(&cfd_->internal_comparator())); immutable_min_heap_.swap(tmp); } for (size_t i = 0; i < imm_iters_.size(); i++) { auto* m = imm_iters_[i]; seek_to_first ? m->SeekToFirst() : m->Seek(internal_key); if (!m->status().ok()) { immutable_status_ = m->status(); } else if (m->Valid()) { immutable_min_heap_.push(m); } } Slice user_key; if (!seek_to_first) { user_key = ExtractUserKey(internal_key); } const VersionStorageInfo* vstorage = sv_->current->storage_info(); const std::vector& l0 = vstorage->LevelFiles(0); for (size_t i = 0; i < l0.size(); ++i) { if (!l0_iters_[i]) { continue; } if (seek_to_first) { l0_iters_[i]->SeekToFirst(); } else { // If the target key passes over the larget key, we are sure Next() // won't go over this file. if (user_comparator_->Compare(user_key, l0[i]->largest.user_key()) > 0) { if (read_options_.iterate_upper_bound != nullptr) { has_iter_trimmed_for_upper_bound_ = true; delete l0_iters_[i]; l0_iters_[i] = nullptr; } continue; } l0_iters_[i]->Seek(internal_key); } if (!l0_iters_[i]->status().ok()) { immutable_status_ = l0_iters_[i]->status(); } else if (l0_iters_[i]->Valid()) { if (!IsOverUpperBound(l0_iters_[i]->key())) { immutable_min_heap_.push(l0_iters_[i]); } else { has_iter_trimmed_for_upper_bound_ = true; delete l0_iters_[i]; l0_iters_[i] = nullptr; } } } int32_t search_left_bound = 0; int32_t search_right_bound = FileIndexer::kLevelMaxIndex; for (int32_t level = 1; level < vstorage->num_levels(); ++level) { const std::vector& level_files = vstorage->LevelFiles(level); if (level_files.empty()) { search_left_bound = 0; search_right_bound = FileIndexer::kLevelMaxIndex; continue; } if (level_iters_[level - 1] == nullptr) { continue; } uint32_t f_idx = 0; const auto& indexer = vstorage->file_indexer(); if (!seek_to_first) { if (search_left_bound == search_right_bound) { f_idx = search_left_bound; } else if (search_left_bound < search_right_bound) { f_idx = FindFileInRange(level_files, internal_key, search_left_bound, search_right_bound == FileIndexer::kLevelMaxIndex ? static_cast(level_files.size()) : search_right_bound); } else { // search_left_bound > search_right_bound // There are only 2 cases this can happen: // (1) target key is smaller than left most file // (2) target key is larger than right most file assert(search_left_bound == (int32_t)level_files.size() || search_right_bound == -1); if (search_right_bound == -1) { assert(search_left_bound == 0); f_idx = 0; } else { indexer.GetNextLevelIndex( level, level_files.size() - 1, 1, 1, &search_left_bound, &search_right_bound); continue; } } // Prepare hints for the next level if (f_idx < level_files.size()) { int cmp_smallest = user_comparator_->Compare( user_key, level_files[f_idx]->smallest.user_key()); assert(user_comparator_->Compare( user_key, level_files[f_idx]->largest.user_key()) <= 0); indexer.GetNextLevelIndex(level, f_idx, cmp_smallest, -1, &search_left_bound, &search_right_bound); } else { indexer.GetNextLevelIndex( level, level_files.size() - 1, 1, 1, &search_left_bound, &search_right_bound); } } // Seek if (f_idx < level_files.size()) { level_iters_[level - 1]->SetFileIndex(f_idx); seek_to_first ? level_iters_[level - 1]->SeekToFirst() : level_iters_[level - 1]->Seek(internal_key); if (!level_iters_[level - 1]->status().ok()) { immutable_status_ = level_iters_[level - 1]->status(); } else if (level_iters_[level - 1]->Valid()) { if (!IsOverUpperBound(level_iters_[level - 1]->key())) { immutable_min_heap_.push(level_iters_[level - 1]); } else { // Nothing in this level is interesting. Remove. has_iter_trimmed_for_upper_bound_ = true; delete level_iters_[level - 1]; level_iters_[level - 1] = nullptr; } } } } if (seek_to_first) { is_prev_set_ = false; } else { prev_key_.SetKey(internal_key); is_prev_set_ = true; is_prev_inclusive_ = true; } TEST_SYNC_POINT_CALLBACK("ForwardIterator::SeekInternal:Immutable", this); } else if (current_ && current_ != mutable_iter_) { // current_ is one of immutable iterators, push it back to the heap immutable_min_heap_.push(current_); } UpdateCurrent(); TEST_SYNC_POINT_CALLBACK("ForwardIterator::SeekInternal:Return", this); } void ForwardIterator::Next() { assert(valid_); bool update_prev_key = false; if (sv_ == nullptr || sv_->version_number != cfd_->GetSuperVersionNumber()) { std::string current_key = key().ToString(); Slice old_key(current_key.data(), current_key.size()); if (sv_ == nullptr) { RebuildIterators(true); } else { RenewIterators(); } SeekInternal(old_key, false); if (!valid_ || key().compare(old_key) != 0) { return; } } else if (current_ != mutable_iter_) { // It is going to advance immutable iterator if (is_prev_set_ && prefix_extractor_) { // advance prev_key_ to current_ only if they share the same prefix update_prev_key = prefix_extractor_->Transform(prev_key_.GetKey()).compare( prefix_extractor_->Transform(current_->key())) == 0; } else { update_prev_key = true; } if (update_prev_key) { prev_key_.SetKey(current_->key()); is_prev_set_ = true; is_prev_inclusive_ = false; } } current_->Next(); if (current_ != mutable_iter_) { if (!current_->status().ok()) { immutable_status_ = current_->status(); } else if ((current_->Valid()) && (!IsOverUpperBound(current_->key()))) { immutable_min_heap_.push(current_); } else { if ((current_->Valid()) && (IsOverUpperBound(current_->key()))) { // remove the current iterator DeleteCurrentIter(); current_ = nullptr; } if (update_prev_key) { mutable_iter_->Seek(prev_key_.GetKey()); } } } UpdateCurrent(); TEST_SYNC_POINT_CALLBACK("ForwardIterator::Next:Return", this); } Slice ForwardIterator::key() const { assert(valid_); return current_->key(); } Slice ForwardIterator::value() const { assert(valid_); return current_->value(); } Status ForwardIterator::status() const { if (!status_.ok()) { return status_; } else if (!mutable_iter_->status().ok()) { return mutable_iter_->status(); } return immutable_status_; } void ForwardIterator::RebuildIterators(bool refresh_sv) { // Clean up Cleanup(refresh_sv); if (refresh_sv) { // New sv_ = cfd_->GetReferencedSuperVersion(&(db_->mutex_)); } mutable_iter_ = sv_->mem->NewIterator(read_options_, &arena_); sv_->imm->AddIterators(read_options_, &imm_iters_, &arena_); has_iter_trimmed_for_upper_bound_ = false; const auto* vstorage = sv_->current->storage_info(); const auto& l0_files = vstorage->LevelFiles(0); l0_iters_.reserve(l0_files.size()); for (const auto* l0 : l0_files) { if ((read_options_.iterate_upper_bound != nullptr) && cfd_->internal_comparator().user_comparator()->Compare( l0->smallest.user_key(), *read_options_.iterate_upper_bound) > 0) { has_iter_trimmed_for_upper_bound_ = true; l0_iters_.push_back(nullptr); continue; } l0_iters_.push_back(cfd_->table_cache()->NewIterator( read_options_, *cfd_->soptions(), cfd_->internal_comparator(), l0->fd)); } BuildLevelIterators(vstorage); current_ = nullptr; is_prev_set_ = false; } void ForwardIterator::RenewIterators() { SuperVersion* svnew; assert(sv_); svnew = cfd_->GetReferencedSuperVersion(&(db_->mutex_)); if (mutable_iter_ != nullptr) { mutable_iter_->~InternalIterator(); } for (auto* m : imm_iters_) { m->~InternalIterator(); } imm_iters_.clear(); mutable_iter_ = svnew->mem->NewIterator(read_options_, &arena_); svnew->imm->AddIterators(read_options_, &imm_iters_, &arena_); const auto* vstorage = sv_->current->storage_info(); const auto& l0_files = vstorage->LevelFiles(0); const auto* vstorage_new = svnew->current->storage_info(); const auto& l0_files_new = vstorage_new->LevelFiles(0); size_t iold, inew; bool found; std::vector l0_iters_new; l0_iters_new.reserve(l0_files_new.size()); for (inew = 0; inew < l0_files_new.size(); inew++) { found = false; for (iold = 0; iold < l0_files.size(); iold++) { if (l0_files[iold] == l0_files_new[inew]) { found = true; break; } } if (found) { if (l0_iters_[iold] == nullptr) { l0_iters_new.push_back(nullptr); TEST_SYNC_POINT_CALLBACK("ForwardIterator::RenewIterators:Null", this); } else { l0_iters_new.push_back(l0_iters_[iold]); l0_iters_[iold] = nullptr; TEST_SYNC_POINT_CALLBACK("ForwardIterator::RenewIterators:Copy", this); } continue; } l0_iters_new.push_back(cfd_->table_cache()->NewIterator( read_options_, *cfd_->soptions(), cfd_->internal_comparator(), l0_files_new[inew]->fd)); } for (auto* f : l0_iters_) { delete f; } l0_iters_.clear(); l0_iters_ = l0_iters_new; for (auto* l : level_iters_) { delete l; } level_iters_.clear(); BuildLevelIterators(vstorage_new); current_ = nullptr; is_prev_set_ = false; SVCleanup(); sv_ = svnew; } void ForwardIterator::BuildLevelIterators(const VersionStorageInfo* vstorage) { level_iters_.reserve(vstorage->num_levels() - 1); for (int32_t level = 1; level < vstorage->num_levels(); ++level) { const auto& level_files = vstorage->LevelFiles(level); if ((level_files.empty()) || ((read_options_.iterate_upper_bound != nullptr) && (user_comparator_->Compare(*read_options_.iterate_upper_bound, level_files[0]->smallest.user_key()) < 0))) { level_iters_.push_back(nullptr); if (!level_files.empty()) { has_iter_trimmed_for_upper_bound_ = true; } } else { level_iters_.push_back( new LevelIterator(cfd_, read_options_, level_files)); } } } void ForwardIterator::ResetIncompleteIterators() { const auto& l0_files = sv_->current->storage_info()->LevelFiles(0); for (size_t i = 0; i < l0_iters_.size(); ++i) { assert(i < l0_files.size()); if (!l0_iters_[i] || !l0_iters_[i]->status().IsIncomplete()) { continue; } delete l0_iters_[i]; l0_iters_[i] = cfd_->table_cache()->NewIterator( read_options_, *cfd_->soptions(), cfd_->internal_comparator(), l0_files[i]->fd); } for (auto* level_iter : level_iters_) { if (level_iter && level_iter->status().IsIncomplete()) { level_iter->Reset(); } } current_ = nullptr; is_prev_set_ = false; } void ForwardIterator::UpdateCurrent() { if (immutable_min_heap_.empty() && !mutable_iter_->Valid()) { current_ = nullptr; } else if (immutable_min_heap_.empty()) { current_ = mutable_iter_; } else if (!mutable_iter_->Valid()) { current_ = immutable_min_heap_.top(); immutable_min_heap_.pop(); } else { current_ = immutable_min_heap_.top(); assert(current_ != nullptr); assert(current_->Valid()); int cmp = cfd_->internal_comparator().InternalKeyComparator::Compare( mutable_iter_->key(), current_->key()); assert(cmp != 0); if (cmp > 0) { immutable_min_heap_.pop(); } else { current_ = mutable_iter_; } } valid_ = (current_ != nullptr); if (!status_.ok()) { status_ = Status::OK(); } // Upper bound doesn't apply to the memtable iterator. We want Valid() to // return false when all iterators are over iterate_upper_bound, but can't // just set valid_ to false, as that would effectively disable the tailing // optimization (Seek() would be called on all immutable iterators regardless // of whether the target key is greater than prev_key_). current_over_upper_bound_ = valid_ && IsOverUpperBound(current_->key()); } bool ForwardIterator::NeedToSeekImmutable(const Slice& target) { // We maintain the interval (prev_key_, immutable_min_heap_.top()->key()) // such that there are no records with keys within that range in // immutable_min_heap_. Since immutable structures (SST files and immutable // memtables) can't change in this version, we don't need to do a seek if // 'target' belongs to that interval (immutable_min_heap_.top() is already // at the correct position). if (!valid_ || !current_ || !is_prev_set_ || !immutable_status_.ok()) { return true; } Slice prev_key = prev_key_.GetKey(); if (prefix_extractor_ && prefix_extractor_->Transform(target).compare( prefix_extractor_->Transform(prev_key)) != 0) { return true; } if (cfd_->internal_comparator().InternalKeyComparator::Compare( prev_key, target) >= (is_prev_inclusive_ ? 1 : 0)) { return true; } if (immutable_min_heap_.empty() && current_ == mutable_iter_) { // Nothing to seek on. return false; } if (cfd_->internal_comparator().InternalKeyComparator::Compare( target, current_ == mutable_iter_ ? immutable_min_heap_.top()->key() : current_->key()) > 0) { return true; } return false; } void ForwardIterator::DeleteCurrentIter() { const VersionStorageInfo* vstorage = sv_->current->storage_info(); const std::vector& l0 = vstorage->LevelFiles(0); for (size_t i = 0; i < l0.size(); ++i) { if (!l0_iters_[i]) { continue; } if (l0_iters_[i] == current_) { has_iter_trimmed_for_upper_bound_ = true; delete l0_iters_[i]; l0_iters_[i] = nullptr; return; } } for (int32_t level = 1; level < vstorage->num_levels(); ++level) { if (level_iters_[level - 1] == nullptr) { continue; } if (level_iters_[level - 1] == current_) { has_iter_trimmed_for_upper_bound_ = true; delete level_iters_[level - 1]; level_iters_[level - 1] = nullptr; } } } bool ForwardIterator::TEST_CheckDeletedIters(int* pdeleted_iters, int* pnum_iters) { bool retval = false; int deleted_iters = 0; int num_iters = 0; const VersionStorageInfo* vstorage = sv_->current->storage_info(); const std::vector& l0 = vstorage->LevelFiles(0); for (size_t i = 0; i < l0.size(); ++i) { if (!l0_iters_[i]) { retval = true; deleted_iters++; } else { num_iters++; } } for (int32_t level = 1; level < vstorage->num_levels(); ++level) { if ((level_iters_[level - 1] == nullptr) && (!vstorage->LevelFiles(level).empty())) { retval = true; deleted_iters++; } else if (!vstorage->LevelFiles(level).empty()) { num_iters++; } } if ((!retval) && num_iters <= 1) { retval = true; } if (pdeleted_iters) { *pdeleted_iters = deleted_iters; } if (pnum_iters) { *pnum_iters = num_iters; } return retval; } uint32_t ForwardIterator::FindFileInRange( const std::vector& files, const Slice& internal_key, uint32_t left, uint32_t right) { while (left < right) { uint32_t mid = (left + right) / 2; const FileMetaData* f = files[mid]; if (cfd_->internal_comparator().InternalKeyComparator::Compare( f->largest.Encode(), internal_key) < 0) { // Key at "mid.largest" is < "target". Therefore all // files at or before "mid" are uninteresting. left = mid + 1; } else { // Key at "mid.largest" is >= "target". Therefore all files // after "mid" are uninteresting. right = mid; } } return right; } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/forward_iterator.h000066400000000000000000000105341267530634000172740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include "rocksdb/db.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "db/dbformat.h" #include "table/internal_iterator.h" #include "util/arena.h" namespace rocksdb { class DBImpl; class Env; struct SuperVersion; class ColumnFamilyData; class LevelIterator; class VersionStorageInfo; struct FileMetaData; class MinIterComparator { public: explicit MinIterComparator(const Comparator* comparator) : comparator_(comparator) {} bool operator()(InternalIterator* a, InternalIterator* b) { return comparator_->Compare(a->key(), b->key()) > 0; } private: const Comparator* comparator_; }; typedef std::priority_queue, MinIterComparator> MinIterHeap; /** * ForwardIterator is a special type of iterator that only supports Seek() * and Next(). It is expected to perform better than TailingIterator by * removing the encapsulation and making all information accessible within * the iterator. At the current implementation, snapshot is taken at the * time Seek() is called. The Next() followed do not see new values after. */ class ForwardIterator : public InternalIterator { public: ForwardIterator(DBImpl* db, const ReadOptions& read_options, ColumnFamilyData* cfd, SuperVersion* current_sv = nullptr); virtual ~ForwardIterator(); void SeekToLast() override { status_ = Status::NotSupported("ForwardIterator::SeekToLast()"); valid_ = false; } void Prev() override { status_ = Status::NotSupported("ForwardIterator::Prev"); valid_ = false; } virtual bool Valid() const override; void SeekToFirst() override; virtual void Seek(const Slice& target) override; virtual void Next() override; virtual Slice key() const override; virtual Slice value() const override; virtual Status status() const override; bool TEST_CheckDeletedIters(int* deleted_iters, int* num_iters); private: void Cleanup(bool release_sv); void SVCleanup(); void RebuildIterators(bool refresh_sv); void RenewIterators(); void BuildLevelIterators(const VersionStorageInfo* vstorage); void ResetIncompleteIterators(); void SeekInternal(const Slice& internal_key, bool seek_to_first); void UpdateCurrent(); bool NeedToSeekImmutable(const Slice& internal_key); void DeleteCurrentIter(); uint32_t FindFileInRange( const std::vector& files, const Slice& internal_key, uint32_t left, uint32_t right); bool IsOverUpperBound(const Slice& internal_key) const; DBImpl* const db_; const ReadOptions read_options_; ColumnFamilyData* const cfd_; const SliceTransform* const prefix_extractor_; const Comparator* user_comparator_; MinIterHeap immutable_min_heap_; SuperVersion* sv_; InternalIterator* mutable_iter_; std::vector imm_iters_; std::vector l0_iters_; std::vector level_iters_; InternalIterator* current_; bool valid_; // Internal iterator status; set only by one of the unsupported methods. Status status_; // Status of immutable iterators, maintained here to avoid iterating over // all of them in status(). Status immutable_status_; // Indicates that at least one of the immutable iterators pointed to a key // larger than iterate_upper_bound and was therefore destroyed. Seek() may // need to rebuild such iterators. bool has_iter_trimmed_for_upper_bound_; // Is current key larger than iterate_upper_bound? If so, makes Valid() // return false. bool current_over_upper_bound_; // Left endpoint of the range of keys that immutable iterators currently // cover. When Seek() is called with a key that's within that range, immutable // iterators don't need to be moved; see NeedToSeekImmutable(). This key is // included in the range after a Seek(), but excluded when advancing the // iterator using Next(). IterKey prev_key_; bool is_prev_set_; bool is_prev_inclusive_; Arena arena_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/forward_iterator_bench.cc000066400000000000000000000263061267530634000205750ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #if !defined(GFLAGS) || defined(ROCKSDB_LITE) #include int main() { fprintf(stderr, "Please install gflags to run rocksdb tools\n"); return 1; } #elif defined(OS_MACOSX) || defined(OS_WIN) // Block forward_iterator_bench under MAC and Windows int main() { return 0; } #else #include #include #include #include #include #include #include #include #include #include #include #include #include "rocksdb/cache.h" #include "rocksdb/db.h" #include "rocksdb/status.h" #include "rocksdb/table.h" #include "util/testharness.h" const int MAX_SHARDS = 100000; DEFINE_int32(writers, 8, ""); DEFINE_int32(readers, 8, ""); DEFINE_int64(rate, 100000, ""); DEFINE_int64(value_size, 300, ""); DEFINE_int64(shards, 1000, ""); DEFINE_int64(memtable_size, 500000000, ""); DEFINE_int64(block_cache_size, 300000000, ""); DEFINE_int64(block_size, 65536, ""); DEFINE_double(runtime, 300.0, ""); DEFINE_bool(cache_only_first, true, ""); DEFINE_bool(iterate_upper_bound, true, ""); struct Stats { char pad1[128] __attribute__((__unused__)); std::atomic written{0}; char pad2[128] __attribute__((__unused__)); std::atomic read{0}; std::atomic cache_misses{0}; char pad3[128] __attribute__((__unused__)); } stats; struct Key { Key() {} Key(uint64_t shard_in, uint64_t seqno_in) : shard_be(htobe64(shard_in)), seqno_be(htobe64(seqno_in)) {} uint64_t shard() const { return be64toh(shard_be); } uint64_t seqno() const { return be64toh(seqno_be); } private: uint64_t shard_be; uint64_t seqno_be; } __attribute__((__packed__)); struct Reader; struct Writer; struct ShardState { char pad1[128] __attribute__((__unused__)); std::atomic last_written{0}; Writer* writer; Reader* reader; char pad2[128] __attribute__((__unused__)); std::atomic last_read{0}; std::unique_ptr it; std::unique_ptr it_cacheonly; Key upper_bound; rocksdb::Slice upper_bound_slice; char pad3[128] __attribute__((__unused__)); }; struct Reader { public: explicit Reader(std::vector* shard_states, rocksdb::DB* db) : shard_states_(shard_states), db_(db) { sem_init(&sem_, 0, 0); thread_ = std::thread(&Reader::run, this); } void run() { while (1) { sem_wait(&sem_); if (done_.load()) { break; } uint64_t shard; { std::lock_guard guard(queue_mutex_); assert(!shards_pending_queue_.empty()); shard = shards_pending_queue_.front(); shards_pending_queue_.pop(); shards_pending_set_.reset(shard); } readOnceFromShard(shard); } } void readOnceFromShard(uint64_t shard) { ShardState& state = (*shard_states_)[shard]; if (!state.it) { // Initialize iterators rocksdb::ReadOptions options; options.tailing = true; if (FLAGS_iterate_upper_bound) { state.upper_bound = Key(shard, std::numeric_limits::max()); state.upper_bound_slice = rocksdb::Slice( (const char*)&state.upper_bound, sizeof(state.upper_bound)); options.iterate_upper_bound = &state.upper_bound_slice; } state.it.reset(db_->NewIterator(options)); if (FLAGS_cache_only_first) { options.read_tier = rocksdb::ReadTier::kBlockCacheTier; state.it_cacheonly.reset(db_->NewIterator(options)); } } const uint64_t upto = state.last_written.load(); for (rocksdb::Iterator* it : {state.it_cacheonly.get(), state.it.get()}) { if (it == nullptr) { continue; } if (state.last_read.load() >= upto) { break; } bool need_seek = true; for (uint64_t seq = state.last_read.load() + 1; seq <= upto; ++seq) { if (need_seek) { Key from(shard, state.last_read.load() + 1); it->Seek(rocksdb::Slice((const char*)&from, sizeof(from))); need_seek = false; } else { it->Next(); } if (it->status().IsIncomplete()) { ++::stats.cache_misses; break; } assert(it->Valid()); assert(it->key().size() == sizeof(Key)); Key key; memcpy(&key, it->key().data(), it->key().size()); // fprintf(stderr, "Expecting (%ld, %ld) read (%ld, %ld)\n", // shard, seq, key.shard(), key.seqno()); assert(key.shard() == shard); assert(key.seqno() == seq); state.last_read.store(seq); ++::stats.read; } } } void onWrite(uint64_t shard) { { std::lock_guard guard(queue_mutex_); if (!shards_pending_set_.test(shard)) { shards_pending_queue_.push(shard); shards_pending_set_.set(shard); sem_post(&sem_); } } } ~Reader() { done_.store(true); sem_post(&sem_); thread_.join(); } private: char pad1[128] __attribute__((__unused__)); std::vector* shard_states_; rocksdb::DB* db_; std::thread thread_; sem_t sem_; std::mutex queue_mutex_; std::bitset shards_pending_set_; std::queue shards_pending_queue_; std::atomic done_{false}; char pad2[128] __attribute__((__unused__)); }; struct Writer { explicit Writer(std::vector* shard_states, rocksdb::DB* db) : shard_states_(shard_states), db_(db) {} void start() { thread_ = std::thread(&Writer::run, this); } void run() { std::queue workq; std::chrono::steady_clock::time_point deadline( std::chrono::steady_clock::now() + std::chrono::nanoseconds((uint64_t)(1000000000 * FLAGS_runtime))); std::vector my_shards; for (int i = 1; i <= FLAGS_shards; ++i) { if ((*shard_states_)[i].writer == this) { my_shards.push_back(i); } } std::mt19937 rng{std::random_device()()}; std::uniform_int_distribution shard_dist( 0, static_cast(my_shards.size()) - 1); std::string value(FLAGS_value_size, '*'); while (1) { auto now = std::chrono::steady_clock::now(); if (FLAGS_runtime >= 0 && now >= deadline) { break; } if (workq.empty()) { for (int i = 0; i < FLAGS_rate; i += FLAGS_writers) { std::chrono::nanoseconds offset(1000000000LL * i / FLAGS_rate); workq.push(now + offset); } } while (!workq.empty() && workq.front() < now) { workq.pop(); uint64_t shard = my_shards[shard_dist(rng)]; ShardState& state = (*shard_states_)[shard]; uint64_t seqno = state.last_written.load() + 1; Key key(shard, seqno); // fprintf(stderr, "Writing (%ld, %ld)\n", shard, seqno); rocksdb::Status status = db_->Put(rocksdb::WriteOptions(), rocksdb::Slice((const char*)&key, sizeof(key)), rocksdb::Slice(value)); assert(status.ok()); state.last_written.store(seqno); state.reader->onWrite(shard); ++::stats.written; } std::this_thread::sleep_for(std::chrono::milliseconds(1)); } // fprintf(stderr, "Writer done\n"); } ~Writer() { thread_.join(); } private: char pad1[128] __attribute__((__unused__)); std::vector* shard_states_; rocksdb::DB* db_; std::thread thread_; char pad2[128] __attribute__((__unused__)); }; struct StatsThread { explicit StatsThread(rocksdb::DB* db) : db_(db), thread_(&StatsThread::run, this) {} void run() { // using namespace std::chrono; auto tstart = std::chrono::steady_clock::now(), tlast = tstart; uint64_t wlast = 0, rlast = 0; while (!done_.load()) { { std::unique_lock lock(cvm_); cv_.wait_for(lock, std::chrono::seconds(1)); } auto now = std::chrono::steady_clock::now(); double elapsed = std::chrono::duration_cast >( now - tlast).count(); uint64_t w = ::stats.written.load(); uint64_t r = ::stats.read.load(); fprintf(stderr, "%s elapsed %4lds | written %10ld | w/s %10.0f | read %10ld | " "r/s %10.0f | cache misses %10ld\n", db_->GetEnv()->TimeToString(time(nullptr)).c_str(), std::chrono::duration_cast(now - tstart) .count(), w, (w - wlast) / elapsed, r, (r - rlast) / elapsed, ::stats.cache_misses.load()); wlast = w; rlast = r; tlast = now; } } ~StatsThread() { { std::lock_guard guard(cvm_); done_.store(true); } cv_.notify_all(); thread_.join(); } private: rocksdb::DB* db_; std::mutex cvm_; std::condition_variable cv_; std::thread thread_; std::atomic done_{false}; }; int main(int argc, char** argv) { GFLAGS::ParseCommandLineFlags(&argc, &argv, true); std::mt19937 rng{std::random_device()()}; rocksdb::Status status; std::string path = rocksdb::test::TmpDir() + "/forward_iterator_test"; fprintf(stderr, "db path is %s\n", path.c_str()); rocksdb::Options options; options.create_if_missing = true; options.compression = rocksdb::CompressionType::kNoCompression; options.compaction_style = rocksdb::CompactionStyle::kCompactionStyleNone; options.level0_slowdown_writes_trigger = 99999; options.level0_stop_writes_trigger = 99999; options.allow_os_buffer = false; options.write_buffer_size = FLAGS_memtable_size; rocksdb::BlockBasedTableOptions table_options; table_options.block_cache = rocksdb::NewLRUCache(FLAGS_block_cache_size); table_options.block_size = FLAGS_block_size; options.table_factory.reset( rocksdb::NewBlockBasedTableFactory(table_options)); status = rocksdb::DestroyDB(path, options); assert(status.ok()); rocksdb::DB* db_raw; status = rocksdb::DB::Open(options, path, &db_raw); assert(status.ok()); std::unique_ptr db(db_raw); std::vector shard_states(FLAGS_shards + 1); std::deque readers; while (static_cast(readers.size()) < FLAGS_readers) { readers.emplace_back(&shard_states, db_raw); } std::deque writers; while (static_cast(writers.size()) < FLAGS_writers) { writers.emplace_back(&shard_states, db_raw); } // Each shard gets a random reader and random writer assigned to it for (int i = 1; i <= FLAGS_shards; ++i) { std::uniform_int_distribution reader_dist(0, FLAGS_readers - 1); std::uniform_int_distribution writer_dist(0, FLAGS_writers - 1); shard_states[i].reader = &readers[reader_dist(rng)]; shard_states[i].writer = &writers[writer_dist(rng)]; } StatsThread stats_thread(db_raw); for (Writer& w : writers) { w.start(); } writers.clear(); readers.clear(); } #endif // !defined(GFLAGS) || defined(ROCKSDB_LITE) rocksdb-4.5.1/db/inlineskiplist.h000066400000000000000000000550701267530634000167640ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional // grant of patent rights can be found in the PATENTS file in the same // directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. Use of // this source code is governed by a BSD-style license that can be found // in the LICENSE file. See the AUTHORS file for names of contributors. // // InlineSkipList is derived from SkipList (skiplist.h), but it optimizes // the memory layout by requiring that the key storage be allocated through // the skip list instance. For the common case of SkipList this saves 1 pointer per skip list node and gives better cache // locality, at the expense of wasted padding from using AllocateAligned // instead of Allocate for the keys. The unused padding will be from // 0 to sizeof(void*)-1 bytes, and the space savings are sizeof(void*) // bytes, so despite the padding the space used is always less than // SkipList. // // Thread safety ------------- // // Writes via Insert require external synchronization, most likely a mutex. // InsertConcurrently can be safely called concurrently with reads and // with other concurrent inserts. Reads require a guarantee that the // InlineSkipList will not be destroyed while the read is in progress. // Apart from that, reads progress without any internal locking or // synchronization. // // Invariants: // // (1) Allocated nodes are never deleted until the InlineSkipList is // destroyed. This is trivially guaranteed by the code since we never // delete any skip list nodes. // // (2) The contents of a Node except for the next/prev pointers are // immutable after the Node has been linked into the InlineSkipList. // Only Insert() modifies the list, and it is careful to initialize a // node and use release-stores to publish the nodes in one or more lists. // // ... prev vs. next pointer ordering ... // #pragma once #include #include #include #include "port/port.h" #include "util/allocator.h" #include "util/random.h" namespace rocksdb { template class InlineSkipList { private: struct Node; public: // Create a new InlineSkipList object that will use "cmp" for comparing // keys, and will allocate memory using "*allocator". Objects allocated // in the allocator must remain allocated for the lifetime of the // skiplist object. explicit InlineSkipList(Comparator cmp, Allocator* allocator, int32_t max_height = 12, int32_t branching_factor = 4); // Allocates a key and a skip-list node, returning a pointer to the key // portion of the node. This method is thread-safe if the allocator // is thread-safe. char* AllocateKey(size_t key_size); // Inserts a key allocated by AllocateKey, after the actual key value // has been filled in. // // REQUIRES: nothing that compares equal to key is currently in the list. // REQUIRES: no concurrent calls to INSERT void Insert(const char* key); // Like Insert, but external synchronization is not required. void InsertConcurrently(const char* key); // Returns true iff an entry that compares equal to key is in the list. bool Contains(const char* key) const; // Return estimated number of entries smaller than `key`. uint64_t EstimateCount(const char* key) const; // Iteration over the contents of a skip list class Iterator { public: // Initialize an iterator over the specified list. // The returned iterator is not valid. explicit Iterator(const InlineSkipList* list); // Change the underlying skiplist used for this iterator // This enables us not changing the iterator without deallocating // an old one and then allocating a new one void SetList(const InlineSkipList* list); // Returns true iff the iterator is positioned at a valid node. bool Valid() const; // Returns the key at the current position. // REQUIRES: Valid() const char* key() const; // Advances to the next position. // REQUIRES: Valid() void Next(); // Advances to the previous position. // REQUIRES: Valid() void Prev(); // Advance to the first entry with a key >= target void Seek(const char* target); // Position at the first entry in list. // Final state of iterator is Valid() iff list is not empty. void SeekToFirst(); // Position at the last entry in list. // Final state of iterator is Valid() iff list is not empty. void SeekToLast(); private: const InlineSkipList* list_; Node* node_; // Intentionally copyable }; private: enum MaxPossibleHeightEnum : uint16_t { kMaxPossibleHeight = 32 }; const uint16_t kMaxHeight_; const uint16_t kBranching_; const uint32_t kScaledInverseBranching_; // Immutable after construction Comparator const compare_; Allocator* const allocator_; // Allocator used for allocations of nodes Node* const head_; // Modified only by Insert(). Read racily by readers, but stale // values are ok. std::atomic max_height_; // Height of the entire list // Used for optimizing sequential insert patterns. Tricky. prev_height_ // of zero means prev_ is undefined. Otherwise: prev_[i] for i up // to max_height_ - 1 (inclusive) is the predecessor of prev_[0], and // prev_height_ is the height of prev_[0]. prev_[0] can only be equal // to head when max_height_ and prev_height_ are both 1. Node** prev_; std::atomic prev_height_; inline int GetMaxHeight() const { return max_height_.load(std::memory_order_relaxed); } int RandomHeight(); Node* AllocateNode(size_t key_size, int height); bool Equal(const char* a, const char* b) const { return (compare_(a, b) == 0); } // Return true if key is greater than the data stored in "n". Null n // is considered infinite. bool KeyIsAfterNode(const char* key, Node* n) const; // Returns the earliest node with a key >= key. // Return nullptr if there is no such node. Node* FindGreaterOrEqual(const char* key) const; // Return the latest node with a key < key. // Return head_ if there is no such node. // Fills prev[level] with pointer to previous node at "level" for every // level in [0..max_height_-1], if prev is non-null. Node* FindLessThan(const char* key, Node** prev = nullptr) const; // Return the last node in the list. // Return head_ if list is empty. Node* FindLast() const; // Traverses a single level of the list, setting *out_prev to the last // node before the key and *out_next to the first node after. Assumes // that the key is not present in the skip list. On entry, before should // point to a node that is before the key, and after should point to // a node that is after the key. after should be nullptr if a good after // node isn't conveniently available. void FindLevelSplice(const char* key, Node* before, Node* after, int level, Node** out_prev, Node** out_next); // No copying allowed InlineSkipList(const InlineSkipList&); InlineSkipList& operator=(const InlineSkipList&); }; // Implementation details follow // The Node data type is more of a pointer into custom-managed memory than // a traditional C++ struct. The key is stored in the bytes immediately // after the struct, and the next_ pointers for nodes with height > 1 are // stored immediately _before_ the struct. This avoids the need to include // any pointer or sizing data, which reduces per-node memory overheads. template struct InlineSkipList::Node { // Stores the height of the node in the memory location normally used for // next_[0]. This is used for passing data from AllocateKey to Insert. void StashHeight(const int height) { assert(sizeof(int) <= sizeof(next_[0])); memcpy(&next_[0], &height, sizeof(int)); } // Retrieves the value passed to StashHeight. Undefined after a call // to SetNext or NoBarrier_SetNext. int UnstashHeight() const { int rv; memcpy(&rv, &next_[0], sizeof(int)); return rv; } const char* Key() const { return reinterpret_cast(&next_[1]); } // Accessors/mutators for links. Wrapped in methods so we can add // the appropriate barriers as necessary, and perform the necessary // addressing trickery for storing links below the Node in memory. Node* Next(int n) { assert(n >= 0); // Use an 'acquire load' so that we observe a fully initialized // version of the returned Node. return (next_[-n].load(std::memory_order_acquire)); } void SetNext(int n, Node* x) { assert(n >= 0); // Use a 'release store' so that anybody who reads through this // pointer observes a fully initialized version of the inserted node. next_[-n].store(x, std::memory_order_release); } bool CASNext(int n, Node* expected, Node* x) { assert(n >= 0); return next_[-n].compare_exchange_strong(expected, x); } // No-barrier variants that can be safely used in a few locations. Node* NoBarrier_Next(int n) { assert(n >= 0); return next_[-n].load(std::memory_order_relaxed); } void NoBarrier_SetNext(int n, Node* x) { assert(n >= 0); next_[-n].store(x, std::memory_order_relaxed); } private: // next_[0] is the lowest level link (level 0). Higher levels are // stored _earlier_, so level 1 is at next_[-1]. std::atomic next_[1]; }; template inline InlineSkipList::Iterator::Iterator( const InlineSkipList* list) { SetList(list); } template inline void InlineSkipList::Iterator::SetList( const InlineSkipList* list) { list_ = list; node_ = nullptr; } template inline bool InlineSkipList::Iterator::Valid() const { return node_ != nullptr; } template inline const char* InlineSkipList::Iterator::key() const { assert(Valid()); return node_->Key(); } template inline void InlineSkipList::Iterator::Next() { assert(Valid()); node_ = node_->Next(0); } template inline void InlineSkipList::Iterator::Prev() { // Instead of using explicit "prev" links, we just search for the // last node that falls before key. assert(Valid()); node_ = list_->FindLessThan(node_->Key()); if (node_ == list_->head_) { node_ = nullptr; } } template inline void InlineSkipList::Iterator::Seek(const char* target) { node_ = list_->FindGreaterOrEqual(target); } template inline void InlineSkipList::Iterator::SeekToFirst() { node_ = list_->head_->Next(0); } template inline void InlineSkipList::Iterator::SeekToLast() { node_ = list_->FindLast(); if (node_ == list_->head_) { node_ = nullptr; } } template int InlineSkipList::RandomHeight() { auto rnd = Random::GetTLSInstance(); // Increase height with probability 1 in kBranching int height = 1; while (height < kMaxHeight_ && height < kMaxPossibleHeight && rnd->Next() < kScaledInverseBranching_) { height++; } assert(height > 0); assert(height <= kMaxHeight_); assert(height <= kMaxPossibleHeight); return height; } template bool InlineSkipList::KeyIsAfterNode(const char* key, Node* n) const { // nullptr n is considered infinite return (n != nullptr) && (compare_(n->Key(), key) < 0); } template typename InlineSkipList::Node* InlineSkipList::FindGreaterOrEqual(const char* key) const { // Note: It looks like we could reduce duplication by implementing // this function as FindLessThan(key)->Next(0), but we wouldn't be able // to exit early on equality and the result wouldn't even be correct. // A concurrent insert might occur after FindLessThan(key) but before // we get a chance to call Next(0). Node* x = head_; int level = GetMaxHeight() - 1; Node* last_bigger = nullptr; while (true) { Node* next = x->Next(level); // Make sure the lists are sorted assert(x == head_ || next == nullptr || KeyIsAfterNode(next->Key(), x)); // Make sure we haven't overshot during our search assert(x == head_ || KeyIsAfterNode(key, x)); int cmp = (next == nullptr || next == last_bigger) ? 1 : compare_(next->Key(), key); if (cmp == 0 || (cmp > 0 && level == 0)) { return next; } else if (cmp < 0) { // Keep searching in this list x = next; } else { // Switch to next list, reuse compare_() result last_bigger = next; level--; } } } template typename InlineSkipList::Node* InlineSkipList::FindLessThan(const char* key, Node** prev) const { Node* x = head_; int level = GetMaxHeight() - 1; // KeyIsAfter(key, last_not_after) is definitely false Node* last_not_after = nullptr; while (true) { Node* next = x->Next(level); assert(x == head_ || next == nullptr || KeyIsAfterNode(next->Key(), x)); assert(x == head_ || KeyIsAfterNode(key, x)); if (next != last_not_after && KeyIsAfterNode(key, next)) { // Keep searching in this list x = next; } else { if (prev != nullptr) { prev[level] = x; } if (level == 0) { return x; } else { // Switch to next list, reuse KeyIUsAfterNode() result last_not_after = next; level--; } } } } template typename InlineSkipList::Node* InlineSkipList::FindLast() const { Node* x = head_; int level = GetMaxHeight() - 1; while (true) { Node* next = x->Next(level); if (next == nullptr) { if (level == 0) { return x; } else { // Switch to next list level--; } } else { x = next; } } } template uint64_t InlineSkipList::EstimateCount(const char* key) const { uint64_t count = 0; Node* x = head_; int level = GetMaxHeight() - 1; while (true) { assert(x == head_ || compare_(x->Key(), key) < 0); Node* next = x->Next(level); if (next == nullptr || compare_(next->Key(), key) >= 0) { if (level == 0) { return count; } else { // Switch to next list count *= kBranching_; level--; } } else { x = next; count++; } } } template InlineSkipList::InlineSkipList(const Comparator cmp, Allocator* allocator, int32_t max_height, int32_t branching_factor) : kMaxHeight_(max_height), kBranching_(branching_factor), kScaledInverseBranching_((Random::kMaxNext + 1) / kBranching_), compare_(cmp), allocator_(allocator), head_(AllocateNode(0, max_height)), max_height_(1), prev_height_(1) { assert(max_height > 0 && kMaxHeight_ == static_cast(max_height)); assert(branching_factor > 1 && kBranching_ == static_cast(branching_factor)); assert(kScaledInverseBranching_ > 0); // Allocate the prev_ Node* array, directly from the passed-in allocator. // prev_ does not need to be freed, as its life cycle is tied up with // the allocator as a whole. prev_ = reinterpret_cast( allocator_->AllocateAligned(sizeof(Node*) * kMaxHeight_)); for (int i = 0; i < kMaxHeight_; i++) { head_->SetNext(i, nullptr); prev_[i] = head_; } } template char* InlineSkipList::AllocateKey(size_t key_size) { return const_cast(AllocateNode(key_size, RandomHeight())->Key()); } template typename InlineSkipList::Node* InlineSkipList::AllocateNode(size_t key_size, int height) { auto prefix = sizeof(std::atomic) * (height - 1); // prefix is space for the height - 1 pointers that we store before // the Node instance (next_[-(height - 1) .. -1]). Node starts at // raw + prefix, and holds the bottom-mode (level 0) skip list pointer // next_[0]. key_size is the bytes for the key, which comes just after // the Node. char* raw = allocator_->AllocateAligned(prefix + sizeof(Node) + key_size); Node* x = reinterpret_cast(raw + prefix); // Once we've linked the node into the skip list we don't actually need // to know its height, because we can implicitly use the fact that we // traversed into a node at level h to known that h is a valid level // for that node. We need to convey the height to the Insert step, // however, so that it can perform the proper links. Since we're not // using the pointers at the moment, StashHeight temporarily borrow // storage from next_[0] for that purpose. x->StashHeight(height); return x; } template void InlineSkipList::Insert(const char* key) { // InsertConcurrently often can't maintain the prev_ invariants, so // it just sets prev_height_ to zero, letting us know that we should // ignore it. A relaxed load suffices here because write thread // synchronization separates Insert calls from InsertConcurrently calls. auto prev_height = prev_height_.load(std::memory_order_relaxed); // fast path for sequential insertion if (prev_height > 0 && !KeyIsAfterNode(key, prev_[0]->NoBarrier_Next(0)) && (prev_[0] == head_ || KeyIsAfterNode(key, prev_[0]))) { assert(prev_[0] != head_ || (prev_height == 1 && GetMaxHeight() == 1)); // Outside of this method prev_[1..max_height_] is the predecessor // of prev_[0], and prev_height_ refers to prev_[0]. Inside Insert // prev_[0..max_height - 1] is the predecessor of key. Switch from // the external state to the internal for (int i = 1; i < prev_height; i++) { prev_[i] = prev_[0]; } } else { // TODO(opt): we could use a NoBarrier predecessor search as an // optimization for architectures where memory_order_acquire needs // a synchronization instruction. Doesn't matter on x86 FindLessThan(key, prev_); } // Our data structure does not allow duplicate insertion assert(prev_[0]->Next(0) == nullptr || !Equal(key, prev_[0]->Next(0)->Key())); // Find the Node that we placed before the key in AllocateKey Node* x = reinterpret_cast(const_cast(key)) - 1; int height = x->UnstashHeight(); assert(height >= 1 && height <= kMaxHeight_); if (height > GetMaxHeight()) { for (int i = GetMaxHeight(); i < height; i++) { prev_[i] = head_; } // It is ok to mutate max_height_ without any synchronization // with concurrent readers. A concurrent reader that observes // the new value of max_height_ will see either the old value of // new level pointers from head_ (nullptr), or a new value set in // the loop below. In the former case the reader will // immediately drop to the next level since nullptr sorts after all // keys. In the latter case the reader will use the new node. max_height_.store(height, std::memory_order_relaxed); } for (int i = 0; i < height; i++) { // NoBarrier_SetNext() suffices since we will add a barrier when // we publish a pointer to "x" in prev[i]. x->NoBarrier_SetNext(i, prev_[i]->NoBarrier_Next(i)); prev_[i]->SetNext(i, x); } prev_[0] = x; prev_height_.store(height, std::memory_order_relaxed); } template void InlineSkipList::FindLevelSplice(const char* key, Node* before, Node* after, int level, Node** out_prev, Node** out_next) { while (true) { Node* next = before->Next(level); assert(before == head_ || next == nullptr || KeyIsAfterNode(next->Key(), before)); assert(before == head_ || KeyIsAfterNode(key, before)); if (next == after || !KeyIsAfterNode(key, next)) { // found it *out_prev = before; *out_next = next; return; } before = next; } } template void InlineSkipList::InsertConcurrently(const char* key) { Node* x = reinterpret_cast(const_cast(key)) - 1; int height = x->UnstashHeight(); assert(height >= 1 && height <= kMaxHeight_); // We don't have a lock-free algorithm for updating prev_, but we do have // the option of invalidating the entire sequential-insertion cache. // prev_'s invariant is that prev_[i] (i > 0) is the predecessor of // prev_[0] at that level. We're only going to violate that if height // > 1 and key lands after prev_[height - 1] but before prev_[0]. // Comparisons are pretty expensive, so an easier version is to just // clear the cache if height > 1. We only write to prev_height_ if the // nobody else has, to avoid invalidating the root of the skip list in // all of the other CPU caches. if (height > 1 && prev_height_.load(std::memory_order_relaxed) != 0) { prev_height_.store(0, std::memory_order_relaxed); } int max_height = max_height_.load(std::memory_order_relaxed); while (height > max_height) { if (max_height_.compare_exchange_strong(max_height, height)) { // successfully updated it max_height = height; break; } // else retry, possibly exiting the loop because somebody else // increased it } assert(max_height <= kMaxPossibleHeight); Node* prev[kMaxPossibleHeight + 1]; Node* next[kMaxPossibleHeight + 1]; prev[max_height] = head_; next[max_height] = nullptr; for (int i = max_height - 1; i >= 0; --i) { FindLevelSplice(key, prev[i + 1], next[i + 1], i, &prev[i], &next[i]); } for (int i = 0; i < height; ++i) { while (true) { x->NoBarrier_SetNext(i, next[i]); if (prev[i]->CASNext(i, next[i], x)) { // success break; } // CAS failed, we need to recompute prev and next. It is unlikely // to be helpful to try to use a different level as we redo the // search, because it should be unlikely that lots of nodes have // been inserted between prev[i] and next[i]. No point in using // next[i] as the after hint, because we know it is stale. FindLevelSplice(key, prev[i], nullptr, i, &prev[i], &next[i]); } } } template bool InlineSkipList::Contains(const char* key) const { Node* x = FindGreaterOrEqual(key); if (x != nullptr && Equal(key, x->Key())) { return true; } else { return false; } } } // namespace rocksdb rocksdb-4.5.1/db/inlineskiplist_test.cc000066400000000000000000000322601267530634000201550ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/inlineskiplist.h" #include #include "rocksdb/env.h" #include "util/concurrent_arena.h" #include "util/hash.h" #include "util/random.h" #include "util/testharness.h" namespace rocksdb { // Our test skip list stores 8-byte unsigned integers typedef uint64_t Key; static const char* Encode(const uint64_t* key) { return reinterpret_cast(key); } static Key Decode(const char* key) { Key rv; memcpy(&rv, key, sizeof(Key)); return rv; } struct TestComparator { int operator()(const char* a, const char* b) const { if (Decode(a) < Decode(b)) { return -1; } else if (Decode(a) > Decode(b)) { return +1; } else { return 0; } } }; class InlineSkipTest : public testing::Test {}; TEST_F(InlineSkipTest, Empty) { Arena arena; TestComparator cmp; InlineSkipList list(cmp, &arena); Key key = 10; ASSERT_TRUE(!list.Contains(Encode(&key))); InlineSkipList::Iterator iter(&list); ASSERT_TRUE(!iter.Valid()); iter.SeekToFirst(); ASSERT_TRUE(!iter.Valid()); key = 100; iter.Seek(Encode(&key)); ASSERT_TRUE(!iter.Valid()); iter.SeekToLast(); ASSERT_TRUE(!iter.Valid()); } TEST_F(InlineSkipTest, InsertAndLookup) { const int N = 2000; const int R = 5000; Random rnd(1000); std::set keys; ConcurrentArena arena; TestComparator cmp; InlineSkipList list(cmp, &arena); for (int i = 0; i < N; i++) { Key key = rnd.Next() % R; if (keys.insert(key).second) { char* buf = list.AllocateKey(sizeof(Key)); memcpy(buf, &key, sizeof(Key)); list.Insert(buf); } } for (Key i = 0; i < R; i++) { if (list.Contains(Encode(&i))) { ASSERT_EQ(keys.count(i), 1U); } else { ASSERT_EQ(keys.count(i), 0U); } } // Simple iterator tests { InlineSkipList::Iterator iter(&list); ASSERT_TRUE(!iter.Valid()); uint64_t zero = 0; iter.Seek(Encode(&zero)); ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*(keys.begin()), Decode(iter.key())); iter.SeekToFirst(); ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*(keys.begin()), Decode(iter.key())); iter.SeekToLast(); ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*(keys.rbegin()), Decode(iter.key())); } // Forward iteration test for (Key i = 0; i < R; i++) { InlineSkipList::Iterator iter(&list); iter.Seek(Encode(&i)); // Compare against model iterator std::set::iterator model_iter = keys.lower_bound(i); for (int j = 0; j < 3; j++) { if (model_iter == keys.end()) { ASSERT_TRUE(!iter.Valid()); break; } else { ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*model_iter, Decode(iter.key())); ++model_iter; iter.Next(); } } } // Backward iteration test { InlineSkipList::Iterator iter(&list); iter.SeekToLast(); // Compare against model iterator for (std::set::reverse_iterator model_iter = keys.rbegin(); model_iter != keys.rend(); ++model_iter) { ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*model_iter, Decode(iter.key())); iter.Prev(); } ASSERT_TRUE(!iter.Valid()); } } // We want to make sure that with a single writer and multiple // concurrent readers (with no synchronization other than when a // reader's iterator is created), the reader always observes all the // data that was present in the skip list when the iterator was // constructor. Because insertions are happening concurrently, we may // also observe new values that were inserted since the iterator was // constructed, but we should never miss any values that were present // at iterator construction time. // // We generate multi-part keys: // // where: // key is in range [0..K-1] // gen is a generation number for key // hash is hash(key,gen) // // The insertion code picks a random key, sets gen to be 1 + the last // generation number inserted for that key, and sets hash to Hash(key,gen). // // At the beginning of a read, we snapshot the last inserted // generation number for each key. We then iterate, including random // calls to Next() and Seek(). For every key we encounter, we // check that it is either expected given the initial snapshot or has // been concurrently added since the iterator started. class ConcurrentTest { public: static const uint32_t K = 8; private: static uint64_t key(Key key) { return (key >> 40); } static uint64_t gen(Key key) { return (key >> 8) & 0xffffffffu; } static uint64_t hash(Key key) { return key & 0xff; } static uint64_t HashNumbers(uint64_t k, uint64_t g) { uint64_t data[2] = {k, g}; return Hash(reinterpret_cast(data), sizeof(data), 0); } static Key MakeKey(uint64_t k, uint64_t g) { assert(sizeof(Key) == sizeof(uint64_t)); assert(k <= K); // We sometimes pass K to seek to the end of the skiplist assert(g <= 0xffffffffu); return ((k << 40) | (g << 8) | (HashNumbers(k, g) & 0xff)); } static bool IsValidKey(Key k) { return hash(k) == (HashNumbers(key(k), gen(k)) & 0xff); } static Key RandomTarget(Random* rnd) { switch (rnd->Next() % 10) { case 0: // Seek to beginning return MakeKey(0, 0); case 1: // Seek to end return MakeKey(K, 0); default: // Seek to middle return MakeKey(rnd->Next() % K, 0); } } // Per-key generation struct State { std::atomic generation[K]; void Set(int k, int v) { generation[k].store(v, std::memory_order_release); } int Get(int k) { return generation[k].load(std::memory_order_acquire); } State() { for (unsigned int k = 0; k < K; k++) { Set(k, 0); } } }; // Current state of the test State current_; ConcurrentArena arena_; // InlineSkipList is not protected by mu_. We just use a single writer // thread to modify it. InlineSkipList list_; public: ConcurrentTest() : list_(TestComparator(), &arena_) {} // REQUIRES: No concurrent calls to WriteStep or ConcurrentWriteStep void WriteStep(Random* rnd) { const uint32_t k = rnd->Next() % K; const int g = current_.Get(k) + 1; const Key new_key = MakeKey(k, g); char* buf = list_.AllocateKey(sizeof(Key)); memcpy(buf, &new_key, sizeof(Key)); list_.Insert(buf); current_.Set(k, g); } // REQUIRES: No concurrent calls for the same k void ConcurrentWriteStep(uint32_t k) { const int g = current_.Get(k) + 1; const Key new_key = MakeKey(k, g); char* buf = list_.AllocateKey(sizeof(Key)); memcpy(buf, &new_key, sizeof(Key)); list_.InsertConcurrently(buf); ASSERT_EQ(g, current_.Get(k) + 1); current_.Set(k, g); } void ReadStep(Random* rnd) { // Remember the initial committed state of the skiplist. State initial_state; for (unsigned int k = 0; k < K; k++) { initial_state.Set(k, current_.Get(k)); } Key pos = RandomTarget(rnd); InlineSkipList::Iterator iter(&list_); iter.Seek(Encode(&pos)); while (true) { Key current; if (!iter.Valid()) { current = MakeKey(K, 0); } else { current = Decode(iter.key()); ASSERT_TRUE(IsValidKey(current)) << current; } ASSERT_LE(pos, current) << "should not go backwards"; // Verify that everything in [pos,current) was not present in // initial_state. while (pos < current) { ASSERT_LT(key(pos), K) << pos; // Note that generation 0 is never inserted, so it is ok if // <*,0,*> is missing. ASSERT_TRUE((gen(pos) == 0U) || (gen(pos) > static_cast(initial_state.Get( static_cast(key(pos)))))) << "key: " << key(pos) << "; gen: " << gen(pos) << "; initgen: " << initial_state.Get(static_cast(key(pos))); // Advance to next key in the valid key space if (key(pos) < key(current)) { pos = MakeKey(key(pos) + 1, 0); } else { pos = MakeKey(key(pos), gen(pos) + 1); } } if (!iter.Valid()) { break; } if (rnd->Next() % 2) { iter.Next(); pos = MakeKey(key(pos), gen(pos) + 1); } else { Key new_target = RandomTarget(rnd); if (new_target > pos) { pos = new_target; iter.Seek(Encode(&new_target)); } } } } }; const uint32_t ConcurrentTest::K; // Simple test that does single-threaded testing of the ConcurrentTest // scaffolding. TEST_F(InlineSkipTest, ConcurrentReadWithoutThreads) { ConcurrentTest test; Random rnd(test::RandomSeed()); for (int i = 0; i < 10000; i++) { test.ReadStep(&rnd); test.WriteStep(&rnd); } } TEST_F(InlineSkipTest, ConcurrentInsertWithoutThreads) { ConcurrentTest test; Random rnd(test::RandomSeed()); for (int i = 0; i < 10000; i++) { test.ReadStep(&rnd); uint32_t base = rnd.Next(); for (int j = 0; j < 4; ++j) { test.ConcurrentWriteStep((base + j) % ConcurrentTest::K); } } } class TestState { public: ConcurrentTest t_; int seed_; std::atomic quit_flag_; std::atomic next_writer_; enum ReaderState { STARTING, RUNNING, DONE }; explicit TestState(int s) : seed_(s), quit_flag_(false), state_(STARTING), pending_writers_(0), state_cv_(&mu_) {} void Wait(ReaderState s) { mu_.Lock(); while (state_ != s) { state_cv_.Wait(); } mu_.Unlock(); } void Change(ReaderState s) { mu_.Lock(); state_ = s; state_cv_.Signal(); mu_.Unlock(); } void AdjustPendingWriters(int delta) { mu_.Lock(); pending_writers_ += delta; if (pending_writers_ == 0) { state_cv_.Signal(); } mu_.Unlock(); } void WaitForPendingWriters() { mu_.Lock(); while (pending_writers_ != 0) { state_cv_.Wait(); } mu_.Unlock(); } private: port::Mutex mu_; ReaderState state_; int pending_writers_; port::CondVar state_cv_; }; static void ConcurrentReader(void* arg) { TestState* state = reinterpret_cast(arg); Random rnd(state->seed_); int64_t reads = 0; state->Change(TestState::RUNNING); while (!state->quit_flag_.load(std::memory_order_acquire)) { state->t_.ReadStep(&rnd); ++reads; } state->Change(TestState::DONE); } static void ConcurrentWriter(void* arg) { TestState* state = reinterpret_cast(arg); uint32_t k = state->next_writer_++ % ConcurrentTest::K; state->t_.ConcurrentWriteStep(k); state->AdjustPendingWriters(-1); } static void RunConcurrentRead(int run) { const int seed = test::RandomSeed() + (run * 100); Random rnd(seed); const int N = 1000; const int kSize = 1000; for (int i = 0; i < N; i++) { if ((i % 100) == 0) { fprintf(stderr, "Run %d of %d\n", i, N); } TestState state(seed + 1); Env::Default()->Schedule(ConcurrentReader, &state); state.Wait(TestState::RUNNING); for (int k = 0; k < kSize; ++k) { state.t_.WriteStep(&rnd); } state.quit_flag_.store(true, std::memory_order_release); state.Wait(TestState::DONE); } } static void RunConcurrentInsert(int run, int write_parallelism = 4) { Env::Default()->SetBackgroundThreads(1 + write_parallelism, Env::Priority::LOW); const int seed = test::RandomSeed() + (run * 100); Random rnd(seed); const int N = 1000; const int kSize = 1000; for (int i = 0; i < N; i++) { if ((i % 100) == 0) { fprintf(stderr, "Run %d of %d\n", i, N); } TestState state(seed + 1); Env::Default()->Schedule(ConcurrentReader, &state); state.Wait(TestState::RUNNING); for (int k = 0; k < kSize; k += write_parallelism) { state.next_writer_ = rnd.Next(); state.AdjustPendingWriters(write_parallelism); for (int p = 0; p < write_parallelism; ++p) { Env::Default()->Schedule(ConcurrentWriter, &state); } state.WaitForPendingWriters(); } state.quit_flag_.store(true, std::memory_order_release); state.Wait(TestState::DONE); } } TEST_F(InlineSkipTest, ConcurrentRead1) { RunConcurrentRead(1); } TEST_F(InlineSkipTest, ConcurrentRead2) { RunConcurrentRead(2); } TEST_F(InlineSkipTest, ConcurrentRead3) { RunConcurrentRead(3); } TEST_F(InlineSkipTest, ConcurrentRead4) { RunConcurrentRead(4); } TEST_F(InlineSkipTest, ConcurrentRead5) { RunConcurrentRead(5); } TEST_F(InlineSkipTest, ConcurrentInsert1) { RunConcurrentInsert(1); } TEST_F(InlineSkipTest, ConcurrentInsert2) { RunConcurrentInsert(2); } TEST_F(InlineSkipTest, ConcurrentInsert3) { RunConcurrentInsert(3); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/internal_stats.cc000066400000000000000000001073161267530634000171140ustar00rootroot00000000000000// This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/internal_stats.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include "db/column_family.h" #include "db/db_impl.h" #include "util/string_util.h" namespace rocksdb { #ifndef ROCKSDB_LITE namespace { const double kMB = 1048576.0; const double kGB = kMB * 1024; const double kMicrosInSec = 1000000.0; void PrintLevelStatsHeader(char* buf, size_t len, const std::string& cf_name) { snprintf( buf, len, "\n** Compaction Stats [%s] **\n" "Level Files Size(MB) Score Read(GB) Rn(GB) Rnp1(GB) " "Write(GB) Wnew(GB) Moved(GB) W-Amp Rd(MB/s) Wr(MB/s) " "Comp(sec) Comp(cnt) Avg(sec) KeyIn KeyDrop\n" "--------------------------------------------------------------------" "-----------------------------------------------------------" "--------------------------------------\n", cf_name.c_str()); } void PrintLevelStats(char* buf, size_t len, const std::string& name, int num_files, int being_compacted, double total_file_size, double score, double w_amp, const InternalStats::CompactionStats& stats) { uint64_t bytes_read = stats.bytes_read_non_output_levels + stats.bytes_read_output_level; int64_t bytes_new = stats.bytes_written - stats.bytes_read_output_level; double elapsed = (stats.micros + 1) / kMicrosInSec; std::string num_input_records = NumberToHumanString(stats.num_input_records); std::string num_dropped_records = NumberToHumanString(stats.num_dropped_records); snprintf(buf, len, "%4s %6d/%-3d %8.2f %5.1f " /* Level, Files, Size(MB), Score */ "%8.1f " /* Read(GB) */ "%7.1f " /* Rn(GB) */ "%8.1f " /* Rnp1(GB) */ "%9.1f " /* Write(GB) */ "%8.1f " /* Wnew(GB) */ "%9.1f " /* Moved(GB) */ "%5.1f " /* W-Amp */ "%8.1f " /* Rd(MB/s) */ "%8.1f " /* Wr(MB/s) */ "%9.0f " /* Comp(sec) */ "%9d " /* Comp(cnt) */ "%8.3f " /* Avg(sec) */ "%7s " /* KeyIn */ "%6s\n", /* KeyDrop */ name.c_str(), num_files, being_compacted, total_file_size / kMB, score, bytes_read / kGB, stats.bytes_read_non_output_levels / kGB, stats.bytes_read_output_level / kGB, stats.bytes_written / kGB, bytes_new / kGB, stats.bytes_moved / kGB, w_amp, bytes_read / kMB / elapsed, stats.bytes_written / kMB / elapsed, stats.micros / kMicrosInSec, stats.count, stats.count == 0 ? 0 : stats.micros / kMicrosInSec / stats.count, num_input_records.c_str(), num_dropped_records.c_str()); } // Assumes that trailing numbers represent an optional argument. This requires // property names to not end with numbers. std::pair GetPropertyNameAndArg(const Slice& property) { Slice name = property, arg = property; size_t sfx_len = 0; while (sfx_len < property.size() && isdigit(property[property.size() - sfx_len - 1])) { ++sfx_len; } name.remove_suffix(sfx_len); arg.remove_prefix(property.size() - sfx_len); return {name, arg}; } } // anonymous namespace static const std::string rocksdb_prefix = "rocksdb."; static const std::string num_files_at_level_prefix = "num-files-at-level"; static const std::string allstats = "stats"; static const std::string sstables = "sstables"; static const std::string cfstats = "cfstats"; static const std::string dbstats = "dbstats"; static const std::string levelstats = "levelstats"; static const std::string num_immutable_mem_table = "num-immutable-mem-table"; static const std::string num_immutable_mem_table_flushed = "num-immutable-mem-table-flushed"; static const std::string mem_table_flush_pending = "mem-table-flush-pending"; static const std::string compaction_pending = "compaction-pending"; static const std::string background_errors = "background-errors"; static const std::string cur_size_active_mem_table = "cur-size-active-mem-table"; static const std::string cur_size_all_mem_tables = "cur-size-all-mem-tables"; static const std::string size_all_mem_tables = "size-all-mem-tables"; static const std::string num_entries_active_mem_table = "num-entries-active-mem-table"; static const std::string num_entries_imm_mem_tables = "num-entries-imm-mem-tables"; static const std::string num_deletes_active_mem_table = "num-deletes-active-mem-table"; static const std::string num_deletes_imm_mem_tables = "num-deletes-imm-mem-tables"; static const std::string estimate_num_keys = "estimate-num-keys"; static const std::string estimate_table_readers_mem = "estimate-table-readers-mem"; static const std::string is_file_deletions_enabled = "is-file-deletions-enabled"; static const std::string num_snapshots = "num-snapshots"; static const std::string oldest_snapshot_time = "oldest-snapshot-time"; static const std::string num_live_versions = "num-live-versions"; static const std::string estimate_live_data_size = "estimate-live-data-size"; static const std::string base_level = "base-level"; static const std::string total_sst_files_size = "total-sst-files-size"; static const std::string estimate_pending_comp_bytes = "estimate-pending-compaction-bytes"; static const std::string aggregated_table_properties = "aggregated-table-properties"; static const std::string aggregated_table_properties_at_level = aggregated_table_properties + "-at-level"; static const std::string num_running_compactions = "num-running-compactions"; static const std::string num_running_flushes = "num-running-flushes"; const std::string DB::Properties::kNumFilesAtLevelPrefix = rocksdb_prefix + num_files_at_level_prefix; const std::string DB::Properties::kStats = rocksdb_prefix + allstats; const std::string DB::Properties::kSSTables = rocksdb_prefix + sstables; const std::string DB::Properties::kCFStats = rocksdb_prefix + cfstats; const std::string DB::Properties::kDBStats = rocksdb_prefix + dbstats; const std::string DB::Properties::kLevelStats = rocksdb_prefix + levelstats; const std::string DB::Properties::kNumImmutableMemTable = rocksdb_prefix + num_immutable_mem_table; const std::string DB::Properties::kNumImmutableMemTableFlushed = rocksdb_prefix + num_immutable_mem_table_flushed; const std::string DB::Properties::kMemTableFlushPending = rocksdb_prefix + mem_table_flush_pending; const std::string DB::Properties::kCompactionPending = rocksdb_prefix + compaction_pending; const std::string DB::Properties::kNumRunningCompactions = rocksdb_prefix + num_running_compactions; const std::string DB::Properties::kNumRunningFlushes = rocksdb_prefix + num_running_flushes; const std::string DB::Properties::kBackgroundErrors = rocksdb_prefix + background_errors; const std::string DB::Properties::kCurSizeActiveMemTable = rocksdb_prefix + cur_size_active_mem_table; const std::string DB::Properties::kCurSizeAllMemTables = rocksdb_prefix + cur_size_all_mem_tables; const std::string DB::Properties::kSizeAllMemTables = rocksdb_prefix + size_all_mem_tables; const std::string DB::Properties::kNumEntriesActiveMemTable = rocksdb_prefix + num_entries_active_mem_table; const std::string DB::Properties::kNumEntriesImmMemTables = rocksdb_prefix + num_entries_imm_mem_tables; const std::string DB::Properties::kNumDeletesActiveMemTable = rocksdb_prefix + num_deletes_active_mem_table; const std::string DB::Properties::kNumDeletesImmMemTables = rocksdb_prefix + num_deletes_imm_mem_tables; const std::string DB::Properties::kEstimateNumKeys = rocksdb_prefix + estimate_num_keys; const std::string DB::Properties::kEstimateTableReadersMem = rocksdb_prefix + estimate_table_readers_mem; const std::string DB::Properties::kIsFileDeletionsEnabled = rocksdb_prefix + is_file_deletions_enabled; const std::string DB::Properties::kNumSnapshots = rocksdb_prefix + num_snapshots; const std::string DB::Properties::kOldestSnapshotTime = rocksdb_prefix + oldest_snapshot_time; const std::string DB::Properties::kNumLiveVersions = rocksdb_prefix + num_live_versions; const std::string DB::Properties::kEstimateLiveDataSize = rocksdb_prefix + estimate_live_data_size; const std::string DB::Properties::kTotalSstFilesSize = rocksdb_prefix + total_sst_files_size; const std::string DB::Properties::kBaseLevel = rocksdb_prefix + base_level; const std::string DB::Properties::kEstimatePendingCompactionBytes = rocksdb_prefix + estimate_pending_comp_bytes; const std::string DB::Properties::kAggregatedTableProperties = rocksdb_prefix + aggregated_table_properties; const std::string DB::Properties::kAggregatedTablePropertiesAtLevel = rocksdb_prefix + aggregated_table_properties_at_level; const std::unordered_map InternalStats::ppt_name_to_info = { {DB::Properties::kNumFilesAtLevelPrefix, {false, &InternalStats::HandleNumFilesAtLevel, nullptr}}, {DB::Properties::kLevelStats, {false, &InternalStats::HandleLevelStats, nullptr}}, {DB::Properties::kStats, {false, &InternalStats::HandleStats, nullptr}}, {DB::Properties::kCFStats, {false, &InternalStats::HandleCFStats, nullptr}}, {DB::Properties::kDBStats, {false, &InternalStats::HandleDBStats, nullptr}}, {DB::Properties::kSSTables, {false, &InternalStats::HandleSsTables, nullptr}}, {DB::Properties::kAggregatedTableProperties, {false, &InternalStats::HandleAggregatedTableProperties, nullptr}}, {DB::Properties::kAggregatedTablePropertiesAtLevel, {false, &InternalStats::HandleAggregatedTablePropertiesAtLevel, nullptr}}, {DB::Properties::kNumImmutableMemTable, {false, nullptr, &InternalStats::HandleNumImmutableMemTable}}, {DB::Properties::kNumImmutableMemTableFlushed, {false, nullptr, &InternalStats::HandleNumImmutableMemTableFlushed}}, {DB::Properties::kMemTableFlushPending, {false, nullptr, &InternalStats::HandleMemTableFlushPending}}, {DB::Properties::kCompactionPending, {false, nullptr, &InternalStats::HandleCompactionPending}}, {DB::Properties::kBackgroundErrors, {false, nullptr, &InternalStats::HandleBackgroundErrors}}, {DB::Properties::kCurSizeActiveMemTable, {false, nullptr, &InternalStats::HandleCurSizeActiveMemTable}}, {DB::Properties::kCurSizeAllMemTables, {false, nullptr, &InternalStats::HandleCurSizeAllMemTables}}, {DB::Properties::kSizeAllMemTables, {false, nullptr, &InternalStats::HandleSizeAllMemTables}}, {DB::Properties::kNumEntriesActiveMemTable, {false, nullptr, &InternalStats::HandleNumEntriesActiveMemTable}}, {DB::Properties::kNumEntriesImmMemTables, {false, nullptr, &InternalStats::HandleNumEntriesImmMemTables}}, {DB::Properties::kNumDeletesActiveMemTable, {false, nullptr, &InternalStats::HandleNumDeletesActiveMemTable}}, {DB::Properties::kNumDeletesImmMemTables, {false, nullptr, &InternalStats::HandleNumDeletesImmMemTables}}, {DB::Properties::kEstimateNumKeys, {false, nullptr, &InternalStats::HandleEstimateNumKeys}}, {DB::Properties::kEstimateTableReadersMem, {true, nullptr, &InternalStats::HandleEstimateTableReadersMem}}, {DB::Properties::kIsFileDeletionsEnabled, {false, nullptr, &InternalStats::HandleIsFileDeletionsEnabled}}, {DB::Properties::kNumSnapshots, {false, nullptr, &InternalStats::HandleNumSnapshots}}, {DB::Properties::kOldestSnapshotTime, {false, nullptr, &InternalStats::HandleOldestSnapshotTime}}, {DB::Properties::kNumLiveVersions, {false, nullptr, &InternalStats::HandleNumLiveVersions}}, {DB::Properties::kEstimateLiveDataSize, {true, nullptr, &InternalStats::HandleEstimateLiveDataSize}}, {DB::Properties::kBaseLevel, {false, nullptr, &InternalStats::HandleBaseLevel}}, {DB::Properties::kTotalSstFilesSize, {false, nullptr, &InternalStats::HandleTotalSstFilesSize}}, {DB::Properties::kEstimatePendingCompactionBytes, {false, nullptr, &InternalStats::HandleEstimatePendingCompactionBytes}}, {DB::Properties::kNumRunningFlushes, {false, nullptr, &InternalStats::HandleNumRunningFlushes}}, {DB::Properties::kNumRunningCompactions, {false, nullptr, &InternalStats::HandleNumRunningCompactions}}, }; const DBPropertyInfo* GetPropertyInfo(const Slice& property) { std::string ppt_name = GetPropertyNameAndArg(property).first.ToString(); auto ppt_info_iter = InternalStats::ppt_name_to_info.find(ppt_name); if (ppt_info_iter == InternalStats::ppt_name_to_info.end()) { return nullptr; } return &ppt_info_iter->second; } bool InternalStats::GetStringProperty(const DBPropertyInfo& property_info, const Slice& property, std::string* value) { assert(value != nullptr); assert(property_info.handle_string != nullptr); Slice arg = GetPropertyNameAndArg(property).second; return (this->*(property_info.handle_string))(value, arg); } bool InternalStats::GetIntProperty(const DBPropertyInfo& property_info, uint64_t* value, DBImpl* db) { assert(value != nullptr); assert(property_info.handle_int != nullptr && !property_info.need_out_of_mutex); db->mutex_.AssertHeld(); return (this->*(property_info.handle_int))(value, db, nullptr /* version */); } bool InternalStats::GetIntPropertyOutOfMutex( const DBPropertyInfo& property_info, Version* version, uint64_t* value) { assert(value != nullptr); assert(property_info.handle_int != nullptr && property_info.need_out_of_mutex); return (this->*(property_info.handle_int))(value, nullptr /* db */, version); } bool InternalStats::HandleNumFilesAtLevel(std::string* value, Slice suffix) { uint64_t level; const auto* vstorage = cfd_->current()->storage_info(); bool ok = ConsumeDecimalNumber(&suffix, &level) && suffix.empty(); if (!ok || static_cast(level) >= number_levels_) { return false; } else { char buf[100]; snprintf(buf, sizeof(buf), "%d", vstorage->NumLevelFiles(static_cast(level))); *value = buf; return true; } } bool InternalStats::HandleLevelStats(std::string* value, Slice suffix) { char buf[1000]; const auto* vstorage = cfd_->current()->storage_info(); snprintf(buf, sizeof(buf), "Level Files Size(MB)\n" "--------------------\n"); value->append(buf); for (int level = 0; level < number_levels_; level++) { snprintf(buf, sizeof(buf), "%3d %8d %8.0f\n", level, vstorage->NumLevelFiles(level), vstorage->NumLevelBytes(level) / kMB); value->append(buf); } return true; } bool InternalStats::HandleStats(std::string* value, Slice suffix) { if (!HandleCFStats(value, suffix)) { return false; } if (!HandleDBStats(value, suffix)) { return false; } return true; } bool InternalStats::HandleCFStats(std::string* value, Slice suffix) { DumpCFStats(value); return true; } bool InternalStats::HandleDBStats(std::string* value, Slice suffix) { DumpDBStats(value); return true; } bool InternalStats::HandleSsTables(std::string* value, Slice suffix) { auto* current = cfd_->current(); *value = current->DebugString(); return true; } bool InternalStats::HandleAggregatedTableProperties(std::string* value, Slice suffix) { std::shared_ptr tp; auto s = cfd_->current()->GetAggregatedTableProperties(&tp); if (!s.ok()) { return false; } *value = tp->ToString(); return true; } bool InternalStats::HandleAggregatedTablePropertiesAtLevel(std::string* value, Slice suffix) { uint64_t level; bool ok = ConsumeDecimalNumber(&suffix, &level) && suffix.empty(); if (!ok || static_cast(level) >= number_levels_) { return false; } std::shared_ptr tp; auto s = cfd_->current()->GetAggregatedTableProperties( &tp, static_cast(level)); if (!s.ok()) { return false; } *value = tp->ToString(); return true; } bool InternalStats::HandleNumImmutableMemTable(uint64_t* value, DBImpl* db, Version* version) { *value = cfd_->imm()->NumNotFlushed(); return true; } bool InternalStats::HandleNumImmutableMemTableFlushed(uint64_t* value, DBImpl* db, Version* version) { *value = cfd_->imm()->NumFlushed(); return true; } bool InternalStats::HandleMemTableFlushPending(uint64_t* value, DBImpl* db, Version* version) { // Return number of mem tables that are ready to flush (made immutable) *value = (cfd_->imm()->IsFlushPending() ? 1 : 0); return true; } bool InternalStats::HandleNumRunningFlushes(uint64_t* value, DBImpl* db, Version* version) { *value = db->num_running_flushes(); return true; } bool InternalStats::HandleCompactionPending(uint64_t* value, DBImpl* db, Version* version) { // 1 if the system already determines at least one compaction is needed. // 0 otherwise, const auto* vstorage = cfd_->current()->storage_info(); *value = (cfd_->compaction_picker()->NeedsCompaction(vstorage) ? 1 : 0); return true; } bool InternalStats::HandleNumRunningCompactions(uint64_t* value, DBImpl* db, Version* version) { *value = db->num_running_compactions_; return true; } bool InternalStats::HandleBackgroundErrors(uint64_t* value, DBImpl* db, Version* version) { // Accumulated number of errors in background flushes or compactions. *value = GetBackgroundErrorCount(); return true; } bool InternalStats::HandleCurSizeActiveMemTable(uint64_t* value, DBImpl* db, Version* version) { // Current size of the active memtable *value = cfd_->mem()->ApproximateMemoryUsage(); return true; } bool InternalStats::HandleCurSizeAllMemTables(uint64_t* value, DBImpl* db, Version* version) { // Current size of the active memtable + immutable memtables *value = cfd_->mem()->ApproximateMemoryUsage() + cfd_->imm()->ApproximateUnflushedMemTablesMemoryUsage(); return true; } bool InternalStats::HandleSizeAllMemTables(uint64_t* value, DBImpl* db, Version* version) { *value = cfd_->mem()->ApproximateMemoryUsage() + cfd_->imm()->ApproximateMemoryUsage(); return true; } bool InternalStats::HandleNumEntriesActiveMemTable(uint64_t* value, DBImpl* db, Version* version) { // Current number of entires in the active memtable *value = cfd_->mem()->num_entries(); return true; } bool InternalStats::HandleNumEntriesImmMemTables(uint64_t* value, DBImpl* db, Version* version) { // Current number of entries in the immutable memtables *value = cfd_->imm()->current()->GetTotalNumEntries(); return true; } bool InternalStats::HandleNumDeletesActiveMemTable(uint64_t* value, DBImpl* db, Version* version) { // Current number of entires in the active memtable *value = cfd_->mem()->num_deletes(); return true; } bool InternalStats::HandleNumDeletesImmMemTables(uint64_t* value, DBImpl* db, Version* version) { // Current number of entries in the immutable memtables *value = cfd_->imm()->current()->GetTotalNumDeletes(); return true; } bool InternalStats::HandleEstimateNumKeys(uint64_t* value, DBImpl* db, Version* version) { // Estimate number of entries in the column family: // Use estimated entries in tables + total entries in memtables. const auto* vstorage = cfd_->current()->storage_info(); *value = cfd_->mem()->num_entries() + cfd_->imm()->current()->GetTotalNumEntries() - (cfd_->mem()->num_deletes() + cfd_->imm()->current()->GetTotalNumDeletes()) * 2 + vstorage->GetEstimatedActiveKeys(); return true; } bool InternalStats::HandleNumSnapshots(uint64_t* value, DBImpl* db, Version* version) { *value = db->snapshots().count(); return true; } bool InternalStats::HandleOldestSnapshotTime(uint64_t* value, DBImpl* db, Version* version) { *value = static_cast(db->snapshots().GetOldestSnapshotTime()); return true; } bool InternalStats::HandleNumLiveVersions(uint64_t* value, DBImpl* db, Version* version) { *value = cfd_->GetNumLiveVersions(); return true; } bool InternalStats::HandleIsFileDeletionsEnabled(uint64_t* value, DBImpl* db, Version* version) { *value = db->IsFileDeletionsEnabled(); return true; } bool InternalStats::HandleBaseLevel(uint64_t* value, DBImpl* db, Version* version) { const auto* vstorage = cfd_->current()->storage_info(); *value = vstorage->base_level(); return true; } bool InternalStats::HandleTotalSstFilesSize(uint64_t* value, DBImpl* db, Version* version) { *value = cfd_->GetTotalSstFilesSize(); return true; } bool InternalStats::HandleEstimatePendingCompactionBytes(uint64_t* value, DBImpl* db, Version* version) { const auto* vstorage = cfd_->current()->storage_info(); *value = vstorage->estimated_compaction_needed_bytes(); return true; } bool InternalStats::HandleEstimateTableReadersMem(uint64_t* value, DBImpl* db, Version* version) { *value = (version == nullptr) ? 0 : version->GetMemoryUsageByTableReaders(); return true; } bool InternalStats::HandleEstimateLiveDataSize(uint64_t* value, DBImpl* db, Version* version) { const auto* vstorage = cfd_->current()->storage_info(); *value = vstorage->EstimateLiveDataSize(); return true; } void InternalStats::DumpDBStats(std::string* value) { char buf[1000]; // DB-level stats, only available from default column family double seconds_up = (env_->NowMicros() - started_at_ + 1) / kMicrosInSec; double interval_seconds_up = seconds_up - db_stats_snapshot_.seconds_up; snprintf(buf, sizeof(buf), "\n** DB Stats **\nUptime(secs): %.1f total, %.1f interval\n", seconds_up, interval_seconds_up); value->append(buf); // Cumulative uint64_t user_bytes_written = GetDBStats(InternalStats::BYTES_WRITTEN); uint64_t num_keys_written = GetDBStats(InternalStats::NUMBER_KEYS_WRITTEN); uint64_t write_other = GetDBStats(InternalStats::WRITE_DONE_BY_OTHER); uint64_t write_self = GetDBStats(InternalStats::WRITE_DONE_BY_SELF); uint64_t wal_bytes = GetDBStats(InternalStats::WAL_FILE_BYTES); uint64_t wal_synced = GetDBStats(InternalStats::WAL_FILE_SYNCED); uint64_t write_with_wal = GetDBStats(InternalStats::WRITE_WITH_WAL); uint64_t write_stall_micros = GetDBStats(InternalStats::WRITE_STALL_MICROS); uint64_t compact_bytes_read = 0; uint64_t compact_bytes_write = 0; uint64_t compact_micros = 0; const int kHumanMicrosLen = 32; char human_micros[kHumanMicrosLen]; // Data // writes: total number of write requests. // keys: total number of key updates issued by all the write requests // batches: number of group commits issued to the DB. Each group can contain // one or more writes. // so writes/keys is the average number of put in multi-put or put // writes/batches is the average group commit size. // // The format is the same for interval stats. snprintf(buf, sizeof(buf), "Cumulative writes: %s writes, %s keys, %s batches, " "%.1f writes per batch, ingest: %.2f GB, %.2f MB/s\n", NumberToHumanString(write_other + write_self).c_str(), NumberToHumanString(num_keys_written).c_str(), NumberToHumanString(write_self).c_str(), (write_other + write_self) / static_cast(write_self + 1), user_bytes_written / kGB, user_bytes_written / kMB / seconds_up); value->append(buf); // WAL snprintf(buf, sizeof(buf), "Cumulative WAL: %s writes, %s syncs, " "%.2f writes per sync, written: %.2f GB, %.2f MB/s\n", NumberToHumanString(write_with_wal).c_str(), NumberToHumanString(wal_synced).c_str(), write_with_wal / static_cast(wal_synced + 1), wal_bytes / kGB, wal_bytes / kMB / seconds_up); value->append(buf); // Compact for (int level = 0; level < number_levels_; level++) { compact_bytes_read += comp_stats_[level].bytes_read_output_level + comp_stats_[level].bytes_read_non_output_levels; compact_bytes_write += comp_stats_[level].bytes_written; compact_micros += comp_stats_[level].micros; } snprintf(buf, sizeof(buf), "Cumulative compaction: %.2f GB write, %.2f MB/s write, " "%.2f GB read, %.2f MB/s read, %.1f seconds\n", compact_bytes_write / kGB, compact_bytes_write / kMB / seconds_up, compact_bytes_read / kGB, compact_bytes_read / kMB / seconds_up, compact_micros / kMicrosInSec); value->append(buf); // Stall AppendHumanMicros(write_stall_micros, human_micros, kHumanMicrosLen, true); snprintf(buf, sizeof(buf), "Cumulative stall: %s, %.1f percent\n", human_micros, // 10000 = divide by 1M to get secs, then multiply by 100 for pct write_stall_micros / 10000.0 / std::max(seconds_up, 0.001)); value->append(buf); // Interval uint64_t interval_write_other = write_other - db_stats_snapshot_.write_other; uint64_t interval_write_self = write_self - db_stats_snapshot_.write_self; uint64_t interval_num_keys_written = num_keys_written - db_stats_snapshot_.num_keys_written; snprintf(buf, sizeof(buf), "Interval writes: %s writes, %s keys, %s batches, " "%.1f writes per batch, ingest: %.2f MB, %.2f MB/s\n", NumberToHumanString( interval_write_other + interval_write_self).c_str(), NumberToHumanString(interval_num_keys_written).c_str(), NumberToHumanString(interval_write_self).c_str(), static_cast(interval_write_other + interval_write_self) / (interval_write_self + 1), (user_bytes_written - db_stats_snapshot_.ingest_bytes) / kMB, (user_bytes_written - db_stats_snapshot_.ingest_bytes) / kMB / std::max(interval_seconds_up, 0.001)), value->append(buf); uint64_t interval_write_with_wal = write_with_wal - db_stats_snapshot_.write_with_wal; uint64_t interval_wal_synced = wal_synced - db_stats_snapshot_.wal_synced; uint64_t interval_wal_bytes = wal_bytes - db_stats_snapshot_.wal_bytes; snprintf(buf, sizeof(buf), "Interval WAL: %s writes, %s syncs, " "%.2f writes per sync, written: %.2f MB, %.2f MB/s\n", NumberToHumanString(interval_write_with_wal).c_str(), NumberToHumanString(interval_wal_synced).c_str(), interval_write_with_wal / static_cast(interval_wal_synced + 1), interval_wal_bytes / kGB, interval_wal_bytes / kMB / std::max(interval_seconds_up, 0.001)); value->append(buf); // Compaction uint64_t interval_compact_bytes_write = compact_bytes_write - db_stats_snapshot_.compact_bytes_write; uint64_t interval_compact_bytes_read = compact_bytes_read - db_stats_snapshot_.compact_bytes_read; uint64_t interval_compact_micros = compact_micros - db_stats_snapshot_.compact_micros; snprintf( buf, sizeof(buf), "Interval compaction: %.2f GB write, %.2f MB/s write, " "%.2f GB read, %.2f MB/s read, %.1f seconds\n", interval_compact_bytes_write / kGB, interval_compact_bytes_write / kMB / std::max(interval_seconds_up, 0.001), interval_compact_bytes_read / kGB, interval_compact_bytes_read / kMB / std::max(interval_seconds_up, 0.001), interval_compact_micros / kMicrosInSec); value->append(buf); // Stall AppendHumanMicros( write_stall_micros - db_stats_snapshot_.write_stall_micros, human_micros, kHumanMicrosLen, true); snprintf(buf, sizeof(buf), "Interval stall: %s, %.1f percent\n", human_micros, // 10000 = divide by 1M to get secs, then multiply by 100 for pct (write_stall_micros - db_stats_snapshot_.write_stall_micros) / 10000.0 / std::max(interval_seconds_up, 0.001)); value->append(buf); for (int level = 0; level < number_levels_; level++) { if (!file_read_latency_[level].Empty()) { char buf2[5000]; snprintf(buf2, sizeof(buf2), "** Level %d read latency histogram (micros):\n%s\n", level, file_read_latency_[level].ToString().c_str()); value->append(buf2); } } db_stats_snapshot_.seconds_up = seconds_up; db_stats_snapshot_.ingest_bytes = user_bytes_written; db_stats_snapshot_.write_other = write_other; db_stats_snapshot_.write_self = write_self; db_stats_snapshot_.num_keys_written = num_keys_written; db_stats_snapshot_.wal_bytes = wal_bytes; db_stats_snapshot_.wal_synced = wal_synced; db_stats_snapshot_.write_with_wal = write_with_wal; db_stats_snapshot_.write_stall_micros = write_stall_micros; db_stats_snapshot_.compact_bytes_write = compact_bytes_write; db_stats_snapshot_.compact_bytes_read = compact_bytes_read; db_stats_snapshot_.compact_micros = compact_micros; } void InternalStats::DumpCFStats(std::string* value) { const VersionStorageInfo* vstorage = cfd_->current()->storage_info(); int num_levels_to_check = (cfd_->ioptions()->compaction_style != kCompactionStyleFIFO) ? vstorage->num_levels() - 1 : 1; // Compaction scores are sorted base on its value. Restore them to the // level order std::vector compaction_score(number_levels_, 0); for (int i = 0; i < num_levels_to_check; ++i) { compaction_score[vstorage->CompactionScoreLevel(i)] = vstorage->CompactionScore(i); } // Count # of files being compacted for each level std::vector files_being_compacted(number_levels_, 0); for (int level = 0; level < number_levels_; ++level) { for (auto* f : vstorage->LevelFiles(level)) { if (f->being_compacted) { ++files_being_compacted[level]; } } } char buf[1000]; // Per-ColumnFamily stats PrintLevelStatsHeader(buf, sizeof(buf), cfd_->GetName()); value->append(buf); CompactionStats stats_sum(0); int total_files = 0; int total_files_being_compacted = 0; double total_file_size = 0; for (int level = 0; level < number_levels_; level++) { int files = vstorage->NumLevelFiles(level); total_files += files; total_files_being_compacted += files_being_compacted[level]; if (comp_stats_[level].micros > 0 || files > 0) { stats_sum.Add(comp_stats_[level]); total_file_size += vstorage->NumLevelBytes(level); double w_amp = (comp_stats_[level].bytes_read_non_output_levels == 0) ? 0.0 : static_cast(comp_stats_[level].bytes_written) / comp_stats_[level].bytes_read_non_output_levels; PrintLevelStats(buf, sizeof(buf), "L" + ToString(level), files, files_being_compacted[level], static_cast(vstorage->NumLevelBytes(level)), compaction_score[level], w_amp, comp_stats_[level]); value->append(buf); } } uint64_t curr_ingest = cf_stats_value_[BYTES_FLUSHED]; // Cumulative summary double w_amp = stats_sum.bytes_written / static_cast(curr_ingest + 1); uint64_t total_stall_count = cf_stats_count_[LEVEL0_SLOWDOWN_TOTAL] + cf_stats_count_[LEVEL0_NUM_FILES_TOTAL] + cf_stats_count_[SOFT_PENDING_COMPACTION_BYTES_LIMIT] + cf_stats_count_[HARD_PENDING_COMPACTION_BYTES_LIMIT] + cf_stats_count_[MEMTABLE_COMPACTION] + cf_stats_count_[MEMTABLE_SLOWDOWN]; // Stats summary across levels PrintLevelStats(buf, sizeof(buf), "Sum", total_files, total_files_being_compacted, total_file_size, 0, w_amp, stats_sum); value->append(buf); // Interval summary uint64_t interval_ingest = curr_ingest - cf_stats_snapshot_.ingest_bytes + 1; CompactionStats interval_stats(stats_sum); interval_stats.Subtract(cf_stats_snapshot_.comp_stats); w_amp = interval_stats.bytes_written / static_cast(interval_ingest); PrintLevelStats(buf, sizeof(buf), "Int", 0, 0, 0, 0, w_amp, interval_stats); value->append(buf); snprintf(buf, sizeof(buf), "Flush(GB): cumulative %.3f, interval %.3f\n", curr_ingest / kGB, interval_ingest / kGB); value->append(buf); snprintf(buf, sizeof(buf), "Stalls(count): %" PRIu64 " level0_slowdown, " "%" PRIu64 " level0_slowdown_with_compaction, " "%" PRIu64 " level0_numfiles, " "%" PRIu64 " level0_numfiles_with_compaction, " "%" PRIu64 " stop for pending_compaction_bytes, " "%" PRIu64 " slowdown for pending_compaction_bytes, " "%" PRIu64 " memtable_compaction, " "%" PRIu64 " memtable_slowdown, " "interval %" PRIu64 " total count\n", cf_stats_count_[LEVEL0_SLOWDOWN_TOTAL], cf_stats_count_[LEVEL0_SLOWDOWN_WITH_COMPACTION], cf_stats_count_[LEVEL0_NUM_FILES_TOTAL], cf_stats_count_[LEVEL0_NUM_FILES_WITH_COMPACTION], cf_stats_count_[HARD_PENDING_COMPACTION_BYTES_LIMIT], cf_stats_count_[SOFT_PENDING_COMPACTION_BYTES_LIMIT], cf_stats_count_[MEMTABLE_COMPACTION], cf_stats_count_[MEMTABLE_SLOWDOWN], total_stall_count - cf_stats_snapshot_.stall_count); value->append(buf); cf_stats_snapshot_.ingest_bytes = curr_ingest; cf_stats_snapshot_.comp_stats = stats_sum; cf_stats_snapshot_.stall_count = total_stall_count; } #else const DBPropertyInfo* GetPropertyInfo(const Slice& property) { return nullptr; } #endif // !ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/internal_stats.h000066400000000000000000000373571267530634000167650ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // #pragma once #include "db/version_set.h" #include #include class ColumnFamilyData; namespace rocksdb { class MemTableList; class DBImpl; // Config for retrieving a property's value. struct DBPropertyInfo { bool need_out_of_mutex; // gcc had an internal error for initializing union of pointer-to-member- // functions. Workaround is to populate exactly one of the following function // pointers with a non-nullptr value. // @param value Value-result argument for storing the property's string value // @param suffix Argument portion of the property. For example, suffix would // be "5" for the property "rocksdb.num-files-at-level5". So far, only // certain string properties take an argument. bool (InternalStats::*handle_string)(std::string* value, Slice suffix); // @param value Value-result argument for storing the property's uint64 value // @param db Many of the int properties rely on DBImpl methods. // @param version Version is needed in case the property is retrieved without // holding db mutex, which is only supported for int properties. bool (InternalStats::*handle_int)(uint64_t* value, DBImpl* db, Version* version); }; extern const DBPropertyInfo* GetPropertyInfo(const Slice& property); #ifndef ROCKSDB_LITE class InternalStats { public: enum InternalCFStatsType { LEVEL0_SLOWDOWN_TOTAL, LEVEL0_SLOWDOWN_WITH_COMPACTION, MEMTABLE_COMPACTION, MEMTABLE_SLOWDOWN, LEVEL0_NUM_FILES_TOTAL, LEVEL0_NUM_FILES_WITH_COMPACTION, SOFT_PENDING_COMPACTION_BYTES_LIMIT, HARD_PENDING_COMPACTION_BYTES_LIMIT, WRITE_STALLS_ENUM_MAX, BYTES_FLUSHED, INTERNAL_CF_STATS_ENUM_MAX, }; enum InternalDBStatsType { WAL_FILE_BYTES, WAL_FILE_SYNCED, BYTES_WRITTEN, NUMBER_KEYS_WRITTEN, WRITE_DONE_BY_OTHER, WRITE_DONE_BY_SELF, WRITE_WITH_WAL, WRITE_STALL_MICROS, INTERNAL_DB_STATS_ENUM_MAX, }; InternalStats(int num_levels, Env* env, ColumnFamilyData* cfd) : db_stats_{}, cf_stats_value_{}, cf_stats_count_{}, comp_stats_(num_levels), file_read_latency_(num_levels), bg_error_count_(0), number_levels_(num_levels), env_(env), cfd_(cfd), started_at_(env->NowMicros()) {} // Per level compaction stats. comp_stats_[level] stores the stats for // compactions that produced data for the specified "level". struct CompactionStats { uint64_t micros; // The number of bytes read from all non-output levels uint64_t bytes_read_non_output_levels; // The number of bytes read from the compaction output level. uint64_t bytes_read_output_level; // Total number of bytes written during compaction uint64_t bytes_written; // Total number of bytes moved to the output level uint64_t bytes_moved; // The number of compaction input files in all non-output levels. int num_input_files_in_non_output_levels; // The number of compaction input files in the output level. int num_input_files_in_output_level; // The number of compaction output files. int num_output_files; // Total incoming entries during compaction between levels N and N+1 uint64_t num_input_records; // Accumulated diff number of entries // (num input entries - num output entires) for compaction levels N and N+1 uint64_t num_dropped_records; // Number of compactions done int count; explicit CompactionStats(int _count = 0) : micros(0), bytes_read_non_output_levels(0), bytes_read_output_level(0), bytes_written(0), bytes_moved(0), num_input_files_in_non_output_levels(0), num_input_files_in_output_level(0), num_output_files(0), num_input_records(0), num_dropped_records(0), count(_count) {} explicit CompactionStats(const CompactionStats& c) : micros(c.micros), bytes_read_non_output_levels(c.bytes_read_non_output_levels), bytes_read_output_level(c.bytes_read_output_level), bytes_written(c.bytes_written), bytes_moved(c.bytes_moved), num_input_files_in_non_output_levels( c.num_input_files_in_non_output_levels), num_input_files_in_output_level( c.num_input_files_in_output_level), num_output_files(c.num_output_files), num_input_records(c.num_input_records), num_dropped_records(c.num_dropped_records), count(c.count) {} void Add(const CompactionStats& c) { this->micros += c.micros; this->bytes_read_non_output_levels += c.bytes_read_non_output_levels; this->bytes_read_output_level += c.bytes_read_output_level; this->bytes_written += c.bytes_written; this->bytes_moved += c.bytes_moved; this->num_input_files_in_non_output_levels += c.num_input_files_in_non_output_levels; this->num_input_files_in_output_level += c.num_input_files_in_output_level; this->num_output_files += c.num_output_files; this->num_input_records += c.num_input_records; this->num_dropped_records += c.num_dropped_records; this->count += c.count; } void Subtract(const CompactionStats& c) { this->micros -= c.micros; this->bytes_read_non_output_levels -= c.bytes_read_non_output_levels; this->bytes_read_output_level -= c.bytes_read_output_level; this->bytes_written -= c.bytes_written; this->bytes_moved -= c.bytes_moved; this->num_input_files_in_non_output_levels -= c.num_input_files_in_non_output_levels; this->num_input_files_in_output_level -= c.num_input_files_in_output_level; this->num_output_files -= c.num_output_files; this->num_input_records -= c.num_input_records; this->num_dropped_records -= c.num_dropped_records; this->count -= c.count; } }; void AddCompactionStats(int level, const CompactionStats& stats) { comp_stats_[level].Add(stats); } void IncBytesMoved(int level, uint64_t amount) { comp_stats_[level].bytes_moved += amount; } void AddCFStats(InternalCFStatsType type, uint64_t value) { cf_stats_value_[type] += value; ++cf_stats_count_[type]; } void AddDBStats(InternalDBStatsType type, uint64_t value) { auto& v = db_stats_[type]; v.store(v.load(std::memory_order_relaxed) + value, std::memory_order_relaxed); } uint64_t GetDBStats(InternalDBStatsType type) { return db_stats_[type].load(std::memory_order_relaxed); } HistogramImpl* GetFileReadHist(int level) { return &file_read_latency_[level]; } uint64_t GetBackgroundErrorCount() const { return bg_error_count_; } uint64_t BumpAndGetBackgroundErrorCount() { return ++bg_error_count_; } bool GetStringProperty(const DBPropertyInfo& property_info, const Slice& property, std::string* value); bool GetIntProperty(const DBPropertyInfo& property_info, uint64_t* value, DBImpl* db); bool GetIntPropertyOutOfMutex(const DBPropertyInfo& property_info, Version* version, uint64_t* value); // Store a mapping from the user-facing DB::Properties string to our // DBPropertyInfo struct used internally for retrieving properties. static const std::unordered_map ppt_name_to_info; private: void DumpDBStats(std::string* value); void DumpCFStats(std::string* value); // Per-DB stats std::atomic db_stats_[INTERNAL_DB_STATS_ENUM_MAX]; // Per-ColumnFamily stats uint64_t cf_stats_value_[INTERNAL_CF_STATS_ENUM_MAX]; uint64_t cf_stats_count_[INTERNAL_CF_STATS_ENUM_MAX]; // Per-ColumnFamily/level compaction stats std::vector comp_stats_; std::vector file_read_latency_; // Used to compute per-interval statistics struct CFStatsSnapshot { // ColumnFamily-level stats CompactionStats comp_stats; uint64_t ingest_bytes; // Bytes written to L0 uint64_t stall_count; // Stall count CFStatsSnapshot() : comp_stats(0), ingest_bytes(0), stall_count(0) {} } cf_stats_snapshot_; struct DBStatsSnapshot { // DB-level stats uint64_t ingest_bytes; // Bytes written by user uint64_t wal_bytes; // Bytes written to WAL uint64_t wal_synced; // Number of times WAL is synced uint64_t write_with_wal; // Number of writes that request WAL // These count the number of writes processed by the calling thread or // another thread. uint64_t write_other; uint64_t write_self; // Stats from compaction jobs - bytes written, bytes read, duration. uint64_t compact_bytes_write; uint64_t compact_bytes_read; uint64_t compact_micros; // Total number of keys written. write_self and write_other measure number // of write requests written, Each of the write request can contain updates // to multiple keys. num_keys_written is total number of keys updated by all // those writes. uint64_t num_keys_written; // Total time writes delayed by stalls. uint64_t write_stall_micros; double seconds_up; DBStatsSnapshot() : ingest_bytes(0), wal_bytes(0), wal_synced(0), write_with_wal(0), write_other(0), write_self(0), compact_bytes_write(0), compact_bytes_read(0), compact_micros(0), num_keys_written(0), write_stall_micros(0), seconds_up(0) {} } db_stats_snapshot_; // Handler functions for getting property values. They use "value" as a value- // result argument, and return true upon successfully setting "value". bool HandleNumFilesAtLevel(std::string* value, Slice suffix); bool HandleLevelStats(std::string* value, Slice suffix); bool HandleStats(std::string* value, Slice suffix); bool HandleCFStats(std::string* value, Slice suffix); bool HandleDBStats(std::string* value, Slice suffix); bool HandleSsTables(std::string* value, Slice suffix); bool HandleAggregatedTableProperties(std::string* value, Slice suffix); bool HandleAggregatedTablePropertiesAtLevel(std::string* value, Slice suffix); bool HandleNumImmutableMemTable(uint64_t* value, DBImpl* db, Version* version); bool HandleNumImmutableMemTableFlushed(uint64_t* value, DBImpl* db, Version* version); bool HandleMemTableFlushPending(uint64_t* value, DBImpl* db, Version* version); bool HandleNumRunningFlushes(uint64_t* value, DBImpl* db, Version* version); bool HandleCompactionPending(uint64_t* value, DBImpl* db, Version* version); bool HandleNumRunningCompactions(uint64_t* value, DBImpl* db, Version* version); bool HandleBackgroundErrors(uint64_t* value, DBImpl* db, Version* version); bool HandleCurSizeActiveMemTable(uint64_t* value, DBImpl* db, Version* version); bool HandleCurSizeAllMemTables(uint64_t* value, DBImpl* db, Version* version); bool HandleSizeAllMemTables(uint64_t* value, DBImpl* db, Version* version); bool HandleNumEntriesActiveMemTable(uint64_t* value, DBImpl* db, Version* version); bool HandleNumEntriesImmMemTables(uint64_t* value, DBImpl* db, Version* version); bool HandleNumDeletesActiveMemTable(uint64_t* value, DBImpl* db, Version* version); bool HandleNumDeletesImmMemTables(uint64_t* value, DBImpl* db, Version* version); bool HandleEstimateNumKeys(uint64_t* value, DBImpl* db, Version* version); bool HandleNumSnapshots(uint64_t* value, DBImpl* db, Version* version); bool HandleOldestSnapshotTime(uint64_t* value, DBImpl* db, Version* version); bool HandleNumLiveVersions(uint64_t* value, DBImpl* db, Version* version); bool HandleIsFileDeletionsEnabled(uint64_t* value, DBImpl* db, Version* version); bool HandleBaseLevel(uint64_t* value, DBImpl* db, Version* version); bool HandleTotalSstFilesSize(uint64_t* value, DBImpl* db, Version* version); bool HandleEstimatePendingCompactionBytes(uint64_t* value, DBImpl* db, Version* version); bool HandleEstimateTableReadersMem(uint64_t* value, DBImpl* db, Version* version); bool HandleEstimateLiveDataSize(uint64_t* value, DBImpl* db, Version* version); // Total number of background errors encountered. Every time a flush task // or compaction task fails, this counter is incremented. The failure can // be caused by any possible reason, including file system errors, out of // resources, or input file corruption. Failing when retrying the same flush // or compaction will cause the counter to increase too. uint64_t bg_error_count_; const int number_levels_; Env* env_; ColumnFamilyData* cfd_; const uint64_t started_at_; }; #else class InternalStats { public: enum InternalCFStatsType { LEVEL0_SLOWDOWN_TOTAL, LEVEL0_SLOWDOWN_WITH_COMPACTION, MEMTABLE_COMPACTION, MEMTABLE_SLOWDOWN, LEVEL0_NUM_FILES_TOTAL, LEVEL0_NUM_FILES_WITH_COMPACTION, SOFT_PENDING_COMPACTION_BYTES_LIMIT, HARD_PENDING_COMPACTION_BYTES_LIMIT, WRITE_STALLS_ENUM_MAX, BYTES_FLUSHED, INTERNAL_CF_STATS_ENUM_MAX, }; enum InternalDBStatsType { WAL_FILE_BYTES, WAL_FILE_SYNCED, BYTES_WRITTEN, NUMBER_KEYS_WRITTEN, WRITE_DONE_BY_OTHER, WRITE_DONE_BY_SELF, WRITE_WITH_WAL, WRITE_STALL_MICROS, INTERNAL_DB_STATS_ENUM_MAX, }; InternalStats(int num_levels, Env* env, ColumnFamilyData* cfd) {} struct CompactionStats { uint64_t micros; uint64_t bytes_read_non_output_levels; uint64_t bytes_read_output_level; uint64_t bytes_written; uint64_t bytes_moved; int num_input_files_in_non_output_levels; int num_input_files_in_output_level; int num_output_files; uint64_t num_input_records; uint64_t num_dropped_records; int count; explicit CompactionStats(int _count = 0) {} explicit CompactionStats(const CompactionStats& c) {} void Add(const CompactionStats& c) {} void Subtract(const CompactionStats& c) {} }; void AddCompactionStats(int level, const CompactionStats& stats) {} void IncBytesMoved(int level, uint64_t amount) {} void AddCFStats(InternalCFStatsType type, uint64_t value) {} void AddDBStats(InternalDBStatsType type, uint64_t value) {} HistogramImpl* GetFileReadHist(int level) { return nullptr; } uint64_t GetBackgroundErrorCount() const { return 0; } uint64_t BumpAndGetBackgroundErrorCount() { return 0; } bool GetStringProperty(const DBPropertyInfo& property_info, const Slice& property, std::string* value) { return false; } bool GetIntProperty(const DBPropertyInfo& property_info, uint64_t* value, DBImpl* db) const { return false; } bool GetIntPropertyOutOfMutex(const DBPropertyInfo& property_info, Version* version, uint64_t* value) const { return false; } }; #endif // !ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/job_context.h000066400000000000000000000073151267530634000162400ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include "db/column_family.h" #include "db/log_writer.h" namespace rocksdb { class MemTable; struct JobContext { inline bool HaveSomethingToDelete() const { return full_scan_candidate_files.size() || sst_delete_files.size() || log_delete_files.size() || new_superversion != nullptr || superversions_to_free.size() > 0 || memtables_to_free.size() > 0 || logs_to_free.size() > 0; } // Structure to store information for candidate files to delete. struct CandidateFileInfo { std::string file_name; uint32_t path_id; CandidateFileInfo(std::string name, uint32_t path) : file_name(std::move(name)), path_id(path) {} bool operator==(const CandidateFileInfo& other) const { return file_name == other.file_name && path_id == other.path_id; } }; // Unique job id int job_id; // a list of all files that we'll consider deleting // (every once in a while this is filled up with all files // in the DB directory) // (filled only if we're doing full scan) std::vector full_scan_candidate_files; // the list of all live sst files that cannot be deleted std::vector sst_live; // a list of sst files that we need to delete std::vector sst_delete_files; // a list of log files that we need to delete std::vector log_delete_files; // a list of memtables to be free autovector memtables_to_free; autovector superversions_to_free; autovector logs_to_free; SuperVersion* new_superversion; // if nullptr no new superversion // the current manifest_file_number, log_number and prev_log_number // that corresponds to the set of files in 'live'. uint64_t manifest_file_number; uint64_t pending_manifest_file_number; uint64_t log_number; uint64_t prev_log_number; uint64_t min_pending_output = 0; explicit JobContext(int _job_id, bool create_superversion = false) { job_id = _job_id; manifest_file_number = 0; pending_manifest_file_number = 0; log_number = 0; prev_log_number = 0; new_superversion = create_superversion ? new SuperVersion() : nullptr; } // For non-empty JobContext Clean() has to be called at least once before // before destruction (see asserts in ~JobContext()). Should be called with // unlocked DB mutex. Destructor doesn't call Clean() to avoid accidentally // doing potentially slow Clean() with locked DB mutex. void Clean() { // free pending memtables for (auto m : memtables_to_free) { delete m; } // free superversions for (auto s : superversions_to_free) { delete s; } for (auto l : logs_to_free) { delete l; } // if new_superversion was not used, it will be non-nullptr and needs // to be freed here delete new_superversion; memtables_to_free.clear(); superversions_to_free.clear(); logs_to_free.clear(); new_superversion = nullptr; } ~JobContext() { assert(memtables_to_free.size() == 0); assert(superversions_to_free.size() == 0); assert(new_superversion == nullptr); assert(logs_to_free.size() == 0); } }; } // namespace rocksdb rocksdb-4.5.1/db/listener_test.cc000066400000000000000000000470751267530634000167530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/db_impl.h" #include "db/db_test_util.h" #include "db/dbformat.h" #include "db/filename.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "memtable/hash_linklist_rep.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/options.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "table/block_based_table_factory.h" #include "table/plain_table_factory.h" #include "util/hash.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/rate_limiter.h" #include "util/statistics.h" #include "util/string_util.h" #include "util/sync_point.h" #include "util/testharness.h" #include "util/testutil.h" #include "utilities/merge_operators.h" #ifndef ROCKSDB_LITE namespace rocksdb { class EventListenerTest : public DBTestBase { public: EventListenerTest() : DBTestBase("/listener_test") {} const size_t k110KB = 110 << 10; }; struct TestPropertiesCollector : public rocksdb::TablePropertiesCollector { virtual rocksdb::Status AddUserKey(const rocksdb::Slice& key, const rocksdb::Slice& value, rocksdb::EntryType type, rocksdb::SequenceNumber seq, uint64_t file_size) override { return Status::OK(); } virtual rocksdb::Status Finish( rocksdb::UserCollectedProperties* properties) override { properties->insert({"0", "1"}); return Status::OK(); } virtual const char* Name() const override { return "TestTablePropertiesCollector"; } rocksdb::UserCollectedProperties GetReadableProperties() const override { rocksdb::UserCollectedProperties ret; ret["2"] = "3"; return ret; } }; class TestPropertiesCollectorFactory : public TablePropertiesCollectorFactory { public: virtual TablePropertiesCollector* CreateTablePropertiesCollector( TablePropertiesCollectorFactory::Context context) override { return new TestPropertiesCollector; } const char* Name() const override { return "TestTablePropertiesCollector"; } }; class TestCompactionListener : public EventListener { public: void OnCompactionCompleted(DB *db, const CompactionJobInfo& ci) override { std::lock_guard lock(mutex_); compacted_dbs_.push_back(db); ASSERT_GT(ci.input_files.size(), 0U); ASSERT_GT(ci.output_files.size(), 0U); ASSERT_EQ(db->GetEnv()->GetThreadID(), ci.thread_id); ASSERT_GT(ci.thread_id, 0U); for (auto fl : {ci.input_files, ci.output_files}) { for (auto fn : fl) { auto it = ci.table_properties.find(fn); ASSERT_NE(it, ci.table_properties.end()); auto tp = it->second; ASSERT_TRUE(tp != nullptr); ASSERT_EQ(tp->user_collected_properties.find("0")->second, "1"); } } } std::vector compacted_dbs_; std::mutex mutex_; }; TEST_F(EventListenerTest, OnSingleDBCompactionTest) { const int kTestKeySize = 16; const int kTestValueSize = 984; const int kEntrySize = kTestKeySize + kTestValueSize; const int kEntriesPerBuffer = 100; const int kNumL0Files = 4; Options options; options.create_if_missing = true; options.write_buffer_size = kEntrySize * kEntriesPerBuffer; options.compaction_style = kCompactionStyleLevel; options.target_file_size_base = options.write_buffer_size; options.max_bytes_for_level_base = options.target_file_size_base * 2; options.max_bytes_for_level_multiplier = 2; options.compression = kNoCompression; #if ROCKSDB_USING_THREAD_STATUS options.enable_thread_tracking = true; #endif // ROCKSDB_USING_THREAD_STATUS options.level0_file_num_compaction_trigger = kNumL0Files; options.table_properties_collector_factories.push_back( std::make_shared()); TestCompactionListener* listener = new TestCompactionListener(); options.listeners.emplace_back(listener); std::vector cf_names = { "pikachu", "ilya", "muromec", "dobrynia", "nikitich", "alyosha", "popovich"}; CreateAndReopenWithCF(cf_names, options); ASSERT_OK(Put(1, "pikachu", std::string(90000, 'p'))); ASSERT_OK(Put(2, "ilya", std::string(90000, 'i'))); ASSERT_OK(Put(3, "muromec", std::string(90000, 'm'))); ASSERT_OK(Put(4, "dobrynia", std::string(90000, 'd'))); ASSERT_OK(Put(5, "nikitich", std::string(90000, 'n'))); ASSERT_OK(Put(6, "alyosha", std::string(90000, 'a'))); ASSERT_OK(Put(7, "popovich", std::string(90000, 'p'))); for (int i = 1; i < 8; ++i) { ASSERT_OK(Flush(i)); const Slice kRangeStart = "a"; const Slice kRangeEnd = "z"; ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), handles_[i], &kRangeStart, &kRangeEnd)); dbfull()->TEST_WaitForFlushMemTable(); dbfull()->TEST_WaitForCompact(); } ASSERT_EQ(listener->compacted_dbs_.size(), cf_names.size()); for (size_t i = 0; i < cf_names.size(); ++i) { ASSERT_EQ(listener->compacted_dbs_[i], db_); } } // This simple Listener can only handle one flush at a time. class TestFlushListener : public EventListener { public: explicit TestFlushListener(Env* env) : slowdown_count(0), stop_count(0), db_closed(), env_(env) { db_closed = false; } void OnTableFileCreated( const TableFileCreationInfo& info) override { // remember the info for later checking the FlushJobInfo. prev_fc_info_ = info; ASSERT_GT(info.db_name.size(), 0U); ASSERT_GT(info.cf_name.size(), 0U); ASSERT_GT(info.file_path.size(), 0U); ASSERT_GT(info.job_id, 0); ASSERT_GT(info.table_properties.data_size, 0U); ASSERT_GT(info.table_properties.raw_key_size, 0U); ASSERT_GT(info.table_properties.raw_value_size, 0U); ASSERT_GT(info.table_properties.num_data_blocks, 0U); ASSERT_GT(info.table_properties.num_entries, 0U); #if ROCKSDB_USING_THREAD_STATUS // Verify the id of the current thread that created this table // file matches the id of any active flush or compaction thread. uint64_t thread_id = env_->GetThreadID(); std::vector thread_list; ASSERT_OK(env_->GetThreadList(&thread_list)); bool found_match = false; for (auto thread_status : thread_list) { if (thread_status.operation_type == ThreadStatus::OP_FLUSH || thread_status.operation_type == ThreadStatus::OP_COMPACTION) { if (thread_id == thread_status.thread_id) { found_match = true; break; } } } ASSERT_TRUE(found_match); #endif // ROCKSDB_USING_THREAD_STATUS } void OnFlushCompleted( DB* db, const FlushJobInfo& info) override { flushed_dbs_.push_back(db); flushed_column_family_names_.push_back(info.cf_name); if (info.triggered_writes_slowdown) { slowdown_count++; } if (info.triggered_writes_stop) { stop_count++; } // verify whether the previously created file matches the flushed file. ASSERT_EQ(prev_fc_info_.db_name, db->GetName()); ASSERT_EQ(prev_fc_info_.cf_name, info.cf_name); ASSERT_EQ(prev_fc_info_.job_id, info.job_id); ASSERT_EQ(prev_fc_info_.file_path, info.file_path); ASSERT_EQ(db->GetEnv()->GetThreadID(), info.thread_id); ASSERT_GT(info.thread_id, 0U); ASSERT_EQ(info.table_properties.user_collected_properties.find("0")->second, "1"); } std::vector flushed_column_family_names_; std::vector flushed_dbs_; int slowdown_count; int stop_count; bool db_closing; std::atomic_bool db_closed; TableFileCreationInfo prev_fc_info_; protected: Env* env_; }; TEST_F(EventListenerTest, OnSingleDBFlushTest) { Options options; options.write_buffer_size = k110KB; #if ROCKSDB_USING_THREAD_STATUS options.enable_thread_tracking = true; #endif // ROCKSDB_USING_THREAD_STATUS TestFlushListener* listener = new TestFlushListener(options.env); options.listeners.emplace_back(listener); std::vector cf_names = { "pikachu", "ilya", "muromec", "dobrynia", "nikitich", "alyosha", "popovich"}; options.table_properties_collector_factories.push_back( std::make_shared()); CreateAndReopenWithCF(cf_names, options); ASSERT_OK(Put(1, "pikachu", std::string(90000, 'p'))); ASSERT_OK(Put(2, "ilya", std::string(90000, 'i'))); ASSERT_OK(Put(3, "muromec", std::string(90000, 'm'))); ASSERT_OK(Put(4, "dobrynia", std::string(90000, 'd'))); ASSERT_OK(Put(5, "nikitich", std::string(90000, 'n'))); ASSERT_OK(Put(6, "alyosha", std::string(90000, 'a'))); ASSERT_OK(Put(7, "popovich", std::string(90000, 'p'))); for (int i = 1; i < 8; ++i) { ASSERT_OK(Flush(i)); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(listener->flushed_dbs_.size(), i); ASSERT_EQ(listener->flushed_column_family_names_.size(), i); } // make sure call-back functions are called in the right order for (size_t i = 0; i < cf_names.size(); ++i) { ASSERT_EQ(listener->flushed_dbs_[i], db_); ASSERT_EQ(listener->flushed_column_family_names_[i], cf_names[i]); } } TEST_F(EventListenerTest, MultiCF) { Options options; options.write_buffer_size = k110KB; #if ROCKSDB_USING_THREAD_STATUS options.enable_thread_tracking = true; #endif // ROCKSDB_USING_THREAD_STATUS TestFlushListener* listener = new TestFlushListener(options.env); options.listeners.emplace_back(listener); options.table_properties_collector_factories.push_back( std::make_shared()); std::vector cf_names = { "pikachu", "ilya", "muromec", "dobrynia", "nikitich", "alyosha", "popovich"}; CreateAndReopenWithCF(cf_names, options); ASSERT_OK(Put(1, "pikachu", std::string(90000, 'p'))); ASSERT_OK(Put(2, "ilya", std::string(90000, 'i'))); ASSERT_OK(Put(3, "muromec", std::string(90000, 'm'))); ASSERT_OK(Put(4, "dobrynia", std::string(90000, 'd'))); ASSERT_OK(Put(5, "nikitich", std::string(90000, 'n'))); ASSERT_OK(Put(6, "alyosha", std::string(90000, 'a'))); ASSERT_OK(Put(7, "popovich", std::string(90000, 'p'))); for (int i = 1; i < 8; ++i) { ASSERT_OK(Flush(i)); ASSERT_EQ(listener->flushed_dbs_.size(), i); ASSERT_EQ(listener->flushed_column_family_names_.size(), i); } // make sure call-back functions are called in the right order for (size_t i = 0; i < cf_names.size(); i++) { ASSERT_EQ(listener->flushed_dbs_[i], db_); ASSERT_EQ(listener->flushed_column_family_names_[i], cf_names[i]); } } TEST_F(EventListenerTest, MultiDBMultiListeners) { Options options; #if ROCKSDB_USING_THREAD_STATUS options.enable_thread_tracking = true; #endif // ROCKSDB_USING_THREAD_STATUS options.table_properties_collector_factories.push_back( std::make_shared()); std::vector listeners; const int kNumDBs = 5; const int kNumListeners = 10; for (int i = 0; i < kNumListeners; ++i) { listeners.emplace_back(new TestFlushListener(options.env)); } std::vector cf_names = { "pikachu", "ilya", "muromec", "dobrynia", "nikitich", "alyosha", "popovich"}; options.create_if_missing = true; for (int i = 0; i < kNumListeners; ++i) { options.listeners.emplace_back(listeners[i]); } DBOptions db_opts(options); ColumnFamilyOptions cf_opts(options); std::vector dbs; std::vector> vec_handles; for (int d = 0; d < kNumDBs; ++d) { ASSERT_OK(DestroyDB(dbname_ + ToString(d), options)); DB* db; std::vector handles; ASSERT_OK(DB::Open(options, dbname_ + ToString(d), &db)); for (size_t c = 0; c < cf_names.size(); ++c) { ColumnFamilyHandle* handle; db->CreateColumnFamily(cf_opts, cf_names[c], &handle); handles.push_back(handle); } vec_handles.push_back(std::move(handles)); dbs.push_back(db); } for (int d = 0; d < kNumDBs; ++d) { for (size_t c = 0; c < cf_names.size(); ++c) { ASSERT_OK(dbs[d]->Put(WriteOptions(), vec_handles[d][c], cf_names[c], cf_names[c])); } } for (size_t c = 0; c < cf_names.size(); ++c) { for (int d = 0; d < kNumDBs; ++d) { ASSERT_OK(dbs[d]->Flush(FlushOptions(), vec_handles[d][c])); reinterpret_cast(dbs[d])->TEST_WaitForFlushMemTable(); } } for (auto* listener : listeners) { int pos = 0; for (size_t c = 0; c < cf_names.size(); ++c) { for (int d = 0; d < kNumDBs; ++d) { ASSERT_EQ(listener->flushed_dbs_[pos], dbs[d]); ASSERT_EQ(listener->flushed_column_family_names_[pos], cf_names[c]); pos++; } } } for (auto handles : vec_handles) { for (auto h : handles) { delete h; } handles.clear(); } vec_handles.clear(); for (auto db : dbs) { delete db; } } TEST_F(EventListenerTest, DisableBGCompaction) { Options options; #if ROCKSDB_USING_THREAD_STATUS options.enable_thread_tracking = true; #endif // ROCKSDB_USING_THREAD_STATUS TestFlushListener* listener = new TestFlushListener(options.env); const int kCompactionTrigger = 1; const int kSlowdownTrigger = 5; const int kStopTrigger = 100; options.level0_file_num_compaction_trigger = kCompactionTrigger; options.level0_slowdown_writes_trigger = kSlowdownTrigger; options.level0_stop_writes_trigger = kStopTrigger; options.max_write_buffer_number = 10; options.listeners.emplace_back(listener); // BG compaction is disabled. Number of L0 files will simply keeps // increasing in this test. options.compaction_style = kCompactionStyleNone; options.compression = kNoCompression; options.write_buffer_size = 100000; // Small write buffer options.table_properties_collector_factories.push_back( std::make_shared()); CreateAndReopenWithCF({"pikachu"}, options); ColumnFamilyMetaData cf_meta; db_->GetColumnFamilyMetaData(handles_[1], &cf_meta); // keep writing until writes are forced to stop. for (int i = 0; static_cast(cf_meta.file_count) < kSlowdownTrigger * 10; ++i) { Put(1, ToString(i), std::string(10000, 'x'), WriteOptions()); db_->Flush(FlushOptions(), handles_[1]); db_->GetColumnFamilyMetaData(handles_[1], &cf_meta); } ASSERT_GE(listener->slowdown_count, kSlowdownTrigger * 9); } class TestCompactionReasonListener : public EventListener { public: void OnCompactionCompleted(DB* db, const CompactionJobInfo& ci) override { std::lock_guard lock(mutex_); compaction_reasons_.push_back(ci.compaction_reason); } std::vector compaction_reasons_; std::mutex mutex_; }; TEST_F(EventListenerTest, CompactionReasonLevel) { Options options; options.create_if_missing = true; options.memtable_factory.reset( new SpecialSkipListFactory(DBTestBase::kNumKeysByGenerateNewRandomFile)); TestCompactionReasonListener* listener = new TestCompactionReasonListener(); options.listeners.emplace_back(listener); options.level0_file_num_compaction_trigger = 4; options.compaction_style = kCompactionStyleLevel; DestroyAndReopen(options); Random rnd(301); // Write 4 files in L0 for (int i = 0; i < 4; i++) { GenerateNewRandomFile(&rnd); } dbfull()->TEST_WaitForCompact(); ASSERT_EQ(listener->compaction_reasons_.size(), 1); ASSERT_EQ(listener->compaction_reasons_[0], CompactionReason::kLevelL0FilesNum); DestroyAndReopen(options); // Write 3 non-overlapping files in L0 for (int k = 1; k <= 30; k++) { ASSERT_OK(Put(Key(k), Key(k))); if (k % 10 == 0) { Flush(); } } // Do a trivial move from L0 -> L1 db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); options.max_bytes_for_level_base = 1; Close(); listener->compaction_reasons_.clear(); Reopen(options); dbfull()->TEST_WaitForCompact(); ASSERT_GT(listener->compaction_reasons_.size(), 1); for (auto compaction_reason : listener->compaction_reasons_) { ASSERT_EQ(compaction_reason, CompactionReason::kLevelMaxLevelSize); } options.disable_auto_compactions = true; Close(); listener->compaction_reasons_.clear(); Reopen(options); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_GT(listener->compaction_reasons_.size(), 0); for (auto compaction_reason : listener->compaction_reasons_) { ASSERT_EQ(compaction_reason, CompactionReason::kManualCompaction); } } TEST_F(EventListenerTest, CompactionReasonUniversal) { Options options; options.create_if_missing = true; options.memtable_factory.reset( new SpecialSkipListFactory(DBTestBase::kNumKeysByGenerateNewRandomFile)); TestCompactionReasonListener* listener = new TestCompactionReasonListener(); options.listeners.emplace_back(listener); options.compaction_style = kCompactionStyleUniversal; Random rnd(301); options.level0_file_num_compaction_trigger = 8; options.compaction_options_universal.max_size_amplification_percent = 100000; options.compaction_options_universal.size_ratio = 100000; DestroyAndReopen(options); listener->compaction_reasons_.clear(); // Write 8 files in L0 for (int i = 0; i < 8; i++) { GenerateNewRandomFile(&rnd); } dbfull()->TEST_WaitForCompact(); ASSERT_GT(listener->compaction_reasons_.size(), 0); for (auto compaction_reason : listener->compaction_reasons_) { ASSERT_EQ(compaction_reason, CompactionReason::kUniversalSortedRunNum); } options.level0_file_num_compaction_trigger = 8; options.compaction_options_universal.max_size_amplification_percent = 1; options.compaction_options_universal.size_ratio = 100000; DestroyAndReopen(options); listener->compaction_reasons_.clear(); // Write 8 files in L0 for (int i = 0; i < 8; i++) { GenerateNewRandomFile(&rnd); } dbfull()->TEST_WaitForCompact(); ASSERT_GT(listener->compaction_reasons_.size(), 0); for (auto compaction_reason : listener->compaction_reasons_) { ASSERT_EQ(compaction_reason, CompactionReason::kUniversalSizeAmplification); } options.disable_auto_compactions = true; Close(); listener->compaction_reasons_.clear(); Reopen(options); db_->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_GT(listener->compaction_reasons_.size(), 0); for (auto compaction_reason : listener->compaction_reasons_) { ASSERT_EQ(compaction_reason, CompactionReason::kManualCompaction); } } TEST_F(EventListenerTest, CompactionReasonFIFO) { Options options; options.create_if_missing = true; options.memtable_factory.reset( new SpecialSkipListFactory(DBTestBase::kNumKeysByGenerateNewRandomFile)); TestCompactionReasonListener* listener = new TestCompactionReasonListener(); options.listeners.emplace_back(listener); options.level0_file_num_compaction_trigger = 4; options.compaction_style = kCompactionStyleFIFO; options.compaction_options_fifo.max_table_files_size = 1; DestroyAndReopen(options); Random rnd(301); // Write 4 files in L0 for (int i = 0; i < 4; i++) { GenerateNewRandomFile(&rnd); } dbfull()->TEST_WaitForCompact(); ASSERT_GT(listener->compaction_reasons_.size(), 0); for (auto compaction_reason : listener->compaction_reasons_) { ASSERT_EQ(compaction_reason, CompactionReason::kFIFOMaxSize); } } } // namespace rocksdb #endif // ROCKSDB_LITE int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/log_format.h000066400000000000000000000025631267530634000160530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Log format information shared by reader and writer. // See ../doc/log_format.txt for more detail. #pragma once namespace rocksdb { namespace log { enum RecordType { // Zero is reserved for preallocated files kZeroType = 0, kFullType = 1, // For fragments kFirstType = 2, kMiddleType = 3, kLastType = 4, // For recycled log files kRecyclableFullType = 5, kRecyclableFirstType = 6, kRecyclableMiddleType = 7, kRecyclableLastType = 8, }; static const int kMaxRecordType = kRecyclableLastType; static const unsigned int kBlockSize = 32768; // Header is checksum (4 bytes), type (1 byte), length (2 bytes). static const int kHeaderSize = 4 + 1 + 2; // Recyclable header is checksum (4 bytes), type (1 byte), log number // (4 bytes), length (2 bytes). static const int kRecyclableHeaderSize = 4 + 1 + 4 + 2; } // namespace log } // namespace rocksdb rocksdb-4.5.1/db/log_reader.cc000066400000000000000000000323521267530634000161620ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/log_reader.h" #include #include "rocksdb/env.h" #include "util/coding.h" #include "util/crc32c.h" #include "util/file_reader_writer.h" namespace rocksdb { namespace log { Reader::Reporter::~Reporter() { } Reader::Reader(std::shared_ptr info_log, unique_ptr&& _file, Reporter* reporter, bool checksum, uint64_t initial_offset, uint64_t log_num) : info_log_(info_log), file_(std::move(_file)), reporter_(reporter), checksum_(checksum), backing_store_(new char[kBlockSize]), buffer_(), eof_(false), read_error_(false), eof_offset_(0), last_record_offset_(0), end_of_buffer_offset_(0), initial_offset_(initial_offset), log_number_(log_num) {} Reader::~Reader() { delete[] backing_store_; } bool Reader::SkipToInitialBlock() { size_t initial_offset_in_block = initial_offset_ % kBlockSize; uint64_t block_start_location = initial_offset_ - initial_offset_in_block; // Don't search a block if we'd be in the trailer if (initial_offset_in_block > kBlockSize - 6) { block_start_location += kBlockSize; } end_of_buffer_offset_ = block_start_location; // Skip to start of first block that can contain the initial record if (block_start_location > 0) { Status skip_status = file_->Skip(block_start_location); if (!skip_status.ok()) { ReportDrop(static_cast(block_start_location), skip_status); return false; } } return true; } // For kAbsoluteConsistency, on clean shutdown we don't expect any error // in the log files. For other modes, we can ignore only incomplete records // in the last log file, which are presumably due to a write in progress // during restart (or from log recycling). // // TODO krad: Evaluate if we need to move to a more strict mode where we // restrict the inconsistency to only the last log bool Reader::ReadRecord(Slice* record, std::string* scratch, WALRecoveryMode wal_recovery_mode) { if (last_record_offset_ < initial_offset_) { if (!SkipToInitialBlock()) { return false; } } scratch->clear(); record->clear(); bool in_fragmented_record = false; // Record offset of the logical record that we're reading // 0 is a dummy value to make compilers happy uint64_t prospective_record_offset = 0; Slice fragment; while (true) { uint64_t physical_record_offset = end_of_buffer_offset_ - buffer_.size(); size_t drop_size; const unsigned int record_type = ReadPhysicalRecord(&fragment, &drop_size); switch (record_type) { case kFullType: case kRecyclableFullType: if (in_fragmented_record && !scratch->empty()) { // Handle bug in earlier versions of log::Writer where // it could emit an empty kFirstType record at the tail end // of a block followed by a kFullType or kFirstType record // at the beginning of the next block. ReportCorruption(scratch->size(), "partial record without end(1)"); } prospective_record_offset = physical_record_offset; scratch->clear(); *record = fragment; last_record_offset_ = prospective_record_offset; return true; case kFirstType: case kRecyclableFirstType: if (in_fragmented_record && !scratch->empty()) { // Handle bug in earlier versions of log::Writer where // it could emit an empty kFirstType record at the tail end // of a block followed by a kFullType or kFirstType record // at the beginning of the next block. ReportCorruption(scratch->size(), "partial record without end(2)"); } prospective_record_offset = physical_record_offset; scratch->assign(fragment.data(), fragment.size()); in_fragmented_record = true; break; case kMiddleType: case kRecyclableMiddleType: if (!in_fragmented_record) { ReportCorruption(fragment.size(), "missing start of fragmented record(1)"); } else { scratch->append(fragment.data(), fragment.size()); } break; case kLastType: case kRecyclableLastType: if (!in_fragmented_record) { ReportCorruption(fragment.size(), "missing start of fragmented record(2)"); } else { scratch->append(fragment.data(), fragment.size()); *record = Slice(*scratch); last_record_offset_ = prospective_record_offset; return true; } break; case kBadHeader: if (wal_recovery_mode == WALRecoveryMode::kAbsoluteConsistency) { // in clean shutdown we don't expect any error in the log files ReportCorruption(drop_size, "truncated header"); } // fall-thru case kEof: if (in_fragmented_record) { if (wal_recovery_mode == WALRecoveryMode::kAbsoluteConsistency) { // in clean shutdown we don't expect any error in the log files ReportCorruption(scratch->size(), "error reading trailing data"); } // This can be caused by the writer dying immediately after // writing a physical record but before completing the next; don't // treat it as a corruption, just ignore the entire logical record. scratch->clear(); } return false; case kOldRecord: if (wal_recovery_mode != WALRecoveryMode::kSkipAnyCorruptedRecords) { // Treat a record from a previous instance of the log as EOF. if (in_fragmented_record) { if (wal_recovery_mode == WALRecoveryMode::kAbsoluteConsistency) { // in clean shutdown we don't expect any error in the log files ReportCorruption(scratch->size(), "error reading trailing data"); } // This can be caused by the writer dying immediately after // writing a physical record but before completing the next; don't // treat it as a corruption, just ignore the entire logical record. scratch->clear(); } return false; } // fall-thru case kBadRecord: if (in_fragmented_record) { ReportCorruption(scratch->size(), "error in middle of record"); in_fragmented_record = false; scratch->clear(); } break; default: { char buf[40]; snprintf(buf, sizeof(buf), "unknown record type %u", record_type); ReportCorruption( (fragment.size() + (in_fragmented_record ? scratch->size() : 0)), buf); in_fragmented_record = false; scratch->clear(); break; } } } return false; } uint64_t Reader::LastRecordOffset() { return last_record_offset_; } void Reader::UnmarkEOF() { if (read_error_) { return; } eof_ = false; if (eof_offset_ == 0) { return; } // If the EOF was in the middle of a block (a partial block was read) we have // to read the rest of the block as ReadPhysicalRecord can only read full // blocks and expects the file position indicator to be aligned to the start // of a block. // // consumed_bytes + buffer_size() + remaining == kBlockSize size_t consumed_bytes = eof_offset_ - buffer_.size(); size_t remaining = kBlockSize - eof_offset_; // backing_store_ is used to concatenate what is left in buffer_ and // the remainder of the block. If buffer_ already uses backing_store_, // we just append the new data. if (buffer_.data() != backing_store_ + consumed_bytes) { // Buffer_ does not use backing_store_ for storage. // Copy what is left in buffer_ to backing_store. memmove(backing_store_ + consumed_bytes, buffer_.data(), buffer_.size()); } Slice read_buffer; Status status = file_->Read(remaining, &read_buffer, backing_store_ + eof_offset_); size_t added = read_buffer.size(); end_of_buffer_offset_ += added; if (!status.ok()) { if (added > 0) { ReportDrop(added, status); } read_error_ = true; return; } if (read_buffer.data() != backing_store_ + eof_offset_) { // Read did not write to backing_store_ memmove(backing_store_ + eof_offset_, read_buffer.data(), read_buffer.size()); } buffer_ = Slice(backing_store_ + consumed_bytes, eof_offset_ + added - consumed_bytes); if (added < remaining) { eof_ = true; eof_offset_ += added; } else { eof_offset_ = 0; } } void Reader::ReportCorruption(size_t bytes, const char* reason) { ReportDrop(bytes, Status::Corruption(reason)); } void Reader::ReportDrop(size_t bytes, const Status& reason) { if (reporter_ != nullptr && end_of_buffer_offset_ - buffer_.size() - bytes >= initial_offset_) { reporter_->Corruption(bytes, reason); } } bool Reader::ReadMore(size_t* drop_size, int *error) { if (!eof_ && !read_error_) { // Last read was a full read, so this is a trailer to skip buffer_.clear(); Status status = file_->Read(kBlockSize, &buffer_, backing_store_); end_of_buffer_offset_ += buffer_.size(); if (!status.ok()) { buffer_.clear(); ReportDrop(kBlockSize, status); read_error_ = true; *error = kEof; return false; } else if (buffer_.size() < (size_t)kBlockSize) { eof_ = true; eof_offset_ = buffer_.size(); } return true; } else { // Note that if buffer_ is non-empty, we have a truncated header at the // end of the file, which can be caused by the writer crashing in the // middle of writing the header. Unless explicitly requested we don't // considering this an error, just report EOF. if (buffer_.size()) { *drop_size = buffer_.size(); buffer_.clear(); *error = kBadHeader; return false; } buffer_.clear(); *error = kEof; return false; } } unsigned int Reader::ReadPhysicalRecord(Slice* result, size_t* drop_size) { while (true) { // We need at least the minimum header size if (buffer_.size() < (size_t)kHeaderSize) { int r; if (!ReadMore(drop_size, &r)) { return r; } continue; } // Parse the header const char* header = buffer_.data(); const uint32_t a = static_cast(header[4]) & 0xff; const uint32_t b = static_cast(header[5]) & 0xff; const unsigned int type = header[6]; const uint32_t length = a | (b << 8); int header_size = kHeaderSize; if (type >= kRecyclableFullType && type <= kRecyclableLastType) { header_size = kRecyclableHeaderSize; // We need enough for the larger header if (buffer_.size() < (size_t)kRecyclableHeaderSize) { int r; if (!ReadMore(drop_size, &r)) { return r; } continue; } const uint32_t log_num = DecodeFixed32(header + 7); if (log_num != log_number_) { return kOldRecord; } } if (header_size + length > buffer_.size()) { *drop_size = buffer_.size(); buffer_.clear(); if (!eof_) { ReportCorruption(*drop_size, "bad record length"); return kBadRecord; } // If the end of the file has been reached without reading |length| bytes // of payload, assume the writer died in the middle of writing the record. // Don't report a corruption unless requested. if (*drop_size) { return kBadHeader; } return kEof; } if (type == kZeroType && length == 0) { // Skip zero length record without reporting any drops since // such records are produced by the mmap based writing code in // env_posix.cc that preallocates file regions. // NOTE: this should never happen in DB written by new RocksDB versions, // since we turn off mmap writes to manifest and log files buffer_.clear(); return kBadRecord; } // Check crc if (checksum_) { uint32_t expected_crc = crc32c::Unmask(DecodeFixed32(header)); uint32_t actual_crc = crc32c::Value(header + 6, length + header_size - 6); if (actual_crc != expected_crc) { // Drop the rest of the buffer since "length" itself may have // been corrupted and if we trust it, we could find some // fragment of a real log record that just happens to look // like a valid log record. *drop_size = buffer_.size(); buffer_.clear(); ReportCorruption(*drop_size, "checksum mismatch"); return kBadRecord; } } buffer_.remove_prefix(header_size + length); // Skip physical record that started before initial_offset_ if (end_of_buffer_offset_ - buffer_.size() - header_size - length < initial_offset_) { result->clear(); return kBadRecord; } *result = Slice(header + header_size, length); return type; } } } // namespace log } // namespace rocksdb rocksdb-4.5.1/db/log_reader.h000066400000000000000000000124041267530634000160200ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include "db/log_format.h" #include "rocksdb/slice.h" #include "rocksdb/status.h" #include "rocksdb/options.h" namespace rocksdb { class SequentialFileReader; class Logger; using std::unique_ptr; namespace log { /** * Reader is a general purpose log stream reader implementation. The actual job * of reading from the device is implemented by the SequentialFile interface. * * Please see Writer for details on the file and record layout. */ class Reader { public: // Interface for reporting errors. class Reporter { public: virtual ~Reporter(); // Some corruption was detected. "size" is the approximate number // of bytes dropped due to the corruption. virtual void Corruption(size_t bytes, const Status& status) = 0; }; // Create a reader that will return log records from "*file". // "*file" must remain live while this Reader is in use. // // If "reporter" is non-nullptr, it is notified whenever some data is // dropped due to a detected corruption. "*reporter" must remain // live while this Reader is in use. // // If "checksum" is true, verify checksums if available. // // The Reader will start reading at the first record located at physical // position >= initial_offset within the file. Reader(std::shared_ptr info_log, unique_ptr&& file, Reporter* reporter, bool checksum, uint64_t initial_offset, uint64_t log_num); ~Reader(); // Read the next record into *record. Returns true if read // successfully, false if we hit end of the input. May use // "*scratch" as temporary storage. The contents filled in *record // will only be valid until the next mutating operation on this // reader or the next mutation to *scratch. bool ReadRecord(Slice* record, std::string* scratch, WALRecoveryMode wal_recovery_mode = WALRecoveryMode::kTolerateCorruptedTailRecords); // Returns the physical offset of the last record returned by ReadRecord. // // Undefined before the first call to ReadRecord. uint64_t LastRecordOffset(); // returns true if the reader has encountered an eof condition. bool IsEOF() { return eof_; } // when we know more data has been written to the file. we can use this // function to force the reader to look again in the file. // Also aligns the file position indicator to the start of the next block // by reading the rest of the data from the EOF position to the end of the // block that was partially read. void UnmarkEOF(); SequentialFileReader* file() { return file_.get(); } private: std::shared_ptr info_log_; const unique_ptr file_; Reporter* const reporter_; bool const checksum_; char* const backing_store_; Slice buffer_; bool eof_; // Last Read() indicated EOF by returning < kBlockSize bool read_error_; // Error occurred while reading from file // Offset of the file position indicator within the last block when an // EOF was detected. size_t eof_offset_; // Offset of the last record returned by ReadRecord. uint64_t last_record_offset_; // Offset of the first location past the end of buffer_. uint64_t end_of_buffer_offset_; // Offset at which to start looking for the first record to return uint64_t const initial_offset_; // which log number this is uint64_t const log_number_; // Extend record types with the following special values enum { kEof = kMaxRecordType + 1, // Returned whenever we find an invalid physical record. // Currently there are three situations in which this happens: // * The record has an invalid CRC (ReadPhysicalRecord reports a drop) // * The record is a 0-length record (No drop is reported) // * The record is below constructor's initial_offset (No drop is reported) kBadRecord = kMaxRecordType + 2, // Returned when we fail to read a valid header. kBadHeader = kMaxRecordType + 3, // Returned when we read an old record from a previous user of the log. kOldRecord = kMaxRecordType + 4, }; // Skips all blocks that are completely before "initial_offset_". // // Returns true on success. Handles reporting. bool SkipToInitialBlock(); // Return type, or one of the preceding special values unsigned int ReadPhysicalRecord(Slice* result, size_t* drop_size); // Read some more bool ReadMore(size_t* drop_size, int *error); // Reports dropped bytes to the reporter. // buffer_ must be updated to remove the dropped bytes prior to invocation. void ReportCorruption(size_t bytes, const char* reason); void ReportDrop(size_t bytes, const Status& reason); // No copying allowed Reader(const Reader&); void operator=(const Reader&); }; } // namespace log } // namespace rocksdb rocksdb-4.5.1/db/log_test.cc000066400000000000000000000511321267530634000156740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/log_reader.h" #include "db/log_writer.h" #include "rocksdb/env.h" #include "util/coding.h" #include "util/crc32c.h" #include "util/file_reader_writer.h" #include "util/random.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { namespace log { // Construct a string of the specified length made out of the supplied // partial string. static std::string BigString(const std::string& partial_string, size_t n) { std::string result; while (result.size() < n) { result.append(partial_string); } result.resize(n); return result; } // Construct a string from a number static std::string NumberString(int n) { char buf[50]; snprintf(buf, sizeof(buf), "%d.", n); return std::string(buf); } // Return a skewed potentially long string static std::string RandomSkewedString(int i, Random* rnd) { return BigString(NumberString(i), rnd->Skewed(17)); } class LogTest : public ::testing::TestWithParam { private: class StringSource : public SequentialFile { public: Slice& contents_; bool force_error_; size_t force_error_position_; bool force_eof_; size_t force_eof_position_; bool returned_partial_; explicit StringSource(Slice& contents) : contents_(contents), force_error_(false), force_error_position_(0), force_eof_(false), force_eof_position_(0), returned_partial_(false) { } virtual Status Read(size_t n, Slice* result, char* scratch) override { EXPECT_TRUE(!returned_partial_) << "must not Read() after eof/error"; if (force_error_) { if (force_error_position_ >= n) { force_error_position_ -= n; } else { *result = Slice(contents_.data(), force_error_position_); contents_.remove_prefix(force_error_position_); force_error_ = false; returned_partial_ = true; return Status::Corruption("read error"); } } if (contents_.size() < n) { n = contents_.size(); returned_partial_ = true; } if (force_eof_) { if (force_eof_position_ >= n) { force_eof_position_ -= n; } else { force_eof_ = false; n = force_eof_position_; returned_partial_ = true; } } // By using scratch we ensure that caller has control over the // lifetime of result.data() memcpy(scratch, contents_.data(), n); *result = Slice(scratch, n); contents_.remove_prefix(n); return Status::OK(); } virtual Status Skip(uint64_t n) override { if (n > contents_.size()) { contents_.clear(); return Status::NotFound("in-memory file skipepd past end"); } contents_.remove_prefix(n); return Status::OK(); } }; class ReportCollector : public Reader::Reporter { public: size_t dropped_bytes_; std::string message_; ReportCollector() : dropped_bytes_(0) { } virtual void Corruption(size_t bytes, const Status& status) override { dropped_bytes_ += bytes; message_.append(status.ToString()); } }; std::string& dest_contents() { auto dest = dynamic_cast(writer_.file()->writable_file()); assert(dest); return dest->contents_; } const std::string& dest_contents() const { auto dest = dynamic_cast(writer_.file()->writable_file()); assert(dest); return dest->contents_; } void reset_source_contents() { auto src = dynamic_cast(reader_.file()->file()); assert(src); src->contents_ = dest_contents(); } Slice reader_contents_; unique_ptr dest_holder_; unique_ptr source_holder_; ReportCollector report_; Writer writer_; Reader reader_; // Record metadata for testing initial offset functionality static size_t initial_offset_record_sizes_[]; uint64_t initial_offset_last_record_offsets_[4]; public: LogTest() : reader_contents_(), dest_holder_(test::GetWritableFileWriter( new test::StringSink(&reader_contents_))), source_holder_( test::GetSequentialFileReader(new StringSource(reader_contents_))), writer_(std::move(dest_holder_), 123, GetParam()), reader_(NULL, std::move(source_holder_), &report_, true /*checksum*/, 0 /*initial_offset*/, 123) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; initial_offset_last_record_offsets_[0] = 0; initial_offset_last_record_offsets_[1] = header_size + 10000; initial_offset_last_record_offsets_[2] = 2 * (header_size + 10000); initial_offset_last_record_offsets_[3] = 2 * (header_size + 10000) + (2 * log::kBlockSize - 1000) + 3 * header_size; } void Write(const std::string& msg) { writer_.AddRecord(Slice(msg)); } size_t WrittenBytes() const { return dest_contents().size(); } std::string Read(const WALRecoveryMode wal_recovery_mode = WALRecoveryMode::kTolerateCorruptedTailRecords) { std::string scratch; Slice record; if (reader_.ReadRecord(&record, &scratch, wal_recovery_mode)) { return record.ToString(); } else { return "EOF"; } } void IncrementByte(int offset, int delta) { dest_contents()[offset] += delta; } void SetByte(int offset, char new_byte) { dest_contents()[offset] = new_byte; } void ShrinkSize(int bytes) { auto dest = dynamic_cast(writer_.file()->writable_file()); assert(dest); dest->Drop(bytes); } void FixChecksum(int header_offset, int len, bool recyclable) { // Compute crc of type/len/data int header_size = recyclable ? kRecyclableHeaderSize : kHeaderSize; uint32_t crc = crc32c::Value(&dest_contents()[header_offset + 6], header_size - 6 + len); crc = crc32c::Mask(crc); EncodeFixed32(&dest_contents()[header_offset], crc); } void ForceError(size_t position = 0) { auto src = dynamic_cast(reader_.file()->file()); src->force_error_ = true; src->force_error_position_ = position; } size_t DroppedBytes() const { return report_.dropped_bytes_; } std::string ReportMessage() const { return report_.message_; } void ForceEOF(size_t position = 0) { auto src = dynamic_cast(reader_.file()->file()); src->force_eof_ = true; src->force_eof_position_ = position; } void UnmarkEOF() { auto src = dynamic_cast(reader_.file()->file()); src->returned_partial_ = false; reader_.UnmarkEOF(); } bool IsEOF() { return reader_.IsEOF(); } // Returns OK iff recorded error message contains "msg" std::string MatchError(const std::string& msg) const { if (report_.message_.find(msg) == std::string::npos) { return report_.message_; } else { return "OK"; } } void WriteInitialOffsetLog() { for (int i = 0; i < 4; i++) { std::string record(initial_offset_record_sizes_[i], static_cast('a' + i)); Write(record); } } void CheckOffsetPastEndReturnsNoRecords(uint64_t offset_past_end) { WriteInitialOffsetLog(); unique_ptr file_reader( test::GetSequentialFileReader(new StringSource(reader_contents_))); unique_ptr offset_reader( new Reader(NULL, std::move(file_reader), &report_, true /*checksum*/, WrittenBytes() + offset_past_end, 123)); Slice record; std::string scratch; ASSERT_TRUE(!offset_reader->ReadRecord(&record, &scratch)); } void CheckInitialOffsetRecord(uint64_t initial_offset, int expected_record_offset) { WriteInitialOffsetLog(); unique_ptr file_reader( test::GetSequentialFileReader(new StringSource(reader_contents_))); unique_ptr offset_reader( new Reader(NULL, std::move(file_reader), &report_, true /*checksum*/, initial_offset, 123)); Slice record; std::string scratch; ASSERT_TRUE(offset_reader->ReadRecord(&record, &scratch)); ASSERT_EQ(initial_offset_record_sizes_[expected_record_offset], record.size()); ASSERT_EQ(initial_offset_last_record_offsets_[expected_record_offset], offset_reader->LastRecordOffset()); ASSERT_EQ((char)('a' + expected_record_offset), record.data()[0]); } }; size_t LogTest::initial_offset_record_sizes_[] = {10000, // Two sizable records in first block 10000, 2 * log::kBlockSize - 1000, // Span three blocks 1}; TEST_P(LogTest, Empty) { ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, ReadWrite) { Write("foo"); Write("bar"); Write(""); Write("xxxx"); ASSERT_EQ("foo", Read()); ASSERT_EQ("bar", Read()); ASSERT_EQ("", Read()); ASSERT_EQ("xxxx", Read()); ASSERT_EQ("EOF", Read()); ASSERT_EQ("EOF", Read()); // Make sure reads at eof work } TEST_P(LogTest, ManyBlocks) { for (int i = 0; i < 100000; i++) { Write(NumberString(i)); } for (int i = 0; i < 100000; i++) { ASSERT_EQ(NumberString(i), Read()); } ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, Fragmentation) { Write("small"); Write(BigString("medium", 50000)); Write(BigString("large", 100000)); ASSERT_EQ("small", Read()); ASSERT_EQ(BigString("medium", 50000), Read()); ASSERT_EQ(BigString("large", 100000), Read()); ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, MarginalTrailer) { // Make a trailer that is exactly the same length as an empty record. int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; const int n = kBlockSize - 2 * header_size; Write(BigString("foo", n)); ASSERT_EQ((unsigned int)(kBlockSize - header_size), WrittenBytes()); Write(""); Write("bar"); ASSERT_EQ(BigString("foo", n), Read()); ASSERT_EQ("", Read()); ASSERT_EQ("bar", Read()); ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, MarginalTrailer2) { // Make a trailer that is exactly the same length as an empty record. int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; const int n = kBlockSize - 2 * header_size; Write(BigString("foo", n)); ASSERT_EQ((unsigned int)(kBlockSize - header_size), WrittenBytes()); Write("bar"); ASSERT_EQ(BigString("foo", n), Read()); ASSERT_EQ("bar", Read()); ASSERT_EQ("EOF", Read()); ASSERT_EQ(0U, DroppedBytes()); ASSERT_EQ("", ReportMessage()); } TEST_P(LogTest, ShortTrailer) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; const int n = kBlockSize - 2 * header_size + 4; Write(BigString("foo", n)); ASSERT_EQ((unsigned int)(kBlockSize - header_size + 4), WrittenBytes()); Write(""); Write("bar"); ASSERT_EQ(BigString("foo", n), Read()); ASSERT_EQ("", Read()); ASSERT_EQ("bar", Read()); ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, AlignedEof) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; const int n = kBlockSize - 2 * header_size + 4; Write(BigString("foo", n)); ASSERT_EQ((unsigned int)(kBlockSize - header_size + 4), WrittenBytes()); ASSERT_EQ(BigString("foo", n), Read()); ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, RandomRead) { const int N = 500; Random write_rnd(301); for (int i = 0; i < N; i++) { Write(RandomSkewedString(i, &write_rnd)); } Random read_rnd(301); for (int i = 0; i < N; i++) { ASSERT_EQ(RandomSkewedString(i, &read_rnd), Read()); } ASSERT_EQ("EOF", Read()); } // Tests of all the error paths in log_reader.cc follow: TEST_P(LogTest, ReadError) { Write("foo"); ForceError(); ASSERT_EQ("EOF", Read()); ASSERT_EQ((unsigned int)kBlockSize, DroppedBytes()); ASSERT_EQ("OK", MatchError("read error")); } TEST_P(LogTest, BadRecordType) { Write("foo"); // Type is stored in header[6] IncrementByte(6, 100); FixChecksum(0, 3, false); ASSERT_EQ("EOF", Read()); ASSERT_EQ(3U, DroppedBytes()); ASSERT_EQ("OK", MatchError("unknown record type")); } TEST_P(LogTest, TruncatedTrailingRecordIsIgnored) { Write("foo"); ShrinkSize(4); // Drop all payload as well as a header byte ASSERT_EQ("EOF", Read()); // Truncated last record is ignored, not treated as an error ASSERT_EQ(0U, DroppedBytes()); ASSERT_EQ("", ReportMessage()); } TEST_P(LogTest, TruncatedTrailingRecordIsNotIgnored) { Write("foo"); ShrinkSize(4); // Drop all payload as well as a header byte ASSERT_EQ("EOF", Read(WALRecoveryMode::kAbsoluteConsistency)); // Truncated last record is ignored, not treated as an error ASSERT_GT(DroppedBytes(), 0U); ASSERT_EQ("OK", MatchError("Corruption: truncated header")); } TEST_P(LogTest, BadLength) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; const int kPayloadSize = kBlockSize - header_size; Write(BigString("bar", kPayloadSize)); Write("foo"); // Least significant size byte is stored in header[4]. IncrementByte(4, 1); ASSERT_EQ("foo", Read()); ASSERT_EQ(kBlockSize, DroppedBytes()); ASSERT_EQ("OK", MatchError("bad record length")); } TEST_P(LogTest, BadLengthAtEndIsIgnored) { Write("foo"); ShrinkSize(1); ASSERT_EQ("EOF", Read()); ASSERT_EQ(0U, DroppedBytes()); ASSERT_EQ("", ReportMessage()); } TEST_P(LogTest, BadLengthAtEndIsNotIgnored) { Write("foo"); ShrinkSize(1); ASSERT_EQ("EOF", Read(WALRecoveryMode::kAbsoluteConsistency)); ASSERT_GT(DroppedBytes(), 0U); ASSERT_EQ("OK", MatchError("Corruption: truncated header")); } TEST_P(LogTest, ChecksumMismatch) { Write("foooooo"); IncrementByte(0, 14); ASSERT_EQ("EOF", Read()); ASSERT_EQ(14U + 4 * !!GetParam(), DroppedBytes()); ASSERT_EQ("OK", MatchError("checksum mismatch")); } TEST_P(LogTest, UnexpectedMiddleType) { Write("foo"); SetByte(6, GetParam() ? kRecyclableMiddleType : kMiddleType); FixChecksum(0, 3, !!GetParam()); ASSERT_EQ("EOF", Read()); ASSERT_EQ(3U, DroppedBytes()); ASSERT_EQ("OK", MatchError("missing start")); } TEST_P(LogTest, UnexpectedLastType) { Write("foo"); SetByte(6, GetParam() ? kRecyclableLastType : kLastType); FixChecksum(0, 3, !!GetParam()); ASSERT_EQ("EOF", Read()); ASSERT_EQ(3U, DroppedBytes()); ASSERT_EQ("OK", MatchError("missing start")); } TEST_P(LogTest, UnexpectedFullType) { Write("foo"); Write("bar"); SetByte(6, GetParam() ? kRecyclableFirstType : kFirstType); FixChecksum(0, 3, !!GetParam()); ASSERT_EQ("bar", Read()); ASSERT_EQ("EOF", Read()); ASSERT_EQ(3U, DroppedBytes()); ASSERT_EQ("OK", MatchError("partial record without end")); } TEST_P(LogTest, UnexpectedFirstType) { Write("foo"); Write(BigString("bar", 100000)); SetByte(6, GetParam() ? kRecyclableFirstType : kFirstType); FixChecksum(0, 3, !!GetParam()); ASSERT_EQ(BigString("bar", 100000), Read()); ASSERT_EQ("EOF", Read()); ASSERT_EQ(3U, DroppedBytes()); ASSERT_EQ("OK", MatchError("partial record without end")); } TEST_P(LogTest, MissingLastIsIgnored) { Write(BigString("bar", kBlockSize)); // Remove the LAST block, including header. ShrinkSize(14); ASSERT_EQ("EOF", Read()); ASSERT_EQ("", ReportMessage()); ASSERT_EQ(0U, DroppedBytes()); } TEST_P(LogTest, MissingLastIsNotIgnored) { Write(BigString("bar", kBlockSize)); // Remove the LAST block, including header. ShrinkSize(14); ASSERT_EQ("EOF", Read(WALRecoveryMode::kAbsoluteConsistency)); ASSERT_GT(DroppedBytes(), 0U); ASSERT_EQ("OK", MatchError("Corruption: error reading trailing data")); } TEST_P(LogTest, PartialLastIsIgnored) { Write(BigString("bar", kBlockSize)); // Cause a bad record length in the LAST block. ShrinkSize(1); ASSERT_EQ("EOF", Read()); ASSERT_EQ("", ReportMessage()); ASSERT_EQ(0U, DroppedBytes()); } TEST_P(LogTest, PartialLastIsNotIgnored) { Write(BigString("bar", kBlockSize)); // Cause a bad record length in the LAST block. ShrinkSize(1); ASSERT_EQ("EOF", Read(WALRecoveryMode::kAbsoluteConsistency)); ASSERT_GT(DroppedBytes(), 0U); ASSERT_EQ("OK", MatchError( "Corruption: truncated headerCorruption: " "error reading trailing data")); } TEST_P(LogTest, ErrorJoinsRecords) { // Consider two fragmented records: // first(R1) last(R1) first(R2) last(R2) // where the middle two fragments disappear. We do not want // first(R1),last(R2) to get joined and returned as a valid record. // Write records that span two blocks Write(BigString("foo", kBlockSize)); Write(BigString("bar", kBlockSize)); Write("correct"); // Wipe the middle block for (unsigned int offset = kBlockSize; offset < 2*kBlockSize; offset++) { SetByte(offset, 'x'); } ASSERT_EQ("correct", Read()); ASSERT_EQ("EOF", Read()); size_t dropped = DroppedBytes(); ASSERT_LE(dropped, 2 * kBlockSize + 100); ASSERT_GE(dropped, 2 * kBlockSize); } TEST_P(LogTest, ReadStart) { CheckInitialOffsetRecord(0, 0); } TEST_P(LogTest, ReadSecondOneOff) { CheckInitialOffsetRecord(1, 1); } TEST_P(LogTest, ReadSecondTenThousand) { CheckInitialOffsetRecord(10000, 1); } TEST_P(LogTest, ReadSecondStart) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; CheckInitialOffsetRecord(10000 + header_size, 1); } TEST_P(LogTest, ReadThirdOneOff) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; CheckInitialOffsetRecord(10000 + header_size + 1, 2); } TEST_P(LogTest, ReadThirdStart) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; CheckInitialOffsetRecord(20000 + 2 * header_size, 2); } TEST_P(LogTest, ReadFourthOneOff) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; CheckInitialOffsetRecord(20000 + 2 * header_size + 1, 3); } TEST_P(LogTest, ReadFourthFirstBlockTrailer) { CheckInitialOffsetRecord(log::kBlockSize - 4, 3); } TEST_P(LogTest, ReadFourthMiddleBlock) { CheckInitialOffsetRecord(log::kBlockSize + 1, 3); } TEST_P(LogTest, ReadFourthLastBlock) { CheckInitialOffsetRecord(2 * log::kBlockSize + 1, 3); } TEST_P(LogTest, ReadFourthStart) { int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; CheckInitialOffsetRecord( 2 * (header_size + 1000) + (2 * log::kBlockSize - 1000) + 3 * header_size, 3); } TEST_P(LogTest, ReadEnd) { CheckOffsetPastEndReturnsNoRecords(0); } TEST_P(LogTest, ReadPastEnd) { CheckOffsetPastEndReturnsNoRecords(5); } TEST_P(LogTest, ClearEofSingleBlock) { Write("foo"); Write("bar"); int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; ForceEOF(3 + header_size + 2); ASSERT_EQ("foo", Read()); UnmarkEOF(); ASSERT_EQ("bar", Read()); ASSERT_TRUE(IsEOF()); ASSERT_EQ("EOF", Read()); Write("xxx"); UnmarkEOF(); ASSERT_EQ("xxx", Read()); ASSERT_TRUE(IsEOF()); } TEST_P(LogTest, ClearEofMultiBlock) { size_t num_full_blocks = 5; int header_size = GetParam() ? kRecyclableHeaderSize : kHeaderSize; size_t n = (kBlockSize - header_size) * num_full_blocks + 25; Write(BigString("foo", n)); Write(BigString("bar", n)); ForceEOF(n + num_full_blocks * header_size + header_size + 3); ASSERT_EQ(BigString("foo", n), Read()); ASSERT_TRUE(IsEOF()); UnmarkEOF(); ASSERT_EQ(BigString("bar", n), Read()); ASSERT_TRUE(IsEOF()); Write(BigString("xxx", n)); UnmarkEOF(); ASSERT_EQ(BigString("xxx", n), Read()); ASSERT_TRUE(IsEOF()); } TEST_P(LogTest, ClearEofError) { // If an error occurs during Read() in UnmarkEOF(), the records contained // in the buffer should be returned on subsequent calls of ReadRecord() // until no more full records are left, whereafter ReadRecord() should return // false to indicate that it cannot read any further. Write("foo"); Write("bar"); UnmarkEOF(); ASSERT_EQ("foo", Read()); ASSERT_TRUE(IsEOF()); Write("xxx"); ForceError(0); UnmarkEOF(); ASSERT_EQ("bar", Read()); ASSERT_EQ("EOF", Read()); } TEST_P(LogTest, ClearEofError2) { Write("foo"); Write("bar"); UnmarkEOF(); ASSERT_EQ("foo", Read()); Write("xxx"); ForceError(3); UnmarkEOF(); ASSERT_EQ("bar", Read()); ASSERT_EQ("EOF", Read()); ASSERT_EQ(3U, DroppedBytes()); ASSERT_EQ("OK", MatchError("read error")); } INSTANTIATE_TEST_CASE_P(bool, LogTest, ::testing::Values(0, 2)); } // namespace log } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/log_writer.cc000066400000000000000000000105611267530634000162320ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/log_writer.h" #include #include "rocksdb/env.h" #include "util/coding.h" #include "util/crc32c.h" #include "util/file_reader_writer.h" namespace rocksdb { namespace log { Writer::Writer(unique_ptr&& dest, uint64_t log_number, bool recycle_log_files) : dest_(std::move(dest)), block_offset_(0), log_number_(log_number), recycle_log_files_(recycle_log_files) { for (int i = 0; i <= kMaxRecordType; i++) { char t = static_cast(i); type_crc_[i] = crc32c::Value(&t, 1); } } Writer::~Writer() { } Status Writer::AddRecord(const Slice& slice) { const char* ptr = slice.data(); size_t left = slice.size(); // Header size varies depending on whether we are recycling or not. const int header_size = recycle_log_files_ ? kRecyclableHeaderSize : kHeaderSize; // Fragment the record if necessary and emit it. Note that if slice // is empty, we still want to iterate once to emit a single // zero-length record Status s; bool begin = true; do { const int64_t leftover = kBlockSize - block_offset_; assert(leftover >= 0); if (leftover < header_size) { // Switch to a new block if (leftover > 0) { // Fill the trailer (literal below relies on kHeaderSize and // kRecyclableHeaderSize being <= 11) assert(header_size <= 11); dest_->Append( Slice("\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00", leftover)); } block_offset_ = 0; } // Invariant: we never leave < header_size bytes in a block. assert(static_cast(kBlockSize - block_offset_) >= header_size); const size_t avail = kBlockSize - block_offset_ - header_size; const size_t fragment_length = (left < avail) ? left : avail; RecordType type; const bool end = (left == fragment_length); if (begin && end) { type = recycle_log_files_ ? kRecyclableFullType : kFullType; } else if (begin) { type = recycle_log_files_ ? kRecyclableFirstType : kFirstType; } else if (end) { type = recycle_log_files_ ? kRecyclableLastType : kLastType; } else { type = recycle_log_files_ ? kRecyclableMiddleType : kMiddleType; } s = EmitPhysicalRecord(type, ptr, fragment_length); ptr += fragment_length; left -= fragment_length; begin = false; } while (s.ok() && left > 0); return s; } Status Writer::EmitPhysicalRecord(RecordType t, const char* ptr, size_t n) { assert(n <= 0xffff); // Must fit in two bytes size_t header_size; char buf[kRecyclableHeaderSize]; // Format the header buf[4] = static_cast(n & 0xff); buf[5] = static_cast(n >> 8); buf[6] = static_cast(t); uint32_t crc = type_crc_[t]; if (t < kRecyclableFullType) { // Legacy record format assert(block_offset_ + kHeaderSize + n <= kBlockSize); header_size = kHeaderSize; } else { // Recyclable record format assert(block_offset_ + kRecyclableHeaderSize + n <= kBlockSize); header_size = kRecyclableHeaderSize; // Only encode low 32-bits of the 64-bit log number. This means // we will fail to detect an old record if we recycled a log from // ~4 billion logs ago, but that is effectively impossible, and // even if it were we'dbe far more likely to see a false positive // on the 32-bit CRC. EncodeFixed32(buf + 7, static_cast(log_number_)); crc = crc32c::Extend(crc, buf + 7, 4); } // Compute the crc of the record type and the payload. crc = crc32c::Extend(crc, ptr, n); crc = crc32c::Mask(crc); // Adjust for storage EncodeFixed32(buf, crc); // Write the header and the payload Status s = dest_->Append(Slice(buf, header_size)); if (s.ok()) { s = dest_->Append(Slice(ptr, n)); if (s.ok()) { s = dest_->Flush(); } } block_offset_ += header_size + n; return s; } } // namespace log } // namespace rocksdb rocksdb-4.5.1/db/log_writer.h000066400000000000000000000067721267530634000161050ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include "db/log_format.h" #include "rocksdb/slice.h" #include "rocksdb/status.h" namespace rocksdb { class WritableFileWriter; using std::unique_ptr; namespace log { /** * Writer is a general purpose log stream writer. It provides an append-only * abstraction for writing data. The details of the how the data is written is * handled by the WriteableFile sub-class implementation. * * File format: * * File is broken down into variable sized records. The format of each record * is described below. * +-----+-------------+--+----+----------+------+-- ... ----+ * File | r0 | r1 |P | r2 | r3 | r4 | | * +-----+-------------+--+----+----------+------+-- ... ----+ * <--- kBlockSize ------>|<-- kBlockSize ------>| * rn = variable size records * P = Padding * * Data is written out in kBlockSize chunks. If next record does not fit * into the space left, the leftover space will be padded with \0. * * Legacy record format: * * +---------+-----------+-----------+--- ... ---+ * |CRC (4B) | Size (2B) | Type (1B) | Payload | * +---------+-----------+-----------+--- ... ---+ * * CRC = 32bit hash computed over the payload using CRC * Size = Length of the payload data * Type = Type of record * (kZeroType, kFullType, kFirstType, kLastType, kMiddleType ) * The type is used to group a bunch of records together to represent * blocks that are larger than kBlockSize * Payload = Byte stream as long as specified by the payload size * * Recyclable record format: * * +---------+-----------+-----------+----------------+--- ... ---+ * |CRC (4B) | Size (2B) | Type (1B) | Log number (4B)| Payload | * +---------+-----------+-----------+----------------+--- ... ---+ * * Same as above, with the addition of * Log number = 32bit log file number, so that we can distinguish between * records written by the most recent log writer vs a previous one. */ class Writer { public: // Create a writer that will append data to "*dest". // "*dest" must be initially empty. // "*dest" must remain live while this Writer is in use. explicit Writer(unique_ptr&& dest, uint64_t log_number, bool recycle_log_files); ~Writer(); Status AddRecord(const Slice& slice); WritableFileWriter* file() { return dest_.get(); } const WritableFileWriter* file() const { return dest_.get(); } private: unique_ptr dest_; size_t block_offset_; // Current offset in block uint64_t log_number_; bool recycle_log_files_; // crc32c values for all supported record types. These are // pre-computed to reduce the overhead of computing the crc of the // record type stored in the header. uint32_t type_crc_[kMaxRecordType + 1]; Status EmitPhysicalRecord(RecordType type, const char* ptr, size_t length); // No copying allowed Writer(const Writer&); void operator=(const Writer&); }; } // namespace log } // namespace rocksdb rocksdb-4.5.1/db/managed_iterator.cc000066400000000000000000000153671267530634000173730ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "db/managed_iterator.h" #include #include #include #include "db/column_family.h" #include "db/db_impl.h" #include "db/db_iter.h" #include "db/dbformat.h" #include "db/xfunc_test_points.h" #include "rocksdb/env.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "table/merger.h" #include "util/xfunc.h" namespace rocksdb { namespace { // Helper class that locks a mutex on construction and unlocks the mutex when // the destructor of the MutexLock object is invoked. // // Typical usage: // // void MyClass::MyMethod() { // MILock l(&mu_); // mu_ is an instance variable // ... some complex code, possibly with multiple return paths ... // } class MILock { public: explicit MILock(std::mutex* mu, ManagedIterator* mi) : mu_(mu), mi_(mi) { this->mu_->lock(); } ~MILock() { this->mu_->unlock(); XFUNC_TEST("managed_xftest_release", "managed_unlock", managed_unlock1, xf_manage_release, mi_); } ManagedIterator* GetManagedIterator() { return mi_; } private: std::mutex* const mu_; ManagedIterator* mi_; // No copying allowed MILock(const MILock&) = delete; void operator=(const MILock&) = delete; }; } // anonymous namespace // // Synchronization between modifiers, releasers, creators // If iterator operation, wait till (!in_use), set in_use, do op, reset in_use // if modifying mutable_iter, atomically exchange in_use: // return if in_use set / otherwise set in use, // atomically replace new iter with old , reset in use // The releaser is the new operation and it holds a lock for a very short time // The existing non-const iterator operations are supposed to be single // threaded and hold the lock for the duration of the operation // The existing const iterator operations use the cached key/values // and don't do any locking. ManagedIterator::ManagedIterator(DBImpl* db, const ReadOptions& read_options, ColumnFamilyData* cfd) : db_(db), read_options_(read_options), cfd_(cfd), svnum_(cfd->GetSuperVersionNumber()), mutable_iter_(nullptr), valid_(false), snapshot_created_(false), release_supported_(true) { read_options_.managed = false; if ((!read_options_.tailing) && (read_options_.snapshot == nullptr)) { assert(read_options_.snapshot = db_->GetSnapshot()); snapshot_created_ = true; } cfh_.SetCFD(cfd); mutable_iter_ = unique_ptr(db->NewIterator(read_options_, &cfh_)); XFUNC_TEST("managed_xftest_dropold", "managed_create", xf_managed_create1, xf_manage_create, this); } ManagedIterator::~ManagedIterator() { Lock(); if (snapshot_created_) { db_->ReleaseSnapshot(read_options_.snapshot); snapshot_created_ = false; read_options_.snapshot = nullptr; } UnLock(); } bool ManagedIterator::Valid() const { return valid_; } void ManagedIterator::SeekToLast() { MILock l(&in_use_, this); if (NeedToRebuild()) { RebuildIterator(); } assert(mutable_iter_ != nullptr); mutable_iter_->SeekToLast(); if (mutable_iter_->status().ok()) { UpdateCurrent(); } } void ManagedIterator::SeekToFirst() { MILock l(&in_use_, this); SeekInternal(Slice(), true); } void ManagedIterator::Seek(const Slice& user_key) { MILock l(&in_use_, this); SeekInternal(user_key, false); } void ManagedIterator::SeekInternal(const Slice& user_key, bool seek_to_first) { if (NeedToRebuild()) { RebuildIterator(); } assert(mutable_iter_ != nullptr); if (seek_to_first) { mutable_iter_->SeekToFirst(); } else { mutable_iter_->Seek(user_key); } UpdateCurrent(); } void ManagedIterator::Prev() { if (!valid_) { status_ = Status::InvalidArgument("Iterator value invalid"); return; } MILock l(&in_use_, this); if (NeedToRebuild()) { std::string current_key = key().ToString(); Slice old_key(current_key); RebuildIterator(); SeekInternal(old_key, false); UpdateCurrent(); if (!valid_) { return; } if (key().compare(old_key) != 0) { valid_ = false; status_ = Status::Incomplete("Cannot do Prev now"); return; } } mutable_iter_->Prev(); if (mutable_iter_->status().ok()) { UpdateCurrent(); status_ = Status::OK(); } else { status_ = mutable_iter_->status(); } } void ManagedIterator::Next() { if (!valid_) { status_ = Status::InvalidArgument("Iterator value invalid"); return; } MILock l(&in_use_, this); if (NeedToRebuild()) { std::string current_key = key().ToString(); Slice old_key(current_key.data(), cached_key_.Size()); RebuildIterator(); SeekInternal(old_key, false); UpdateCurrent(); if (!valid_) { return; } if (key().compare(old_key) != 0) { valid_ = false; status_ = Status::Incomplete("Cannot do Next now"); return; } } mutable_iter_->Next(); UpdateCurrent(); } Slice ManagedIterator::key() const { assert(valid_); return cached_key_.GetKey(); } Slice ManagedIterator::value() const { assert(valid_); return cached_value_.GetKey(); } Status ManagedIterator::status() const { return status_; } void ManagedIterator::RebuildIterator() { svnum_ = cfd_->GetSuperVersionNumber(); mutable_iter_ = unique_ptr(db_->NewIterator(read_options_, &cfh_)); } void ManagedIterator::UpdateCurrent() { assert(mutable_iter_ != nullptr); if (!(valid_ = mutable_iter_->Valid())) { status_ = mutable_iter_->status(); return; } status_ = Status::OK(); cached_key_.SetKey(mutable_iter_->key()); cached_value_.SetKey(mutable_iter_->value()); } void ManagedIterator::ReleaseIter(bool only_old) { if ((mutable_iter_ == nullptr) || (!release_supported_)) { return; } if (svnum_ != cfd_->GetSuperVersionNumber() || !only_old) { if (!TryLock()) { // Don't release iter if in use return; } mutable_iter_ = nullptr; // in_use for a very short time UnLock(); } } bool ManagedIterator::NeedToRebuild() { if ((mutable_iter_ == nullptr) || (status_.IsIncomplete()) || (!only_drop_old_ && (svnum_ != cfd_->GetSuperVersionNumber()))) { return true; } return false; } void ManagedIterator::Lock() { in_use_.lock(); return; } bool ManagedIterator::TryLock() { return in_use_.try_lock(); } void ManagedIterator::UnLock() { in_use_.unlock(); XFUNC_TEST("managed_xftest_release", "managed_unlock", managed_unlock1, xf_manage_release, this); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/managed_iterator.h000066400000000000000000000046261267530634000172310ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include #include "db/column_family.h" #include "rocksdb/db.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "util/arena.h" namespace rocksdb { class DBImpl; struct SuperVersion; class ColumnFamilyData; /** * ManagedIterator is a special type of iterator that supports freeing the * underlying iterator and still being able to access the current key/value * pair. This is done by copying the key/value pair so that clients can * continue to access the data without getting a SIGSEGV. * The underlying iterator can be freed manually through the call to * ReleaseIter or automatically (as needed on space pressure or age.) * The iterator is recreated using the saved original arguments. */ class ManagedIterator : public Iterator { public: ManagedIterator(DBImpl* db, const ReadOptions& read_options, ColumnFamilyData* cfd); virtual ~ManagedIterator(); virtual void SeekToLast() override; virtual void Prev() override; virtual bool Valid() const override; void SeekToFirst() override; virtual void Seek(const Slice& target) override; virtual void Next() override; virtual Slice key() const override; virtual Slice value() const override; virtual Status status() const override; void ReleaseIter(bool only_old); void SetDropOld(bool only_old) { only_drop_old_ = read_options_.tailing || only_old; } private: void RebuildIterator(); void UpdateCurrent(); void SeekInternal(const Slice& user_key, bool seek_to_first); bool NeedToRebuild(); void Lock(); bool TryLock(); void UnLock(); DBImpl* const db_; ReadOptions read_options_; ColumnFamilyData* const cfd_; ColumnFamilyHandleInternal cfh_; uint64_t svnum_; std::unique_ptr mutable_iter_; // internal iterator status Status status_; bool valid_; IterKey cached_key_; IterKey cached_value_; bool only_drop_old_ = true; bool snapshot_created_; bool release_supported_; std::mutex in_use_; // is managed iterator in use }; } // namespace rocksdb #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/manual_compaction_test.cc000066400000000000000000000107211267530634000206030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Test for issue 178: a manual compaction causes deleted data to reappear. #include #include #include #include "rocksdb/db.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/slice.h" #include "rocksdb/write_batch.h" #include "util/testharness.h" #include "port/port.h" using namespace rocksdb; namespace { const int kNumKeys = 1100000; std::string Key1(int i) { char buf[100]; snprintf(buf, sizeof(buf), "my_key_%d", i); return buf; } std::string Key2(int i) { return Key1(i) + "_xxx"; } class ManualCompactionTest : public testing::Test { public: ManualCompactionTest() { // Get rid of any state from an old run. dbname_ = rocksdb::test::TmpDir() + "/rocksdb_cbug_test"; DestroyDB(dbname_, rocksdb::Options()); } std::string dbname_; }; class DestroyAllCompactionFilter : public CompactionFilter { public: DestroyAllCompactionFilter() {} virtual bool Filter(int level, const Slice& key, const Slice& existing_value, std::string* new_value, bool* value_changed) const override { return existing_value.ToString() == "destroy"; } virtual const char* Name() const override { return "DestroyAllCompactionFilter"; } }; TEST_F(ManualCompactionTest, CompactTouchesAllKeys) { for (int iter = 0; iter < 2; ++iter) { DB* db; Options options; if (iter == 0) { // level compaction options.num_levels = 3; options.compaction_style = kCompactionStyleLevel; } else { // universal compaction options.compaction_style = kCompactionStyleUniversal; } options.create_if_missing = true; options.compression = rocksdb::kNoCompression; options.compaction_filter = new DestroyAllCompactionFilter(); ASSERT_OK(DB::Open(options, dbname_, &db)); db->Put(WriteOptions(), Slice("key1"), Slice("destroy")); db->Put(WriteOptions(), Slice("key2"), Slice("destroy")); db->Put(WriteOptions(), Slice("key3"), Slice("value3")); db->Put(WriteOptions(), Slice("key4"), Slice("destroy")); Slice key4("key4"); db->CompactRange(CompactRangeOptions(), nullptr, &key4); Iterator* itr = db->NewIterator(ReadOptions()); itr->SeekToFirst(); ASSERT_TRUE(itr->Valid()); ASSERT_EQ("key3", itr->key().ToString()); itr->Next(); ASSERT_TRUE(!itr->Valid()); delete itr; delete options.compaction_filter; delete db; DestroyDB(dbname_, options); } } TEST_F(ManualCompactionTest, Test) { // Open database. Disable compression since it affects the creation // of layers and the code below is trying to test against a very // specific scenario. rocksdb::DB* db; rocksdb::Options db_options; db_options.create_if_missing = true; db_options.compression = rocksdb::kNoCompression; ASSERT_OK(rocksdb::DB::Open(db_options, dbname_, &db)); // create first key range rocksdb::WriteBatch batch; for (int i = 0; i < kNumKeys; i++) { batch.Put(Key1(i), "value for range 1 key"); } ASSERT_OK(db->Write(rocksdb::WriteOptions(), &batch)); // create second key range batch.Clear(); for (int i = 0; i < kNumKeys; i++) { batch.Put(Key2(i), "value for range 2 key"); } ASSERT_OK(db->Write(rocksdb::WriteOptions(), &batch)); // delete second key range batch.Clear(); for (int i = 0; i < kNumKeys; i++) { batch.Delete(Key2(i)); } ASSERT_OK(db->Write(rocksdb::WriteOptions(), &batch)); // compact database std::string start_key = Key1(0); std::string end_key = Key1(kNumKeys - 1); rocksdb::Slice least(start_key.data(), start_key.size()); rocksdb::Slice greatest(end_key.data(), end_key.size()); // commenting out the line below causes the example to work correctly db->CompactRange(CompactRangeOptions(), &least, &greatest); // count the keys rocksdb::Iterator* iter = db->NewIterator(rocksdb::ReadOptions()); int num_keys = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { num_keys++; } delete iter; ASSERT_EQ(kNumKeys, num_keys) << "Bad number of keys"; // close database delete db; DestroyDB(dbname_, rocksdb::Options()); } } // anonymous namespace int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/memtable.cc000066400000000000000000000702721267530634000156500ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/memtable.h" #include #include #include #include "db/dbformat.h" #include "db/merge_context.h" #include "db/writebuffer.h" #include "rocksdb/comparator.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "rocksdb/merge_operator.h" #include "rocksdb/slice_transform.h" #include "table/internal_iterator.h" #include "table/merger.h" #include "util/arena.h" #include "util/coding.h" #include "util/murmurhash.h" #include "util/mutexlock.h" #include "util/perf_context_imp.h" #include "util/statistics.h" #include "util/stop_watch.h" namespace rocksdb { MemTableOptions::MemTableOptions( const ImmutableCFOptions& ioptions, const MutableCFOptions& mutable_cf_options) : write_buffer_size(mutable_cf_options.write_buffer_size), arena_block_size(mutable_cf_options.arena_block_size), memtable_prefix_bloom_bits(mutable_cf_options.memtable_prefix_bloom_bits), memtable_prefix_bloom_probes( mutable_cf_options.memtable_prefix_bloom_probes), memtable_prefix_bloom_huge_page_tlb_size( mutable_cf_options.memtable_prefix_bloom_huge_page_tlb_size), inplace_update_support(ioptions.inplace_update_support), inplace_update_num_locks(mutable_cf_options.inplace_update_num_locks), inplace_callback(ioptions.inplace_callback), max_successive_merges(mutable_cf_options.max_successive_merges), filter_deletes(mutable_cf_options.filter_deletes), statistics(ioptions.statistics), merge_operator(ioptions.merge_operator), info_log(ioptions.info_log) {} MemTable::MemTable(const InternalKeyComparator& cmp, const ImmutableCFOptions& ioptions, const MutableCFOptions& mutable_cf_options, WriteBuffer* write_buffer, SequenceNumber earliest_seq) : comparator_(cmp), moptions_(ioptions, mutable_cf_options), refs_(0), kArenaBlockSize(OptimizeBlockSize(moptions_.arena_block_size)), arena_(moptions_.arena_block_size, 0), allocator_(&arena_, write_buffer), table_(ioptions.memtable_factory->CreateMemTableRep( comparator_, &allocator_, ioptions.prefix_extractor, ioptions.info_log)), data_size_(0), num_entries_(0), num_deletes_(0), flush_in_progress_(false), flush_completed_(false), file_number_(0), first_seqno_(0), earliest_seqno_(earliest_seq), mem_next_logfile_number_(0), locks_(moptions_.inplace_update_support ? moptions_.inplace_update_num_locks : 0), prefix_extractor_(ioptions.prefix_extractor), flush_state_(FLUSH_NOT_REQUESTED), env_(ioptions.env) { UpdateFlushState(); // something went wrong if we need to flush before inserting anything assert(!ShouldScheduleFlush()); if (prefix_extractor_ && moptions_.memtable_prefix_bloom_bits > 0) { prefix_bloom_.reset(new DynamicBloom( &allocator_, moptions_.memtable_prefix_bloom_bits, ioptions.bloom_locality, moptions_.memtable_prefix_bloom_probes, nullptr, moptions_.memtable_prefix_bloom_huge_page_tlb_size, ioptions.info_log)); } } MemTable::~MemTable() { assert(refs_ == 0); } size_t MemTable::ApproximateMemoryUsage() { size_t arena_usage = arena_.ApproximateMemoryUsage(); size_t table_usage = table_->ApproximateMemoryUsage(); // let MAX_USAGE = std::numeric_limits::max() // then if arena_usage + total_usage >= MAX_USAGE, return MAX_USAGE. // the following variation is to avoid numeric overflow. if (arena_usage >= std::numeric_limits::max() - table_usage) { return std::numeric_limits::max(); } // otherwise, return the actual usage return arena_usage + table_usage; } bool MemTable::ShouldFlushNow() const { // In a lot of times, we cannot allocate arena blocks that exactly matches the // buffer size. Thus we have to decide if we should over-allocate or // under-allocate. // This constant variable can be interpreted as: if we still have more than // "kAllowOverAllocationRatio * kArenaBlockSize" space left, we'd try to over // allocate one more block. const double kAllowOverAllocationRatio = 0.6; // If arena still have room for new block allocation, we can safely say it // shouldn't flush. auto allocated_memory = table_->ApproximateMemoryUsage() + arena_.MemoryAllocatedBytes(); // if we can still allocate one more block without exceeding the // over-allocation ratio, then we should not flush. if (allocated_memory + kArenaBlockSize < moptions_.write_buffer_size + kArenaBlockSize * kAllowOverAllocationRatio) { return false; } // if user keeps adding entries that exceeds moptions.write_buffer_size, // we need to flush earlier even though we still have much available // memory left. if (allocated_memory > moptions_.write_buffer_size + kArenaBlockSize * kAllowOverAllocationRatio) { return true; } // In this code path, Arena has already allocated its "last block", which // means the total allocatedmemory size is either: // (1) "moderately" over allocated the memory (no more than `0.6 * arena // block size`. Or, // (2) the allocated memory is less than write buffer size, but we'll stop // here since if we allocate a new arena block, we'll over allocate too much // more (half of the arena block size) memory. // // In either case, to avoid over-allocate, the last block will stop allocation // when its usage reaches a certain ratio, which we carefully choose "0.75 // full" as the stop condition because it addresses the following issue with // great simplicity: What if the next inserted entry's size is // bigger than AllocatedAndUnused()? // // The answer is: if the entry size is also bigger than 0.25 * // kArenaBlockSize, a dedicated block will be allocated for it; otherwise // arena will anyway skip the AllocatedAndUnused() and allocate a new, empty // and regular block. In either case, we *overly* over-allocated. // // Therefore, setting the last block to be at most "0.75 full" avoids both // cases. // // NOTE: the average percentage of waste space of this approach can be counted // as: "arena block size * 0.25 / write buffer size". User who specify a small // write buffer size and/or big arena block size may suffer. return arena_.AllocatedAndUnused() < kArenaBlockSize / 4; } void MemTable::UpdateFlushState() { auto state = flush_state_.load(std::memory_order_relaxed); if (state == FLUSH_NOT_REQUESTED && ShouldFlushNow()) { // ignore CAS failure, because that means somebody else requested // a flush flush_state_.compare_exchange_strong(state, FLUSH_REQUESTED, std::memory_order_relaxed, std::memory_order_relaxed); } } int MemTable::KeyComparator::operator()(const char* prefix_len_key1, const char* prefix_len_key2) const { // Internal keys are encoded as length-prefixed strings. Slice k1 = GetLengthPrefixedSlice(prefix_len_key1); Slice k2 = GetLengthPrefixedSlice(prefix_len_key2); return comparator.Compare(k1, k2); } int MemTable::KeyComparator::operator()(const char* prefix_len_key, const Slice& key) const { // Internal keys are encoded as length-prefixed strings. Slice a = GetLengthPrefixedSlice(prefix_len_key); return comparator.Compare(a, key); } Slice MemTableRep::UserKey(const char* key) const { Slice slice = GetLengthPrefixedSlice(key); return Slice(slice.data(), slice.size() - 8); } KeyHandle MemTableRep::Allocate(const size_t len, char** buf) { *buf = allocator_->Allocate(len); return static_cast(*buf); } // Encode a suitable internal key target for "target" and return it. // Uses *scratch as scratch space, and the returned pointer will point // into this scratch space. const char* EncodeKey(std::string* scratch, const Slice& target) { scratch->clear(); PutVarint32(scratch, static_cast(target.size())); scratch->append(target.data(), target.size()); return scratch->data(); } class MemTableIterator : public InternalIterator { public: MemTableIterator( const MemTable& mem, const ReadOptions& read_options, Arena* arena) : bloom_(nullptr), prefix_extractor_(mem.prefix_extractor_), valid_(false), arena_mode_(arena != nullptr) { if (prefix_extractor_ != nullptr && !read_options.total_order_seek) { bloom_ = mem.prefix_bloom_.get(); iter_ = mem.table_->GetDynamicPrefixIterator(arena); } else { iter_ = mem.table_->GetIterator(arena); } } ~MemTableIterator() { if (arena_mode_) { iter_->~Iterator(); } else { delete iter_; } } virtual bool Valid() const override { return valid_; } virtual void Seek(const Slice& k) override { PERF_TIMER_GUARD(seek_on_memtable_time); PERF_COUNTER_ADD(seek_on_memtable_count, 1); if (bloom_ != nullptr) { if (!bloom_->MayContain( prefix_extractor_->Transform(ExtractUserKey(k)))) { PERF_COUNTER_ADD(bloom_memtable_miss_count, 1); valid_ = false; return; } else { PERF_COUNTER_ADD(bloom_memtable_hit_count, 1); } } iter_->Seek(k, nullptr); valid_ = iter_->Valid(); } virtual void SeekToFirst() override { iter_->SeekToFirst(); valid_ = iter_->Valid(); } virtual void SeekToLast() override { iter_->SeekToLast(); valid_ = iter_->Valid(); } virtual void Next() override { assert(Valid()); iter_->Next(); valid_ = iter_->Valid(); } virtual void Prev() override { assert(Valid()); iter_->Prev(); valid_ = iter_->Valid(); } virtual Slice key() const override { assert(Valid()); return GetLengthPrefixedSlice(iter_->key()); } virtual Slice value() const override { assert(Valid()); Slice key_slice = GetLengthPrefixedSlice(iter_->key()); return GetLengthPrefixedSlice(key_slice.data() + key_slice.size()); } virtual Status status() const override { return Status::OK(); } virtual Status PinData() override { // memtable data is always pinned return Status::OK(); } virtual Status ReleasePinnedData() override { // memtable data is always pinned return Status::OK(); } virtual bool IsKeyPinned() const override { // memtable data is always pinned return true; } private: DynamicBloom* bloom_; const SliceTransform* const prefix_extractor_; MemTableRep::Iterator* iter_; bool valid_; bool arena_mode_; // No copying allowed MemTableIterator(const MemTableIterator&); void operator=(const MemTableIterator&); }; InternalIterator* MemTable::NewIterator(const ReadOptions& read_options, Arena* arena) { assert(arena != nullptr); auto mem = arena->AllocateAligned(sizeof(MemTableIterator)); return new (mem) MemTableIterator(*this, read_options, arena); } port::RWMutex* MemTable::GetLock(const Slice& key) { static murmur_hash hash; return &locks_[hash(key) % locks_.size()]; } uint64_t MemTable::ApproximateSize(const Slice& start_ikey, const Slice& end_ikey) { uint64_t entry_count = table_->ApproximateNumEntries(start_ikey, end_ikey); if (entry_count == 0) { return 0; } uint64_t n = num_entries_.load(std::memory_order_relaxed); if (n == 0) { return 0; } if (entry_count > n) { // table_->ApproximateNumEntries() is just an estimate so it can be larger // than actual entries we have. Cap it to entries we have to limit the // inaccuracy. entry_count = n; } uint64_t data_size = data_size_.load(std::memory_order_relaxed); return entry_count * (data_size / n); } void MemTable::Add(SequenceNumber s, ValueType type, const Slice& key, /* user key */ const Slice& value, bool allow_concurrent) { // Format of an entry is concatenation of: // key_size : varint32 of internal_key.size() // key bytes : char[internal_key.size()] // value_size : varint32 of value.size() // value bytes : char[value.size()] uint32_t key_size = static_cast(key.size()); uint32_t val_size = static_cast(value.size()); uint32_t internal_key_size = key_size + 8; const uint32_t encoded_len = VarintLength(internal_key_size) + internal_key_size + VarintLength(val_size) + val_size; char* buf = nullptr; KeyHandle handle = table_->Allocate(encoded_len, &buf); char* p = EncodeVarint32(buf, internal_key_size); memcpy(p, key.data(), key_size); p += key_size; uint64_t packed = PackSequenceAndType(s, type); EncodeFixed64(p, packed); p += 8; p = EncodeVarint32(p, val_size); memcpy(p, value.data(), val_size); assert((unsigned)(p + val_size - buf) == (unsigned)encoded_len); if (!allow_concurrent) { table_->Insert(handle); // this is a bit ugly, but is the way to avoid locked instructions // when incrementing an atomic num_entries_.store(num_entries_.load(std::memory_order_relaxed) + 1, std::memory_order_relaxed); data_size_.store(data_size_.load(std::memory_order_relaxed) + encoded_len, std::memory_order_relaxed); if (type == kTypeDeletion) { num_deletes_.store(num_deletes_.load(std::memory_order_relaxed) + 1, std::memory_order_relaxed); } if (prefix_bloom_) { assert(prefix_extractor_); prefix_bloom_->Add(prefix_extractor_->Transform(key)); } // The first sequence number inserted into the memtable assert(first_seqno_ == 0 || s > first_seqno_); if (first_seqno_ == 0) { first_seqno_.store(s, std::memory_order_relaxed); if (earliest_seqno_ == kMaxSequenceNumber) { earliest_seqno_.store(GetFirstSequenceNumber(), std::memory_order_relaxed); } assert(first_seqno_.load() >= earliest_seqno_.load()); } } else { table_->InsertConcurrently(handle); num_entries_.fetch_add(1, std::memory_order_relaxed); data_size_.fetch_add(encoded_len, std::memory_order_relaxed); if (type == kTypeDeletion) { num_deletes_.fetch_add(1, std::memory_order_relaxed); } if (prefix_bloom_) { assert(prefix_extractor_); prefix_bloom_->AddConcurrently(prefix_extractor_->Transform(key)); } // atomically update first_seqno_ and earliest_seqno_. uint64_t cur_seq_num = first_seqno_.load(std::memory_order_relaxed); while ((cur_seq_num == 0 || s < cur_seq_num) && !first_seqno_.compare_exchange_weak(cur_seq_num, s)) { } uint64_t cur_earliest_seqno = earliest_seqno_.load(std::memory_order_relaxed); while ( (cur_earliest_seqno == kMaxSequenceNumber || s < cur_earliest_seqno) && !first_seqno_.compare_exchange_weak(cur_earliest_seqno, s)) { } } UpdateFlushState(); } // Callback from MemTable::Get() namespace { struct Saver { Status* status; const LookupKey* key; bool* found_final_value; // Is value set correctly? Used by KeyMayExist bool* merge_in_progress; std::string* value; SequenceNumber seq; const MergeOperator* merge_operator; // the merge operations encountered; MergeContext* merge_context; MemTable* mem; Logger* logger; Statistics* statistics; bool inplace_update_support; Env* env_; }; } // namespace static bool SaveValue(void* arg, const char* entry) { Saver* s = reinterpret_cast(arg); MergeContext* merge_context = s->merge_context; const MergeOperator* merge_operator = s->merge_operator; assert(s != nullptr && merge_context != nullptr); // entry format is: // klength varint32 // userkey char[klength-8] // tag uint64 // vlength varint32 // value char[vlength] // Check that it belongs to same user key. We do not check the // sequence number since the Seek() call above should have skipped // all entries with overly large sequence numbers. uint32_t key_length; const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length); if (s->mem->GetInternalKeyComparator().user_comparator()->Equal( Slice(key_ptr, key_length - 8), s->key->user_key())) { // Correct user key const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8); ValueType type; UnPackSequenceAndType(tag, &s->seq, &type); switch (type) { case kTypeValue: { if (s->inplace_update_support) { s->mem->GetLock(s->key->user_key())->ReadLock(); } Slice v = GetLengthPrefixedSlice(key_ptr + key_length); *(s->status) = Status::OK(); if (*(s->merge_in_progress)) { assert(merge_operator); bool merge_success = false; { StopWatchNano timer(s->env_, s->statistics != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); merge_success = merge_operator->FullMerge( s->key->user_key(), &v, merge_context->GetOperands(), s->value, s->logger); RecordTick(s->statistics, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } if (!merge_success) { RecordTick(s->statistics, NUMBER_MERGE_FAILURES); *(s->status) = Status::Corruption("Error: Could not perform merge."); } } else if (s->value != nullptr) { s->value->assign(v.data(), v.size()); } if (s->inplace_update_support) { s->mem->GetLock(s->key->user_key())->ReadUnlock(); } *(s->found_final_value) = true; return false; } case kTypeDeletion: case kTypeSingleDeletion: { if (*(s->merge_in_progress)) { assert(merge_operator != nullptr); *(s->status) = Status::OK(); bool merge_success = false; { StopWatchNano timer(s->env_, s->statistics != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); merge_success = merge_operator->FullMerge( s->key->user_key(), nullptr, merge_context->GetOperands(), s->value, s->logger); RecordTick(s->statistics, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } if (!merge_success) { RecordTick(s->statistics, NUMBER_MERGE_FAILURES); *(s->status) = Status::Corruption("Error: Could not perform merge."); } } else { *(s->status) = Status::NotFound(); } *(s->found_final_value) = true; return false; } case kTypeMerge: { if (!merge_operator) { *(s->status) = Status::InvalidArgument( "merge_operator is not properly initialized."); // Normally we continue the loop (return true) when we see a merge // operand. But in case of an error, we should stop the loop // immediately and pretend we have found the value to stop further // seek. Otherwise, the later call will override this error status. *(s->found_final_value) = true; return false; } Slice v = GetLengthPrefixedSlice(key_ptr + key_length); *(s->merge_in_progress) = true; merge_context->PushOperand(v); return true; } default: assert(false); return true; } } // s->state could be Corrupt, merge or notfound return false; } bool MemTable::Get(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq) { // The sequence number is updated synchronously in version_set.h if (IsEmpty()) { // Avoiding recording stats for speed. return false; } PERF_TIMER_GUARD(get_from_memtable_time); Slice user_key = key.user_key(); bool found_final_value = false; bool merge_in_progress = s->IsMergeInProgress(); bool const may_contain = nullptr == prefix_bloom_ ? false : prefix_bloom_->MayContain(prefix_extractor_->Transform(user_key)); if (prefix_bloom_ && !may_contain) { // iter is null if prefix bloom says the key does not exist PERF_COUNTER_ADD(bloom_memtable_miss_count, 1); *seq = kMaxSequenceNumber; } else { if (prefix_bloom_) { PERF_COUNTER_ADD(bloom_memtable_hit_count, 1); } Saver saver; saver.status = s; saver.found_final_value = &found_final_value; saver.merge_in_progress = &merge_in_progress; saver.key = &key; saver.value = value; saver.seq = kMaxSequenceNumber; saver.mem = this; saver.merge_context = merge_context; saver.merge_operator = moptions_.merge_operator; saver.logger = moptions_.info_log; saver.inplace_update_support = moptions_.inplace_update_support; saver.statistics = moptions_.statistics; saver.env_ = env_; table_->Get(key, &saver, SaveValue); *seq = saver.seq; } // No change to value, since we have not yet found a Put/Delete if (!found_final_value && merge_in_progress) { *s = Status::MergeInProgress(); } PERF_COUNTER_ADD(get_from_memtable_count, 1); return found_final_value; } void MemTable::Update(SequenceNumber seq, const Slice& key, const Slice& value) { LookupKey lkey(key, seq); Slice mem_key = lkey.memtable_key(); std::unique_ptr iter( table_->GetDynamicPrefixIterator()); iter->Seek(lkey.internal_key(), mem_key.data()); if (iter->Valid()) { // entry format is: // key_length varint32 // userkey char[klength-8] // tag uint64 // vlength varint32 // value char[vlength] // Check that it belongs to same user key. We do not check the // sequence number since the Seek() call above should have skipped // all entries with overly large sequence numbers. const char* entry = iter->key(); uint32_t key_length = 0; const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length); if (comparator_.comparator.user_comparator()->Equal( Slice(key_ptr, key_length - 8), lkey.user_key())) { // Correct user key const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8); ValueType type; SequenceNumber unused; UnPackSequenceAndType(tag, &unused, &type); switch (type) { case kTypeValue: { Slice prev_value = GetLengthPrefixedSlice(key_ptr + key_length); uint32_t prev_size = static_cast(prev_value.size()); uint32_t new_size = static_cast(value.size()); // Update value, if new value size <= previous value size if (new_size <= prev_size ) { char* p = EncodeVarint32(const_cast(key_ptr) + key_length, new_size); WriteLock wl(GetLock(lkey.user_key())); memcpy(p, value.data(), value.size()); assert((unsigned)((p + value.size()) - entry) == (unsigned)(VarintLength(key_length) + key_length + VarintLength(value.size()) + value.size())); return; } } default: // If the latest value is kTypeDeletion, kTypeMerge or kTypeLogData // we don't have enough space for update inplace Add(seq, kTypeValue, key, value); return; } } } // key doesn't exist Add(seq, kTypeValue, key, value); } bool MemTable::UpdateCallback(SequenceNumber seq, const Slice& key, const Slice& delta) { LookupKey lkey(key, seq); Slice memkey = lkey.memtable_key(); std::unique_ptr iter( table_->GetDynamicPrefixIterator()); iter->Seek(lkey.internal_key(), memkey.data()); if (iter->Valid()) { // entry format is: // key_length varint32 // userkey char[klength-8] // tag uint64 // vlength varint32 // value char[vlength] // Check that it belongs to same user key. We do not check the // sequence number since the Seek() call above should have skipped // all entries with overly large sequence numbers. const char* entry = iter->key(); uint32_t key_length = 0; const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length); if (comparator_.comparator.user_comparator()->Equal( Slice(key_ptr, key_length - 8), lkey.user_key())) { // Correct user key const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8); ValueType type; uint64_t unused; UnPackSequenceAndType(tag, &unused, &type); switch (type) { case kTypeValue: { Slice prev_value = GetLengthPrefixedSlice(key_ptr + key_length); uint32_t prev_size = static_cast(prev_value.size()); char* prev_buffer = const_cast(prev_value.data()); uint32_t new_prev_size = prev_size; std::string str_value; WriteLock wl(GetLock(lkey.user_key())); auto status = moptions_.inplace_callback(prev_buffer, &new_prev_size, delta, &str_value); if (status == UpdateStatus::UPDATED_INPLACE) { // Value already updated by callback. assert(new_prev_size <= prev_size); if (new_prev_size < prev_size) { // overwrite the new prev_size char* p = EncodeVarint32(const_cast(key_ptr) + key_length, new_prev_size); if (VarintLength(new_prev_size) < VarintLength(prev_size)) { // shift the value buffer as well. memcpy(p, prev_buffer, new_prev_size); } } RecordTick(moptions_.statistics, NUMBER_KEYS_UPDATED); UpdateFlushState(); return true; } else if (status == UpdateStatus::UPDATED) { Add(seq, kTypeValue, key, Slice(str_value)); RecordTick(moptions_.statistics, NUMBER_KEYS_WRITTEN); UpdateFlushState(); return true; } else if (status == UpdateStatus::UPDATE_FAILED) { // No action required. Return. UpdateFlushState(); return true; } } default: break; } } } // If the latest value is not kTypeValue // or key doesn't exist return false; } size_t MemTable::CountSuccessiveMergeEntries(const LookupKey& key) { Slice memkey = key.memtable_key(); // A total ordered iterator is costly for some memtablerep (prefix aware // reps). By passing in the user key, we allow efficient iterator creation. // The iterator only needs to be ordered within the same user key. std::unique_ptr iter( table_->GetDynamicPrefixIterator()); iter->Seek(key.internal_key(), memkey.data()); size_t num_successive_merges = 0; for (; iter->Valid(); iter->Next()) { const char* entry = iter->key(); uint32_t key_length = 0; const char* iter_key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length); if (!comparator_.comparator.user_comparator()->Equal( Slice(iter_key_ptr, key_length - 8), key.user_key())) { break; } const uint64_t tag = DecodeFixed64(iter_key_ptr + key_length - 8); ValueType type; uint64_t unused; UnPackSequenceAndType(tag, &unused, &type); if (type != kTypeMerge) { break; } ++num_successive_merges; } return num_successive_merges; } void MemTableRep::Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) { auto iter = GetDynamicPrefixIterator(); for (iter->Seek(k.internal_key(), k.memtable_key().data()); iter->Valid() && callback_func(callback_args, iter->key()); iter->Next()) { } } } // namespace rocksdb rocksdb-4.5.1/db/memtable.h000066400000000000000000000341441267530634000155100ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include #include "db/dbformat.h" #include "db/skiplist.h" #include "db/version_edit.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "rocksdb/immutable_options.h" #include "db/memtable_allocator.h" #include "util/concurrent_arena.h" #include "util/dynamic_bloom.h" #include "util/instrumented_mutex.h" #include "util/mutable_cf_options.h" namespace rocksdb { class Mutex; class MemTableIterator; class MergeContext; class WriteBuffer; class InternalIterator; struct MemTableOptions { explicit MemTableOptions( const ImmutableCFOptions& ioptions, const MutableCFOptions& mutable_cf_options); size_t write_buffer_size; size_t arena_block_size; uint32_t memtable_prefix_bloom_bits; uint32_t memtable_prefix_bloom_probes; size_t memtable_prefix_bloom_huge_page_tlb_size; bool inplace_update_support; size_t inplace_update_num_locks; UpdateStatus (*inplace_callback)(char* existing_value, uint32_t* existing_value_size, Slice delta_value, std::string* merged_value); size_t max_successive_merges; bool filter_deletes; Statistics* statistics; MergeOperator* merge_operator; Logger* info_log; }; // Note: Many of the methods in this class have comments indicating that // external synchromization is required as these methods are not thread-safe. // It is up to higher layers of code to decide how to prevent concurrent // invokation of these methods. This is usually done by acquiring either // the db mutex or the single writer thread. // // Some of these methods are documented to only require external // synchronization if this memtable is immutable. Calling MarkImmutable() is // not sufficient to guarantee immutability. It is up to higher layers of // code to determine if this MemTable can still be modified by other threads. // Eg: The Superversion stores a pointer to the current MemTable (that can // be modified) and a separate list of the MemTables that can no longer be // written to (aka the 'immutable memtables'). class MemTable { public: struct KeyComparator : public MemTableRep::KeyComparator { const InternalKeyComparator comparator; explicit KeyComparator(const InternalKeyComparator& c) : comparator(c) { } virtual int operator()(const char* prefix_len_key1, const char* prefix_len_key2) const override; virtual int operator()(const char* prefix_len_key, const Slice& key) const override; }; // MemTables are reference counted. The initial reference count // is zero and the caller must call Ref() at least once. // // earliest_seq should be the current SequenceNumber in the db such that any // key inserted into this memtable will have an equal or larger seq number. // (When a db is first created, the earliest sequence number will be 0). // If the earliest sequence number is not known, kMaxSequenceNumber may be // used, but this may prevent some transactions from succeeding until the // first key is inserted into the memtable. explicit MemTable(const InternalKeyComparator& comparator, const ImmutableCFOptions& ioptions, const MutableCFOptions& mutable_cf_options, WriteBuffer* write_buffer, SequenceNumber earliest_seq); // Do not delete this MemTable unless Unref() indicates it not in use. ~MemTable(); // Increase reference count. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. void Ref() { ++refs_; } // Drop reference count. // If the refcount goes to zero return this memtable, otherwise return null. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. MemTable* Unref() { --refs_; assert(refs_ >= 0); if (refs_ <= 0) { return this; } return nullptr; } // Returns an estimate of the number of bytes of data in use by this // data structure. // // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable (unless this Memtable is immutable). size_t ApproximateMemoryUsage(); // This method heuristically determines if the memtable should continue to // host more data. bool ShouldScheduleFlush() const { return flush_state_.load(std::memory_order_relaxed) == FLUSH_REQUESTED; } // Returns true if a flush should be scheduled and the caller should // be the one to schedule it bool MarkFlushScheduled() { auto before = FLUSH_REQUESTED; return flush_state_.compare_exchange_strong(before, FLUSH_SCHEDULED, std::memory_order_relaxed, std::memory_order_relaxed); } // Return an iterator that yields the contents of the memtable. // // The caller must ensure that the underlying MemTable remains live // while the returned iterator is live. The keys returned by this // iterator are internal keys encoded by AppendInternalKey in the // db/dbformat.{h,cc} module. // // By default, it returns an iterator for prefix seek if prefix_extractor // is configured in Options. // arena: If not null, the arena needs to be used to allocate the Iterator. // Calling ~Iterator of the iterator will destroy all the states but // those allocated in arena. InternalIterator* NewIterator(const ReadOptions& read_options, Arena* arena); // Add an entry into memtable that maps key to value at the // specified sequence number and with the specified type. // Typically value will be empty if type==kTypeDeletion. // // REQUIRES: if allow_concurrent = false, external synchronization to prevent // simultaneous operations on the same MemTable. void Add(SequenceNumber seq, ValueType type, const Slice& key, const Slice& value, bool allow_concurrent = false); // If memtable contains a value for key, store it in *value and return true. // If memtable contains a deletion for key, store a NotFound() error // in *status and return true. // If memtable contains Merge operation as the most recent entry for a key, // and the merge process does not stop (not reaching a value or delete), // prepend the current merge operand to *operands. // store MergeInProgress in s, and return false. // Else, return false. // If any operation was found, its most recent sequence number // will be stored in *seq on success (regardless of whether true/false is // returned). Otherwise, *seq will be set to kMaxSequenceNumber. // On success, *s may be set to OK, NotFound, or MergeInProgress. Any other // status returned indicates a corruption or other unexpected error. bool Get(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq); bool Get(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context) { SequenceNumber seq; return Get(key, value, s, merge_context, &seq); } // Attempts to update the new_value inplace, else does normal Add // Pseudocode // if key exists in current memtable && prev_value is of type kTypeValue // if new sizeof(new_value) <= sizeof(prev_value) // update inplace // else add(key, new_value) // else add(key, new_value) // // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. void Update(SequenceNumber seq, const Slice& key, const Slice& value); // If prev_value for key exists, attempts to update it inplace. // else returns false // Pseudocode // if key exists in current memtable && prev_value is of type kTypeValue // new_value = delta(prev_value) // if sizeof(new_value) <= sizeof(prev_value) // update inplace // else add(key, new_value) // else return false // // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. bool UpdateCallback(SequenceNumber seq, const Slice& key, const Slice& delta); // Returns the number of successive merge entries starting from the newest // entry for the key up to the last non-merge entry or last entry for the // key in the memtable. size_t CountSuccessiveMergeEntries(const LookupKey& key); // Get total number of entries in the mem table. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable (unless this Memtable is immutable). uint64_t num_entries() const { return num_entries_.load(std::memory_order_relaxed); } // Get total number of deletes in the mem table. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable (unless this Memtable is immutable). uint64_t num_deletes() const { return num_deletes_.load(std::memory_order_relaxed); } // Returns the edits area that is needed for flushing the memtable VersionEdit* GetEdits() { return &edit_; } // Returns if there is no entry inserted to the mem table. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable (unless this Memtable is immutable). bool IsEmpty() const { return first_seqno_ == 0; } // Returns the sequence number of the first element that was inserted // into the memtable. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable (unless this Memtable is immutable). SequenceNumber GetFirstSequenceNumber() { return first_seqno_.load(std::memory_order_relaxed); } // Returns the sequence number that is guaranteed to be smaller than or equal // to the sequence number of any key that could be inserted into this // memtable. It can then be assumed that any write with a larger(or equal) // sequence number will be present in this memtable or a later memtable. // // If the earliest sequence number could not be determined, // kMaxSequenceNumber will be returned. SequenceNumber GetEarliestSequenceNumber() { return earliest_seqno_.load(std::memory_order_relaxed); } // Returns the next active logfile number when this memtable is about to // be flushed to storage // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. uint64_t GetNextLogNumber() { return mem_next_logfile_number_; } // Sets the next active logfile number when this memtable is about to // be flushed to storage // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. void SetNextLogNumber(uint64_t num) { mem_next_logfile_number_ = num; } // Notify the underlying storage that no more items will be added. // REQUIRES: external synchronization to prevent simultaneous // operations on the same MemTable. // After MarkImmutable() is called, you should not attempt to // write anything to this MemTable(). (Ie. do not call Add() or Update()). void MarkImmutable() { table_->MarkReadOnly(); allocator_.DoneAllocating(); } // return true if the current MemTableRep supports merge operator. bool IsMergeOperatorSupported() const { return table_->IsMergeOperatorSupported(); } // return true if the current MemTableRep supports snapshots. // inplace update prevents snapshots, bool IsSnapshotSupported() const { return table_->IsSnapshotSupported() && !moptions_.inplace_update_support; } uint64_t ApproximateSize(const Slice& start_ikey, const Slice& end_ikey); // Get the lock associated for the key port::RWMutex* GetLock(const Slice& key); const InternalKeyComparator& GetInternalKeyComparator() const { return comparator_.comparator; } const MemTableOptions* GetMemTableOptions() const { return &moptions_; } private: enum FlushStateEnum { FLUSH_NOT_REQUESTED, FLUSH_REQUESTED, FLUSH_SCHEDULED }; friend class MemTableIterator; friend class MemTableBackwardIterator; friend class MemTableList; KeyComparator comparator_; const MemTableOptions moptions_; int refs_; const size_t kArenaBlockSize; ConcurrentArena arena_; MemTableAllocator allocator_; unique_ptr table_; // Total data size of all data inserted std::atomic data_size_; std::atomic num_entries_; std::atomic num_deletes_; // These are used to manage memtable flushes to storage bool flush_in_progress_; // started the flush bool flush_completed_; // finished the flush uint64_t file_number_; // filled up after flush is complete // The updates to be applied to the transaction log when this // memtable is flushed to storage. VersionEdit edit_; // The sequence number of the kv that was inserted first std::atomic first_seqno_; // The db sequence number at the time of creation or kMaxSequenceNumber // if not set. std::atomic earliest_seqno_; // The log files earlier than this number can be deleted. uint64_t mem_next_logfile_number_; // rw locks for inplace updates std::vector locks_; const SliceTransform* const prefix_extractor_; std::unique_ptr prefix_bloom_; std::atomic flush_state_; Env* env_; // Returns a heuristic flush decision bool ShouldFlushNow() const; // Updates flush_state_ using ShouldFlushNow() void UpdateFlushState(); // No copying allowed MemTable(const MemTable&); MemTable& operator=(const MemTable&); }; extern const char* EncodeKey(std::string* scratch, const Slice& target); } // namespace rocksdb rocksdb-4.5.1/db/memtable_allocator.cc000066400000000000000000000033621267530634000177040ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/memtable_allocator.h" #include #include "db/writebuffer.h" #include "util/arena.h" namespace rocksdb { MemTableAllocator::MemTableAllocator(Allocator* allocator, WriteBuffer* write_buffer) : allocator_(allocator), write_buffer_(write_buffer), bytes_allocated_(0) {} MemTableAllocator::~MemTableAllocator() { DoneAllocating(); } char* MemTableAllocator::Allocate(size_t bytes) { assert(write_buffer_ != nullptr); bytes_allocated_.fetch_add(bytes, std::memory_order_relaxed); write_buffer_->ReserveMem(bytes); return allocator_->Allocate(bytes); } char* MemTableAllocator::AllocateAligned(size_t bytes, size_t huge_page_size, Logger* logger) { assert(write_buffer_ != nullptr); bytes_allocated_.fetch_add(bytes, std::memory_order_relaxed); write_buffer_->ReserveMem(bytes); return allocator_->AllocateAligned(bytes, huge_page_size, logger); } void MemTableAllocator::DoneAllocating() { if (write_buffer_ != nullptr) { write_buffer_->FreeMem(bytes_allocated_.load(std::memory_order_relaxed)); write_buffer_ = nullptr; } } size_t MemTableAllocator::BlockSize() const { return allocator_->BlockSize(); } } // namespace rocksdb rocksdb-4.5.1/db/memtable_allocator.h000066400000000000000000000030131267530634000175370ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // This is used by the MemTable to allocate write buffer memory. It connects // to WriteBuffer so we can track and enforce overall write buffer limits. #pragma once #include #include "util/allocator.h" namespace rocksdb { class Logger; class WriteBuffer; class MemTableAllocator : public Allocator { public: explicit MemTableAllocator(Allocator* allocator, WriteBuffer* write_buffer); ~MemTableAllocator(); // Allocator interface char* Allocate(size_t bytes) override; char* AllocateAligned(size_t bytes, size_t huge_page_size = 0, Logger* logger = nullptr) override; size_t BlockSize() const override; // Call when we're finished allocating memory so we can free it from // the write buffer's limit. void DoneAllocating(); private: Allocator* allocator_; WriteBuffer* write_buffer_; std::atomic bytes_allocated_; // No copying allowed MemTableAllocator(const MemTableAllocator&); void operator=(const MemTableAllocator&); }; } // namespace rocksdb rocksdb-4.5.1/db/memtable_list.cc000066400000000000000000000321521267530634000166760ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include "db/memtable_list.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include "rocksdb/db.h" #include "db/memtable.h" #include "db/version_set.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "table/merger.h" #include "util/coding.h" #include "util/log_buffer.h" #include "util/thread_status_util.h" namespace rocksdb { class InternalKeyComparator; class Mutex; class VersionSet; void MemTableListVersion::AddMemTable(MemTable* m) { memlist_.push_front(m); *parent_memtable_list_memory_usage_ += m->ApproximateMemoryUsage(); } void MemTableListVersion::UnrefMemTable(autovector* to_delete, MemTable* m) { if (m->Unref()) { to_delete->push_back(m); assert(*parent_memtable_list_memory_usage_ >= m->ApproximateMemoryUsage()); *parent_memtable_list_memory_usage_ -= m->ApproximateMemoryUsage(); } else { } } MemTableListVersion::MemTableListVersion( size_t* parent_memtable_list_memory_usage, MemTableListVersion* old) : max_write_buffer_number_to_maintain_( old->max_write_buffer_number_to_maintain_), parent_memtable_list_memory_usage_(parent_memtable_list_memory_usage) { if (old != nullptr) { memlist_ = old->memlist_; for (auto& m : memlist_) { m->Ref(); } memlist_history_ = old->memlist_history_; for (auto& m : memlist_history_) { m->Ref(); } } } MemTableListVersion::MemTableListVersion( size_t* parent_memtable_list_memory_usage, int max_write_buffer_number_to_maintain) : max_write_buffer_number_to_maintain_(max_write_buffer_number_to_maintain), parent_memtable_list_memory_usage_(parent_memtable_list_memory_usage) {} void MemTableListVersion::Ref() { ++refs_; } // called by superversion::clean() void MemTableListVersion::Unref(autovector* to_delete) { assert(refs_ >= 1); --refs_; if (refs_ == 0) { // if to_delete is equal to nullptr it means we're confident // that refs_ will not be zero assert(to_delete != nullptr); for (const auto& m : memlist_) { UnrefMemTable(to_delete, m); } for (const auto& m : memlist_history_) { UnrefMemTable(to_delete, m); } delete this; } } int MemTableList::NumNotFlushed() const { int size = static_cast(current_->memlist_.size()); assert(num_flush_not_started_ <= size); return size; } int MemTableList::NumFlushed() const { return static_cast(current_->memlist_history_.size()); } // Search all the memtables starting from the most recent one. // Return the most recent value found, if any. // Operands stores the list of merge operations to apply, so far. bool MemTableListVersion::Get(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq) { return GetFromList(&memlist_, key, value, s, merge_context, seq); } bool MemTableListVersion::GetFromHistory(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq) { return GetFromList(&memlist_history_, key, value, s, merge_context, seq); } bool MemTableListVersion::GetFromList(std::list* list, const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq) { *seq = kMaxSequenceNumber; for (auto& memtable : *list) { SequenceNumber current_seq = kMaxSequenceNumber; bool done = memtable->Get(key, value, s, merge_context, ¤t_seq); if (*seq == kMaxSequenceNumber) { // Store the most recent sequence number of any operation on this key. // Since we only care about the most recent change, we only need to // return the first operation found when searching memtables in // reverse-chronological order. *seq = current_seq; } if (done) { assert(*seq != kMaxSequenceNumber); return true; } } return false; } void MemTableListVersion::AddIterators( const ReadOptions& options, std::vector* iterator_list, Arena* arena) { for (auto& m : memlist_) { iterator_list->push_back(m->NewIterator(options, arena)); } } void MemTableListVersion::AddIterators( const ReadOptions& options, MergeIteratorBuilder* merge_iter_builder) { for (auto& m : memlist_) { merge_iter_builder->AddIterator( m->NewIterator(options, merge_iter_builder->GetArena())); } } uint64_t MemTableListVersion::GetTotalNumEntries() const { uint64_t total_num = 0; for (auto& m : memlist_) { total_num += m->num_entries(); } return total_num; } uint64_t MemTableListVersion::ApproximateSize(const Slice& start_ikey, const Slice& end_ikey) { uint64_t total_size = 0; for (auto& m : memlist_) { total_size += m->ApproximateSize(start_ikey, end_ikey); } return total_size; } uint64_t MemTableListVersion::GetTotalNumDeletes() const { uint64_t total_num = 0; for (auto& m : memlist_) { total_num += m->num_deletes(); } return total_num; } SequenceNumber MemTableListVersion::GetEarliestSequenceNumber( bool include_history) const { if (include_history && !memlist_history_.empty()) { return memlist_history_.back()->GetEarliestSequenceNumber(); } else if (!memlist_.empty()) { return memlist_.back()->GetEarliestSequenceNumber(); } else { return kMaxSequenceNumber; } } // caller is responsible for referencing m void MemTableListVersion::Add(MemTable* m, autovector* to_delete) { assert(refs_ == 1); // only when refs_ == 1 is MemTableListVersion mutable AddMemTable(m); TrimHistory(to_delete); } // Removes m from list of memtables not flushed. Caller should NOT Unref m. void MemTableListVersion::Remove(MemTable* m, autovector* to_delete) { assert(refs_ == 1); // only when refs_ == 1 is MemTableListVersion mutable memlist_.remove(m); if (max_write_buffer_number_to_maintain_ > 0) { memlist_history_.push_front(m); TrimHistory(to_delete); } else { UnrefMemTable(to_delete, m); } } // Make sure we don't use up too much space in history void MemTableListVersion::TrimHistory(autovector* to_delete) { while (memlist_.size() + memlist_history_.size() > static_cast(max_write_buffer_number_to_maintain_) && !memlist_history_.empty()) { MemTable* x = memlist_history_.back(); memlist_history_.pop_back(); UnrefMemTable(to_delete, x); } } // Returns true if there is at least one memtable on which flush has // not yet started. bool MemTableList::IsFlushPending() const { if ((flush_requested_ && num_flush_not_started_ >= 1) || (num_flush_not_started_ >= min_write_buffer_number_to_merge_)) { assert(imm_flush_needed.load(std::memory_order_relaxed)); return true; } return false; } // Returns the memtables that need to be flushed. void MemTableList::PickMemtablesToFlush(autovector* ret) { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_PICK_MEMTABLES_TO_FLUSH); const auto& memlist = current_->memlist_; for (auto it = memlist.rbegin(); it != memlist.rend(); ++it) { MemTable* m = *it; if (!m->flush_in_progress_) { assert(!m->flush_completed_); num_flush_not_started_--; if (num_flush_not_started_ == 0) { imm_flush_needed.store(false, std::memory_order_release); } m->flush_in_progress_ = true; // flushing will start very soon ret->push_back(m); } } flush_requested_ = false; // start-flush request is complete } void MemTableList::RollbackMemtableFlush(const autovector& mems, uint64_t file_number) { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_MEMTABLE_ROLLBACK); assert(!mems.empty()); // If the flush was not successful, then just reset state. // Maybe a succeeding attempt to flush will be successful. for (MemTable* m : mems) { assert(m->flush_in_progress_); assert(m->file_number_ == 0); m->flush_in_progress_ = false; m->flush_completed_ = false; m->edit_.Clear(); num_flush_not_started_++; } imm_flush_needed.store(true, std::memory_order_release); } // Record a successful flush in the manifest file Status MemTableList::InstallMemtableFlushResults( ColumnFamilyData* cfd, const MutableCFOptions& mutable_cf_options, const autovector& mems, VersionSet* vset, InstrumentedMutex* mu, uint64_t file_number, autovector* to_delete, Directory* db_directory, LogBuffer* log_buffer) { AutoThreadOperationStageUpdater stage_updater( ThreadStatus::STAGE_MEMTABLE_INSTALL_FLUSH_RESULTS); mu->AssertHeld(); // flush was successful for (size_t i = 0; i < mems.size(); ++i) { // All the edits are associated with the first memtable of this batch. assert(i == 0 || mems[i]->GetEdits()->NumEntries() == 0); mems[i]->flush_completed_ = true; mems[i]->file_number_ = file_number; } // if some other thread is already committing, then return Status s; if (commit_in_progress_) { return s; } // Only a single thread can be executing this piece of code commit_in_progress_ = true; // scan all memtables from the earliest, and commit those // (in that order) that have finished flushing. Memetables // are always committed in the order that they were created. while (!current_->memlist_.empty() && s.ok()) { MemTable* m = current_->memlist_.back(); // get the last element if (!m->flush_completed_) { break; } LogToBuffer(log_buffer, "[%s] Level-0 commit table #%" PRIu64 " started", cfd->GetName().c_str(), m->file_number_); // this can release and reacquire the mutex. s = vset->LogAndApply(cfd, mutable_cf_options, &m->edit_, mu, db_directory); // we will be changing the version in the next code path, // so we better create a new one, since versions are immutable InstallNewVersion(); // All the later memtables that have the same filenum // are part of the same batch. They can be committed now. uint64_t mem_id = 1; // how many memtables have been flushed. do { if (s.ok()) { // commit new state LogToBuffer(log_buffer, "[%s] Level-0 commit table #%" PRIu64 ": memtable #%" PRIu64 " done", cfd->GetName().c_str(), m->file_number_, mem_id); assert(m->file_number_ > 0); current_->Remove(m, to_delete); } else { // commit failed. setup state so that we can flush again. LogToBuffer(log_buffer, "Level-0 commit table #%" PRIu64 ": memtable #%" PRIu64 " failed", m->file_number_, mem_id); m->flush_completed_ = false; m->flush_in_progress_ = false; m->edit_.Clear(); num_flush_not_started_++; m->file_number_ = 0; imm_flush_needed.store(true, std::memory_order_release); } ++mem_id; } while (!current_->memlist_.empty() && (m = current_->memlist_.back()) && m->file_number_ == file_number); } commit_in_progress_ = false; return s; } // New memtables are inserted at the front of the list. void MemTableList::Add(MemTable* m, autovector* to_delete) { assert(static_cast(current_->memlist_.size()) >= num_flush_not_started_); InstallNewVersion(); // this method is used to move mutable memtable into an immutable list. // since mutable memtable is already refcounted by the DBImpl, // and when moving to the imutable list we don't unref it, // we don't have to ref the memtable here. we just take over the // reference from the DBImpl. current_->Add(m, to_delete); m->MarkImmutable(); num_flush_not_started_++; if (num_flush_not_started_ == 1) { imm_flush_needed.store(true, std::memory_order_release); } } // Returns an estimate of the number of bytes of data in use. size_t MemTableList::ApproximateUnflushedMemTablesMemoryUsage() { size_t total_size = 0; for (auto& memtable : current_->memlist_) { total_size += memtable->ApproximateMemoryUsage(); } return total_size; } size_t MemTableList::ApproximateMemoryUsage() { return current_memory_usage_; } void MemTableList::InstallNewVersion() { if (current_->refs_ == 1) { // we're the only one using the version, just keep using it } else { // somebody else holds the current version, we need to create new one MemTableListVersion* version = current_; current_ = new MemTableListVersion(¤t_memory_usage_, current_); current_->Ref(); version->Unref(); } } } // namespace rocksdb rocksdb-4.5.1/db/memtable_list.h000066400000000000000000000210351267530634000165360ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #pragma once #include #include #include #include #include #include "db/dbformat.h" #include "db/filename.h" #include "db/memtable.h" #include "db/skiplist.h" #include "rocksdb/db.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "rocksdb/types.h" #include "util/autovector.h" #include "util/instrumented_mutex.h" #include "util/log_buffer.h" namespace rocksdb { class ColumnFamilyData; class InternalKeyComparator; class InstrumentedMutex; class MergeIteratorBuilder; // keeps a list of immutable memtables in a vector. the list is immutable // if refcount is bigger than one. It is used as a state for Get() and // Iterator code paths // // This class is not thread-safe. External synchronization is required // (such as holding the db mutex or being on the write thread). class MemTableListVersion { public: explicit MemTableListVersion(size_t* parent_memtable_list_memory_usage, MemTableListVersion* old = nullptr); explicit MemTableListVersion(size_t* parent_memtable_list_memory_usage, int max_write_buffer_number_to_maintain); void Ref(); void Unref(autovector* to_delete = nullptr); // Search all the memtables starting from the most recent one. // Return the most recent value found, if any. // // If any operation was found for this key, its most recent sequence number // will be stored in *seq on success (regardless of whether true/false is // returned). Otherwise, *seq will be set to kMaxSequenceNumber. bool Get(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq); bool Get(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context) { SequenceNumber seq; return Get(key, value, s, merge_context, &seq); } // Similar to Get(), but searches the Memtable history of memtables that // have already been flushed. Should only be used from in-memory only // queries (such as Transaction validation) as the history may contain // writes that are also present in the SST files. bool GetFromHistory(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq); bool GetFromHistory(const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context) { SequenceNumber seq; return GetFromHistory(key, value, s, merge_context, &seq); } void AddIterators(const ReadOptions& options, std::vector* iterator_list, Arena* arena); void AddIterators(const ReadOptions& options, MergeIteratorBuilder* merge_iter_builder); uint64_t GetTotalNumEntries() const; uint64_t GetTotalNumDeletes() const; uint64_t ApproximateSize(const Slice& start_ikey, const Slice& end_ikey); // Returns the value of MemTable::GetEarliestSequenceNumber() on the most // recent MemTable in this list or kMaxSequenceNumber if the list is empty. // If include_history=true, will also search Memtables in MemTableList // History. SequenceNumber GetEarliestSequenceNumber(bool include_history = false) const; private: // REQUIRE: m is an immutable memtable void Add(MemTable* m, autovector* to_delete); // REQUIRE: m is an immutable memtable void Remove(MemTable* m, autovector* to_delete); void TrimHistory(autovector* to_delete); bool GetFromList(std::list* list, const LookupKey& key, std::string* value, Status* s, MergeContext* merge_context, SequenceNumber* seq); void AddMemTable(MemTable* m); void UnrefMemTable(autovector* to_delete, MemTable* m); friend class MemTableList; // Immutable MemTables that have not yet been flushed. std::list memlist_; // MemTables that have already been flushed // (used during Transaction validation) std::list memlist_history_; // Maximum number of MemTables to keep in memory (including both flushed // and not-yet-flushed tables). const int max_write_buffer_number_to_maintain_; int refs_ = 0; size_t* parent_memtable_list_memory_usage_; }; // This class stores references to all the immutable memtables. // The memtables are flushed to L0 as soon as possible and in // any order. If there are more than one immutable memtable, their // flushes can occur concurrently. However, they are 'committed' // to the manifest in FIFO order to maintain correctness and // recoverability from a crash. // // // Other than imm_flush_needed, this class is not thread-safe and requires // external synchronization (such as holding the db mutex or being on the // write thread.) class MemTableList { public: // A list of memtables. explicit MemTableList(int min_write_buffer_number_to_merge, int max_write_buffer_number_to_maintain) : imm_flush_needed(false), min_write_buffer_number_to_merge_(min_write_buffer_number_to_merge), current_(new MemTableListVersion(¤t_memory_usage_, max_write_buffer_number_to_maintain)), num_flush_not_started_(0), commit_in_progress_(false), flush_requested_(false) { current_->Ref(); current_memory_usage_ = 0; } // Should not delete MemTableList without making sure MemTableList::current() // is Unref()'d. ~MemTableList() {} MemTableListVersion* current() { return current_; } // so that background threads can detect non-nullptr pointer to // determine whether there is anything more to start flushing. std::atomic imm_flush_needed; // Returns the total number of memtables in the list that haven't yet // been flushed and logged. int NumNotFlushed() const; // Returns total number of memtables in the list that have been // completely flushed and logged. int NumFlushed() const; // Returns true if there is at least one memtable on which flush has // not yet started. bool IsFlushPending() const; // Returns the earliest memtables that needs to be flushed. The returned // memtables are guaranteed to be in the ascending order of created time. void PickMemtablesToFlush(autovector* mems); // Reset status of the given memtable list back to pending state so that // they can get picked up again on the next round of flush. void RollbackMemtableFlush(const autovector& mems, uint64_t file_number); // Commit a successful flush in the manifest file Status InstallMemtableFlushResults( ColumnFamilyData* cfd, const MutableCFOptions& mutable_cf_options, const autovector& m, VersionSet* vset, InstrumentedMutex* mu, uint64_t file_number, autovector* to_delete, Directory* db_directory, LogBuffer* log_buffer); // New memtables are inserted at the front of the list. // Takes ownership of the referenced held on *m by the caller of Add(). void Add(MemTable* m, autovector* to_delete); // Returns an estimate of the number of bytes of data in use. size_t ApproximateMemoryUsage(); // Returns an estimate of the number of bytes of data used by // the unflushed mem-tables. size_t ApproximateUnflushedMemTablesMemoryUsage(); // Request a flush of all existing memtables to storage. This will // cause future calls to IsFlushPending() to return true if this list is // non-empty (regardless of the min_write_buffer_number_to_merge // parameter). This flush request will persist until the next time // PickMemtablesToFlush() is called. void FlushRequested() { flush_requested_ = true; } // Copying allowed // MemTableList(const MemTableList&); // void operator=(const MemTableList&); size_t* current_memory_usage() { return ¤t_memory_usage_; } private: // DB mutex held void InstallNewVersion(); const int min_write_buffer_number_to_merge_; MemTableListVersion* current_; // the number of elements that still need flushing int num_flush_not_started_; // committing in progress bool commit_in_progress_; // Requested a flush of all memtables to storage bool flush_requested_; // The current memory usage. size_t current_memory_usage_; }; } // namespace rocksdb rocksdb-4.5.1/db/memtable_list_test.cc000066400000000000000000000470701267530634000177420ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include "db/memtable_list.h" #include "db/merge_context.h" #include "db/version_set.h" #include "db/write_controller.h" #include "db/writebuffer.h" #include "rocksdb/db.h" #include "rocksdb/status.h" #include "util/testutil.h" #include "util/string_util.h" #include "util/testharness.h" namespace rocksdb { class MemTableListTest : public testing::Test { public: std::string dbname; DB* db; Options options; MemTableListTest() : db(nullptr) { dbname = test::TmpDir() + "/memtable_list_test"; } // Create a test db if not yet created void CreateDB() { if (db == nullptr) { options.create_if_missing = true; DestroyDB(dbname, options); Status s = DB::Open(options, dbname, &db); EXPECT_OK(s); } } ~MemTableListTest() { if (db) { delete db; DestroyDB(dbname, options); } } // Calls MemTableList::InstallMemtableFlushResults() and sets up all // structures needed to call this function. Status Mock_InstallMemtableFlushResults( MemTableList* list, const MutableCFOptions& mutable_cf_options, const autovector& m, autovector* to_delete) { // Create a mock Logger test::NullLogger logger; LogBuffer log_buffer(DEBUG_LEVEL, &logger); // Create a mock VersionSet DBOptions db_options; EnvOptions env_options; shared_ptr table_cache(NewLRUCache(50000, 16)); WriteBuffer write_buffer(db_options.db_write_buffer_size); WriteController write_controller(10000000u); CreateDB(); VersionSet versions(dbname, &db_options, env_options, table_cache.get(), &write_buffer, &write_controller); // Create mock default ColumnFamilyData ColumnFamilyOptions cf_options; std::vector column_families; column_families.emplace_back(kDefaultColumnFamilyName, cf_options); EXPECT_OK(versions.Recover(column_families, false)); auto column_family_set = versions.GetColumnFamilySet(); auto cfd = column_family_set->GetColumnFamily(0); EXPECT_TRUE(cfd != nullptr); // Create dummy mutex. InstrumentedMutex mutex; InstrumentedMutexLock l(&mutex); return list->InstallMemtableFlushResults(cfd, mutable_cf_options, m, &versions, &mutex, 1, to_delete, nullptr, &log_buffer); } }; TEST_F(MemTableListTest, Empty) { // Create an empty MemTableList and validate basic functions. MemTableList list(1, 0); ASSERT_EQ(0, list.NumNotFlushed()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); ASSERT_FALSE(list.IsFlushPending()); autovector mems; list.PickMemtablesToFlush(&mems); ASSERT_EQ(0, mems.size()); autovector to_delete; list.current()->Unref(&to_delete); ASSERT_EQ(0, to_delete.size()); } TEST_F(MemTableListTest, GetTest) { // Create MemTableList int min_write_buffer_number_to_merge = 2; int max_write_buffer_number_to_maintain = 0; MemTableList list(min_write_buffer_number_to_merge, max_write_buffer_number_to_maintain); SequenceNumber seq = 1; std::string value; Status s; MergeContext merge_context; autovector to_delete; LookupKey lkey("key1", seq); bool found = list.current()->Get(lkey, &value, &s, &merge_context); ASSERT_FALSE(found); // Create a MemTable InternalKeyComparator cmp(BytewiseComparator()); auto factory = std::make_shared(); options.memtable_factory = factory; ImmutableCFOptions ioptions(options); WriteBuffer wb(options.db_write_buffer_size); MemTable* mem = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb, kMaxSequenceNumber); mem->Ref(); // Write some keys to this memtable. mem->Add(++seq, kTypeDeletion, "key1", ""); mem->Add(++seq, kTypeValue, "key2", "value2"); mem->Add(++seq, kTypeValue, "key1", "value1"); mem->Add(++seq, kTypeValue, "key2", "value2.2"); // Fetch the newly written keys merge_context.Clear(); found = mem->Get(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_TRUE(s.ok() && found); ASSERT_EQ(value, "value1"); merge_context.Clear(); found = mem->Get(LookupKey("key1", 2), &value, &s, &merge_context); // MemTable found out that this key is *not* found (at this sequence#) ASSERT_TRUE(found && s.IsNotFound()); merge_context.Clear(); found = mem->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_TRUE(s.ok() && found); ASSERT_EQ(value, "value2.2"); ASSERT_EQ(4, mem->num_entries()); ASSERT_EQ(1, mem->num_deletes()); // Add memtable to list list.Add(mem, &to_delete); SequenceNumber saved_seq = seq; // Create another memtable and write some keys to it WriteBuffer wb2(options.db_write_buffer_size); MemTable* mem2 = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb2, kMaxSequenceNumber); mem2->Ref(); mem2->Add(++seq, kTypeDeletion, "key1", ""); mem2->Add(++seq, kTypeValue, "key2", "value2.3"); // Add second memtable to list list.Add(mem2, &to_delete); // Fetch keys via MemTableList merge_context.Clear(); found = list.current()->Get(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_TRUE(found && s.IsNotFound()); merge_context.Clear(); found = list.current()->Get(LookupKey("key1", saved_seq), &value, &s, &merge_context); ASSERT_TRUE(s.ok() && found); ASSERT_EQ("value1", value); merge_context.Clear(); found = list.current()->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_TRUE(s.ok() && found); ASSERT_EQ(value, "value2.3"); merge_context.Clear(); found = list.current()->Get(LookupKey("key2", 1), &value, &s, &merge_context); ASSERT_FALSE(found); ASSERT_EQ(2, list.NumNotFlushed()); list.current()->Unref(&to_delete); for (MemTable* m : to_delete) { delete m; } } TEST_F(MemTableListTest, GetFromHistoryTest) { // Create MemTableList int min_write_buffer_number_to_merge = 2; int max_write_buffer_number_to_maintain = 2; MemTableList list(min_write_buffer_number_to_merge, max_write_buffer_number_to_maintain); SequenceNumber seq = 1; std::string value; Status s; MergeContext merge_context; autovector to_delete; LookupKey lkey("key1", seq); bool found = list.current()->Get(lkey, &value, &s, &merge_context); ASSERT_FALSE(found); // Create a MemTable InternalKeyComparator cmp(BytewiseComparator()); auto factory = std::make_shared(); options.memtable_factory = factory; ImmutableCFOptions ioptions(options); WriteBuffer wb(options.db_write_buffer_size); MemTable* mem = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb, kMaxSequenceNumber); mem->Ref(); // Write some keys to this memtable. mem->Add(++seq, kTypeDeletion, "key1", ""); mem->Add(++seq, kTypeValue, "key2", "value2"); mem->Add(++seq, kTypeValue, "key2", "value2.2"); // Fetch the newly written keys merge_context.Clear(); found = mem->Get(LookupKey("key1", seq), &value, &s, &merge_context); // MemTable found out that this key is *not* found (at this sequence#) ASSERT_TRUE(found && s.IsNotFound()); merge_context.Clear(); found = mem->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_TRUE(s.ok() && found); ASSERT_EQ(value, "value2.2"); // Add memtable to list list.Add(mem, &to_delete); ASSERT_EQ(0, to_delete.size()); // Fetch keys via MemTableList merge_context.Clear(); found = list.current()->Get(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_TRUE(found && s.IsNotFound()); merge_context.Clear(); found = list.current()->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_TRUE(s.ok() && found); ASSERT_EQ("value2.2", value); // Flush this memtable from the list. // (It will then be a part of the memtable history). autovector to_flush; list.PickMemtablesToFlush(&to_flush); ASSERT_EQ(1, to_flush.size()); s = Mock_InstallMemtableFlushResults( &list, MutableCFOptions(options, ioptions), to_flush, &to_delete); ASSERT_OK(s); ASSERT_EQ(0, list.NumNotFlushed()); ASSERT_EQ(1, list.NumFlushed()); ASSERT_EQ(0, to_delete.size()); // Verify keys are no longer in MemTableList merge_context.Clear(); found = list.current()->Get(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_FALSE(found); merge_context.Clear(); found = list.current()->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_FALSE(found); // Verify keys are present in history merge_context.Clear(); found = list.current()->GetFromHistory(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_TRUE(found && s.IsNotFound()); merge_context.Clear(); found = list.current()->GetFromHistory(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_TRUE(found); ASSERT_EQ("value2.2", value); // Create another memtable and write some keys to it WriteBuffer wb2(options.db_write_buffer_size); MemTable* mem2 = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb2, kMaxSequenceNumber); mem2->Ref(); mem2->Add(++seq, kTypeDeletion, "key1", ""); mem2->Add(++seq, kTypeValue, "key3", "value3"); // Add second memtable to list list.Add(mem2, &to_delete); ASSERT_EQ(0, to_delete.size()); to_flush.clear(); list.PickMemtablesToFlush(&to_flush); ASSERT_EQ(1, to_flush.size()); // Flush second memtable s = Mock_InstallMemtableFlushResults( &list, MutableCFOptions(options, ioptions), to_flush, &to_delete); ASSERT_OK(s); ASSERT_EQ(0, list.NumNotFlushed()); ASSERT_EQ(2, list.NumFlushed()); ASSERT_EQ(0, to_delete.size()); // Add a third memtable to push the first memtable out of the history WriteBuffer wb3(options.db_write_buffer_size); MemTable* mem3 = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb3, kMaxSequenceNumber); mem3->Ref(); list.Add(mem3, &to_delete); ASSERT_EQ(1, list.NumNotFlushed()); ASSERT_EQ(1, list.NumFlushed()); ASSERT_EQ(1, to_delete.size()); // Verify keys are no longer in MemTableList merge_context.Clear(); found = list.current()->Get(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_FALSE(found); merge_context.Clear(); found = list.current()->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_FALSE(found); merge_context.Clear(); found = list.current()->Get(LookupKey("key3", seq), &value, &s, &merge_context); ASSERT_FALSE(found); // Verify that the second memtable's keys are in the history merge_context.Clear(); found = list.current()->GetFromHistory(LookupKey("key1", seq), &value, &s, &merge_context); ASSERT_TRUE(found && s.IsNotFound()); merge_context.Clear(); found = list.current()->GetFromHistory(LookupKey("key3", seq), &value, &s, &merge_context); ASSERT_TRUE(found); ASSERT_EQ("value3", value); // Verify that key2 from the first memtable is no longer in the history merge_context.Clear(); found = list.current()->Get(LookupKey("key2", seq), &value, &s, &merge_context); ASSERT_FALSE(found); // Cleanup list.current()->Unref(&to_delete); ASSERT_EQ(3, to_delete.size()); for (MemTable* m : to_delete) { delete m; } } TEST_F(MemTableListTest, FlushPendingTest) { const int num_tables = 5; SequenceNumber seq = 1; Status s; auto factory = std::make_shared(); options.memtable_factory = factory; ImmutableCFOptions ioptions(options); InternalKeyComparator cmp(BytewiseComparator()); WriteBuffer wb(options.db_write_buffer_size); autovector to_delete; // Create MemTableList int min_write_buffer_number_to_merge = 3; int max_write_buffer_number_to_maintain = 7; MemTableList list(min_write_buffer_number_to_merge, max_write_buffer_number_to_maintain); // Create some MemTables std::vector tables; MutableCFOptions mutable_cf_options(options, ioptions); for (int i = 0; i < num_tables; i++) { MemTable* mem = new MemTable(cmp, ioptions, mutable_cf_options, &wb, kMaxSequenceNumber); mem->Ref(); std::string value; MergeContext merge_context; mem->Add(++seq, kTypeValue, "key1", ToString(i)); mem->Add(++seq, kTypeValue, "keyN" + ToString(i), "valueN"); mem->Add(++seq, kTypeValue, "keyX" + ToString(i), "value"); mem->Add(++seq, kTypeValue, "keyM" + ToString(i), "valueM"); mem->Add(++seq, kTypeDeletion, "keyX" + ToString(i), ""); tables.push_back(mem); } // Nothing to flush ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); autovector to_flush; list.PickMemtablesToFlush(&to_flush); ASSERT_EQ(0, to_flush.size()); // Request a flush even though there is nothing to flush list.FlushRequested(); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Attempt to 'flush' to clear request for flush list.PickMemtablesToFlush(&to_flush); ASSERT_EQ(0, to_flush.size()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Request a flush again list.FlushRequested(); // No flush pending since the list is empty. ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Add 2 tables list.Add(tables[0], &to_delete); list.Add(tables[1], &to_delete); ASSERT_EQ(2, list.NumNotFlushed()); ASSERT_EQ(0, to_delete.size()); // Even though we have less than the minimum to flush, a flush is // pending since we had previously requested a flush and never called // PickMemtablesToFlush() to clear the flush. ASSERT_TRUE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); // Pick tables to flush list.PickMemtablesToFlush(&to_flush); ASSERT_EQ(2, to_flush.size()); ASSERT_EQ(2, list.NumNotFlushed()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Revert flush list.RollbackMemtableFlush(to_flush, 0); ASSERT_FALSE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); to_flush.clear(); // Add another table list.Add(tables[2], &to_delete); // We now have the minimum to flush regardles of whether FlushRequested() // was called. ASSERT_TRUE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); ASSERT_EQ(0, to_delete.size()); // Pick tables to flush list.PickMemtablesToFlush(&to_flush); ASSERT_EQ(3, to_flush.size()); ASSERT_EQ(3, list.NumNotFlushed()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Pick tables to flush again autovector to_flush2; list.PickMemtablesToFlush(&to_flush2); ASSERT_EQ(0, to_flush2.size()); ASSERT_EQ(3, list.NumNotFlushed()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Add another table list.Add(tables[3], &to_delete); ASSERT_FALSE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); ASSERT_EQ(0, to_delete.size()); // Request a flush again list.FlushRequested(); ASSERT_TRUE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); // Pick tables to flush again list.PickMemtablesToFlush(&to_flush2); ASSERT_EQ(1, to_flush2.size()); ASSERT_EQ(4, list.NumNotFlushed()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Rollback first pick of tables list.RollbackMemtableFlush(to_flush, 0); ASSERT_TRUE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); to_flush.clear(); // Add another tables list.Add(tables[4], &to_delete); ASSERT_EQ(5, list.NumNotFlushed()); // We now have the minimum to flush regardles of whether FlushRequested() ASSERT_TRUE(list.IsFlushPending()); ASSERT_TRUE(list.imm_flush_needed.load(std::memory_order_acquire)); ASSERT_EQ(0, to_delete.size()); // Pick tables to flush list.PickMemtablesToFlush(&to_flush); // Should pick 4 of 5 since 1 table has been picked in to_flush2 ASSERT_EQ(4, to_flush.size()); ASSERT_EQ(5, list.NumNotFlushed()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Pick tables to flush again autovector to_flush3; ASSERT_EQ(0, to_flush3.size()); // nothing not in progress of being flushed ASSERT_EQ(5, list.NumNotFlushed()); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Flush the 4 memtables that were picked in to_flush s = Mock_InstallMemtableFlushResults( &list, MutableCFOptions(options, ioptions), to_flush, &to_delete); ASSERT_OK(s); // Note: now to_flush contains tables[0,1,2,4]. to_flush2 contains // tables[3]. // Current implementation will only commit memtables in the order they were // created. So InstallMemtableFlushResults will install the first 3 tables // in to_flush and stop when it encounters a table not yet flushed. ASSERT_EQ(2, list.NumNotFlushed()); int num_in_history = std::min(3, max_write_buffer_number_to_maintain); ASSERT_EQ(num_in_history, list.NumFlushed()); ASSERT_EQ(5 - list.NumNotFlushed() - num_in_history, to_delete.size()); // Request a flush again. Should be nothing to flush list.FlushRequested(); ASSERT_FALSE(list.IsFlushPending()); ASSERT_FALSE(list.imm_flush_needed.load(std::memory_order_acquire)); // Flush the 1 memtable that was picked in to_flush2 s = MemTableListTest::Mock_InstallMemtableFlushResults( &list, MutableCFOptions(options, ioptions), to_flush2, &to_delete); ASSERT_OK(s); // This will actually install 2 tables. The 1 we told it to flush, and also // tables[4] which has been waiting for tables[3] to commit. ASSERT_EQ(0, list.NumNotFlushed()); num_in_history = std::min(5, max_write_buffer_number_to_maintain); ASSERT_EQ(num_in_history, list.NumFlushed()); ASSERT_EQ(5 - list.NumNotFlushed() - num_in_history, to_delete.size()); for (const auto& m : to_delete) { // Refcount should be 0 after calling InstallMemtableFlushResults. // Verify this, by Ref'ing then UnRef'ing: m->Ref(); ASSERT_EQ(m, m->Unref()); delete m; } to_delete.clear(); list.current()->Unref(&to_delete); int to_delete_size = std::min(5, max_write_buffer_number_to_maintain); ASSERT_EQ(to_delete_size, to_delete.size()); for (const auto& m : to_delete) { // Refcount should be 0 after calling InstallMemtableFlushResults. // Verify this, by Ref'ing then UnRef'ing: m->Ref(); ASSERT_EQ(m, m->Unref()); delete m; } to_delete.clear(); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/memtablerep_bench.cc000066400000000000000000000602361267530634000175150ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #define __STDC_FORMAT_MACROS #ifndef GFLAGS #include int main() { fprintf(stderr, "Please install gflags to run rocksdb tools\n"); return 1; } #else #include #include #include #include #include #include #include #include "db/dbformat.h" #include "db/memtable.h" #include "db/writebuffer.h" #include "port/port.h" #include "port/stack_trace.h" #include "rocksdb/comparator.h" #include "rocksdb/memtablerep.h" #include "rocksdb/options.h" #include "rocksdb/slice_transform.h" #include "util/arena.h" #include "util/mutexlock.h" #include "util/stop_watch.h" #include "util/testutil.h" using GFLAGS::ParseCommandLineFlags; using GFLAGS::RegisterFlagValidator; using GFLAGS::SetUsageMessage; DEFINE_string(benchmarks, "fillrandom", "Comma-separated list of benchmarks to run. Options:\n" "\tfillrandom -- write N random values\n" "\tfillseq -- write N values in sequential order\n" "\treadrandom -- read N values in random order\n" "\treadseq -- scan the DB\n" "\treadwrite -- 1 thread writes while N - 1 threads " "do random\n" "\t reads\n" "\tseqreadwrite -- 1 thread writes while N - 1 threads " "do scans\n"); DEFINE_string(memtablerep, "skiplist", "Which implementation of memtablerep to use. See " "include/memtablerep.h for\n" " more details. Options:\n" "\tskiplist -- backed by a skiplist\n" "\tvector -- backed by an std::vector\n" "\thashskiplist -- backed by a hash skip list\n" "\thashlinklist -- backed by a hash linked list\n" "\tcuckoo -- backed by a cuckoo hash table"); DEFINE_int64(bucket_count, 1000000, "bucket_count parameter to pass into NewHashSkiplistRepFactory or " "NewHashLinkListRepFactory"); DEFINE_int32( hashskiplist_height, 4, "skiplist_height parameter to pass into NewHashSkiplistRepFactory"); DEFINE_int32( hashskiplist_branching_factor, 4, "branching_factor parameter to pass into NewHashSkiplistRepFactory"); DEFINE_int32( huge_page_tlb_size, 0, "huge_page_tlb_size parameter to pass into NewHashLinkListRepFactory"); DEFINE_int32(bucket_entries_logging_threshold, 4096, "bucket_entries_logging_threshold parameter to pass into " "NewHashLinkListRepFactory"); DEFINE_bool(if_log_bucket_dist_when_flash, true, "if_log_bucket_dist_when_flash parameter to pass into " "NewHashLinkListRepFactory"); DEFINE_int32( threshold_use_skiplist, 256, "threshold_use_skiplist parameter to pass into NewHashLinkListRepFactory"); DEFINE_int64( write_buffer_size, 256, "write_buffer_size parameter to pass into NewHashCuckooRepFactory"); DEFINE_int64( average_data_size, 64, "average_data_size parameter to pass into NewHashCuckooRepFactory"); DEFINE_int64( hash_function_count, 4, "hash_function_count parameter to pass into NewHashCuckooRepFactory"); DEFINE_int32( num_threads, 1, "Number of concurrent threads to run. If the benchmark includes writes,\n" "then at most one thread will be a writer"); DEFINE_int32(num_operations, 1000000, "Number of operations to do for write and random read benchmarks"); DEFINE_int32(num_scans, 10, "Number of times for each thread to scan the memtablerep for " "sequential read " "benchmarks"); DEFINE_int32(item_size, 100, "Number of bytes each item should be"); DEFINE_int32(prefix_length, 8, "Prefix length to pass into NewFixedPrefixTransform"); /* VectorRep settings */ DEFINE_int64(vectorrep_count, 0, "Number of entries to reserve on VectorRep initialization"); DEFINE_int64(seed, 0, "Seed base for random number generators. " "When 0 it is deterministic."); namespace rocksdb { namespace { struct CallbackVerifyArgs { bool found; LookupKey* key; MemTableRep* table; InternalKeyComparator* comparator; }; } // namespace // Helper for quickly generating random data. class RandomGenerator { private: std::string data_; unsigned int pos_; public: RandomGenerator() { Random rnd(301); auto size = (unsigned)std::max(1048576, FLAGS_item_size); test::RandomString(&rnd, size, &data_); pos_ = 0; } Slice Generate(unsigned int len) { assert(len <= data_.size()); if (pos_ + len > data_.size()) { pos_ = 0; } pos_ += len; return Slice(data_.data() + pos_ - len, len); } }; enum WriteMode { SEQUENTIAL, RANDOM, UNIQUE_RANDOM }; class KeyGenerator { public: KeyGenerator(Random64* rand, WriteMode mode, uint64_t num) : rand_(rand), mode_(mode), num_(num), next_(0) { if (mode_ == UNIQUE_RANDOM) { // NOTE: if memory consumption of this approach becomes a concern, // we can either break it into pieces and only random shuffle a section // each time. Alternatively, use a bit map implementation // (https://reviews.facebook.net/differential/diff/54627/) values_.resize(num_); for (uint64_t i = 0; i < num_; ++i) { values_[i] = i; } std::shuffle( values_.begin(), values_.end(), std::default_random_engine(static_cast(FLAGS_seed))); } } uint64_t Next() { switch (mode_) { case SEQUENTIAL: return next_++; case RANDOM: return rand_->Next() % num_; case UNIQUE_RANDOM: return values_[next_++]; } assert(false); return std::numeric_limits::max(); } private: Random64* rand_; WriteMode mode_; const uint64_t num_; uint64_t next_; std::vector values_; }; class BenchmarkThread { public: explicit BenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits) : table_(table), key_gen_(key_gen), bytes_written_(bytes_written), bytes_read_(bytes_read), sequence_(sequence), num_ops_(num_ops), read_hits_(read_hits) {} virtual void operator()() = 0; virtual ~BenchmarkThread() {} protected: MemTableRep* table_; KeyGenerator* key_gen_; uint64_t* bytes_written_; uint64_t* bytes_read_; uint64_t* sequence_; uint64_t num_ops_; uint64_t* read_hits_; RandomGenerator generator_; }; class FillBenchmarkThread : public BenchmarkThread { public: FillBenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits) : BenchmarkThread(table, key_gen, bytes_written, bytes_read, sequence, num_ops, read_hits) {} void FillOne() { char* buf = nullptr; auto internal_key_size = 16; auto encoded_len = FLAGS_item_size + VarintLength(internal_key_size) + internal_key_size; KeyHandle handle = table_->Allocate(encoded_len, &buf); assert(buf != nullptr); char* p = EncodeVarint32(buf, internal_key_size); auto key = key_gen_->Next(); EncodeFixed64(p, key); p += 8; EncodeFixed64(p, ++(*sequence_)); p += 8; Slice bytes = generator_.Generate(FLAGS_item_size); memcpy(p, bytes.data(), FLAGS_item_size); p += FLAGS_item_size; assert(p == buf + encoded_len); table_->Insert(handle); *bytes_written_ += encoded_len; } void operator()() override { for (unsigned int i = 0; i < num_ops_; ++i) { FillOne(); } } }; class ConcurrentFillBenchmarkThread : public FillBenchmarkThread { public: ConcurrentFillBenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits, std::atomic_int* threads_done) : FillBenchmarkThread(table, key_gen, bytes_written, bytes_read, sequence, num_ops, read_hits) { threads_done_ = threads_done; } void operator()() override { // # of read threads will be total threads - write threads (always 1). Loop // while all reads complete. while ((*threads_done_).load() < (FLAGS_num_threads - 1)) { FillOne(); } } private: std::atomic_int* threads_done_; }; class ReadBenchmarkThread : public BenchmarkThread { public: ReadBenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits) : BenchmarkThread(table, key_gen, bytes_written, bytes_read, sequence, num_ops, read_hits) {} static bool callback(void* arg, const char* entry) { CallbackVerifyArgs* callback_args = static_cast(arg); assert(callback_args != nullptr); uint32_t key_length; const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length); if ((callback_args->comparator) ->user_comparator() ->Equal(Slice(key_ptr, key_length - 8), callback_args->key->user_key())) { callback_args->found = true; } return false; } void ReadOne() { std::string user_key; auto key = key_gen_->Next(); PutFixed64(&user_key, key); LookupKey lookup_key(user_key, *sequence_); InternalKeyComparator internal_key_comp(BytewiseComparator()); CallbackVerifyArgs verify_args; verify_args.found = false; verify_args.key = &lookup_key; verify_args.table = table_; verify_args.comparator = &internal_key_comp; table_->Get(lookup_key, &verify_args, callback); if (verify_args.found) { *bytes_read_ += VarintLength(16) + 16 + FLAGS_item_size; ++*read_hits_; } } void operator()() override { for (unsigned int i = 0; i < num_ops_; ++i) { ReadOne(); } } }; class SeqReadBenchmarkThread : public BenchmarkThread { public: SeqReadBenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits) : BenchmarkThread(table, key_gen, bytes_written, bytes_read, sequence, num_ops, read_hits) {} void ReadOneSeq() { std::unique_ptr iter(table_->GetIterator()); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { // pretend to read the value *bytes_read_ += VarintLength(16) + 16 + FLAGS_item_size; } ++*read_hits_; } void operator()() override { for (unsigned int i = 0; i < num_ops_; ++i) { { ReadOneSeq(); } } } }; class ConcurrentReadBenchmarkThread : public ReadBenchmarkThread { public: ConcurrentReadBenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits, std::atomic_int* threads_done) : ReadBenchmarkThread(table, key_gen, bytes_written, bytes_read, sequence, num_ops, read_hits) { threads_done_ = threads_done; } void operator()() override { for (unsigned int i = 0; i < num_ops_; ++i) { ReadOne(); } ++*threads_done_; } private: std::atomic_int* threads_done_; }; class SeqConcurrentReadBenchmarkThread : public SeqReadBenchmarkThread { public: SeqConcurrentReadBenchmarkThread(MemTableRep* table, KeyGenerator* key_gen, uint64_t* bytes_written, uint64_t* bytes_read, uint64_t* sequence, uint64_t num_ops, uint64_t* read_hits, std::atomic_int* threads_done) : SeqReadBenchmarkThread(table, key_gen, bytes_written, bytes_read, sequence, num_ops, read_hits) { threads_done_ = threads_done; } void operator()() override { for (unsigned int i = 0; i < num_ops_; ++i) { ReadOneSeq(); } ++*threads_done_; } private: std::atomic_int* threads_done_; }; class Benchmark { public: explicit Benchmark(MemTableRep* table, KeyGenerator* key_gen, uint64_t* sequence, uint32_t num_threads) : table_(table), key_gen_(key_gen), sequence_(sequence), num_threads_(num_threads) {} virtual ~Benchmark() {} virtual void Run() { std::cout << "Number of threads: " << num_threads_ << std::endl; std::vector threads; uint64_t bytes_written = 0; uint64_t bytes_read = 0; uint64_t read_hits = 0; StopWatchNano timer(Env::Default(), true); RunThreads(&threads, &bytes_written, &bytes_read, true, &read_hits); auto elapsed_time = static_cast(timer.ElapsedNanos() / 1000); std::cout << "Elapsed time: " << static_cast(elapsed_time) << " us" << std::endl; if (bytes_written > 0) { auto MiB_written = static_cast(bytes_written) / (1 << 20); auto write_throughput = MiB_written / (elapsed_time / 1000000); std::cout << "Total bytes written: " << MiB_written << " MiB" << std::endl; std::cout << "Write throughput: " << write_throughput << " MiB/s" << std::endl; auto us_per_op = elapsed_time / num_write_ops_per_thread_; std::cout << "write us/op: " << us_per_op << std::endl; } if (bytes_read > 0) { auto MiB_read = static_cast(bytes_read) / (1 << 20); auto read_throughput = MiB_read / (elapsed_time / 1000000); std::cout << "Total bytes read: " << MiB_read << " MiB" << std::endl; std::cout << "Read throughput: " << read_throughput << " MiB/s" << std::endl; auto us_per_op = elapsed_time / num_read_ops_per_thread_; std::cout << "read us/op: " << us_per_op << std::endl; } } virtual void RunThreads(std::vector* threads, uint64_t* bytes_written, uint64_t* bytes_read, bool write, uint64_t* read_hits) = 0; protected: MemTableRep* table_; KeyGenerator* key_gen_; uint64_t* sequence_; uint64_t num_write_ops_per_thread_; uint64_t num_read_ops_per_thread_; const uint32_t num_threads_; }; class FillBenchmark : public Benchmark { public: explicit FillBenchmark(MemTableRep* table, KeyGenerator* key_gen, uint64_t* sequence) : Benchmark(table, key_gen, sequence, 1) { num_write_ops_per_thread_ = FLAGS_num_operations; } void RunThreads(std::vector* threads, uint64_t* bytes_written, uint64_t* bytes_read, bool write, uint64_t* read_hits) override { FillBenchmarkThread(table_, key_gen_, bytes_written, bytes_read, sequence_, num_write_ops_per_thread_, read_hits)(); } }; class ReadBenchmark : public Benchmark { public: explicit ReadBenchmark(MemTableRep* table, KeyGenerator* key_gen, uint64_t* sequence) : Benchmark(table, key_gen, sequence, FLAGS_num_threads) { num_read_ops_per_thread_ = FLAGS_num_operations / FLAGS_num_threads; } void RunThreads(std::vector* threads, uint64_t* bytes_written, uint64_t* bytes_read, bool write, uint64_t* read_hits) override { for (int i = 0; i < FLAGS_num_threads; ++i) { threads->emplace_back( ReadBenchmarkThread(table_, key_gen_, bytes_written, bytes_read, sequence_, num_read_ops_per_thread_, read_hits)); } for (auto& thread : *threads) { thread.join(); } std::cout << "read hit%: " << (static_cast(*read_hits) / FLAGS_num_operations) * 100 << std::endl; } }; class SeqReadBenchmark : public Benchmark { public: explicit SeqReadBenchmark(MemTableRep* table, uint64_t* sequence) : Benchmark(table, nullptr, sequence, FLAGS_num_threads) { num_read_ops_per_thread_ = FLAGS_num_scans; } void RunThreads(std::vector* threads, uint64_t* bytes_written, uint64_t* bytes_read, bool write, uint64_t* read_hits) override { for (int i = 0; i < FLAGS_num_threads; ++i) { threads->emplace_back(SeqReadBenchmarkThread( table_, key_gen_, bytes_written, bytes_read, sequence_, num_read_ops_per_thread_, read_hits)); } for (auto& thread : *threads) { thread.join(); } } }; template class ReadWriteBenchmark : public Benchmark { public: explicit ReadWriteBenchmark(MemTableRep* table, KeyGenerator* key_gen, uint64_t* sequence) : Benchmark(table, key_gen, sequence, FLAGS_num_threads) { num_read_ops_per_thread_ = FLAGS_num_threads <= 1 ? 0 : (FLAGS_num_operations / (FLAGS_num_threads - 1)); num_write_ops_per_thread_ = FLAGS_num_operations; } void RunThreads(std::vector* threads, uint64_t* bytes_written, uint64_t* bytes_read, bool write, uint64_t* read_hits) override { std::atomic_int threads_done; threads_done.store(0); threads->emplace_back(ConcurrentFillBenchmarkThread( table_, key_gen_, bytes_written, bytes_read, sequence_, num_write_ops_per_thread_, read_hits, &threads_done)); for (int i = 1; i < FLAGS_num_threads; ++i) { threads->emplace_back( ReadThreadType(table_, key_gen_, bytes_written, bytes_read, sequence_, num_read_ops_per_thread_, read_hits, &threads_done)); } for (auto& thread : *threads) { thread.join(); } } }; } // namespace rocksdb void PrintWarnings() { #if defined(__GNUC__) && !defined(__OPTIMIZE__) fprintf(stdout, "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); #endif #ifndef NDEBUG fprintf(stdout, "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); #endif } int main(int argc, char** argv) { rocksdb::port::InstallStackTraceHandler(); SetUsageMessage(std::string("\nUSAGE:\n") + std::string(argv[0]) + " [OPTIONS]..."); ParseCommandLineFlags(&argc, &argv, true); PrintWarnings(); rocksdb::Options options; std::unique_ptr factory; if (FLAGS_memtablerep == "skiplist") { factory.reset(new rocksdb::SkipListFactory); #ifndef ROCKSDB_LITE } else if (FLAGS_memtablerep == "vector") { factory.reset(new rocksdb::VectorRepFactory); } else if (FLAGS_memtablerep == "hashskiplist") { factory.reset(rocksdb::NewHashSkipListRepFactory( FLAGS_bucket_count, FLAGS_hashskiplist_height, FLAGS_hashskiplist_branching_factor)); options.prefix_extractor.reset( rocksdb::NewFixedPrefixTransform(FLAGS_prefix_length)); } else if (FLAGS_memtablerep == "hashlinklist") { factory.reset(rocksdb::NewHashLinkListRepFactory( FLAGS_bucket_count, FLAGS_huge_page_tlb_size, FLAGS_bucket_entries_logging_threshold, FLAGS_if_log_bucket_dist_when_flash, FLAGS_threshold_use_skiplist)); options.prefix_extractor.reset( rocksdb::NewFixedPrefixTransform(FLAGS_prefix_length)); } else if (FLAGS_memtablerep == "cuckoo") { factory.reset(rocksdb::NewHashCuckooRepFactory( FLAGS_write_buffer_size, FLAGS_average_data_size, static_cast(FLAGS_hash_function_count))); options.prefix_extractor.reset( rocksdb::NewFixedPrefixTransform(FLAGS_prefix_length)); #endif // ROCKSDB_LITE } else { fprintf(stdout, "Unknown memtablerep: %s\n", FLAGS_memtablerep.c_str()); exit(1); } rocksdb::InternalKeyComparator internal_key_comp( rocksdb::BytewiseComparator()); rocksdb::MemTable::KeyComparator key_comp(internal_key_comp); rocksdb::Arena arena; rocksdb::WriteBuffer wb(FLAGS_write_buffer_size); rocksdb::MemTableAllocator memtable_allocator(&arena, &wb); uint64_t sequence; auto createMemtableRep = [&] { sequence = 0; return factory->CreateMemTableRep(key_comp, &memtable_allocator, options.prefix_extractor.get(), options.info_log.get()); }; std::unique_ptr memtablerep; rocksdb::Random64 rng(FLAGS_seed); const char* benchmarks = FLAGS_benchmarks.c_str(); while (benchmarks != nullptr) { std::unique_ptr key_gen; const char* sep = strchr(benchmarks, ','); rocksdb::Slice name; if (sep == nullptr) { name = benchmarks; benchmarks = nullptr; } else { name = rocksdb::Slice(benchmarks, sep - benchmarks); benchmarks = sep + 1; } std::unique_ptr benchmark; if (name == rocksdb::Slice("fillseq")) { memtablerep.reset(createMemtableRep()); key_gen.reset(new rocksdb::KeyGenerator(&rng, rocksdb::SEQUENTIAL, FLAGS_num_operations)); benchmark.reset(new rocksdb::FillBenchmark(memtablerep.get(), key_gen.get(), &sequence)); } else if (name == rocksdb::Slice("fillrandom")) { memtablerep.reset(createMemtableRep()); key_gen.reset(new rocksdb::KeyGenerator(&rng, rocksdb::UNIQUE_RANDOM, FLAGS_num_operations)); benchmark.reset(new rocksdb::FillBenchmark(memtablerep.get(), key_gen.get(), &sequence)); } else if (name == rocksdb::Slice("readrandom")) { key_gen.reset(new rocksdb::KeyGenerator(&rng, rocksdb::RANDOM, FLAGS_num_operations)); benchmark.reset(new rocksdb::ReadBenchmark(memtablerep.get(), key_gen.get(), &sequence)); } else if (name == rocksdb::Slice("readseq")) { key_gen.reset(new rocksdb::KeyGenerator(&rng, rocksdb::SEQUENTIAL, FLAGS_num_operations)); benchmark.reset( new rocksdb::SeqReadBenchmark(memtablerep.get(), &sequence)); } else if (name == rocksdb::Slice("readwrite")) { memtablerep.reset(createMemtableRep()); key_gen.reset(new rocksdb::KeyGenerator(&rng, rocksdb::RANDOM, FLAGS_num_operations)); benchmark.reset(new rocksdb::ReadWriteBenchmark< rocksdb::ConcurrentReadBenchmarkThread>(memtablerep.get(), key_gen.get(), &sequence)); } else if (name == rocksdb::Slice("seqreadwrite")) { memtablerep.reset(createMemtableRep()); key_gen.reset(new rocksdb::KeyGenerator(&rng, rocksdb::RANDOM, FLAGS_num_operations)); benchmark.reset(new rocksdb::ReadWriteBenchmark< rocksdb::SeqConcurrentReadBenchmarkThread>(memtablerep.get(), key_gen.get(), &sequence)); } else { std::cout << "WARNING: skipping unknown benchmark '" << name.ToString() << std::endl; continue; } std::cout << "Running " << name.ToString() << std::endl; benchmark->Run(); } return 0; } #endif // GFLAGS rocksdb-4.5.1/db/merge_context.h000066400000000000000000000036721267530634000165670ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #pragma once #include "db/dbformat.h" #include "rocksdb/slice.h" #include #include namespace rocksdb { const std::deque empty_operand_list; // The merge context for merging a user key. // When doing a Get(), DB will create such a class and pass it when // issuing Get() operation to memtables and version_set. The operands // will be fetched from the context when issuing partial of full merge. class MergeContext { public: // Clear all the operands void Clear() { if (operand_list) { operand_list->clear(); } } // Replace all operands with merge_result, which are expected to be the // merge result of them. void PushPartialMergeResult(std::string& merge_result) { assert (operand_list); operand_list->clear(); operand_list->push_front(std::move(merge_result)); } // Push a merge operand void PushOperand(const Slice& operand_slice) { Initialize(); operand_list->push_front(operand_slice.ToString()); } // return total number of operands in the list size_t GetNumOperands() const { if (!operand_list) { return 0; } return operand_list->size(); } // Get the operand at the index. Slice GetOperand(int index) const { assert (operand_list); return (*operand_list)[index]; } // Return all the operands. const std::deque& GetOperands() const { if (!operand_list) { return empty_operand_list; } return *operand_list; } private: void Initialize() { if (!operand_list) { operand_list.reset(new std::deque()); } } std::unique_ptr> operand_list; }; } // namespace rocksdb rocksdb-4.5.1/db/merge_helper.cc000066400000000000000000000272561267530634000165240ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/merge_helper.h" #include #include #include "db/dbformat.h" #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/merge_operator.h" #include "table/internal_iterator.h" #include "util/perf_context_imp.h" #include "util/statistics.h" namespace rocksdb { // TODO(agiardullo): Clean up merge callsites to use this func Status MergeHelper::TimedFullMerge(const Slice& key, const Slice* value, const std::deque& operands, const MergeOperator* merge_operator, Statistics* statistics, Env* env, Logger* logger, std::string* result) { if (operands.size() == 0) { result->assign(value->data(), value->size()); return Status::OK(); } if (merge_operator == nullptr) { return Status::NotSupported("Provide a merge_operator when opening DB"); } // Setup to time the merge StopWatchNano timer(env, statistics != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); // Do the merge bool success = merge_operator->FullMerge(key, value, operands, result, logger); RecordTick(statistics, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanosSafe()); if (!success) { RecordTick(statistics, NUMBER_MERGE_FAILURES); return Status::Corruption("Error: Could not perform merge."); } return Status::OK(); } // PRE: iter points to the first merge type entry // POST: iter points to the first entry beyond the merge process (or the end) // keys_, operands_ are updated to reflect the merge result. // keys_ stores the list of keys encountered while merging. // operands_ stores the list of merge operands encountered while merging. // keys_[i] corresponds to operands_[i] for each i. Status MergeHelper::MergeUntil(InternalIterator* iter, const SequenceNumber stop_before, const bool at_bottom) { // Get a copy of the internal key, before it's invalidated by iter->Next() // Also maintain the list of merge operands seen. assert(HasOperator()); keys_.clear(); operands_.clear(); assert(user_merge_operator_); bool first_key = true; // We need to parse the internal key again as the parsed key is // backed by the internal key! // Assume no internal key corruption as it has been successfully parsed // by the caller. // original_key_is_iter variable is just caching the information: // original_key_is_iter == (iter->key().ToString() == original_key) bool original_key_is_iter = true; std::string original_key = iter->key().ToString(); // Important: // orig_ikey is backed by original_key if keys_.empty() // orig_ikey is backed by keys_.back() if !keys_.empty() ParsedInternalKey orig_ikey; ParseInternalKey(original_key, &orig_ikey); Status s; bool hit_the_next_user_key = false; for (; iter->Valid(); iter->Next(), original_key_is_iter = false) { ParsedInternalKey ikey; assert(keys_.size() == operands_.size()); if (!ParseInternalKey(iter->key(), &ikey)) { // stop at corrupted key if (assert_valid_internal_key_) { assert(!"Corrupted internal key not expected."); return Status::Corruption("Corrupted internal key not expected."); } break; } else if (first_key) { assert(user_comparator_->Equal(ikey.user_key, orig_ikey.user_key)); first_key = false; } else if (!user_comparator_->Equal(ikey.user_key, orig_ikey.user_key)) { // hit a different user key, stop right here hit_the_next_user_key = true; break; } else if (stop_before && ikey.sequence <= stop_before) { // hit an entry that's visible by the previous snapshot, can't touch that break; } // At this point we are guaranteed that we need to process this key. assert(IsValueType(ikey.type)); if (ikey.type != kTypeMerge) { if (ikey.type != kTypeValue && ikey.type != kTypeDeletion) { // Merges operands can only be used with puts and deletions, single // deletions are not supported. assert(false); // release build doesn't have asserts, so we return error status return Status::InvalidArgument( " Merges operands can only be used with puts and deletions, single " "deletions are not supported."); } // hit a put/delete // => merge the put value or a nullptr with operands_ // => store result in operands_.back() (and update keys_.back()) // => change the entry type to kTypeValue for keys_.back() // We are done! Success! // If there are no operands, just return the Status::OK(). That will cause // the compaction iterator to write out the key we're currently at, which // is the put/delete we just encountered. if (keys_.empty()) { return Status::OK(); } // TODO(noetzli) If the merge operator returns false, we are currently // (almost) silently dropping the put/delete. That's probably not what we // want. const Slice val = iter->value(); const Slice* val_ptr = (kTypeValue == ikey.type) ? &val : nullptr; std::string merge_result; s = TimedFullMerge(ikey.user_key, val_ptr, operands_, user_merge_operator_, stats_, env_, logger_, &merge_result); // We store the result in keys_.back() and operands_.back() // if nothing went wrong (i.e.: no operand corruption on disk) if (s.ok()) { // The original key encountered original_key = std::move(keys_.back()); orig_ikey.type = kTypeValue; UpdateInternalKey(&original_key, orig_ikey.sequence, orig_ikey.type); keys_.clear(); operands_.clear(); keys_.emplace_front(std::move(original_key)); operands_.emplace_front(std::move(merge_result)); } // move iter to the next entry iter->Next(); return s; } else { // hit a merge // => if there is a compaction filter, apply it. // => merge the operand into the front of the operands_ list // if not filtered // => then continue because we haven't yet seen a Put/Delete. // // Keep queuing keys and operands until we either meet a put / delete // request or later did a partial merge. Slice value_slice = iter->value(); // add an operand to the list if: // 1) it's included in one of the snapshots. in that case we *must* write // it out, no matter what compaction filter says // 2) it's not filtered by a compaction filter if (ikey.sequence <= latest_snapshot_ || !FilterMerge(orig_ikey.user_key, value_slice)) { if (original_key_is_iter) { // this is just an optimization that saves us one memcpy keys_.push_front(std::move(original_key)); } else { keys_.push_front(iter->key().ToString()); } if (keys_.size() == 1) { // we need to re-anchor the orig_ikey because it was anchored by // original_key before ParseInternalKey(keys_.back(), &orig_ikey); } operands_.push_front(value_slice.ToString()); } } } if (operands_.size() == 0) { // we filtered out all the merge operands return Status::OK(); } // We are sure we have seen this key's entire history if we are at the // last level and exhausted all internal keys of this user key. // NOTE: !iter->Valid() does not necessarily mean we hit the // beginning of a user key, as versions of a user key might be // split into multiple files (even files on the same level) // and some files might not be included in the compaction/merge. // // There are also cases where we have seen the root of history of this // key without being sure of it. Then, we simply miss the opportunity // to combine the keys. Since VersionSet::SetupOtherInputs() always makes // sure that all merge-operands on the same level get compacted together, // this will simply lead to these merge operands moving to the next level. // // So, we only perform the following logic (to merge all operands together // without a Put/Delete) if we are certain that we have seen the end of key. bool surely_seen_the_beginning = hit_the_next_user_key && at_bottom; if (surely_seen_the_beginning) { // do a final merge with nullptr as the existing value and say // bye to the merge type (it's now converted to a Put) assert(kTypeMerge == orig_ikey.type); assert(operands_.size() >= 1); assert(operands_.size() == keys_.size()); std::string merge_result; s = TimedFullMerge(orig_ikey.user_key, nullptr, operands_, user_merge_operator_, stats_, env_, logger_, &merge_result); if (s.ok()) { // The original key encountered // We are certain that keys_ is not empty here (see assertions couple of // lines before). original_key = std::move(keys_.back()); orig_ikey.type = kTypeValue; UpdateInternalKey(&original_key, orig_ikey.sequence, orig_ikey.type); keys_.clear(); operands_.clear(); keys_.emplace_front(std::move(original_key)); operands_.emplace_front(std::move(merge_result)); } } else { // We haven't seen the beginning of the key nor a Put/Delete. // Attempt to use the user's associative merge function to // merge the stacked merge operands into a single operand. // // TODO(noetzli) The docblock of MergeUntil suggests that a successful // partial merge returns Status::OK(). Should we change the status code // after a successful partial merge? s = Status::MergeInProgress(); if (operands_.size() >= 2 && operands_.size() >= min_partial_merge_operands_) { bool merge_success = false; std::string merge_result; { StopWatchNano timer(env_, stats_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); merge_success = user_merge_operator_->PartialMergeMulti( orig_ikey.user_key, std::deque(operands_.begin(), operands_.end()), &merge_result, logger_); RecordTick(stats_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanosSafe()); } if (merge_success) { // Merging of operands (associative merge) was successful. // Replace operands with the merge result operands_.clear(); operands_.emplace_front(std::move(merge_result)); keys_.erase(keys_.begin(), keys_.end() - 1); } } } return s; } MergeOutputIterator::MergeOutputIterator(const MergeHelper* merge_helper) : merge_helper_(merge_helper) { it_keys_ = merge_helper_->keys().rend(); it_values_ = merge_helper_->values().rend(); } void MergeOutputIterator::SeekToFirst() { const auto& keys = merge_helper_->keys(); const auto& values = merge_helper_->values(); assert(keys.size() == values.size()); it_keys_ = keys.rbegin(); it_values_ = values.rbegin(); } void MergeOutputIterator::Next() { ++it_keys_; ++it_values_; } bool MergeHelper::FilterMerge(const Slice& user_key, const Slice& value_slice) { if (compaction_filter_ == nullptr) { return false; } if (stats_ != nullptr) { filter_timer_.Start(); } bool to_delete = compaction_filter_->FilterMergeOperand(level_, user_key, value_slice); total_filter_time_ += filter_timer_.ElapsedNanosSafe(); return to_delete; } } // namespace rocksdb rocksdb-4.5.1/db/merge_helper.h000066400000000000000000000153431267530634000163600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #ifndef MERGE_HELPER_H #define MERGE_HELPER_H #include #include #include "db/dbformat.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/env.h" #include "rocksdb/slice.h" #include "util/stop_watch.h" namespace rocksdb { class Comparator; class Iterator; class Logger; class MergeOperator; class Statistics; class InternalIterator; class MergeHelper { public: MergeHelper(Env* env, const Comparator* user_comparator, const MergeOperator* user_merge_operator, const CompactionFilter* compaction_filter, Logger* logger, unsigned min_partial_merge_operands, bool assert_valid_internal_key, SequenceNumber latest_snapshot, int level = 0, Statistics* stats = nullptr) : env_(env), user_comparator_(user_comparator), user_merge_operator_(user_merge_operator), compaction_filter_(compaction_filter), logger_(logger), min_partial_merge_operands_(min_partial_merge_operands), assert_valid_internal_key_(assert_valid_internal_key), latest_snapshot_(latest_snapshot), level_(level), keys_(), operands_(), filter_timer_(env_), total_filter_time_(0U), stats_(stats) { assert(user_comparator_ != nullptr); } // Wrapper around MergeOperator::FullMerge() that records perf statistics. // Result of merge will be written to result if status returned is OK. // If operands is empty, the value will simply be copied to result. // Returns one of the following statuses: // - OK: Entries were successfully merged. // - Corruption: Merge operator reported unsuccessful merge. // - NotSupported: Merge operator is missing. static Status TimedFullMerge(const Slice& key, const Slice* value, const std::deque& operands, const MergeOperator* merge_operator, Statistics* statistics, Env* env, Logger* logger, std::string* result); // Merge entries until we hit // - a corrupted key // - a Put/Delete, // - a different user key, // - a specific sequence number (snapshot boundary), // or - the end of iteration // iter: (IN) points to the first merge type entry // (OUT) points to the first entry not included in the merge process // stop_before: (IN) a sequence number that merge should not cross. // 0 means no restriction // at_bottom: (IN) true if the iterator covers the bottem level, which means // we could reach the start of the history of this user key. // // Returns one of the following statuses: // - OK: Entries were successfully merged. // - MergeInProgress: Put/Delete not encountered and unable to merge operands. // - Corruption: Merge operator reported unsuccessful merge or a corrupted // key has been encountered and not expected (applies only when compiling // with asserts removed). // // REQUIRED: The first key in the input is not corrupted. Status MergeUntil(InternalIterator* iter, const SequenceNumber stop_before = 0, const bool at_bottom = false); // Filters a merge operand using the compaction filter specified // in the constructor. Returns true if the operand should be filtered out. bool FilterMerge(const Slice& user_key, const Slice& value_slice); // Query the merge result // These are valid until the next MergeUntil call // If the merging was successful: // - keys() contains a single element with the latest sequence number of // the merges. The type will be Put or Merge. See IMPORTANT 1 note, below. // - values() contains a single element with the result of merging all the // operands together // // IMPORTANT 1: the key type could change after the MergeUntil call. // Put/Delete + Merge + ... + Merge => Put // Merge + ... + Merge => Merge // // If the merge operator is not associative, and if a Put/Delete is not found // then the merging will be unsuccessful. In this case: // - keys() contains the list of internal keys seen in order of iteration. // - values() contains the list of values (merges) seen in the same order. // values() is parallel to keys() so that the first entry in // keys() is the key associated with the first entry in values() // and so on. These lists will be the same length. // All of these pairs will be merges over the same user key. // See IMPORTANT 2 note below. // // IMPORTANT 2: The entries were traversed in order from BACK to FRONT. // So keys().back() was the first key seen by iterator. // TODO: Re-style this comment to be like the first one const std::deque& keys() const { return keys_; } const std::deque& values() const { return operands_; } uint64_t TotalFilterTime() const { return total_filter_time_; } bool HasOperator() const { return user_merge_operator_ != nullptr; } private: Env* env_; const Comparator* user_comparator_; const MergeOperator* user_merge_operator_; const CompactionFilter* compaction_filter_; Logger* logger_; unsigned min_partial_merge_operands_; bool assert_valid_internal_key_; // enforce no internal key corruption? SequenceNumber latest_snapshot_; int level_; // the scratch area that holds the result of MergeUntil // valid up to the next MergeUntil call std::deque keys_; // Keeps track of the sequence of keys seen std::deque operands_; // Parallel with keys_; stores the values StopWatchNano filter_timer_; uint64_t total_filter_time_; Statistics* stats_; }; // MergeOutputIterator can be used to iterate over the result of a merge. class MergeOutputIterator { public: // The MergeOutputIterator is bound to a MergeHelper instance. explicit MergeOutputIterator(const MergeHelper* merge_helper); // Seeks to the first record in the output. void SeekToFirst(); // Advances to the next record in the output. void Next(); Slice key() { return Slice(*it_keys_); } Slice value() { return Slice(*it_values_); } bool Valid() { return it_keys_ != merge_helper_->keys().rend(); } private: const MergeHelper* merge_helper_; std::deque::const_reverse_iterator it_keys_; std::deque::const_reverse_iterator it_values_; }; } // namespace rocksdb #endif rocksdb-4.5.1/db/merge_helper_test.cc000066400000000000000000000262051267530634000175540ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include "db/merge_helper.h" #include "rocksdb/comparator.h" #include "util/coding.h" #include "util/testharness.h" #include "util/testutil.h" #include "utilities/merge_operators.h" namespace rocksdb { class MergeHelperTest : public testing::Test { public: MergeHelperTest() { env_ = Env::Default(); } ~MergeHelperTest() = default; Status Run(SequenceNumber stop_before, bool at_bottom, SequenceNumber latest_snapshot = 0) { iter_.reset(new test::VectorIterator(ks_, vs_)); iter_->SeekToFirst(); merge_helper_.reset(new MergeHelper(env_, BytewiseComparator(), merge_op_.get(), filter_.get(), nullptr, 2U, false, latest_snapshot)); return merge_helper_->MergeUntil(iter_.get(), stop_before, at_bottom); } void AddKeyVal(const std::string& user_key, const SequenceNumber& seq, const ValueType& t, const std::string& val, bool corrupt = false) { InternalKey ikey(user_key, seq, t); if (corrupt) { test::CorruptKeyType(&ikey); } ks_.push_back(ikey.Encode().ToString()); vs_.push_back(val); } Env* env_; std::unique_ptr iter_; std::shared_ptr merge_op_; std::unique_ptr merge_helper_; std::vector ks_; std::vector vs_; std::unique_ptr filter_; }; // If MergeHelper encounters a new key on the last level, we know that // the key has no more history and it can merge keys. TEST_F(MergeHelperTest, MergeAtBottomSuccess) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); AddKeyVal("a", 20, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 10, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("b", 10, kTypeMerge, test::EncodeInt(4U)); // <- iter_ after merge ASSERT_TRUE(Run(0, true).ok()); ASSERT_EQ(ks_[2], iter_->key()); ASSERT_EQ(test::KeyStr("a", 20, kTypeValue), merge_helper_->keys()[0]); ASSERT_EQ(test::EncodeInt(4U), merge_helper_->values()[0]); ASSERT_EQ(1U, merge_helper_->keys().size()); ASSERT_EQ(1U, merge_helper_->values().size()); } // Merging with a value results in a successful merge. TEST_F(MergeHelperTest, MergeValue) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); AddKeyVal("a", 40, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 20, kTypeValue, test::EncodeInt(4U)); // <- iter_ after merge AddKeyVal("a", 10, kTypeMerge, test::EncodeInt(1U)); ASSERT_TRUE(Run(0, false).ok()); ASSERT_EQ(ks_[3], iter_->key()); ASSERT_EQ(test::KeyStr("a", 40, kTypeValue), merge_helper_->keys()[0]); ASSERT_EQ(test::EncodeInt(8U), merge_helper_->values()[0]); ASSERT_EQ(1U, merge_helper_->keys().size()); ASSERT_EQ(1U, merge_helper_->values().size()); } // Merging stops before a snapshot. TEST_F(MergeHelperTest, SnapshotBeforeValue) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); AddKeyVal("a", 50, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 40, kTypeMerge, test::EncodeInt(3U)); // <- iter_ after merge AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 20, kTypeValue, test::EncodeInt(4U)); AddKeyVal("a", 10, kTypeMerge, test::EncodeInt(1U)); ASSERT_TRUE(Run(31, true).IsMergeInProgress()); ASSERT_EQ(ks_[2], iter_->key()); ASSERT_EQ(test::KeyStr("a", 50, kTypeMerge), merge_helper_->keys()[0]); ASSERT_EQ(test::EncodeInt(4U), merge_helper_->values()[0]); ASSERT_EQ(1U, merge_helper_->keys().size()); ASSERT_EQ(1U, merge_helper_->values().size()); } // MergeHelper preserves the operand stack for merge operators that // cannot do a partial merge. TEST_F(MergeHelperTest, NoPartialMerge) { merge_op_ = MergeOperators::CreateStringAppendTESTOperator(); AddKeyVal("a", 50, kTypeMerge, "v2"); AddKeyVal("a", 40, kTypeMerge, "v"); // <- iter_ after merge AddKeyVal("a", 30, kTypeMerge, "v"); ASSERT_TRUE(Run(31, true).IsMergeInProgress()); ASSERT_EQ(ks_[2], iter_->key()); ASSERT_EQ(test::KeyStr("a", 40, kTypeMerge), merge_helper_->keys()[0]); ASSERT_EQ("v", merge_helper_->values()[0]); ASSERT_EQ(test::KeyStr("a", 50, kTypeMerge), merge_helper_->keys()[1]); ASSERT_EQ("v2", merge_helper_->values()[1]); ASSERT_EQ(2U, merge_helper_->keys().size()); ASSERT_EQ(2U, merge_helper_->values().size()); } // A single operand can not be merged. TEST_F(MergeHelperTest, SingleOperand) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); AddKeyVal("a", 50, kTypeMerge, test::EncodeInt(1U)); ASSERT_TRUE(Run(31, true).IsMergeInProgress()); ASSERT_FALSE(iter_->Valid()); ASSERT_EQ(test::KeyStr("a", 50, kTypeMerge), merge_helper_->keys()[0]); ASSERT_EQ(test::EncodeInt(1U), merge_helper_->values()[0]); ASSERT_EQ(1U, merge_helper_->keys().size()); ASSERT_EQ(1U, merge_helper_->values().size()); } // Merging with a deletion turns the deletion into a value TEST_F(MergeHelperTest, MergeDeletion) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 20, kTypeDeletion, ""); ASSERT_TRUE(Run(15, false).ok()); ASSERT_FALSE(iter_->Valid()); ASSERT_EQ(test::KeyStr("a", 30, kTypeValue), merge_helper_->keys()[0]); ASSERT_EQ(test::EncodeInt(3U), merge_helper_->values()[0]); ASSERT_EQ(1U, merge_helper_->keys().size()); ASSERT_EQ(1U, merge_helper_->values().size()); } // The merge helper stops upon encountering a corrupt key TEST_F(MergeHelperTest, CorruptKey) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 25, kTypeMerge, test::EncodeInt(1U)); // Corrupt key AddKeyVal("a", 20, kTypeDeletion, "", true); // <- iter_ after merge ASSERT_TRUE(Run(15, false).IsMergeInProgress()); ASSERT_EQ(ks_[2], iter_->key()); ASSERT_EQ(test::KeyStr("a", 30, kTypeMerge), merge_helper_->keys()[0]); ASSERT_EQ(test::EncodeInt(4U), merge_helper_->values()[0]); ASSERT_EQ(1U, merge_helper_->keys().size()); ASSERT_EQ(1U, merge_helper_->values().size()); } // The compaction filter is called on every merge operand TEST_F(MergeHelperTest, FilterMergeOperands) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); filter_.reset(new test::FilterNumber(5U)); AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 29, kTypeMerge, test::EncodeInt(5U)); // Filtered AddKeyVal("a", 28, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 27, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 26, kTypeMerge, test::EncodeInt(5U)); // Filtered AddKeyVal("a", 25, kTypeValue, test::EncodeInt(1U)); ASSERT_TRUE(Run(15, false).ok()); ASSERT_FALSE(iter_->Valid()); MergeOutputIterator merge_output_iter(merge_helper_.get()); merge_output_iter.SeekToFirst(); ASSERT_EQ(test::KeyStr("a", 30, kTypeValue), merge_output_iter.key().ToString()); ASSERT_EQ(test::EncodeInt(8U), merge_output_iter.value().ToString()); merge_output_iter.Next(); ASSERT_FALSE(merge_output_iter.Valid()); } TEST_F(MergeHelperTest, FilterAllMergeOperands) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); filter_.reset(new test::FilterNumber(5U)); AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 29, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 28, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 27, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 26, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 25, kTypeMerge, test::EncodeInt(5U)); // filtered out all ASSERT_TRUE(Run(15, false).ok()); ASSERT_FALSE(iter_->Valid()); MergeOutputIterator merge_output_iter(merge_helper_.get()); merge_output_iter.SeekToFirst(); ASSERT_FALSE(merge_output_iter.Valid()); // we have one operand that will survive because it's a delete AddKeyVal("a", 24, kTypeDeletion, test::EncodeInt(5U)); AddKeyVal("b", 23, kTypeValue, test::EncodeInt(5U)); ASSERT_TRUE(Run(15, true).ok()); merge_output_iter = MergeOutputIterator(merge_helper_.get()); ASSERT_TRUE(iter_->Valid()); merge_output_iter.SeekToFirst(); ASSERT_FALSE(merge_output_iter.Valid()); // when all merge operands are filtered out, we leave the iterator pointing to // the Put/Delete that survived ASSERT_EQ(test::KeyStr("a", 24, kTypeDeletion), iter_->key().ToString()); ASSERT_EQ(test::EncodeInt(5U), iter_->value().ToString()); } // Make sure that merge operands are filtered at the beginning TEST_F(MergeHelperTest, FilterFirstMergeOperand) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); filter_.reset(new test::FilterNumber(5U)); AddKeyVal("a", 31, kTypeMerge, test::EncodeInt(5U)); // Filtered AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(5U)); // Filtered AddKeyVal("a", 29, kTypeMerge, test::EncodeInt(2U)); AddKeyVal("a", 28, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 27, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 26, kTypeMerge, test::EncodeInt(5U)); // Filtered AddKeyVal("a", 25, kTypeMerge, test::EncodeInt(5U)); // Filtered AddKeyVal("b", 24, kTypeValue, test::EncodeInt(5U)); // next user key ASSERT_OK(Run(15, true)); ASSERT_TRUE(iter_->Valid()); MergeOutputIterator merge_output_iter(merge_helper_.get()); merge_output_iter.SeekToFirst(); // sequence number is 29 here, because the first merge operand got filtered // out ASSERT_EQ(test::KeyStr("a", 29, kTypeValue), merge_output_iter.key().ToString()); ASSERT_EQ(test::EncodeInt(6U), merge_output_iter.value().ToString()); merge_output_iter.Next(); ASSERT_FALSE(merge_output_iter.Valid()); // make sure that we're passing user keys into the filter ASSERT_EQ("a", filter_->last_merge_operand_key()); } // Make sure that merge operands are not filtered out if there's a snapshot // pointing at them TEST_F(MergeHelperTest, DontFilterMergeOperandsBeforeSnapshotTest) { merge_op_ = MergeOperators::CreateUInt64AddOperator(); filter_.reset(new test::FilterNumber(5U)); AddKeyVal("a", 31, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 30, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 29, kTypeMerge, test::EncodeInt(2U)); AddKeyVal("a", 28, kTypeMerge, test::EncodeInt(1U)); AddKeyVal("a", 27, kTypeMerge, test::EncodeInt(3U)); AddKeyVal("a", 26, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("a", 25, kTypeMerge, test::EncodeInt(5U)); AddKeyVal("b", 24, kTypeValue, test::EncodeInt(5U)); ASSERT_OK(Run(15, true, 32)); ASSERT_TRUE(iter_->Valid()); MergeOutputIterator merge_output_iter(merge_helper_.get()); merge_output_iter.SeekToFirst(); ASSERT_EQ(test::KeyStr("a", 31, kTypeValue), merge_output_iter.key().ToString()); ASSERT_EQ(test::EncodeInt(26U), merge_output_iter.value().ToString()); merge_output_iter.Next(); ASSERT_FALSE(merge_output_iter.Valid()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/merge_operator.cc000066400000000000000000000050311267530634000170630ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // /** * Back-end implementation details specific to the Merge Operator. */ #include "rocksdb/merge_operator.h" namespace rocksdb { // The default implementation of PartialMergeMulti, which invokes // PartialMerge multiple times internally and merges two operands at // a time. bool MergeOperator::PartialMergeMulti(const Slice& key, const std::deque& operand_list, std::string* new_value, Logger* logger) const { assert(operand_list.size() >= 2); // Simply loop through the operands Slice temp_slice(operand_list[0]); for (size_t i = 1; i < operand_list.size(); ++i) { auto& operand = operand_list[i]; std::string temp_value; if (!PartialMerge(key, temp_slice, operand, &temp_value, logger)) { return false; } swap(temp_value, *new_value); temp_slice = Slice(*new_value); } // The result will be in *new_value. All merges succeeded. return true; } // Given a "real" merge from the library, call the user's // associative merge function one-by-one on each of the operands. // NOTE: It is assumed that the client's merge-operator will handle any errors. bool AssociativeMergeOperator::FullMerge( const Slice& key, const Slice* existing_value, const std::deque& operand_list, std::string* new_value, Logger* logger) const { // Simply loop through the operands Slice temp_existing; for (const auto& operand : operand_list) { Slice value(operand); std::string temp_value; if (!Merge(key, existing_value, value, &temp_value, logger)) { return false; } swap(temp_value, *new_value); temp_existing = Slice(*new_value); existing_value = &temp_existing; } // The result will be in *new_value. All merges succeeded. return true; } // Call the user defined simple merge on the operands; // NOTE: It is assumed that the client's merge-operator will handle any errors. bool AssociativeMergeOperator::PartialMerge( const Slice& key, const Slice& left_operand, const Slice& right_operand, std::string* new_value, Logger* logger) const { return Merge(key, &left_operand, right_operand, new_value, logger); } } // namespace rocksdb rocksdb-4.5.1/db/merge_test.cc000066400000000000000000000341121267530634000162110ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include #include #include #include "port/stack_trace.h" #include "rocksdb/cache.h" #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/merge_operator.h" #include "rocksdb/utilities/db_ttl.h" #include "db/dbformat.h" #include "db/db_impl.h" #include "db/write_batch_internal.h" #include "utilities/merge_operators.h" #include "util/testharness.h" using namespace rocksdb; namespace { size_t num_merge_operator_calls; void resetNumMergeOperatorCalls() { num_merge_operator_calls = 0; } size_t num_partial_merge_calls; void resetNumPartialMergeCalls() { num_partial_merge_calls = 0; } } class CountMergeOperator : public AssociativeMergeOperator { public: CountMergeOperator() { mergeOperator_ = MergeOperators::CreateUInt64AddOperator(); } virtual bool Merge(const Slice& key, const Slice* existing_value, const Slice& value, std::string* new_value, Logger* logger) const override { assert(new_value->empty()); ++num_merge_operator_calls; if (existing_value == nullptr) { new_value->assign(value.data(), value.size()); return true; } return mergeOperator_->PartialMerge( key, *existing_value, value, new_value, logger); } virtual bool PartialMergeMulti(const Slice& key, const std::deque& operand_list, std::string* new_value, Logger* logger) const override { assert(new_value->empty()); ++num_partial_merge_calls; return mergeOperator_->PartialMergeMulti(key, operand_list, new_value, logger); } virtual const char* Name() const override { return "UInt64AddOperator"; } private: std::shared_ptr mergeOperator_; }; namespace { std::shared_ptr OpenDb(const std::string& dbname, const bool ttl = false, const size_t max_successive_merges = 0, const uint32_t min_partial_merge_operands = 2) { DB* db; Options options; options.create_if_missing = true; options.merge_operator = std::make_shared(); options.max_successive_merges = max_successive_merges; options.min_partial_merge_operands = min_partial_merge_operands; Status s; DestroyDB(dbname, Options()); // DBWithTTL is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE if (ttl) { std::cout << "Opening database with TTL\n"; DBWithTTL* db_with_ttl; s = DBWithTTL::Open(options, dbname, &db_with_ttl); db = db_with_ttl; } else { s = DB::Open(options, dbname, &db); } #else assert(!ttl); s = DB::Open(options, dbname, &db); #endif // !ROCKSDB_LITE if (!s.ok()) { std::cerr << s.ToString() << std::endl; assert(false); } return std::shared_ptr(db); } } // namespace // Imagine we are maintaining a set of uint64 counters. // Each counter has a distinct name. And we would like // to support four high level operations: // set, add, get and remove // This is a quick implementation without a Merge operation. class Counters { protected: std::shared_ptr db_; WriteOptions put_option_; ReadOptions get_option_; WriteOptions delete_option_; uint64_t default_; public: explicit Counters(std::shared_ptr db, uint64_t defaultCount = 0) : db_(db), put_option_(), get_option_(), delete_option_(), default_(defaultCount) { assert(db_); } virtual ~Counters() {} // public interface of Counters. // All four functions return false // if the underlying level db operation failed. // mapped to a levedb Put bool set(const std::string& key, uint64_t value) { // just treat the internal rep of int64 as the string Slice slice((char *)&value, sizeof(value)); auto s = db_->Put(put_option_, key, slice); if (s.ok()) { return true; } else { std::cerr << s.ToString() << std::endl; return false; } } // mapped to a rocksdb Delete bool remove(const std::string& key) { auto s = db_->Delete(delete_option_, key); if (s.ok()) { return true; } else { std::cerr << s.ToString() << std::endl; return false; } } // mapped to a rocksdb Get bool get(const std::string& key, uint64_t* value) { std::string str; auto s = db_->Get(get_option_, key, &str); if (s.IsNotFound()) { // return default value if not found; *value = default_; return true; } else if (s.ok()) { // deserialization if (str.size() != sizeof(uint64_t)) { std::cerr << "value corruption\n"; return false; } *value = DecodeFixed64(&str[0]); return true; } else { std::cerr << s.ToString() << std::endl; return false; } } // 'add' is implemented as get -> modify -> set // An alternative is a single merge operation, see MergeBasedCounters virtual bool add(const std::string& key, uint64_t value) { uint64_t base = default_; return get(key, &base) && set(key, base + value); } // convenience functions for testing void assert_set(const std::string& key, uint64_t value) { assert(set(key, value)); } void assert_remove(const std::string& key) { assert(remove(key)); } uint64_t assert_get(const std::string& key) { uint64_t value = default_; int result = get(key, &value); assert(result); if (result == 0) exit(1); // Disable unused variable warning. return value; } void assert_add(const std::string& key, uint64_t value) { int result = add(key, value); assert(result); if (result == 0) exit(1); // Disable unused variable warning. } }; // Implement 'add' directly with the new Merge operation class MergeBasedCounters : public Counters { private: WriteOptions merge_option_; // for merge public: explicit MergeBasedCounters(std::shared_ptr db, uint64_t defaultCount = 0) : Counters(db, defaultCount), merge_option_() { } // mapped to a rocksdb Merge operation virtual bool add(const std::string& key, uint64_t value) override { char encoded[sizeof(uint64_t)]; EncodeFixed64(encoded, value); Slice slice(encoded, sizeof(uint64_t)); auto s = db_->Merge(merge_option_, key, slice); if (s.ok()) { return true; } else { std::cerr << s.ToString() << std::endl; return false; } } }; namespace { void dumpDb(DB* db) { auto it = unique_ptr(db->NewIterator(ReadOptions())); for (it->SeekToFirst(); it->Valid(); it->Next()) { uint64_t value = DecodeFixed64(it->value().data()); std::cout << it->key().ToString() << ": " << value << std::endl; } assert(it->status().ok()); // Check for any errors found during the scan } void testCounters(Counters& counters, DB* db, bool test_compaction) { FlushOptions o; o.wait = true; counters.assert_set("a", 1); if (test_compaction) db->Flush(o); assert(counters.assert_get("a") == 1); counters.assert_remove("b"); // defaut value is 0 if non-existent assert(counters.assert_get("b") == 0); counters.assert_add("a", 2); if (test_compaction) db->Flush(o); // 1+2 = 3 assert(counters.assert_get("a")== 3); dumpDb(db); std::cout << "1\n"; // 1+...+49 = ? uint64_t sum = 0; for (int i = 1; i < 50; i++) { counters.assert_add("b", i); sum += i; } assert(counters.assert_get("b") == sum); std::cout << "2\n"; dumpDb(db); std::cout << "3\n"; if (test_compaction) { db->Flush(o); std::cout << "Compaction started ...\n"; db->CompactRange(CompactRangeOptions(), nullptr, nullptr); std::cout << "Compaction ended\n"; dumpDb(db); assert(counters.assert_get("a")== 3); assert(counters.assert_get("b") == sum); } } void testSuccessiveMerge(Counters& counters, size_t max_num_merges, size_t num_merges) { counters.assert_remove("z"); uint64_t sum = 0; for (size_t i = 1; i <= num_merges; ++i) { resetNumMergeOperatorCalls(); counters.assert_add("z", i); sum += i; if (i % (max_num_merges + 1) == 0) { assert(num_merge_operator_calls == max_num_merges + 1); } else { assert(num_merge_operator_calls == 0); } resetNumMergeOperatorCalls(); assert(counters.assert_get("z") == sum); assert(num_merge_operator_calls == i % (max_num_merges + 1)); } } void testPartialMerge(Counters* counters, DB* db, size_t max_merge, size_t min_merge, size_t count) { FlushOptions o; o.wait = true; // Test case 1: partial merge should be called when the number of merge // operands exceeds the threshold. uint64_t tmp_sum = 0; resetNumPartialMergeCalls(); for (size_t i = 1; i <= count; i++) { counters->assert_add("b", i); tmp_sum += i; } db->Flush(o); db->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(tmp_sum, counters->assert_get("b")); if (count > max_merge) { // in this case, FullMerge should be called instead. ASSERT_EQ(num_partial_merge_calls, 0U); } else { // if count >= min_merge, then partial merge should be called once. ASSERT_EQ((count >= min_merge), (num_partial_merge_calls == 1)); } // Test case 2: partial merge should not be called when a put is found. resetNumPartialMergeCalls(); tmp_sum = 0; db->Put(rocksdb::WriteOptions(), "c", "10"); for (size_t i = 1; i <= count; i++) { counters->assert_add("c", i); tmp_sum += i; } db->Flush(o); db->CompactRange(CompactRangeOptions(), nullptr, nullptr); ASSERT_EQ(tmp_sum, counters->assert_get("c")); ASSERT_EQ(num_partial_merge_calls, 0U); } void testSingleBatchSuccessiveMerge(DB* db, size_t max_num_merges, size_t num_merges) { assert(num_merges > max_num_merges); Slice key("BatchSuccessiveMerge"); uint64_t merge_value = 1; Slice merge_value_slice((char *)&merge_value, sizeof(merge_value)); // Create the batch WriteBatch batch; for (size_t i = 0; i < num_merges; ++i) { batch.Merge(key, merge_value_slice); } // Apply to memtable and count the number of merges resetNumMergeOperatorCalls(); { Status s = db->Write(WriteOptions(), &batch); assert(s.ok()); } ASSERT_EQ( num_merge_operator_calls, static_cast(num_merges - (num_merges % (max_num_merges + 1)))); // Get the value resetNumMergeOperatorCalls(); std::string get_value_str; { Status s = db->Get(ReadOptions(), key, &get_value_str); assert(s.ok()); } assert(get_value_str.size() == sizeof(uint64_t)); uint64_t get_value = DecodeFixed64(&get_value_str[0]); ASSERT_EQ(get_value, num_merges * merge_value); ASSERT_EQ(num_merge_operator_calls, static_cast((num_merges % (max_num_merges + 1)))); } void runTest(int argc, const std::string& dbname, const bool use_ttl = false) { bool compact = false; if (argc > 1) { compact = true; std::cout << "Turn on Compaction\n"; } { auto db = OpenDb(dbname, use_ttl); { std::cout << "Test read-modify-write counters... \n"; Counters counters(db, 0); testCounters(counters, db.get(), true); } { std::cout << "Test merge-based counters... \n"; MergeBasedCounters counters(db, 0); testCounters(counters, db.get(), compact); } } DestroyDB(dbname, Options()); { std::cout << "Test merge in memtable... \n"; size_t max_merge = 5; auto db = OpenDb(dbname, use_ttl, max_merge); MergeBasedCounters counters(db, 0); testCounters(counters, db.get(), compact); testSuccessiveMerge(counters, max_merge, max_merge * 2); testSingleBatchSuccessiveMerge(db.get(), 5, 7); DestroyDB(dbname, Options()); } { std::cout << "Test Partial-Merge\n"; size_t max_merge = 100; for (uint32_t min_merge = 5; min_merge < 25; min_merge += 5) { for (uint32_t count = min_merge - 1; count <= min_merge + 1; count++) { auto db = OpenDb(dbname, use_ttl, max_merge, min_merge); MergeBasedCounters counters(db, 0); testPartialMerge(&counters, db.get(), max_merge, min_merge, count); DestroyDB(dbname, Options()); } { auto db = OpenDb(dbname, use_ttl, max_merge, min_merge); MergeBasedCounters counters(db, 0); testPartialMerge(&counters, db.get(), max_merge, min_merge, min_merge * 10); DestroyDB(dbname, Options()); } } } { std::cout << "Test merge-operator not set after reopen\n"; { auto db = OpenDb(dbname); MergeBasedCounters counters(db, 0); counters.add("test-key", 1); counters.add("test-key", 1); counters.add("test-key", 1); db->CompactRange(CompactRangeOptions(), nullptr, nullptr); } DB* reopen_db; ASSERT_OK(DB::Open(Options(), dbname, &reopen_db)); std::string value; ASSERT_TRUE(!(reopen_db->Get(ReadOptions(), "test-key", &value).ok())); delete reopen_db; DestroyDB(dbname, Options()); } /* Temporary remove this test { std::cout << "Test merge-operator not set after reopen (recovery case)\n"; { auto db = OpenDb(dbname); MergeBasedCounters counters(db, 0); counters.add("test-key", 1); counters.add("test-key", 1); counters.add("test-key", 1); } DB* reopen_db; ASSERT_TRUE(DB::Open(Options(), dbname, &reopen_db).IsInvalidArgument()); } */ } } // namespace int main(int argc, char *argv[]) { //TODO: Make this test like a general rocksdb unit-test rocksdb::port::InstallStackTraceHandler(); runTest(argc, test::TmpDir() + "/merge_testdb"); // DBWithTTL is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE runTest(argc, test::TmpDir() + "/merge_testdbttl", true); // Run test on TTL database #endif // !ROCKSDB_LITE printf("Passed all tests!\n"); return 0; } rocksdb-4.5.1/db/options_file_test.cc000066400000000000000000000071141267530634000176060ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include #include "db/db_impl.h" #include "db/db_test_util.h" #include "rocksdb/options.h" #include "rocksdb/table.h" #include "util/testharness.h" namespace rocksdb { class OptionsFileTest : public testing::Test { public: OptionsFileTest() : dbname_(test::TmpDir() + "/options_file_test") {} std::string dbname_; }; namespace { void UpdateOptionsFiles(DB* db, std::unordered_set* filename_history, int* options_files_count) { std::vector filenames; db->GetEnv()->GetChildren(db->GetName(), &filenames); uint64_t number; FileType type; *options_files_count = 0; for (auto filename : filenames) { if (ParseFileName(filename, &number, &type) && type == kOptionsFile) { filename_history->insert(filename); (*options_files_count)++; } } } // Verify whether the current Options Files are the latest ones. void VerifyOptionsFileName( DB* db, const std::unordered_set& past_filenames) { std::vector filenames; std::unordered_set current_filenames; db->GetEnv()->GetChildren(db->GetName(), &filenames); uint64_t number; FileType type; for (auto filename : filenames) { if (ParseFileName(filename, &number, &type) && type == kOptionsFile) { current_filenames.insert(filename); } } for (auto past_filename : past_filenames) { if (current_filenames.find(past_filename) != current_filenames.end()) { continue; } for (auto filename : current_filenames) { ASSERT_GT(filename, past_filename); } } } } // namespace TEST_F(OptionsFileTest, NumberOfOptionsFiles) { const int kReopenCount = 20; Options opt; opt.create_if_missing = true; DestroyDB(dbname_, opt); std::unordered_set filename_history; DB* db; for (int i = 0; i < kReopenCount; ++i) { ASSERT_OK(DB::Open(opt, dbname_, &db)); int num_options_files = 0; UpdateOptionsFiles(db, &filename_history, &num_options_files); ASSERT_GT(num_options_files, 0); ASSERT_LE(num_options_files, 2); // Make sure we always keep the latest option files. VerifyOptionsFileName(db, filename_history); delete db; } } TEST_F(OptionsFileTest, OptionsFileName) { const uint64_t kOptionsFileNum = 12345; uint64_t number; FileType type; auto options_file_name = OptionsFileName("", kOptionsFileNum); ASSERT_TRUE(ParseFileName(options_file_name, &number, &type, nullptr)); ASSERT_EQ(type, kOptionsFile); ASSERT_EQ(number, kOptionsFileNum); const uint64_t kTempOptionsFileNum = 54352; auto temp_options_file_name = TempOptionsFileName("", kTempOptionsFileNum); ASSERT_TRUE(ParseFileName(temp_options_file_name, &number, &type, nullptr)); ASSERT_NE(temp_options_file_name.find(kTempFileNameSuffix), std::string::npos); ASSERT_EQ(type, kTempFile); ASSERT_EQ(number, kTempOptionsFileNum); } } // namespace rocksdb int main(int argc, char** argv) { #if !(defined NDEBUG) || !defined(OS_WIN) ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); #else return 0; #endif // !(defined NDEBUG) || !defined(OS_WIN) } #else #include int main(int argc, char** argv) { printf("Skipped as Options file is not supported in RocksDBLite.\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/perf_context_test.cc000066400000000000000000000510731267530634000176170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include #include #include #include #include "rocksdb/db.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice_transform.h" #include "rocksdb/memtablerep.h" #include "util/histogram.h" #include "util/instrumented_mutex.h" #include "util/stop_watch.h" #include "util/testharness.h" #include "util/thread_status_util.h" #include "util/string_util.h" bool FLAGS_random_key = false; bool FLAGS_use_set_based_memetable = false; int FLAGS_total_keys = 100; int FLAGS_write_buffer_size = 1000000000; int FLAGS_max_write_buffer_number = 8; int FLAGS_min_write_buffer_number_to_merge = 7; bool FLAGS_verbose = false; // Path to the database on file system const std::string kDbName = rocksdb::test::TmpDir() + "/perf_context_test"; namespace rocksdb { std::shared_ptr OpenDb(bool read_only = false) { DB* db; Options options; options.create_if_missing = true; options.write_buffer_size = FLAGS_write_buffer_size; options.max_write_buffer_number = FLAGS_max_write_buffer_number; options.min_write_buffer_number_to_merge = FLAGS_min_write_buffer_number_to_merge; if (FLAGS_use_set_based_memetable) { #ifndef ROCKSDB_LITE options.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(0)); options.memtable_factory.reset(NewHashSkipListRepFactory()); #endif // ROCKSDB_LITE } Status s; if (!read_only) { s = DB::Open(options, kDbName, &db); } else { s = DB::OpenForReadOnly(options, kDbName, &db); } EXPECT_OK(s); return std::shared_ptr(db); } class PerfContextTest : public testing::Test {}; TEST_F(PerfContextTest, SeekIntoDeletion) { DestroyDB(kDbName, Options()); auto db = OpenDb(); WriteOptions write_options; ReadOptions read_options; for (int i = 0; i < FLAGS_total_keys; ++i) { std::string key = "k" + ToString(i); std::string value = "v" + ToString(i); db->Put(write_options, key, value); } for (int i = 0; i < FLAGS_total_keys -1 ; ++i) { std::string key = "k" + ToString(i); db->Delete(write_options, key); } HistogramImpl hist_get; HistogramImpl hist_get_time; for (int i = 0; i < FLAGS_total_keys - 1; ++i) { std::string key = "k" + ToString(i); std::string value; perf_context.Reset(); StopWatchNano timer(Env::Default()); timer.Start(); auto status = db->Get(read_options, key, &value); auto elapsed_nanos = timer.ElapsedNanos(); ASSERT_TRUE(status.IsNotFound()); hist_get.Add(perf_context.user_key_comparison_count); hist_get_time.Add(elapsed_nanos); } if (FLAGS_verbose) { std::cout << "Get user key comparison: \n" << hist_get.ToString() << "Get time: \n" << hist_get_time.ToString(); } { HistogramImpl hist_seek_to_first; std::unique_ptr iter(db->NewIterator(read_options)); perf_context.Reset(); StopWatchNano timer(Env::Default(), true); iter->SeekToFirst(); hist_seek_to_first.Add(perf_context.user_key_comparison_count); auto elapsed_nanos = timer.ElapsedNanos(); if (FLAGS_verbose) { std::cout << "SeekToFirst uesr key comparison: \n" << hist_seek_to_first.ToString() << "ikey skipped: " << perf_context.internal_key_skipped_count << "\n" << "idelete skipped: " << perf_context.internal_delete_skipped_count << "\n" << "elapsed: " << elapsed_nanos << "\n"; } } HistogramImpl hist_seek; for (int i = 0; i < FLAGS_total_keys; ++i) { std::unique_ptr iter(db->NewIterator(read_options)); std::string key = "k" + ToString(i); perf_context.Reset(); StopWatchNano timer(Env::Default(), true); iter->Seek(key); auto elapsed_nanos = timer.ElapsedNanos(); hist_seek.Add(perf_context.user_key_comparison_count); if (FLAGS_verbose) { std::cout << "seek cmp: " << perf_context.user_key_comparison_count << " ikey skipped " << perf_context.internal_key_skipped_count << " idelete skipped " << perf_context.internal_delete_skipped_count << " elapsed: " << elapsed_nanos << "ns\n"; } perf_context.Reset(); ASSERT_TRUE(iter->Valid()); StopWatchNano timer2(Env::Default(), true); iter->Next(); auto elapsed_nanos2 = timer2.ElapsedNanos(); if (FLAGS_verbose) { std::cout << "next cmp: " << perf_context.user_key_comparison_count << "elapsed: " << elapsed_nanos2 << "ns\n"; } } if (FLAGS_verbose) { std::cout << "Seek uesr key comparison: \n" << hist_seek.ToString(); } } TEST_F(PerfContextTest, StopWatchNanoOverhead) { // profile the timer cost by itself! const int kTotalIterations = 1000000; std::vector timings(kTotalIterations); StopWatchNano timer(Env::Default(), true); for (auto& timing : timings) { timing = timer.ElapsedNanos(true /* reset */); } HistogramImpl histogram; for (const auto timing : timings) { histogram.Add(timing); } if (FLAGS_verbose) { std::cout << histogram.ToString(); } } TEST_F(PerfContextTest, StopWatchOverhead) { // profile the timer cost by itself! const int kTotalIterations = 1000000; uint64_t elapsed = 0; std::vector timings(kTotalIterations); StopWatch timer(Env::Default(), nullptr, 0, &elapsed); for (auto& timing : timings) { timing = elapsed; } HistogramImpl histogram; uint64_t prev_timing = 0; for (const auto timing : timings) { histogram.Add(timing - prev_timing); prev_timing = timing; } if (FLAGS_verbose) { std::cout << histogram.ToString(); } } void ProfileQueries(bool enabled_time = false) { DestroyDB(kDbName, Options()); // Start this test with a fresh DB auto db = OpenDb(); WriteOptions write_options; ReadOptions read_options; HistogramImpl hist_put; HistogramImpl hist_get; HistogramImpl hist_get_snapshot; HistogramImpl hist_get_memtable; HistogramImpl hist_get_files; HistogramImpl hist_get_post_process; HistogramImpl hist_num_memtable_checked; HistogramImpl hist_mget; HistogramImpl hist_mget_snapshot; HistogramImpl hist_mget_memtable; HistogramImpl hist_mget_files; HistogramImpl hist_mget_post_process; HistogramImpl hist_mget_num_memtable_checked; HistogramImpl hist_write_pre_post; HistogramImpl hist_write_wal_time; HistogramImpl hist_write_memtable_time; uint64_t total_db_mutex_nanos = 0; if (FLAGS_verbose) { std::cout << "Inserting " << FLAGS_total_keys << " key/value pairs\n...\n"; } std::vector keys; const int kFlushFlag = -1; for (int i = 0; i < FLAGS_total_keys; ++i) { keys.push_back(i); if (i == FLAGS_total_keys / 2) { // Issuing a flush in the middle. keys.push_back(kFlushFlag); } } if (FLAGS_random_key) { std::random_shuffle(keys.begin(), keys.end()); } #ifndef NDEBUG ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 1U); #endif int num_mutex_waited = 0; for (const int i : keys) { if (i == kFlushFlag) { FlushOptions fo; db->Flush(fo); continue; } std::string key = "k" + ToString(i); std::string value = "v" + ToString(i); std::vector values; perf_context.Reset(); db->Put(write_options, key, value); if (++num_mutex_waited > 3) { #ifndef NDEBUG ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0U); #endif } hist_write_pre_post.Add(perf_context.write_pre_and_post_process_time); hist_write_wal_time.Add(perf_context.write_wal_time); hist_write_memtable_time.Add(perf_context.write_memtable_time); hist_put.Add(perf_context.user_key_comparison_count); total_db_mutex_nanos += perf_context.db_mutex_lock_nanos; } #ifndef NDEBUG ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0U); #endif for (const int i : keys) { std::string key = "k" + ToString(i); std::string value = "v" + ToString(i); std::vector multiget_keys = {Slice(key)}; std::vector values; perf_context.Reset(); db->Get(read_options, key, &value); hist_get_snapshot.Add(perf_context.get_snapshot_time); hist_get_memtable.Add(perf_context.get_from_memtable_time); hist_get_files.Add(perf_context.get_from_output_files_time); hist_num_memtable_checked.Add(perf_context.get_from_memtable_count); hist_get_post_process.Add(perf_context.get_post_process_time); hist_get.Add(perf_context.user_key_comparison_count); perf_context.Reset(); db->MultiGet(read_options, multiget_keys, &values); hist_mget_snapshot.Add(perf_context.get_snapshot_time); hist_mget_memtable.Add(perf_context.get_from_memtable_time); hist_mget_files.Add(perf_context.get_from_output_files_time); hist_mget_num_memtable_checked.Add(perf_context.get_from_memtable_count); hist_mget_post_process.Add(perf_context.get_post_process_time); hist_mget.Add(perf_context.user_key_comparison_count); } if (FLAGS_verbose) { std::cout << "Put uesr key comparison: \n" << hist_put.ToString() << "Get uesr key comparison: \n" << hist_get.ToString() << "MultiGet uesr key comparison: \n" << hist_get.ToString(); std::cout << "Put(): Pre and Post Process Time: \n" << hist_write_pre_post.ToString() << " Writing WAL time: \n" << hist_write_wal_time.ToString() << "\n" << " Writing Mem Table time: \n" << hist_write_memtable_time.ToString() << "\n" << " Total DB mutex nanos: \n" << total_db_mutex_nanos << "\n"; std::cout << "Get(): Time to get snapshot: \n" << hist_get_snapshot.ToString() << " Time to get value from memtables: \n" << hist_get_memtable.ToString() << "\n" << " Time to get value from output files: \n" << hist_get_files.ToString() << "\n" << " Number of memtables checked: \n" << hist_num_memtable_checked.ToString() << "\n" << " Time to post process: \n" << hist_get_post_process.ToString() << "\n"; std::cout << "MultiGet(): Time to get snapshot: \n" << hist_mget_snapshot.ToString() << " Time to get value from memtables: \n" << hist_mget_memtable.ToString() << "\n" << " Time to get value from output files: \n" << hist_mget_files.ToString() << "\n" << " Number of memtables checked: \n" << hist_mget_num_memtable_checked.ToString() << "\n" << " Time to post process: \n" << hist_mget_post_process.ToString() << "\n"; } if (enabled_time) { ASSERT_GT(hist_get.Average(), 0); ASSERT_GT(hist_get_snapshot.Average(), 0); ASSERT_GT(hist_get_memtable.Average(), 0); ASSERT_GT(hist_get_files.Average(), 0); ASSERT_GT(hist_get_post_process.Average(), 0); ASSERT_GT(hist_num_memtable_checked.Average(), 0); ASSERT_GT(hist_mget.Average(), 0); ASSERT_GT(hist_mget_snapshot.Average(), 0); ASSERT_GT(hist_mget_memtable.Average(), 0); ASSERT_GT(hist_mget_files.Average(), 0); ASSERT_GT(hist_mget_post_process.Average(), 0); ASSERT_GT(hist_mget_num_memtable_checked.Average(), 0); #ifndef NDEBUG ASSERT_GT(total_db_mutex_nanos, 2000U); #endif } db.reset(); db = OpenDb(true); hist_get.Clear(); hist_get_snapshot.Clear(); hist_get_memtable.Clear(); hist_get_files.Clear(); hist_get_post_process.Clear(); hist_num_memtable_checked.Clear(); hist_mget.Clear(); hist_mget_snapshot.Clear(); hist_mget_memtable.Clear(); hist_mget_files.Clear(); hist_mget_post_process.Clear(); hist_mget_num_memtable_checked.Clear(); for (const int i : keys) { std::string key = "k" + ToString(i); std::string value = "v" + ToString(i); std::vector multiget_keys = {Slice(key)}; std::vector values; perf_context.Reset(); db->Get(read_options, key, &value); hist_get_snapshot.Add(perf_context.get_snapshot_time); hist_get_memtable.Add(perf_context.get_from_memtable_time); hist_get_files.Add(perf_context.get_from_output_files_time); hist_num_memtable_checked.Add(perf_context.get_from_memtable_count); hist_get_post_process.Add(perf_context.get_post_process_time); hist_get.Add(perf_context.user_key_comparison_count); perf_context.Reset(); db->MultiGet(read_options, multiget_keys, &values); hist_mget_snapshot.Add(perf_context.get_snapshot_time); hist_mget_memtable.Add(perf_context.get_from_memtable_time); hist_mget_files.Add(perf_context.get_from_output_files_time); hist_mget_num_memtable_checked.Add(perf_context.get_from_memtable_count); hist_mget_post_process.Add(perf_context.get_post_process_time); hist_mget.Add(perf_context.user_key_comparison_count); } if (FLAGS_verbose) { std::cout << "ReadOnly Get uesr key comparison: \n" << hist_get.ToString() << "ReadOnly MultiGet uesr key comparison: \n" << hist_mget.ToString(); std::cout << "ReadOnly Get(): Time to get snapshot: \n" << hist_get_snapshot.ToString() << " Time to get value from memtables: \n" << hist_get_memtable.ToString() << "\n" << " Time to get value from output files: \n" << hist_get_files.ToString() << "\n" << " Number of memtables checked: \n" << hist_num_memtable_checked.ToString() << "\n" << " Time to post process: \n" << hist_get_post_process.ToString() << "\n"; std::cout << "ReadOnly MultiGet(): Time to get snapshot: \n" << hist_mget_snapshot.ToString() << " Time to get value from memtables: \n" << hist_mget_memtable.ToString() << "\n" << " Time to get value from output files: \n" << hist_mget_files.ToString() << "\n" << " Number of memtables checked: \n" << hist_mget_num_memtable_checked.ToString() << "\n" << " Time to post process: \n" << hist_mget_post_process.ToString() << "\n"; } if (enabled_time) { ASSERT_GT(hist_get.Average(), 0); ASSERT_GT(hist_get_memtable.Average(), 0); ASSERT_GT(hist_get_files.Average(), 0); ASSERT_GT(hist_num_memtable_checked.Average(), 0); // In read-only mode Get(), no super version operation is needed ASSERT_EQ(hist_get_post_process.Average(), 0); ASSERT_EQ(hist_get_snapshot.Average(), 0); ASSERT_GT(hist_mget.Average(), 0); ASSERT_GT(hist_mget_snapshot.Average(), 0); ASSERT_GT(hist_mget_memtable.Average(), 0); ASSERT_GT(hist_mget_files.Average(), 0); ASSERT_GT(hist_mget_post_process.Average(), 0); ASSERT_GT(hist_mget_num_memtable_checked.Average(), 0); } } #ifndef ROCKSDB_LITE TEST_F(PerfContextTest, KeyComparisonCount) { SetPerfLevel(kEnableCount); ProfileQueries(); SetPerfLevel(kDisable); ProfileQueries(); SetPerfLevel(kEnableTime); ProfileQueries(true); } #endif // ROCKSDB_LITE // make perf_context_test // export ROCKSDB_TESTS=PerfContextTest.SeekKeyComparison // For one memtable: // ./perf_context_test --write_buffer_size=500000 --total_keys=10000 // For two memtables: // ./perf_context_test --write_buffer_size=250000 --total_keys=10000 // Specify --random_key=1 to shuffle the key before insertion // Results show that, for sequential insertion, worst-case Seek Key comparison // is close to the total number of keys (linear), when there is only one // memtable. When there are two memtables, even the avg Seek Key comparison // starts to become linear to the input size. TEST_F(PerfContextTest, SeekKeyComparison) { DestroyDB(kDbName, Options()); auto db = OpenDb(); WriteOptions write_options; ReadOptions read_options; if (FLAGS_verbose) { std::cout << "Inserting " << FLAGS_total_keys << " key/value pairs\n...\n"; } std::vector keys; for (int i = 0; i < FLAGS_total_keys; ++i) { keys.push_back(i); } if (FLAGS_random_key) { std::random_shuffle(keys.begin(), keys.end()); } HistogramImpl hist_put_time; HistogramImpl hist_wal_time; HistogramImpl hist_time_diff; SetPerfLevel(kEnableTime); StopWatchNano timer(Env::Default()); for (const int i : keys) { std::string key = "k" + ToString(i); std::string value = "v" + ToString(i); perf_context.Reset(); timer.Start(); db->Put(write_options, key, value); auto put_time = timer.ElapsedNanos(); hist_put_time.Add(put_time); hist_wal_time.Add(perf_context.write_wal_time); hist_time_diff.Add(put_time - perf_context.write_wal_time); } if (FLAGS_verbose) { std::cout << "Put time:\n" << hist_put_time.ToString() << "WAL time:\n" << hist_wal_time.ToString() << "time diff:\n" << hist_time_diff.ToString(); } HistogramImpl hist_seek; HistogramImpl hist_next; for (int i = 0; i < FLAGS_total_keys; ++i) { std::string key = "k" + ToString(i); std::string value = "v" + ToString(i); std::unique_ptr iter(db->NewIterator(read_options)); perf_context.Reset(); iter->Seek(key); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(iter->value().ToString(), value); hist_seek.Add(perf_context.user_key_comparison_count); } std::unique_ptr iter(db->NewIterator(read_options)); for (iter->SeekToFirst(); iter->Valid();) { perf_context.Reset(); iter->Next(); hist_next.Add(perf_context.user_key_comparison_count); } if (FLAGS_verbose) { std::cout << "Seek:\n" << hist_seek.ToString() << "Next:\n" << hist_next.ToString(); } } TEST_F(PerfContextTest, DBMutexLockCounter) { int stats_code[] = {0, static_cast(DB_MUTEX_WAIT_MICROS)}; for (PerfLevel perf_level : {PerfLevel::kEnableTimeExceptForMutex, PerfLevel::kEnableTime}) { for (int c = 0; c < 2; ++c) { InstrumentedMutex mutex(nullptr, Env::Default(), stats_code[c]); mutex.Lock(); std::thread child_thread([&] { SetPerfLevel(perf_level); perf_context.Reset(); ASSERT_EQ(perf_context.db_mutex_lock_nanos, 0); mutex.Lock(); mutex.Unlock(); if (perf_level == PerfLevel::kEnableTimeExceptForMutex || stats_code[c] != DB_MUTEX_WAIT_MICROS) { ASSERT_EQ(perf_context.db_mutex_lock_nanos, 0); } else { // increment the counter only when it's a DB Mutex ASSERT_GT(perf_context.db_mutex_lock_nanos, 0); } }); Env::Default()->SleepForMicroseconds(100); mutex.Unlock(); child_thread.join(); } } } TEST_F(PerfContextTest, FalseDBMutexWait) { SetPerfLevel(kEnableTime); int stats_code[] = {0, static_cast(DB_MUTEX_WAIT_MICROS)}; for (int c = 0; c < 2; ++c) { InstrumentedMutex mutex(nullptr, Env::Default(), stats_code[c]); InstrumentedCondVar lock(&mutex); perf_context.Reset(); mutex.Lock(); lock.TimedWait(100); mutex.Unlock(); if (stats_code[c] == static_cast(DB_MUTEX_WAIT_MICROS)) { // increment the counter only when it's a DB Mutex ASSERT_GT(perf_context.db_condition_wait_nanos, 0); } else { ASSERT_EQ(perf_context.db_condition_wait_nanos, 0); } } } TEST_F(PerfContextTest, ToString) { perf_context.Reset(); perf_context.block_read_count = 12345; std::string zero_included = perf_context.ToString(); ASSERT_NE(std::string::npos, zero_included.find("= 0")); ASSERT_NE(std::string::npos, zero_included.find("= 12345")); std::string zero_excluded = perf_context.ToString(true); ASSERT_EQ(std::string::npos, zero_excluded.find("= 0")); ASSERT_NE(std::string::npos, zero_excluded.find("= 12345")); } } int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); for (int i = 1; i < argc; i++) { int n; char junk; if (sscanf(argv[i], "--write_buffer_size=%d%c", &n, &junk) == 1) { FLAGS_write_buffer_size = n; } if (sscanf(argv[i], "--total_keys=%d%c", &n, &junk) == 1) { FLAGS_total_keys = n; } if (sscanf(argv[i], "--random_key=%d%c", &n, &junk) == 1 && (n == 0 || n == 1)) { FLAGS_random_key = n; } if (sscanf(argv[i], "--use_set_based_memetable=%d%c", &n, &junk) == 1 && (n == 0 || n == 1)) { FLAGS_use_set_based_memetable = n; } if (sscanf(argv[i], "--verbose=%d%c", &n, &junk) == 1 && (n == 0 || n == 1)) { FLAGS_verbose = n; } } if (FLAGS_verbose) { std::cout << kDbName << "\n"; } return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/plain_table_db_test.cc000066400000000000000000001161531267530634000200370ustar00rootroot00000000000000// Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include #include #include "db/db_impl.h" #include "db/filename.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "table/meta_blocks.h" #include "table/bloom_block.h" #include "table/table_builder.h" #include "table/plain_table_factory.h" #include "table/plain_table_key_coding.h" #include "table/plain_table_reader.h" #include "util/hash.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" #include "utilities/merge_operators.h" using std::unique_ptr; namespace rocksdb { class PlainTableKeyDecoderTest : public testing::Test {}; TEST_F(PlainTableKeyDecoderTest, ReadNonMmap) { std::string tmp; Random rnd(301); const uint32_t kLength = 2222; Slice contents = test::RandomString(&rnd, kLength, &tmp); test::StringSource* string_source = new test::StringSource(contents, 0, false); unique_ptr file_reader( test::GetRandomAccessFileReader(string_source)); unique_ptr file_info(new PlainTableReaderFileInfo( std::move(file_reader), EnvOptions(), kLength)); { PlainTableFileReader reader(file_info.get()); const uint32_t kReadSize = 77; for (uint32_t pos = 0; pos < kLength; pos += kReadSize) { uint32_t read_size = std::min(kLength - pos, kReadSize); Slice out; ASSERT_TRUE(reader.Read(pos, read_size, &out)); ASSERT_EQ(0, out.compare(tmp.substr(pos, read_size))); } ASSERT_LT(uint32_t(string_source->total_reads()), kLength / kReadSize / 2); } std::vector>> reads = { {{600, 30}, {590, 30}, {600, 20}, {600, 40}}, {{800, 20}, {100, 20}, {500, 20}, {1500, 20}, {100, 20}, {80, 20}}, {{1000, 20}, {500, 20}, {1000, 50}}, {{1000, 20}, {500, 20}, {500, 20}}, {{1000, 20}, {500, 20}, {200, 20}, {500, 20}}, {{1000, 20}, {500, 20}, {200, 20}, {1000, 50}}, {{600, 500}, {610, 20}, {100, 20}}, {{500, 100}, {490, 100}, {550, 50}}, }; std::vector num_file_reads = {2, 6, 2, 2, 4, 3, 2, 2}; for (size_t i = 0; i < reads.size(); i++) { string_source->set_total_reads(0); PlainTableFileReader reader(file_info.get()); for (auto p : reads[i]) { Slice out; ASSERT_TRUE(reader.Read(p.first, p.second, &out)); ASSERT_EQ(0, out.compare(tmp.substr(p.first, p.second))); } ASSERT_EQ(num_file_reads[i], string_source->total_reads()); } } class PlainTableDBTest : public testing::Test, public testing::WithParamInterface { protected: private: std::string dbname_; Env* env_; DB* db_; bool mmap_mode_; Options last_options_; public: PlainTableDBTest() : env_(Env::Default()) {} ~PlainTableDBTest() { delete db_; EXPECT_OK(DestroyDB(dbname_, Options())); } void SetUp() override { mmap_mode_ = GetParam(); dbname_ = test::TmpDir() + "/plain_table_db_test"; EXPECT_OK(DestroyDB(dbname_, Options())); db_ = nullptr; Reopen(); } // Return the current option configuration. Options CurrentOptions() { Options options; PlainTableOptions plain_table_options; plain_table_options.user_key_len = 0; plain_table_options.bloom_bits_per_key = 2; plain_table_options.hash_table_ratio = 0.8; plain_table_options.index_sparseness = 3; plain_table_options.huge_page_tlb_size = 0; plain_table_options.encoding_type = kPrefix; plain_table_options.full_scan_mode = false; plain_table_options.store_index_in_file = false; options.table_factory.reset(NewPlainTableFactory(plain_table_options)); options.memtable_factory.reset(NewHashLinkListRepFactory(4, 0, 3, true)); options.prefix_extractor.reset(NewFixedPrefixTransform(8)); options.allow_mmap_reads = mmap_mode_; return options; } DBImpl* dbfull() { return reinterpret_cast(db_); } void Reopen(Options* options = nullptr) { ASSERT_OK(TryReopen(options)); } void Close() { delete db_; db_ = nullptr; } void DestroyAndReopen(Options* options = nullptr) { //Destroy using last options Destroy(&last_options_); ASSERT_OK(TryReopen(options)); } void Destroy(Options* options) { delete db_; db_ = nullptr; ASSERT_OK(DestroyDB(dbname_, *options)); } Status PureReopen(Options* options, DB** db) { return DB::Open(*options, dbname_, db); } Status TryReopen(Options* options = nullptr) { delete db_; db_ = nullptr; Options opts; if (options != nullptr) { opts = *options; } else { opts = CurrentOptions(); opts.create_if_missing = true; } last_options_ = opts; return DB::Open(opts, dbname_, &db_); } Status Put(const Slice& k, const Slice& v) { return db_->Put(WriteOptions(), k, v); } Status Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); } std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) { ReadOptions options; options.snapshot = snapshot; std::string result; Status s = db_->Get(options, k, &result); if (s.IsNotFound()) { result = "NOT_FOUND"; } else if (!s.ok()) { result = s.ToString(); } return result; } int NumTableFilesAtLevel(int level) { std::string property; EXPECT_TRUE(db_->GetProperty( "rocksdb.num-files-at-level" + NumberToString(level), &property)); return atoi(property.c_str()); } // Return spread of files per level std::string FilesPerLevel() { std::string result; size_t last_non_zero_offset = 0; for (int level = 0; level < db_->NumberLevels(); level++) { int f = NumTableFilesAtLevel(level); char buf[100]; snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); result += buf; if (f > 0) { last_non_zero_offset = result.size(); } } result.resize(last_non_zero_offset); return result; } std::string IterStatus(Iterator* iter) { std::string result; if (iter->Valid()) { result = iter->key().ToString() + "->" + iter->value().ToString(); } else { result = "(invalid)"; } return result; } }; TEST_P(PlainTableDBTest, Empty) { ASSERT_TRUE(dbfull() != nullptr); ASSERT_EQ("NOT_FOUND", Get("0000000000000foo")); } extern const uint64_t kPlainTableMagicNumber; class TestPlainTableReader : public PlainTableReader { public: TestPlainTableReader(const EnvOptions& env_options, const InternalKeyComparator& icomparator, EncodingType encoding_type, uint64_t file_size, int bloom_bits_per_key, double hash_table_ratio, size_t index_sparseness, const TableProperties* table_properties, unique_ptr&& file, const ImmutableCFOptions& ioptions, bool* expect_bloom_not_match, bool store_index_in_file) : PlainTableReader(ioptions, std::move(file), env_options, icomparator, encoding_type, file_size, table_properties), expect_bloom_not_match_(expect_bloom_not_match) { Status s = MmapDataIfNeeded(); EXPECT_TRUE(s.ok()); s = PopulateIndex(const_cast(table_properties), bloom_bits_per_key, hash_table_ratio, index_sparseness, 2 * 1024 * 1024); EXPECT_TRUE(s.ok()); TableProperties* props = const_cast(table_properties); if (store_index_in_file) { auto bloom_version_ptr = props->user_collected_properties.find( PlainTablePropertyNames::kBloomVersion); EXPECT_TRUE(bloom_version_ptr != props->user_collected_properties.end()); EXPECT_EQ(bloom_version_ptr->second, std::string("1")); if (ioptions.bloom_locality > 0) { auto num_blocks_ptr = props->user_collected_properties.find( PlainTablePropertyNames::kNumBloomBlocks); EXPECT_TRUE(num_blocks_ptr != props->user_collected_properties.end()); } } } virtual ~TestPlainTableReader() {} private: virtual bool MatchBloom(uint32_t hash) const override { bool ret = PlainTableReader::MatchBloom(hash); if (*expect_bloom_not_match_) { EXPECT_TRUE(!ret); } else { EXPECT_TRUE(ret); } return ret; } bool* expect_bloom_not_match_; }; extern const uint64_t kPlainTableMagicNumber; class TestPlainTableFactory : public PlainTableFactory { public: explicit TestPlainTableFactory(bool* expect_bloom_not_match, const PlainTableOptions& options) : PlainTableFactory(options), bloom_bits_per_key_(options.bloom_bits_per_key), hash_table_ratio_(options.hash_table_ratio), index_sparseness_(options.index_sparseness), store_index_in_file_(options.store_index_in_file), expect_bloom_not_match_(expect_bloom_not_match) {} Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table) const override { TableProperties* props = nullptr; auto s = ReadTableProperties(file.get(), file_size, kPlainTableMagicNumber, table_reader_options.ioptions.env, table_reader_options.ioptions.info_log, &props); EXPECT_TRUE(s.ok()); if (store_index_in_file_) { BlockHandle bloom_block_handle; s = FindMetaBlock(file.get(), file_size, kPlainTableMagicNumber, table_reader_options.ioptions.env, BloomBlockBuilder::kBloomBlock, &bloom_block_handle); EXPECT_TRUE(s.ok()); BlockHandle index_block_handle; s = FindMetaBlock(file.get(), file_size, kPlainTableMagicNumber, table_reader_options.ioptions.env, PlainTableIndexBuilder::kPlainTableIndexBlock, &index_block_handle); EXPECT_TRUE(s.ok()); } auto& user_props = props->user_collected_properties; auto encoding_type_prop = user_props.find(PlainTablePropertyNames::kEncodingType); assert(encoding_type_prop != user_props.end()); EncodingType encoding_type = static_cast( DecodeFixed32(encoding_type_prop->second.c_str())); std::unique_ptr new_reader(new TestPlainTableReader( table_reader_options.env_options, table_reader_options.internal_comparator, encoding_type, file_size, bloom_bits_per_key_, hash_table_ratio_, index_sparseness_, props, std::move(file), table_reader_options.ioptions, expect_bloom_not_match_, store_index_in_file_)); *table = std::move(new_reader); return s; } private: int bloom_bits_per_key_; double hash_table_ratio_; size_t index_sparseness_; bool store_index_in_file_; bool* expect_bloom_not_match_; }; TEST_P(PlainTableDBTest, Flush) { for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024; huge_page_tlb_size += 2 * 1024 * 1024) { for (EncodingType encoding_type : {kPlain, kPrefix}) { for (int bloom_bits = 0; bloom_bits <= 117; bloom_bits += 117) { for (int total_order = 0; total_order <= 1; total_order++) { for (int store_index_in_file = 0; store_index_in_file <= 1; ++store_index_in_file) { if (!bloom_bits && store_index_in_file) { continue; } Options options = CurrentOptions(); options.create_if_missing = true; // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 if (total_order) { options.prefix_extractor.reset(); PlainTableOptions plain_table_options; plain_table_options.user_key_len = 0; plain_table_options.bloom_bits_per_key = bloom_bits; plain_table_options.hash_table_ratio = 0; plain_table_options.index_sparseness = 2; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; plain_table_options.encoding_type = encoding_type; plain_table_options.full_scan_mode = false; plain_table_options.store_index_in_file = store_index_in_file; options.table_factory.reset( NewPlainTableFactory(plain_table_options)); } else { PlainTableOptions plain_table_options; plain_table_options.user_key_len = 0; plain_table_options.bloom_bits_per_key = bloom_bits; plain_table_options.hash_table_ratio = 0.75; plain_table_options.index_sparseness = 16; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; plain_table_options.encoding_type = encoding_type; plain_table_options.full_scan_mode = false; plain_table_options.store_index_in_file = store_index_in_file; options.table_factory.reset( NewPlainTableFactory(plain_table_options)); } DestroyAndReopen(&options); uint64_t int_num; ASSERT_TRUE(dbfull()->GetIntProperty( "rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_EQ(int_num, 0U); ASSERT_OK(Put("1000000000000foo", "v1")); ASSERT_OK(Put("0000000000000bar", "v2")); ASSERT_OK(Put("1000000000000foo", "v3")); dbfull()->TEST_FlushMemTable(); ASSERT_TRUE(dbfull()->GetIntProperty( "rocksdb.estimate-table-readers-mem", &int_num)); ASSERT_GT(int_num, 0U); TablePropertiesCollection ptc; reinterpret_cast(dbfull())->GetPropertiesOfAllTables(&ptc); ASSERT_EQ(1U, ptc.size()); auto row = ptc.begin(); auto tp = row->second; if (!store_index_in_file) { ASSERT_EQ(total_order ? "4" : "12", (tp->user_collected_properties) .at("plain_table_hash_table_size")); ASSERT_EQ("0", (tp->user_collected_properties) .at("plain_table_sub_index_size")); } else { ASSERT_EQ("0", (tp->user_collected_properties) .at("plain_table_hash_table_size")); ASSERT_EQ("0", (tp->user_collected_properties) .at("plain_table_sub_index_size")); } ASSERT_EQ("v3", Get("1000000000000foo")); ASSERT_EQ("v2", Get("0000000000000bar")); } } } } } } TEST_P(PlainTableDBTest, Flush2) { for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024; huge_page_tlb_size += 2 * 1024 * 1024) { for (EncodingType encoding_type : {kPlain, kPrefix}) { for (int bloom_bits = 0; bloom_bits <= 117; bloom_bits += 117) { for (int total_order = 0; total_order <= 1; total_order++) { for (int store_index_in_file = 0; store_index_in_file <= 1; ++store_index_in_file) { if (encoding_type == kPrefix && total_order) { continue; } if (!bloom_bits && store_index_in_file) { continue; } if (total_order && store_index_in_file) { continue; } bool expect_bloom_not_match = false; Options options = CurrentOptions(); options.create_if_missing = true; // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 PlainTableOptions plain_table_options; if (total_order) { options.prefix_extractor = nullptr; plain_table_options.hash_table_ratio = 0; plain_table_options.index_sparseness = 2; } else { plain_table_options.hash_table_ratio = 0.75; plain_table_options.index_sparseness = 16; } plain_table_options.user_key_len = kPlainTableVariableLength; plain_table_options.bloom_bits_per_key = bloom_bits; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; plain_table_options.encoding_type = encoding_type; plain_table_options.store_index_in_file = store_index_in_file; options.table_factory.reset(new TestPlainTableFactory( &expect_bloom_not_match, plain_table_options)); DestroyAndReopen(&options); ASSERT_OK(Put("0000000000000bar", "b")); ASSERT_OK(Put("1000000000000foo", "v1")); dbfull()->TEST_FlushMemTable(); ASSERT_OK(Put("1000000000000foo", "v2")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v2", Get("1000000000000foo")); ASSERT_OK(Put("0000000000000eee", "v3")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v3", Get("0000000000000eee")); ASSERT_OK(Delete("0000000000000bar")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("NOT_FOUND", Get("0000000000000bar")); ASSERT_OK(Put("0000000000000eee", "v5")); ASSERT_OK(Put("9000000000000eee", "v5")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v5", Get("0000000000000eee")); // Test Bloom Filter if (bloom_bits > 0) { // Neither key nor value should exist. expect_bloom_not_match = true; ASSERT_EQ("NOT_FOUND", Get("5_not00000000bar")); // Key doesn't exist any more but prefix exists. if (total_order) { ASSERT_EQ("NOT_FOUND", Get("1000000000000not")); ASSERT_EQ("NOT_FOUND", Get("0000000000000not")); } expect_bloom_not_match = false; } } } } } } } TEST_P(PlainTableDBTest, Iterator) { for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024; huge_page_tlb_size += 2 * 1024 * 1024) { for (EncodingType encoding_type : {kPlain, kPrefix}) { for (int bloom_bits = 0; bloom_bits <= 117; bloom_bits += 117) { for (int total_order = 0; total_order <= 1; total_order++) { if (encoding_type == kPrefix && total_order == 1) { continue; } bool expect_bloom_not_match = false; Options options = CurrentOptions(); options.create_if_missing = true; // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 if (total_order) { options.prefix_extractor = nullptr; PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = bloom_bits; plain_table_options.hash_table_ratio = 0; plain_table_options.index_sparseness = 2; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; plain_table_options.encoding_type = encoding_type; options.table_factory.reset(new TestPlainTableFactory( &expect_bloom_not_match, plain_table_options)); } else { PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = bloom_bits; plain_table_options.hash_table_ratio = 0.75; plain_table_options.index_sparseness = 16; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; plain_table_options.encoding_type = encoding_type; options.table_factory.reset(new TestPlainTableFactory( &expect_bloom_not_match, plain_table_options)); } DestroyAndReopen(&options); ASSERT_OK(Put("1000000000foo002", "v_2")); ASSERT_OK(Put("0000000000000bar", "random")); ASSERT_OK(Put("1000000000foo001", "v1")); ASSERT_OK(Put("3000000000000bar", "bar_v")); ASSERT_OK(Put("1000000000foo003", "v__3")); ASSERT_OK(Put("1000000000foo004", "v__4")); ASSERT_OK(Put("1000000000foo005", "v__5")); ASSERT_OK(Put("1000000000foo007", "v__7")); ASSERT_OK(Put("1000000000foo008", "v__8")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get("1000000000foo001")); ASSERT_EQ("v__3", Get("1000000000foo003")); Iterator* iter = dbfull()->NewIterator(ReadOptions()); iter->Seek("1000000000foo000"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo001", iter->key().ToString()); ASSERT_EQ("v1", iter->value().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo002", iter->key().ToString()); ASSERT_EQ("v_2", iter->value().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo003", iter->key().ToString()); ASSERT_EQ("v__3", iter->value().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo004", iter->key().ToString()); ASSERT_EQ("v__4", iter->value().ToString()); iter->Seek("3000000000000bar"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("3000000000000bar", iter->key().ToString()); ASSERT_EQ("bar_v", iter->value().ToString()); iter->Seek("1000000000foo000"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo001", iter->key().ToString()); ASSERT_EQ("v1", iter->value().ToString()); iter->Seek("1000000000foo005"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo005", iter->key().ToString()); ASSERT_EQ("v__5", iter->value().ToString()); iter->Seek("1000000000foo006"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo007", iter->key().ToString()); ASSERT_EQ("v__7", iter->value().ToString()); iter->Seek("1000000000foo008"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo008", iter->key().ToString()); ASSERT_EQ("v__8", iter->value().ToString()); if (total_order == 0) { iter->Seek("1000000000foo009"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("3000000000000bar", iter->key().ToString()); } // Test Bloom Filter if (bloom_bits > 0) { if (!total_order) { // Neither key nor value should exist. expect_bloom_not_match = true; iter->Seek("2not000000000bar"); ASSERT_TRUE(!iter->Valid()); ASSERT_EQ("NOT_FOUND", Get("2not000000000bar")); expect_bloom_not_match = false; } else { expect_bloom_not_match = true; ASSERT_EQ("NOT_FOUND", Get("2not000000000bar")); expect_bloom_not_match = false; } } delete iter; } } } } } namespace { std::string MakeLongKey(size_t length, char c) { return std::string(length, c); } } // namespace TEST_P(PlainTableDBTest, IteratorLargeKeys) { Options options = CurrentOptions(); PlainTableOptions plain_table_options; plain_table_options.user_key_len = 0; plain_table_options.bloom_bits_per_key = 0; plain_table_options.hash_table_ratio = 0; options.table_factory.reset(NewPlainTableFactory(plain_table_options)); options.create_if_missing = true; options.prefix_extractor.reset(); DestroyAndReopen(&options); std::string key_list[] = { MakeLongKey(30, '0'), MakeLongKey(16, '1'), MakeLongKey(32, '2'), MakeLongKey(60, '3'), MakeLongKey(90, '4'), MakeLongKey(50, '5'), MakeLongKey(26, '6') }; for (size_t i = 0; i < 7; i++) { ASSERT_OK(Put(key_list[i], ToString(i))); } dbfull()->TEST_FlushMemTable(); Iterator* iter = dbfull()->NewIterator(ReadOptions()); iter->Seek(key_list[0]); for (size_t i = 0; i < 7; i++) { ASSERT_TRUE(iter->Valid()); ASSERT_EQ(key_list[i], iter->key().ToString()); ASSERT_EQ(ToString(i), iter->value().ToString()); iter->Next(); } ASSERT_TRUE(!iter->Valid()); delete iter; } namespace { std::string MakeLongKeyWithPrefix(size_t length, char c) { return "00000000" + std::string(length - 8, c); } } // namespace TEST_P(PlainTableDBTest, IteratorLargeKeysWithPrefix) { Options options = CurrentOptions(); PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = 0; plain_table_options.hash_table_ratio = 0.8; plain_table_options.index_sparseness = 3; plain_table_options.huge_page_tlb_size = 0; plain_table_options.encoding_type = kPrefix; options.table_factory.reset(NewPlainTableFactory(plain_table_options)); options.create_if_missing = true; DestroyAndReopen(&options); std::string key_list[] = { MakeLongKeyWithPrefix(30, '0'), MakeLongKeyWithPrefix(16, '1'), MakeLongKeyWithPrefix(32, '2'), MakeLongKeyWithPrefix(60, '3'), MakeLongKeyWithPrefix(90, '4'), MakeLongKeyWithPrefix(50, '5'), MakeLongKeyWithPrefix(26, '6')}; for (size_t i = 0; i < 7; i++) { ASSERT_OK(Put(key_list[i], ToString(i))); } dbfull()->TEST_FlushMemTable(); Iterator* iter = dbfull()->NewIterator(ReadOptions()); iter->Seek(key_list[0]); for (size_t i = 0; i < 7; i++) { ASSERT_TRUE(iter->Valid()); ASSERT_EQ(key_list[i], iter->key().ToString()); ASSERT_EQ(ToString(i), iter->value().ToString()); iter->Next(); } ASSERT_TRUE(!iter->Valid()); delete iter; } TEST_P(PlainTableDBTest, IteratorReverseSuffixComparator) { Options options = CurrentOptions(); options.create_if_missing = true; // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 test::SimpleSuffixReverseComparator comp; options.comparator = ∁ DestroyAndReopen(&options); ASSERT_OK(Put("1000000000foo002", "v_2")); ASSERT_OK(Put("0000000000000bar", "random")); ASSERT_OK(Put("1000000000foo001", "v1")); ASSERT_OK(Put("3000000000000bar", "bar_v")); ASSERT_OK(Put("1000000000foo003", "v__3")); ASSERT_OK(Put("1000000000foo004", "v__4")); ASSERT_OK(Put("1000000000foo005", "v__5")); ASSERT_OK(Put("1000000000foo007", "v__7")); ASSERT_OK(Put("1000000000foo008", "v__8")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get("1000000000foo001")); ASSERT_EQ("v__3", Get("1000000000foo003")); Iterator* iter = dbfull()->NewIterator(ReadOptions()); iter->Seek("1000000000foo009"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo008", iter->key().ToString()); ASSERT_EQ("v__8", iter->value().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo007", iter->key().ToString()); ASSERT_EQ("v__7", iter->value().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo005", iter->key().ToString()); ASSERT_EQ("v__5", iter->value().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo004", iter->key().ToString()); ASSERT_EQ("v__4", iter->value().ToString()); iter->Seek("3000000000000bar"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("3000000000000bar", iter->key().ToString()); ASSERT_EQ("bar_v", iter->value().ToString()); iter->Seek("1000000000foo005"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo005", iter->key().ToString()); ASSERT_EQ("v__5", iter->value().ToString()); iter->Seek("1000000000foo006"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo005", iter->key().ToString()); ASSERT_EQ("v__5", iter->value().ToString()); iter->Seek("1000000000foo008"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("1000000000foo008", iter->key().ToString()); ASSERT_EQ("v__8", iter->value().ToString()); iter->Seek("1000000000foo000"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("3000000000000bar", iter->key().ToString()); delete iter; } TEST_P(PlainTableDBTest, HashBucketConflict) { for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024; huge_page_tlb_size += 2 * 1024 * 1024) { for (unsigned char i = 1; i <= 3; i++) { Options options = CurrentOptions(); options.create_if_missing = true; // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = 0; plain_table_options.hash_table_ratio = 0; plain_table_options.index_sparseness = 2 ^ i; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; options.table_factory.reset(NewPlainTableFactory(plain_table_options)); DestroyAndReopen(&options); ASSERT_OK(Put("5000000000000fo0", "v1")); ASSERT_OK(Put("5000000000000fo1", "v2")); ASSERT_OK(Put("5000000000000fo2", "v")); ASSERT_OK(Put("2000000000000fo0", "v3")); ASSERT_OK(Put("2000000000000fo1", "v4")); ASSERT_OK(Put("2000000000000fo2", "v")); ASSERT_OK(Put("2000000000000fo3", "v")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get("5000000000000fo0")); ASSERT_EQ("v2", Get("5000000000000fo1")); ASSERT_EQ("v3", Get("2000000000000fo0")); ASSERT_EQ("v4", Get("2000000000000fo1")); ASSERT_EQ("NOT_FOUND", Get("5000000000000bar")); ASSERT_EQ("NOT_FOUND", Get("2000000000000bar")); ASSERT_EQ("NOT_FOUND", Get("5000000000000fo8")); ASSERT_EQ("NOT_FOUND", Get("2000000000000fo8")); ReadOptions ro; Iterator* iter = dbfull()->NewIterator(ro); iter->Seek("5000000000000fo0"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo0", iter->key().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo1", iter->key().ToString()); iter->Seek("5000000000000fo1"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo1", iter->key().ToString()); iter->Seek("2000000000000fo0"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo0", iter->key().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo1", iter->key().ToString()); iter->Seek("2000000000000fo1"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo1", iter->key().ToString()); iter->Seek("2000000000000bar"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo0", iter->key().ToString()); iter->Seek("5000000000000bar"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo0", iter->key().ToString()); iter->Seek("2000000000000fo8"); ASSERT_TRUE(!iter->Valid() || options.comparator->Compare(iter->key(), "20000001") > 0); iter->Seek("5000000000000fo8"); ASSERT_TRUE(!iter->Valid()); iter->Seek("1000000000000fo2"); ASSERT_TRUE(!iter->Valid()); iter->Seek("3000000000000fo2"); ASSERT_TRUE(!iter->Valid()); iter->Seek("8000000000000fo2"); ASSERT_TRUE(!iter->Valid()); delete iter; } } } TEST_P(PlainTableDBTest, HashBucketConflictReverseSuffixComparator) { for (size_t huge_page_tlb_size = 0; huge_page_tlb_size <= 2 * 1024 * 1024; huge_page_tlb_size += 2 * 1024 * 1024) { for (unsigned char i = 1; i <= 3; i++) { Options options = CurrentOptions(); options.create_if_missing = true; test::SimpleSuffixReverseComparator comp; options.comparator = ∁ // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = 0; plain_table_options.hash_table_ratio = 0; plain_table_options.index_sparseness = 2 ^ i; plain_table_options.huge_page_tlb_size = huge_page_tlb_size; options.table_factory.reset(NewPlainTableFactory(plain_table_options)); DestroyAndReopen(&options); ASSERT_OK(Put("5000000000000fo0", "v1")); ASSERT_OK(Put("5000000000000fo1", "v2")); ASSERT_OK(Put("5000000000000fo2", "v")); ASSERT_OK(Put("2000000000000fo0", "v3")); ASSERT_OK(Put("2000000000000fo1", "v4")); ASSERT_OK(Put("2000000000000fo2", "v")); ASSERT_OK(Put("2000000000000fo3", "v")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get("5000000000000fo0")); ASSERT_EQ("v2", Get("5000000000000fo1")); ASSERT_EQ("v3", Get("2000000000000fo0")); ASSERT_EQ("v4", Get("2000000000000fo1")); ASSERT_EQ("NOT_FOUND", Get("5000000000000bar")); ASSERT_EQ("NOT_FOUND", Get("2000000000000bar")); ASSERT_EQ("NOT_FOUND", Get("5000000000000fo8")); ASSERT_EQ("NOT_FOUND", Get("2000000000000fo8")); ReadOptions ro; Iterator* iter = dbfull()->NewIterator(ro); iter->Seek("5000000000000fo1"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo1", iter->key().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo0", iter->key().ToString()); iter->Seek("5000000000000fo1"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo1", iter->key().ToString()); iter->Seek("2000000000000fo1"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo1", iter->key().ToString()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo0", iter->key().ToString()); iter->Seek("2000000000000fo1"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo1", iter->key().ToString()); iter->Seek("2000000000000var"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("2000000000000fo3", iter->key().ToString()); iter->Seek("5000000000000var"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo2", iter->key().ToString()); std::string seek_key = "2000000000000bar"; iter->Seek(seek_key); ASSERT_TRUE(!iter->Valid() || options.prefix_extractor->Transform(iter->key()) != options.prefix_extractor->Transform(seek_key)); iter->Seek("1000000000000fo2"); ASSERT_TRUE(!iter->Valid()); iter->Seek("3000000000000fo2"); ASSERT_TRUE(!iter->Valid()); iter->Seek("8000000000000fo2"); ASSERT_TRUE(!iter->Valid()); delete iter; } } } TEST_P(PlainTableDBTest, NonExistingKeyToNonEmptyBucket) { Options options = CurrentOptions(); options.create_if_missing = true; // Set only one bucket to force bucket conflict. // Test index interval for the same prefix to be 1, 2 and 4 PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = 0; plain_table_options.hash_table_ratio = 0; plain_table_options.index_sparseness = 5; options.table_factory.reset(NewPlainTableFactory(plain_table_options)); DestroyAndReopen(&options); ASSERT_OK(Put("5000000000000fo0", "v1")); ASSERT_OK(Put("5000000000000fo1", "v2")); ASSERT_OK(Put("5000000000000fo2", "v3")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v1", Get("5000000000000fo0")); ASSERT_EQ("v2", Get("5000000000000fo1")); ASSERT_EQ("v3", Get("5000000000000fo2")); ASSERT_EQ("NOT_FOUND", Get("8000000000000bar")); ASSERT_EQ("NOT_FOUND", Get("1000000000000bar")); Iterator* iter = dbfull()->NewIterator(ReadOptions()); iter->Seek("5000000000000bar"); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("5000000000000fo0", iter->key().ToString()); iter->Seek("5000000000000fo8"); ASSERT_TRUE(!iter->Valid()); iter->Seek("1000000000000fo2"); ASSERT_TRUE(!iter->Valid()); iter->Seek("8000000000000fo2"); ASSERT_TRUE(!iter->Valid()); delete iter; } static std::string Key(int i) { char buf[100]; snprintf(buf, sizeof(buf), "key_______%06d", i); return std::string(buf); } static std::string RandomString(Random* rnd, int len) { std::string r; test::RandomString(rnd, len, &r); return r; } TEST_P(PlainTableDBTest, CompactionTrigger) { Options options = CurrentOptions(); options.write_buffer_size = 120 << 10; // 100KB options.num_levels = 3; options.level0_file_num_compaction_trigger = 3; Reopen(&options); Random rnd(301); for (int num = 0; num < options.level0_file_num_compaction_trigger - 1; num++) { std::vector values; // Write 120KB (10 values, each 12K) for (int i = 0; i < 10; i++) { values.push_back(RandomString(&rnd, 12000)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Put(Key(999), "")); dbfull()->TEST_WaitForFlushMemTable(); ASSERT_EQ(NumTableFilesAtLevel(0), num + 1); } //generate one more file in level-0, and should trigger level-0 compaction std::vector values; for (int i = 0; i < 12; i++) { values.push_back(RandomString(&rnd, 10000)); ASSERT_OK(Put(Key(i), values[i])); } ASSERT_OK(Put(Key(999), "")); dbfull()->TEST_WaitForCompact(); ASSERT_EQ(NumTableFilesAtLevel(0), 0); ASSERT_EQ(NumTableFilesAtLevel(1), 1); } TEST_P(PlainTableDBTest, AdaptiveTable) { Options options = CurrentOptions(); options.create_if_missing = true; options.table_factory.reset(NewPlainTableFactory()); DestroyAndReopen(&options); ASSERT_OK(Put("1000000000000foo", "v1")); ASSERT_OK(Put("0000000000000bar", "v2")); ASSERT_OK(Put("1000000000000foo", "v3")); dbfull()->TEST_FlushMemTable(); options.create_if_missing = false; std::shared_ptr dummy_factory; std::shared_ptr block_based_factory( NewBlockBasedTableFactory()); options.table_factory.reset(NewAdaptiveTableFactory( block_based_factory, dummy_factory, dummy_factory)); Reopen(&options); ASSERT_EQ("v3", Get("1000000000000foo")); ASSERT_EQ("v2", Get("0000000000000bar")); ASSERT_OK(Put("2000000000000foo", "v4")); ASSERT_OK(Put("3000000000000bar", "v5")); dbfull()->TEST_FlushMemTable(); ASSERT_EQ("v4", Get("2000000000000foo")); ASSERT_EQ("v5", Get("3000000000000bar")); Reopen(&options); ASSERT_EQ("v3", Get("1000000000000foo")); ASSERT_EQ("v2", Get("0000000000000bar")); ASSERT_EQ("v4", Get("2000000000000foo")); ASSERT_EQ("v5", Get("3000000000000bar")); options.table_factory.reset(NewBlockBasedTableFactory()); Reopen(&options); ASSERT_NE("v3", Get("1000000000000foo")); options.table_factory.reset(NewPlainTableFactory()); Reopen(&options); ASSERT_NE("v5", Get("3000000000000bar")); } INSTANTIATE_TEST_CASE_P(PlainTableDBTest, PlainTableDBTest, ::testing::Bool()); } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as plain table is not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/prefix_test.cc000066400000000000000000000441771267530634000164230ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #ifndef GFLAGS #include int main() { fprintf(stderr, "Please install gflags to run this test... Skipping...\n"); return 0; } #else #include #include #include #include #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/filter_policy.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice_transform.h" #include "rocksdb/memtablerep.h" #include "rocksdb/table.h" #include "util/histogram.h" #include "util/stop_watch.h" #include "util/string_util.h" #include "util/testharness.h" using GFLAGS::ParseCommandLineFlags; DEFINE_bool(trigger_deadlock, false, "issue delete in range scan to trigger PrefixHashMap deadlock"); DEFINE_int32(bucket_count, 100000, "number of buckets"); DEFINE_uint64(num_locks, 10001, "number of locks"); DEFINE_bool(random_prefix, false, "randomize prefix"); DEFINE_uint64(total_prefixes, 100000, "total number of prefixes"); DEFINE_uint64(items_per_prefix, 1, "total number of values per prefix"); DEFINE_int64(write_buffer_size, 33554432, ""); DEFINE_int32(max_write_buffer_number, 2, ""); DEFINE_int32(min_write_buffer_number_to_merge, 1, ""); DEFINE_int32(skiplist_height, 4, ""); DEFINE_int32(memtable_prefix_bloom_bits, 10000000, ""); DEFINE_int32(memtable_prefix_bloom_probes, 10, ""); DEFINE_int32(memtable_prefix_bloom_huge_page_tlb_size, 2 * 1024 * 1024, ""); DEFINE_int32(value_size, 40, ""); // Path to the database on file system const std::string kDbName = rocksdb::test::TmpDir() + "/prefix_test"; namespace rocksdb { struct TestKey { uint64_t prefix; uint64_t sorted; TestKey(uint64_t _prefix, uint64_t _sorted) : prefix(_prefix), sorted(_sorted) {} }; // return a slice backed by test_key inline Slice TestKeyToSlice(const TestKey& test_key) { return Slice((const char*)&test_key, sizeof(test_key)); } inline const TestKey* SliceToTestKey(const Slice& slice) { return (const TestKey*)slice.data(); } class TestKeyComparator : public Comparator { public: // Compare needs to be aware of the possibility of a and/or b is // prefix only virtual int Compare(const Slice& a, const Slice& b) const override { const TestKey* key_a = SliceToTestKey(a); const TestKey* key_b = SliceToTestKey(b); if (key_a->prefix != key_b->prefix) { if (key_a->prefix < key_b->prefix) return -1; if (key_a->prefix > key_b->prefix) return 1; } else { EXPECT_TRUE(key_a->prefix == key_b->prefix); // note, both a and b could be prefix only if (a.size() != b.size()) { // one of them is prefix EXPECT_TRUE( (a.size() == sizeof(uint64_t) && b.size() == sizeof(TestKey)) || (b.size() == sizeof(uint64_t) && a.size() == sizeof(TestKey))); if (a.size() < b.size()) return -1; if (a.size() > b.size()) return 1; } else { // both a and b are prefix if (a.size() == sizeof(uint64_t)) { return 0; } // both a and b are whole key EXPECT_TRUE(a.size() == sizeof(TestKey) && b.size() == sizeof(TestKey)); if (key_a->sorted < key_b->sorted) return -1; if (key_a->sorted > key_b->sorted) return 1; if (key_a->sorted == key_b->sorted) return 0; } } return 0; } virtual const char* Name() const override { return "TestKeyComparator"; } virtual void FindShortestSeparator(std::string* start, const Slice& limit) const override {} virtual void FindShortSuccessor(std::string* key) const override {} }; namespace { void PutKey(DB* db, WriteOptions write_options, uint64_t prefix, uint64_t suffix, const Slice& value) { TestKey test_key(prefix, suffix); Slice key = TestKeyToSlice(test_key); ASSERT_OK(db->Put(write_options, key, value)); } void SeekIterator(Iterator* iter, uint64_t prefix, uint64_t suffix) { TestKey test_key(prefix, suffix); Slice key = TestKeyToSlice(test_key); iter->Seek(key); } const std::string kNotFoundResult = "NOT_FOUND"; std::string Get(DB* db, const ReadOptions& read_options, uint64_t prefix, uint64_t suffix) { TestKey test_key(prefix, suffix); Slice key = TestKeyToSlice(test_key); std::string result; Status s = db->Get(read_options, key, &result); if (s.IsNotFound()) { result = kNotFoundResult; } else if (!s.ok()) { result = s.ToString(); } return result; } } // namespace class PrefixTest : public testing::Test { public: std::shared_ptr OpenDb() { DB* db; options.create_if_missing = true; options.write_buffer_size = FLAGS_write_buffer_size; options.max_write_buffer_number = FLAGS_max_write_buffer_number; options.min_write_buffer_number_to_merge = FLAGS_min_write_buffer_number_to_merge; options.memtable_prefix_bloom_bits = FLAGS_memtable_prefix_bloom_bits; options.memtable_prefix_bloom_probes = FLAGS_memtable_prefix_bloom_probes; options.memtable_prefix_bloom_huge_page_tlb_size = FLAGS_memtable_prefix_bloom_huge_page_tlb_size; options.prefix_extractor.reset(NewFixedPrefixTransform(8)); BlockBasedTableOptions bbto; bbto.filter_policy.reset(NewBloomFilterPolicy(10, false)); bbto.whole_key_filtering = false; options.table_factory.reset(NewBlockBasedTableFactory(bbto)); Status s = DB::Open(options, kDbName, &db); EXPECT_OK(s); return std::shared_ptr(db); } void FirstOption() { option_config_ = kBegin; } bool NextOptions(int bucket_count) { // skip some options option_config_++; if (option_config_ < kEnd) { options.prefix_extractor.reset(NewFixedPrefixTransform(8)); switch(option_config_) { case kHashSkipList: options.memtable_factory.reset( NewHashSkipListRepFactory(bucket_count, FLAGS_skiplist_height)); return true; case kHashLinkList: options.memtable_factory.reset( NewHashLinkListRepFactory(bucket_count)); return true; case kHashLinkListHugePageTlb: options.memtable_factory.reset( NewHashLinkListRepFactory(bucket_count, 2 * 1024 * 1024)); return true; case kHashLinkListTriggerSkipList: options.memtable_factory.reset( NewHashLinkListRepFactory(bucket_count, 0, 3)); return true; default: return false; } } return false; } PrefixTest() : option_config_(kBegin) { options.comparator = new TestKeyComparator(); } ~PrefixTest() { delete options.comparator; } protected: enum OptionConfig { kBegin, kHashSkipList, kHashLinkList, kHashLinkListHugePageTlb, kHashLinkListTriggerSkipList, kEnd }; int option_config_; Options options; }; TEST_F(PrefixTest, TestResult) { for (int num_buckets = 1; num_buckets <= 2; num_buckets++) { FirstOption(); while (NextOptions(num_buckets)) { std::cout << "*** Mem table: " << options.memtable_factory->Name() << " number of buckets: " << num_buckets << std::endl; DestroyDB(kDbName, Options()); auto db = OpenDb(); WriteOptions write_options; ReadOptions read_options; // 1. Insert one row. Slice v16("v16"); PutKey(db.get(), write_options, 1, 6, v16); std::unique_ptr iter(db->NewIterator(read_options)); SeekIterator(iter.get(), 1, 6); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); SeekIterator(iter.get(), 1, 5); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); SeekIterator(iter.get(), 1, 5); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); iter->Next(); ASSERT_TRUE(!iter->Valid()); SeekIterator(iter.get(), 2, 0); ASSERT_TRUE(!iter->Valid()); ASSERT_EQ(v16.ToString(), Get(db.get(), read_options, 1, 6)); ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 1, 5)); ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 1, 7)); ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 0, 6)); ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 2, 6)); // 2. Insert an entry for the same prefix as the last entry in the bucket. Slice v17("v17"); PutKey(db.get(), write_options, 1, 7, v17); iter.reset(db->NewIterator(read_options)); SeekIterator(iter.get(), 1, 7); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); SeekIterator(iter.get(), 1, 6); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); iter->Next(); ASSERT_TRUE(!iter->Valid()); SeekIterator(iter.get(), 2, 0); ASSERT_TRUE(!iter->Valid()); // 3. Insert an entry for the same prefix as the head of the bucket. Slice v15("v15"); PutKey(db.get(), write_options, 1, 5, v15); iter.reset(db->NewIterator(read_options)); SeekIterator(iter.get(), 1, 7); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); SeekIterator(iter.get(), 1, 5); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v15 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); SeekIterator(iter.get(), 1, 5); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v15 == iter->value()); ASSERT_EQ(v15.ToString(), Get(db.get(), read_options, 1, 5)); ASSERT_EQ(v16.ToString(), Get(db.get(), read_options, 1, 6)); ASSERT_EQ(v17.ToString(), Get(db.get(), read_options, 1, 7)); // 4. Insert an entry with a larger prefix Slice v22("v22"); PutKey(db.get(), write_options, 2, 2, v22); iter.reset(db->NewIterator(read_options)); SeekIterator(iter.get(), 2, 2); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v22 == iter->value()); SeekIterator(iter.get(), 2, 0); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v22 == iter->value()); SeekIterator(iter.get(), 1, 5); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v15 == iter->value()); SeekIterator(iter.get(), 1, 7); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); // 5. Insert an entry with a smaller prefix Slice v02("v02"); PutKey(db.get(), write_options, 0, 2, v02); iter.reset(db->NewIterator(read_options)); SeekIterator(iter.get(), 0, 2); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v02 == iter->value()); SeekIterator(iter.get(), 0, 0); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v02 == iter->value()); SeekIterator(iter.get(), 2, 0); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v22 == iter->value()); SeekIterator(iter.get(), 1, 5); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v15 == iter->value()); SeekIterator(iter.get(), 1, 7); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); // 6. Insert to the beginning and the end of the first prefix Slice v13("v13"); Slice v18("v18"); PutKey(db.get(), write_options, 1, 3, v13); PutKey(db.get(), write_options, 1, 8, v18); iter.reset(db->NewIterator(read_options)); SeekIterator(iter.get(), 1, 7); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); SeekIterator(iter.get(), 1, 3); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v13 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v15 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v18 == iter->value()); SeekIterator(iter.get(), 0, 0); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v02 == iter->value()); SeekIterator(iter.get(), 2, 0); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v22 == iter->value()); ASSERT_EQ(v22.ToString(), Get(db.get(), read_options, 2, 2)); ASSERT_EQ(v02.ToString(), Get(db.get(), read_options, 0, 2)); ASSERT_EQ(v13.ToString(), Get(db.get(), read_options, 1, 3)); ASSERT_EQ(v15.ToString(), Get(db.get(), read_options, 1, 5)); ASSERT_EQ(v16.ToString(), Get(db.get(), read_options, 1, 6)); ASSERT_EQ(v17.ToString(), Get(db.get(), read_options, 1, 7)); ASSERT_EQ(v18.ToString(), Get(db.get(), read_options, 1, 8)); } } } // Show results in prefix TEST_F(PrefixTest, PrefixValid) { for (int num_buckets = 1; num_buckets <= 2; num_buckets++) { FirstOption(); while (NextOptions(num_buckets)) { std::cout << "*** Mem table: " << options.memtable_factory->Name() << " number of buckets: " << num_buckets << std::endl; DestroyDB(kDbName, Options()); auto db = OpenDb(); WriteOptions write_options; ReadOptions read_options; // Insert keys with common prefix and one key with different Slice v16("v16"); Slice v17("v17"); Slice v18("v18"); Slice v19("v19"); PutKey(db.get(), write_options, 12345, 6, v16); PutKey(db.get(), write_options, 12345, 7, v17); PutKey(db.get(), write_options, 12345, 8, v18); PutKey(db.get(), write_options, 12345, 9, v19); PutKey(db.get(), write_options, 12346, 8, v16); db->Flush(FlushOptions()); db->Delete(write_options, TestKeyToSlice(TestKey(12346, 8))); db->Flush(FlushOptions()); read_options.prefix_same_as_start = true; std::unique_ptr iter(db->NewIterator(read_options)); SeekIterator(iter.get(), 12345, 6); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v16 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v17 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v18 == iter->value()); iter->Next(); ASSERT_TRUE(iter->Valid()); ASSERT_TRUE(v19 == iter->value()); iter->Next(); ASSERT_FALSE(iter->Valid()); ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 12346, 8)); } } } TEST_F(PrefixTest, DynamicPrefixIterator) { while (NextOptions(FLAGS_bucket_count)) { std::cout << "*** Mem table: " << options.memtable_factory->Name() << std::endl; DestroyDB(kDbName, Options()); auto db = OpenDb(); WriteOptions write_options; ReadOptions read_options; std::vector prefixes; for (uint64_t i = 0; i < FLAGS_total_prefixes; ++i) { prefixes.push_back(i); } if (FLAGS_random_prefix) { std::random_shuffle(prefixes.begin(), prefixes.end()); } HistogramImpl hist_put_time; HistogramImpl hist_put_comparison; // insert x random prefix, each with y continuous element. for (auto prefix : prefixes) { for (uint64_t sorted = 0; sorted < FLAGS_items_per_prefix; sorted++) { TestKey test_key(prefix, sorted); Slice key = TestKeyToSlice(test_key); std::string value(FLAGS_value_size, 0); perf_context.Reset(); StopWatchNano timer(Env::Default(), true); ASSERT_OK(db->Put(write_options, key, value)); hist_put_time.Add(timer.ElapsedNanos()); hist_put_comparison.Add(perf_context.user_key_comparison_count); } } std::cout << "Put key comparison: \n" << hist_put_comparison.ToString() << "Put time: \n" << hist_put_time.ToString(); // test seek existing keys HistogramImpl hist_seek_time; HistogramImpl hist_seek_comparison; std::unique_ptr iter(db->NewIterator(read_options)); for (auto prefix : prefixes) { TestKey test_key(prefix, FLAGS_items_per_prefix / 2); Slice key = TestKeyToSlice(test_key); std::string value = "v" + ToString(0); perf_context.Reset(); StopWatchNano timer(Env::Default(), true); auto key_prefix = options.prefix_extractor->Transform(key); uint64_t total_keys = 0; for (iter->Seek(key); iter->Valid() && iter->key().starts_with(key_prefix); iter->Next()) { if (FLAGS_trigger_deadlock) { std::cout << "Behold the deadlock!\n"; db->Delete(write_options, iter->key()); } total_keys++; } hist_seek_time.Add(timer.ElapsedNanos()); hist_seek_comparison.Add(perf_context.user_key_comparison_count); ASSERT_EQ(total_keys, FLAGS_items_per_prefix - FLAGS_items_per_prefix/2); } std::cout << "Seek key comparison: \n" << hist_seek_comparison.ToString() << "Seek time: \n" << hist_seek_time.ToString(); // test non-existing keys HistogramImpl hist_no_seek_time; HistogramImpl hist_no_seek_comparison; for (auto prefix = FLAGS_total_prefixes; prefix < FLAGS_total_prefixes + 10000; prefix++) { TestKey test_key(prefix, 0); Slice key = TestKeyToSlice(test_key); perf_context.Reset(); StopWatchNano timer(Env::Default(), true); iter->Seek(key); hist_no_seek_time.Add(timer.ElapsedNanos()); hist_no_seek_comparison.Add(perf_context.user_key_comparison_count); ASSERT_TRUE(!iter->Valid()); } std::cout << "non-existing Seek key comparison: \n" << hist_no_seek_comparison.ToString() << "non-existing Seek time: \n" << hist_no_seek_time.ToString(); } } } int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); ParseCommandLineFlags(&argc, &argv, true); std::cout << kDbName << "\n"; return RUN_ALL_TESTS(); } #endif // GFLAGS #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as HashSkipList and HashLinkList are not supported in " "ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/repair.cc000066400000000000000000000365441267530634000153500ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Repairer does best effort recovery to recover as much data as possible after // a disaster without compromising consistency. It does not guarantee bringing // the database to a time consistent state. // // Repair process is broken into 4 phases: // (a) Find files // (b) Convert logs to tables // (c) Extract metadata // (d) Write Descriptor // // (a) Find files // // The repairer goes through all the files in the directory, and classifies them // based on their file name. Any file that cannot be identified by name will be // ignored. // // (b) Convert logs to table // // Every log file that is active is replayed. All sections of the file where the // checksum does not match is skipped over. We intentionally give preference to // data consistency. // // (c) Extract metadata // // We scan every table to compute // (1) smallest/largest for the table // (2) largest sequence number in the table // // If we are unable to scan the file, then we ignore the table. // // (d) Write Descriptor // // We generate descriptor contents: // - log number is set to zero // - next-file-number is set to 1 + largest file number we found // - last-sequence-number is set to largest sequence# found across // all tables (see 2c) // - compaction pointers are cleared // - every table file is added at level 0 // // Possible optimization 1: // (a) Compute total size and use to pick appropriate max-level M // (b) Sort tables by largest sequence# in the table // (c) For each table: if it overlaps earlier table, place in level-0, // else place in level-M. // (d) We can provide options for time consistent recovery and unsafe recovery // (ignore checksum failure when applicable) // Possible optimization 2: // Store per-table metadata (smallest, largest, largest-seq#, ...) // in the table's meta section to speed up ScanTable. #ifndef ROCKSDB_LITE #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include "db/builder.h" #include "db/db_impl.h" #include "db/dbformat.h" #include "db/filename.h" #include "db/log_reader.h" #include "db/log_writer.h" #include "db/memtable.h" #include "db/table_cache.h" #include "db/version_edit.h" #include "db/writebuffer.h" #include "db/write_batch_internal.h" #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/immutable_options.h" #include "table/scoped_arena_iterator.h" #include "util/file_reader_writer.h" namespace rocksdb { namespace { class Repairer { public: Repairer(const std::string& dbname, const Options& options) : dbname_(dbname), env_(options.env), icmp_(options.comparator), options_(SanitizeOptions(dbname, &icmp_, options)), ioptions_(options_), raw_table_cache_( // TableCache can be small since we expect each table to be opened // once. NewLRUCache(10, options_.table_cache_numshardbits)), next_file_number_(1) { GetIntTblPropCollectorFactory(options, &int_tbl_prop_collector_factories_); table_cache_ = new TableCache(ioptions_, env_options_, raw_table_cache_.get()); edit_ = new VersionEdit(); } ~Repairer() { delete table_cache_; raw_table_cache_.reset(); delete edit_; } Status Run() { Status status = FindFiles(); if (status.ok()) { ConvertLogFilesToTables(); ExtractMetaData(); status = WriteDescriptor(); } if (status.ok()) { uint64_t bytes = 0; for (size_t i = 0; i < tables_.size(); i++) { bytes += tables_[i].meta.fd.GetFileSize(); } Log(InfoLogLevel::WARN_LEVEL, options_.info_log, "**** Repaired rocksdb %s; " "recovered %" ROCKSDB_PRIszt " files; %" PRIu64 "bytes. " "Some data may have been lost. " "****", dbname_.c_str(), tables_.size(), bytes); } return status; } private: struct TableInfo { FileMetaData meta; SequenceNumber min_sequence; SequenceNumber max_sequence; }; std::string const dbname_; Env* const env_; const InternalKeyComparator icmp_; std::vector> int_tbl_prop_collector_factories_; const Options options_; const ImmutableCFOptions ioptions_; std::shared_ptr raw_table_cache_; TableCache* table_cache_; VersionEdit* edit_; std::vector manifests_; std::vector table_fds_; std::vector logs_; std::vector tables_; uint64_t next_file_number_; const EnvOptions env_options_; Status FindFiles() { std::vector filenames; bool found_file = false; for (size_t path_id = 0; path_id < options_.db_paths.size(); path_id++) { Status status = env_->GetChildren(options_.db_paths[path_id].path, &filenames); if (!status.ok()) { return status; } if (!filenames.empty()) { found_file = true; } uint64_t number; FileType type; for (size_t i = 0; i < filenames.size(); i++) { if (ParseFileName(filenames[i], &number, &type)) { if (type == kDescriptorFile) { assert(path_id == 0); manifests_.push_back(filenames[i]); } else { if (number + 1 > next_file_number_) { next_file_number_ = number + 1; } if (type == kLogFile) { assert(path_id == 0); logs_.push_back(number); } else if (type == kTableFile) { table_fds_.emplace_back(number, static_cast(path_id), 0); } else { // Ignore other files } } } } } if (!found_file) { return Status::Corruption(dbname_, "repair found no files"); } return Status::OK(); } void ConvertLogFilesToTables() { for (size_t i = 0; i < logs_.size(); i++) { std::string logname = LogFileName(dbname_, logs_[i]); Status status = ConvertLogToTable(logs_[i]); if (!status.ok()) { Log(InfoLogLevel::WARN_LEVEL, options_.info_log, "Log #%" PRIu64 ": ignoring conversion error: %s", logs_[i], status.ToString().c_str()); } ArchiveFile(logname); } } Status ConvertLogToTable(uint64_t log) { struct LogReporter : public log::Reader::Reporter { Env* env; std::shared_ptr info_log; uint64_t lognum; virtual void Corruption(size_t bytes, const Status& s) override { // We print error messages for corruption, but continue repairing. Log(InfoLogLevel::ERROR_LEVEL, info_log, "Log #%" PRIu64 ": dropping %d bytes; %s", lognum, static_cast(bytes), s.ToString().c_str()); } }; // Open the log file std::string logname = LogFileName(dbname_, log); unique_ptr lfile; Status status = env_->NewSequentialFile(logname, &lfile, env_options_); if (!status.ok()) { return status; } unique_ptr lfile_reader( new SequentialFileReader(std::move(lfile))); // Create the log reader. LogReporter reporter; reporter.env = env_; reporter.info_log = options_.info_log; reporter.lognum = log; // We intentially make log::Reader do checksumming so that // corruptions cause entire commits to be skipped instead of // propagating bad information (like overly large sequence // numbers). log::Reader reader(options_.info_log, std::move(lfile_reader), &reporter, true /*enable checksum*/, 0 /*initial_offset*/, log); // Read all the records and add to a memtable std::string scratch; Slice record; WriteBatch batch; WriteBuffer wb(options_.db_write_buffer_size); MemTable* mem = new MemTable(icmp_, ioptions_, MutableCFOptions(options_, ioptions_), &wb, kMaxSequenceNumber); auto cf_mems_default = new ColumnFamilyMemTablesDefault(mem); mem->Ref(); int counter = 0; while (reader.ReadRecord(&record, &scratch)) { if (record.size() < 12) { reporter.Corruption( record.size(), Status::Corruption("log record too small")); continue; } WriteBatchInternal::SetContents(&batch, record); status = WriteBatchInternal::InsertInto(&batch, cf_mems_default, nullptr); if (status.ok()) { counter += WriteBatchInternal::Count(&batch); } else { Log(InfoLogLevel::WARN_LEVEL, options_.info_log, "Log #%" PRIu64 ": ignoring %s", log, status.ToString().c_str()); status = Status::OK(); // Keep going with rest of file } } // Do not record a version edit for this conversion to a Table // since ExtractMetaData() will also generate edits. FileMetaData meta; meta.fd = FileDescriptor(next_file_number_++, 0, 0); { ReadOptions ro; ro.total_order_seek = true; Arena arena; ScopedArenaIterator iter(mem->NewIterator(ro, &arena)); status = BuildTable( dbname_, env_, ioptions_, env_options_, table_cache_, iter.get(), &meta, icmp_, &int_tbl_prop_collector_factories_, TablePropertiesCollectorFactory::Context::kUnknownColumnFamily, {}, kMaxSequenceNumber, kNoCompression, CompressionOptions(), false, nullptr); } delete mem->Unref(); delete cf_mems_default; mem = nullptr; if (status.ok()) { if (meta.fd.GetFileSize() > 0) { table_fds_.push_back(meta.fd); } } Log(InfoLogLevel::INFO_LEVEL, options_.info_log, "Log #%" PRIu64 ": %d ops saved to Table #%" PRIu64 " %s", log, counter, meta.fd.GetNumber(), status.ToString().c_str()); return status; } void ExtractMetaData() { for (size_t i = 0; i < table_fds_.size(); i++) { TableInfo t; t.meta.fd = table_fds_[i]; Status status = ScanTable(&t); if (!status.ok()) { std::string fname = TableFileName( options_.db_paths, t.meta.fd.GetNumber(), t.meta.fd.GetPathId()); char file_num_buf[kFormatFileNumberBufSize]; FormatFileNumber(t.meta.fd.GetNumber(), t.meta.fd.GetPathId(), file_num_buf, sizeof(file_num_buf)); Log(InfoLogLevel::WARN_LEVEL, options_.info_log, "Table #%s: ignoring %s", file_num_buf, status.ToString().c_str()); ArchiveFile(fname); } else { tables_.push_back(t); } } } Status ScanTable(TableInfo* t) { std::string fname = TableFileName(options_.db_paths, t->meta.fd.GetNumber(), t->meta.fd.GetPathId()); int counter = 0; uint64_t file_size; Status status = env_->GetFileSize(fname, &file_size); t->meta.fd = FileDescriptor(t->meta.fd.GetNumber(), t->meta.fd.GetPathId(), file_size); if (status.ok()) { InternalIterator* iter = table_cache_->NewIterator( ReadOptions(), env_options_, icmp_, t->meta.fd); bool empty = true; ParsedInternalKey parsed; t->min_sequence = 0; t->max_sequence = 0; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { Slice key = iter->key(); if (!ParseInternalKey(key, &parsed)) { Log(InfoLogLevel::ERROR_LEVEL, options_.info_log, "Table #%" PRIu64 ": unparsable key %s", t->meta.fd.GetNumber(), EscapeString(key).c_str()); continue; } counter++; if (empty) { empty = false; t->meta.smallest.DecodeFrom(key); } t->meta.largest.DecodeFrom(key); if (parsed.sequence < t->min_sequence) { t->min_sequence = parsed.sequence; } if (parsed.sequence > t->max_sequence) { t->max_sequence = parsed.sequence; } } if (!iter->status().ok()) { status = iter->status(); } delete iter; } Log(InfoLogLevel::INFO_LEVEL, options_.info_log, "Table #%" PRIu64 ": %d entries %s", t->meta.fd.GetNumber(), counter, status.ToString().c_str()); return status; } Status WriteDescriptor() { std::string tmp = TempFileName(dbname_, 1); unique_ptr file; EnvOptions env_options = env_->OptimizeForManifestWrite(env_options_); Status status = env_->NewWritableFile(tmp, &file, env_options); if (!status.ok()) { return status; } SequenceNumber max_sequence = 0; for (size_t i = 0; i < tables_.size(); i++) { if (max_sequence < tables_[i].max_sequence) { max_sequence = tables_[i].max_sequence; } } edit_->SetComparatorName(icmp_.user_comparator()->Name()); edit_->SetLogNumber(0); edit_->SetNextFile(next_file_number_); edit_->SetLastSequence(max_sequence); for (size_t i = 0; i < tables_.size(); i++) { // TODO(opt): separate out into multiple levels const TableInfo& t = tables_[i]; edit_->AddFile(0, t.meta.fd.GetNumber(), t.meta.fd.GetPathId(), t.meta.fd.GetFileSize(), t.meta.smallest, t.meta.largest, t.min_sequence, t.max_sequence, t.meta.marked_for_compaction); } //fprintf(stderr, "NewDescriptor:\n%s\n", edit_.DebugString().c_str()); { unique_ptr file_writer( new WritableFileWriter(std::move(file), env_options)); log::Writer log(std::move(file_writer), 0, false); std::string record; edit_->EncodeTo(&record); status = log.AddRecord(record); } if (!status.ok()) { env_->DeleteFile(tmp); } else { // Discard older manifests for (size_t i = 0; i < manifests_.size(); i++) { ArchiveFile(dbname_ + "/" + manifests_[i]); } // Install new manifest status = env_->RenameFile(tmp, DescriptorFileName(dbname_, 1)); if (status.ok()) { status = SetCurrentFile(env_, dbname_, 1, nullptr); } else { env_->DeleteFile(tmp); } } return status; } void ArchiveFile(const std::string& fname) { // Move into another directory. E.g., for // dir/foo // rename to // dir/lost/foo const char* slash = strrchr(fname.c_str(), '/'); std::string new_dir; if (slash != nullptr) { new_dir.assign(fname.data(), slash - fname.data()); } new_dir.append("/lost"); env_->CreateDir(new_dir); // Ignore error std::string new_file = new_dir; new_file.append("/"); new_file.append((slash == nullptr) ? fname.c_str() : slash + 1); Status s = env_->RenameFile(fname, new_file); Log(InfoLogLevel::INFO_LEVEL, options_.info_log, "Archiving %s: %s\n", fname.c_str(), s.ToString().c_str()); } }; } // namespace Status RepairDB(const std::string& dbname, const Options& options) { Repairer repairer(dbname, options); return repairer.Run(); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/skiplist.h000066400000000000000000000361661267530634000155720ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Thread safety // ------------- // // Writes require external synchronization, most likely a mutex. // Reads require a guarantee that the SkipList will not be destroyed // while the read is in progress. Apart from that, reads progress // without any internal locking or synchronization. // // Invariants: // // (1) Allocated nodes are never deleted until the SkipList is // destroyed. This is trivially guaranteed by the code since we // never delete any skip list nodes. // // (2) The contents of a Node except for the next/prev pointers are // immutable after the Node has been linked into the SkipList. // Only Insert() modifies the list, and it is careful to initialize // a node and use release-stores to publish the nodes in one or // more lists. // // ... prev vs. next pointer ordering ... // #pragma once #include #include #include #include "port/port.h" #include "util/allocator.h" #include "util/random.h" namespace rocksdb { template class SkipList { private: struct Node; public: // Create a new SkipList object that will use "cmp" for comparing keys, // and will allocate memory using "*allocator". Objects allocated in the // allocator must remain allocated for the lifetime of the skiplist object. explicit SkipList(Comparator cmp, Allocator* allocator, int32_t max_height = 12, int32_t branching_factor = 4); // Insert key into the list. // REQUIRES: nothing that compares equal to key is currently in the list. void Insert(const Key& key); // Returns true iff an entry that compares equal to key is in the list. bool Contains(const Key& key) const; // Return estimated number of entries smaller than `key`. uint64_t EstimateCount(const Key& key) const; // Iteration over the contents of a skip list class Iterator { public: // Initialize an iterator over the specified list. // The returned iterator is not valid. explicit Iterator(const SkipList* list); // Change the underlying skiplist used for this iterator // This enables us not changing the iterator without deallocating // an old one and then allocating a new one void SetList(const SkipList* list); // Returns true iff the iterator is positioned at a valid node. bool Valid() const; // Returns the key at the current position. // REQUIRES: Valid() const Key& key() const; // Advances to the next position. // REQUIRES: Valid() void Next(); // Advances to the previous position. // REQUIRES: Valid() void Prev(); // Advance to the first entry with a key >= target void Seek(const Key& target); // Position at the first entry in list. // Final state of iterator is Valid() iff list is not empty. void SeekToFirst(); // Position at the last entry in list. // Final state of iterator is Valid() iff list is not empty. void SeekToLast(); private: const SkipList* list_; Node* node_; // Intentionally copyable }; private: const uint16_t kMaxHeight_; const uint16_t kBranching_; const uint32_t kScaledInverseBranching_; // Immutable after construction Comparator const compare_; Allocator* const allocator_; // Allocator used for allocations of nodes Node* const head_; // Modified only by Insert(). Read racily by readers, but stale // values are ok. std::atomic max_height_; // Height of the entire list // Used for optimizing sequential insert patterns. Tricky. prev_[i] for // i up to max_height_ is the predecessor of prev_[0] and prev_height_ // is the height of prev_[0]. prev_[0] can only be equal to head before // insertion, in which case max_height_ and prev_height_ are 1. Node** prev_; int32_t prev_height_; inline int GetMaxHeight() const { return max_height_.load(std::memory_order_relaxed); } Node* NewNode(const Key& key, int height); int RandomHeight(); bool Equal(const Key& a, const Key& b) const { return (compare_(a, b) == 0); } // Return true if key is greater than the data stored in "n" bool KeyIsAfterNode(const Key& key, Node* n) const; // Returns the earliest node with a key >= key. // Return nullptr if there is no such node. Node* FindGreaterOrEqual(const Key& key) const; // Return the latest node with a key < key. // Return head_ if there is no such node. // Fills prev[level] with pointer to previous node at "level" for every // level in [0..max_height_-1], if prev is non-null. Node* FindLessThan(const Key& key, Node** prev = nullptr) const; // Return the last node in the list. // Return head_ if list is empty. Node* FindLast() const; // No copying allowed SkipList(const SkipList&); void operator=(const SkipList&); }; // Implementation details follow template struct SkipList::Node { explicit Node(const Key& k) : key(k) { } Key const key; // Accessors/mutators for links. Wrapped in methods so we can // add the appropriate barriers as necessary. Node* Next(int n) { assert(n >= 0); // Use an 'acquire load' so that we observe a fully initialized // version of the returned Node. return (next_[n].load(std::memory_order_acquire)); } void SetNext(int n, Node* x) { assert(n >= 0); // Use a 'release store' so that anybody who reads through this // pointer observes a fully initialized version of the inserted node. next_[n].store(x, std::memory_order_release); } // No-barrier variants that can be safely used in a few locations. Node* NoBarrier_Next(int n) { assert(n >= 0); return next_[n].load(std::memory_order_relaxed); } void NoBarrier_SetNext(int n, Node* x) { assert(n >= 0); next_[n].store(x, std::memory_order_relaxed); } private: // Array of length equal to the node height. next_[0] is lowest level link. std::atomic next_[1]; }; template typename SkipList::Node* SkipList::NewNode(const Key& key, int height) { char* mem = allocator_->AllocateAligned( sizeof(Node) + sizeof(std::atomic) * (height - 1)); return new (mem) Node(key); } template inline SkipList::Iterator::Iterator(const SkipList* list) { SetList(list); } template inline void SkipList::Iterator::SetList(const SkipList* list) { list_ = list; node_ = nullptr; } template inline bool SkipList::Iterator::Valid() const { return node_ != nullptr; } template inline const Key& SkipList::Iterator::key() const { assert(Valid()); return node_->key; } template inline void SkipList::Iterator::Next() { assert(Valid()); node_ = node_->Next(0); } template inline void SkipList::Iterator::Prev() { // Instead of using explicit "prev" links, we just search for the // last node that falls before key. assert(Valid()); node_ = list_->FindLessThan(node_->key); if (node_ == list_->head_) { node_ = nullptr; } } template inline void SkipList::Iterator::Seek(const Key& target) { node_ = list_->FindGreaterOrEqual(target); } template inline void SkipList::Iterator::SeekToFirst() { node_ = list_->head_->Next(0); } template inline void SkipList::Iterator::SeekToLast() { node_ = list_->FindLast(); if (node_ == list_->head_) { node_ = nullptr; } } template int SkipList::RandomHeight() { auto rnd = Random::GetTLSInstance(); // Increase height with probability 1 in kBranching int height = 1; while (height < kMaxHeight_ && rnd->Next() < kScaledInverseBranching_) { height++; } assert(height > 0); assert(height <= kMaxHeight_); return height; } template bool SkipList::KeyIsAfterNode(const Key& key, Node* n) const { // nullptr n is considered infinite return (n != nullptr) && (compare_(n->key, key) < 0); } template typename SkipList::Node* SkipList:: FindGreaterOrEqual(const Key& key) const { // Note: It looks like we could reduce duplication by implementing // this function as FindLessThan(key)->Next(0), but we wouldn't be able // to exit early on equality and the result wouldn't even be correct. // A concurrent insert might occur after FindLessThan(key) but before // we get a chance to call Next(0). Node* x = head_; int level = GetMaxHeight() - 1; Node* last_bigger = nullptr; while (true) { Node* next = x->Next(level); // Make sure the lists are sorted assert(x == head_ || next == nullptr || KeyIsAfterNode(next->key, x)); // Make sure we haven't overshot during our search assert(x == head_ || KeyIsAfterNode(key, x)); int cmp = (next == nullptr || next == last_bigger) ? 1 : compare_(next->key, key); if (cmp == 0 || (cmp > 0 && level == 0)) { return next; } else if (cmp < 0) { // Keep searching in this list x = next; } else { // Switch to next list, reuse compare_() result last_bigger = next; level--; } } } template typename SkipList::Node* SkipList::FindLessThan(const Key& key, Node** prev) const { Node* x = head_; int level = GetMaxHeight() - 1; // KeyIsAfter(key, last_not_after) is definitely false Node* last_not_after = nullptr; while (true) { Node* next = x->Next(level); assert(x == head_ || next == nullptr || KeyIsAfterNode(next->key, x)); assert(x == head_ || KeyIsAfterNode(key, x)); if (next != last_not_after && KeyIsAfterNode(key, next)) { // Keep searching in this list x = next; } else { if (prev != nullptr) { prev[level] = x; } if (level == 0) { return x; } else { // Switch to next list, reuse KeyIUsAfterNode() result last_not_after = next; level--; } } } } template typename SkipList::Node* SkipList::FindLast() const { Node* x = head_; int level = GetMaxHeight() - 1; while (true) { Node* next = x->Next(level); if (next == nullptr) { if (level == 0) { return x; } else { // Switch to next list level--; } } else { x = next; } } } template uint64_t SkipList::EstimateCount(const Key& key) const { uint64_t count = 0; Node* x = head_; int level = GetMaxHeight() - 1; while (true) { assert(x == head_ || compare_(x->key, key) < 0); Node* next = x->Next(level); if (next == nullptr || compare_(next->key, key) >= 0) { if (level == 0) { return count; } else { // Switch to next list count *= kBranching_; level--; } } else { x = next; count++; } } } template SkipList::SkipList(const Comparator cmp, Allocator* allocator, int32_t max_height, int32_t branching_factor) : kMaxHeight_(max_height), kBranching_(branching_factor), kScaledInverseBranching_((Random::kMaxNext + 1) / kBranching_), compare_(cmp), allocator_(allocator), head_(NewNode(0 /* any key will do */, max_height)), max_height_(1), prev_height_(1) { assert(max_height > 0 && kMaxHeight_ == static_cast(max_height)); assert(branching_factor > 0 && kBranching_ == static_cast(branching_factor)); assert(kScaledInverseBranching_ > 0); // Allocate the prev_ Node* array, directly from the passed-in allocator. // prev_ does not need to be freed, as its life cycle is tied up with // the allocator as a whole. prev_ = reinterpret_cast( allocator_->AllocateAligned(sizeof(Node*) * kMaxHeight_)); for (int i = 0; i < kMaxHeight_; i++) { head_->SetNext(i, nullptr); prev_[i] = head_; } } template void SkipList::Insert(const Key& key) { // fast path for sequential insertion if (!KeyIsAfterNode(key, prev_[0]->NoBarrier_Next(0)) && (prev_[0] == head_ || KeyIsAfterNode(key, prev_[0]))) { assert(prev_[0] != head_ || (prev_height_ == 1 && GetMaxHeight() == 1)); // Outside of this method prev_[1..max_height_] is the predecessor // of prev_[0], and prev_height_ refers to prev_[0]. Inside Insert // prev_[0..max_height - 1] is the predecessor of key. Switch from // the external state to the internal for (int i = 1; i < prev_height_; i++) { prev_[i] = prev_[0]; } } else { // TODO(opt): we could use a NoBarrier predecessor search as an // optimization for architectures where memory_order_acquire needs // a synchronization instruction. Doesn't matter on x86 FindLessThan(key, prev_); } // Our data structure does not allow duplicate insertion assert(prev_[0]->Next(0) == nullptr || !Equal(key, prev_[0]->Next(0)->key)); int height = RandomHeight(); if (height > GetMaxHeight()) { for (int i = GetMaxHeight(); i < height; i++) { prev_[i] = head_; } //fprintf(stderr, "Change height from %d to %d\n", max_height_, height); // It is ok to mutate max_height_ without any synchronization // with concurrent readers. A concurrent reader that observes // the new value of max_height_ will see either the old value of // new level pointers from head_ (nullptr), or a new value set in // the loop below. In the former case the reader will // immediately drop to the next level since nullptr sorts after all // keys. In the latter case the reader will use the new node. max_height_.store(height, std::memory_order_relaxed); } Node* x = NewNode(key, height); for (int i = 0; i < height; i++) { // NoBarrier_SetNext() suffices since we will add a barrier when // we publish a pointer to "x" in prev[i]. x->NoBarrier_SetNext(i, prev_[i]->NoBarrier_Next(i)); prev_[i]->SetNext(i, x); } prev_[0] = x; prev_height_ = height; } template bool SkipList::Contains(const Key& key) const { Node* x = FindGreaterOrEqual(key); if (x != nullptr && Equal(key, x->key)) { return true; } else { return false; } } } // namespace rocksdb rocksdb-4.5.1/db/skiplist_test.cc000066400000000000000000000237221267530634000167610ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/skiplist.h" #include #include "rocksdb/env.h" #include "util/arena.h" #include "util/hash.h" #include "util/random.h" #include "util/testharness.h" namespace rocksdb { typedef uint64_t Key; struct TestComparator { int operator()(const Key& a, const Key& b) const { if (a < b) { return -1; } else if (a > b) { return +1; } else { return 0; } } }; class SkipTest : public testing::Test {}; TEST_F(SkipTest, Empty) { Arena arena; TestComparator cmp; SkipList list(cmp, &arena); ASSERT_TRUE(!list.Contains(10)); SkipList::Iterator iter(&list); ASSERT_TRUE(!iter.Valid()); iter.SeekToFirst(); ASSERT_TRUE(!iter.Valid()); iter.Seek(100); ASSERT_TRUE(!iter.Valid()); iter.SeekToLast(); ASSERT_TRUE(!iter.Valid()); } TEST_F(SkipTest, InsertAndLookup) { const int N = 2000; const int R = 5000; Random rnd(1000); std::set keys; Arena arena; TestComparator cmp; SkipList list(cmp, &arena); for (int i = 0; i < N; i++) { Key key = rnd.Next() % R; if (keys.insert(key).second) { list.Insert(key); } } for (int i = 0; i < R; i++) { if (list.Contains(i)) { ASSERT_EQ(keys.count(i), 1U); } else { ASSERT_EQ(keys.count(i), 0U); } } // Simple iterator tests { SkipList::Iterator iter(&list); ASSERT_TRUE(!iter.Valid()); iter.Seek(0); ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*(keys.begin()), iter.key()); iter.SeekToFirst(); ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*(keys.begin()), iter.key()); iter.SeekToLast(); ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*(keys.rbegin()), iter.key()); } // Forward iteration test for (int i = 0; i < R; i++) { SkipList::Iterator iter(&list); iter.Seek(i); // Compare against model iterator std::set::iterator model_iter = keys.lower_bound(i); for (int j = 0; j < 3; j++) { if (model_iter == keys.end()) { ASSERT_TRUE(!iter.Valid()); break; } else { ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*model_iter, iter.key()); ++model_iter; iter.Next(); } } } // Backward iteration test { SkipList::Iterator iter(&list); iter.SeekToLast(); // Compare against model iterator for (std::set::reverse_iterator model_iter = keys.rbegin(); model_iter != keys.rend(); ++model_iter) { ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*model_iter, iter.key()); iter.Prev(); } ASSERT_TRUE(!iter.Valid()); } } // We want to make sure that with a single writer and multiple // concurrent readers (with no synchronization other than when a // reader's iterator is created), the reader always observes all the // data that was present in the skip list when the iterator was // constructor. Because insertions are happening concurrently, we may // also observe new values that were inserted since the iterator was // constructed, but we should never miss any values that were present // at iterator construction time. // // We generate multi-part keys: // // where: // key is in range [0..K-1] // gen is a generation number for key // hash is hash(key,gen) // // The insertion code picks a random key, sets gen to be 1 + the last // generation number inserted for that key, and sets hash to Hash(key,gen). // // At the beginning of a read, we snapshot the last inserted // generation number for each key. We then iterate, including random // calls to Next() and Seek(). For every key we encounter, we // check that it is either expected given the initial snapshot or has // been concurrently added since the iterator started. class ConcurrentTest { private: static const uint32_t K = 4; static uint64_t key(Key key) { return (key >> 40); } static uint64_t gen(Key key) { return (key >> 8) & 0xffffffffu; } static uint64_t hash(Key key) { return key & 0xff; } static uint64_t HashNumbers(uint64_t k, uint64_t g) { uint64_t data[2] = { k, g }; return Hash(reinterpret_cast(data), sizeof(data), 0); } static Key MakeKey(uint64_t k, uint64_t g) { assert(sizeof(Key) == sizeof(uint64_t)); assert(k <= K); // We sometimes pass K to seek to the end of the skiplist assert(g <= 0xffffffffu); return ((k << 40) | (g << 8) | (HashNumbers(k, g) & 0xff)); } static bool IsValidKey(Key k) { return hash(k) == (HashNumbers(key(k), gen(k)) & 0xff); } static Key RandomTarget(Random* rnd) { switch (rnd->Next() % 10) { case 0: // Seek to beginning return MakeKey(0, 0); case 1: // Seek to end return MakeKey(K, 0); default: // Seek to middle return MakeKey(rnd->Next() % K, 0); } } // Per-key generation struct State { std::atomic generation[K]; void Set(int k, int v) { generation[k].store(v, std::memory_order_release); } int Get(int k) { return generation[k].load(std::memory_order_acquire); } State() { for (unsigned int k = 0; k < K; k++) { Set(k, 0); } } }; // Current state of the test State current_; Arena arena_; // SkipList is not protected by mu_. We just use a single writer // thread to modify it. SkipList list_; public: ConcurrentTest() : list_(TestComparator(), &arena_) {} // REQUIRES: External synchronization void WriteStep(Random* rnd) { const uint32_t k = rnd->Next() % K; const int g = current_.Get(k) + 1; const Key new_key = MakeKey(k, g); list_.Insert(new_key); current_.Set(k, g); } void ReadStep(Random* rnd) { // Remember the initial committed state of the skiplist. State initial_state; for (unsigned int k = 0; k < K; k++) { initial_state.Set(k, current_.Get(k)); } Key pos = RandomTarget(rnd); SkipList::Iterator iter(&list_); iter.Seek(pos); while (true) { Key current; if (!iter.Valid()) { current = MakeKey(K, 0); } else { current = iter.key(); ASSERT_TRUE(IsValidKey(current)) << current; } ASSERT_LE(pos, current) << "should not go backwards"; // Verify that everything in [pos,current) was not present in // initial_state. while (pos < current) { ASSERT_LT(key(pos), K) << pos; // Note that generation 0 is never inserted, so it is ok if // <*,0,*> is missing. ASSERT_TRUE((gen(pos) == 0U) || (gen(pos) > static_cast(initial_state.Get( static_cast(key(pos)))))) << "key: " << key(pos) << "; gen: " << gen(pos) << "; initgen: " << initial_state.Get(static_cast(key(pos))); // Advance to next key in the valid key space if (key(pos) < key(current)) { pos = MakeKey(key(pos) + 1, 0); } else { pos = MakeKey(key(pos), gen(pos) + 1); } } if (!iter.Valid()) { break; } if (rnd->Next() % 2) { iter.Next(); pos = MakeKey(key(pos), gen(pos) + 1); } else { Key new_target = RandomTarget(rnd); if (new_target > pos) { pos = new_target; iter.Seek(new_target); } } } } }; const uint32_t ConcurrentTest::K; // Simple test that does single-threaded testing of the ConcurrentTest // scaffolding. TEST_F(SkipTest, ConcurrentWithoutThreads) { ConcurrentTest test; Random rnd(test::RandomSeed()); for (int i = 0; i < 10000; i++) { test.ReadStep(&rnd); test.WriteStep(&rnd); } } class TestState { public: ConcurrentTest t_; int seed_; std::atomic quit_flag_; enum ReaderState { STARTING, RUNNING, DONE }; explicit TestState(int s) : seed_(s), quit_flag_(false), state_(STARTING), state_cv_(&mu_) {} void Wait(ReaderState s) { mu_.Lock(); while (state_ != s) { state_cv_.Wait(); } mu_.Unlock(); } void Change(ReaderState s) { mu_.Lock(); state_ = s; state_cv_.Signal(); mu_.Unlock(); } private: port::Mutex mu_; ReaderState state_; port::CondVar state_cv_; }; static void ConcurrentReader(void* arg) { TestState* state = reinterpret_cast(arg); Random rnd(state->seed_); int64_t reads = 0; state->Change(TestState::RUNNING); while (!state->quit_flag_.load(std::memory_order_acquire)) { state->t_.ReadStep(&rnd); ++reads; } state->Change(TestState::DONE); } static void RunConcurrent(int run) { const int seed = test::RandomSeed() + (run * 100); Random rnd(seed); const int N = 1000; const int kSize = 1000; for (int i = 0; i < N; i++) { if ((i % 100) == 0) { fprintf(stderr, "Run %d of %d\n", i, N); } TestState state(seed + 1); Env::Default()->Schedule(ConcurrentReader, &state); state.Wait(TestState::RUNNING); for (int k = 0; k < kSize; k++) { state.t_.WriteStep(&rnd); } state.quit_flag_.store(true, std::memory_order_release); state.Wait(TestState::DONE); } } TEST_F(SkipTest, Concurrent1) { RunConcurrent(1); } TEST_F(SkipTest, Concurrent2) { RunConcurrent(2); } TEST_F(SkipTest, Concurrent3) { RunConcurrent(3); } TEST_F(SkipTest, Concurrent4) { RunConcurrent(4); } TEST_F(SkipTest, Concurrent5) { RunConcurrent(5); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/slice.cc000066400000000000000000000013171267530634000151530ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "rocksdb/slice.h" namespace rocksdb { Slice::Slice(const SliceParts& parts, std::string* buf) { size_t length = 0; for (int i = 0; i < parts.num_parts; ++i) { length += parts.parts[i].size(); } buf->reserve(length); for (int i = 0; i < parts.num_parts; ++i) { buf->append(parts.parts[i].data(), parts.parts[i].size()); } data_ = buf->data(); size_ = buf->size(); } } // namespace rocksdb rocksdb-4.5.1/db/snapshot_impl.cc000066400000000000000000000014351267530634000167350ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "rocksdb/snapshot.h" #include "rocksdb/db.h" namespace rocksdb { ManagedSnapshot::ManagedSnapshot(DB* db) : db_(db), snapshot_(db->GetSnapshot()) {} ManagedSnapshot::ManagedSnapshot(DB* db, const Snapshot* _snapshot) : db_(db), snapshot_(_snapshot) {} ManagedSnapshot::~ManagedSnapshot() { if (snapshot_) { db_->ReleaseSnapshot(snapshot_); } } const Snapshot* ManagedSnapshot::snapshot() { return snapshot_;} } // namespace rocksdb rocksdb-4.5.1/db/snapshot_impl.h000066400000000000000000000070621267530634000166010ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include "rocksdb/db.h" namespace rocksdb { class SnapshotList; // Snapshots are kept in a doubly-linked list in the DB. // Each SnapshotImpl corresponds to a particular sequence number. class SnapshotImpl : public Snapshot { public: SequenceNumber number_; // const after creation virtual SequenceNumber GetSequenceNumber() const override { return number_; } private: friend class SnapshotList; // SnapshotImpl is kept in a doubly-linked circular list SnapshotImpl* prev_; SnapshotImpl* next_; SnapshotList* list_; // just for sanity checks int64_t unix_time_; // Will this snapshot be used by a Transaction to do write-conflict checking? bool is_write_conflict_boundary_; }; class SnapshotList { public: SnapshotList() { list_.prev_ = &list_; list_.next_ = &list_; list_.number_ = 0xFFFFFFFFL; // placeholder marker, for debugging count_ = 0; } bool empty() const { return list_.next_ == &list_; } SnapshotImpl* oldest() const { assert(!empty()); return list_.next_; } SnapshotImpl* newest() const { assert(!empty()); return list_.prev_; } const SnapshotImpl* New(SnapshotImpl* s, SequenceNumber seq, uint64_t unix_time, bool is_write_conflict_boundary) { s->number_ = seq; s->unix_time_ = unix_time; s->is_write_conflict_boundary_ = is_write_conflict_boundary; s->list_ = this; s->next_ = &list_; s->prev_ = list_.prev_; s->prev_->next_ = s; s->next_->prev_ = s; count_++; return s; } // Do not responsible to free the object. void Delete(const SnapshotImpl* s) { assert(s->list_ == this); s->prev_->next_ = s->next_; s->next_->prev_ = s->prev_; count_--; } // retrieve all snapshot numbers. They are sorted in ascending order. std::vector GetAll( SequenceNumber* oldest_write_conflict_snapshot = nullptr) { std::vector ret; if (oldest_write_conflict_snapshot != nullptr) { *oldest_write_conflict_snapshot = kMaxSequenceNumber; } if (empty()) { return ret; } SnapshotImpl* s = &list_; while (s->next_ != &list_) { ret.push_back(s->next_->number_); if (oldest_write_conflict_snapshot != nullptr && *oldest_write_conflict_snapshot == kMaxSequenceNumber && s->next_->is_write_conflict_boundary_) { // If this is the first write-conflict boundary snapshot in the list, // it is the oldest *oldest_write_conflict_snapshot = s->next_->number_; } s = s->next_; } return ret; } // get the sequence number of the most recent snapshot SequenceNumber GetNewest() { if (empty()) { return 0; } return newest()->number_; } int64_t GetOldestSnapshotTime() const { if (empty()) { return 0; } else { return oldest()->unix_time_; } } uint64_t count() const { return count_; } private: // Dummy head of doubly-linked list of snapshots SnapshotImpl list_; uint64_t count_; }; } // namespace rocksdb rocksdb-4.5.1/db/table_cache.cc000066400000000000000000000303401267530634000162640ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/table_cache.h" #include "db/dbformat.h" #include "db/filename.h" #include "db/version_edit.h" #include "rocksdb/statistics.h" #include "table/internal_iterator.h" #include "table/iterator_wrapper.h" #include "table/table_builder.h" #include "table/table_reader.h" #include "table/get_context.h" #include "util/coding.h" #include "util/file_reader_writer.h" #include "util/perf_context_imp.h" #include "util/stop_watch.h" #include "util/sync_point.h" namespace rocksdb { namespace { template static void DeleteEntry(const Slice& key, void* value) { T* typed_value = reinterpret_cast(value); delete typed_value; } static void UnrefEntry(void* arg1, void* arg2) { Cache* cache = reinterpret_cast(arg1); Cache::Handle* h = reinterpret_cast(arg2); cache->Release(h); } static void DeleteTableReader(void* arg1, void* arg2) { TableReader* table_reader = reinterpret_cast(arg1); delete table_reader; } static Slice GetSliceForFileNumber(const uint64_t* file_number) { return Slice(reinterpret_cast(file_number), sizeof(*file_number)); } #ifndef ROCKSDB_LITE void AppendVarint64(IterKey* key, uint64_t v) { char buf[10]; auto ptr = EncodeVarint64(buf, v); key->TrimAppend(key->Size(), buf, ptr - buf); } #endif // ROCKSDB_LITE } // namespace TableCache::TableCache(const ImmutableCFOptions& ioptions, const EnvOptions& env_options, Cache* const cache) : ioptions_(ioptions), env_options_(env_options), cache_(cache) { if (ioptions_.row_cache) { // If the same cache is shared by multiple instances, we need to // disambiguate its entries. PutVarint64(&row_cache_id_, ioptions_.row_cache->NewId()); } } TableCache::~TableCache() { } TableReader* TableCache::GetTableReaderFromHandle(Cache::Handle* handle) { return reinterpret_cast(cache_->Value(handle)); } void TableCache::ReleaseHandle(Cache::Handle* handle) { cache_->Release(handle); } Status TableCache::GetTableReader( const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd, bool sequential_mode, bool record_read_stats, HistogramImpl* file_read_hist, unique_ptr* table_reader, bool skip_filters) { std::string fname = TableFileName(ioptions_.db_paths, fd.GetNumber(), fd.GetPathId()); unique_ptr file; Status s = ioptions_.env->NewRandomAccessFile(fname, &file, env_options); if (sequential_mode && ioptions_.compaction_readahead_size > 0) { file = NewReadaheadRandomAccessFile(std::move(file), ioptions_.compaction_readahead_size); } RecordTick(ioptions_.statistics, NO_FILE_OPENS); if (s.ok()) { if (!sequential_mode && ioptions_.advise_random_on_open) { file->Hint(RandomAccessFile::RANDOM); } StopWatch sw(ioptions_.env, ioptions_.statistics, TABLE_OPEN_IO_MICROS); std::unique_ptr file_reader( new RandomAccessFileReader(std::move(file), ioptions_.env, ioptions_.statistics, record_read_stats, file_read_hist)); s = ioptions_.table_factory->NewTableReader( TableReaderOptions(ioptions_, env_options, internal_comparator, skip_filters), std::move(file_reader), fd.GetFileSize(), table_reader); TEST_SYNC_POINT("TableCache::GetTableReader:0"); } return s; } Status TableCache::FindTable(const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd, Cache::Handle** handle, const bool no_io, bool record_read_stats, HistogramImpl* file_read_hist, bool skip_filters) { PERF_TIMER_GUARD(find_table_nanos); Status s; uint64_t number = fd.GetNumber(); Slice key = GetSliceForFileNumber(&number); *handle = cache_->Lookup(key); TEST_SYNC_POINT_CALLBACK("TableCache::FindTable:0", const_cast(&no_io)); if (*handle == nullptr) { if (no_io) { // Don't do IO and return a not-found status return Status::Incomplete("Table not found in table_cache, no_io is set"); } unique_ptr table_reader; s = GetTableReader(env_options, internal_comparator, fd, false /* sequential mode */, record_read_stats, file_read_hist, &table_reader, skip_filters); if (!s.ok()) { assert(table_reader == nullptr); RecordTick(ioptions_.statistics, NO_FILE_ERRORS); // We do not cache error results so that if the error is transient, // or somebody repairs the file, we recover automatically. } else { *handle = cache_->Insert(key, table_reader.release(), 1, &DeleteEntry); } } return s; } InternalIterator* TableCache::NewIterator( const ReadOptions& options, const EnvOptions& env_options, const InternalKeyComparator& icomparator, const FileDescriptor& fd, TableReader** table_reader_ptr, HistogramImpl* file_read_hist, bool for_compaction, Arena* arena, bool skip_filters) { PERF_TIMER_GUARD(new_table_iterator_nanos); if (table_reader_ptr != nullptr) { *table_reader_ptr = nullptr; } TableReader* table_reader = nullptr; Cache::Handle* handle = nullptr; bool create_new_table_reader = (for_compaction && ioptions_.new_table_reader_for_compaction_inputs); if (create_new_table_reader) { unique_ptr table_reader_unique_ptr; Status s = GetTableReader( env_options, icomparator, fd, /* sequential mode */ true, /* record stats */ false, nullptr, &table_reader_unique_ptr); if (!s.ok()) { return NewErrorInternalIterator(s, arena); } table_reader = table_reader_unique_ptr.release(); } else { table_reader = fd.table_reader; if (table_reader == nullptr) { Status s = FindTable(env_options, icomparator, fd, &handle, options.read_tier == kBlockCacheTier /* no_io */, !for_compaction /* record read_stats */, file_read_hist, skip_filters); if (!s.ok()) { return NewErrorInternalIterator(s, arena); } table_reader = GetTableReaderFromHandle(handle); } } InternalIterator* result = table_reader->NewIterator(options, arena, skip_filters); if (create_new_table_reader) { assert(handle == nullptr); result->RegisterCleanup(&DeleteTableReader, table_reader, nullptr); } else if (handle != nullptr) { result->RegisterCleanup(&UnrefEntry, cache_, handle); } if (for_compaction) { table_reader->SetupForCompaction(); } if (table_reader_ptr != nullptr) { *table_reader_ptr = table_reader; } return result; } Status TableCache::Get(const ReadOptions& options, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd, const Slice& k, GetContext* get_context, HistogramImpl* file_read_hist, bool skip_filters) { TableReader* t = fd.table_reader; Status s; Cache::Handle* handle = nullptr; std::string* row_cache_entry = nullptr; #ifndef ROCKSDB_LITE IterKey row_cache_key; std::string row_cache_entry_buffer; // Check row cache if enabled. Since row cache does not currently store // sequence numbers, we cannot use it if we need to fetch the sequence. if (ioptions_.row_cache && !get_context->NeedToReadSequence()) { uint64_t fd_number = fd.GetNumber(); auto user_key = ExtractUserKey(k); // We use the user key as cache key instead of the internal key, // otherwise the whole cache would be invalidated every time the // sequence key increases. However, to support caching snapshot // reads, we append the sequence number (incremented by 1 to // distinguish from 0) only in this case. uint64_t seq_no = options.snapshot == nullptr ? 0 : 1 + GetInternalKeySeqno(k); // Compute row cache key. row_cache_key.TrimAppend(row_cache_key.Size(), row_cache_id_.data(), row_cache_id_.size()); AppendVarint64(&row_cache_key, fd_number); AppendVarint64(&row_cache_key, seq_no); row_cache_key.TrimAppend(row_cache_key.Size(), user_key.data(), user_key.size()); if (auto row_handle = ioptions_.row_cache->Lookup(row_cache_key.GetKey())) { auto found_row_cache_entry = static_cast( ioptions_.row_cache->Value(row_handle)); replayGetContextLog(*found_row_cache_entry, user_key, get_context); ioptions_.row_cache->Release(row_handle); RecordTick(ioptions_.statistics, ROW_CACHE_HIT); return Status::OK(); } // Not found, setting up the replay log. RecordTick(ioptions_.statistics, ROW_CACHE_MISS); row_cache_entry = &row_cache_entry_buffer; } #endif // ROCKSDB_LITE if (!t) { s = FindTable(env_options_, internal_comparator, fd, &handle, options.read_tier == kBlockCacheTier /* no_io */, true /* record_read_stats */, file_read_hist, skip_filters); if (s.ok()) { t = GetTableReaderFromHandle(handle); } } if (s.ok()) { get_context->SetReplayLog(row_cache_entry); // nullptr if no cache. s = t->Get(options, k, get_context, skip_filters); get_context->SetReplayLog(nullptr); if (handle != nullptr) { ReleaseHandle(handle); } } else if (options.read_tier == kBlockCacheTier && s.IsIncomplete()) { // Couldn't find Table in cache but treat as kFound if no_io set get_context->MarkKeyMayExist(); return Status::OK(); } #ifndef ROCKSDB_LITE // Put the replay log in row cache only if something was found. if (s.ok() && row_cache_entry && !row_cache_entry->empty()) { size_t charge = row_cache_key.Size() + row_cache_entry->size() + sizeof(std::string); void* row_ptr = new std::string(std::move(*row_cache_entry)); auto row_handle = ioptions_.row_cache->Insert( row_cache_key.GetKey(), row_ptr, charge, &DeleteEntry); ioptions_.row_cache->Release(row_handle); } #endif // ROCKSDB_LITE return s; } Status TableCache::GetTableProperties( const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd, std::shared_ptr* properties, bool no_io) { Status s; auto table_reader = fd.table_reader; // table already been pre-loaded? if (table_reader) { *properties = table_reader->GetTableProperties(); return s; } Cache::Handle* table_handle = nullptr; s = FindTable(env_options, internal_comparator, fd, &table_handle, no_io); if (!s.ok()) { return s; } assert(table_handle); auto table = GetTableReaderFromHandle(table_handle); *properties = table->GetTableProperties(); ReleaseHandle(table_handle); return s; } size_t TableCache::GetMemoryUsageByTableReader( const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd) { Status s; auto table_reader = fd.table_reader; // table already been pre-loaded? if (table_reader) { return table_reader->ApproximateMemoryUsage(); } Cache::Handle* table_handle = nullptr; s = FindTable(env_options, internal_comparator, fd, &table_handle, true); if (!s.ok()) { return 0; } assert(table_handle); auto table = GetTableReaderFromHandle(table_handle); auto ret = table->ApproximateMemoryUsage(); ReleaseHandle(table_handle); return ret; } void TableCache::Evict(Cache* cache, uint64_t file_number) { cache->Erase(GetSliceForFileNumber(&file_number)); } } // namespace rocksdb rocksdb-4.5.1/db/table_cache.h000066400000000000000000000113601267530634000161270ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Thread-safe (provides internal synchronization) #pragma once #include #include #include #include "db/dbformat.h" #include "port/port.h" #include "rocksdb/cache.h" #include "rocksdb/env.h" #include "rocksdb/table.h" #include "rocksdb/options.h" #include "table/table_reader.h" namespace rocksdb { class Env; class Arena; struct FileDescriptor; class GetContext; class HistogramImpl; class InternalIterator; class TableCache { public: TableCache(const ImmutableCFOptions& ioptions, const EnvOptions& storage_options, Cache* cache); ~TableCache(); // Return an iterator for the specified file number (the corresponding // file length must be exactly "file_size" bytes). If "tableptr" is // non-nullptr, also sets "*tableptr" to point to the Table object // underlying the returned iterator, or nullptr if no Table object underlies // the returned iterator. The returned "*tableptr" object is owned by // the cache and should not be deleted, and is valid for as long as the // returned iterator is live. // @param skip_filters Disables loading/accessing the filter block InternalIterator* NewIterator( const ReadOptions& options, const EnvOptions& toptions, const InternalKeyComparator& internal_comparator, const FileDescriptor& file_fd, TableReader** table_reader_ptr = nullptr, HistogramImpl* file_read_hist = nullptr, bool for_compaction = false, Arena* arena = nullptr, bool skip_filters = false); // If a seek to internal key "k" in specified file finds an entry, // call (*handle_result)(arg, found_key, found_value) repeatedly until // it returns false. // @param skip_filters Disables loading/accessing the filter block Status Get(const ReadOptions& options, const InternalKeyComparator& internal_comparator, const FileDescriptor& file_fd, const Slice& k, GetContext* get_context, HistogramImpl* file_read_hist = nullptr, bool skip_filters = false); // Evict any entry for the specified file number static void Evict(Cache* cache, uint64_t file_number); // Find table reader // @param skip_filters Disables loading/accessing the filter block Status FindTable(const EnvOptions& toptions, const InternalKeyComparator& internal_comparator, const FileDescriptor& file_fd, Cache::Handle**, const bool no_io = false, bool record_read_stats = true, HistogramImpl* file_read_hist = nullptr, bool skip_filters = false); // Get TableReader from a cache handle. TableReader* GetTableReaderFromHandle(Cache::Handle* handle); // Get the table properties of a given table. // @no_io: indicates if we should load table to the cache if it is not present // in table cache yet. // @returns: `properties` will be reset on success. Please note that we will // return Status::Incomplete() if table is not present in cache and // we set `no_io` to be true. Status GetTableProperties(const EnvOptions& toptions, const InternalKeyComparator& internal_comparator, const FileDescriptor& file_meta, std::shared_ptr* properties, bool no_io = false); // Return total memory usage of the table reader of the file. // 0 if table reader of the file is not loaded. size_t GetMemoryUsageByTableReader( const EnvOptions& toptions, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd); // Release the handle from a cache void ReleaseHandle(Cache::Handle* handle); private: // Build a table reader Status GetTableReader(const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, const FileDescriptor& fd, bool sequential_mode, bool record_read_stats, HistogramImpl* file_read_hist, unique_ptr* table_reader, bool skip_filters = false); const ImmutableCFOptions& ioptions_; const EnvOptions& env_options_; Cache* const cache_; std::string row_cache_id_; }; } // namespace rocksdb rocksdb-4.5.1/db/table_properties_collector.cc000066400000000000000000000057421267530634000214730ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/table_properties_collector.h" #include "db/dbformat.h" #include "util/coding.h" #include "util/string_util.h" namespace rocksdb { Status InternalKeyPropertiesCollector::InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) { ParsedInternalKey ikey; if (!ParseInternalKey(key, &ikey)) { return Status::InvalidArgument("Invalid internal key"); } // Note: We count both, deletions and single deletions here. if (ikey.type == ValueType::kTypeDeletion || ikey.type == ValueType::kTypeSingleDeletion) { ++deleted_keys_; } return Status::OK(); } Status InternalKeyPropertiesCollector::Finish( UserCollectedProperties* properties) { assert(properties); assert(properties->find( InternalKeyTablePropertiesNames::kDeletedKeys) == properties->end()); std::string val; PutVarint64(&val, deleted_keys_); properties->insert({ InternalKeyTablePropertiesNames::kDeletedKeys, val }); return Status::OK(); } UserCollectedProperties InternalKeyPropertiesCollector::GetReadableProperties() const { return { { "kDeletedKeys", ToString(deleted_keys_) } }; } namespace { EntryType GetEntryType(ValueType value_type) { switch (value_type) { case kTypeValue: return kEntryPut; case kTypeDeletion: return kEntryDelete; case kTypeSingleDeletion: return kEntrySingleDelete; case kTypeMerge: return kEntryMerge; default: return kEntryOther; } } } // namespace Status UserKeyTablePropertiesCollector::InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) { ParsedInternalKey ikey; if (!ParseInternalKey(key, &ikey)) { return Status::InvalidArgument("Invalid internal key"); } return collector_->AddUserKey(ikey.user_key, value, GetEntryType(ikey.type), ikey.sequence, file_size); } Status UserKeyTablePropertiesCollector::Finish( UserCollectedProperties* properties) { return collector_->Finish(properties); } UserCollectedProperties UserKeyTablePropertiesCollector::GetReadableProperties() const { return collector_->GetReadableProperties(); } const std::string InternalKeyTablePropertiesNames::kDeletedKeys = "rocksdb.deleted.keys"; uint64_t GetDeletedKeys( const UserCollectedProperties& props) { auto pos = props.find(InternalKeyTablePropertiesNames::kDeletedKeys); if (pos == props.end()) { return 0; } Slice raw = pos->second; uint64_t val = 0; return GetVarint64(&raw, &val) ? val : 0; } } // namespace rocksdb rocksdb-4.5.1/db/table_properties_collector.h000066400000000000000000000103551267530634000213310ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file defines a collection of statistics collectors. #pragma once #include "rocksdb/table_properties.h" #include #include #include namespace rocksdb { struct InternalKeyTablePropertiesNames { static const std::string kDeletedKeys; }; // Base class for internal table properties collector. class IntTblPropCollector { public: virtual ~IntTblPropCollector() {} virtual Status Finish(UserCollectedProperties* properties) = 0; virtual const char* Name() const = 0; // @params key the user key that is inserted into the table. // @params value the value that is inserted into the table. virtual Status InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) = 0; virtual UserCollectedProperties GetReadableProperties() const = 0; virtual bool NeedCompact() const { return false; } }; // Factory for internal table properties collector. class IntTblPropCollectorFactory { public: virtual ~IntTblPropCollectorFactory() {} // has to be thread-safe virtual IntTblPropCollector* CreateIntTblPropCollector( uint32_t column_family_id) = 0; // The name of the properties collector can be used for debugging purpose. virtual const char* Name() const = 0; }; // Collecting the statistics for internal keys. Visible only by internal // rocksdb modules. class InternalKeyPropertiesCollector : public IntTblPropCollector { public: virtual Status InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) override; virtual Status Finish(UserCollectedProperties* properties) override; virtual const char* Name() const override { return "InternalKeyPropertiesCollector"; } UserCollectedProperties GetReadableProperties() const override; private: uint64_t deleted_keys_ = 0; }; class InternalKeyPropertiesCollectorFactory : public IntTblPropCollectorFactory { public: virtual IntTblPropCollector* CreateIntTblPropCollector( uint32_t column_family_id) override { return new InternalKeyPropertiesCollector(); } virtual const char* Name() const override { return "InternalKeyPropertiesCollectorFactory"; } }; // When rocksdb creates a new table, it will encode all "user keys" into // "internal keys", which contains meta information of a given entry. // // This class extracts user key from the encoded internal key when Add() is // invoked. class UserKeyTablePropertiesCollector : public IntTblPropCollector { public: // transfer of ownership explicit UserKeyTablePropertiesCollector(TablePropertiesCollector* collector) : collector_(collector) {} virtual ~UserKeyTablePropertiesCollector() {} virtual Status InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) override; virtual Status Finish(UserCollectedProperties* properties) override; virtual const char* Name() const override { return collector_->Name(); } UserCollectedProperties GetReadableProperties() const override; virtual bool NeedCompact() const override { return collector_->NeedCompact(); } protected: std::unique_ptr collector_; }; class UserKeyTablePropertiesCollectorFactory : public IntTblPropCollectorFactory { public: explicit UserKeyTablePropertiesCollectorFactory( std::shared_ptr user_collector_factory) : user_collector_factory_(user_collector_factory) {} virtual IntTblPropCollector* CreateIntTblPropCollector( uint32_t column_family_id) override { TablePropertiesCollectorFactory::Context context; context.column_family_id = column_family_id; return new UserKeyTablePropertiesCollector( user_collector_factory_->CreateTablePropertiesCollector(context)); } virtual const char* Name() const override { return user_collector_factory_->Name(); } private: std::shared_ptr user_collector_factory_; }; } // namespace rocksdb rocksdb-4.5.1/db/table_properties_collector_test.cc000066400000000000000000000430261267530634000225270ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include #include #include "db/db_impl.h" #include "db/dbformat.h" #include "db/table_properties_collector.h" #include "rocksdb/immutable_options.h" #include "rocksdb/table.h" #include "table/block_based_table_factory.h" #include "table/meta_blocks.h" #include "table/plain_table_factory.h" #include "table/table_builder.h" #include "util/coding.h" #include "util/file_reader_writer.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class TablePropertiesTest : public testing::Test, public testing::WithParamInterface { public: virtual void SetUp() override { backward_mode_ = GetParam(); } bool backward_mode_; }; // Utilities test functions namespace { static const uint32_t kTestColumnFamilyId = 66; void MakeBuilder(const Options& options, const ImmutableCFOptions& ioptions, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, std::unique_ptr* writable, std::unique_ptr* builder) { unique_ptr wf(new test::StringSink); writable->reset(new WritableFileWriter(std::move(wf), EnvOptions())); builder->reset(NewTableBuilder( ioptions, internal_comparator, int_tbl_prop_collector_factories, kTestColumnFamilyId /* column_family_id */, writable->get(), options.compression, options.compression_opts)); } } // namespace // Collects keys that starts with "A" in a table. class RegularKeysStartWithA: public TablePropertiesCollector { public: const char* Name() const override { return "RegularKeysStartWithA"; } Status Finish(UserCollectedProperties* properties) override { std::string encoded; std::string encoded_num_puts; std::string encoded_num_deletes; std::string encoded_num_single_deletes; std::string encoded_num_size_changes; PutVarint32(&encoded, count_); PutVarint32(&encoded_num_puts, num_puts_); PutVarint32(&encoded_num_deletes, num_deletes_); PutVarint32(&encoded_num_single_deletes, num_single_deletes_); PutVarint32(&encoded_num_size_changes, num_size_changes_); *properties = UserCollectedProperties{ {"TablePropertiesTest", message_}, {"Count", encoded}, {"NumPuts", encoded_num_puts}, {"NumDeletes", encoded_num_deletes}, {"NumSingleDeletes", encoded_num_single_deletes}, {"NumSizeChanges", encoded_num_size_changes}, }; return Status::OK(); } Status AddUserKey(const Slice& user_key, const Slice& value, EntryType type, SequenceNumber seq, uint64_t file_size) override { // simply asssume all user keys are not empty. if (user_key.data()[0] == 'A') { ++count_; } if (type == kEntryPut) { num_puts_++; } else if (type == kEntryDelete) { num_deletes_++; } else if (type == kEntrySingleDelete) { num_single_deletes_++; } if (file_size < file_size_) { message_ = "File size should not decrease."; } else if (file_size != file_size_) { num_size_changes_++; } return Status::OK(); } virtual UserCollectedProperties GetReadableProperties() const override { return UserCollectedProperties{}; } private: std::string message_ = "Rocksdb"; uint32_t count_ = 0; uint32_t num_puts_ = 0; uint32_t num_deletes_ = 0; uint32_t num_single_deletes_ = 0; uint32_t num_size_changes_ = 0; uint64_t file_size_ = 0; }; // Collects keys that starts with "A" in a table. Backward compatible mode // It is also used to test internal key table property collector class RegularKeysStartWithABackwardCompatible : public TablePropertiesCollector { public: const char* Name() const override { return "RegularKeysStartWithA"; } Status Finish(UserCollectedProperties* properties) override { std::string encoded; PutVarint32(&encoded, count_); *properties = UserCollectedProperties{{"TablePropertiesTest", "Rocksdb"}, {"Count", encoded}}; return Status::OK(); } Status Add(const Slice& user_key, const Slice& value) override { // simply asssume all user keys are not empty. if (user_key.data()[0] == 'A') { ++count_; } return Status::OK(); } virtual UserCollectedProperties GetReadableProperties() const override { return UserCollectedProperties{}; } private: uint32_t count_ = 0; }; class RegularKeysStartWithAInternal : public IntTblPropCollector { public: const char* Name() const override { return "RegularKeysStartWithA"; } Status Finish(UserCollectedProperties* properties) override { std::string encoded; PutVarint32(&encoded, count_); *properties = UserCollectedProperties{{"TablePropertiesTest", "Rocksdb"}, {"Count", encoded}}; return Status::OK(); } Status InternalAdd(const Slice& user_key, const Slice& value, uint64_t file_size) override { // simply asssume all user keys are not empty. if (user_key.data()[0] == 'A') { ++count_; } return Status::OK(); } virtual UserCollectedProperties GetReadableProperties() const override { return UserCollectedProperties{}; } private: uint32_t count_ = 0; }; class RegularKeysStartWithAFactory : public IntTblPropCollectorFactory, public TablePropertiesCollectorFactory { public: explicit RegularKeysStartWithAFactory(bool backward_mode) : backward_mode_(backward_mode) {} virtual TablePropertiesCollector* CreateTablePropertiesCollector( TablePropertiesCollectorFactory::Context context) override { EXPECT_EQ(kTestColumnFamilyId, context.column_family_id); if (!backward_mode_) { return new RegularKeysStartWithA(); } else { return new RegularKeysStartWithABackwardCompatible(); } } virtual IntTblPropCollector* CreateIntTblPropCollector( uint32_t column_family_id) override { return new RegularKeysStartWithAInternal(); } const char* Name() const override { return "RegularKeysStartWithA"; } bool backward_mode_; }; class FlushBlockEveryThreePolicy : public FlushBlockPolicy { public: virtual bool Update(const Slice& key, const Slice& value) override { return (++count_ % 3U == 0); } private: uint64_t count_ = 0; }; class FlushBlockEveryThreePolicyFactory : public FlushBlockPolicyFactory { public: explicit FlushBlockEveryThreePolicyFactory() {} const char* Name() const override { return "FlushBlockEveryThreePolicyFactory"; } FlushBlockPolicy* NewFlushBlockPolicy( const BlockBasedTableOptions& table_options, const BlockBuilder& data_block_builder) const override { return new FlushBlockEveryThreePolicy; } }; extern const uint64_t kBlockBasedTableMagicNumber; extern const uint64_t kPlainTableMagicNumber; namespace { void TestCustomizedTablePropertiesCollector( bool backward_mode, uint64_t magic_number, bool test_int_tbl_prop_collector, const Options& options, const InternalKeyComparator& internal_comparator) { // make sure the entries will be inserted with order. std::map, std::string> kvs = { {{"About ", kTypeValue}, "val5"}, // starts with 'A' {{"Abstract", kTypeValue}, "val2"}, // starts with 'A' {{"Around ", kTypeValue}, "val7"}, // starts with 'A' {{"Beyond ", kTypeValue}, "val3"}, {{"Builder ", kTypeValue}, "val1"}, {{"Love ", kTypeDeletion}, ""}, {{"Cancel ", kTypeValue}, "val4"}, {{"Find ", kTypeValue}, "val6"}, {{"Rocks ", kTypeDeletion}, ""}, {{"Foo ", kTypeSingleDeletion}, ""}, }; // -- Step 1: build table std::unique_ptr builder; std::unique_ptr writer; const ImmutableCFOptions ioptions(options); std::vector> int_tbl_prop_collector_factories; if (test_int_tbl_prop_collector) { int_tbl_prop_collector_factories.emplace_back( new RegularKeysStartWithAFactory(backward_mode)); } else { GetIntTblPropCollectorFactory(options, &int_tbl_prop_collector_factories); } MakeBuilder(options, ioptions, internal_comparator, &int_tbl_prop_collector_factories, &writer, &builder); SequenceNumber seqNum = 0U; for (const auto& kv : kvs) { InternalKey ikey(kv.first.first, seqNum++, kv.first.second); builder->Add(ikey.Encode(), kv.second); } ASSERT_OK(builder->Finish()); writer->Flush(); // -- Step 2: Read properties test::StringSink* fwf = static_cast(writer->writable_file()); std::unique_ptr fake_file_reader( test::GetRandomAccessFileReader( new test::StringSource(fwf->contents()))); TableProperties* props; Status s = ReadTableProperties(fake_file_reader.get(), fwf->contents().size(), magic_number, Env::Default(), nullptr, &props); std::unique_ptr props_guard(props); ASSERT_OK(s); auto user_collected = props->user_collected_properties; ASSERT_NE(user_collected.find("TablePropertiesTest"), user_collected.end()); ASSERT_EQ("Rocksdb", user_collected.at("TablePropertiesTest")); uint32_t starts_with_A = 0; ASSERT_NE(user_collected.find("Count"), user_collected.end()); Slice key(user_collected.at("Count")); ASSERT_TRUE(GetVarint32(&key, &starts_with_A)); ASSERT_EQ(3u, starts_with_A); if (!backward_mode && !test_int_tbl_prop_collector) { uint32_t num_puts; ASSERT_NE(user_collected.find("NumPuts"), user_collected.end()); Slice key_puts(user_collected.at("NumPuts")); ASSERT_TRUE(GetVarint32(&key_puts, &num_puts)); ASSERT_EQ(7u, num_puts); uint32_t num_deletes; ASSERT_NE(user_collected.find("NumDeletes"), user_collected.end()); Slice key_deletes(user_collected.at("NumDeletes")); ASSERT_TRUE(GetVarint32(&key_deletes, &num_deletes)); ASSERT_EQ(2u, num_deletes); uint32_t num_single_deletes; ASSERT_NE(user_collected.find("NumSingleDeletes"), user_collected.end()); Slice key_single_deletes(user_collected.at("NumSingleDeletes")); ASSERT_TRUE(GetVarint32(&key_single_deletes, &num_single_deletes)); ASSERT_EQ(1u, num_single_deletes); uint32_t num_size_changes; ASSERT_NE(user_collected.find("NumSizeChanges"), user_collected.end()); Slice key_size_changes(user_collected.at("NumSizeChanges")); ASSERT_TRUE(GetVarint32(&key_size_changes, &num_size_changes)); ASSERT_GE(num_size_changes, 2u); } } } // namespace TEST_P(TablePropertiesTest, CustomizedTablePropertiesCollector) { // Test properties collectors with internal keys or regular keys // for block based table for (bool encode_as_internal : { true, false }) { Options options; BlockBasedTableOptions table_options; table_options.flush_block_policy_factory = std::make_shared(); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); test::PlainInternalKeyComparator ikc(options.comparator); std::shared_ptr collector_factory( new RegularKeysStartWithAFactory(backward_mode_)); options.table_properties_collector_factories.resize(1); options.table_properties_collector_factories[0] = collector_factory; TestCustomizedTablePropertiesCollector(backward_mode_, kBlockBasedTableMagicNumber, encode_as_internal, options, ikc); #ifndef ROCKSDB_LITE // PlainTable is not supported in Lite // test plain table PlainTableOptions plain_table_options; plain_table_options.user_key_len = 8; plain_table_options.bloom_bits_per_key = 8; plain_table_options.hash_table_ratio = 0; options.table_factory = std::make_shared(plain_table_options); TestCustomizedTablePropertiesCollector(backward_mode_, kPlainTableMagicNumber, encode_as_internal, options, ikc); #endif // !ROCKSDB_LITE } } namespace { void TestInternalKeyPropertiesCollector( bool backward_mode, uint64_t magic_number, bool sanitized, std::shared_ptr table_factory) { InternalKey keys[] = { InternalKey("A ", 0, ValueType::kTypeValue), InternalKey("B ", 1, ValueType::kTypeValue), InternalKey("C ", 2, ValueType::kTypeValue), InternalKey("W ", 3, ValueType::kTypeDeletion), InternalKey("X ", 4, ValueType::kTypeDeletion), InternalKey("Y ", 5, ValueType::kTypeDeletion), InternalKey("Z ", 6, ValueType::kTypeDeletion), InternalKey("a ", 7, ValueType::kTypeSingleDeletion), }; std::unique_ptr builder; std::unique_ptr writable; Options options; test::PlainInternalKeyComparator pikc(options.comparator); std::vector> int_tbl_prop_collector_factories; options.table_factory = table_factory; if (sanitized) { options.table_properties_collector_factories.emplace_back( new RegularKeysStartWithAFactory(backward_mode)); // with sanitization, even regular properties collector will be able to // handle internal keys. auto comparator = options.comparator; // HACK: Set options.info_log to avoid writing log in // SanitizeOptions(). options.info_log = std::make_shared(); options = SanitizeOptions("db", // just a place holder &pikc, options); GetIntTblPropCollectorFactory(options, &int_tbl_prop_collector_factories); options.comparator = comparator; } else { int_tbl_prop_collector_factories.emplace_back( new InternalKeyPropertiesCollectorFactory); } const ImmutableCFOptions ioptions(options); for (int iter = 0; iter < 2; ++iter) { MakeBuilder(options, ioptions, pikc, &int_tbl_prop_collector_factories, &writable, &builder); for (const auto& k : keys) { builder->Add(k.Encode(), "val"); } ASSERT_OK(builder->Finish()); writable->Flush(); test::StringSink* fwf = static_cast(writable->writable_file()); unique_ptr reader(test::GetRandomAccessFileReader( new test::StringSource(fwf->contents()))); TableProperties* props; Status s = ReadTableProperties(reader.get(), fwf->contents().size(), magic_number, Env::Default(), nullptr, &props); ASSERT_OK(s); std::unique_ptr props_guard(props); auto user_collected = props->user_collected_properties; uint64_t deleted = GetDeletedKeys(user_collected); ASSERT_EQ(5u, deleted); // deletes + single-deletes if (sanitized) { uint32_t starts_with_A = 0; ASSERT_NE(user_collected.find("Count"), user_collected.end()); Slice key(user_collected.at("Count")); ASSERT_TRUE(GetVarint32(&key, &starts_with_A)); ASSERT_EQ(1u, starts_with_A); if (!backward_mode) { uint32_t num_puts; ASSERT_NE(user_collected.find("NumPuts"), user_collected.end()); Slice key_puts(user_collected.at("NumPuts")); ASSERT_TRUE(GetVarint32(&key_puts, &num_puts)); ASSERT_EQ(3u, num_puts); uint32_t num_deletes; ASSERT_NE(user_collected.find("NumDeletes"), user_collected.end()); Slice key_deletes(user_collected.at("NumDeletes")); ASSERT_TRUE(GetVarint32(&key_deletes, &num_deletes)); ASSERT_EQ(4u, num_deletes); uint32_t num_single_deletes; ASSERT_NE(user_collected.find("NumSingleDeletes"), user_collected.end()); Slice key_single_deletes(user_collected.at("NumSingleDeletes")); ASSERT_TRUE(GetVarint32(&key_single_deletes, &num_single_deletes)); ASSERT_EQ(1u, num_single_deletes); } } } } } // namespace TEST_P(TablePropertiesTest, InternalKeyPropertiesCollector) { TestInternalKeyPropertiesCollector( backward_mode_, kBlockBasedTableMagicNumber, true /* sanitize */, std::make_shared()); if (backward_mode_) { TestInternalKeyPropertiesCollector( backward_mode_, kBlockBasedTableMagicNumber, false /* not sanitize */, std::make_shared()); } #ifndef ROCKSDB_LITE // PlainTable is not supported in Lite PlainTableOptions plain_table_options; plain_table_options.user_key_len = 8; plain_table_options.bloom_bits_per_key = 8; plain_table_options.hash_table_ratio = 0; TestInternalKeyPropertiesCollector( backward_mode_, kPlainTableMagicNumber, false /* not sanitize */, std::make_shared(plain_table_options)); #endif // !ROCKSDB_LITE } INSTANTIATE_TEST_CASE_P(InternalKeyPropertiesCollector, TablePropertiesTest, ::testing::Bool()); INSTANTIATE_TEST_CASE_P(CustomizedTablePropertiesCollector, TablePropertiesTest, ::testing::Bool()); } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/transaction_log_impl.cc000066400000000000000000000216231267530634000202650ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include "db/transaction_log_impl.h" #include "db/write_batch_internal.h" #include "util/file_reader_writer.h" namespace rocksdb { TransactionLogIteratorImpl::TransactionLogIteratorImpl( const std::string& dir, const DBOptions* options, const TransactionLogIterator::ReadOptions& read_options, const EnvOptions& soptions, const SequenceNumber seq, std::unique_ptr files, VersionSet const* const versions) : dir_(dir), options_(options), read_options_(read_options), soptions_(soptions), startingSequenceNumber_(seq), files_(std::move(files)), started_(false), isValid_(false), currentFileIndex_(0), currentBatchSeq_(0), currentLastSeq_(0), versions_(versions) { assert(files_ != nullptr); assert(versions_ != nullptr); reporter_.env = options_->env; reporter_.info_log = options_->info_log.get(); SeekToStartSequence(); // Seek till starting sequence } Status TransactionLogIteratorImpl::OpenLogFile( const LogFile* logFile, unique_ptr* file_reader) { Env* env = options_->env; unique_ptr file; Status s; if (logFile->Type() == kArchivedLogFile) { std::string fname = ArchivedLogFileName(dir_, logFile->LogNumber()); s = env->NewSequentialFile(fname, &file, soptions_); } else { std::string fname = LogFileName(dir_, logFile->LogNumber()); s = env->NewSequentialFile(fname, &file, soptions_); if (!s.ok()) { // If cannot open file in DB directory. // Try the archive dir, as it could have moved in the meanwhile. fname = ArchivedLogFileName(dir_, logFile->LogNumber()); s = env->NewSequentialFile(fname, &file, soptions_); } } if (s.ok()) { file_reader->reset(new SequentialFileReader(std::move(file))); } return s; } BatchResult TransactionLogIteratorImpl::GetBatch() { assert(isValid_); // cannot call in a non valid state. BatchResult result; result.sequence = currentBatchSeq_; result.writeBatchPtr = std::move(currentBatch_); return result; } Status TransactionLogIteratorImpl::status() { return currentStatus_; } bool TransactionLogIteratorImpl::Valid() { return started_ && isValid_; } bool TransactionLogIteratorImpl::RestrictedRead( Slice* record, std::string* scratch) { // Don't read if no more complete entries to read from logs if (currentLastSeq_ >= versions_->LastSequence()) { return false; } return currentLogReader_->ReadRecord(record, scratch); } void TransactionLogIteratorImpl::SeekToStartSequence( uint64_t startFileIndex, bool strict) { std::string scratch; Slice record; started_ = false; isValid_ = false; if (files_->size() <= startFileIndex) { return; } Status s = OpenLogReader(files_->at(startFileIndex).get()); if (!s.ok()) { currentStatus_ = s; reporter_.Info(currentStatus_.ToString().c_str()); return; } while (RestrictedRead(&record, &scratch)) { if (record.size() < 12) { reporter_.Corruption( record.size(), Status::Corruption("very small log record")); continue; } UpdateCurrentWriteBatch(record); if (currentLastSeq_ >= startingSequenceNumber_) { if (strict && currentBatchSeq_ != startingSequenceNumber_) { currentStatus_ = Status::Corruption("Gap in sequence number. Could not " "seek to required sequence number"); reporter_.Info(currentStatus_.ToString().c_str()); return; } else if (strict) { reporter_.Info("Could seek required sequence number. Iterator will " "continue."); } isValid_ = true; started_ = true; // set started_ as we could seek till starting sequence return; } else { isValid_ = false; } } // Could not find start sequence in first file. Normally this must be the // only file. Otherwise log the error and let the iterator return next entry // If strict is set, we want to seek exactly till the start sequence and it // should have been present in the file we scanned above if (strict) { currentStatus_ = Status::Corruption("Gap in sequence number. Could not " "seek to required sequence number"); reporter_.Info(currentStatus_.ToString().c_str()); } else if (files_->size() != 1) { currentStatus_ = Status::Corruption("Start sequence was not found, " "skipping to the next available"); reporter_.Info(currentStatus_.ToString().c_str()); // Let NextImpl find the next available entry. started_ remains false // because we don't want to check for gaps while moving to start sequence NextImpl(true); } } void TransactionLogIteratorImpl::Next() { return NextImpl(false); } void TransactionLogIteratorImpl::NextImpl(bool internal) { std::string scratch; Slice record; isValid_ = false; if (!internal && !started_) { // Runs every time until we can seek to the start sequence return SeekToStartSequence(); } while(true) { assert(currentLogReader_); if (currentLogReader_->IsEOF()) { currentLogReader_->UnmarkEOF(); } while (RestrictedRead(&record, &scratch)) { if (record.size() < 12) { reporter_.Corruption( record.size(), Status::Corruption("very small log record")); continue; } else { // started_ should be true if called by application assert(internal || started_); // started_ should be false if called internally assert(!internal || !started_); UpdateCurrentWriteBatch(record); if (internal && !started_) { started_ = true; } return; } } // Open the next file if (currentFileIndex_ < files_->size() - 1) { ++currentFileIndex_; Status s = OpenLogReader(files_->at(currentFileIndex_).get()); if (!s.ok()) { isValid_ = false; currentStatus_ = s; return; } } else { isValid_ = false; if (currentLastSeq_ == versions_->LastSequence()) { currentStatus_ = Status::OK(); } else { currentStatus_ = Status::Corruption("NO MORE DATA LEFT"); } return; } } } bool TransactionLogIteratorImpl::IsBatchExpected( const WriteBatch* batch, const SequenceNumber expectedSeq) { assert(batch); SequenceNumber batchSeq = WriteBatchInternal::Sequence(batch); if (batchSeq != expectedSeq) { char buf[200]; snprintf(buf, sizeof(buf), "Discontinuity in log records. Got seq=%" PRIu64 ", Expected seq=%" PRIu64 ", Last flushed seq=%" PRIu64 ".Log iterator will reseek the correct batch.", batchSeq, expectedSeq, versions_->LastSequence()); reporter_.Info(buf); return false; } return true; } void TransactionLogIteratorImpl::UpdateCurrentWriteBatch(const Slice& record) { std::unique_ptr batch(new WriteBatch()); WriteBatchInternal::SetContents(batch.get(), record); SequenceNumber expectedSeq = currentLastSeq_ + 1; // If the iterator has started, then confirm that we get continuous batches if (started_ && !IsBatchExpected(batch.get(), expectedSeq)) { // Seek to the batch having expected sequence number if (expectedSeq < files_->at(currentFileIndex_)->StartSequence()) { // Expected batch must lie in the previous log file // Avoid underflow. if (currentFileIndex_ != 0) { currentFileIndex_--; } } startingSequenceNumber_ = expectedSeq; // currentStatus_ will be set to Ok if reseek succeeds currentStatus_ = Status::NotFound("Gap in sequence numbers"); return SeekToStartSequence(currentFileIndex_, true); } currentBatchSeq_ = WriteBatchInternal::Sequence(batch.get()); currentLastSeq_ = currentBatchSeq_ + WriteBatchInternal::Count(batch.get()) - 1; // currentBatchSeq_ can only change here assert(currentLastSeq_ <= versions_->LastSequence()); currentBatch_ = move(batch); isValid_ = true; currentStatus_ = Status::OK(); } Status TransactionLogIteratorImpl::OpenLogReader(const LogFile* logFile) { unique_ptr file; Status s = OpenLogFile(logFile, &file); if (!s.ok()) { return s; } assert(file); currentLogReader_.reset(new log::Reader(options_->info_log, std::move(file), &reporter_, read_options_.verify_checksums_, 0, logFile->LogNumber())); return Status::OK(); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/transaction_log_impl.h000066400000000000000000000103361267530634000201260ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #pragma once #include #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/types.h" #include "rocksdb/transaction_log.h" #include "db/version_set.h" #include "db/log_reader.h" #include "db/filename.h" #include "port/port.h" namespace rocksdb { class LogFileImpl : public LogFile { public: LogFileImpl(uint64_t logNum, WalFileType logType, SequenceNumber startSeq, uint64_t sizeBytes) : logNumber_(logNum), type_(logType), startSequence_(startSeq), sizeFileBytes_(sizeBytes) { } std::string PathName() const override { if (type_ == kArchivedLogFile) { return ArchivedLogFileName("", logNumber_); } return LogFileName("", logNumber_); } uint64_t LogNumber() const override { return logNumber_; } WalFileType Type() const override { return type_; } SequenceNumber StartSequence() const override { return startSequence_; } uint64_t SizeFileBytes() const override { return sizeFileBytes_; } bool operator < (const LogFile& that) const { return LogNumber() < that.LogNumber(); } private: uint64_t logNumber_; WalFileType type_; SequenceNumber startSequence_; uint64_t sizeFileBytes_; }; class TransactionLogIteratorImpl : public TransactionLogIterator { public: TransactionLogIteratorImpl( const std::string& dir, const DBOptions* options, const TransactionLogIterator::ReadOptions& read_options, const EnvOptions& soptions, const SequenceNumber seqNum, std::unique_ptr files, VersionSet const* const versions); virtual bool Valid() override; virtual void Next() override; virtual Status status() override; virtual BatchResult GetBatch() override; private: const std::string& dir_; const DBOptions* options_; const TransactionLogIterator::ReadOptions read_options_; const EnvOptions& soptions_; SequenceNumber startingSequenceNumber_; std::unique_ptr files_; bool started_; bool isValid_; // not valid when it starts of. Status currentStatus_; size_t currentFileIndex_; std::unique_ptr currentBatch_; unique_ptr currentLogReader_; Status OpenLogFile(const LogFile* logFile, unique_ptr* file); struct LogReporter : public log::Reader::Reporter { Env* env; Logger* info_log; virtual void Corruption(size_t bytes, const Status& s) override { Log(InfoLogLevel::ERROR_LEVEL, info_log, "dropping %" ROCKSDB_PRIszt " bytes; %s", bytes, s.ToString().c_str()); } virtual void Info(const char* s) { Log(InfoLogLevel::INFO_LEVEL, info_log, "%s", s); } } reporter_; SequenceNumber currentBatchSeq_; // sequence number at start of current batch SequenceNumber currentLastSeq_; // last sequence in the current batch // Used only to get latest seq. num // TODO(icanadi) can this be just a callback? VersionSet const* const versions_; // Reads from transaction log only if the writebatch record has been written bool RestrictedRead(Slice* record, std::string* scratch); // Seeks to startingSequenceNumber reading from startFileIndex in files_. // If strict is set,then must get a batch starting with startingSequenceNumber void SeekToStartSequence(uint64_t startFileIndex = 0, bool strict = false); // Implementation of Next. SeekToStartSequence calls it internally with // internal=true to let it find next entry even if it has to jump gaps because // the iterator may start off from the first available entry but promises to // be continuous after that void NextImpl(bool internal = false); // Check if batch is expected, else return false bool IsBatchExpected(const WriteBatch* batch, SequenceNumber expectedSeq); // Update current batch if a continuous batch is found, else return false void UpdateCurrentWriteBatch(const Slice& record); Status OpenLogReader(const LogFile* file); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/db/version_builder.cc000066400000000000000000000303571267530634000172550ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/version_builder.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include #include #include #include #include "db/dbformat.h" #include "db/internal_stats.h" #include "db/table_cache.h" #include "db/version_set.h" #include "table/table_reader.h" namespace rocksdb { bool NewestFirstBySeqNo(FileMetaData* a, FileMetaData* b) { if (a->smallest_seqno != b->smallest_seqno) { return a->smallest_seqno > b->smallest_seqno; } if (a->largest_seqno != b->largest_seqno) { return a->largest_seqno > b->largest_seqno; } // Break ties by file number return a->fd.GetNumber() > b->fd.GetNumber(); } namespace { bool BySmallestKey(FileMetaData* a, FileMetaData* b, const InternalKeyComparator* cmp) { int r = cmp->Compare(a->smallest, b->smallest); if (r != 0) { return (r < 0); } // Break ties by file number return (a->fd.GetNumber() < b->fd.GetNumber()); } } // namespace class VersionBuilder::Rep { private: // Helper to sort files_ in v // kLevel0 -- NewestFirstBySeqNo // kLevelNon0 -- BySmallestKey struct FileComparator { enum SortMethod { kLevel0 = 0, kLevelNon0 = 1, } sort_method; const InternalKeyComparator* internal_comparator; bool operator()(FileMetaData* f1, FileMetaData* f2) const { switch (sort_method) { case kLevel0: return NewestFirstBySeqNo(f1, f2); case kLevelNon0: return BySmallestKey(f1, f2, internal_comparator); } assert(false); return false; } }; struct LevelState { std::unordered_set deleted_files; // Map from file number to file meta data. std::unordered_map added_files; }; const EnvOptions& env_options_; Logger* info_log_; TableCache* table_cache_; VersionStorageInfo* base_vstorage_; LevelState* levels_; FileComparator level_zero_cmp_; FileComparator level_nonzero_cmp_; public: Rep(const EnvOptions& env_options, Logger* info_log, TableCache* table_cache, VersionStorageInfo* base_vstorage) : env_options_(env_options), info_log_(info_log), table_cache_(table_cache), base_vstorage_(base_vstorage) { levels_ = new LevelState[base_vstorage_->num_levels()]; level_zero_cmp_.sort_method = FileComparator::kLevel0; level_nonzero_cmp_.sort_method = FileComparator::kLevelNon0; level_nonzero_cmp_.internal_comparator = base_vstorage_->InternalComparator(); } ~Rep() { for (int level = 0; level < base_vstorage_->num_levels(); level++) { const auto& added = levels_[level].added_files; for (auto& pair : added) { UnrefFile(pair.second); } } delete[] levels_; } void UnrefFile(FileMetaData* f) { f->refs--; if (f->refs <= 0) { if (f->table_reader_handle) { assert(table_cache_ != nullptr); table_cache_->ReleaseHandle(f->table_reader_handle); f->table_reader_handle = nullptr; } delete f; } } void CheckConsistency(VersionStorageInfo* vstorage) { #ifndef NDEBUG // make sure the files are sorted correctly for (int level = 0; level < vstorage->num_levels(); level++) { auto& level_files = vstorage->LevelFiles(level); for (size_t i = 1; i < level_files.size(); i++) { auto f1 = level_files[i - 1]; auto f2 = level_files[i]; if (level == 0) { assert(level_zero_cmp_(f1, f2)); assert(f1->largest_seqno > f2->largest_seqno || // We can have multiple files with seqno = 0 as a result of // using DB::AddFile() (f1->largest_seqno == 0 && f2->largest_seqno == 0)); } else { assert(level_nonzero_cmp_(f1, f2)); // Make sure there is no overlap in levels > 0 if (vstorage->InternalComparator()->Compare(f1->largest, f2->smallest) >= 0) { fprintf(stderr, "overlapping ranges in same level %s vs. %s\n", (f1->largest).DebugString().c_str(), (f2->smallest).DebugString().c_str()); abort(); } } } } #endif } void CheckConsistencyForDeletes(VersionEdit* edit, uint64_t number, int level) { #ifndef NDEBUG // a file to be deleted better exist in the previous version bool found = false; for (int l = 0; !found && l < base_vstorage_->num_levels(); l++) { const std::vector& base_files = base_vstorage_->LevelFiles(l); for (size_t i = 0; i < base_files.size(); i++) { FileMetaData* f = base_files[i]; if (f->fd.GetNumber() == number) { found = true; break; } } } // if the file did not exist in the previous version, then it // is possibly moved from lower level to higher level in current // version for (int l = level + 1; !found && l < base_vstorage_->num_levels(); l++) { auto& level_added = levels_[l].added_files; auto got = level_added.find(number); if (got != level_added.end()) { found = true; break; } } // maybe this file was added in a previous edit that was Applied if (!found) { auto& level_added = levels_[level].added_files; auto got = level_added.find(number); if (got != level_added.end()) { found = true; } } if (!found) { fprintf(stderr, "not found %" PRIu64 "\n", number); } assert(found); #endif } // Apply all of the edits in *edit to the current state. void Apply(VersionEdit* edit) { CheckConsistency(base_vstorage_); // Delete files const VersionEdit::DeletedFileSet& del = edit->GetDeletedFiles(); for (const auto& del_file : del) { const auto level = del_file.first; const auto number = del_file.second; levels_[level].deleted_files.insert(number); CheckConsistencyForDeletes(edit, number, level); auto exising = levels_[level].added_files.find(number); if (exising != levels_[level].added_files.end()) { UnrefFile(exising->second); levels_[level].added_files.erase(number); } } // Add new files for (const auto& new_file : edit->GetNewFiles()) { const int level = new_file.first; FileMetaData* f = new FileMetaData(new_file.second); f->refs = 1; assert(levels_[level].added_files.find(f->fd.GetNumber()) == levels_[level].added_files.end()); levels_[level].deleted_files.erase(f->fd.GetNumber()); levels_[level].added_files[f->fd.GetNumber()] = f; } } // Save the current state in *v. void SaveTo(VersionStorageInfo* vstorage) { CheckConsistency(base_vstorage_); CheckConsistency(vstorage); for (int level = 0; level < base_vstorage_->num_levels(); level++) { const auto& cmp = (level == 0) ? level_zero_cmp_ : level_nonzero_cmp_; // Merge the set of added files with the set of pre-existing files. // Drop any deleted files. Store the result in *v. const auto& base_files = base_vstorage_->LevelFiles(level); auto base_iter = base_files.begin(); auto base_end = base_files.end(); const auto& unordered_added_files = levels_[level].added_files; vstorage->Reserve(level, base_files.size() + unordered_added_files.size()); // Sort added files for the level. std::vector added_files; added_files.reserve(unordered_added_files.size()); for (const auto& pair : unordered_added_files) { added_files.push_back(pair.second); } std::sort(added_files.begin(), added_files.end(), cmp); #ifndef NDEBUG FileMetaData* prev_file = nullptr; #endif for (const auto& added : added_files) { #ifndef NDEBUG if (level > 0 && prev_file != nullptr) { assert(base_vstorage_->InternalComparator()->Compare( prev_file->smallest, added->smallest) <= 0); } prev_file = added; #endif // Add all smaller files listed in base_ for (auto bpos = std::upper_bound(base_iter, base_end, added, cmp); base_iter != bpos; ++base_iter) { MaybeAddFile(vstorage, level, *base_iter); } MaybeAddFile(vstorage, level, added); } // Add remaining base files for (; base_iter != base_end; ++base_iter) { MaybeAddFile(vstorage, level, *base_iter); } } CheckConsistency(vstorage); } void LoadTableHandlers(InternalStats* internal_stats, int max_threads) { assert(table_cache_ != nullptr); // std::vector> files_meta; for (int level = 0; level < base_vstorage_->num_levels(); level++) { for (auto& file_meta_pair : levels_[level].added_files) { auto* file_meta = file_meta_pair.second; assert(!file_meta->table_reader_handle); files_meta.emplace_back(file_meta, level); } } std::atomic next_file_meta_idx(0); std::function load_handlers_func = [&]() { while (true) { size_t file_idx = next_file_meta_idx.fetch_add(1); if (file_idx >= files_meta.size()) { break; } auto* file_meta = files_meta[file_idx].first; int level = files_meta[file_idx].second; table_cache_->FindTable(env_options_, *(base_vstorage_->InternalComparator()), file_meta->fd, &file_meta->table_reader_handle, false /*no_io */, true /* record_read_stats */, internal_stats->GetFileReadHist(level)); if (file_meta->table_reader_handle != nullptr) { // Load table_reader file_meta->fd.table_reader = table_cache_->GetTableReaderFromHandle( file_meta->table_reader_handle); } } }; if (max_threads <= 1) { load_handlers_func(); } else { std::vector threads; for (int i = 0; i < max_threads; i++) { threads.emplace_back(load_handlers_func); } for (auto& t : threads) { t.join(); } } } void MaybeAddFile(VersionStorageInfo* vstorage, int level, FileMetaData* f) { if (levels_[level].deleted_files.count(f->fd.GetNumber()) > 0) { // f is to-be-delected table file vstorage->RemoveCurrentStats(f); } else { vstorage->AddFile(level, f, info_log_); } } }; VersionBuilder::VersionBuilder(const EnvOptions& env_options, TableCache* table_cache, VersionStorageInfo* base_vstorage, Logger* info_log) : rep_(new Rep(env_options, info_log, table_cache, base_vstorage)) {} VersionBuilder::~VersionBuilder() { delete rep_; } void VersionBuilder::CheckConsistency(VersionStorageInfo* vstorage) { rep_->CheckConsistency(vstorage); } void VersionBuilder::CheckConsistencyForDeletes(VersionEdit* edit, uint64_t number, int level) { rep_->CheckConsistencyForDeletes(edit, number, level); } void VersionBuilder::Apply(VersionEdit* edit) { rep_->Apply(edit); } void VersionBuilder::SaveTo(VersionStorageInfo* vstorage) { rep_->SaveTo(vstorage); } void VersionBuilder::LoadTableHandlers(InternalStats* internal_stats, int max_threads) { rep_->LoadTableHandlers(internal_stats, max_threads); } void VersionBuilder::MaybeAddFile(VersionStorageInfo* vstorage, int level, FileMetaData* f) { rep_->MaybeAddFile(vstorage, level, f); } } // namespace rocksdb rocksdb-4.5.1/db/version_builder.h000066400000000000000000000031171267530634000171110ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // #pragma once #include "rocksdb/env.h" namespace rocksdb { class TableCache; class VersionStorageInfo; class VersionEdit; struct FileMetaData; class InternalStats; // A helper class so we can efficiently apply a whole sequence // of edits to a particular state without creating intermediate // Versions that contain full copies of the intermediate state. class VersionBuilder { public: VersionBuilder(const EnvOptions& env_options, TableCache* table_cache, VersionStorageInfo* base_vstorage, Logger* info_log = nullptr); ~VersionBuilder(); void CheckConsistency(VersionStorageInfo* vstorage); void CheckConsistencyForDeletes(VersionEdit* edit, uint64_t number, int level); void Apply(VersionEdit* edit); void SaveTo(VersionStorageInfo* vstorage); void LoadTableHandlers(InternalStats* internal_stats, int max_threads = 1); void MaybeAddFile(VersionStorageInfo* vstorage, int level, FileMetaData* f); private: class Rep; Rep* rep_; }; extern bool NewestFirstBySeqNo(FileMetaData* a, FileMetaData* b); } // namespace rocksdb rocksdb-4.5.1/db/version_builder_test.cc000066400000000000000000000245571267530634000203210ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include "db/version_edit.h" #include "db/version_set.h" #include "util/logging.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class VersionBuilderTest : public testing::Test { public: const Comparator* ucmp_; InternalKeyComparator icmp_; Options options_; ImmutableCFOptions ioptions_; MutableCFOptions mutable_cf_options_; VersionStorageInfo vstorage_; uint32_t file_num_; CompactionOptionsFIFO fifo_options_; std::vector size_being_compacted_; VersionBuilderTest() : ucmp_(BytewiseComparator()), icmp_(ucmp_), ioptions_(options_), mutable_cf_options_(options_, ioptions_), vstorage_(&icmp_, ucmp_, options_.num_levels, kCompactionStyleLevel, nullptr), file_num_(1) { mutable_cf_options_.RefreshDerivedOptions(ioptions_); size_being_compacted_.resize(options_.num_levels); } ~VersionBuilderTest() { for (int i = 0; i < vstorage_.num_levels(); i++) { for (auto* f : vstorage_.LevelFiles(i)) { if (--f->refs == 0) { delete f; } } } } InternalKey GetInternalKey(const char* ukey, SequenceNumber smallest_seq = 100) { return InternalKey(ukey, smallest_seq, kTypeValue); } void Add(int level, uint32_t file_number, const char* smallest, const char* largest, uint64_t file_size = 0, uint32_t path_id = 0, SequenceNumber smallest_seq = 100, SequenceNumber largest_seq = 100, uint64_t num_entries = 0, uint64_t num_deletions = 0, bool sampled = false, SequenceNumber smallest_seqno = 0, SequenceNumber largest_seqno = 0) { assert(level < vstorage_.num_levels()); FileMetaData* f = new FileMetaData; f->fd = FileDescriptor(file_number, path_id, file_size); f->smallest = GetInternalKey(smallest, smallest_seq); f->largest = GetInternalKey(largest, largest_seq); f->smallest_seqno = smallest_seqno; f->largest_seqno = largest_seqno; f->compensated_file_size = file_size; f->refs = 0; f->num_entries = num_entries; f->num_deletions = num_deletions; vstorage_.AddFile(level, f); if (sampled) { f->init_stats_from_file = true; vstorage_.UpdateAccumulatedStats(f); } } void UpdateVersionStorageInfo() { vstorage_.UpdateFilesByCompactionPri(mutable_cf_options_); vstorage_.UpdateNumNonEmptyLevels(); vstorage_.GenerateFileIndexer(); vstorage_.GenerateLevelFilesBrief(); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); vstorage_.GenerateLevel0NonOverlapping(); vstorage_.SetFinalized(); } }; void UnrefFilesInVersion(VersionStorageInfo* new_vstorage) { for (int i = 0; i < new_vstorage->num_levels(); i++) { for (auto* f : new_vstorage->LevelFiles(i)) { if (--f->refs == 0) { delete f; } } } } TEST_F(VersionBuilderTest, ApplyAndSaveTo) { Add(0, 1U, "150", "200", 100U); Add(1, 66U, "150", "200", 100U); Add(1, 88U, "201", "300", 100U); Add(2, 6U, "150", "179", 100U); Add(2, 7U, "180", "220", 100U); Add(2, 8U, "221", "300", 100U); Add(3, 26U, "150", "170", 100U); Add(3, 27U, "171", "179", 100U); Add(3, 28U, "191", "220", 100U); Add(3, 29U, "221", "300", 100U); UpdateVersionStorageInfo(); VersionEdit version_edit; version_edit.AddFile(2, 666, 0, 100U, GetInternalKey("301"), GetInternalKey("350"), 200, 200, false); version_edit.DeleteFile(3, 27U); EnvOptions env_options; VersionBuilder version_builder(env_options, nullptr, &vstorage_); VersionStorageInfo new_vstorage(&icmp_, ucmp_, options_.num_levels, kCompactionStyleLevel, nullptr); version_builder.Apply(&version_edit); version_builder.SaveTo(&new_vstorage); ASSERT_EQ(400U, new_vstorage.NumLevelBytes(2)); ASSERT_EQ(300U, new_vstorage.NumLevelBytes(3)); UnrefFilesInVersion(&new_vstorage); } TEST_F(VersionBuilderTest, ApplyAndSaveToDynamic) { ioptions_.level_compaction_dynamic_level_bytes = true; Add(0, 1U, "150", "200", 100U, 0, 200U, 200U, 0, 0, false, 200U, 200U); Add(0, 88U, "201", "300", 100U, 0, 100U, 100U, 0, 0, false, 100U, 100U); Add(4, 6U, "150", "179", 100U); Add(4, 7U, "180", "220", 100U); Add(4, 8U, "221", "300", 100U); Add(5, 26U, "150", "170", 100U); Add(5, 27U, "171", "179", 100U); UpdateVersionStorageInfo(); VersionEdit version_edit; version_edit.AddFile(3, 666, 0, 100U, GetInternalKey("301"), GetInternalKey("350"), 200, 200, false); version_edit.DeleteFile(0, 1U); version_edit.DeleteFile(0, 88U); EnvOptions env_options; VersionBuilder version_builder(env_options, nullptr, &vstorage_); VersionStorageInfo new_vstorage(&icmp_, ucmp_, options_.num_levels, kCompactionStyleLevel, nullptr); version_builder.Apply(&version_edit); version_builder.SaveTo(&new_vstorage); ASSERT_EQ(0U, new_vstorage.NumLevelBytes(0)); ASSERT_EQ(100U, new_vstorage.NumLevelBytes(3)); ASSERT_EQ(300U, new_vstorage.NumLevelBytes(4)); ASSERT_EQ(200U, new_vstorage.NumLevelBytes(5)); UnrefFilesInVersion(&new_vstorage); } TEST_F(VersionBuilderTest, ApplyAndSaveToDynamic2) { ioptions_.level_compaction_dynamic_level_bytes = true; Add(0, 1U, "150", "200", 100U, 0, 200U, 200U, 0, 0, false, 200U, 200U); Add(0, 88U, "201", "300", 100U, 0, 100U, 100U, 0, 0, false, 100U, 100U); Add(4, 6U, "150", "179", 100U); Add(4, 7U, "180", "220", 100U); Add(4, 8U, "221", "300", 100U); Add(5, 26U, "150", "170", 100U); Add(5, 27U, "171", "179", 100U); UpdateVersionStorageInfo(); VersionEdit version_edit; version_edit.AddFile(4, 666, 0, 100U, GetInternalKey("301"), GetInternalKey("350"), 200, 200, false); version_edit.DeleteFile(0, 1U); version_edit.DeleteFile(0, 88U); version_edit.DeleteFile(4, 6U); version_edit.DeleteFile(4, 7U); version_edit.DeleteFile(4, 8U); EnvOptions env_options; VersionBuilder version_builder(env_options, nullptr, &vstorage_); VersionStorageInfo new_vstorage(&icmp_, ucmp_, options_.num_levels, kCompactionStyleLevel, nullptr); version_builder.Apply(&version_edit); version_builder.SaveTo(&new_vstorage); ASSERT_EQ(0U, new_vstorage.NumLevelBytes(0)); ASSERT_EQ(100U, new_vstorage.NumLevelBytes(4)); ASSERT_EQ(200U, new_vstorage.NumLevelBytes(5)); UnrefFilesInVersion(&new_vstorage); } TEST_F(VersionBuilderTest, ApplyMultipleAndSaveTo) { UpdateVersionStorageInfo(); VersionEdit version_edit; version_edit.AddFile(2, 666, 0, 100U, GetInternalKey("301"), GetInternalKey("350"), 200, 200, false); version_edit.AddFile(2, 676, 0, 100U, GetInternalKey("401"), GetInternalKey("450"), 200, 200, false); version_edit.AddFile(2, 636, 0, 100U, GetInternalKey("601"), GetInternalKey("650"), 200, 200, false); version_edit.AddFile(2, 616, 0, 100U, GetInternalKey("501"), GetInternalKey("550"), 200, 200, false); version_edit.AddFile(2, 606, 0, 100U, GetInternalKey("701"), GetInternalKey("750"), 200, 200, false); EnvOptions env_options; VersionBuilder version_builder(env_options, nullptr, &vstorage_); VersionStorageInfo new_vstorage(&icmp_, ucmp_, options_.num_levels, kCompactionStyleLevel, nullptr); version_builder.Apply(&version_edit); version_builder.SaveTo(&new_vstorage); ASSERT_EQ(500U, new_vstorage.NumLevelBytes(2)); UnrefFilesInVersion(&new_vstorage); } TEST_F(VersionBuilderTest, ApplyDeleteAndSaveTo) { UpdateVersionStorageInfo(); EnvOptions env_options; VersionBuilder version_builder(env_options, nullptr, &vstorage_); VersionStorageInfo new_vstorage(&icmp_, ucmp_, options_.num_levels, kCompactionStyleLevel, nullptr); VersionEdit version_edit; version_edit.AddFile(2, 666, 0, 100U, GetInternalKey("301"), GetInternalKey("350"), 200, 200, false); version_edit.AddFile(2, 676, 0, 100U, GetInternalKey("401"), GetInternalKey("450"), 200, 200, false); version_edit.AddFile(2, 636, 0, 100U, GetInternalKey("601"), GetInternalKey("650"), 200, 200, false); version_edit.AddFile(2, 616, 0, 100U, GetInternalKey("501"), GetInternalKey("550"), 200, 200, false); version_edit.AddFile(2, 606, 0, 100U, GetInternalKey("701"), GetInternalKey("750"), 200, 200, false); version_builder.Apply(&version_edit); VersionEdit version_edit2; version_edit.AddFile(2, 808, 0, 100U, GetInternalKey("901"), GetInternalKey("950"), 200, 200, false); version_edit2.DeleteFile(2, 616); version_edit2.DeleteFile(2, 636); version_edit.AddFile(2, 806, 0, 100U, GetInternalKey("801"), GetInternalKey("850"), 200, 200, false); version_builder.Apply(&version_edit2); version_builder.SaveTo(&new_vstorage); ASSERT_EQ(300U, new_vstorage.NumLevelBytes(2)); UnrefFilesInVersion(&new_vstorage); } TEST_F(VersionBuilderTest, EstimatedActiveKeys) { const uint32_t kTotalSamples = 20; const uint32_t kNumLevels = 5; const uint32_t kFilesPerLevel = 8; const uint32_t kNumFiles = kNumLevels * kFilesPerLevel; const uint32_t kEntriesPerFile = 1000; const uint32_t kDeletionsPerFile = 100; for (uint32_t i = 0; i < kNumFiles; ++i) { Add(static_cast(i / kFilesPerLevel), i + 1, ToString((i + 100) * 1000).c_str(), ToString((i + 100) * 1000 + 999).c_str(), 100U, 0, 100, 100, kEntriesPerFile, kDeletionsPerFile, (i < kTotalSamples)); } // minus 2X for the number of deletion entries because: // 1x for deletion entry does not count as a data entry. // 1x for each deletion entry will actually remove one data entry. ASSERT_EQ(vstorage_.GetEstimatedActiveKeys(), (kEntriesPerFile - 2 * kDeletionsPerFile) * kNumFiles); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/version_edit.cc000066400000000000000000000411231267530634000165450ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/version_edit.h" #include "db/version_set.h" #include "util/coding.h" #include "util/event_logger.h" #include "util/sync_point.h" #include "rocksdb/slice.h" namespace rocksdb { // Tag numbers for serialized VersionEdit. These numbers are written to // disk and should not be changed. enum Tag { kComparator = 1, kLogNumber = 2, kNextFileNumber = 3, kLastSequence = 4, kCompactPointer = 5, kDeletedFile = 6, kNewFile = 7, // 8 was used for large value refs kPrevLogNumber = 9, // these are new formats divergent from open source leveldb kNewFile2 = 100, kNewFile3 = 102, kNewFile4 = 103, // 4th (the latest) format version of adding files kColumnFamily = 200, // specify column family for version edit kColumnFamilyAdd = 201, kColumnFamilyDrop = 202, kMaxColumnFamily = 203, }; enum CustomTag { kTerminate = 1, // The end of customized fields kNeedCompaction = 2, kPathId = 65, }; // If this bit for the custom tag is set, opening DB should fail if // we don't know this field. uint32_t kCustomTagNonSafeIgnoreMask = 1 << 6; uint64_t PackFileNumberAndPathId(uint64_t number, uint64_t path_id) { assert(number <= kFileNumberMask); return number | (path_id * (kFileNumberMask + 1)); } void VersionEdit::Clear() { comparator_.clear(); max_level_ = 0; log_number_ = 0; prev_log_number_ = 0; last_sequence_ = 0; next_file_number_ = 0; max_column_family_ = 0; has_comparator_ = false; has_log_number_ = false; has_prev_log_number_ = false; has_next_file_number_ = false; has_last_sequence_ = false; has_max_column_family_ = false; deleted_files_.clear(); new_files_.clear(); column_family_ = 0; is_column_family_add_ = 0; is_column_family_drop_ = 0; column_family_name_.clear(); } bool VersionEdit::EncodeTo(std::string* dst) const { if (has_comparator_) { PutVarint32(dst, kComparator); PutLengthPrefixedSlice(dst, comparator_); } if (has_log_number_) { PutVarint32(dst, kLogNumber); PutVarint64(dst, log_number_); } if (has_prev_log_number_) { PutVarint32(dst, kPrevLogNumber); PutVarint64(dst, prev_log_number_); } if (has_next_file_number_) { PutVarint32(dst, kNextFileNumber); PutVarint64(dst, next_file_number_); } if (has_last_sequence_) { PutVarint32(dst, kLastSequence); PutVarint64(dst, last_sequence_); } if (has_max_column_family_) { PutVarint32(dst, kMaxColumnFamily); PutVarint32(dst, max_column_family_); } for (const auto& deleted : deleted_files_) { PutVarint32(dst, kDeletedFile); PutVarint32(dst, deleted.first /* level */); PutVarint64(dst, deleted.second /* file number */); } for (size_t i = 0; i < new_files_.size(); i++) { const FileMetaData& f = new_files_[i].second; if (!f.smallest.Valid() || !f.largest.Valid()) { return false; } bool has_customized_fields = false; if (f.marked_for_compaction) { PutVarint32(dst, kNewFile4); has_customized_fields = true; } else if (f.fd.GetPathId() == 0) { // Use older format to make sure user can roll back the build if they // don't config multiple DB paths. PutVarint32(dst, kNewFile2); } else { PutVarint32(dst, kNewFile3); } PutVarint32(dst, new_files_[i].first); // level PutVarint64(dst, f.fd.GetNumber()); if (f.fd.GetPathId() != 0 && !has_customized_fields) { // kNewFile3 PutVarint32(dst, f.fd.GetPathId()); } PutVarint64(dst, f.fd.GetFileSize()); PutLengthPrefixedSlice(dst, f.smallest.Encode()); PutLengthPrefixedSlice(dst, f.largest.Encode()); PutVarint64(dst, f.smallest_seqno); PutVarint64(dst, f.largest_seqno); if (has_customized_fields) { // Customized fields' format: // +-----------------------------+ // | 1st field's tag (varint32) | // +-----------------------------+ // | 1st field's size (varint32) | // +-----------------------------+ // | bytes for 1st field | // | (based on size decoded) | // +-----------------------------+ // | | // | ...... | // | | // +-----------------------------+ // | last field's size (varint32)| // +-----------------------------+ // | bytes for last field | // | (based on size decoded) | // +-----------------------------+ // | terminating tag (varint32) | // +-----------------------------+ // // Customized encoding for fields: // tag kPathId: 1 byte as path_id // tag kNeedCompaction: // now only can take one char value 1 indicating need-compaction // if (f.fd.GetPathId() != 0) { PutVarint32(dst, CustomTag::kPathId); char p = static_cast(f.fd.GetPathId()); PutLengthPrefixedSlice(dst, Slice(&p, 1)); } if (f.marked_for_compaction) { PutVarint32(dst, CustomTag::kNeedCompaction); char p = static_cast(1); PutLengthPrefixedSlice(dst, Slice(&p, 1)); } TEST_SYNC_POINT_CALLBACK("VersionEdit::EncodeTo:NewFile4:CustomizeFields", dst); PutVarint32(dst, CustomTag::kTerminate); } } // 0 is default and does not need to be explicitly written if (column_family_ != 0) { PutVarint32(dst, kColumnFamily); PutVarint32(dst, column_family_); } if (is_column_family_add_) { PutVarint32(dst, kColumnFamilyAdd); PutLengthPrefixedSlice(dst, Slice(column_family_name_)); } if (is_column_family_drop_) { PutVarint32(dst, kColumnFamilyDrop); } return true; } static bool GetInternalKey(Slice* input, InternalKey* dst) { Slice str; if (GetLengthPrefixedSlice(input, &str)) { dst->DecodeFrom(str); return dst->Valid(); } else { return false; } } bool VersionEdit::GetLevel(Slice* input, int* level, const char** msg) { uint32_t v; if (GetVarint32(input, &v)) { *level = v; if (max_level_ < *level) { max_level_ = *level; } return true; } else { return false; } } const char* VersionEdit::DecodeNewFile4From(Slice* input) { const char* msg = nullptr; int level; FileMetaData f; uint64_t number; uint32_t path_id = 0; uint64_t file_size; if (GetLevel(input, &level, &msg) && GetVarint64(input, &number) && GetVarint64(input, &file_size) && GetInternalKey(input, &f.smallest) && GetInternalKey(input, &f.largest) && GetVarint64(input, &f.smallest_seqno) && GetVarint64(input, &f.largest_seqno)) { // See comments in VersionEdit::EncodeTo() for format of customized fields while (true) { uint32_t custom_tag; Slice field; if (!GetVarint32(input, &custom_tag)) { return "new-file4 custom field"; } if (custom_tag == kTerminate) { break; } if (!GetLengthPrefixedSlice(input, &field)) { return "new-file4 custom field lenth prefixed slice error"; } switch (custom_tag) { case kPathId: if (field.size() != 1) { return "path_id field wrong size"; } path_id = field[0]; if (path_id > 3) { return "path_id wrong vaue"; } break; case kNeedCompaction: if (field.size() != 1) { return "need_compaction field wrong size"; } f.marked_for_compaction = (field[0] == 1); break; default: if ((custom_tag & kCustomTagNonSafeIgnoreMask) != 0) { // Should not proceed if cannot understand it return "new-file4 custom field not supported"; } break; } } } else { return "new-file4 entry"; } f.fd = FileDescriptor(number, path_id, file_size); new_files_.push_back(std::make_pair(level, f)); return nullptr; } Status VersionEdit::DecodeFrom(const Slice& src) { Clear(); Slice input = src; const char* msg = nullptr; uint32_t tag; // Temporary storage for parsing int level; FileMetaData f; Slice str; InternalKey key; while (msg == nullptr && GetVarint32(&input, &tag)) { switch (tag) { case kComparator: if (GetLengthPrefixedSlice(&input, &str)) { comparator_ = str.ToString(); has_comparator_ = true; } else { msg = "comparator name"; } break; case kLogNumber: if (GetVarint64(&input, &log_number_)) { has_log_number_ = true; } else { msg = "log number"; } break; case kPrevLogNumber: if (GetVarint64(&input, &prev_log_number_)) { has_prev_log_number_ = true; } else { msg = "previous log number"; } break; case kNextFileNumber: if (GetVarint64(&input, &next_file_number_)) { has_next_file_number_ = true; } else { msg = "next file number"; } break; case kLastSequence: if (GetVarint64(&input, &last_sequence_)) { has_last_sequence_ = true; } else { msg = "last sequence number"; } break; case kMaxColumnFamily: if (GetVarint32(&input, &max_column_family_)) { has_max_column_family_ = true; } else { msg = "max column family"; } break; case kCompactPointer: if (GetLevel(&input, &level, &msg) && GetInternalKey(&input, &key)) { // we don't use compact pointers anymore, // but we should not fail if they are still // in manifest } else { if (!msg) { msg = "compaction pointer"; } } break; case kDeletedFile: { uint64_t number; if (GetLevel(&input, &level, &msg) && GetVarint64(&input, &number)) { deleted_files_.insert(std::make_pair(level, number)); } else { if (!msg) { msg = "deleted file"; } } break; } case kNewFile: { uint64_t number; uint64_t file_size; if (GetLevel(&input, &level, &msg) && GetVarint64(&input, &number) && GetVarint64(&input, &file_size) && GetInternalKey(&input, &f.smallest) && GetInternalKey(&input, &f.largest)) { f.fd = FileDescriptor(number, 0, file_size); new_files_.push_back(std::make_pair(level, f)); } else { if (!msg) { msg = "new-file entry"; } } break; } case kNewFile2: { uint64_t number; uint64_t file_size; if (GetLevel(&input, &level, &msg) && GetVarint64(&input, &number) && GetVarint64(&input, &file_size) && GetInternalKey(&input, &f.smallest) && GetInternalKey(&input, &f.largest) && GetVarint64(&input, &f.smallest_seqno) && GetVarint64(&input, &f.largest_seqno)) { f.fd = FileDescriptor(number, 0, file_size); new_files_.push_back(std::make_pair(level, f)); } else { if (!msg) { msg = "new-file2 entry"; } } break; } case kNewFile3: { uint64_t number; uint32_t path_id; uint64_t file_size; if (GetLevel(&input, &level, &msg) && GetVarint64(&input, &number) && GetVarint32(&input, &path_id) && GetVarint64(&input, &file_size) && GetInternalKey(&input, &f.smallest) && GetInternalKey(&input, &f.largest) && GetVarint64(&input, &f.smallest_seqno) && GetVarint64(&input, &f.largest_seqno)) { f.fd = FileDescriptor(number, path_id, file_size); new_files_.push_back(std::make_pair(level, f)); } else { if (!msg) { msg = "new-file3 entry"; } } break; } case kNewFile4: { msg = DecodeNewFile4From(&input); break; } case kColumnFamily: if (!GetVarint32(&input, &column_family_)) { if (!msg) { msg = "set column family id"; } } break; case kColumnFamilyAdd: if (GetLengthPrefixedSlice(&input, &str)) { is_column_family_add_ = true; column_family_name_ = str.ToString(); } else { if (!msg) { msg = "column family add"; } } break; case kColumnFamilyDrop: is_column_family_drop_ = true; break; default: msg = "unknown tag"; break; } } if (msg == nullptr && !input.empty()) { msg = "invalid tag"; } Status result; if (msg != nullptr) { result = Status::Corruption("VersionEdit", msg); } return result; } std::string VersionEdit::DebugString(bool hex_key) const { std::string r; r.append("VersionEdit {"); if (has_comparator_) { r.append("\n Comparator: "); r.append(comparator_); } if (has_log_number_) { r.append("\n LogNumber: "); AppendNumberTo(&r, log_number_); } if (has_prev_log_number_) { r.append("\n PrevLogNumber: "); AppendNumberTo(&r, prev_log_number_); } if (has_next_file_number_) { r.append("\n NextFileNumber: "); AppendNumberTo(&r, next_file_number_); } if (has_last_sequence_) { r.append("\n LastSeq: "); AppendNumberTo(&r, last_sequence_); } for (DeletedFileSet::const_iterator iter = deleted_files_.begin(); iter != deleted_files_.end(); ++iter) { r.append("\n DeleteFile: "); AppendNumberTo(&r, iter->first); r.append(" "); AppendNumberTo(&r, iter->second); } for (size_t i = 0; i < new_files_.size(); i++) { const FileMetaData& f = new_files_[i].second; r.append("\n AddFile: "); AppendNumberTo(&r, new_files_[i].first); r.append(" "); AppendNumberTo(&r, f.fd.GetNumber()); r.append(" "); AppendNumberTo(&r, f.fd.GetFileSize()); r.append(" "); r.append(f.smallest.DebugString(hex_key)); r.append(" .. "); r.append(f.largest.DebugString(hex_key)); } r.append("\n ColumnFamily: "); AppendNumberTo(&r, column_family_); if (is_column_family_add_) { r.append("\n ColumnFamilyAdd: "); r.append(column_family_name_); } if (is_column_family_drop_) { r.append("\n ColumnFamilyDrop"); } if (has_max_column_family_) { r.append("\n MaxColumnFamily: "); AppendNumberTo(&r, max_column_family_); } r.append("\n}\n"); return r; } std::string VersionEdit::DebugJSON(int edit_num, bool hex_key) const { JSONWriter jw; jw << "EditNumber" << edit_num; if (has_comparator_) { jw << "Comparator" << comparator_; } if (has_log_number_) { jw << "LogNumber" << log_number_; } if (has_prev_log_number_) { jw << "PrevLogNumber" << prev_log_number_; } if (has_next_file_number_) { jw << "NextFileNumber" << next_file_number_; } if (has_last_sequence_) { jw << "LastSeq" << last_sequence_; } if (!deleted_files_.empty()) { jw << "DeletedFiles"; jw.StartArray(); for (DeletedFileSet::const_iterator iter = deleted_files_.begin(); iter != deleted_files_.end(); ++iter) { jw.StartArrayedObject(); jw << "Level" << iter->first; jw << "FileNumber" << iter->second; jw.EndArrayedObject(); } jw.EndArray(); } if (!new_files_.empty()) { jw << "AddedFiles"; jw.StartArray(); for (size_t i = 0; i < new_files_.size(); i++) { jw.StartArrayedObject(); jw << "Level" << new_files_[i].first; const FileMetaData& f = new_files_[i].second; jw << "FileNumber" << f.fd.GetNumber(); jw << "FileSize" << f.fd.GetFileSize(); jw << "SmallestIKey" << f.smallest.DebugString(hex_key); jw << "LargestIKey" << f.largest.DebugString(hex_key); jw.EndArrayedObject(); } jw.EndArray(); } jw << "ColumnFamily" << column_family_; if (is_column_family_add_) { jw << "ColumnFamilyAdd" << column_family_name_; } if (is_column_family_drop_) { jw << "ColumnFamilyDrop" << column_family_name_; } if (has_max_column_family_) { jw << "MaxColumnFamily" << max_column_family_; } jw.EndObject(); return jw.Get(); } } // namespace rocksdb rocksdb-4.5.1/db/version_edit.h000066400000000000000000000222151267530634000164100ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include "rocksdb/cache.h" #include "db/dbformat.h" #include "util/arena.h" #include "util/autovector.h" namespace rocksdb { class VersionSet; const uint64_t kFileNumberMask = 0x3FFFFFFFFFFFFFFF; extern uint64_t PackFileNumberAndPathId(uint64_t number, uint64_t path_id); // A copyable structure contains information needed to read data from an SST // file. It can contains a pointer to a table reader opened for the file, or // file number and size, which can be used to create a new table reader for it. // The behavior is undefined when a copied of the structure is used when the // file is not in any live version any more. struct FileDescriptor { // Table reader in table_reader_handle TableReader* table_reader; uint64_t packed_number_and_path_id; uint64_t file_size; // File size in bytes FileDescriptor() : FileDescriptor(0, 0, 0) {} FileDescriptor(uint64_t number, uint32_t path_id, uint64_t _file_size) : table_reader(nullptr), packed_number_and_path_id(PackFileNumberAndPathId(number, path_id)), file_size(_file_size) {} FileDescriptor& operator=(const FileDescriptor& fd) { table_reader = fd.table_reader; packed_number_and_path_id = fd.packed_number_and_path_id; file_size = fd.file_size; return *this; } uint64_t GetNumber() const { return packed_number_and_path_id & kFileNumberMask; } uint32_t GetPathId() const { return static_cast( packed_number_and_path_id / (kFileNumberMask + 1)); } uint64_t GetFileSize() const { return file_size; } }; struct FileMetaData { int refs; FileDescriptor fd; InternalKey smallest; // Smallest internal key served by table InternalKey largest; // Largest internal key served by table bool being_compacted; // Is this file undergoing compaction? SequenceNumber smallest_seqno; // The smallest seqno in this file SequenceNumber largest_seqno; // The largest seqno in this file // Needs to be disposed when refs becomes 0. Cache::Handle* table_reader_handle; // Stats for compensating deletion entries during compaction // File size compensated by deletion entry. // This is updated in Version::UpdateAccumulatedStats() first time when the // file is created or loaded. After it is updated (!= 0), it is immutable. uint64_t compensated_file_size; // These values can mutate, but they can only be read or written from // single-threaded LogAndApply thread uint64_t num_entries; // the number of entries. uint64_t num_deletions; // the number of deletion entries. uint64_t raw_key_size; // total uncompressed key size. uint64_t raw_value_size; // total uncompressed value size. bool init_stats_from_file; // true if the data-entry stats of this file // has initialized from file. bool marked_for_compaction; // True if client asked us nicely to compact this // file. FileMetaData() : refs(0), being_compacted(false), smallest_seqno(kMaxSequenceNumber), largest_seqno(0), table_reader_handle(nullptr), compensated_file_size(0), num_entries(0), num_deletions(0), raw_key_size(0), raw_value_size(0), init_stats_from_file(false), marked_for_compaction(false) {} // REQUIRED: Keys must be given to the function in sorted order (it expects // the last key to be the largest). void UpdateBoundaries(const Slice& key, SequenceNumber seqno) { if (smallest.size() == 0) { smallest.DecodeFrom(key); } largest.DecodeFrom(key); smallest_seqno = std::min(smallest_seqno, seqno); largest_seqno = std::max(largest_seqno, seqno); } }; // A compressed copy of file meta data that just contain // smallest and largest key's slice struct FdWithKeyRange { FileDescriptor fd; Slice smallest_key; // slice that contain smallest key Slice largest_key; // slice that contain largest key FdWithKeyRange() : fd(), smallest_key(), largest_key() { } FdWithKeyRange(FileDescriptor _fd, Slice _smallest_key, Slice _largest_key) : fd(_fd), smallest_key(_smallest_key), largest_key(_largest_key) {} }; // Data structure to store an array of FdWithKeyRange in one level // Actual data is guaranteed to be stored closely struct LevelFilesBrief { size_t num_files; FdWithKeyRange* files; LevelFilesBrief() { num_files = 0; files = nullptr; } }; class VersionEdit { public: VersionEdit() { Clear(); } ~VersionEdit() { } void Clear(); void SetComparatorName(const Slice& name) { has_comparator_ = true; comparator_ = name.ToString(); } void SetLogNumber(uint64_t num) { has_log_number_ = true; log_number_ = num; } void SetPrevLogNumber(uint64_t num) { has_prev_log_number_ = true; prev_log_number_ = num; } void SetNextFile(uint64_t num) { has_next_file_number_ = true; next_file_number_ = num; } void SetLastSequence(SequenceNumber seq) { has_last_sequence_ = true; last_sequence_ = seq; } void SetMaxColumnFamily(uint32_t max_column_family) { has_max_column_family_ = true; max_column_family_ = max_column_family; } // Add the specified file at the specified number. // REQUIRES: This version has not been saved (see VersionSet::SaveTo) // REQUIRES: "smallest" and "largest" are smallest and largest keys in file void AddFile(int level, uint64_t file, uint32_t file_path_id, uint64_t file_size, const InternalKey& smallest, const InternalKey& largest, const SequenceNumber& smallest_seqno, const SequenceNumber& largest_seqno, bool marked_for_compaction) { assert(smallest_seqno <= largest_seqno); FileMetaData f; f.fd = FileDescriptor(file, file_path_id, file_size); f.smallest = smallest; f.largest = largest; f.smallest_seqno = smallest_seqno; f.largest_seqno = largest_seqno; f.marked_for_compaction = marked_for_compaction; new_files_.emplace_back(level, f); } void AddFile(int level, const FileMetaData& f) { assert(f.smallest_seqno <= f.largest_seqno); new_files_.emplace_back(level, f); } // Delete the specified "file" from the specified "level". void DeleteFile(int level, uint64_t file) { deleted_files_.insert({level, file}); } // Number of edits size_t NumEntries() { return new_files_.size() + deleted_files_.size(); } bool IsColumnFamilyManipulation() { return is_column_family_add_ || is_column_family_drop_; } void SetColumnFamily(uint32_t column_family_id) { column_family_ = column_family_id; } // set column family ID by calling SetColumnFamily() void AddColumnFamily(const std::string& name) { assert(!is_column_family_drop_); assert(!is_column_family_add_); assert(NumEntries() == 0); is_column_family_add_ = true; column_family_name_ = name; } // set column family ID by calling SetColumnFamily() void DropColumnFamily() { assert(!is_column_family_drop_); assert(!is_column_family_add_); assert(NumEntries() == 0); is_column_family_drop_ = true; } // return true on success. bool EncodeTo(std::string* dst) const; Status DecodeFrom(const Slice& src); const char* DecodeNewFile4From(Slice* input); typedef std::set> DeletedFileSet; const DeletedFileSet& GetDeletedFiles() { return deleted_files_; } const std::vector>& GetNewFiles() { return new_files_; } std::string DebugString(bool hex_key = false) const; std::string DebugJSON(int edit_num, bool hex_key = false) const; private: friend class VersionSet; friend class Version; bool GetLevel(Slice* input, int* level, const char** msg); int max_level_; std::string comparator_; uint64_t log_number_; uint64_t prev_log_number_; uint64_t next_file_number_; uint32_t max_column_family_; SequenceNumber last_sequence_; bool has_comparator_; bool has_log_number_; bool has_prev_log_number_; bool has_next_file_number_; bool has_last_sequence_; bool has_max_column_family_; DeletedFileSet deleted_files_; std::vector> new_files_; // Each version edit record should have column_family_id set // If it's not set, it is default (0) uint32_t column_family_; // a version edit can be either column_family add or // column_family drop. If it's column family add, // it also includes column family name. bool is_column_family_drop_; bool is_column_family_add_; std::string column_family_name_; }; } // namespace rocksdb rocksdb-4.5.1/db/version_edit_test.cc000066400000000000000000000145161267530634000176120ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/version_edit.h" #include "util/sync_point.h" #include "util/testharness.h" namespace rocksdb { static void TestEncodeDecode(const VersionEdit& edit) { std::string encoded, encoded2; edit.EncodeTo(&encoded); VersionEdit parsed; Status s = parsed.DecodeFrom(encoded); ASSERT_TRUE(s.ok()) << s.ToString(); parsed.EncodeTo(&encoded2); ASSERT_EQ(encoded, encoded2); } class VersionEditTest : public testing::Test {}; TEST_F(VersionEditTest, EncodeDecode) { static const uint64_t kBig = 1ull << 50; static const uint32_t kBig32Bit = 1ull << 30; VersionEdit edit; for (int i = 0; i < 4; i++) { TestEncodeDecode(edit); edit.AddFile(3, kBig + 300 + i, kBig32Bit + 400 + i, 0, InternalKey("foo", kBig + 500 + i, kTypeValue), InternalKey("zoo", kBig + 600 + i, kTypeDeletion), kBig + 500 + i, kBig + 600 + i, false); edit.DeleteFile(4, kBig + 700 + i); } edit.SetComparatorName("foo"); edit.SetLogNumber(kBig + 100); edit.SetNextFile(kBig + 200); edit.SetLastSequence(kBig + 1000); TestEncodeDecode(edit); } TEST_F(VersionEditTest, EncodeDecodeNewFile4) { static const uint64_t kBig = 1ull << 50; VersionEdit edit; edit.AddFile(3, 300, 3, 100, InternalKey("foo", kBig + 500, kTypeValue), InternalKey("zoo", kBig + 600, kTypeDeletion), kBig + 500, kBig + 600, true); edit.AddFile(4, 301, 3, 100, InternalKey("foo", kBig + 501, kTypeValue), InternalKey("zoo", kBig + 601, kTypeDeletion), kBig + 501, kBig + 601, false); edit.AddFile(5, 302, 0, 100, InternalKey("foo", kBig + 502, kTypeValue), InternalKey("zoo", kBig + 602, kTypeDeletion), kBig + 502, kBig + 602, true); edit.DeleteFile(4, 700); edit.SetComparatorName("foo"); edit.SetLogNumber(kBig + 100); edit.SetNextFile(kBig + 200); edit.SetLastSequence(kBig + 1000); TestEncodeDecode(edit); std::string encoded, encoded2; edit.EncodeTo(&encoded); VersionEdit parsed; Status s = parsed.DecodeFrom(encoded); ASSERT_TRUE(s.ok()) << s.ToString(); auto& new_files = parsed.GetNewFiles(); ASSERT_TRUE(new_files[0].second.marked_for_compaction); ASSERT_TRUE(!new_files[1].second.marked_for_compaction); ASSERT_TRUE(new_files[2].second.marked_for_compaction); ASSERT_EQ(3, new_files[0].second.fd.GetPathId()); ASSERT_EQ(3, new_files[1].second.fd.GetPathId()); ASSERT_EQ(0, new_files[2].second.fd.GetPathId()); } TEST_F(VersionEditTest, ForwardCompatibleNewFile4) { static const uint64_t kBig = 1ull << 50; VersionEdit edit; edit.AddFile(3, 300, 3, 100, InternalKey("foo", kBig + 500, kTypeValue), InternalKey("zoo", kBig + 600, kTypeDeletion), kBig + 500, kBig + 600, true); edit.AddFile(4, 301, 3, 100, InternalKey("foo", kBig + 501, kTypeValue), InternalKey("zoo", kBig + 601, kTypeDeletion), kBig + 501, kBig + 601, false); edit.DeleteFile(4, 700); edit.SetComparatorName("foo"); edit.SetLogNumber(kBig + 100); edit.SetNextFile(kBig + 200); edit.SetLastSequence(kBig + 1000); std::string encoded; // Call back function to add extra customized builds. bool first = true; rocksdb::SyncPoint::GetInstance()->SetCallBack( "VersionEdit::EncodeTo:NewFile4:CustomizeFields", [&](void* arg) { std::string* str = reinterpret_cast(arg); PutVarint32(str, 33); const std::string str1 = "random_string"; PutLengthPrefixedSlice(str, str1); if (first) { first = false; PutVarint32(str, 22); const std::string str2 = "s"; PutLengthPrefixedSlice(str, str2); } }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); edit.EncodeTo(&encoded); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); VersionEdit parsed; Status s = parsed.DecodeFrom(encoded); ASSERT_TRUE(s.ok()) << s.ToString(); ASSERT_TRUE(!first); auto& new_files = parsed.GetNewFiles(); ASSERT_TRUE(new_files[0].second.marked_for_compaction); ASSERT_TRUE(!new_files[1].second.marked_for_compaction); ASSERT_EQ(3, new_files[0].second.fd.GetPathId()); ASSERT_EQ(3, new_files[1].second.fd.GetPathId()); ASSERT_EQ(1u, parsed.GetDeletedFiles().size()); } TEST_F(VersionEditTest, NewFile4NotSupportedField) { static const uint64_t kBig = 1ull << 50; VersionEdit edit; edit.AddFile(3, 300, 3, 100, InternalKey("foo", kBig + 500, kTypeValue), InternalKey("zoo", kBig + 600, kTypeDeletion), kBig + 500, kBig + 600, true); edit.SetComparatorName("foo"); edit.SetLogNumber(kBig + 100); edit.SetNextFile(kBig + 200); edit.SetLastSequence(kBig + 1000); std::string encoded; // Call back function to add extra customized builds. rocksdb::SyncPoint::GetInstance()->SetCallBack( "VersionEdit::EncodeTo:NewFile4:CustomizeFields", [&](void* arg) { std::string* str = reinterpret_cast(arg); const std::string str1 = "s"; PutLengthPrefixedSlice(str, str1); }); rocksdb::SyncPoint::GetInstance()->EnableProcessing(); edit.EncodeTo(&encoded); rocksdb::SyncPoint::GetInstance()->DisableProcessing(); VersionEdit parsed; Status s = parsed.DecodeFrom(encoded); ASSERT_NOK(s); } TEST_F(VersionEditTest, EncodeEmptyFile) { VersionEdit edit; edit.AddFile(0, 0, 0, 0, InternalKey(), InternalKey(), 0, 0, false); std::string buffer; ASSERT_TRUE(!edit.EncodeTo(&buffer)); } TEST_F(VersionEditTest, ColumnFamilyTest) { VersionEdit edit; edit.SetColumnFamily(2); edit.AddColumnFamily("column_family"); edit.SetMaxColumnFamily(5); TestEncodeDecode(edit); edit.Clear(); edit.SetColumnFamily(3); edit.DropColumnFamily(); TestEncodeDecode(edit); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/version_set.cc000066400000000000000000003606121267530634000164220ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/version_set.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include #include #include #include #include "db/filename.h" #include "db/internal_stats.h" #include "db/log_reader.h" #include "db/log_writer.h" #include "db/memtable.h" #include "db/merge_context.h" #include "db/table_cache.h" #include "db/compaction.h" #include "db/version_builder.h" #include "db/writebuffer.h" #include "rocksdb/env.h" #include "rocksdb/merge_operator.h" #include "table/internal_iterator.h" #include "table/table_reader.h" #include "table/merger.h" #include "table/two_level_iterator.h" #include "table/format.h" #include "table/plain_table_factory.h" #include "table/meta_blocks.h" #include "table/get_context.h" #include "util/coding.h" #include "util/file_reader_writer.h" #include "util/logging.h" #include "util/stop_watch.h" #include "util/sync_point.h" namespace rocksdb { namespace { // Find File in LevelFilesBrief data structure // Within an index range defined by left and right int FindFileInRange(const InternalKeyComparator& icmp, const LevelFilesBrief& file_level, const Slice& key, uint32_t left, uint32_t right) { while (left < right) { uint32_t mid = (left + right) / 2; const FdWithKeyRange& f = file_level.files[mid]; if (icmp.InternalKeyComparator::Compare(f.largest_key, key) < 0) { // Key at "mid.largest" is < "target". Therefore all // files at or before "mid" are uninteresting. left = mid + 1; } else { // Key at "mid.largest" is >= "target". Therefore all files // after "mid" are uninteresting. right = mid; } } return right; } // Class to help choose the next file to search for the particular key. // Searches and returns files level by level. // We can search level-by-level since entries never hop across // levels. Therefore we are guaranteed that if we find data // in a smaller level, later levels are irrelevant (unless we // are MergeInProgress). class FilePicker { public: FilePicker(std::vector* files, const Slice& user_key, const Slice& ikey, autovector* file_levels, unsigned int num_levels, FileIndexer* file_indexer, const Comparator* user_comparator, const InternalKeyComparator* internal_comparator) : num_levels_(num_levels), curr_level_(-1), hit_file_level_(-1), search_left_bound_(0), search_right_bound_(FileIndexer::kLevelMaxIndex), #ifndef NDEBUG files_(files), #endif level_files_brief_(file_levels), is_hit_file_last_in_level_(false), user_key_(user_key), ikey_(ikey), file_indexer_(file_indexer), user_comparator_(user_comparator), internal_comparator_(internal_comparator) { // Setup member variables to search first level. search_ended_ = !PrepareNextLevel(); if (!search_ended_) { // Prefetch Level 0 table data to avoid cache miss if possible. for (unsigned int i = 0; i < (*level_files_brief_)[0].num_files; ++i) { auto* r = (*level_files_brief_)[0].files[i].fd.table_reader; if (r) { r->Prepare(ikey); } } } } FdWithKeyRange* GetNextFile() { while (!search_ended_) { // Loops over different levels. while (curr_index_in_curr_level_ < curr_file_level_->num_files) { // Loops over all files in current level. FdWithKeyRange* f = &curr_file_level_->files[curr_index_in_curr_level_]; hit_file_level_ = curr_level_; is_hit_file_last_in_level_ = curr_index_in_curr_level_ == curr_file_level_->num_files - 1; int cmp_largest = -1; // Do key range filtering of files or/and fractional cascading if: // (1) not all the files are in level 0, or // (2) there are more than 3 Level 0 files // If there are only 3 or less level 0 files in the system, we skip // the key range filtering. In this case, more likely, the system is // highly tuned to minimize number of tables queried by each query, // so it is unlikely that key range filtering is more efficient than // querying the files. if (num_levels_ > 1 || curr_file_level_->num_files > 3) { // Check if key is within a file's range. If search left bound and // right bound point to the same find, we are sure key falls in // range. assert( curr_level_ == 0 || curr_index_in_curr_level_ == start_index_in_curr_level_ || user_comparator_->Compare(user_key_, ExtractUserKey(f->smallest_key)) <= 0); int cmp_smallest = user_comparator_->Compare(user_key_, ExtractUserKey(f->smallest_key)); if (cmp_smallest >= 0) { cmp_largest = user_comparator_->Compare(user_key_, ExtractUserKey(f->largest_key)); } // Setup file search bound for the next level based on the // comparison results if (curr_level_ > 0) { file_indexer_->GetNextLevelIndex(curr_level_, curr_index_in_curr_level_, cmp_smallest, cmp_largest, &search_left_bound_, &search_right_bound_); } // Key falls out of current file's range if (cmp_smallest < 0 || cmp_largest > 0) { if (curr_level_ == 0) { ++curr_index_in_curr_level_; continue; } else { // Search next level. break; } } } #ifndef NDEBUG // Sanity check to make sure that the files are correctly sorted if (prev_file_) { if (curr_level_ != 0) { int comp_sign = internal_comparator_->Compare( prev_file_->largest_key, f->smallest_key); assert(comp_sign < 0); } else { // level == 0, the current file cannot be newer than the previous // one. Use compressed data structure, has no attribute seqNo assert(curr_index_in_curr_level_ > 0); assert(!NewestFirstBySeqNo(files_[0][curr_index_in_curr_level_], files_[0][curr_index_in_curr_level_-1])); } } prev_file_ = f; #endif if (curr_level_ > 0 && cmp_largest < 0) { // No more files to search in this level. search_ended_ = !PrepareNextLevel(); } else { ++curr_index_in_curr_level_; } return f; } // Start searching next level. search_ended_ = !PrepareNextLevel(); } // Search ended. return nullptr; } // getter for current file level // for GET_HIT_L0, GET_HIT_L1 & GET_HIT_L2_AND_UP counts unsigned int GetHitFileLevel() { return hit_file_level_; } // Returns true if the most recent "hit file" (i.e., one returned by // GetNextFile()) is at the last index in its level. bool IsHitFileLastInLevel() { return is_hit_file_last_in_level_; } private: unsigned int num_levels_; unsigned int curr_level_; unsigned int hit_file_level_; int32_t search_left_bound_; int32_t search_right_bound_; #ifndef NDEBUG std::vector* files_; #endif autovector* level_files_brief_; bool search_ended_; bool is_hit_file_last_in_level_; LevelFilesBrief* curr_file_level_; unsigned int curr_index_in_curr_level_; unsigned int start_index_in_curr_level_; Slice user_key_; Slice ikey_; FileIndexer* file_indexer_; const Comparator* user_comparator_; const InternalKeyComparator* internal_comparator_; #ifndef NDEBUG FdWithKeyRange* prev_file_; #endif // Setup local variables to search next level. // Returns false if there are no more levels to search. bool PrepareNextLevel() { curr_level_++; while (curr_level_ < num_levels_) { curr_file_level_ = &(*level_files_brief_)[curr_level_]; if (curr_file_level_->num_files == 0) { // When current level is empty, the search bound generated from upper // level must be [0, -1] or [0, FileIndexer::kLevelMaxIndex] if it is // also empty. assert(search_left_bound_ == 0); assert(search_right_bound_ == -1 || search_right_bound_ == FileIndexer::kLevelMaxIndex); // Since current level is empty, it will need to search all files in // the next level search_left_bound_ = 0; search_right_bound_ = FileIndexer::kLevelMaxIndex; curr_level_++; continue; } // Some files may overlap each other. We find // all files that overlap user_key and process them in order from // newest to oldest. In the context of merge-operator, this can occur at // any level. Otherwise, it only occurs at Level-0 (since Put/Deletes // are always compacted into a single entry). int32_t start_index; if (curr_level_ == 0) { // On Level-0, we read through all files to check for overlap. start_index = 0; } else { // On Level-n (n>=1), files are sorted. Binary search to find the // earliest file whose largest key >= ikey. Search left bound and // right bound are used to narrow the range. if (search_left_bound_ == search_right_bound_) { start_index = search_left_bound_; } else if (search_left_bound_ < search_right_bound_) { if (search_right_bound_ == FileIndexer::kLevelMaxIndex) { search_right_bound_ = static_cast(curr_file_level_->num_files) - 1; } start_index = FindFileInRange(*internal_comparator_, *curr_file_level_, ikey_, static_cast(search_left_bound_), static_cast(search_right_bound_)); } else { // search_left_bound > search_right_bound, key does not exist in // this level. Since no comparison is done in this level, it will // need to search all files in the next level. search_left_bound_ = 0; search_right_bound_ = FileIndexer::kLevelMaxIndex; curr_level_++; continue; } } start_index_in_curr_level_ = start_index; curr_index_in_curr_level_ = start_index; #ifndef NDEBUG prev_file_ = nullptr; #endif return true; } // curr_level_ = num_levels_. So, no more levels to search. return false; } }; } // anonymous namespace VersionStorageInfo::~VersionStorageInfo() { delete[] files_; } Version::~Version() { assert(refs_ == 0); // Remove from linked list prev_->next_ = next_; next_->prev_ = prev_; // Drop references to files for (int level = 0; level < storage_info_.num_levels_; level++) { for (size_t i = 0; i < storage_info_.files_[level].size(); i++) { FileMetaData* f = storage_info_.files_[level][i]; assert(f->refs > 0); f->refs--; if (f->refs <= 0) { if (f->table_reader_handle) { cfd_->table_cache()->ReleaseHandle(f->table_reader_handle); f->table_reader_handle = nullptr; } vset_->obsolete_files_.push_back(f); } } } } int FindFile(const InternalKeyComparator& icmp, const LevelFilesBrief& file_level, const Slice& key) { return FindFileInRange(icmp, file_level, key, 0, static_cast(file_level.num_files)); } void DoGenerateLevelFilesBrief(LevelFilesBrief* file_level, const std::vector& files, Arena* arena) { assert(file_level); assert(arena); size_t num = files.size(); file_level->num_files = num; char* mem = arena->AllocateAligned(num * sizeof(FdWithKeyRange)); file_level->files = new (mem)FdWithKeyRange[num]; for (size_t i = 0; i < num; i++) { Slice smallest_key = files[i]->smallest.Encode(); Slice largest_key = files[i]->largest.Encode(); // Copy key slice to sequential memory size_t smallest_size = smallest_key.size(); size_t largest_size = largest_key.size(); mem = arena->AllocateAligned(smallest_size + largest_size); memcpy(mem, smallest_key.data(), smallest_size); memcpy(mem + smallest_size, largest_key.data(), largest_size); FdWithKeyRange& f = file_level->files[i]; f.fd = files[i]->fd; f.smallest_key = Slice(mem, smallest_size); f.largest_key = Slice(mem + smallest_size, largest_size); } } static bool AfterFile(const Comparator* ucmp, const Slice* user_key, const FdWithKeyRange* f) { // nullptr user_key occurs before all keys and is therefore never after *f return (user_key != nullptr && ucmp->Compare(*user_key, ExtractUserKey(f->largest_key)) > 0); } static bool BeforeFile(const Comparator* ucmp, const Slice* user_key, const FdWithKeyRange* f) { // nullptr user_key occurs after all keys and is therefore never before *f return (user_key != nullptr && ucmp->Compare(*user_key, ExtractUserKey(f->smallest_key)) < 0); } bool SomeFileOverlapsRange( const InternalKeyComparator& icmp, bool disjoint_sorted_files, const LevelFilesBrief& file_level, const Slice* smallest_user_key, const Slice* largest_user_key) { const Comparator* ucmp = icmp.user_comparator(); if (!disjoint_sorted_files) { // Need to check against all files for (size_t i = 0; i < file_level.num_files; i++) { const FdWithKeyRange* f = &(file_level.files[i]); if (AfterFile(ucmp, smallest_user_key, f) || BeforeFile(ucmp, largest_user_key, f)) { // No overlap } else { return true; // Overlap } } return false; } // Binary search over file list uint32_t index = 0; if (smallest_user_key != nullptr) { // Find the earliest possible internal key for smallest_user_key InternalKey small; small.SetMaxPossibleForUserKey(*smallest_user_key); index = FindFile(icmp, file_level, small.Encode()); } if (index >= file_level.num_files) { // beginning of range is after all files, so no overlap. return false; } return !BeforeFile(ucmp, largest_user_key, &file_level.files[index]); } namespace { // An internal iterator. For a given version/level pair, yields // information about the files in the level. For a given entry, key() // is the largest key that occurs in the file, and value() is an // 16-byte value containing the file number and file size, both // encoded using EncodeFixed64. class LevelFileNumIterator : public InternalIterator { public: LevelFileNumIterator(const InternalKeyComparator& icmp, const LevelFilesBrief* flevel) : icmp_(icmp), flevel_(flevel), index_(static_cast(flevel->num_files)), current_value_(0, 0, 0) { // Marks as invalid } virtual bool Valid() const override { return index_ < flevel_->num_files; } virtual void Seek(const Slice& target) override { index_ = FindFile(icmp_, *flevel_, target); } virtual void SeekToFirst() override { index_ = 0; } virtual void SeekToLast() override { index_ = (flevel_->num_files == 0) ? 0 : static_cast(flevel_->num_files) - 1; } virtual void Next() override { assert(Valid()); index_++; } virtual void Prev() override { assert(Valid()); if (index_ == 0) { index_ = static_cast(flevel_->num_files); // Marks as invalid } else { index_--; } } Slice key() const override { assert(Valid()); return flevel_->files[index_].largest_key; } Slice value() const override { assert(Valid()); auto file_meta = flevel_->files[index_]; current_value_ = file_meta.fd; return Slice(reinterpret_cast(¤t_value_), sizeof(FileDescriptor)); } virtual Status status() const override { return Status::OK(); } private: const InternalKeyComparator icmp_; const LevelFilesBrief* flevel_; uint32_t index_; mutable FileDescriptor current_value_; }; class LevelFileIteratorState : public TwoLevelIteratorState { public: // @param skip_filters Disables loading/accessing the filter block LevelFileIteratorState(TableCache* table_cache, const ReadOptions& read_options, const EnvOptions& env_options, const InternalKeyComparator& icomparator, HistogramImpl* file_read_hist, bool for_compaction, bool prefix_enabled, bool skip_filters) : TwoLevelIteratorState(prefix_enabled), table_cache_(table_cache), read_options_(read_options), env_options_(env_options), icomparator_(icomparator), file_read_hist_(file_read_hist), for_compaction_(for_compaction), skip_filters_(skip_filters) {} InternalIterator* NewSecondaryIterator(const Slice& meta_handle) override { if (meta_handle.size() != sizeof(FileDescriptor)) { return NewErrorInternalIterator( Status::Corruption("FileReader invoked with unexpected value")); } else { const FileDescriptor* fd = reinterpret_cast(meta_handle.data()); return table_cache_->NewIterator( read_options_, env_options_, icomparator_, *fd, nullptr /* don't need reference to table*/, file_read_hist_, for_compaction_, nullptr /* arena */, skip_filters_); } } bool PrefixMayMatch(const Slice& internal_key) override { return true; } private: TableCache* table_cache_; const ReadOptions read_options_; const EnvOptions& env_options_; const InternalKeyComparator& icomparator_; HistogramImpl* file_read_hist_; bool for_compaction_; bool skip_filters_; }; // A wrapper of version builder which references the current version in // constructor and unref it in the destructor. // Both of the constructor and destructor need to be called inside DB Mutex. class BaseReferencedVersionBuilder { public: explicit BaseReferencedVersionBuilder(ColumnFamilyData* cfd) : version_builder_(new VersionBuilder( cfd->current()->version_set()->env_options(), cfd->table_cache(), cfd->current()->storage_info(), cfd->ioptions()->info_log)), version_(cfd->current()) { version_->Ref(); } ~BaseReferencedVersionBuilder() { delete version_builder_; version_->Unref(); } VersionBuilder* version_builder() { return version_builder_; } private: VersionBuilder* version_builder_; Version* version_; }; } // anonymous namespace Status Version::GetTableProperties(std::shared_ptr* tp, const FileMetaData* file_meta, const std::string* fname) const { auto table_cache = cfd_->table_cache(); auto ioptions = cfd_->ioptions(); Status s = table_cache->GetTableProperties( vset_->env_options_, cfd_->internal_comparator(), file_meta->fd, tp, true /* no io */); if (s.ok()) { return s; } // We only ignore error type `Incomplete` since it's by design that we // disallow table when it's not in table cache. if (!s.IsIncomplete()) { return s; } // 2. Table is not present in table cache, we'll read the table properties // directly from the properties block in the file. std::unique_ptr file; if (fname != nullptr) { s = ioptions->env->NewRandomAccessFile( *fname, &file, vset_->env_options_); } else { s = ioptions->env->NewRandomAccessFile( TableFileName(vset_->db_options_->db_paths, file_meta->fd.GetNumber(), file_meta->fd.GetPathId()), &file, vset_->env_options_); } if (!s.ok()) { return s; } TableProperties* raw_table_properties; // By setting the magic number to kInvalidTableMagicNumber, we can by // pass the magic number check in the footer. std::unique_ptr file_reader( new RandomAccessFileReader(std::move(file))); s = ReadTableProperties( file_reader.get(), file_meta->fd.GetFileSize(), Footer::kInvalidTableMagicNumber /* table's magic number */, vset_->env_, ioptions->info_log, &raw_table_properties); if (!s.ok()) { return s; } RecordTick(ioptions->statistics, NUMBER_DIRECT_LOAD_TABLE_PROPERTIES); *tp = std::shared_ptr(raw_table_properties); return s; } Status Version::GetPropertiesOfAllTables(TablePropertiesCollection* props) { Status s; for (int level = 0; level < storage_info_.num_levels_; level++) { s = GetPropertiesOfAllTables(props, level); if (!s.ok()) { return s; } } return Status::OK(); } Status Version::GetPropertiesOfAllTables(TablePropertiesCollection* props, int level) { for (const auto& file_meta : storage_info_.files_[level]) { auto fname = TableFileName(vset_->db_options_->db_paths, file_meta->fd.GetNumber(), file_meta->fd.GetPathId()); // 1. If the table is already present in table cache, load table // properties from there. std::shared_ptr table_properties; Status s = GetTableProperties(&table_properties, file_meta, &fname); if (s.ok()) { props->insert({fname, table_properties}); } else { return s; } } return Status::OK(); } Status Version::GetPropertiesOfTablesInRange( const Range* range, std::size_t n, TablePropertiesCollection* props) const { for (int level = 0; level < storage_info_.num_non_empty_levels(); level++) { for (decltype(n) i = 0; i < n; i++) { // Convert user_key into a corresponding internal key. InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek); InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek); std::vector files; storage_info_.GetOverlappingInputs(level, &k1, &k2, &files, -1, nullptr, false); for (const auto& file_meta : files) { auto fname = TableFileName(vset_->db_options_->db_paths, file_meta->fd.GetNumber(), file_meta->fd.GetPathId()); if (props->count(fname) == 0) { // 1. If the table is already present in table cache, load table // properties from there. std::shared_ptr table_properties; Status s = GetTableProperties(&table_properties, file_meta, &fname); if (s.ok()) { props->insert({fname, table_properties}); } else { return s; } } } } } return Status::OK(); } Status Version::GetAggregatedTableProperties( std::shared_ptr* tp, int level) { TablePropertiesCollection props; Status s; if (level < 0) { s = GetPropertiesOfAllTables(&props); } else { s = GetPropertiesOfAllTables(&props, level); } if (!s.ok()) { return s; } auto* new_tp = new TableProperties(); for (const auto& item : props) { new_tp->Add(*item.second); } tp->reset(new_tp); return Status::OK(); } size_t Version::GetMemoryUsageByTableReaders() { size_t total_usage = 0; for (auto& file_level : storage_info_.level_files_brief_) { for (size_t i = 0; i < file_level.num_files; i++) { total_usage += cfd_->table_cache()->GetMemoryUsageByTableReader( vset_->env_options_, cfd_->internal_comparator(), file_level.files[i].fd); } } return total_usage; } void Version::GetColumnFamilyMetaData(ColumnFamilyMetaData* cf_meta) { assert(cf_meta); assert(cfd_); cf_meta->name = cfd_->GetName(); cf_meta->size = 0; cf_meta->file_count = 0; cf_meta->levels.clear(); auto* ioptions = cfd_->ioptions(); auto* vstorage = storage_info(); for (int level = 0; level < cfd_->NumberLevels(); level++) { uint64_t level_size = 0; cf_meta->file_count += vstorage->LevelFiles(level).size(); std::vector files; for (const auto& file : vstorage->LevelFiles(level)) { uint32_t path_id = file->fd.GetPathId(); std::string file_path; if (path_id < ioptions->db_paths.size()) { file_path = ioptions->db_paths[path_id].path; } else { assert(!ioptions->db_paths.empty()); file_path = ioptions->db_paths.back().path; } files.emplace_back( MakeTableFileName("", file->fd.GetNumber()), file_path, file->fd.GetFileSize(), file->smallest_seqno, file->largest_seqno, file->smallest.user_key().ToString(), file->largest.user_key().ToString(), file->being_compacted); level_size += file->fd.GetFileSize(); } cf_meta->levels.emplace_back( level, level_size, std::move(files)); cf_meta->size += level_size; } } uint64_t VersionStorageInfo::GetEstimatedActiveKeys() const { // Estimation will be inaccurate when: // (1) there exist merge keys // (2) keys are directly overwritten // (3) deletion on non-existing keys // (4) low number of samples if (current_num_samples_ == 0) { return 0; } if (current_num_non_deletions_ <= current_num_deletions_) { return 0; } uint64_t est = current_num_non_deletions_ - current_num_deletions_; uint64_t file_count = 0; for (int level = 0; level < num_levels_; ++level) { file_count += files_[level].size(); } if (current_num_samples_ < file_count) { // casting to avoid overflowing return static_cast( (est * static_cast(file_count) / current_num_samples_) ); } else { return est; } } void Version::AddIterators(const ReadOptions& read_options, const EnvOptions& soptions, MergeIteratorBuilder* merge_iter_builder) { assert(storage_info_.finalized_); if (storage_info_.num_non_empty_levels() == 0) { // No file in the Version. return; } auto* arena = merge_iter_builder->GetArena(); // Merge all level zero files together since they may overlap for (size_t i = 0; i < storage_info_.LevelFilesBrief(0).num_files; i++) { const auto& file = storage_info_.LevelFilesBrief(0).files[i]; merge_iter_builder->AddIterator(cfd_->table_cache()->NewIterator( read_options, soptions, cfd_->internal_comparator(), file.fd, nullptr, cfd_->internal_stats()->GetFileReadHist(0), false, arena)); } // For levels > 0, we can use a concatenating iterator that sequentially // walks through the non-overlapping files in the level, opening them // lazily. for (int level = 1; level < storage_info_.num_non_empty_levels(); level++) { if (storage_info_.LevelFilesBrief(level).num_files != 0) { auto* mem = arena->AllocateAligned(sizeof(LevelFileIteratorState)); auto* state = new (mem) LevelFileIteratorState(cfd_->table_cache(), read_options, soptions, cfd_->internal_comparator(), cfd_->internal_stats()->GetFileReadHist(level), false /* for_compaction */, cfd_->ioptions()->prefix_extractor != nullptr, IsFilterSkipped(level)); mem = arena->AllocateAligned(sizeof(LevelFileNumIterator)); auto* first_level_iter = new (mem) LevelFileNumIterator( cfd_->internal_comparator(), &storage_info_.LevelFilesBrief(level)); merge_iter_builder->AddIterator( NewTwoLevelIterator(state, first_level_iter, arena, false)); } } } VersionStorageInfo::VersionStorageInfo( const InternalKeyComparator* internal_comparator, const Comparator* user_comparator, int levels, CompactionStyle compaction_style, VersionStorageInfo* ref_vstorage) : internal_comparator_(internal_comparator), user_comparator_(user_comparator), // cfd is nullptr if Version is dummy num_levels_(levels), num_non_empty_levels_(0), file_indexer_(user_comparator), compaction_style_(compaction_style), files_(new std::vector[num_levels_]), base_level_(num_levels_ == 1 ? -1 : 1), files_by_compaction_pri_(num_levels_), level0_non_overlapping_(false), next_file_to_compact_by_size_(num_levels_), compaction_score_(num_levels_), compaction_level_(num_levels_), l0_delay_trigger_count_(0), accumulated_file_size_(0), accumulated_raw_key_size_(0), accumulated_raw_value_size_(0), accumulated_num_non_deletions_(0), accumulated_num_deletions_(0), current_num_non_deletions_(0), current_num_deletions_(0), current_num_samples_(0), estimated_compaction_needed_bytes_(0), finalized_(false) { if (ref_vstorage != nullptr) { accumulated_file_size_ = ref_vstorage->accumulated_file_size_; accumulated_raw_key_size_ = ref_vstorage->accumulated_raw_key_size_; accumulated_raw_value_size_ = ref_vstorage->accumulated_raw_value_size_; accumulated_num_non_deletions_ = ref_vstorage->accumulated_num_non_deletions_; accumulated_num_deletions_ = ref_vstorage->accumulated_num_deletions_; current_num_non_deletions_ = ref_vstorage->current_num_non_deletions_; current_num_deletions_ = ref_vstorage->current_num_deletions_; current_num_samples_ = ref_vstorage->current_num_samples_; } } Version::Version(ColumnFamilyData* column_family_data, VersionSet* vset, uint64_t version_number) : env_(vset->env_), cfd_(column_family_data), info_log_((cfd_ == nullptr) ? nullptr : cfd_->ioptions()->info_log), db_statistics_((cfd_ == nullptr) ? nullptr : cfd_->ioptions()->statistics), table_cache_((cfd_ == nullptr) ? nullptr : cfd_->table_cache()), merge_operator_((cfd_ == nullptr) ? nullptr : cfd_->ioptions()->merge_operator), storage_info_((cfd_ == nullptr) ? nullptr : &cfd_->internal_comparator(), (cfd_ == nullptr) ? nullptr : cfd_->user_comparator(), cfd_ == nullptr ? 0 : cfd_->NumberLevels(), cfd_ == nullptr ? kCompactionStyleLevel : cfd_->ioptions()->compaction_style, (cfd_ == nullptr || cfd_->current() == nullptr) ? nullptr : cfd_->current()->storage_info()), vset_(vset), next_(this), prev_(this), refs_(0), version_number_(version_number) {} void Version::Get(const ReadOptions& read_options, const LookupKey& k, std::string* value, Status* status, MergeContext* merge_context, bool* value_found, bool* key_exists, SequenceNumber* seq) { Slice ikey = k.internal_key(); Slice user_key = k.user_key(); assert(status->ok() || status->IsMergeInProgress()); if (key_exists != nullptr) { // will falsify below if not found *key_exists = true; } GetContext get_context( user_comparator(), merge_operator_, info_log_, db_statistics_, status->ok() ? GetContext::kNotFound : GetContext::kMerge, user_key, value, value_found, merge_context, this->env_, seq); FilePicker fp( storage_info_.files_, user_key, ikey, &storage_info_.level_files_brief_, storage_info_.num_non_empty_levels_, &storage_info_.file_indexer_, user_comparator(), internal_comparator()); FdWithKeyRange* f = fp.GetNextFile(); while (f != nullptr) { *status = table_cache_->Get( read_options, *internal_comparator(), f->fd, ikey, &get_context, cfd_->internal_stats()->GetFileReadHist(fp.GetHitFileLevel()), IsFilterSkipped(static_cast(fp.GetHitFileLevel()), fp.IsHitFileLastInLevel())); // TODO: examine the behavior for corrupted key if (!status->ok()) { return; } switch (get_context.State()) { case GetContext::kNotFound: // Keep searching in other files break; case GetContext::kFound: if (fp.GetHitFileLevel() == 0) { RecordTick(db_statistics_, GET_HIT_L0); } else if (fp.GetHitFileLevel() == 1) { RecordTick(db_statistics_, GET_HIT_L1); } else if (fp.GetHitFileLevel() >= 2) { RecordTick(db_statistics_, GET_HIT_L2_AND_UP); } return; case GetContext::kDeleted: // Use empty error message for speed *status = Status::NotFound(); return; case GetContext::kCorrupt: *status = Status::Corruption("corrupted key for ", user_key); return; case GetContext::kMerge: break; } f = fp.GetNextFile(); } if (GetContext::kMerge == get_context.State()) { if (!merge_operator_) { *status = Status::InvalidArgument( "merge_operator is not properly initialized."); return; } // merge_operands are in saver and we hit the beginning of the key history // do a final merge of nullptr and operands; if (merge_operator_->FullMerge(user_key, nullptr, merge_context->GetOperands(), value, info_log_)) { *status = Status::OK(); } else { RecordTick(db_statistics_, NUMBER_MERGE_FAILURES); *status = Status::Corruption("could not perform end-of-key merge for ", user_key); } } else { if (key_exists != nullptr) { *key_exists = false; } *status = Status::NotFound(); // Use an empty error message for speed } } bool Version::IsFilterSkipped(int level, bool is_file_last_in_level) { // Reaching the bottom level implies misses at all upper levels, so we'll // skip checking the filters when we predict a hit. return cfd_->ioptions()->optimize_filters_for_hits && (level > 0 || is_file_last_in_level) && level == storage_info_.num_non_empty_levels() - 1; } void VersionStorageInfo::GenerateLevelFilesBrief() { level_files_brief_.resize(num_non_empty_levels_); for (int level = 0; level < num_non_empty_levels_; level++) { DoGenerateLevelFilesBrief( &level_files_brief_[level], files_[level], &arena_); } } void Version::PrepareApply( const MutableCFOptions& mutable_cf_options, bool update_stats) { UpdateAccumulatedStats(update_stats); storage_info_.UpdateNumNonEmptyLevels(); storage_info_.CalculateBaseBytes(*cfd_->ioptions(), mutable_cf_options); storage_info_.UpdateFilesByCompactionPri(mutable_cf_options); storage_info_.GenerateFileIndexer(); storage_info_.GenerateLevelFilesBrief(); storage_info_.GenerateLevel0NonOverlapping(); } bool Version::MaybeInitializeFileMetaData(FileMetaData* file_meta) { if (file_meta->init_stats_from_file || file_meta->compensated_file_size > 0) { return false; } std::shared_ptr tp; Status s = GetTableProperties(&tp, file_meta); file_meta->init_stats_from_file = true; if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, vset_->db_options_->info_log, "Unable to load table properties for file %" PRIu64 " --- %s\n", file_meta->fd.GetNumber(), s.ToString().c_str()); return false; } if (tp.get() == nullptr) return false; file_meta->num_entries = tp->num_entries; file_meta->num_deletions = GetDeletedKeys(tp->user_collected_properties); file_meta->raw_value_size = tp->raw_value_size; file_meta->raw_key_size = tp->raw_key_size; return true; } void VersionStorageInfo::UpdateAccumulatedStats(FileMetaData* file_meta) { assert(file_meta->init_stats_from_file); accumulated_file_size_ += file_meta->fd.GetFileSize(); accumulated_raw_key_size_ += file_meta->raw_key_size; accumulated_raw_value_size_ += file_meta->raw_value_size; accumulated_num_non_deletions_ += file_meta->num_entries - file_meta->num_deletions; accumulated_num_deletions_ += file_meta->num_deletions; current_num_non_deletions_ += file_meta->num_entries - file_meta->num_deletions; current_num_deletions_ += file_meta->num_deletions; current_num_samples_++; } void VersionStorageInfo::RemoveCurrentStats(FileMetaData* file_meta) { if (file_meta->init_stats_from_file) { current_num_non_deletions_ -= file_meta->num_entries - file_meta->num_deletions; current_num_deletions_ -= file_meta->num_deletions; current_num_samples_--; } } void Version::UpdateAccumulatedStats(bool update_stats) { if (update_stats) { // maximum number of table properties loaded from files. const int kMaxInitCount = 20; int init_count = 0; // here only the first kMaxInitCount files which haven't been // initialized from file will be updated with num_deletions. // The motivation here is to cap the maximum I/O per Version creation. // The reason for choosing files from lower-level instead of higher-level // is that such design is able to propagate the initialization from // lower-level to higher-level: When the num_deletions of lower-level // files are updated, it will make the lower-level files have accurate // compensated_file_size, making lower-level to higher-level compaction // will be triggered, which creates higher-level files whose num_deletions // will be updated here. for (int level = 0; level < storage_info_.num_levels_ && init_count < kMaxInitCount; ++level) { for (auto* file_meta : storage_info_.files_[level]) { if (MaybeInitializeFileMetaData(file_meta)) { // each FileMeta will be initialized only once. storage_info_.UpdateAccumulatedStats(file_meta); if (++init_count >= kMaxInitCount) { break; } } } } // In case all sampled-files contain only deletion entries, then we // load the table-property of a file in higher-level to initialize // that value. for (int level = storage_info_.num_levels_ - 1; storage_info_.accumulated_raw_value_size_ == 0 && level >= 0; --level) { for (int i = static_cast(storage_info_.files_[level].size()) - 1; storage_info_.accumulated_raw_value_size_ == 0 && i >= 0; --i) { if (MaybeInitializeFileMetaData(storage_info_.files_[level][i])) { storage_info_.UpdateAccumulatedStats(storage_info_.files_[level][i]); } } } } storage_info_.ComputeCompensatedSizes(); } void VersionStorageInfo::ComputeCompensatedSizes() { static const int kDeletionWeightOnCompaction = 2; uint64_t average_value_size = GetAverageValueSize(); // compute the compensated size for (int level = 0; level < num_levels_; level++) { for (auto* file_meta : files_[level]) { // Here we only compute compensated_file_size for those file_meta // which compensated_file_size is uninitialized (== 0). This is true only // for files that have been created right now and no other thread has // access to them. That's why we can safely mutate compensated_file_size. if (file_meta->compensated_file_size == 0) { file_meta->compensated_file_size = file_meta->fd.GetFileSize(); // Here we only boost the size of deletion entries of a file only // when the number of deletion entries is greater than the number of // non-deletion entries in the file. The motivation here is that in // a stable workload, the number of deletion entries should be roughly // equal to the number of non-deletion entries. If we compensate the // size of deletion entries in a stable workload, the deletion // compensation logic might introduce unwanted effet which changes the // shape of LSM tree. if (file_meta->num_deletions * 2 >= file_meta->num_entries) { file_meta->compensated_file_size += (file_meta->num_deletions * 2 - file_meta->num_entries) * average_value_size * kDeletionWeightOnCompaction; } } } } } int VersionStorageInfo::MaxInputLevel() const { if (compaction_style_ == kCompactionStyleLevel) { return num_levels() - 2; } return 0; } void VersionStorageInfo::EstimateCompactionBytesNeeded( const MutableCFOptions& mutable_cf_options) { // Only implemented for level-based compaction if (compaction_style_ != kCompactionStyleLevel) { estimated_compaction_needed_bytes_ = 0; return; } // Start from Level 0, if level 0 qualifies compaction to level 1, // we estimate the size of compaction. // Then we move on to the next level and see whether it qualifies compaction // to the next level. The size of the level is estimated as the actual size // on the level plus the input bytes from the previous level if there is any. // If it exceeds, take the exceeded bytes as compaction input and add the size // of the compaction size to tatal size. // We keep doing it to Level 2, 3, etc, until the last level and return the // accumulated bytes. uint64_t bytes_compact_to_next_level = 0; // Level 0 bool level0_compact_triggered = false; if (static_cast(files_[0].size()) > mutable_cf_options.level0_file_num_compaction_trigger) { level0_compact_triggered = true; for (auto* f : files_[0]) { bytes_compact_to_next_level += f->fd.GetFileSize(); } estimated_compaction_needed_bytes_ = bytes_compact_to_next_level; } else { estimated_compaction_needed_bytes_ = 0; } // Level 1 and up. for (int level = base_level(); level <= MaxInputLevel(); level++) { uint64_t level_size = 0; for (auto* f : files_[level]) { level_size += f->fd.GetFileSize(); } if (level == base_level() && level0_compact_triggered) { // Add base level size to compaction if level0 compaction triggered. estimated_compaction_needed_bytes_ += level_size; } // Add size added by previous compaction level_size += bytes_compact_to_next_level; bytes_compact_to_next_level = 0; uint64_t level_target = MaxBytesForLevel(level); if (level_size > level_target) { bytes_compact_to_next_level = level_size - level_target; // Simplify to assume the actual compaction fan-out ratio is always // mutable_cf_options.max_bytes_for_level_multiplier. estimated_compaction_needed_bytes_ += bytes_compact_to_next_level * (1 + mutable_cf_options.max_bytes_for_level_multiplier); } } } void VersionStorageInfo::ComputeCompactionScore( const MutableCFOptions& mutable_cf_options, const CompactionOptionsFIFO& compaction_options_fifo) { for (int level = 0; level <= MaxInputLevel(); level++) { double score; if (level == 0) { // We treat level-0 specially by bounding the number of files // instead of number of bytes for two reasons: // // (1) With larger write-buffer sizes, it is nice not to do too // many level-0 compactions. // // (2) The files in level-0 are merged on every read and // therefore we wish to avoid too many files when the individual // file size is small (perhaps because of a small write-buffer // setting, or very high compression ratios, or lots of // overwrites/deletions). int num_sorted_runs = 0; uint64_t total_size = 0; for (auto* f : files_[level]) { if (!f->being_compacted) { total_size += f->compensated_file_size; num_sorted_runs++; } } if (compaction_style_ == kCompactionStyleUniversal) { // For universal compaction, we use level0 score to indicate // compaction score for the whole DB. Adding other levels as if // they are L0 files. for (int i = 1; i < num_levels(); i++) { if (!files_[i].empty() && !files_[i][0]->being_compacted) { num_sorted_runs++; } } } if (compaction_style_ == kCompactionStyleFIFO) { score = static_cast(total_size) / compaction_options_fifo.max_table_files_size; } else { score = static_cast(num_sorted_runs) / mutable_cf_options.level0_file_num_compaction_trigger; } } else { // Compute the ratio of current size to size limit. uint64_t level_bytes_no_compacting = 0; for (auto f : files_[level]) { if (!f->being_compacted) { level_bytes_no_compacting += f->compensated_file_size; } } score = static_cast(level_bytes_no_compacting) / MaxBytesForLevel(level); } compaction_level_[level] = level; compaction_score_[level] = score; } // sort all the levels based on their score. Higher scores get listed // first. Use bubble sort because the number of entries are small. for (int i = 0; i < num_levels() - 2; i++) { for (int j = i + 1; j < num_levels() - 1; j++) { if (compaction_score_[i] < compaction_score_[j]) { double score = compaction_score_[i]; int level = compaction_level_[i]; compaction_score_[i] = compaction_score_[j]; compaction_level_[i] = compaction_level_[j]; compaction_score_[j] = score; compaction_level_[j] = level; } } } ComputeFilesMarkedForCompaction(); EstimateCompactionBytesNeeded(mutable_cf_options); } void VersionStorageInfo::ComputeFilesMarkedForCompaction() { files_marked_for_compaction_.clear(); int last_qualify_level = 0; // Do not include files from the last level with data // If table properties collector suggests a file on the last level, // we should not move it to a new level. for (int level = num_levels() - 1; level >= 1; level--) { if (!files_[level].empty()) { last_qualify_level = level - 1; break; } } for (int level = 0; level <= last_qualify_level; level++) { for (auto* f : files_[level]) { if (!f->being_compacted && f->marked_for_compaction) { files_marked_for_compaction_.emplace_back(level, f); } } } } namespace { // used to sort files by size struct Fsize { size_t index; FileMetaData* file; }; // Compator that is used to sort files based on their size // In normal mode: descending size bool CompareCompensatedSizeDescending(const Fsize& first, const Fsize& second) { return (first.file->compensated_file_size > second.file->compensated_file_size); } } // anonymous namespace void VersionStorageInfo::AddFile(int level, FileMetaData* f, Logger* info_log) { auto* level_files = &files_[level]; // Must not overlap #ifndef NDEBUG if (level > 0 && !level_files->empty() && internal_comparator_->Compare( (*level_files)[level_files->size() - 1]->largest, f->smallest) >= 0) { auto* f2 = (*level_files)[level_files->size() - 1]; if (info_log != nullptr) { Error(info_log, "Adding new file %" PRIu64 " range (%s, %s) to level %d but overlapping " "with existing file %" PRIu64 " %s %s", f->fd.GetNumber(), f->smallest.DebugString(true).c_str(), f->largest.DebugString(true).c_str(), level, f2->fd.GetNumber(), f2->smallest.DebugString(true).c_str(), f2->largest.DebugString(true).c_str()); LogFlush(info_log); } assert(false); } #endif f->refs++; level_files->push_back(f); } // Version::PrepareApply() need to be called before calling the function, or // following functions called: // 1. UpdateNumNonEmptyLevels(); // 2. CalculateBaseBytes(); // 3. UpdateFilesByCompactionPri(); // 4. GenerateFileIndexer(); // 5. GenerateLevelFilesBrief(); // 6. GenerateLevel0NonOverlapping(); void VersionStorageInfo::SetFinalized() { finalized_ = true; #ifndef NDEBUG if (compaction_style_ != kCompactionStyleLevel) { // Not level based compaction. return; } assert(base_level_ < 0 || num_levels() == 1 || (base_level_ >= 1 && base_level_ < num_levels())); // Verify all levels newer than base_level are empty except L0 for (int level = 1; level < base_level(); level++) { assert(NumLevelBytes(level) == 0); } uint64_t max_bytes_prev_level = 0; for (int level = base_level(); level < num_levels() - 1; level++) { if (LevelFiles(level).size() == 0) { continue; } assert(MaxBytesForLevel(level) >= max_bytes_prev_level); max_bytes_prev_level = MaxBytesForLevel(level); } int num_empty_non_l0_level = 0; for (int level = 0; level < num_levels(); level++) { assert(LevelFiles(level).size() == 0 || LevelFiles(level).size() == LevelFilesBrief(level).num_files); if (level > 0 && NumLevelBytes(level) > 0) { num_empty_non_l0_level++; } if (LevelFiles(level).size() > 0) { assert(level < num_non_empty_levels()); } } assert(compaction_level_.size() > 0); assert(compaction_level_.size() == compaction_score_.size()); #endif } void VersionStorageInfo::UpdateNumNonEmptyLevels() { num_non_empty_levels_ = num_levels_; for (int i = num_levels_ - 1; i >= 0; i--) { if (files_[i].size() != 0) { return; } else { num_non_empty_levels_ = i; } } } void VersionStorageInfo::UpdateFilesByCompactionPri( const MutableCFOptions& mutable_cf_options) { if (compaction_style_ == kCompactionStyleFIFO || compaction_style_ == kCompactionStyleUniversal) { // don't need this return; } // No need to sort the highest level because it is never compacted. for (int level = 0; level < num_levels() - 1; level++) { const std::vector& files = files_[level]; auto& files_by_compaction_pri = files_by_compaction_pri_[level]; assert(files_by_compaction_pri.size() == 0); // populate a temp vector for sorting based on size std::vector temp(files.size()); for (size_t i = 0; i < files.size(); i++) { temp[i].index = i; temp[i].file = files[i]; } // sort the top number_of_files_to_sort_ based on file size size_t num = VersionStorageInfo::kNumberFilesToSort; if (num > temp.size()) { num = temp.size(); } switch (mutable_cf_options.compaction_pri) { case kByCompensatedSize: std::partial_sort(temp.begin(), temp.begin() + num, temp.end(), CompareCompensatedSizeDescending); break; case kOldestLargestSeqFirst: std::sort(temp.begin(), temp.end(), [this](const Fsize& f1, const Fsize& f2) -> bool { return f1.file->largest_seqno < f2.file->largest_seqno; }); break; case kOldestSmallestSeqFirst: std::sort(temp.begin(), temp.end(), [this](const Fsize& f1, const Fsize& f2) -> bool { return f1.file->smallest_seqno < f2.file->smallest_seqno; }); break; default: assert(false); } assert(temp.size() == files.size()); // initialize files_by_compaction_pri_ for (size_t i = 0; i < temp.size(); i++) { files_by_compaction_pri.push_back(static_cast(temp[i].index)); } next_file_to_compact_by_size_[level] = 0; assert(files_[level].size() == files_by_compaction_pri_[level].size()); } } void VersionStorageInfo::GenerateLevel0NonOverlapping() { assert(!finalized_); level0_non_overlapping_ = true; if (level_files_brief_.size() == 0) { return; } // A copy of L0 files sorted by smallest key std::vector level0_sorted_file( level_files_brief_[0].files, level_files_brief_[0].files + level_files_brief_[0].num_files); sort(level0_sorted_file.begin(), level0_sorted_file.end(), [this](const FdWithKeyRange & f1, const FdWithKeyRange & f2)->bool { return (internal_comparator_->Compare(f1.smallest_key, f2.smallest_key) < 0); }); for (size_t i = 1; i < level0_sorted_file.size(); ++i) { FdWithKeyRange& f = level0_sorted_file[i]; FdWithKeyRange& prev = level0_sorted_file[i - 1]; if (internal_comparator_->Compare(prev.largest_key, f.smallest_key) >= 0) { level0_non_overlapping_ = false; break; } } } void Version::Ref() { ++refs_; } bool Version::Unref() { assert(refs_ >= 1); --refs_; if (refs_ == 0) { delete this; return true; } return false; } bool VersionStorageInfo::OverlapInLevel(int level, const Slice* smallest_user_key, const Slice* largest_user_key) { if (level >= num_non_empty_levels_) { // empty level, no overlap return false; } return SomeFileOverlapsRange(*internal_comparator_, (level > 0), level_files_brief_[level], smallest_user_key, largest_user_key); } // Store in "*inputs" all files in "level" that overlap [begin,end] // If hint_index is specified, then it points to a file in the // overlapping range. // The file_index returns a pointer to any file in an overlapping range. void VersionStorageInfo::GetOverlappingInputs( int level, const InternalKey* begin, const InternalKey* end, std::vector* inputs, int hint_index, int* file_index, bool expand_range) const { if (level >= num_non_empty_levels_) { // this level is empty, no overlapping inputs return; } inputs->clear(); Slice user_begin, user_end; if (begin != nullptr) { user_begin = begin->user_key(); } if (end != nullptr) { user_end = end->user_key(); } if (file_index) { *file_index = -1; } const Comparator* user_cmp = user_comparator_; if (begin != nullptr && end != nullptr && level > 0) { GetOverlappingInputsBinarySearch(level, user_begin, user_end, inputs, hint_index, file_index); return; } for (size_t i = 0; i < level_files_brief_[level].num_files; ) { FdWithKeyRange* f = &(level_files_brief_[level].files[i++]); const Slice file_start = ExtractUserKey(f->smallest_key); const Slice file_limit = ExtractUserKey(f->largest_key); if (begin != nullptr && user_cmp->Compare(file_limit, user_begin) < 0) { // "f" is completely before specified range; skip it } else if (end != nullptr && user_cmp->Compare(file_start, user_end) > 0) { // "f" is completely after specified range; skip it } else { inputs->push_back(files_[level][i-1]); if (level == 0 && expand_range) { // Level-0 files may overlap each other. So check if the newly // added file has expanded the range. If so, restart search. if (begin != nullptr && user_cmp->Compare(file_start, user_begin) < 0) { user_begin = file_start; inputs->clear(); i = 0; } else if (end != nullptr && user_cmp->Compare(file_limit, user_end) > 0) { user_end = file_limit; inputs->clear(); i = 0; } } else if (file_index) { *file_index = static_cast(i) - 1; } } } } // Store in "*inputs" all files in "level" that overlap [begin,end] // Employ binary search to find at least one file that overlaps the // specified range. From that file, iterate backwards and // forwards to find all overlapping files. void VersionStorageInfo::GetOverlappingInputsBinarySearch( int level, const Slice& user_begin, const Slice& user_end, std::vector* inputs, int hint_index, int* file_index) const { assert(level > 0); int min = 0; int mid = 0; int max = static_cast(files_[level].size()) - 1; bool foundOverlap = false; const Comparator* user_cmp = user_comparator_; // if the caller already knows the index of a file that has overlap, // then we can skip the binary search. if (hint_index != -1) { mid = hint_index; foundOverlap = true; } while (!foundOverlap && min <= max) { mid = (min + max)/2; FdWithKeyRange* f = &(level_files_brief_[level].files[mid]); const Slice file_start = ExtractUserKey(f->smallest_key); const Slice file_limit = ExtractUserKey(f->largest_key); if (user_cmp->Compare(file_limit, user_begin) < 0) { min = mid + 1; } else if (user_cmp->Compare(user_end, file_start) < 0) { max = mid - 1; } else { foundOverlap = true; break; } } // If there were no overlapping files, return immediately. if (!foundOverlap) { return; } // returns the index where an overlap is found if (file_index) { *file_index = mid; } ExtendOverlappingInputs(level, user_begin, user_end, inputs, mid); } // Store in "*inputs" all files in "level" that overlap [begin,end] // The midIndex specifies the index of at least one file that // overlaps the specified range. From that file, iterate backward // and forward to find all overlapping files. // Use FileLevel in searching, make it faster void VersionStorageInfo::ExtendOverlappingInputs( int level, const Slice& user_begin, const Slice& user_end, std::vector* inputs, unsigned int midIndex) const { const Comparator* user_cmp = user_comparator_; const FdWithKeyRange* files = level_files_brief_[level].files; #ifndef NDEBUG { // assert that the file at midIndex overlaps with the range assert(midIndex < level_files_brief_[level].num_files); const FdWithKeyRange* f = &files[midIndex]; const Slice fstart = ExtractUserKey(f->smallest_key); const Slice flimit = ExtractUserKey(f->largest_key); if (user_cmp->Compare(fstart, user_begin) >= 0) { assert(user_cmp->Compare(fstart, user_end) <= 0); } else { assert(user_cmp->Compare(flimit, user_begin) >= 0); } } #endif int startIndex = midIndex + 1; int endIndex = midIndex; int count __attribute__((unused)) = 0; // check backwards from 'mid' to lower indices for (int i = midIndex; i >= 0 ; i--) { const FdWithKeyRange* f = &files[i]; const Slice file_limit = ExtractUserKey(f->largest_key); if (user_cmp->Compare(file_limit, user_begin) >= 0) { startIndex = i; assert((count++, true)); } else { break; } } // check forward from 'mid+1' to higher indices for (unsigned int i = midIndex+1; i < level_files_brief_[level].num_files; i++) { const FdWithKeyRange* f = &files[i]; const Slice file_start = ExtractUserKey(f->smallest_key); if (user_cmp->Compare(file_start, user_end) <= 0) { assert((count++, true)); endIndex = i; } else { break; } } assert(count == endIndex - startIndex + 1); // insert overlapping files into vector for (int i = startIndex; i <= endIndex; i++) { FileMetaData* f = files_[level][i]; inputs->push_back(f); } } // Returns true iff the first or last file in inputs contains // an overlapping user key to the file "just outside" of it (i.e. // just after the last file, or just before the first file) // REQUIRES: "*inputs" is a sorted list of non-overlapping files bool VersionStorageInfo::HasOverlappingUserKey( const std::vector* inputs, int level) { // If inputs empty, there is no overlap. // If level == 0, it is assumed that all needed files were already included. if (inputs->empty() || level == 0){ return false; } const Comparator* user_cmp = user_comparator_; const rocksdb::LevelFilesBrief& file_level = level_files_brief_[level]; const FdWithKeyRange* files = level_files_brief_[level].files; const size_t kNumFiles = file_level.num_files; // Check the last file in inputs against the file after it size_t last_file = FindFile(*internal_comparator_, file_level, inputs->back()->largest.Encode()); assert(last_file < kNumFiles); // File should exist! if (last_file < kNumFiles-1) { // If not the last file const Slice last_key_in_input = ExtractUserKey( files[last_file].largest_key); const Slice first_key_after = ExtractUserKey( files[last_file+1].smallest_key); if (user_cmp->Equal(last_key_in_input, first_key_after)) { // The last user key in input overlaps with the next file's first key return true; } } // Check the first file in inputs against the file just before it size_t first_file = FindFile(*internal_comparator_, file_level, inputs->front()->smallest.Encode()); assert(first_file <= last_file); // File should exist! if (first_file > 0) { // If not first file const Slice& first_key_in_input = ExtractUserKey( files[first_file].smallest_key); const Slice& last_key_before = ExtractUserKey( files[first_file-1].largest_key); if (user_cmp->Equal(first_key_in_input, last_key_before)) { // The first user key in input overlaps with the previous file's last key return true; } } return false; } uint64_t VersionStorageInfo::NumLevelBytes(int level) const { assert(level >= 0); assert(level < num_levels()); return TotalFileSize(files_[level]); } const char* VersionStorageInfo::LevelSummary( LevelSummaryStorage* scratch) const { int len = 0; if (compaction_style_ == kCompactionStyleLevel && num_levels() > 1) { assert(base_level_ < static_cast(level_max_bytes_.size())); len = snprintf(scratch->buffer, sizeof(scratch->buffer), "base level %d max bytes base %" PRIu64 " ", base_level_, level_max_bytes_[base_level_]); } len += snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, "files["); for (int i = 0; i < num_levels(); i++) { int sz = sizeof(scratch->buffer) - len; int ret = snprintf(scratch->buffer + len, sz, "%d ", int(files_[i].size())); if (ret < 0 || ret >= sz) break; len += ret; } if (len > 0) { // overwrite the last space --len; } len += snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, "] max score %.2f", compaction_score_[0]); if (!files_marked_for_compaction_.empty()) { snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, " (%" ROCKSDB_PRIszt " files need compaction)", files_marked_for_compaction_.size()); } return scratch->buffer; } const char* VersionStorageInfo::LevelFileSummary(FileSummaryStorage* scratch, int level) const { int len = snprintf(scratch->buffer, sizeof(scratch->buffer), "files_size["); for (const auto& f : files_[level]) { int sz = sizeof(scratch->buffer) - len; char sztxt[16]; AppendHumanBytes(f->fd.GetFileSize(), sztxt, sizeof(sztxt)); int ret = snprintf(scratch->buffer + len, sz, "#%" PRIu64 "(seq=%" PRIu64 ",sz=%s,%d) ", f->fd.GetNumber(), f->smallest_seqno, sztxt, static_cast(f->being_compacted)); if (ret < 0 || ret >= sz) break; len += ret; } // overwrite the last space (only if files_[level].size() is non-zero) if (files_[level].size() && len > 0) { --len; } snprintf(scratch->buffer + len, sizeof(scratch->buffer) - len, "]"); return scratch->buffer; } int64_t VersionStorageInfo::MaxNextLevelOverlappingBytes() { uint64_t result = 0; std::vector overlaps; for (int level = 1; level < num_levels() - 1; level++) { for (const auto& f : files_[level]) { GetOverlappingInputs(level + 1, &f->smallest, &f->largest, &overlaps); const uint64_t sum = TotalFileSize(overlaps); if (sum > result) { result = sum; } } } return result; } uint64_t VersionStorageInfo::MaxBytesForLevel(int level) const { // Note: the result for level zero is not really used since we set // the level-0 compaction threshold based on number of files. assert(level >= 0); assert(level < static_cast(level_max_bytes_.size())); return level_max_bytes_[level]; } void VersionStorageInfo::CalculateBaseBytes(const ImmutableCFOptions& ioptions, const MutableCFOptions& options) { // Special logic to set number of sorted runs. // It is to match the previous behavior when all files are in L0. int num_l0_count = static_cast(files_[0].size()); if (compaction_style_ == kCompactionStyleUniversal) { // For universal compaction, we use level0 score to indicate // compaction score for the whole DB. Adding other levels as if // they are L0 files. for (int i = 1; i < num_levels(); i++) { if (!files_[i].empty()) { num_l0_count++; } } } set_l0_delay_trigger_count(num_l0_count); level_max_bytes_.resize(ioptions.num_levels); if (!ioptions.level_compaction_dynamic_level_bytes) { base_level_ = (ioptions.compaction_style == kCompactionStyleLevel) ? 1 : -1; // Calculate for static bytes base case for (int i = 0; i < ioptions.num_levels; ++i) { if (i == 0 && ioptions.compaction_style == kCompactionStyleUniversal) { level_max_bytes_[i] = options.max_bytes_for_level_base; } else if (i > 1) { level_max_bytes_[i] = MultiplyCheckOverflow( MultiplyCheckOverflow(level_max_bytes_[i - 1], options.max_bytes_for_level_multiplier), options.MaxBytesMultiplerAdditional(i - 1)); } else { level_max_bytes_[i] = options.max_bytes_for_level_base; } } } else { uint64_t max_level_size = 0; int first_non_empty_level = -1; // Find size of non-L0 level of most data. // Cannot use the size of the last level because it can be empty or less // than previous levels after compaction. for (int i = 1; i < num_levels_; i++) { uint64_t total_size = 0; for (const auto& f : files_[i]) { total_size += f->fd.GetFileSize(); } if (total_size > 0 && first_non_empty_level == -1) { first_non_empty_level = i; } if (total_size > max_level_size) { max_level_size = total_size; } } // Prefill every level's max bytes to disallow compaction from there. for (int i = 0; i < num_levels_; i++) { level_max_bytes_[i] = std::numeric_limits::max(); } if (max_level_size == 0) { // No data for L1 and up. L0 compacts to last level directly. // No compaction from L1+ needs to be scheduled. base_level_ = num_levels_ - 1; } else { uint64_t base_bytes_max = options.max_bytes_for_level_base; uint64_t base_bytes_min = base_bytes_max / options.max_bytes_for_level_multiplier; // Try whether we can make last level's target size to be max_level_size uint64_t cur_level_size = max_level_size; for (int i = num_levels_ - 2; i >= first_non_empty_level; i--) { // Round up after dividing cur_level_size /= options.max_bytes_for_level_multiplier; } // Calculate base level and its size. uint64_t base_level_size; if (cur_level_size <= base_bytes_min) { // Case 1. If we make target size of last level to be max_level_size, // target size of the first non-empty level would be smaller than // base_bytes_min. We set it be base_bytes_min. base_level_size = base_bytes_min + 1U; base_level_ = first_non_empty_level; Warn(ioptions.info_log, "More existing levels in DB than needed. " "max_bytes_for_level_multiplier may not be guaranteed."); } else { // Find base level (where L0 data is compacted to). base_level_ = first_non_empty_level; while (base_level_ > 1 && cur_level_size > base_bytes_max) { --base_level_; cur_level_size = cur_level_size / options.max_bytes_for_level_multiplier; } if (cur_level_size > base_bytes_max) { // Even L1 will be too large assert(base_level_ == 1); base_level_size = base_bytes_max; } else { base_level_size = cur_level_size; } } uint64_t level_size = base_level_size; for (int i = base_level_; i < num_levels_; i++) { if (i > base_level_) { level_size = MultiplyCheckOverflow( level_size, options.max_bytes_for_level_multiplier); } level_max_bytes_[i] = level_size; } } } } uint64_t VersionStorageInfo::EstimateLiveDataSize() const { // Estimate the live data size by adding up the size of the last level for all // key ranges. Note: Estimate depends on the ordering of files in level 0 // because files in level 0 can be overlapping. uint64_t size = 0; auto ikey_lt = [this](InternalKey* x, InternalKey* y) { return internal_comparator_->Compare(*x, *y) < 0; }; // (Ordered) map of largest keys in non-overlapping files std::map ranges(ikey_lt); for (int l = num_levels_ - 1; l >= 0; l--) { bool found_end = false; for (auto file : files_[l]) { // Find the first file where the largest key is larger than the smallest // key of the current file. If this file does not overlap with the // current file, none of the files in the map does. If there is // no potential overlap, we can safely insert the rest of this level // (if the level is not 0) into the map without checking again because // the elements in the level are sorted and non-overlapping. auto lb = (found_end && l != 0) ? ranges.end() : ranges.lower_bound(&file->smallest); found_end = (lb == ranges.end()); if (found_end || internal_comparator_->Compare( file->largest, (*lb).second->smallest) < 0) { ranges.emplace_hint(lb, &file->largest, file); size += file->fd.file_size; } } } return size; } void Version::AddLiveFiles(std::vector* live) { for (int level = 0; level < storage_info_.num_levels(); level++) { const std::vector& files = storage_info_.files_[level]; for (const auto& file : files) { live->push_back(file->fd); } } } std::string Version::DebugString(bool hex) const { std::string r; for (int level = 0; level < storage_info_.num_levels_; level++) { // E.g., // --- level 1 --- // 17:123['a' .. 'd'] // 20:43['e' .. 'g'] r.append("--- level "); AppendNumberTo(&r, level); r.append(" --- version# "); AppendNumberTo(&r, version_number_); r.append(" ---\n"); const std::vector& files = storage_info_.files_[level]; for (size_t i = 0; i < files.size(); i++) { r.push_back(' '); AppendNumberTo(&r, files[i]->fd.GetNumber()); r.push_back(':'); AppendNumberTo(&r, files[i]->fd.GetFileSize()); r.append("["); r.append(files[i]->smallest.DebugString(hex)); r.append(" .. "); r.append(files[i]->largest.DebugString(hex)); r.append("]\n"); } } return r; } // this is used to batch writes to the manifest file struct VersionSet::ManifestWriter { Status status; bool done; InstrumentedCondVar cv; ColumnFamilyData* cfd; VersionEdit* edit; explicit ManifestWriter(InstrumentedMutex* mu, ColumnFamilyData* _cfd, VersionEdit* e) : done(false), cv(mu), cfd(_cfd), edit(e) {} }; VersionSet::VersionSet(const std::string& dbname, const DBOptions* db_options, const EnvOptions& storage_options, Cache* table_cache, WriteBuffer* write_buffer, WriteController* write_controller) : column_family_set_(new ColumnFamilySet( dbname, db_options, storage_options, table_cache, write_buffer, write_controller)), env_(db_options->env), dbname_(dbname), db_options_(db_options), next_file_number_(2), manifest_file_number_(0), // Filled by Recover() pending_manifest_file_number_(0), last_sequence_(0), prev_log_number_(0), current_version_number_(0), manifest_file_size_(0), env_options_(storage_options), env_options_compactions_(env_options_) {} VersionSet::~VersionSet() { // we need to delete column_family_set_ because its destructor depends on // VersionSet column_family_set_.reset(); for (auto file : obsolete_files_) { delete file; } obsolete_files_.clear(); } void VersionSet::AppendVersion(ColumnFamilyData* column_family_data, Version* v) { // compute new compaction score v->storage_info()->ComputeCompactionScore( *column_family_data->GetLatestMutableCFOptions(), column_family_data->ioptions()->compaction_options_fifo); // Mark v finalized v->storage_info_.SetFinalized(); // Make "v" current assert(v->refs_ == 0); Version* current = column_family_data->current(); assert(v != current); if (current != nullptr) { assert(current->refs_ > 0); current->Unref(); } column_family_data->SetCurrent(v); v->Ref(); // Append to linked list v->prev_ = column_family_data->dummy_versions()->prev_; v->next_ = column_family_data->dummy_versions(); v->prev_->next_ = v; v->next_->prev_ = v; } Status VersionSet::LogAndApply(ColumnFamilyData* column_family_data, const MutableCFOptions& mutable_cf_options, VersionEdit* edit, InstrumentedMutex* mu, Directory* db_directory, bool new_descriptor_log, const ColumnFamilyOptions* new_cf_options) { mu->AssertHeld(); // column_family_data can be nullptr only if this is column_family_add. // in that case, we also need to specify ColumnFamilyOptions if (column_family_data == nullptr) { assert(edit->is_column_family_add_); assert(new_cf_options != nullptr); } // queue our request ManifestWriter w(mu, column_family_data, edit); manifest_writers_.push_back(&w); while (!w.done && &w != manifest_writers_.front()) { w.cv.Wait(); } if (w.done) { return w.status; } if (column_family_data != nullptr && column_family_data->IsDropped()) { // if column family is dropped by the time we get here, no need to write // anything to the manifest manifest_writers_.pop_front(); // Notify new head of write queue if (!manifest_writers_.empty()) { manifest_writers_.front()->cv.Signal(); } // we steal this code to also inform about cf-drop return Status::ShutdownInProgress(); } std::vector batch_edits; Version* v = nullptr; std::unique_ptr builder_guard(nullptr); // process all requests in the queue ManifestWriter* last_writer = &w; assert(!manifest_writers_.empty()); assert(manifest_writers_.front() == &w); if (edit->IsColumnFamilyManipulation()) { // no group commits for column family add or drop LogAndApplyCFHelper(edit); batch_edits.push_back(edit); } else { v = new Version(column_family_data, this, current_version_number_++); builder_guard.reset(new BaseReferencedVersionBuilder(column_family_data)); auto* builder = builder_guard->version_builder(); for (const auto& writer : manifest_writers_) { if (writer->edit->IsColumnFamilyManipulation() || writer->cfd->GetID() != column_family_data->GetID()) { // no group commits for column family add or drop // also, group commits across column families are not supported break; } last_writer = writer; LogAndApplyHelper(column_family_data, builder, v, last_writer->edit, mu); batch_edits.push_back(last_writer->edit); } builder->SaveTo(v->storage_info()); } // Initialize new descriptor log file if necessary by creating // a temporary file that contains a snapshot of the current version. uint64_t new_manifest_file_size = 0; Status s; assert(pending_manifest_file_number_ == 0); if (!descriptor_log_ || manifest_file_size_ > db_options_->max_manifest_file_size) { pending_manifest_file_number_ = NewFileNumber(); batch_edits.back()->SetNextFile(next_file_number_.load()); new_descriptor_log = true; } else { pending_manifest_file_number_ = manifest_file_number_; } if (new_descriptor_log) { // if we're writing out new snapshot make sure to persist max column family if (column_family_set_->GetMaxColumnFamily() > 0) { edit->SetMaxColumnFamily(column_family_set_->GetMaxColumnFamily()); } } // Unlock during expensive operations. New writes cannot get here // because &w is ensuring that all new writes get queued. { mu->Unlock(); TEST_SYNC_POINT("VersionSet::LogAndApply:WriteManifest"); if (!edit->IsColumnFamilyManipulation() && db_options_->max_open_files == -1) { // unlimited table cache. Pre-load table handle now. // Need to do it out of the mutex. builder_guard->version_builder()->LoadTableHandlers( column_family_data->internal_stats(), column_family_data->ioptions()->optimize_filters_for_hits); } // This is fine because everything inside of this block is serialized -- // only one thread can be here at the same time if (new_descriptor_log) { // create manifest file Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "Creating manifest %" PRIu64 "\n", pending_manifest_file_number_); unique_ptr descriptor_file; EnvOptions opt_env_opts = env_->OptimizeForManifestWrite(env_options_); s = NewWritableFile( env_, DescriptorFileName(dbname_, pending_manifest_file_number_), &descriptor_file, opt_env_opts); if (s.ok()) { descriptor_file->SetPreallocationBlockSize( db_options_->manifest_preallocation_size); unique_ptr file_writer( new WritableFileWriter(std::move(descriptor_file), opt_env_opts)); descriptor_log_.reset(new log::Writer(std::move(file_writer), 0, false)); s = WriteSnapshot(descriptor_log_.get()); } } if (!edit->IsColumnFamilyManipulation()) { // This is cpu-heavy operations, which should be called outside mutex. v->PrepareApply(mutable_cf_options, true); } // Write new record to MANIFEST log if (s.ok()) { for (auto& e : batch_edits) { std::string record; if (!e->EncodeTo(&record)) { s = Status::Corruption( "Unable to Encode VersionEdit:" + e->DebugString(true)); break; } TEST_KILL_RANDOM("VersionSet::LogAndApply:BeforeAddRecord", rocksdb_kill_odds * REDUCE_ODDS2); s = descriptor_log_->AddRecord(record); if (!s.ok()) { break; } } if (s.ok()) { s = SyncManifest(env_, db_options_, descriptor_log_->file()); } if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options_->info_log, "MANIFEST write: %s\n", s.ToString().c_str()); } } // If we just created a new descriptor file, install it by writing a // new CURRENT file that points to it. if (s.ok() && new_descriptor_log) { s = SetCurrentFile(env_, dbname_, pending_manifest_file_number_, db_options_->disableDataSync ? nullptr : db_directory); if (s.ok() && pending_manifest_file_number_ > manifest_file_number_) { // delete old manifest file Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "Deleting manifest %" PRIu64 " current manifest %" PRIu64 "\n", manifest_file_number_, pending_manifest_file_number_); // we don't care about an error here, PurgeObsoleteFiles will take care // of it later env_->DeleteFile(DescriptorFileName(dbname_, manifest_file_number_)); } } if (s.ok()) { // find offset in manifest file where this version is stored. new_manifest_file_size = descriptor_log_->file()->GetFileSize(); } if (edit->is_column_family_drop_) { TEST_SYNC_POINT("VersionSet::LogAndApply::ColumnFamilyDrop:0"); TEST_SYNC_POINT("VersionSet::LogAndApply::ColumnFamilyDrop:1"); TEST_SYNC_POINT("VersionSet::LogAndApply::ColumnFamilyDrop:2"); } LogFlush(db_options_->info_log); mu->Lock(); } // Install the new version if (s.ok()) { if (edit->is_column_family_add_) { // no group commit on column family add assert(batch_edits.size() == 1); assert(new_cf_options != nullptr); CreateColumnFamily(*new_cf_options, edit); } else if (edit->is_column_family_drop_) { assert(batch_edits.size() == 1); column_family_data->SetDropped(); if (column_family_data->Unref()) { delete column_family_data; } } else { uint64_t max_log_number_in_batch = 0; for (auto& e : batch_edits) { if (e->has_log_number_) { max_log_number_in_batch = std::max(max_log_number_in_batch, e->log_number_); } } if (max_log_number_in_batch != 0) { assert(column_family_data->GetLogNumber() <= max_log_number_in_batch); column_family_data->SetLogNumber(max_log_number_in_batch); } AppendVersion(column_family_data, v); } manifest_file_number_ = pending_manifest_file_number_; manifest_file_size_ = new_manifest_file_size; prev_log_number_ = edit->prev_log_number_; } else { Log(InfoLogLevel::ERROR_LEVEL, db_options_->info_log, "Error in committing version %lu to [%s]", (unsigned long)v->GetVersionNumber(), column_family_data ? column_family_data->GetName().c_str() : ""); delete v; if (new_descriptor_log) { Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "Deleting manifest %" PRIu64 " current manifest %" PRIu64 "\n", manifest_file_number_, pending_manifest_file_number_); descriptor_log_.reset(); env_->DeleteFile( DescriptorFileName(dbname_, pending_manifest_file_number_)); } } pending_manifest_file_number_ = 0; // wake up all the waiting writers while (true) { ManifestWriter* ready = manifest_writers_.front(); manifest_writers_.pop_front(); if (ready != &w) { ready->status = s; ready->done = true; ready->cv.Signal(); } if (ready == last_writer) break; } // Notify new head of write queue if (!manifest_writers_.empty()) { manifest_writers_.front()->cv.Signal(); } return s; } void VersionSet::LogAndApplyCFHelper(VersionEdit* edit) { assert(edit->IsColumnFamilyManipulation()); edit->SetNextFile(next_file_number_.load()); edit->SetLastSequence(last_sequence_); if (edit->is_column_family_drop_) { // if we drop column family, we have to make sure to save max column family, // so that we don't reuse existing ID edit->SetMaxColumnFamily(column_family_set_->GetMaxColumnFamily()); } } void VersionSet::LogAndApplyHelper(ColumnFamilyData* cfd, VersionBuilder* builder, Version* v, VersionEdit* edit, InstrumentedMutex* mu) { mu->AssertHeld(); assert(!edit->IsColumnFamilyManipulation()); if (edit->has_log_number_) { assert(edit->log_number_ >= cfd->GetLogNumber()); assert(edit->log_number_ < next_file_number_.load()); } if (!edit->has_prev_log_number_) { edit->SetPrevLogNumber(prev_log_number_); } edit->SetNextFile(next_file_number_.load()); edit->SetLastSequence(last_sequence_); builder->Apply(edit); } Status VersionSet::Recover( const std::vector& column_families, bool read_only) { std::unordered_map cf_name_to_options; for (auto cf : column_families) { cf_name_to_options.insert({cf.name, cf.options}); } // keeps track of column families in manifest that were not found in // column families parameters. if those column families are not dropped // by subsequent manifest records, Recover() will return failure status std::unordered_map column_families_not_found; // Read "CURRENT" file, which contains a pointer to the current manifest file std::string manifest_filename; Status s = ReadFileToString( env_, CurrentFileName(dbname_), &manifest_filename ); if (!s.ok()) { return s; } if (manifest_filename.empty() || manifest_filename.back() != '\n') { return Status::Corruption("CURRENT file does not end with newline"); } // remove the trailing '\n' manifest_filename.resize(manifest_filename.size() - 1); FileType type; bool parse_ok = ParseFileName(manifest_filename, &manifest_file_number_, &type); if (!parse_ok || type != kDescriptorFile) { return Status::Corruption("CURRENT file corrupted"); } Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "Recovering from manifest file: %s\n", manifest_filename.c_str()); manifest_filename = dbname_ + "/" + manifest_filename; unique_ptr manifest_file_reader; { unique_ptr manifest_file; s = env_->NewSequentialFile(manifest_filename, &manifest_file, env_options_); if (!s.ok()) { return s; } manifest_file_reader.reset( new SequentialFileReader(std::move(manifest_file))); } uint64_t current_manifest_file_size; s = env_->GetFileSize(manifest_filename, ¤t_manifest_file_size); if (!s.ok()) { return s; } bool have_log_number = false; bool have_prev_log_number = false; bool have_next_file = false; bool have_last_sequence = false; uint64_t next_file = 0; uint64_t last_sequence = 0; uint64_t log_number = 0; uint64_t previous_log_number = 0; uint32_t max_column_family = 0; std::unordered_map builders; // add default column family auto default_cf_iter = cf_name_to_options.find(kDefaultColumnFamilyName); if (default_cf_iter == cf_name_to_options.end()) { return Status::InvalidArgument("Default column family not specified"); } VersionEdit default_cf_edit; default_cf_edit.AddColumnFamily(kDefaultColumnFamilyName); default_cf_edit.SetColumnFamily(0); ColumnFamilyData* default_cfd = CreateColumnFamily(default_cf_iter->second, &default_cf_edit); builders.insert({0, new BaseReferencedVersionBuilder(default_cfd)}); { VersionSet::LogReporter reporter; reporter.status = &s; log::Reader reader(NULL, std::move(manifest_file_reader), &reporter, true /*checksum*/, 0 /*initial_offset*/, 0); Slice record; std::string scratch; while (reader.ReadRecord(&record, &scratch) && s.ok()) { VersionEdit edit; s = edit.DecodeFrom(record); if (!s.ok()) { break; } // Not found means that user didn't supply that column // family option AND we encountered column family add // record. Once we encounter column family drop record, // we will delete the column family from // column_families_not_found. bool cf_in_not_found = column_families_not_found.find(edit.column_family_) != column_families_not_found.end(); // in builders means that user supplied that column family // option AND that we encountered column family add record bool cf_in_builders = builders.find(edit.column_family_) != builders.end(); // they can't both be true assert(!(cf_in_not_found && cf_in_builders)); ColumnFamilyData* cfd = nullptr; if (edit.is_column_family_add_) { if (cf_in_builders || cf_in_not_found) { s = Status::Corruption( "Manifest adding the same column family twice"); break; } auto cf_options = cf_name_to_options.find(edit.column_family_name_); if (cf_options == cf_name_to_options.end()) { column_families_not_found.insert( {edit.column_family_, edit.column_family_name_}); } else { cfd = CreateColumnFamily(cf_options->second, &edit); builders.insert( {edit.column_family_, new BaseReferencedVersionBuilder(cfd)}); } } else if (edit.is_column_family_drop_) { if (cf_in_builders) { auto builder = builders.find(edit.column_family_); assert(builder != builders.end()); delete builder->second; builders.erase(builder); cfd = column_family_set_->GetColumnFamily(edit.column_family_); if (cfd->Unref()) { delete cfd; cfd = nullptr; } else { // who else can have reference to cfd!? assert(false); } } else if (cf_in_not_found) { column_families_not_found.erase(edit.column_family_); } else { s = Status::Corruption( "Manifest - dropping non-existing column family"); break; } } else if (!cf_in_not_found) { if (!cf_in_builders) { s = Status::Corruption( "Manifest record referencing unknown column family"); break; } cfd = column_family_set_->GetColumnFamily(edit.column_family_); // this should never happen since cf_in_builders is true assert(cfd != nullptr); if (edit.max_level_ >= cfd->current()->storage_info()->num_levels()) { s = Status::InvalidArgument( "db has more levels than options.num_levels"); break; } // if it is not column family add or column family drop, // then it's a file add/delete, which should be forwarded // to builder auto builder = builders.find(edit.column_family_); assert(builder != builders.end()); builder->second->version_builder()->Apply(&edit); } if (cfd != nullptr) { if (edit.has_log_number_) { if (cfd->GetLogNumber() > edit.log_number_) { Log(InfoLogLevel::WARN_LEVEL, db_options_->info_log, "MANIFEST corruption detected, but ignored - Log numbers in " "records NOT monotonically increasing"); } else { cfd->SetLogNumber(edit.log_number_); have_log_number = true; } } if (edit.has_comparator_ && edit.comparator_ != cfd->user_comparator()->Name()) { s = Status::InvalidArgument( cfd->user_comparator()->Name(), "does not match existing comparator " + edit.comparator_); break; } } if (edit.has_prev_log_number_) { previous_log_number = edit.prev_log_number_; have_prev_log_number = true; } if (edit.has_next_file_number_) { next_file = edit.next_file_number_; have_next_file = true; } if (edit.has_max_column_family_) { max_column_family = edit.max_column_family_; } if (edit.has_last_sequence_) { last_sequence = edit.last_sequence_; have_last_sequence = true; } } } if (s.ok()) { if (!have_next_file) { s = Status::Corruption("no meta-nextfile entry in descriptor"); } else if (!have_log_number) { s = Status::Corruption("no meta-lognumber entry in descriptor"); } else if (!have_last_sequence) { s = Status::Corruption("no last-sequence-number entry in descriptor"); } if (!have_prev_log_number) { previous_log_number = 0; } column_family_set_->UpdateMaxColumnFamily(max_column_family); MarkFileNumberUsedDuringRecovery(previous_log_number); MarkFileNumberUsedDuringRecovery(log_number); } // there were some column families in the MANIFEST that weren't specified // in the argument. This is OK in read_only mode if (read_only == false && !column_families_not_found.empty()) { std::string list_of_not_found; for (const auto& cf : column_families_not_found) { list_of_not_found += ", " + cf.second; } list_of_not_found = list_of_not_found.substr(2); s = Status::InvalidArgument( "You have to open all column families. Column families not opened: " + list_of_not_found); } if (s.ok()) { for (auto cfd : *column_family_set_) { if (cfd->IsDropped()) { continue; } auto builders_iter = builders.find(cfd->GetID()); assert(builders_iter != builders.end()); auto* builder = builders_iter->second->version_builder(); if (db_options_->max_open_files == -1) { // unlimited table cache. Pre-load table handle now. // Need to do it out of the mutex. builder->LoadTableHandlers(cfd->internal_stats(), db_options_->max_file_opening_threads); } Version* v = new Version(cfd, this, current_version_number_++); builder->SaveTo(v->storage_info()); // Install recovered version v->PrepareApply(*cfd->GetLatestMutableCFOptions(), !(db_options_->skip_stats_update_on_db_open)); AppendVersion(cfd, v); } manifest_file_size_ = current_manifest_file_size; next_file_number_.store(next_file + 1); last_sequence_ = last_sequence; prev_log_number_ = previous_log_number; Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "Recovered from manifest file:%s succeeded," "manifest_file_number is %lu, next_file_number is %lu, " "last_sequence is %lu, log_number is %lu," "prev_log_number is %lu," "max_column_family is %u\n", manifest_filename.c_str(), (unsigned long)manifest_file_number_, (unsigned long)next_file_number_.load(), (unsigned long)last_sequence_, (unsigned long)log_number, (unsigned long)prev_log_number_, column_family_set_->GetMaxColumnFamily()); for (auto cfd : *column_family_set_) { if (cfd->IsDropped()) { continue; } Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "Column family [%s] (ID %u), log number is %" PRIu64 "\n", cfd->GetName().c_str(), cfd->GetID(), cfd->GetLogNumber()); } } for (auto builder : builders) { delete builder.second; } return s; } Status VersionSet::ListColumnFamilies(std::vector* column_families, const std::string& dbname, Env* env) { // these are just for performance reasons, not correcntes, // so we're fine using the defaults EnvOptions soptions; // Read "CURRENT" file, which contains a pointer to the current manifest file std::string current; Status s = ReadFileToString(env, CurrentFileName(dbname), ¤t); if (!s.ok()) { return s; } if (current.empty() || current[current.size()-1] != '\n') { return Status::Corruption("CURRENT file does not end with newline"); } current.resize(current.size() - 1); std::string dscname = dbname + "/" + current; unique_ptr file_reader; { unique_ptr file; s = env->NewSequentialFile(dscname, &file, soptions); if (!s.ok()) { return s; } file_reader.reset(new SequentialFileReader(std::move(file))); } std::map column_family_names; // default column family is always implicitly there column_family_names.insert({0, kDefaultColumnFamilyName}); VersionSet::LogReporter reporter; reporter.status = &s; log::Reader reader(NULL, std::move(file_reader), &reporter, true /*checksum*/, 0 /*initial_offset*/, 0); Slice record; std::string scratch; while (reader.ReadRecord(&record, &scratch) && s.ok()) { VersionEdit edit; s = edit.DecodeFrom(record); if (!s.ok()) { break; } if (edit.is_column_family_add_) { if (column_family_names.find(edit.column_family_) != column_family_names.end()) { s = Status::Corruption("Manifest adding the same column family twice"); break; } column_family_names.insert( {edit.column_family_, edit.column_family_name_}); } else if (edit.is_column_family_drop_) { if (column_family_names.find(edit.column_family_) == column_family_names.end()) { s = Status::Corruption( "Manifest - dropping non-existing column family"); break; } column_family_names.erase(edit.column_family_); } } column_families->clear(); if (s.ok()) { for (const auto& iter : column_family_names) { column_families->push_back(iter.second); } } return s; } #ifndef ROCKSDB_LITE Status VersionSet::ReduceNumberOfLevels(const std::string& dbname, const Options* options, const EnvOptions& env_options, int new_levels) { if (new_levels <= 1) { return Status::InvalidArgument( "Number of levels needs to be bigger than 1"); } ColumnFamilyOptions cf_options(*options); std::shared_ptr tc(NewLRUCache(options->max_open_files - 10, options->table_cache_numshardbits)); WriteController wc(options->delayed_write_rate); WriteBuffer wb(options->db_write_buffer_size); VersionSet versions(dbname, options, env_options, tc.get(), &wb, &wc); Status status; std::vector dummy; ColumnFamilyDescriptor dummy_descriptor(kDefaultColumnFamilyName, ColumnFamilyOptions(*options)); dummy.push_back(dummy_descriptor); status = versions.Recover(dummy); if (!status.ok()) { return status; } Version* current_version = versions.GetColumnFamilySet()->GetDefault()->current(); auto* vstorage = current_version->storage_info(); int current_levels = vstorage->num_levels(); if (current_levels <= new_levels) { return Status::OK(); } // Make sure there are file only on one level from // (new_levels-1) to (current_levels-1) int first_nonempty_level = -1; int first_nonempty_level_filenum = 0; for (int i = new_levels - 1; i < current_levels; i++) { int file_num = vstorage->NumLevelFiles(i); if (file_num != 0) { if (first_nonempty_level < 0) { first_nonempty_level = i; first_nonempty_level_filenum = file_num; } else { char msg[255]; snprintf(msg, sizeof(msg), "Found at least two levels containing files: " "[%d:%d],[%d:%d].\n", first_nonempty_level, first_nonempty_level_filenum, i, file_num); return Status::InvalidArgument(msg); } } } // we need to allocate an array with the old number of levels size to // avoid SIGSEGV in WriteSnapshot() // however, all levels bigger or equal to new_levels will be empty std::vector* new_files_list = new std::vector[current_levels]; for (int i = 0; i < new_levels - 1; i++) { new_files_list[i] = vstorage->LevelFiles(i); } if (first_nonempty_level > 0) { new_files_list[new_levels - 1] = vstorage->LevelFiles(first_nonempty_level); } delete[] vstorage -> files_; vstorage->files_ = new_files_list; vstorage->num_levels_ = new_levels; MutableCFOptions mutable_cf_options(*options, ImmutableCFOptions(*options)); VersionEdit ve; InstrumentedMutex dummy_mutex; InstrumentedMutexLock l(&dummy_mutex); return versions.LogAndApply( versions.GetColumnFamilySet()->GetDefault(), mutable_cf_options, &ve, &dummy_mutex, nullptr, true); } Status VersionSet::DumpManifest(Options& options, std::string& dscname, bool verbose, bool hex, bool json) { // Open the specified manifest file. unique_ptr file_reader; Status s; { unique_ptr file; s = options.env->NewSequentialFile(dscname, &file, env_options_); if (!s.ok()) { return s; } file_reader.reset(new SequentialFileReader(std::move(file))); } bool have_prev_log_number = false; bool have_next_file = false; bool have_last_sequence = false; uint64_t next_file = 0; uint64_t last_sequence = 0; uint64_t previous_log_number = 0; int count = 0; std::unordered_map comparators; std::unordered_map builders; // add default column family VersionEdit default_cf_edit; default_cf_edit.AddColumnFamily(kDefaultColumnFamilyName); default_cf_edit.SetColumnFamily(0); ColumnFamilyData* default_cfd = CreateColumnFamily(ColumnFamilyOptions(options), &default_cf_edit); builders.insert({0, new BaseReferencedVersionBuilder(default_cfd)}); { VersionSet::LogReporter reporter; reporter.status = &s; log::Reader reader(NULL, std::move(file_reader), &reporter, true /*checksum*/, 0 /*initial_offset*/, 0); Slice record; std::string scratch; while (reader.ReadRecord(&record, &scratch) && s.ok()) { VersionEdit edit; s = edit.DecodeFrom(record); if (!s.ok()) { break; } // Write out each individual edit if (verbose && !json) { printf("%s\n", edit.DebugString(hex).c_str()); } else if (json) { printf("%s\n", edit.DebugJSON(count, hex).c_str()); } count++; bool cf_in_builders = builders.find(edit.column_family_) != builders.end(); if (edit.has_comparator_) { comparators.insert({edit.column_family_, edit.comparator_}); } ColumnFamilyData* cfd = nullptr; if (edit.is_column_family_add_) { if (cf_in_builders) { s = Status::Corruption( "Manifest adding the same column family twice"); break; } cfd = CreateColumnFamily(ColumnFamilyOptions(options), &edit); builders.insert( {edit.column_family_, new BaseReferencedVersionBuilder(cfd)}); } else if (edit.is_column_family_drop_) { if (!cf_in_builders) { s = Status::Corruption( "Manifest - dropping non-existing column family"); break; } auto builder_iter = builders.find(edit.column_family_); delete builder_iter->second; builders.erase(builder_iter); comparators.erase(edit.column_family_); cfd = column_family_set_->GetColumnFamily(edit.column_family_); assert(cfd != nullptr); cfd->Unref(); delete cfd; cfd = nullptr; } else { if (!cf_in_builders) { s = Status::Corruption( "Manifest record referencing unknown column family"); break; } cfd = column_family_set_->GetColumnFamily(edit.column_family_); // this should never happen since cf_in_builders is true assert(cfd != nullptr); // if it is not column family add or column family drop, // then it's a file add/delete, which should be forwarded // to builder auto builder = builders.find(edit.column_family_); assert(builder != builders.end()); builder->second->version_builder()->Apply(&edit); } if (cfd != nullptr && edit.has_log_number_) { cfd->SetLogNumber(edit.log_number_); } if (edit.has_prev_log_number_) { previous_log_number = edit.prev_log_number_; have_prev_log_number = true; } if (edit.has_next_file_number_) { next_file = edit.next_file_number_; have_next_file = true; } if (edit.has_last_sequence_) { last_sequence = edit.last_sequence_; have_last_sequence = true; } if (edit.has_max_column_family_) { column_family_set_->UpdateMaxColumnFamily(edit.max_column_family_); } } } file_reader.reset(); if (s.ok()) { if (!have_next_file) { s = Status::Corruption("no meta-nextfile entry in descriptor"); printf("no meta-nextfile entry in descriptor"); } else if (!have_last_sequence) { printf("no last-sequence-number entry in descriptor"); s = Status::Corruption("no last-sequence-number entry in descriptor"); } if (!have_prev_log_number) { previous_log_number = 0; } } if (s.ok()) { for (auto cfd : *column_family_set_) { if (cfd->IsDropped()) { continue; } auto builders_iter = builders.find(cfd->GetID()); assert(builders_iter != builders.end()); auto builder = builders_iter->second->version_builder(); Version* v = new Version(cfd, this, current_version_number_++); builder->SaveTo(v->storage_info()); v->PrepareApply(*cfd->GetLatestMutableCFOptions(), false); printf("--------------- Column family \"%s\" (ID %u) --------------\n", cfd->GetName().c_str(), (unsigned int)cfd->GetID()); printf("log number: %lu\n", (unsigned long)cfd->GetLogNumber()); auto comparator = comparators.find(cfd->GetID()); if (comparator != comparators.end()) { printf("comparator: %s\n", comparator->second.c_str()); } else { printf("comparator: \n"); } printf("%s \n", v->DebugString(hex).c_str()); delete v; } // Free builders for (auto& builder : builders) { delete builder.second; } next_file_number_.store(next_file + 1); last_sequence_ = last_sequence; prev_log_number_ = previous_log_number; printf( "next_file_number %lu last_sequence " "%lu prev_log_number %lu max_column_family %u\n", (unsigned long)next_file_number_.load(), (unsigned long)last_sequence, (unsigned long)previous_log_number, column_family_set_->GetMaxColumnFamily()); } return s; } #endif // ROCKSDB_LITE void VersionSet::MarkFileNumberUsedDuringRecovery(uint64_t number) { // only called during recovery which is single threaded, so this works because // there can't be concurrent calls if (next_file_number_.load(std::memory_order_relaxed) <= number) { next_file_number_.store(number + 1, std::memory_order_relaxed); } } Status VersionSet::WriteSnapshot(log::Writer* log) { // TODO: Break up into multiple records to reduce memory usage on recovery? // WARNING: This method doesn't hold a mutex!! // This is done without DB mutex lock held, but only within single-threaded // LogAndApply. Column family manipulations can only happen within LogAndApply // (the same single thread), so we're safe to iterate. for (auto cfd : *column_family_set_) { if (cfd->IsDropped()) { continue; } { // Store column family info VersionEdit edit; if (cfd->GetID() != 0) { // default column family is always there, // no need to explicitly write it edit.AddColumnFamily(cfd->GetName()); edit.SetColumnFamily(cfd->GetID()); } edit.SetComparatorName( cfd->internal_comparator().user_comparator()->Name()); std::string record; if (!edit.EncodeTo(&record)) { return Status::Corruption( "Unable to Encode VersionEdit:" + edit.DebugString(true)); } Status s = log->AddRecord(record); if (!s.ok()) { return s; } } { // Save files VersionEdit edit; edit.SetColumnFamily(cfd->GetID()); for (int level = 0; level < cfd->NumberLevels(); level++) { for (const auto& f : cfd->current()->storage_info()->LevelFiles(level)) { edit.AddFile(level, f->fd.GetNumber(), f->fd.GetPathId(), f->fd.GetFileSize(), f->smallest, f->largest, f->smallest_seqno, f->largest_seqno, f->marked_for_compaction); } } edit.SetLogNumber(cfd->GetLogNumber()); std::string record; if (!edit.EncodeTo(&record)) { return Status::Corruption( "Unable to Encode VersionEdit:" + edit.DebugString(true)); } Status s = log->AddRecord(record); if (!s.ok()) { return s; } } } return Status::OK(); } // TODO(aekmekji): in CompactionJob::GenSubcompactionBoundaries(), this // function is called repeatedly with consecutive pairs of slices. For example // if the slice list is [a, b, c, d] this function is called with arguments // (a,b) then (b,c) then (c,d). Knowing this, an optimization is possible where // we avoid doing binary search for the keys b and c twice and instead somehow // maintain state of where they first appear in the files. uint64_t VersionSet::ApproximateSize(Version* v, const Slice& start, const Slice& end, int start_level, int end_level) { // pre-condition assert(v->cfd_->internal_comparator().Compare(start, end) <= 0); uint64_t size = 0; const auto* vstorage = v->storage_info(); end_level = end_level == -1 ? vstorage->num_non_empty_levels() : std::min(end_level, vstorage->num_non_empty_levels()); assert(start_level <= end_level); for (int level = start_level; level < end_level; level++) { const LevelFilesBrief& files_brief = vstorage->LevelFilesBrief(level); if (!files_brief.num_files) { // empty level, skip exploration continue; } if (!level) { // level 0 data is sorted order, handle the use case explicitly size += ApproximateSizeLevel0(v, files_brief, start, end); continue; } assert(level > 0); assert(files_brief.num_files > 0); // identify the file position for starting key const uint64_t idx_start = FindFileInRange( v->cfd_->internal_comparator(), files_brief, start, /*start=*/0, static_cast(files_brief.num_files - 1)); assert(idx_start < files_brief.num_files); // scan all files from the starting position until the ending position // inferred from the sorted order for (uint64_t i = idx_start; i < files_brief.num_files; i++) { uint64_t val; val = ApproximateSize(v, files_brief.files[i], end); if (!val) { // the files after this will not have the range break; } size += val; if (i == idx_start) { // subtract the bytes needed to be scanned to get to the starting // key val = ApproximateSize(v, files_brief.files[i], start); assert(size >= val); size -= val; } } } return size; } uint64_t VersionSet::ApproximateSizeLevel0(Version* v, const LevelFilesBrief& files_brief, const Slice& key_start, const Slice& key_end) { // level 0 files are not in sorted order, we need to iterate through // the list to compute the total bytes that require scanning uint64_t size = 0; for (size_t i = 0; i < files_brief.num_files; i++) { const uint64_t start = ApproximateSize(v, files_brief.files[i], key_start); const uint64_t end = ApproximateSize(v, files_brief.files[i], key_end); assert(end >= start); size += end - start; } return size; } uint64_t VersionSet::ApproximateSize(Version* v, const FdWithKeyRange& f, const Slice& key) { // pre-condition assert(v); uint64_t result = 0; if (v->cfd_->internal_comparator().Compare(f.largest_key, key) <= 0) { // Entire file is before "key", so just add the file size result = f.fd.GetFileSize(); } else if (v->cfd_->internal_comparator().Compare(f.smallest_key, key) > 0) { // Entire file is after "key", so ignore result = 0; } else { // "key" falls in the range for this table. Add the // approximate offset of "key" within the table. TableReader* table_reader_ptr; InternalIterator* iter = v->cfd_->table_cache()->NewIterator( ReadOptions(), env_options_, v->cfd_->internal_comparator(), f.fd, &table_reader_ptr); if (table_reader_ptr != nullptr) { result = table_reader_ptr->ApproximateOffsetOf(key); } delete iter; } return result; } void VersionSet::AddLiveFiles(std::vector* live_list) { // pre-calculate space requirement int64_t total_files = 0; for (auto cfd : *column_family_set_) { Version* dummy_versions = cfd->dummy_versions(); for (Version* v = dummy_versions->next_; v != dummy_versions; v = v->next_) { const auto* vstorage = v->storage_info(); for (int level = 0; level < vstorage->num_levels(); level++) { total_files += vstorage->LevelFiles(level).size(); } } } // just one time extension to the right size live_list->reserve(live_list->size() + static_cast(total_files)); for (auto cfd : *column_family_set_) { auto* current = cfd->current(); bool found_current = false; Version* dummy_versions = cfd->dummy_versions(); for (Version* v = dummy_versions->next_; v != dummy_versions; v = v->next_) { v->AddLiveFiles(live_list); if (v == current) { found_current = true; } } if (!found_current && current != nullptr) { // Should never happen unless it is a bug. assert(false); current->AddLiveFiles(live_list); } } } InternalIterator* VersionSet::MakeInputIterator(Compaction* c) { auto cfd = c->column_family_data(); ReadOptions read_options; read_options.verify_checksums = c->mutable_cf_options()->verify_checksums_in_compaction; read_options.fill_cache = false; if (c->ShouldFormSubcompactions()) { read_options.total_order_seek = true; } // Level-0 files have to be merged together. For other levels, // we will make a concatenating iterator per level. // TODO(opt): use concatenating iterator for level-0 if there is no overlap const size_t space = (c->level() == 0 ? c->input_levels(0)->num_files + c->num_input_levels() - 1 : c->num_input_levels()); InternalIterator** list = new InternalIterator* [space]; size_t num = 0; for (size_t which = 0; which < c->num_input_levels(); which++) { if (c->input_levels(which)->num_files != 0) { if (c->level(which) == 0) { const LevelFilesBrief* flevel = c->input_levels(which); for (size_t i = 0; i < flevel->num_files; i++) { list[num++] = cfd->table_cache()->NewIterator( read_options, env_options_compactions_, cfd->internal_comparator(), flevel->files[i].fd, nullptr, nullptr, /* no per level latency histogram*/ true /* for compaction */); } } else { // Create concatenating iterator for the files from this level list[num++] = NewTwoLevelIterator( new LevelFileIteratorState( cfd->table_cache(), read_options, env_options_, cfd->internal_comparator(), nullptr /* no per level latency histogram */, true /* for_compaction */, false /* prefix enabled */, false /* skip_filters */), new LevelFileNumIterator(cfd->internal_comparator(), c->input_levels(which))); } } } assert(num <= space); InternalIterator* result = NewMergingIterator(&c->column_family_data()->internal_comparator(), list, static_cast(num)); delete[] list; return result; } // verify that the files listed in this compaction are present // in the current version bool VersionSet::VerifyCompactionFileConsistency(Compaction* c) { #ifndef NDEBUG Version* version = c->column_family_data()->current(); const VersionStorageInfo* vstorage = version->storage_info(); if (c->input_version() != version) { Log(InfoLogLevel::INFO_LEVEL, db_options_->info_log, "[%s] compaction output being applied to a different base version from" " input version", c->column_family_data()->GetName().c_str()); if (vstorage->compaction_style_ == kCompactionStyleLevel && c->start_level() == 0 && c->num_input_levels() > 2U) { // We are doing a L0->base_level compaction. The assumption is if // base level is not L1, levels from L1 to base_level - 1 is empty. // This is ensured by having one compaction from L0 going on at the // same time in level-based compaction. So that during the time, no // compaction/flush can put files to those levels. for (int l = c->start_level() + 1; l < c->output_level(); l++) { if (vstorage->NumLevelFiles(l) != 0) { return false; } } } } for (size_t input = 0; input < c->num_input_levels(); ++input) { int level = c->level(input); for (size_t i = 0; i < c->num_input_files(input); ++i) { uint64_t number = c->input(input, i)->fd.GetNumber(); bool found = false; for (size_t j = 0; j < vstorage->files_[level].size(); j++) { FileMetaData* f = vstorage->files_[level][j]; if (f->fd.GetNumber() == number) { found = true; break; } } if (!found) { return false; // input files non existent in current version } } } #endif return true; // everything good } Status VersionSet::GetMetadataForFile(uint64_t number, int* filelevel, FileMetaData** meta, ColumnFamilyData** cfd) { for (auto cfd_iter : *column_family_set_) { Version* version = cfd_iter->current(); const auto* vstorage = version->storage_info(); for (int level = 0; level < vstorage->num_levels(); level++) { for (const auto& file : vstorage->LevelFiles(level)) { if (file->fd.GetNumber() == number) { *meta = file; *filelevel = level; *cfd = cfd_iter; return Status::OK(); } } } } return Status::NotFound("File not present in any level"); } void VersionSet::GetLiveFilesMetaData(std::vector* metadata) { for (auto cfd : *column_family_set_) { if (cfd->IsDropped()) { continue; } for (int level = 0; level < cfd->NumberLevels(); level++) { for (const auto& file : cfd->current()->storage_info()->LevelFiles(level)) { LiveFileMetaData filemetadata; filemetadata.column_family_name = cfd->GetName(); uint32_t path_id = file->fd.GetPathId(); if (path_id < db_options_->db_paths.size()) { filemetadata.db_path = db_options_->db_paths[path_id].path; } else { assert(!db_options_->db_paths.empty()); filemetadata.db_path = db_options_->db_paths.back().path; } filemetadata.name = MakeTableFileName("", file->fd.GetNumber()); filemetadata.level = level; filemetadata.size = file->fd.GetFileSize(); filemetadata.smallestkey = file->smallest.user_key().ToString(); filemetadata.largestkey = file->largest.user_key().ToString(); filemetadata.smallest_seqno = file->smallest_seqno; filemetadata.largest_seqno = file->largest_seqno; metadata->push_back(filemetadata); } } } } void VersionSet::GetObsoleteFiles(std::vector* files, uint64_t min_pending_output) { std::vector pending_files; for (auto f : obsolete_files_) { if (f->fd.GetNumber() < min_pending_output) { files->push_back(f); } else { pending_files.push_back(f); } } obsolete_files_.swap(pending_files); } ColumnFamilyData* VersionSet::CreateColumnFamily( const ColumnFamilyOptions& cf_options, VersionEdit* edit) { assert(edit->is_column_family_add_); Version* dummy_versions = new Version(nullptr, this); // Ref() dummy version once so that later we can call Unref() to delete it // by avoiding calling "delete" explicitly (~Version is private) dummy_versions->Ref(); auto new_cfd = column_family_set_->CreateColumnFamily( edit->column_family_name_, edit->column_family_, dummy_versions, cf_options); Version* v = new Version(new_cfd, this, current_version_number_++); // Fill level target base information. v->storage_info()->CalculateBaseBytes(*new_cfd->ioptions(), *new_cfd->GetLatestMutableCFOptions()); AppendVersion(new_cfd, v); // GetLatestMutableCFOptions() is safe here without mutex since the // cfd is not available to client new_cfd->CreateNewMemtable(*new_cfd->GetLatestMutableCFOptions(), LastSequence()); new_cfd->SetLogNumber(edit->log_number_); return new_cfd; } uint64_t VersionSet::GetNumLiveVersions(Version* dummy_versions) { uint64_t count = 0; for (Version* v = dummy_versions->next_; v != dummy_versions; v = v->next_) { count++; } return count; } uint64_t VersionSet::GetTotalSstFilesSize(Version* dummy_versions) { std::unordered_set unique_files; uint64_t total_files_size = 0; for (Version* v = dummy_versions->next_; v != dummy_versions; v = v->next_) { VersionStorageInfo* storage_info = v->storage_info(); for (int level = 0; level < storage_info->num_levels_; level++) { for (const auto& file_meta : storage_info->LevelFiles(level)) { if (unique_files.find(file_meta->fd.packed_number_and_path_id) == unique_files.end()) { unique_files.insert(file_meta->fd.packed_number_and_path_id); total_files_size += file_meta->fd.GetFileSize(); } } } } return total_files_size; } } // namespace rocksdb rocksdb-4.5.1/db/version_set.h000066400000000000000000000730001267530634000162540ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // The representation of a DBImpl consists of a set of Versions. The // newest version is called "current". Older versions may be kept // around to provide a consistent view to live iterators. // // Each Version keeps track of a set of Table files per level. The // entire set of versions is maintained in a VersionSet. // // Version,VersionSet are thread-compatible, but require external // synchronization on all accesses. #pragma once #include #include #include #include #include #include #include #include #include "db/dbformat.h" #include "db/version_builder.h" #include "db/version_edit.h" #include "port/port.h" #include "db/table_cache.h" #include "db/compaction.h" #include "db/compaction_picker.h" #include "db/column_family.h" #include "db/log_reader.h" #include "db/file_indexer.h" #include "db/write_controller.h" #include "rocksdb/env.h" #include "util/instrumented_mutex.h" namespace rocksdb { namespace log { class Writer; } class Compaction; class InternalIterator; class LogBuffer; class LookupKey; class MemTable; class Version; class VersionSet; class WriteBuffer; class MergeContext; class ColumnFamilyData; class ColumnFamilySet; class TableCache; class MergeIteratorBuilder; // Return the smallest index i such that file_level.files[i]->largest >= key. // Return file_level.num_files if there is no such file. // REQUIRES: "file_level.files" contains a sorted list of // non-overlapping files. extern int FindFile(const InternalKeyComparator& icmp, const LevelFilesBrief& file_level, const Slice& key); // Returns true iff some file in "files" overlaps the user key range // [*smallest,*largest]. // smallest==nullptr represents a key smaller than all keys in the DB. // largest==nullptr represents a key largest than all keys in the DB. // REQUIRES: If disjoint_sorted_files, file_level.files[] // contains disjoint ranges in sorted order. extern bool SomeFileOverlapsRange(const InternalKeyComparator& icmp, bool disjoint_sorted_files, const LevelFilesBrief& file_level, const Slice* smallest_user_key, const Slice* largest_user_key); // Generate LevelFilesBrief from vector // Would copy smallest_key and largest_key data to sequential memory // arena: Arena used to allocate the memory extern void DoGenerateLevelFilesBrief(LevelFilesBrief* file_level, const std::vector& files, Arena* arena); class VersionStorageInfo { public: VersionStorageInfo(const InternalKeyComparator* internal_comparator, const Comparator* user_comparator, int num_levels, CompactionStyle compaction_style, VersionStorageInfo* src_vstorage); ~VersionStorageInfo(); void Reserve(int level, size_t size) { files_[level].reserve(size); } void AddFile(int level, FileMetaData* f, Logger* info_log = nullptr); void SetFinalized(); // Update num_non_empty_levels_. void UpdateNumNonEmptyLevels(); void GenerateFileIndexer() { file_indexer_.UpdateIndex(&arena_, num_non_empty_levels_, files_); } // Update the accumulated stats from a file-meta. void UpdateAccumulatedStats(FileMetaData* file_meta); // Decrease the current stat form a to-be-delected file-meta void RemoveCurrentStats(FileMetaData* file_meta); void ComputeCompensatedSizes(); // Updates internal structures that keep track of compaction scores // We use compaction scores to figure out which compaction to do next // REQUIRES: db_mutex held!! // TODO find a better way to pass compaction_options_fifo. void ComputeCompactionScore( const MutableCFOptions& mutable_cf_options, const CompactionOptionsFIFO& compaction_options_fifo); // Estimate est_comp_needed_bytes_ void EstimateCompactionBytesNeeded( const MutableCFOptions& mutable_cf_options); // This computes files_marked_for_compaction_ and is called by // ComputeCompactionScore() void ComputeFilesMarkedForCompaction(); // Generate level_files_brief_ from files_ void GenerateLevelFilesBrief(); // Sort all files for this version based on their file size and // record results in files_by_compaction_pri_. The largest files are listed // first. void UpdateFilesByCompactionPri(const MutableCFOptions& mutable_cf_options); void GenerateLevel0NonOverlapping(); bool level0_non_overlapping() const { return level0_non_overlapping_; } int MaxInputLevel() const; // Return level number that has idx'th highest score int CompactionScoreLevel(int idx) const { return compaction_level_[idx]; } // Return idx'th highest score double CompactionScore(int idx) const { return compaction_score_[idx]; } void GetOverlappingInputs( int level, const InternalKey* begin, // nullptr means before all keys const InternalKey* end, // nullptr means after all keys std::vector* inputs, int hint_index = -1, // index of overlap file int* file_index = nullptr, // return index of overlap file bool expand_range = true) // if set, returns files which overlap the const; // range and overlap each other. If false, // then just files intersecting the range void GetOverlappingInputsBinarySearch( int level, const Slice& begin, // nullptr means before all keys const Slice& end, // nullptr means after all keys std::vector* inputs, int hint_index, // index of overlap file int* file_index) const; // return index of overlap file void ExtendOverlappingInputs( int level, const Slice& begin, // nullptr means before all keys const Slice& end, // nullptr means after all keys std::vector* inputs, unsigned int index) const; // start extending from this index // Returns true iff some file in the specified level overlaps // some part of [*smallest_user_key,*largest_user_key]. // smallest_user_key==NULL represents a key smaller than all keys in the DB. // largest_user_key==NULL represents a key largest than all keys in the DB. bool OverlapInLevel(int level, const Slice* smallest_user_key, const Slice* largest_user_key); // Returns true iff the first or last file in inputs contains // an overlapping user key to the file "just outside" of it (i.e. // just after the last file, or just before the first file) // REQUIRES: "*inputs" is a sorted list of non-overlapping files bool HasOverlappingUserKey(const std::vector* inputs, int level); int num_levels() const { return num_levels_; } // REQUIRES: This version has been saved (see VersionSet::SaveTo) int num_non_empty_levels() const { assert(finalized_); return num_non_empty_levels_; } // REQUIRES: This version has been finalized. // (CalculateBaseBytes() is called) // This may or may not return number of level files. It is to keep backward // compatible behavior in universal compaction. int l0_delay_trigger_count() const { return l0_delay_trigger_count_; } void set_l0_delay_trigger_count(int v) { l0_delay_trigger_count_ = v; } // REQUIRES: This version has been saved (see VersionSet::SaveTo) int NumLevelFiles(int level) const { assert(finalized_); return static_cast(files_[level].size()); } // Return the combined file size of all files at the specified level. uint64_t NumLevelBytes(int level) const; // REQUIRES: This version has been saved (see VersionSet::SaveTo) const std::vector& LevelFiles(int level) const { return files_[level]; } const rocksdb::LevelFilesBrief& LevelFilesBrief(int level) const { assert(level < static_cast(level_files_brief_.size())); return level_files_brief_[level]; } // REQUIRES: This version has been saved (see VersionSet::SaveTo) const std::vector& FilesByCompactionPri(int level) const { assert(finalized_); return files_by_compaction_pri_[level]; } // REQUIRES: This version has been saved (see VersionSet::SaveTo) // REQUIRES: DB mutex held during access const autovector>& FilesMarkedForCompaction() const { assert(finalized_); return files_marked_for_compaction_; } int base_level() const { return base_level_; } // REQUIRES: lock is held // Set the index that is used to offset into files_by_compaction_pri_ to find // the next compaction candidate file. void SetNextCompactionIndex(int level, int index) { next_file_to_compact_by_size_[level] = index; } // REQUIRES: lock is held int NextCompactionIndex(int level) const { return next_file_to_compact_by_size_[level]; } // REQUIRES: This version has been saved (see VersionSet::SaveTo) const FileIndexer& file_indexer() const { assert(finalized_); return file_indexer_; } // Only the first few entries of files_by_compaction_pri_ are sorted. // There is no need to sort all the files because it is likely // that on a running system, we need to look at only the first // few largest files because a new version is created every few // seconds/minutes (because of concurrent compactions). static const size_t kNumberFilesToSort = 50; // Return a human-readable short (single-line) summary of the number // of files per level. Uses *scratch as backing store. struct LevelSummaryStorage { char buffer[1000]; }; struct FileSummaryStorage { char buffer[3000]; }; const char* LevelSummary(LevelSummaryStorage* scratch) const; // Return a human-readable short (single-line) summary of files // in a specified level. Uses *scratch as backing store. const char* LevelFileSummary(FileSummaryStorage* scratch, int level) const; // Return the maximum overlapping data (in bytes) at next level for any // file at a level >= 1. int64_t MaxNextLevelOverlappingBytes(); // Return a human readable string that describes this version's contents. std::string DebugString(bool hex = false) const; uint64_t GetAverageValueSize() const { if (accumulated_num_non_deletions_ == 0) { return 0; } assert(accumulated_raw_key_size_ + accumulated_raw_value_size_ > 0); assert(accumulated_file_size_ > 0); return accumulated_raw_value_size_ / accumulated_num_non_deletions_ * accumulated_file_size_ / (accumulated_raw_key_size_ + accumulated_raw_value_size_); } uint64_t GetEstimatedActiveKeys() const; // re-initializes the index that is used to offset into // files_by_compaction_pri_ // to find the next compaction candidate file. void ResetNextCompactionIndex(int level) { next_file_to_compact_by_size_[level] = 0; } const InternalKeyComparator* InternalComparator() { return internal_comparator_; } // Returns maximum total bytes of data on a given level. uint64_t MaxBytesForLevel(int level) const; // Must be called after any change to MutableCFOptions. void CalculateBaseBytes(const ImmutableCFOptions& ioptions, const MutableCFOptions& options); // Returns an estimate of the amount of live data in bytes. uint64_t EstimateLiveDataSize() const; uint64_t estimated_compaction_needed_bytes() const { return estimated_compaction_needed_bytes_; } void TEST_set_estimated_compaction_needed_bytes(uint64_t v) { estimated_compaction_needed_bytes_ = v; } private: const InternalKeyComparator* internal_comparator_; const Comparator* user_comparator_; int num_levels_; // Number of levels int num_non_empty_levels_; // Number of levels. Any level larger than it // is guaranteed to be empty. // Per-level max bytes std::vector level_max_bytes_; // A short brief metadata of files per level autovector level_files_brief_; FileIndexer file_indexer_; Arena arena_; // Used to allocate space for file_levels_ CompactionStyle compaction_style_; // List of files per level, files in each level are arranged // in increasing order of keys std::vector* files_; // Level that L0 data should be compacted to. All levels < base_level_ should // be empty. -1 if it is not level-compaction so it's not applicable. int base_level_; // A list for the same set of files that are stored in files_, // but files in each level are now sorted based on file // size. The file with the largest size is at the front. // This vector stores the index of the file from files_. std::vector> files_by_compaction_pri_; // If true, means that files in L0 have keys with non overlapping ranges bool level0_non_overlapping_; // An index into files_by_compaction_pri_ that specifies the first // file that is not yet compacted std::vector next_file_to_compact_by_size_; // Only the first few entries of files_by_compaction_pri_ are sorted. // There is no need to sort all the files because it is likely // that on a running system, we need to look at only the first // few largest files because a new version is created every few // seconds/minutes (because of concurrent compactions). static const size_t number_of_files_to_sort_ = 50; // This vector contains list of files marked for compaction and also not // currently being compacted. It is protected by DB mutex. It is calculated in // ComputeCompactionScore() autovector> files_marked_for_compaction_; // Level that should be compacted next and its compaction score. // Score < 1 means compaction is not strictly needed. These fields // are initialized by Finalize(). // The most critical level to be compacted is listed first // These are used to pick the best compaction level std::vector compaction_score_; std::vector compaction_level_; int l0_delay_trigger_count_ = 0; // Count used to trigger slow down and stop // for number of L0 files. // the following are the sampled temporary stats. // the current accumulated size of sampled files. uint64_t accumulated_file_size_; // the current accumulated size of all raw keys based on the sampled files. uint64_t accumulated_raw_key_size_; // the current accumulated size of all raw keys based on the sampled files. uint64_t accumulated_raw_value_size_; // total number of non-deletion entries uint64_t accumulated_num_non_deletions_; // total number of deletion entries uint64_t accumulated_num_deletions_; // current number of non_deletion entries uint64_t current_num_non_deletions_; // current number of delection entries uint64_t current_num_deletions_; // current number of file samples uint64_t current_num_samples_; // Estimated bytes needed to be compacted until all levels' size is down to // target sizes. uint64_t estimated_compaction_needed_bytes_; bool finalized_; friend class Version; friend class VersionSet; // No copying allowed VersionStorageInfo(const VersionStorageInfo&) = delete; void operator=(const VersionStorageInfo&) = delete; }; class Version { public: // Append to *iters a sequence of iterators that will // yield the contents of this Version when merged together. // REQUIRES: This version has been saved (see VersionSet::SaveTo) void AddIterators(const ReadOptions&, const EnvOptions& soptions, MergeIteratorBuilder* merger_iter_builder); // Lookup the value for key. If found, store it in *val and // return OK. Else return a non-OK status. // Uses *operands to store merge_operator operations to apply later. // // If the ReadOptions.read_tier is set to do a read-only fetch, then // *value_found will be set to false if it cannot be determined whether // this value exists without doing IO. // // If the key is Deleted, *status will be set to NotFound and // *key_exists will be set to true. // If no key was found, *status will be set to NotFound and // *key_exists will be set to false. // If seq is non-null, *seq will be set to the sequence number found // for the key if a key was found. // // REQUIRES: lock is not held void Get(const ReadOptions&, const LookupKey& key, std::string* val, Status* status, MergeContext* merge_context, bool* value_found = nullptr, bool* key_exists = nullptr, SequenceNumber* seq = nullptr); // Loads some stats information from files. Call without mutex held. It needs // to be called before applying the version to the version set. void PrepareApply(const MutableCFOptions& mutable_cf_options, bool update_stats); // Reference count management (so Versions do not disappear out from // under live iterators) void Ref(); // Decrease reference count. Delete the object if no reference left // and return true. Otherwise, return false. bool Unref(); // Add all files listed in the current version to *live. void AddLiveFiles(std::vector* live); // Return a human readable string that describes this version's contents. std::string DebugString(bool hex = false) const; // Returns the version nuber of this version uint64_t GetVersionNumber() const { return version_number_; } // REQUIRES: lock is held // On success, "tp" will contains the table properties of the file // specified in "file_meta". If the file name of "file_meta" is // known ahread, passing it by a non-null "fname" can save a // file-name conversion. Status GetTableProperties(std::shared_ptr* tp, const FileMetaData* file_meta, const std::string* fname = nullptr) const; // REQUIRES: lock is held // On success, *props will be populated with all SSTables' table properties. // The keys of `props` are the sst file name, the values of `props` are the // tables' propertis, represented as shared_ptr. Status GetPropertiesOfAllTables(TablePropertiesCollection* props); Status GetPropertiesOfAllTables(TablePropertiesCollection* props, int level); Status GetPropertiesOfTablesInRange(const Range* range, std::size_t n, TablePropertiesCollection* props) const; // REQUIRES: lock is held // On success, "tp" will contains the aggregated table property amoug // the table properties of all sst files in this version. Status GetAggregatedTableProperties( std::shared_ptr* tp, int level = -1); uint64_t GetEstimatedActiveKeys() { return storage_info_.GetEstimatedActiveKeys(); } size_t GetMemoryUsageByTableReaders(); ColumnFamilyData* cfd() const { return cfd_; } // Return the next Version in the linked list. Used for debug only Version* TEST_Next() const { return next_; } VersionStorageInfo* storage_info() { return &storage_info_; } VersionSet* version_set() { return vset_; } void GetColumnFamilyMetaData(ColumnFamilyMetaData* cf_meta); private: Env* env_; friend class VersionSet; const InternalKeyComparator* internal_comparator() const { return storage_info_.internal_comparator_; } const Comparator* user_comparator() const { return storage_info_.user_comparator_; } bool PrefixMayMatch(const ReadOptions& read_options, InternalIterator* level_iter, const Slice& internal_prefix) const; // Returns true if the filter blocks in the specified level will not be // checked during read operations. In certain cases (trivial move or preload), // the filter block may already be cached, but we still do not access it such // that it eventually expires from the cache. bool IsFilterSkipped(int level, bool is_file_last_in_level = false); // The helper function of UpdateAccumulatedStats, which may fill the missing // fields of file_mata from its associated TableProperties. // Returns true if it does initialize FileMetaData. bool MaybeInitializeFileMetaData(FileMetaData* file_meta); // Update the accumulated stats associated with the current version. // This accumulated stats will be used in compaction. void UpdateAccumulatedStats(bool update_stats); // Sort all files for this version based on their file size and // record results in files_by_compaction_pri_. The largest files are listed // first. void UpdateFilesByCompactionPri(); ColumnFamilyData* cfd_; // ColumnFamilyData to which this Version belongs Logger* info_log_; Statistics* db_statistics_; TableCache* table_cache_; const MergeOperator* merge_operator_; VersionStorageInfo storage_info_; VersionSet* vset_; // VersionSet to which this Version belongs Version* next_; // Next version in linked list Version* prev_; // Previous version in linked list int refs_; // Number of live refs to this version // A version number that uniquely represents this version. This is // used for debugging and logging purposes only. uint64_t version_number_; Version(ColumnFamilyData* cfd, VersionSet* vset, uint64_t version_number = 0); ~Version(); // No copying allowed Version(const Version&); void operator=(const Version&); }; class VersionSet { public: VersionSet(const std::string& dbname, const DBOptions* db_options, const EnvOptions& env_options, Cache* table_cache, WriteBuffer* write_buffer, WriteController* write_controller); ~VersionSet(); // Apply *edit to the current version to form a new descriptor that // is both saved to persistent state and installed as the new // current version. Will release *mu while actually writing to the file. // column_family_options has to be set if edit is column family add // REQUIRES: *mu is held on entry. // REQUIRES: no other thread concurrently calls LogAndApply() Status LogAndApply( ColumnFamilyData* column_family_data, const MutableCFOptions& mutable_cf_options, VersionEdit* edit, InstrumentedMutex* mu, Directory* db_directory = nullptr, bool new_descriptor_log = false, const ColumnFamilyOptions* column_family_options = nullptr); // Recover the last saved descriptor from persistent storage. // If read_only == true, Recover() will not complain if some column families // are not opened Status Recover(const std::vector& column_families, bool read_only = false); // Reads a manifest file and returns a list of column families in // column_families. static Status ListColumnFamilies(std::vector* column_families, const std::string& dbname, Env* env); #ifndef ROCKSDB_LITE // Try to reduce the number of levels. This call is valid when // only one level from the new max level to the old // max level containing files. // The call is static, since number of levels is immutable during // the lifetime of a RocksDB instance. It reduces number of levels // in a DB by applying changes to manifest. // For example, a db currently has 7 levels [0-6], and a call to // to reduce to 5 [0-4] can only be executed when only one level // among [4-6] contains files. static Status ReduceNumberOfLevels(const std::string& dbname, const Options* options, const EnvOptions& env_options, int new_levels); // printf contents (for debugging) Status DumpManifest(Options& options, std::string& manifestFileName, bool verbose, bool hex = false, bool json = false); #endif // ROCKSDB_LITE // Return the current manifest file number uint64_t manifest_file_number() const { return manifest_file_number_; } uint64_t pending_manifest_file_number() const { return pending_manifest_file_number_; } uint64_t current_next_file_number() const { return next_file_number_.load(); } // Allocate and return a new file number uint64_t NewFileNumber() { return next_file_number_.fetch_add(1); } // Return the last sequence number. uint64_t LastSequence() const { return last_sequence_.load(std::memory_order_acquire); } // Set the last sequence number to s. void SetLastSequence(uint64_t s) { assert(s >= last_sequence_); last_sequence_.store(s, std::memory_order_release); } // Mark the specified file number as used. // REQUIRED: this is only called during single-threaded recovery void MarkFileNumberUsedDuringRecovery(uint64_t number); // Return the log file number for the log file that is currently // being compacted, or zero if there is no such log file. uint64_t prev_log_number() const { return prev_log_number_; } // Returns the minimum log number such that all // log numbers less than or equal to it can be deleted uint64_t MinLogNumber() const { uint64_t min_log_num = std::numeric_limits::max(); for (auto cfd : *column_family_set_) { // It's safe to ignore dropped column families here: // cfd->IsDropped() becomes true after the drop is persisted in MANIFEST. if (min_log_num > cfd->GetLogNumber() && !cfd->IsDropped()) { min_log_num = cfd->GetLogNumber(); } } return min_log_num; } // Create an iterator that reads over the compaction inputs for "*c". // The caller should delete the iterator when no longer needed. InternalIterator* MakeInputIterator(Compaction* c); // Add all files listed in any live version to *live. void AddLiveFiles(std::vector* live_list); // Return the approximate size of data to be scanned for range [start, end) // in levels [start_level, end_level). If end_level == 0 it will search // through all non-empty levels uint64_t ApproximateSize(Version* v, const Slice& start, const Slice& end, int start_level = 0, int end_level = -1); // Return the size of the current manifest file uint64_t manifest_file_size() const { return manifest_file_size_; } // verify that the files that we started with for a compaction // still exist in the current version and in the same original level. // This ensures that a concurrent compaction did not erroneously // pick the same files to compact. bool VerifyCompactionFileConsistency(Compaction* c); Status GetMetadataForFile(uint64_t number, int* filelevel, FileMetaData** metadata, ColumnFamilyData** cfd); // This function doesn't support leveldb SST filenames void GetLiveFilesMetaData(std::vector *metadata); void GetObsoleteFiles(std::vector* files, uint64_t min_pending_output); ColumnFamilySet* GetColumnFamilySet() { return column_family_set_.get(); } const EnvOptions& env_options() { return env_options_; } static uint64_t GetNumLiveVersions(Version* dummy_versions); static uint64_t GetTotalSstFilesSize(Version* dummy_versions); private: struct ManifestWriter; friend class Version; friend class DBImpl; struct LogReporter : public log::Reader::Reporter { Status* status; virtual void Corruption(size_t bytes, const Status& s) override { if (this->status->ok()) *this->status = s; } }; // ApproximateSize helper uint64_t ApproximateSizeLevel0(Version* v, const LevelFilesBrief& files_brief, const Slice& start, const Slice& end); uint64_t ApproximateSize(Version* v, const FdWithKeyRange& f, const Slice& key); // Save current contents to *log Status WriteSnapshot(log::Writer* log); void AppendVersion(ColumnFamilyData* column_family_data, Version* v); ColumnFamilyData* CreateColumnFamily(const ColumnFamilyOptions& cf_options, VersionEdit* edit); std::unique_ptr column_family_set_; Env* const env_; const std::string dbname_; const DBOptions* const db_options_; std::atomic next_file_number_; uint64_t manifest_file_number_; uint64_t pending_manifest_file_number_; std::atomic last_sequence_; uint64_t prev_log_number_; // 0 or backing store for memtable being compacted // Opened lazily unique_ptr descriptor_log_; // generates a increasing version number for every new version uint64_t current_version_number_; // Queue of writers to the manifest file std::deque manifest_writers_; // Current size of manifest file uint64_t manifest_file_size_; std::vector obsolete_files_; // env options for all reads and writes except compactions const EnvOptions& env_options_; // env options used for compactions. This is a copy of // env_options_ but with readaheads set to readahead_compactions_. const EnvOptions env_options_compactions_; // No copying allowed VersionSet(const VersionSet&); void operator=(const VersionSet&); void LogAndApplyCFHelper(VersionEdit* edit); void LogAndApplyHelper(ColumnFamilyData* cfd, VersionBuilder* b, Version* v, VersionEdit* edit, InstrumentedMutex* mu); }; } // namespace rocksdb rocksdb-4.5.1/db/version_set_test.cc000066400000000000000000000351741267530634000174630ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/version_set.h" #include "util/logging.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class GenerateLevelFilesBriefTest : public testing::Test { public: std::vector files_; LevelFilesBrief file_level_; Arena arena_; GenerateLevelFilesBriefTest() { } ~GenerateLevelFilesBriefTest() { for (size_t i = 0; i < files_.size(); i++) { delete files_[i]; } } void Add(const char* smallest, const char* largest, SequenceNumber smallest_seq = 100, SequenceNumber largest_seq = 100) { FileMetaData* f = new FileMetaData; f->fd = FileDescriptor(files_.size() + 1, 0, 0); f->smallest = InternalKey(smallest, smallest_seq, kTypeValue); f->largest = InternalKey(largest, largest_seq, kTypeValue); files_.push_back(f); } int Compare() { int diff = 0; for (size_t i = 0; i < files_.size(); i++) { if (file_level_.files[i].fd.GetNumber() != files_[i]->fd.GetNumber()) { diff++; } } return diff; } }; TEST_F(GenerateLevelFilesBriefTest, Empty) { DoGenerateLevelFilesBrief(&file_level_, files_, &arena_); ASSERT_EQ(0u, file_level_.num_files); ASSERT_EQ(0, Compare()); } TEST_F(GenerateLevelFilesBriefTest, Single) { Add("p", "q"); DoGenerateLevelFilesBrief(&file_level_, files_, &arena_); ASSERT_EQ(1u, file_level_.num_files); ASSERT_EQ(0, Compare()); } TEST_F(GenerateLevelFilesBriefTest, Multiple) { Add("150", "200"); Add("200", "250"); Add("300", "350"); Add("400", "450"); DoGenerateLevelFilesBrief(&file_level_, files_, &arena_); ASSERT_EQ(4u, file_level_.num_files); ASSERT_EQ(0, Compare()); } class CountingLogger : public Logger { public: CountingLogger() : log_count(0) {} using Logger::Logv; virtual void Logv(const char* format, va_list ap) override { log_count++; } int log_count; }; Options GetOptionsWithNumLevels(int num_levels, std::shared_ptr logger) { Options opt; opt.num_levels = num_levels; opt.info_log = logger; return opt; } class VersionStorageInfoTest : public testing::Test { public: const Comparator* ucmp_; InternalKeyComparator icmp_; std::shared_ptr logger_; Options options_; ImmutableCFOptions ioptions_; MutableCFOptions mutable_cf_options_; VersionStorageInfo vstorage_; InternalKey GetInternalKey(const char* ukey, SequenceNumber smallest_seq = 100) { return InternalKey(ukey, smallest_seq, kTypeValue); } VersionStorageInfoTest() : ucmp_(BytewiseComparator()), icmp_(ucmp_), logger_(new CountingLogger()), options_(GetOptionsWithNumLevels(6, logger_)), ioptions_(options_), mutable_cf_options_(options_, ioptions_), vstorage_(&icmp_, ucmp_, 6, kCompactionStyleLevel, nullptr) {} ~VersionStorageInfoTest() { for (int i = 0; i < vstorage_.num_levels(); i++) { for (auto* f : vstorage_.LevelFiles(i)) { if (--f->refs == 0) { delete f; } } } } void Add(int level, uint32_t file_number, const char* smallest, const char* largest, uint64_t file_size = 0) { assert(level < vstorage_.num_levels()); FileMetaData* f = new FileMetaData; f->fd = FileDescriptor(file_number, 0, file_size); f->smallest = GetInternalKey(smallest, 0); f->largest = GetInternalKey(largest, 0); f->compensated_file_size = file_size; f->refs = 0; f->num_entries = 0; f->num_deletions = 0; vstorage_.AddFile(level, f); } }; TEST_F(VersionStorageInfoTest, MaxBytesForLevelStatic) { ioptions_.level_compaction_dynamic_level_bytes = false; mutable_cf_options_.max_bytes_for_level_base = 10; mutable_cf_options_.max_bytes_for_level_multiplier = 5; Add(4, 100U, "1", "2"); Add(5, 101U, "1", "2"); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(vstorage_.MaxBytesForLevel(1), 10U); ASSERT_EQ(vstorage_.MaxBytesForLevel(2), 50U); ASSERT_EQ(vstorage_.MaxBytesForLevel(3), 250U); ASSERT_EQ(vstorage_.MaxBytesForLevel(4), 1250U); ASSERT_EQ(0, logger_->log_count); } TEST_F(VersionStorageInfoTest, MaxBytesForLevelDynamic) { ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.max_bytes_for_level_base = 1000; mutable_cf_options_.max_bytes_for_level_multiplier = 5; Add(5, 1U, "1", "2", 500U); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(0, logger_->log_count); ASSERT_EQ(vstorage_.base_level(), 5); Add(5, 2U, "3", "4", 550U); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(0, logger_->log_count); ASSERT_EQ(vstorage_.MaxBytesForLevel(4), 210U); ASSERT_EQ(vstorage_.base_level(), 4); Add(4, 3U, "3", "4", 550U); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(0, logger_->log_count); ASSERT_EQ(vstorage_.MaxBytesForLevel(4), 210U); ASSERT_EQ(vstorage_.base_level(), 4); Add(3, 4U, "3", "4", 250U); Add(3, 5U, "5", "7", 300U); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(1, logger_->log_count); ASSERT_EQ(vstorage_.MaxBytesForLevel(4), 1005U); ASSERT_EQ(vstorage_.MaxBytesForLevel(3), 201U); ASSERT_EQ(vstorage_.base_level(), 3); Add(1, 6U, "3", "4", 5U); Add(1, 7U, "8", "9", 5U); logger_->log_count = 0; vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(1, logger_->log_count); ASSERT_GT(vstorage_.MaxBytesForLevel(4), 1005U); ASSERT_GT(vstorage_.MaxBytesForLevel(3), 1005U); ASSERT_EQ(vstorage_.MaxBytesForLevel(2), 1005U); ASSERT_EQ(vstorage_.MaxBytesForLevel(1), 201U); ASSERT_EQ(vstorage_.base_level(), 1); } TEST_F(VersionStorageInfoTest, MaxBytesForLevelDynamicLotsOfData) { ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.max_bytes_for_level_base = 100; mutable_cf_options_.max_bytes_for_level_multiplier = 2; Add(0, 1U, "1", "2", 50U); Add(1, 2U, "1", "2", 50U); Add(2, 3U, "1", "2", 500U); Add(3, 4U, "1", "2", 500U); Add(4, 5U, "1", "2", 1700U); Add(5, 6U, "1", "2", 500U); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(vstorage_.MaxBytesForLevel(4), 800U); ASSERT_EQ(vstorage_.MaxBytesForLevel(3), 400U); ASSERT_EQ(vstorage_.MaxBytesForLevel(2), 200U); ASSERT_EQ(vstorage_.MaxBytesForLevel(1), 100U); ASSERT_EQ(vstorage_.base_level(), 1); ASSERT_EQ(0, logger_->log_count); } TEST_F(VersionStorageInfoTest, MaxBytesForLevelDynamicLargeLevel) { uint64_t kOneGB = 1000U * 1000U * 1000U; ioptions_.level_compaction_dynamic_level_bytes = true; mutable_cf_options_.max_bytes_for_level_base = 10U * kOneGB; mutable_cf_options_.max_bytes_for_level_multiplier = 10; Add(0, 1U, "1", "2", 50U); Add(3, 4U, "1", "2", 32U * kOneGB); Add(4, 5U, "1", "2", 500U * kOneGB); Add(5, 6U, "1", "2", 3000U * kOneGB); vstorage_.CalculateBaseBytes(ioptions_, mutable_cf_options_); ASSERT_EQ(vstorage_.MaxBytesForLevel(5), 3000U * kOneGB); ASSERT_EQ(vstorage_.MaxBytesForLevel(4), 300U * kOneGB); ASSERT_EQ(vstorage_.MaxBytesForLevel(3), 30U * kOneGB); ASSERT_EQ(vstorage_.MaxBytesForLevel(2), 3U * kOneGB); ASSERT_EQ(vstorage_.base_level(), 2); ASSERT_EQ(0, logger_->log_count); } TEST_F(VersionStorageInfoTest, EstimateLiveDataSize) { // Test whether the overlaps are detected as expected Add(1, 1U, "4", "7", 1U); // Perfect overlap with last level Add(2, 2U, "3", "5", 1U); // Partial overlap with last level Add(2, 3U, "6", "8", 1U); // Partial overlap with last level Add(3, 4U, "1", "9", 1U); // Contains range of last level Add(4, 5U, "4", "5", 1U); // Inside range of last level Add(4, 5U, "6", "7", 1U); // Inside range of last level Add(5, 6U, "4", "7", 10U); ASSERT_EQ(10U, vstorage_.EstimateLiveDataSize()); } TEST_F(VersionStorageInfoTest, EstimateLiveDataSize2) { Add(0, 1U, "9", "9", 1U); // Level 0 is not ordered Add(0, 1U, "5", "6", 1U); // Ignored because of [5,6] in l1 Add(1, 1U, "1", "2", 1U); // Ignored because of [2,3] in l2 Add(1, 2U, "3", "4", 1U); // Ignored because of [2,3] in l2 Add(1, 3U, "5", "6", 1U); Add(2, 4U, "2", "3", 1U); Add(3, 5U, "7", "8", 1U); ASSERT_EQ(4U, vstorage_.EstimateLiveDataSize()); } class FindLevelFileTest : public testing::Test { public: LevelFilesBrief file_level_; bool disjoint_sorted_files_; Arena arena_; FindLevelFileTest() : disjoint_sorted_files_(true) { } ~FindLevelFileTest() { } void LevelFileInit(size_t num = 0) { char* mem = arena_.AllocateAligned(num * sizeof(FdWithKeyRange)); file_level_.files = new (mem)FdWithKeyRange[num]; file_level_.num_files = 0; } void Add(const char* smallest, const char* largest, SequenceNumber smallest_seq = 100, SequenceNumber largest_seq = 100) { InternalKey smallest_key = InternalKey(smallest, smallest_seq, kTypeValue); InternalKey largest_key = InternalKey(largest, largest_seq, kTypeValue); Slice smallest_slice = smallest_key.Encode(); Slice largest_slice = largest_key.Encode(); char* mem = arena_.AllocateAligned( smallest_slice.size() + largest_slice.size()); memcpy(mem, smallest_slice.data(), smallest_slice.size()); memcpy(mem + smallest_slice.size(), largest_slice.data(), largest_slice.size()); // add to file_level_ size_t num = file_level_.num_files; auto& file = file_level_.files[num]; file.fd = FileDescriptor(num + 1, 0, 0); file.smallest_key = Slice(mem, smallest_slice.size()); file.largest_key = Slice(mem + smallest_slice.size(), largest_slice.size()); file_level_.num_files++; } int Find(const char* key) { InternalKey target(key, 100, kTypeValue); InternalKeyComparator cmp(BytewiseComparator()); return FindFile(cmp, file_level_, target.Encode()); } bool Overlaps(const char* smallest, const char* largest) { InternalKeyComparator cmp(BytewiseComparator()); Slice s(smallest != nullptr ? smallest : ""); Slice l(largest != nullptr ? largest : ""); return SomeFileOverlapsRange(cmp, disjoint_sorted_files_, file_level_, (smallest != nullptr ? &s : nullptr), (largest != nullptr ? &l : nullptr)); } }; TEST_F(FindLevelFileTest, LevelEmpty) { LevelFileInit(0); ASSERT_EQ(0, Find("foo")); ASSERT_TRUE(! Overlaps("a", "z")); ASSERT_TRUE(! Overlaps(nullptr, "z")); ASSERT_TRUE(! Overlaps("a", nullptr)); ASSERT_TRUE(! Overlaps(nullptr, nullptr)); } TEST_F(FindLevelFileTest, LevelSingle) { LevelFileInit(1); Add("p", "q"); ASSERT_EQ(0, Find("a")); ASSERT_EQ(0, Find("p")); ASSERT_EQ(0, Find("p1")); ASSERT_EQ(0, Find("q")); ASSERT_EQ(1, Find("q1")); ASSERT_EQ(1, Find("z")); ASSERT_TRUE(! Overlaps("a", "b")); ASSERT_TRUE(! Overlaps("z1", "z2")); ASSERT_TRUE(Overlaps("a", "p")); ASSERT_TRUE(Overlaps("a", "q")); ASSERT_TRUE(Overlaps("a", "z")); ASSERT_TRUE(Overlaps("p", "p1")); ASSERT_TRUE(Overlaps("p", "q")); ASSERT_TRUE(Overlaps("p", "z")); ASSERT_TRUE(Overlaps("p1", "p2")); ASSERT_TRUE(Overlaps("p1", "z")); ASSERT_TRUE(Overlaps("q", "q")); ASSERT_TRUE(Overlaps("q", "q1")); ASSERT_TRUE(! Overlaps(nullptr, "j")); ASSERT_TRUE(! Overlaps("r", nullptr)); ASSERT_TRUE(Overlaps(nullptr, "p")); ASSERT_TRUE(Overlaps(nullptr, "p1")); ASSERT_TRUE(Overlaps("q", nullptr)); ASSERT_TRUE(Overlaps(nullptr, nullptr)); } TEST_F(FindLevelFileTest, LevelMultiple) { LevelFileInit(4); Add("150", "200"); Add("200", "250"); Add("300", "350"); Add("400", "450"); ASSERT_EQ(0, Find("100")); ASSERT_EQ(0, Find("150")); ASSERT_EQ(0, Find("151")); ASSERT_EQ(0, Find("199")); ASSERT_EQ(0, Find("200")); ASSERT_EQ(1, Find("201")); ASSERT_EQ(1, Find("249")); ASSERT_EQ(1, Find("250")); ASSERT_EQ(2, Find("251")); ASSERT_EQ(2, Find("299")); ASSERT_EQ(2, Find("300")); ASSERT_EQ(2, Find("349")); ASSERT_EQ(2, Find("350")); ASSERT_EQ(3, Find("351")); ASSERT_EQ(3, Find("400")); ASSERT_EQ(3, Find("450")); ASSERT_EQ(4, Find("451")); ASSERT_TRUE(! Overlaps("100", "149")); ASSERT_TRUE(! Overlaps("251", "299")); ASSERT_TRUE(! Overlaps("451", "500")); ASSERT_TRUE(! Overlaps("351", "399")); ASSERT_TRUE(Overlaps("100", "150")); ASSERT_TRUE(Overlaps("100", "200")); ASSERT_TRUE(Overlaps("100", "300")); ASSERT_TRUE(Overlaps("100", "400")); ASSERT_TRUE(Overlaps("100", "500")); ASSERT_TRUE(Overlaps("375", "400")); ASSERT_TRUE(Overlaps("450", "450")); ASSERT_TRUE(Overlaps("450", "500")); } TEST_F(FindLevelFileTest, LevelMultipleNullBoundaries) { LevelFileInit(4); Add("150", "200"); Add("200", "250"); Add("300", "350"); Add("400", "450"); ASSERT_TRUE(! Overlaps(nullptr, "149")); ASSERT_TRUE(! Overlaps("451", nullptr)); ASSERT_TRUE(Overlaps(nullptr, nullptr)); ASSERT_TRUE(Overlaps(nullptr, "150")); ASSERT_TRUE(Overlaps(nullptr, "199")); ASSERT_TRUE(Overlaps(nullptr, "200")); ASSERT_TRUE(Overlaps(nullptr, "201")); ASSERT_TRUE(Overlaps(nullptr, "400")); ASSERT_TRUE(Overlaps(nullptr, "800")); ASSERT_TRUE(Overlaps("100", nullptr)); ASSERT_TRUE(Overlaps("200", nullptr)); ASSERT_TRUE(Overlaps("449", nullptr)); ASSERT_TRUE(Overlaps("450", nullptr)); } TEST_F(FindLevelFileTest, LevelOverlapSequenceChecks) { LevelFileInit(1); Add("200", "200", 5000, 3000); ASSERT_TRUE(! Overlaps("199", "199")); ASSERT_TRUE(! Overlaps("201", "300")); ASSERT_TRUE(Overlaps("200", "200")); ASSERT_TRUE(Overlaps("190", "200")); ASSERT_TRUE(Overlaps("200", "210")); } TEST_F(FindLevelFileTest, LevelOverlappingFiles) { LevelFileInit(2); Add("150", "600"); Add("400", "500"); disjoint_sorted_files_ = false; ASSERT_TRUE(! Overlaps("100", "149")); ASSERT_TRUE(! Overlaps("601", "700")); ASSERT_TRUE(Overlaps("100", "150")); ASSERT_TRUE(Overlaps("100", "200")); ASSERT_TRUE(Overlaps("100", "300")); ASSERT_TRUE(Overlaps("100", "400")); ASSERT_TRUE(Overlaps("100", "500")); ASSERT_TRUE(Overlaps("375", "400")); ASSERT_TRUE(Overlaps("450", "450")); ASSERT_TRUE(Overlaps("450", "500")); ASSERT_TRUE(Overlaps("450", "700")); ASSERT_TRUE(Overlaps("600", "700")); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/wal_manager.cc000066400000000000000000000374361267530634000163440ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/wal_manager.h" #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include "db/filename.h" #include "db/transaction_log_impl.h" #include "db/log_reader.h" #include "db/log_writer.h" #include "db/write_batch_internal.h" #include "port/port.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/write_batch.h" #include "util/coding.h" #include "util/file_reader_writer.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/sync_point.h" #include "util/string_util.h" namespace rocksdb { #ifndef ROCKSDB_LITE Status WalManager::GetSortedWalFiles(VectorLogPtr& files) { // First get sorted files in db dir, then get sorted files from archived // dir, to avoid a race condition where a log file is moved to archived // dir in between. Status s; // list wal files in main db dir. VectorLogPtr logs; s = GetSortedWalsOfType(db_options_.wal_dir, logs, kAliveLogFile); if (!s.ok()) { return s; } // Reproduce the race condition where a log file is moved // to archived dir, between these two sync points, used in // (DBTest,TransactionLogIteratorRace) TEST_SYNC_POINT("WalManager::GetSortedWalFiles:1"); TEST_SYNC_POINT("WalManager::GetSortedWalFiles:2"); files.clear(); // list wal files in archive dir. std::string archivedir = ArchivalDirectory(db_options_.wal_dir); Status exists = env_->FileExists(archivedir); if (exists.ok()) { s = GetSortedWalsOfType(archivedir, files, kArchivedLogFile); if (!s.ok()) { return s; } } else if (!exists.IsNotFound()) { assert(s.IsIOError()); return s; } uint64_t latest_archived_log_number = 0; if (!files.empty()) { latest_archived_log_number = files.back()->LogNumber(); Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Latest Archived log: %" PRIu64, latest_archived_log_number); } files.reserve(files.size() + logs.size()); for (auto& log : logs) { if (log->LogNumber() > latest_archived_log_number) { files.push_back(std::move(log)); } else { // When the race condition happens, we could see the // same log in both db dir and archived dir. Simply // ignore the one in db dir. Note that, if we read // archived dir first, we would have missed the log file. Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "%s already moved to archive", log->PathName().c_str()); } } return s; } Status WalManager::GetUpdatesSince( SequenceNumber seq, std::unique_ptr* iter, const TransactionLogIterator::ReadOptions& read_options, VersionSet* version_set) { // Get all sorted Wal Files. // Do binary search and open files and find the seq number. std::unique_ptr wal_files(new VectorLogPtr); Status s = GetSortedWalFiles(*wal_files); if (!s.ok()) { return s; } s = RetainProbableWalFiles(*wal_files, seq); if (!s.ok()) { return s; } iter->reset(new TransactionLogIteratorImpl( db_options_.wal_dir, &db_options_, read_options, env_options_, seq, std::move(wal_files), version_set)); return (*iter)->status(); } // 1. Go through all archived files and // a. if ttl is enabled, delete outdated files // b. if archive size limit is enabled, delete empty files, // compute file number and size. // 2. If size limit is enabled: // a. compute how many files should be deleted // b. get sorted non-empty archived logs // c. delete what should be deleted void WalManager::PurgeObsoleteWALFiles() { bool const ttl_enabled = db_options_.WAL_ttl_seconds > 0; bool const size_limit_enabled = db_options_.WAL_size_limit_MB > 0; if (!ttl_enabled && !size_limit_enabled) { return; } int64_t current_time; Status s = env_->GetCurrentTime(¤t_time); if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Can't get current time: %s", s.ToString().c_str()); assert(false); return; } uint64_t const now_seconds = static_cast(current_time); uint64_t const time_to_check = (ttl_enabled && !size_limit_enabled) ? db_options_.WAL_ttl_seconds / 2 : kDefaultIntervalToDeleteObsoleteWAL; if (purge_wal_files_last_run_ + time_to_check > now_seconds) { return; } purge_wal_files_last_run_ = now_seconds; std::string archival_dir = ArchivalDirectory(db_options_.wal_dir); std::vector files; s = env_->GetChildren(archival_dir, &files); if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Can't get archive files: %s", s.ToString().c_str()); assert(false); return; } size_t log_files_num = 0; uint64_t log_file_size = 0; for (auto& f : files) { uint64_t number; FileType type; if (ParseFileName(f, &number, &type) && type == kLogFile) { std::string const file_path = archival_dir + "/" + f; if (ttl_enabled) { uint64_t file_m_time; s = env_->GetFileModificationTime(file_path, &file_m_time); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Can't get file mod time: %s: %s", file_path.c_str(), s.ToString().c_str()); continue; } if (now_seconds - file_m_time > db_options_.WAL_ttl_seconds) { s = env_->DeleteFile(file_path); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Can't delete file: %s: %s", file_path.c_str(), s.ToString().c_str()); continue; } else { MutexLock l(&read_first_record_cache_mutex_); read_first_record_cache_.erase(number); } continue; } } if (size_limit_enabled) { uint64_t file_size; s = env_->GetFileSize(file_path, &file_size); if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "Unable to get file size: %s: %s", file_path.c_str(), s.ToString().c_str()); return; } else { if (file_size > 0) { log_file_size = std::max(log_file_size, file_size); ++log_files_num; } else { s = env_->DeleteFile(file_path); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Unable to delete file: %s: %s", file_path.c_str(), s.ToString().c_str()); continue; } else { MutexLock l(&read_first_record_cache_mutex_); read_first_record_cache_.erase(number); } } } } } } if (0 == log_files_num || !size_limit_enabled) { return; } size_t const files_keep_num = db_options_.WAL_size_limit_MB * 1024 * 1024 / log_file_size; if (log_files_num <= files_keep_num) { return; } size_t files_del_num = log_files_num - files_keep_num; VectorLogPtr archived_logs; GetSortedWalsOfType(archival_dir, archived_logs, kArchivedLogFile); if (files_del_num > archived_logs.size()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Trying to delete more archived log files than " "exist. Deleting all"); files_del_num = archived_logs.size(); } for (size_t i = 0; i < files_del_num; ++i) { std::string const file_path = archived_logs[i]->PathName(); s = env_->DeleteFile(db_options_.wal_dir + "/" + file_path); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log, "Unable to delete file: %s: %s", file_path.c_str(), s.ToString().c_str()); continue; } else { MutexLock l(&read_first_record_cache_mutex_); read_first_record_cache_.erase(archived_logs[i]->LogNumber()); } } } void WalManager::ArchiveWALFile(const std::string& fname, uint64_t number) { auto archived_log_name = ArchivedLogFileName(db_options_.wal_dir, number); // The sync point below is used in (DBTest,TransactionLogIteratorRace) TEST_SYNC_POINT("WalManager::PurgeObsoleteFiles:1"); Status s = env_->RenameFile(fname, archived_log_name); // The sync point below is used in (DBTest,TransactionLogIteratorRace) TEST_SYNC_POINT("WalManager::PurgeObsoleteFiles:2"); Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log, "Move log file %s to %s -- %s\n", fname.c_str(), archived_log_name.c_str(), s.ToString().c_str()); } namespace { struct CompareLogByPointer { bool operator()(const std::unique_ptr& a, const std::unique_ptr& b) { LogFileImpl* a_impl = dynamic_cast(a.get()); LogFileImpl* b_impl = dynamic_cast(b.get()); return *a_impl < *b_impl; } }; } Status WalManager::GetSortedWalsOfType(const std::string& path, VectorLogPtr& log_files, WalFileType log_type) { std::vector all_files; const Status status = env_->GetChildren(path, &all_files); if (!status.ok()) { return status; } log_files.reserve(all_files.size()); for (const auto& f : all_files) { uint64_t number; FileType type; if (ParseFileName(f, &number, &type) && type == kLogFile) { SequenceNumber sequence; Status s = ReadFirstRecord(log_type, number, &sequence); if (!s.ok()) { return s; } if (sequence == 0) { // empty file continue; } // Reproduce the race condition where a log file is moved // to archived dir, between these two sync points, used in // (DBTest,TransactionLogIteratorRace) TEST_SYNC_POINT("WalManager::GetSortedWalsOfType:1"); TEST_SYNC_POINT("WalManager::GetSortedWalsOfType:2"); uint64_t size_bytes; s = env_->GetFileSize(LogFileName(path, number), &size_bytes); // re-try in case the alive log file has been moved to archive. std::string archived_file = ArchivedLogFileName(path, number); if (!s.ok() && log_type == kAliveLogFile && env_->FileExists(archived_file).ok()) { s = env_->GetFileSize(archived_file, &size_bytes); if (!s.ok() && env_->FileExists(archived_file).IsNotFound()) { // oops, the file just got deleted from archived dir! move on s = Status::OK(); continue; } } if (!s.ok()) { return s; } log_files.push_back(std::unique_ptr( new LogFileImpl(number, log_type, sequence, size_bytes))); } } CompareLogByPointer compare_log_files; std::sort(log_files.begin(), log_files.end(), compare_log_files); return status; } Status WalManager::RetainProbableWalFiles(VectorLogPtr& all_logs, const SequenceNumber target) { int64_t start = 0; // signed to avoid overflow when target is < first file. int64_t end = static_cast(all_logs.size()) - 1; // Binary Search. avoid opening all files. while (end >= start) { int64_t mid = start + (end - start) / 2; // Avoid overflow. SequenceNumber current_seq_num = all_logs.at(mid)->StartSequence(); if (current_seq_num == target) { end = mid; break; } else if (current_seq_num < target) { start = mid + 1; } else { end = mid - 1; } } // end could be -ve. size_t start_index = std::max(static_cast(0), end); // The last wal file is always included all_logs.erase(all_logs.begin(), all_logs.begin() + start_index); return Status::OK(); } Status WalManager::ReadFirstRecord(const WalFileType type, const uint64_t number, SequenceNumber* sequence) { *sequence = 0; if (type != kAliveLogFile && type != kArchivedLogFile) { Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log, "[WalManger] Unknown file type %s", ToString(type).c_str()); return Status::NotSupported( "File Type Not Known " + ToString(type)); } { MutexLock l(&read_first_record_cache_mutex_); auto itr = read_first_record_cache_.find(number); if (itr != read_first_record_cache_.end()) { *sequence = itr->second; return Status::OK(); } } Status s; if (type == kAliveLogFile) { std::string fname = LogFileName(db_options_.wal_dir, number); s = ReadFirstLine(fname, sequence); if (env_->FileExists(fname).ok() && !s.ok()) { // return any error that is not caused by non-existing file return s; } } if (type == kArchivedLogFile || !s.ok()) { // check if the file got moved to archive. std::string archived_file = ArchivedLogFileName(db_options_.wal_dir, number); s = ReadFirstLine(archived_file, sequence); // maybe the file was deleted from archive dir. If that's the case, return // Status::OK(). The caller with identify this as empty file because // *sequence == 0 if (!s.ok() && env_->FileExists(archived_file).IsNotFound()) { return Status::OK(); } } if (s.ok() && *sequence != 0) { MutexLock l(&read_first_record_cache_mutex_); read_first_record_cache_.insert({number, *sequence}); } return s; } // the function returns status.ok() and sequence == 0 if the file exists, but is // empty Status WalManager::ReadFirstLine(const std::string& fname, SequenceNumber* sequence) { struct LogReporter : public log::Reader::Reporter { Env* env; Logger* info_log; const char* fname; Status* status; bool ignore_error; // true if db_options_.paranoid_checks==false virtual void Corruption(size_t bytes, const Status& s) override { Log(InfoLogLevel::WARN_LEVEL, info_log, "[WalManager] %s%s: dropping %d bytes; %s", (this->ignore_error ? "(ignoring error) " : ""), fname, static_cast(bytes), s.ToString().c_str()); if (this->status->ok()) { // only keep the first error *this->status = s; } } }; std::unique_ptr file; Status status = env_->NewSequentialFile(fname, &file, env_options_); unique_ptr file_reader( new SequentialFileReader(std::move(file))); if (!status.ok()) { return status; } LogReporter reporter; reporter.env = env_; reporter.info_log = db_options_.info_log.get(); reporter.fname = fname.c_str(); reporter.status = &status; reporter.ignore_error = !db_options_.paranoid_checks; log::Reader reader(db_options_.info_log, std::move(file_reader), &reporter, true /*checksum*/, 0 /*initial_offset*/, *sequence); std::string scratch; Slice record; if (reader.ReadRecord(&record, &scratch) && (status.ok() || !db_options_.paranoid_checks)) { if (record.size() < 12) { reporter.Corruption(record.size(), Status::Corruption("log record too small")); // TODO read record's till the first no corrupt entry? } else { WriteBatch batch; WriteBatchInternal::SetContents(&batch, record); *sequence = WriteBatchInternal::Sequence(&batch); return Status::OK(); } } // ReadRecord returns false on EOF, which means that the log file is empty. we // return status.ok() in that case and set sequence number to 0 *sequence = 0; return status; } #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/wal_manager.h000066400000000000000000000061241267530634000161740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include #include #include #include "port/port.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/types.h" #include "rocksdb/transaction_log.h" #include "rocksdb/status.h" #include "db/version_set.h" namespace rocksdb { #ifndef ROCKSDB_LITE class WalManager { public: WalManager(const DBOptions& db_options, const EnvOptions& env_options) : db_options_(db_options), env_options_(env_options), env_(db_options.env), purge_wal_files_last_run_(0) {} Status GetSortedWalFiles(VectorLogPtr& files); Status GetUpdatesSince( SequenceNumber seq_number, std::unique_ptr* iter, const TransactionLogIterator::ReadOptions& read_options, VersionSet* version_set); void PurgeObsoleteWALFiles(); void ArchiveWALFile(const std::string& fname, uint64_t number); Status TEST_ReadFirstRecord(const WalFileType type, const uint64_t number, SequenceNumber* sequence) { return ReadFirstRecord(type, number, sequence); } Status TEST_ReadFirstLine(const std::string& fname, SequenceNumber* sequence) { return ReadFirstLine(fname, sequence); } private: Status GetSortedWalsOfType(const std::string& path, VectorLogPtr& log_files, WalFileType type); // Requires: all_logs should be sorted with earliest log file first // Retains all log files in all_logs which contain updates with seq no. // Greater Than or Equal to the requested SequenceNumber. Status RetainProbableWalFiles(VectorLogPtr& all_logs, const SequenceNumber target); Status ReadFirstRecord(const WalFileType type, const uint64_t number, SequenceNumber* sequence); Status ReadFirstLine(const std::string& fname, SequenceNumber* sequence); // ------- state from DBImpl ------ const DBOptions& db_options_; const EnvOptions& env_options_; Env* env_; // ------- WalManager state ------- // cache for ReadFirstRecord() calls std::unordered_map read_first_record_cache_; port::Mutex read_first_record_cache_mutex_; // last time when PurgeObsoleteWALFiles ran. uint64_t purge_wal_files_last_run_; // obsolete files will be deleted every this seconds if ttl deletion is // enabled and archive size_limit is disabled. static const uint64_t kDefaultIntervalToDeleteObsoleteWAL = 600; }; #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/db/wal_manager_test.cc000066400000000000000000000225401267530634000173710ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include #include #include "rocksdb/cache.h" #include "rocksdb/write_batch.h" #include "db/wal_manager.h" #include "db/log_writer.h" #include "db/column_family.h" #include "db/version_set.h" #include "db/writebuffer.h" #include "util/file_reader_writer.h" #include "util/mock_env.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" #include "table/mock_table.h" #include "db/db_impl.h" namespace rocksdb { // TODO(icanadi) mock out VersionSet // TODO(icanadi) move other WalManager-specific tests from db_test here class WalManagerTest : public testing::Test { public: WalManagerTest() : env_(new MockEnv(Env::Default())), dbname_(test::TmpDir() + "/wal_manager_test"), table_cache_(NewLRUCache(50000, 16)), write_buffer_(db_options_.db_write_buffer_size), current_log_number_(0) { DestroyDB(dbname_, Options()); } void Init() { ASSERT_OK(env_->CreateDirIfMissing(dbname_)); ASSERT_OK(env_->CreateDirIfMissing(ArchivalDirectory(dbname_))); db_options_.db_paths.emplace_back(dbname_, std::numeric_limits::max()); db_options_.wal_dir = dbname_; db_options_.env = env_.get(); versions_.reset(new VersionSet(dbname_, &db_options_, env_options_, table_cache_.get(), &write_buffer_, &write_controller_)); wal_manager_.reset(new WalManager(db_options_, env_options_)); } void Reopen() { wal_manager_.reset(new WalManager(db_options_, env_options_)); } // NOT thread safe void Put(const std::string& key, const std::string& value) { assert(current_log_writer_.get() != nullptr); uint64_t seq = versions_->LastSequence() + 1; WriteBatch batch; batch.Put(key, value); WriteBatchInternal::SetSequence(&batch, seq); current_log_writer_->AddRecord(WriteBatchInternal::Contents(&batch)); versions_->SetLastSequence(seq); } // NOT thread safe void RollTheLog(bool archived) { current_log_number_++; std::string fname = ArchivedLogFileName(dbname_, current_log_number_); unique_ptr file; ASSERT_OK(env_->NewWritableFile(fname, &file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(file), env_options_)); current_log_writer_.reset(new log::Writer(std::move(file_writer), 0, false)); } void CreateArchiveLogs(int num_logs, int entries_per_log) { for (int i = 1; i <= num_logs; ++i) { RollTheLog(true); for (int k = 0; k < entries_per_log; ++k) { Put(ToString(k), std::string(1024, 'a')); } } } std::unique_ptr OpenTransactionLogIter( const SequenceNumber seq) { unique_ptr iter; Status status = wal_manager_->GetUpdatesSince( seq, &iter, TransactionLogIterator::ReadOptions(), versions_.get()); EXPECT_OK(status); return iter; } std::unique_ptr env_; std::string dbname_; WriteController write_controller_; EnvOptions env_options_; std::shared_ptr table_cache_; DBOptions db_options_; WriteBuffer write_buffer_; std::unique_ptr versions_; std::unique_ptr wal_manager_; std::unique_ptr current_log_writer_; uint64_t current_log_number_; }; TEST_F(WalManagerTest, ReadFirstRecordCache) { Init(); std::string path = dbname_ + "/000001.log"; unique_ptr file; ASSERT_OK(env_->NewWritableFile(path, &file, EnvOptions())); SequenceNumber s; ASSERT_OK(wal_manager_->TEST_ReadFirstLine(path, &s)); ASSERT_EQ(s, 0U); ASSERT_OK(wal_manager_->TEST_ReadFirstRecord(kAliveLogFile, 1, &s)); ASSERT_EQ(s, 0U); unique_ptr file_writer( new WritableFileWriter(std::move(file), EnvOptions())); log::Writer writer(std::move(file_writer), 1, db_options_.recycle_log_file_num > 0); WriteBatch batch; batch.Put("foo", "bar"); WriteBatchInternal::SetSequence(&batch, 10); writer.AddRecord(WriteBatchInternal::Contents(&batch)); // TODO(icanadi) move SpecialEnv outside of db_test, so we can reuse it here. // Waiting for lei to finish with db_test // env_->count_sequential_reads_ = true; // sequential_read_counter_ sanity test // ASSERT_EQ(env_->sequential_read_counter_.Read(), 0); ASSERT_OK(wal_manager_->TEST_ReadFirstRecord(kAliveLogFile, 1, &s)); ASSERT_EQ(s, 10U); // did a read // TODO(icanadi) move SpecialEnv outside of db_test, so we can reuse it here // ASSERT_EQ(env_->sequential_read_counter_.Read(), 1); ASSERT_OK(wal_manager_->TEST_ReadFirstRecord(kAliveLogFile, 1, &s)); ASSERT_EQ(s, 10U); // no new reads since the value is cached // TODO(icanadi) move SpecialEnv outside of db_test, so we can reuse it here // ASSERT_EQ(env_->sequential_read_counter_.Read(), 1); } namespace { uint64_t GetLogDirSize(std::string dir_path, Env* env) { uint64_t dir_size = 0; std::vector files; env->GetChildren(dir_path, &files); for (auto& f : files) { uint64_t number; FileType type; if (ParseFileName(f, &number, &type) && type == kLogFile) { std::string const file_path = dir_path + "/" + f; uint64_t file_size; env->GetFileSize(file_path, &file_size); dir_size += file_size; } } return dir_size; } std::vector ListSpecificFiles( Env* env, const std::string& path, const FileType expected_file_type) { std::vector files; std::vector file_numbers; env->GetChildren(path, &files); uint64_t number; FileType type; for (size_t i = 0; i < files.size(); ++i) { if (ParseFileName(files[i], &number, &type)) { if (type == expected_file_type) { file_numbers.push_back(number); } } } return file_numbers; } int CountRecords(TransactionLogIterator* iter) { int count = 0; SequenceNumber lastSequence = 0; BatchResult res; while (iter->Valid()) { res = iter->GetBatch(); EXPECT_TRUE(res.sequence > lastSequence); ++count; lastSequence = res.sequence; EXPECT_OK(iter->status()); iter->Next(); } return count; } } // namespace TEST_F(WalManagerTest, WALArchivalSizeLimit) { db_options_.WAL_ttl_seconds = 0; db_options_.WAL_size_limit_MB = 1000; Init(); // TEST : Create WalManager with huge size limit and no ttl. // Create some archived files and call PurgeObsoleteWALFiles(). // Count the archived log files that survived. // Assert that all of them did. // Change size limit. Re-open WalManager. // Assert that archive is not greater than WAL_size_limit_MB after // PurgeObsoleteWALFiles() // Set ttl and time_to_check_ to small values. Re-open db. // Assert that there are no archived logs left. std::string archive_dir = ArchivalDirectory(dbname_); CreateArchiveLogs(20, 5000); std::vector log_files = ListSpecificFiles(env_.get(), archive_dir, kLogFile); ASSERT_EQ(log_files.size(), 20U); db_options_.WAL_size_limit_MB = 8; Reopen(); wal_manager_->PurgeObsoleteWALFiles(); uint64_t archive_size = GetLogDirSize(archive_dir, env_.get()); ASSERT_TRUE(archive_size <= db_options_.WAL_size_limit_MB * 1024 * 1024); db_options_.WAL_ttl_seconds = 1; env_->FakeSleepForMicroseconds(2 * 1000 * 1000); Reopen(); wal_manager_->PurgeObsoleteWALFiles(); log_files = ListSpecificFiles(env_.get(), archive_dir, kLogFile); ASSERT_TRUE(log_files.empty()); } TEST_F(WalManagerTest, WALArchivalTtl) { db_options_.WAL_ttl_seconds = 1000; Init(); // TEST : Create WalManager with a ttl and no size limit. // Create some archived log files and call PurgeObsoleteWALFiles(). // Assert that files are not deleted // Reopen db with small ttl. // Assert that all archived logs was removed. std::string archive_dir = ArchivalDirectory(dbname_); CreateArchiveLogs(20, 5000); std::vector log_files = ListSpecificFiles(env_.get(), archive_dir, kLogFile); ASSERT_GT(log_files.size(), 0U); db_options_.WAL_ttl_seconds = 1; env_->FakeSleepForMicroseconds(3 * 1000 * 1000); Reopen(); wal_manager_->PurgeObsoleteWALFiles(); log_files = ListSpecificFiles(env_.get(), archive_dir, kLogFile); ASSERT_TRUE(log_files.empty()); } TEST_F(WalManagerTest, TransactionLogIteratorMoveOverZeroFiles) { Init(); RollTheLog(false); Put("key1", std::string(1024, 'a')); // Create a zero record WAL file. RollTheLog(false); RollTheLog(false); Put("key2", std::string(1024, 'a')); auto iter = OpenTransactionLogIter(0); ASSERT_EQ(2, CountRecords(iter.get())); } TEST_F(WalManagerTest, TransactionLogIteratorJustEmptyFile) { Init(); RollTheLog(false); auto iter = OpenTransactionLogIter(0); // Check that an empty iterator is returned ASSERT_TRUE(!iter->Valid()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as WalManager is not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/write_batch.cc000066400000000000000000000723311267530634000163530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // WriteBatch::rep_ := // sequence: fixed64 // count: fixed32 // data: record[count] // record := // kTypeValue varstring varstring // kTypeDeletion varstring // kTypeSingleDeletion varstring // kTypeMerge varstring varstring // kTypeColumnFamilyValue varint32 varstring varstring // kTypeColumnFamilyDeletion varint32 varstring varstring // kTypeColumnFamilySingleDeletion varint32 varstring varstring // kTypeColumnFamilyMerge varint32 varstring varstring // varstring := // len: varint32 // data: uint8[len] #include "rocksdb/write_batch.h" #include #include #include #include "db/column_family.h" #include "db/db_impl.h" #include "db/dbformat.h" #include "db/flush_scheduler.h" #include "db/memtable.h" #include "db/snapshot_impl.h" #include "db/write_batch_internal.h" #include "rocksdb/merge_operator.h" #include "util/coding.h" #include "util/perf_context_imp.h" #include "util/statistics.h" namespace rocksdb { // anon namespace for file-local types namespace { enum ContentFlags : uint32_t { DEFERRED = 1, HAS_PUT = 2, HAS_DELETE = 4, HAS_SINGLE_DELETE = 8, HAS_MERGE = 16, }; struct BatchContentClassifier : public WriteBatch::Handler { uint32_t content_flags = 0; Status PutCF(uint32_t, const Slice&, const Slice&) override { content_flags |= ContentFlags::HAS_PUT; return Status::OK(); } Status DeleteCF(uint32_t, const Slice&) override { content_flags |= ContentFlags::HAS_DELETE; return Status::OK(); } Status SingleDeleteCF(uint32_t, const Slice&) override { content_flags |= ContentFlags::HAS_SINGLE_DELETE; return Status::OK(); } Status MergeCF(uint32_t, const Slice&, const Slice&) override { content_flags |= ContentFlags::HAS_MERGE; return Status::OK(); } }; } // anon namespace // WriteBatch header has an 8-byte sequence number followed by a 4-byte count. static const size_t kHeader = 12; struct SavePoint { size_t size; // size of rep_ int count; // count of elements in rep_ uint32_t content_flags; }; struct SavePoints { std::stack stack; }; WriteBatch::WriteBatch(size_t reserved_bytes) : save_points_(nullptr), content_flags_(0), rep_() { rep_.reserve((reserved_bytes > kHeader) ? reserved_bytes : kHeader); rep_.resize(kHeader); } WriteBatch::WriteBatch(const std::string& rep) : save_points_(nullptr), content_flags_(ContentFlags::DEFERRED), rep_(rep) {} WriteBatch::WriteBatch(const WriteBatch& src) : save_points_(src.save_points_), content_flags_(src.content_flags_.load(std::memory_order_relaxed)), rep_(src.rep_) {} WriteBatch::WriteBatch(WriteBatch&& src) : save_points_(std::move(src.save_points_)), content_flags_(src.content_flags_.load(std::memory_order_relaxed)), rep_(std::move(src.rep_)) {} WriteBatch& WriteBatch::operator=(const WriteBatch& src) { if (&src != this) { this->~WriteBatch(); new (this) WriteBatch(src); } return *this; } WriteBatch& WriteBatch::operator=(WriteBatch&& src) { if (&src != this) { this->~WriteBatch(); new (this) WriteBatch(std::move(src)); } return *this; } WriteBatch::~WriteBatch() { delete save_points_; } WriteBatch::Handler::~Handler() { } void WriteBatch::Handler::LogData(const Slice& blob) { // If the user has not specified something to do with blobs, then we ignore // them. } bool WriteBatch::Handler::Continue() { return true; } void WriteBatch::Clear() { rep_.clear(); rep_.resize(kHeader); content_flags_.store(0, std::memory_order_relaxed); if (save_points_ != nullptr) { while (!save_points_->stack.empty()) { save_points_->stack.pop(); } } } int WriteBatch::Count() const { return WriteBatchInternal::Count(this); } uint32_t WriteBatch::ComputeContentFlags() const { auto rv = content_flags_.load(std::memory_order_relaxed); if ((rv & ContentFlags::DEFERRED) != 0) { BatchContentClassifier classifier; Iterate(&classifier); rv = classifier.content_flags; // this method is conceptually const, because it is performing a lazy // computation that doesn't affect the abstract state of the batch. // content_flags_ is marked mutable so that we can perform the // following assignment content_flags_.store(rv, std::memory_order_relaxed); } return rv; } bool WriteBatch::HasPut() const { return (ComputeContentFlags() & ContentFlags::HAS_PUT) != 0; } bool WriteBatch::HasDelete() const { return (ComputeContentFlags() & ContentFlags::HAS_DELETE) != 0; } bool WriteBatch::HasSingleDelete() const { return (ComputeContentFlags() & ContentFlags::HAS_SINGLE_DELETE) != 0; } bool WriteBatch::HasMerge() const { return (ComputeContentFlags() & ContentFlags::HAS_MERGE) != 0; } Status ReadRecordFromWriteBatch(Slice* input, char* tag, uint32_t* column_family, Slice* key, Slice* value, Slice* blob) { assert(key != nullptr && value != nullptr); *tag = (*input)[0]; input->remove_prefix(1); *column_family = 0; // default switch (*tag) { case kTypeColumnFamilyValue: if (!GetVarint32(input, column_family)) { return Status::Corruption("bad WriteBatch Put"); } // intentional fallthrough case kTypeValue: if (!GetLengthPrefixedSlice(input, key) || !GetLengthPrefixedSlice(input, value)) { return Status::Corruption("bad WriteBatch Put"); } break; case kTypeColumnFamilyDeletion: case kTypeColumnFamilySingleDeletion: if (!GetVarint32(input, column_family)) { return Status::Corruption("bad WriteBatch Delete"); } // intentional fallthrough case kTypeDeletion: case kTypeSingleDeletion: if (!GetLengthPrefixedSlice(input, key)) { return Status::Corruption("bad WriteBatch Delete"); } break; case kTypeColumnFamilyMerge: if (!GetVarint32(input, column_family)) { return Status::Corruption("bad WriteBatch Merge"); } // intentional fallthrough case kTypeMerge: if (!GetLengthPrefixedSlice(input, key) || !GetLengthPrefixedSlice(input, value)) { return Status::Corruption("bad WriteBatch Merge"); } break; case kTypeLogData: assert(blob != nullptr); if (!GetLengthPrefixedSlice(input, blob)) { return Status::Corruption("bad WriteBatch Blob"); } break; default: return Status::Corruption("unknown WriteBatch tag"); } return Status::OK(); } Status WriteBatch::Iterate(Handler* handler) const { Slice input(rep_); if (input.size() < kHeader) { return Status::Corruption("malformed WriteBatch (too small)"); } input.remove_prefix(kHeader); Slice key, value, blob; int found = 0; Status s; while (s.ok() && !input.empty() && handler->Continue()) { char tag = 0; uint32_t column_family = 0; // default s = ReadRecordFromWriteBatch(&input, &tag, &column_family, &key, &value, &blob); if (!s.ok()) { return s; } switch (tag) { case kTypeColumnFamilyValue: case kTypeValue: assert(content_flags_.load(std::memory_order_relaxed) & (ContentFlags::DEFERRED | ContentFlags::HAS_PUT)); s = handler->PutCF(column_family, key, value); found++; break; case kTypeColumnFamilyDeletion: case kTypeDeletion: assert(content_flags_.load(std::memory_order_relaxed) & (ContentFlags::DEFERRED | ContentFlags::HAS_DELETE)); s = handler->DeleteCF(column_family, key); found++; break; case kTypeColumnFamilySingleDeletion: case kTypeSingleDeletion: assert(content_flags_.load(std::memory_order_relaxed) & (ContentFlags::DEFERRED | ContentFlags::HAS_SINGLE_DELETE)); s = handler->SingleDeleteCF(column_family, key); found++; break; case kTypeColumnFamilyMerge: case kTypeMerge: assert(content_flags_.load(std::memory_order_relaxed) & (ContentFlags::DEFERRED | ContentFlags::HAS_MERGE)); s = handler->MergeCF(column_family, key, value); found++; break; case kTypeLogData: handler->LogData(blob); break; default: return Status::Corruption("unknown WriteBatch tag"); } } if (!s.ok()) { return s; } if (found != WriteBatchInternal::Count(this)) { return Status::Corruption("WriteBatch has wrong count"); } else { return Status::OK(); } } int WriteBatchInternal::Count(const WriteBatch* b) { return DecodeFixed32(b->rep_.data() + 8); } void WriteBatchInternal::SetCount(WriteBatch* b, int n) { EncodeFixed32(&b->rep_[8], n); } SequenceNumber WriteBatchInternal::Sequence(const WriteBatch* b) { return SequenceNumber(DecodeFixed64(b->rep_.data())); } void WriteBatchInternal::SetSequence(WriteBatch* b, SequenceNumber seq) { EncodeFixed64(&b->rep_[0], seq); } size_t WriteBatchInternal::GetFirstOffset(WriteBatch* b) { return kHeader; } void WriteBatchInternal::Put(WriteBatch* b, uint32_t column_family_id, const Slice& key, const Slice& value) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeValue)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilyValue)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSlice(&b->rep_, key); PutLengthPrefixedSlice(&b->rep_, value); b->content_flags_.store( b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_PUT, std::memory_order_relaxed); } void WriteBatch::Put(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) { WriteBatchInternal::Put(this, GetColumnFamilyID(column_family), key, value); } void WriteBatchInternal::Put(WriteBatch* b, uint32_t column_family_id, const SliceParts& key, const SliceParts& value) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeValue)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilyValue)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSliceParts(&b->rep_, key); PutLengthPrefixedSliceParts(&b->rep_, value); b->content_flags_.store( b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_PUT, std::memory_order_relaxed); } void WriteBatch::Put(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) { WriteBatchInternal::Put(this, GetColumnFamilyID(column_family), key, value); } void WriteBatchInternal::Delete(WriteBatch* b, uint32_t column_family_id, const Slice& key) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeDeletion)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilyDeletion)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSlice(&b->rep_, key); b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_DELETE, std::memory_order_relaxed); } void WriteBatch::Delete(ColumnFamilyHandle* column_family, const Slice& key) { WriteBatchInternal::Delete(this, GetColumnFamilyID(column_family), key); } void WriteBatchInternal::Delete(WriteBatch* b, uint32_t column_family_id, const SliceParts& key) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeDeletion)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilyDeletion)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSliceParts(&b->rep_, key); b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_DELETE, std::memory_order_relaxed); } void WriteBatch::Delete(ColumnFamilyHandle* column_family, const SliceParts& key) { WriteBatchInternal::Delete(this, GetColumnFamilyID(column_family), key); } void WriteBatchInternal::SingleDelete(WriteBatch* b, uint32_t column_family_id, const Slice& key) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeSingleDeletion)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilySingleDeletion)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSlice(&b->rep_, key); b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_SINGLE_DELETE, std::memory_order_relaxed); } void WriteBatch::SingleDelete(ColumnFamilyHandle* column_family, const Slice& key) { WriteBatchInternal::SingleDelete(this, GetColumnFamilyID(column_family), key); } void WriteBatchInternal::SingleDelete(WriteBatch* b, uint32_t column_family_id, const SliceParts& key) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeSingleDeletion)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilySingleDeletion)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSliceParts(&b->rep_, key); b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_SINGLE_DELETE, std::memory_order_relaxed); } void WriteBatch::SingleDelete(ColumnFamilyHandle* column_family, const SliceParts& key) { WriteBatchInternal::SingleDelete(this, GetColumnFamilyID(column_family), key); } void WriteBatchInternal::Merge(WriteBatch* b, uint32_t column_family_id, const Slice& key, const Slice& value) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeMerge)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilyMerge)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSlice(&b->rep_, key); PutLengthPrefixedSlice(&b->rep_, value); b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_MERGE, std::memory_order_relaxed); } void WriteBatch::Merge(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) { WriteBatchInternal::Merge(this, GetColumnFamilyID(column_family), key, value); } void WriteBatchInternal::Merge(WriteBatch* b, uint32_t column_family_id, const SliceParts& key, const SliceParts& value) { WriteBatchInternal::SetCount(b, WriteBatchInternal::Count(b) + 1); if (column_family_id == 0) { b->rep_.push_back(static_cast(kTypeMerge)); } else { b->rep_.push_back(static_cast(kTypeColumnFamilyMerge)); PutVarint32(&b->rep_, column_family_id); } PutLengthPrefixedSliceParts(&b->rep_, key); PutLengthPrefixedSliceParts(&b->rep_, value); b->content_flags_.store(b->content_flags_.load(std::memory_order_relaxed) | ContentFlags::HAS_MERGE, std::memory_order_relaxed); } void WriteBatch::Merge(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) { WriteBatchInternal::Merge(this, GetColumnFamilyID(column_family), key, value); } void WriteBatch::PutLogData(const Slice& blob) { rep_.push_back(static_cast(kTypeLogData)); PutLengthPrefixedSlice(&rep_, blob); } void WriteBatch::SetSavePoint() { if (save_points_ == nullptr) { save_points_ = new SavePoints(); } // Record length and count of current batch of writes. save_points_->stack.push(SavePoint{ GetDataSize(), Count(), content_flags_.load(std::memory_order_relaxed)}); } Status WriteBatch::RollbackToSavePoint() { if (save_points_ == nullptr || save_points_->stack.size() == 0) { return Status::NotFound(); } // Pop the most recent savepoint off the stack SavePoint savepoint = save_points_->stack.top(); save_points_->stack.pop(); assert(savepoint.size <= rep_.size()); assert(savepoint.count <= Count()); if (savepoint.size == rep_.size()) { // No changes to rollback } else if (savepoint.size == 0) { // Rollback everything Clear(); } else { rep_.resize(savepoint.size); WriteBatchInternal::SetCount(this, savepoint.count); content_flags_.store(savepoint.content_flags, std::memory_order_relaxed); } return Status::OK(); } namespace { class MemTableInserter : public WriteBatch::Handler { public: SequenceNumber sequence_; ColumnFamilyMemTables* const cf_mems_; FlushScheduler* const flush_scheduler_; const bool ignore_missing_column_families_; const uint64_t log_number_; DBImpl* db_; const bool dont_filter_deletes_; const bool concurrent_memtable_writes_; // cf_mems should not be shared with concurrent inserters MemTableInserter(SequenceNumber sequence, ColumnFamilyMemTables* cf_mems, FlushScheduler* flush_scheduler, bool ignore_missing_column_families, uint64_t log_number, DB* db, const bool dont_filter_deletes, bool concurrent_memtable_writes) : sequence_(sequence), cf_mems_(cf_mems), flush_scheduler_(flush_scheduler), ignore_missing_column_families_(ignore_missing_column_families), log_number_(log_number), db_(reinterpret_cast(db)), dont_filter_deletes_(dont_filter_deletes), concurrent_memtable_writes_(concurrent_memtable_writes) { assert(cf_mems_); if (!dont_filter_deletes_) { assert(db_); } } bool SeekToColumnFamily(uint32_t column_family_id, Status* s) { // If we are in a concurrent mode, it is the caller's responsibility // to clone the original ColumnFamilyMemTables so that each thread // has its own instance. Otherwise, it must be guaranteed that there // is no concurrent access bool found = cf_mems_->Seek(column_family_id); if (!found) { if (ignore_missing_column_families_) { *s = Status::OK(); } else { *s = Status::InvalidArgument( "Invalid column family specified in write batch"); } return false; } if (log_number_ != 0 && log_number_ < cf_mems_->GetLogNumber()) { // This is true only in recovery environment (log_number_ is always 0 in // non-recovery, regular write code-path) // * If log_number_ < cf_mems_->GetLogNumber(), this means that column // family already contains updates from this log. We can't apply updates // twice because of update-in-place or merge workloads -- ignore the // update *s = Status::OK(); return false; } return true; } virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { Status seek_status; if (!SeekToColumnFamily(column_family_id, &seek_status)) { ++sequence_; return seek_status; } MemTable* mem = cf_mems_->GetMemTable(); auto* moptions = mem->GetMemTableOptions(); if (!moptions->inplace_update_support) { mem->Add(sequence_, kTypeValue, key, value, concurrent_memtable_writes_); } else if (moptions->inplace_callback == nullptr) { assert(!concurrent_memtable_writes_); mem->Update(sequence_, key, value); RecordTick(moptions->statistics, NUMBER_KEYS_UPDATED); } else { assert(!concurrent_memtable_writes_); if (mem->UpdateCallback(sequence_, key, value)) { } else { // key not found in memtable. Do sst get, update, add SnapshotImpl read_from_snapshot; read_from_snapshot.number_ = sequence_; ReadOptions ropts; ropts.snapshot = &read_from_snapshot; std::string prev_value; std::string merged_value; auto cf_handle = cf_mems_->GetColumnFamilyHandle(); if (cf_handle == nullptr) { cf_handle = db_->DefaultColumnFamily(); } Status s = db_->Get(ropts, cf_handle, key, &prev_value); char* prev_buffer = const_cast(prev_value.c_str()); uint32_t prev_size = static_cast(prev_value.size()); auto status = moptions->inplace_callback(s.ok() ? prev_buffer : nullptr, s.ok() ? &prev_size : nullptr, value, &merged_value); if (status == UpdateStatus::UPDATED_INPLACE) { // prev_value is updated in-place with final value. mem->Add(sequence_, kTypeValue, key, Slice(prev_buffer, prev_size)); RecordTick(moptions->statistics, NUMBER_KEYS_WRITTEN); } else if (status == UpdateStatus::UPDATED) { // merged_value contains the final value. mem->Add(sequence_, kTypeValue, key, Slice(merged_value)); RecordTick(moptions->statistics, NUMBER_KEYS_WRITTEN); } } } // Since all Puts are logged in trasaction logs (if enabled), always bump // sequence number. Even if the update eventually fails and does not result // in memtable add/update. sequence_++; CheckMemtableFull(); return Status::OK(); } Status DeleteImpl(uint32_t column_family_id, const Slice& key, ValueType delete_type) { Status seek_status; if (!SeekToColumnFamily(column_family_id, &seek_status)) { ++sequence_; return seek_status; } MemTable* mem = cf_mems_->GetMemTable(); auto* moptions = mem->GetMemTableOptions(); if (!dont_filter_deletes_ && moptions->filter_deletes) { assert(!concurrent_memtable_writes_); SnapshotImpl read_from_snapshot; read_from_snapshot.number_ = sequence_; ReadOptions ropts; ropts.snapshot = &read_from_snapshot; std::string value; auto cf_handle = cf_mems_->GetColumnFamilyHandle(); if (cf_handle == nullptr) { cf_handle = db_->DefaultColumnFamily(); } if (!db_->KeyMayExist(ropts, cf_handle, key, &value)) { RecordTick(moptions->statistics, NUMBER_FILTERED_DELETES); return Status::OK(); } } mem->Add(sequence_, delete_type, key, Slice(), concurrent_memtable_writes_); sequence_++; CheckMemtableFull(); return Status::OK(); } virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) override { return DeleteImpl(column_family_id, key, kTypeDeletion); } virtual Status SingleDeleteCF(uint32_t column_family_id, const Slice& key) override { return DeleteImpl(column_family_id, key, kTypeSingleDeletion); } virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { assert(!concurrent_memtable_writes_); Status seek_status; if (!SeekToColumnFamily(column_family_id, &seek_status)) { ++sequence_; return seek_status; } MemTable* mem = cf_mems_->GetMemTable(); auto* moptions = mem->GetMemTableOptions(); bool perform_merge = false; if (moptions->max_successive_merges > 0 && db_ != nullptr) { LookupKey lkey(key, sequence_); // Count the number of successive merges at the head // of the key in the memtable size_t num_merges = mem->CountSuccessiveMergeEntries(lkey); if (num_merges >= moptions->max_successive_merges) { perform_merge = true; } } if (perform_merge) { // 1) Get the existing value std::string get_value; // Pass in the sequence number so that we also include previous merge // operations in the same batch. SnapshotImpl read_from_snapshot; read_from_snapshot.number_ = sequence_; ReadOptions read_options; read_options.snapshot = &read_from_snapshot; auto cf_handle = cf_mems_->GetColumnFamilyHandle(); if (cf_handle == nullptr) { cf_handle = db_->DefaultColumnFamily(); } db_->Get(read_options, cf_handle, key, &get_value); Slice get_value_slice = Slice(get_value); // 2) Apply this merge auto merge_operator = moptions->merge_operator; assert(merge_operator); std::deque operands; operands.push_front(value.ToString()); std::string new_value; bool merge_success = false; { StopWatchNano timer(Env::Default(), moptions->statistics != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); merge_success = merge_operator->FullMerge( key, &get_value_slice, operands, &new_value, moptions->info_log); RecordTick(moptions->statistics, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanos()); } if (!merge_success) { // Failed to merge! RecordTick(moptions->statistics, NUMBER_MERGE_FAILURES); // Store the delta in memtable perform_merge = false; } else { // 3) Add value to memtable mem->Add(sequence_, kTypeValue, key, new_value); } } if (!perform_merge) { // Add merge operator to memtable mem->Add(sequence_, kTypeMerge, key, value); } sequence_++; CheckMemtableFull(); return Status::OK(); } void CheckMemtableFull() { if (flush_scheduler_ != nullptr) { auto* cfd = cf_mems_->current(); assert(cfd != nullptr); if (cfd->mem()->ShouldScheduleFlush() && cfd->mem()->MarkFlushScheduled()) { // MarkFlushScheduled only returns true if we are the one that // should take action, so no need to dedup further flush_scheduler_->ScheduleFlush(cfd); } } } }; } // namespace // This function can only be called in these conditions: // 1) During Recovery() // 2) During Write(), in a single-threaded write thread // 3) During Write(), in a concurrent context where memtables has been cloned // The reason is that it calls memtables->Seek(), which has a stateful cache Status WriteBatchInternal::InsertInto( const autovector& batches, SequenceNumber sequence, ColumnFamilyMemTables* memtables, FlushScheduler* flush_scheduler, bool ignore_missing_column_families, uint64_t log_number, DB* db, const bool dont_filter_deletes, bool concurrent_memtable_writes) { MemTableInserter inserter(sequence, memtables, flush_scheduler, ignore_missing_column_families, log_number, db, dont_filter_deletes, concurrent_memtable_writes); Status rv = Status::OK(); for (size_t i = 0; i < batches.size() && rv.ok(); ++i) { rv = batches[i]->Iterate(&inserter); } return rv; } Status WriteBatchInternal::InsertInto(const WriteBatch* batch, ColumnFamilyMemTables* memtables, FlushScheduler* flush_scheduler, bool ignore_missing_column_families, uint64_t log_number, DB* db, const bool dont_filter_deletes, bool concurrent_memtable_writes) { MemTableInserter inserter(WriteBatchInternal::Sequence(batch), memtables, flush_scheduler, ignore_missing_column_families, log_number, db, dont_filter_deletes, concurrent_memtable_writes); return batch->Iterate(&inserter); } void WriteBatchInternal::SetContents(WriteBatch* b, const Slice& contents) { assert(contents.size() >= kHeader); b->rep_.assign(contents.data(), contents.size()); b->content_flags_.store(ContentFlags::DEFERRED, std::memory_order_relaxed); } void WriteBatchInternal::Append(WriteBatch* dst, const WriteBatch* src) { SetCount(dst, Count(dst) + Count(src)); assert(src->rep_.size() >= kHeader); dst->rep_.append(src->rep_.data() + kHeader, src->rep_.size() - kHeader); dst->content_flags_.store( dst->content_flags_.load(std::memory_order_relaxed) | src->content_flags_.load(std::memory_order_relaxed), std::memory_order_relaxed); } size_t WriteBatchInternal::AppendedByteSize(size_t leftByteSize, size_t rightByteSize) { if (leftByteSize == 0 || rightByteSize == 0) { return leftByteSize + rightByteSize; } else { return leftByteSize + rightByteSize - kHeader; } } } // namespace rocksdb rocksdb-4.5.1/db/write_batch_base.cc000066400000000000000000000044241267530634000173430ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "rocksdb/write_batch_base.h" #include #include "rocksdb/slice.h" namespace rocksdb { // Simple implementation of SlicePart variants of Put(). Child classes // can override these method with more performant solutions if they choose. void WriteBatchBase::Put(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) { std::string key_buf, value_buf; Slice key_slice(key, &key_buf); Slice value_slice(value, &value_buf); Put(column_family, key_slice, value_slice); } void WriteBatchBase::Put(const SliceParts& key, const SliceParts& value) { std::string key_buf, value_buf; Slice key_slice(key, &key_buf); Slice value_slice(value, &value_buf); Put(key_slice, value_slice); } void WriteBatchBase::Delete(ColumnFamilyHandle* column_family, const SliceParts& key) { std::string key_buf; Slice key_slice(key, &key_buf); Delete(column_family, key_slice); } void WriteBatchBase::Delete(const SliceParts& key) { std::string key_buf; Slice key_slice(key, &key_buf); Delete(key_slice); } void WriteBatchBase::SingleDelete(ColumnFamilyHandle* column_family, const SliceParts& key) { std::string key_buf; Slice key_slice(key, &key_buf); SingleDelete(column_family, key_slice); } void WriteBatchBase::SingleDelete(const SliceParts& key) { std::string key_buf; Slice key_slice(key, &key_buf); SingleDelete(key_slice); } void WriteBatchBase::Merge(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) { std::string key_buf, value_buf; Slice key_slice(key, &key_buf); Slice value_slice(value, &value_buf); Merge(column_family, key_slice, value_slice); } void WriteBatchBase::Merge(const SliceParts& key, const SliceParts& value) { std::string key_buf, value_buf; Slice key_slice(key, &key_buf); Slice value_slice(value, &value_buf); Merge(key_slice, value_slice); } } // namespace rocksdb rocksdb-4.5.1/db/write_batch_internal.h000066400000000000000000000141101267530634000201000ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include "rocksdb/types.h" #include "rocksdb/write_batch.h" #include "rocksdb/db.h" #include "rocksdb/options.h" #include "util/autovector.h" namespace rocksdb { class MemTable; class FlushScheduler; class ColumnFamilyData; class ColumnFamilyMemTables { public: virtual ~ColumnFamilyMemTables() {} virtual bool Seek(uint32_t column_family_id) = 0; // returns true if the update to memtable should be ignored // (useful when recovering from log whose updates have already // been processed) virtual uint64_t GetLogNumber() const = 0; virtual MemTable* GetMemTable() const = 0; virtual ColumnFamilyHandle* GetColumnFamilyHandle() = 0; virtual ColumnFamilyData* current() { return nullptr; } }; class ColumnFamilyMemTablesDefault : public ColumnFamilyMemTables { public: explicit ColumnFamilyMemTablesDefault(MemTable* mem) : ok_(false), mem_(mem) {} bool Seek(uint32_t column_family_id) override { ok_ = (column_family_id == 0); return ok_; } uint64_t GetLogNumber() const override { return 0; } MemTable* GetMemTable() const override { assert(ok_); return mem_; } ColumnFamilyHandle* GetColumnFamilyHandle() override { return nullptr; } private: bool ok_; MemTable* mem_; }; // WriteBatchInternal provides static methods for manipulating a // WriteBatch that we don't want in the public WriteBatch interface. class WriteBatchInternal { public: // WriteBatch methods with column_family_id instead of ColumnFamilyHandle* static void Put(WriteBatch* batch, uint32_t column_family_id, const Slice& key, const Slice& value); static void Put(WriteBatch* batch, uint32_t column_family_id, const SliceParts& key, const SliceParts& value); static void Delete(WriteBatch* batch, uint32_t column_family_id, const SliceParts& key); static void Delete(WriteBatch* batch, uint32_t column_family_id, const Slice& key); static void SingleDelete(WriteBatch* batch, uint32_t column_family_id, const SliceParts& key); static void SingleDelete(WriteBatch* batch, uint32_t column_family_id, const Slice& key); static void Merge(WriteBatch* batch, uint32_t column_family_id, const Slice& key, const Slice& value); static void Merge(WriteBatch* batch, uint32_t column_family_id, const SliceParts& key, const SliceParts& value); // Return the number of entries in the batch. static int Count(const WriteBatch* batch); // Set the count for the number of entries in the batch. static void SetCount(WriteBatch* batch, int n); // Return the seqeunce number for the start of this batch. static SequenceNumber Sequence(const WriteBatch* batch); // Store the specified number as the seqeunce number for the start of // this batch. static void SetSequence(WriteBatch* batch, SequenceNumber seq); // Returns the offset of the first entry in the batch. // This offset is only valid if the batch is not empty. static size_t GetFirstOffset(WriteBatch* batch); static Slice Contents(const WriteBatch* batch) { return Slice(batch->rep_); } static size_t ByteSize(const WriteBatch* batch) { return batch->rep_.size(); } static void SetContents(WriteBatch* batch, const Slice& contents); // Inserts batches[i] into memtable, for i in 0..num_batches-1 inclusive. // // If dont_filter_deletes is false AND options.filter_deletes is true // AND db->KeyMayExist is false, then a Delete won't modify the memtable. // // If ignore_missing_column_families == true. WriteBatch // referencing non-existing column family will be ignored. // If ignore_missing_column_families == false, processing of the // batches will be stopped if a reference is found to a non-existing // column family and InvalidArgument() will be returned. The writes // in batches may be only partially applied at that point. // // If log_number is non-zero, the memtable will be updated only if // memtables->GetLogNumber() >= log_number. // // If flush_scheduler is non-null, it will be invoked if the memtable // should be flushed. // // Under concurrent use, the caller is responsible for making sure that // the memtables object itself is thread-local. static Status InsertInto(const autovector& batches, SequenceNumber sequence, ColumnFamilyMemTables* memtables, FlushScheduler* flush_scheduler, bool ignore_missing_column_families = false, uint64_t log_number = 0, DB* db = nullptr, const bool dont_filter_deletes = true, bool concurrent_memtable_writes = false); // Convenience form of InsertInto when you have only one batch static Status InsertInto(const WriteBatch* batch, ColumnFamilyMemTables* memtables, FlushScheduler* flush_scheduler, bool ignore_missing_column_families = false, uint64_t log_number = 0, DB* db = nullptr, const bool dont_filter_deletes = true, bool concurrent_memtable_writes = false); static void Append(WriteBatch* dst, const WriteBatch* src); // Returns the byte size of appending a WriteBatch with ByteSize // leftByteSize and a WriteBatch with ByteSize rightByteSize static size_t AppendedByteSize(size_t leftByteSize, size_t rightByteSize); }; } // namespace rocksdb rocksdb-4.5.1/db/write_batch_test.cc000066400000000000000000000544651267530634000174220ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "rocksdb/db.h" #include #include "db/memtable.h" #include "db/column_family.h" #include "db/write_batch_internal.h" #include "db/writebuffer.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "rocksdb/utilities/write_batch_with_index.h" #include "table/scoped_arena_iterator.h" #include "util/logging.h" #include "util/string_util.h" #include "util/testharness.h" namespace rocksdb { static std::string PrintContents(WriteBatch* b) { InternalKeyComparator cmp(BytewiseComparator()); auto factory = std::make_shared(); Options options; options.memtable_factory = factory; ImmutableCFOptions ioptions(options); WriteBuffer wb(options.db_write_buffer_size); MemTable* mem = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb, kMaxSequenceNumber); mem->Ref(); std::string state; ColumnFamilyMemTablesDefault cf_mems_default(mem); Status s = WriteBatchInternal::InsertInto(b, &cf_mems_default, nullptr); int count = 0; int put_count = 0; int delete_count = 0; int single_delete_count = 0; int merge_count = 0; Arena arena; ScopedArenaIterator iter(mem->NewIterator(ReadOptions(), &arena)); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ParsedInternalKey ikey; memset((void *)&ikey, 0, sizeof(ikey)); EXPECT_TRUE(ParseInternalKey(iter->key(), &ikey)); switch (ikey.type) { case kTypeValue: state.append("Put("); state.append(ikey.user_key.ToString()); state.append(", "); state.append(iter->value().ToString()); state.append(")"); count++; put_count++; break; case kTypeDeletion: state.append("Delete("); state.append(ikey.user_key.ToString()); state.append(")"); count++; delete_count++; break; case kTypeSingleDeletion: state.append("SingleDelete("); state.append(ikey.user_key.ToString()); state.append(")"); count++; single_delete_count++; break; case kTypeMerge: state.append("Merge("); state.append(ikey.user_key.ToString()); state.append(", "); state.append(iter->value().ToString()); state.append(")"); count++; merge_count++; break; default: assert(false); break; } state.append("@"); state.append(NumberToString(ikey.sequence)); } EXPECT_EQ(b->HasPut(), put_count > 0); EXPECT_EQ(b->HasDelete(), delete_count > 0); EXPECT_EQ(b->HasSingleDelete(), single_delete_count > 0); EXPECT_EQ(b->HasMerge(), merge_count > 0); if (!s.ok()) { state.append(s.ToString()); } else if (count != WriteBatchInternal::Count(b)) { state.append("CountMismatch()"); } delete mem->Unref(); return state; } class WriteBatchTest : public testing::Test {}; TEST_F(WriteBatchTest, Empty) { WriteBatch batch; ASSERT_EQ("", PrintContents(&batch)); ASSERT_EQ(0, WriteBatchInternal::Count(&batch)); ASSERT_EQ(0, batch.Count()); } TEST_F(WriteBatchTest, Multiple) { WriteBatch batch; batch.Put(Slice("foo"), Slice("bar")); batch.Delete(Slice("box")); batch.Put(Slice("baz"), Slice("boo")); WriteBatchInternal::SetSequence(&batch, 100); ASSERT_EQ(100U, WriteBatchInternal::Sequence(&batch)); ASSERT_EQ(3, WriteBatchInternal::Count(&batch)); ASSERT_EQ("Put(baz, boo)@102" "Delete(box)@101" "Put(foo, bar)@100", PrintContents(&batch)); ASSERT_EQ(3, batch.Count()); } TEST_F(WriteBatchTest, Corruption) { WriteBatch batch; batch.Put(Slice("foo"), Slice("bar")); batch.Delete(Slice("box")); WriteBatchInternal::SetSequence(&batch, 200); Slice contents = WriteBatchInternal::Contents(&batch); WriteBatchInternal::SetContents(&batch, Slice(contents.data(),contents.size()-1)); ASSERT_EQ("Put(foo, bar)@200" "Corruption: bad WriteBatch Delete", PrintContents(&batch)); } TEST_F(WriteBatchTest, Append) { WriteBatch b1, b2; WriteBatchInternal::SetSequence(&b1, 200); WriteBatchInternal::SetSequence(&b2, 300); WriteBatchInternal::Append(&b1, &b2); ASSERT_EQ("", PrintContents(&b1)); ASSERT_EQ(0, b1.Count()); b2.Put("a", "va"); WriteBatchInternal::Append(&b1, &b2); ASSERT_EQ("Put(a, va)@200", PrintContents(&b1)); ASSERT_EQ(1, b1.Count()); b2.Clear(); b2.Put("b", "vb"); WriteBatchInternal::Append(&b1, &b2); ASSERT_EQ("Put(a, va)@200" "Put(b, vb)@201", PrintContents(&b1)); ASSERT_EQ(2, b1.Count()); b2.Delete("foo"); WriteBatchInternal::Append(&b1, &b2); ASSERT_EQ("Put(a, va)@200" "Put(b, vb)@202" "Put(b, vb)@201" "Delete(foo)@203", PrintContents(&b1)); ASSERT_EQ(4, b1.Count()); } TEST_F(WriteBatchTest, SingleDeletion) { WriteBatch batch; WriteBatchInternal::SetSequence(&batch, 100); ASSERT_EQ("", PrintContents(&batch)); ASSERT_EQ(0, batch.Count()); batch.Put("a", "va"); ASSERT_EQ("Put(a, va)@100", PrintContents(&batch)); ASSERT_EQ(1, batch.Count()); batch.SingleDelete("a"); ASSERT_EQ( "SingleDelete(a)@101" "Put(a, va)@100", PrintContents(&batch)); ASSERT_EQ(2, batch.Count()); } namespace { struct TestHandler : public WriteBatch::Handler { std::string seen; virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { if (column_family_id == 0) { seen += "Put(" + key.ToString() + ", " + value.ToString() + ")"; } else { seen += "PutCF(" + ToString(column_family_id) + ", " + key.ToString() + ", " + value.ToString() + ")"; } return Status::OK(); } virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) override { if (column_family_id == 0) { seen += "Delete(" + key.ToString() + ")"; } else { seen += "DeleteCF(" + ToString(column_family_id) + ", " + key.ToString() + ")"; } return Status::OK(); } virtual Status SingleDeleteCF(uint32_t column_family_id, const Slice& key) override { if (column_family_id == 0) { seen += "SingleDelete(" + key.ToString() + ")"; } else { seen += "SingleDeleteCF(" + ToString(column_family_id) + ", " + key.ToString() + ")"; } return Status::OK(); } virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { if (column_family_id == 0) { seen += "Merge(" + key.ToString() + ", " + value.ToString() + ")"; } else { seen += "MergeCF(" + ToString(column_family_id) + ", " + key.ToString() + ", " + value.ToString() + ")"; } return Status::OK(); } virtual void LogData(const Slice& blob) override { seen += "LogData(" + blob.ToString() + ")"; } }; } TEST_F(WriteBatchTest, PutNotImplemented) { WriteBatch batch; batch.Put(Slice("k1"), Slice("v1")); ASSERT_EQ(1, batch.Count()); ASSERT_EQ("Put(k1, v1)@0", PrintContents(&batch)); WriteBatch::Handler handler; ASSERT_OK(batch.Iterate(&handler)); } TEST_F(WriteBatchTest, DeleteNotImplemented) { WriteBatch batch; batch.Delete(Slice("k2")); ASSERT_EQ(1, batch.Count()); ASSERT_EQ("Delete(k2)@0", PrintContents(&batch)); WriteBatch::Handler handler; ASSERT_OK(batch.Iterate(&handler)); } TEST_F(WriteBatchTest, SingleDeleteNotImplemented) { WriteBatch batch; batch.SingleDelete(Slice("k2")); ASSERT_EQ(1, batch.Count()); ASSERT_EQ("SingleDelete(k2)@0", PrintContents(&batch)); WriteBatch::Handler handler; ASSERT_OK(batch.Iterate(&handler)); } TEST_F(WriteBatchTest, MergeNotImplemented) { WriteBatch batch; batch.Merge(Slice("foo"), Slice("bar")); ASSERT_EQ(1, batch.Count()); ASSERT_EQ("Merge(foo, bar)@0", PrintContents(&batch)); WriteBatch::Handler handler; ASSERT_OK(batch.Iterate(&handler)); } TEST_F(WriteBatchTest, Blob) { WriteBatch batch; batch.Put(Slice("k1"), Slice("v1")); batch.Put(Slice("k2"), Slice("v2")); batch.Put(Slice("k3"), Slice("v3")); batch.PutLogData(Slice("blob1")); batch.Delete(Slice("k2")); batch.SingleDelete(Slice("k3")); batch.PutLogData(Slice("blob2")); batch.Merge(Slice("foo"), Slice("bar")); ASSERT_EQ(6, batch.Count()); ASSERT_EQ( "Merge(foo, bar)@5" "Put(k1, v1)@0" "Delete(k2)@3" "Put(k2, v2)@1" "SingleDelete(k3)@4" "Put(k3, v3)@2", PrintContents(&batch)); TestHandler handler; batch.Iterate(&handler); ASSERT_EQ( "Put(k1, v1)" "Put(k2, v2)" "Put(k3, v3)" "LogData(blob1)" "Delete(k2)" "SingleDelete(k3)" "LogData(blob2)" "Merge(foo, bar)", handler.seen); } // It requires more than 30GB of memory to run the test. With single memory // allocation of more than 30GB. // Not all platform can run it. Also it runs a long time. So disable it. TEST_F(WriteBatchTest, DISABLED_ManyUpdates) { // Insert key and value of 3GB and push total batch size to 12GB. static const size_t kKeyValueSize = 4u; static const uint32_t kNumUpdates = 3 << 30; std::string raw(kKeyValueSize, 'A'); WriteBatch batch(kNumUpdates * (4 + kKeyValueSize * 2) + 1024u); char c = 'A'; for (uint32_t i = 0; i < kNumUpdates; i++) { if (c > 'Z') { c = 'A'; } raw[0] = c; raw[raw.length() - 1] = c; c++; batch.Put(raw, raw); } ASSERT_EQ(kNumUpdates, batch.Count()); struct NoopHandler : public WriteBatch::Handler { uint32_t num_seen = 0; char expected_char = 'A'; virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { EXPECT_EQ(kKeyValueSize, key.size()); EXPECT_EQ(kKeyValueSize, value.size()); EXPECT_EQ(expected_char, key[0]); EXPECT_EQ(expected_char, value[0]); EXPECT_EQ(expected_char, key[kKeyValueSize - 1]); EXPECT_EQ(expected_char, value[kKeyValueSize - 1]); expected_char++; if (expected_char > 'Z') { expected_char = 'A'; } ++num_seen; return Status::OK(); } virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) override { EXPECT_TRUE(false); return Status::OK(); } virtual Status SingleDeleteCF(uint32_t column_family_id, const Slice& key) override { EXPECT_TRUE(false); return Status::OK(); } virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { EXPECT_TRUE(false); return Status::OK(); } virtual void LogData(const Slice& blob) override { EXPECT_TRUE(false); } virtual bool Continue() override { return num_seen < kNumUpdates; } } handler; batch.Iterate(&handler); ASSERT_EQ(kNumUpdates, handler.num_seen); } // The test requires more than 18GB memory to run it, with single memory // allocation of more than 12GB. Not all the platform can run it. So disable it. TEST_F(WriteBatchTest, DISABLED_LargeKeyValue) { // Insert key and value of 3GB and push total batch size to 12GB. static const size_t kKeyValueSize = 3221225472u; std::string raw(kKeyValueSize, 'A'); WriteBatch batch(12884901888u + 1024u); for (char i = 0; i < 2; i++) { raw[0] = 'A' + i; raw[raw.length() - 1] = 'A' - i; batch.Put(raw, raw); } ASSERT_EQ(2, batch.Count()); struct NoopHandler : public WriteBatch::Handler { int num_seen = 0; virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { EXPECT_EQ(kKeyValueSize, key.size()); EXPECT_EQ(kKeyValueSize, value.size()); EXPECT_EQ('A' + num_seen, key[0]); EXPECT_EQ('A' + num_seen, value[0]); EXPECT_EQ('A' - num_seen, key[kKeyValueSize - 1]); EXPECT_EQ('A' - num_seen, value[kKeyValueSize - 1]); ++num_seen; return Status::OK(); } virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) override { EXPECT_TRUE(false); return Status::OK(); } virtual Status SingleDeleteCF(uint32_t column_family_id, const Slice& key) override { EXPECT_TRUE(false); return Status::OK(); } virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { EXPECT_TRUE(false); return Status::OK(); } virtual void LogData(const Slice& blob) override { EXPECT_TRUE(false); } virtual bool Continue() override { return num_seen < 2; } } handler; batch.Iterate(&handler); ASSERT_EQ(2, handler.num_seen); } TEST_F(WriteBatchTest, Continue) { WriteBatch batch; struct Handler : public TestHandler { int num_seen = 0; virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { ++num_seen; return TestHandler::PutCF(column_family_id, key, value); } virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) override { ++num_seen; return TestHandler::DeleteCF(column_family_id, key); } virtual Status SingleDeleteCF(uint32_t column_family_id, const Slice& key) override { ++num_seen; return TestHandler::SingleDeleteCF(column_family_id, key); } virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { ++num_seen; return TestHandler::MergeCF(column_family_id, key, value); } virtual void LogData(const Slice& blob) override { ++num_seen; TestHandler::LogData(blob); } virtual bool Continue() override { return num_seen < 5; } } handler; batch.Put(Slice("k1"), Slice("v1")); batch.Put(Slice("k2"), Slice("v2")); batch.PutLogData(Slice("blob1")); batch.Delete(Slice("k1")); batch.SingleDelete(Slice("k2")); batch.PutLogData(Slice("blob2")); batch.Merge(Slice("foo"), Slice("bar")); batch.Iterate(&handler); ASSERT_EQ( "Put(k1, v1)" "Put(k2, v2)" "LogData(blob1)" "Delete(k1)" "SingleDelete(k2)", handler.seen); } TEST_F(WriteBatchTest, PutGatherSlices) { WriteBatch batch; batch.Put(Slice("foo"), Slice("bar")); { // Try a write where the key is one slice but the value is two Slice key_slice("baz"); Slice value_slices[2] = { Slice("header"), Slice("payload") }; batch.Put(SliceParts(&key_slice, 1), SliceParts(value_slices, 2)); } { // One where the key is composite but the value is a single slice Slice key_slices[3] = { Slice("key"), Slice("part2"), Slice("part3") }; Slice value_slice("value"); batch.Put(SliceParts(key_slices, 3), SliceParts(&value_slice, 1)); } WriteBatchInternal::SetSequence(&batch, 100); ASSERT_EQ("Put(baz, headerpayload)@101" "Put(foo, bar)@100" "Put(keypart2part3, value)@102", PrintContents(&batch)); ASSERT_EQ(3, batch.Count()); } namespace { class ColumnFamilyHandleImplDummy : public ColumnFamilyHandleImpl { public: explicit ColumnFamilyHandleImplDummy(int id) : ColumnFamilyHandleImpl(nullptr, nullptr, nullptr), id_(id) {} uint32_t GetID() const override { return id_; } const Comparator* user_comparator() const override { return BytewiseComparator(); } private: uint32_t id_; }; } // namespace anonymous TEST_F(WriteBatchTest, ColumnFamiliesBatchTest) { WriteBatch batch; ColumnFamilyHandleImplDummy zero(0), two(2), three(3), eight(8); batch.Put(&zero, Slice("foo"), Slice("bar")); batch.Put(&two, Slice("twofoo"), Slice("bar2")); batch.Put(&eight, Slice("eightfoo"), Slice("bar8")); batch.Delete(&eight, Slice("eightfoo")); batch.SingleDelete(&two, Slice("twofoo")); batch.Merge(&three, Slice("threethree"), Slice("3three")); batch.Put(&zero, Slice("foo"), Slice("bar")); batch.Merge(Slice("omom"), Slice("nom")); TestHandler handler; batch.Iterate(&handler); ASSERT_EQ( "Put(foo, bar)" "PutCF(2, twofoo, bar2)" "PutCF(8, eightfoo, bar8)" "DeleteCF(8, eightfoo)" "SingleDeleteCF(2, twofoo)" "MergeCF(3, threethree, 3three)" "Put(foo, bar)" "Merge(omom, nom)", handler.seen); } #ifndef ROCKSDB_LITE TEST_F(WriteBatchTest, ColumnFamiliesBatchWithIndexTest) { WriteBatchWithIndex batch; ColumnFamilyHandleImplDummy zero(0), two(2), three(3), eight(8); batch.Put(&zero, Slice("foo"), Slice("bar")); batch.Put(&two, Slice("twofoo"), Slice("bar2")); batch.Put(&eight, Slice("eightfoo"), Slice("bar8")); batch.Delete(&eight, Slice("eightfoo")); batch.SingleDelete(&two, Slice("twofoo")); batch.Merge(&three, Slice("threethree"), Slice("3three")); batch.Put(&zero, Slice("foo"), Slice("bar")); batch.Merge(Slice("omom"), Slice("nom")); std::unique_ptr iter; iter.reset(batch.NewIterator(&eight)); iter->Seek("eightfoo"); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kPutRecord, iter->Entry().type); ASSERT_EQ("eightfoo", iter->Entry().key.ToString()); ASSERT_EQ("bar8", iter->Entry().value.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kDeleteRecord, iter->Entry().type); ASSERT_EQ("eightfoo", iter->Entry().key.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(!iter->Valid()); iter.reset(batch.NewIterator(&two)); iter->Seek("twofoo"); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kPutRecord, iter->Entry().type); ASSERT_EQ("twofoo", iter->Entry().key.ToString()); ASSERT_EQ("bar2", iter->Entry().value.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kSingleDeleteRecord, iter->Entry().type); ASSERT_EQ("twofoo", iter->Entry().key.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(!iter->Valid()); iter.reset(batch.NewIterator()); iter->Seek("gggg"); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kMergeRecord, iter->Entry().type); ASSERT_EQ("omom", iter->Entry().key.ToString()); ASSERT_EQ("nom", iter->Entry().value.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(!iter->Valid()); iter.reset(batch.NewIterator(&zero)); iter->Seek("foo"); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kPutRecord, iter->Entry().type); ASSERT_EQ("foo", iter->Entry().key.ToString()); ASSERT_EQ("bar", iter->Entry().value.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kPutRecord, iter->Entry().type); ASSERT_EQ("foo", iter->Entry().key.ToString()); ASSERT_EQ("bar", iter->Entry().value.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ(WriteType::kMergeRecord, iter->Entry().type); ASSERT_EQ("omom", iter->Entry().key.ToString()); ASSERT_EQ("nom", iter->Entry().value.ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(!iter->Valid()); TestHandler handler; batch.GetWriteBatch()->Iterate(&handler); ASSERT_EQ( "Put(foo, bar)" "PutCF(2, twofoo, bar2)" "PutCF(8, eightfoo, bar8)" "DeleteCF(8, eightfoo)" "SingleDeleteCF(2, twofoo)" "MergeCF(3, threethree, 3three)" "Put(foo, bar)" "Merge(omom, nom)", handler.seen); } #endif // !ROCKSDB_LITE TEST_F(WriteBatchTest, SavePointTest) { Status s; WriteBatch batch; batch.SetSavePoint(); batch.Put("A", "a"); batch.Put("B", "b"); batch.SetSavePoint(); batch.Put("C", "c"); batch.Delete("A"); batch.SetSavePoint(); batch.SetSavePoint(); ASSERT_OK(batch.RollbackToSavePoint()); ASSERT_EQ( "Delete(A)@3" "Put(A, a)@0" "Put(B, b)@1" "Put(C, c)@2", PrintContents(&batch)); ASSERT_OK(batch.RollbackToSavePoint()); ASSERT_OK(batch.RollbackToSavePoint()); ASSERT_EQ( "Put(A, a)@0" "Put(B, b)@1", PrintContents(&batch)); batch.Delete("A"); batch.Put("B", "bb"); ASSERT_OK(batch.RollbackToSavePoint()); ASSERT_EQ("", PrintContents(&batch)); s = batch.RollbackToSavePoint(); ASSERT_TRUE(s.IsNotFound()); ASSERT_EQ("", PrintContents(&batch)); batch.Put("D", "d"); batch.Delete("A"); batch.SetSavePoint(); batch.Put("A", "aaa"); ASSERT_OK(batch.RollbackToSavePoint()); ASSERT_EQ( "Delete(A)@1" "Put(D, d)@0", PrintContents(&batch)); batch.SetSavePoint(); batch.Put("D", "d"); batch.Delete("A"); ASSERT_OK(batch.RollbackToSavePoint()); ASSERT_EQ( "Delete(A)@1" "Put(D, d)@0", PrintContents(&batch)); s = batch.RollbackToSavePoint(); ASSERT_TRUE(s.IsNotFound()); ASSERT_EQ( "Delete(A)@1" "Put(D, d)@0", PrintContents(&batch)); WriteBatch batch2; s = batch2.RollbackToSavePoint(); ASSERT_TRUE(s.IsNotFound()); ASSERT_EQ("", PrintContents(&batch2)); batch2.Delete("A"); batch2.SetSavePoint(); s = batch2.RollbackToSavePoint(); ASSERT_OK(s); ASSERT_EQ("Delete(A)@0", PrintContents(&batch2)); batch2.Clear(); ASSERT_EQ("", PrintContents(&batch2)); batch2.SetSavePoint(); batch2.Delete("B"); ASSERT_EQ("Delete(B)@0", PrintContents(&batch2)); batch2.SetSavePoint(); s = batch2.RollbackToSavePoint(); ASSERT_OK(s); ASSERT_EQ("Delete(B)@0", PrintContents(&batch2)); s = batch2.RollbackToSavePoint(); ASSERT_OK(s); ASSERT_EQ("", PrintContents(&batch2)); s = batch2.RollbackToSavePoint(); ASSERT_TRUE(s.IsNotFound()); ASSERT_EQ("", PrintContents(&batch2)); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/write_callback.h000066400000000000000000000013051267530634000166610ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include "rocksdb/status.h" namespace rocksdb { class DB; class WriteCallback { public: virtual ~WriteCallback() {} // Will be called while on the write thread before the write executes. If // this function returns a non-OK status, the write will be aborted and this // status will be returned to the caller of DB::Write(). virtual Status Callback(DB* db) = 0; }; } // namespace rocksdb rocksdb-4.5.1/db/write_callback_test.cc000066400000000000000000000053601267530634000200630ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include #include "db/db_impl.h" #include "db/write_callback.h" #include "rocksdb/db.h" #include "rocksdb/write_batch.h" #include "util/logging.h" #include "util/testharness.h" using std::string; namespace rocksdb { class WriteCallbackTest : public testing::Test { public: string dbname; WriteCallbackTest() { dbname = test::TmpDir() + "/write_callback_testdb"; } }; class WriteCallbackTestWriteCallback1 : public WriteCallback { public: bool was_called = false; Status Callback(DB *db) override { was_called = true; // Make sure db is a DBImpl DBImpl* db_impl = dynamic_cast (db); if (db_impl == nullptr) { return Status::InvalidArgument(""); } return Status::OK(); } }; class WriteCallbackTestWriteCallback2 : public WriteCallback { public: Status Callback(DB *db) override { return Status::Busy(); } }; TEST_F(WriteCallbackTest, WriteCallBackTest) { Options options; WriteOptions write_options; ReadOptions read_options; string value; DB* db; DBImpl* db_impl; options.create_if_missing = true; Status s = DB::Open(options, dbname, &db); ASSERT_OK(s); db_impl = dynamic_cast (db); ASSERT_TRUE(db_impl); WriteBatch wb; wb.Put("a", "value.a"); wb.Delete("x"); // Test a simple Write s = db->Write(write_options, &wb); ASSERT_OK(s); s = db->Get(read_options, "a", &value); ASSERT_OK(s); ASSERT_EQ("value.a", value); // Test WriteWithCallback WriteCallbackTestWriteCallback1 callback1; WriteBatch wb2; wb2.Put("a", "value.a2"); s = db_impl->WriteWithCallback(write_options, &wb2, &callback1); ASSERT_OK(s); ASSERT_TRUE(callback1.was_called); s = db->Get(read_options, "a", &value); ASSERT_OK(s); ASSERT_EQ("value.a2", value); // Test WriteWithCallback for a callback that fails WriteCallbackTestWriteCallback2 callback2; WriteBatch wb3; wb3.Put("a", "value.a3"); s = db_impl->WriteWithCallback(write_options, &wb3, &callback2); ASSERT_NOK(s); s = db->Get(read_options, "a", &value); ASSERT_OK(s); ASSERT_EQ("value.a2", value); delete db; DestroyDB(dbname, options); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as WriteWithCallback is not supported in ROCKSDB_LITE\n"); return 0; } #endif // !ROCKSDB_LITE rocksdb-4.5.1/db/write_controller.cc000066400000000000000000000075111267530634000174530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/write_controller.h" #include #include #include "rocksdb/env.h" namespace rocksdb { std::unique_ptr WriteController::GetStopToken() { ++total_stopped_; return std::unique_ptr(new StopWriteToken(this)); } std::unique_ptr WriteController::GetDelayToken( uint64_t write_rate) { total_delayed_++; // Reset counters. last_refill_time_ = 0; bytes_left_ = 0; set_delayed_write_rate(write_rate); return std::unique_ptr(new DelayWriteToken(this)); } std::unique_ptr WriteController::GetCompactionPressureToken() { ++total_compaction_pressure_; return std::unique_ptr( new CompactionPressureToken(this)); } bool WriteController::IsStopped() const { return total_stopped_ > 0; } // This is inside DB mutex, so we can't sleep and need to minimize // frequency to get time. // If it turns out to be a performance issue, we can redesign the thread // synchronization model here. // The function trust caller will sleep micros returned. uint64_t WriteController::GetDelay(Env* env, uint64_t num_bytes) { if (total_stopped_ > 0) { return 0; } if (total_delayed_ == 0) { return 0; } const uint64_t kMicrosPerSecond = 1000000; const uint64_t kRefillInterval = 1024U; if (bytes_left_ >= num_bytes) { bytes_left_ -= num_bytes; return 0; } // The frequency to get time inside DB mutex is less than one per refill // interval. auto time_now = env->NowMicros(); uint64_t sleep_debt = 0; uint64_t time_since_last_refill = 0; if (last_refill_time_ != 0) { if (last_refill_time_ > time_now) { sleep_debt = last_refill_time_ - time_now; } else { time_since_last_refill = time_now - last_refill_time_; bytes_left_ += static_cast(static_cast(time_since_last_refill) / kMicrosPerSecond * delayed_write_rate_); if (time_since_last_refill >= kRefillInterval && bytes_left_ > num_bytes) { // If refill interval already passed and we have enough bytes // return without extra sleeping. last_refill_time_ = time_now; bytes_left_ -= num_bytes; return 0; } } } uint64_t single_refill_amount = delayed_write_rate_ * kRefillInterval / kMicrosPerSecond; if (bytes_left_ + single_refill_amount >= num_bytes) { // Wait until a refill interval // Never trigger expire for less than one refill interval to avoid to get // time. bytes_left_ = bytes_left_ + single_refill_amount - num_bytes; last_refill_time_ = time_now + kRefillInterval; return kRefillInterval + sleep_debt; } // Need to refill more than one interval. Need to sleep longer. Check // whether expiration will hit // Sleep just until `num_bytes` is allowed. uint64_t sleep_amount = static_cast(num_bytes / static_cast(delayed_write_rate_) * kMicrosPerSecond) + sleep_debt; last_refill_time_ = time_now + sleep_amount; return sleep_amount; } StopWriteToken::~StopWriteToken() { assert(controller_->total_stopped_ >= 1); --controller_->total_stopped_; } DelayWriteToken::~DelayWriteToken() { controller_->total_delayed_--; assert(controller_->total_delayed_ >= 0); } CompactionPressureToken::~CompactionPressureToken() { controller_->total_compaction_pressure_--; assert(controller_->total_compaction_pressure_ >= 0); } } // namespace rocksdb rocksdb-4.5.1/db/write_controller.h000066400000000000000000000073471267530634000173240ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include namespace rocksdb { class Env; class WriteControllerToken; // WriteController is controlling write stalls in our write code-path. Write // stalls happen when compaction can't keep up with write rate. // All of the methods here (including WriteControllerToken's destructors) need // to be called while holding DB mutex class WriteController { public: explicit WriteController(uint64_t _delayed_write_rate = 1024u * 1024u * 32u) : total_stopped_(0), total_delayed_(0), total_compaction_pressure_(0), bytes_left_(0), last_refill_time_(0) { set_delayed_write_rate(_delayed_write_rate); } ~WriteController() = default; // When an actor (column family) requests a stop token, all writes will be // stopped until the stop token is released (deleted) std::unique_ptr GetStopToken(); // When an actor (column family) requests a delay token, total delay for all // writes to the DB will be controlled under the delayed write rate. Every // write needs to call GetDelay() with number of bytes writing to the DB, // which returns number of microseconds to sleep. std::unique_ptr GetDelayToken( uint64_t delayed_write_rate); // When an actor (column family) requests a moderate token, compaction // threads will be increased std::unique_ptr GetCompactionPressureToken(); // these three metods are querying the state of the WriteController bool IsStopped() const; bool NeedsDelay() const { return total_delayed_ > 0; } bool NeedSpeedupCompaction() const { return IsStopped() || NeedsDelay() || total_compaction_pressure_ > 0; } // return how many microseconds the caller needs to sleep after the call // num_bytes: how many number of bytes to put into the DB. // Prerequisite: DB mutex held. uint64_t GetDelay(Env* env, uint64_t num_bytes); void set_delayed_write_rate(uint64_t write_rate) { // avoid divide 0 if (write_rate == 0) { write_rate = 1u; } delayed_write_rate_ = write_rate; } uint64_t delayed_write_rate() const { return delayed_write_rate_; } private: friend class WriteControllerToken; friend class StopWriteToken; friend class DelayWriteToken; friend class CompactionPressureToken; int total_stopped_; int total_delayed_; int total_compaction_pressure_; uint64_t bytes_left_; uint64_t last_refill_time_; uint64_t delayed_write_rate_; }; class WriteControllerToken { public: explicit WriteControllerToken(WriteController* controller) : controller_(controller) {} virtual ~WriteControllerToken() {} protected: WriteController* controller_; private: // no copying allowed WriteControllerToken(const WriteControllerToken&) = delete; void operator=(const WriteControllerToken&) = delete; }; class StopWriteToken : public WriteControllerToken { public: explicit StopWriteToken(WriteController* controller) : WriteControllerToken(controller) {} virtual ~StopWriteToken(); }; class DelayWriteToken : public WriteControllerToken { public: explicit DelayWriteToken(WriteController* controller) : WriteControllerToken(controller) {} virtual ~DelayWriteToken(); }; class CompactionPressureToken : public WriteControllerToken { public: explicit CompactionPressureToken(WriteController* controller) : WriteControllerToken(controller) {} virtual ~CompactionPressureToken(); }; } // namespace rocksdb rocksdb-4.5.1/db/write_controller_test.cc000066400000000000000000000110051267530634000205030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include "db/write_controller.h" #include "rocksdb/env.h" #include "util/testharness.h" namespace rocksdb { class WriteControllerTest : public testing::Test {}; class TimeSetEnv : public EnvWrapper { public: explicit TimeSetEnv() : EnvWrapper(nullptr) {} uint64_t now_micros_ = 6666; virtual uint64_t NowMicros() override { return now_micros_; } }; TEST_F(WriteControllerTest, ChangeDelayRateTest) { TimeSetEnv env; WriteController controller(10000000u); auto delay_token_0 = controller.GetDelayToken(controller.delayed_write_rate()); ASSERT_EQ(static_cast(2000000), controller.GetDelay(&env, 20000000u)); auto delay_token_1 = controller.GetDelayToken(2000000u); ASSERT_EQ(static_cast(10000000), controller.GetDelay(&env, 20000000u)); auto delay_token_2 = controller.GetDelayToken(1000000u); ASSERT_EQ(static_cast(20000000), controller.GetDelay(&env, 20000000u)); auto delay_token_3 = controller.GetDelayToken(20000000u); ASSERT_EQ(static_cast(1000000), controller.GetDelay(&env, 20000000u)); auto delay_token_4 = controller.GetDelayToken(controller.delayed_write_rate() * 2); ASSERT_EQ(static_cast(500000), controller.GetDelay(&env, 20000000u)); } TEST_F(WriteControllerTest, SanityTest) { WriteController controller(10000000u); auto stop_token_1 = controller.GetStopToken(); auto stop_token_2 = controller.GetStopToken(); ASSERT_TRUE(controller.IsStopped()); stop_token_1.reset(); ASSERT_TRUE(controller.IsStopped()); stop_token_2.reset(); ASSERT_FALSE(controller.IsStopped()); TimeSetEnv env; auto delay_token_1 = controller.GetDelayToken(10000000u); ASSERT_EQ(static_cast(2000000), controller.GetDelay(&env, 20000000u)); env.now_micros_ += 1999900u; // sleep debt 1000 auto delay_token_2 = controller.GetDelayToken(10000000u); // Rate reset after changing the token. ASSERT_EQ(static_cast(2000000), controller.GetDelay(&env, 20000000u)); env.now_micros_ += 1999900u; // sleep debt 1000 // One refill: 10240 bytes allowed, 1000 used, 9240 left ASSERT_EQ(static_cast(1124), controller.GetDelay(&env, 1000u)); env.now_micros_ += 1124u; // sleep debt 0 delay_token_2.reset(); // 1000 used, 8240 left ASSERT_EQ(static_cast(0), controller.GetDelay(&env, 1000u)); env.now_micros_ += 100u; // sleep credit 100 // 1000 used, 7240 left ASSERT_EQ(static_cast(0), controller.GetDelay(&env, 1000u)); env.now_micros_ += 100u; // sleep credit 200 // One refill: 10240 fileed, sleep credit generates 2000. 8000 used // 7240 + 10240 + 2000 - 8000 = 11480 left ASSERT_EQ(static_cast(1024u), controller.GetDelay(&env, 8000u)); env.now_micros_ += 200u; // sleep debt 824 // 1000 used, 10480 left. ASSERT_EQ(static_cast(0), controller.GetDelay(&env, 1000u)); env.now_micros_ += 200u; // sleep debt 624 // Out of bound sleep, still 10480 left ASSERT_EQ(static_cast(3000624u), controller.GetDelay(&env, 30000000u)); env.now_micros_ += 3000724u; // sleep credit 100 // 6000 used, 4480 left. ASSERT_EQ(static_cast(0), controller.GetDelay(&env, 6000u)); env.now_micros_ += 200u; // sleep credit 300 // One refill, credit 4480 balance + 3000 credit + 10240 refill // Use 8000, 9720 left ASSERT_EQ(static_cast(1024u), controller.GetDelay(&env, 8000u)); env.now_micros_ += 3024u; // sleep credit 2000 // 1720 left ASSERT_EQ(static_cast(0u), controller.GetDelay(&env, 8000u)); // 1720 balance + 20000 credit = 20170 left // Use 8000, 12170 left ASSERT_EQ(static_cast(0u), controller.GetDelay(&env, 8000u)); // 4170 left ASSERT_EQ(static_cast(0u), controller.GetDelay(&env, 8000u)); // Need a refill ASSERT_EQ(static_cast(1024u), controller.GetDelay(&env, 9000u)); delay_token_1.reset(); ASSERT_EQ(static_cast(0), controller.GetDelay(&env, 30000000u)); delay_token_1.reset(); ASSERT_FALSE(controller.IsStopped()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/db/write_thread.cc000066400000000000000000000376161267530634000165500ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/write_thread.h" #include #include #include #include "db/column_family.h" #include "port/port.h" #include "util/sync_point.h" namespace rocksdb { WriteThread::WriteThread(uint64_t max_yield_usec, uint64_t slow_yield_usec) : max_yield_usec_(max_yield_usec), slow_yield_usec_(slow_yield_usec), newest_writer_(nullptr) {} uint8_t WriteThread::BlockingAwaitState(Writer* w, uint8_t goal_mask) { // We're going to block. Lazily create the mutex. We guarantee // propagation of this construction to the waker via the // STATE_LOCKED_WAITING state. The waker won't try to touch the mutex // or the condvar unless they CAS away the STATE_LOCKED_WAITING that // we install below. w->CreateMutex(); auto state = w->state.load(std::memory_order_acquire); assert(state != STATE_LOCKED_WAITING); if ((state & goal_mask) == 0 && w->state.compare_exchange_strong(state, STATE_LOCKED_WAITING)) { // we have permission (and an obligation) to use StateMutex std::unique_lock guard(w->StateMutex()); w->StateCV().wait(guard, [w] { return w->state.load(std::memory_order_relaxed) != STATE_LOCKED_WAITING; }); state = w->state.load(std::memory_order_relaxed); } // else tricky. Goal is met or CAS failed. In the latter case the waker // must have changed the state, and compare_exchange_strong has updated // our local variable with the new one. At the moment WriteThread never // waits for a transition across intermediate states, so we know that // since a state change has occurred the goal must have been met. assert((state & goal_mask) != 0); return state; } uint8_t WriteThread::AwaitState(Writer* w, uint8_t goal_mask, AdaptationContext* ctx) { uint8_t state; // On a modern Xeon each loop takes about 7 nanoseconds (most of which // is the effect of the pause instruction), so 200 iterations is a bit // more than a microsecond. This is long enough that waits longer than // this can amortize the cost of accessing the clock and yielding. for (uint32_t tries = 0; tries < 200; ++tries) { state = w->state.load(std::memory_order_acquire); if ((state & goal_mask) != 0) { return state; } port::AsmVolatilePause(); } // If we're only going to end up waiting a short period of time, // it can be a lot more efficient to call std::this_thread::yield() // in a loop than to block in StateMutex(). For reference, on my 4.0 // SELinux test server with support for syscall auditing enabled, the // minimum latency between FUTEX_WAKE to returning from FUTEX_WAIT is // 2.7 usec, and the average is more like 10 usec. That can be a big // drag on RockDB's single-writer design. Of course, spinning is a // bad idea if other threads are waiting to run or if we're going to // wait for a long time. How do we decide? // // We break waiting into 3 categories: short-uncontended, // short-contended, and long. If we had an oracle, then we would always // spin for short-uncontended, always block for long, and our choice for // short-contended might depend on whether we were trying to optimize // RocksDB throughput or avoid being greedy with system resources. // // Bucketing into short or long is easy by measuring elapsed time. // Differentiating short-uncontended from short-contended is a bit // trickier, but not too bad. We could look for involuntary context // switches using getrusage(RUSAGE_THREAD, ..), but it's less work // (portability code and CPU) to just look for yield calls that take // longer than we expect. sched_yield() doesn't actually result in any // context switch overhead if there are no other runnable processes // on the current core, in which case it usually takes less than // a microsecond. // // There are two primary tunables here: the threshold between "short" // and "long" waits, and the threshold at which we suspect that a yield // is slow enough to indicate we should probably block. If these // thresholds are chosen well then CPU-bound workloads that don't // have more threads than cores will experience few context switches // (voluntary or involuntary), and the total number of context switches // (voluntary and involuntary) will not be dramatically larger (maybe // 2x) than the number of voluntary context switches that occur when // --max_yield_wait_micros=0. // // There's another constant, which is the number of slow yields we will // tolerate before reversing our previous decision. Solitary slow // yields are pretty common (low-priority small jobs ready to run), // so this should be at least 2. We set this conservatively to 3 so // that we can also immediately schedule a ctx adaptation, rather than // waiting for the next update_ctx. const size_t kMaxSlowYieldsWhileSpinning = 3; bool update_ctx = false; bool would_spin_again = false; if (max_yield_usec_ > 0) { update_ctx = Random::GetTLSInstance()->OneIn(256); if (update_ctx || ctx->value.load(std::memory_order_relaxed) >= 0) { // we're updating the adaptation statistics, or spinning has > // 50% chance of being shorter than max_yield_usec_ and causing no // involuntary context switches auto spin_begin = std::chrono::steady_clock::now(); // this variable doesn't include the final yield (if any) that // causes the goal to be met size_t slow_yield_count = 0; auto iter_begin = spin_begin; while ((iter_begin - spin_begin) <= std::chrono::microseconds(max_yield_usec_)) { std::this_thread::yield(); state = w->state.load(std::memory_order_acquire); if ((state & goal_mask) != 0) { // success would_spin_again = true; break; } auto now = std::chrono::steady_clock::now(); if (now == iter_begin || now - iter_begin >= std::chrono::microseconds(slow_yield_usec_)) { // conservatively count it as a slow yield if our clock isn't // accurate enough to measure the yield duration ++slow_yield_count; if (slow_yield_count >= kMaxSlowYieldsWhileSpinning) { // Not just one ivcsw, but several. Immediately update ctx // and fall back to blocking update_ctx = true; break; } } iter_begin = now; } } } if ((state & goal_mask) == 0) { state = BlockingAwaitState(w, goal_mask); } if (update_ctx) { auto v = ctx->value.load(std::memory_order_relaxed); // fixed point exponential decay with decay constant 1/1024, with +1 // and -1 scaled to avoid overflow for int32_t v = v + (v / 1024) + (would_spin_again ? 1 : -1) * 16384; ctx->value.store(v, std::memory_order_relaxed); } assert((state & goal_mask) != 0); return state; } void WriteThread::SetState(Writer* w, uint8_t new_state) { auto state = w->state.load(std::memory_order_acquire); if (state == STATE_LOCKED_WAITING || !w->state.compare_exchange_strong(state, new_state)) { assert(state == STATE_LOCKED_WAITING); std::lock_guard guard(w->StateMutex()); assert(w->state.load(std::memory_order_relaxed) != new_state); w->state.store(new_state, std::memory_order_relaxed); w->StateCV().notify_one(); } } void WriteThread::LinkOne(Writer* w, bool* linked_as_leader) { assert(w->state == STATE_INIT); Writer* writers = newest_writer_.load(std::memory_order_relaxed); while (true) { w->link_older = writers; if (newest_writer_.compare_exchange_strong(writers, w)) { if (writers == nullptr) { // this isn't part of the WriteThread machinery, but helps with // debugging and is checked by an assert in WriteImpl w->state.store(STATE_GROUP_LEADER, std::memory_order_relaxed); } *linked_as_leader = (writers == nullptr); return; } } } void WriteThread::CreateMissingNewerLinks(Writer* head) { while (true) { Writer* next = head->link_older; if (next == nullptr || next->link_newer != nullptr) { assert(next == nullptr || next->link_newer == head); break; } next->link_newer = head; head = next; } } void WriteThread::JoinBatchGroup(Writer* w) { static AdaptationContext ctx("JoinBatchGroup"); assert(w->batch != nullptr); bool linked_as_leader; LinkOne(w, &linked_as_leader); if (!linked_as_leader) { AwaitState(w, STATE_GROUP_LEADER | STATE_PARALLEL_FOLLOWER | STATE_COMPLETED, &ctx); } } size_t WriteThread::EnterAsBatchGroupLeader( Writer* leader, WriteThread::Writer** last_writer, autovector* write_batch_group) { assert(leader->link_older == nullptr); assert(leader->batch != nullptr); size_t size = WriteBatchInternal::ByteSize(leader->batch); write_batch_group->push_back(leader->batch); // Allow the group to grow up to a maximum size, but if the // original write is small, limit the growth so we do not slow // down the small write too much. size_t max_size = 1 << 20; if (size <= (128 << 10)) { max_size = size + (128 << 10); } *last_writer = leader; if (leader->has_callback) { // TODO(agiardullo:) Batching not currently supported as this write may // fail if the callback function decides to abort this write. return size; } Writer* newest_writer = newest_writer_.load(std::memory_order_acquire); // This is safe regardless of any db mutex status of the caller. Previous // calls to ExitAsGroupLeader either didn't call CreateMissingNewerLinks // (they emptied the list and then we added ourself as leader) or had to // explicitly wake us up (the list was non-empty when we added ourself, // so we have already received our MarkJoined). CreateMissingNewerLinks(newest_writer); // Tricky. Iteration start (leader) is exclusive and finish // (newest_writer) is inclusive. Iteration goes from old to new. Writer* w = leader; while (w != newest_writer) { w = w->link_newer; if (w->sync && !leader->sync) { // Do not include a sync write into a batch handled by a non-sync write. break; } if (!w->disableWAL && leader->disableWAL) { // Do not include a write that needs WAL into a batch that has // WAL disabled. break; } if (w->has_callback) { // Do not include writes which may be aborted if the callback does not // succeed. break; } if (w->batch == nullptr) { // Do not include those writes with nullptr batch. Those are not writes, // those are something else. They want to be alone break; } auto batch_size = WriteBatchInternal::ByteSize(w->batch); if (size + batch_size > max_size) { // Do not make batch too big break; } size += batch_size; write_batch_group->push_back(w->batch); w->in_batch_group = true; *last_writer = w; } return size; } void WriteThread::LaunchParallelFollowers(ParallelGroup* pg, SequenceNumber sequence) { // EnterAsBatchGroupLeader already created the links from leader to // newer writers in the group pg->leader->parallel_group = pg; Writer* w = pg->leader; w->sequence = sequence; while (w != pg->last_writer) { sequence += WriteBatchInternal::Count(w->batch); w = w->link_newer; w->sequence = sequence; w->parallel_group = pg; SetState(w, STATE_PARALLEL_FOLLOWER); } } bool WriteThread::CompleteParallelWorker(Writer* w) { static AdaptationContext ctx("CompleteParallelWorker"); auto* pg = w->parallel_group; if (!w->status.ok()) { std::lock_guard guard(w->StateMutex()); pg->status = w->status; } auto leader = pg->leader; auto early_exit_allowed = pg->early_exit_allowed; if (pg->running.load(std::memory_order_acquire) > 1 && pg->running-- > 1) { // we're not the last one AwaitState(w, STATE_COMPLETED, &ctx); // Caller only needs to perform exit duties if early exit doesn't // apply and this is the leader. Can't touch pg here. Whoever set // our state to STATE_COMPLETED copied pg->status to w.status for us. return w == leader && !(early_exit_allowed && w->status.ok()); } // else we're the last parallel worker if (w == leader || (early_exit_allowed && pg->status.ok())) { // this thread should perform exit duties w->status = pg->status; return true; } else { // We're the last parallel follower but early commit is not // applicable. Wake up the leader and then wait for it to exit. assert(w->state == STATE_PARALLEL_FOLLOWER); SetState(leader, STATE_COMPLETED); AwaitState(w, STATE_COMPLETED, &ctx); return false; } } void WriteThread::EarlyExitParallelGroup(Writer* w) { auto* pg = w->parallel_group; assert(w->state == STATE_PARALLEL_FOLLOWER); assert(pg->status.ok()); ExitAsBatchGroupLeader(pg->leader, pg->last_writer, pg->status); assert(w->state == STATE_COMPLETED); assert(w->status.ok()); SetState(pg->leader, STATE_COMPLETED); } void WriteThread::ExitAsBatchGroupLeader(Writer* leader, Writer* last_writer, Status status) { assert(leader->link_older == nullptr); Writer* head = newest_writer_.load(std::memory_order_acquire); if (head != last_writer || !newest_writer_.compare_exchange_strong(head, nullptr)) { // Either w wasn't the head during the load(), or it was the head // during the load() but somebody else pushed onto the list before // we did the compare_exchange_strong (causing it to fail). In the // latter case compare_exchange_strong has the effect of re-reading // its first param (head). No need to retry a failing CAS, because // only a departing leader (which we are at the moment) can remove // nodes from the list. assert(head != last_writer); // After walking link_older starting from head (if not already done) // we will be able to traverse w->link_newer below. This function // can only be called from an active leader, only a leader can // clear newest_writer_, we didn't, and only a clear newest_writer_ // could cause the next leader to start their work without a call // to MarkJoined, so we can definitely conclude that no other leader // work is going on here (with or without db mutex). CreateMissingNewerLinks(head); assert(last_writer->link_newer->link_older == last_writer); last_writer->link_newer->link_older = nullptr; // Next leader didn't self-identify, because newest_writer_ wasn't // nullptr when they enqueued (we were definitely enqueued before them // and are still in the list). That means leader handoff occurs when // we call MarkJoined SetState(last_writer->link_newer, STATE_GROUP_LEADER); } // else nobody else was waiting, although there might already be a new // leader now while (last_writer != leader) { last_writer->status = status; // we need to read link_older before calling SetState, because as soon // as it is marked committed the other thread's Await may return and // deallocate the Writer. auto next = last_writer->link_older; SetState(last_writer, STATE_COMPLETED); last_writer = next; } } void WriteThread::EnterUnbatched(Writer* w, InstrumentedMutex* mu) { static AdaptationContext ctx("EnterUnbatched"); assert(w->batch == nullptr); bool linked_as_leader; LinkOne(w, &linked_as_leader); if (!linked_as_leader) { mu->Unlock(); TEST_SYNC_POINT("WriteThread::EnterUnbatched:Wait"); AwaitState(w, STATE_GROUP_LEADER, &ctx); mu->Lock(); } } void WriteThread::ExitUnbatched(Writer* w) { Status dummy_status; ExitAsBatchGroupLeader(w, w, dummy_status); } } // namespace rocksdb rocksdb-4.5.1/db/write_thread.h000066400000000000000000000225701267530634000164030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include #include #include #include #include #include "db/write_batch_internal.h" #include "rocksdb/status.h" #include "util/autovector.h" #include "util/instrumented_mutex.h" namespace rocksdb { class WriteThread { public: enum State : uint8_t { // The initial state of a writer. This is a Writer that is // waiting in JoinBatchGroup. This state can be left when another // thread informs the waiter that it has become a group leader // (-> STATE_GROUP_LEADER), when a leader that has chosen to be // non-parallel informs a follower that its writes have been committed // (-> STATE_COMPLETED), or when a leader that has chosen to perform // updates in parallel and needs this Writer to apply its batch (-> // STATE_PARALLEL_FOLLOWER). STATE_INIT = 1, // The state used to inform a waiting Writer that it has become the // leader, and it should now build a write batch group. Tricky: // this state is not used if newest_writer_ is empty when a writer // enqueues itself, because there is no need to wait (or even to // create the mutex and condvar used to wait) in that case. This is // a terminal state unless the leader chooses to make this a parallel // batch, in which case the last parallel worker to finish will move // the leader to STATE_COMPLETED. STATE_GROUP_LEADER = 2, // A Writer that has returned as a follower in a parallel group. // It should apply its batch to the memtable and then call // CompleteParallelWorker. When someone calls ExitAsBatchGroupLeader // or EarlyExitParallelGroup this state will get transitioned to // STATE_COMPLETED. STATE_PARALLEL_FOLLOWER = 4, // A follower whose writes have been applied, or a parallel leader // whose followers have all finished their work. This is a terminal // state. STATE_COMPLETED = 8, // A state indicating that the thread may be waiting using StateMutex() // and StateCondVar() STATE_LOCKED_WAITING = 16, }; struct Writer; struct ParallelGroup { Writer* leader; Writer* last_writer; bool early_exit_allowed; // before running goes to zero, status needs leader->StateMutex() Status status; std::atomic running; }; // Information kept for every waiting writer. struct Writer { WriteBatch* batch; bool sync; bool disableWAL; bool in_batch_group; bool has_callback; bool made_waitable; // records lazy construction of mutex and cv std::atomic state; // write under StateMutex() or pre-link ParallelGroup* parallel_group; SequenceNumber sequence; // the sequence number to use Status status; std::aligned_storage::type state_mutex_bytes; std::aligned_storage::type state_cv_bytes; Writer* link_older; // read/write only before linking, or as leader Writer* link_newer; // lazy, read/write only before linking, or as leader Writer() : batch(nullptr), sync(false), disableWAL(false), in_batch_group(false), has_callback(false), made_waitable(false), state(STATE_INIT), link_older(nullptr), link_newer(nullptr) {} ~Writer() { if (made_waitable) { StateMutex().~mutex(); StateCV().~condition_variable(); } } void CreateMutex() { if (!made_waitable) { // Note that made_waitable is tracked separately from state // transitions, because we can't atomically create the mutex and // link into the list. made_waitable = true; new (&state_mutex_bytes) std::mutex; new (&state_cv_bytes) std::condition_variable; } } // No other mutexes may be acquired while holding StateMutex(), it is // always last in the order std::mutex& StateMutex() { assert(made_waitable); return *static_cast(static_cast(&state_mutex_bytes)); } std::condition_variable& StateCV() { assert(made_waitable); return *static_cast( static_cast(&state_cv_bytes)); } }; WriteThread(uint64_t max_yield_usec, uint64_t slow_yield_usec); // IMPORTANT: None of the methods in this class rely on the db mutex // for correctness. All of the methods except JoinBatchGroup and // EnterUnbatched may be called either with or without the db mutex held. // Correctness is maintained by ensuring that only a single thread is // a leader at a time. // Registers w as ready to become part of a batch group, waits until the // caller should perform some work, and returns the current state of the // writer. If w has become the leader of a write batch group, returns // STATE_GROUP_LEADER. If w has been made part of a sequential batch // group and the leader has performed the write, returns STATE_DONE. // If w has been made part of a parallel batch group and is reponsible // for updating the memtable, returns STATE_PARALLEL_FOLLOWER. // // The db mutex SHOULD NOT be held when calling this function, because // it will block. // // Writer* w: Writer to be executed as part of a batch group void JoinBatchGroup(Writer* w); // Constructs a write batch group led by leader, which should be a // Writer passed to JoinBatchGroup on the current thread. // // Writer* leader: Writer that is STATE_GROUP_LEADER // Writer** last_writer: Out-param that identifies the last follower // autovector* write_batch_group: Out-param of group members // returns: Total batch group byte size size_t EnterAsBatchGroupLeader(Writer* leader, Writer** last_writer, autovector* write_batch_group); // Causes JoinBatchGroup to return STATE_PARALLEL_FOLLOWER for all of the // non-leader members of this write batch group. Sets Writer::sequence // before waking them up. // // ParallalGroup* pg: Extra state used to coordinate the parallel add // SequenceNumber sequence: Starting sequence number to assign to Writer-s void LaunchParallelFollowers(ParallelGroup* pg, SequenceNumber sequence); // Reports the completion of w's batch to the parallel group leader, and // waits for the rest of the parallel batch to complete. Returns true // if this thread is the last to complete, and hence should advance // the sequence number and then call EarlyExitParallelGroup, false if // someone else has already taken responsibility for that. bool CompleteParallelWorker(Writer* w); // This method performs an early completion of a parallel write group, // where the cleanup work of the leader is performed by a follower who // happens to be the last parallel worker to complete. void EarlyExitParallelGroup(Writer* w); // Unlinks the Writer-s in a batch group, wakes up the non-leaders, // and wakes up the next leader (if any). // // Writer* leader: From EnterAsBatchGroupLeader // Writer* last_writer: Value of out-param of EnterAsBatchGroupLeader // Status status: Status of write operation void ExitAsBatchGroupLeader(Writer* leader, Writer* last_writer, Status status); // Waits for all preceding writers (unlocking mu while waiting), then // registers w as the currently proceeding writer. // // Writer* w: A Writer not eligible for batching // InstrumentedMutex* mu: The db mutex, to unlock while waiting // REQUIRES: db mutex held void EnterUnbatched(Writer* w, InstrumentedMutex* mu); // Completes a Writer begun with EnterUnbatched, unblocking subsequent // writers. void ExitUnbatched(Writer* w); struct AdaptationContext { const char* name; std::atomic value; explicit AdaptationContext(const char* name0) : name(name0), value(0) {} }; private: uint64_t max_yield_usec_; uint64_t slow_yield_usec_; // Points to the newest pending Writer. Only leader can remove // elements, adding can be done lock-free by anybody std::atomic newest_writer_; // Waits for w->state & goal_mask using w->StateMutex(). Returns // the state that satisfies goal_mask. uint8_t BlockingAwaitState(Writer* w, uint8_t goal_mask); // Blocks until w->state & goal_mask, returning the state value // that satisfied the predicate. Uses ctx to adaptively use // std::this_thread::yield() to avoid mutex overheads. ctx should be // a context-dependent static. uint8_t AwaitState(Writer* w, uint8_t goal_mask, AdaptationContext* ctx); void SetState(Writer* w, uint8_t new_state); // Links w into the newest_writer_ list. Sets *linked_as_leader to // true if w was linked directly into the leader position. Safe to // call from multiple threads without external locking. void LinkOne(Writer* w, bool* linked_as_leader); // Computes any missing link_newer links. Should not be called // concurrently with itself. void CreateMissingNewerLinks(Writer* head); }; } // namespace rocksdb rocksdb-4.5.1/db/writebuffer.h000066400000000000000000000027631267530634000162500ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // WriteBuffer is for managing memory allocation for one or more MemTables. #pragma once #include namespace rocksdb { class WriteBuffer { public: explicit WriteBuffer(size_t _buffer_size) : buffer_size_(_buffer_size), memory_used_(0) {} ~WriteBuffer() {} size_t memory_usage() const { return memory_used_.load(std::memory_order_relaxed); } size_t buffer_size() const { return buffer_size_; } // Should only be called from write thread bool ShouldFlush() const { return buffer_size() > 0 && memory_usage() >= buffer_size(); } // Should only be called from write thread void ReserveMem(size_t mem) { memory_used_.fetch_add(mem, std::memory_order_relaxed); } void FreeMem(size_t mem) { memory_used_.fetch_sub(mem, std::memory_order_relaxed); } private: const size_t buffer_size_; std::atomic memory_used_; // No copying allowed WriteBuffer(const WriteBuffer&); void operator=(const WriteBuffer&); }; } // namespace rocksdb rocksdb-4.5.1/db/xfunc_test_points.cc000066400000000000000000000104111267530634000176250ustar00rootroot00000000000000// Copyright (c) 2016, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "db/xfunc_test_points.h" #include "util/xfunc.h" namespace rocksdb { #ifdef XFUNC void xf_manage_release(ManagedIterator* iter) { if (!(XFuncPoint::GetSkip() & kSkipNoPrefix)) { iter->ReleaseIter(false); } } void xf_manage_create(ManagedIterator* iter) { iter->SetDropOld(false); } void xf_manage_new(DBImpl* db, ReadOptions* read_options, bool is_snapshot_supported) { if ((!XFuncPoint::Check("managed_xftest_dropold") && (!XFuncPoint::Check("managed_xftest_release"))) || (!read_options->managed)) { return; } if ((!read_options->tailing) && (read_options->snapshot == nullptr) && (!is_snapshot_supported)) { read_options->managed = false; return; } if (db->GetOptions().prefix_extractor != nullptr) { if (strcmp(db->GetOptions().table_factory.get()->Name(), "PlainTable")) { if (!(XFuncPoint::GetSkip() & kSkipNoPrefix)) { read_options->total_order_seek = true; } } else { read_options->managed = false; } } } class XFTransactionWriteHandler : public WriteBatch::Handler { public: Transaction* txn_; DBImpl* db_impl_; XFTransactionWriteHandler(Transaction* txn, DBImpl* db_impl) : txn_(txn), db_impl_(db_impl) {} virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { InstrumentedMutexLock l(&db_impl_->mutex_); ColumnFamilyHandle* cfh = db_impl_->GetColumnFamilyHandle(column_family_id); if (cfh == nullptr) { return Status::InvalidArgument( "XFUNC test could not find column family " "handle for id ", ToString(column_family_id)); } txn_->Put(cfh, key, value); return Status::OK(); } virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) override { InstrumentedMutexLock l(&db_impl_->mutex_); ColumnFamilyHandle* cfh = db_impl_->GetColumnFamilyHandle(column_family_id); if (cfh == nullptr) { return Status::InvalidArgument( "XFUNC test could not find column family " "handle for id ", ToString(column_family_id)); } txn_->Merge(cfh, key, value); return Status::OK(); } virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) override { InstrumentedMutexLock l(&db_impl_->mutex_); ColumnFamilyHandle* cfh = db_impl_->GetColumnFamilyHandle(column_family_id); if (cfh == nullptr) { return Status::InvalidArgument( "XFUNC test could not find column family " "handle for id ", ToString(column_family_id)); } txn_->Delete(cfh, key); return Status::OK(); } virtual void LogData(const Slice& blob) override { txn_->PutLogData(blob); } }; // Whenever DBImpl::Write is called, create a transaction and do the write via // the transaction. void xf_transaction_write(const WriteOptions& write_options, const DBOptions& db_options, WriteBatch* my_batch, WriteCallback* callback, DBImpl* db_impl, Status* s, bool* write_attempted) { if (callback != nullptr) { // We may already be in a transaction, don't force a transaction *write_attempted = false; return; } OptimisticTransactionDB* txn_db = new OptimisticTransactionDB(db_impl); Transaction* txn = Transaction::BeginTransaction(txn_db, write_options); XFTransactionWriteHandler handler(txn, db_impl); *s = my_batch->Iterate(&handler); if (!s->ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options.info_log, "XFUNC test could not iterate batch. status: $s\n", s->ToString().c_str()); } *s = txn->Commit(); if (!s->ok()) { Log(InfoLogLevel::ERROR_LEVEL, db_options.info_log, "XFUNC test could not commit transaction. status: $s\n", s->ToString().c_str()); } *write_attempted = true; delete txn; delete txn_db; } #endif // XFUNC } // namespace rocksdb rocksdb-4.5.1/db/xfunc_test_points.h000066400000000000000000000022071267530634000174730ustar00rootroot00000000000000// Copyright (c) 2016, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include "db/db_impl.h" #include "db/managed_iterator.h" #include "db/write_callback.h" #include "rocksdb/options.h" #include "rocksdb/write_batch.h" #include "util/xfunc.h" namespace rocksdb { #ifdef XFUNC // DB-specific test points for the cross-functional test framework (see // util/xfunc.h). void xf_manage_release(ManagedIterator* iter); void xf_manage_create(ManagedIterator* iter); void xf_manage_new(DBImpl* db, ReadOptions* readoptions, bool is_snapshot_supported); void xf_transaction_write(const WriteOptions& write_options, const DBOptions& db_options, class WriteBatch* my_batch, class WriteCallback* callback, DBImpl* db_impl, Status* success, bool* write_attempted); #endif // XFUNC } // namespace rocksdb rocksdb-4.5.1/doc/000077500000000000000000000000001267530634000137235ustar00rootroot00000000000000rocksdb-4.5.1/doc/doc.css000066400000000000000000000021471267530634000152060ustar00rootroot00000000000000body { margin-left: 0.5in; margin-right: 0.5in; background: white; color: black; } h1 { margin-left: -0.2in; font-size: 14pt; } h2 { margin-left: -0in; font-size: 12pt; } h3 { margin-left: -0in; } h4 { margin-left: -0in; } hr { margin-left: -0in; } /* Definition lists: definition term bold */ dt { font-weight: bold; } address { text-align: center; } code,samp,var { color: blue; } kbd { color: #600000; } div.note p { float: right; width: 3in; margin-right: 0%; padding: 1px; border: 2px solid #6060a0; background-color: #fffff0; } ul { margin-top: -0em; margin-bottom: -0em; } ol { margin-top: -0em; margin-bottom: -0em; } UL.nobullets { list-style-type: none; list-style-image: none; margin-left: -1em; } p { margin: 1em 0 1em 0; padding: 0 0 0 0; } pre { line-height: 1.3em; padding: 0.4em 0 0.8em 0; margin: 0 0 0 0; border: 0 0 0 0; color: blue; } .datatable { margin-left: auto; margin-right: auto; margin-top: 2em; margin-bottom: 2em; border: 1px solid; } .datatable td,th { padding: 0 0.5em 0 0.5em; text-align: right; } rocksdb-4.5.1/doc/index.html000066400000000000000000001056261267530634000157320ustar00rootroot00000000000000 RocksDB

RocksDB

The Facebook Database Engineering Team
Build on earlier work on leveldb by Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com)

The rocksdb library provides a persistent key value store. Keys and values are arbitrary byte arrays. The keys are ordered within the key value store according to a user-specified comparator function.

Opening A Database

A rocksdb database has a name which corresponds to a file system directory. All of the contents of database are stored in this directory. The following example shows how to open a database, creating it if necessary: 

  #include <assert>
  #include "rocksdb/db.h"

  rocksdb::DB* db;
  rocksdb::Options options;
  options.create_if_missing = true;
  rocksdb::Status status = rocksdb::DB::Open(options, "/tmp/testdb", &db);
  assert(status.ok());
  ...
If you want to raise an error if the database already exists, add the following line before the rocksdb::DB::Open call: 
  options.error_if_exists = true;

Status

You may have noticed the rocksdb::Status type above. Values of this type are returned by most functions in rocksdb that may encounter an error. You can check if such a result is ok, and also print an associated error message: 

   rocksdb::Status s = ...;
   if (!s.ok()) cerr << s.ToString() << endl;

Closing A Database

When you are done with a database, just delete the database object. Example: 

  ... open the db as described above ...
  ... do something with db ...
  delete db;

Reads And Writes

The database provides Put, Delete, and Get methods to modify/query the database. For example, the following code moves the value stored under key1 to key2. 

  std::string value;
  rocksdb::Status s = db->Get(rocksdb::ReadOptions(), key1, &value);
  if (s.ok()) s = db->Put(rocksdb::WriteOptions(), key2, value);
  if (s.ok()) s = db->Delete(rocksdb::WriteOptions(), key1);

Atomic Updates

Note that if the process dies after the Put of key2 but before the delete of key1, the same value may be left stored under multiple keys. Such problems can be avoided by using the WriteBatch class to atomically apply a set of updates: 

  #include "rocksdb/write_batch.h"
  ...
  std::string value;
  rocksdb::Status s = db->Get(rocksdb::ReadOptions(), key1, &value);
  if (s.ok()) {
    rocksdb::WriteBatch batch;
    batch.Delete(key1);
    batch.Put(key2, value);
    s = db->Write(rocksdb::WriteOptions(), &batch);
  }
The WriteBatch holds a sequence of edits to be made to the database, and these edits within the batch are applied in order. Note that we called Delete before Put so that if key1 is identical to key2, we do not end up erroneously dropping the value entirely.

Apart from its atomicity benefits, WriteBatch may also be used to speed up bulk updates by placing lots of individual mutations into the same batch.

Synchronous Writes

By default, each write to leveldb is asynchronous: it returns after pushing the write from the process into the operating system. The transfer from operating system memory to the underlying persistent storage happens asynchronously. The sync flag can be turned on for a particular write to make the write operation not return until the data being written has been pushed all the way to persistent storage. (On Posix systems, this is implemented by calling either fsync(...) or fdatasync(...) or msync(..., MS_SYNC) before the write operation returns.) 
  rocksdb::WriteOptions write_options;
  write_options.sync = true;
  db->Put(write_options, ...);
Asynchronous writes are often more than a thousand times as fast as synchronous writes. The downside of asynchronous writes is that a crash of the machine may cause the last few updates to be lost. Note that a crash of just the writing process (i.e., not a reboot) will not cause any loss since even when sync is false, an update is pushed from the process memory into the operating system before it is considered done.

Asynchronous writes can often be used safely. For example, when loading a large amount of data into the database you can handle lost updates by restarting the bulk load after a crash. A hybrid scheme is also possible where every Nth write is synchronous, and in the event of a crash, the bulk load is restarted just after the last synchronous write finished by the previous run. (The synchronous write can update a marker that describes where to restart on a crash.)

WriteBatch provides an alternative to asynchronous writes. Multiple updates may be placed in the same WriteBatch and applied together using a synchronous write (i.e., write_options.sync is set to true). The extra cost of the synchronous write will be amortized across all of the writes in the batch.

We also provide a way to completely disable Write Ahead Log for a particular write. If you set write_option.disableWAL to true, the write will not go to the log at all and may be lost in an event of process crash.

When opening a DB, you can disable syncing of data files by setting Options::disableDataSync to true. This can be useful when doing bulk-loading or big idempotent operations. Once the operation is finished, you can manually call sync() to flush all dirty buffers to stable storage.

RocksDB by default uses faster fdatasync() to sync files. If you want to use fsync(), you can set Options::use_fsync to true. You should set this to true on filesystems like ext3 that can lose files after a reboot.

Concurrency

A database may only be opened by one process at a time. The rocksdb implementation acquires a lock from the operating system to prevent misuse. Within a single process, the same rocksdb::DB object may be safely shared by multiple concurrent threads. I.e., different threads may write into or fetch iterators or call Get on the same database without any external synchronization (the leveldb implementation will automatically do the required synchronization). However other objects (like Iterator and WriteBatch) may require external synchronization. If two threads share such an object, they must protect access to it using their own locking protocol. More details are available in the public header files.

Merge operators

Merge operators provide efficient support for read-modify-write operation. More on the interface and implementation can be found on:

Merge Operator

Merge Operator Implementation

Iteration

The following example demonstrates how to print all key,value pairs in a database. 

  rocksdb::Iterator* it = db->NewIterator(rocksdb::ReadOptions());
  for (it->SeekToFirst(); it->Valid(); it->Next()) {
    cout << it->key().ToString() << ": "  << it->value().ToString() << endl;
  }
  assert(it->status().ok());  // Check for any errors found during the scan
  delete it;
The following variation shows how to process just the keys in the range [start,limit): 

  for (it->Seek(start);
       it->Valid() && it->key().ToString() < limit;
       it->Next()) {
    ...
  }
You can also process entries in reverse order. (Caveat: reverse iteration may be somewhat slower than forward iteration.) 

  for (it->SeekToLast(); it->Valid(); it->Prev()) {
    ...
  }

Snapshots

Snapshots provide consistent read-only views over the entire state of the key-value store. ReadOptions::snapshot may be non-NULL to indicate that a read should operate on a particular version of the DB state. If ReadOptions::snapshot is NULL, the read will operate on an implicit snapshot of the current state.

Snapshots are created by the DB::GetSnapshot() method: 

  rocksdb::ReadOptions options;
  options.snapshot = db->GetSnapshot();
  ... apply some updates to db ...
  rocksdb::Iterator* iter = db->NewIterator(options);
  ... read using iter to view the state when the snapshot was created ...
  delete iter;
  db->ReleaseSnapshot(options.snapshot);
Note that when a snapshot is no longer needed, it should be released using the DB::ReleaseSnapshot interface. This allows the implementation to get rid of state that was being maintained just to support reading as of that snapshot.

Slice

The return value of the it->key() and it->value() calls above are instances of the rocksdb::Slice type. Slice is a simple structure that contains a length and a pointer to an external byte array. Returning a Slice is a cheaper alternative to returning a std::string since we do not need to copy potentially large keys and values. In addition, rocksdb methods do not return null-terminated C-style strings since rocksdb keys and values are allowed to contain '\0' bytes.

C++ strings and null-terminated C-style strings can be easily converted to a Slice: 

   rocksdb::Slice s1 = "hello";

   std::string str("world");
   rocksdb::Slice s2 = str;
A Slice can be easily converted back to a C++ string: 
   std::string str = s1.ToString();
   assert(str == std::string("hello"));
Be careful when using Slices since it is up to the caller to ensure that the external byte array into which the Slice points remains live while the Slice is in use. For example, the following is buggy: 

   rocksdb::Slice slice;
   if (...) {
     std::string str = ...;
     slice = str;
   }
   Use(slice);
When the if statement goes out of scope, str will be destroyed and the backing storage for slice will disappear.

Comparators

The preceding examples used the default ordering function for key, which orders bytes lexicographically. You can however supply a custom comparator when opening a database. For example, suppose each database key consists of two numbers and we should sort by the first number, breaking ties by the second number. First, define a proper subclass of rocksdb::Comparator that expresses these rules: 

  class TwoPartComparator : public rocksdb::Comparator {
   public:
    // Three-way comparison function:
    //   if a < b: negative result
    //   if a > b: positive result
    //   else: zero result
    int Compare(const rocksdb::Slice& a, const rocksdb::Slice& b) const {
      int a1, a2, b1, b2;
      ParseKey(a, &a1, &a2);
      ParseKey(b, &b1, &b2);
      if (a1 < b1) return -1;
      if (a1 > b1) return +1;
      if (a2 < b2) return -1;
      if (a2 > b2) return +1;
      return 0;
    }

    // Ignore the following methods for now:
    const char* Name() const { return "TwoPartComparator"; }
    void FindShortestSeparator(std::string*, const rocksdb::Slice&) const { }
    void FindShortSuccessor(std::string*) const { }
  };
Now create a database using this custom comparator: 

  TwoPartComparator cmp;
  rocksdb::DB* db;
  rocksdb::Options options;
  options.create_if_missing = true;
  options.comparator = &cmp;
  rocksdb::Status status = rocksdb::DB::Open(options, "/tmp/testdb", &db);
  ...

Backwards compatibility

The result of the comparator's Name method is attached to the database when it is created, and is checked on every subsequent database open. If the name changes, the rocksdb::DB::Open call will fail. Therefore, change the name if and only if the new key format and comparison function are incompatible with existing databases, and it is ok to discard the contents of all existing databases.

You can however still gradually evolve your key format over time with a little bit of pre-planning. For example, you could store a version number at the end of each key (one byte should suffice for most uses). When you wish to switch to a new key format (e.g., adding an optional third part to the keys processed by TwoPartComparator), (a) keep the same comparator name (b) increment the version number for new keys (c) change the comparator function so it uses the version numbers found in the keys to decide how to interpret them.

MemTable and Table factories

By default, we keep the data in memory in skiplist memtable and the data on disk in a table format described here: RocksDB Table Format.

Since one of the goals of RocksDB is to have different parts of the system easily pluggable, we support different implementations of both memtable and table format. You can supply your own memtable factory by setting Options::memtable_factory and your own table factory by setting Options::table_factory. For available memtable factories, please refer to rocksdb/memtablerep.h and for table factores to rocksdb/table.h. These features are both in active development and please be wary of any API changes that might break your application going forward.

You can also read more about memtables here: Memtables wiki

Performance

Performance can be tuned by changing the default values of the types defined in include/rocksdb/options.h.

Block size

rocksdb groups adjacent keys together into the same block and such a block is the unit of transfer to and from persistent storage. The default block size is approximately 4096 uncompressed bytes. Applications that mostly do bulk scans over the contents of the database may wish to increase this size. Applications that do a lot of point reads of small values may wish to switch to a smaller block size if performance measurements indicate an improvement. There isn't much benefit in using blocks smaller than one kilobyte, or larger than a few megabytes. Also note that compression will be more effective with larger block sizes. To change block size parameter, use Options::block_size.

Write buffer

Options::write_buffer_size specifies the amount of data to build up in memory before converting to a sorted on-disk file. Larger values increase performance, especially during bulk loads. Up to max_write_buffer_number write buffers may be held in memory at the same time, so you may wish to adjust this parameter to control memory usage. Also, a larger write buffer will result in a longer recovery time the next time the database is opened. Related option is Options::max_write_buffer_number, which is maximum number of write buffers that are built up in memory. The default is 2, so that when 1 write buffer is being flushed to storage, new writes can continue to the other write buffer. Options::min_write_buffer_number_to_merge is the minimum number of write buffers that will be merged together before writing to storage. If set to 1, then all write buffers are flushed to L0 as individual files and this increases read amplification because a get request has to check in all of these files. Also, an in-memory merge may result in writing lesser data to storage if there are duplicate records in each of these individual write buffers. Default: 1

Compression

Each block is individually compressed before being written to persistent storage. Compression is on by default since the default compression method is very fast, and is automatically disabled for uncompressible data. In rare cases, applications may want to disable compression entirely, but should only do so if benchmarks show a performance improvement: 

  rocksdb::Options options;
  options.compression = rocksdb::kNoCompression;
  ... rocksdb::DB::Open(options, name, ...) ....

Cache

The contents of the database are stored in a set of files in the filesystem and each file stores a sequence of compressed blocks. If options.block_cache is non-NULL, it is used to cache frequently used uncompressed block contents. If options.block_cache_compressed is non-NULL, it is used to cache frequently used compressed blocks. Compressed cache is an alternative to OS cache, which also caches compressed blocks. If compressed cache is used, the OS cache will be disabled automatically by setting options.allow_os_buffer to false. 

  #include "rocksdb/cache.h"

  rocksdb::Options options;
  options.block_cache = rocksdb::NewLRUCache(100 * 1048576);  // 100MB uncompressed cache
  options.block_cache_compressed = rocksdb::NewLRUCache(100 * 1048576);  // 100MB compressed cache
  rocksdb::DB* db;
  rocksdb::DB::Open(options, name, &db);
  ... use the db ...
  delete db
  delete options.block_cache;
  delete options.block_cache_compressed;

When performing a bulk read, the application may wish to disable caching so that the data processed by the bulk read does not end up displacing most of the cached contents. A per-iterator option can be used to achieve this: 

  rocksdb::ReadOptions options;
  options.fill_cache = false;
  rocksdb::Iterator* it = db->NewIterator(options);
  for (it->SeekToFirst(); it->Valid(); it->Next()) {
    ...
  }

You can also disable block cache by setting options.no_block_cache to true.

Key Layout

Note that the unit of disk transfer and caching is a block. Adjacent keys (according to the database sort order) will usually be placed in the same block. Therefore the application can improve its performance by placing keys that are accessed together near each other and placing infrequently used keys in a separate region of the key space.

For example, suppose we are implementing a simple file system on top of rocksdb. The types of entries we might wish to store are: 

   filename -> permission-bits, length, list of file_block_ids
   file_block_id -> data
We might want to prefix filename keys with one letter (say '/') and the file_block_id keys with a different letter (say '0') so that scans over just the metadata do not force us to fetch and cache bulky file contents.

Filters

Because of the way rocksdb data is organized on disk, a single Get() call may involve multiple reads from disk. The optional FilterPolicy mechanism can be used to reduce the number of disk reads substantially. 

   rocksdb::Options options;
   options.filter_policy = NewBloomFilter(10);
   rocksdb::DB* db;
   rocksdb::DB::Open(options, "/tmp/testdb", &db);
   ... use the database ...
   delete db;
   delete options.filter_policy;
The preceding code associates a Bloom filter based filtering policy with the database. Bloom filter based filtering relies on keeping some number of bits of data in memory per key (in this case 10 bits per key since that is the argument we passed to NewBloomFilter). This filter will reduce the number of unnecessary disk reads needed for Get() calls by a factor of approximately a 100. Increasing the bits per key will lead to a larger reduction at the cost of more memory usage. We recommend that applications whose working set does not fit in memory and that do a lot of random reads set a filter policy.

If you are using a custom comparator, you should ensure that the filter policy you are using is compatible with your comparator. For example, consider a comparator that ignores trailing spaces when comparing keys. NewBloomFilter must not be used with such a comparator. Instead, the application should provide a custom filter policy that also ignores trailing spaces. For example: 

  class CustomFilterPolicy : public rocksdb::FilterPolicy {
   private:
    FilterPolicy* builtin_policy_;
   public:
    CustomFilterPolicy() : builtin_policy_(NewBloomFilter(10)) { }
    ~CustomFilterPolicy() { delete builtin_policy_; }

    const char* Name() const { return "IgnoreTrailingSpacesFilter"; }

    void CreateFilter(const Slice* keys, int n, std::string* dst) const {
      // Use builtin bloom filter code after removing trailing spaces
      std::vector<Slice> trimmed(n);
      for (int i = 0; i < n; i++) {
        trimmed[i] = RemoveTrailingSpaces(keys[i]);
      }
      return builtin_policy_->CreateFilter(&trimmed[i], n, dst);
    }

    bool KeyMayMatch(const Slice& key, const Slice& filter) const {
      // Use builtin bloom filter code after removing trailing spaces
      return builtin_policy_->KeyMayMatch(RemoveTrailingSpaces(key), filter);
    }
  };

Advanced applications may provide a filter policy that does not use a bloom filter but uses some other mechanism for summarizing a set of keys. See rocksdb/filter_policy.h for detail.

Checksums

rocksdb associates checksums with all data it stores in the file system. There are two separate controls provided over how aggressively these checksums are verified:

  • ReadOptions::verify_checksums may be set to true to force checksum verification of all data that is read from the file system on behalf of a particular read. By default, no such verification is done.

  • Options::paranoid_checks may be set to true before opening a database to make the database implementation raise an error as soon as it detects an internal corruption. Depending on which portion of the database has been corrupted, the error may be raised when the database is opened, or later by another database operation. By default, paranoid checking is off so that the database can be used even if parts of its persistent storage have been corrupted.

    If a database is corrupted (perhaps it cannot be opened when paranoid checking is turned on), the rocksdb::RepairDB function may be used to recover as much of the data as possible.

Compaction

You can read more on Compactions here: Multi-threaded compactions

Here we give overview of the options that impact behavior of Compactions:

  • Options::compaction_style - RocksDB currently supports two compaction algorithms - Universal style and Level style. This option switches between the two. Can be kCompactionStyleUniversal or kCompactionStyleLevel. If this is kCompactionStyleUniversal, then you can configure universal style parameters with Options::compaction_options_universal.

  • Options::disable_auto_compactions - Disable automatic compactions. Manual compactions can still be issued on this database.

  • Options::compaction_filter - Allows an application to modify/delete a key-value during background compaction. The client must provide compaction_filter_factory if it requires a new compaction filter to be used for different compaction processes. Client should specify only one of filter or factory.

  • Options::compaction_filter_factory - a factory that provides compaction filter objects which allow an application to modify/delete a key-value during background compaction.

Other options impacting performance of compactions and when they get triggered are:

  • Options::access_hint_on_compaction_start - Specify the file access pattern once a compaction is started. It will be applied to all input files of a compaction. Default: NORMAL

  • Options::level0_file_num_compaction_trigger - Number of files to trigger level-0 compaction. A negative value means that level-0 compaction will not be triggered by number of files at all.

  • Options::max_mem_compaction_level - Maximum level to which a new compacted memtable is pushed if it does not create overlap. We try to push to level 2 to avoid the relatively expensive level 0=>1 compactions and to avoid some expensive manifest file operations. We do not push all the way to the largest level since that can generate a lot of wasted disk space if the same key space is being repeatedly overwritten.

  • Options::target_file_size_base and Options::target_file_size_multiplier - Target file size for compaction. target_file_size_base is per-file size for level-1. Target file size for level L can be calculated by target_file_size_base * (target_file_size_multiplier ^ (L-1)) For example, if target_file_size_base is 2MB and target_file_size_multiplier is 10, then each file on level-1 will be 2MB, and each file on level 2 will be 20MB, and each file on level-3 will be 200MB. Default target_file_size_base is 2MB and default target_file_size_multiplier is 1.

  • Options::expanded_compaction_factor - Maximum number of bytes in all compacted files. We avoid expanding the lower level file set of a compaction if it would make the total compaction cover more than (expanded_compaction_factor * targetFileSizeLevel()) many bytes.

  • Options::source_compaction_factor - Maximum number of bytes in all source files to be compacted in a single compaction run. We avoid picking too many files in the source level so that we do not exceed the total source bytes for compaction to exceed (source_compaction_factor * targetFileSizeLevel()) many bytes. Default:1, i.e. pick maxfilesize amount of data as the source of a compaction.

  • Options::max_grandparent_overlap_factor - Control maximum bytes of overlaps in grandparent (i.e., level+2) before we stop building a single file in a level->level+1 compaction.

  • Options::max_background_compactions - Maximum number of concurrent background jobs, submitted to the default LOW priority thread pool

You can learn more about all of those options in rocksdb/options.h

Universal style compaction specific settings

If you're using Universal style compaction, there is an object CompactionOptionsUniversal that hold all the different options for that compaction. The exact definition is in rocksdb/universal_compaction.h and you can set it in Options::compaction_options_universal. Here we give short overview of options in CompactionOptionsUniversal:

  • CompactionOptionsUniversal::size_ratio - Percentage flexibility while comparing file size. If the candidate file(s) size is 1% smaller than the next file's size, then include next file into this candidate set. Default: 1

  • CompactionOptionsUniversal::min_merge_width - The minimum number of files in a single compaction run. Default: 2

  • CompactionOptionsUniversal::max_merge_width - The maximum number of files in a single compaction run. Default: UINT_MAX

  • CompactionOptionsUniversal::max_size_amplification_percent - The size amplification is defined as the amount (in percentage) of additional storage needed to store a single byte of data in the database. For example, a size amplification of 2% means that a database that contains 100 bytes of user-data may occupy upto 102 bytes of physical storage. By this definition, a fully compacted database has a size amplification of 0%. Rocksdb uses the following heuristic to calculate size amplification: it assumes that all files excluding the earliest file contribute to the size amplification. Default: 200, which means that a 100 byte database could require upto 300 bytes of storage.

  • CompactionOptionsUniversal::compression_size_percent - If this option is set to be -1 (the default value), all the output files will follow compression type specified. If this option is not negative, we will try to make sure compressed size is just above this value. In normal cases, at least this percentage of data will be compressed. When we are compacting to a new file, here is the criteria whether it needs to be compressed: assuming here are the list of files sorted by generation time: [ A1...An B1...Bm C1...Ct ], where A1 is the newest and Ct is the oldest, and we are going to compact B1...Bm, we calculate the total size of all the files as total_size, as well as the total size of C1...Ct as total_C, the compaction output file will be compressed iff total_C / total_size < this percentage

  • CompactionOptionsUniversal::stop_style - The algorithm used to stop picking files into a single compaction run. Can be kCompactionStopStyleSimilarSize (pick files of similar size) or kCompactionStopStyleTotalSize (total size of picked files > next file). Default: kCompactionStopStyleTotalSize

Thread pools

A thread pool is associated with Env environment object. The client has to create a thread pool by setting the number of background threads using method Env::SetBackgroundThreads() defined in rocksdb/env.h. We use the thread pool for compactions and memtable flushes. Since memtable flushes are in critical code path (stalling memtable flush can stall writes, increasing p99), we suggest having two thread pools - with priorities HIGH and LOW. Memtable flushes can be set up to be scheduled on HIGH thread pool. There are two options available for configuration of background compactions and flushes:

  • Options::max_background_compactions - Maximum number of concurrent background jobs, submitted to the default LOW priority thread pool

  • Options::max_background_flushes - Maximum number of concurrent background memtable flush jobs, submitted to the HIGH priority thread pool. By default, all background jobs (major compaction and memtable flush) go to the LOW priority pool. If this option is set to a positive number, memtable flush jobs will be submitted to the HIGH priority pool. It is important when the same Env is shared by multiple db instances. Without a separate pool, long running major compaction jobs could potentially block memtable flush jobs of other db instances, leading to unnecessary Put stalls. 

  #include "rocksdb/env.h"
  #include "rocksdb/db.h"

  auto env = rocksdb::Env::Default();
  env->SetBackgroundThreads(2, rocksdb::Env::LOW);
  env->SetBackgroundThreads(1, rocksdb::Env::HIGH);
  rocksdb::DB* db;
  rocksdb::Options options;
  options.env = env;
  options.max_background_compactions = 2;
  options.max_background_flushes = 1;
  rocksdb::Status status = rocksdb::DB::Open(options, "/tmp/testdb", &db);
  assert(status.ok());
  ...

Approximate Sizes

The GetApproximateSizes method can used to get the approximate number of bytes of file system space used by one or more key ranges. 

   rocksdb::Range ranges[2];
   ranges[0] = rocksdb::Range("a", "c");
   ranges[1] = rocksdb::Range("x", "z");
   uint64_t sizes[2];
   rocksdb::Status s = db->GetApproximateSizes(ranges, 2, sizes);
The preceding call will set sizes[0] to the approximate number of bytes of file system space used by the key range [a..c) and sizes[1] to the approximate number of bytes used by the key range [x..z).

Environment

All file operations (and other operating system calls) issued by the rocksdb implementation are routed through a rocksdb::Env object. Sophisticated clients may wish to provide their own Env implementation to get better control. For example, an application may introduce artificial delays in the file IO paths to limit the impact of rocksdb on other activities in the system. 

  class SlowEnv : public rocksdb::Env {
    .. implementation of the Env interface ...
  };

  SlowEnv env;
  rocksdb::Options options;
  options.env = &env;
  Status s = rocksdb::DB::Open(options, ...);

Porting

rocksdb may be ported to a new platform by providing platform specific implementations of the types/methods/functions exported by rocksdb/port/port.h. See rocksdb/port/port_example.h for more details.

In addition, the new platform may need a new default rocksdb::Env implementation. See rocksdb/util/env_posix.h for an example.

Statistics

To be able to efficiently tune your application, it is always helpful if you have access to usage statistics. You can collect those statistics by setting Options::table_properties_collectors or Options::statistics. For more information, refer to rocksdb/table_properties.h and rocksdb/statistics.h. These should not add significant overhead to your application and we recommend exporting them to other monitoring tools.

Purging WAL files

By default, old write-ahead logs are deleted automatically when they fall out of scope and application doesn't need them anymore. There are options that enable the user to archive the logs and then delete them lazily, either in TTL fashion or based on size limit. The options are Options::WAL_ttl_seconds and Options::WAL_size_limit_MB. Here is how they can be used:

  • If both set to 0, logs will be deleted asap and will never get into the archive.

  • If WAL_ttl_seconds is 0 and WAL_size_limit_MB is not 0, WAL files will be checked every 10 min and if total size is greater then WAL_size_limit_MB, they will be deleted starting with the earliest until size_limit is met. All empty files will be deleted.

  • If WAL_ttl_seconds is not 0 and WAL_size_limit_MB is 0, then WAL files will be checked every WAL_ttl_seconds / 2 and those that are older than WAL_ttl_seconds will be deleted.

  • If both are not 0, WAL files will be checked every 10 min and both checks will be performed with ttl being first.

Other Information

Details about the rocksdb implementation may be found in the following documents:

rocksdb-4.5.1/doc/log_format.txt000066400000000000000000000053211267530634000166160ustar00rootroot00000000000000The log file contents are a sequence of 32KB blocks. The only exception is that the tail of the file may contain a partial block. Each block consists of a sequence of records: block := record* trailer? record := checksum: uint32 // crc32c of type and data[] length: uint16 type: uint8 // One of FULL, FIRST, MIDDLE, LAST data: uint8[length] A record never starts within the last six bytes of a block (since it won't fit). Any leftover bytes here form the trailer, which must consist entirely of zero bytes and must be skipped by readers. Aside: if exactly seven bytes are left in the current block, and a new non-zero length record is added, the writer must emit a FIRST record (which contains zero bytes of user data) to fill up the trailing seven bytes of the block and then emit all of the user data in subsequent blocks. More types may be added in the future. Some Readers may skip record types they do not understand, others may report that some data was skipped. FULL == 1 FIRST == 2 MIDDLE == 3 LAST == 4 The FULL record contains the contents of an entire user record. FIRST, MIDDLE, LAST are types used for user records that have been split into multiple fragments (typically because of block boundaries). FIRST is the type of the first fragment of a user record, LAST is the type of the last fragment of a user record, and MID is the type of all interior fragments of a user record. Example: consider a sequence of user records: A: length 1000 B: length 97270 C: length 8000 A will be stored as a FULL record in the first block. B will be split into three fragments: first fragment occupies the rest of the first block, second fragment occupies the entirety of the second block, and the third fragment occupies a prefix of the third block. This will leave six bytes free in the third block, which will be left empty as the trailer. C will be stored as a FULL record in the fourth block. =================== Some benefits over the recordio format: (1) We do not need any heuristics for resyncing - just go to next block boundary and scan. If there is a corruption, skip to the next block. As a side-benefit, we do not get confused when part of the contents of one log file are embedded as a record inside another log file. (2) Splitting at approximate boundaries (e.g., for mapreduce) is simple: find the next block boundary and skip records until we hit a FULL or FIRST record. (3) We do not need extra buffering for large records. Some downsides compared to recordio format: (1) No packing of tiny records. This could be fixed by adding a new record type, so it is a shortcoming of the current implementation, not necessarily the format. (2) No compression. Again, this could be fixed by adding new record types. rocksdb-4.5.1/doc/rockslogo.jpg000066400000000000000000004140201267530634000164300ustar00rootroot00000000000000ÿØÿàJFIF``ÿâ XICC_PROFILE HLinomntrRGB XYZ Î 1acspMSFTIEC sRGBöÖÓ-HP cprtP3desc„lwtptðbkptrXYZgXYZ,bXYZ@dmndTpdmddĈvuedL†viewÔ$lumiømeas $tech0 rTRC< gTRC< bTRC< textCopyright (c) 1998 Hewlett-Packard CompanydescsRGB IEC61966-2.1sRGB IEC61966-2.1XYZ óQÌXYZ XYZ o¢8õXYZ b™·…ÚXYZ $ „¶ÏdescIEC http://www.iec.chIEC http://www.iec.chdesc.IEC 61966-2.1 Default RGB colour space - sRGB.IEC 61966-2.1 Default RGB colour space - sRGBdesc,Reference Viewing Condition in IEC61966-2.1,Reference Viewing Condition in IEC61966-2.1view¤þ_.ÏíÌ \žXYZ L VPWçmeassig CRT curv #(-27;@EJOTY^chmrw|†‹•šŸ¤©®²·¼ÁÆËÐÕÛàåëðöû %+28>ELRY`gnu|ƒ‹’š¡©±¹ÁÉÑÙáéòú &/8AKT]gqz„Ž˜¢¬¶ÁËÕàëõ !-8COZfr~Š–¢®ºÇÓàìù -;HUcq~Œš¨¶ÄÓáðþ +:IXgw†–¦µÅÕåö'7HYj{Œ¯ÀÑãõ+=Oat†™¬¿Òåø 2FZn‚–ª¾Òçû  % : O d y ¤ º Ï å û  ' = T j ˜ ® Å Ü ó " 9 Q i € ˜ ° È á ù  * C \ u Ž § À Ù ó & @ Z t Ž © Ã Þ ø.Id›¶Òî %A^z–³Ïì &Ca~›¹×õ1OmŒªÉè&Ed„£Ãã#Ccƒ¤Åå'Ij‹­Îð4Vx›½à&Il²ÖúAe‰®Ò÷@eŠ¯Õú Ek‘·Ý*QwžÅì;cŠ²Ú*R{£ÌõGp™Ãì@j”¾é>i”¿ê  A l ˜ Ä ð!!H!u!¡!Î!û"'"U"‚"¯"Ý# #8#f#”#Â#ð$$M$|$«$Ú% %8%h%—%Ç%÷&'&W&‡&·&è''I'z'«'Ü( (?(q(¢(Ô))8)k))Ð**5*h*›*Ï++6+i++Ñ,,9,n,¢,×- -A-v-«-á..L.‚.·.î/$/Z/‘/Ç/þ050l0¤0Û11J1‚1º1ò2*2c2›2Ô3 3F33¸3ñ4+4e4ž4Ø55M5‡5Â5ý676r6®6é7$7`7œ7×88P8Œ8È99B99¼9ù:6:t:²:ï;-;k;ª;è<' >`> >à?!?a?¢?â@#@d@¦@çA)AjA¬AîB0BrBµB÷C:C}CÀDDGDŠDÎEEUEšEÞF"FgF«FðG5G{GÀHHKH‘H×IIcI©IðJ7J}JÄK KSKšKâL*LrLºMMJM“MÜN%NnN·OOIO“OÝP'PqP»QQPQ›QæR1R|RÇSS_SªSöTBTTÛU(UuUÂVV\V©V÷WDW’WàX/X}XËYYiY¸ZZVZ¦Zõ[E[•[å\5\†\Ö]']x]É^^l^½__a_³``W`ª`üaOa¢aõbIbœbðcCc—cëd@d”dée=e’eçf=f’fèg=g“géh?h–hìiCišiñjHjŸj÷kOk§kÿlWl¯mm`m¹nnknÄooxoÑp+p†pàq:q•qðrKr¦ss]s¸ttptÌu(u…uáv>v›vøwVw³xxnxÌy*y‰yçzFz¥{{c{Â|!||á}A}¡~~b~Â#„å€G€¨ kÍ‚0‚’‚ôƒWƒº„„€„ã…G…«††r†×‡;‡ŸˆˆiˆÎ‰3‰™‰þŠdŠÊ‹0‹–‹üŒcŒÊ1˜ÿŽfŽÎ6žnÖ‘?‘¨’’z’ã“M“¶” ”Š”ô•_•É–4–Ÿ— —u—à˜L˜¸™$™™üšhšÕ›B›¯œœ‰œ÷dÒž@ž®ŸŸ‹Ÿú i Ø¡G¡¶¢&¢–££v£æ¤V¤Ç¥8¥©¦¦‹¦ý§n§à¨R¨Ä©7©©ªª««u«é¬\¬Ð­D­¸®-®¡¯¯‹°°u°ê±`±Ö²K²Â³8³®´%´œµµŠ¶¶y¶ð·h·à¸Y¸Ñ¹J¹Âº;ºµ».»§¼!¼›½½¾ ¾„¾ÿ¿z¿õÀpÀìÁgÁãÂ_ÂÛÃXÃÔÄQÄÎÅKÅÈÆFÆÃÇAÇ¿È=ȼÉ:ɹÊ8Ê·Ë6˶Ì5̵Í5͵Î6ζÏ7ϸÐ9кÑ<ѾÒ?ÒÁÓDÓÆÔIÔËÕNÕÑÖUÖØ×\×àØdØèÙlÙñÚvÚûÛ€ÜÜŠÝÝ–ÞÞ¢ß)߯à6à½áDáÌâSâÛãcãëäsäüå„æ æ–çç©è2è¼éFéÐê[êåëpëûì†ííœî(î´ï@ïÌðXðåñrñÿòŒóó§ô4ôÂõPõÞömöû÷Šøø¨ù8ùÇúWúçûwüü˜ý)ýºþKþÜÿmÿÿÿáŒExifMM*JR(‡iZ%d%d  € 8ÿÛC      ÿÛC  ÿÀ8€"ÿÄ ÿĵ}!1AQa"q2‘¡#B±ÁRÑð$3br‚ %&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyzƒ„…†‡ˆ‰Š’“”•–—˜™š¢£¤¥¦§¨©ª²³´µ¶·¸¹ºÂÃÄÅÆÇÈÉÊÒÓÔÕÖ×ØÙÚáâãäåæçèéêñòóôõö÷øùúÿÄ ÿĵw!1AQaq"2B‘¡±Á #3RðbrÑ $4á%ñ&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz‚ƒ„…†‡ˆ‰Š’“”•–—˜™š¢£¤¥¦§¨©ª²³´µ¶·¸¹ºÂÃÄÅÆÇÈÉÊÒÓÔÕÖ×ØÙÚâãäåæçèéêòóôõö÷øùúÿÚ ?ý ¢Š+ŒŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š)@É£pŠw—ïG—ïWìåØÑNòýèòýèörìh§y~ôy~ô{9v´S¼¿z `Pá$h¢Š€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ f—ÐÓ“¥-mi«€ÌCF¡§ÑOÙ ƒèhÁô4ú(öH`ú0} >Š=’<UûÔúnÜ='O•¦€uQ[QEQER7CKAèi=€ŽŠ(®@ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (§¯ÝªŒy€e%§²ó:*J(ö^`GEIEËÌè©)®;ÔÊŸ*¸ ¢Š+0 (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€)i¥-uCdEUQEQEQEQEQEQEQEÖLt¦Ô”ÖLô¬gOª´QEbEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPO_»L§¯Ý­inÑE¸Q@Q@fŠ(20y¢žWwZ6zÁÒwÐQOØ=èØ=é{9Ê)û½5† 'X ETQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEôéKHŸv–º¡²¢Š*€(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢ŠF\ÓjJÏZÎtùµ@GE8§¥!u¬\ZÜ¢Š*@(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š*@1QÔ•µ. QElEPEPEPEPEPMqÞA2\ÊÀGEW(Q@Q@Q@=~í2ž¿vµ¥¸ EVàQEQEŽ¹éKE&¹•€ŽŠs)'ŠM†¹œ{”PF:ÑREPEP@äÑJ:Škp°{Ñ°{ÒÑ]<‘ìlôlô´QÉÀ&ÁïFÁïKE‘ìlô›:Š9#ØÈÁ¢žËšesÎ<¬Š(©¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(ë÷iiîÒ×Tv@QEPQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@S4Ò¤u§ÑQ*j@GE<¦zSÇZÂPq¢Š*@(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š*JŽ¤­©u¢Š+` (¢€ (¢€ (¢€ (¢€ (¢€ (¢€¨'š6 Z)r®À&ÁFÁKE«° dãŠmILdÅeRP EV õû´ÊzýÚÖ–à-Q[€QEQEQEQEžµ5%# ³œ9µ@2Š(®p (¢€ QÔRRŽ¢šÜÑEÖEPEPEPHËšZ)4žŒÈÇZ)ì7 esÎ<¬Š(¨¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(ë÷iiîÒ×Tv@QEPQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@h¢€¦M4® >Š‰SL觕˜F:Ö2ƒˆQPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPRTu%mK¨Q[QEQEQEQEQEQEQEQEQEŒ™éL©)®½ë*ê€m=~í2ž¿v¦–à-Q[€QEQEQEQEQEŒ¹¦T”×^ⲩ ê€mQXRŽ¢’”uÖà>Š(®° (¢€ (¢€ (¢€ (¢€ kŽôê*ee`#¢•† %s5` (¢Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@ª2iõ¤!ͨ M.ÃN¢´öHì4l4ê(öq»  :Š=œ@nÃFÃN¢g°Ñ°Ó¨£ÙÄì4l5ã?kÑðKÇCEmLý™. Æ÷È1nùM6õϨí\…—üb/a]KÂ2Ån΢W‹R:&y*¦% @ÎaŸQÖ¿<Ì|RáL«S‹ÆrÕƒå’ä¨ì×K¨8üïcë°œ žãððÅPÃ^WOš¯G+þÒ»  Rðωì}¾²›L½šÛR†[{‹wh¥ŠT(ñ:œe<‚ ƒÒ«5Êx‹ÃœuJ2J~ï=9N1šßÝ’åæ[;tê“'Ê8à Š0«ùf£'ç{?é6½?gOŽPütð9½’(­5K_[£‚ð‘Kßœnm»_œžø›¯ü1ÔšïÀz¥Ö+ÿ¬T!£›€ó#`Qñ½±¸‘ƒ]ü6GÄúÿòŸkÿÆ«÷®úB`0ùu*YÕ²ÄEZR„`Ô­´µœlÚÝZ×Û²ü³9ð—WR¦[R¢õJNI®ëHÊétÖö>ꢾÿ†Èø‘ÿCþSíøÕGwû_üF½µ–¼I I£ìíãpÇÊËe>à‚;W¹/¤WYÛ^ÿá§ÿËO1xCœ_ZÔ¿ð)ÿòÝÕÌÿÂéðoý ¾ÿÁ¤ü]~zøÇö¨ºÐKé¿>#ÜYÈû]CÄ?>&Êœ¤’ Èpæ®7þ·‚?èrð¯þ ­ÿøºñ±¾<ãë¨O+Ê&â÷ræwínUnýOG áf›”q¸ø©.‘·ÎüÌý<ÿ…ÓàìóâÏ ÿàÒþ.ø]^ ?ó6øgÿñuù‡ÿ ÛÁô9xWÿÖÿü]_ðßįxÆùí|#¯èš­Ìq™Z+;è§u@@,U2Ê3î=kÌ«ãÎB¥L¥(­Ûö‰/ŽØxY•T’Œ1í·Ñrÿ™ú[ÿ §Á§§‹|3ÿƒH?øº_ø\þÿ¡³Ã_ø3ƒÿ‹¯Îν)JíÛ¼ªïåw }3Ö¼ÿø˜üÇþ€)ÿàR:¿âàÿè*_r?L¬¯aÔ¬â¸Ó¥ŠâÞá‘KIŒ†VAR×ç…ñ#Vð5¼–š'‰5m2>sÛØ_IÞ@`¤Ä§×[à¯Û'Ç>·´ZÕÕÅ…›‰6ßE’] åŠ<Ž¦@§8Îü€@p1õ™oÒ3+¬¢±Ø*Ñ]ÅÆjÿ7nº]ù3ÁÆøAŽ§«ba.ÊW‹·Ë™_ðó>ð¢¼Káïíéàj¶šä—Z ÍÜQ²µÔ_èÆV8dYTœ¨9;Ü Ú cŸm¯Üò^!ËxŠ“­–â#V+~Wª½íÌ·ìít®~c˜åÌ¢jž2“ƒ{]hýÏäØQEìžpQEU›;uXe¸½Ba!C9 Ä€:ƒ×¶9Í%´-•ÅÚ¬n-†¸$`B:A==k2HäÔµ(´ý?uü«?™s<àÈ­& g•³µU©8ë@iRòÎ?2‰¢ò¤xßÌó'ßÀ\ÑrXãŒck F5”¨mì%º¹¾’êh‘&³•,1³1ÚcD;’A r[‘SØ)²¸uo=’ã3î~|²J©ÞyÁw$Œœ’[8=@/QEQEQEQEQEQEÖ\ò)µ%5×½cRP ¢Š+ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ zœŠe*œ­\%f袊颊(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¦ƒO¦¸ç5•UupFph¢°N´SQ³Å:ºâù•À(¢Š`QEQEQEQEQE5׊mHyëLaƒXT…µ@%QYQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE«÷©ôÅûÔú襰QZQEQEQE>áöÀŒí× 3@ ®cãgü‘ŸØóÿD=u"&eÊ© œqUuÍßÄúEΓª¨š-R'´–Ý\‰J‘Ç OYêZdž¦‹WÐmx OÛ ‡/* Bt,¤ñómQ¾3_£Ú†¡§xŠ;­.A6©lmów±ž&ƒ.ÆÝŸ00 3žãÂk¿³‚\²‚6áxÜÖ¾=þËÞýü-Æ/ø*'‹õOŽ_õË¢h-›=Ê´wÚÞɵ†g—s:­°Šà/Ùw¼q¿èÏ„îø?GÓ< Iálàé-æ³bP°¸·gŽ¢PT˜Ø§Ë¶3‚ õ”UÎT’Z3ó/öÄÿ‚YþÊ?°·€¬µ¿Ž¾8ø»-Χqö{K¾ÓgÔoˆ#Ìh‘ívƬ™™T«»s¢·7â/ø"—Ç¿„>ñ'ìQÿ džuFâàjZOÅK)­mÖI#ŽMÖÖùGÌ%öm2L„2•Ãï|>Ñ­à«ðU_x¦÷DðÞ£ðÇàÂ"þ&6:øŠæäYÉ/ï”܉¤77Y¡1À¸`ûŸôCâ§Æ ü ðnŸ©|tñ¬ÞÑïnàÓÅÍì©d³M4„$$¢ƒ‚ÌÊ‘£³ªî &'9FÊúŸ‹_´_üÓãìÚ᯴#Æ0}'i|1tnåBîSËR,wnW†fXY¶KpÁ|KÂ?µ‡Ž<®Ûêzˆ5fUP ¾¥;ÝÁ2îVÚë#°T«…'f¿£ý Æ ,Vp[Ã<ó4ìʬ]Ãc‰Ë* ß6Œ_k˜Þ[‘ ä°/ÌŸáÜ8${ÏõÁ˜dø,ÚŸ²ÆQHÙ«J*[ï¿s« šâp2ç¡7Þ-­Á}þ Wy'Ùcñ† µ-å­ÕÆp¥— ’HÒXÎOÞ*…ý‹zÃ¿Û À?fQ×5 2ç÷‚+m]Dɵ7iT´`˜ ÒHÀ®gÚÚE/½HÙÉÏò¿Æïø"‡À]_ÚønãÀ·ÓµMyá«¥°‹ËŽ ´Šö±FpŒÅ"F,»‹|Ï»òÌÛÀ¾ÇÁ¬59P—xI½|Ôù´òVÓf·>ßâ†q†’öòUi%ÿ¶òýîÿ3æca|ä)6f6 n—åÏBî3œk¸ðÏÇox>[VÐø‡âÅü@¸ŠýìÑ.!¹¤N ¥   GƒŒŒ¶IÈÛïUù›à/‰·Ö_Ú_ üC4vתWíZ]ù 8V ñ6uàƒ]ü.ÏÿÐÝâüOÿÅ×ØpŸŽßØm<oF­jðrNnJïÞm'Íï]m®º9Ÿx]ý«ž+RéI+E-6Z«ig¾‡èùÝÿ ³Æô7xŸÿ“ÿñtÂìñŸý Þ'ÿÁ¤ÿü]}'üL~[ÿ@5?ð(ž?üAÜgýCîg螯r-nìmG–ñÙBúÉ`dÁ**'¶6òzqYz½¤’xlÙ­¤×¿ÚV2äEUá–V¥ÙUä…9ÏJñ/Ø“ã¹ãOx†ÓÆúœ7rEw†öîRn\8”ìy\æER€¨ä®â3Œë~(¹·Ôô¶ú\0¤~Tq¥ÄJF£'²ªqÆqÁç"¿oán!¡ÅyU × TOGºi¸µ§šv}V§æYæQW"ÇUÀÖiÊ j¶wI¯Á£ ‡M ñ\HË(,w[¦Ø¢aNIË8P¼(95NûNVÓíç†$†ÚÉ$Éó|µ‘ .@e*åä61MÔ57T¼g}DùCe¼QÛ;²#˜Ôÿä… ç‚yÝ0éî––º.™yä@>Í÷Jmá„r æø8Âc,0w <¢æ|/bÉ8¹oS÷]pOÊÜsú‚ŠÇ×#û%È‘. Me4X‘ßV8xù¶U{œãÖ­´âêÚ9T2‰0 Ôdw ÑEQEQEQEQEPy¢Ša4•!ëL+Šçœ9u@%QY€QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE 9R!â–ºâî®ES¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¤a‘KE \è§:çšmrÊ<®À*ýê}1~õ>¶¥°QZQEQEQEQEQER¸RÑCW20y¢žë‘L®YÇ•€QE QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE*ýê}1~õ>º)lEV€QE>Þ¹#‚Œô8æœ+éðÜ«K™¼¨ü°3È:ži.ueÓ­-à‡k-ܘ¹f`¯@SÇ^:ïçµgÿiMuÕÍÀµDs •„Í ‰IE¬`œ`d±^¥ Åj\E䉒à›v´ÎÒÛduGRIãÎk5¼K§]y;..õui0‘[DLsœãpX°x<–c€dèzLÚ®š‚x¯u;÷K©Æë§pCm Ç+–@ ~e-šÕÿ„ræHã‹R»šTEòÖÛJ€ð‹ü cã·`H ¸ñR³Å•h¯*þé0 厴§l+·?0S#u‘ƒÃüiøýkðך‹éº®—¦jvÆiôøe‰æ—P›kF¬cf‡~>wÚ©è¨Mjüdø™§| ð¬:¯ŠãÔ'·¸ºK6i IÚÍ´ne ¤FÀwr1íð—ÄïÝüJñ¶¡­ëÏq#OpòÁÅɸû E›lQ–û¨¡Š€¡G^M~7⧊¸2Ô°mK5³¿¸šÒNÖïîë¿¡ú'pMN#«õœEãB=t÷š ¿ã¡Ìê:ŒÚö­=ÑçΘÌ[ÉŽÞÙwí@9à”ôWüçÄW°j~*С–gž{Xõ 8ÄæS"±I^Pƒpÿ[2!ÜxQ_?ƒ‘‘ÎkêÿØ×öwñÃ_¿ˆüeµœŽ˜öéhÎßkˆ´‘:™nÕùcmÿˆÓêÚÿó!÷ª€?t©×—Z^‹KXÞ{’@;•†AÁùÏs¹¶¯N{Wáß„¿àô:_é‹ãÙâÿMÒZî!{wcãt¼ºµƒxó$†ÓâY¤UÜËKb.€î° ~=xwö—ø+¦üBø¯ÅâŸøÜ­M)™-ï.7]Ï$*Œ#–3DËæ”Èü_øãð–ÚÃÀ^øå¦éúßÄ/²]øƒÃ?iÐÒüéž[å~Í&¶ó9µ•Ó’Ög%XÂ$ñŸø,§í7©|ý¯4©íôØî~"4Þ‹Kšÿ_f’ EÅÜiö)E²¢<ðà¥úþ/éƒÄ0j×vv—:•ŠË …̶±yºl2ÄÛº 1ÆæfüÌbp1ùÉñƒá®·ûaÁ{-4¿¶ézEÁk#Yó|™kËKg¶¾òöîÃJ÷:—»(«͆dÛ$½ i¤ÝßCëø&Ÿì”?c?ÙÃÞÔáø†øc_²þeMÑ|²HŸ¹!·ÝþFðsX?ðQ‰ß F·ð«áGí9૯ißüD¶¾TžHÒ¥¢.ªé*¿›woòÙO•%ÆXãÊ—éšñŸ~ÍW>|7¸ýºtÿ x§âWÃûäÕ-õ}ÍQ^E.bž&&ùÒ;w–"¢‘?ÕíD í¥‘*W|ÌòÏØgàOÅŸÙCö’ñ_Ã뚟¾ ÚhjžÕµbÒÜé™E¼:jÈĶ&ßA£QœÆßZ]$¯l©e)‰–A bK?Þ œé‚r¡©kçÏÚ³þ ³û:þĺ„Ö´·Å¿ è:Í­ÔV—:5¼¯©êöo,&xÌú}šKs mHñªaãù³"±2—3¹ï"ØÆC8º•Ãn® nl“¹‰ÆâsÎ}1Ð ³o¦G®Höúª]M§›å¼ÅÑä_™Lá°Fì7Ë2N+áø‰Wö(ÿ¢Óÿ–†½ÿÈ_ö}ÿƒ“¿fÏÚöÎÑ>üÔÊÎm?^º‡lÓÅrP­Ã.<ùU>ÄÉ!?vb<ã<M\¨ÍÁÝΟì“ðóÅÿ´?Æ_|-øq®êºe¿ŒoŒiÉX"·ØZêw„ʉ.Ëxär…qÁ’@¯Ðíþ ¿ð컋Ä?5«­Y÷ýšæßEŠ xr '™Lí&$âDÈ ¤n>ûørÓÂ_ð^ÛÍ'ÃÖvšu†™ãZ[ZZÆ‘Ak[êH‘Ɖ…TUB¯~ª~Ø'¹Ð?dŸ‹‡…/î,5M#ÂZ¼ð\ÚÌÐÜYϔ̒#© Ž¬¡•€Aèk`¨ûÍFÍ»¶½ÖÚîÕ›ùÕ1•“ŠRº¶—Ö×í{Øü;Ð/cýœl[|ñÿü'~¶× °:…ºKk§ê­2Ç ò-¸••¶3º¤_ÉGRÊE}¯_ÿÁ=¼þ$øË>·2Ê-¼9hÎ@óæ$VSóc3ž1‚ƒ'±ûj¿¼~Åa*ñ (aà”áMsµ»rnI?5Þ¯›Èþ‡ð¾xeR©VMÆR|©öI&׫M|§àœã|12¤ÙðzËzúgÉM嶮æÆN98éùWÌÿðN_õ~0úÙízúj¿¡|ÿ’7ÿq?ôíCñÿÿä¢ÅÛŸúnŠ(¢¿Q> νÓRy¦6¶E¬is;¦Z{‡OáásÀã£æÖQZ€ê)¢OZváëZ©'°n´U^à2drŽm\E# n##Û ßž¢ª‹ém#TÔãh-`fhű–v|·Bª£o¯þ¢íDxžÛʸXâÔ— |µ·#ݶqÿëU|Añ'Jð~¨jš„ÑXAknÎïwÀÜÄ‚»Ø ÀbxÍlWǶǹ<ây|-¤Gåizã Y—縹MѱöU˨®I=±ñ}Ƹ~Ê媹ª?vœš^}’Z·ÛMÚ>—…xn·ã–ÐZÎ]£þoe÷ì™æŸ¾;Iñ£ÇU¸“N³Ûǵ¶àâÕTd€À.~v‘ò@?6; qqꯪ\Ë—,o!æ›É%&ù˜lVÈy8lg½_S~Äl×F¶ñψŒw7S¼ƒL‰Kjž'wìÌH`PÎOËüQÃyeâ^{*\÷«Qº•*JÖŠºæ—*Jú´£ev—ºµ_ÒYÆgƒà̱Tåµ8%A_WmÝû6Û»²oW£ôÙáŧƒ~醕œ-¬éVsÜ`*óHb ZIÙ›Ûjª»›šôª(¯ïì§.¥”`èà¨$£N**ÉEhµvZ+½}Yü§ÆTÌ115r“oWw¯›ÞÛ~ÿÁd¿l¿ÿðYÿ‹Ÿþ Á04[ÿ|øAjš§‰oü//›‹.#òò¾Áss$©meo曧µ{˜þÐ#è?ø-¿ü[ý·~6i°§üK]ûOÄjriÞ9ÖàÔ£´Ó,m"‚g»Òe¸ÚÎûcG–ïÈ;•-Í®ÙÞY ôWöýˆ¼#ÿõý•ü/ðÇá‚G£ÚÆú¾¥ojmäñ¤cE¹Ô'VyÌ™Ó!ZG Ž$"8ÑG rÆ ý9ÁYÿàÚ¿†ß·ž¡«xëöo¸°ø[ñgRº¹Ôõ+¡ ³i)¸xp«w¾-$i’7k«t$™nHn$ueþy?l?Ùcâìeñ2Çáÿí9àø@Ìlø´G0 À Œ®‹ïs^Qÿ…ýªSà?ìc¯iwVzgöÇÄ'‡lm¤Ÿs¼sF~Õpb ²Å *|ÅŽYmÃ×ÖÚfœ4û£“d’¬j²H)”ÄZü=ÿ‚ã|Oñ‹ÿoßøÄš­ÕÖ‰á,­ô{&!a°Yìm®&( ]ä–s–!QIÚˆdìiGžZ›?ðOˆ^Óö~”¶™ig=Ö½u:_B—Œ^Eº,m)xcxæÙŽ䛃–îËy ‰ZI0Äd¨F8÷tüs¾ÅðËá~‹¡[„‚âÂÉažKy”Ìß4Î…ÇF»r —ðkவñ‹ÅRi>Ô'ŠÝ®®ä "•^J¡cË(Àÿ/óÊ1•x߉q5°qs•ZA-Ú^ì;ý”ºéè[dÔ!Ãy=(bŒiÆòodÞ²ü[=ãþ Õ{ºxÈÜ-Wìd³vÿ_Ôuž¤b¾Šæ+–²È“*6Ýèw#pÊÝsÔgœŽ ×•üýœíþø0XEmm©_ëÜêò³±Ù[H, ™1ë¸nR¡onÐL@ÀŒ(ÁÈË·9$míÐþÿløo‘b¸k†ð™v1%V M¥­¹§)¥ÓT¤¯çsù«ŒsJ9ÎqˆÆaïÉ&­(¨þkO"z(¢¾àù¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¤eÏN´´Ri=ÀŽŠ(®@ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ )TdóNØ*ã%t(§ìl^É€Ê)ûÉ€Ê)ûÉ€Ê)å8¦TJ.;€T˜•IZRêéF¥V¶@”QE0 (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ F\ÒÑI¤ô`0Œu¤©:ÓJzV2¤úÚ(#h¬¶¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢ŠP¹¦“{”ž”𘥭.à3a¥ ëN¢´Tâ€MƒÞƒÞ–Š|‘ìlôl´QÉÀ&ÁFÁKE>UØØ(Ø)h£•v6 6 Z(å]€M‚‚–Š9W`ã)´é;SkžzHŠ(¨CÞ¾dø÷ûmOø‹Ã^°¸²žZÊ=]nÔ:”p²2Fã8uVaƒÀÐý«jØô)5ox6ÞËPk‹G´¿¼iY–Ùœ:I PTîÜ@$©\‚ʵüÏâ׋UpÕOb,×4jÎ6ìâáã£[¹ÁÝ=M3ößø ³Ö•Ó³§÷ó4ž·þY+5º;ÝSöñ>·ð¡¼«ÜAs¦³&f’2n|´`Ëüãhe¦xÆqÅpTTÚ~Ÿ>¯ ®— ·77.#Š(Ô³ÈÄà*É$ö¯æÜvgŽÎgMâêʤ£ݶÔUí嫲ó?gÂà°Ùtf¨AAI¹;+&ÞíýÇWðWà®­ñ¿Å‹§øy|›Xp÷·®¤Çg=O÷˜à…Låˆ=f_Ð 3KµÑ,"´Ñ­ ´µvÇ 1ˆãŒz*ŽúWç÷„þ$ø³à'ˆ5;/ _>—y­m{ÈæBñ±R`ÊH €ÃÔààœýÙðÃZ¹ñ/Ã_j:É.õ 2Úæw ;¼JÌp8$ð+ú‹è÷W+£K…¥Nk£ªä—-”šŒc­Õ¯y]&ßV’Kñ¡©:ç8¼6ÐI»Ý¤Û––wéfôõwÜ¢Š+úPühñŸ€¿ðOƒ³Ç/|Kø#à{ ÇŸ.¦»ñ´÷W7wW5ÃÜÌ#7¸¶ŽIŸÌxà£*|¸öú—iFXüÍíò©#àø5£öñrüø™ûH|iñ=†»¬þÓ:ëjÒAoh!’²¾Ô£¸¸œ¢¤k$÷W7DCDŽ2™ q2ÔúòïÙ ö|Öÿf¿„‡@ø‰ñÅ?u™ï¥½¸ÖuÉä’A¸*¬0¬’HÑD©†GË´jñŸ¶ÇßÚ;Ç:Þ'ì;ñÁžÓ­à™uH5}<<÷–]Ž²½½Ê² ˆ©ÜKI½DY_?e?ŽzOÅoø»ö‘ý ®¼W‡à¸kï éÞ‹MÓîåžÕ¢dy­äŒ\¤RxÞh ýÞBÆ\á_Rí¦çÓ5ò‡üïöºÿ†+ÿ‚büTñf‘¨}ƒÄzž˜|9áæ‡Xþ˽÷çìÉ5œ ií£’kÀ‘üåm†@ ‰õ}~6ÁÍúŽ·û\þÙŸ³ì{àÍbÂOêqx“\–-";ëkXî'{ kàQ¼æûëS̀ljՙة0²ÿ‚pÁ¤zoí'ð?áÅ/ÚWã ý‡‡|o¡Áâ Í Ãš2¥Ä÷vækDS¸wQ(ImÞE6Œ3æF¤€&¯²þÁ«Ÿ²ÃUüQ¢ø÷âMµÙˆE/ŠÞÿ…ÙàÏú¼1ÿƒH?øºã¼]ûjxÂÛÖ×PºÖgŠs Åan[Î\I&ÈÙr1•cœ‚29Ñ_›c¾‘YÝj|¸\5*o»æ—Ü®—ßCì°¾å´çÍ^´ä»h¾ýû¬}Mÿ±Tÿ‰w…®ålt’ùcGpÛ?ç§{û}xÖîÎh ³ðí³ÈŒ‹4vÒ—ˆ‘€ÊR¹FArJñ +àñþ-ñf`ýüt£þoþ§øéÒÇÓá|?È0Ÿ?ñ9KómŸS¬ñ×Ç_|I·0xË^½º¶eöèV$· ñFXƒƒ’ àz ä袾˜b³:®¾2¬ªOù¥''÷¶ÙõX\% =–š„{E$¾å QEz—Â?Ù?Ä?âÑïÐÇkáýU%fÔ’CjQ¤M­ uf%ãåÈÞY6Gâ BÂeÔeR§h­“j7oeÚ¼’½ÛHÃ1Í0™EˆÆTP‡wÕÙ».­Ù;%«Ù+œïÀß„7?|{‹cqö8„2\\\ùbO³Æ£ì,»²ì‹€r7g 5ö'Á¿Ù·Ãß5ÛÏ É¨]]^Ä°´·’#´hJ¦Ä\;IÎ~êÖ÷ÿ…ZÂí ÒÇÂV1D-#dûC¨k‰Kí23ÉŒ’Æ4' ù¨eøwá^ „hÄjc/ÍÏ«å¼mËé¥ß½k»ß²_Íüaǘž «*8i8a­n]=íoyu×M/e·vÊ’£©+ö*]OÏ‚Š(­€+ÈÿlßÛ7Âÿ°ßÂûüY°×õ;QÕ#Ò#H‚§Y^¥ ÂYc]›mÜœ•ã©¹UuÍËÄú%æ›â[;]GNÔ`{k«[˜–h.bu*ñÈŒ º2’ Aƒ@Õ¯©Æ|ý¨~þÔž“Vøâ­/Ä–°cí ´w6yy|ûiËã›|Ä]áw.G5ÁhðMo„žý¥ôŠ¿ ü?ÿˆô´¢è̶Úeïjmyi´¤{#g#Èò²îÌûɯ-Ñ?à>ø?ûHxsÇÿ²wüSðâ×O¾‚m_B†iníµkH¥·‘¬Öc2J‘J`c"Ì× ÅÆB?fRßr›QøYù'ÿ[ê·'í%û4~Æ¿ ®ì"Ö|{®¿‹õÙžÑÞëC±‚)í⽤–(&mÿ·%{pæW6qª”.¢_Ôÿ…ß 4Oƒ ü;à_Ù¾ðž™m£iVžt“}–ÒÞ%†üÉöÆŠ»™Ž2I95ùCiãølÿø;O¹øi6…ý•û6øëOÕn£ÔþÕý«þq ËŽ2±Ï çˆÖÞH]¾_±ÌKÄCõþÛ´÷Žÿf @Ô>|×þ-AªO5½òén“élª­hb·šGI—  jæEƾÿÁP>-øûâ—†´/~Éÿ|=¦kZ­­…æ«q%畦C,ÊpûôÔ]±«;™FäÍ}«E+2”•­`½ØÛC¥soboãbSs(R_2? ¡IÈQœrO ÈOØ:8ÿnïø9Ãöƒø±sm¯ëø¥ŸèwZ†¨ÑTUM4Çkn³³›yÅ¿ˆäEÈÆGH¦‘qúèÚm»ß§‚rPFe(7í!wuÆyÅ|™ÿïÿ‚MØþÁŸ~8x¿Røâø‡ã‰›]Ô.î­ Ó<ˆ„·3FŽ–¡RIüË릒TDù@–ð…!™'½|eøñ¥üøsâ­~òÒóYÃv7zõ®”«ÙmM$Mpì7H6ù.¿9cØóÅŸðX Zÿãí÷â}vOêI‡¼-¦™,´æe1 Ų8eŠËli0“4²DååDZO]ÿ‚êê÷rüø_ðáT×>"ø–+k+[iÒ [è­Õb[YݤPw\ÞÙ:‡7D]™J)?j|ø?¢~Ïÿ <=࿇py7†ìc²·ÊF²M´|ÓK媫K#–‘Ü(ÜîÍŒšÙª÷#~¬>|ðìÿà¨<;ð_ÃÚW†ôx6Ÿ³Ù@#óœF‘ù³?ÞšR‘ id,í´bk©¢¼gößý·ü%û |$|D¶ê—»áÑ4He s¬\(ª’—nS$Ä€Ã;GÖÆi9;çüƒöüÒÿa‚Í©ÙÅa¬xËWZhºD·(¿;+ŸµOq)¶aÜP|ÎÑǹ7ï_Éÿ|ñ_í“ñ§Tø¡ûH¬¶ö&ºmYãóŸuðv>\‡v’dET]Í‘ƨH!Öž??à¤?´…ïhYe¾Òb“eÏ”íokm‚ÐéökÉXÔ°,ݵÙüÉ7?×µüë㊕ø}ÿce3^ÞIûI­é§´c®“kV߬Ҽ“ì^p5,Éhc¢ýšk–=$Öíé¬SÙ-ÝÓÑY˜EÂÚÁkk±ÃmA *8 Q€‘ PŠ(¯£¿g_Ø¢}Zcªüq³–ÞË`û6˜e),å”ò”;£<&CnvÐ0ÿÍ<5¹¯c¾­€ƒœ›¼§+òÆ÷|Ó•¯gÝÉè“z±gY¾Û%¥îÅZîÝ"´¿O%Õ¤w_°”ÖŸ/$º†XÒçWšXYÐ*¡BÊOÞ‘×#º‘Ø׸U}'J·Ð´«k&%‚ÖÎ$‚פh *¨Ï UŠÿ@8W$|9”arÇ>gJ*-íwÖ˵öò?”3ÜÉg…|j*œ›·eÐ(¢ŠúÉ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€#¢Š+ŒŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( 'Zu5:Ó«¢–ÀQEhEPEPEPFzÑEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEs¾:ø·á¯†‘3xãZ°ÓÝQe<›®¶†XW.à òô>‡ؼfI×ÅTŒ ·”šŠ^­Ù#l>®.j•9Iì’m¿’Ô訯üÿÒtã,4‹JQæ¢ÜÞ7‘aÂ: ËȤä•o,àÁø™†G‡uÒý8KÿÄÐ~x”0úîOAö yÿÇiÄÀcèG?üÿùHÿâa¿ègüòÃônŠüåo…~(DÜþׂú>lzÿv‘~x•×)áÝtƒÜXKÿÄÑÿÿ´þßþ —ÿ)üBŒ7ý ãÿ€/þX~Ñ_œ‡á‡‰C|;®äô`—Ÿüv±ì¯¦Ó/a¹Óf–ÞâÝÖX¥‰Ê[3àLï+sö˜w(ÇíCÞMwÓ_=RijÒ=b 5™ªZëv]è×0]Úλ£š‘È=UÁJž¿AŒ£V*QwOTÏ‘”\[Œ•šTŠJ’®j%K°†QJTŠJÅ«QEQEõû´´‹÷ik®; (¢˜Q@Q@Q@WÍ¿ÿn?ìmJM/à«ÚÝ4>l77óAæG» í˜>ió2YJ’Êr~gŠ¸¿,àÜ'Öó*–OHÅk)>¼ªêö½ÛÙu{ÖGÃøÞ"¯õ|.Öíéú¾—éÕŸITW×Ði–S\êSEoon†Ie•Â$jK3I¯†¿á²>$ÐÇÿ”û_þ5\w>&ëÿõ%»ñæ©u¨Ê™òÕÈXáÈP|¸ÔLì\í$dä×ã™Ò7(§A¼¬êtSä„}oMü¹uîÑ0~檖+üy¤þ梾wÓ³>œø·ûxé^¿¹ÓþØ®·qxä’ì´à \™—%9@våIðo‰ß´Ï‹~(ß5 B].Ýí~É-ž›<Ð[N¤±%ã2ć*sÔ+Ïè¯Àx›Å!â—8âq4›Ò÷b—m=çÿo6þZªä¼”dj.)¯µ-[óì¾IôÃ_Ø+Pñ^“k©x§_±³³¼†˜Î&¸’HÝwŶ8+‚7ŽO§>Ù¥~Çßô™­¥O,òÛ`ÓÝÏ ‘—.…ö0$r¥vž˜Çïäñ6yIWœ!FÍ:’Õ§Õ()?ü –÷Vº<¼×ÄÌ—,›¥J¤•ïÈ´OÍÉÅ}×·Sá:+ô»FÑ,¼9¦ÇgáëK[ 8såÁoÅd’pª$Ÿ©5j¿A§ôk“ŠsÍ,úÚÕüŸµWõ²ôGÉKÆX©>\ŸõòßûcüÙùE~œQWÿÕÿS_ü£ÿÝIÿˆÍÿP?ùSÿ¹Ÿ˜õØ|'øâ/Œº‡ÂV…mAe’úá]-!ePv´OÌw/Ê?08ÆHý“µ6º0?G<6c3RšÞ1§Èß—7<­÷_³G6/Æ*õ)J8|"„ÞÍÏ™/—,o÷Ÿ%7üßÄÙ{×\Ñ î÷DJ"Îç <͹ÀQg%˜q´ô¿‡?±w†ì¼)n¿4ûmW\é4¶×ꀈ¨±ùyùK3)fvrIÍ{Uú>Uá' e½µ<*ž–µGí"ÿ½Ë+®n‰ôMÙj|f?Ä ó0§ìç_—[Þä½/;u·].rÞ ø'áO‡Ú›^ø?C±²»hÒ10R°m,IRGÞ#Ï-“ÍtvZ|dV›6ñi6Ds1fl噘žä“Þ¦¢¾÷—arè*xZ1§tŒTVº½[½ûŸ)‰ÆWÆKž½IM鬛oM·íÐ(¢Šì9‚¤¨êJÚ—P (¢¶¬¿x_þoêú(Ô5m'ûbÊ{·is/¬<Ä)çA.—*îܯƒµ”qZ”û{§³vx.Q”g¶å#<ñ‘œþå±ìÕÿ û5xoáÑÖ¿á!>ûWüL>ÇöO´y×S\ªónß;oÞ9Ûž3ÒüyøŦ~Ï|gñÆ°_ÝhÞЯ¼C ’#ÝMoin÷,*ìªÒ‚†e‘’5ÐÞÛKjËs$Ä4H±fF,>O-ÔN27dŸÿ‚‡è:—ÄOø'oÇm#Áf£ªë:ÏïYXéÖ–ï=ÝÝĺeÂG Q -$ŽÌªª –,6ç# Û»»> ÿƒXg]C]ø'ñ+öŸøëköÿˆß¼O}$ẕÙ¥XG9{‰mÖkçêMz%‹(ö+fXÕQÿWkó7þ Uý«<#ñ‡þ ™¥|4ð”×ëâ߃×WVþ ¶¸¶)ÍCP¾¼´ž A+$n,dd:½¼›)äý2 Gòáÿÿ‚ŠþÖ°¿üƒã†ƒñâN…&…ãNïFÑ®ü@Ú¾‘m¦Þ´“اÎóZþÅw$m0’Ÿ*I äÿ¿à¶¿µ‡íá˜/ü}ûFø³KºÐ.–ÞÓMðô­á»«ä¸GigfÒ­à†hâ6ЩoSr¾J0iÙ>¾ÿƒ°à~&ðíp~?øK×u¯øûL·ÿ„Žò+inmü9f¶Ö ö‰#€Ekñ5‚ÅæJÏ$Âä ‹_”?ð«¼Mÿ 3þ¯øGußøL¿´ÿ±?°~Á/öŸÛüß'ìŸeÛæùþoîü­»÷ü¸Ïûÿª~Ù_ÿhø(WŒ´_Úâ·ÄŸèÖ¿ïoa°ñ‰¯u;hnRÓfX®%eYK"‡ ; àšýþ¯ÍÏø6Ëþ ]ÿ )û#§Ž~7øWûãÄ%Õÿö…Ÿ•©èNå6ÚsfGò÷ùBîUÛ ›æŽ)“uªmý#<O~žôó‡íeû1jlÏÙ×ÄRhŸÚ¾ø}¬j:ÅÇÛ°\yòX>Ï1d“ý*Ú3„ >L8ÚN~¢²üsãM3á¿‚µxÒçìZ6c6£q弞E¼1´’>Ä›¬p “ŒMm»#‚ý°k/ þÆ_o¼añ*_ùék¥Zm—þ&—þD³CkæGžO™ä2ù®»©ìã_†l<[ÿ#øߪ|Ký¤õ?µi°Î-¾Í˜ãUO,mcÜLè$œîbäîiä¿i¾$ÿ‚­þ×ÒêzkV_´Y ;© Çå?È¥ÍÃFÏÞ6v¡wŽËô¦…¢ZøgC³ÓtH¼‹->¶·qo.4Pª¹bIÀd’kð_.‡Ìðö•û¹¢o1>Û+#mD‘ ãgÊí†ÈÊ a²>ׯ‚ðƒÂøñø‡>=)7É6ý¤®ù§>ñ¾É»ÊWr÷Ræú_xÞYGü$åo–i.i-9HÇÎÛ¿²¬–·åŠÊÊ2Êm6­ííÑbŠ("DŠ0TpJ–Š+úÚ1QJ1VHü ·'w¸QEÄQEQEQEQEQEQEQEQEQEQEQEGEWQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@N´êju§WE-€(¢ŠÐŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Šà|SûPøÁšíÆ™âCí£™"·šqªZ4eÜg ðpk̾$þßö:>¸m¾马Z"×—ð,Žy!#+»¦[9ÀÀ¾+9ñ†ò(JX¬};Åò¸ÆJsO³Œ/%kkt’Ùê}&]ÂÎi$¨agf®œ“Œmþ)Y?+=O unËÚl—¾!¼µ°³‡d÷¬Q¦H,Ä’@ú‘_?|Xý¾¬´{‡µøGe¬ÛTýºìI³’¢+·%yìG_øáûMë¿ \4š^™!%°·Œ7¿v÷¾ê`6BíÈÁ'<_…üªø×X·°ð­Œ÷·wN#Š8Ç$ú’x ’ÇI ü¼qÇæ˜ìþMAÙsò·RMô„]ùV¶^ï;{rŸ«p׆X\ /­ç­9-yy­¥ÖMnþ|¶ÞçEâÚ/ÆÞ(×îµ ¼I¬Y§Þ-ì¯%··„t ‘«`ÉêI$“Ä×Óß`¨.t¨/¾2\^Ãvϸé¶ïX¹âIT¶ðêü…H®z{†~x7ÂZÃ:4RY8’ ä¶Y®#`ÛƒyÏ—$„¶F: ò2Ï8¯‹#fqˆöj^òU§SÞÕÞ?e½ÚmI;&“½½ oˆÙBÞ/¥Ïm=Ä£4^÷[ti5mSïñG€ÿgo|D½¶CЯ᷺E™o.âk{_)ŠâA#€aƒa71 W³x3þ áƒm/Ä/gï}¢×OƒêdòÀIÌ~ z¾œ¢¿]È<á̪ÓÆ©b'ýæãŸf-_ÒNIßT~šø§œc½Ü5¨ÇÉ]ýò¿Þ’gŸ|3ý˜|ðªö ÏiòO©ÛîÙ{w1–Q¸žƒ„RÛ•Pq×'$ú Wë™^QÉ(,6_B4©öŠI_k»nôÕ½_V|;0ÄæU}¶.¬§.òmü•ö^K@¢Š+Ñ8Š( Š( ¹¯üð§‹mxwH[ø¾Ì©"òÊê.HÁ®–ŠåÅàpØø:xšQœ{I)/¹¦oC[ .z3q}Óiþ™?ìuðâG,Þc“‹û?!/ÉüEýü;âK»i>ÞÉᤠÏïRcžo”2‘Èê /ƒu RÜ9’K)YžÖrp xóŒª7 6W¸|>ÿ‚…ÿ­OŠº'«G>’?ÝÂ4R¿ûçxA·©«^<ÿ‚zY>š¯ðÃ[ºŽñ3º-P«Ç6Jô’$< çî¾NËɯø¯ð3Äõ+Æ6Möf*±_@íf$1 ²`a¾VùXÀÎ1‚-•?<+…Ó’ÃǵªR×ïåÛûºú¦þáO„øêV÷]gë é÷_ñü¾ìðÄÝâ~š×~Õ-u“ýb¡+,9,™é’Àd ŒŠÝ¯Í\½ðæ§î¬/!Ï—=¼­‰AÃ)d>„×¼üý¹u‹,?îEö“r¢3qª$¶oŸ¾D`nNNá‚xzoÕxGé€ÌÝ<6wKØÕm.xëKÕÝóA_O¶–îImð¹ÿ…¼l¶§´‚WåzOÑYZOÿì“gÖ´…3Ò¹O‡¼+ñ^öæÛÀ:¼W׈%’#¸Bq¸,Š¥€8Œã+œdg¬¯Þ°Xì&kEb0ucR›½¥ )'mšmh÷?+Åa+àª:8Šn]$šsÔaëIRu¦”ô­eM­Žq´QEf×îÒÒ/Ý¥®¸ì€(¢Š`QES..#µåºtŽ(Ô»»$“Ð ñŸŽ?¶~‡ðàIcàSoâ döHEµ¡ÙÎàbO˜±f”€‡Ÿñ.YÃg‹Ìëªp]÷o´bµ“òIévôMžžU“csºÊ† “œ¼¶^mì—«òÜöšó‹¿µ—…>ùö¿iþÙÖ"Ü¿b²`Þ[ãÉ÷cÃ&Öºä„WÈþ:ý ¼eñ'O[Oë×w€ÐF©o x±*‡ÁPFìã¶2k¯ç'úDÔ­K!Ã8_íÕ³kÒ ¸ÝtnR]â~Å’øE rU3ZÜßÝ…íó“Iü’^§mñwöñ/ƹc_]D–P?› •²yvñ>Ð¥°If<]››Šâh¢¿œ3,Ïœb%‹ÇU•J’ÞRm¿Ç¢Ù-’Ñh~ÃÁaòú1¡†¦¡²JËúî÷}BŠ–ÊÊmNöm6n..bŠ(»ÊìpTrI$Z÷6«¯Ù‹‰×Òè/ÖÖ³Ž8®x@RYfIJ¡9uS¹ÁÎ+ÖáÎÍø²«£•aÝKnôQ¬¢¼•îí¢gqÄ~CMTÇVP¾Ëvý"®ß›µ—SǾ|'ñÅ{Û‹‡údº„–ˆ$˜‡HÒ N]ÈPNrpp88ú;àì/e¡ý‡Vø»/Û¯“lßÙhÛBß7É3sçc(HW*AÞ§Ÿ ´Ý: N·´ÒâH-­cXa‰4P¨€ š¿­xGÀ¼— œ19„ž&²þejiÞé¨jݶ÷¥$÷åNÖü?ñ?2Íc*8D¨Ó}µ›^ré}ýÔšÚï©EWíçæaEPEPdíM§IÚ›\Õ>&ETQEQET•I[RêEVÀQE8"ωqó™Ѹ“í·éŒÿœVô’é2·] Hc‹³†‘Tc÷vîãíõÎ+”ý¦?k?þÈ^¸ñÆÍvÛNT‚I¬¬T:Ž®ÈP¬à,WÝ,@‘…Má‘rÂÇÁ¯Ú;Á´Ç­|Yð+ÄúÖ—%ãZEql’Dâá[†EY"~U¶È ”tq•`ÔÎ×?&/>(\ÿÁÿà¿1oŠ?Úð/ö¾ºV:µÌ°Ã©OxÌ÷7sÁb;;«»Ñ$)0YêM#K""7íL–’ÅfgxßÊÒKÿº:·ášùgþ û é¿ðRoØCÆ^ :f£¬x“HµŸÄÚ”v,šõ½¬âÕZIää2½»‰¸Ù;°1¸ISóßöÿƒ—ü=û9Á"èßþøƒÅÿo›áß iWzÞ©v!’²YÚÄÓO)Ž%gm¨¬ÛUK6POóì£ã_Øëþ mâÝgâïìñ¤|#ñ÷Š¼%©Ú>§â;¿Eˆt˘¢Fµ¸v¼µKÅÚ*Ã>6“nÉæ øËñâwíïÿ ìX|4Ðo6'!c÷Y‰($ž>‡åø*§íuñgöCøg¿g/øgTð}”SÙëwZ赆ém¡µ’#¹†Vó%šUݶLmRvŽOáwíiñþ ÿTø±¥ÜünøÝã @x¾ÚâÛFÔï¼L<]¥ê‹ ÚKpÑYê¢V‚HšâüÙ-â|™lÈX·Ó<ÿ‚|Xÿ‚àZüøW‡¼'àùõ‘ ô¶pê$:¾¨¶sI-éžá|Ø`Š×í2­ªy¬«æ × ±á7b髳õûöý£5ÚËöDðWüY§Ûéš®»m*^A—‹Î‚â[i22ªípœí·scqüðÿ‚Â~Üßµ7Æ[O€¿³ÝΩ“£ê²XxŠB+}Zþ9yn †ÞÑ£‘™‰Øî ì"¤o¡?à¢_µŒ_ðNØŸ þk–º—Ä;¯ÛxnÒd’KK»(­M´šª¤LZÝØ‘q#S ‚E?Á7þAy6½â[OŠY-dŽËN»rÂåXÎsò¯ÝŽŒ@8,+ä8뉣ÂYµåh­5œšŒ}RnîÚò§cèøW$yöeK ´[»òŠÕþ ËÍ£é‡>µøið×Dðχ൶²Ò żF5»¹eQ=Ü™™¥•Y› ©â8ãEܲ²›S½†ÛM†[‹‹‡X¢Š$.ò»U’IÖ¢¯¬?cOÙÆÓMðü^(øƒ¥¤šÄ±ÜiFYTŠºLª¤€ä±ûß2ìRž¿Ä|3ÃÙ§‰¹Û„ê9Noš¥I]¤»»'gÒ E²VKOéë7Àð^Z¦ ”cî ڿ›ùÉê÷z³Ø¾|/²ø?à;=E"_ žãË×R·Þ‘€üÉ$*¨ÉÅtÔQ_è C-ÃÓÂá £N F)tIY#ù?‰«Œ­:õ¥Í96ÛîÞáEWQ€QEQEQEQEQEQEQEPN:ÑM“µLŸ*¸ ¼zѼzÓ(¬½«ûÇ­Ç­2Š=«ûÇ­Ç­2Š=«ûÇ­Ç­2Š=«¢Š+ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€iÔÀpiàäVôž–¢Š+P (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ Šúú2ÊkJh­ííÑ¥–Y\"DŠ2Y˜ð’OJâ¾7üÑ~è¾v´Eî£7iðÌ‚wÈ}®ÊNV-Ñ•.`ö'Šøûâwí#âߊ7)ªj÷–ú\²Jc±Äq¤NxÌj¾p—ÏSŸË8ëÅŒ§‚›Ã;ÕÄÛà­Üòéè®ü’ÔûŽà<%YZ™õÿ ëêì¼ú»ñ3þ ,Zœü"Ó-äµT&yµH›tŒO9©$’z e¼§â_íK㊚mîŸâÛXt«ýže½ª,caVv 'Þ@ß|ó‘ÓŠâ<=áëßk¶šo‡­ÞêúúU†”€]‰Àäð©$9$ úcá¿üÿO¹ð¤|SÔ585y¾w‚Âh„vÊ@Âdmî9Ë·' noç<cÇž*N½,5yû'~dŸ³¥f­Éu½ÓÖ »­Zê~ÁŠÁð·F”ëÓ´_ ×<÷¿6»Y¯‹Kl»0XéלÆ=6 ®$vØ»c×½sÂ?°çŽÞ¸ÚØ`AÈþ øaû=xWá è»ðU„±_µ¯Ù%º–æItÊ’YKl²);T{8®Ú¿Há_£Ö•?kÄœ¦ö…7h¯Y4›~I$»³ã³ßëTŸ&SO–?Í5vývKÕ¶û#營°—e¢£üJÔ®çÕK±eÓäU·Eþ ‘îcÆIÂõÆ8Éö¿|3Ð~ÄËàí2ÞȾAuËÈAÇ˽‰m¿(8Î23[´Wí™ä|1ÿgác$—3W›·^gwwÖÖüÍs^&Ìó¦þ·]´þÍí¹iép¢Š+êÏ(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Šò/~Å>ñm•ÉÐìåÐoçv™n-%fErÂä sµ}ÐQ_;|Lý¹ñí¯à? ÌÑZ^Þk2Ç+E"éöå‚mãpy #®G³ÔqÍyWÄø(¥¬é·Ÿ t¯ìiLÿº¿žUž_(ÿ,JlVo—9.,N|÷ee6§{ ¶› ·±EH]åv8 ª9$’­uš7ìùãwRŽÒËšêK.vµÅ£ÛF0 æIB¢ôîFNäŠü›âçñJ•pŒ½ÛQ¦Û×D”½é)vqißky…àÈÚ©ŠjMkz“IiÖÞìZõM[s/ÇŸuÿ‰ú’ÝøóTºÔeLùjä,pä(>\j&v.v’2rk ½ÿÂ?ðOŸê›ÆZÆ™¤Å$öB­w4r~í×äN2ÙevdwøSû1øOá™ôkGÔnå•ezˆŽi¡Û‚¢2†ì€ 8É8\NMàßqN%âsDé)k)Õ|Óz-ù›Ûâåõv+1ñ"Èèª8+TkE+E|íÊ—¥Ï™~~Æž,ø…sm6¿ðþ‘q ˜]Üy÷U`$~m£9<îÞý†üáý`ñ”ø‚ü»;ݼ²ÚŒŠ±Ç&¥‰$œã{5ýÃ~ ðÏE9áÕzeVÒZÛh[‘j´|®Jís3òçÄ\ë6•£WÙC¢§î÷Þ_óÕ'Øóøc†ÿô.åBëÿŽÑÿ oðßþ…Ïü¨]ñÚôê+ë?Ônÿ¡eüOÿ‘<õŸ8ÿ Ú¿ø2æfxGÁzO€´dÓü§ÚéÖiƒåÂw°P»œõv!TbXã’kNŠBø¯££F–œiRŠŒ‰%d—d–ˆñêÕyº•$Üží»·êŤ,Zibi(u{ó½!“Ò›EC©&¼ÃG˜i´Rç—pæBæ’Š9›êNzÑE QEQEQE%GNó iNJ7¸¢›æ<ÃZ{H€ê)¾a£Ì4{Hò7ü[ö3ø}ûEøkEñÆ=_Ç^¡¤Ç>•¥>øçnÓjQeëK¹“R´D²6ͪ'ýò ©Á!€\+('>áؾ*ÿÁ6þ&'ŠþÜ¥ÝVÝw2Ke*n(‚î0G–ÿ:‘ó`ï(Áu?_Â>-åœV”)Ë“+ZŒÚI믳Ÿ/¼ÚÙ;6ÒVŠ¼ ?ðÿ’{Ò\ÔUÿyߧ4o¢¾í^Ë»ÐýîðŸƒôxjÇFð.—§hº>™o¥…²[[ZC,qÅH¢"""ª€¨~tüÿ‚~Κgü¯RðWÃi¾.iÞ4øc=¯ŒÖnì_BEIlîb·‰Ú6¹t_µÀ˜r…|ÈN»?ƒ_ðpŸxÞçãF›®øWÄ”¹²µ·kø ‡ŠU J°ÁÃ(e;—·O'üƒàÇ“¨ø‘½s£È?Æ¿J©›Ð¦í8Îëµ:ñŒZ~©´|]<¾»W‹Ž¿ß‚üOï±õÇ?žý¥þëÞøï Øx›Â^&µ6š–›v¤Ç:d2ÊCG":£¤ˆUãtGFVUañ§Å_ø%×ìYÿýðÎñSÆ^Ô´¯øWšž«áäoë—’JÙĶ‘ÚZO~b–PÑåQ—`Ubø\‹ÿðpÏÁ+K颷ðÿÄ˨⑑f‹L³ 0”=â°¨ÜçðÏüCþ gÿ éâmKøw§êš<7ºâÞÞú]·:•Üˆ¡§¸†9%òÀhâ³ò¶ÿÞìOGÚ)EI‘¡.k3Ì<1§øƒöýý©õÝwâ%îÛfyµ½fhîmm÷ª,6Ëq!rŠd‚Ð q€v”±ö¿¼cðóÂv‰áÈü»-2‚<ª†|]ö€ ±Ë1e˜žõÂ~ÈßÓá?Á½9u ­5½V1w¨¾Ö1bÍI¼R‘²©L­¿’OÐÿ¾ Ëñ×Æòèðj ¦,w$Í ”…VE¦FI2/R8ÏÐÿø“Å8ßsøäùZr¥ rSãiÍ9'RîÉ_ewe}hþ•á— Â9[ÇãZS”y¤ìï´šµnÝl·vÖÈìÿfÙB÷âf©©ñÎòÇÑ*Ëpb}D¬3†òŠ¶Lƒ¯E9É_²ll`Òìa¶ÓaŠÞÚÝ(¢B¤h£T¦éze¾‰¦[YiQ,¶‘,0Æ¿v4P¨öS×õ/pp^à ¹ªÊÎs{ÉöÙZ+¢ù½[? âž)ÅqF'ÚÖv„~­’ïêú¿Ð(¢Šû“æŠ( Š( Š( Š( Š( Š( Š( Š( ›'ju6NÕ>h¢Šæ¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(§!ÁÅ6Šiò»%ŠÙµÔÕÀ(¢Š`QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQXþ:ø£ü4ÐSñÅüZ}’ºÆÃ1wnŠ¨ ³§=#N&Ž”«â&¡«¹I¤’[¶Þ‰y³J4jb'T¢å'¢I]·Ù%«6+æÿÚöÝ·±µ¸Ò¾ \HÚŒ7$ÔÌ %¾À2ÞFâwØ]Ì›q¸®r­^sûDþÖ÷ŸôÖÑô 'Ò´A.÷.鯱)æc…u¶ Ø`>c^K¥i7z‡kqyw9Û0Fd’CŒáUA'€zWò¿ˆ¾7UÇMå|1'ÈôuW2”›ÒÔÕ“¿›âoᵯ/ÜøCÃHa¢±ÙÒ\ËUMÙÅ%ÖoTÿò[ÞöSøŸÄ÷þ3×îµO]K{zþdÓHycÐp8Q€ÜüýšõŽ:¡“÷šn‡L×ï`çœG8ÞÙó…žv«{Á¯Ø2ÖÊ?¶|k”^Mc°´–¥9ód1`Ç¢ϼÁ°>‰ÓtËmÂ+]Þ KXdPÃŽ8Ç¢¨à¥aÀ¾csJë2âvãù6ß´›½ß;û)îõszß•êkÅ&á°4ž $³’Vç²äŠþêê×M9WžÇ?ð»àþðwF{?Ùù~Ãq<Œ^k¦UÀgcø£ 6É®žŠàü[ûMø#ÀºýÆ—â½i¬ïíNÙak†)‘‘ÊÆG ‚yÁý7S”ð®:“§‡ ½Ø¦ãN7ÕÙ]¥}ßw«?…~{^s„gZ«ÕÙ9˵ݮí²üòŠóøl†ÿô1ÿå>ëÿQÿ ‘ðßþ†?ü§Ýñªàÿ^xoþ†t?ðu?þHêÿV3ú«ÿ‚çþG¢Ø_Iwu{Öï ÚÌ"Û8LhûÇ2ì½ùCôk”‡ãw…õ ßj¾Ö,uXôý=õ9mí'G¹¢o9ˆ°e=n â¼Ùà ž fé¾&POSmþF¨Çq¶E“¨}sÏwtÓWèãu¦Åaxk4̾¯…“å²jÚ§n©ë®çºQEõ‡‚QEQEQEQEQEQEQEQEQEQEQEq>!ýœ< âk%ƒRð®#‰´€Z>@#—‡kÉùIÇCŽqZÏìà}SR’{)µÝ:'ÆÛ{{¤hãÀƒ,nü‘žXòN00µÑ_+˜p7fºâ°¤ûòE?½$íå{î‰ó|”1sK·3kîw_=Ï ÿ‡}x3þ‚~'ÿÀˆ?øÍðï¯ÐOÄÿøÿ¯u¢¼Ïø…Ü)ÿBêsÿ3·ýwÏ¿è2‡ùÿúðgýüOÿñš?áß^ ÿ Ÿ‰ÿð"þ3^ëEñ ¸Sþ…Ôþçþaþ»çßô?Ãü ÿ‡}x3þ‚~'ÿÀˆ?øÍtV?±ëK(b¸Ð庒4 ÓKpR 0GUÉêpô½NŠèÃxqÂøVÜ2ÚNÿÍ/ý*öùÖã îºJXÊŸ)8þV3<9àÁÞwü":F™¥}§o›ö;Xàóvçní€g›é“ëZtQ_aB…<4:1QŠÙ$’ï²>~­YÖ“I6ßVîŠ(­LŠ)=©7epÀw¤2zSh¬VöIÏZJ(¬÷¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢ŠùŸöÅÿ‚¬ü*ý‘tY£}Z×Æž*ÜÙ&ƒ¡ÞÁ<ö·© ·Î‹Dl·“,JÆûh5'd}1_”߶Ÿüßþ+Üx/ö?øe£kúäzÅ›ø†î\C¨Jd’[hì¦Uš3²)ÒêiJíߺ jà?hßø*GÅÿø(”cÁ?´9| áMFÔØë°Û]}°Ý ±7WÞB!Ù_*0¦@eV2‡®§ìÏû3éß|<^s÷ˆï£öôp|˜sÈŒ2x.@'UOÊ|PãL‡°3ÂãiCˆ’÷iI)%}¥?åŠß¤ž–î¿Càn̳,Lq'*TÖó‹kmÒ¶íüÒëÙú…Q_Ã'ôáæºßì}ðÛĤ·—þ´ŽY±¹m§šÚ!€ËN¨¼ÀdäžI5Sþ—áý ?ùQ»ÿãµõv—ûøçWðdZ½­­šË2î]6iZ Ð7íù–E§6 ƒ~+éÙ{áüÿ~[YkÒÙê\M5Ü.É sucBH9È8"¿zàîã|÷&+ŠÂÒö|ñ“•Wh”R犾¿ ÓK¡ù_gü5•Puèa¨×©ÏÊ┾­·îÉÛMífúŸžßcÿxÞþßDð€]½½Å¤1G’XíE’I8îI¯¼è¯Òéx…ÅQäͳ:õ¥{Ý5ÿà3U÷×›®Ýþ"§Š˜Š59°*Tãkj›|y4ÛK|Ï‘tïø'׊¥¿‰u}cÃð[Y!y¥‘¹T1¨cìX}kêø3JðŒšƒl-´ëHð|¸P.ò.ç=]ˆU˜’qÉ5§E~‡Âžä¼*•2êMNv¼¤ù•ôMìµÖÛé{Ù[ãóî-̸Ž0†2§»’VW}_wÚûtÝ º–˜§Ÿ_{ s#æ‚Š(«¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¦ÉÚ‚ø4ŒÛ«)Í5`Š(¬(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢ŠU84úŽœž+ZR¶€:Š(­À(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(ª'ñ=‡ƒ4 ­SÅQYXY'™4ÒtI$€d’@’|£ûFþÙ—=¶ºÐ¾‰¬4–¤×ûÊÏ…\»#%™vƒ´Sñ\iǹWa\mKÕjð¦¾9vÓìÆëY=¬®ìŸÒðç 㸚²†6‚~ôßÃó}¢µôZ¯tø¿ûSx[á-“«]Ŭê›ä„XØÌŽñH€äNAýÐ …9¹8VÚqñ¿Å/ŒÿÆ-i/|syçùż¨HmU›%QGà7± ¹'¹©ç{™žK—i$‘‹;±Ë1<’IêkØþþÇZ÷Ä=XOãëKÝE·“ËœNoy7ÈOîQÐðÀY€œn ü¡žqwx½ŒŽ]ƒ¤Õ¯ÝÂü«ûÕe³µ®œ¬—E}ÿwË8%ðÿñ˜‰§Rß­ÍéÓ{;jú»/Âoƒ:÷Æqì¼nŒ°({‹™˜¤ÀôÞÀIÎN Æ#ë߀¿²Öð>y/–áõmfE1}²Hü¡gTq%~ñ$öÈç¹ðW4‡:é¾ °ƒO²V/åÇ’]ŽfbK3`–$àØV½Ax}àþ[§‹Æ%[µæ×–û‰Ûoæ’澫—còn,ñ Äx|;tðïN_µ%ýçú'nŽû…E}}™e5Î¥4VööèÒË,®"E,ÌxI'¥sþ6øká ‘“ƺ”Q\2oŠÎ?Þ\ÏÃclc FPí„Ï…|iñ»öŒ×~2ëWââ{ —ÉO‚y#YѦMÅ^NW-€>UÀ¯W{8Që6·ÿׂêzwíûjnÞûAøHÒEnÍüNa¸–\ ®Â%Y>a´±'#v5ó”ó½ÌÏ%Ë´’HÅØ嘞I$õ5sÃ^¿ñŽ½k¦xfÖ[Ûû×òá†1ËŸä’N’@õwÀ_Ø¢ÃÁÒiºßÄyZûVˆy§Ohâ{H !$7˜Ê[!”€ g?ËølxÑ™¼EMag+rÓ¥ïËÕ¤ö»›VrosöÚجÌ£ $ÕÒÞsvÝöNÛéö]Ÿ¼#û3øëÇ:2jðõÓÙ˃“K·š¥CA+)e! SØðj‡ÄOÞ*øQemsãí"[{·1G(–9¸ÚZ6`¤Œ3†ÆpqúE~ÉWèå”ù¯Å8û¬oPŽ.½xÔ¥)òEÅ4þï$ô‹vÑ'+몶¿¤ð§a¸ª«¡N“„ãf›Mn—ºÖ­k«j=4>Øø}¡ðSÂrXMèºE¬E£pÀ:D¨ë‘ÝYYHìA¥u•ã?°ˆfÖ¾‹k¥‰SHÔ'´„  ²“Ù<Ó8ãÇR}š¿³¸/1ŽkàqqVæ¥ ù5šûÓ±üåĘGÍqTIËî½×àQE}1â…Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Uø§Lð•’Üø¯Q°Ó-ÝÄK-ÝÂBŒä3 “cYÕ« u*I(­ÛvKæË…9U’„Ûèµeú+Ãß|µrZY°T.5ß×;AÀ98ÏŠÅÐÀÒ•|ME qÕÊM$—vÝ’ùšÐ¡WQR£)=’M·è–¬Ý®ã§í¢ü ÓTj§íº½Ây–Ú|oµä\ãs¶sžHçàãÀ~0þÝzÇ‹ìÞÇá¼Þ¶g‘d»2,—SÆAUåÄ'’T³·k Þ îÒ¹i fc’IÉ&¿œøßÇÜ63ÂpäyæÕ½¬•£ýصy5®²´SW´Ñû øUZ³Ž#8|±ÿŸiûÏüRNÑ^—vëuþ7x—âõé“ƺ”²Û«ïŠÎ/ÝÚÁËclc‚@vPí—ÇcøGÁz·u¤Óü§Ýj7ƒåÂ…¶)`»œôE”bg’+ÕgïØóTøœÐj~=[­#óÁçA"™8]ŠrQ‹s/#nÐCn[x/áö‰ðïMû'‚tË=6?’˜yvô.çæsÉå‰<×Áð„¹ßÔY¶yZTéO^iÝÕšï-¢úJZ[X¦©Ïøû,áh}C,§Î:Z6P‹ìÚÝ÷K®¦p?³·ìµ§|‡íú¼‘jž ™tæ%òìÎÖ ¶äÃ!Ê–$p¼Šõz+ â?Ä}'áW….5Üy6°ü¨‹ƒ-Ä„±Æ¹ù˜àñÐI Gõ–[–e|–û 4cGM6Ûz.®R“Ý÷mŸƒc1¸î"Æ{JÍÔ­6’ýKo$Úùãö˜ý²ÂÏ6…ð†á¿¶-® W·¯m”´(ì¯,ƒ û—¶•Úx$œ¯“~Ð_µn¯ñ˜Í¦iJt¿,Å£IݨÆÓpÀpA`‹ò‚FwV¯0ÑôkÏê1Yè—7×sdG¼M,’`vª‚N'è+ù¿ÄjæyW &”½ßj®¥+ô¦•œoÒOÞ}vgì\'á”0œ¸ìêÍ­}žŽ+üogþ»·±7‰üO¨xÏ^¹ÔüQw-íýã™%šCË@FÙ|ýœ5o|ñ£lÒ-ÇÙ¼ ÞT‰$¤F®Œ9Kï㟗·ZøªËöž×¾êVzç|]«ÿei—PÏr—ºÔÛÌ‚EýÛ»±P…ä½Ðô¯ÌøËÄìæ42ÚÔeRu~ïÙNN+K6ïg¢»ÓmQöÜ7Àøž$ÁÕÆR¨£6µêһצëW¡úzRì5àÂaàÍ#Wý›ûVÊÏ+~ÿ+Ì_nì ãv3ŸJÔ¯Ópò§‰¥ÐwŒ’iù5u¾§ÆÕ§*3•9«4Ú~¨fÃFÃO¢·öHÌfÃ_‘_ðR_x›þ QûIj^øs¢Úèþ øªO êZ½íÔKr÷W˜å@e *Ó]bŽ%fÄlò2ÛßßðQ?ÛLý†gûÏoÒï<[¨ÿ¢øsH»•Çö…Æå!XÆã(þcò€á#Þ*üçý‰¾¯€¿gý2K›Xá¿×µK‰ q]óQwü¤GXÝ›cK)K°¯Î¼NãÁ4ñTµ¯7ÉMi¤šo™§Ò)7³MòÅÛšçÙð?>"ÌcNý-öOeæöòWkkÿïiß ¼§èYVÇNŒ¤fWÞîKfcêÌÌǜۢŠþÄâ*ã*νy9NMÊMêÛní·Ý½YýSF”(B4©«F)$–É-^‡¼üý‹¼I7Ä >+i?cÑ-÷Kqþiˆ")…Ù²X‚~ïʧ œ ú7áWÁ-àßö¯ü!Q\Gý­8š_6MþZŒì8EÜØÎ[æ9'Œ[ø'ÿ$gÂ?ö³ÿÑ ]zpO‡Ù' ahbptœªKßSŸ,§x$⤣-¦ËwÜþUâž,Ìóšõ(â*Z ÝqÔ_,›MŹk>‹°QEúñ¡EPEPEPEPÔäRÓPóN®¨K™QE@QEQEQEQEQEQEQE0õ4”§©¤®FER¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(§ƒ‘L§!ÁÅiNVvÔQEtQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQExßÇÿÛJøKq>•áxWXñDÇ,LY ²;+;c÷‡,¿"Ì )>.ÄYw a73¬©ÓZk»}¢–²~I={&ÏK*Ê1yÞ!a°TÜæþä»·²^o®›³ªý¡>5Û|ð×û­fÕ®?u§ZJÄ}¢LŒ± ÉTqè ¸ðÏŽ|}«üIñêž6¾–úñÔ gÀXÔg Š0rIÀd“Ô“QøËÆš§ÄϪøÂíïoî0VU\€*€tW¡~Ì_³çÅÿÛ^xŠÊu𵫹™˜Ä.ˆÎ"±–Ë 1\mPß0b¹þ6âÎ.ͼ^ÎieÙl%Jîí¾µjZêé=wPŽ‰·w/蜇 Àx—O’umyOåðBö~›9=]´Kšø]ðÅ·Éà?ͳ†t‚k¹dX¡…›®K¶ÐC˜8ägêï„Ÿ±ß…¾êúПZÕ­Y$Šâëˆà. GñÔî·!H9>¡h>Òa°ð好•¸"( Œ"&I'z’I=É&­×ô'x7“p¬aˆÅGÛâUŸ4’qŒ—òFÚ[MewtšåØü—‰|DÌs×*4²£ª´~&Ÿó?ÑYYÙßp¢£ººŠÆÖIïdŽaRòHìQ@É$žœ×È´çíw'ÄhfÐ>ý¢ÏHI˜Mz³:I©&Â¥6 mˆî|«d°ÚH^AúÎ6㜻°YÆJó•ù ¾)µÛGdºÉè¶Ý¤üáŒgâ}ŽZ+â“Ú+õo¢Z¿DÚöµv…ðRûû7È—XÖü±!µ†ED€0&ça*KŽ$ÊOÈ?~1xƒã°·~8½3ˆw {xÔ$ÊÌNÔAø Í–!W,p+œ²²›R¼ŠßNŠ[‹‹‡Åj]äbpTrI${¯ÀØŽÿÇv#Sø¥%öƒb\¬vb-—“€,wÝ Ûq¹X°…Xÿ)f¼KÅ~1ã^Mª¿³‹´"•µ©7nf´~÷_‚)»?Ý°96CáæëX™§WùÚ¼›íëkë·OŠM+žcð‹à¾¹ñ¯_’ÇÁ±Dºy—S±K{qÎÝì9b0O' GÙŸg=à]¤Ï¢´·Ú¥Üj—Ó…Þ@)÷q–¶äž™fÚ1Õø;ÁOÃý4ïX[iÖhwyq.7¶Üìyv!@,Ä““Zµýáç„™F¬Mªã?ŸìÃ¥ Ÿ–ŽOWä´?%âî>Åñ¥Béáÿ—¬¼äýz-›Ô(¢Šýtø¢Š(¢Š(¯;ý©>ÂÖøCkn÷BóKÝ©ZG~c\Kn[:¶àÌ£†*yÆÓè”W›œeT3Ì l%^H¸¿ŸùnŽÌ¿W,ÄÓÅQv”kä~c×è·Â_7Äo†º.·p-ÄÚ…ªÉ0³ÉÑÂòHÃ$‘Œ‘_~Õ> ¶ðÇ-bËAÓΧH"¸µŒ+,l¯—hóü>g˜8àTcÇÿÿø¥ö½7Pð}ò]<¶»õ+iZMѤDÆS÷0í¼c —rpGÍüŸàÞ:\Ř¼ƒ?Êž—³©NO•­/f¹í{nÞo#å¾ý¥t[¿Úã×Ã?ƒž´¼šOê¶÷WrÛÂÏ$³¼&E+» BêY BQ[8 _N|lÿ’Íâïû ^è÷¯'ýž¿å3Ÿì}ÿ¤Z¥!p~2\_â=LV7xûNTŸÃȹ#oE¯øµÜþ‡ÏpÑáî… 6Ï–÷[ó>g]½4ØýXð‡cð‡„ô½&ÖGš-.Ò+D‘À â4 Çs¶´h¢¿¯¨Ñ†œiSVŒRIvKD=Ô©*³s›»nïÕ…|¹ÿÿ‚ŸøKö$ðUå‡n´¿|JŸ÷Z \ ?³ÜÆ®'Ôt1’7J¼Û”&|±öðR> ý™¿bx×ád–°xƒJ‚Ú+9®!%»\]Ále~VtYÙÔ0+¹Wr²åOæìÇð‚êê8>$|aÖoü]ãoÚEr5FæK¹­ h•c_6R]¤ò¶)r~Uå¿Æñßa8/úæ&.S“´ ´r—kÙ¨¥»m;tMÙ?¥á^¯Å¯aIòÅk)>‹Ó«ì¿µqà_ˆ_µïÄm?â/í›®Ëz>Õ-í§„¥ŠCa¦Dè¡#Š&Ûqû¸KGµ™ÄKç3;>=cÅ^*Ó¼áë½[Å—qXéÖ1ù“M!áAÀä’H@$’(ñWŠ´ïxzïVñeÜV:uŒ~dÓHxAÐp9$’P $€$ ø×öÓý§4iú'ÃÉeŸDÓ¤ûT×/–.ç+…Ú®¡Ô"³Œœn.~\*±þQËðÙÿ9å:˜îo«Å´åÚ£ñ8ÅÚÎné+óIèåxÇOÞ1Ur¿òÙà ojÒi7ïMì›ëʵnÖŠÕ+7¯Kð×öèñÏÄO^“¦ØiK£k7ÐÛ½‚YyòÁ–˜Oþ·zÄϸ‚"!C4yPGÖµó×ìû?OàIâßù_nñ ¤bÊ4bÆÞѱ&Xƒ´™1c¨AÈ,Ê>…¯ÅLVK,Ý`²:1…*än;JIÝúò·Ë}nÓwi¦wðMÁ`>±™TrGÌ“Þ+§ß½º++#ôGàŸü‘ŸÿØÏÿD%tuÎ|ÿ’3áûÙÿè„®Ž¿¸²ùáëÜ?ô”/fï•ÿÇ/Í…Q]ÇQEQEQEQEƒRnµqŸ(gÞŒç½GEWµ`I¸zѸzÔtQíXgÞŒç½GFqGµJ("ŠÝ;€QEQEQEQEÃÔÒRž¦’¹QHŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ©È¥¦!Á§×T%Ì€(¢Š  (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (®'ãoÇmàn€·>"óg½»I œjÙ»tÚß‚¨˜¤“Û8 Fe™ár|-Ln6¢…(+ÊMÙ.Ÿ{vI-[i$ÛGV _1¯ 6S–‰-ßü2Õ½’Õèp¿¶oí7Ã=Á·RÛkÚª d™æÚÔîRÈùdfR Œ”ì5ñ½wvº'‹?jOŠw¶VŸj¿ÔgO´Î‘í,WnÌ`ÅT9fÙÆæ<ýUðŸöGð¯ÃH´Ë«ûHµ}{Mv”jyˆ÷1Fò Œ€¨*ÁåC ŸÈ¸Ì—>ñÃ8«˜aÿwƒå„§uÒ÷b¯ÍRKÞ•­®YM{·ýÿ™e~eÐÂU÷ñ2W’®ån¯KA=õê¢õ<[àìO{ãkgÔ~.&¡£X0d·µ_ÜÞ;†{«¡ØŸ|pǃÀÁo¬ôÏÃútVz ­µ•¤ ˆà·‰cŽ<’NÕPä“Ç­Y¢¿¥872¾Ã{ /Q¥Í7ñJÝû+ì—âõ?â>)Çq5oiŠ•¢¾¯†?æû·øfø³Æ:_4Iu_[iöPƒ™&|n µGVb¨$ã€køñûEèÿôØÆ¢>ß«Ü®û}=$Øî›°]ÛË^¸$|ÄÁÇÄ¿þ#êß|Yq¬xÆãκ›åD\ˆíãíŽ5ÏÊ£'Ž¤’I$’~OÄ_p<Þ •l_Xß݆š9µ×´µn)«ûü!À®$ÿh®ÝñÈ@?k|ý¼;ð8]ÉáT¹¹¼¼ùd»»ey„|~íJª…\ŒIÆIÂãùÿ‡87ˆ<]Ìešfu¥t¥Q«ifùiFÊ:yh¹¹Û×õ|㈲Ÿðk‚‚u,Ü`õï7{ýú»YY-0gÙSKø0"Õ5v—‰99‚Ԝ© ]¹ m Ê}fŠ+û'"Èp7ƒ†.¤¡N=íõmîÛêÞ§ó¶išâ³œD±XɹMþ ²]ì‚Š(¯\óŠ( Š( Š( Š( œ¿oŸ„¯ªè¶¾2Ó¤•ŸLT±»„íØ°³±I$‰$ @Ýàñ´çæø²ëÀž-ÓµIb¹ÓçY”Æû ~eÏ<2åH ‚ q_£÷ÖPêvS[jPÅqop±JÒTa‚¬§‚$z×çoÅoËðËâ.¯¡]–oìû‚‘³ZHˆ ¼ÈÈØíœWòŽü+,‹4£ÄxâªÉs5öjÇXÉ?ï%tº87×Oè ³Øæ˜*™>&ÍÁiç£Vòo~ÒK¡úàïi_4(µ/ß[_ÚJZ' cb¡¶8£€Ë•8#<ŠÕ¯?a¯‰žƒñbßðLÒi:à•¥ŽV9R&q*ÌD{N:‚3«²ëúî3äñÌ9jFN]9’OÝßF¤ŸÍ®‡äÜ_Æ3„Ræ‹JQ}y[k_;¦¿¢Š+îÏ— (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (ª)ñ >ðΣªêK+Ûé–²Ýʱ]’4.B‚@' q’>µZ°¡ T¨í›o²Z²éÂUd¡vô^¬ùƒþ â«MGÅ^ÑíX½Þ™×*¢bƒØ„’2‘œñó­t¾#Oñgâ¥â BÞ;G¿uÛlXDŠŠŠ¹=NÕ8œp0l¬¦Ôïa¶Óa–ââáÖ(¢‰ ¼®ÇUG$’@u¯ó«Ž³ÕÅ\C‹Ìhë“´tzÆ)BÛÝÆ)µ½ÏëÞÊÞG”áð•4”cyyI·)kÙ6þGß¿³‡‡¡ðÏÀŸ ÛX4®’éñÝ“!ï˜yÎ8€Ò0ÀdžµÛVg‚|9ÿwƒ4#ÎûOöU”6~nÍžo—¦í¹8ÎÜã'ëZuþƒd˜?ìü» …qåä§Û·,R·ÊÇòfeˆúÞ.µ{ßšRwïvÝŠ(¯Lâ (¢€ (¢€ (¢€ (¢€ ó?ÚçÀgÇŸ5a ‚9´uþÔsa[ÊV.œÆdÀãæÛ’kÓ(¯/;Ê©g™~#/­ðÕŒ¢ü®­–çv[ŽžY‹¥‹§ñBJ_sÛç±ù_B~ÀÿæÒ|csá=Bh–ÃTI.íPÄK›¥UÜŽ€ÄŒNî?v¸ ’ö¸ýœ#øU­Áª|?°¹_ÝCûÕ_2e°‘J©ÞìÕré·s[x ñ›k©l®[9$†T9WF*Ê}A+ø+ ,Ï®&ƒÄG÷”eï$ÚH=ìì¯GfÖŽ×WV?©ëÇÇY,•){•¥xÉmuÝ=Òz­ÏÓJ+‰øñ¢Ãã_à¿°–!©[¢G©Zªì6ócœ)$ùlClœŽ ÜÚ¿¾òÌË œaiãp“R§4šk·ù­šèò¦7[/¯<6"<³‹³Oúù§ÕjQEw FEFF IMuïYTÕÀmQXPzR…&œ*ã '­(éKEn ¢EUQEùÝñ³þK7‹¿ì5yÿ£Þ¾}økñ³Â¿ ?à®ÿ µŸˆZÕ®¤èv2éÚ…åÈd‚Îk‹KÕˆHø¡7vùüˆ–e Ä}ñ³þK7‹¿ì5yÿ£Þ¾Cý½ÿg¤×tGñ×…#º“T²òàÔm`¶C–Ád&éÜ0}ñâ4#kå$ ˆîþ&ðÃ2ÁåœuVXÉòóʬ"î’çrÑ;÷I¥ýçúcŒ°uñ¼3Ðì ä·|©koMòLýÍ¢¾\ÿ‚Q~Þ?ðÛ¿³ùÿ„ÒãÎøá.ÓÄ[,¾ÏǘÒ}žæ=¤¡óc‰·…Û‰LFˆcÝõl§sù¢IÅÙŸ ÿÁ}¾1Éá_ÙGEð‡dóuŸ‰Ü6âÅm^iní-JÍ!‰”a\\èWøs iš¨Qu§i¶ö³mÊ"Ulã óZõþ˜e”åG†§5i(E5Ù¨«£ø³0œjb«N.éÊM}ì(¢Šë8Š( Š( Š( Š( Š( Š( Š( Š(  ƒKæShªSk`æQæSh§í%Üy”y”Ú(ö’î¼Ê<Êm{IwÞee6Š=¤»€“ETQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQESÁȦSóZS•˜¢Š+ Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ŠŠúú2ÊkJh­ííÑ¥–Y\"DŠ2Y˜ð’OJùö†ý´.¯lšÃà·›—vëM ‘¼’Œém» «¨f#póP¸Wo”âÞ3Ë83ñY„õû0ZÎ^‹µÚNNÑWWwi?{ áÌo×Tp‘Ó¬ž‘«ïÙ-]–ŒîÿiOÚÒƒW­¢øRÖ-CÄ#™ŒÄ5­ª1$‰8s&Õ/ËÄŠÙ=‡ü%øâÚgÅŸð‘xê{¨ô{éÙîõ9 ‰nJ p'þ8o–H)°îþÏŸ²-ïÅ_'Å?®.¢Ó¯'ûJÛÈXÜêªrÆGȬÄܳ‚ÄmʹúÒÊÊ2Êm6­ííÑbŠ("DŠ0TpJü—*áŒÛÅLTsŽ&槀Oší̺JvW³Z]ÚROÝå¹¾÷à8ƒËò[OÕªUµìú¨ùß[+Å[[»Û3À~Ó>øfÛJð¥´VööèªÌ±¢=ê*e(i(ËcšØ¢›,« ló2¢ ,ÌÇ@îM~ý‡ÃÒÁÒ1Q„U’Z$‘ùMZµ15J¹7vÞ펯 ý¥ÿkè>Ëw |:"ãÄmI®^ ööa•³´“óJ¿!Æ ó’ לþÓ?¶Þ;†óÿ XÛh®Æ9K¨&PŽ2w9,Îe>_Ìž%xà£í2¾—”«/šjŸáûËÛ~TÕ¤~ÕÁžß“œGÍSû¬çÿÈýïtK}}6§{5Î¥4·ÒË,®]åv9,Ìy$’I'­z¿ìéû(ê£:¦»<ºO‡ãp«0Œ4·Ä0±p‡˜A¸°Á}öný‹PCk¯|g·c.åžÛI|mœ}©H䓵¼¼ñŒ>rÈ>™¯;êf<™¯&©½cIÝJWë7tâºòüOíZÖ}|câ\0|ا5£©£KÊ+füö]/º«£h–^Óc²ðõ¥­…œ9òà·‰b2I8U ’OÔšµEýaNœiEB ÑZ$´I.ˆüR”äå'vŠ(«$(¢Š(¢Š(¢Š(¢Š(¢¸ïôŸ‚mCÄ-ç]M”²²F[Éè?º£ ³ã êJ«qf9Ž)ÃTÆc*(R‚¼¤öKúÑ%«z+¶tàðu±õ¡‡ÃÁÊrvIuþº½’Õ”h/vü2’Þ$³êš’Lšt+ôi3ºS¹qfŒß7à‘ð¯Š`DUÁÎà »#Ûðw‰ëðV}W‡óEìáV\­5¬j­#ªé/‡µÜdšW¿›âKK‰2¸fØÍ(.em¥Mêôî·ï£V½­õ½Q_ÙGó¨QEQEQEQEQEQEWÊ߶çí·(ð—‚5’Þ’=i 弉"Œ; Û‘£bvà|Àdòiû[~Ó¶Þѯü3à‹«¤ñ4ÛbšX”Çý*9`ì¿320g+’Û•”ñå1xÛât! œ7•Í7%jÓ‹øuÖšiîíj‹UfàõrKö¿ x*S”3Œtl–´âÖúi6šÛ[í×2Ò×+Üÿao†šŽ·ñ2?Å.—¢ù‘<Ç™,‘2…NH ä`ëƒá•÷ïìíðd|øxºT÷l½¹˜ÝÝÈ£‰YUJÇÀ;@E<““£ó_8V|CÄÅÎ/Øáš›i¥ïÞô×wv›Ó¤^¨û?sØå9L¨Eþòµâ•¾ÏÛ$íó;º(¢¿¹Ï梊(¢Š(¢Š(¢Š(¢Š(¢Š(—ˆü7aâýçMñ5¬7¶7k²Xe\«Œä} G €F¯ˆi߀w|hòiñƒáýRG“OubÞHêa|’A\ŒNåÁÎwû®¹o‹ß 4¯Œž—Lñ,JdPïiqófÒbŒ« ˸ ÙØN{Wæ~'p8Ëd¨Á,\5§'§¬³|­^ˤ¬ôÖÿiÁ\WS†q©Ô“t%ñÇûy+î´õW]ˆ~ ükÕ¾x±u7k6öÉ؈ï#¡þë ’¯Œ©'¨,­÷Ão‹zÅ}ÏjNÏ–KRê.mrHı‚JAèqH Ÿ€þ xÿ᧌oô?¬K{§¸G1>ôpT2²ŸFVVóÈ Gàßj_üIm«øFç욦ï*_-dÙ¹J7Êà©Ê³Gzþdà³ëÏ-ÇSséZTÞ“¦Ó´ù/×{ÁÙ9-âÛoö®+à¬ÒŽ7 5Î7Œ—Ã5kÇšÝ6´•Ú]ˆý$¢¹ƒÿ¬¾1x Ï\Ñ“ÈóóöæEwµ•NÁ†@%YNq]=nàq´3,=<^|Ô攢×TÕÓÿ‡×¹üÑŠÃUÁÖ Ñ´âÚk³[…Q]Fyâœih¨TÒw¢Š*À(¢Š(¢Š(¢Šù›öúøKµ±ñvd‚O4Ûê³&òî ¨…Ø ¨UØÈXàåÐsÆ>`¯Ó;ëu;)­µ(b¸·¸FŠX¤@é*0ÁVSÁ5ð¿í-ðóáŒîçÓ´ëˆ<1yq·O¤¯(ÆH%—°]ø,Ÿ›×ò/ŽžTÁb¥Äx(Þ•F½¤T~ [ãvV啽æíï½[rÓ÷ÿ 8¶š 'ÄÊÓ‚|¿‰_á×[Ç¢W÷WEàÈôßÿÁ2~=øgâOÁKP¼ðú]Gk{j÷I×Ð’{ƒ°«G2DÅdŸ-•[ãFoÒŸÚ÷þ ñðßà§ìÞ±dí9ƒ$±:ÑJ´älH*Íú'ƒÞ&>(ÃffuWÖá³z:‘ïÙÊ?in×½üÍ|¿ˆ¢²Æá!û‰n–Ñ}»Ùôìôì{Oì—ñ§Äß´7íuâ|aÔ¿¶°ÿ‚4þÒÚÇí3ûÚÝxò TÔ|'ªOáÓ5µ¼°\EpÍ—moq[¤q\Ç€6~l೚喭ÿOøgo¥^ZÜÏ¥éš5µäqJ®ö’JæQ £˜æ‰ö¶Ùº0'§ÿƒsuË/ hŸ5/ÞZéÚvsuus*ôHº“<’;¨Š ’Ä€$×Ζ^6ÿ†³ÿ‚žxÇÆËi®é)ª^ê6WÚù>}„'ì¶"1©‘›B ª¿î÷1ß×ú¯‹±ôò܉«´iMúû®Ëæì¾gá™X¬Ö…{¹Çó_’ÔúŽ»€ æ½ñ—ÃCG³¹¼š­ÍÀŠ# †7Hø*ŒŒ“Ç"¸êú¯öøY6“¤j>,Õá‰F¨‚ÓOq).bWo8”]#?7îÛ€[ø[Ã~«ÅC†ÂBê1|òkìÆÝú»EyÉ.§ôÏçÉ2šØ‰n×,Ww--òW~‰ŸFÔu%G_èE^‡ò0QEˆQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@"–˜­ƒO®˜K™QE`QEQEQEQEQEQEQEQEQEQEQEQEQEQEõô:e”×:”Ñ[ÛÛ£K,²¸D‰d³1à$ž”__C¦YMs©M½½º4²Ë+„H‘FK3Ié_(|]ø»®~Ö^8Á¿ã•t%}ÒHÙŒ]…#3Îq”…N ¡'i ¹D_ã.2Ãð–뫉ªùiRÅ9~‘WW%vÒEü;[?­/{’Œ5©Qü1ùö_¥Ùƒñóö›ãoŽ/t½ëX›Ã¡>˦XY!Œêw W™I%‡˜7¨*IŠÙ×Ö?fïØ⇷¶~ ø’ñ^ë1¢ËPÐéÒäŲD²·ªÛˆÜB¸í¾þÎ:OÀ6si/ö–±w•ŸPx„lcÎDq®NÅà2K0É8 Ñ+â8KÃJÕñ¿ëµW'Ío Ní¥{µ.UeðBÚsKß>›?ã:t°ßÙpÃ%g-¥5k>Š×Õ·ñK­—ºQ\ÇÅŠúOÂ/ ]j>#ºµIÒ ÒÑæ -ôŠ#Œ`±Ë2ÁHPÙÃaªã*Æ…¹NNÉ-Û-|Gø¤ü*ð¥Æ±ã&Ö•pe¸ƒ¶8×?3: $Hø£ã÷í«|wÔ ‘fèö˜h4ô”È¢L`É#`onH*œ’ų>5|kÕ¾7ø±µ·“kRÊɘìã' þóÏŒ±  «Gá§ÂwâÖ¼– ²’àïUšà©ZÎWÆaXú¤('Šþ3ñÄÜÇñØÙ,eõVì£ïÕk«[òõQÿ·¥­”¢øG‚°œ)‡þÑÌšöé]¶ýÚk²é~òùGM^O‡<7âýrÛMðͬ׷×m²(b\³œdý’ON¯µ?goÙoJø9em¨ë1Ç}⃠nKŽÛwaR>_”¹Ž[muß ~ è?ti¬ülÑý¦O2yåmóNyÚ±ÑA €2ORÄõ5û7†¾a¸MÇ0Íjâº+^úû·Þ_Þ²¶Ñ]_ç\gâlù<& ð¡×_z~¶Ú?ÝÖû¾È¢Š+÷ó ¢Š(¢Š(¢Š(¢Š(¢Š(¢Š¡âØx3@ºÕøÁñsSøÇã S^’d¤&ÖÏÎi"²Bm@x„]ĸ‚p:V×íñòçã¿‹c!6šF›¾;0+ºIþ6Ú¼…“–lÏ‚_õ/þ/n„Éo*%»º–ñÇ?*äg#ø¿Äž<ÆxšC$É–I(%£«?æwµ¢µåNÉ/zVzGú;ƒx[Áø)fy“Q­kɽ¡å^o­ºû«Î÷ÀoÙëUøñ«Ü&•4VuŽßµ^J»‚言‚ì@'ÉýÏáéÞðí¶“á;e³ÓìÁ›fX±å‰'%‰É=é|#áKøfËHðÜ+oeaŠ$ ã«6,Ç,OrI<šÑ¯èŸ<7Áð›I⦗´ŸãËÑ]zͤÞÑQü‡ŒxÇÅ5íðЋ÷cøsK»}?•6—VügöÜøaŒþK­%ÿ†‘¥·Š d’H„­"í$…D-F0Iâ¾Tø'ÿ%›Â?ö³ÿÑé_¢4WÆ>Pâœò–yKì¥Nhòss¸=ùãgËhìôŠ=ñ¦G–O,iÍÊù­Ê¤µVåwÖïu«aEWìgçEPEPEPEPEó—ÅÛÙ<7¯êzgÃ.ÇT[WXàÔäºg·”€¥Ï”ª¥€;Ôø8 渟‹òžÃÇšÖäRmEYÊRi^É$ßÍÚ)´›WG³’ðþ?ˆjº8|ÍjÝÒI^Ú·o¹jõ²vgѵá?´Gí•iðëS¼Ð<ÔuXá’)ïUéÓB€ °•ÔòÊpI'p_™>!|gñGÅSñæ±s}8)(C Ø)Mÿ; ØÎ3Ž+˜¯æ¾0ú@âs2Âä4]¿âI§;uJ)8žüÒvÚÎÍ~ËÃÞÑÂT|Ö¢©o°®£9hÚò²óºÑË}}6§{5Î¥4·ÒË,®]åv9,Ìy$’I'­EEt? ü'Äïˆú>…nX Bà,¬¬$@‘†î2\߯ç¬.¶e‰§‡¢¹ªT’Š]\¤ì¾öÏÖëÖ§ƒ£*Õ¡ÛòI]ýÈõŸØSáx¯Æ·>#ñ§›c¢û!‘cÝÊqµŒj¤‘œ©xÏ¥}}YÞð­ü5e¤øn·²°ˆE(ÕŽ:±9bz’I<šÑ¯ôÃî§Á9=<jU½9%ni?Ò*Ñ^JöW?’ø³ˆgĹ„ñm5 ¢»E~¯wæìQE}¹óAEPEPEPEPEPEPEPŽþÖ_³•ÏÆ­6ÓQ𜖱êúD2ªÂñÚ‚’¥c3dmÛ‰ ‚̇•šøÏYÑ/|9©IgâK« ÈqæAqE*d2¬¡ú]\ŸÅÿ‚ú'ÆÏ¥‡Œ"‘ZÝüË{¨Yí¶1a€©P¤~âgƒtx²­L×-Ÿ&-¯z/à¨ÒI‚VVº¼]•ÒmÈýC‚üD©Ó†j è×Å »¿ñ+½·]I#á_‡|Cð¢öæãáþ§.ž÷h#˜I@FQÔ®G8lde€<œý;ðwöèÒ¼k©Ã§|D³‹@¹’5 yö€m$p¤¾âØ0‚@Ú ~¸-ÆO–ühýŠõÿjÜü9·¸×4eìUa%â’mѪß6~è8R3Бâ÷Ö3éw³[jpËosnæ9b• § Y…E‡¯­Ôô§§i·m¿›•ßE}/ùøk™å3•L$j]9u—Î+_üêÚ÷·ºÑT<3â{躧…°½O2£<0èx<‚ ©Á@ Š¿_±Ò« ðZRRŒ’i§tÓÕ4Ö5³?<©NT¤á5i-z4×FQEh@QEQEQEVg<#eãß j7ˆS}ž£C&K&G»”á”àá€=«NŠÊ½ xšr£V7Œ“M=šjÍ?TiJ¬èÎ5)»I4ÓìÖÌø+ãçìå«|Ôá7²iix[}B8Œj_1ºäìn $0† ÀÚ_M§Êd°šX‘â-•%J:ävefR;‚Aë_£_| añ3Á·ú‰–Sg¨ W1¾×B2²ŸUeR3q‚ȯψ^‡Â>>×4­=ä’ 3P¸´¤ »,r2Øg 3€+ø‡Å¿¡ÀØÊx¼½Û Uû©¶å -\nõqJÎ2nýÚ»þ—à/—áå‡Å¯ßCwm%³ìžé«[ªÞËŠ(¯ÆÑOÏß|XÔÿe/|WÐ<&ÿiºñ&‰{à‡¾Â'•n÷°äò\2iaÛWÏÞ¯”úKöøK?ÃÏ„ÒjšìE}ây#¼FIK“hº§ç•øçÇ#jôŸcï‡ß£Ô.u="ëN×u Ø®äÔôû¶‰ŠªH²Fa`Ñ1ž7g tYÏÎÙô Dµð·g¦è‘yZ| moâÞ\h¡UrÄ“€É$×ïÞ#x±‡â¾ÃeØ7%Rvu”£mcÒéÙ§/yYl—Âô?.á.«’f•±x„¹UÕ;;ï×Uuhé¿}֦·á»Ïkz~Ÿ¥Å¾ëT•!¶V`¢FgØ9<»Œšý øaáø@¾hš3¥ªK§YE ßg\Fò…cŽw>æÉ’I<šó³™ð½çÃmNŽÒïÃ:|ñjQÉ)yæØ¢•Ê³M1È=SÅzõ~·à÷‡ux>8Œ^6þÚ|‘]”\!7n·ç“ƒ½µù߈\]!tpøoáÇ™¿ñsJ+ËáJJßÌIQ×í5zšQX€QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQES•³Á¦ÑUr»%€äR×RwÔŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ý­þøßâœZ6‘ðéâmåÏöŠŽv<®[t‘Œ–ˆpP“¸ì Úüø+¤üð¢éþ_:êl=íë¨ÞHSýÕ!S8PORY›±¢¾_ Â9~9­Ÿ5)â&”S“æTâ•š¦¾Ê–ï}[µ“höëq.¶]O+V¶ÚŠ³›n÷›ëm–Û+ݤ¢¾¾‡L²šçRš+{{tie–W‘"Œ–f<$“Ò±¾!|KÑ>è/¨øÞú+8¤î–vàm.y:N&¾@ý¢jíKãJ/E…ô¯$…¼ùš÷äiˆà€Â1’X…#Äã¿r¾¡%Zjx–¯Iêû9,|Þêü©½O…ø;ÄÕS§Z7÷¦ö]Òþgä¾m‹ñSöþk R{?„úm¥Ô0Èê…ÊN»Ní°¥~b0Åü¼ñówŠ}ø'û=ëßõ1ý‡¶Òa˜Ew¨È—oò–© ÈØÇʽ7.â æ¾Ýøyð»BøW£‹/éÐY!P%”.f¸ ±IÌäldàgŠÛ²²‡M³ŠßNŠ+{{tÅj#P0TpK_Óøa—p%'R/ÚâeñTiiýØ-ycßVä÷v²_‹ñgc8¢jÜ¢¶‚{ùÉý§ÛK.Š÷l¢Š+ôÃâŠ( Š( Š( Š( Š( Š+Éþ)~ØþørðÃ¥ÜÝNž`]6hå†!»¼”6?1 79O‘œçùwPxœÊ¼iüžþ‹w¾É3Ð˲¬foWØàé9Ë˧«Ù|ÙéúαmáíîÿX”Cic ÜO!ùhŠY›$àx¯‹jïÚ1~4ëÐØx[ÍÃúS±‰˜²›éšÈxB7ÌN7m^wãWí ¯ünÔ[û~o³é1Nf´Ó¢Ç—oÆÐY° ŒüÍÝ›hPq\Ÿ†|3ã=~×K𽬷·÷¯åà c–=O'€–8I_É>%ø¹[Œï’ä°k&“v÷ê»è’Õ¨·k/ŠZ^ÚÄýóƒ8ŸÛ2̤d›JþìµmõinöZÚú2÷Ç·Å_[èþ·ó®¦ùÛ";xÁ¤‘±ò¨Èç©$ »þ|Ó> ø>ßLÐ’)nwz`Hæ½|³eÊŒ °PIÚ¸<“û8ü¶øáIayþ׬j[PIòË(;cOð®öùˆÜÄ’p0«è•û/„¾Cƒð«0ÇÆøÚ‹^ªœ_Ù_ÞjÜÏþÝZ&åù×q¬¸†»Âaeþͧ÷ßó?%öWÍëdŠ(¢¿h?8 (¢€ (¢€ (¢€ (¢€ * OTµÑ,%»Ön`´µwI4ÒãŒz³€>µóÆÛÑm¤šÃà¼N ý«p´e|¨X¹XŒ3äeHØA¾WŠxÓ(àê ¶gYE¿†+YËü1ÞÝÛ´V—jè÷r>Ì8Š¯³ÁSº[ÉéêÿM[裵Zß@Òn¯õyD–P¼óÈA"4U,ÍÏÅx¿ÄoÛ»ÂÞ°·o‡É'‰n¦r$LIi Vi#ËHÀð$påüCñ‹¬ÒßÅz汩ÛÆþbÅw{$È­‚7v 3îk¿šø—é ™c²Éh*1kYKßïöWÀ»{Ñ•ï}Ùro 0xwÏ™Uuþû±·ŸÚ}ôqùÿÅOÚcÅ¿ìå³ñ=ì0é’È’ý†Ö ¨ÀùŽ]†~l3» pQ_‚晶7:®ñXúÒ©Qõ“mÛ{+ì•ôKEÑ©à°l¶’¡…¦¡ÑI/]7}ÛÕ…Q^yÖåÿmߊ_±Äí\ø9i£IáëËV‚ìjrÜÃ}9ßû™eVY¢ r¨ãÞÊw_¹¼ûoø +ßkWZv±SEZå,rrÉ2¹ í´¨ lØÙûàW|Qÿ‚}ø³Ä ªèš¯‰tœÅæý¦’ê3‚­±äƒƒÇ*Ë•c€ÇöÎá>&àLÇŸWʧZ›û0\óŠ’jü±mÆJ:®emy[Rz~ožgù/á1]åîÔM]¿uüVZ·¥ÕöüO;àüÇ$\õ©{ŸÌ½èyj¶ùÙ¾‡îuñ/ìÃÿÕøWñÖûMÒ>"éúÏ€üE!‹Ê¸Åí|ɱc¹ŒbÊ‘Œ4)óÈwcu}«e}§e Λ4W÷²Å,N%F ¬8 ‚#­}„1tjU•ÉsÆÍ®¶{;uOUu¥ÓWºi|ÕL5ZPU%uìú]n¯ßË{4öh–Š(®ƒ¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(®kâGÂ?üWÒ&´ñžo;ÉŠ;¥EÕ·9)pJpqÐô ‚Aéh®\f ˜Ñ–MNœ•œd“My¦o‡ÄÕÁÔjqšÕ4ìÑò?ÄÿØ\Ñ®åŸámÔ:ÅŽ¥½Ä« Ø$àŒ#`ì®yxçÂõíûÂú¼Ö#´¸±½· Kñ”tÈdPA¸ ×éer?>ø_âÕ¬‹ã.ºuÂÞÂWQ¬ªDƒ–q![rä ©Å?q€teˆÈ'ìªÞü’mÓ{蜢ïæÕ´²Üýc‡¼WÅa¤©f±öþh«Mzê“ü›?=k{Â_|Gà1ø;\Õ4袘\a¹eœc–‹;;@! ŠõßþÀ#ЭŒÞÔ¬µðª Bëö9Ù‹c™ç,ëÐàŒù7Ž¾x—᤬¾8ÑoôôWX„ïëwv]ÁVeÊ1À<=¡ÇóžeÂ|KÁÕ=¶' V‹ÛùWýăq¾›s\ýŸdÜCgF´*_ì»_ÿ•Ÿ^ÇÐ_³¿í³6µ©Ùè_Bî¤hâÕ—ljYˆØ“FªFK.õÀû¹_¼õô¥~SüNø£ü"ð…Æ·ã["Ò•@2ÜÈAÛK‘¹Î$…Ž£á×í?®|QðVŸ¯øWÅ~(kF2ñ‰oçGB«+ ýU•”àqHÁ?±ðo¶I”{\ï SIIBv×–î“^ô¬®›¼ž·lü÷ˆ¼7ÀæXþL¶´iMÇšT÷ÒÿKe­­¶ÖHý0¢¾7øCûpëÌxf\ب^›vŒÓM=/êº4¶vºW6(¢Šû³åŠ( Š( ¾ý²þIàO7×vöþV¯ÿ§Ûºïei<àY†7y›œ¨$‘:d÷y/íð—Pø«ðÎÜx>Î ÍSJ»*§g„£+¤LGRLlW#>_rËãÃÜñd™¼I%N¢ä•Ý’¾©öÑ¥«èÙñ -í”ÚmäÖúŒRÛÜ[¹ŽX¤R‚¬§Aj*þ ”\[Œ–¨þ§MI]l뱧ÇñßÆ›+»‹7OÐ?ÓçvÞª$ê@eãw™µÂ’¿\|ž¾˜ÿ‚vkÖ‘ÝxŸLu…/¦X.‘†ÿ2hÔº°<í ŒéŒI”äß¾ð·-ÃæÜS€ÃâšPçæ׫‚sŒ{k(¤ÓÝi»Hù^8ÆVÀdxª´½Ëm:)5þI·å¹õQ_èIü”9Ÿ=)µÏRJ[QE˜Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@ ­ƒO¨éÈsÁ­iË ¢Š+p (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¦Ï:[Bò\ºÇjYÝŽ@ä’OAI´•Ø%}êó?ß´ÖðSKxmÞßTñ‘éé0Ì9·ÏŒ˜×k)òÙã,¾GûEþÚ‘øJmàÌ÷¶ñÊÄ\ê%Þ\+FÈm\2ƒµÃoPH8îkæúþnñÇJX<»‡m9ÛZ×¼bßò+ZM3|©ô‘û'øa%ãåG7Î1|´ê$Ô§)N£ƒW‹ŠÙ'}škWËßôLËò^u2ü¾…åÓQJ0潚o{«jÔ]ö¿o˜> þÄZï`ŠÿÇòIáÝ8¹D·Û¥ ÀÝ°0Fð²r ì*A?Pü8ø9áÏ…VÃàÝ.Ö ã‡É{ÆZêpH-æKÇ,ÇÝOEFð†¹'AK Kšµµ©-dý:E_¤Róosñþ!ã<ψäÕzœ´úB:GçÖ_?•‚Š(¯¿>P(¢Š(¢Š(¢Š(®OÄÿüàèîOˆAê6ñƒœ`×–xóþ  hZ’Áà=*ë_‰sæ\<¦Î3‘å†Fvä°;•0WÀæ¾K9ãÎáø¹c±Ôâÿ•>ivø!Í+i½¬ºžþ]ÂÙ¶lí…ÃI®írÇÿ•£ò¹ïõå?j½àõ­öŸ¥ºê(„*ÇdQÄp—]Êò¾ÚjÇ*8²üáñöÄñ·Ä+;xò- ;w2¤4¶Ï1Æ÷2Àç€@9Éc}}>©y-Î¥4·±y%•Ë¼Œz–cÉ'Ô×áÚé-šYvs€ÒlUä•VCƒ´ØÏ]|&øáïƒzZÃáK(ÍáB³_Ê¡®gÎÜ‚øÈRUNÁ…ÈÎ3’{ +úo‚<-Éø*«N>×k:’Zîß»µÚke«gâÜMLJIÂRä£} Ÿæ윻ë§dQE~”|aÇ|Gøá?Šv÷ð•hö¿l¸ùô!» ¢·š£-€F÷/ àWοÿà›Wzåµý®•y¢ëzL­¹,µHy•Aܪãk#0!pHËÛëº+ḇ̃‰ª*ø¬?-TïÏÉ;÷n;»ê›½™ôùGf¹,*¯MéË/z6íg·m,~#þÒðKƒáoÜÚø-ßÃZ´Ké—ò™íH!0Ru,ê1½‰ýîX€6Ç àÚ«ö‘ýÛOŠË^ñ Ž‡†(tíJEÕt-]˜[ bñÀX#äBÑÉ·$Á¯ÝŸü8Ð~ Û¼k¤iúòÚ%yáV’%n¾\Ÿyºr¯Ÿ>:Á4¼?ñÃö>¹Í¾¡¹fÓµBd¶ugRª’ ó"Ø2UŽöÈC¹HÝ_.²þ0áJ–§(æDö›¶&0ºvRvŒä·ÕóI¥n]÷þ¿Ãùì?y…®ÖñÖ‹—œuqOm4Ië}O–ÿfø8•//¤²ý®¼¬rÈL:§…‘ŠB U%´¸•˜ûçiRRxE–¯¾¾þ׿ iËHäøãmÄ3É—&Æ++P†$“Êi%³“lñ&ò£s¢ƒ½Èe'òãŸüCľÔï4'¾Ñ-דiiªXJ-Q³óF· °pÈT…bBŽO/_;üIøãoÙ“Äv©¨lžÚé%Óµ2å‚Çp›dVŽEÛ$R)RBœÆÅr5õy7äÙÕxáib9kµü*‘tæ®×,’m«ëÊä´vm&Ï1á$k6XE–Á–×HiÆ9#7–l"³¬‘C,n6¾ +ùÅûJÿÁW¾7~Ô:~¡¦x»Å_Ø^Ô°%Ñ´~ÃlÈa0¼O &âX¤åâ–WF-÷pª\Ò6Hý¶ø£ûa|*ø+wªÚüTø‹àÍQÑ 7šuƯö„Kå‰@ŒÎì…YQP³î] ägå/ÚGþ ÅðKÃÚΛðßFÕ~)M<J±=™°Ò¤o;æŠi.ÓÍWE_0·‘I(7¸§ä?Ãß…~"ø­ª5ŸÃí&ïS–ÒÌ‹L€FÒ`:BKI& 6ÝÎÎÛr@ÝÀ«¢hV>Òâ±ð妟e|»{hV(£É,v¢€I'ŽäÖ…í”ÚmìÖÚ”2ÛÜ[»G,R!GÔà«)äAþGãß'Æ4iá0ØeC Nr’Œ~Ó–ÎVI)%Í·ó=÷?záŽÔ•zÕJÓŠW}Ý+¶í{oÙ –&‚VI•‘Еea‚¤uW¼~Ê¿µ{ü=‹Oð§ŒàŠM[¡½ç˜±8HÄ·™œ{$T9aµGägÆÚÇVñÿÆý;Å~XÛø~Aý™o,Îw(b]¦ Àn”®Œ¦–ÆãöM‡íà]FÆ‹øicž5‘DºŒQ8dnG`ÊyåX: {3áŽ%ðª¾6Áó:•bùãórꛥ;s&š³¾žò|ºÅHó¿µò~8¥‰Àbl¡¹[v¾)ÆöjÎúk£WÒM¨?ð»<ÿCw†?ðiÿ[zˆl?ý¶¾øf†=N]ræ9-*18PWvá+‰€à®H'àã¥ø/ñûÃß4»›Év²ØíûUµÌ9m÷— ¸Œ£n±ùY°1œ+õxÉÄÙ~ûK0ÈšÃéï^pÞÖm¸ÊÑ{&Õ›jÏU†^äغÿTÂæiÕíhËmÕ”–«µî¬ô?[h¯¾þÓž2ø_,k¤j²ßYF‹²Ô®-ÕJ¢ 't`g¢2çjç b¾™ýŸ?k/âû7Ä鋯(P‘s ñÀÄ[£nÏîÉ'`·Í·íxCÆ\‹Šêà &èW–Š3پњÑߢ|­½z_æxƒÃ¬Ó"„«¤ªR[Ê;¥çýmt–­ïo`¢Š+õ³àBŠ( ý°¿f¹>"ØÿÂIà;TmnÊ2.­â÷º’ ¡H;€.Ša…ÏÊ¢¾>¯ÓŠù³öÚýžc¹ÓOŒ< e¶{|RÞÚUxÉwk¦Û‚X1ߘCb™ügð±bÕn#Ê£ï¥z°Kâ¶óf–³_jÜßù¿hðçŽ]O'Ç?u»BO§h¿+éÛmk|µ]×ìãñWþÿÅ{ Níöé׳¿ã8ÈËp¬ß#“ 2vmÏ&¸Z+ù{*Ìñ62Ž? +T§%(¿4ï¯tökªº?nÇ`©f8z˜Zêðšiú?׳èÏÓ‚q֘Ϟ•àß²§íMkã&ÇÃ.æMzÝ G{w4a/ÆõXÐ1!ŒÇx]¸bÛ $‘^ñ_è—ñN ‹00Ç`'xÉ+®±vÖ2ó[v{¦Ñüd¸¬‡,.*6kgÒK£^Oïï¨QEížHQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEP P™ÍÔnÔêê‹æW¢Š*€(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢¹Ÿ‰¼=ð§IžçÆZ•¼E¢È¦êägEAlž3ÐrI9q¸Ü>]FXŒUE qWr“I%êÍðØj¸Ê‘£BR{$®Ù©âÏi~Ñ%Ô|_}o§ÙB i&ld€NÕYˆ  “Øùö¡ýªn>,^É£xY­|3á˜e$Ô˜Ž¢?î¡ëÃ78Tæ Õ¼{¬¦ŸàÝ>ëQ»|~î-° ¹ÏDPYrÌ@äŠûgá/ì¹áo‡6šeåÎgq¯ÛÛD³Ü»½ÄbáB³IÉÂëÁA¦2Ew>ð†—à}!,\„ pæÇ,Ä(Ë“ŽMiWèx;€á âqÎ8ŒCµ›‚å…¿’÷wþöM>S‹|CÅq¨áS¥I_i;ÊÿÍk+ykêŠ(¯ÙO΂Š( Š( Š( Š( Š( Š( Š( ;ãÿ.~ü×u­syk¤ ‘û©$u‰dÁ¥Ãm#n;æ¿>+êÿø(Ä3¦xkJðݧإ:£5ÍÈc™íÖ6O-”ò†&A’v;׉þË·ñ÷Ç-Ë[±ký>#%ÍÌe Æ#fS Æ6<µ ðw=q_Çž3âjq_ar,4þ v jT•äô_Êáͽ¹|™ý áÍd\=_4¯‹š~nVKïR·©öÇÃÿÂðëDÑ-R]:Ê(fû8Äo(Qæ8àgsîl $“É­Ú(¯ë¼&ž …<=%hÁ(¯D¬¿ùþ½iâjʵGïI¶ý[» (¢º ‚Š( Š( Šço¾/xOK½šÛSñG‡mîmÜÇ,RêP£ÆÀà«)l‚ ƒ^wñ+öáðƒb’/ ´¾$¿GhÌvùŠÝJ°t̸ ‚ÅZ0àíê¾o5ã “$¥*ØÜe8¥ýäÞUäß’MžÆ‡³<Ê¢§†ÃÎMù4¾mÙ%æÚG«ø§Ä0øKÃŽ«¨¬²[é–²Ýʱ€]–4.B‚@Îã$WÇÿ¿moøøÝXø/v¥ý <A#Ç~È£yð7˜ªŽ8›¶‡ÄïÛ»Tñφ.ô@±Ómõi­nÞ{†¹vIoîÈ€[¨lätÇ> _̾+ø»ý±*x.ÅÉP³çj.MôNIJÖz«-{è~ÕÀœýž¥‰Î0ëÚßÝM©Yw²n7¾ÛèQ_Y|ýŠ4&ð,ß-f¿Ôõ8ÒámüÙmÅ‚ÈB«9ù·pÚ·äœÁŸb¥…Ë”W*¼¥&Ôbº]¤ÝÛÑ$Ÿ}“k︋‰p\3B5ñûÎÉE]¾ö»KN·kï²>R²²›S½†ÛM†[‹‹‡X¢Š$.ò»U’IÖ½'þßâGý ŸùPµÿãµ÷Uý–ý2ÊJ_Úê•–äŒiÛ½ù½­ü­kyŸc6hºG‹­¾×§]ùþl^cÇ¿l:üÈC 2©àö¯,ÿƒ’uË+oàþ›å«ê6j÷3Ú¬ªg†)ZÉc‘Ó;• ˜+†18Úqôì£[jŸežú=òéºd×6çq\…㈜ÏÉ+ŒŽsÔ ðx«Ã\Ÿ‡8ƒ‡ðy};Æ­Wí=£æSŒ%I´ÓVÕ9+Y'{=_$ã,Ã7ÉóLN.vp‡»Ê¹y\”ÕÓßF“½î­¡õ¯|£ü4ÐLð=ŒZ}’»JQ 1wn¬ÎijÉ'€@Ø¢Šþ¦Ãa¨àéF…(B*Ê1I$–É%¢^Hü.µj˜‰Ê­Y9IêÛwm÷mêŠ(­ŒÂŠ( Š( Š( Š( ¯¬¡Ô즶ԡŠâÞá)b•¤¨ÃYOH õ¯"ñ7ì1à=qbþɃSÑŒA²m.ÙĹÆ7yÞgLmÛÔç‹öhÿ¢“ÿ–ö«ÿȵù-ÿâñÇý|+ÿ7übøwŽ?è+á_ü ¸ÿãÄRáoúSûßùú‰œÿÐ,ÿõ§þEû4ÑIÿË{UÿäZÀø‰ÿÒýž|¢Euá­_ñ|òN"k=#Dž)âR¬L¬oDìB¶]p¤dÌO ÿÁ6¼Gu|ëâÿè–6Â2VK4–íËä`uˆÇ;@1ÎGmà¯ø&þ¤ß|u¯jÌi"T– pNx5íŸ `cZŠ;Ÿ‰ÚŒZ*oRlíÂÜ\2!Ã8;#$T3ïr0~´¢½Þú>ä™l•Lʬ±]?‡-"Ü´ÿŸUÐñ³osÜgø|ݹïŒñÒ¾ý¯ÿoë(m>5ÙÉk Q½Î‘ ³2 W̘¼v¢.Oe°®² &…è¼%(ÓTªÆÊ*ÊÎ.-Yiüº½mø_šâ+ç•"nn¤%vÝÝÓR½ßÏï?¾:h_ðŒügñUŠYÿgÅ«sä[ˆ|¥Ž#4{P©\q´ŒqŠû+öÿ‚3hŸ¶gÀï|@Ô>)}ŠÖêú[}_D°Ò£šæËɸeh ɸ")^ª^´N±Æ7Qý¾þxo^øu⤼_¢Þ[‘¬}ªV3X–1?³îò‡ïnÞnÒÿ&·¬àÝßù2¯Øíwÿ¤}~áÏPâÌ’†*“÷¢”&µÒqŠº»Jú4µÛLùŽ0Ê+dY…Z2Ù¾h½5‹nÚ&íªk[=67¾ÿÁ~x+[–ëÄ©ã?Á$%³Õõq1e"U6QA&ð¨ÊáÛ*Nø Å_ |Sÿý§/¬>0øwþo x‚Æxtû˜o%Ó Ö"òî`‘w¬7P9@ñJ“8Ã,±Ìtë–øÃðKÂ_´‚§ðïÆŸi~$ѧÜ~Ï{“ÉsÇæÂÿzBHáeŒ«®âU¯±ÇeølÊ„°Øªjt壌’iú§÷úŸ3„Ç×ÁUU©MÆKfšôhüÄðÏí ð÷ÇÞ[xÆ:]Ù›ìê-o3§Þ¤³ïÙ‚ly²¯–Cý¦Y” z–ëc‘¢‘^&*Êr}A­Ú·þ ìÑü[­Þë²/‰-|*+‡utžâÌJ,Ž·Ûäš4)ó„hæ>`a¹U€å?Ùÿã‰ÿgÿ‰Òüý¦tíCMÕ4Û¥ÓmEÂîšÆS W"H2åR˵”‚Ѳ²*x…àoöN¦eÊRŒu•'«Q[¸KwoåzÚömèÿuá/ÖcV8<Í%'¢šÑ7ÑIl¯Ýi~‰j~Ì~Ï?¬>.ü9²¸±ºóu+"ƒR…ø–)‚à±’UÈfVÉÈÈûÁ€îëó¿àçÄ»¿…?tÝZÆæî h§A}&æßpó#*Ä+esŒô8 ‚¡vWÐêvPÜé³Eqop‹,RÄáÒTaÊ ‚:×ëþq÷úë–:x…lEXÏ´“^ì×]lùº_U£²üë¸Wý[Æ©Ñw£Vî>O¬~WÓËÒîZ(¢¿X> (¢Šù·ã_ì%'ˆüXڇ 3Mµ»ËÜY]3Çg¬;°­’v‘Ç*ùÅßÙ[Å?´õM|X_X3ùrÍc#È-IÀ_02)‰Àa‘ž ¨?wÔWÖPêvS[jPÅqop±JÒTa‚¬§‚$z×ã!ðË2Ê«…ýõ5ÙZKþÝ»½»¦ûÙõEUѵ»/i±Þx~îÖþÎlùsÛʲÆø$2’#ê Z¯ÙaR5b§tõMjš}QùÄ£(IÆJÍQTHQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEƒRu¨éÈkZR³°¢Š+p (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€<¿âßÇOF³K‚Zß‹/¦Tu½¶„Üé°Çr¼‘6L€(ù2¸­žÇ¬d?ˆßSŠ­«*õæå9jÛÕ°¢Š+¨À(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š+'ÇÞ!“Â>ÖµkTI%Ó,'»D|íclàÆWÖµ«ç/ø(Ĥ²Ð4¿ XI*Ü^¸Ô.Â;(òWrÆŒ1‡ û›ùL ‘È#å8߈©ð®I‰ÌfõŒm{79i´}]övI½“=ÞÊ'žætpqZ7wå«onž{è|ÇâßøÏÄz§‰î¤¼¿½É4®ycÐ:€})ÿ÷ðövÚ爵M>H£ºXí,.Ÿ#Í@Ìf 3ÊîX¾lc*@9 „|øwÅŠºF©\Kko|îe’%Â$m! ž!çÎ0A´Ý: N·´ÒâH-­cXa‰4P¨€ þmð'„*çY”ø›;Æ”¤•ìÜêJ>ó•îýÕ5+îäÓOF~ÉâSËpqÉhF΢MÛE'¢VÓW­²Iéª&¢Š+úôþ~ ( œu ¼Ûö‡ý¢ôï‚qjÖ÷Þ ¸mlLƒäÈ8–`áDZcÀ#–_1ý¢ÿmai(Ò¾^J·V—Gí:˜Š)-å@¤lˆ8mà³rø_õcnàÙ¯š5½fçÄzÕ桬Éç^_Î÷É´.ù‹1À ’N¿|Eñà –Æ®[>zû:º8G¿&üÒ^œ©õ–ÇëÜ!álk§ŒÍW--Ô5æ—ø¿•?[¿-͉ÿõ‹Þ&:¯fŠ[…O&%Š% ‹{8@ä€]°X–õ&¹ã#Â%T’xã'ô•{/ìéû#_üW”ê>;[ýÃê¢`ž]ÅñeL[Ô1‚sÉNSã‡ìÿ­|Ö„ZÏún70_à ‚Ë8Tv#j˶2ÅlÔõ¯ç\Ó…¸Š¦ýbÇÒ›§U»ÎNò{{ÒOÞ³èÞŽÝ­×ð9æQ Wö>¤Tà•¢¶ôOk®©wï{|éâÚãÂ~ ø¸þñQÔ,nâ’8ä¾–4Q™#Y³ïÜ£çU,Wœ’¯{ð×À¿x¾[Q øgY–;ÔA;Û46ò)]Á¼ç#K`äc¨¯š¿kÙR×ã‡q«ø&ÒÒE±¼ÒÅ?´#Eaä±Ü9q#~ERBò½÷üŸþ ­}ð#T°ø9ûgOw§èpyvº±©«E.<»[Âø"Ô‚¾\­þ¤­ûœ?Rà ¸_°”ka±U)Õ„R­OÝrrÒóïhË[;5²²i§ñ|MÅù× V© ”#8I·NZÙ-mk^KK«§÷«}CàßØÅšßÙ¥ñuA.ï6=æææd/ÈŸ»l§‰8 ž£mzŸÃØSÃ^ ½~5¹—Ä·¾è£–/"ÕpT‚Ñbç*À†b„6 qš÷ +÷“ÁÎÉ% ‘ÂûYÇiT“—mã¤<þ Þ-ùneâ&{™)A×ä‹é£øë//‹b®¢YxsMŽÏÃÖ–¶pçË‚Þ%Š4É$áT2I?RjÕWéÔéÆ”T ¬–‰-K¢>*R”äå'vŠ(«$ü(ýªuX?iŸø+Ï‹ÿ´ìⱶƒÄϧKo..Òå4¨E»n(ÄÂÄ’¤žn>}¹oÑOø'×ü–mOþÀ²ÿèø+à/ØAyÿJøÛ%Ü1K%®½â ag@Æ:¦ÂÊOÝ;]×#³Кý<ÿ‚{øfÁ>êúȵ‹ûR]AìÉq Ç„û£s’qŒás£Ï|Aíó¿rü,%haéóÙ÷÷œ­n­r­{_×õìµÓËx/ZJò«.]>I_ÓÞ3èJ(¢¿¡ÈBŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ¾Yÿ‚ŠèÖÐk^Ô"—P\ÛË&ãóGFÈ1œ ¤9'w=>¦¯˜ÿà£ó&ÿÛïþÛ×å^5ÓŒø7ä®×³kÉûX+®Ú6½>ëÃiJUûhX½ßì·âù#*¬v’¾zo­Ó|¸ýkÐàÛù­÷ÿÜ|ÇÑâ.<;]¾µçÿ¦éŸAâÛO4§n”ãÿ¥ÌýF¢Š+÷ÃòP¯Ê¯ø/÷ì£gà/xwã€'—OÔõÍF-#ZÅܛ于ßuÄ ·÷dEi"9£÷pMÆF?ªµáÿðQ/ÙJËöÂý•ÆS ígzrÖ÷[Áí¯ÚKF•Ú?Q¢ò8£ícVÛÝZqõÚKò}.}ÍðÓö¯ðoÄ뇷±¾“J¼]Åmõ%XE~e`ÅÞû»·|¬qšôKø5KUŸMž+ˆ;d‰Ã£`ààŽ ŠüЭxÿXøk¯.§à‹él/U2胣uVVXp<€z€GéyÒG’žo„SWÖp|®Ýù]Ók]9¢Ÿ‘ñ9§„*sO/Ä8¾‘’º¿nef—ÊMyŸ£”WÊÞÿ‚‡ê–ÖŽ¾+ðÕ…ìåò¯itö¨‚®²sžr:Ž8Éè¼ûÙx‡Åz}‡ˆ<=ý•gy2Ã%áÔ–E¶ÜpƒFƒh$n;†O8Áý{ã c%GË)4’”*+6ì®ùyW­ìº³óÌG‡k5ü§»|Sý¿¯5k)­>iÒée je$˜ °a‘In2Yþ\ð |Äþ'¿ñž¿uªx¢ê[Ûû×ó&šC˃ÀŒ«>ø{®|A½0x+I¿ÔÝ]#s%’ä…óîÆÌÄ Ï¾¡ø]ûhšiþ(Ý~Ya@-aó-¡·~®w«‡“žù2J俆aðkã~z“r¢žò|”cñZÉ-^ñ¼c)íÌí©úulW øyO–J£è½ê’ÛvÞ‹gfãÈÀÿ‚{|=;Yñ\Ó/—ƒ¤ÅêOîåvn8ê€Áç-03ôõUÑ´K/i±ÙøzÒÖÂÎùp[ıF™$œ*€I'êMZ¯ëÎáxpvOC+Œ¹¥¹K]dÝÛÖút^Hþâlò\E˜ÕÆÉY=ì–‰~¯Ì(¢ŠúÃÁ ø³ö£ý¨çø¹&àÙe·ðÄ2pQõ&Sî!åPúaœöŸÛw⦥ðóáõŸ…g–ÎëZ¸*×PÊñMGµÏ–ÊA‰PN~îáŽr>:Ñ´‹jö¶D~uÝìÉo{‚ïw`ª2H$’q_Ëž:ñæ%W\1—6®—µ¶òæIÆšëfä—Åt¶æOöÿ øZ‹¤ó¼bOWìï²å¿4Ÿô]¬ßf£²²›S½†ÛM†[‹‹‡X¢Š$.ò»U’IÖ¾¹ý–ÿe[ ø^âÿâΛ¤êwzºA4·VbV°M…Š°•I r@ã`äöì~þͺ7Á;îíÄ×:ôöæ«·µŠ3F±ƒ´(d^y<pq^“^ÿ†> C œ3LîÓ¯gËOxÂêÞ÷IJÍ«ZÉìÛÔò¸×Äif±–-¼i_Yìåg}:¥uêúÛ`ª'ð͇Œô ­/ű^Ø^§—42uŽA`‚ _¢¿ *Ò…xJ•X©FI¦šºièÓOtú£òŠu%JJpv’Õ5£MuGÇß´7ìc?Ã=MoáÝÅÞ«¥Û ×pL ÜÚ 2å@&s» ŒFæ_‡¿jOÙ×ã†Ý_ÁòZiž'jI,¹X/£”¨$:ºà´äm)ûE^3ñ»ö3Ðþ&ÞêZdž¦—H×®Óx GØç›9/*, ô,¤sómcßÏœIáF;‡±ë>à™û:‹zMèÕÔ\š–žä´¾±’²Kõ¬—pÙ®åœIx=ªu]®—U¯¼µ¶;¶ÿ1àžŸðQŸÿÁ8¼^~þÔ6¥Ï€.¿Ò" Ïq É!-çZÛe·wÝæD­€û0âD›öáïÄM âǃlÜ2œgn×PWäýYø³ÿÆøcâ?…Ú쫦]]+?’ò/Yùmõ l€ÄÇ$„I]ìÑHK/Öð/‰Ô8Š§öveûŒ|oRi«Ê;¸7ó¼rVÒé]ø¼QÁ52È}sý¦Ù©¦ž¤­øI+;÷v_Ðð_Áø87á'Ž?³m~0èž)ð5õ×›ö»)u=2ËnòŸ¾‡o ƒ‹nð~P^½ö“ÿ‚Èüøð·M×¼¯YüCÔõ茚v“¢]ÆfOÝ |[æ²\´hVDóAcˆ›d›?XrI\øe;ÚÇÕôWä7ÄOø8Ï⧭Å'Âo‡þ Ñ4áY Õæ¹Õ'yw6]e‰í•P©@¡ «Ç /{û-ÁñjÚÚ鿶…mt¸.g µá¸äh-š%{9]ä( Í#KŽØ « ’s¡ºJö>`½ø¥¡|$ÿ‚”|oÔ¾!_gÙOâzÙ$òd—t‡TÜ«ˆÇ8ÇõŸì½ûp¼67’|ñ ¶£§Ar­{e5³2À,’*È¡€Æå*f2v=sâwü+öwý²õñçõ RßþëÍFïWð¯ˆîÛZ¸–berÓ‹ˆ†ÙVQˆvYÁ/ÎÿàÜ›û;ûËŸ„¿b¼±ß4°Yj:N˨ã1Eç,Â9¤ÇÊ\¬*HÎ/ä¼eáÅ<Ó,ÿ ­‡Å%¥M9¤–Ÿœ®´iJÝÓWGßð÷¼9U|=:´[m©¾]_÷¤ÜUž×Wó¹öï„ÿoÿ êÉmŠôý[H¸•ˆ™•VæÞ“ƒ½HvÆqA8ç>ƒáïÚ?Àž&²k7ÅZ<¶¸R¥¿çå?ÌýÂÔž}ðËò>'ý¼|d¾ýœµ+'µûAñÕ¶œ$ó6}”¬¢ç~0wäZ”ÆGßÎxÁú#þ ×ø3{á_^:ñƪn¡ƒÆ:¤pKhѤ‘X¤™¸ŠRq*4—’ÅòŒ+[8ÜNBü«ÿ ¿‚?ƒÚ5¬“D·3k),qÝ ƒ0^¤è í¹}E~Áå_ ¿î)ÿ§[ÊËÀJ*Ÿ ÆI[š¤ß®Ñ¿áo‘Óâ¥W,âQ¾ÑŠôÝþ§ÔtQE~Ö~b`|Qø£áÿ‚¿õ_|TÕmtOè‹ËË‚vD¹,îÌUUveUˆñãöùÿ‚¬øÇöíñm—¿eˆ|Iá¿Ý¡ìÖQo©x䈬«xbª[*4‹äï(Wt’“ò¬>ÿ~ÓÐüBñÿ…~ ü<ŸP¼¼ðí×ö–»f–@¤—³C±Ž'#Ìyç%Sä?iA–u"<ÏÙóàvð+ÀVú~Ÿ GT¹$Ôî•·›™öó† -I`‹É™‰üÏį(ðPö•ê]B7²ÓyK¯*ºÑk'¥Ö­}Ïð„øž»æ—-8ÙÉõ×d¼Þº½×]Ÿš~Íÿ± ÿ¼k§x›ÅúäWZŒs³³„ù*î­&gÃ8òÙŽ6/ÌG$/Íô-zï¯ØÏÅ?´ -^I¬46õÁSt_í½HÂà‚ *—v3ÂÇ_ö¾ýœôoƒÚ^©xn"·º•í®cšc'ÏÈË‘ÇÁç²àu5ü³Ä™qV¯ç0ýÝ8ÆÎIE¸JNÜ‘KXÇšíË[;ÞZŸ·e®A’âiäy|½ù9^ÍÉ)%¯3owkYuV²=öøËi­x5|©¿—©éFImFÓ‹‹vmÇæÉ•†8ùJã8b>€¯Î…¿o~xïO×4BåìåXƒ1óÄIË‘œÈú'e}§e Λ4W÷²Å,N%F ¬8 ‚#­CxƈrW—×½ÃrÇÖ›^ãójÎ/É+êÏȼNáõ”æ_[¤¿wZòôýkÕö%¢Š+öÃóP¢Š(¢Š(¢Š(¯;ø¹û0xWã Ü×Úå´¶z¼±¬nµ}ŽvýÒêr¯ÇË’3´‘D¢¼Ü×'Àç˜w…Ì(Æ¥7ÒI5~ë³WvjÍtgf0Äå•U|%G ®©Ûäû¯'¡ùýñÓàn¥ð+ÅXëSGymyMiul©*†+´î²‚ؼœ×:ó_¤¼ÄêÚ%ù‰SSµxÉ”Bä|’l$d£maÈåFë_|uý”5Ï‚–?ÚBxµÌ›¨cdx hÎvĨ!˜d Xüsâw„X®«<~UNSÁÚï«¥äýç)G¯5¬–’Úïú#‚¼@¡ÓŽ5FË´üÖ‰'Ó–úî»/6µÑ®o´Û»»HüÈ,vÊ°-±Ú®wmÝ…-¡™ ºƒV¬éšÅÞ‹,¤ÜÍlÓFÐÈcr¾dl0ÈØûÊGOu¯Qñ÷ì·ªØ|7ÒüaàÈã¼Ò®´»K›«h·›‹ghWÍ“iÎôÜ7’WyùB©jü³Ⳝ=ZØ R›£j‰khÝûÉod¬¤µk}S|¿q‹ÍheÕ¡O5QÚ éw§ºüßNmí~¿ö,øÿ}Š¾9ÕÕ´ÉíÄz`ºÉxçVP¤•”° ÜeQWᾫ¯ÌºúWö\ý¯/îµÛ_ üX¸’õo^;{ ö\ˇ±Í— qûÖ NâAÊ~÷áŠÔ°´éäYÄÞ®Ôê7t¯´$ÛºIé²NÎÉ\üŸÄ.©ˆœó\º+kÎ Yé¼£mݵ’ßK«·céú(¢¿©Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ’£§#v5­9YØQE¸Q@Q@Q@Q@Q@!ëKAæ“¢Š)€QEQEQEQEQEQEQEQEQEUMo]²ðÖ—-÷ˆ®í¬làÉ=Ä‚8Ó$–c’@¤\ßÆZ?Á ÿhx™Ì³ÎJZYÆG›vãÆz(È,Ç‘Ô•SñGÅïŽÞ!øѪ™¼Wtc³R¦–Ö €Á 9™²ç'æ# `ÊüBñW.àdðÑ^×ÕÔÒ7Z9¾‹ª_]“¹÷<%À¸¾'~Ù¾J ë'׺Šëë²ózEüSý¼ôÆ`øcnÚýÓÆOÚ$´ A!”<˜ £h ðùÎ>Vñ—Œõ/ˆ$¹Õü]sö½Bïo›/–±ïÚ¡å@aUGµKà_‡úÏĽytÏØK¨^²4…ªª(êÌìB¨è2HÉ u W½ü)ý€îÞúÒ÷âåõ¼vÈRY4ëBÏ$£1É/0pÍٰƯæü^#|d«©·AKD—%=›rIõršM¨­lÿc¡G†ü;„›šUZÖ—D’Ù6º(Å´›z]xwÃÏ„ž"ø¯ss€4É5´ f"Dcݹg`9ÚØ籯¡~ ~Âi¢êN±ñNîÞîH‡>äo‰_ªò‡Ã…ùK(]¤‚2ËË} ¡è¶ÞÑm4í/&η‚=Ŷ"(U$“€$“V«÷>ð'%ÈÝ¡ã-B×N³L2g ½‚–Úƒ«± Ä*‚Ç_3ü}ýµl|yðþM+áŠkuÅë˜oîÚ “Z´r#Æ÷ ’ÉÈðyþ7‹xï)àêž6ªö¼­ÆškžVè—KßwçkØú,‡…ñüCR1ÃS|—IͧʿÏÑ~›~Ó¿?ápüPººÒo.®4;•®G?üEw¬YG†û,‚ÓÌ`ÀáßÌs°€À…ÚÜä0Ç?¹½k•ø‘ðSÃ?bŒxãJ‚êh¶ì¸LÇpª7a<ÅÃlùÛå'9ÆpGÄOÃN*Éðupù&}.W´*Gnö©yÊw~äV»ë©õ0ãl“ˆ…\Ë+W[Êû¯ EK·¼ßè~pøoÂzWƒ¬^×Â:fŸ¥[I!•¢³·Hœ€ @8Uö•Çü]ý™?âcdD7Gî™°VO•1[hÎÜu¯»¼MÿòÒõ Ny|-â+­6ÞGÝZ‹Ÿ){¨mêHÏByÇ'š½àïØúD7Kã=JïYi‚y/C·tv »+צÞù¯Ç²ß x÷™{|?îê]þ÷ÛFÏÖÍÍ©Y+8ëuÌ’½¿@Åñÿ bp~ί¿?wìßê”SW½Ô½ì~Hx_âwÆÏø%ߊmî¾øšY<1tÒV q¤j ½ %Å«ñ òGo2FVM›•%á±úûÁY>þØ)a¦XÞËáOݺ[ÿ`ê¤ '˜Ä$o²Ì¿$ñî"“²C³&4Ü¡¸ÿŒ°wˆ-u›»?éëâOÞBT‹™ VØÀ†ŠT¨|Œô ô+ó¿ãü‚m:òö„w2Áq¸—FÔ²¯äǤdB I©i+öL“ÄÌvRÿ³x‹ÃW¿,kJØÔ³ÖÍY--ª“Ž·|¿ üÿ3àÜ&d¾¹TU©Úî KÚBëK§võè×6–WÜýݯý¬àš?ÿly§¿øŸáöÓüI8U> Ñ䚉ÛåÞ6ÖŽs²%ˆ’MˆHM§~yþËÿð[ŠŸ³Mîá¯ÛBÔ<]¢ªFæò²×­`Ìq‡êíU"œâP$’G;®¯¹~ ÿÁd¿gïŒÿÙ°Âgÿ¦©©y¿è>#µ{²ìÞ}uóZ&åMËûóÊ¿|ì¯ÚéV¥‰‚© )'³NëäÏÍ*P«†“Œ“OúÜø/ö§ÿ‚/üRý“5Éümû+ßÝø×FÓo‘¬­ì yK‹8'ã:*žcNó_ Ք㾊VÕ+¿vJQêÕõ>ƒ!âìO› ?uï¬_ª¾þi§Òö>:ø9ñòÚêñ5Ÿ‚~'…î"Ž9${^4rRxú€Js‹ÎÒàŠ÷/þÞ>5ж'ˆ×L×"óÃÈóAäÌcã1£DU¡Ãb s‘_.þÒ¿ðo?‰¼áCVý™|Yÿ ¥Òß³hö‘X\ý‘œ‹ÆœE,¨ nÜ«€ì¸!coŸüEñ§ã§ì‰­Á¤þÒÕ"‡ÏkH·4÷ƒíklÁ&û%â—9™›3)ÜŒ ó~7‰ð—‰øM:œ)šK–÷ör|½½iÍï~hÂ˹ú-:É8òçx(ómΕÿÊq^Ž_#÷#á/Å­'ã?…?µü!ö¤'kyb¸Œ$°È §*ÊÙRFwÈ=~G~ÊŸðR ;­vX<ªÜøK[½‚(ä·¾´7ŽÌ— }Ë#+*®C£Àýý™¿k+/‰ºlZOÄ;›[ö4weŠ=S$*˜ú)$êNTc*Ÿ}¾%ûlE,“ˆèË ¶¼É*s}9%w¬–¶Ú÷I·d|Ž}Án)æYEE[ ~¹Åy[dô¾öÕ¤®Ïk¢Š+õ£àŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ¾7ý¿¯¡»ø×g¬ÑJöÚD1L¨áŒNe™Â°títlÌq^¯ñ{öâÐþëñéþ ´‹Å ©¾ââ ѼDત\HqÉÇ’wùã'Äø[5/ýû?ûCÊÿGó|ß/dIßÚ¹ÎÌôkù—Æþ=ɳ,¦Y6ªWUcÌ’•’JMµ+r;;&”›Õövý§Ã>ÌpXøæ8ª.œ+m]¶ÒK–üËK½RÛÍÁG|eý«ñ/EÑ ’ÒH´{;ùm™cšvù’Nxù"…€À8|òÇíì;ÿ&UðþÄÿH!¯çûö—ñþñCㆽ­øQ¥“N»’$…äM†A)m§ BFppF@9úý‡äʾÿØ“¢ÿé5û?‡ÙKÈø{ƒ”\eqrMY©KÞ’kº”ÏÏøó,׈Œ®œÚMlÒ÷b×É#Ôkçø)ÿí¿oû~Í÷ZŽŽûücâo7JðìI,"K{ƒzÑÊ|Vÿ#Áw†6Ú$ܾãñ7⇇þ øûÄ¿µ[MBÓ|¿µ^ܱX¡ß"Æ›ˆõwEú°¯ÃoÚgãÞ§ÿIý·’òÒ=VÃÂ0í¶Ólçdó4­6,dcYemï–ÞæŽ-쨆¾‹2Ìðù] ¸ŠóJ4âå-uQI»ÛÎÍ-¯­µV<\¿WV©Å·'ʼޚ_ç¯o™/ì5ðŠ÷Ç&¼øŸñ ÷UºÔZúfµžy™äÔ¥‘$[™çi´¹ixpù.²néÏé'ìmû8Eñý¼KãëI_G±}†µ A;‹då£B#YŽ !]OšþËß³Èø“â-7ÃÚ$7V~ÑàŽ;™âùþÇn‰¶4 ç–mw­† Õ÷ö‰£[xsF´Óôhü›;Þ÷Ù(U$“€É$×ó§äu¼RϪqnoOý’›å£N[K–öO¼`ß4ºJ£kT¤×x›5§Á9\2¾¿š½I.—ßÑÊÖ]TuݦZ¬‰Þµø¡à-OBÕÈHõJ$˜'È|É&íp­Œ€qƒÁ5½EJbð´qÔ*a±R§4ã$öi«4ýQøÅ õ0µcZ”­(´ÓìÖ©ýçæ=}û:þÙoÃ?‡)£|EMsQšÒvoN°Ûí]±îiœ7™Œç€Ë|uû:øÏᶀڧŒ´9mlÖ7™'Šaná±PNã’r@ožøs˜ûztåF³œjBJñ}ã+]]hû£úç0Ëò¾0Â{9ÍT¦ï 'f»5{;=|™öü<ÁŸô ñ?þÁÿÇ«Oÿ·G€µ¿;ûJ}SGò¶íû]™o;9Îß$ÉÓ;±Ôc<ãâš+îhøûÅ4¦¥'JK³†ÿ’s>b§…Yâãtû©kø¦¿ïß ~Ó~ñ_ý›â2#nï¶9³Îs¾p]Ývçgê<9ãmÆ>wü":¾™ªý›o›ö;¨çò·gní„ã;[ëƒé_›TWÒ`¾‘ù•>_®`iÏ{òÊP¿k_žÝ/½ü¯§‰ð{;ý_8öæJ^··-ÿ™úqE~tÙ|_ñn™e ¶›â[ÛÛ¢ÅQjS"DŠ0T6é]߃¿m¿øNÈÁysa­"¢GÔ-Ë/‚3ì¶Ú¦VZÞ-JÖëî6×ôú3;,þº¿‰>!êwÿ$ð]:ɬ“L†"Qwü«÷ àp>†øká¹üðóCÒu7ŠK2ÆYZ"J3"%I‘‘Ü Û¢½àL£†1ØŒÃ/ƒŒë_™_ÝW—5£¢“Ù-–‡&mÅüï G ‹’q§ki®Š×oví»î|Ýû\þʳkׇÄß ôÑ%Ü›ßS³€á§<,Qã—?6àXí!K'åŠý8¯‘?nO‚1x;Ä1x¯ÃÊVÏ\’ò?”,7$ ¼äù€;8(ÄŸ˜ø?~ÒÂÓ«Ä™\l¯z°IY^Ëž)yük]ù´IŸ¨ømÆÓ¯(døç}-NO}5å/‡Ó—]Mý?hvø«¤·‡üJ¬5Ñ\4¯+_ÆcHÅÚÃ1–%‹’ÚÈÇZüåøo㋆Þ<ÒµÝ5KɦÜ,¦0Á|äèñî ãr\àãvkôá¿Ží>'xM×tPV Fûs‚UÐ’vº²ç8ÈàŠûøíñN^òÜuK⨯œé蔞–¼[å}~îÛgÌx‘‹$Ŭf¡SîŒõmz;s/šZ#bŠVLt¤¯ÙÚksó0¢Š)QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEP P€äQMCƒŠuuB\ÈŠ(ª¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¯ý¹|S©øKá6sáMFÿL¸}^8š[K‡…Ù 3’¥‚FT{ öj¯©éVºÝ“Ûk6Ö÷vÒc|SF$FÁÈÊ°ÁÁþáq6SW=ʱ }œêE¥-t}ôiþ'©’ãá•ã¨âªÃž0wqî¾w?6un÷Äz”—¾!¼º¿¼›d÷´²¾,Ä“€ú]ÿìéû:_üt×ËÌe²ðý“yx,x>TYàÈA<„œª·Ùkðk 7˜¾ðØ“$îdçßmtvpéö‘[ØE@‚8ãB¤j¨_ä_GµKVs‹Ui§w¦œÞþô›ºMïm_u¹ú¦iâ״ºvÓ›VRmZ+É%¿n‹³Øç~ü Ð>hòYøÈ[ý£k\Lîdšå•pØþ'h‚̀2k¦¢ŠþÁ`pùma°”Ô)ÇE¤’ôKO?SñìN&¶2¬«W›”Þí»·óaEyÅ/Úß µ¤Óõ ®µ[ÏœMš#›ìŒ­´¬¥B¶C ¹,6òF~høûaøßÆ··ë¦ê’èºmÓƒ­˜T{tÀQ!?(É 3“ÀS´~qÅž/pÿ 7FUjÉ´áNÎÍ+ûÍ´’ÙhÛ¿Ùvvû ‡ÃüÛ=J¤aìéïÍ;««ôVm÷è¼õGÜo2 %Ù@rkŠ×ÿiø{H¸¼ŸÄÚEÊÛŒùV·)<ÒàD$¶O~É&¿?ÕK0 '€zé¿áIøÏþ…ÿà®þ"¿'«ô‚ͳ5•åJéjÜ¥RÍìíBÝt{ùy ð9'ŽÇhöÒ0½·ÞRÿæzïÄ¿ÛûXÖb’ÛáŽöòà­ÅÃ(`Tª²2@!ó>÷“ÄÿÃd|Hÿ¡ÿ)ö¿üj­hß±GÄ=SRŽ í&×N‰óºââúŽ<y3¿$c…<‘œ ‘î¾ý„üáÍ5—ÆŸjñäŸzW‘ícLÿWNÈ*ænW#"¼L¿*ñ?1R­S[¶nr.ÉB 7ê ÿ½+žž/Á<7B4áJgÚ*5eêå&Òôr^HøóYÖï|G©I{âË«ûɱæOq+K+à2ÌI8 ³à„^&ø“"èš…ür;F'Xö[«*î*Ó6N1ÁaÔ¤WÚ:7ìðóBÔ£»²ðÍ«Ëv­Äó\Æræ9]‘º÷rz%{Y/Ñ×^¬ªç¸å«½©^NWݹÍ+?ûr]ïÐós/¨R‚†W†{o;$»%·uÿo.ÇÏÿ³—ìk/Ã_I¬üNm/R¸´Új[HòG ™;¤ut_™p»z€I8ÈR>€¢Šþ‰áž˸G°9m>X]¶Þ²mõ“릋²I‘gYæ3ˆ1ZÆNò²K¢ItK§wæŠ(¯¡<€¢Š(¢Š(¢Š(¢Š(¢Š(¯-øåû(h?¯Ž¤f›HÖ¼±º’|ƒ4gÈ€C)Á’@õ*+Éβ<ae‚̨ª”ŸGߺjÍ?4Óó;òÜÏ”WXœG ®«òkf¼šhüþøóûëðéŸân‹¡øƒBµd™æÂ]AŒJ.c•Cù±¸.>|g’+䟉ßðO kú]Ä¿ ^ïAÔ– ¶Ñ5ÃMfòNeA¸¹VÂðvœ߶õæß?dïüC²¶‰´¸´G¶râm"(­]Á*à!VÈÈÇdƒø–;Â\ㆪýgƒsSŽÒœŸ+i÷øZ·IEµ¯¼ù´ý3 â8‡±â$döö‘Z¥nÛ§~±-5üqøñËö“ÿ‚rèÁ|¯Ë?‚ln /g+&§¤°ÞÒ˜Ò)G›i,‹#D .Ì>mÛ¾Ïý—¿àà_xú(¬j"_ßCkºmNÐM¨i÷S*ÆÈ¢¦‡{HB$ ª‘‰¯gøÿ÷{K+û¯†Úä·/µ°¼…CÊ@”Ü “óc( 9jøëã/üFO‹Þ}×ü þ'Ñ5©w?Û¬t™“Ís¼îš-›d˾æn¶¼ ô0¾!ñV§‚⼺Qoj´W3çtr\ª2Yu9)V—ÃÍ­]öÙªqwn^óZ{¿²àrºå•3 k‹ÅÉ5MognûÍì—ºžºúŸÁ„V<¡I-Á.gº¸“ƒs3 ûrB *€£ Q’NXõ”Q_Óx +ÃSÂa ¡N (¥²KúÝê÷zŸˆâñU±Õ§ˆ¯.iÉÝ·Õ°¢Š+°ç ÆñÿøºñnŸp뽕dqÊ:’­ŒŒàœ5‡_¦¦•k­ØÉk­[[Þ[KðÏ’7ÁeXyþ'Á?ÌI“Â^%ŽIþ̃$ýv×ÃæG žÞO/ǯgÑN>òòn:?[GÐúlŒPöIb𯟫‹Ñú'ªô»õ?;¨¯¿¼Eû5øÄzÕ„žÒìÅÊÚ[$3Â{28pzA ü½ûE~Êú‡Â}sÎð=®¯«ø}­¼ö¹0‰ZЮ|Å•£}åTa±ÎÒkó¾0ðo;á,7×/Ô•¹œ/xß¼Z½¶ÕwÖÇ×ð÷ˆ¹n[êÖtê=”­gèï¿—ÿµÎ½ðßSÓ4ÿÞI©xbÙ>ÌÖï™m£'†ŽLo;:bFß”ò•úçáßÄý⦆·þ¿Šî,2<í–Üóòȇ•<¼d9¯Îj¿áŸßø3_µÕgàï{Z1Ò-¸_UR”›Vo£¶ª’ºÒÍÿKàë`¼CÈù*=ZJVÑÆ¢IÝy_UѧgÕ/ÓŠk&zW”þÇ_î~(|( â;¯µjÚ4æÒy$˜<ÓG€ÑÈ㌂S'%ŒLI$œzÅndYƉ2ê–à«%}×tüÓºvÒëK£ù§5Ëjåʸ:ÿO³ìדZ¢:)åwSHÇZï”O2þÞ+áÏͦü'²°ÔÒÕöI¨\;<Â$B¥€8Ä›°ppÛ|cñ \øƒ{öj×ú›«¼ˆ'˜²B\‚ÞZ}ØÁÀùTÀàWáÛj•Æ™uös›ÿ´o¦Ø©PÇ’ò!&1ü[ät 1ó5ð7 b8÷;Žs—+¼êÏv£»w{ÊRj)ë«»M&~ÏÄ™Ý.ËxÅ]Z0ŽÉ¾‹N‰kÓEm46üoÿ)ÿ¨¾xkûŸgºÔ§ÿt¶ûxÿà@b_CþÍx/įž.ø¿cmkñY—P¶´ËBá@äcqXÕC29Æ[ÉÏÐÿ±ü#âGí'ñm,þ3xwÅ?<¥ìŸU¿ÕtÉlnn“ˆ,’td¯µ|2D>fÉÙ¿¬¿³/ü›àÿì•wo¨|"ð¨ñGÖõ÷PfXÞ6•$•·yY<Án±+nÁ]¡@þ×áïø†Z©€ÂET_nWœÓµ®¥+¸Ýn£Ê¼ç,ãŒ3<Ù8bk·öW»^öiY?+ÝŸÿÁ$¿à±k–—þ;ý·ütˆ“ÄžÐ5s%»–ŠBÒÜ^Y•V(Y#DŽSµÔÌ^&V‰ê…UÖõÛ/ i’ÞøŠîÚÂÎ y“ÜJ±Æ™ ³I{+ó»þ Áÿ~Ó¼áÝSá·ì‘¯É7‹gdƒVñŸµíô¸_͆Öå_"ì(Uaw Ë2þëì½µ(OÙs.k^Ýlݯmí~»1ìêÖ\ö÷v¿OK÷<›þ ûv^|[øœŸ~ \ê¤hKˆ>Å:K½¨-¢µ Y…»åY ÎHhÃ@Œgøð/Gø áÓ|8¾}Üû^úùÐ od©ëµHT…òX³7—þÄß³ ÿ ì_Ŭ¢[¾Œ yPùÚlDŲp²Hcr(Á »¨ûOö|ý™µo:Ô3ê]iÞOžk振Ú1S¹a‡bUnU0IÉ·òwŠ#ŠqkUrÆ*ÑŠè¿Íõù.EWÜ2QEQEQEQEQEQEQEQE|ûL~È^{ïü7í: ~úkE%§±– cæ‰x9Îå‘…-^ _¦·VÑÞÛI ìi43)I#u ®¤`‚Ú¾Gý¬¿eoøA|=ŽI4y]æ¿·Ähºyiß-F Œ—ÀP§`^NÉ>-øB²ÈÏ:È©þå^U)­y5ÖQ_Ⱥ¯²“zEiûçø€ñ®9ni?Þm ¿µÚ2ÍÙý§¦û¯ìAñÎ/koá_Kmo§j³y¶¯äÈóIy#CG•È TH#ï•õÕ~fY_M¦^Ãs¦Í-½Å»¬±K”xNC+Aé_wþ̇ƆV·:­åµÆ¹dñ_uzµ{¤—ÃÌÞǑ⧠}^¢Î0ëÝ“´üŸIm³Ù¶÷·s¿¾²‡S²šÛR†+‹{„h¥ŠT’£ e<A ƒÖ¾ý¢ü aðÛã>¹£x]eK WáIyŒIHWqä€\œœ’NIý¯›¿o¿…¯¡Ùø»Gµ){ëi¨;pÛθ{ë”`L<¹Ûû^ÎÍM'êã&¶´{Øðü/ÎV[›}^¤Ÿ%UÊ—N{§×ÉÅ‹±ä?²ßÆ„ø3ñ*;jiSDÔ“ì×áUœ ê’„d£wà “_wƒ‘‘Îkó¾ûý›>)·ÅÏ…6Zë3ß[±Þ¹EA$ȪK¤€e=ºžÅ|OÑ㊥(â2 òÑ~òšÿÓŠ÷Úü²JÝdî}/‹™‹¥šÒŽþäÿö×ù¦ïü§{EWõ øˆÒ™éM#jJÏZÎTÓØè§ô¦Ö.;€QE€(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢ŠzœŠZbœ}tS•ÐQZQEQEQEQER?JZGéS?…€ æŠbš}(K™QE`QEQEQEQEQEQ_~Ú_uûÏ‹ZφæÕ.†‡gäyvq‘mºd>`P<ÏŸæ÷m=1_ÇÜmC€òŘV¤ê9IB1M+É©I]½•¢õJNöÓ·Òð¯ UâœkÂS¨¡h¹6Óz&–‹«»Z6–úŸ@|Uý¬<#ðÓIÙßÛk÷ÓH­tëˆå!•s™\#\í9nr°qòwÆÏÚ]øß«—×$6ºdRo¶Óâra‡¨ ßß| :œÇèÚ%ïˆõ(ìü=iuy6|¸-âie|NA'ô½_àßìiâ?‰mu/ŠÅ×…¬íþD{Ë&ó§~&(v€yr@ÎÏÍ·ù[9ã.ñb¢À`¨µFO৤[J²ó´šŽÚ^ÌýÏ.á܃€àñxš‰Ô_jZ»7ov*þ—I½õ±æ>ðn«ã½j=?Áö:…ä¤b8S;A`»˜ôE†Yˆ<‘_Nü8ÿ‚é6Zm¼ßõ «ÍE'ß$´hÁŒ³'˜ÙåCó`tÜ}“áOÂí3áƒmô ¦>yå?zæRi“‚p8èt®’¿eào2Ìž”qÚUë´½×ðAõI}§ÒïNË©ùßøÌfèå­Ò¤ŸÄ¾)yÿuy-|Êð¶™á+&¶ð¦a¦[»™Z+Kt…ȱT…>¯ÑE~íJ”(AS§¢¶IY/‘ùtêJ¬œæîßW« (¢´ (¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š+Žø§û?ø;ãXQñGBµÕÂÀöÇÌwO2&ûѾÆÓ“ò¶@ÜØûÇ=ËŒÀá³N†.”jCGË$¤®Ó³Mhõ]™¾[?kBnïÓ×Gªîp?eï~ÌÖö¿|9iáËmRcqu¼²ºÊ噳‡vÆ7À FTúŠ+JzXhòQ‚ŒnÝ’I]»·eÕ¶Ûîõ"­j•åÏRM½5nïMÜ´AEVÆaEPEPEPEPEPEPAºóEÓ”éJ4½i„m5›¼5[ú)¡ýiÕq’–Àxÿí±ã /Jøªi7÷Öñêz™·Ö»³,¡gG-´rßæ8ÎHó ö²ý¨nÿgètÛO iq^ê:¬o"Mt$öáEJHgUÔ§ÊHÃ~êÿ‚‡Z:xïÃó´l#’ÁãWÇ VBHÔo÷Я>øû0\~кœÓ]7ö~Ÿ¤fHµ ,ÅÂ¥É6F ©åÉ&YIÂœoþ@ãl]N!ñajåÿYö1TãIK“(ÊiÊOD“ÞѲIÝ^ÿÐœ3B?ûxb½´|îmsrݨµ·mFËww§K|à…´çí¬Xx[NÒ|_›«ÝKFçH}3L¶Yó>×w*b6$ÆKÊ…GrŠ@`Oø"O‡ÿeˆ7øÓ®ZøãÄÚtͧ٥‰‡Oѯq™&RîZéÑ°"‘’-¸2y{ö¾³ñ'ÄχŸ³Fƒ¤iž=ñ?„|e2:iñjº¾ž.vm2üÖ]ä±æ@O-ÏÉŸà¿¿þÚIÁk-â&¢`ŽXZ(KÓ÷6¼RÍr¢du@_+nêw"îbŸÔÙ&S‡Éð±¥J…:RvrT¢£n¶I+ú½_–ÇâY–c_3­)J¤æ¶Nm¹[¥ÛÛÑuW•þÕ?µÇƒÿe?]ê5Í?OÔ^ÖI4ûiòÆâ@­°R « †`Ž3 WæÄÿø.ŸÇ~-ƒDýš´}Á"öê3e­ºjš“³Ü¯c,ZBÂÊ*(-…rÞQã¿?ÿk_¶µûMø†I.#‘TM©^¤âÜüâÒÚß1DÐ|±å)${‘åqO`2Z.–'N„š½äÓŸ+Òð…ï'½šM]ÆI4wdY+0¨§O *±Od½ÛïiJÖKºÑµ¢ks·ý³ÿà®þ5ý¦ø[áY0é·sà±òî~bSf2®å-‚ä¹2ð±•SNý€àžZ÷ˆõKoëÚ Î£­¦&²ÓÙǦpYf¹gÂ$§iجFÒ;ɾþÃ?ðL­*=Xj¾²|7ÁnµmNxžî%P‚H-¶G”c¬s°ÜC1Àô7áŸÂá˜öÞ±Þr¢Ï+9ynJîÃ;§.ÇŒžù¶W„ÅñÅ PË}¥Oââ*7õŠêÏH6šP»µöJê1I¸¿±Çâ°ü1QOÉS‚”Rö4Ö²³M˯vìäÛ³^û?~Ãï§ÞÚë?V&x™4a¶T$±¦‘X«ùc ü»‰ù’¾’²²‡L²†ÛM†+{{tX¢Š$‘"ŒUÒ¥¢¿OáN ʸ3 õ\¶•¯ñIë9>ò¢J+¢GÂç¼EŽâ:þß;öŠÒ1]’ý]ÛêØQEõ'†QEQEQEQEQEQEQEQEQEåœ:…¤¶÷ñG<¡ŽHäPÉ"‘‚¬H Ô”R”T“Mh4ÚwGµOÂ+ƒ_ŽŸáy%k ëU¾†)96¡ÔǸœ°2A<à€rAcÏ|%ø·«ü ñU¾¡á»›…ƒÎ¯-]±ßF§”pA‚à6 ]ÄŽkì¿Ú#ö~‹ãî§[¾©&•6›3H²êêË‚ îSœ… ç±àç#âø;Pð‰îôB°_Ù0YQ]\ €À†RAÈ þ5ü7â_ cø=y¦]K)ޔᢌ­wq³ºIÙI'edÒþ›àÌÿ ÅyZÀâä§YFÕ#-[W²–Ê÷Òí]¦ÕÝÚoô á§Å=âׇcÔ¼v“#(2ÀÌ¢{RIe@IC•lv8È$s[:Æ“o¯é7V:¼B{KØ^ ã$":•eÈçHâ¿=þüUÔ¾øÎ _ÃýØ®¢Âÿ¥Aæ#¼[™[fí€n#µ~x_Ä0ø·Ã:v«¦¬©o©ÚÅwÊuI8  0Î ú×ôw†>"ÒñRŽ* 8Ši{HÛÝ’zs+ô{8½¼Ó?ã^©Âx˜Ô¡&èÍû¯í&µ³ó]_SàßÚá|ø“q£Å4—RF·VRÉ·{Âĸ/ñ++©8Ûœ@®³ö*ø¬Ÿ~'¾©}¥í™­hQËr<<ó˜cMÅa)isœá KáË&¾KÑ5›Ÿë6š†'“ya2\A&ÐÛ$F §p@8 Šþk✠ü.ã_oƒ‹(ÉT§ç /z*ïT¯*z½m®çì¹&&„*Á;¥8ÆI;5{I5{6¯½›:ðxüN_7S VP“V¼dâíÚé­4ZyÇÃïƒ>øWæŸèÖ¶Íóe¥˜©Û•ód,ûrŠvçnFqšé袵Á`pÙu%C J4é­£¢—¢I#ðæ¥%Ÿˆ|E¡X^C2 ‹ø¢•2V`FAèEsâqT0‘çÄTQ‹Òòi+öÔÖ ¸‰rÒ‹“òMþFÑ4+ã­y'Æ/Û#Âß ì^? ÉŠ5Y-dšÚ+”6¾`ËIîw–† U‘”ÛÊò+öý«ÿhø(^¹gá¯Ø]øoE‡WšV™iw§i &ä¸óoÄ®æGP uWf  hÉd®uã m?ܤçï+E=u{}Ûlõ=eÙ©Iáä•M#î½}çêwíYÿBø?ûkWº'ÄÝzêÿÅV0Cpú‘h×W›ea´3°DûšRYQ¼½¬Þ›¿<¾=ÿÁp~/~Ô3ÛxOöeÐbø~5tû4†Æçíú­Ã2ʲîÞ8ÖÞ=Ž½#Y#h·‰”d?áïüŸÃÞÕãâ¯wâ(—]´p›( ˜UÙÛ’¤mdÁ^wŠ÷ÿèv¾ðÆ›£xz/³izDf³¶V%-ãÞÎBä“Ë»¹=Y˜ä±'ò^%úAe8ÊžONUçm$Ó„kÍs»=×,oÒ]OÑ2o ñµÚža5N=R´¥ø{ªýÝ»&è¿ðOŸü@”ëß|Uck©koquz²4ÚŽ¦·^^å›lneËɽ&ò7ÄãÚ>~Æ^øsæ¿öWöõĹn°ëb¿*Ç´F0W!¶îù˜nÁÅ}ðÃà—‰~/Þˆü¦Ë-º¾Éo%ýÝ´®wHx$V(¹|r×¼|,ý€ ²®>0ê vc‘LVºt¬±H£ù’2ç‘…ÚF3»ž?9Žg⊠}YJ–Z^7¥KÏÞÖrZmÍ-vHúéá¸O‚“öÍNªèí9ù{ºE}ÑÓv|¿á M}%¦“àÍ:IPEmgenX„EácEUèz úáì!¬j·©yñeâÒmín£&À¸{èÁ’)ˆò‚ o¼p07}/à?†ZÃ5­<¥ÚéÑ?úÆ@ZY°X2F%ßÛ‰À8»_¢ðŸ€9~QÅgÕ}½]Ü#uNþo㞺ßÜOgϼVÅâ“¡•ÃÙCngg?’øcÛí>©¦UÑ´K/i±ÙøzÒÖÂÎùp[ıF™$œ*€I'êMZ¢Šþ‚§N4¢¡d´Ih’]ù4¥)ÉÊNí…QVHQEQEQEQEQEQEQEQEQEQEQEWË?ðPo‡ e©é)Ó`ÙöÌØ_H»ZE¡$™˜¨9DƒŽ3õ5s<~ëZö}®£xöS›æDm—&'XÙ ðŒ `6F3ÔWÄø‰ÃQ⼃€·¿nh;]©GUe®úÇMZmKÂ9ËÈsZ8¯³{K§»-ý4~¨ü?c_‰Iãïƒ6v—2«jÆŸ2eCyj?rÛG!var@ËFýpkäoˆÿüIð“ìð°tßìÿí ÿgÿHŠ_3fÝßêÙ±ë×kÖ?àžºÍÌu½>)1gu¥›‰cÚ>i#–5Cœd`M À8;¹è1ü©á?¼]KŠ¥(:©Óœ$œd®¹ ÜefµIÝý–í{ëû¯ˆ8\6{OBq’¦ùã$ÓNÏ–I5~—ù¥}­/¬¡Ô즶ԡŠâÞá)b•¤¨ÃYOH õ¯Íë)´ËÙ­µ(e·¸·vŠX¥B©ÁVSÈ ‚=+ôξ ý¬tkm ö†ñ4T~TO„äwñ_jQ_šeßGŒ Qˈ©U[E¤uït®}ž3ÅÌδhR„}eòÔù_ÀŸðO‹×Ô™¾'kV±Ú'Ý‹K,òËÝ^T<‡î¾FGËÁ¯1øïÿ ýž¿`W^ñ·€üc«x‡ÃÐ[´—öú}är™"I•ÃÏt‡̤‹ƒ¸÷©\×à‡ü#ø)7ÿë×MÿÒ+ö¼"á|¿ì#„R÷e.i{Òºqµœ¯k_ì¥÷ÝŸ–xƒâ>,¾X•ˆi§e¤lôz+_çsîëø8Ãàe¤";O |RŠ5è©¥éꣿ_WMð‡þ ÙðãOÅOøCÂÞø“o©x›QƒLµ–ïO²H#’i)‘–ñ˜(,2B“ŽÆ¿ ë×`Où=ÿ„Ÿö7iŸúS~üø/,„t‹ÓÌü> æó’¼£¯÷Oé.Š(¯ÈOÝ‚Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( !ý²>êŸ| ¦Eà}55 VÊü8&XâhàhÜ>FQ‚Â,Œçé\_ìwð Å¿ þ&ßj<Ò~Ág6—%ºIö¨ĕXX Fìz#ãWÒTWÀc¼8Ë1üCO‰g9ªðåÑ8ò>UeuÈäôþ÷D}^Œq¸\¢y4cJWÕ©s+»»>koäð¯í‘ÿ'!â?ûvÿÒXkîªøWöÈÿ“ñý»é,5ðHŸù&ðÿõþ?ún©õ>ÿÈâ¯ýz—þ—ÝàŸ_òFu?û Kÿ¢ ¯u¯ÿ‚zk6Óü3Öôø¤Í宧ö‰cÚ~Xä‰qƒ“ ƒämç¨Ï¿×ÝøWR58O.pw\–Óºm5òjÏÌù~9Œ¡ŸbÔ•½ïÍ&¾ôQE~€|˜QEQEQEQET:•òizuÅÍÀc¼m+êB‚N=ø©«;Åÿò)êŸõé/þ€j*ÉÂ’è(ÅN¤bömÿÄG??èVø«ÿ‚ËþN£þ"9ø!ÿB·Å_üXòu~+Q_±ÿ©yoi}çà_ñsæþB¿±'ü À·¿‹µ½àöã 6ëA³KÛ†Ö--¡ÑŸ`a¸“ŸP;׺üIñݧÂï‡z÷‰¼AÌÖÓ®5;˜íÕZgŽÚF€X…8ž¤WäÏüiÿ%ãâ?ý€ ÿÒ_§_¶üšWÅûµ_ý#–¿6âl¶ŽWöå²zë¹ú§f•ó¬¯‰k›™­´Ðù#þ"6ø#ÿB·Å_üØòuñ·Áú¾*ÿà¶Ãÿ“«ñbŠýýGÊÿ–_øù_üDßù£ÿ€ŸÑGì;ÿðOíûoâI>i~)ÓWÂín·_Û6ÐBd3 ù~LòçSg8ê1žÞù_–¿ðmü‚¾/ÿ×]+ù]×êU~iÄÙ}¯1©†¡ð¥_]ãÿ~«Ây|ã,§ŠÄµÎÜ–ŠÛI¥ø ¢Š+Á>(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¬?‰_ôƒ¾ÕèÿògïÿZ+ðþiûNÿÑLÿËwIÿäJ?áöŸ´ïýÏü·tŸþD­¿ÔLóÃï—ÿ"ñ²Ïù÷Sîÿ&~ÿQ_€?ðûOÚwþŠgþ[ºOÿ"Qÿ´ý§è¦廤ÿò%?õüðûåÿȇüD¬³þ}Ôû£ÿÉŸ¿ÔWàü>Óöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJ?áöŸ´ïýÏü·tŸþD£ýDÇÿ<>ùò!ÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úšã½~ÿÃí?ißú)Ÿùné?ü‰Aÿ‚ÙþÓ§¯ÄÏü·tŸþD¨Ÿcä¾8}òÿäCþ"VYÿ>ê}ÑÿäÏߪz¶kðþcûNÑLÿËwIÿäJüËö>&廤ÿò%Dx 1‹þ$>ùò!ÿ+,ÿŸu>èÿògïýøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Vßê&?ùá÷Ëÿ‘ø‰Ygüû©÷Gÿ“?¨¬‡š¬úï€4;ÝVO6êóO·žgÚ{´jÌp0I<+b¾6­7JrƒÝ;}ÇßQª«ÓHìÒxQE|Ñÿ#ÿ‚‘øoöøp8¶Ö|{¬ÂÇEÑKñÝ~Ós´å R£HÀªã ÉT(TÅT*Q¼ž‰/ëï{%«ÐœF"–”«V’Œ"®Ûéý~;#éz+ðoø-¿í9$¬Éñ)c I ¾Ò°¾Ã6¤þdÓáöŸ´ïýÏü·tŸþD¯¯ÿQqÿϾ_ü‰ðÄœ±;{:Ÿtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘)ÿ¨˜ÿç‡ß/þD_ñ²Ïù÷Sîÿ&~ÿQ_€?ðûOÚwþŠgþ[ºOÿ"Qÿ´ý§è¦廤ÿò%ê&?ùá÷Ëÿ‘ø‰Ygüû©÷Gÿ“?¨¯Àø}§í;ÿE3ÿ-Ý'ÿ‘(ÿ‡Ú~Ó¿ôS?òÝÒùõüðûåÿȇüD¬³þ}Ôû£ÿÉŸ¿ÔWàü>Óöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJ?áöŸ´ïýÏü·tŸþD£ýDÇÿ<>ùò!ÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘(ÿQ1ÿϾ_üˆÄJË?çÝOº?ü™ûýE|{ÿ²ÿ‚¦é·„SÃßÞÓIø¤ÁºîÑp‘k¨ææØz÷xÇ*y/O°«å±ø ÙmgBº´—Ü×uÝ?ëSìòÌÏ›áã‰ÃJñz}S]ÿ‚®šaEWÞQA8ÐE~QÿÁL?๺æ‡ñøCöÖm¬­´9Ùu?}’ Á(ʘm’ttòTç2í%Ø „(Ý'Ë¿ðûOÚwþŠgþ[ºOÿ"WÕ`¸?£Ë–)ì¤Úvïdžÿyñ™—å¹n"Xi)MÇwš¿Uw%ªëÐýþ¢¿áöŸ´ïýÏü·tŸþD£þiûNÿÑLÿËwIÿäJëÿQ1ÿϾ_ü‰Áÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘(ÿQ1ÿϾ_üˆÄJË?çÝOº?ü™ûýE~ÿÃí?ißú)Ÿùné?ü‰Gü>Óöÿ¢™ÿ–î“ÿÈ”¨˜ÿç‡ß/þD?â%eŸóî§ÝþLýþ¢¿áöŸ´ïýÏü·tŸþD£þiûNÿÑLÿËwIÿäJ?ÔLóÃï—ÿ"ñ²Ïù÷Sîÿ&~ÿQ_€?ðûOÚwþŠgþ[ºOÿ"Qÿ´ý§è¦廤ÿò%ê&?ùá÷Ëÿ‘ø‰Ygüû©÷Gÿ“?¨¯Àø}§í;ÿE3ÿ-Ý'ÿ‘(ÿ‡Ú~Ó¿ôS?òÝÒùõüðûåÿȇüD¬³þ}Ôû£ÿÉŸ¿ÔWàü>Óöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJýgÿ‚IþÐ^/ý§?bÝÅ¿5í¿]j°KuöXm·$sAåÀˆƒc!r{×™ð¾/) ñ¥¥£w×Õ#ÕÉøÏâV„&¥fõJÚzIþGÒôWÉ_ðYŸÚoDz‡ìŸ§ø›à·ý­Ïâ+k¹û½Öè^†dÙÓöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJÝø]ÿ›ý¥èÿògï ~ÿÁr?å$þ7ÿ¯m7ÿH`¯ß üèÿ‚ÏÿÁ*.¾>%ïÅ¿Ùö »ÏÚ@ƒYÑÕŒ‡W‚$ ²Û)ùò!ÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘(ÿQ1ÿϾ_üˆÄJË?çÝOº?ü™ûýE~ÿÃí?ißú)Ÿùné?ü‰Gü>Óöÿ¢™ÿ–î“ÿÈ”¨˜ÿç‡ß/þD?â%eŸóî§ÝþLýþ¢¿áöŸ´ïýÏü·tŸþD£þiûNÿÑLÿËwIÿäJ?ÔLóÃï—ÿ"ñ²Ïù÷Sîÿ&~ÿQ_€?ðûOÚwþŠgþ[ºOÿ"Qÿ´ý§è¦廤ÿò%ê&?ùá÷Ëÿ‘ø‰Ygüû©÷Gÿ“?¨¯Àø}§í;ÿE3ÿ-Ý'ÿ‘*[/ø-ßí5i{ ³üFŠå#uv†_ia%䫶 ƒÐí óÁhÿQ1ÿóò|¿ùÿˆ•–ϺŸtù3÷æŠùãþ áÿð¿íïð¿íº1‡Jñ~•®·¢4™{fÓöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰]\Qàßúß‚ú†dâéó)+JI¦¯fŸ.š6½žÉÓöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJ?áöŸ´ïýÏü·tŸþD£ýDÇÿ<>ùò!ÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘(ÿQ1ÿϾ_üˆÄJË?çÝOº?ü™ûýYÞ/ÿ‘OTÿ¯Iô_‚_ðûOÚwþŠgþ[ºOÿ"TwŸðZŸÚfþÒX.þ%ïŠd1ºÿÂ=¥ ÊFÈ´ô5NÌ'R®òÿäM)x™•Âq“§SFºGÿ“>Y¢Š+õ³ðÓô“þ ´ÿ’ññþÀé@¯Ó¯Û þM+â‡ýŠZ¯þ‘Ë_˜¿ðm§ü—ˆÿöƒÿJ~~Ø_òi_?ìRÕôŽZügÿäkÿnÄýëÃùÇüoô?š(¢¿f??V¿àÚ?ù|_ÿ®ºWò»¯Ôªþl¿fÛ{â‡ì}°Ÿ³§‰ÿá]xÄ×ãû:ÒïÏ1oÙÿÉ·cýÜg<ç½KþcûNÑLÿËwIÿäJüK›ãêb¨Î*2QÝ»éžÑ}WsôîãLI—C^rNOEk&ÖòOgØýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJñ?âæ?óò|¿ùè?â%eŸóî§ÝþLýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJ?âæ?óò|¿ùÿˆ•–ϺŸtù3÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘(ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“?~è¯ÀOø}í9ÿE3ÿ-Ý'ÿ‘(ÿ‡ØþÓŸôS?òÝÒùø‡ùüü‡ß/þD?â%eŸóî§ÝþLýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJ?âæ?óò|¿ùÿˆ•–ϺŸtù3÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘(ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ñ_öBÿ‚ôüPðWÅëCûZëð™x2û]ˆ´»K[­7'þ>!ûÔ­5}Wnlï-Ÿ|SÆÝ?¡AWƒœpþ/$qöé8½š»WíªZü½:ŸE‘ñ. ˆ–µ(ïY;wÑ»®›éÖ×WÖ¢Š+Ä>€(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š+âWĽàï€õOüNÕ-t]E€Ü^^\6v嘒UAfbH|Jø— üð©â‰Ú¥®‹ è°‹ËˆÂBƒŽÜ³Bª¨,ÌB€I¿à¦ÿðS}{öññçØ4µhß4iËiZS6íÆGÚîÀ8iH'jä¬jJŒ’îþÞE‘WÏ+òCH/Š]¼—vú/›<,ÿ?ÃðþÚÕÖOáY?Ñ.¯§«IŸðSø)¾½ûxøóìÚ´o‡4å´­)›vã#íw`4¤µrV5%FIw•h®¿á¯ìùãïŒöwW?üâÿ[Ù8Žâ]F¹¿Kw#!]¡F Hä_·àð˜l§¨ÒJ0õvû³ùû0Çâ³ÌS¯Y¹N[%ÑvK²ÿ‚õ»9 +Õá…~7Ñø«ÿ„ÿÿ£þWãwýÏŠ¿øIßÿñš×ë¸ùù½}Cÿ>¥ÿ€¿ò<ªŠõ_øa_ßôG>*ÿá'ÿÆhÿ†øÞæüUÿÂNÿÿŒÑõÜ?üüÞƒê¯ùõ/üÿ‘åTWªÿà üoÿ¢7ñWÿ +ÿþ3Gü0ŸÆÿú#ð’¿ÿã4}sÿ?#÷ ú†+þ}KÿäyUê¿ðÂÿèüUÿÂJÿÿŒÑÿ 'ñ¿þˆßÅ_ü$¯ÿøÍ\ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0ŸÆÿú#ð’¿ÿã4à üoÿ¢7ñWÿ +ÿþ3G×0ÿóò?z¨b¿çÔ¿ðþG•Q^«ÿ 'ñ¿þˆßÅ_ü$¯ÿøÍðÂÿèüUÿÂJÿÿŒÑõÌ?üüÞƒê¯ùõ/üÿ‘åTWªÿà üoÿ¢7ñWÿ +ÿþ3Gü0ŸÆÿú#ð’¿ÿã4}sÿ?#÷ ú†+þ}KÿäyUê¿ðÂÿèüUÿÂJÿÿŒÑÿ )ñ¿þˆßÅ_ü$¯ÿøÍ\ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0¯Æïú#Ÿð“¿ÿã4à |oÿ¢7ñWÿ ;ÿþ3G×pÿóò?z¨b¿çÔ¿ðþGôgð›þI_†ìkÿ¢Rº Ãøci-‡ÃoAÏ™m‘È¥^6(*Àò#5àŸðR?ø)†ÿ`o‡‹mgǺÌ,t]¿×í7;NR ú4Œ ®0ÌŸ…U£S‹•*1攤ì—]_õ}’ÕèJP¯O…ZòåŒb›o¦‹ú¶÷Ópÿ‚‘ÿÁHü7û|8[k>=Öac¢è¥øî¿i¹Úr)Ѥ`Uq†düøÁñƒÄŸþ#ê¾-øµªÜë:þ³1šææcÉìTpˆ TPTŒ«âß‹Z­Î³¯ë3®nf<žÁUGŠUEU@+š¯Ö¸{‡©äÔù¥­W»íä¼»½ÛÕôKðþ*⪹ý_gNñ¡¢ïýéyö]=nÙEz~ŸûühÕ¬ ºÒ¾|O¹µ¹e†h¼+|ñÊŒ2¬¬"à Ž57ü0ŸÆÿú#ð’¿ÿã5î<]ìê/½4°8™+ªR·£<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf—×0ÿóò?zÔ1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ£þOãý¿Š¿øI_ÿñš>¹‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf®aÿçä~ôPÅÏ©à/ü*¢½WþOãý¿Š¿øI_ÿñš?á„þ7ÿÑø«ÿ„•ÿÿ£ë˜ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ þŸÀçàßÅ_ü$ïÿøÍ]ïùy½Ô1_óê_ø ÿ#ϼã}_á·‹´í{ÀZÞ“¬é3­Íå´…%·‘NC)Ë¡ŠýÒÿ‚YÁStÛ‡Â)áïˆoi¤üNÒ`Ýwh¸HµˆÔssl={¼c•<—§àÜÑ=¼­êÈèJ²°ÁR:‚;ÒðOõ†Þ.Óµïj7zN³¤Î·6w–Ò–ÞE9 ¤.„d+ƒ9ɨçT}M$¾uOõOªýlÏO!ϱ<=‰ö”õ‹ÒQ{5ú5ÑôôºÔ¥ñïüËþ ›¤~Ü>O|C{M'âv“ë»EÂE¬F£››aëÝã©ä|½>Â'šü_€­–Ö• êÍ}Íw^_ÓÔþƒË3<>o‡Ž' +ÅýéõMtkþ ºi8×äüþ ÿ ÁÕ>þÉú§üI>k_ø†ÖOù v{[WòèyGúÎUw“)ÿƒÿ‚ÁÿÂpuO„ÿ²~©ÿOš×Ä>!µ“þB]žÖÕÇü°êQþ³•_ÝäËù_mÂü/ÏËÆÇMãù¿Ñ|Ùù÷q—Õù° {ûJK§÷cçÝôÙk±Et?>x¯ã³6ð‹Ã!ñV¡o ¸–×GÓ¦¾š8Á ]’b,£qɽvŸðÂÿèüUÿÂJÿÿŒ×è•1©;Ni?6‘ù5<-jËšœ^I³Ê¨¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆj>¹‡ÿŸ‘ûѧÔ1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ£þOãý¿Š¿øI_ÿñš>¹‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf®aÿçä~ôPÅÏ©à/ü*¢½WþOãý¿Š¿øI_ÿñš?á„þ7ÿÑø«ÿ„•ÿÿ£ë˜ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ£þOãý¿Š¿øI_ÿñš>¹‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf®aÿçä~ôPÅÏ©à/ü*¯Ýïø Ïü££ÃÿöÔô¡«ñÛþOãý¿Š¿øI_ÿñšý¢ÿ‚+|9ñ¿Ø/CÑþ'è:φõxu;ù$±Õ,¥³¹EiÉV1JªÀÈ8æ¾[Œ1ªåÎ0šnëf´à,-z9´eR›K–[¦?ÿƒˆäÅt¯ûlÿôžê¿+ößþ!ÿ“Ò¿ìm³ÿÒ{ªüH©àO÷ ÿÿé16ñ3þFTëÚÿÒæýÁåÿ ë•ïþ—ÜWóó_Ð7ü‡þQ¯ð×þ¹^ÿé}ÅmÇò+_õò?úLÎOäpÿëÜ¿ô¨QQEù û°QEQEQE|Uÿ\ÿ‚®iß±‡fð ­u?Š:œU K‡¢qÄó¯F”ƒ˜â>ÎãnÕ“£ „«Ž«#y?êï²]YÍŒÆQÀQ–#.XGwýnßE»aÿ\ÿ‚®iß±‡fð ­u?Š:œU K‡¢qÄó¯F”ƒ˜â>ÎãnÕ“ðóÅž,Õ¦›âê>5ñö¯âûë­OUÔç{›»»™L³\Êä³;»rÌI$“Tköl‹"¥’Ò²Ö£ø¥ú.ÉÁgàMÄÕ¸†·òÑÃÕ÷oîKEÕ²Šè~|$ñ_Æ=fm;á†zžµeÍN ¯$ÙåTWªÿà üoÿ¢7ñWÿ +ÿþ3Gü0ŸÆÿú#ð’¿ÿã5\ÃÿÏÈýèÓê¯ùõ/üÿ‘åTWªÿà üoÿ¢7ñWÿ +ÿþ3Gü0ŸÆÿú#ð’¿ÿã4}sÿ?#÷ ú†+þ}KÿäyUtÿÿä³xGþÃVú=+®ÿ†øßÿDoâ¯þWÿüfº/„±Æ3âÏ…îu/„?-ííõ{Ie–_ ß"D‹2ÌÆ,$“ÒªÌ?2ýä~ôLð®Wû©à/üèÆŠ(¯ç£úœüÍÿ‚ÇÿÁ#?á`Á¨üYý•´Œø…7\ø‹B´NuEê×VÑÌã’ñõƒ, É‘'äMTï÷kòóþ ÿ‰oÉ©ü\ý•t¥þÑÃ]x“@´›ÃË=å²ùiÞHÀùùp7î÷\/Å+ãÆKÜû2gÉÿw³û;|?æœiÁï͘`cûϵö¿¼¿½Ý}­×½ñ~MÑEú‘øÐQEQEQEQEQEQEQEÚ~Ïÿ´Š¿f/ŠšgŒ¾jrišÖ˜ùïEsÆøfN’FÀ`©ö#?}ÿàŸ·÷…ÿo„ßÚþÛ¦x—JT‹\Ñ]÷Ia+‡Cüp¾Ö*þÄ0"¿JîgOÚ/ÅŸ²·Å7Ƶ'Óµ9°ÊrÐ^DHßé‘æDø_`A ªÃÁÏr*YÕ+mQ|2ý—åºìþ›†¸–·׿ÅJ_Uæ¿ŸF¿¦ú+Ã?`ÛÃÂß·ŸÁô×üWO×4íkš,’†ŸLœŽè^Ó0V ¬«îuøÎ+ WVT+FÒ[¯ëð}QýƒÆQÌ(LJ—4%³þº÷]Š(®s¨(¢Š(¢Š(¢Š(¢¾*ÿ‚®ÁW4ïؿóx?áÖºŸÅN ª%‡ÃÑ8ây×£JAÌqgq·jÉÑ„ÂUÇU ¼ŸõwÙ.¬æÆc(à(ˈ—,#»þ·o¢Ý°ÿ‚®ÁW4ïؿóx?áÖºŸÅN ª%‡ÃÑ8ây×£JAÌqgq·jÉøuâê>5ñö¯âûë­OUÔç{›»»™L³\Êä³;»rÌI$“G‰üO¨ø×ÄwÚ¿‹ï®µ=WSînîîe2Ís+’ÌîíË1$’MV²²›S½†ÛN†[‹‹‡XâŠ4.ò;U’IÖ¿fȲ*Y-+-j?Š_¢ì—ü~ÄÜM[ˆ+-ü1ý_vþä´][ŠŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf½_®aÿçä~ôx_PÅÏ©à/ü*¢½WþOãý¿Š¿øI_ÿñš?á„þ7ÿÑø«ÿ„•ÿÿ£ë˜ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ£þOãý¿Š¿øI_ÿñš>¹‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf®aÿçä~ôPÅÏ©à/ü*¢½WþOãý¿Š¿øI_ÿñš?á„þ7ÿÑø«ÿ„•ÿÿ£ë˜ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ£þOãý¿Š¿øI_ÿñš>¹‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf®aÿçä~ôPÅÏ©à/ü³ÿàÛOù/ÿìþ” ý:ý°¿äÒ¾(Ø¥ªÿéµùëÿÿ~Ï^?ø3ñ§Ç÷_< ã ZÞhÅo6³£\ØÇ;‰Á*2(b8â¿B¿l/ù4¯Šö)j¿úG-~?Æu!W4æƒMrÇc÷/)NŽ[Ô‹Oî­Øþhh¢Šý¤ü(®—á·Á¯|f¿¹µø?áOø®êÎ1-Ä:>™=ô!8 ë 1PO*ÿá'ÿÆk˜š4Ÿ,æ“óiÓÂÖ¬¹©Áµä›<ªŠõ_øa_ßôG>*ÿá'ÿÆhÿ†øÝÿDsâ¯þwÿüf£ë¸ùù½}Cÿ>¥ÿ€¿ò<ªŠõ_øa_ßôG>*ÿá'ÿÆhÿ†øÝÿDsâ¯þwÿüf®áÿçä~ôPÅÏ©à/ü*¢½WþWãwýÏŠ¿øIßÿñš?á…~7Ñø«ÿ„ÿÿ£ë¸ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øÝÿDsâ¯þwÿüføa_ßôG>*ÿá'ÿÆhúîþ~GïAõ Wüú—þÿÈòª+Õá…~7Ñø«ÿ„ÿÿ£þWãwýÏŠ¿øIßÿñš>»‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa_ßôG>*ÿá'ÿÆhÿ†øÝÿDsâ¯þwÿüf®áÿçä~ôPÅÏ©à/ü*¢½WþWãwýÏŠ¿øIßÿñš?á…~7Ñø«ÿ„ÿÿ£ë¸ùù½Ô1_óê_ø ÿ#Ê«ë¿ø%ÏüXý†übºŽžïWøg«ÏºöÉNùt©ƒslïÇÑÀìÀùÃâ_À_|ŠÍþ1ø/ÅžM@ºÚ¶³¤\X ’˜Ü#3"ïÛ¹sŒãpÏZäêkÐÃæt%J¢R„¿­û¯Á•‡¯‰Ê±­M¸TŽ«£ûŸGÕlÑýGxÇz?Äïi¾ ø}©Zjú.¯ÜÙÞ[>ø§ºB ‚­jüÿ‚\ÿÁQµØoÆ+¡xéîõ†z¼û¯l”ï—J‘¸76À÷þü}̯Ý?xïGøàí7ĵ+M_EÕà[›;Ëgßñ·BèAäAÁøŸpý\Ž¯zOá—èüÿ=×T¿|áž&£Ä–¬~(þ«ËñOGѽj(¢¾xúp¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(æ_ø,wü£kâwýzÚéu½>µýÁc¿å_¿ëÖÓÿK­ëùõ¯Õü;ÿs¯þ?ý¶'ãþ'¼a¿Ã/Í~¼ÿÁ¶òF¾&ÿØjÓÿD5~CWëÏüaÿ$kâoý†­?ôCW»Æò'ÄÛ¿ú\O™àŸùa¿íÿý73ô®Š(¯ÂÏè`¥’ŠiØ (¦¡È§WL_2¢Š*€(¢Š(¢Š(¢Š(¢Šc àÓ˜n¦:×4—#à×ã/üÿ'oàÏû£ÿÒËšý™Sšüfÿƒÿäíüÿb”úYs_UÂù­û{ÿIgËq·üˆñ^ÿÓ?<¨¢Šý¬þx?§¯ÙÿþH?‚ìcÿ¤ñ×]\ìÿÿ$Á?ö±ÿÒx뮯çìûÍ_ñKógõYþçGü1üQEÈv…Q@Q@Q@Q@Q@6oõMô4êlßê›èk:ß½ÖçòÓâÏùµ?úû—ÿC5ŸZ,ÿ‘«Sÿ¯¹ô3YõýGøqôGòÞeþ÷[üRüÙöü;þR!¤ÿØPÿÑB¿ufÿTßC_…_ð@ïùH†“ÿ`]CÿE ýÕ›ýS} ~UâûÜ?ëßþÝ3ö ¿ä]Wþ¾¿ý"òÓâÏùµ?úû—ÿC5ŸZ,ÿ‘«Sÿ¯¹ô3Yõú­áÇÑæ_ïu¿Å/ÍŸ ÿðn7üž'‹ìQ›ÿJíkö†¿¿àÜoù£ã_ßjþ/¾ºÔõ]Nw¹»»¹”Ë5Ì®K3»·,Ä’I4xŸÄú|G}«ø¾úëSÕu9ÞæîîæS,×2¹,ÎîܳI$Õý›"È©d´¬µ¨þ)~‹²_ðYøq5n!­ü´cðÇõ}Ûû’Ñul¯Dý—eßþ×ßl<ðnÀÝ꼸¸“+m§@qpà‘®G©$…PY€'ì»û.ø¿ö¾ø»aà߃vïP»ýåÅÄ™[m:@{‹‡ìr=I$*‚Ìýýý‡¿aï~¿¢ðßÃx…Þ¥w¶mgYš0·:´à}æë²5ˈÉffiÏóúY-.õ_ÃÕù~-躴¸g†kq o壊_ûl{ÉýÉjú&~Ãß°÷„?a_„Qxoá¼BïR»Û6³¬Í[Zp>óuÙå‚D  ä³37³ÑE~5‰ÄÕÆU•jÒ¼žïúüÉhè Ž_F8|$Øý(û_¿ðm‡ü—‰öƒÿJ~ÁWä/ÿäO?ñGó>G®Ÿà§?|%ÿa«?ý•ÌWOðOþK7„ì5gÿ£Ò¿bÄÀj|,þŸ"”IëJ˺˜F:×óKæ¦Yž”Tt¡ÈëVª® >Š@àÒÖ‰§°QLŠ( Š( Š( Š( Š( ?Þ¯8ý°¿äÒ¾(Ø¥ªÿéµèï÷«Î?l/ù4¯Šö)j¿úG-s}¿™tþ5ê44QEJŸÉ‡é'üiÿ%ãâ?ý€ ÿÒ_°øÿÿÚÉxøÿ`?ô Wì~1Ç?ò4á‰û¿‡Ÿò)ÿ·¥úQ_}ÈQEQEQEQEQEQEùyÿ-ÿȳð‡þ¾µ_ýÖ¿'«õ‡þ[ÿ‘gáý}j¿ú­~OWí\ ÿ"ŠâŸþ–ÏÂ|Dÿ‘»ÿP¯è‹þ Qÿ(ðøQÿ`Qÿ£d¯çv¿¢/ø%Gü£ÃáGýGþ’¹xÿþEÐÿ¯‹ÿI™·†ßò5©ÿ^åÿ¥@ú Š(¯ÇÏÜŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ™à±ßò¯‰ßõëiÿ¥Öõüú×ôÿŽÿ”m|Nÿ¯[Oý.·¯çÖ¿WðïýοøÿöØŸøþñ†ÿ ¿4úóÿØÉø›ÿa«OýÕù _¯?ðm‡ü‘¯‰¿ö´ÿÑ ^ïÈŸÿnÿéq>g‚äy†ÿ·ÿôÜÏÒº(¢¿ ?¡‚Š( ŠŽœƒZS•€uQ]QEQEQEQE×\ò)ÔTÊ<ÈÁÁ¯Æø8ûþNÛÁö)Gÿ¥—5û0˃_Œßðqçü¯‚ÿìRÿK.kßàë¬ÞŠÞÿÒYòÜmÿ"¬øœ»3Äå5¾±„Ÿ,ìÕ잎ÍèÓ]ú©ÿ4ÕÿËÃÿ¸(ÿˆš?ê‰ÿåáÿÜùWEyê–Sÿ>òiòG»þ¼gôÿ’Ãÿ‘?U?â&ú¢ùx÷ñGýQ?ü¼?û‚¿*è£ýRÊçÏþM/þH?׌ïþ‚?òXò'é¿Ä¿ø9?^ñ€õK†? í|5¯]Àb³Õ.Îm#0Ú¶îÁ!€*5¼Oâ}Gƾ#¾Õü_}u©êºœïswws)–k™\–gwnY‰$’jèà2|Väð´ù[ÝÝ·÷¶ôò<¼Ï?Ì3˜ÆÊ®J:¥d•ûÙ%wë·Ì(y¢Šôý>ý…?િ³/ì#ðr øŸ/õ[½³ëZÔº>œ—µÀ’ñÙå„qd…ä³3»{_üDsðCþ…oŠ¿ø,°ÿäêüV¢¾o¸eYW­Ìå-Ýÿ¯’Ù-õØ>5Ìrú1Ãáùcì¹­^íõgíOüDsðCþ…oŠ¿ø,°ÿäê?â#Ÿ‚ô+|UÿÁe‡ÿ'WâµÏþ¥å½¥÷?ñsæþ~ÔÿÄG??èVø«ÿ‚ËþN£þ"9ø!ÿB·Å_üXòu~+QGú—–ö—Þñsæþ~ÔÿÄG??èVø«ÿ‚ËþN¯£¿b¿ø(OïÛÃ@Ôn¾ Üê—úCí¼ÒuX£‚þ?v]‘Èêѱà2±ÁàààWó‘]gÁŽ>)ýœþ%iÞ.ø9«Ühºî˜ÄÅ#ÿØý(û_ŒqÏüøb~ïáçüŠíé~EWÇrQEQEQEQEQEQE~^ÁËò,ü!ÿ¯­Wÿ@µ¯Éêýaÿƒ–ÿäYøCÿ_Z¯þk_“ÕûWÈ¢Ÿø§ÿ¥³ðŸ?änÿÁÔ+ú"ÿ‚TÊ<>ØèÙ+ùݯè‹þ Qÿ(ðøQÿ`Qÿ£d®^?ÿ‘t?ëâÿÒfmá·üj×¹éP>‚¢Š+ñó÷¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(æ_ø,wü£kâwýzÚéu½>µýÁc¿å_¿ëÖÓÿK­ëùõ¯Õü;ÿs¯þ?ý¶'ãþ'¼a¿Ã/Í~¼ÿÁ¶òF¾&ÿØjÓÿD5~CWëÏüaÿ$kâoý†­?ôCW»Æò'ÄÛ¿ú\O™àŸùa¿íÿý73ô®Š(¯ÂÏè`¢Š(¢Š(êr)iŠpiõÓ s (¢¬Š( Š( Š( Š( Œõ¯Æ?ø8ôcöµðgýŠQÿéeÍ~ÎWÆ¿ðV¿ø&~ܾƒÄß æû/Ä Z,£šb¶º½¸f²¾ã¶'ÜÎRNX«ü¤4~¿âéeù,Egh«¦û]5Mu<.&ÀÖÌòºøZ ó’V]í(Êß4´óì~ÑW¼OáGÁ^#¾Ò<_cu¦jºdïmwisŠkiP•dtnUj~é)¥(»¦7N¥' «5£Otû3úzýŸÿäƒø'þÀ?úOuÕüûxþ -ûIx[A²Ó4/ˆþEŽo­¼_ðémåÆŠWsZ’p$š·ÿ´ý§è¦廤ÿò%~_‰àœujÓ©ÂÍ·¼º¿ðŸ²`üDËpô)Ò•:—ŒRÚ=¿˜ýþ¢¿áöŸ´ïýÏü·tŸþD£þiûNÿÑLÿËwIÿäJÇýDÇÿ<>ùò'GüD¬³þ}Ôû£ÿÉŸ¿ÔWàü>Óöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJ?áöŸ´ïýÏü·tŸþD£ýDÇÿ<>ùò!ÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘(ÿQ1ÿϾ_üˆÄJË?çÝOº?ü™ûýE~ÿÃí?ißú)Ÿùné?ü‰Gü>Óöÿ¢™ÿ–î“ÿÈ”¨˜ÿç‡ß/þD?â%eŸóî§ÝþLýþ¦Íþ©¾†¿ÿáöŸ´ïýÏü·tŸþD ÿÁlÿiÖ‰œú—tŸþD¨©Ày„¢Ò©¾_üˆ×‰Ybé÷Gÿ“>eñgüZŸý}Ëÿ¡šÏ©.»–{¦ß,Î]Ûd“’p=él¬¦Ôïa¶Óa–ââáÖ(¢‰ ¼®ÇUG$’@u¯ÕéÅÆ*,ücZ5ëÔ«¤ÛûÝϳ?àßò‘ 'þÀº‡þŠû«7ú¦úøþéÿ§¸ý•ìáø•ñáfÇÚ¥«Gg¦,„&‰o ù„À<ì1rq÷²Gߤn†¿ãLmÇjºŒy[é{ÉéÝk¹ûŸåØŒ·-X+œÜ’ëg¥~ÏKÛµ¯®‡òÏâÏùµ?úû—ÿC5Ÿ_wÁ\?à“wß²–»yñàdz—ÃmJ}÷P–i§ðä®ßrF9g·f8INH$#œíi>¯Ôò¬ÂŽe†jëgÝ>©ùÿíùe¸Œ³VxÚí´ú4Þ?êÏG©úÿãÉâx·þÅ¿ô®Ö¿hkù™ýœjþÉ>1»×ÿgÍ{þý^úÐØO?Ø­î÷Â]\¦ÛˆÝGÍ€z×´Ãí?ißú)Ÿùné?ü‰_/Ä|3ŠÎ1J½E.Tµnû¾É÷>«„¸¿ेÄBnNMû©5f’ë%Øýþ¢¿áöŸ´ïýÏü·tŸþD£þiûNÿÑLÿËwIÿäJðÔLóÃï—ÿ"}GüD¬³þ}Ôû£ÿÉŸ¿ÔWàü>Óöÿ¢™ÿ–î“ÿÈ”Ãí?ißú)Ÿùné?ü‰Gú‰þx}òÿäCþ"VYÿ>ê}ÑÿäÏßê+ðþiûNÿÑLÿËwIÿäJ?áöŸ´ïýÏü·tŸþD£ýDÇÿ<>ùò!ÿ+,ÿŸu>èÿògïõøÿ´ý§è¦廤ÿò%ðûOÚwþŠgþ[ºOÿ"Qþ¢cÿž|¿ùÿˆ•–ϺŸtù3÷úŠüÿ‡Ú~Ó¿ôS?òÝÒùø}§í;ÿE3ÿ-Ý'ÿ‘(ÿQ1ÿϾ_üˆÄJË?çÝOº?ü™ûýE~ÿÃí?ißú)Ÿùné?ü‰Gü>Óöÿ¢™ÿ–î“ÿÈ”¨˜ÿç‡ß/þD?â%eŸóî§ÝþLýþ¢¿áöŸ´ïýÏü·tŸþD£þiûNÿÑLÿËwIÿäJ?ÔLóÃï—ÿ"ñ²Ïù÷Sîÿ&~ÿQ_€?ðûOÚwþŠgþ[ºOÿ"Qÿ´ý§è¦廤ÿò%ê&?ùá÷Ëÿ‘ø‰Ygüû©÷Gÿ“?¨¯Àø}§í;ÿE3ÿ-Ý'ÿ‘+õŸþ 'ûAx¿öœý‹tüpÕÿ¶üAu¨^Á-×Ùa¶Ü‘ÌU—" Œ…Éï^vgÂø¼¦ƒÄV”\n–ß_TW'ã<w‰XZš•›Õ+ié'ùKÿÿÉŠé_ö6Ùÿé=Õ~$Wí¿üCÿ&+¥ØÛgÿ¤÷Uø‘_iÀŸîÿÿÒb|‰Ÿò2£ÿ^×þ—0¢Šý„ÿ‚dÁ,þ~ﱂ<_ñƒÀ¿Úþ"Ö#ºk»¿í­BßÎ)w4kû¸nŠ8QÓ=kèó\Ò–OCë“qºŽ–½Úo«]ÉrzÙî'ê¸v”¹\½ë¥dÒ螺®‡ãÝûýÿKý˜¿è™ÿåÅ«ò]ðä¿Ù‹þ‰Ÿþ\Z·ÿ%×ίxäŸÝþHúÏø†¹Ÿüü§÷Ëÿ?h¯ßïør_ìÅÿDÏÿ.-[ÿ’èÿ‡%þÌ_ôLÿòâÕ¿ù.õïü“û£ÿÉüC\Ïþ~SûåÿÈ€4Wï÷ü93öbÏü“?ü¸µoþK£þ—û1Ñ3ÿË‹Vÿäº_ëÞù'÷Gÿ’ø†¹Ÿüü§÷Ëÿ?h¯ßïør_ìÅÿDÏÿ.-[ÿ’èÿ‡%þÌ_ôLÿòâÕ¿ù.Ÿú÷€þIýÑÿäƒþ!®gÿ?)ýòÿäÀ+÷ûþ—û1Ñ3ÿË‹Vÿäº?áɳý?ü¸µoþK£ý{À$þèÿòAÿ×3ÿŸ”þùòà ûýÿKý˜¿è™ÿåÅ«ò]ðä¿Ù‹þ‰Ÿþ\Z·ÿ%Ñþ½à?’tù ÿˆk™ÿÏÊ|¿ùðŠýþÿ‡%þÌ_ôLÿòâÕ¿ù.ør_ìÅÿDÏÿ.-[ÿ’èÿ^ðÉ?º?üÄ5Ìÿçå?¾_üøE~ÿÃ’ÿf/ú&ùqjßü—_–ŸðYÙŸÁ²íieៀZ'ö‡7‡moÞÛí—Y™æ¸V}óÈïÈ3Ž:rk»-â¬&iˆŽ”d¤ïºVÑ_¤™ææüŽÉ°²ÅלU¶r¾®Ýb¿3çOƒ_¼Kû?|JÒ¼]ðU¸Ñõí_6Þâ#ø2:ž¬Œ`H#¿{¿àœ_ðQï ~ß? L}ŸFñÎÿmè›þçAö‹|œ¼ Hõ(HVê¬ÿÏMzì£ñ#]øOûGx3[øqª]餽´kqnØ%EGF†FVee`A‚®¼ó#¥Ñä–•Ã.ÞOÉþ®·óø{ˆëðíi i¿Š=ü×i.®Ï£_Ó6áëFáëQ×äwüWþ yñËömý·Y’Ö)ù“[»œ³1åŽ3ÇøÖS€­œâ>­E¥+7­ídÒèŸs÷Ìã6£’áž*ºn)¥¥¯¯«_™úèÇ+Ö™_€ŸðûÚsþŠgþ[ºOÿ"Qÿ±ý§?è¦廤ÿò%}¸1“¿´§÷Ëÿ‘>Gþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD©ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“?~è¯ÀOø}í9ÿE3ÿ-Ý'ÿ‘(ÿ‡ØþÓŸôS?òÝÒùø‡ùüü‡ß/þD?â%eŸóî§ÝþLýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJ?âæ?óò|¿ùÿˆ•–ϺŸtù3÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘(ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“?bžzWàü>Çöœÿ¢™ÿ–î“ÿÈ”ø-—í8:|LÿËwIÿäJÒ˜Çþ^Cï—ÿ"ñ²Ïù÷Sîÿ&~ÿל~ØŸòi?ìRÕôŽZüHÿ‡Ú~Ó¿ôS?òÝÒù³|cÿ†ý£|áSBñoÄOµéZÕ¤¶7`i‘ùÐÊ…7%¨eʱRÏVŸê.?ùá÷Ëÿ‘*%åq’~ΧÝþLùžŠ+CžÔüuâk Á¶z¦­ªN–Ö––Ñ&¹•ÎG$’@Å~®~~‹Á¶ò^>$Øý(û_ÿÁ%¿àšöø}w­|C¹û_ÄÛƺ”p˺×K„Ëm8‘Á9y9€åŸì ü[Š±´qÙ„§BWŠI_£k{w^}zi©ýÁY}|»+„1å“nV{¤öºèúÛu³³ºEQ_8}hQEQEQEWàü#þROãúöÓô† ýð¯ÎÏø,¯ü~ïöˆkß‹³ÌW7~5µ·A¬hÁÚC­CY-û³¢(XâEQ´“èx[G/ÌcV»´\\oÑ6Õ¯Ùi¿Nº]¯˜ã º¾g•Ô£‡çtíÕÙê—Ÿe×Ôüj®ŸàŸü–oÿØjÏÿG¥sSDöò´s«#¡*ÊÃHêìjƉ¬ÜøsZ´Ô4i<›Ë Òâ 6†Ù"0e8 ƒ‚ÁWí‘viŸÎ³‹³‹Üþ§¨¯Àø}§í;ÿE3ÿ-Ý'ÿ‘(ÿ‡Ú~Ó¿ôS?òÝÒù¿'ÿQ1ÿϾ_ü‰ûüD¬³þ}Ôû£ÿÉŸ¿g©¤¯ÀOø}—í8æ¦廤ÿò%ðûÚsþŠgþ[ºOÿ"W?ú˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“?~è¯ÀOø}í9ÿE3ÿ-Ý'ÿ‘(ÿ‡ØþÓŸôS?òÝÒùø‡ùüü‡ß/þD?â%eŸóî§ÝþLýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJ?âæ?óò|¿ùÿˆ•–ϺŸtù3÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘(ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“?~è¯ÀOø}í9ÿE3ÿ-Ý'ÿ‘(ÿ‡ØþÓŸôS?òÝÒùø‡ùüü‡ß/þD?â%eŸóî§ÝþLýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJ?âæ?óò|¿ùÿˆ•–ϺŸtù3÷î¼çöÂÿ“Jø¡ÿb–«ÿ¤r×Ä¿ðDÛÛâÏíyñkÆÚgíâ¿øHlt"+«H¿³,í<©`¥·[Â…¸ã‘_¡Þ1ð–Ÿãïêš‹-þ×¥kV’ØÞA½£ó¡• :îBr¬FAg‚+æsL²®Mˆú½fœ¬ž—¶¾©]‘ç43ª A5Û[_OFÿ3ùm¢¾ªÿ‚›ÿÁ25ïØ;ÇŸoÐ~Õ­|8Ö§+¥j¬¹{G9?d» ae­€²(,0C¢|«_¼à1ô3*Äaåx¿ÃÉöhþsÌòÌFQˆ–I}ÍtiõOþ³M¤Ÿðm§ü—ˆÿöƒÿJ~ÀWóIû6~×ÿd={RÔÿgÿÂ=}«Û­­Ü¿`¶»ócVÜmÄNžr5ìðûÚsþŠgþ[ºOÿ"WÅq' bóŒcÄQ”Tl–­ßOH³î¸WŒ°YêÕá7.fýÔšÖÝä¿#÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘+Áÿˆ˜ÿÏÈ}òÿäO¤ÿˆ•–ϺŸtù3÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘(ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“?~è¯ÀOø}í9ÿE3ÿ-Ý'ÿ‘(ÿ‡ØþÓŸôS?òÝÒùø‡ùüü‡ß/þD?â%eŸóî§ÝþLýû¢¿?áö?´çýÏü·tŸþD£þcûNÑLÿËwIÿäJ?âæ?óò|¿ùÿˆ•–ϺŸtù3÷îŠüÿ‡ØþÓŸôS?òÝÒùø}í9ÿE3ÿ-Ý'ÿ‘(ÿˆ˜ÿÏÈ}òÿäCþ"VYÿ>ê}ÑÿäÏߺ+ðþcûNÑLÿËwIÿäJ?áö?´çýÏü·tŸþD£þ!þcÿ?!÷Ëÿ‘ø‰Ygüû©÷Gÿ“>¸ÿƒ–ÿäYøCÿ_Z¯þk_“Õë´¿íÅñGöÁ¶Ò!ý¢üQÿ z Jö#û6ÒÓÈ2„so {³å§ÞÎ1Æ2kÉëô.Ë*äøák4䜞—¶²oª]ûšñVqG<Ç}j‚j<©{ÖOOFÿ0¯è‹þ Qÿ(ðøQÿ`Qÿ£d¯Åÿø'ÇüãÅ?·¯Å!§ø|K¥øSKumo[hó¢|¸óÃÌÃ;S·ÞlÏïçÁŸ„Z'À_…šƒ¾Á5¾‰áëU´´IeidÚ9%¹$’Ií“À¾cñô`Ô¯S™I®ÊÍkæï¢í¯kýW†Ùn!bjcœmO•Å>íÊ/Né[WßNö騢Šü¬ýˆ(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Šù—þ ÿ(Úøÿ^¶Ÿú]o_Ï­C¿ðUÏë?àŸßto‡ÚF§®ëÖö«mc§Z½ÕÍÁ1 `³aU‰À<{Wá¿ü0¯Æïú#Ÿð“¿ÿã5ú—‡øŠTp•ÕI¥ïõi}˜Ÿ’ø“‡«^¾Óƒ~ëÙ7Ôòªýyÿƒl?ä|Mÿ°Õ§þˆjüÛÿ†øÝÿDsâ¯þwÿüf¿Rà߯‚þ1ø3ðŸâ%·Æ ø——ºµ´–ñk:\öp‚‘fE,à‘^×â¨ÔÊkÆMû»5üñ>oƒ0uégXiNœ’\Ú´ÿ’Gè5Q_‰Ÿ¿Q@Q@9[±¦ÑN2åwJ)äR×RwW¢Š)€QEQEQEQEñOüwþ G§~Ùþ›ÅÿaµÓ>(iaX‘>!‰ËíÑeb9²9Ûµ£ü;ñ?†5xŽûHñ}Ö™ªé“½µÝ¥ÌF)­¥BU‘ѹVA¯êhŒ×ÅðU¿ø%&ûgøvoü!†×Lø¡¦A…bDPø†$,·E”ˆå=8G;v´]û>~Q^¯/ìñÂYàçÅ2P•%|)|ÃB"ÁúŠoü0ŸÆÿú#ð’¿ÿã5úªÆáÞÕ#÷£ñw—â¢ìèËÿäyUê¿ðÂÿèüUÿÂJÿÿŒÑÿ 'ñ¿þˆßÅ_ü$¯ÿøÍ\ÃÿÏÈýè_PÅÏ©à/ü*¢½WþOãý¿Š¿øI_ÿñš?á„þ7ÿÑø«ÿ„•ÿÿ£ë˜ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòª+Õá„þ7ÿÑø«ÿ„•ÿÿ£þOãý¿Š¿øI_ÿñš>¹‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa?ÿôFþ*ÿá%ÿÆhÿ†øßÿDoâ¯þWÿüf®aÿçä~ôPÅÏ©à/ü*¢½WþOãý¿Š¿øI_ÿñš?á„þ7ÿÑø«ÿ„•ÿÿ£ë˜ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øßÿDoâ¯þWÿüføa?ÿôFþ*ÿá%ÿÆhúæþ~GïAõ Wüú—þÿÈòû+)µ;Øm´Øe¸¸¸uŠ(¢Bï+±ÀUQÉ$kösþ ÿ‰‡öu²±ø•ûJXÅqãë„éz\ :xupî:’ý³è>lÁ"à‘0þζV?¿iK®<}p‚]/K”O£ÇCrAÿ¶}Í’?Aëà8£Š9ù°x9i´¤ºù/.ï®ËMÿPàÞ äåÌ3ë¼"úv”—~˦ï]Š(¢¿;?U+k:5Ÿˆô‹­?Ä6¶×Öн½ÍµÄk,7¸*ÈèÀ†V‚Á¿?à­_ðI[ÏÙ/XºñïÀ[[›ï†wÓfâÜn–o Hí„äµ¹$ò¤„s­'íåVÖtk?éZˆmm¯¬/¡{{›kˆÖXn#pU‘Ñ ¬ ‚ z™Nm_'¯í¨íÕt’ÿ>ϧšm?#:Épùîáñ ü2ëÝ~«¯­šþX(¯½ÿà¦ðF?ü øˆuïÙ;ÃÚ÷Œ<®ÎÆ=7L´–úûB”åŒMaàë²\ckóµ¤ùwþOãý¿Š¿øI_ÿñšý—œa1ôcZD“èÚM>ÍKªº?̲ vYˆ–¥6í³I´×FŸŸÞ¶zžUEz¯ü0ŸÆÿú#ð’¿ÿã4à üoÿ¢7ñWÿ +ÿþ3]_\ÃÿÏÈýèàú†+þ}KÿäyUê¿ðÂÿèüUÿÂJÿÿŒÑÿ 'ñ¿þˆßÅ_ü$¯ÿøÍ\ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0ŸÆÿú#ð’¿ÿã4à üoÿ¢7ñWÿ +ÿþ3G×0ÿóò?z¨b¿çÔ¿ðþG•Q^«ÿ 'ñ¿þˆßÅ_ü$¯ÿøÍðÂÿèüUÿÂJÿÿŒÑõÌ?üüÞƒê¯ùõ/üÿ‘åTWªÿà üoÿ¢7ñWÿ +ÿþ3Gü0ŸÆÿú#ð’¿ÿã4}sÿ?#÷ ú†+þ}KÿäyUê¿ðÂÿèüUÿÂJÿÿŒÑÿ 'ñ¿þˆßÅ_ü$¯ÿøÍ\ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0ŸÆÿú#ð’¿ÿã4à üoÿ¢7ñWÿ +ÿþ3G×0ÿóò?z¨b¿çÔ¿ðþG•Q^«ÿ 'ñ¿þˆßÅ_ü$¯ÿøÍðÂÿèüUÿÂJÿÿŒÑõÌ?üüÞƒê¯ùõ/üÿ‘åUû½ÿÿ”txþºþ”5~;à üoÿ¢7ñWÿ +ÿþ3_´_ðEo‡>!øWûèz?ÄýYðÞ¯§$–:¥”¶w(­9*Æ)UX9×Ëq†"•\¹ÆMÝlÑöœ…¯G6ŒªSirËtÑçÿðqü˜®•ÿcmŸþ“ÝWâE~êÿÁw~xŸâÿìe¦é ¼9¯x£SÄö· i¤iòÞαˆ.AsJÌPN1’=kñÿþOãý¿Š¿øI_ÿñšçàŒE*X©Í'Î÷i}˜ž#á«WÌiJœ^Íl›ûs<ª¿ oø#ü£_á¯ýr½ÿÒûŠüNÿ†øßÿDoâ¯þWÿüf¿qà’žÖþÿÁ?>è¿4}SAÖ,£¼¤–·0½—|RË•e##Aï[q®"•\±FMûHì×òÌåðû ZŽnåR /g-Ó_jÑÔQE~N~ÞQE#ôÏ¥ Û…÷i€àñYJ\²J("ŠÖ÷¢Š(¢Š(¢Š(¯Ä_ø8þOÇMÿ±JËÿJ.«öê¿?໳7ċ߶žŸªü'ø}ãiiá{Kv¼Ò4+«ØEžä²"”0 ¤Œänµô¼%8ÓÌéÊnÊÒßÑŸ#Ç4çW'«&Ýã¢×í#ójº‚ŸòY|%ÿa«?ý•×ØSãƒð“¿ÿã5Ðü"ýˆ¾4i¿'ÿÁ@|a¬ü5øoãïè÷6úzÃ}¦x~îîÚR¶p«–8ʶpx ŠüwêB–iÍ6’ä–úu‰û·Òl¢q„[|ÑÛ^§ÁTWªÿà ünÿ¢9ñWÿ ;ÿþ3Gü0¯Æïú#Ÿð“¿ÿã5û×pÿóò?z? ú†+þ}KÿäyUê¿ð¿¿èŽ|UÿÂNÿÿŒÑÿ +ñ»þˆçÅ_ü$ïÿøÍ]ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0¯Æïú#Ÿð“¿ÿã4à ünÿ¢9ñWÿ ;ÿþ3G×pÿóò?z¨b¿çÔ¿ðþG•Q^«ÿ +ñ»þˆçÅ_ü$ïÿøÍð¿¿èŽ|UÿÂNÿÿŒÑõÜ?üüÞƒê¯ùõ/üÿ‘åTWªÿà ünÿ¢9ñWÿ ;ÿþ3Gü0¯Æïú#Ÿð“¿ÿã4}wÿ?#÷ ú†+þ}KÿäyUê¿ð¿¿èŽ|UÿÂNÿÿŒÑÿ +ñ»þˆçÅ_ü$ïÿøÍ]ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0¯Æïú#Ÿð“¿ÿã4à ünÿ¢9ñWÿ ;ÿþ3G×pÿóò?z¨b¿çÔ¿ðþG•Q^«ÿ +ñ»þˆçÅ_ü$ïÿøÍð¿¿èŽ|UÿÂNÿÿŒÑõÜ?üüÞƒê¯ùõ/üÿ‘åTWªÿà ünÿ¢9ñWÿ ;ÿþ3RY~Á?u Ø`ƒàïÅyÝcS'…ïb@IÀÜïUòIu&®áÿçä~ôPÅÏ©à/ü6ð§…5?xšÃFðm…Þ©«j“¥µ¥¥´FI®es…DQÉ$1_¹?ðJø%>™ûøj/|XŠÓTø£ªA‰$KŽmíÛ¡ƒ‰%yE;rdwüËþ K¤~Äzx§âp´Ö¾&êPm’áT<l¿4Ç»œópHùW »Ù¹Í~qÄÜT±|Ø<½Í¥%ö¼—÷{¿µ·Ãñ~¯Á¼õW~axAýžÒ—÷»/³»÷­ÊQEðgéÁEPEPEPEPEPçüûþ ŸâÔ¾)þËZj§Œ5ι¡Û&º-qn£¥×vAþ»¨ýçúßÇi¢{yZ9ՑЕea‚¤uv5ýSWçü÷þ ¿#Ô¾)~Ë:j§ŒP5ƹ¡Û&º-=ºùzîÈ?×u¼ÿ[÷\/ÄÿSåÁâßîöŒ¿—Éÿw³û?áø9ãYŠ–;Þï(¯·æ¿½ÿ¥zïøéEz¯ü0ŸÆÿú#ð’¿ÿã4à üoÿ¢7ñWÿ +ÿþ3_¤ýsÿ?#÷£ò¨b¿çÔ¿ðþG•Q^«ÿ )ñ»þˆçÅ_ü$ïÿøÍð¿¿èŽ|UÿÂNÿÿŒÑõÜ?üüÞƒê¯ùõ/üÿ‘åTWªÿà ünÿ¢9ñWÿ ;ÿþ3Gü0¯Æïú#Ÿð“¿ÿã4}wÿ?#÷ ú†+þ}KÿäyUê¿ð¿¿èŽ|UÿÂNÿÿŒÑÿ +ñ»þˆçÅ_ü$ïÿøÍ]ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0¯Æïú#Ÿð“¿ÿã4à ünÿ¢9ñWÿ ;ÿþ3G×pÿóò?z¨b¿çÔ¿ðþG•Q^«ÿ +ñ»þˆçÅ_ü$ïÿøÍð¿¿èŽ|UÿÂNÿÿŒÑõÜ?üüÞƒê¯ùõ/üÿ‘åTWªÿà ünÿ¢9ñWÿ ;ÿþ3Gü0¯Æïú#Ÿð“¿ÿã4}wÿ?#÷ ú†+þ}KÿäyUê¿ð¿¿èŽ|UÿÂNÿÿŒÑÿ +ñ»þˆçÅ_ü$ïÿøÍ]ÃÿÏÈýè>¡ŠÿŸRÿÀ_ùUEz¯ü0¯Æïú#Ÿð“¿ÿã4à ünÿ¢9ñWÿ ;ÿþ3G×pÿóò?z¨b¿çÔ¿ðþGÙÿðm§ü—ˆÿöƒÿJ~ÀWåüûû=xÿàÏÆŸÝ|`ð?Œ<)ky¢C¼ÚÆscî'ª4È¡ˆàsŠýP¯Ç¸Ö¤*æNPi®U¶§íüJtr®Z‘ióKunÆį†ºÆ/êžø¥ÚëZµ·¼³¸\¤ÈyíÊ° 2²ÊÀ0 €kð‡þ oÿÈ׿`ï}¿AûVµðãZœ®•ª²åíäý’ì…”v¶È °Á‰ûõX¾è?¼ªxcâv—k­h:ÔÞòÎár“!ç·*À€ÊÊC+À‚¯3"Ïkäuùá¬Åþk³]ÉžÎaøƒìªé%ðˬ_êŸU×Õ&.ôWØÿ·ŸüÛâ/ìÙñeâøáÏ|Að^¬^m2ëJÓe¿»²Py‚í BU×#`,ƒ‘ƒ¹Ãá…~7Ñø«ÿ„ÿÿ¯Úð™¾J5©ÕV}ÚOÑ®Œü ‘ã²úòÃդÚ~iõOþÏCʨ¯Uÿ†øÝÿDsâ¯þwÿüføa_ßôG>*ÿá'ÿÆk§ë¸ùù½ŸPÅÏ©à/ü*¢½WþWãwýÏŠ¿øIßÿñš?á…~7Ñø«ÿ„ÿÿ£ë¸ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øÝÿDsâ¯þwÿüføa_ßôG>*ÿá'ÿÆhúîþ~GïAõ Wüú—þÿÈòª+Õá…~7Ñø«ÿ„ÿÿ£þWãwýÏŠ¿øIßÿñš>»‡ÿŸ‘ûÐ}Cÿ>¥ÿ€¿ò<ªŠõ_øa_ßôG>*ÿá'ÿÆhÿ†øÝÿDsâ¯þwÿüf®áÿçä~ôPÅÏ©à/ü*¢½WþWãwýÏŠ¿øIßÿñš?á…~7Ñø«ÿ„ÿÿ£ë¸ùù½Ô1_óê_ø ÿ#ʨ¯Uÿ†øÝÿDsâ¯þwÿüføa_ßôG>*ÿá'ÿÆhúîþ~GïAõ Wüú—þÿÈòªú þ ñÿøñOíëñHiþé~ÒÝ[[ÖÚ<Çh‡Ÿ.<ðó0ÎÔí÷›sÒþÈ_ðI‹_´wÅëMℼUðÿÃb}OVÖôyì„pƒÊ[¬È¾t­Ð(È[sû™ðà/…¿fŸ…ºgƒ¾i‘iz.–˜DÉ;Ÿ½,¯ÕäcÉcúùŽ#âªY]?g†’•YmÕGÍùö_7¦ÿYÂü[8«í1Qq£­ôr}——wòZì|ø áoÙ§án™àïƒÚdZ^‹¥¦2NçïK+õyòXþ€;(¯ÆêÕyº•äõmõ?t£FžœiRŠQJÉ-’ (¢³5 (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€pj@sQÓ•±ÅiNVvÔQEtQEQEQEQER8È¥¢¦QæV0pj@sLeÁ¡McÊìÀ}u¢º(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢ŠGû´Ê{ýÚeaWp[ŸQÓ•ñÖŠs¶ŒQI¸zÒäzÖ×@Q¸zÑŸz.€(£põ£#Ö‹  (Ü=hϽ@6NÔÚsœô¦×=O‰€QEQEQEQEQEQEQEQEQEQE ã¥%^À<6êZŽœÖ·KèÀuQZ€QEQEQEQEQEQEÃÔÒRž¦’¹QHŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( «g­-F <ŠÞœï£h¢ŠÔŠ( Š( Š( Š( ŒŠŒŒ’‘—"³© ê€æ–£¤Šr¾€QEhEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEP?ݦSßîÓ+ »€QEQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@ ­Šx9éQÑWŽ IE4?­8ô­Ô”¶¢Š*€(¢Š(¢Š(¢Š('h¤»I»+€ÃÉ¢Š+Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( •[’ŠiÙÜ (¤C‘K]Q|ÊàQE0 (¢€ (¢€ (¢€ (¢€ëÜP¦La´Ö3\¯™ú)äRÖ©ÜŠ(¦EPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEP?ݦSßîÓ+ »€QEQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@ëJôÊ+EQ $Ü=hÜ=j:)ûV›‡­‡­GEÕ&áëFáëQÑGµ`I¸zÒ1Êõ¦QCªÚ°QYQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEƒRšŽœØÖ”åg`EW@Q@Q@Q@Q@g­Q¸ ksO¤qži沺ù@uQZ€QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE#¶:Ro•\þí2”±#šJ眔ÐQPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPÔî´Å84ú鄹QVEPEPEPEPMa´äS¨##š™G™š)ªppiÔEÝQE@QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEÙ;S©²v¨©ð°EW0Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@=E2€qÒª2åwJ("Šê½À(¢Š(¢Š(¢Š(¢Š(¢ŠGäu¡NE-4ü§>µÝwÔQEXQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@ÐÓwSQ)ò€ú)›©£qõ5>Õú)›©£qõ4{TéŽrhÜ}M'Z™ÔæV¢Š+ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¢€[ŸQÓ‘»Öœí£ÔQEnEPEPEPEPAÑE€Õ884êF\ò:Э¸TGOu€´QEXQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@1—ŸAëS8ó #¢”ŒJåjÀQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE=[u-0Z鄹QVEPEPEPEPMo•³N¢”•À:ÑMk`ô§RŒ®EUQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE×ëM§?ZmsOâ`QE@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@9±¦ÑM>Wt”R+dR×RwW¢Š)€QEQEQEQEŒ7 ElpiÔ×^õM{ÈQH­ž´µIó+  (¢˜Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@ ~´Úsõ¦×4þ&ETQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE àÓ󚎜­Žµ¥9ÛF¨¢Šè¢Š(¢Š(¢Š(¢Š(¢Š(Œ6ž)ÊÙ¤g­0ü­Y?Ý»­€}š+P (¢€ (¢€ (¢€ (¢€ (¢€ (¢€ (¤ÝÎ(nÀ-Q@Q@Q@Q@Q@Q@Q@ ~´Úsõ¦×4þ&ETQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEõlŠZŒŠè§;«QE Q@Q@Q@Q@dQE&¯  iÁ§R2î¡x5÷_+h¢ŠÐŠ( Š( Š( Š( Š( šýiÔ×ëQSá7Z7ŸZJ+Ÿ™÷wŸZ7ŸZJ)ó>à.óëNVÜ)”SŒÚz%²9¥Ü=k¡I0 (Ü=hÜ=hº¢ÃÖÃÖ‹  (Ü=hÜ=hº¯Ö›Jç&’¹§ñ0 (¢¤Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( •[’ŠiÙÜ (¦£g­:º¢ù•À(¢Š`QEQEQEQEŒ;ŠZ)Isš(P€(¢Š`QEQEQEQE×ëN¦¿ZŠŸ ¢Š+˜Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ¤"£¥SƒW r°EWHQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@5Ç4êÈ©’æV:(¢¹@(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Š(¢Šr·cN¨éÊÙë[SŸF¨¢ŠØŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ¸ïM©=j20k ‘³¸QYQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEõlýij0piêÛ…oNwÑ€´QEjEPEPEPEPEPEPEPEPEPEPMqžiÔR’æV:)X`ÒW+VЊ(¤EPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPJ % 9¢˜­´Óóšé„ùQVEPEPEPEPEPEPEPEPEPEP0ȦT”Æ5Xõ(¢ŠÄŠ( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( •[š(¦Ð¢Š+¬Š( Š( Š( Š( Š( Š( Š( Š( Š( Š( ‘ÆETÏTÀeQ\ QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEÿÙrocksdb-4.5.1/doc/rockslogo.png000066400000000000000000001704071267530634000164440ustar00rootroot00000000000000‰PNG  IHDR€8èÓÁCsRGB®ÎégAMA± üa pHYsÄÄ•+ðœIDATx^ìý $÷}øþ{fzn`pçà NÒ4eSKšZ®èÖšMQ0 °$»-Ë!›Ö¾Ýol—¶w_„ì µŸ×‚EY¶Éǧ§•Ä =z_’‰žD‘ ˆƒ8¸/3À sõ«oV&ÐÓèî©êº³>ŸˆìÊÌʪÊÊʪÎÊoýþÿ…§Ÿ~t©0ó6Õ—Ì80@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€ZB Ð`€–´„ %À-!h 0@K€Zbáé§]ªÇ¨l­/çÕÑú˜5*€ZB Ð`€–´„ %À¬°µ¾f %À-!h 0@K,<ýô£Kõ8Ðj[ëËy—S¢Çº£k²­J9Z_³D0Ì……ú’î¶ðí$æTBàÅî(@K€€9¦2h00ç4 ´‡æ‚SëS ´ƒ£>˜ 'ëKV§/` ÀMA³O ð!00ÛÀ§³K À*á0‹À¬A ³F À:ëK`€XÇBgP ³B s!! "!°j`˜v`˜ N2 ª`Ú9ꀹp¢¾„aP Ójáé§]ªÇ‰|šçf¶ŽÖ—LV¾› ìßý`š¨`š„€i"`Ä| €öóý¦…`ä´´ïK Ìßÿ`€FÎG-0/„À0iR †(!ðbw;0câÀ0?:ƒs0 `€‘;Y_â‹Ìç`ÜÀŒÆI À˜ùÒ0›ëK˜ç`\ÀL@ú`¼NÔ—0)B`0°X_óE £&`Œ„À0J`&Ä—=˜_9/àÜŒ‚`ä|Ô®Í=˜oÎ À°I%`.,Ô—0m„À0L`&Ì—< ç28e ƒòß`ä–êKÖ&=ßO™[:ƒó00ÀÈù‚ À4p*YÒTýZxúéG%#µØô³4›ŽÖ—m°¹`Vùž½ð³?X“_ÀôP ½Àº|±€éÒÁ¾³ÀjÀ#§¥}€ÑÀJ`fÜòªàÅÌ€¹µðôÓ*M©-ÁïmfÛÑú’þÍû¯°í;L“ÍõóÄ÷2æD`ä|ÔÂüj~}Àd4ßË|7`~H%FÆ— ¦‰†™wÎÕ04 02¾P´‹&£6Æûàuö¡á˜÷}Ê~ BÐð:ÇÖ´— `è‰ ˜ù,òylÔB} 8¶ ÍÀcáÄÀôñ] €öŒMN,¨º˜.ªh0ôÌ—A†a±3Ø—”}†¯ ‚o0ÛÀ0“œ”€Ñq¼ Àì@_|dXìK ƒýF'ÇÛŽ¹˜=`˜iiV&0À0ÓêK€ÓYª/0ýÀÐ7_öûÃb_€ñÉñ·–x˜^`h!0p:ZáÉûÉ18ÓI âKÃ’}i#¬Æ¾¬çd} cp¦`&å$ƒ&ÇXPÀZœ „ÑÈ1x3Àä9ê€ óÅŽIkšÛRMÁë|>Àd4A°&ט0ÀÌË!À•ì09i±© ƒ`¼ž~úÑ¥z€¡r€Ï¸œ¨‡iä}0~Gë˶²Oõ¯íûл´£&Ãq9ã㈠U¤¬”}Â~“µü˜¼9Fw¬Àh€`æmî ¾0²œýÕØ/`²¾Œ‡ZÃIàt|FÀ|ÓÀ<´’ µdßHÿÀüY¨/é²üû›ïr N ÐZN°–| °ÀôÈñùÊctÇìlŒ Õœ0`=ö˜>«…ÁÐ;0´Îb} Íõ%¬FsÐ0?NÔ—ÀìhÂ`0½@ëè׋•À¬GsÐ0?œ „Ù&  7Žú •|d%û§c€Ùáø€µ € µòePsÐ,ç@{ä;žïy¼‘Z-ÍAçË &€i89Àzì0[ŽÖ—ð:0Ì…À‚ >+`väøÝ1<§À\ñ¥È~êpÚA ÀëÀ0w|)$Ò?´¦ÁY‹Ï˜MÍw~ÇôóL sË—A;d-ªÄf‘3~0×ü*˜-õ%¬”*q ]–êK`>ø¾0¯À@‡/…óÍëÏZìÐ.Zø˜` &è™o^€öÃüÉw½ÕÚláé§Õö ÀH8˜f–­/å}0{†õÚŠ}j2¦}¿`üRS + €ÙçX T°Œ~_Yà€•„¿³Oø ÐV``žù¥Y@`¥®˜}>ÏÚJ ¬ÁÁùåµg5ö €vK+ ù¬_m ½|δ‘X‡jÐùå$«ÑÜ'@{9î˜_~ðÐ6``›ëK擬äë@û$øþ@›8ƒ¬Ã¡B`V²O´Kš~ÎŽõÚbáé§]ªÇ*Í´ÉÑú²_Þí°Ñ×ìSÓašö ãëX¸¿”ßü¡z¢G·>RŒƒc}€6PÖôÀ‰Áùæõg%ûlÈë‘>컢Çúm z¤?àùæ$+Ù' o©Þ!00V9Öof‘èQ`_þæ›×` oúzd„ÀôHÀ##(¡ÍzíÈû }NÔäا¦>Âf›ÿ­cõÌ×JùòÍõÄ-î*å#÷ÔÃæ˜`Ö¨6À‰Âù•JpͳÜb} ḺC¥|ñõÀ°90kTŒŒƒcÚ®—_{´W!uGÇÊ>5TÌ.ÿ[ÇbÒÍ7ßúH=ãsþùç—”'žx¢\pÁåì³Ï.>ø`uɬrÜ0+Tädá|[è öö˜jU­„ñ:yòd9~üxÙ³gOÙ¼ys9|øp¹ôÒKË

¬|ÝGÉ>5ƹO0ÿ[Gn«no}¤™Ÿk)Ç•²¸ëõËã‡KÙ²³”ÇJÙ\MÍxsÙ,³ò¶½^.¿ïÓùÈ=õ«I¨› ßsÎ9§êû7ïóÏ?_¯¾újÕ ô®]»ª@8×=óÌ3UUpš…nnˬrì0À#ãD ó¬ùè}0r8y¬;:rö©Ùà$Àìð¿uäÀ뛵f©§-<Ÿçw^ð¦ÂwÓ¦MeÏž=Uø{ðàÁ*ôMÈ»uëÖ²}ûörâĉjxòÉ'«pøú믯ã?þã?.—]vYu{f‘ã€i#'R˜wùè}0?Æõ…ß>5;NÔÓÍÿÖ‘›æ€saæ—ÞWÊ ÷Ö-7‡ápÂß—^z©ÜÿýUÀ{Ýu×U¿©ðMÓÏê&ôMÀguVYXX(Ï>ûìk¡qúfV ¦‰`dœH托µ80Ýüo¹~à……RnY§_ÔY«–#yévl](KKKåÄÉÎЉÒ/eË?úZŸ¸m«nMø›*ß|°œ{î¹UŸ¿yþ §¹ç;w–Ǽ w3žÐ7Í=ŸþùÕåsÏ=WÇ Š{ì±rá…VMB3«ûL mjÀ4X/ü[¬G˜´¾gí\(gïZ(ÇN”*ô=v|©ê¥º¾3œÛ¹îÕ¯þLr&ÔLÉmóÄO”7½éMUœÐ÷Å_¬*‚z§ê÷ᇮ‚Þ‹.º¨ }Sœ*ß3Î8£ê 8CšˆŽÌK³ÐÌ2?*˜`†À}Öb߀¡YØRÊ™W×LÒ¹»ÊâæR^:²Tvn+eWg8v²”‡ž]*‡–*~õD)Ûûµ²åÿù–ê6—^ziyú駫ñ6H°› 7•¾©üM˜›°;Ï3}þ¦ŸßTgH…o–O?¿MUpÂà4ýœ8UÀé/8Ź/f™ã€i  h€‘qÀ Ì›caÔ‡–>[g—æওÿ­#×O³Í½ö!ûŸïŒÍ™;ª*ßM wÍ–Ržy©œLÏ›Kyõx)OX*gu–‰sw—r¼³l¦Rœjá 9TÎò÷^ñ©rÍ»nª–E©à=vìXð¦²9C‚àC‡U¼'Ož,û÷﯂à4 ªàHðû½ï}¯ ~¯¼òʪR8Ë&Îmç~yä‘rñÅW·aV9ö˜4À ‰CKÖÓ=! ÁÛ>Þ ‹{ ŒéËž eÛ–R6woöfHµïbç2¿÷?}²úÙc®?p¸”'_\*»¶¦àRÎØÞíû÷•£¥œè\&ø}ùÕRu†L?òÜR9çOÿQÙtç3Û§sšvNµoª|Ó¯qªw#!n*xÓpª‚ÓäsÂᄼ¹Lhœ è½{÷VU¿ €§éè4!Ý,³ÐÙh™Ç,óã"€IS 02vy4ê_zûlm*¦ÿ­#7Š àÕÌH˜xûoü@=6Zw|à÷ê±µmÚÔý‰Z*{ã¼3Ê‹‡—ÊÓ—ÊέÝ8n.ãÑïuO£ž³{¡œìÜæà+iþ¹Ûtà¼[:÷yðH)Û;·Iè{èèR9~¢{›3¶•ò½ÃÝÇ|楥rÕyõäf,ÈOpªx_xá…*¬Meoª#M='üM(œp8Añw¾óª8˦B8— ‰s™e3ïèÑ£e÷îÝåÙgŸ­æ}öÙ塇ª.™UŽý&I 02N¤ój”_ô}¶Î>'‚¦‹ÿ­#×Âx\!î¨m®«S郿[UífHµî«Ç–ÊöÅ…ªÂ7U¿™Ÿ07apÞ¾iþyÇbw<ÁoBä,—ù©~ùHwÞ+ûÚ±u¡ìÞVÊ ‡K9v|©\vΦª’ø¡ç–ÊÕçÏVœªÞ„µßýîw«Jà /¼°jÂùðáÃU€›À7M97M;ÿáþa¹úê«Ë9çœSõœ*ß¾©~Ó›ÞTßwß}媫®ªúÎíÓLtúN5pšÎ2̢Λ¡ê&€IŒL¾Äu¾ ÌQöì$õì;QLÿ[Gn\pŒ8žµà7aã‘W_-[;ßͪP±Ž[+©PÍQk‚Ê\·©³ÌÉÎeÆá¿_Uþ&ÌMosË U?ÀGêøà‘¥òÜKKeËænÅpîïÈÑ¥räx©ªˆ³ìžiz¡xe©œ±=} /•{Ÿ\*gï,å¬ §¬Ó)¦0N ûè£V!púéM8›¦œÖ&ζ»üòË«ñ{ u¯¿þúòüóÏWÍ>§BøÀUqÂä¾Y&×G^³Üw–K…p†¼NÌ*?þ˜0ÀH9™Ì«Q}ÑÏkÖ‚s»n¸¡<öØc嬳ΪîƒYä˜`À#ç„ 0Àô"%2Ýj&ÅÿÕ‘ë%xW 7@¼V|Ç~¯ëÃóHx˜P6Cƒ‡*<ð@yË[ÞR‰û÷ï/ÿà'²O9îÒRÙ¶uk5ó¦MÝPòÕW«€wqóæ*ÄM°˜ëþæ²i:ÃÁ—_®æ-vnŸåšesB5ËÿÒ¿»b+¤iç£'JyòÅ¥ræŽRÎÛ½Pžyi©êKøœ]Ý&¡¿ûüÉòÒ+¥\~nÀñ¶ØþQ=1> €Ì>þøãU›~{úæuHp›>~Ó'pBõ„Ä©ìͼK/½´Úî ìgù„¿©βY.áo¶m^§\ÿÄOT÷Ÿjî¼ÖM0œë™5Žù&A 0NªóFL¯TL–ÿ«#7Mpcéx)w^]OœÞjáo/•ºÃ–懦z4¡ïOßv[,& LPøo?÷¹òÔSOU!a‚ÇT—^yå•UX™€1aåߺùæׂà'O–ôû›0øåÎýíè,Ÿû©ú²í<^w®?c×®×–‹„– ‡o]H6x*€]*—rÁ™ém å‘ç»ÕÀ €s÷=u²9VÊÕLœð7Ïûᇮ¶÷%—\RUb÷»ß}-ÜÍeù„ñM¿Ê]tQU)üòË/¿Þç;ß©^›„Å™Ÿ!Û½ ‡Óôsî+—!·I¨Ùe—Ò„7³$ï¨nU>ã!‹|™K¿•óbÔ_ðøi!0ÀäøŸ:r½VÝN¢Yß×my<Ìà÷ö[o­ÇºVÞwúm‚ĸ Îfø{·ÝÖ e;CU%ÚYîŸý‹Q…… ú&0>xð`*& Ìý$´Ìt‚ËüñWâÎ2gìÞ]Uùxé¥×š"Ží ó8ñá&ôMåðéd~ùG~§<¨”KÏî¼ßyædÙ¹u¡œ½³š,ßx|©3]ÊE{¦?NÏ©þMðžmž×&ÛóÑG­BÞl³TôæõJ…p‚àLgû&°Ï6Ïí2žù é›à7¯IÂù„½™Ÿë³|^»ú ›³Ü÷¾÷½ê5eV ÆI 0``2ÔËgjŸ'ʘrB`€ÉÜ4ÀÑO³ÐCZÇ•Áocyœ6!cš N¨˜p0ÁoÆSù›ð1ˤš÷¿ÿ¹Ÿ«BÄ7½éMUœù²L¦›ŠÒôU›êÓ¦¯à¦"õý½¿WvîØQ…“[ëàwPÍã'0þüG~¯¼rt©2Ý㹟ÜG¶Íƒ>XUö¦¯à4ÝësY6Á Œ3ž×6ã©:Þ»wo½vÌžÀ" €QŒM¾Ñõû­`–:Ðsº}„ÀãçÿéÈM0NUf¹TX&MEeBÑ¥ A«Š×­[«ªË,{þÿç?)gl_(O\*»ê]cë–ÎèãÝñAÖ±—à7f=ö¥ùßÈ:g=SáûããoT¡jÂß-›7—ÏýÿG&èM˜Ø4'œçœ¦ˆ›1÷‘€1AbžwcØ–?ÇVçºýïþ]õÜÓ¬s‚Å4KœûÈsNh˜1·OóÅ©,ÍmD6¡ln›e^¦Â5›ùÃðwüÇ«ÇW;¯Oìؾ½œ8~¼ ‚ów|à÷ÊÃÏ-•«Î›þnMÎÚµPž:°T¶m)e÷O>ÓÙçŽUû]ä2Û/Û® Þ›m˜ý4apª®³}³½›æóšä6iF:˧J8¯O^¿,›ý!²7¯K+÷Ÿ×?töñ„ËÌ2Çþ£";'X€y‘ÃÌQ÷ïä3µ}NÔãáéXôR¼FÜ£ ¾š€¶ 9Œ&kšzμTóÆóÏ?_¶oß^Ý>Ah–Kˆ–Û4!]†¦YèåÓ ò²|‚½&€K¼¹¾«…¿+ƒßÜwÖ1Á`†„€ ³î—_~yõ\óÜøý£Ÿú©²¸eKu›CeÿÉÏý\*&äm*HóÜ2$,Lß´y¾yU…óùçW÷ßÈmó³L–ÏvÌø°¤Â5û_ý½¿W5[]…ŸuËsÈzê<þýÊö)ÿMã+CúM—~[g=On=·,ÜôGÕ>’ý1Û:¡z¶ed~¦óZdÊþ˜mš}3ûV‚øT£g~n“0¸Y>ãY&Ë&äÍt^—ìóyœF¶knûÈ#¨n!0À(è8 €G5ƒ“í“6…Ì—4±›*Ýc Àx%¼LЕ 7•¯©ÈM€˜P-¡X³„£,ŸÐtyø™&tSM™Û'4mšém‚·¦b3k3/÷ŸûôñÚ„¿ ûúµ<ìÍøÊð7~òÇ~¬ºÌóÎz$üÍc^|ñÅÕsN3Ðy^ÿÍÏüLšæ'†/uÖñŸÿ/ÿKõÓí›ßüæj½›m—m“m—ç—ð7÷›mÑl«Õäv ›ePæö•í•õHùË_øBÕgqlêÌËp²³Ýó˜û×þ\¹ý7~ ºnZíX,å؉RŽw†ÎKS­wöÅì; wœgÛçukúbÎvÌþ” ߌçµËþ”íšý0¯W¶O–f<÷—m—û˾×.·Íu¹ÿ¦’¸ š³.Ì:Çþ£ `ìˆh²˜'ãi4h'?Àèù:lMð˜`*—UpºïŠ²sÛB9|t©ª¤Ü±¸PŽXª.·t®>¹TÊÁ›öWýØ&N@–à+![€%„m 3?!Yä1 %pk*'#óÒe^dùÜgªi3/a\æ%¤k¼„ ²L7Móf}ò|Ò,sªm‡åïßv[µ~¹ïþ¿ýoU°—uJX›m—à:žY·àÕ¼æ9ÿâ¾}ÕúFÖ?M 'N…iSÅ›ë³ý›ð1ã (³ "Ë7“Ûç¾³ír»ŒgûåúlÈk™uK€žûËöû¿ýý¿_­ÇáÎ:ì¬ÃÏXþÓÉ4=ÎܱP^Ù|nyåý¿_½VMÓâ s³Í›}-—"ûQ*¼³lóC„f{f[絉ì× û›}8Ûê]ïzWuÛÜÏvÍ>“ûʼ×çõI@à 7T×å5o$˜OpœÇËceÿOØ›ûÊ~‘×4÷mû¢ |0ÀD8ÉÌ“œ0ížØŸ«íåÀhùº–„y ëæ%tJˆ•ð*Á]F–I8•Ë„R `%¬jÍCÿjwùîóKå²sÊÑΡÑîm eqs)÷>y²œ³k¡júÀ+Ke©,”W?tO¦yÛ„”¹ïÏñ‰'ž¨î3¯O‚âåÍ'„L ™ë3?Û½©lÍt^ƒf[f;äúæ‡YßÜ6¯O–ÉuM ýÇüÇÕvÌc6ËGîû¿è¬ãÖÎí²|îsyðj¦! Þ¼ë¼rðƯTûQöÃ<ç<ÿìGÙ73ç™ç“éÇ{¬ q³&|ë[ßZ½7²ýs}®[OÞ{yeûçvÙVyo>ûì³Õãfòšem’ÈâÔÚÐ0ÀÄ8ÑÌUÀ B 0:þ®”à.Vª„Z:%lM`š€/aT‚§„~¹.— LV¥É⯹& =üo.,Û¶”ª¹çÎâÛ–òä¥* >w×Bu¤ºòøÉR6ßüÍ* Mð˜ÇÉ0-ÁZœåq2äþ¢eù,wÏ=÷TÁiÇûÙêqšP;Ï;Ë5ëû“?ñÕíSù½„¿±<ŽÜ.“ç—í’õhúM€˜yÑ·iJ:¯uÊ,—æ‰3uË>‘ñ¼Æ¹.ëœñÜWäuÉ>ñû¿ÿûU0™>Šs]#·‹À™Îíêoý­ªxkgÎʸßþƒ‘·Ÿ[ÊMwW£y²Oæ9D¶i^Ãff¼¹.ËF¶4¯ùéd™ì빟ü8 ¯U^“ÐyìÍõ]tQ5êÌ‹/¾¸zÌaË:¤*9Û!Û9á]Ö%A^žKÖ-Aß%—\R]&LÏëæªÿù?ûgeϲêÎ^àXGnŸ×'“uÈãçõËvÌvŠlïHhm™õËöÊöI™ù™Îe^æö‘×"·Éë‘3Ímç¹'\Ït‚ßåAqä>³Mþé?þÇÕô¸l( ^K^«ì7Ù¶Ùç›×¬ ñ³OûÛß.—^zi} ÚÃ÷€LŒ-À¼õ—wŸ«íæäÀhÌçÿÏ„v ìšjÌp©VLДp© {¨&$L€—*•ž™—ð/alÃŒgùÜ6óšøþàÊõ×__úåòÈ×¾P^yá‘râ•ËÛK¹è¬…²kkgùW—ÊÖ-¥~µ³‡–Êùg,T•Àß}×ÿYU'@κ}÷»ß­ß‹ M›jÇ„Š‘õK°˜ç•07Ï-C‚ÆL'(˺f½"ã Yó\h' ¦Ü¶Wî7÷Ÿíœ;ë”uÌõÏöÌzgHåræß{ï½Õ6N ÷Îw¾ó M9oÄjÕÀëÉvJ¸›í–¾i³žMŸ×»ÙîyÍó<"Û:Ï9Û4r]^£Tg^î#f^˦J<Ï;÷•í“Jìôy¼Zh=* €7užßÇî¯'Æ'Áoöüh!Û6¯C¶íå—_^½VßúÖ·ª°6ò=`#À#¨æÑ(¿¼û\m¿Tfùú0\óóÿ3U‚Mà›Ë“ ì$EšAN›ð/áh½Lç2S‚Ô„x îâ%ŒÍtÀ„S©XͲ /S½›P*óâ¾w•½ç,”'Kþ¦Ïß#ÇKyòÅôù›^Ê‹¯,• Ï\(ç±P~öd9ïï|· z³Ž<ð@yÓ›ÞôZè؆Y¦Ú4Ás/¶™×„Å™N€œõÊzæ¹åy¤Yç½{÷V÷7lYïf=ž6Í`g[ggâeò\êe;æ¹>òÈ#U@œmyõÕW÷Ü®§ŸûÊzeݬg;f²Í›í˜@7—"ûGžWçù7ÍXçye›ç>Sžç›JÕ¼.Ù÷rÿ ‚ó„fÅFƒà;nëì\ËÜþ™õ›•î;ž@pdŸN•t³Ý2ž}$Á|æ=øàƒUßÖ´• _`€‰VóFÌ rÒ¾{â€aiçÿÏv Ó¬%ÄM™jÁsMp—êÚ4#›yiz8ó²LB¼„t ìr™ŽTÝ&̦€êå¶ {âå6ÒlqúÌÍx‚ä‹þß+;Ê+Ç–Ê–ÍݦžŸxq©l[,Uð™;ºýÿnïLßûÄRÕð÷ý·OWÁác=V­CÄŽy¹Ìºä¾Ó·lÖ¯ 3¾|]Š*è])ÁoBélǸٶYßlçl§„©Y¿<Ÿ<‡zM8œm–õÏ:gû&Ôû™ÿò¿¬ï¹«ßð7êÊÛ,UOwYÿìYç„¿Yÿ¬s¦öf½›@8á{žW³þy¾ÍsÏóÊëuÅW”‡z¨º4ÓÝóÙòZ&ÔÌí³½„Àýl—Õî¯ï¸óüË-×ã•ý*}1g;¥º>¯Oöýü !Û00m•6tp@oºí¾ÀXtûUƒÙÜýœ*ZºFÂÄr …¢¦5— î¸eH(šªÝy©®MH— 7ÁR*xs]* S雦yÓq*3ÓDoš*Îò™ÿ;¿ó;U•€6¡_–Kø›e³N _9ZÊs//•¯”òÔ‹Kåñ„¿[:ë½m¡ìÚVÊÓ—Êæ…Rõ |õ åú Ê‘_¼°ªŠMÀ› +ϱ©:MÈš¦…s]äùdf~ÂÈÌK 8®ð7f¶s¶Yª\#¡zÖ+Á\¶WÒl£„çÿ;§Z>{úN°—*ælÇl¿¦ îj‚Εg?aiÖ=²Í³ï¤¢7²m³îÙ¾r}Ââ¼>ÙŸ² òÚç9äº4}ûj–Ë2 }sÙ4e³ß΄mçÔ#ã—× ïÁìûy¯æuȾ•×!ï5ÚlýJvÞH À¥‚F€Ç ìC¼.AP¶„nM(”P-äæºl R‘ÙT¤&ðM(™À.á\‚Ó„©Ö¼ï¾ûªªÔ¦ª³êË·syÏ=÷T!oš*Nà÷®w½« ,d¦º3Ah‚¼ÿà_–Ç¿ýË«û:·ÛÜYm¥¼xx©<óÒR¹ìì…ª¯ß‹ K9wW·Â¸ûw6åù§23f‚ßT!§z9ágÂë„Ù™—ísë‡>T^ê¼>?öáWAh¶w½·½ímõ½½Q¿Õ¿§ÓÏý%üÍYß y> ³_e<ûY$Nõyæglú„ξ•},C–ÏrÙ·²¯æ~²/eÈr© nœn=­þÝ#Ï×#ã—×!Û¿‘í˜ý.Û0?, í„ÀýL”>L€y5˧8€i@6ª%Ji6&HËx¢¦ú´©ÌêàTé&œËuôGTÍËò ™â¥z6áîE]T…r÷Þ{o`&¬¼êª«ªÇÈc¦ià}Y>áòËÿûÙeñOþûrõ=?QöìèöýûÇ–Êg,”+Ïkç1PÏl«¡y-RÀ7¡\Ž&hÏpìøñjÛÑy}²ß÷}ßk¯echaå*†&¯'û`Sñ›ý/ûHö¯ô/œªÕlŸ<÷Œg~öÍl·å†±¾CÝžéx‚²ïD¶Y~xm›÷]¶3mç5è‡OM&@SÐ J0Àùä“U¨›Jàu×\sMÕ¼sš!nª†³l*{s_ ~3?A^BßT¦Ùå\æ1Ò”ô‡¿YN,•ªÙ焾‰{;WWM;·U¶A¶UBò ©dM3½ åjf^šÒζÌëö?ýë]m³ÿâïÿýrÝu×Õ÷òºAšmž6yžMSÏù¡A¶GBðÌKSãOÙÙ~™—ý´ 8Ç¡ïþc‚ÀË¥Š?ï[•¿óÆw€^ €˜_Þ”}`^%dL˜–À,ã !Ê&øMЖ@6M17U•qÿý÷—o}ë[U(÷o|£šwÅW”n¸¡j~8Á[š,Nà›À2AoÂË™¹ÿ413— 2sÿ¹¯T'À;òoÎ/÷üâJY*eó¦R®»p>Z<ÉöÍvHÿ¶ÙVÙ6 3̧¹ìò—_~yU5¦³¢ÿÛ_ýÕªùìQ™–Ð8Õ½Ùg²Oæy7?DÈt4Apö©ìo¹>ÁæJƒ<Ÿ¡WSO¸x¹¼¿™G¾ôB Àùòô'ÁkªwS=™Ð6ao¤æô»šªà„ ~øá*\KÏ ØÞò–·T·Iµo*}#Áe‚ÉôE›f *%¬l*53d™ ¹ÏÜ‚½¿¹ÿ4Ù›ÇHSÏ—Ÿ;T¶KBËÁ ãš'øMeuª¢¿ÿû¿¿Ú> ß3¤J{ØV 9šN:ξ”ý2ûLö£&èM`žêÕ4éT约ùò©6%ÀÌ;ß#NG 0qú€sò`¥â¶ ¬¥r7ÓiÂ9Óéƒ6amúùMf.¯½öÚ*¬=pà@Ê% L0—à7nßô÷›jß,“þSÑšÀ2Ëå1Ópî?¡]‚ºT¶Rªê넽 ÄS…¾nßùÎwVÛ3á{‚м6ñ/þÅú«¤Zuè•®C ·i.<ûgö³ì?ÙŸ2/ûfª~óc‚TgßK€>jþ9¦¨˜y7ü“´‰€ à1(ûÀv_yui{ÕGnÖ/ýʤÂßY«÷vzâ‰'ª}6ûTÂßìc o#ûbö‹4sÞTÿfÿËõY6Cæ§z8ûr®Ëý2ûlÂâæºf¿ÎøOþÍ¿ÙÙ»ÇÔÛ:÷ùo?ÿùòüÿã‚òìKKeï9h*zSç3ãc÷×0­T¨`J©f£|Íh»\©nL•ãsÏ=W» i›jß_i67ó›à7ZÂâ„o ÙRyÙ„Ã Ïþæ>¸5Íî6UšyœÜ.ÁïÓ—ª~}_8TÊs/—òÝç—ªð7ýÿn^(åȱRxz©|íá“UÓЙŸfŸ}ú`ù“?ù“*$½øâ‹ëg2|*{3‰í”À6û[dÿËþÕ„¹Ù÷š%d_=ãŒ3ªŠÞLgÌx†fÙìŸÙ/³/gßÌr_r}Þë曫#léìÿ©>ÒÙÿË¿{SU|Ñž…òø‹¨ ÚzF=Ó̃TL¨oÔ믷}†²’_þœÞìÿÿL –ì{ßû^Ü&XKvÉ%—ToὩ¼ô·šð8ÁZg^nÓ„j ßRM™ŠÍôÓzéü+U_¿ÇN.•Ã¯vþÃœ(e÷¶ôû[ÊÎÎ&ܵm¡¼ph©œÓYæ»ÏŸ,çîz½²òÄR7.·>RM'œKÕæ(L:ü• àIm§^x¡zý³O¦™æ샗_~yõÃ…ì¿Ù/ÒÜx.g~–É2ž!oöÛüP!ÕÄÙo³ÌóÏ?_íÿÙwóþñÇ?^U7û\îç_ü¥¯v¦K¹pO·J½ §º}Sg_Í:»z%»nöáÌÏíT3;|æ›`ª/V×Ë—wŸ¡ô+Í!v›D˜_íùÿ™ð«©øMS¹ †3/AYæ%$Keeæ% K8œË Y&áZn ‘SuÙ„Ç›ö]ZURVXG°8Ö™N3Й¿es74û“ý'ËÞs7•3·w—yM’¸7“ÚNùBöѸùaÁÛÞö¶jÿKÓÙך0dßLXœ08§Z8•ãM¸›ý2÷•Û$<β'8N(üwâ'ºûüâbÕtÆOv†;oº«ÚOéöM½µ3,néV®g¿Þ¾Øyßtè,ZíóÎ]”w–ÙqNYøàW«Häý”õÎãæ2‘¦‡˜_`€©#ÀXÝé¾¼ûüdNóªÿ?SÕ›`,ÁY²U ê&4KÈÖU ®š+aÜùçŸ_ÈMh¼éó××ÁX·b2áYn™¾·mé†fTS>sp©ü'W¬Rá;âxe¨Ù„±ã ;Àk»çž{Ê\PUä>öØcUõnöµ„½Mˆ›}õÜsÏ­öÇì«MuošsÎ>™å³Oç‡ Íšý4ûõ3ÏO°›@÷ /¬BÝÌKÕo~¬p8¿ [s]‚ÝFóã…„Ç ^3þ·?úÑêºM©hï Í~ñ¯ÿÇ*ð}õX)/IÓåÝ}8p~Ìй‹ªëØÑùhÈþž!ÍœgŸÏ¾~ds·8v*™›~Œ³î ‚›uiªéaòÛói£R˜&9A½Ø…‘Ê¾Ö Ì›„rçw^¹ü?¾·\òÍ¿_.Ý ~'þƬTæ¶YªÐæ& Mh›4Í”GásÎ9給ÇSœ"$øMÕpÂÝ4°µÑTþ6Á™ÎíNÑ™—ÇÈí^<Ümê9•êçì*åÌ Uà›êÞÁi z÷öRÎÚÙ } Ÿ¹£s}g×~ùHš^Jž\ÑYǬσ>X­wáZ…Ç ‚s›„¯MóÏ —À~ê'~¢:bÎЄ±›;óÓü󞥼ph©úáB‚ߌ'§­*{–rð•nóÏ©ú}ùÕnÕûáÎe楹èübÏÖnßÙ °ÿ÷µ~‘çͺ“ã?13(§´`\²¿%ö〹ñ¥÷Õ#tÝmëów=YFH<wß}wùæ7¿Y.¿üòzÎëš*ßµé7w÷îÝUUo߸ÍuÏ>ûlÕÜóÞ½{«þ~Ó tšYNÜ‹Ûo½µ{£;>ð{)®úùMÓÏ[6•rüd)ßyæd9ðÊRyþÐRyòÀRyñðRÙÿ½ÎôËÝñ丩úÝvþ[Êâåµük_«š«N•üUW]U­¦§º9Ï/ë L–€¤)h&%!° h“´@ÔËfgÓ9h†–ûÊí¥¼po=±AïøD=2Ö z­G…íFjÝî½÷Þò'ò'Uÿ·×\sMÕ'îþýûëk»R)›€4Í;' N5o.›ÊÝÌOˆš&¢¯¿þújþÁƒ«Ûdþ¶mÛê{®]Cš·\¼©jÚüì] åªó»Í™gú‚3Ê9y‡^íì/—Êw®ûtyôÍÿ´ ¯S‰œ :ý§Ùç4ÿœuM }ÆgTUÊÀd €˜QB8&© ‚í‡À¼Xê éÓ3CáLgh™„¿¹žØ WþÆFªW:]50½I›¦šóš¤ÏÞ4áÜTõ6ì&;~›~~›`7Í4g~ª~£i^ùÛßþvuß©µÅM¥l^¥‘ÝU¼xÏBµ^xaUá›p:Ï;Ï+aud}óüR­ÜÌ&G 0•rB€ÓS Ì4H¼¥; 07Rœ¡Eß]¾ñ/?úízdt†þ.7!ð(+“¯¸âŠªïÛ4ßœ!¡nª`S¹ÛH•lÓTÉ&ÜÍõ©šÍ¼T 'Ì„Àƒ¬g”U¬a„ÀãXÏIK3Ï©„Müä“OV¡í9çœSUÀ¾üòË婧žª*f”f~ª‚SœJß4›œÐ8ãY.apBá¦8Í(§àÜWæl„`"T¬ —˜i‘Øþ°1«…†à®T¬ðÅw”²ïŠî0¨·ýÃzdñÄeóæÍUÀ›Û'HN œåöïß_.¾øâ*@NÐiútÖ ~}À7~®” ÞYO@8&æ“ `€‘H0áe0»š÷°áôôyÞC…pÞc‡JyùÑRž¹«”/½¯¾¢ã+·×#³!¡j30Z?üpyä‘GªJÜôÑ›p6C*ySÅ»gÏžª¹æ¹©úM(|ÁTÍ6'N@œ 7Ëæ6 rò>ôÐCÕôW\Q-ûè£ý²#ýÿæ¾r»4-ËÀ§3ôð7„¿Ð `€‘JXlÌÊàqœ³£ ‚O'~áÞnEp‚áÇ¿\_ÁFŒ"võïédÝS›p6•»™N…nò&¬MÕnšsNEoBÚ—^z©ªð=ï¼óª7óR±›ûHo.¯¼òÊj~î+ÓO œûù³öÏVÕÄ©üMß¿é#8·Ëzä1,7òzNþnˆ¦Ÿ 54 °&'}€a;M8,ãx|zåTÒ±îè´Ùwm=2FB³Ê´5Q+›^¹^[¶l©BØô½›*ÞFª}Äf^Âà4Õœfž§)çmÛ¶UýûfHÅnúNx›Jߦ©çôå›7¡o*}sŸ{÷î-G­Âææqî¹çžª9è¬kî'Ap–[O/áíË´’ão`>©ón¡34¶§ OÂ_Ö±Z(ÊÝTÛ&èM€ ~Óôs*t›êÞŹ>Í<ïر£ ŒÓ4t*„sûT §ç,—€7CšyNøûØcUáq*}#÷“ªà½é[8aqBè4ùœ*á‰Tþ­#+áódy ›aZúŽe0 òþζóÏ?¿ ~ô&ÄM5n¦Ó¼ógœQ5ãœÀ7U¾Y&ÁoÂÛ„¶ Sù{á…VÍ@'Ì}öÙg«°8·Í²¹Ýƒ>XUûÞpà ÕüÁ¹ÿ,›¦¤ ¿õ­o­Bç{ï½·j:!ò 6Tý ´Š& ÖäÄ,0jš¤£ÍO·×6½ÿÿ,åwÿA=13þ6Õ¢ãè[wÚš€^«ùç„´ mû†RÞ÷¥zbº- GOsܬOúçMÕm*ySi›`6Õ¸©ØM0œ 6Í;§R7næ'Nh›Êß4 æšÞ¦’7UÀiÚ9}ÿ¦YçÜWd:.?ðÀUóÐO<ñDyûÛß^ÝG#o3äqOçtð†«~oy°”…Á«Žaz9Îæ“& &.¡™àŒY”&›ý×>Ìœ8ör)_¹½ž˜³ÔwlÖu˜ë›p7¼ uú&tM8›~ws]BÝT§J÷©§žªâ½{÷–‹.º¨Z>}ó¦Z7ü&¾öÚk«þz³lÓtšrN(i&úî»ï.—]vYFç6‘å󸹟„ÁÃ"ü€VR ­“/óÞ¯Ö°~Ùç0 ÍW´)k^NáXy¾Íyðr‹»JÙvNçr÷TV¯Ž¢x­æ–×Òk ¹Ñu]¹>©èMSιLóÍ©ìM › 8•À cÞ¦Š7Áoštβi²9Í9§ 8}'¬Mpœà6•Ä¹MªyÓÔs3dþã?^Ý.¡snûo|£¼å-oÙPÿ¾½l« WÿêÇš¹ ˜O`h'bÀ@»8iÅ´Icb*ÆæÛ”À_ù©Rÿ¿ê‰)0¥aÚZâ0ƒà~à~ªY‡ÿéŸþiÕ s*vÂ&´MnÜ\¦ç¦ èT§š7óSÍûæ7¿¹ªÚ}饗^ s»<æ¤êæM›ªu}ì±ÇªÛ¦¢8arúúÍý%î¥Òw¥‘†¿!f.8–æ“ZGhÙ5Œƒ;Û˜&N^1IË[ê?Ä m&O²âw=S¦2î·ú7V[ŸaÒ s¸6¾ùÍo–k®¹æµà6Í@§B7¡nªzSíÛ¿© Noïܹ³ rìf¹»¹Û>úÑr¬3?ãÛ¶n-Ç;ãÿôŸÿój¹w¾óÕýeÈcnÔÈÃß3CóI ­#´|Ý x¶%0mœÀbÜ^l<À ­À«š0m½PqÕ¶½„¿£”æ–W†®iŠ9M=§/߬_*ySí{ÁT!múý=tèP'ôM¿¿¹|á…ª*àÆyŽ÷ǼZ&ð+y‡_y¥ìÚ¹³ºïÿñSŸªî7ÍGgÒðgœQ]·Q§ €…¿Ð+ÇÏÀ|Cë-_—ƒ|ÄÙ–À´r"‹qp<ÌZ&O"üÝ}Y)ü§ì Ô†o¤úw¶&´M_¾éãw¹³ sSÝ{î¹çV×'N œ€7}ýæy¤‰è„»™Îò ’3/U¿7½ï}U•oઊøèѲ{×®òï>ÿùòâ‹/Vó²\Seœuĺ¯Ñ~½óö;TO @ÌÜpÜ Ì'0´Ž“4§:ÞNvGûf[ÓîD=À09¦8¡>Éð7Z7N[aÚC<-ðŽ;ªp6· ƒ÷Ýw_5o×®]Õå§â7amšuμíÛ·WÁï£>ZÍKß½©þØüH9øÒKÕ²ç2>sçUЛÇͲ•ÓýZ7ü´ê7¿Ì00ŸÀÐ:NÖ¼ÑFn[³ÄÉ-6Êq/1¤ÏœimÎ9n½¿éÅl¿Ö × §!üM¿ mî¦éå4眦©ÎM ›fœOžÿüó«€6Í8§¢wëÖ­U³Î©Ü]î‰'žx-ÜM%pS œà6¿Ï<óLçv¹.¿GÓgpçú,{êjâ\÷‹wÞYʺáo^‡}WÔS3wóI ­ãDÎÚÀÀòÇÝñžµë}Õ„‘½Àã Sé›Ð7ýý&¼MóÍ{öì©šg^^ýÛ„¿YÏć®–ÏrO=õTU¼{÷îê¶iÖùýù/—cÇw›{Þ´©yõÕrvçºÿáSŸªîwØÕ¾+­Y…=ŒfŸ—3wóI ­ãdÎÚÀÀ¼I@ã+ Ç· Ë*𬄼k髹çÕÌ×ûkœÕ¿guVyá…ª6Õ»=ôPu™à6n$ÔMðe—]VMÇ“O>ùZ0|ðàÁjHSÐ }SAœ8UÀiú¯ÿЕ=gžYUÇâ–-UÅo*ó£ ×­üvø`掘O`h'uÖ×ÏAŸm ´… xþl®’»?YDçóäñÿPÊžk:—ÿW=oF ü6æ7ÞHø»^è¹Ü–-[Ê¿¹óÎ*Þ¹sgÕ‡o*€SÑ›°6¡o.â¦øÜsÏ­oYªeÓ'ð¶mÛªø•W^)—^ziê¦8Ó©"þÎw¾SþçŸû¹²³sûÆ/üʯŒ¼â7Æþ†˜¹#æ“ZGhyz½øÙ–@Ûäë_ª›| l·„uS¶0ˆaõ9:ÍZ/×kÊy#Æ—……ò¿þ›S&ôM5p8p  }S{ÆgTÕ¼‹‹§~î}÷»ß­®ËíÒt®O¨›ð7ÕÁ™—øçþÉ?);¶o¯îoËæÍCßVk™Hø`掘O›êKZ¯ ýЭ]òz.„¿lPßåCÛ -ü~ {]Oc}ÿö²Ö}ü&à=û쳫êÝT§ê7ÙgžY»<ð@}ƒ®Ĺ.ˤ¿à¿OóÑú§Z50ùÏü™?Svuî{Sçú„¿ã³ú±üŽ4ü­øÌÀÐ:Næõ¦—_ÿÙ–ÀXUò^}õÕåᇮ*ßõ®wÕK–òÔSOU¾ ÓðÑ£GË]wÝU…ÅoyË[ªûyöÙg«&¡ÿÎýX}«áWJ¯jßåößøzâu£~;Tÿ2—ëóI00§œ0èÊq‘c£Ùáõ¢O©äþžêC_­GÚ-aæZæ°ªÇ)ï‹/¾X¹,W\qEUé»wïÞrùå—×Ku¥ÏßT 'NðïüÎïTañ÷}ß÷UÕ¿©(Îí–‡¿cqô€ð ÀÐ:Nõg½_Ú–À¼É×C}O'ǦôišpÞˆ‘5û<ÝïÑaVOª¸‘jÞôë»ÒòõúÖ·¾U5ÊàHÿ¾×^{m5žæ cå6GõïZ¯ÃÈà+´”wÿËzæ ``> €¡uœêO>uGßÀ¶æf“•¾|»A¼Á7~¾”—÷wÇwïí^fkiŸ¿³ñýqµrÖàÕ,_§¸÷Þ{ËÎ;Ë–-[Ê+¯¼R¶mÛV'üT<‘ðWÕ/t8žæ“ZGhÙ¿µmK€.'ÎÆËq(+4Õ¼ uO^K'»—œÞösK9òüˆÃߘ­÷íò0² p4ëµÚ:¥¸ ~ƒlƒ~ a’ÇóI ­ãdQÿÀ§—À)ÍC3ZŽAYæÀý¥üæÕlÈȃßÆ|¼wWVÚNSܯqÀk…¿1²Xø Ë€ùTÿtÖ“¯ 84I<Ù¶Â_j‡ŸìVý ûw˽õ£Ö–ðw”„¿À¤¨†ÖqÒhcVû5 m °¾|<Öeƒs²LÓÔ37¶Šß•æã½<­Í?÷kÕ¿c ÓÔùMw×À©TóI0@ʼn7€þ¥8ÇQ›«)z•mÖ Ì½„¾ÍSleóϬmì•¿Â_`ÀÐ:N" fù¯mK€þ¨°èã̹óÅw”räùz‚¡¹ä/—òÞÕ?ð@)¿ùÃ*€×Ð’ƒT»êÿ·7# 5í äø˜O`h'“×Ú–ãDÛúg¶žjÞù蟖ò«¶ž¨mZ,ådÝÔ|sýÙ×—ò¾_ïΛÓû¾VØ9hóÏ«U¯u?½T@*þÂ4r\ Ì'0´Ž“IÑ ‹ÝQ6èD=p*Çì­'ÌĪxÑîxÕ¿½„ºýF<Žðwà&Ÿ…¿0 00Ÿô °*á/ÀàÒ7pB'-öÌ×NíÇWø;˜™ §Û°›8$xm«‘„¿ ~…¿À €A0-Ó„½_¾¹žÁi¥9ç&àÍåu?ÑoäzF"!ð°ƒ`ºFþ @ÐÐ:Nª0íò5´îÓs.9fŸiw^UÊ®KKyy=ƒuÍMEïô¼¯wíÚU^xá…²¸¸Xê¹7ŒæŸ£Ÿ& ó§[~Zš€^-þÂ4Ñ40ŸÀÐ:N&0+šxÞ¾–:fŸYšwîÝâ®R~ès¥ì¹¦žÑvÓ÷¾>~üxyå•WÊgœQ6oÞ\NœØXŸì˃ØA›f?À“nŠZå/Ì 00Ÿ4 À„¤¹×-Ë.a 8¢oß~¥ê÷#¦õój*> ò>çfÜW3}ïë¾gžyfUœàwëÖ­U³Ð_ÿú×Ëå—_^/uz*€ßH³Ï0KTóI ­ãd0¥„¿Ã×ÊÆ'ëËãõe9fŸjÓøY5uïuðMçûú‚ .(/¾øbyòÉ'Ëž={ª¦  ?ñÄåâ‹/®—Zß4ÀÓþ†¦•˜O`h'“€)"ôNS¢9_S›¡ÕÀŽÙÇn>‹®»­”w|¢ž˜và7šÎ÷õùçŸ_¾ûÝï–-[¶TCúNœþ€;VvîÜY…Â뙶¸Uá¯àÆ@ Ì'}Ñ€eåÀtšú×faÙ¾×'`U³ôY”àwfÂ_fÉ¡C‡ªfŸònß¾½ºÌ¼mÛ¶Uó8P/9“ oGe Ê_€Q ­£š“GJùÕëë fÚÿv){®­'¦MS iºi"z–9f‹i~gºòOðMçûúÈ‘#U•ïÑ£G«Šßô œêßô œéÂi"ú¬³Îª*ƒW3Ì à¤ ¸UÕ¿½·”Í;ê `tTóI ­ãd0bŸk)ÇÕ´ÚTT˃àYïØ1ûHMkð{É¥¼÷ŽzbÖ €ßh:ß× u>\vïÞ]¿/¿ür5¾k×®òÜsÏUp¦yä‘5û^ØN2žT¬ég˜e``>i8½åM© çGóšo¡«KeZó¶iºi&jÙ_פÒ7aï¦M›ª°÷ì³Ï®æ¥è;vTó €¡uœL0©åÆÏ•rÁ;ë‰9vßgJyywn}¸ógÒÕ·Í×Ù\&Îå,|ÅuÌ>4³TñÛºÐÇ õSM÷û:MA§8aoÂàL7}þ6aé9çœS^yå•òì³Ï®Æ.Ÿ7ª¦˜W{ܵ+ÕÌYßôa¼eË–²mÛ¶ê2¡vÂíŒgHÿÇkõo|:«†¿ûömü³'ëqKþããÿ0ŸÀÐ:N&=˜–ÐäìJyëÏ”²÷ýõ Þ¨ó•í+?YʱƒÝÉgÿ¨{¹ž¥1VþçÓSðú5_k—‡ÀÓüU×1ûP'Ì õSMÿû:AèÖ­[˱cǪ*ÙÁ‘À4RœJÚ„Äi"ú‚ .¨æ7Æ÷*MV§?ã¬GÂÞLïÙ³§jÞú©§žªžkÖ?•Îéë8áw¿†þnêì#»¿žÆÇÿ+`> €¡uœLNcR¡ÉösKÙ²«”ÃO–ò‘o”²ùõæ/½¯”î­'Fì’KyïõÄ$-—ÓÆ1û@f)øV†¿y_ëŽR›÷u*f”FÂÒ„Á‘8pà >úhÕ¼òòæ WVçN2Þ½{wÙ¿ðf<pžÏ /¼P Yï‹/¾¸š÷È#”'Ÿ|²\sÍ5õ­OoèáoœÿÎRþêçê `|ÀÀ|Cë8™¬pàþR~ó‡ê‰ ÑßÝdL2(›èkÞ¿M%ô´õ옽oKÇK¹óêzbF´úsOüF³õ¾NÓÈiö9•²‘¾3/éÁƒ«P5Í&'L]üNCpÂÝ4[*åÀiº:•ÍY÷TaöOßv[½t)/w–KóпôÙÏöÜôÐûý ÇB0!``> €¡uœLjwýìäúu’s:M"žØ¾Ð|ÕÆf¡³÷eÖÂß¹øüË{I|ªÙz_§ÙäTú¦ê7Ái.'>묳ª&“§r6Ãu×]WÝnUÀ©XnÖ/Í>§29Ã3Ï €¡uœL€¹öù·–rìP=1‚ßö8z ”/¼½ž±©‚'Ù,´cöÓš•à÷Ã_＜ÝþTçKÞC*€O5{ïëTûîر£ XSa¹<|øpª&lMõ綾^Z]×è%N0›ûIÀœës›Ü_î7ÁíüðÃUX`:qBß矾|ï{ß+W\qEÕ$tnÿâ‹/–[nº©ª >»îËx½xµð7ÀÌ'`h!'“`.3,qsNt¾*‘±ïKÍ×ßå}Oâ+±cö5ÍRÅï\ €ßhöÞ×M𛦠¨&(ÍjÙ· ‡ª6ámäºHx»\BÞ,“ªâÜošjÎ}æþøFÆ› ãŒçþÓ4sšx^)nn—8Í:§©ç³Ï>»ºßÜ+Âi:}'(ÎÊáÜúNÀýcþpõ˜;;Ï1ú €…¿Ì&0`h!'“`.#4qsþŒ+Œ›HÜ 1î Ø1ûkN)åÿûãm²~>ky¿€O5›ïë° ZS­ÛHpš 8×%TM؛鄱ÑÇ e/¼ðB=VªûÊ2Í}DÂàH@œà¶ifú²Ë.«æ¯'apBÞÜ.ÁtÖ)·}饗ª@:¡o¦S]Ü\Ÿñ[ôGËÑÎãd™Àª™`ÀÐBN&À\¹óªR–†Ü”íösK¹éîzj£ƒw•ò‘{ê‰qiBàækñ¸‚`Çì•Yªö÷|º”½ï¯'æ]Þ'àSÍæû:Õ± V›ÊÙ&¨Íxæ7MAg~‚âTÖ&`M¨zá…VË6ÒTs‚Øܶ asyŒÌÏíЦÉæ4Ó¼òöýJŸ¿ ¡X§8ëvñÅ—Ý»wWó~ä?ÿÏËÉÎznÙ¼¹ìè<öZðÈÂß+´”wÿËzÆI €ZÈÉ$˜Ÿ½¦”“C<ÿÿ]Êžkë XÅ8»³o¨G:Þ÷¥zd”–‡¿‘é!ÿ¨â æô˜}Ößå6-–ò±ê À«™Ý÷uÂÚ„³©¨MXÛHÅoÓDt*p›€8ájn“e—WdzÏ>ûÚõMXœ*à¿ ’Wö'<,?þxÕïp‚æ4ýÑù‘rèðárÆîÝe¡s}Öãßôþªúe¢À€¡…ÀÐ*û«{ùÊÓ¥=ÐoBÚßýéîå nü\)¼³ž€>#ÐËIôå_‹Ot†Ä£¬žÃcöYC˜³ éf÷} 7aïÎ;«ÊÞfH€›Ë„ÃiÎ9!j–IÀ›ÊÞ\f:}7Ä&ðMPœ06ã Žþ® ‹‡-ý8p Üþc?VŽtÖ}Wgݶ..VÏaSg]V«Iõ¯c+&Îç3`h!0´Æ(''¦Qî«—ÜXÊ{ï¨'Fmù×㦠xA𜳧ß_½¾ž˜1·Ð9¾«/€IIHRLˆQWË ¦œìn†ô“:L¹”Ï¿µ3²ü=2*©üÍÍ×åLûê¼!³þÆÉãõ«åûqKš 6¾©–M3Î;vì¨àf^ÆS]ûÜsÏUÓ ˆÓïnš[^)k–O³Ï½„¿ßüæ7Ë·¾õ­òï|§<ðÀÕåóÏ?_­Çc=V¾ò•¯”?üÃ?,÷ß}‹Õý•ÿô?­Ö7áoúŽµÂ_€¶S ­£fÂWnï[£°°©3lîœYϨÝtw=c0óÍC7_•7aà°è6³w¶Ï¾+ëñ£j¯yèøT³ù¾^Yû ¿ò+Ueo‚ÝTð>õÔS墋.z­?ß]»vUÀ VÓüs†Tç²/¾øbÙ¿UUœ8UÆ/¼ðB5ð8÷Ÿ>‰_ýõUøÜTgH¿+½ÿ/ý¥r¬s?iö9M>ÇjÍ>ÇHª?zo)›wÔ0I*€C €`ê"kBßØÜ9ؼ½”-»JYÌpF·Ég˜„QÁ#횯Ìà ‚[xÌ~÷'»—÷}¦{9k„¿}¿Ñl½¯× @ëÐ4Õ·©¾âŠ+ªöàÁƒU¥pÂׄđ鄹i"úÒK/­æ­çßøÆkÄ “/¸à‚*ø}衇Ê%—\RöìÙS¿ œB§ø]ïzW}ëR…Íyìåý7>ðWþÊkm8ÄÿñþCyúé§ë©×­÷ÜCÌì †ÀTUÖ¿ }#ÁïŽ º¡ïâ™CƒÎðîOu¯ƒIÕû`,áÝò¯Î':à !pKŽÙÇQå=*ß €ß¨=p#Ap*rŸ|òÉò§ú§åÝï~wÊ6¡oªvÓÌs‚Ü,÷Ö·¦¹þµ¥â÷‰'ž(W^yj+ uÿÎÿø)áírkUñÆò ø¯ÿ•¿Ru%Žå!ðÈÂßÐw8SC €ZH Si”U¿Ë+~w^(øeºµ"NL±<üê÷kõŒ³ ~ç@~äp²34ý_g_ÏåzN5;ïë^ÂßÆzáëJ«UåÆjU¸+õR‘¼–µ·Ñ<þHÃß½³ ì©'`ÒÀ€¡…À0qKÇK¹óêzbDþnÙñzð«©gfÕ°ÃıÁŽ5Ap?_­gì˜}¯ vš}8Á¯€·³õ¾uÜKè»ÒZëÔÏã¯gd°Ï+¦Ž0´ÆjÁȦÅnÕïGî©g@ ó½4¶“ñ 梧 x†ŽÙ…¿âäýì™Ýï⽆Á½± 7ÒzÏqÃÁï%7–ò}ÿM){®­gÀ´ð?0´ÆæΫJYj 1i*…¿´ÑL†À|dh¾b¯÷U{Ù³¾ ß©ê[¸³gö¿‹+¤ xTÖ¬.¤Ùç[îüY«çb˜$0`h!0ŒÔ]?[Ê¡ÇJyækõŒŠ0w:_O—êàëäñR~õúîø0¥¹ô‘þ€"_±;ë^Yëëö”³ß÷™Rîþd=1!sõ9ç„|»ÍØwñ¯t>co¨'Nuûoü@=¶¾SÂÓw|¢”ën«'6nyP;Ž¦Ÿ7¿çÓ¥ì}=ÓÆÿÀÐB`©qTÇ ~áu£èSû–;߆·Ô£ÒÁ«}åžà1û$*|—»ò¦RÞù?vû/Ÿ NÂÏû.Þãgk¯apcÐÐvØðºáoî#Ÿ‰ŽÓ˜zþ÷Ðí¨èŨ‚“"©šÉà¤"œ*Aíÿv=1$ =Ü_OŒJšÍWî\NA¡ùüštø› ÁwjÂßœxo˜mýVÈ®¸ŽÛi×¥ßÏÄ£9Nf„ `hÀ0tG”òÔ,åwºž1dN&Âé*¼¼äÆRÞ{G=1*Ë+s9æcöý¿5ºÏ¯ÓÉ\^¼·”÷}©žÑfßùÖÎ à•znúƒTÊž7Oô½?ô¦Ÿ¯13ü?@ -$€¡GÅœŠpz_¾¹”ê¾{ê^ŽÒÐß—';C¾~çr13FcÒU¾Ëµú³-¯c‚}hÌGÜè%¾ãÃ÷•rîŸÃlÞhèáïÛþagøGõL;0`h©ôg¦…wØ°Q( ÿÑ·óœTgPÍBz=×ô'~¢š\ýø‡Ÿ*å×þ³îø€z ‚o;>¶xèտᘎ™â`h9!0ôe˜uþzç_ñžz !ðäµ>É)“cÝQh“¥¥Ü9XðJ=ÁÿÓÇ»#翳{¹ÜRçý¶PÔ/ïï^6vï­GÖ·^ø}Àcé#FA €æ@¾<°1hƒ_ÿÁ7žh„*˜¼io"yVÍÍç[B_§Kh¹»ÿY)÷ýr=1<§ ‚7Z…ÛKÀ¼ž¾×ñ3K €æ„J`X×°‚"' aº‡cî>Ûœ8gÎ컶žAÃÚQ3?ü@ s"ÀM?CÀk†½ã¥\w[=L !poæ:èHµo¾/åôˆS$Ì¡UÇ´Áª:3³À€aŽ¨€S¨ú…ù# >ÕÜ~~5§ANv†ÝQ`$•ÀqÁm^ú5›KùØ=õD8—Â, †9âK ¼Fø T:_‡÷]YÏŸY9’Š_`]# ƒ ‚wOçÖûë‘•œOaրÜñ¥€96ÌÊ?! ´Û,W øëÃþ=õ§JÅïñî(УüPæºz¼ÅÖ Χ0KÀ€aùÒÀV˜såM¥¼ûSõ³!¡oÂ_ ·ßú·ë±1TßNš˜VPʦúÚiÿoÕ#JÕï4…¿wýl=À©râ{ù ü…~$ø]þ¶Ú•ê!ü 0[TÃ\òËUæ@Û›|N¬`! lÔz¡ïÌW÷òöÂùfÿ…€aNm®h¡ÃO–òkï®'†`ZûûÔœè†aX+¾ãG¿^ʱCõÔ”ùðÖ#+lÞVʦÅRN뎗…îü¡3íü_@ sÌZjX•¿oûxw`Š8© £°fø;Í•¿C«êÝçT˜fþW †9ç -2Ì&Ÿßñ‰R®»­ž`òœÌ†Q™ªðwû¹¥ÜtW=1ÍœOašùŸ @)›êK˜]_¾¹ÐBçð8M> ¦À‰Î“ØNdèLMø›à7ÃEÿY=cÚù\¦› `˜{~µ ÀŒ»ï3¥ÜýÉzbÓÚ×/ÀܬÀ¨­üÆØÃßôÕ»õÌRvœ_Ê΋Ky￪¯˜Ω0ü@ T|a`FÝõ³¥<üÅzbªfï®'˜Ç:ƒÓ0 SþFàÍ[KÙvN)üõÌY✠ÓF €¨ø²ÀŒ¹óªR–NÖPõ 0ÅœÀ†aX/ð]n"áo¤ Ž-;JÙ}Y)ïûõzæ,qN…qð?€þ€š/,̈}WÔ#tåM¥¼ûSõDËœnÛœ}C)ïûR=0 Žw†!üàæÐT¿·Þ_Ôö]Û½Lðyo/åÆ}Ýé™â| ã  ?``_Z˜bƒ¿qTý¦iêC•òÌ×êC¶Ösèe©zfŽæ ¡_Sþ¾ígºÃZ¾|ëŒÀ a±; ## ?``0Shéx)w^]O à=Ÿ.eïûë‰1V`}:»÷–òÁ¯vÇÀ@+¥8•ÀÀzz }cìÍ=ßro) ›ë‰6r>…QÐ0°Œ_­0…†¤¾íãÝa¾ñóÝaœÒlâ΋Jyy=ã4„ÀÀÌÉ©‹T«é%üX¿|ÿŒVööC̨ €èX!p‚`˜³Ôìó¸*~‡A Ì,'Àa¹©~Óßïá§JÙya=£ÍÀŒšÿôgS} Ó#aê°*…-"dÆéÂß¿ sþLÀÀ*œT`‚†üž}C)—þçãiöù™¯•òå›ë‰!fš*(X/üxè{É_.å½ÿªž˜Σ0jþ÷Ð0°_^˜€a„¿ã 7¿r{)¹ž˜1``æ9ÎüZ+üxð›&Ÿçšó(Œ’ÿ{ôGÐLÞ0š|~ϧÇlÎjø Ð NgÀríç·˜*€uøõ*£Ö9Ýwe=>€q¿iî9Í>jaS)·øòÀi »êwœÁï0ÖûÃ_ß:¯ç‚wÖ#À,[µùçQôû»y{=²Ì;þ»z„s˜Œ…§Ÿ~TYп¨` à ~Ç ~þ­Ã;‰šðwëžzb ôòzLCX ´Ûé>‹Fö9$taö­þÞñ߫džèìëKyñþR–NÖ3:6•rË·ë ã< ÃàÿýQ ôÉ'«Fø›`\äÑ¥üú/üÍzOSøÛ«g¾VLÈ°~<ôfÛª•¿£²¸ûÔð7„¿Ct¢¾0°B`j9q?èÉû÷|z¼•¨KÇKùÂÛKyy=cã ­ûuö õÀ¼CO>Ü­n,gÀÀø €èßW ¯jkïûë‘kÂê;¯®g hZƒßÆû¾T̳-õ%̾‘4ÿ|å‡J¹é®Îqï—²ýÜnÓÏgÈü/0°A~Á 0·¢®l&p#Þñ‰ñ…¨Ã «/¹qº«~ûõå›ë€¶Êi•ÀÌ–±õýïþ¹z¤#A°¦ŸGèX} 0z``ƒÀsç??¼ 5êu·Õ#v×ÏÖ#CðÞ;ê€áKø«‰ç Yª/FO Àé%üÍ0 㬞M`ýð뉵¥ê`.©fú­ò÷…ºÊ÷Öû»—´ÎÂÓO?êçgÀœHh½aUýƸBÔY\çQèe;¶QšºÏ!ýp²Òô|§½ýÖ[ë±RîØ·¯;2¬cšÓ¾Ï¼7ÆÃ96Ê{€þ¨`mÃ<é8®üKÇë‘}ø볎ö²þÃl"`¥^>‡†ù£Ó¾°Ü”ïc}o0B<`<ÀÀ€|yh/ßÜ=á8Œ“Ž—Ü8¾õ×°»Îw^]Ï@Öyëžz¢å†ÕD6Ì”ÍõåtX^ý[¹óÍõÈ“æŠ& !Xè ‹ÝQfÛ°N¾çÓ¥ì}=1b©øFèûÿ]Êžkë‰éå5õJg`ºMå粶ÊgßZÊÉcõÄ*Néë6µ[º£Sä”æŸåßwŽm®ª§†hÝ÷™÷Äøi‘€~xÐÀÀø ÀÌËÉùa…¿9¹8®ð7†þFÃ_Ö xi•Í½¾žYnúÂß7øÜ õlŒ¿D˜Yà ~SA;Î ®a…Ömèëw=7~®€–Ú¼½YK‚ßé ýO©þýð¥œxµž¢Ýü˜0À€W£e£ÑHeo!oÿÃNéÿwß¾zl<ÖêïwTáo_÷»îûË>>šæéOÔ—Ëå½´ü4o¦›ÚŸ“a­SÀ+o·ž~_K Ãã= @Tc’/ Õ]?;{áoNÜ6Ã0 ?hžlËfH@Ôn h§"ü¥%¼®¾ÆÊ7ÓÍòë¼½†¿Ñïã÷sßÀ0 €fÑ3_+åá/Ö}zϧÇþ~é}à }#}ù©Pݘþízä4†ùzA¿–ÿXdµ&B<˜&ôm§Tþ®¬þбú700&ÇëK–ðãË7×}Jhº÷ýõĸ,•ò½õøä9\ðÎz‚¾í¹VxÎôð2õ„ÀýkwðÃìï·¡ú—öÐt1L‚#ý$ùÐ]õÈ8-uÖy€fª—»åAÁ%´šf"[éð“õó¥ }Û˜"ü…öq>ÆM 0 î¼jðª¸Õ#còù·'üMè›aaK=ƒ±Q‰ jë™õH›´?ÔÌ|oŸa„¿ª”3¿úè[B¸¥“õÄ»r6}þ;TO @Åïä ›…ú’VÙ²«é‘Ïœ'ü”ð€aL«œôDøî½ãQý‡Óç¯ðælj#‚?fÌ<ÿ`!Ï}å°¹3¨ŒÞ¨„¾Í°á/³MAŒ“˜ý§ußgꑼçÓ¥|ð«õĘ|ùæR^Þ_OlÀG¿]ʇ¿.üÛ˜iò«××#Ð1?X¬/çIÞ yî+‡ÀlÄéB_hçƒ`\ÀÓ&'¾ïþd=±AgßPÊÞ÷×c°t¼»ÞÏ|­ž± $7o/eëžz#·ØcÓ¬*3•ì[ö/fÖÇ©©ü„ðw6,NÓ佘©ÊðôÛiÊYÅo_>ÿÖz¤G[vvûúý•óë«»cß¾z Ú¬sl ŒŒ` z¢;}ç&HwŸ¿ƒTþ6ëÌäÜtw=3ÄçÆl¤EŠ™ œÎº<‡¶ÙãsûÿëíõXonÿµï«ÇV—à7Ãí·ÞZ϶SЛ¥úz'&ÌÁ>ÀÀ®»­¿¦|‡!/ƒœ˜¿ñsã_gV×O˜6hßÔ­5‹³Ë«}…¿£vºþ~Uý2¿r^È°þp¬3@À“6Hšð'Æ$ë{ìP=±·Œ“ð—^ ^MâÿÙ\J?¿íïë·_£jz3ôK LUÀÀœ˜öÀlÐõûè·Kùð×äŸUïøD=Ò‡»?Y@üh€?Wôi¦öITß&ôÍ㎱«ˆ)µfŸ¼#î ¦…`œz9y=© ©ô±8Œ“ë›·—²uO=Á̹î¶z¤÷}¦óg©;ëô3ÆKÚá‚wÖ#m·¹¾ÁïJ£jjYóÏÌ00eT-Ökø; wýl·ÅA gÚa÷Þz†h¦ª7¹þ¿˜©ýh\ð|öõ{Zkì+ìž©æŸßöñz€y ¦h™œ€œÖðwéxwÝþb=cêCJáo{|ð«õÀš©x”álî[ø»MA/#˜+`€IKp:‘ðt©”;¯®Ç7èÖ‡K¹å!á/0z>gÚgÿs7¡¯à·W“ªÒ½ý7~ €ñSêX} 0Ãz©üT Q­Û•õÄU뮿AjƒîO´Û að—™7ŒÀVè»ÑÏ‚ÕBàQVO]øû¡»êæ…`z ~'þâ–…1m÷ë?XÀœ8RlÐÙ7Ô#´Vþ§\ðÎz¢ƒþ?›)Âß×¼ãõÈàÆÙôûöÕc°ó¢z€y!¦ÔR} 0ƒ¦ù„ô0ÖmaK=B+eyy=Êþô«××ôÊ3õ­vãç6ö㢙 7â.váï)®»­³¯<\O ®×8Ëõ²ìZÕ¿½Þ†A L±£õ%ÀŒÈIèÓˆžTåïÝŸÎIr•¿í6SA ÓoH?è;÷ÏÕ#Ì…Kn¬GúÐÊ8ËeÐÕÂêúß.iz£ý n˜5`€aèåäó$ÃÓû>SlФ‚kf‹™åN¼Z@Þ{G=Ò§™úüIUïZšà—Qé·?à~ÃßÓ…ÌÂdÆA L9UÀÀ”Ë çi{]¿õ|øëõ­5Œý¤1¬ûaö Úôs¤¯Ï‚Ì®þÏœ™ÏŸÕªW¿}àت×Jà†µ­4€aŒÒ$+g‡q"<ë¾uO=A+"0Ï·»~v8ûÀïìöõÉ|j},ì´•!í°+ssÿÍ0Q»÷Ö#Ì00TSêt'™'ü=ÐíówP“Z`¶=üÅzd@?ø õsëm¯Gú”ÿÏíþ`äšX¼a·)üÈ–R>rO=Ñ¿^«sïØ·¯{Ý •½“ —˜``F¨¦DN$Ÿîdò$ÃßA ~çÇ8B‘T£3?¦>hcæhœ^¥xƒz rW¶ãlÖyâMH0“ÀÀ 9Q_LH/áƬ†¿ ÃBáïüØèþrö õH¾ñóõ­7Šðwë™õs-ÿ›ýÿ”ýsû(Ó#•À„ÀýX­xX†Þü3sK Ì00!½ž8žÄ‰¶÷~RûCw•rËCõ­7Èþò¾/Õ#=šúþ7Š¥ãõÈÝúpçÏBw  oü\=ÑŸ¾ªlý˜€ fÌÉú`Lz ~'þ~ñ¥üæÕôžO—²ó¢z‚ÖÛèIëüH ÙÇûÝ×(o¿;¯®G†¤ÚÇ„¿¬0Œÿ³>ÚoÛYõHÿ4µ @›€ÖÒkø; w²”#Ï×”ußûþz‚ÖÛhð‘ýdåúÝï….í5ì×vRŸ©Ì†aìw^UÐJ‹gÔ#sºxÜ!ñ†Ïg)ÀÜ3fM ¬”@cšÃß/½o°¦uw918oFÀ~øëõs©×ÏI¶Aÿ-iQ¨Õòƒ¥ì#ì'“ªZÿ¿Žñè3h©¾•ÆZ'Ñ&urmÿo•ò½õÄ}äžz„¹0HH·Þ~¾uO=Ò#a!0-|͇!‡Àƒç»íÐÂ_¨-<ýô£’`m­/†¨—“ ~‡q²zRëÎä=PÊÞ^Oô©×ý¥Ÿ}Ó>Ø£ ÏìôÃÿEz5è¾’ýdûÛí¿ñõØëîØ·¯º\-óÏGëKÚ@00£|9†('õz9±7©ÅNr³Ùo6þ^yS=+ ãóh5?üÛõôhÿ×Fµ?3]ÝWfi?q¼@M Ì·^OêÍjøûžO;8žùZ=²Ù_Þý©z–U’Ï©=×Ö0fÙ¯›&`¨Õ¿P3L00 •'{× Jg5ü½õáRö¾¿ž`n,/åË7×}ZØÀ׃¼?üÈ ÝÜßmN|T|N1-„ÀíåÿsFÀÀŒÓ00€å'zsb°×@xÔ†q:uB•ù³ÿ·JùÝŸ®'úô¶w‡êe¿u~öŒ:»à¥Üø¹z6àÄ‘R~õúzbH6-v/?ö@÷’ö¤oü!X­àÕŒ¿ÿßðk€6Q Ì8_R Zj´)üM˜"ü?9©½Ñð7ûû á/í4êð7„¿ jóöR>úízbHNëyÜýÉz&­°uO)þz=16þæÇ4~dÀ2``þœ.Ô˜õð7'™?­hr˜Ռ#ü…aI|˃õÄÝ÷™në ´GBàaÿh`Hz Ͼ¡©m?×ix0Ъ€>´9üͺ çÓF÷Ÿën«G†@Üã`Øg¦…-Ý}jûUZWÈûbùÀlåFí¬0¬BÀ@‹è8õNØN2ˆôD²e¾mtÿ±ßИD˜eÿcF¹oÛ‡Ûa}H¯ãt}÷Tœ}ï×°”—÷¿>=~X Ð&*€€ù ü¥?YôÉ~CäóGøK›r_Ë{çÀýõ3+•Àc´¡þ}W³óÂîåâ®î%¬ hUÀÀ*¦1ü]:^ÊW×}ºò¦R^¼·”÷}©žÁÜÚHx—fŸßñ‰z‚¹óÙk:‡Kg–räùzƘ½íãÝÆiÔ?tð£†Ù6æ*à×tö›Ûo½µžè1^þ?üóo-åø+¥ÜòPwz`*€ÚD ´XfÃßäd´ÍÄFö¡íç–rÓÝõså‹ï˜\èÛxϧKÙûþzÆ,-&{©”o*å o¯gŽ€ÿѳéèRŽ.å×Þ]σÝÕÿã-ß¿¾ô¾n3й§ž1(0@›€µµ²Y ~N.³‘ýhaÓ«„˜zÃø¬6Ÿ]L‹/ß\Ê3_«'FÄþ>›¦ñ³s¹åûÕ]?[Ê¡ÇJ¹ñsõŒA €ÚDÀ@ ùâ tLcø; N(sçUõH.¹QøËd}C=S YšÑesä9IU)ŒÊ»?5Äð€¶Q ´˜J`˜[Sþv¹ö]YoÀî½¥|ð«õsk£•I~80_¦±‚Í>È4ûõì^¦9ÝQó^˜nÓøùÙù¾ã‡Ôm¢h—i ³Nƒ„¿Ywá/=)~Wi¿¥ãÝašÃ ˜Vù›!ÿowÕ3GÄ{À@Ë©†¹2máo*Š†QM4ÉðšéðÅw”räùz¢OöŸö;p)¿ùCõIJ2‹îþd)÷}¦ž3­~LδôcùU Ð&` å¶tÀ\˜¶ðw'…&ÄF÷¥½”­{ê ZmZ‹ð9FLÃ{Ì{i<ÒÏþÒÉzbŠ€è“˜ª€¡õ¦)üF%ÞÛ>Þ 6< ÚmšCßenÿ¨.ïØ·¯º„™5Mï9Ÿï£5mŸ¯c{½Àm"æ„ZkZÂßaž,tb—ÆF÷+ûP;M{SÏ+4áïr‚`fÞ7~¾;LšÏùÑÐ``Nl® •Vž¨׉²/ß\Ê3_«'†àìJyß—ê æ[ç}ß•õxŸ„í3maDV €—3Ó¦å}é3ø¦é3wqW)¹§ž50@›è˜':Ãöå ߸N„¦úg˜áoiçØŠßçÎhøÛ‹Ûo½µƒ”ÏÛ 7~®ž1!-þŒ˜˜iú_:¶ð€¶Q Ì™…ΰØب/½¯”î­'†@hGã×°”—÷×}ºõáΟüŸcf »U)pº*àåT3óîúÙRþb=1F»÷–òÁ¯Ö Å\Vx«hÀ=K³¼W ‡¿Ù„¿3¯eáoÜñß«Çzp÷'ë˜QïþÔdþ¯ç‡÷ÏpMÃñ™cD ˜S[ëK€> «d¬ý¹1Ù·œ žmG”ò…·×í¥˜¹rßg&È~è®zd;/ªG8­IWõÿ»ê_€¶sL ôaX'…u¬ôù·–rìP=±ö©Ù2é@a‚ú ãŽ÷‹¥lÞ^OÁŒš¶÷ü¦ÅR>ö@=ÁiMòõËÿ÷õ@Û€9&z4Œ“¼³”?WO@Ç0ö+áïl™ãðw¹¾ƒ`Á´Á´½ÿýÿèͤ^·±½>*ÚJ Ì9!0°Žý¿UÊïþt=±AN°òéKúÊz|ö­ÙÑæà·Ù7ø…ÁÌ!ðì™äk6Ò×çXgp: ÍÀÀœÛÒ6uG–ô„ß{>]ÊÞ÷×ÐÑô™¾!å¤ýô{æk×ü¥¼po=£eÖÛ7ðù)fî|åöRÿr=1%üoy£VÀª~æ@0°Ü7~¾; "M=§ÉgˆaŸ¶O?a° ˜Véósdå{E<iÅc-xô#?&Ì ‡Fø 0OÀ0̵aœÔs²”Æ(À…M¥ÜòP=ÁTFë³`Ÿu#îüYèŽÃ¬;M5ýjïðˆœ8Rʯ^_OŒÉÐ^ á/À¼¼F sk“¥l°É۞ؿ¦×(ÿi3ìýïèR¾ðöz¢7}Á·<Øù¶»¥ž3nÏðŒós(¯EŽMŽwG˜`€Sa® 랥Œâdðÿv){®­'˜:m~w•ò‘{ê‰ëc[öÇý÷UÏÓN÷ÍZïðŒëÀÀ¯…ð`^ €N!†¹ðå›KyækõÄ€œ$/i.|÷Þz¢#ÓÖûÿàWë ¦RÛÃßën+埨'ƤϦUû ‚ãŽOüÕîsƒwû­·Öco4pøÛp|szSkö`ž €Þ@ íÕ9ì¹ûŸ Þ߯€nþ|öšRN«'FèCw•²ó¢z‚©Óöà7ÞóéRö¾¿ž˜€>›‡ÞHܸcß¾z fÃzÁochp×7®ÿ jÒ^ø 0ïÀ«C+ ëD¢ógÔ'yíSÓ¯­áo*b_¸§”ÃOM×[¾ñó¥¼¼¿;þì½>¾A0mÖKðC —ó?jmù¬Ê0j}½Â_ÀkC»twö]YÀ Ð93¤ýf=ö©ÙÐÖðwœýüCš\?M Á ϼ'™½†¾ê‡ £ü¬òÞXÛ8þGœvûçÔÞZ*`f€Ö$†ÖƉ9'>çϨ›}¶OM·ÂÆ™7«û`úpO_*˜YÒoàÛ8e_e¹¡fˆçȯÿài[*ØšŸÙÇ;ÃÉî(ÔÀëÃLü2ˆQœH·?ÍŽqTtMÒösK¹éîzbõЇ0˜aÛhػ܆š{¾ò¦Rþb=1BïøD·yùy3êÿ«C¨ü`u`€u €a& óœÀn~ s?úÐ]¥ì¼¨ž`ª~²”¯ü¥¼po=£EÚÊìÿ­R~÷§ë‰õ ‡ ˜A #ôml¸¯ß³o8å³­y_Œ¬ïàÕ´ý¸j”!ð¶¾~X›à´ÒÔÙ¦î(0݆uÒm^+WxÝ°ö¥÷|º”½ï¯'˜Z£üõR¶î©'ZêèR¾ðözbuÃÂ`z5¬àw(!íBç;ÍR·©à•ï‡±†À¶†Á=|mXµÍRñëTëôD%0L¿Î!;+ëñ‰‚ªÅgS[Ãßyî¯ó4¯© ˜Q›ªÐw S7Úøÿs”ÿ[n½¿€µ €z²Ð»£ÀôÖI6~÷)ûÎìøì5¥œly‰öÇ×­ñ^fÜ3Í<÷`­ý¢!p£MŸ_£ Àô@ Ð3UÀ0µ†q‚M`Âr½îS?üۥ칶ž`*êü4ð¹Õ»Óìà „…ÀóiV‚ßXo_ŸŠ¸Ñ–Ï·‘…À÷uþäGÊ°:0@ÏÀ0Uîþd)÷}¦ž…å?Yʯ½»žXÇ­wþ8ù:†Ô$ü4òy5˜ØaV …Ûi˜¡oc’áoläñW»Ï¡=6|ÞúGHªXƒ /B`˜ Ã:™–¾~Óç/,×Ëþµi±”=PO0QGt/¿ðöîe}ôÛ¥lÞ^O0°>þ‡ 3Ž;~å—;ßÂç´oæ–˜Åà·ÑËþÜ˺ôú¾ÊóúÐ]¥ì¼¨ž˜Q£ ?ôÕÎöùÿ·ww±›\÷}Øg¹»¬JË¡U´ ULèJ”Yb¦h/!”P•`mJ¤a#*Ø ËU´p‹4íEk}¡zAT‚”ì0BP‹Í KH²€bpU¯k1‚¼Bš±¥2Ü·>¿}Îá>ûßçefžy93óù‡3óç¾ü÷?óÌœ9ßùùé´×€Ãèþðo¯Û1î|¨ªîýÍ´'€ËûèÔMUuårú ©ôÎÏ|×!ð&ÕÁÓ1‰à÷‘ïTÕ3ïL7êóXÞ¦³ßúm}†Àq=|äiÖÀ a4_~¬ª¾÷bÚhI¨Â!à²õ9ˆ>4磲Ô<¶úÑžzêœ~•ãŒuþöVé›Ï%W.Î3o¿¿ª>ôTÚ˜¨>¯_¦‚à0@+B`ܱƒf‚êªs¬9ž†ïûŽ÷~ωc¨L-¯5]k*ƒÇ×uÅo§aèÎóÇá÷ŸO:~ÿ§×mÊ„À D ЊxÙ.­Úéõ*ÃxîîªzýOÒFM}¡ªÎþxU]ø³ªºõ®ôE8 îÀ¬ðnXmÎ¥qÌLÓ3?sôTã]nááü†½ç’áàÍcsÛÏîÍÿßEðùîOVÕݧ 00@kB`èÝ© ÷zôåÕN­×¡.p9ú ƒcfú^ûVUý·ÓF;džo‚àîõñnß0lð»áùªêÕ¯§µ=ë{»~–†Às9— è™ 5ð²E5ÐMëUúsÌàØ#ßYõtΤ ¨¡îñvjõÙä_¤ :uéõõòóïY/'"*{ð|tÜt]…Ù¥%„Ì}…¾a˜©žØsŒÖ9öŽ9öýl¯þ¹Ç~~æô€_ç’„À‹'8ŠzsÌ ˜ …6êsŽ¯þô9Þ¥càdر3üqÜÌÓ‘ÇlÉAp6õ@¸Ï°wSÁo¶ç¸ì;Cà ŸwX×ij?VUÿÁ߬ªwÿJúK!8ÊÙU3Å,tê˜Á.•™£î±'ÈëOƒÝÇj¸ÀlÕâXžBÌ: ~ôTÚ˜¹>®Â`€YM Gi3 uÛ=UuÿÒtà»_ªªô©´±‡¯_`–¢£c] Ü¿£CßlèsÀcl¨8ëµx)ç×¾®‘‚`€YtB¼<—Víôz•öž ª^ýzÚ¨ámï­ªžOС:🬪;>’6èÜkߪªßùpÚèÐ/üAZY¹ø£îÞ±™Ô €ã{¸ùÖ´É¥×ÓÊÊŸ7­¬´ø‚ë,Èmâ·ªêÿEççZ……)4ÝÌvoª wˆóÆçß“6:ôHêŸr0`€N€¡‘ßz_U]øaÚ¨á-?YU½”6 c/>\U?øjÚØCg?ú¬úýø7ªêô[ÒF?jÀŽšˆóQœ—Ž$!ôÀ÷‡ÿr½~ö/Œ;=ñ¡sk̤rÛ=£À¡óxiçÙ¾®·ý•Õ±Ñï¾``€Î¡¶ºƒVg¬ª~ñkizR÷xâ§Ï w›ªnÀô"ÞKþÆkUõòsé ã˜Bˆ¯«ïþ•ªºûo¥ ¦H ЙS«vv½ ìVg°JðËÀý:ø Tþf`z³f„KoTÕå ëuÆuSêó¿ïתêýŸ^¯—fBp¶³Y¼[ÓYušò~`€Éº)-8šçià :ƒ‘1x'üe(QaG÷ⳞÛ aqMŒö‰o¯¯‘(W<`û"Z©áïDí Mg¾Gœâœ¯KéÃÓw­“£ s¦‚†ë|ù±ªúÞ‹ic·½·ªx>mÀ"üýí¤.×ë»âèž«ª[nOýÚV­¦˜AõÅÜÍá3ÛàØ(¥8«}nÝféçÛsŸéêxUÁ“ èœÞTgð1*z)mÀ@®\¬ªgÞ™6j˜ë€ò¡ÏèÛï¯÷ÇFÞ'¦*C[lüòê?ñjžûµªþü»i£g|}õã>6(‰) :w)-aá 3ŽçîN+5ă ÷!mÌH|FT‡®àРHGž3wÍr0„£«§t}ëÛ@¯D¨^þâºy_0@1À3³> ÁÚ¡A·¨¦ŒÊ„=•¾ŠwR¿þ'i£†7þuZ™ Á/°-®]Ìõ>ødZH„À›Me0À(À½x#-a)®¬C¥}ÁÚ©U·#Þ':ÇjJ¦ãµo¥•š._H+3Pzð›ÃÞÍV¨£ÞO @ÿ¦ô°Sßîøȸ×ר ~é¿I åÔùó¯(UèÅÍi 3Wg€íìUÕ/~-mÀXâA…†ï¼élU}âÛicÂ~ë}Uuá‡i£'+± ¼‡4€'üo&¬fº«²ö詘;pÔ7ñ‡‡^JܨE¿¨+6| €ÆTôF0 !Y¼×÷mï]WùnVDþR„SiÙÀÂßsŸé<üÝæàÔœž»þ¼}jòʇEZõ‹Æºçé¡èhîùÖïR`)Ž"$‹÷ú>ð¼wû2Qµ>éòJU}÷KëꯗŸK_ëß›!py7Û-·¯ÿPŽÍku<ÈyÛ=ë6„czèÍ@'L Ð;SA¡ÉûctÊx÷ᾊߦ í“) I˜éСÑkœƒ3@Ÿb'SE´¦ w—Ò€Ñ4¼üà“ieJ†¦sßàû®û¹þ@L К w`€I¹ã#ieBž¾srç6>w…ŸK ·þ¥“*ä1+Xç ®ec^ϼ/ 0À .¤%óv9-ÉZr`×ï€cm ÷½†`+!ðñJ ‚£ýèÓØE qY…"Õô}ä;iebºÔÞÐm1¸k:hXˆ±ƒàðÅ¿z} ¼¯,”‘j€AÄ»‡?#sµº¦DðÛuø›µÜLл?™V¦GðÈ¢Ÿð๴Q0!0°P`æk®ƒ½ñÎß.t\Á# ¼ ˜îÝrûº/ñöûÓ(…`0ñ`UÀÅ:u&­ìÒëÝUýÖ ëþº ï ‚…À0CzªqaPª€€ êbZ2oo¤%0ª&AiɃ–Yü{>ÿž´q„6ÿÖ> ‚zöÝ/¥•^ýZZ)Û®ëFãi _|8­Ð !0@1ÀƒR¼ 7§%0š¹Mý|Ì¿'ÞÏ÷ño¬Û#ßI_l¡á î¾é Ã"B`A0–ô©´rÀÙ¿VÊÔùCC?øjZ¡7¾œV !ðþÏi`ÞÀƒSº Â~\„¤¹Ížü¿'ÞÏwú-ëvì4×Bàf"hÈÇä7?—¾P·¾#­”g ÍÒ©uá£/TÕŸL_+HÀª€ŒB88§ÒèßêœzLHz÷ãi¥0o¼VU_¼7m´tÛ=i¥C1¨íí÷§/ì§8yé‰ãÃ|€…¨{m8tÙê¹»Ó ½ºõ®ªºã#ë>CôµÎþXú…3wêüùW¤£0Mð|Å¥õòª¾ºô¤«0­aUë ®\¬ªgÞ™6Ž0Ä¿­æ~8ôžÆ§ž~:­•ïd0Ñ*€(ñ¸æãÈsó˜çäCáo«sîIñZ„˜ƒaýƒ¿ZUþÝ´Ñ@„Ç~˜6:ôè·Ò Àü€FsvÕT‰.K\rís8Z—U”¥†pÇþ‡þw=û®ªº|!mì·/¾:¨?`´“ø¤÷zݺ0Áxàw“qÆÓe_î``ÆÀ£‰ 0B`æIØ ½™{ÜÅ¿oŒWÃï{gè0¸×©I7 '€¶&þfž[¿+ŸauÕ×éäÏóäÀ£‰€Ð38ó%ü…NÅ_n샡c…† ÿÞ]ƒùFÔ XK·¯Ú¹–ÍãÞñ d1áoxå¦ÕE_%®‰ü9Þ Ì“ `€Q™`¯>¯±‚ÒCÚü›Kû·4ø7 IK|7p×uãp£Ôc(GÍóð¶sðçÝ}çÓ^ƒßìîÇ«êÝŸLŒâ˜¾^Þ]ôU3£`Tñ¾DÏá\çÊŪúÁW»ÌÛ&³Ò´ø7Ç¿·¯óÐü| ðG80—Šà]®V=7©žËq°Ã áoø¿þ‡´ÂhŽé“½ôÄúšxÛ=é GP ÌŒ `€Ñå÷«¨~ôýªúâ½i£C¥¿ãïתê·?6jŠÁÎû¿6 Õ ¨¬€ªÞCU0°©æùwÌ à]çÏÖçÁ¨ ÿöOUÕ?úÔz½Ž›V÷cŸøvÚ`]=Øý½.ú‘ª€áæ´X°¾ªKÆ¢ú÷™w¦šÞÿéu›’ûwèøPxÛöïê"n€l¢ðQ•¿'ÏMú¾¼úrGSZ„ÀÀÄ €Šp:5æçUðÃNK }7Åô…ßü\Ú¨aÊï+<"8©q@±ú?öÄï¦fZ‡!{*»ëü³FˆJ6 ,-ÞvÝoÒÇpî,å׫êóïI-ä_ºê_ € #¦‚öÔù|]\µ3ëU`Mø»vî3Uõòsi£†>YUw|$mLT‡Ap¶/°è¢"7Šlùw7ù7†ÚáÈ¥••{3­‹qÄyvˆøèð·Îµþ»_j6ôÔús÷[﫪 ?L5]÷Ê+«ÏÁiýB``¢ÀEQ) ,@_Áo˜âàí—«ªï½˜6(ý]ÆMôg›áEWáoÖy0òÚ·ªêw>Ü,ìnS!—ß¹é®_^Ýz8 f©àx×y¹ö¹­îµ^<}MûŒ7ô“„ÀÀr €Š¢ xžò4ÐV-ö1,Ðw·®ª|5mð °:Üš6f Ç¸o}$MÃêÖS¥nòÞK˜§ç×]çÖ1àÎÃß×Ö¸ÆÖuꦪzä_¤ ŠÑ¤¹ëA¹ç¨ªW¿ž6Zxð+«?÷§ÓÀ4€Š°Âùñ`êÙw¥••ËñDÇæR©Ód`sŽÕI +´º ƒ÷uþž>C’AÃ`Uo0?-àbÃß·ÞQUûJÚh ͲPÛG_¨ª[ïJ¡I?aßL)?ú~U}ñÞ´ÑÂ#ßYÝ®_Nå%à˜ŠQ%ÎüY &Áf]s ªêþŒ–е<^šÂMƒÒQƒào~®zì‰ßM‡A0ÌGðQS?s~jz}q.,ס}YçU-ûW]}@ ¦”6” ”O PUÀó$fAŽXÛg©áoXÊ`ôÇN­ öˆ€tÔ x¥IEðÑAp&éªq>2Þw;xÎêâ\ÔôúâüW®}û²NÚö7®«û;€r €Št:5æ#™¹#¼½Þýɪºûñ´1uVK„>òÚhtˆ® ‚ûž>5Ô ƒ»ü7¿I Óqà\ºë\6d\ë<ÕÕy§éµÅù®\ûöeÝ8´íoü¬nénMA ”I P¤oZ5SAÏG¼ÿ4.¹`f蛟«ª—žHšûàkÝÇ¥B¿øpUýà«i£,%„Àqüú>BçA°P¦áÀ5fȸõÔÏÿÆê¶è-i£M?çºríÛMà稪W¿ž6zÛ{«êçÓFˆw_\¯B P,UÀóchfæÒëUõù÷¤Ž ¯1Ýìç5 "‚à•Q*‚—P¶BàÖáoèú<3ôµ$ì]º ¶iïÜgªêåçÒFC[OÕÀ@9ÀÅRã-à žˆýæÍ™®Máªêæ[ÓÆ‚ûseË»÷žygZF­ øÿøì Ç~Ó 8kô…ªºõ®´áÀµfxõ=lû{‡¿=>XÒu_f úýjsŒìÓ €qHŠ¶e0›‰‹êßدÂ_&"Âú0]bøÛ”AÞfN¹±ÅÏpÀŸcðô±¿ñ«Uõ£ï§­þ´ qê„Ø×i;m&0kÏ%oÒGÞ+÷Ëv5Žs÷ãi¥?þÇie³?ãP 0 gVÍ3;󑧎K°A. ÖÇ@âRCÍ6?Kp·¾û¥ªzË¿»^ñáõ²Žû¿PU¯~m½þ¯þ°ª^~n½¾GÝà£ëwmÒ¦*¸ÕÑŽ]×kNßÀÅMýœ-- u.nçùV×ý¯§šjý¬/¥0 0À$DA¡°p>âæÿôzJáؾš6:´äȦƒÎž«ª[nOôâË¥••ï½˜V’·½w½üóïVÕ/¦ðwÓKOTÕ7?—6v›S…Á‚Æ¥××Ë›â~!90ýýðèáoXZ¼Ë#ßY¿;ÿì¥/œ3h,]Óc¥öñk:h`8`€É8»jàùÈïVÙMa";†µuÓê¼uû‡ªêCO¥/,Õª»ýôi½&áYùÎ}¦V5p(5Κµƒ`Ç1ô«e Ùgܪúw¨s…¸¾CïbîãgYÒ5#¬øü{ÒFMo¿¿AŸW ôO 0ª€žu9˜·äàçØŸ£Ðl:>0Q7³© vNÇzÝïsLIG×›>àɇ¿»<û®ªº|!mPŒ1Ž›ÁBà º#˜$UÀóL9܇ôV†%†9~–Ô²ºÍjñŽÎ¦að`AðÆqßä{Ü9Ö¡ž¯7µàè(žÝŒ ì¡Gc;M?‹G}B` `€I:³j šº -ßzGU}ì+icFú vëˆÍÓ•‹UõÌ;ÓFsņÁ+GO 혇ëÅùâŸý·ëõW¿VU?øêz½ÚhpYLø»I\–¡Ž¥Aà Ž'˜,UÀó£ ˜un>úòê?3;÷Œ=ÍãG_¨ª[ïJÌÖ‹èôàeu^i] ìÝápÍ‘‹ÒÇùc‘áï6c><Æ°×’&ûº“ï+ú¦†l€öÀ“uzÕnZ5!0PC”s´}îîªzýOÒÆÈæ6Î~G|Wÿ§×+÷a½ìCú÷ÔýÞn‡ÿ,Z»©â›h|Þ8"^\ø[—¸_·ÝÓïu.kº;ù¾Tí €&Mð¼\Zµö¡Ç>ÎeÐöÜgªêåçÒÆH^„#§mTÑ÷_ÝSUïÿtÚêÁÆù¥Î÷u]HäóÀR  nûLÆgp×gµNܪú×g½pý?ŒÐ«wr½¼ûñõ²/M?·ñ}uò= ‚€æÀ“'j8f yŠƒ¶1Õî[ß‘6’/Þ›VFbð›]¾ù¹ªz鉴Ñ\£0¸¯é¡Oü}OB`o€¸É¹!;tŽþR=ÿ@U½úõ´Q ÏUÕ-·§ŽµùÜvöý€fÀ“—ß+öh;Ð\ä íªûzéߤõ ÿæÕõrì 7{Û{«êçÓpdÔ&è Oýÿ}½rÓ™UWbÕŽ±åß°³ÊPÌ únjz^Øþî ~ÃÞð7øŒÏÏ‘.õîí÷WÕÿ—ª:ý–ô…µùwú®bA0P`nNK€bºÙ¯ügëõ VUúÍUOðtU]¾°þZˆÀ2\øaU}ì+ëõ <€Þ˜nŽÑÁñÝ8ôÉáÍ£/¯þÓÁe'þ Û¾0‹AYfiý@È–x_ð†¿Ág|Þ^|xÝ· %Vw}üµ½Nwú}€ÃÀ³;7­WæJðËqÜ7 ®…ÀÇ{Bà#Ÿælàk—ð—¢\zýÚ¬-ÿßùõ2œýñ´²2ÄÌ-]‡Ç|¦;ÿ<‚€ÝÀ³á]ÀóriÕN¯Wa©J | ^3¶#? m¡NÞ\÷ûöc®–þŸoêâ³"F 0+¦‚fbàAó­ ZS²Ÿ‘Q‚à:ߧÏsóåǪê{/¦atþ†]píà7ølÓTT þ=i£%„À.bXwã•.‰`VT30vøkÀš)jø¹i €Y’sŸ©ª—ŸKÃè:ø Â_Æueuí¸3­÷¤«ã³˜8¨®'˜UÀÀÄŒøf«™ºŸ¥A‚àCß—Ïs0𵬊þ}Ó> Tߟ§.Ñ6ßkoŸ!0p`vT2ð€ùU¦™³#>Su¥§~ùÕªzàù´UÓ¡ïÉg’©úæçÖË—žX/Ò׃*)ÊóTÕ«_Oëò8}îîªzýOÒFM½NÁ€`¦T%ëqj¿·ÞQUûJÚ€¥Úv‹‡Õûìí —®AM®ÀÌÕÀ1 üá/Eèú³Öõ±Úæûëýó’ß lØ–J 0K1È•ÀLßÅU;³^…)ë{ü£/¬N{?^U·Üž¾lÕ೸+ljú;…FLMß×´ }…¾›:©ü½ížªºÿ izÐåçnpþ9 –D 0[¦‚Ftéõªúü{ÒFOb ùÕ¯UÕ¯UÕû?¾ìŸ—ßþ@Ú8l[øÔª øGßO++¯üêzÛûªê­ï¨ª VU·Þ•þL€ðw;rз/?VUß{1m©ãµè8354,‰`ÖL=q©æ3!}Žðɪºã#ihìµoUÕÿý¿UÕËÏ¥/ì·5n6Á¤õ|}køfÇ|; Uÿ2”®ª|5máÁsýÌÓôñÈwV·ycÌö$€%¸)-˜%ÏøÌC„¿14L@ßáoTú á8Qq{ïoVÕÇ¿‘¾ìõü½^ß"ø:üÝ¥qø{ê&á/Éc-88Öïíç3Ý´¢÷™w¦•¡Åƒâ0€¹Ìšwý̇÷3}‡¿ïþ¤©ž¡K§ß²~6°[\Û^ýzÚèF|Û¿]ØVýÛ8ü ÿñ?N+0ßöÞ´A{ñÊ(3†Àœ™`ö"8ô¼Ïô]ZµÓëU(͹ÏÔžJ¶‘G_^ýGu âÊÅ•HÛ*S@3{=<ÔÔeØ[ÄÔÏÁ»KÃwÚoUÌ»€KèóšæF"0{1u°g}¦-nÆ…¿"µN¶^ÂßþÂ`â„ñ¹ËmC«P¸N)áo§úÏ ®›o=þÌ}Ù±=}gZ“i¡`nÀ‹`*èiËSs òÑdQý°%|€Áôp½ë*üà÷Øð÷èêß|v­¦q,ÞÓO\T4δÐ0'`€ÅN[Tç'²íKÖWðûà¹ɪ ÿFZù­÷¥•îtþvü†N§~†’Üþ¡´B7"ÀÔ €Cð´Å øåõªŒ!ä ¨>ÂßøÞr{ú09}=cú·~2­tãØð·«à7tþ^}X ô¡§Ž;>»¼¦Íês"€),ŠÊÑi‹÷9Ãú v Ã,\ ¶ž}WÚ‚ˆkߟ7mŒ'‡¾]¿Aå/‹ñ¶÷¦•ÆŸû+i¥4¹ØCÈ05`€EQõÁÀ1Àrœý±uüÍÏ¥/îJ÷ýª&Ó?÷ü^µ'L2õ3³¯)á˜0z–"Ž 8a d`€E2ô4åé·‚K8GzãµuðÛ·€3p °,o½c]ýöÒý¾V P%„¿QùûØ÷Õ´u=á/‹pÛ=i¥ ?\Ý*8ãR<´ñåÇÒFébJè|o ”Èè1À"™ zº¢ 8Wq ò9Âo ­t䦳×ÂÞÍLÓMgÒÊnMB.*_"^`|H§áoÌêw­Ú¾iŸ…¿,Æý_hw ?óÎõ牚6RJrêüùWŒ,RLÙOí2ün»¹ŽK¹}I QÑpêtU=}gúBG¢Â"Ù€ùˆóE ‚oØø>õÙߨª[ïJ[°C"x¼*>*~…¿°ß‡žZOK?–6Ÿ©g~&­LQ<äï Ʀ`ÑâÆÌÍÙtlV_Z5ûŽâ]bѺdpbu«¸å¡‘™Þ “>ú‚ `8üÆCl1+ÍÈŽ¨®úQý{tøëÚÎܵùœ¿ýþu€|¬®ªl'÷N³øLÎïa˜0À¢ÅD‡+|(Åæ ´÷,QùÏTÕËÏ¥Ž à(Û‘û55ƒ¡6¾½„¿'¾_á/ÔÔöa.>#Ͼ«ª.ÇlN Ìæ³)€1€/`o˜†|ãœIc[E0;QÍtƒÂt*]Ã#Á®™m‚ßÐiø»å{<ô} a‹6}ä®>'MÿîÙ}>ã5F^EC,^À1ýž÷È–o[pÜD ðÙÐUð Ћxp-ÚI#½ždǵ³˜Šßpâ{<:üug©.½^UŸOÚh “ÏÌöW+ìôño¬N‰oIs£*ú&`e¤Á6:çÁÓ€sˆ/ ^,Á/“³-ˆþL~@q{ &poÁoxîîªzýO®®Öùž„¿pÀ3?SUWâAÚ†ºxï}Ó÷¿õŽªúØWÒÆ ‚ /`Vrø+.ÛÉ›ã\Óhy—ó¢µy§Ø.¾¼úÊ®Áÿ¦ƒî ú··ð·a𶆿_¸^Ûø¦³Uõ‰o§#4ypsöp&€® €Hrp\„?eÚ7@áßÙ«[,@ÓÊ:ö裵ÂßøÞ¶†¿ïÿôºéÂ<\‰Wþ,Mœ¿âºà<m©àU¤åÚ5ðû,öW Ī}7M*šºÿ UuÛ=iJP'ä¾NýœsŸ©ª—ŸK5õp팰·‚ßü}Å÷ÖEpÖæÁ¦ýüEVoêø! X0'ÄägÖ«fßMoî£BÔ»€g£ëøÔêóï;»ó¡ªº÷7Ó $uù÷U ×Ð&î(€é,ôÝ´øpZˆ×«„cgÙióùkójŸóA0Ô%` UÀe:Çÿ?r0”²t?x®ªn¹=m@Éš ð·ìûÔ €;\Ú¾›ö†¿A0Ç9¶ßÝæ3Øôï|Û{«êçÓÆÒÅ««–856Ô'`‹}U‘–çÐ`hݬâOZWðÇ¿QU§ß’6 tM+¼Z„À=Àƒ„¾Að Ýh3+ÀIM?múú>ó[4½fÀ2€Ø!¦‚ŽàðôÕ-JP÷Æ6A/­š}7y]ÀÂ_&«É ~Ãw¿ôDU}ósic‡†AË1¡o­°w“º÷Ú·ªêw>œ6Zhó¹lÚß÷ÙßC ›Àì‘«)bz¥„WÚ“UÀ*€'Í€‹×d@¿Á,6o¼VU¿ý´±EkjU¾¡qø\ó¡{cÀ¡IŸßg¿A00{l¢ ÇW÷F6ï7Áýä `ñšä·˜ºcCßVaïI®ùПcfà$ˆ6ä À² €8@p9ê€Æ>‹_Û  †2Õ 2Àì· ~; |3×|FÛø˜ÏhÝ¿Óy ¥¦×˜>05äA4UÀãjrÓÚoõ À°Ú Þß*"ø BNÛø¶{ªêþ/¤†À°`jˆÁ³‹«vfÕ„Àãiǯ?À”]J­©v!p›à·óÐw“Іsèáû´ý¬ €GcÑb¬æG @MyðÌTÐãiú´òñU/åˆá‹x¯cSñJŒÃ0 }7µ­Ž à¨nJ\Â̇€šò Y  Å¥CððÚÜŒÆ~ò``NÚÐo8®iðûÔgcÕ½úñªúâ½é+=òÀ¸ÚÀ¡ÍçW\¨¨Ž‡à`zÀ4ƒf1õ^„Šª€‡×¶8*er,¼¦.ãÛLÙyíÁ¸Æ¡oŸ•¾á¦Õ÷õ‰o§  mCàG_^ý§á=—xBT 0 Fïh ntâ}²qùðüPùbm†¿Aø L]Û¡Œè¿^)/üýø7„¿0'Oß™VìNH­Ý¨÷ð÷j ³9¼¢³!¦W‚`úa´€–â&%¹"Ht9Ʊam¾±Œ}–ƒdƒ“À5dz #ˆy3Œ‰~UnÀläÏùfƒÎăڹ@wTp„c€+*J#6t¿.­Zü¼­Ž?'Ø_@ÚVÛ%Q Ü{ÅïÇ¿±êJ½%m‹ÓW¥® `®Š1CöGÉGˆ›’§¦_qÙŽ›Àc‚Û¨Žß-ï?€‚Üt\ß²—ð÷ƒO^k¾èM\Up0GŠ@1‡Š¹²”~lN)xìÔ‚±Ïâ¦ÒSÅ@a®ŒÜ§|÷¯\k~køÞñ‘k Aä X @3`:oD.§e,ö#¿gV­máoÞgÇL' ЃS#¼¦âÔMëïÝëZ»*?ðx(œ€Ä€X¾QQšÅc+TÙ.~®qùÎaû1!p´Øg›Á=ÀÂDèûÈ7Öm/}&`Àt$‚Èsåïæ:ÝŠŸu´c«c¢ 8{ƒ™@A†šúêôΫV›™S€i8uþü+FçèHžVøªEeiˆËLÛ*Uv‹ŸkncÙörž÷Y ´Ž0Ý"ÀIϾkÕ½‰þdKñnΦg­;…óUO|ê…ê³i« Wúý_ÿ¹êi˜‰§ÿRZÙ!ÞÞÔ¡?3´ùsÏ?ÿÝ꧟lò°Ña÷=ôpõìý·¦­}ú¸¦Ý³º¦}À5 `Tб<­p®Üþö#~®›—ñc~Îy o€üåÿ:­´ðÑv¼›3?œБb÷5€ €èP‰¹¢tsZaáb¿ŽÐÌûìÌÕ-€Ñ½û“ÛÔ:íÖ»ÒrR<,—û©ó& c(ªF®ŽŸs¬·ý9Çï}0wB``þÀô Ö"PÌUÀ*€û“ƒßüþÞ6!p샡ÀRD¿Ç”ÐÀ| €èAžV8Ö.®ZÛÊTꉟo„í9n*~oì3A=°ÑŠþªa`~ÜéÐ#Ó #0ã’žÜ6{ €Æï»°jùÏsw&5€ùГ\jù]ÀA¨ØŸpÛNg˜+€ã÷ç¹M 0514’û­Ó' G*FÛœšX¨Øø¹F‹°=¯7•OTç÷7,E ‘Â]z=­ì& gQMa¢i…û—ƒß¸7 ó”Ý1ð©‹Ð Öúü{Ò ÀnFwèY„Šq¹ÉadÓP’fâ烌m«WbEp4ŽçH`X`zÀ!‚Å‹ëUU=Š8.ïùgÜ&p}ïÎÁÁ>ho³_Ð/0=˃]*橉]~úÁo„¸mÄ>‹ÀwóÝÍöÀq„ÀÀoùÉ´p£¹ ¦½»´j¦F„¿Ñâg—ú6UÀñû¢å0€ãEßJ ìðÐKià8`¡oT“æé 3ïCëO„¾9Ä MCà˜Œßûl³»ûÌ~8Žè€äj‡˜–8W·©L¥™¨Þ=³^m,ï³ÍÊØgöÀñNö³º!`@¹x³‚4¦)¦ÔÆ¥>OáÜ&¸}mspÒ>èŽØðèÿs¢½|mý‘﬷0 ÜŠð7Å\SCÓ¯~£òºØgy ïÌ>Æƃ¼§bv'32‡ €XܬnVzŸl¿âçæqÙo3Xáq„¾9Àö@w¢Ÿ¥è†€ÅÀV„‰*ƺ§—û?ã¸äçË~ÓŸyŒŒÊíü{c!pž^€ã €ã €A‡1¸¢÷É'ÂÚ˜2¬Ø_¹e9X ;Ñß2Û ÐÞ©óç_qWÀT=ñ©ªÏ¦-è¯~êתÇ6m îåÕ1ý%Çô‚Ý÷Ð/UÏÞÿik~ðϪOüƹê÷Ò&t¡þyºûsê•;î­~ÿ×®zGÚ¦ Z=ýßÿÝê3ßM› Ôìܯ¯@·f×÷ çºwßC¯Î)·¦­}ŒçнúÇ06e;ìáf¹1CUýÞs·úÄ‹š¶&Nø $ümîÎêñ'?RýjÚ‚cͪïAî=€ùù½ç¾°º¦½–¶€’ €ØAøËÜ€ášY Ä ƒ„¿í é–xiÜ{ó%†i°…ð—¹1Ã&=+üþïÎêñ߸·º/mÁ±„ÀKáÞ˜?!0”O À Â_æÆ »Mr Vø $üíÌm?W=+¦CBà¹sï,‡Ê&`ƒð—¹1ÃaÓˆ]ÓÂ_`/áoç„ÀtL0Àâ™0ôgœØ1 âøYw%ú8û½òâß«þÃçþUÚ‚ã\©îªþþ“­úù´MWúê'Æýt<u8uî`—Ã×›ëÍuŒˆñ8¦`JÀ³2çŽX/¬ZŠ¡nˆÈñrwa3®+öÝføÛÄØaó3ö1%,cnÓÝêçµsG—šȗȃ–ýÑŸŽ¾öÕÉûiŸ7Ú36ÇL‰) ˜ˆ¸Ñ¹´j9ü ÂßáÄÏ:‡¾M»›áo¾a ?°&üeLÑGSÈÏ[¾/臀 9½j›Á¡AªaEœ»±Þ$€Ïƒç*îÍ0–LøK ¢&”‚áä{¤Í{\€n€˜¸9ÎÁaLC¬xXùç݇6ám„¿ñ.çÜäá/åðâþ(Z|ö<ä tC ÀÄä㸄¹9^c:îØM«€£Š;‚àæÀ’ )UôÕ¢ÃR‰tC ÀÄÄ q~ŸlT“2ŽèBänD“8~m hä}(Ä`©„¿LAôÛÌÚÃËaäÐŒ€ Šà£‚T€8¼øæ}ÐDì»þÆzüY1¨h`€%þ2%Ñç@A{]Ì|ŸAŸC >0ÁaLCœ«€ÁÊŸÞÑÈ¡pyðbs*hÓA°Â_¦*÷áê¶ü_´è3Æò䯥ÈÁ+>7q|ùêÀ.`&*n|ã}²Æ€R“’nÄÏ<º±B“} ›¡½ `–@øË’D_/·è¯o{`3¾.Èb)º ã3!°)€ÝÀLXÜøæ 8^„¾±šv)b¿Å àf0Ì™ðn”ƒa!0K÷=gÖ«70añÔs~Šz[EÈ8Wÿ6©ŽnÈføkà€¹þÂnç)mõé™»¸_êk86ÁÂ`@ Àäå›ÛxŠZ€8ŽúÆ>Ø ƒ‰}~üúXnI ú™€9þB=G_PÌt5ô.‚`X:03•uƒGº—ÃÛ&pÈy*èø½¦ƒ`.„¿ÐL„Àd)âh¨ Øç –F À Ä m¼8ÂCïOs«ëä ‡Á æ@ø íå)¡a úC<¬›Ã`` ÀÌD¿96X4ŽÍð7¦ƒ®+ö[ì³\¬€©þÂqòû€õëYŠ!Bà,Á¦\€90ç›f7±ãÉp ÖE7#Öëˆ_ƒ98VÉ ÀT ¡Ñô^`–$îg‡ª™—¢Å}07`f$߸ªßfp8ö]ðˆõ؇!ý`*„¿Ð½¨„À,EÜC1#R܃åæ³s `†¢ZÀ%n<¹ 8öCrëÈy*èÐd*i“ðúz0¥¨;‹R_rUp´X¦Èè837©9t4ðxbÐ"ºñôx„ë8ùî+}LÁ€áouWõ÷…¿,Rô+óC‚0w'ï‹Æ’?w¹S!`†"ø'•#6}Õ¸"øm:tì¿\¬€Ò þ>ùתŸOÛ°LÑWôº– î‰J ‚3aðxüÌhæÔùó¯˜?ªžøÔ @Þ÷ÐÃÕ³÷ßš¶¦"ß(7†˜nE7#·¨«ÓíÈáoÆÿ²—Aõûú¥Õ1ýi‹eé'¨©Luÿ÷› –q ~.ð·då‡ýûè+N¡®aì¾õ•¼¯J?Ïă»êŒúÓvÿ#¢kŽ)˜0À¬èˆ]/‡À<ºOîjäéœë†ÀáÛ=ÓÌ–K ÝÓ½é]§Àã‰Ð)‚à:3ÍŒE<%﫸§Ûëqòç6îw­gC}ÆÇúû7ÿ®.#¢kŽ)˜£áÌ\T»Ü+n–ã†6WjÔ©ÑL Kó{ÏýÝê/þiÚ‚}"‹~f_á ”bŽ÷´›ŸÝ]ëYþZß-l®g›¿¦Ý3"ÀŒE€˜§Ž ˜ñlv9ê†À¥½ï Š˜ú"¾°^®0sy@(‚`OÖŽ+‚ßØu+#À©Î€%ÓLôùõ €˜¹:»^5 4²úF_· ¢«K&¦¾ü°§>?U`¢’4¨.Cî~D |¨8*¸€%‹ø§þÎËi ö‰¾c´õûX.0 ÓÂÕ™z˜þäŸ,se6À~§Î}ILG,`™À,DT› º ¹ò÷ÒªEWD(&¦™ãPX0 @Ñâò§`\úÆ~È]!0p˜˜æâAP€°,`$BߨöN°rä}P˜v„À,‡€…‰ €\y*Wž :ãb¨êÓLL̓ ,ƒ€…Š© ŽãË!ðé«[ö P—˜æ"B`æM ÀEpžvØTpã‹Ð7WÔ'¦¹x4š€ù2Ò ÀB埸ü_®Î  !0í˜/0 ƒ=Qœ§c\9ô© £ €ú„À´!pÌó"`Áb°ç̪EØèéÿñånIì 0ÐŒ˜ö"€ùÀÕ*`c9b_äPŸ˜vâÄÜ`úÀ,\®|Æ·þÆTÐÍi'ª€U0`¸:У ¸964ÐÎ:þöj-öë»1/ö)Ó'€«AcT_Në”!º)ª€€vN{¡ú©¿óGi šˆX50Ó%€«b'.‹*NË÷C„òíi/ U À4:þ¥N³ñGÕŸz¡úlÚêÊ}=\={ÿ­ikÎn^µxÒÿìÕ-Æ–»(/¯Žëç;?®¡+÷=ôK«säO¤­:â˜þ’cš¢ýê§~­zügÓÆ^ÝÏWúý_ÿ¹êi›úç¿[ýô“ßJݨߟí§? ]ºrïGªó¿|gÚ¢[úi”¯Ùëe[Θ# O0¼)žðÏá¯ÊÓñå*à;«ÇŸ| úÕ´%iþ†8¦?â˜þöM?¹ù‹«cúÃŽiŠ$ü…~ €à:QÍ%² 9Ž›Ö¿î¦•¢´ 3ƒ‹0m¦…†'üŠ~s#¦<Â_èŸÑm¸NL;UÀWíR|ÑE]–ßM+Å8.üÍ .´y›0áïÐôÓ˜!0åþÂ0Àpƒ¨ê9³^¥Ç´Ü?㦕Ñuþfašn^µ˜1 Âß±è§17B`Æ'ü…á€`§U—#BàÓ«1nZG·áofp¦+S ôGø;6ý4æFÌx„¿0,0lº1t\*…Àeˆ8Ù£:û/¹iepý„¿™ÁE˜ž¨Q ìÀ@÷„¿¥ÐOcn„À Oø ÃÀN1˜›«N)Cž :º0*N¿áofpXþ–F?¹3á/ŒC µ¨.Gå£X%0ý&üÍ .Ât©º!ü-•~s#¦Â_öŠÁÜU—#Oû$ÖÝ´ÒŸaÃßÌà"LGž:Çþ–N?¹q?M„¿0.0p¼ßOp9"ønL^ºi¥{ã„¿™ÁE˜.!0ÐŽðw*ôÓ˜÷ÓtOø ã@-¹âT\Ž¼O‚J`º5nø›\„i8Yô€f„¿S£ŸÆܸŸ¦;Â_(ƒj‹Šž8^NKÆ“§‚ÎUÀÁM+Ç+#üÍ .Â4 €ú„¿S¥ŸÆܸŸæxÂ_(‡‹é ]B˱YÜ´Ò^Yáofp¦ÉTÐÀaÂß©ÓOcnÜOÓžðÊbô‰Áܳ«•§ª{Æ·­ 8¸i¥¹2ÃßÌà"LSôô€í„¿s¡ŸÆܸŸ¦9á/”G Å`n®6e|›áo®nZ©¯ìð73¸Ó$n$üý4æÆý4õ ¡L`h-ÞlP· Q|r*èকæþfwVÿƽÕ}i ˜ ÓA×\¹÷áï, ™÷Ó&ü…r € •<t\J…ÀãÛ5tpÓÊnÓ “Û~®zV $"üýpuþ—ÿ½´Åü™÷Óì&ü…² € µȽ¸^¥›áo,U³ß$ÃßL %†%‹òó¿üÓVœœæIÌܸŸæFÂ_(ŸŽUÀ1õ°*à2DðÉð7sÓÊ5“3!0LÆîr!ð< ™÷Ó\#ü…iÀQbÀ&ÂßüþYÆÁo·ÁnZÙ ã‰Ã {`i”«ž'!0sã~á/LÉ©óç_™ÁÈŒ)BÆ€/¯Ú™ø£ÊïŽ`>öI8ÔݹyÕòïÛ´íkpR›à&®=G`Æ97°‹sPŸ `8Z„„ñ.àþæБqD`›ƒÛ&] «–UbF€,ü /òmâÜàüÀqÀЙ1¸–«€c?æ_/Ì`lÑ?U À~`èM†¹1¾r›ò1¸–«~cM0}R Ô•ÀÎì&€ÞDðUÀ Ç—Ãß:UÀYì»Í*`è“@h"Ϊ¸‘zƒ2þª-Ct}r÷§N¨û/¦‚yPþ⪠ôè‹hB507@ïòå6‚CÆ—§s®[Õj9Ž¶3«&РOqݹ¼^¨%ÎR`M ½ËOåGpLÓ6®\pÝ8ÂßØoyß\ oñàXôêŠ>ªó`PÁ Fè(WÞMª€ã×F¼9¥·ý@ß„9@SqÞˆ{–J ƒÉ¡c‚O£5é„ß}MÅý†sÀR €`P1svÕâláqm†¿u‚ÜØwQÕ¿›ƒiö#CˆkOnø9@]ÎKtêüùW” À "@¼¼j>æ©„Gtƒ¢ÅþˆÖD„ÁJW¢¯±ëa¥xêä@üú|mÜfÛïÙg쿸FÿuÚù@}*€`p1UÀ1x©zt\1ˆÝ¡¼¬#œå˜‹«f?Pª}רmAjüú}E5 _ÇþûkâU4Þ °`E„‡gÖ«Œ,’s\Gì»<tˆýë7­"€¡D¿77æL £ŠÀPõè¸"øÍ-‡º‡DåDì·ÍiØ"Öµ tª€æÎ(%Œ&ÂØ zß»ñNîÕ©Žª‰¨Ž¶ÛLAôa7û±Ì‰Fƒ.ñ¾Kòø"Ôð·î¾ÈƒfùÀ!W› €)ˆþ¬i¡æÆh3Œ.ª€#xŒwÈ2žÍð·n)¾'«'t±˜Šx 5š`.ŒNÀèr%i¼C–qÅ WtêVç÷ç*àˆ 8Þó›ßõ»Oì»ü>ç¼cÚLQîËæ¾-S!€"Å“öª€Ç—à&b€,¿8~®&€©Š>® `*ÀPœX‰JÒ¸L{÷Vb_Ô™z3üÍd‚|æB 0`(R ¬¨-GÞuºN±ï"¸÷N`æ(ú»‚`€’‘€¢Eõ¨*à2D·©îtÐñëb¿åªaï`n"Ž(Š*1tÓwÐÒŸ¨Ž}rHì»ØoñëcP,W ‚˜“èßæŠàh”@ E‹ð0š*àrD˜›Ý}bŸåÿb|a¥Î”@ E‹'ês°¸uª€cÅ{€CLåT0w›Áú¿c@ñòÀ‰© Ë’ÃÝ}ò¾‹_»94,Aô£O ÀÀ0 98T<=±ïb¿ ðXª\, ‚&!”ËR· 8ö[t»¢"3K$è›&#ÂÃhª€Ë!ð¡`>öY tÅ»ƒóþ34K}ãhîmú €ÉˆŠÑ#l4P2-±¿rµðå´€¥ËJª è’&%F¢ ØTÐe‰ýqh:èØw9Ä¿_Þ”ƒ`¯J8Ö©óç_Ñ«€I‰1¦Ž§åÁåÙW½!q„¿ñ žçðæcß¹†eä&'¿ Xø;=œYµè‚©º'€ÉŠØ»€Ësh*èüž³‚í? [`˜¤ü‡¨"–g_û+ÿÿËi Ð 0LV„À¦‚ž¦ü®ÈxŸ³èŽfAˆXžCSAïº%€I;Yœ§†¦ ûBàü`ÓxÝÀ,D%i„ˆ§®n1±ß"ÝÀäEÕo´ó:åØW|yÕòÿu€ã€`6b*è¨V\ž³«¶k¿Dpˆ_#Ž#€Yجú5•py"üÍïiÞ%|ÕÛÀqÀ0Fˆ˜ƒFabY¢Ûµk:討 `ïŽ#€Ù¹¸j"š ºL»*#¼ 8¿Ë 90ÌJ‡ÑT—kWû*W«ÚÀ,E5iP\¦}SA‡3«&¼šÀìl‡*I§)|ûhN ³!¢© Ëµ¯ øìªE°hF ³”ߣ™ ºL»Bàï#Þõ¾`€íÀ0[Gø« xzb_åpøbZ&€Ù‹jÒ  ¸Lû¦‚1´ð`qþ×Òm-´ò eÀìåð0©‰eÚ‡00¼ôæ ød‹óóåU€r€`bð:¦‚Ž*à°f¢ øìªE°€8L ‹«”¢™ ºLû¦‚Ž¦Ûf$#Âߨª€Ë´«k–ÃaSAû €`Qb*è  ¸L1Õó¶}À!øÛ €`QråoT’ª.S¼ówWо*``70,N„ˆ0ªž–¨ÎáðåøÀ À°8Qù{qÕT—+BÞ{OŠ8š.°ÑCX¤¨ ÍÓ ËÚ7¯W¯“¿f*hàF`X¬˜ :BFSAOKT‡]ï –L ‹•+c© ¸\Ûª€Cøª€úãÚÀ4 €`ÑTOÃÉ.[TÇ;‚£ X@Ð×F¦I ‹ááÅU‹JRAb¹Î¤å¦£ (€kÀ°x—W-*I#H—kÛTÐùk¦‚èžk"Ó$Vb*è štZ¢8D€@·\˜&0°²Yå¤â©\Ûª€ƒi¼ºçœ À4 €€Dð4œ £ 8Þ¬  [®‡L“H¢ÒI%é4E¾°–N l¸¼jQIA¢¸\Û¦‚>›–àp¼¸&Àô€€¢’4½U“–íd7.ª€ƒ© `ÉÀÀ Qù›»ª€ËïýÝ&*€UÀR €€-¢ 8BDUÀe;9tTGp4á=,‘Ø"ÂÃh §%Bà<7°4``‡ F°¸\'«€ÃÙU‹*`SA´w1-`ZÀÀ› º|Û¦‚K"öجüU<=±ÏL Ð\L¥Ó$ˆ `UÀå‹*àÍ}UÀ±Ý=!0,…¨IˆX¾“]»¨`‹¦ËKa4¨!*€M]¾xçïæ{Má K#ˆ ˜²E|v½ú¦‹«f o O]?dÞ¶¶Ë¾ÿ×F× çÔù󯸳jŠñLZR®èÞÅÔÏY~ðfu0@—â½ã@ T œ )S¾7¯W¯Šý–ŸùóìÌ™hAˆ8 ›!pˆ)¼UoÀœ €€TOÏåÔ„÷@N>pŒE ´ä}Ópr*há=Ì™8‚jÒi8–YL Ì‘8‚jÒiØGx/À€¹G2ô4œ|?g„÷§Ö«ð`(`±T]ÀØÀ@TOö*``NÀ@G„‰Óp2Þ]S c1ðtÙw@ׄÀ00С¨&&–oÛTÐöÌèØÅU»¼^¥`§Ó2]S c‹ ñÒz•‚ €#¸]ÀÐÀ@"HŒé )Û¶© €)°A0Ð5UÀ0$0Ð#UÀå‹`æÔzõª¨6…7Ð5!0 E ôL\¾³©eÛo@×N¾{胀D¼ù° Ð00Õ¤å‹i ONÑj¿]Ûv®º$"Lœ!00!0ôE 0``@ªI§A0L•˜0qN§ev!-€’ €Øâd|%-€’ €€Dðåõ*34LÉÅU»´^eB÷P200"pùNVGp”J ŒÌ´Âå34L…€®¤%P0P¥å‹*àÍ®ã…U3…7”F B\¾3«vv½zUÀö”D 0`  ªIËwjÕb:èMö”B F˜8 §Ó2³ß `Z8%RM: ¦‚€Ò€€B]NK¦%B`A0ŒE êbZR¶“UÀÀ˜À@ÁT’Nöؾ€1€è‰†& 'DœSA@ ÀÀD|e½JÁL cгÁ000Víâz•‚Eðéõ*0800!—WM< Ûº™ª€ o``b"¦lQ|f½z!0ôI Lqb:èm®¤%Ð50‹÷9}¥ xvUÇþ³ k``ˆӰ+Þé ]0¢‹i tA Lœ*àiØWl@WÀÀ \HKʶ/¾”–À1ÀÀ \IK¦K ]3aái84´ýÇ0°C]PSz@[``FTNÙ´ÜÅ”ÞЖ˜!ð4ÄTЇ¦ƒš0¢Ói¹š3$8œŽ}0Д˜)!ðtš Ú¾€ºÀàP%°ê3¡áåõ*…34tA Ì\ÀWÖ«.¦‚>»^ÝJ0"f.à‹ëU&âLZns)-€mÀÀD°êÑi8•Ú.àeñ¹hJ @a"64™ý ЄXAÒtD¼o*h–Åg .0°0‚¤éØ×Uµ—çBZ°€‚ÅTл*#öNàåˆwy  ¤zt:b*hÓA“EŸßÜ8éÔùó¯Ä( ÀÂD èY¸é8TýysZ2/S y‹õ ñ F=€…º˜–LGL½‹ ‰R8€q €€ÔLGLLÍ8·ã0ù}À»Ý@ ,œÐpzb*è]ÝØxW0” Î-ŽG`x`!ð„ä*àèÆžŽ/œp!-¡q<:¿Ã0AÑÝ5´ÀÒÄ1鸆!¸J83=ŸY¯W €ÞÓµ^^¯2¹xÛTÐ}J¤èŸàMWÒ’éˆîì®.­ R9×ý\çâªE%0ÓrvÕv½JãôG pƒ¨ÎS¡79ø®íÉé US*Ç&Ð0ÀV*ô¦%B`SA35ùa“C-lûz´°íkáä×s Û¾¾«…m_¶}-œüznÁçúrêüùWò87¸9-)_îÖ^^µKëÕ7ÙÓ$$hJ03UÀ¹™ €eì%8œžÀò€OK€Ï\ݺÆ~`þÀÌPc*èÍé /¤%Ì“ Õ£Ót²»{%-`žÀµ©–üàèònV ó˜/0@mªG§'O6C`˜'0@#ªG§)º½9ö#ó$h,*UOK®>sukM Àü€‹w €§gs*hÝ`æÉÈ@+WMéôD¼9ô¥´€y° 9ønpT_¾ºs!8Š*àé‰x3€ù0âp4!ðôÄ;œsW8Þé ó `rp´˜ æA Ð UÀÓ”»Ã—Ò¦M ÐS OSTë0Fº:ï•ez"—Ó¦K Ð)SAOS~°© ˜60¼I÷€i3ÂÐ9UÀÓ”«€Må Àt €z!ž®üN`ÆwsZP—`&À½‰*àKëU %UÀM€z°é á8B`€ºÀ3!„*`8ÎÙ´`0À`¼è—`0`hïÔª^¯°“`P¦‚†öÀ‡€fB 08UÀÐÞÍi À6`€Q¡=!0À.`€™Œ&ª€¯¬W†Tl#Õ…U34Ð 00Qª€NA0´#Ø$(†wÇG ͨÈÀE¹–Í €Š!°wC3ª€‚ X¦‚†f„À`€bE%°¨O P4!04£ X60@ñ"¾¼^ØC P¼‹i©êQ ,—`"Žé zN§%À²€&åRZû¹Õ–ɨÀ¤DSA›ö;•À²€&'¦‚ÃagÓ`9À“•À¦ƒ®'˜¤\¬ö»9-–A 0YQÓAA0 ˜7VíÔzØB0°`€Y¸œ°X0À,\LË`:hX*0ÀlDï64l§ ˜?0Àì˜ v;“–ó$˜•¨ÎÓA› nä˜7£³ôƪ™ ¶340_`€ÙŠjàhÀR€f+‡¿¦‚†©æI 0k_^¯ ‚ᄳi 0`€Eˆ Ø;áz>Àü€f/¿ 8Wט ˜0Àb\LË`:h˜#0À¢äj`SßÂ5ª€€ù,J„¿ÁtÐp=·FÀ<åXœóЦ‚†5UñÀ<€)‚_SAÃ5§Ó`ÚÀ‹S@ç÷kÞ LŸ`Ñ6`ÓAÀÔ €ïâªEøk:hP L`ñb:è<%40e`V"üÍS@› š¥S L—€$‚ߘ:¦‚À €H"ôÍSA{0L‘€6߬˜¥2 40M`¶¸–¦ƒ€)°E~°é Y2UÀÀô€Ø!Þœ©€)°GT¿±jª€` ÀÔA0,‘i €iPCÀ1%´©  d`jº¸j`–èLZ”O @G%0,‰Û&`:ŒdÐ@L­˜%:•–%«ªÿ(µï¦À‡jBIEND®B`‚rocksdb-4.5.1/examples/000077500000000000000000000000001267530634000147745ustar00rootroot00000000000000rocksdb-4.5.1/examples/.gitignore000066400000000000000000000002331267530634000167620ustar00rootroot00000000000000c_simple_example column_families_example compact_files_example compaction_filter_example optimistic_transaction_example simple_example transaction_example rocksdb-4.5.1/examples/Makefile000066400000000000000000000036431267530634000164420ustar00rootroot00000000000000include ../make_config.mk .PHONY: clean librocksdb all: simple_example column_families_example compact_files_example c_simple_example optimistic_transaction_example transaction_example simple_example: librocksdb simple_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) column_families_example: librocksdb column_families_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) compaction_filter_example: librocksdb compaction_filter_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) compact_files_example: librocksdb compact_files_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) .c.o: $(CC) $(CFLAGS) -c $< -o $@ -I../include c_simple_example: librocksdb c_simple_example.o $(CXX) $@.o -o$@ ../librocksdb.a $(PLATFORM_LDFLAGS) $(EXEC_LDFLAGS) optimistic_transaction_example: librocksdb optimistic_transaction_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) transaction_example: librocksdb transaction_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) options_file_example: librocksdb options_file_example.cc $(CXX) $(CXXFLAGS) $@.cc -o$@ ../librocksdb.a -I../include -O2 -std=c++11 $(PLATFORM_LDFLAGS) $(PLATFORM_CXXFLAGS) $(EXEC_LDFLAGS) clean: rm -rf ./simple_example ./column_families_example ./compact_files_example ./compaction_filter_example ./c_simple_example c_simple_example.o ./optimistic_transaction_example ./transaction_example librocksdb: cd .. && $(MAKE) static_lib rocksdb-4.5.1/examples/README.md000066400000000000000000000001701267530634000162510ustar00rootroot000000000000001. Compile RocksDB first by executing `make static_lib` in parent dir 2. Compile all examples: `cd examples/; make all` rocksdb-4.5.1/examples/c_simple_example.c000066400000000000000000000045241267530634000204530ustar00rootroot00000000000000#include #include #include #include #include "rocksdb/c.h" #include // sysconf() - get CPU count const char DBPath[] = "/tmp/rocksdb_simple_example"; const char DBBackupPath[] = "/tmp/rocksdb_simple_example_backup"; int main(int argc, char **argv) { rocksdb_t *db; rocksdb_backup_engine_t *be; rocksdb_options_t *options = rocksdb_options_create(); // Optimize RocksDB. This is the easiest way to // get RocksDB to perform well long cpus = sysconf(_SC_NPROCESSORS_ONLN); // get # of online cores rocksdb_options_increase_parallelism(options, (int)(cpus)); rocksdb_options_optimize_level_style_compaction(options, 0); // create the DB if it's not already present rocksdb_options_set_create_if_missing(options, 1); // open DB char *err = NULL; db = rocksdb_open(options, DBPath, &err); assert(!err); // open Backup Engine that we will use for backing up our database be = rocksdb_backup_engine_open(options, DBBackupPath, &err); assert(!err); // Put key-value rocksdb_writeoptions_t *writeoptions = rocksdb_writeoptions_create(); const char key[] = "key"; const char *value = "value"; rocksdb_put(db, writeoptions, key, strlen(key), value, strlen(value) + 1, &err); assert(!err); // Get value rocksdb_readoptions_t *readoptions = rocksdb_readoptions_create(); size_t len; char *returned_value = rocksdb_get(db, readoptions, key, strlen(key), &len, &err); assert(!err); assert(strcmp(returned_value, "value") == 0); free(returned_value); // create new backup in a directory specified by DBBackupPath rocksdb_backup_engine_create_new_backup(be, db, &err); assert(!err); rocksdb_close(db); // If something is wrong, you might want to restore data from last backup rocksdb_restore_options_t *restore_options = rocksdb_restore_options_create(); rocksdb_backup_engine_restore_db_from_latest_backup(be, DBPath, DBPath, restore_options, &err); assert(!err); rocksdb_restore_options_destroy(restore_options); db = rocksdb_open(options, DBPath, &err); assert(!err); // cleanup rocksdb_writeoptions_destroy(writeoptions); rocksdb_readoptions_destroy(readoptions); rocksdb_options_destroy(options); rocksdb_backup_engine_close(be); rocksdb_close(db); return 0; } rocksdb-4.5.1/examples/column_families_example.cc000066400000000000000000000037451267530634000221750ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include "rocksdb/db.h" #include "rocksdb/slice.h" #include "rocksdb/options.h" using namespace rocksdb; std::string kDBPath = "/tmp/rocksdb_column_families_example"; int main() { // open DB Options options; options.create_if_missing = true; DB* db; Status s = DB::Open(options, kDBPath, &db); assert(s.ok()); // create column family ColumnFamilyHandle* cf; s = db->CreateColumnFamily(ColumnFamilyOptions(), "new_cf", &cf); assert(s.ok()); // close DB delete cf; delete db; // open DB with two column families std::vector column_families; // have to open default column family column_families.push_back(ColumnFamilyDescriptor( kDefaultColumnFamilyName, ColumnFamilyOptions())); // open the new one, too column_families.push_back(ColumnFamilyDescriptor( "new_cf", ColumnFamilyOptions())); std::vector handles; s = DB::Open(DBOptions(), kDBPath, column_families, &handles, &db); assert(s.ok()); // put and get from non-default column family s = db->Put(WriteOptions(), handles[1], Slice("key"), Slice("value")); assert(s.ok()); std::string value; s = db->Get(ReadOptions(), handles[1], Slice("key"), &value); assert(s.ok()); // atomic write WriteBatch batch; batch.Put(handles[0], Slice("key2"), Slice("value2")); batch.Put(handles[1], Slice("key3"), Slice("value3")); batch.Delete(handles[0], Slice("key")); s = db->Write(WriteOptions(), &batch); assert(s.ok()); // drop column family s = db->DropColumnFamily(handles[1]); assert(s.ok()); // close db for (auto handle : handles) { delete handle; } delete db; return 0; } rocksdb-4.5.1/examples/compact_files_example.cc000066400000000000000000000133661267530634000216370ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // An example code demonstrating how to use CompactFiles, EventListener, // and GetColumnFamilyMetaData APIs to implement custom compaction algorithm. #include #include #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/options.h" using namespace rocksdb; std::string kDBPath = "/tmp/rocksdb_compact_files_example"; struct CompactionTask; // This is an example interface of external-compaction algorithm. // Compaction algorithm can be implemented outside the core-RocksDB // code by using the pluggable compaction APIs that RocksDb provides. class Compactor : public EventListener { public: // Picks and returns a compaction task given the specified DB // and column family. It is the caller's responsibility to // destroy the returned CompactionTask. Returns "nullptr" // if it cannot find a proper compaction task. virtual CompactionTask* PickCompaction( DB* db, const std::string& cf_name) = 0; // Schedule and run the specified compaction task in background. virtual void ScheduleCompaction(CompactionTask *task) = 0; }; // Example structure that describes a compaction task. struct CompactionTask { CompactionTask( DB* _db, Compactor* _compactor, const std::string& _column_family_name, const std::vector& _input_file_names, const int _output_level, const CompactionOptions& _compact_options, bool _retry_on_fail) : db(_db), compactor(_compactor), column_family_name(_column_family_name), input_file_names(_input_file_names), output_level(_output_level), compact_options(_compact_options), retry_on_fail(_retry_on_fail) {} DB* db; Compactor* compactor; const std::string& column_family_name; std::vector input_file_names; int output_level; CompactionOptions compact_options; bool retry_on_fail; }; // A simple compaction algorithm that always compacts everything // to the highest level whenever possible. class FullCompactor : public Compactor { public: explicit FullCompactor(const Options options) : options_(options) { compact_options_.compression = options_.compression; compact_options_.output_file_size_limit = options_.target_file_size_base; } // When flush happens, it determines whether to trigger compaction. If // triggered_writes_stop is true, it will also set the retry flag of // compaction-task to true. void OnFlushCompleted( DB* db, const FlushJobInfo& info) override { CompactionTask* task = PickCompaction(db, info.cf_name); if (task != nullptr) { if (info.triggered_writes_stop) { task->retry_on_fail = true; } // Schedule compaction in a different thread. ScheduleCompaction(task); } } // Always pick a compaction which includes all files whenever possible. CompactionTask* PickCompaction( DB* db, const std::string& cf_name) override { ColumnFamilyMetaData cf_meta; db->GetColumnFamilyMetaData(&cf_meta); std::vector input_file_names; for (auto level : cf_meta.levels) { for (auto file : level.files) { if (file.being_compacted) { return nullptr; } input_file_names.push_back(file.name); } } return new CompactionTask( db, this, cf_name, input_file_names, options_.num_levels - 1, compact_options_, false); } // Schedule the specified compaction task in background. void ScheduleCompaction(CompactionTask* task) override { options_.env->Schedule(&FullCompactor::CompactFiles, task); } static void CompactFiles(void* arg) { std::unique_ptr task( reinterpret_cast(arg)); assert(task); assert(task->db); Status s = task->db->CompactFiles( task->compact_options, task->input_file_names, task->output_level); printf("CompactFiles() finished with status %s\n", s.ToString().c_str()); if (!s.ok() && !s.IsIOError() && task->retry_on_fail) { // If a compaction task with its retry_on_fail=true failed, // try to schedule another compaction in case the reason // is not an IO error. CompactionTask* new_task = task->compactor->PickCompaction( task->db, task->column_family_name); task->compactor->ScheduleCompaction(new_task); } } private: Options options_; CompactionOptions compact_options_; }; int main() { Options options; options.create_if_missing = true; // Disable RocksDB background compaction. options.compaction_style = kCompactionStyleNone; // Small slowdown and stop trigger for experimental purpose. options.level0_slowdown_writes_trigger = 3; options.level0_stop_writes_trigger = 5; options.IncreaseParallelism(5); options.listeners.emplace_back(new FullCompactor(options)); DB* db = nullptr; DestroyDB(kDBPath, options); Status s = DB::Open(options, kDBPath, &db); assert(s.ok()); assert(db); // if background compaction is not working, write will stall // because of options.level0_stop_writes_trigger for (int i = 1000; i < 99999; ++i) { db->Put(WriteOptions(), std::to_string(i), std::string(500, 'a' + (i % 26))); } // verify the values are still there std::string value; for (int i = 1000; i < 99999; ++i) { db->Get(ReadOptions(), std::to_string(i), &value); assert(value == std::string(500, 'a' + (i % 26))); } // close the db. delete db; return 0; } rocksdb-4.5.1/examples/compaction_filter_example.cc000066400000000000000000000054611267530634000225250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include class MyMerge : public rocksdb::MergeOperator { public: bool FullMerge(const rocksdb::Slice& key, const rocksdb::Slice* existing_value, const std::deque& operand_list, std::string* new_value, rocksdb::Logger* logger) const override { new_value->clear(); if (existing_value != nullptr) { new_value->assign(existing_value->data(), existing_value->size()); } for (const std::string& m : operand_list) { fprintf(stderr, "Merge(%s)\n", m.c_str()); assert(m != "bad"); // the compaction filter filters out bad values new_value->assign(m); } return true; } const char* Name() const override { return "MyMerge"; } }; class MyFilter : public rocksdb::CompactionFilter { public: bool Filter(int level, const rocksdb::Slice& key, const rocksdb::Slice& existing_value, std::string* new_value, bool* value_changed) const override { fprintf(stderr, "Filter(%s)\n", key.ToString().c_str()); ++count_; assert(*value_changed == false); return false; } bool FilterMergeOperand(int level, const rocksdb::Slice& key, const rocksdb::Slice& existing_value) const override { fprintf(stderr, "FilterMerge(%s)\n", key.ToString().c_str()); ++merge_count_; return existing_value == "bad"; } const char* Name() const override { return "MyFilter"; } mutable int count_ = 0; mutable int merge_count_ = 0; }; int main() { rocksdb::DB* raw_db; rocksdb::Status status; MyFilter filter; system("rm -rf /tmp/rocksmergetest"); rocksdb::Options options; options.create_if_missing = true; options.merge_operator.reset(new MyMerge); options.compaction_filter = &filter; status = rocksdb::DB::Open(options, "/tmp/rocksmergetest", &raw_db); assert(status.ok()); std::unique_ptr db(raw_db); rocksdb::WriteOptions wopts; db->Merge(wopts, "0", "bad"); // This is filtered out db->Merge(wopts, "1", "data1"); db->Merge(wopts, "1", "bad"); db->Merge(wopts, "1", "data2"); db->Merge(wopts, "1", "bad"); db->Merge(wopts, "3", "data3"); db->CompactRange(rocksdb::CompactRangeOptions(), nullptr, nullptr); fprintf(stderr, "filter.count_ = %d\n", filter.count_); assert(filter.count_ == 1); fprintf(stderr, "filter.merge_count_ = %d\n", filter.merge_count_); assert(filter.merge_count_ == 5); } rocksdb-4.5.1/examples/optimistic_transaction_example.cc000066400000000000000000000101511267530634000236050ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "rocksdb/db.h" #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/utilities/transaction.h" #include "rocksdb/utilities/optimistic_transaction_db.h" using namespace rocksdb; std::string kDBPath = "/tmp/rocksdb_transaction_example"; int main() { // open DB Options options; options.create_if_missing = true; DB* db; OptimisticTransactionDB* txn_db; Status s = OptimisticTransactionDB::Open(options, kDBPath, &txn_db); assert(s.ok()); db = txn_db->GetBaseDB(); WriteOptions write_options; ReadOptions read_options; OptimisticTransactionOptions txn_options; std::string value; //////////////////////////////////////////////////////// // // Simple OptimisticTransaction Example ("Read Committed") // //////////////////////////////////////////////////////// // Start a transaction Transaction* txn = txn_db->BeginTransaction(write_options); assert(txn); // Read a key in this transaction s = txn->Get(read_options, "abc", &value); assert(s.IsNotFound()); // Write a key in this transaction txn->Put("abc", "def"); // Read a key OUTSIDE this transaction. Does not affect txn. s = db->Get(read_options, "abc", &value); // Write a key OUTSIDE of this transaction. // Does not affect txn since this is an unrelated key. If we wrote key 'abc' // here, the transaction would fail to commit. s = db->Put(write_options, "xyz", "zzz"); // Commit transaction s = txn->Commit(); assert(s.ok()); delete txn; //////////////////////////////////////////////////////// // // "Repeatable Read" (Snapshot Isolation) Example // -- Using a single Snapshot // //////////////////////////////////////////////////////// // Set a snapshot at start of transaction by setting set_snapshot=true txn_options.set_snapshot = true; txn = txn_db->BeginTransaction(write_options, txn_options); const Snapshot* snapshot = txn->GetSnapshot(); // Write a key OUTSIDE of transaction db->Put(write_options, "abc", "xyz"); // Read a key using the snapshot read_options.snapshot = snapshot; s = txn->GetForUpdate(read_options, "abc", &value); assert(value == "def"); // Attempt to commit transaction s = txn->Commit(); // Transaction could not commit since the write outside of the txn conflicted // with the read! assert(s.IsBusy()); delete txn; // Clear snapshot from read options since it is no longer valid read_options.snapshot = nullptr; snapshot = nullptr; //////////////////////////////////////////////////////// // // "Read Committed" (Monotonic Atomic Views) Example // --Using multiple Snapshots // //////////////////////////////////////////////////////// // In this example, we set the snapshot multiple times. This is probably // only necessary if you have very strict isolation requirements to // implement. // Set a snapshot at start of transaction txn_options.set_snapshot = true; txn = txn_db->BeginTransaction(write_options, txn_options); // Do some reads and writes to key "x" read_options.snapshot = db->GetSnapshot(); s = txn->Get(read_options, "x", &value); txn->Put("x", "x"); // Do a write outside of the transaction to key "y" s = db->Put(write_options, "y", "y"); // Set a new snapshot in the transaction txn->SetSnapshot(); read_options.snapshot = db->GetSnapshot(); // Do some reads and writes to key "y" s = txn->GetForUpdate(read_options, "y", &value); txn->Put("y", "y"); // Commit. Since the snapshot was advanced, the write done outside of the // transaction does not prevent this transaction from Committing. s = txn->Commit(); assert(s.ok()); delete txn; // Clear snapshot from read options since it is no longer valid read_options.snapshot = nullptr; // Cleanup delete txn_db; DestroyDB(kDBPath, options); return 0; } #endif // ROCKSDB_LITE rocksdb-4.5.1/examples/options_file_example.cc000066400000000000000000000077541267530634000215250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file demonstrates how to use the utility functions defined in // rocksdb/utilities/options_util.h to open a rocksdb database without // remembering all the rocksdb options. #include #include #include #include "rocksdb/cache.h" #include "rocksdb/compaction_filter.h" #include "rocksdb/db.h" #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/table.h" #include "rocksdb/utilities/options_util.h" using namespace rocksdb; std::string kDBPath = "/tmp/rocksdb_options_file_example"; namespace { // A dummy compaction filter class DummyCompactionFilter : public CompactionFilter { public: virtual ~DummyCompactionFilter() {} virtual bool Filter(int level, const Slice& key, const Slice& existing_value, std::string* new_value, bool* value_changed) const { return false; } virtual const char* Name() const { return "DummyCompactionFilter"; } }; } // namespace int main() { DBOptions db_opt; db_opt.create_if_missing = true; std::vector cf_descs; cf_descs.push_back({kDefaultColumnFamilyName, ColumnFamilyOptions()}); cf_descs.push_back({"new_cf", ColumnFamilyOptions()}); // initialize BlockBasedTableOptions auto cache = NewLRUCache(1 * 1024 * 1024 * 1024); BlockBasedTableOptions bbt_opts; bbt_opts.block_size = 32 * 1024; bbt_opts.block_cache = cache; // initialize column families options std::unique_ptr compaction_filter; compaction_filter.reset(new DummyCompactionFilter()); cf_descs[0].options.table_factory.reset(NewBlockBasedTableFactory(bbt_opts)); cf_descs[0].options.compaction_filter = compaction_filter.get(); cf_descs[1].options.table_factory.reset(NewBlockBasedTableFactory(bbt_opts)); // destroy and open DB DB* db; Status s = DestroyDB(kDBPath, Options(db_opt, cf_descs[0].options)); assert(s.ok()); s = DB::Open(Options(db_opt, cf_descs[0].options), kDBPath, &db); assert(s.ok()); // Create column family, and rocksdb will persist the options. ColumnFamilyHandle* cf; s = db->CreateColumnFamily(ColumnFamilyOptions(), "new_cf", &cf); assert(s.ok()); // close DB delete cf; delete db; // In the following code, we will reopen the rocksdb instance using // the options file stored in the db directory. // Load the options file. DBOptions loaded_db_opt; std::vector loaded_cf_descs; s = LoadLatestOptions(kDBPath, Env::Default(), &loaded_db_opt, &loaded_cf_descs); assert(s.ok()); assert(loaded_db_opt.create_if_missing == db_opt.create_if_missing); // Initialize pointer options for each column family for (size_t i = 0; i < loaded_cf_descs.size(); ++i) { auto* loaded_bbt_opt = reinterpret_cast( loaded_cf_descs[0].options.table_factory->GetOptions()); // Expect the same as BlockBasedTableOptions will be loaded form file. assert(loaded_bbt_opt->block_size == bbt_opts.block_size); // However, block_cache needs to be manually initialized as documented // in rocksdb/utilities/options_util.h. loaded_bbt_opt->block_cache = cache; } // In addition, as pointer options are initialized with default value, // we need to properly initialized all the pointer options if non-defalut // values are used before calling DB::Open(). assert(loaded_cf_descs[0].options.compaction_filter == nullptr); loaded_cf_descs[0].options.compaction_filter = compaction_filter.get(); // reopen the db using the loaded options. std::vector handles; s = DB::Open(loaded_db_opt, kDBPath, loaded_cf_descs, &handles, &db); assert(s.ok()); // close DB for (auto* handle : handles) { delete handle; } delete db; } rocksdb-4.5.1/examples/rocksdb_option_file_example.ini000066400000000000000000000107511267530634000232320ustar00rootroot00000000000000# This is a RocksDB option file. # # A typical RocksDB options file has four sections, which are # Version section, DBOptions section, at least one CFOptions # section, and one TableOptions section for each column family. # The RocksDB options file in general follows the basic INI # file format with the following extensions / modifications: # # * Escaped characters # We escaped the following characters: # - \n -- line feed - new line # - \r -- carriage return # - \\ -- backslash \ # - \: -- colon symbol : # - \# -- hash tag # # * Comments # We support # style comments. Comments can appear at the ending # part of a line. # * Statements # A statement is of the form option_name = value. # Each statement contains a '=', where extra white-spaces # are supported. However, we don't support multi-lined statement. # Furthermore, each line can only contain at most one statement. # * Sections # Sections are of the form [SecitonTitle "SectionArgument"], # where section argument is optional. # * List # We use colon-separated string to represent a list. # For instance, n1:n2:n3:n4 is a list containing four values. # # Below is an example of a RocksDB options file: [Version] rocksdb_version=4.3.0 options_file_version=1.1 [DBOptions] stats_dump_period_sec=600 max_manifest_file_size=18446744073709551615 bytes_per_sync=8388608 delayed_write_rate=2097152 WAL_ttl_seconds=0 WAL_size_limit_MB=0 max_subcompactions=1 wal_dir= wal_bytes_per_sync=0 db_write_buffer_size=0 keep_log_file_num=1000 table_cache_numshardbits=4 max_file_opening_threads=1 writable_file_max_buffer_size=1048576 random_access_max_buffer_size=1048576 use_fsync=false max_total_wal_size=0 max_open_files=-1 skip_stats_update_on_db_open=false max_background_compactions=16 manifest_preallocation_size=4194304 max_background_flushes=7 is_fd_close_on_exec=true max_log_file_size=0 advise_random_on_open=true create_missing_column_families=false paranoid_checks=true delete_obsolete_files_period_micros=21600000000 disable_data_sync=false log_file_time_to_roll=0 compaction_readahead_size=0 create_if_missing=false use_adaptive_mutex=false enable_thread_tracking=false disableDataSync=false allow_fallocate=true error_if_exists=false recycle_log_file_num=0 skip_log_error_on_recovery=false allow_mmap_reads=false allow_os_buffer=true db_log_dir= new_table_reader_for_compaction_inputs=true allow_mmap_writes=false [CFOptions "default"] compaction_style=kCompactionStyleLevel compaction_filter=nullptr num_levels=6 table_factory=BlockBasedTable comparator=leveldb.BytewiseComparator max_sequential_skip_in_iterations=8 soft_rate_limit=0.000000 max_bytes_for_level_base=1073741824 memtable_prefix_bloom_probes=6 memtable_prefix_bloom_bits=0 memtable_prefix_bloom_huge_page_tlb_size=0 max_successive_merges=0 arena_block_size=16777216 min_write_buffer_number_to_merge=1 target_file_size_multiplier=1 source_compaction_factor=1 max_bytes_for_level_multiplier=8 compaction_filter_factory=nullptr max_write_buffer_number=8 level0_stop_writes_trigger=20 compression=kSnappyCompression level0_file_num_compaction_trigger=4 purge_redundant_kvs_while_flush=true max_write_buffer_number_to_maintain=0 memtable_factory=SkipListFactory max_grandparent_overlap_factor=8 expanded_compaction_factor=25 hard_pending_compaction_bytes_limit=137438953472 inplace_update_num_locks=10000 level_compaction_dynamic_level_bytes=true level0_slowdown_writes_trigger=12 filter_deletes=false verify_checksums_in_compaction=true min_partial_merge_operands=2 paranoid_file_checks=false target_file_size_base=134217728 optimize_filters_for_hits=false merge_operator=PutOperator compression_per_level=kNoCompression:kNoCompression:kNoCompression:kSnappyCompression:kSnappyCompression:kSnappyCompression compaction_measure_io_stats=false prefix_extractor=nullptr bloom_locality=0 write_buffer_size=134217728 disable_auto_compactions=false inplace_update_support=false [TableOptions/BlockBasedTable "default"] format_version=2 whole_key_filtering=true skip_table_builder_flush=false no_block_cache=false checksum=kCRC32c filter_policy=rocksdb.BuiltinBloomFilter block_size_deviation=10 block_size=8192 block_restart_interval=16 cache_index_and_filter_blocks=false index_type=kBinarySearch hash_index_allow_collision=true flush_block_policy_factory=FlushBlockBySizePolicyFactory rocksdb-4.5.1/examples/simple_example.cc000066400000000000000000000025661267530634000203200ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include "rocksdb/db.h" #include "rocksdb/slice.h" #include "rocksdb/options.h" using namespace rocksdb; std::string kDBPath = "/tmp/rocksdb_simple_example"; int main() { DB* db; Options options; // Optimize RocksDB. This is the easiest way to get RocksDB to perform well options.IncreaseParallelism(); options.OptimizeLevelStyleCompaction(); // create the DB if it's not already present options.create_if_missing = true; // open DB Status s = DB::Open(options, kDBPath, &db); assert(s.ok()); // Put key-value s = db->Put(WriteOptions(), "key1", "value"); assert(s.ok()); std::string value; // get value s = db->Get(ReadOptions(), "key1", &value); assert(s.ok()); assert(value == "value"); // atomically apply a set of updates { WriteBatch batch; batch.Delete("key1"); batch.Put("key2", value); s = db->Write(WriteOptions(), &batch); } s = db->Get(ReadOptions(), "key1", &value); assert(s.IsNotFound()); db->Get(ReadOptions(), "key2", &value); assert(value == "value"); delete db; return 0; } rocksdb-4.5.1/examples/transaction_example.cc000066400000000000000000000102641267530634000213460ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "rocksdb/db.h" #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/utilities/transaction.h" #include "rocksdb/utilities/transaction_db.h" using namespace rocksdb; std::string kDBPath = "/tmp/rocksdb_transaction_example"; int main() { // open DB Options options; TransactionDBOptions txn_db_options; options.create_if_missing = true; TransactionDB* txn_db; Status s = TransactionDB::Open(options, txn_db_options, kDBPath, &txn_db); assert(s.ok()); WriteOptions write_options; ReadOptions read_options; TransactionOptions txn_options; std::string value; //////////////////////////////////////////////////////// // // Simple OptimisticTransaction Example ("Read Committed") // //////////////////////////////////////////////////////// // Start a transaction Transaction* txn = txn_db->BeginTransaction(write_options); assert(txn); // Read a key in this transaction s = txn->Get(read_options, "abc", &value); assert(s.IsNotFound()); // Write a key in this transaction s = txn->Put("abc", "def"); assert(s.ok()); // Read a key OUTSIDE this transaction. Does not affect txn. s = txn_db->Get(read_options, "abc", &value); // Write a key OUTSIDE of this transaction. // Does not affect txn since this is an unrelated key. If we wrote key 'abc' // here, the transaction would fail to commit. s = txn_db->Put(write_options, "xyz", "zzz"); // Commit transaction s = txn->Commit(); assert(s.ok()); delete txn; //////////////////////////////////////////////////////// // // "Repeatable Read" (Snapshot Isolation) Example // -- Using a single Snapshot // //////////////////////////////////////////////////////// // Set a snapshot at start of transaction by setting set_snapshot=true txn_options.set_snapshot = true; txn = txn_db->BeginTransaction(write_options, txn_options); const Snapshot* snapshot = txn->GetSnapshot(); // Write a key OUTSIDE of transaction s = txn_db->Put(write_options, "abc", "xyz"); assert(s.ok()); // Attempt to read a key using the snapshot. This will fail since // the previous write outside this txn conflicts with this read. read_options.snapshot = snapshot; s = txn->GetForUpdate(read_options, "abc", &value); assert(s.IsBusy()); txn->Rollback(); delete txn; // Clear snapshot from read options since it is no longer valid read_options.snapshot = nullptr; snapshot = nullptr; //////////////////////////////////////////////////////// // // "Read Committed" (Monotonic Atomic Views) Example // --Using multiple Snapshots // //////////////////////////////////////////////////////// // In this example, we set the snapshot multiple times. This is probably // only necessary if you have very strict isolation requirements to // implement. // Set a snapshot at start of transaction txn_options.set_snapshot = true; txn = txn_db->BeginTransaction(write_options, txn_options); // Do some reads and writes to key "x" read_options.snapshot = txn_db->GetSnapshot(); s = txn->Get(read_options, "x", &value); txn->Put("x", "x"); // Do a write outside of the transaction to key "y" s = txn_db->Put(write_options, "y", "y"); // Set a new snapshot in the transaction txn->SetSnapshot(); txn->SetSavePoint(); read_options.snapshot = txn_db->GetSnapshot(); // Do some reads and writes to key "y" // Since the snapshot was advanced, the write done outside of the // transaction does not conflict. s = txn->GetForUpdate(read_options, "y", &value); txn->Put("y", "y"); // Decide we want to revert the last write from this transaction. txn->RollbackToSavePoint(); // Commit. s = txn->Commit(); assert(s.ok()); delete txn; // Clear snapshot from read options since it is no longer valid read_options.snapshot = nullptr; // Cleanup delete txn_db; DestroyDB(kDBPath, options); return 0; } #endif // ROCKSDB_LITE rocksdb-4.5.1/hdfs/000077500000000000000000000000001267530634000141025ustar00rootroot00000000000000rocksdb-4.5.1/hdfs/README000066400000000000000000000016621267530634000147670ustar00rootroot00000000000000This directory contains the hdfs extensions needed to make rocksdb store files in HDFS. It has been compiled and testing against CDH 4.4 (2.0.0+1475-1.cdh4.4.0.p0.23~precise-cdh4.4.0). The configuration assumes that packages libhdfs0, libhdfs0-dev are installed which basically means that hdfs.h is in /usr/include and libhdfs in /usr/lib The env_hdfs.h file defines the rocksdb objects that are needed to talk to an underlying filesystem. If you want to compile rocksdb with hdfs support, please set the following environment variables appropriately (also defined in setup.sh for convenience) USE_HDFS=1 JAVA_HOME=/usr/local/jdk-6u22-64 LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/jdk-6u22-64/jre/lib/amd64/server:/usr/local/jdk-6u22-64/jre/lib/amd64/:./snappy/libs make clean all db_bench To run dbbench, set CLASSPATH to include your hadoop distribution db_bench --hdfs="hdfs://hbaseudbperf001.snc1.facebook.com:9000" rocksdb-4.5.1/hdfs/env_hdfs.h000066400000000000000000000260231267530634000160520ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #pragma once #include #include #include #include #include "port/sys_time.h" #include "rocksdb/env.h" #include "rocksdb/status.h" #ifdef USE_HDFS #include namespace rocksdb { // Thrown during execution when there is an issue with the supplied // arguments. class HdfsUsageException : public std::exception { }; // A simple exception that indicates something went wrong that is not // recoverable. The intention is for the message to be printed (with // nothing else) and the process terminate. class HdfsFatalException : public std::exception { public: explicit HdfsFatalException(const std::string& s) : what_(s) { } virtual ~HdfsFatalException() throw() { } virtual const char* what() const throw() { return what_.c_str(); } private: const std::string what_; }; // // The HDFS environment for rocksdb. This class overrides all the // file/dir access methods and delegates the thread-mgmt methods to the // default posix environment. // class HdfsEnv : public Env { public: explicit HdfsEnv(const std::string& fsname) : fsname_(fsname) { posixEnv = Env::Default(); fileSys_ = connectToPath(fsname_); } virtual ~HdfsEnv() { fprintf(stderr, "Destroying HdfsEnv::Default()\n"); hdfsDisconnect(fileSys_); } virtual Status NewSequentialFile(const std::string& fname, std::unique_ptr* result, const EnvOptions& options); virtual Status NewRandomAccessFile(const std::string& fname, std::unique_ptr* result, const EnvOptions& options); virtual Status NewWritableFile(const std::string& fname, std::unique_ptr* result, const EnvOptions& options); virtual Status NewDirectory(const std::string& name, std::unique_ptr* result); virtual Status FileExists(const std::string& fname); virtual Status GetChildren(const std::string& path, std::vector* result); virtual Status DeleteFile(const std::string& fname); virtual Status CreateDir(const std::string& name); virtual Status CreateDirIfMissing(const std::string& name); virtual Status DeleteDir(const std::string& name); virtual Status GetFileSize(const std::string& fname, uint64_t* size); virtual Status GetFileModificationTime(const std::string& fname, uint64_t* file_mtime); virtual Status RenameFile(const std::string& src, const std::string& target); virtual Status LinkFile(const std::string& src, const std::string& target) { return Status::NotSupported(); // not supported } virtual Status LockFile(const std::string& fname, FileLock** lock); virtual Status UnlockFile(FileLock* lock); virtual Status NewLogger(const std::string& fname, std::shared_ptr* result); virtual void Schedule(void (*function)(void* arg), void* arg, Priority pri = LOW, void* tag = nullptr, void (*unschedFunction)(void* arg) = 0) { posixEnv->Schedule(function, arg, pri, tag, unschedFunction); } virtual int UnSchedule(void* tag, Priority pri) { posixEnv->UnSchedule(tag, pri); } virtual void StartThread(void (*function)(void* arg), void* arg) { posixEnv->StartThread(function, arg); } virtual void WaitForJoin() { posixEnv->WaitForJoin(); } virtual unsigned int GetThreadPoolQueueLen(Priority pri = LOW) const override { return posixEnv->GetThreadPoolQueueLen(pri); } virtual Status GetTestDirectory(std::string* path) { return posixEnv->GetTestDirectory(path); } virtual uint64_t NowMicros() { return posixEnv->NowMicros(); } virtual void SleepForMicroseconds(int micros) { posixEnv->SleepForMicroseconds(micros); } virtual Status GetHostName(char* name, uint64_t len) { return posixEnv->GetHostName(name, len); } virtual Status GetCurrentTime(int64_t* unix_time) { return posixEnv->GetCurrentTime(unix_time); } virtual Status GetAbsolutePath(const std::string& db_path, std::string* output_path) { return posixEnv->GetAbsolutePath(db_path, output_path); } virtual void SetBackgroundThreads(int number, Priority pri = LOW) { posixEnv->SetBackgroundThreads(number, pri); } virtual void IncBackgroundThreadsIfNeeded(int number, Priority pri) override { posixEnv->IncBackgroundThreadsIfNeeded(number, pri); } virtual std::string TimeToString(uint64_t number) { return posixEnv->TimeToString(number); } static uint64_t gettid() { assert(sizeof(pthread_t) <= sizeof(uint64_t)); return (uint64_t)pthread_self(); } virtual uint64_t GetThreadID() const override { return HdfsEnv::gettid(); } private: std::string fsname_; // string of the form "hdfs://hostname:port/" hdfsFS fileSys_; // a single FileSystem object for all files Env* posixEnv; // This object is derived from Env, but not from // posixEnv. We have posixnv as an encapsulated // object here so that we can use posix timers, // posix threads, etc. static const std::string kProto; static const std::string pathsep; /** * If the URI is specified of the form hdfs://server:port/path, * then connect to the specified cluster * else connect to default. */ hdfsFS connectToPath(const std::string& uri) { if (uri.empty()) { return nullptr; } if (uri.find(kProto) != 0) { // uri doesn't start with hdfs:// -> use default:0, which is special // to libhdfs. return hdfsConnectNewInstance("default", 0); } const std::string hostport = uri.substr(kProto.length()); std::vector parts; split(hostport, ':', parts); if (parts.size() != 2) { throw HdfsFatalException("Bad uri for hdfs " + uri); } // parts[0] = hosts, parts[1] = port/xxx/yyy std::string host(parts[0]); std::string remaining(parts[1]); int rem = remaining.find(pathsep); std::string portStr = (rem == 0 ? remaining : remaining.substr(0, rem)); tPort port; port = atoi(portStr.c_str()); if (port == 0) { throw HdfsFatalException("Bad host-port for hdfs " + uri); } hdfsFS fs = hdfsConnectNewInstance(host.c_str(), port); return fs; } void split(const std::string &s, char delim, std::vector &elems) { elems.clear(); size_t prev = 0; size_t pos = s.find(delim); while (pos != std::string::npos) { elems.push_back(s.substr(prev, pos)); prev = pos + 1; pos = s.find(delim, prev); } elems.push_back(s.substr(prev, s.size())); } }; } // namespace rocksdb #else // USE_HDFS namespace rocksdb { static const Status notsup; class HdfsEnv : public Env { public: explicit HdfsEnv(const std::string& fsname) { fprintf(stderr, "You have not build rocksdb with HDFS support\n"); fprintf(stderr, "Please see hdfs/README for details\n"); abort(); } virtual ~HdfsEnv() { } virtual Status NewSequentialFile(const std::string& fname, unique_ptr* result, const EnvOptions& options) override; virtual Status NewRandomAccessFile(const std::string& fname, unique_ptr* result, const EnvOptions& options) override { return notsup; } virtual Status NewWritableFile(const std::string& fname, unique_ptr* result, const EnvOptions& options) override { return notsup; } virtual Status NewDirectory(const std::string& name, unique_ptr* result) override { return notsup; } virtual Status FileExists(const std::string& fname) override { return notsup; } virtual Status GetChildren(const std::string& path, std::vector* result) override { return notsup; } virtual Status DeleteFile(const std::string& fname) override { return notsup; } virtual Status CreateDir(const std::string& name) override { return notsup; } virtual Status CreateDirIfMissing(const std::string& name) override { return notsup; } virtual Status DeleteDir(const std::string& name) override { return notsup; } virtual Status GetFileSize(const std::string& fname, uint64_t* size) override { return notsup; } virtual Status GetFileModificationTime(const std::string& fname, uint64_t* time) override { return notsup; } virtual Status RenameFile(const std::string& src, const std::string& target) override { return notsup; } virtual Status LinkFile(const std::string& src, const std::string& target) override { return notsup; } virtual Status LockFile(const std::string& fname, FileLock** lock) override { return notsup; } virtual Status UnlockFile(FileLock* lock) override { return notsup; } virtual Status NewLogger(const std::string& fname, shared_ptr* result) override { return notsup; } virtual void Schedule(void (*function)(void* arg), void* arg, Priority pri = LOW, void* tag = nullptr, void (*unschedFunction)(void* arg) = 0) override {} virtual int UnSchedule(void* tag, Priority pri) override { return 0; } virtual void StartThread(void (*function)(void* arg), void* arg) override {} virtual void WaitForJoin() override {} virtual unsigned int GetThreadPoolQueueLen( Priority pri = LOW) const override { return 0; } virtual Status GetTestDirectory(std::string* path) override { return notsup; } virtual uint64_t NowMicros() override { return 0; } virtual void SleepForMicroseconds(int micros) override {} virtual Status GetHostName(char* name, uint64_t len) override { return notsup; } virtual Status GetCurrentTime(int64_t* unix_time) override { return notsup; } virtual Status GetAbsolutePath(const std::string& db_path, std::string* outputpath) override { return notsup; } virtual void SetBackgroundThreads(int number, Priority pri = LOW) override {} virtual void IncBackgroundThreadsIfNeeded(int number, Priority pri) override { } virtual std::string TimeToString(uint64_t number) override { return ""; } virtual uint64_t GetThreadID() const override { return 0; } }; } #endif // USE_HDFS rocksdb-4.5.1/hdfs/setup.sh000066400000000000000000000006301267530634000155750ustar00rootroot00000000000000export USE_HDFS=1 export LD_LIBRARY_PATH=$JAVA_HOME/jre/lib/amd64/server:$JAVA_HOME/jre/lib/amd64:/usr/lib/hadoop/lib/native export CLASSPATH= for f in `find /usr/lib/hadoop-hdfs | grep jar`; do export CLASSPATH=$CLASSPATH:$f; done for f in `find /usr/lib/hadoop | grep jar`; do export CLASSPATH=$CLASSPATH:$f; done for f in `find /usr/lib/hadoop/client | grep jar`; do export CLASSPATH=$CLASSPATH:$f; done rocksdb-4.5.1/include/000077500000000000000000000000001267530634000146015ustar00rootroot00000000000000rocksdb-4.5.1/include/rocksdb/000077500000000000000000000000001267530634000162305ustar00rootroot00000000000000rocksdb-4.5.1/include/rocksdb/c.h000066400000000000000000001257101267530634000166310ustar00rootroot00000000000000/* Copyright (c) 2013, Facebook, Inc. All rights reserved. This source code is licensed under the BSD-style license found in the LICENSE file in the root directory of this source tree. An additional grant of patent rights can be found in the PATENTS file in the same directory. Copyright (c) 2011 The LevelDB Authors. All rights reserved. Use of this source code is governed by a BSD-style license that can be found in the LICENSE file. See the AUTHORS file for names of contributors. C bindings for rocksdb. May be useful as a stable ABI that can be used by programs that keep rocksdb in a shared library, or for a JNI api. Does not support: . getters for the option types . custom comparators that implement key shortening . capturing post-write-snapshot . custom iter, db, env, cache implementations using just the C bindings Some conventions: (1) We expose just opaque struct pointers and functions to clients. This allows us to change internal representations without having to recompile clients. (2) For simplicity, there is no equivalent to the Slice type. Instead, the caller has to pass the pointer and length as separate arguments. (3) Errors are represented by a null-terminated c string. NULL means no error. All operations that can raise an error are passed a "char** errptr" as the last argument. One of the following must be true on entry: *errptr == NULL *errptr points to a malloc()ed null-terminated error message On success, a leveldb routine leaves *errptr unchanged. On failure, leveldb frees the old value of *errptr and set *errptr to a malloc()ed error message. (4) Bools have the type unsigned char (0 == false; rest == true) (5) All of the pointer arguments must be non-NULL. */ #ifndef STORAGE_ROCKSDB_INCLUDE_C_H_ #define STORAGE_ROCKSDB_INCLUDE_C_H_ #pragma once #ifdef _WIN32 #ifdef ROCKSDB_DLL #ifdef ROCKSDB_LIBRARY_EXPORTS #define ROCKSDB_LIBRARY_API __declspec(dllexport) #else #define ROCKSDB_LIBRARY_API __declspec(dllimport) #endif #else #define ROCKSDB_LIBRARY_API #endif #else #define ROCKSDB_LIBRARY_API #endif #ifdef __cplusplus extern "C" { #endif #include #include #include /* Exported types */ typedef struct rocksdb_t rocksdb_t; typedef struct rocksdb_backup_engine_t rocksdb_backup_engine_t; typedef struct rocksdb_backup_engine_info_t rocksdb_backup_engine_info_t; typedef struct rocksdb_restore_options_t rocksdb_restore_options_t; typedef struct rocksdb_cache_t rocksdb_cache_t; typedef struct rocksdb_compactionfilter_t rocksdb_compactionfilter_t; typedef struct rocksdb_compactionfiltercontext_t rocksdb_compactionfiltercontext_t; typedef struct rocksdb_compactionfilterfactory_t rocksdb_compactionfilterfactory_t; typedef struct rocksdb_comparator_t rocksdb_comparator_t; typedef struct rocksdb_env_t rocksdb_env_t; typedef struct rocksdb_fifo_compaction_options_t rocksdb_fifo_compaction_options_t; typedef struct rocksdb_filelock_t rocksdb_filelock_t; typedef struct rocksdb_filterpolicy_t rocksdb_filterpolicy_t; typedef struct rocksdb_flushoptions_t rocksdb_flushoptions_t; typedef struct rocksdb_iterator_t rocksdb_iterator_t; typedef struct rocksdb_logger_t rocksdb_logger_t; typedef struct rocksdb_mergeoperator_t rocksdb_mergeoperator_t; typedef struct rocksdb_options_t rocksdb_options_t; typedef struct rocksdb_block_based_table_options_t rocksdb_block_based_table_options_t; typedef struct rocksdb_cuckoo_table_options_t rocksdb_cuckoo_table_options_t; typedef struct rocksdb_randomfile_t rocksdb_randomfile_t; typedef struct rocksdb_readoptions_t rocksdb_readoptions_t; typedef struct rocksdb_seqfile_t rocksdb_seqfile_t; typedef struct rocksdb_slicetransform_t rocksdb_slicetransform_t; typedef struct rocksdb_snapshot_t rocksdb_snapshot_t; typedef struct rocksdb_writablefile_t rocksdb_writablefile_t; typedef struct rocksdb_writebatch_t rocksdb_writebatch_t; typedef struct rocksdb_writeoptions_t rocksdb_writeoptions_t; typedef struct rocksdb_universal_compaction_options_t rocksdb_universal_compaction_options_t; typedef struct rocksdb_livefiles_t rocksdb_livefiles_t; typedef struct rocksdb_column_family_handle_t rocksdb_column_family_handle_t; /* DB operations */ extern ROCKSDB_LIBRARY_API rocksdb_t* rocksdb_open( const rocksdb_options_t* options, const char* name, char** errptr); extern ROCKSDB_LIBRARY_API rocksdb_t* rocksdb_open_for_read_only( const rocksdb_options_t* options, const char* name, unsigned char error_if_log_file_exist, char** errptr); extern ROCKSDB_LIBRARY_API rocksdb_backup_engine_t* rocksdb_backup_engine_open( const rocksdb_options_t* options, const char* path, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_backup_engine_create_new_backup( rocksdb_backup_engine_t* be, rocksdb_t* db, char** errptr); extern ROCKSDB_LIBRARY_API rocksdb_restore_options_t* rocksdb_restore_options_create(); extern ROCKSDB_LIBRARY_API void rocksdb_restore_options_destroy( rocksdb_restore_options_t* opt); extern ROCKSDB_LIBRARY_API void rocksdb_restore_options_set_keep_log_files( rocksdb_restore_options_t* opt, int v); extern ROCKSDB_LIBRARY_API void rocksdb_backup_engine_restore_db_from_latest_backup( rocksdb_backup_engine_t* be, const char* db_dir, const char* wal_dir, const rocksdb_restore_options_t* restore_options, char** errptr); extern ROCKSDB_LIBRARY_API const rocksdb_backup_engine_info_t* rocksdb_backup_engine_get_backup_info(rocksdb_backup_engine_t* be); extern ROCKSDB_LIBRARY_API int rocksdb_backup_engine_info_count( const rocksdb_backup_engine_info_t* info); extern ROCKSDB_LIBRARY_API int64_t rocksdb_backup_engine_info_timestamp(const rocksdb_backup_engine_info_t* info, int index); extern ROCKSDB_LIBRARY_API uint32_t rocksdb_backup_engine_info_backup_id(const rocksdb_backup_engine_info_t* info, int index); extern ROCKSDB_LIBRARY_API uint64_t rocksdb_backup_engine_info_size(const rocksdb_backup_engine_info_t* info, int index); extern ROCKSDB_LIBRARY_API uint32_t rocksdb_backup_engine_info_number_files( const rocksdb_backup_engine_info_t* info, int index); extern ROCKSDB_LIBRARY_API void rocksdb_backup_engine_info_destroy( const rocksdb_backup_engine_info_t* info); extern ROCKSDB_LIBRARY_API void rocksdb_backup_engine_close( rocksdb_backup_engine_t* be); extern ROCKSDB_LIBRARY_API rocksdb_t* rocksdb_open_column_families( const rocksdb_options_t* options, const char* name, int num_column_families, const char** column_family_names, const rocksdb_options_t** column_family_options, rocksdb_column_family_handle_t** column_family_handles, char** errptr); extern ROCKSDB_LIBRARY_API rocksdb_t* rocksdb_open_for_read_only_column_families( const rocksdb_options_t* options, const char* name, int num_column_families, const char** column_family_names, const rocksdb_options_t** column_family_options, rocksdb_column_family_handle_t** column_family_handles, unsigned char error_if_log_file_exist, char** errptr); extern ROCKSDB_LIBRARY_API char** rocksdb_list_column_families( const rocksdb_options_t* options, const char* name, size_t* lencf, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_list_column_families_destroy( char** list, size_t len); extern ROCKSDB_LIBRARY_API rocksdb_column_family_handle_t* rocksdb_create_column_family(rocksdb_t* db, const rocksdb_options_t* column_family_options, const char* column_family_name, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_drop_column_family( rocksdb_t* db, rocksdb_column_family_handle_t* handle, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_column_family_handle_destroy( rocksdb_column_family_handle_t*); extern ROCKSDB_LIBRARY_API void rocksdb_close(rocksdb_t* db); extern ROCKSDB_LIBRARY_API void rocksdb_put( rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_put_cf( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_delete( rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key, size_t keylen, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_delete_cf( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_merge( rocksdb_t* db, const rocksdb_writeoptions_t* options, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_merge_cf( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, const char* val, size_t vallen, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_write( rocksdb_t* db, const rocksdb_writeoptions_t* options, rocksdb_writebatch_t* batch, char** errptr); /* Returns NULL if not found. A malloc()ed array otherwise. Stores the length of the array in *vallen. */ extern ROCKSDB_LIBRARY_API char* rocksdb_get( rocksdb_t* db, const rocksdb_readoptions_t* options, const char* key, size_t keylen, size_t* vallen, char** errptr); extern ROCKSDB_LIBRARY_API char* rocksdb_get_cf( rocksdb_t* db, const rocksdb_readoptions_t* options, rocksdb_column_family_handle_t* column_family, const char* key, size_t keylen, size_t* vallen, char** errptr); // if values_list[i] == NULL and errs[i] == NULL, // then we got status.IsNotFound(), which we will not return. // all errors except status status.ok() and status.IsNotFound() are returned. // // errs, values_list and values_list_sizes must be num_keys in length, // allocated by the caller. // errs is a list of strings as opposed to the conventional one error, // where errs[i] is the status for retrieval of keys_list[i]. // each non-NULL errs entry is a malloc()ed, null terminated string. // each non-NULL values_list entry is a malloc()ed array, with // the length for each stored in values_list_sizes[i]. extern ROCKSDB_LIBRARY_API void rocksdb_multi_get( rocksdb_t* db, const rocksdb_readoptions_t* options, size_t num_keys, const char* const* keys_list, const size_t* keys_list_sizes, char** values_list, size_t* values_list_sizes, char** errs); extern ROCKSDB_LIBRARY_API void rocksdb_multi_get_cf( rocksdb_t* db, const rocksdb_readoptions_t* options, const rocksdb_column_family_handle_t* const* column_families, size_t num_keys, const char* const* keys_list, const size_t* keys_list_sizes, char** values_list, size_t* values_list_sizes, char** errs); extern ROCKSDB_LIBRARY_API rocksdb_iterator_t* rocksdb_create_iterator( rocksdb_t* db, const rocksdb_readoptions_t* options); extern ROCKSDB_LIBRARY_API rocksdb_iterator_t* rocksdb_create_iterator_cf( rocksdb_t* db, const rocksdb_readoptions_t* options, rocksdb_column_family_handle_t* column_family); extern ROCKSDB_LIBRARY_API const rocksdb_snapshot_t* rocksdb_create_snapshot( rocksdb_t* db); extern ROCKSDB_LIBRARY_API void rocksdb_release_snapshot( rocksdb_t* db, const rocksdb_snapshot_t* snapshot); /* Returns NULL if property name is unknown. Else returns a pointer to a malloc()-ed null-terminated value. */ extern ROCKSDB_LIBRARY_API char* rocksdb_property_value(rocksdb_t* db, const char* propname); extern ROCKSDB_LIBRARY_API char* rocksdb_property_value_cf( rocksdb_t* db, rocksdb_column_family_handle_t* column_family, const char* propname); extern ROCKSDB_LIBRARY_API void rocksdb_approximate_sizes( rocksdb_t* db, int num_ranges, const char* const* range_start_key, const size_t* range_start_key_len, const char* const* range_limit_key, const size_t* range_limit_key_len, uint64_t* sizes); extern ROCKSDB_LIBRARY_API void rocksdb_approximate_sizes_cf( rocksdb_t* db, rocksdb_column_family_handle_t* column_family, int num_ranges, const char* const* range_start_key, const size_t* range_start_key_len, const char* const* range_limit_key, const size_t* range_limit_key_len, uint64_t* sizes); extern ROCKSDB_LIBRARY_API void rocksdb_compact_range(rocksdb_t* db, const char* start_key, size_t start_key_len, const char* limit_key, size_t limit_key_len); extern ROCKSDB_LIBRARY_API void rocksdb_compact_range_cf( rocksdb_t* db, rocksdb_column_family_handle_t* column_family, const char* start_key, size_t start_key_len, const char* limit_key, size_t limit_key_len); extern ROCKSDB_LIBRARY_API void rocksdb_delete_file(rocksdb_t* db, const char* name); extern ROCKSDB_LIBRARY_API const rocksdb_livefiles_t* rocksdb_livefiles( rocksdb_t* db); extern ROCKSDB_LIBRARY_API void rocksdb_flush( rocksdb_t* db, const rocksdb_flushoptions_t* options, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_disable_file_deletions(rocksdb_t* db, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_enable_file_deletions( rocksdb_t* db, unsigned char force, char** errptr); /* Management operations */ extern ROCKSDB_LIBRARY_API void rocksdb_destroy_db( const rocksdb_options_t* options, const char* name, char** errptr); extern ROCKSDB_LIBRARY_API void rocksdb_repair_db( const rocksdb_options_t* options, const char* name, char** errptr); /* Iterator */ extern ROCKSDB_LIBRARY_API void rocksdb_iter_destroy(rocksdb_iterator_t*); extern ROCKSDB_LIBRARY_API unsigned char rocksdb_iter_valid( const rocksdb_iterator_t*); extern ROCKSDB_LIBRARY_API void rocksdb_iter_seek_to_first(rocksdb_iterator_t*); extern ROCKSDB_LIBRARY_API void rocksdb_iter_seek_to_last(rocksdb_iterator_t*); extern ROCKSDB_LIBRARY_API void rocksdb_iter_seek(rocksdb_iterator_t*, const char* k, size_t klen); extern ROCKSDB_LIBRARY_API void rocksdb_iter_next(rocksdb_iterator_t*); extern ROCKSDB_LIBRARY_API void rocksdb_iter_prev(rocksdb_iterator_t*); extern ROCKSDB_LIBRARY_API const char* rocksdb_iter_key( const rocksdb_iterator_t*, size_t* klen); extern ROCKSDB_LIBRARY_API const char* rocksdb_iter_value( const rocksdb_iterator_t*, size_t* vlen); extern ROCKSDB_LIBRARY_API void rocksdb_iter_get_error( const rocksdb_iterator_t*, char** errptr); /* Write batch */ extern ROCKSDB_LIBRARY_API rocksdb_writebatch_t* rocksdb_writebatch_create(); extern ROCKSDB_LIBRARY_API rocksdb_writebatch_t* rocksdb_writebatch_create_from( const char* rep, size_t size); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_destroy( rocksdb_writebatch_t*); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_clear(rocksdb_writebatch_t*); extern ROCKSDB_LIBRARY_API int rocksdb_writebatch_count(rocksdb_writebatch_t*); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put(rocksdb_writebatch_t*, const char* key, size_t klen, const char* val, size_t vlen); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put_cf( rocksdb_writebatch_t*, rocksdb_column_family_handle_t* column_family, const char* key, size_t klen, const char* val, size_t vlen); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_putv( rocksdb_writebatch_t* b, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_putv_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_merge(rocksdb_writebatch_t*, const char* key, size_t klen, const char* val, size_t vlen); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_merge_cf( rocksdb_writebatch_t*, rocksdb_column_family_handle_t* column_family, const char* key, size_t klen, const char* val, size_t vlen); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_mergev( rocksdb_writebatch_t* b, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_mergev_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes, int num_values, const char* const* values_list, const size_t* values_list_sizes); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_delete(rocksdb_writebatch_t*, const char* key, size_t klen); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_delete_cf( rocksdb_writebatch_t*, rocksdb_column_family_handle_t* column_family, const char* key, size_t klen); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_deletev( rocksdb_writebatch_t* b, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_deletev_cf( rocksdb_writebatch_t* b, rocksdb_column_family_handle_t* column_family, int num_keys, const char* const* keys_list, const size_t* keys_list_sizes); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_put_log_data( rocksdb_writebatch_t*, const char* blob, size_t len); extern ROCKSDB_LIBRARY_API void rocksdb_writebatch_iterate( rocksdb_writebatch_t*, void* state, void (*put)(void*, const char* k, size_t klen, const char* v, size_t vlen), void (*deleted)(void*, const char* k, size_t klen)); extern ROCKSDB_LIBRARY_API const char* rocksdb_writebatch_data( rocksdb_writebatch_t*, size_t* size); /* Block based table options */ extern ROCKSDB_LIBRARY_API rocksdb_block_based_table_options_t* rocksdb_block_based_options_create(); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_destroy( rocksdb_block_based_table_options_t* options); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_block_size( rocksdb_block_based_table_options_t* options, size_t block_size); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_block_size_deviation( rocksdb_block_based_table_options_t* options, int block_size_deviation); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_block_restart_interval( rocksdb_block_based_table_options_t* options, int block_restart_interval); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_filter_policy( rocksdb_block_based_table_options_t* options, rocksdb_filterpolicy_t* filter_policy); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_no_block_cache( rocksdb_block_based_table_options_t* options, unsigned char no_block_cache); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_block_cache( rocksdb_block_based_table_options_t* options, rocksdb_cache_t* block_cache); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_block_cache_compressed( rocksdb_block_based_table_options_t* options, rocksdb_cache_t* block_cache_compressed); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_whole_key_filtering( rocksdb_block_based_table_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_format_version( rocksdb_block_based_table_options_t*, int); enum { rocksdb_block_based_table_index_type_binary_search = 0, rocksdb_block_based_table_index_type_hash_search = 1, }; extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_index_type( rocksdb_block_based_table_options_t*, int); // uses one of the above enums extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_hash_index_allow_collision( rocksdb_block_based_table_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_cache_index_and_filter_blocks( rocksdb_block_based_table_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_block_based_options_set_skip_table_builder_flush( rocksdb_block_based_table_options_t* options, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_block_based_table_factory( rocksdb_options_t* opt, rocksdb_block_based_table_options_t* table_options); /* Cuckoo table options */ extern ROCKSDB_LIBRARY_API rocksdb_cuckoo_table_options_t* rocksdb_cuckoo_options_create(); extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_destroy( rocksdb_cuckoo_table_options_t* options); extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_set_hash_ratio( rocksdb_cuckoo_table_options_t* options, double v); extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_set_max_search_depth( rocksdb_cuckoo_table_options_t* options, uint32_t v); extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_set_cuckoo_block_size( rocksdb_cuckoo_table_options_t* options, uint32_t v); extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_set_identity_as_first_hash( rocksdb_cuckoo_table_options_t* options, unsigned char v); extern ROCKSDB_LIBRARY_API void rocksdb_cuckoo_options_set_use_module_hash( rocksdb_cuckoo_table_options_t* options, unsigned char v); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_cuckoo_table_factory( rocksdb_options_t* opt, rocksdb_cuckoo_table_options_t* table_options); /* Options */ extern ROCKSDB_LIBRARY_API rocksdb_options_t* rocksdb_options_create(); extern ROCKSDB_LIBRARY_API void rocksdb_options_destroy(rocksdb_options_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_increase_parallelism( rocksdb_options_t* opt, int total_threads); extern ROCKSDB_LIBRARY_API void rocksdb_options_optimize_for_point_lookup( rocksdb_options_t* opt, uint64_t block_cache_size_mb); extern ROCKSDB_LIBRARY_API void rocksdb_options_optimize_level_style_compaction( rocksdb_options_t* opt, uint64_t memtable_memory_budget); extern ROCKSDB_LIBRARY_API void rocksdb_options_optimize_universal_style_compaction( rocksdb_options_t* opt, uint64_t memtable_memory_budget); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_compaction_filter( rocksdb_options_t*, rocksdb_compactionfilter_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_compaction_filter_factory( rocksdb_options_t*, rocksdb_compactionfilterfactory_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_comparator( rocksdb_options_t*, rocksdb_comparator_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_merge_operator( rocksdb_options_t*, rocksdb_mergeoperator_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_uint64add_merge_operator( rocksdb_options_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_compression_per_level( rocksdb_options_t* opt, int* level_values, size_t num_levels); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_create_if_missing( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_create_missing_column_families(rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_error_if_exists( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_paranoid_checks( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_env(rocksdb_options_t*, rocksdb_env_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_info_log(rocksdb_options_t*, rocksdb_logger_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_info_log_level( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_write_buffer_size( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_open_files( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_total_wal_size( rocksdb_options_t* opt, uint64_t n); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_compression_options( rocksdb_options_t*, int, int, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_prefix_extractor( rocksdb_options_t*, rocksdb_slicetransform_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_num_levels( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_level0_file_num_compaction_trigger(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_level0_slowdown_writes_trigger(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_level0_stop_writes_trigger( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_mem_compaction_level( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_target_file_size_base( rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_target_file_size_multiplier( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_bytes_for_level_base( rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_bytes_for_level_multiplier(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_expanded_compaction_factor( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_grandparent_overlap_factor(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_bytes_for_level_multiplier_additional( rocksdb_options_t*, int* level_values, size_t num_levels); extern ROCKSDB_LIBRARY_API void rocksdb_options_enable_statistics( rocksdb_options_t*); /* returns a pointer to a malloc()-ed, null terminated string */ extern ROCKSDB_LIBRARY_API char* rocksdb_options_statistics_get_string( rocksdb_options_t* opt); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_write_buffer_number( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_min_write_buffer_number_to_merge(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_write_buffer_number_to_maintain(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_background_compactions( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_background_flushes( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_log_file_size( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_log_file_time_to_roll( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_keep_log_file_num( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_recycle_log_file_num( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_soft_rate_limit( rocksdb_options_t*, double); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_hard_rate_limit( rocksdb_options_t*, double); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_rate_limit_delay_max_milliseconds(rocksdb_options_t*, unsigned int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_manifest_file_size( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_table_cache_numshardbits( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_table_cache_remove_scan_count_limit(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_arena_block_size( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_use_fsync( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_db_log_dir( rocksdb_options_t*, const char*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_wal_dir(rocksdb_options_t*, const char*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_WAL_ttl_seconds( rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_WAL_size_limit_MB( rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_manifest_preallocation_size( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_purge_redundant_kvs_while_flush(rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_allow_os_buffer( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_allow_mmap_reads( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_allow_mmap_writes( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_is_fd_close_on_exec( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_skip_log_error_on_recovery( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_stats_dump_period_sec( rocksdb_options_t*, unsigned int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_advise_random_on_open( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_access_hint_on_compaction_start(rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_use_adaptive_mutex( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_bytes_per_sync( rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_verify_checksums_in_compaction(rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_filter_deletes( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_sequential_skip_in_iterations(rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_disable_data_sync( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_disable_auto_compactions( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_delete_obsolete_files_period_micros(rocksdb_options_t*, uint64_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_source_compaction_factor( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_prepare_for_bulk_load( rocksdb_options_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_memtable_vector_rep( rocksdb_options_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_hash_skip_list_rep( rocksdb_options_t*, size_t, int32_t, int32_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_hash_link_list_rep( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_plain_table_factory( rocksdb_options_t*, uint32_t, int, double, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_min_level_to_compress( rocksdb_options_t* opt, int level); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_memtable_prefix_bloom_bits( rocksdb_options_t*, uint32_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_memtable_prefix_bloom_probes(rocksdb_options_t*, uint32_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_memtable_prefix_bloom_huge_page_tlb_size(rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_max_successive_merges( rocksdb_options_t*, size_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_min_partial_merge_operands( rocksdb_options_t*, uint32_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_bloom_locality( rocksdb_options_t*, uint32_t); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_inplace_update_support( rocksdb_options_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_inplace_update_num_locks( rocksdb_options_t*, size_t); enum { rocksdb_no_compression = 0, rocksdb_snappy_compression = 1, rocksdb_zlib_compression = 2, rocksdb_bz2_compression = 3, rocksdb_lz4_compression = 4, rocksdb_lz4hc_compression = 5 }; extern ROCKSDB_LIBRARY_API void rocksdb_options_set_compression( rocksdb_options_t*, int); enum { rocksdb_level_compaction = 0, rocksdb_universal_compaction = 1, rocksdb_fifo_compaction = 2 }; extern ROCKSDB_LIBRARY_API void rocksdb_options_set_compaction_style( rocksdb_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_universal_compaction_options( rocksdb_options_t*, rocksdb_universal_compaction_options_t*); extern ROCKSDB_LIBRARY_API void rocksdb_options_set_fifo_compaction_options( rocksdb_options_t* opt, rocksdb_fifo_compaction_options_t* fifo); /* Compaction Filter */ extern ROCKSDB_LIBRARY_API rocksdb_compactionfilter_t* rocksdb_compactionfilter_create( void* state, void (*destructor)(void*), unsigned char (*filter)(void*, int level, const char* key, size_t key_length, const char* existing_value, size_t value_length, char** new_value, size_t* new_value_length, unsigned char* value_changed), const char* (*name)(void*)); extern ROCKSDB_LIBRARY_API void rocksdb_compactionfilter_destroy( rocksdb_compactionfilter_t*); /* Compaction Filter Context */ extern ROCKSDB_LIBRARY_API unsigned char rocksdb_compactionfiltercontext_is_full_compaction( rocksdb_compactionfiltercontext_t* context); extern ROCKSDB_LIBRARY_API unsigned char rocksdb_compactionfiltercontext_is_manual_compaction( rocksdb_compactionfiltercontext_t* context); /* Compaction Filter Factory */ extern ROCKSDB_LIBRARY_API rocksdb_compactionfilterfactory_t* rocksdb_compactionfilterfactory_create( void* state, void (*destructor)(void*), rocksdb_compactionfilter_t* (*create_compaction_filter)( void*, rocksdb_compactionfiltercontext_t* context), const char* (*name)(void*)); extern ROCKSDB_LIBRARY_API void rocksdb_compactionfilterfactory_destroy( rocksdb_compactionfilterfactory_t*); /* Comparator */ extern ROCKSDB_LIBRARY_API rocksdb_comparator_t* rocksdb_comparator_create( void* state, void (*destructor)(void*), int (*compare)(void*, const char* a, size_t alen, const char* b, size_t blen), const char* (*name)(void*)); extern ROCKSDB_LIBRARY_API void rocksdb_comparator_destroy( rocksdb_comparator_t*); /* Filter policy */ extern ROCKSDB_LIBRARY_API rocksdb_filterpolicy_t* rocksdb_filterpolicy_create( void* state, void (*destructor)(void*), char* (*create_filter)(void*, const char* const* key_array, const size_t* key_length_array, int num_keys, size_t* filter_length), unsigned char (*key_may_match)(void*, const char* key, size_t length, const char* filter, size_t filter_length), void (*delete_filter)(void*, const char* filter, size_t filter_length), const char* (*name)(void*)); extern ROCKSDB_LIBRARY_API void rocksdb_filterpolicy_destroy( rocksdb_filterpolicy_t*); extern ROCKSDB_LIBRARY_API rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom(int bits_per_key); extern ROCKSDB_LIBRARY_API rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom_full(int bits_per_key); /* Merge Operator */ extern ROCKSDB_LIBRARY_API rocksdb_mergeoperator_t* rocksdb_mergeoperator_create( void* state, void (*destructor)(void*), char* (*full_merge)(void*, const char* key, size_t key_length, const char* existing_value, size_t existing_value_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length), char* (*partial_merge)(void*, const char* key, size_t key_length, const char* const* operands_list, const size_t* operands_list_length, int num_operands, unsigned char* success, size_t* new_value_length), void (*delete_value)(void*, const char* value, size_t value_length), const char* (*name)(void*)); extern ROCKSDB_LIBRARY_API void rocksdb_mergeoperator_destroy( rocksdb_mergeoperator_t*); /* Read options */ extern ROCKSDB_LIBRARY_API rocksdb_readoptions_t* rocksdb_readoptions_create(); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_destroy( rocksdb_readoptions_t*); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_verify_checksums( rocksdb_readoptions_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_fill_cache( rocksdb_readoptions_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_snapshot( rocksdb_readoptions_t*, const rocksdb_snapshot_t*); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_iterate_upper_bound( rocksdb_readoptions_t*, const char* key, size_t keylen); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_read_tier( rocksdb_readoptions_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_readoptions_set_tailing( rocksdb_readoptions_t*, unsigned char); /* Write options */ extern ROCKSDB_LIBRARY_API rocksdb_writeoptions_t* rocksdb_writeoptions_create(); extern ROCKSDB_LIBRARY_API void rocksdb_writeoptions_destroy( rocksdb_writeoptions_t*); extern ROCKSDB_LIBRARY_API void rocksdb_writeoptions_set_sync( rocksdb_writeoptions_t*, unsigned char); extern ROCKSDB_LIBRARY_API void rocksdb_writeoptions_disable_WAL( rocksdb_writeoptions_t* opt, int disable); /* Flush options */ extern ROCKSDB_LIBRARY_API rocksdb_flushoptions_t* rocksdb_flushoptions_create(); extern ROCKSDB_LIBRARY_API void rocksdb_flushoptions_destroy( rocksdb_flushoptions_t*); extern ROCKSDB_LIBRARY_API void rocksdb_flushoptions_set_wait( rocksdb_flushoptions_t*, unsigned char); /* Cache */ extern ROCKSDB_LIBRARY_API rocksdb_cache_t* rocksdb_cache_create_lru( size_t capacity); extern ROCKSDB_LIBRARY_API void rocksdb_cache_destroy(rocksdb_cache_t* cache); /* Env */ extern ROCKSDB_LIBRARY_API rocksdb_env_t* rocksdb_create_default_env(); extern ROCKSDB_LIBRARY_API void rocksdb_env_set_background_threads( rocksdb_env_t* env, int n); extern ROCKSDB_LIBRARY_API void rocksdb_env_set_high_priority_background_threads(rocksdb_env_t* env, int n); extern ROCKSDB_LIBRARY_API void rocksdb_env_join_all_threads( rocksdb_env_t* env); extern ROCKSDB_LIBRARY_API void rocksdb_env_destroy(rocksdb_env_t*); /* SliceTransform */ extern ROCKSDB_LIBRARY_API rocksdb_slicetransform_t* rocksdb_slicetransform_create( void* state, void (*destructor)(void*), char* (*transform)(void*, const char* key, size_t length, size_t* dst_length), unsigned char (*in_domain)(void*, const char* key, size_t length), unsigned char (*in_range)(void*, const char* key, size_t length), const char* (*name)(void*)); extern ROCKSDB_LIBRARY_API rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t); extern ROCKSDB_LIBRARY_API rocksdb_slicetransform_t* rocksdb_slicetransform_create_noop(); extern ROCKSDB_LIBRARY_API void rocksdb_slicetransform_destroy( rocksdb_slicetransform_t*); /* Universal Compaction options */ enum { rocksdb_similar_size_compaction_stop_style = 0, rocksdb_total_size_compaction_stop_style = 1 }; extern ROCKSDB_LIBRARY_API rocksdb_universal_compaction_options_t* rocksdb_universal_compaction_options_create(); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_set_size_ratio( rocksdb_universal_compaction_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_set_min_merge_width( rocksdb_universal_compaction_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_set_max_merge_width( rocksdb_universal_compaction_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_set_max_size_amplification_percent( rocksdb_universal_compaction_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_set_compression_size_percent( rocksdb_universal_compaction_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_set_stop_style( rocksdb_universal_compaction_options_t*, int); extern ROCKSDB_LIBRARY_API void rocksdb_universal_compaction_options_destroy( rocksdb_universal_compaction_options_t*); extern ROCKSDB_LIBRARY_API rocksdb_fifo_compaction_options_t* rocksdb_fifo_compaction_options_create(); extern ROCKSDB_LIBRARY_API void rocksdb_fifo_compaction_options_set_max_table_files_size( rocksdb_fifo_compaction_options_t* fifo_opts, uint64_t size); extern ROCKSDB_LIBRARY_API void rocksdb_fifo_compaction_options_destroy( rocksdb_fifo_compaction_options_t* fifo_opts); extern ROCKSDB_LIBRARY_API int rocksdb_livefiles_count( const rocksdb_livefiles_t*); extern ROCKSDB_LIBRARY_API const char* rocksdb_livefiles_name( const rocksdb_livefiles_t*, int index); extern ROCKSDB_LIBRARY_API int rocksdb_livefiles_level( const rocksdb_livefiles_t*, int index); extern ROCKSDB_LIBRARY_API size_t rocksdb_livefiles_size(const rocksdb_livefiles_t*, int index); extern ROCKSDB_LIBRARY_API const char* rocksdb_livefiles_smallestkey( const rocksdb_livefiles_t*, int index, size_t* size); extern ROCKSDB_LIBRARY_API const char* rocksdb_livefiles_largestkey( const rocksdb_livefiles_t*, int index, size_t* size); extern ROCKSDB_LIBRARY_API void rocksdb_livefiles_destroy( const rocksdb_livefiles_t*); /* Utility Helpers */ extern ROCKSDB_LIBRARY_API void rocksdb_get_options_from_string( const rocksdb_options_t* base_options, const char* opts_str, rocksdb_options_t* new_options, char** errptr); // referring to convention (3), this should be used by client // to free memory that was malloc()ed extern ROCKSDB_LIBRARY_API void rocksdb_free(void* ptr); #ifdef __cplusplus } /* end extern "C" */ #endif #endif /* STORAGE_ROCKSDB_INCLUDE_C_H_ */ rocksdb-4.5.1/include/rocksdb/cache.h000066400000000000000000000127361267530634000174550ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // A Cache is an interface that maps keys to values. It has internal // synchronization and may be safely accessed concurrently from // multiple threads. It may automatically evict entries to make room // for new entries. Values have a specified charge against the cache // capacity. For example, a cache where the values are variable // length strings, may use the length of the string as the charge for // the string. // // A builtin cache implementation with a least-recently-used eviction // policy is provided. Clients may use their own implementations if // they want something more sophisticated (like scan-resistance, a // custom eviction policy, variable cache sizing, etc.) #ifndef STORAGE_ROCKSDB_INCLUDE_CACHE_H_ #define STORAGE_ROCKSDB_INCLUDE_CACHE_H_ #include #include #include "rocksdb/slice.h" namespace rocksdb { using std::shared_ptr; class Cache; // Create a new cache with a fixed size capacity. The cache is sharded // to 2^numShardBits shards, by hash of the key. The total capacity // is divided and evenly assigned to each shard. // // The functions without parameter numShardBits uses default value, which is 4 extern shared_ptr NewLRUCache(size_t capacity); extern shared_ptr NewLRUCache(size_t capacity, int numShardBits); class Cache { public: Cache() { } // Destroys all existing entries by calling the "deleter" // function that was passed via the Insert() function. // // @See Insert virtual ~Cache(); // Opaque handle to an entry stored in the cache. struct Handle { }; // Insert a mapping from key->value into the cache and assign it // the specified charge against the total cache capacity. // // Returns a handle that corresponds to the mapping. The caller // must call this->Release(handle) when the returned mapping is no // longer needed. // // When the inserted entry is no longer needed, the key and // value will be passed to "deleter". virtual Handle* Insert(const Slice& key, void* value, size_t charge, void (*deleter)(const Slice& key, void* value)) = 0; // If the cache has no mapping for "key", returns nullptr. // // Else return a handle that corresponds to the mapping. The caller // must call this->Release(handle) when the returned mapping is no // longer needed. virtual Handle* Lookup(const Slice& key) = 0; // Release a mapping returned by a previous Lookup(). // REQUIRES: handle must not have been released yet. // REQUIRES: handle must have been returned by a method on *this. virtual void Release(Handle* handle) = 0; // Return the value encapsulated in a handle returned by a // successful Lookup(). // REQUIRES: handle must not have been released yet. // REQUIRES: handle must have been returned by a method on *this. virtual void* Value(Handle* handle) = 0; // If the cache contains entry for key, erase it. Note that the // underlying entry will be kept around until all existing handles // to it have been released. virtual void Erase(const Slice& key) = 0; // Return a new numeric id. May be used by multiple clients who are // sharing the same cache to partition the key space. Typically the // client will allocate a new id at startup and prepend the id to // its cache keys. virtual uint64_t NewId() = 0; // sets the maximum configured capacity of the cache. When the new // capacity is less than the old capacity and the existing usage is // greater than new capacity, the implementation will do its best job to // purge the released entries from the cache in order to lower the usage virtual void SetCapacity(size_t capacity) = 0; // returns the maximum configured capacity of the cache virtual size_t GetCapacity() const = 0; // returns the memory size for the entries residing in the cache. virtual size_t GetUsage() const = 0; // returns the memory size for a specific entry in the cache. virtual size_t GetUsage(Handle* handle) const = 0; // returns the memory size for the entries in use by the system virtual size_t GetPinnedUsage() const = 0; // Call this on shutdown if you want to speed it up. Cache will disown // any underlying data and will not free it on delete. This call will leak // memory - call this only if you're shutting down the process. // Any attempts of using cache after this call will fail terribly. // Always delete the DB object before calling this method! virtual void DisownData() { // default implementation is noop }; // Apply callback to all entries in the cache // If thread_safe is true, it will also lock the accesses. Otherwise, it will // access the cache without the lock held virtual void ApplyToAllCacheEntries(void (*callback)(void*, size_t), bool thread_safe) = 0; private: void LRU_Remove(Handle* e); void LRU_Append(Handle* e); void Unref(Handle* e); // No copying allowed Cache(const Cache&); void operator=(const Cache&); }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_UTIL_CACHE_H_ rocksdb-4.5.1/include/rocksdb/compaction_filter.h000066400000000000000000000137371267530634000221150ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2013 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef STORAGE_ROCKSDB_INCLUDE_COMPACTION_FILTER_H_ #define STORAGE_ROCKSDB_INCLUDE_COMPACTION_FILTER_H_ #include #include #include namespace rocksdb { class Slice; class SliceTransform; // Context information of a compaction run struct CompactionFilterContext { // Does this compaction run include all data files bool is_full_compaction; // Is this compaction requested by the client (true), // or is it occurring as an automatic compaction process bool is_manual_compaction; }; // CompactionFilter allows an application to modify/delete a key-value at // the time of compaction. class CompactionFilter { public: // Context information of a compaction run struct Context { // Does this compaction run include all data files bool is_full_compaction; // Is this compaction requested by the client (true), // or is it occurring as an automatic compaction process bool is_manual_compaction; // Which column family this compaction is for. uint32_t column_family_id; }; virtual ~CompactionFilter() {} // The compaction process invokes this // method for kv that is being compacted. A return value // of false indicates that the kv should be preserved in the // output of this compaction run and a return value of true // indicates that this key-value should be removed from the // output of the compaction. The application can inspect // the existing value of the key and make decision based on it. // // Key-Values that are results of merge operation during compaction are not // passed into this function. Currently, when you have a mix of Put()s and // Merge()s on a same key, we only guarantee to process the merge operands // through the compaction filters. Put()s might be processed, or might not. // // When the value is to be preserved, the application has the option // to modify the existing_value and pass it back through new_value. // value_changed needs to be set to true in this case. // // If you use snapshot feature of RocksDB (i.e. call GetSnapshot() API on a // DB* object), CompactionFilter might not be very useful for you. Due to // guarantees we need to maintain, compaction process will not call Filter() // on any keys that were written before the latest snapshot. In other words, // compaction will only call Filter() on keys written after your most recent // call to GetSnapshot(). In most cases, Filter() will not be called very // often. This is something we're fixing. See the discussion at: // https://www.facebook.com/groups/mysqlonrocksdb/permalink/999723240091865/ // // If multithreaded compaction is being used *and* a single CompactionFilter // instance was supplied via Options::compaction_filter, this method may be // called from different threads concurrently. The application must ensure // that the call is thread-safe. // // If the CompactionFilter was created by a factory, then it will only ever // be used by a single thread that is doing the compaction run, and this // call does not need to be thread-safe. However, multiple filters may be // in existence and operating concurrently. // // The last paragraph is not true if you set max_subcompactions to more than // 1. In that case, subcompaction from multiple threads may call a single // CompactionFilter concurrently. virtual bool Filter(int level, const Slice& key, const Slice& existing_value, std::string* new_value, bool* value_changed) const = 0; // The compaction process invokes this method on every merge operand. If this // method returns true, the merge operand will be ignored and not written out // in the compaction output // // Note: If you are using a TransactionDB, it is not recommended to implement // FilterMergeOperand(). If a Merge operation is filtered out, TransactionDB // may not realize there is a write conflict and may allow a Transaction to // Commit that should have failed. Instead, it is better to implement any // Merge filtering inside the MergeOperator. virtual bool FilterMergeOperand(int level, const Slice& key, const Slice& operand) const { return false; } // By default, compaction will only call Filter() on keys written after the // most recent call to GetSnapshot(). However, if the compaction filter // overrides IgnoreSnapshots to make it return false, the compaction filter // will be called even if the keys were written before the last snapshot. // This behavior is to be used only when we want to delete a set of keys // irrespective of snapshots. In particular, care should be taken // to understand that the values of thesekeys will change even if we are // using a snapshot. virtual bool IgnoreSnapshots() const { return false; } // Returns a name that identifies this compaction filter. // The name will be printed to LOG file on start up for diagnosis. virtual const char* Name() const = 0; }; // Each compaction will create a new CompactionFilter allowing the // application to know about different compactions class CompactionFilterFactory { public: virtual ~CompactionFilterFactory() { } virtual std::unique_ptr CreateCompactionFilter( const CompactionFilter::Context& context) = 0; // Returns a name that identifies this compaction filter factory. virtual const char* Name() const = 0; }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_COMPACTION_FILTER_H_ rocksdb-4.5.1/include/rocksdb/compaction_job_stats.h000066400000000000000000000056061267530634000226140ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include namespace rocksdb { struct CompactionJobStats { CompactionJobStats() { Reset(); } void Reset(); // Aggregate the CompactionJobStats from another instance with this one void Add(const CompactionJobStats& stats); // the elapsed time in micro of this compaction. uint64_t elapsed_micros; // the number of compaction input records. uint64_t num_input_records; // the number of compaction input files. size_t num_input_files; // the number of compaction input files at the output level. size_t num_input_files_at_output_level; // the number of compaction output records. uint64_t num_output_records; // the number of compaction output files. size_t num_output_files; // true if the compaction is a manual compaction bool is_manual_compaction; // the size of the compaction input in bytes. uint64_t total_input_bytes; // the size of the compaction output in bytes. uint64_t total_output_bytes; // number of records being replaced by newer record associated with same key. // this could be a new value or a deletion entry for that key so this field // sums up all updated and deleted keys uint64_t num_records_replaced; // the sum of the uncompressed input keys in bytes. uint64_t total_input_raw_key_bytes; // the sum of the uncompressed input values in bytes. uint64_t total_input_raw_value_bytes; // the number of deletion entries before compaction. Deletion entries // can disappear after compaction because they expired uint64_t num_input_deletion_records; // number of deletion records that were found obsolete and discarded // because it is not possible to delete any more keys with this entry // (i.e. all possible deletions resulting from it have been completed) uint64_t num_expired_deletion_records; // number of corrupt keys (ParseInternalKey returned false when applied to // the key) encountered and written out. uint64_t num_corrupt_keys; // Following counters are only populated if // options.compaction_measure_io_stats = true; // Time spent on file's Append() call. uint64_t file_write_nanos; // Time spent on sync file range. uint64_t file_range_sync_nanos; // Time spent on file fsync. uint64_t file_fsync_nanos; // Time spent on preparing file write (falocate, etc) uint64_t file_prepare_write_nanos; // 0-terminated strings storing the first 8 bytes of the smallest and // largest key in the output. static const size_t kMaxPrefixLength = 8; std::string smallest_output_key_prefix; std::string largest_output_key_prefix; }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/comparator.h000066400000000000000000000061131267530634000205510ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef STORAGE_ROCKSDB_INCLUDE_COMPARATOR_H_ #define STORAGE_ROCKSDB_INCLUDE_COMPARATOR_H_ #include namespace rocksdb { class Slice; // A Comparator object provides a total order across slices that are // used as keys in an sstable or a database. A Comparator implementation // must be thread-safe since rocksdb may invoke its methods concurrently // from multiple threads. class Comparator { public: virtual ~Comparator(); // Three-way comparison. Returns value: // < 0 iff "a" < "b", // == 0 iff "a" == "b", // > 0 iff "a" > "b" virtual int Compare(const Slice& a, const Slice& b) const = 0; // Compares two slices for equality. The following invariant should always // hold (and is the default implementation): // Equal(a, b) iff Compare(a, b) == 0 // Overwrite only if equality comparisons can be done more efficiently than // three-way comparisons. virtual bool Equal(const Slice& a, const Slice& b) const { return Compare(a, b) == 0; } // The name of the comparator. Used to check for comparator // mismatches (i.e., a DB created with one comparator is // accessed using a different comparator. // // The client of this package should switch to a new name whenever // the comparator implementation changes in a way that will cause // the relative ordering of any two keys to change. // // Names starting with "rocksdb." are reserved and should not be used // by any clients of this package. virtual const char* Name() const = 0; // Advanced functions: these are used to reduce the space requirements // for internal data structures like index blocks. // If *start < limit, changes *start to a short string in [start,limit). // Simple comparator implementations may return with *start unchanged, // i.e., an implementation of this method that does nothing is correct. virtual void FindShortestSeparator( std::string* start, const Slice& limit) const = 0; // Changes *key to a short string >= *key. // Simple comparator implementations may return with *key unchanged, // i.e., an implementation of this method that does nothing is correct. virtual void FindShortSuccessor(std::string* key) const = 0; }; // Return a builtin comparator that uses lexicographic byte-wise // ordering. The result remains the property of this module and // must not be deleted. extern const Comparator* BytewiseComparator(); // Return a builtin comparator that uses reverse lexicographic byte-wise // ordering. extern const Comparator* ReverseBytewiseComparator(); } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_COMPARATOR_H_ rocksdb-4.5.1/include/rocksdb/convenience.h000066400000000000000000000104511267530634000206760ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "rocksdb/db.h" #include "rocksdb/options.h" #include "rocksdb/table.h" namespace rocksdb { #ifndef ROCKSDB_LITE // Take a map of option name and option value, apply them into the // base_options, and return the new options as a result. // // If input_strings_escaped is set to true, then each escaped characters // prefixed by '\' in the the values of the opts_map will be further // converted back to the raw string before assigning to the associated // options. Status GetColumnFamilyOptionsFromMap( const ColumnFamilyOptions& base_options, const std::unordered_map& opts_map, ColumnFamilyOptions* new_options, bool input_strings_escaped = false); // Take a map of option name and option value, apply them into the // base_options, and return the new options as a result. // // If input_strings_escaped is set to true, then each escaped characters // prefixed by '\' in the the values of the opts_map will be further // converted back to the raw string before assigning to the associated // options. Status GetDBOptionsFromMap( const DBOptions& base_options, const std::unordered_map& opts_map, DBOptions* new_options, bool input_strings_escaped = false); Status GetBlockBasedTableOptionsFromMap( const BlockBasedTableOptions& table_options, const std::unordered_map& opts_map, BlockBasedTableOptions* new_table_options, bool input_strings_escaped = false); Status GetPlainTableOptionsFromMap( const PlainTableOptions& table_options, const std::unordered_map& opts_map, PlainTableOptions* new_table_options, bool input_strings_escaped = false); // Take a string representation of option names and values, apply them into the // base_options, and return the new options as a result. The string has the // following format: // "write_buffer_size=1024;max_write_buffer_number=2" // Nested options config is also possible. For example, you can define // BlockBasedTableOptions as part of the string for block-based table factory: // "write_buffer_size=1024;block_based_table_factory={block_size=4k};" // "max_write_buffer_num=2" Status GetColumnFamilyOptionsFromString( const ColumnFamilyOptions& base_options, const std::string& opts_str, ColumnFamilyOptions* new_options); Status GetDBOptionsFromString( const DBOptions& base_options, const std::string& opts_str, DBOptions* new_options); Status GetStringFromDBOptions(std::string* opts_str, const DBOptions& db_options, const std::string& delimiter = "; "); Status GetStringFromColumnFamilyOptions(std::string* opts_str, const ColumnFamilyOptions& db_options, const std::string& delimiter = "; "); Status GetBlockBasedTableOptionsFromString( const BlockBasedTableOptions& table_options, const std::string& opts_str, BlockBasedTableOptions* new_table_options); Status GetPlainTableOptionsFromString( const PlainTableOptions& table_options, const std::string& opts_str, PlainTableOptions* new_table_options); Status GetMemTableRepFactoryFromString( const std::string& opts_str, std::unique_ptr* new_mem_factory); Status GetOptionsFromString(const Options& base_options, const std::string& opts_str, Options* new_options); // Request stopping background work, if wait is true wait until it's done void CancelAllBackgroundWork(DB* db, bool wait = false); // Delete files which are entirely in the given range // Could leave some keys in the range which are in files which are not // entirely in the range. // Snapshots before the delete might not see the data in the given range. Status DeleteFilesInRange(DB* db, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end); #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/db.h000066400000000000000000001127761267530634000170040ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef STORAGE_ROCKSDB_INCLUDE_DB_H_ #define STORAGE_ROCKSDB_INCLUDE_DB_H_ #include #include #include #include #include #include #include "rocksdb/immutable_options.h" #include "rocksdb/iterator.h" #include "rocksdb/listener.h" #include "rocksdb/metadata.h" #include "rocksdb/options.h" #include "rocksdb/snapshot.h" #include "rocksdb/thread_status.h" #include "rocksdb/transaction_log.h" #include "rocksdb/types.h" #include "rocksdb/version.h" #ifdef _WIN32 // Windows API macro interference #undef DeleteFile #endif namespace rocksdb { struct Options; struct DBOptions; struct ColumnFamilyOptions; struct ReadOptions; struct WriteOptions; struct FlushOptions; struct CompactionOptions; struct CompactRangeOptions; struct TableProperties; struct ExternalSstFileInfo; class WriteBatch; class Env; class EventListener; using std::unique_ptr; extern const std::string kDefaultColumnFamilyName; struct ColumnFamilyDescriptor { std::string name; ColumnFamilyOptions options; ColumnFamilyDescriptor() : name(kDefaultColumnFamilyName), options(ColumnFamilyOptions()) {} ColumnFamilyDescriptor(const std::string& _name, const ColumnFamilyOptions& _options) : name(_name), options(_options) {} }; class ColumnFamilyHandle { public: virtual ~ColumnFamilyHandle() {} // Returns the name of the column family associated with the current handle. virtual const std::string& GetName() const = 0; // Returns the ID of the column family associated with the current handle. virtual uint32_t GetID() const = 0; // Fills "*desc" with the up-to-date descriptor of the column family // associated with this handle. Since it fills "*desc" with the up-to-date // information, this call might internally lock and release DB mutex to // access the up-to-date CF options. In addition, all the pointer-typed // options cannot be referenced any longer than the original options exist. // // Note that this function is not supported in RocksDBLite. virtual Status GetDescriptor(ColumnFamilyDescriptor* desc) = 0; }; static const int kMajorVersion = __ROCKSDB_MAJOR__; static const int kMinorVersion = __ROCKSDB_MINOR__; // A range of keys struct Range { Slice start; // Included in the range Slice limit; // Not included in the range Range() { } Range(const Slice& s, const Slice& l) : start(s), limit(l) { } }; // A collections of table properties objects, where // key: is the table's file name. // value: the table properties object of the given table. typedef std::unordered_map> TablePropertiesCollection; // A DB is a persistent ordered map from keys to values. // A DB is safe for concurrent access from multiple threads without // any external synchronization. class DB { public: // Open the database with the specified "name". // Stores a pointer to a heap-allocated database in *dbptr and returns // OK on success. // Stores nullptr in *dbptr and returns a non-OK status on error. // Caller should delete *dbptr when it is no longer needed. static Status Open(const Options& options, const std::string& name, DB** dbptr); // Open the database for read only. All DB interfaces // that modify data, like put/delete, will return error. // If the db is opened in read only mode, then no compactions // will happen. // // Not supported in ROCKSDB_LITE, in which case the function will // return Status::NotSupported. static Status OpenForReadOnly(const Options& options, const std::string& name, DB** dbptr, bool error_if_log_file_exist = false); // Open the database for read only with column families. When opening DB with // read only, you can specify only a subset of column families in the // database that should be opened. However, you always need to specify default // column family. The default column family name is 'default' and it's stored // in rocksdb::kDefaultColumnFamilyName // // Not supported in ROCKSDB_LITE, in which case the function will // return Status::NotSupported. static Status OpenForReadOnly( const DBOptions& db_options, const std::string& name, const std::vector& column_families, std::vector* handles, DB** dbptr, bool error_if_log_file_exist = false); // Open DB with column families. // db_options specify database specific options // column_families is the vector of all column families in the database, // containing column family name and options. You need to open ALL column // families in the database. To get the list of column families, you can use // ListColumnFamilies(). Also, you can open only a subset of column families // for read-only access. // The default column family name is 'default' and it's stored // in rocksdb::kDefaultColumnFamilyName. // If everything is OK, handles will on return be the same size // as column_families --- handles[i] will be a handle that you // will use to operate on column family column_family[i] static Status Open(const DBOptions& db_options, const std::string& name, const std::vector& column_families, std::vector* handles, DB** dbptr); // ListColumnFamilies will open the DB specified by argument name // and return the list of all column families in that DB // through column_families argument. The ordering of // column families in column_families is unspecified. static Status ListColumnFamilies(const DBOptions& db_options, const std::string& name, std::vector* column_families); DB() { } virtual ~DB(); // Create a column_family and return the handle of column family // through the argument handle. virtual Status CreateColumnFamily(const ColumnFamilyOptions& options, const std::string& column_family_name, ColumnFamilyHandle** handle); // Drop a column family specified by column_family handle. This call // only records a drop record in the manifest and prevents the column // family from flushing and compacting. virtual Status DropColumnFamily(ColumnFamilyHandle* column_family); // Set the database entry for "key" to "value". // If "key" already exists, it will be overwritten. // Returns OK on success, and a non-OK status on error. // Note: consider setting options.sync = true. virtual Status Put(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual Status Put(const WriteOptions& options, const Slice& key, const Slice& value) { return Put(options, DefaultColumnFamily(), key, value); } // Remove the database entry (if any) for "key". Returns OK on // success, and a non-OK status on error. It is not an error if "key" // did not exist in the database. // Note: consider setting options.sync = true. virtual Status Delete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual Status Delete(const WriteOptions& options, const Slice& key) { return Delete(options, DefaultColumnFamily(), key); } // Remove the database entry for "key". Requires that the key exists // and was not overwritten. Returns OK on success, and a non-OK status // on error. It is not an error if "key" did not exist in the database. // // If a key is overwritten (by calling Put() multiple times), then the result // of calling SingleDelete() on this key is undefined. SingleDelete() only // behaves correctly if there has been only one Put() for this key since the // previous call to SingleDelete() for this key. // // This feature is currently an experimental performance optimization // for a very specific workload. It is up to the caller to ensure that // SingleDelete is only used for a key that is not deleted using Delete() or // written using Merge(). Mixing SingleDelete operations with Deletes and // Merges can result in undefined behavior. // // Note: consider setting options.sync = true. virtual Status SingleDelete(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual Status SingleDelete(const WriteOptions& options, const Slice& key) { return SingleDelete(options, DefaultColumnFamily(), key); } // Merge the database entry for "key" with "value". Returns OK on success, // and a non-OK status on error. The semantics of this operation is // determined by the user provided merge_operator when opening DB. // Note: consider setting options.sync = true. virtual Status Merge(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual Status Merge(const WriteOptions& options, const Slice& key, const Slice& value) { return Merge(options, DefaultColumnFamily(), key, value); } // Apply the specified updates to the database. // If `updates` contains no update, WAL will still be synced if // options.sync=true. // Returns OK on success, non-OK on failure. // Note: consider setting options.sync = true. virtual Status Write(const WriteOptions& options, WriteBatch* updates) = 0; // If the database contains an entry for "key" store the // corresponding value in *value and return OK. // // If there is no entry for "key" leave *value unchanged and return // a status for which Status::IsNotFound() returns true. // // May return some other Status on an error. virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) = 0; virtual Status Get(const ReadOptions& options, const Slice& key, std::string* value) { return Get(options, DefaultColumnFamily(), key, value); } // If keys[i] does not exist in the database, then the i'th returned // status will be one for which Status::IsNotFound() is true, and // (*values)[i] will be set to some arbitrary value (often ""). Otherwise, // the i'th returned status will have Status::ok() true, and (*values)[i] // will store the value associated with keys[i]. // // (*values) will always be resized to be the same size as (keys). // Similarly, the number of returned statuses will be the number of keys. // Note: keys will not be "de-duplicated". Duplicate keys will return // duplicate values in order. virtual std::vector MultiGet( const ReadOptions& options, const std::vector& column_family, const std::vector& keys, std::vector* values) = 0; virtual std::vector MultiGet(const ReadOptions& options, const std::vector& keys, std::vector* values) { return MultiGet(options, std::vector( keys.size(), DefaultColumnFamily()), keys, values); } // If the key definitely does not exist in the database, then this method // returns false, else true. If the caller wants to obtain value when the key // is found in memory, a bool for 'value_found' must be passed. 'value_found' // will be true on return if value has been set properly. // This check is potentially lighter-weight than invoking DB::Get(). One way // to make this lighter weight is to avoid doing any IOs. // Default implementation here returns true and sets 'value_found' to false virtual bool KeyMayExist(const ReadOptions& /*options*/, ColumnFamilyHandle* /*column_family*/, const Slice& /*key*/, std::string* /*value*/, bool* value_found = nullptr) { if (value_found != nullptr) { *value_found = false; } return true; } virtual bool KeyMayExist(const ReadOptions& options, const Slice& key, std::string* value, bool* value_found = nullptr) { return KeyMayExist(options, DefaultColumnFamily(), key, value, value_found); } // Return a heap-allocated iterator over the contents of the database. // The result of NewIterator() is initially invalid (caller must // call one of the Seek methods on the iterator before using it). // // Caller should delete the iterator when it is no longer needed. // The returned iterator should be deleted before this db is deleted. virtual Iterator* NewIterator(const ReadOptions& options, ColumnFamilyHandle* column_family) = 0; virtual Iterator* NewIterator(const ReadOptions& options) { return NewIterator(options, DefaultColumnFamily()); } // Returns iterators from a consistent database state across multiple // column families. Iterators are heap allocated and need to be deleted // before the db is deleted virtual Status NewIterators( const ReadOptions& options, const std::vector& column_families, std::vector* iterators) = 0; // Return a handle to the current DB state. Iterators created with // this handle will all observe a stable snapshot of the current DB // state. The caller must call ReleaseSnapshot(result) when the // snapshot is no longer needed. // // nullptr will be returned if the DB fails to take a snapshot or does // not support snapshot. virtual const Snapshot* GetSnapshot() = 0; // Release a previously acquired snapshot. The caller must not // use "snapshot" after this call. virtual void ReleaseSnapshot(const Snapshot* snapshot) = 0; #ifndef ROCKSDB_LITE // Contains all valid property arguments for GetProperty(). // // NOTE: Property names cannot end in numbers since those are interpreted as // arguments, e.g., see kNumFilesAtLevelPrefix. struct Properties { // "rocksdb.num-files-at-level" - returns string containing the number // of files at level , where is an ASCII representation of a // level number (e.g., "0"). static const std::string kNumFilesAtLevelPrefix; // "rocksdb.stats" - returns a multi-line string containing the data // described by kCFStats followed by the data described by kDBStats. static const std::string kStats; // "rocksdb.sstables" - returns a multi-line string summarizing current // SST files. static const std::string kSSTables; // "rocksdb.cfstats" - returns a multi-line string with general column // family stats per-level over db's lifetime ("L"), aggregated over // db's lifetime ("Sum"), and aggregated over the interval since the // last retrieval ("Int"). static const std::string kCFStats; // "rocksdb.dbstats" - returns a multi-line string with general database // stats, both cumulative (over the db's lifetime) and interval (since // the last retrieval of kDBStats). static const std::string kDBStats; // "rocksdb.levelstats" - returns multi-line string containing the number // of files per level and total size of each level (MB). static const std::string kLevelStats; // "rocksdb.num-immutable-mem-table" - returns number of immutable // memtables that have not yet been flushed. static const std::string kNumImmutableMemTable; // "rocksdb.num-immutable-mem-table-flushed" - returns number of immutable // memtables that have already been flushed. static const std::string kNumImmutableMemTableFlushed; // "rocksdb.mem-table-flush-pending" - returns 1 if a memtable flush is // pending; otherwise, returns 0. static const std::string kMemTableFlushPending; // "rocksdb.num-running-flushes" - returns the number of currently running // flushes. static const std::string kNumRunningFlushes; // "rocksdb.compaction-pending" - returns 1 if at least one compaction is // pending; otherwise, returns 0. static const std::string kCompactionPending; // "rocksdb.num-running-compactions" - returns the number of currently // running compactions. static const std::string kNumRunningCompactions; // "rocksdb.background-errors" - returns accumulated number of background // errors. static const std::string kBackgroundErrors; // "rocksdb.cur-size-active-mem-table" - returns approximate size of active // memtable (bytes). static const std::string kCurSizeActiveMemTable; // "rocksdb.cur-size-all-mem-tables" - returns approximate size of active // and unflushed immutable memtables (bytes). static const std::string kCurSizeAllMemTables; // "rocksdb.size-all-mem-tables" - returns approximate size of active, // unflushed immutable, and pinned immutable memtables (bytes). static const std::string kSizeAllMemTables; // "rocksdb.num-entries-active-mem-table" - returns total number of entries // in the active memtable. static const std::string kNumEntriesActiveMemTable; // "rocksdb.num-entries-imm-mem-tables" - returns total number of entries // in the unflushed immutable memtables. static const std::string kNumEntriesImmMemTables; // "rocksdb.num-deletes-active-mem-table" - returns total number of delete // entries in the active memtable. static const std::string kNumDeletesActiveMemTable; // "rocksdb.num-deletes-imm-mem-tables" - returns total number of delete // entries in the unflushed immutable memtables. static const std::string kNumDeletesImmMemTables; // "rocksdb.estimate-num-keys" - returns estimated number of total keys in // the active and unflushed immutable memtables. static const std::string kEstimateNumKeys; // "rocksdb.estimate-table-readers-mem" - returns estimated memory used for // reading SST tables, excluding memory used in block cache (e.g., // filter and index blocks). static const std::string kEstimateTableReadersMem; // "rocksdb.is-file-deletions-enabled" - returns 0 if deletion of obsolete // files is enabled; otherwise, returns a non-zero number. static const std::string kIsFileDeletionsEnabled; // "rocksdb.num-snapshots" - returns number of unreleased snapshots of the // database. static const std::string kNumSnapshots; // "rocksdb.oldest-snapshot-time" - returns number representing unix // timestamp of oldest unreleased snapshot. static const std::string kOldestSnapshotTime; // "rocksdb.num-live-versions" - returns number of live versions. `Version` // is an internal data structure. See version_set.h for details. More // live versions often mean more SST files are held from being deleted, // by iterators or unfinished compactions. static const std::string kNumLiveVersions; // "rocksdb.estimate-live-data-size" - returns an estimate of the amount of // live data in bytes. static const std::string kEstimateLiveDataSize; // "rocksdb.total-sst-files-size" - returns total size (bytes) of all SST // files. // WARNING: may slow down online queries if there are too many files. static const std::string kTotalSstFilesSize; // "rocksdb.base-level" - returns number of level to which L0 data will be // compacted. static const std::string kBaseLevel; // "rocksdb.estimate-pending-compaction-bytes" - returns estimated total // number of bytes compaction needs to rewrite to get all levels down // to under target size. Not valid for other compactions than level- // based. static const std::string kEstimatePendingCompactionBytes; // "rocksdb.aggregated-table-properties" - returns a string representation // of the aggregated table properties of the target column family. static const std::string kAggregatedTableProperties; // "rocksdb.aggregated-table-properties-at-level", same as the previous // one but only returns the aggregated table properties of the // specified level "N" at the target column family. static const std::string kAggregatedTablePropertiesAtLevel; }; #endif /* ROCKSDB_LITE */ // DB implementations can export properties about their state via this method. // If "property" is a valid property understood by this DB implementation (see // Properties struct above for valid options), fills "*value" with its current // value and returns true. Otherwise, returns false. virtual bool GetProperty(ColumnFamilyHandle* column_family, const Slice& property, std::string* value) = 0; virtual bool GetProperty(const Slice& property, std::string* value) { return GetProperty(DefaultColumnFamily(), property, value); } // Similar to GetProperty(), but only works for a subset of properties whose // return value is an integer. Return the value by integer. Supported // properties: // "rocksdb.num-immutable-mem-table" // "rocksdb.mem-table-flush-pending" // "rocksdb.compaction-pending" // "rocksdb.background-errors" // "rocksdb.cur-size-active-mem-table" // "rocksdb.cur-size-all-mem-tables" // "rocksdb.size-all-mem-tables" // "rocksdb.num-entries-active-mem-table" // "rocksdb.num-entries-imm-mem-tables" // "rocksdb.num-deletes-active-mem-table" // "rocksdb.num-deletes-imm-mem-tables" // "rocksdb.estimate-num-keys" // "rocksdb.estimate-table-readers-mem" // "rocksdb.is-file-deletions-enabled" // "rocksdb.num-snapshots" // "rocksdb.oldest-snapshot-time" // "rocksdb.num-live-versions" // "rocksdb.estimate-live-data-size" // "rocksdb.total-sst-files-size" // "rocksdb.base-level" // "rocksdb.estimate-pending-compaction-bytes" // "rocksdb.num-running-compactions" // "rocksdb.num-running-flushes" virtual bool GetIntProperty(ColumnFamilyHandle* column_family, const Slice& property, uint64_t* value) = 0; virtual bool GetIntProperty(const Slice& property, uint64_t* value) { return GetIntProperty(DefaultColumnFamily(), property, value); } // Same as GetIntProperty(), but this one returns the aggregated int // property from all column families. virtual bool GetAggregatedIntProperty(const Slice& property, uint64_t* value) = 0; // For each i in [0,n-1], store in "sizes[i]", the approximate // file system space used by keys in "[range[i].start .. range[i].limit)". // // Note that the returned sizes measure file system space usage, so // if the user data compresses by a factor of ten, the returned // sizes will be one-tenth the size of the corresponding user data size. // // If include_memtable is set to true, then the result will also // include those recently written data in the mem-tables if // the mem-table type supports it. virtual void GetApproximateSizes(ColumnFamilyHandle* column_family, const Range* range, int n, uint64_t* sizes, bool include_memtable = false) = 0; virtual void GetApproximateSizes(const Range* range, int n, uint64_t* sizes, bool include_memtable = false) { GetApproximateSizes(DefaultColumnFamily(), range, n, sizes, include_memtable); } // Compact the underlying storage for the key range [*begin,*end]. // The actual compaction interval might be superset of [*begin, *end]. // In particular, deleted and overwritten versions are discarded, // and the data is rearranged to reduce the cost of operations // needed to access the data. This operation should typically only // be invoked by users who understand the underlying implementation. // // begin==nullptr is treated as a key before all keys in the database. // end==nullptr is treated as a key after all keys in the database. // Therefore the following call will compact the entire database: // db->CompactRange(options, nullptr, nullptr); // Note that after the entire database is compacted, all data are pushed // down to the last level containing any data. If the total data size after // compaction is reduced, that level might not be appropriate for hosting all // the files. In this case, client could set options.change_level to true, to // move the files back to the minimum level capable of holding the data set // or a given level (specified by non-negative options.target_level). virtual Status CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) = 0; virtual Status CompactRange(const CompactRangeOptions& options, const Slice* begin, const Slice* end) { return CompactRange(options, DefaultColumnFamily(), begin, end); } #if defined(__GNUC__) || defined(__clang__) __attribute__((deprecated)) #elif _WIN32 __declspec(deprecated) #endif virtual Status CompactRange(ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end, bool change_level = false, int target_level = -1, uint32_t target_path_id = 0) { CompactRangeOptions options; options.change_level = change_level; options.target_level = target_level; options.target_path_id = target_path_id; return CompactRange(options, column_family, begin, end); } #if defined(__GNUC__) || defined(__clang__) __attribute__((deprecated)) #elif _WIN32 __declspec(deprecated) #endif virtual Status CompactRange(const Slice* begin, const Slice* end, bool change_level = false, int target_level = -1, uint32_t target_path_id = 0) { CompactRangeOptions options; options.change_level = change_level; options.target_level = target_level; options.target_path_id = target_path_id; return CompactRange(options, DefaultColumnFamily(), begin, end); } virtual Status SetOptions( ColumnFamilyHandle* /*column_family*/, const std::unordered_map& /*new_options*/) { return Status::NotSupported("Not implemented"); } virtual Status SetOptions( const std::unordered_map& new_options) { return SetOptions(DefaultColumnFamily(), new_options); } // CompactFiles() inputs a list of files specified by file numbers and // compacts them to the specified level. Note that the behavior is different // from CompactRange() in that CompactFiles() performs the compaction job // using the CURRENT thread. // // @see GetDataBaseMetaData // @see GetColumnFamilyMetaData virtual Status CompactFiles( const CompactionOptions& compact_options, ColumnFamilyHandle* column_family, const std::vector& input_file_names, const int output_level, const int output_path_id = -1) = 0; virtual Status CompactFiles( const CompactionOptions& compact_options, const std::vector& input_file_names, const int output_level, const int output_path_id = -1) { return CompactFiles(compact_options, DefaultColumnFamily(), input_file_names, output_level, output_path_id); } // This function will wait until all currently running background processes // finish. After it returns, no background process will be run until // UnblockBackgroundWork is called virtual Status PauseBackgroundWork() = 0; virtual Status ContinueBackgroundWork() = 0; // This function will enable automatic compactions for the given column // families if they were previously disabled. The function will first set the // disable_auto_compactions option for each column family to 'false', after // which it will schedule a flush/compaction. // // NOTE: Setting disable_auto_compactions to 'false' through SetOptions() API // does NOT schedule a flush/compaction afterwards, and only changes the // parameter itself within the column family option. // virtual Status EnableAutoCompaction( const std::vector& column_family_handles) = 0; // Number of levels used for this DB. virtual int NumberLevels(ColumnFamilyHandle* column_family) = 0; virtual int NumberLevels() { return NumberLevels(DefaultColumnFamily()); } // Maximum level to which a new compacted memtable is pushed if it // does not create overlap. virtual int MaxMemCompactionLevel(ColumnFamilyHandle* column_family) = 0; virtual int MaxMemCompactionLevel() { return MaxMemCompactionLevel(DefaultColumnFamily()); } // Number of files in level-0 that would stop writes. virtual int Level0StopWriteTrigger(ColumnFamilyHandle* column_family) = 0; virtual int Level0StopWriteTrigger() { return Level0StopWriteTrigger(DefaultColumnFamily()); } // Get DB name -- the exact same name that was provided as an argument to // DB::Open() virtual const std::string& GetName() const = 0; // Get Env object from the DB virtual Env* GetEnv() const = 0; // Get DB Options that we use. During the process of opening the // column family, the options provided when calling DB::Open() or // DB::CreateColumnFamily() will have been "sanitized" and transformed // in an implementation-defined manner. virtual const Options& GetOptions(ColumnFamilyHandle* column_family) const = 0; virtual const Options& GetOptions() const { return GetOptions(DefaultColumnFamily()); } virtual const DBOptions& GetDBOptions() const = 0; // Flush all mem-table data. virtual Status Flush(const FlushOptions& options, ColumnFamilyHandle* column_family) = 0; virtual Status Flush(const FlushOptions& options) { return Flush(options, DefaultColumnFamily()); } // Sync the wal. Note that Write() followed by SyncWAL() is not exactly the // same as Write() with sync=true: in the latter case the changes won't be // visible until the sync is done. // Currently only works if allow_mmap_writes = false in Options. virtual Status SyncWAL() = 0; // The sequence number of the most recent transaction. virtual SequenceNumber GetLatestSequenceNumber() const = 0; #ifndef ROCKSDB_LITE // Prevent file deletions. Compactions will continue to occur, // but no obsolete files will be deleted. Calling this multiple // times have the same effect as calling it once. virtual Status DisableFileDeletions() = 0; // Allow compactions to delete obsolete files. // If force == true, the call to EnableFileDeletions() will guarantee that // file deletions are enabled after the call, even if DisableFileDeletions() // was called multiple times before. // If force == false, EnableFileDeletions will only enable file deletion // after it's been called at least as many times as DisableFileDeletions(), // enabling the two methods to be called by two threads concurrently without // synchronization -- i.e., file deletions will be enabled only after both // threads call EnableFileDeletions() virtual Status EnableFileDeletions(bool force = true) = 0; // GetLiveFiles followed by GetSortedWalFiles can generate a lossless backup // Retrieve the list of all files in the database. The files are // relative to the dbname and are not absolute paths. The valid size of the // manifest file is returned in manifest_file_size. The manifest file is an // ever growing file, but only the portion specified by manifest_file_size is // valid for this snapshot. // Setting flush_memtable to true does Flush before recording the live files. // Setting flush_memtable to false is useful when we don't want to wait for // flush which may have to wait for compaction to complete taking an // indeterminate time. // // In case you have multiple column families, even if flush_memtable is true, // you still need to call GetSortedWalFiles after GetLiveFiles to compensate // for new data that arrived to already-flushed column families while other // column families were flushing virtual Status GetLiveFiles(std::vector&, uint64_t* manifest_file_size, bool flush_memtable = true) = 0; // Retrieve the sorted list of all wal files with earliest file first virtual Status GetSortedWalFiles(VectorLogPtr& files) = 0; // Sets iter to an iterator that is positioned at a write-batch containing // seq_number. If the sequence number is non existent, it returns an iterator // at the first available seq_no after the requested seq_no // Returns Status::OK if iterator is valid // Must set WAL_ttl_seconds or WAL_size_limit_MB to large values to // use this api, else the WAL files will get // cleared aggressively and the iterator might keep getting invalid before // an update is read. virtual Status GetUpdatesSince( SequenceNumber seq_number, unique_ptr* iter, const TransactionLogIterator::ReadOptions& read_options = TransactionLogIterator::ReadOptions()) = 0; // Windows API macro interference #undef DeleteFile // Delete the file name from the db directory and update the internal state to // reflect that. Supports deletion of sst and log files only. 'name' must be // path relative to the db directory. eg. 000001.sst, /archive/000003.log virtual Status DeleteFile(std::string name) = 0; // Returns a list of all table files with their level, start key // and end key virtual void GetLiveFilesMetaData( std::vector* /*metadata*/) {} // Obtains the meta data of the specified column family of the DB. // Status::NotFound() will be returned if the current DB does not have // any column family match the specified name. // // If cf_name is not specified, then the metadata of the default // column family will be returned. virtual void GetColumnFamilyMetaData(ColumnFamilyHandle* /*column_family*/, ColumnFamilyMetaData* /*metadata*/) {} // Get the metadata of the default column family. void GetColumnFamilyMetaData( ColumnFamilyMetaData* metadata) { GetColumnFamilyMetaData(DefaultColumnFamily(), metadata); } // Load table file located at "file_path" into "column_family", a pointer to // ExternalSstFileInfo can be used instead of "file_path" to do a blind add // that wont need to read the file, move_file can be set to true to // move the file instead of copying it. // // Current Requirements: // (1) Key range in loaded table file don't overlap with // existing keys or tombstones in DB. // (2) No other writes happen during AddFile call, otherwise // DB may get corrupted. // (3) No snapshots are held. virtual Status AddFile(ColumnFamilyHandle* column_family, const std::string& file_path, bool move_file = false) = 0; virtual Status AddFile(const std::string& file_path, bool move_file = false) { return AddFile(DefaultColumnFamily(), file_path, move_file); } // Load table file with information "file_info" into "column_family" virtual Status AddFile(ColumnFamilyHandle* column_family, const ExternalSstFileInfo* file_info, bool move_file = false) = 0; virtual Status AddFile(const ExternalSstFileInfo* file_info, bool move_file = false) { return AddFile(DefaultColumnFamily(), file_info, move_file); } #endif // ROCKSDB_LITE // Sets the globally unique ID created at database creation time by invoking // Env::GenerateUniqueId(), in identity. Returns Status::OK if identity could // be set properly virtual Status GetDbIdentity(std::string& identity) const = 0; // Returns default column family handle virtual ColumnFamilyHandle* DefaultColumnFamily() const = 0; #ifndef ROCKSDB_LITE virtual Status GetPropertiesOfAllTables(ColumnFamilyHandle* column_family, TablePropertiesCollection* props) = 0; virtual Status GetPropertiesOfAllTables(TablePropertiesCollection* props) { return GetPropertiesOfAllTables(DefaultColumnFamily(), props); } virtual Status GetPropertiesOfTablesInRange( ColumnFamilyHandle* column_family, const Range* range, std::size_t n, TablePropertiesCollection* props) = 0; #endif // ROCKSDB_LITE // Needed for StackableDB virtual DB* GetRootDB() { return this; } private: // No copying allowed DB(const DB&); void operator=(const DB&); }; // Destroy the contents of the specified database. // Be very careful using this method. Status DestroyDB(const std::string& name, const Options& options); #ifndef ROCKSDB_LITE // If a DB cannot be opened, you may attempt to call this method to // resurrect as much of the contents of the database as possible. // Some data may be lost, so be careful when calling this function // on a database that contains important information. Status RepairDB(const std::string& dbname, const Options& options); #endif } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_DB_H_ rocksdb-4.5.1/include/rocksdb/db_dump_tool.h000066400000000000000000000023051267530634000210500ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/db.h" namespace rocksdb { struct DumpOptions { // Database that will be dumped std::string db_path; // File location that will contain dump output std::string dump_location; // Dont include db information header in the dump bool anonymous = false; }; class DbDumpTool { public: bool Run(const DumpOptions& dump_options, rocksdb::Options options = rocksdb::Options()); }; struct UndumpOptions { // Database that we will load the dumped file into std::string db_path; // File location of the dumped file that will be loaded std::string dump_location; // Compact the db after loading the dumped file bool compact_db = false; }; class DbUndumpTool { public: bool Run(const UndumpOptions& undump_options, rocksdb::Options options = rocksdb::Options()); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/env.h000066400000000000000000001064661267530634000172060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // An Env is an interface used by the rocksdb implementation to access // operating system functionality like the filesystem etc. Callers // may wish to provide a custom Env object when opening a database to // get fine gain control; e.g., to rate limit file system operations. // // All Env implementations are safe for concurrent access from // multiple threads without any external synchronization. #ifndef STORAGE_ROCKSDB_INCLUDE_ENV_H_ #define STORAGE_ROCKSDB_INCLUDE_ENV_H_ #include #include #include #include #include #include #include "rocksdb/status.h" #include "rocksdb/thread_status.h" #ifdef _WIN32 // Windows API macro interference #undef DeleteFile #undef GetCurrentTime #endif namespace rocksdb { class FileLock; class Logger; class RandomAccessFile; class SequentialFile; class Slice; class WritableFile; class Directory; struct DBOptions; class RateLimiter; class ThreadStatusUpdater; struct ThreadStatus; using std::unique_ptr; using std::shared_ptr; // Options while opening a file to read/write struct EnvOptions { // construct with default Options EnvOptions(); // construct from Options explicit EnvOptions(const DBOptions& options); // If true, then allow caching of data in environment buffers bool use_os_buffer = true; // If true, then use mmap to read data bool use_mmap_reads = false; // If true, then use mmap to write data bool use_mmap_writes = true; // If false, fallocate() calls are bypassed bool allow_fallocate = true; // If true, set the FD_CLOEXEC on open fd. bool set_fd_cloexec = true; // Allows OS to incrementally sync files to disk while they are being // written, in the background. Issue one request for every bytes_per_sync // written. 0 turns it off. // Default: 0 uint64_t bytes_per_sync = 0; // If true, we will preallocate the file with FALLOC_FL_KEEP_SIZE flag, which // means that file size won't change as part of preallocation. // If false, preallocation will also change the file size. This option will // improve the performance in workloads where you sync the data on every // write. By default, we set it to true for MANIFEST writes and false for // WAL writes bool fallocate_with_keep_size = true; // See DBOPtions doc size_t compaction_readahead_size; // See DBOPtions doc size_t random_access_max_buffer_size; // See DBOptions doc size_t writable_file_max_buffer_size = 1024 * 1024; // If not nullptr, write rate limiting is enabled for flush and compaction RateLimiter* rate_limiter = nullptr; }; class Env { public: Env() : thread_status_updater_(nullptr) {} virtual ~Env(); // Return a default environment suitable for the current operating // system. Sophisticated users may wish to provide their own Env // implementation instead of relying on this default environment. // // The result of Default() belongs to rocksdb and must never be deleted. static Env* Default(); // Create a brand new sequentially-readable file with the specified name. // On success, stores a pointer to the new file in *result and returns OK. // On failure stores nullptr in *result and returns non-OK. If the file does // not exist, returns a non-OK status. // // The returned file will only be accessed by one thread at a time. virtual Status NewSequentialFile(const std::string& fname, unique_ptr* result, const EnvOptions& options) = 0; // Create a brand new random access read-only file with the // specified name. On success, stores a pointer to the new file in // *result and returns OK. On failure stores nullptr in *result and // returns non-OK. If the file does not exist, returns a non-OK // status. // // The returned file may be concurrently accessed by multiple threads. virtual Status NewRandomAccessFile(const std::string& fname, unique_ptr* result, const EnvOptions& options) = 0; // Create an object that writes to a new file with the specified // name. Deletes any existing file with the same name and creates a // new file. On success, stores a pointer to the new file in // *result and returns OK. On failure stores nullptr in *result and // returns non-OK. // // The returned file will only be accessed by one thread at a time. virtual Status NewWritableFile(const std::string& fname, unique_ptr* result, const EnvOptions& options) = 0; // Reuse an existing file by renaming it and opening it as writable. virtual Status ReuseWritableFile(const std::string& fname, const std::string& old_fname, unique_ptr* result, const EnvOptions& options); // Create an object that represents a directory. Will fail if directory // doesn't exist. If the directory exists, it will open the directory // and create a new Directory object. // // On success, stores a pointer to the new Directory in // *result and returns OK. On failure stores nullptr in *result and // returns non-OK. virtual Status NewDirectory(const std::string& name, unique_ptr* result) = 0; // Returns OK if the named file exists. // NotFound if the named file does not exist, // the calling process does not have permission to determine // whether this file exists, or if the path is invalid. // IOError if an IO Error was encountered virtual Status FileExists(const std::string& fname) = 0; // Store in *result the names of the children of the specified directory. // The names are relative to "dir". // Original contents of *results are dropped. virtual Status GetChildren(const std::string& dir, std::vector* result) = 0; // Delete the named file. virtual Status DeleteFile(const std::string& fname) = 0; // Create the specified directory. Returns error if directory exists. virtual Status CreateDir(const std::string& dirname) = 0; // Creates directory if missing. Return Ok if it exists, or successful in // Creating. virtual Status CreateDirIfMissing(const std::string& dirname) = 0; // Delete the specified directory. virtual Status DeleteDir(const std::string& dirname) = 0; // Store the size of fname in *file_size. virtual Status GetFileSize(const std::string& fname, uint64_t* file_size) = 0; // Store the last modification time of fname in *file_mtime. virtual Status GetFileModificationTime(const std::string& fname, uint64_t* file_mtime) = 0; // Rename file src to target. virtual Status RenameFile(const std::string& src, const std::string& target) = 0; // Hard Link file src to target. virtual Status LinkFile(const std::string& src, const std::string& target) { return Status::NotSupported("LinkFile is not supported for this Env"); } // Lock the specified file. Used to prevent concurrent access to // the same db by multiple processes. On failure, stores nullptr in // *lock and returns non-OK. // // On success, stores a pointer to the object that represents the // acquired lock in *lock and returns OK. The caller should call // UnlockFile(*lock) to release the lock. If the process exits, // the lock will be automatically released. // // If somebody else already holds the lock, finishes immediately // with a failure. I.e., this call does not wait for existing locks // to go away. // // May create the named file if it does not already exist. virtual Status LockFile(const std::string& fname, FileLock** lock) = 0; // Release the lock acquired by a previous successful call to LockFile. // REQUIRES: lock was returned by a successful LockFile() call // REQUIRES: lock has not already been unlocked. virtual Status UnlockFile(FileLock* lock) = 0; // Priority for scheduling job in thread pool enum Priority { LOW, HIGH, TOTAL }; // Priority for requesting bytes in rate limiter scheduler enum IOPriority { IO_LOW = 0, IO_HIGH = 1, IO_TOTAL = 2 }; // Arrange to run "(*function)(arg)" once in a background thread, in // the thread pool specified by pri. By default, jobs go to the 'LOW' // priority thread pool. // "function" may run in an unspecified thread. Multiple functions // added to the same Env may run concurrently in different threads. // I.e., the caller may not assume that background work items are // serialized. // When the UnSchedule function is called, the unschedFunction // registered at the time of Schedule is invoked with arg as a parameter. virtual void Schedule(void (*function)(void* arg), void* arg, Priority pri = LOW, void* tag = nullptr, void (*unschedFunction)(void* arg) = 0) = 0; // Arrange to remove jobs for given arg from the queue_ if they are not // already scheduled. Caller is expected to have exclusive lock on arg. virtual int UnSchedule(void* arg, Priority pri) { return 0; } // Start a new thread, invoking "function(arg)" within the new thread. // When "function(arg)" returns, the thread will be destroyed. virtual void StartThread(void (*function)(void* arg), void* arg) = 0; // Wait for all threads started by StartThread to terminate. virtual void WaitForJoin() {} // Get thread pool queue length for specific thrad pool. virtual unsigned int GetThreadPoolQueueLen(Priority pri = LOW) const { return 0; } // *path is set to a temporary directory that can be used for testing. It may // or many not have just been created. The directory may or may not differ // between runs of the same process, but subsequent calls will return the // same directory. virtual Status GetTestDirectory(std::string* path) = 0; // Create and return a log file for storing informational messages. virtual Status NewLogger(const std::string& fname, shared_ptr* result) = 0; // Returns the number of micro-seconds since some fixed point in time. Only // useful for computing deltas of time. // However, it is often used as system time such as in GenericRateLimiter // and other places so a port needs to return system time in order to work. virtual uint64_t NowMicros() = 0; // Returns the number of nano-seconds since some fixed point in time. Only // useful for computing deltas of time in one run. // Default implementation simply relies on NowMicros virtual uint64_t NowNanos() { return NowMicros() * 1000; } // Sleep/delay the thread for the perscribed number of micro-seconds. virtual void SleepForMicroseconds(int micros) = 0; // Get the current host name. virtual Status GetHostName(char* name, uint64_t len) = 0; // Get the number of seconds since the Epoch, 1970-01-01 00:00:00 (UTC). virtual Status GetCurrentTime(int64_t* unix_time) = 0; // Get full directory name for this db. virtual Status GetAbsolutePath(const std::string& db_path, std::string* output_path) = 0; // The number of background worker threads of a specific thread pool // for this environment. 'LOW' is the default pool. // default number: 1 virtual void SetBackgroundThreads(int number, Priority pri = LOW) = 0; // Enlarge number of background worker threads of a specific thread pool // for this environment if it is smaller than specified. 'LOW' is the default // pool. virtual void IncBackgroundThreadsIfNeeded(int number, Priority pri) = 0; // Lower IO priority for threads from the specified pool. virtual void LowerThreadPoolIOPriority(Priority pool = LOW) {} // Converts seconds-since-Jan-01-1970 to a printable string virtual std::string TimeToString(uint64_t time) = 0; // Generates a unique id that can be used to identify a db virtual std::string GenerateUniqueId(); // OptimizeForLogWrite will create a new EnvOptions object that is a copy of // the EnvOptions in the parameters, but is optimized for writing log files. // Default implementation returns the copy of the same object. virtual EnvOptions OptimizeForLogWrite(const EnvOptions& env_options, const DBOptions& db_options) const; // OptimizeForManifestWrite will create a new EnvOptions object that is a copy // of the EnvOptions in the parameters, but is optimized for writing manifest // files. Default implementation returns the copy of the same object. virtual EnvOptions OptimizeForManifestWrite(const EnvOptions& env_options) const; // Returns the status of all threads that belong to the current Env. virtual Status GetThreadList(std::vector* thread_list) { return Status::NotSupported("Not supported."); } // Returns the pointer to ThreadStatusUpdater. This function will be // used in RocksDB internally to update thread status and supports // GetThreadList(). virtual ThreadStatusUpdater* GetThreadStatusUpdater() const { return thread_status_updater_; } // Returns the ID of the current thread. virtual uint64_t GetThreadID() const; protected: // The pointer to an internal structure that will update the // status of each thread. ThreadStatusUpdater* thread_status_updater_; private: // No copying allowed Env(const Env&); void operator=(const Env&); }; // The factory function to construct a ThreadStatusUpdater. Any Env // that supports GetThreadList() feature should call this function in its // constructor to initialize thread_status_updater_. ThreadStatusUpdater* CreateThreadStatusUpdater(); // A file abstraction for reading sequentially through a file class SequentialFile { public: SequentialFile() { } virtual ~SequentialFile(); // Read up to "n" bytes from the file. "scratch[0..n-1]" may be // written by this routine. Sets "*result" to the data that was // read (including if fewer than "n" bytes were successfully read). // May set "*result" to point at data in "scratch[0..n-1]", so // "scratch[0..n-1]" must be live when "*result" is used. // If an error was encountered, returns a non-OK status. // // REQUIRES: External synchronization virtual Status Read(size_t n, Slice* result, char* scratch) = 0; // Skip "n" bytes from the file. This is guaranteed to be no // slower that reading the same data, but may be faster. // // If end of file is reached, skipping will stop at the end of the // file, and Skip will return OK. // // REQUIRES: External synchronization virtual Status Skip(uint64_t n) = 0; // Remove any kind of caching of data from the offset to offset+length // of this file. If the length is 0, then it refers to the end of file. // If the system is not caching the file contents, then this is a noop. virtual Status InvalidateCache(size_t offset, size_t length) { return Status::NotSupported("InvalidateCache not supported."); } }; // A file abstraction for randomly reading the contents of a file. class RandomAccessFile { public: RandomAccessFile() { } virtual ~RandomAccessFile(); // Read up to "n" bytes from the file starting at "offset". // "scratch[0..n-1]" may be written by this routine. Sets "*result" // to the data that was read (including if fewer than "n" bytes were // successfully read). May set "*result" to point at data in // "scratch[0..n-1]", so "scratch[0..n-1]" must be live when // "*result" is used. If an error was encountered, returns a non-OK // status. // // Safe for concurrent use by multiple threads. virtual Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const = 0; // Used by the file_reader_writer to decide if the ReadAhead wrapper // should simply forward the call and do not enact buffering or locking. virtual bool ShouldForwardRawRequest() const { return false; } // For cases when read-ahead is implemented in the platform dependent // layer virtual void EnableReadAhead() {} // Tries to get an unique ID for this file that will be the same each time // the file is opened (and will stay the same while the file is open). // Furthermore, it tries to make this ID at most "max_size" bytes. If such an // ID can be created this function returns the length of the ID and places it // in "id"; otherwise, this function returns 0, in which case "id" // may not have been modified. // // This function guarantees, for IDs from a given environment, two unique ids // cannot be made equal to eachother by adding arbitrary bytes to one of // them. That is, no unique ID is the prefix of another. // // This function guarantees that the returned ID will not be interpretable as // a single varint. // // Note: these IDs are only valid for the duration of the process. virtual size_t GetUniqueId(char* id, size_t max_size) const { return 0; // Default implementation to prevent issues with backwards // compatibility. }; enum AccessPattern { NORMAL, RANDOM, SEQUENTIAL, WILLNEED, DONTNEED }; virtual void Hint(AccessPattern pattern) {} // Remove any kind of caching of data from the offset to offset+length // of this file. If the length is 0, then it refers to the end of file. // If the system is not caching the file contents, then this is a noop. virtual Status InvalidateCache(size_t offset, size_t length) { return Status::NotSupported("InvalidateCache not supported."); } }; // A file abstraction for sequential writing. The implementation // must provide buffering since callers may append small fragments // at a time to the file. class WritableFile { public: WritableFile() : last_preallocated_block_(0), preallocation_block_size_(0), io_priority_(Env::IO_TOTAL) { } virtual ~WritableFile(); // Indicates if the class makes use of unbuffered I/O virtual bool UseOSBuffer() const { return true; } const size_t c_DefaultPageSize = 4 * 1024; // This is needed when you want to allocate // AlignedBuffer for use with file I/O classes // Used for unbuffered file I/O when UseOSBuffer() returns false virtual size_t GetRequiredBufferAlignment() const { return c_DefaultPageSize; } virtual Status Append(const Slice& data) = 0; // Positioned write for unbuffered access default forward // to simple append as most of the tests are buffered by default virtual Status PositionedAppend(const Slice& /* data */, uint64_t /* offset */) { return Status::NotSupported(); } // Truncate is necessary to trim the file to the correct size // before closing. It is not always possible to keep track of the file // size due to whole pages writes. The behavior is undefined if called // with other writes to follow. virtual Status Truncate(uint64_t size) { return Status::OK(); } virtual Status Close() = 0; virtual Status Flush() = 0; virtual Status Sync() = 0; // sync data /* * Sync data and/or metadata as well. * By default, sync only data. * Override this method for environments where we need to sync * metadata as well. */ virtual Status Fsync() { return Sync(); } // true if Sync() and Fsync() are safe to call concurrently with Append() // and Flush(). virtual bool IsSyncThreadSafe() const { return false; } // Indicates the upper layers if the current WritableFile implementation // uses direct IO. virtual bool UseDirectIO() const { return false; } /* * Change the priority in rate limiter if rate limiting is enabled. * If rate limiting is not enabled, this call has no effect. */ virtual void SetIOPriority(Env::IOPriority pri) { io_priority_ = pri; } virtual Env::IOPriority GetIOPriority() { return io_priority_; } /* * Get the size of valid data in the file. */ virtual uint64_t GetFileSize() { return 0; } /* * Get and set the default pre-allocation block size for writes to * this file. If non-zero, then Allocate will be used to extend the * underlying storage of a file (generally via fallocate) if the Env * instance supports it. */ void SetPreallocationBlockSize(size_t size) { preallocation_block_size_ = size; } virtual void GetPreallocationStatus(size_t* block_size, size_t* last_allocated_block) { *last_allocated_block = last_preallocated_block_; *block_size = preallocation_block_size_; } // For documentation, refer to RandomAccessFile::GetUniqueId() virtual size_t GetUniqueId(char* id, size_t max_size) const { return 0; // Default implementation to prevent issues with backwards } // Remove any kind of caching of data from the offset to offset+length // of this file. If the length is 0, then it refers to the end of file. // If the system is not caching the file contents, then this is a noop. // This call has no effect on dirty pages in the cache. virtual Status InvalidateCache(size_t offset, size_t length) { return Status::NotSupported("InvalidateCache not supported."); } // Sync a file range with disk. // offset is the starting byte of the file range to be synchronized. // nbytes specifies the length of the range to be synchronized. // This asks the OS to initiate flushing the cached data to disk, // without waiting for completion. // Default implementation does nothing. virtual Status RangeSync(uint64_t offset, uint64_t nbytes) { return Status::OK(); } // PrepareWrite performs any necessary preparation for a write // before the write actually occurs. This allows for pre-allocation // of space on devices where it can result in less file // fragmentation and/or less waste from over-zealous filesystem // pre-allocation. void PrepareWrite(size_t offset, size_t len) { if (preallocation_block_size_ == 0) { return; } // If this write would cross one or more preallocation blocks, // determine what the last preallocation block necesessary to // cover this write would be and Allocate to that point. const auto block_size = preallocation_block_size_; size_t new_last_preallocated_block = (offset + len + block_size - 1) / block_size; if (new_last_preallocated_block > last_preallocated_block_) { size_t num_spanned_blocks = new_last_preallocated_block - last_preallocated_block_; Allocate(block_size * last_preallocated_block_, block_size * num_spanned_blocks); last_preallocated_block_ = new_last_preallocated_block; } } protected: /* * Pre-allocate space for a file. */ virtual Status Allocate(uint64_t offset, uint64_t len) { return Status::OK(); } size_t preallocation_block_size() { return preallocation_block_size_; } private: size_t last_preallocated_block_; size_t preallocation_block_size_; // No copying allowed WritableFile(const WritableFile&); void operator=(const WritableFile&); protected: friend class WritableFileWrapper; friend class WritableFileMirror; Env::IOPriority io_priority_; }; // Directory object represents collection of files and implements // filesystem operations that can be executed on directories. class Directory { public: virtual ~Directory() {} // Fsync directory. Can be called concurrently from multiple threads. virtual Status Fsync() = 0; }; enum InfoLogLevel : unsigned char { DEBUG_LEVEL = 0, INFO_LEVEL, WARN_LEVEL, ERROR_LEVEL, FATAL_LEVEL, HEADER_LEVEL, NUM_INFO_LOG_LEVELS, }; // An interface for writing log messages. class Logger { public: size_t kDoNotSupportGetLogFileSize = std::numeric_limits::max(); explicit Logger(const InfoLogLevel log_level = InfoLogLevel::INFO_LEVEL) : log_level_(log_level) {} virtual ~Logger(); // Write a header to the log file with the specified format // It is recommended that you log all header information at the start of the // application. But it is not enforced. virtual void LogHeader(const char* format, va_list ap) { // Default implementation does a simple INFO level log write. // Please override as per the logger class requirement. Logv(format, ap); } // Write an entry to the log file with the specified format. virtual void Logv(const char* format, va_list ap) = 0; // Write an entry to the log file with the specified log level // and format. Any log with level under the internal log level // of *this (see @SetInfoLogLevel and @GetInfoLogLevel) will not be // printed. virtual void Logv(const InfoLogLevel log_level, const char* format, va_list ap); virtual size_t GetLogFileSize() const { return kDoNotSupportGetLogFileSize; } // Flush to the OS buffers virtual void Flush() {} virtual InfoLogLevel GetInfoLogLevel() const { return log_level_; } virtual void SetInfoLogLevel(const InfoLogLevel log_level) { log_level_ = log_level; } private: // No copying allowed Logger(const Logger&); void operator=(const Logger&); InfoLogLevel log_level_; }; // Identifies a locked file. class FileLock { public: FileLock() { } virtual ~FileLock(); private: // No copying allowed FileLock(const FileLock&); void operator=(const FileLock&); }; extern void LogFlush(const shared_ptr& info_log); extern void Log(const InfoLogLevel log_level, const shared_ptr& info_log, const char* format, ...); // a set of log functions with different log levels. extern void Header(const shared_ptr& info_log, const char* format, ...); extern void Debug(const shared_ptr& info_log, const char* format, ...); extern void Info(const shared_ptr& info_log, const char* format, ...); extern void Warn(const shared_ptr& info_log, const char* format, ...); extern void Error(const shared_ptr& info_log, const char* format, ...); extern void Fatal(const shared_ptr& info_log, const char* format, ...); // Log the specified data to *info_log if info_log is non-nullptr. // The default info log level is InfoLogLevel::ERROR. extern void Log(const shared_ptr& info_log, const char* format, ...) # if defined(__GNUC__) || defined(__clang__) __attribute__((__format__ (__printf__, 2, 3))) # endif ; extern void LogFlush(Logger *info_log); extern void Log(const InfoLogLevel log_level, Logger* info_log, const char* format, ...); // The default info log level is InfoLogLevel::ERROR. extern void Log(Logger* info_log, const char* format, ...) # if defined(__GNUC__) || defined(__clang__) __attribute__((__format__ (__printf__, 2, 3))) # endif ; // a set of log functions with different log levels. extern void Header(Logger* info_log, const char* format, ...); extern void Debug(Logger* info_log, const char* format, ...); extern void Info(Logger* info_log, const char* format, ...); extern void Warn(Logger* info_log, const char* format, ...); extern void Error(Logger* info_log, const char* format, ...); extern void Fatal(Logger* info_log, const char* format, ...); // A utility routine: write "data" to the named file. extern Status WriteStringToFile(Env* env, const Slice& data, const std::string& fname, bool should_sync = false); // A utility routine: read contents of named file into *data extern Status ReadFileToString(Env* env, const std::string& fname, std::string* data); // An implementation of Env that forwards all calls to another Env. // May be useful to clients who wish to override just part of the // functionality of another Env. class EnvWrapper : public Env { public: // Initialize an EnvWrapper that delegates all calls to *t explicit EnvWrapper(Env* t) : target_(t) { } virtual ~EnvWrapper(); // Return the target to which this Env forwards all calls Env* target() const { return target_; } // The following text is boilerplate that forwards all methods to target() Status NewSequentialFile(const std::string& f, unique_ptr* r, const EnvOptions& options) override { return target_->NewSequentialFile(f, r, options); } Status NewRandomAccessFile(const std::string& f, unique_ptr* r, const EnvOptions& options) override { return target_->NewRandomAccessFile(f, r, options); } Status NewWritableFile(const std::string& f, unique_ptr* r, const EnvOptions& options) override { return target_->NewWritableFile(f, r, options); } Status ReuseWritableFile(const std::string& fname, const std::string& old_fname, unique_ptr* r, const EnvOptions& options) override { return target_->ReuseWritableFile(fname, old_fname, r, options); } virtual Status NewDirectory(const std::string& name, unique_ptr* result) override { return target_->NewDirectory(name, result); } Status FileExists(const std::string& f) override { return target_->FileExists(f); } Status GetChildren(const std::string& dir, std::vector* r) override { return target_->GetChildren(dir, r); } Status DeleteFile(const std::string& f) override { return target_->DeleteFile(f); } Status CreateDir(const std::string& d) override { return target_->CreateDir(d); } Status CreateDirIfMissing(const std::string& d) override { return target_->CreateDirIfMissing(d); } Status DeleteDir(const std::string& d) override { return target_->DeleteDir(d); } Status GetFileSize(const std::string& f, uint64_t* s) override { return target_->GetFileSize(f, s); } Status GetFileModificationTime(const std::string& fname, uint64_t* file_mtime) override { return target_->GetFileModificationTime(fname, file_mtime); } Status RenameFile(const std::string& s, const std::string& t) override { return target_->RenameFile(s, t); } Status LinkFile(const std::string& s, const std::string& t) override { return target_->LinkFile(s, t); } Status LockFile(const std::string& f, FileLock** l) override { return target_->LockFile(f, l); } Status UnlockFile(FileLock* l) override { return target_->UnlockFile(l); } void Schedule(void (*f)(void* arg), void* a, Priority pri, void* tag = nullptr, void (*u)(void* arg) = 0) override { return target_->Schedule(f, a, pri, tag, u); } int UnSchedule(void* tag, Priority pri) override { return target_->UnSchedule(tag, pri); } void StartThread(void (*f)(void*), void* a) override { return target_->StartThread(f, a); } void WaitForJoin() override { return target_->WaitForJoin(); } virtual unsigned int GetThreadPoolQueueLen( Priority pri = LOW) const override { return target_->GetThreadPoolQueueLen(pri); } virtual Status GetTestDirectory(std::string* path) override { return target_->GetTestDirectory(path); } virtual Status NewLogger(const std::string& fname, shared_ptr* result) override { return target_->NewLogger(fname, result); } uint64_t NowMicros() override { return target_->NowMicros(); } void SleepForMicroseconds(int micros) override { target_->SleepForMicroseconds(micros); } Status GetHostName(char* name, uint64_t len) override { return target_->GetHostName(name, len); } Status GetCurrentTime(int64_t* unix_time) override { return target_->GetCurrentTime(unix_time); } Status GetAbsolutePath(const std::string& db_path, std::string* output_path) override { return target_->GetAbsolutePath(db_path, output_path); } void SetBackgroundThreads(int num, Priority pri) override { return target_->SetBackgroundThreads(num, pri); } void IncBackgroundThreadsIfNeeded(int num, Priority pri) override { return target_->IncBackgroundThreadsIfNeeded(num, pri); } void LowerThreadPoolIOPriority(Priority pool = LOW) override { target_->LowerThreadPoolIOPriority(pool); } std::string TimeToString(uint64_t time) override { return target_->TimeToString(time); } Status GetThreadList(std::vector* thread_list) override { return target_->GetThreadList(thread_list); } ThreadStatusUpdater* GetThreadStatusUpdater() const override { return target_->GetThreadStatusUpdater(); } uint64_t GetThreadID() const override { return target_->GetThreadID(); } private: Env* target_; }; // An implementation of WritableFile that forwards all calls to another // WritableFile. May be useful to clients who wish to override just part of the // functionality of another WritableFile. // It's declared as friend of WritableFile to allow forwarding calls to // protected virtual methods. class WritableFileWrapper : public WritableFile { public: explicit WritableFileWrapper(WritableFile* t) : target_(t) { } Status Append(const Slice& data) override { return target_->Append(data); } Status PositionedAppend(const Slice& data, uint64_t offset) override { return target_->PositionedAppend(data, offset); } Status Truncate(uint64_t size) override { return target_->Truncate(size); } Status Close() override { return target_->Close(); } Status Flush() override { return target_->Flush(); } Status Sync() override { return target_->Sync(); } Status Fsync() override { return target_->Fsync(); } bool IsSyncThreadSafe() const override { return target_->IsSyncThreadSafe(); } void SetIOPriority(Env::IOPriority pri) override { target_->SetIOPriority(pri); } Env::IOPriority GetIOPriority() override { return target_->GetIOPriority(); } uint64_t GetFileSize() override { return target_->GetFileSize(); } void GetPreallocationStatus(size_t* block_size, size_t* last_allocated_block) override { target_->GetPreallocationStatus(block_size, last_allocated_block); } size_t GetUniqueId(char* id, size_t max_size) const override { return target_->GetUniqueId(id, max_size); } Status InvalidateCache(size_t offset, size_t length) override { return target_->InvalidateCache(offset, length); } protected: Status Allocate(uint64_t offset, uint64_t len) override { return target_->Allocate(offset, len); } Status RangeSync(uint64_t offset, uint64_t nbytes) override { return target_->RangeSync(offset, nbytes); } private: WritableFile* target_; }; // Returns a new environment that stores its data in memory and delegates // all non-file-storage tasks to base_env. The caller must delete the result // when it is no longer needed. // *base_env must remain live while the result is in use. Env* NewMemEnv(Env* base_env); // Returns a new environment that is used for HDFS environment. // This is a factory method for HdfsEnv declared in hdfs/env_hdfs.h Status NewHdfsEnv(Env** hdfs_env, const std::string& fsname); } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_ENV_H_ rocksdb-4.5.1/include/rocksdb/experimental.h000066400000000000000000000022301267530634000210730ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include "rocksdb/db.h" #include "rocksdb/status.h" namespace rocksdb { namespace experimental { // Supported only for Leveled compaction Status SuggestCompactRange(DB* db, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end); Status SuggestCompactRange(DB* db, const Slice* begin, const Slice* end); // Move all L0 files to target_level skipping compaction. // This operation succeeds only if the files in L0 have disjoint ranges; this // is guaranteed to happen, for instance, if keys are inserted in sorted // order. Furthermore, all levels between 1 and target_level must be empty. // If any of the above condition is violated, InvalidArgument will be // returned. Status PromoteL0(DB* db, ColumnFamilyHandle* column_family, int target_level = 1); } // namespace experimental } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/filter_policy.h000066400000000000000000000126251267530634000212530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2012 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // A database can be configured with a custom FilterPolicy object. // This object is responsible for creating a small filter from a set // of keys. These filters are stored in rocksdb and are consulted // automatically by rocksdb to decide whether or not to read some // information from disk. In many cases, a filter can cut down the // number of disk seeks form a handful to a single disk seek per // DB::Get() call. // // Most people will want to use the builtin bloom filter support (see // NewBloomFilterPolicy() below). #ifndef STORAGE_ROCKSDB_INCLUDE_FILTER_POLICY_H_ #define STORAGE_ROCKSDB_INCLUDE_FILTER_POLICY_H_ #include #include namespace rocksdb { class Slice; // A class that takes a bunch of keys, then generates filter class FilterBitsBuilder { public: virtual ~FilterBitsBuilder() {} // Add Key to filter, you could use any way to store the key. // Such as: storing hashes or original keys // Keys are in sorted order and duplicated keys are possible. virtual void AddKey(const Slice& key) = 0; // Generate the filter using the keys that are added // The return value of this function would be the filter bits, // The ownership of actual data is set to buf virtual Slice Finish(std::unique_ptr* buf) = 0; }; // A class that checks if a key can be in filter // It should be initialized by Slice generated by BitsBuilder class FilterBitsReader { public: virtual ~FilterBitsReader() {} // Check if the entry match the bits in filter virtual bool MayMatch(const Slice& entry) = 0; }; // We add a new format of filter block called full filter block // This new interface gives you more space of customization // // For the full filter block, you can plug in your version by implement // the FilterBitsBuilder and FilterBitsReader // // There are two sets of interface in FilterPolicy // Set 1: CreateFilter, KeyMayMatch: used for blockbased filter // Set 2: GetFilterBitsBuilder, GetFilterBitsReader, they are used for // full filter. // Set 1 MUST be implemented correctly, Set 2 is optional // RocksDB would first try using functions in Set 2. if they return nullptr, // it would use Set 1 instead. // You can choose filter type in NewBloomFilterPolicy class FilterPolicy { public: virtual ~FilterPolicy(); // Return the name of this policy. Note that if the filter encoding // changes in an incompatible way, the name returned by this method // must be changed. Otherwise, old incompatible filters may be // passed to methods of this type. virtual const char* Name() const = 0; // keys[0,n-1] contains a list of keys (potentially with duplicates) // that are ordered according to the user supplied comparator. // Append a filter that summarizes keys[0,n-1] to *dst. // // Warning: do not change the initial contents of *dst. Instead, // append the newly constructed filter to *dst. virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const = 0; // "filter" contains the data appended by a preceding call to // CreateFilter() on this class. This method must return true if // the key was in the list of keys passed to CreateFilter(). // This method may return true or false if the key was not on the // list, but it should aim to return false with a high probability. virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const = 0; // Get the FilterBitsBuilder, which is ONLY used for full filter block // It contains interface to take individual key, then generate filter virtual FilterBitsBuilder* GetFilterBitsBuilder() const { return nullptr; } // Get the FilterBitsReader, which is ONLY used for full filter block // It contains interface to tell if key can be in filter // The input slice should NOT be deleted by FilterPolicy virtual FilterBitsReader* GetFilterBitsReader(const Slice& contents) const { return nullptr; } }; // Return a new filter policy that uses a bloom filter with approximately // the specified number of bits per key. // // bits_per_key: bits per key in bloom filter. A good value for bits_per_key // is 10, which yields a filter with ~ 1% false positive rate. // use_block_based_builder: use block based filter rather than full fiter. // If you want to builder full filter, it needs to be set to false. // // Callers must delete the result after any database that is using the // result has been closed. // // Note: if you are using a custom comparator that ignores some parts // of the keys being compared, you must not use NewBloomFilterPolicy() // and must provide your own FilterPolicy that also ignores the // corresponding parts of the keys. For example, if the comparator // ignores trailing spaces, it would be incorrect to use a // FilterPolicy (like NewBloomFilterPolicy) that does not ignore // trailing spaces in keys. extern const FilterPolicy* NewBloomFilterPolicy(int bits_per_key, bool use_block_based_builder = true); } #endif // STORAGE_ROCKSDB_INCLUDE_FILTER_POLICY_H_ rocksdb-4.5.1/include/rocksdb/flush_block_policy.h000066400000000000000000000035121267530634000222540ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include "rocksdb/table.h" namespace rocksdb { class Slice; class BlockBuilder; struct Options; // FlushBlockPolicy provides a configurable way to determine when to flush a // block in the block based tables, class FlushBlockPolicy { public: // Keep track of the key/value sequences and return the boolean value to // determine if table builder should flush current data block. virtual bool Update(const Slice& key, const Slice& value) = 0; virtual ~FlushBlockPolicy() { } }; class FlushBlockPolicyFactory { public: // Return the name of the flush block policy. virtual const char* Name() const = 0; // Return a new block flush policy that flushes data blocks by data size. // FlushBlockPolicy may need to access the metadata of the data block // builder to determine when to flush the blocks. // // Callers must delete the result after any database that is using the // result has been closed. virtual FlushBlockPolicy* NewFlushBlockPolicy( const BlockBasedTableOptions& table_options, const BlockBuilder& data_block_builder) const = 0; virtual ~FlushBlockPolicyFactory() { } }; class FlushBlockBySizePolicyFactory : public FlushBlockPolicyFactory { public: FlushBlockBySizePolicyFactory() {} virtual const char* Name() const override { return "FlushBlockBySizePolicyFactory"; } virtual FlushBlockPolicy* NewFlushBlockPolicy( const BlockBasedTableOptions& table_options, const BlockBuilder& data_block_builder) const override; }; } // rocksdb rocksdb-4.5.1/include/rocksdb/immutable_options.h000066400000000000000000000055501267530634000221400ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "rocksdb/options.h" namespace rocksdb { // ImmutableCFOptions is a data struct used by RocksDB internal. It contains a // subset of Options that should not be changed during the entire lifetime // of DB. You shouldn't need to access this data structure unless you are // implementing a new TableFactory. Raw pointers defined in this struct do // not have ownership to the data they point to. Options contains shared_ptr // to these data. struct ImmutableCFOptions { explicit ImmutableCFOptions(const Options& options); CompactionStyle compaction_style; CompactionOptionsUniversal compaction_options_universal; CompactionOptionsFIFO compaction_options_fifo; const SliceTransform* prefix_extractor; const Comparator* comparator; MergeOperator* merge_operator; const CompactionFilter* compaction_filter; CompactionFilterFactory* compaction_filter_factory; bool inplace_update_support; UpdateStatus (*inplace_callback)(char* existing_value, uint32_t* existing_value_size, Slice delta_value, std::string* merged_value); Logger* info_log; Statistics* statistics; InfoLogLevel info_log_level; Env* env; uint64_t delayed_write_rate; // Allow the OS to mmap file for reading sst tables. Default: false bool allow_mmap_reads; // Allow the OS to mmap file for writing. Default: false bool allow_mmap_writes; std::vector db_paths; MemTableRepFactory* memtable_factory; TableFactory* table_factory; Options::TablePropertiesCollectorFactories table_properties_collector_factories; bool advise_random_on_open; // This options is required by PlainTableReader. May need to move it // to PlainTalbeOptions just like bloom_bits_per_key uint32_t bloom_locality; bool purge_redundant_kvs_while_flush; uint32_t min_partial_merge_operands; bool disable_data_sync; bool use_fsync; CompressionType compression; std::vector compression_per_level; CompressionOptions compression_opts; bool level_compaction_dynamic_level_bytes; Options::AccessHint access_hint_on_compaction_start; bool new_table_reader_for_compaction_inputs; size_t compaction_readahead_size; int num_levels; bool optimize_filters_for_hits; // A vector of EventListeners which call-back functions will be called // when specific RocksDB event happens. std::vector> listeners; std::shared_ptr row_cache; }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/iostats_context.h000066400000000000000000000030421267530634000216320ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "rocksdb/perf_level.h" // A thread local context for gathering io-stats efficiently and transparently. // Use SetPerfLevel(PerfLevel::kEnableTime) to enable time stats. namespace rocksdb { struct IOStatsContext { // reset all io-stats counter to zero void Reset(); std::string ToString() const; // the thread pool id uint64_t thread_pool_id; // number of bytes that has been written. uint64_t bytes_written; // number of bytes that has been read. uint64_t bytes_read; // time spent in open() and fopen(). uint64_t open_nanos; // time spent in fallocate(). uint64_t allocate_nanos; // time spent in write() and pwrite(). uint64_t write_nanos; // time spent in read() and pread() uint64_t read_nanos; // time spent in sync_file_range(). uint64_t range_sync_nanos; // time spent in fsync uint64_t fsync_nanos; // time spent in preparing write (fallocate etc). uint64_t prepare_write_nanos; // time spent in Logger::Logv(). uint64_t logger_nanos; }; #ifndef IOS_CROSS_COMPILE # ifdef _WIN32 extern __declspec(thread) IOStatsContext iostats_context; # else extern __thread IOStatsContext iostats_context; # endif #endif // IOS_CROSS_COMPILE } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/iterator.h000066400000000000000000000105511267530634000202340ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // An iterator yields a sequence of key/value pairs from a source. // The following class defines the interface. Multiple implementations // are provided by this library. In particular, iterators are provided // to access the contents of a Table or a DB. // // Multiple threads can invoke const methods on an Iterator without // external synchronization, but if any of the threads may call a // non-const method, all threads accessing the same Iterator must use // external synchronization. #ifndef STORAGE_ROCKSDB_INCLUDE_ITERATOR_H_ #define STORAGE_ROCKSDB_INCLUDE_ITERATOR_H_ #include "rocksdb/slice.h" #include "rocksdb/status.h" namespace rocksdb { class Cleanable { public: Cleanable(); ~Cleanable(); // Clients are allowed to register function/arg1/arg2 triples that // will be invoked when this iterator is destroyed. // // Note that unlike all of the preceding methods, this method is // not abstract and therefore clients should not override it. typedef void (*CleanupFunction)(void* arg1, void* arg2); void RegisterCleanup(CleanupFunction function, void* arg1, void* arg2); protected: struct Cleanup { CleanupFunction function; void* arg1; void* arg2; Cleanup* next; }; Cleanup cleanup_; }; class Iterator : public Cleanable { public: Iterator() {} virtual ~Iterator() {} // An iterator is either positioned at a key/value pair, or // not valid. This method returns true iff the iterator is valid. virtual bool Valid() const = 0; // Position at the first key in the source. The iterator is Valid() // after this call iff the source is not empty. virtual void SeekToFirst() = 0; // Position at the last key in the source. The iterator is // Valid() after this call iff the source is not empty. virtual void SeekToLast() = 0; // Position at the first key in the source that at or past target // The iterator is Valid() after this call iff the source contains // an entry that comes at or past target. virtual void Seek(const Slice& target) = 0; // Moves to the next entry in the source. After this call, Valid() is // true iff the iterator was not positioned at the last entry in the source. // REQUIRES: Valid() virtual void Next() = 0; // Moves to the previous entry in the source. After this call, Valid() is // true iff the iterator was not positioned at the first entry in source. // REQUIRES: Valid() virtual void Prev() = 0; // Return the key for the current entry. The underlying storage for // the returned slice is valid only until the next modification of // the iterator. // REQUIRES: Valid() virtual Slice key() const = 0; // Return the value for the current entry. The underlying storage for // the returned slice is valid only until the next modification of // the iterator. // REQUIRES: !AtEnd() && !AtStart() virtual Slice value() const = 0; // If an error has occurred, return it. Else return an ok status. // If non-blocking IO is requested and this operation cannot be // satisfied without doing some IO, then this returns Status::Incomplete(). virtual Status status() const = 0; // If true, this means that the Slice returned by key() is valid as long // as the iterator is not deleted and ReleasePinnedData() is not called. // // IsKeyPinned() is guaranteed to always return true if // - Iterator created with ReadOptions::pin_data = true // - DB tables were created with BlockBasedTableOptions::use_delta_encoding // set to false. virtual bool IsKeyPinned() const { return false; } private: // No copying allowed Iterator(const Iterator&); void operator=(const Iterator&); }; // Return an empty iterator (yields nothing). extern Iterator* NewEmptyIterator(); // Return an empty iterator with the specified status. extern Iterator* NewErrorIterator(const Status& status); } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_ITERATOR_H_ rocksdb-4.5.1/include/rocksdb/ldb_tool.h000066400000000000000000000022561267530634000202040ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #pragma once #include #include #include "rocksdb/db.h" #include "rocksdb/options.h" namespace rocksdb { // An interface for converting a slice to a readable string class SliceFormatter { public: virtual ~SliceFormatter() {} virtual std::string Format(const Slice& s) const = 0; }; // Options for customizing ldb tool (beyond the DB Options) struct LDBOptions { // Create LDBOptions with default values for all fields LDBOptions(); // Key formatter that converts a slice to a readable string. // Default: Slice::ToString() std::shared_ptr key_formatter; }; class LDBTool { public: void Run( int argc, char** argv, Options db_options = Options(), const LDBOptions& ldb_options = LDBOptions(), const std::vector* column_families = nullptr); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/listener.h000066400000000000000000000214241267530634000202310ustar00rootroot00000000000000// Copyright (c) 2014 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include "rocksdb/compaction_job_stats.h" #include "rocksdb/status.h" #include "rocksdb/table_properties.h" namespace rocksdb { typedef std::unordered_map> TablePropertiesCollection; class DB; class Status; struct CompactionJobStats; struct TableFileCreationInfo { TableFileCreationInfo() = default; explicit TableFileCreationInfo(TableProperties&& prop) : table_properties(prop) {} // the name of the database where the file was created std::string db_name; // the name of the column family where the file was created. std::string cf_name; // the path to the created file. std::string file_path; // the size of the file. uint64_t file_size; // the id of the job (which could be flush or compaction) that // created the file. int job_id; // Detailed properties of the created file. TableProperties table_properties; }; enum class CompactionReason { kUnknown, // [Level] number of L0 files > level0_file_num_compaction_trigger kLevelL0FilesNum, // [Level] total size of level > MaxBytesForLevel() kLevelMaxLevelSize, // [Universal] Compacting for size amplification kUniversalSizeAmplification, // [Universal] Compacting for size ratio kUniversalSizeRatio, // [Universal] number of sorted runs > level0_file_num_compaction_trigger kUniversalSortedRunNum, // [FIFO] total size > max_table_files_size kFIFOMaxSize, // Manual compaction kManualCompaction, // DB::SuggestCompactRange() marked files for compaction kFilesMarkedForCompaction, }; #ifndef ROCKSDB_LITE struct TableFileDeletionInfo { // The name of the database where the file was deleted. std::string db_name; // The path to the deleted file. std::string file_path; // The id of the job which deleted the file. int job_id; // The status indicating whether the deletion was successful or not. Status status; }; struct FlushJobInfo { // the name of the column family std::string cf_name; // the path to the newly created file std::string file_path; // the id of the thread that completed this flush job. uint64_t thread_id; // the job id, which is unique in the same thread. int job_id; // If true, then rocksdb is currently slowing-down all writes to prevent // creating too many Level 0 files as compaction seems not able to // catch up the write request speed. This indicates that there are // too many files in Level 0. bool triggered_writes_slowdown; // If true, then rocksdb is currently blocking any writes to prevent // creating more L0 files. This indicates that there are too many // files in level 0. Compactions should try to compact L0 files down // to lower levels as soon as possible. bool triggered_writes_stop; // The smallest sequence number in the newly created file SequenceNumber smallest_seqno; // The largest sequence number in the newly created file SequenceNumber largest_seqno; // Table properties of the table being flushed TableProperties table_properties; }; struct CompactionJobInfo { CompactionJobInfo() = default; explicit CompactionJobInfo(const CompactionJobStats& _stats) : stats(_stats) {} // the name of the column family where the compaction happened. std::string cf_name; // the status indicating whether the compaction was successful or not. Status status; // the id of the thread that completed this compaction job. uint64_t thread_id; // the job id, which is unique in the same thread. int job_id; // the smallest input level of the compaction. int base_input_level; // the output level of the compaction. int output_level; // the names of the compaction input files. std::vector input_files; // the names of the compaction output files. std::vector output_files; // Table properties for input and output tables. // The map is keyed by values from input_files and output_files. TablePropertiesCollection table_properties; // Reason to run the compaction CompactionReason compaction_reason; // If non-null, this variable stores detailed information // about this compaction. CompactionJobStats stats; }; // EventListener class contains a set of call-back functions that will // be called when specific RocksDB event happens such as flush. It can // be used as a building block for developing custom features such as // stats-collector or external compaction algorithm. // // Note that call-back functions should not run for an extended period of // time before the function returns, otherwise RocksDB may be blocked. // For example, it is not suggested to do DB::CompactFiles() (as it may // run for a long while) or issue many of DB::Put() (as Put may be blocked // in certain cases) in the same thread in the EventListener callback. // However, doing DB::CompactFiles() and DB::Put() in another thread is // considered safe. // // [Threading] All EventListener callback will be called using the // actual thread that involves in that specific event. For example, it // is the RocksDB background flush thread that does the actual flush to // call EventListener::OnFlushCompleted(). // // [Locking] All EventListener callbacks are designed to be called without // the current thread holding any DB mutex. This is to prevent potential // deadlock and performance issue when using EventListener callback // in a complex way. However, all EventListener call-back functions // should not run for an extended period of time before the function // returns, otherwise RocksDB may be blocked. For example, it is not // suggested to do DB::CompactFiles() (as it may run for a long while) // or issue many of DB::Put() (as Put may be blocked in certain cases) // in the same thread in the EventListener callback. However, doing // DB::CompactFiles() and DB::Put() in a thread other than the // EventListener callback thread is considered safe. class EventListener { public: // A call-back function to RocksDB which will be called whenever a // registered RocksDB flushes a file. The default implementation is // no-op. // // Note that the this function must be implemented in a way such that // it should not run for an extended period of time before the function // returns. Otherwise, RocksDB may be blocked. virtual void OnFlushCompleted(DB* /*db*/, const FlushJobInfo& /*flush_job_info*/) {} // A call-back function for RocksDB which will be called whenever // a SST file is deleted. Different from OnCompactionCompleted and // OnFlushCompleted, this call-back is designed for external logging // service and thus only provide string parameters instead // of a pointer to DB. Applications that build logic basic based // on file creations and deletions is suggested to implement // OnFlushCompleted and OnCompactionCompleted. // // Note that if applications would like to use the passed reference // outside this function call, they should make copies from the // returned value. virtual void OnTableFileDeleted(const TableFileDeletionInfo& /*info*/) {} // A call-back function for RocksDB which will be called whenever // a registered RocksDB compacts a file. The default implementation // is a no-op. // // Note that this function must be implemented in a way such that // it should not run for an extended period of time before the function // returns. Otherwise, RocksDB may be blocked. // // @param db a pointer to the rocksdb instance which just compacted // a file. // @param ci a reference to a CompactionJobInfo struct. 'ci' is released // after this function is returned, and must be copied if it is needed // outside of this function. virtual void OnCompactionCompleted(DB* /*db*/, const CompactionJobInfo& /*ci*/) {} // A call-back function for RocksDB which will be called whenever // a SST file is created. Different from OnCompactionCompleted and // OnFlushCompleted, this call-back is designed for external logging // service and thus only provide string parameters instead // of a pointer to DB. Applications that build logic basic based // on file creations and deletions is suggested to implement // OnFlushCompleted and OnCompactionCompleted. // // Note that if applications would like to use the passed reference // outside this function call, they should make copies from these // returned value. virtual void OnTableFileCreated(const TableFileCreationInfo& /*info*/) {} virtual ~EventListener() {} }; #else class EventListener { }; #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/memtablerep.h000066400000000000000000000337231267530634000207060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file contains the interface that must be implemented by any collection // to be used as the backing store for a MemTable. Such a collection must // satisfy the following properties: // (1) It does not store duplicate items. // (2) It uses MemTableRep::KeyComparator to compare items for iteration and // equality. // (3) It can be accessed concurrently by multiple readers and can support // during reads. However, it needn't support multiple concurrent writes. // (4) Items are never deleted. // The liberal use of assertions is encouraged to enforce (1). // // The factory will be passed an MemTableAllocator object when a new MemTableRep // is requested. // // Users can implement their own memtable representations. We include three // types built in: // - SkipListRep: This is the default; it is backed by a skip list. // - HashSkipListRep: The memtable rep that is best used for keys that are // structured like "prefix:suffix" where iteration within a prefix is // common and iteration across different prefixes is rare. It is backed by // a hash map where each bucket is a skip list. // - VectorRep: This is backed by an unordered std::vector. On iteration, the // vector is sorted. It is intelligent about sorting; once the MarkReadOnly() // has been called, the vector will only be sorted once. It is optimized for // random-write-heavy workloads. // // The last four implementations are designed for situations in which // iteration over the entire collection is rare since doing so requires all the // keys to be copied into a sorted data structure. #pragma once #include #include #include #include namespace rocksdb { class Arena; class MemTableAllocator; class LookupKey; class Slice; class SliceTransform; class Logger; typedef void* KeyHandle; class MemTableRep { public: // KeyComparator provides a means to compare keys, which are internal keys // concatenated with values. class KeyComparator { public: // Compare a and b. Return a negative value if a is less than b, 0 if they // are equal, and a positive value if a is greater than b virtual int operator()(const char* prefix_len_key1, const char* prefix_len_key2) const = 0; virtual int operator()(const char* prefix_len_key, const Slice& key) const = 0; virtual ~KeyComparator() { } }; explicit MemTableRep(MemTableAllocator* allocator) : allocator_(allocator) {} // Allocate a buf of len size for storing key. The idea is that a // specific memtable representation knows its underlying data structure // better. By allowing it to allocate memory, it can possibly put // correlated stuff in consecutive memory area to make processor // prefetching more efficient. virtual KeyHandle Allocate(const size_t len, char** buf); // Insert key into the collection. (The caller will pack key and value into a // single buffer and pass that in as the parameter to Insert). // REQUIRES: nothing that compares equal to key is currently in the // collection, and no concurrent modifications to the table in progress virtual void Insert(KeyHandle handle) = 0; // Like Insert(handle), but may be called concurrent with other calls // to InsertConcurrently for other handles virtual void InsertConcurrently(KeyHandle handle) { #ifndef ROCKSDB_LITE throw std::runtime_error("concurrent insert not supported"); #else abort(); #endif } // Returns true iff an entry that compares equal to key is in the collection. virtual bool Contains(const char* key) const = 0; // Notify this table rep that it will no longer be added to. By default, // does nothing. After MarkReadOnly() is called, this table rep will // not be written to (ie No more calls to Allocate(), Insert(), // or any writes done directly to entries accessed through the iterator.) virtual void MarkReadOnly() { } // Look up key from the mem table, since the first key in the mem table whose // user_key matches the one given k, call the function callback_func(), with // callback_args directly forwarded as the first parameter, and the mem table // key as the second parameter. If the return value is false, then terminates. // Otherwise, go through the next key. // // It's safe for Get() to terminate after having finished all the potential // key for the k.user_key(), or not. // // Default: // Get() function with a default value of dynamically construct an iterator, // seek and call the call back function. virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)); virtual uint64_t ApproximateNumEntries(const Slice& start_ikey, const Slice& end_key) { return 0; } // Report an approximation of how much memory has been used other than memory // that was allocated through the allocator. Safe to call from any thread. virtual size_t ApproximateMemoryUsage() = 0; virtual ~MemTableRep() { } // Iteration over the contents of a skip collection class Iterator { public: // Initialize an iterator over the specified collection. // The returned iterator is not valid. // explicit Iterator(const MemTableRep* collection); virtual ~Iterator() {} // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const = 0; // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const = 0; // Advances to the next position. // REQUIRES: Valid() virtual void Next() = 0; // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() = 0; // Advance to the first entry with a key >= target virtual void Seek(const Slice& internal_key, const char* memtable_key) = 0; // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() = 0; // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() = 0; }; // Return an iterator over the keys in this representation. // arena: If not null, the arena needs to be used to allocate the Iterator. // When destroying the iterator, the caller will not call "delete" // but Iterator::~Iterator() directly. The destructor needs to destroy // all the states but those allocated in arena. virtual Iterator* GetIterator(Arena* arena = nullptr) = 0; // Return an iterator that has a special Seek semantics. The result of // a Seek might only include keys with the same prefix as the target key. // arena: If not null, the arena is used to allocate the Iterator. // When destroying the iterator, the caller will not call "delete" // but Iterator::~Iterator() directly. The destructor needs to destroy // all the states but those allocated in arena. virtual Iterator* GetDynamicPrefixIterator(Arena* arena = nullptr) { return GetIterator(arena); } // Return true if the current MemTableRep supports merge operator. // Default: true virtual bool IsMergeOperatorSupported() const { return true; } // Return true if the current MemTableRep supports snapshot // Default: true virtual bool IsSnapshotSupported() const { return true; } protected: // When *key is an internal key concatenated with the value, returns the // user key. virtual Slice UserKey(const char* key) const; MemTableAllocator* allocator_; }; // This is the base class for all factories that are used by RocksDB to create // new MemTableRep objects class MemTableRepFactory { public: virtual ~MemTableRepFactory() {} virtual MemTableRep* CreateMemTableRep(const MemTableRep::KeyComparator&, MemTableAllocator*, const SliceTransform*, Logger* logger) = 0; virtual const char* Name() const = 0; // Return true if the current MemTableRep supports concurrent inserts // Default: false virtual bool IsInsertConcurrentlySupported() const { return false; } }; // This uses a skip list to store keys. It is the default. // // Parameters: // lookahead: If non-zero, each iterator's seek operation will start the // search from the previously visited record (doing at most 'lookahead' // steps). This is an optimization for the access pattern including many // seeks with consecutive keys. class SkipListFactory : public MemTableRepFactory { public: explicit SkipListFactory(size_t lookahead = 0) : lookahead_(lookahead) {} virtual MemTableRep* CreateMemTableRep(const MemTableRep::KeyComparator&, MemTableAllocator*, const SliceTransform*, Logger* logger) override; virtual const char* Name() const override { return "SkipListFactory"; } bool IsInsertConcurrentlySupported() const override { return true; } private: const size_t lookahead_; }; #ifndef ROCKSDB_LITE // This creates MemTableReps that are backed by an std::vector. On iteration, // the vector is sorted. This is useful for workloads where iteration is very // rare and writes are generally not issued after reads begin. // // Parameters: // count: Passed to the constructor of the underlying std::vector of each // VectorRep. On initialization, the underlying array will be at least count // bytes reserved for usage. class VectorRepFactory : public MemTableRepFactory { const size_t count_; public: explicit VectorRepFactory(size_t count = 0) : count_(count) { } virtual MemTableRep* CreateMemTableRep(const MemTableRep::KeyComparator&, MemTableAllocator*, const SliceTransform*, Logger* logger) override; virtual const char* Name() const override { return "VectorRepFactory"; } }; // This class contains a fixed array of buckets, each // pointing to a skiplist (null if the bucket is empty). // bucket_count: number of fixed array buckets // skiplist_height: the max height of the skiplist // skiplist_branching_factor: probabilistic size ratio between adjacent // link lists in the skiplist extern MemTableRepFactory* NewHashSkipListRepFactory( size_t bucket_count = 1000000, int32_t skiplist_height = 4, int32_t skiplist_branching_factor = 4 ); // The factory is to create memtables based on a hash table: // it contains a fixed array of buckets, each pointing to either a linked list // or a skip list if number of entries inside the bucket exceeds // threshold_use_skiplist. // @bucket_count: number of fixed array buckets // @huge_page_tlb_size: if <=0, allocate the hash table bytes from malloc. // Otherwise from huge page TLB. The user needs to reserve // huge pages for it to be allocated, like: // sysctl -w vm.nr_hugepages=20 // See linux doc Documentation/vm/hugetlbpage.txt // @bucket_entries_logging_threshold: if number of entries in one bucket // exceeds this number, log about it. // @if_log_bucket_dist_when_flash: if true, log distribution of number of // entries when flushing. // @threshold_use_skiplist: a bucket switches to skip list if number of // entries exceed this parameter. extern MemTableRepFactory* NewHashLinkListRepFactory( size_t bucket_count = 50000, size_t huge_page_tlb_size = 0, int bucket_entries_logging_threshold = 4096, bool if_log_bucket_dist_when_flash = true, uint32_t threshold_use_skiplist = 256); // This factory creates a cuckoo-hashing based mem-table representation. // Cuckoo-hash is a closed-hash strategy, in which all key/value pairs // are stored in the bucket array itself intead of in some data structures // external to the bucket array. In addition, each key in cuckoo hash // has a constant number of possible buckets in the bucket array. These // two properties together makes cuckoo hash more memory efficient and // a constant worst-case read time. Cuckoo hash is best suitable for // point-lookup workload. // // When inserting a key / value, it first checks whether one of its possible // buckets is empty. If so, the key / value will be inserted to that vacant // bucket. Otherwise, one of the keys originally stored in one of these // possible buckets will be "kicked out" and move to one of its possible // buckets (and possibly kicks out another victim.) In the current // implementation, such "kick-out" path is bounded. If it cannot find a // "kick-out" path for a specific key, this key will be stored in a backup // structure, and the current memtable to be forced to immutable. // // Note that currently this mem-table representation does not support // snapshot (i.e., it only queries latest state) and iterators. In addition, // MultiGet operation might also lose its atomicity due to the lack of // snapshot support. // // Parameters: // write_buffer_size: the write buffer size in bytes. // average_data_size: the average size of key + value in bytes. This value // together with write_buffer_size will be used to compute the number // of buckets. // hash_function_count: the number of hash functions that will be used by // the cuckoo-hash. The number also equals to the number of possible // buckets each key will have. extern MemTableRepFactory* NewHashCuckooRepFactory( size_t write_buffer_size, size_t average_data_size = 64, unsigned int hash_function_count = 4); #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/merge_operator.h000066400000000000000000000215711267530634000214210ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef STORAGE_ROCKSDB_INCLUDE_MERGE_OPERATOR_H_ #define STORAGE_ROCKSDB_INCLUDE_MERGE_OPERATOR_H_ #include #include #include #include "rocksdb/slice.h" namespace rocksdb { class Slice; class Logger; // The Merge Operator // // Essentially, a MergeOperator specifies the SEMANTICS of a merge, which only // client knows. It could be numeric addition, list append, string // concatenation, edit data structure, ... , anything. // The library, on the other hand, is concerned with the exercise of this // interface, at the right time (during get, iteration, compaction...) // // To use merge, the client needs to provide an object implementing one of // the following interfaces: // a) AssociativeMergeOperator - for most simple semantics (always take // two values, and merge them into one value, which is then put back // into rocksdb); numeric addition and string concatenation are examples; // // b) MergeOperator - the generic class for all the more abstract / complex // operations; one method (FullMerge) to merge a Put/Delete value with a // merge operand; and another method (PartialMerge) that merges multiple // operands together. this is especially useful if your key values have // complex structures but you would still like to support client-specific // incremental updates. // // AssociativeMergeOperator is simpler to implement. MergeOperator is simply // more powerful. // // Refer to rocksdb-merge wiki for more details and example implementations. // class MergeOperator { public: virtual ~MergeOperator() {} // Gives the client a way to express the read -> modify -> write semantics // key: (IN) The key that's associated with this merge operation. // Client could multiplex the merge operator based on it // if the key space is partitioned and different subspaces // refer to different types of data which have different // merge operation semantics // existing: (IN) null indicates that the key does not exist before this op // operand_list:(IN) the sequence of merge operations to apply, front() first. // new_value:(OUT) Client is responsible for filling the merge result here. // The string that new_value is pointing to will be empty. // logger: (IN) Client could use this to log errors during merge. // // Return true on success. // All values passed in will be client-specific values. So if this method // returns false, it is because client specified bad data or there was // internal corruption. This will be treated as an error by the library. // // Also make use of the *logger for error messages. virtual bool FullMerge(const Slice& key, const Slice* existing_value, const std::deque& operand_list, std::string* new_value, Logger* logger) const = 0; // This function performs merge(left_op, right_op) // when both the operands are themselves merge operation types // that you would have passed to a DB::Merge() call in the same order // (i.e.: DB::Merge(key,left_op), followed by DB::Merge(key,right_op)). // // PartialMerge should combine them into a single merge operation that is // saved into *new_value, and then it should return true. // *new_value should be constructed such that a call to // DB::Merge(key, *new_value) would yield the same result as a call // to DB::Merge(key, left_op) followed by DB::Merge(key, right_op). // // The string that new_value is pointing to will be empty. // // The default implementation of PartialMergeMulti will use this function // as a helper, for backward compatibility. Any successor class of // MergeOperator should either implement PartialMerge or PartialMergeMulti, // although implementing PartialMergeMulti is suggested as it is in general // more effective to merge multiple operands at a time instead of two // operands at a time. // // If it is impossible or infeasible to combine the two operations, // leave new_value unchanged and return false. The library will // internally keep track of the operations, and apply them in the // correct order once a base-value (a Put/Delete/End-of-Database) is seen. // // TODO: Presently there is no way to differentiate between error/corruption // and simply "return false". For now, the client should simply return // false in any case it cannot perform partial-merge, regardless of reason. // If there is corruption in the data, handle it in the FullMerge() function, // and return false there. The default implementation of PartialMerge will // always return false. virtual bool PartialMerge(const Slice& key, const Slice& left_operand, const Slice& right_operand, std::string* new_value, Logger* logger) const { return false; } // This function performs merge when all the operands are themselves merge // operation types that you would have passed to a DB::Merge() call in the // same order (front() first) // (i.e. DB::Merge(key, operand_list[0]), followed by // DB::Merge(key, operand_list[1]), ...) // // PartialMergeMulti should combine them into a single merge operation that is // saved into *new_value, and then it should return true. *new_value should // be constructed such that a call to DB::Merge(key, *new_value) would yield // the same result as subquential individual calls to DB::Merge(key, operand) // for each operand in operand_list from front() to back(). // // The string that new_value is pointing to will be empty. // // The PartialMergeMulti function will be called only when the list of // operands are long enough. The minimum amount of operands that will be // passed to the function are specified by the "min_partial_merge_operands" // option. // // In the default implementation, PartialMergeMulti will invoke PartialMerge // multiple times, where each time it only merges two operands. Developers // should either implement PartialMergeMulti, or implement PartialMerge which // is served as the helper function of the default PartialMergeMulti. virtual bool PartialMergeMulti(const Slice& key, const std::deque& operand_list, std::string* new_value, Logger* logger) const; // The name of the MergeOperator. Used to check for MergeOperator // mismatches (i.e., a DB created with one MergeOperator is // accessed using a different MergeOperator) // TODO: the name is currently not stored persistently and thus // no checking is enforced. Client is responsible for providing // consistent MergeOperator between DB opens. virtual const char* Name() const = 0; }; // The simpler, associative merge operator. class AssociativeMergeOperator : public MergeOperator { public: virtual ~AssociativeMergeOperator() {} // Gives the client a way to express the read -> modify -> write semantics // key: (IN) The key that's associated with this merge operation. // existing_value:(IN) null indicates the key does not exist before this op // value: (IN) the value to update/merge the existing_value with // new_value: (OUT) Client is responsible for filling the merge result // here. The string that new_value is pointing to will be empty. // logger: (IN) Client could use this to log errors during merge. // // Return true on success. // All values passed in will be client-specific values. So if this method // returns false, it is because client specified bad data or there was // internal corruption. The client should assume that this will be treated // as an error by the library. virtual bool Merge(const Slice& key, const Slice* existing_value, const Slice& value, std::string* new_value, Logger* logger) const = 0; private: // Default implementations of the MergeOperator functions virtual bool FullMerge(const Slice& key, const Slice* existing_value, const std::deque& operand_list, std::string* new_value, Logger* logger) const override; virtual bool PartialMerge(const Slice& key, const Slice& left_operand, const Slice& right_operand, std::string* new_value, Logger* logger) const override; }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_MERGE_OPERATOR_H_ rocksdb-4.5.1/include/rocksdb/metadata.h000066400000000000000000000060631267530634000201660ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include #include "rocksdb/types.h" namespace rocksdb { struct ColumnFamilyMetaData; struct LevelMetaData; struct SstFileMetaData; // The metadata that describes a column family. struct ColumnFamilyMetaData { ColumnFamilyMetaData() : size(0), name("") {} ColumnFamilyMetaData(const std::string& _name, uint64_t _size, const std::vector&& _levels) : size(_size), name(_name), levels(_levels) {} // The size of this column family in bytes, which is equal to the sum of // the file size of its "levels". uint64_t size; // The number of files in this column family. size_t file_count; // The name of the column family. std::string name; // The metadata of all levels in this column family. std::vector levels; }; // The metadata that describes a level. struct LevelMetaData { LevelMetaData(int _level, uint64_t _size, const std::vector&& _files) : level(_level), size(_size), files(_files) {} // The level which this meta data describes. const int level; // The size of this level in bytes, which is equal to the sum of // the file size of its "files". const uint64_t size; // The metadata of all sst files in this level. const std::vector files; }; // The metadata that describes a SST file. struct SstFileMetaData { SstFileMetaData() {} SstFileMetaData(const std::string& _file_name, const std::string& _path, uint64_t _size, SequenceNumber _smallest_seqno, SequenceNumber _largest_seqno, const std::string& _smallestkey, const std::string& _largestkey, bool _being_compacted) : size(_size), name(_file_name), db_path(_path), smallest_seqno(_smallest_seqno), largest_seqno(_largest_seqno), smallestkey(_smallestkey), largestkey(_largestkey), being_compacted(_being_compacted) {} // File size in bytes. uint64_t size; // The name of the file. std::string name; // The full path where the file locates. std::string db_path; SequenceNumber smallest_seqno; // Smallest sequence number in file. SequenceNumber largest_seqno; // Largest sequence number in file. std::string smallestkey; // Smallest user defined key in the file. std::string largestkey; // Largest user defined key in the file. bool being_compacted; // true if the file is currently being compacted. }; // The full set of metadata associated with each SST file. struct LiveFileMetaData : SstFileMetaData { std::string column_family_name; // Name of the column family int level; // Level at which this file resides. }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/options.h000066400000000000000000001724051267530634000201050ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_ #define STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_ #include #include #include #include #include #include #include #include "rocksdb/version.h" #include "rocksdb/listener.h" #include "rocksdb/universal_compaction.h" #ifdef max #undef max #endif namespace rocksdb { class Cache; class CompactionFilter; class CompactionFilterFactory; class Comparator; class Env; enum InfoLogLevel : unsigned char; class SstFileManager; class FilterPolicy; class Logger; class MergeOperator; class Snapshot; class TableFactory; class MemTableRepFactory; class TablePropertiesCollectorFactory; class RateLimiter; class Slice; class SliceTransform; class Statistics; class InternalKeyComparator; class WalFilter; // DB contents are stored in a set of blocks, each of which holds a // sequence of key,value pairs. Each block may be compressed before // being stored in a file. The following enum describes which // compression method (if any) is used to compress a block. enum CompressionType : char { // NOTE: do not change the values of existing entries, as these are // part of the persistent format on disk. kNoCompression = 0x0, kSnappyCompression = 0x1, kZlibCompression = 0x2, kBZip2Compression = 0x3, kLZ4Compression = 0x4, kLZ4HCCompression = 0x5, // zstd format is not finalized yet so it's subject to changes. kZSTDNotFinalCompression = 0x40, }; enum CompactionStyle : char { // level based compaction style kCompactionStyleLevel = 0x0, // Universal compaction style // Not supported in ROCKSDB_LITE. kCompactionStyleUniversal = 0x1, // FIFO compaction style // Not supported in ROCKSDB_LITE kCompactionStyleFIFO = 0x2, // Disable background compaction. Compaction jobs are submitted // via CompactFiles(). // Not supported in ROCKSDB_LITE kCompactionStyleNone = 0x3, }; enum CompactionPri : char { // Slightly Priotize larger files by size compensated by #deletes kByCompensatedSize = 0x0, // First compact files whose data's latest update time is oldest. // Try this if you only update some hot keys in small ranges. kOldestLargestSeqFirst = 0x1, // First compact files whose range hasn't been compacted to the next level // for the longest. If your updates are random across the key space, // write amplification is slightly better with this option. kOldestSmallestSeqFirst = 0x2, }; enum class WALRecoveryMode : char { // Original levelDB recovery // We tolerate incomplete record in trailing data on all logs // Use case : This is legacy behavior (default) kTolerateCorruptedTailRecords = 0x00, // Recover from clean shutdown // We don't expect to find any corruption in the WAL // Use case : This is ideal for unit tests and rare applications that // can require high consistency guarantee kAbsoluteConsistency = 0x01, // Recover to point-in-time consistency // We stop the WAL playback on discovering WAL inconsistency // Use case : Ideal for systems that have disk controller cache like // hard disk, SSD without super capacitor that store related data kPointInTimeRecovery = 0x02, // Recovery after a disaster // We ignore any corruption in the WAL and try to salvage as much data as // possible // Use case : Ideal for last ditch effort to recover data or systems that // operate with low grade unrelated data kSkipAnyCorruptedRecords = 0x03, }; struct CompactionOptionsFIFO { // once the total sum of table files reaches this, we will delete the oldest // table file // Default: 1GB uint64_t max_table_files_size; CompactionOptionsFIFO() : max_table_files_size(1 * 1024 * 1024 * 1024) {} }; // Compression options for different compression algorithms like Zlib struct CompressionOptions { int window_bits; int level; int strategy; CompressionOptions() : window_bits(-14), level(-1), strategy(0) {} CompressionOptions(int wbits, int _lev, int _strategy) : window_bits(wbits), level(_lev), strategy(_strategy) {} }; enum UpdateStatus { // Return status For inplace update callback UPDATE_FAILED = 0, // Nothing to update UPDATED_INPLACE = 1, // Value updated inplace UPDATED = 2, // No inplace update. Merged value set }; struct DbPath { std::string path; uint64_t target_size; // Target size of total files under the path, in byte. DbPath() : target_size(0) {} DbPath(const std::string& p, uint64_t t) : path(p), target_size(t) {} }; struct Options; struct ColumnFamilyOptions { // Some functions that make it easier to optimize RocksDB // Use this if you don't need to keep the data sorted, i.e. you'll never use // an iterator, only Put() and Get() API calls // // Not supported in ROCKSDB_LITE ColumnFamilyOptions* OptimizeForPointLookup( uint64_t block_cache_size_mb); // Default values for some parameters in ColumnFamilyOptions are not // optimized for heavy workloads and big datasets, which means you might // observe write stalls under some conditions. As a starting point for tuning // RocksDB options, use the following two functions: // * OptimizeLevelStyleCompaction -- optimizes level style compaction // * OptimizeUniversalStyleCompaction -- optimizes universal style compaction // Universal style compaction is focused on reducing Write Amplification // Factor for big data sets, but increases Space Amplification. You can learn // more about the different styles here: // https://github.com/facebook/rocksdb/wiki/Rocksdb-Architecture-Guide // Make sure to also call IncreaseParallelism(), which will provide the // biggest performance gains. // Note: we might use more memory than memtable_memory_budget during high // write rate period // // OptimizeUniversalStyleCompaction is not supported in ROCKSDB_LITE ColumnFamilyOptions* OptimizeLevelStyleCompaction( uint64_t memtable_memory_budget = 512 * 1024 * 1024); ColumnFamilyOptions* OptimizeUniversalStyleCompaction( uint64_t memtable_memory_budget = 512 * 1024 * 1024); // ------------------- // Parameters that affect behavior // Comparator used to define the order of keys in the table. // Default: a comparator that uses lexicographic byte-wise ordering // // REQUIRES: The client must ensure that the comparator supplied // here has the same name and orders keys *exactly* the same as the // comparator provided to previous open calls on the same DB. const Comparator* comparator; // REQUIRES: The client must provide a merge operator if Merge operation // needs to be accessed. Calling Merge on a DB without a merge operator // would result in Status::NotSupported. The client must ensure that the // merge operator supplied here has the same name and *exactly* the same // semantics as the merge operator provided to previous open calls on // the same DB. The only exception is reserved for upgrade, where a DB // previously without a merge operator is introduced to Merge operation // for the first time. It's necessary to specify a merge operator when // openning the DB in this case. // Default: nullptr std::shared_ptr merge_operator; // A single CompactionFilter instance to call into during compaction. // Allows an application to modify/delete a key-value during background // compaction. // // If the client requires a new compaction filter to be used for different // compaction runs, it can specify compaction_filter_factory instead of this // option. The client should specify only one of the two. // compaction_filter takes precedence over compaction_filter_factory if // client specifies both. // // If multithreaded compaction is being used, the supplied CompactionFilter // instance may be used from different threads concurrently and so should be // thread-safe. // // Default: nullptr const CompactionFilter* compaction_filter; // This is a factory that provides compaction filter objects which allow // an application to modify/delete a key-value during background compaction. // // A new filter will be created on each compaction run. If multithreaded // compaction is being used, each created CompactionFilter will only be used // from a single thread and so does not need to be thread-safe. // // Default: nullptr std::shared_ptr compaction_filter_factory; // ------------------- // Parameters that affect performance // Amount of data to build up in memory (backed by an unsorted log // on disk) before converting to a sorted on-disk file. // // Larger values increase performance, especially during bulk loads. // Up to max_write_buffer_number write buffers may be held in memory // at the same time, // so you may wish to adjust this parameter to control memory usage. // Also, a larger write buffer will result in a longer recovery time // the next time the database is opened. // // Note that write_buffer_size is enforced per column family. // See db_write_buffer_size for sharing memory across column families. // // Default: 4MB // // Dynamically changeable through SetOptions() API size_t write_buffer_size; // The maximum number of write buffers that are built up in memory. // The default and the minimum number is 2, so that when 1 write buffer // is being flushed to storage, new writes can continue to the other // write buffer. // If max_write_buffer_number > 3, writing will be slowed down to // options.delayed_write_rate if we are writing to the last write buffer // allowed. // // Default: 2 // // Dynamically changeable through SetOptions() API int max_write_buffer_number; // The minimum number of write buffers that will be merged together // before writing to storage. If set to 1, then // all write buffers are fushed to L0 as individual files and this increases // read amplification because a get request has to check in all of these // files. Also, an in-memory merge may result in writing lesser // data to storage if there are duplicate records in each of these // individual write buffers. Default: 1 int min_write_buffer_number_to_merge; // The total maximum number of write buffers to maintain in memory including // copies of buffers that have already been flushed. Unlike // max_write_buffer_number, this parameter does not affect flushing. // This controls the minimum amount of write history that will be available // in memory for conflict checking when Transactions are used. // // When using an OptimisticTransactionDB: // If this value is too low, some transactions may fail at commit time due // to not being able to determine whether there were any write conflicts. // // When using a TransactionDB: // If Transaction::SetSnapshot is used, TransactionDB will read either // in-memory write buffers or SST files to do write-conflict checking. // Increasing this value can reduce the number of reads to SST files // done for conflict detection. // // Setting this value to 0 will cause write buffers to be freed immediately // after they are flushed. // If this value is set to -1, 'max_write_buffer_number' will be used. // // Default: // If using a TransactionDB/OptimisticTransactionDB, the default value will // be set to the value of 'max_write_buffer_number' if it is not explicitly // set by the user. Otherwise, the default is 0. int max_write_buffer_number_to_maintain; // Compress blocks using the specified compression algorithm. This // parameter can be changed dynamically. // // Default: kSnappyCompression, if it's supported. If snappy is not linked // with the library, the default is kNoCompression. // // Typical speeds of kSnappyCompression on an Intel(R) Core(TM)2 2.4GHz: // ~200-500MB/s compression // ~400-800MB/s decompression // Note that these speeds are significantly faster than most // persistent storage speeds, and therefore it is typically never // worth switching to kNoCompression. Even if the input data is // incompressible, the kSnappyCompression implementation will // efficiently detect that and will switch to uncompressed mode. CompressionType compression; // Different levels can have different compression policies. There // are cases where most lower levels would like to use quick compression // algorithms while the higher levels (which have more data) use // compression algorithms that have better compression but could // be slower. This array, if non-empty, should have an entry for // each level of the database; these override the value specified in // the previous field 'compression'. // // NOTICE if level_compaction_dynamic_level_bytes=true, // compression_per_level[0] still determines L0, but other elements // of the array are based on base level (the level L0 files are merged // to), and may not match the level users see from info log for metadata. // If L0 files are merged to level-n, then, for i>0, compression_per_level[i] // determines compaction type for level n+i-1. // For example, if we have three 5 levels, and we determine to merge L0 // data to L4 (which means L1..L3 will be empty), then the new files go to // L4 uses compression type compression_per_level[1]. // If now L0 is merged to L2. Data goes to L2 will be compressed // according to compression_per_level[1], L3 using compression_per_level[2] // and L4 using compression_per_level[3]. Compaction for each level can // change when data grows. std::vector compression_per_level; // different options for compression algorithms CompressionOptions compression_opts; // If non-nullptr, use the specified function to determine the // prefixes for keys. These prefixes will be placed in the filter. // Depending on the workload, this can reduce the number of read-IOP // cost for scans when a prefix is passed via ReadOptions to // db.NewIterator(). For prefix filtering to work properly, // "prefix_extractor" and "comparator" must be such that the following // properties hold: // // 1) key.starts_with(prefix(key)) // 2) Compare(prefix(key), key) <= 0. // 3) If Compare(k1, k2) <= 0, then Compare(prefix(k1), prefix(k2)) <= 0 // 4) prefix(prefix(key)) == prefix(key) // // Default: nullptr std::shared_ptr prefix_extractor; // Number of levels for this database int num_levels; // Number of files to trigger level-0 compaction. A value <0 means that // level-0 compaction will not be triggered by number of files at all. // // Default: 4 // // Dynamically changeable through SetOptions() API int level0_file_num_compaction_trigger; // Soft limit on number of level-0 files. We start slowing down writes at this // point. A value <0 means that no writing slow down will be triggered by // number of files in level-0. // // Dynamically changeable through SetOptions() API int level0_slowdown_writes_trigger; // Maximum number of level-0 files. We stop writes at this point. // // Dynamically changeable through SetOptions() API int level0_stop_writes_trigger; // This does not do anything anymore. Deprecated. int max_mem_compaction_level; // Target file size for compaction. // target_file_size_base is per-file size for level-1. // Target file size for level L can be calculated by // target_file_size_base * (target_file_size_multiplier ^ (L-1)) // For example, if target_file_size_base is 2MB and // target_file_size_multiplier is 10, then each file on level-1 will // be 2MB, and each file on level 2 will be 20MB, // and each file on level-3 will be 200MB. // // Default: 2MB. // // Dynamically changeable through SetOptions() API uint64_t target_file_size_base; // By default target_file_size_multiplier is 1, which means // by default files in different levels will have similar size. // // Dynamically changeable through SetOptions() API int target_file_size_multiplier; // Control maximum total data size for a level. // max_bytes_for_level_base is the max total for level-1. // Maximum number of bytes for level L can be calculated as // (max_bytes_for_level_base) * (max_bytes_for_level_multiplier ^ (L-1)) // For example, if max_bytes_for_level_base is 20MB, and if // max_bytes_for_level_multiplier is 10, total data size for level-1 // will be 20MB, total file size for level-2 will be 200MB, // and total file size for level-3 will be 2GB. // // Default: 10MB. // // Dynamically changeable through SetOptions() API uint64_t max_bytes_for_level_base; // If true, RocksDB will pick target size of each level dynamically. // We will pick a base level b >= 1. L0 will be directly merged into level b, // instead of always into level 1. Level 1 to b-1 need to be empty. // We try to pick b and its target size so that // 1. target size is in the range of // (max_bytes_for_level_base / max_bytes_for_level_multiplier, // max_bytes_for_level_base] // 2. target size of the last level (level num_levels-1) equals to extra size // of the level. // At the same time max_bytes_for_level_multiplier and // max_bytes_for_level_multiplier_additional are still satisfied. // // With this option on, from an empty DB, we make last level the base level, // which means merging L0 data into the last level, until it exceeds // max_bytes_for_level_base. And then we make the second last level to be // base level, to start to merge L0 data to second last level, with its // target size to be 1/max_bytes_for_level_multiplier of the last level's // extra size. After the data accumulates more so that we need to move the // base level to the third last one, and so on. // // For example, assume max_bytes_for_level_multiplier=10, num_levels=6, // and max_bytes_for_level_base=10MB. // Target sizes of level 1 to 5 starts with: // [- - - - 10MB] // with base level is level. Target sizes of level 1 to 4 are not applicable // because they will not be used. // Until the size of Level 5 grows to more than 10MB, say 11MB, we make // base target to level 4 and now the targets looks like: // [- - - 1.1MB 11MB] // While data are accumulated, size targets are tuned based on actual data // of level 5. When level 5 has 50MB of data, the target is like: // [- - - 5MB 50MB] // Until level 5's actual size is more than 100MB, say 101MB. Now if we keep // level 4 to be the base level, its target size needs to be 10.1MB, which // doesn't satisfy the target size range. So now we make level 3 the target // size and the target sizes of the levels look like: // [- - 1.01MB 10.1MB 101MB] // In the same way, while level 5 further grows, all levels' targets grow, // like // [- - 5MB 50MB 500MB] // Until level 5 exceeds 1000MB and becomes 1001MB, we make level 2 the // base level and make levels' target sizes like this: // [- 1.001MB 10.01MB 100.1MB 1001MB] // and go on... // // By doing it, we give max_bytes_for_level_multiplier a priority against // max_bytes_for_level_base, for a more predictable LSM tree shape. It is // useful to limit worse case space amplification. // // max_bytes_for_level_multiplier_additional is ignored with this flag on. // // Turning this feature on or off for an existing DB can cause unexpected // LSM tree structure so it's not recommended. // // NOTE: this option is experimental // // Default: false bool level_compaction_dynamic_level_bytes; // Default: 10. // // Dynamically changeable through SetOptions() API int max_bytes_for_level_multiplier; // Different max-size multipliers for different levels. // These are multiplied by max_bytes_for_level_multiplier to arrive // at the max-size of each level. // // Default: 1 // // Dynamically changeable through SetOptions() API std::vector max_bytes_for_level_multiplier_additional; // Maximum number of bytes in all compacted files. We avoid expanding // the lower level file set of a compaction if it would make the // total compaction cover more than // (expanded_compaction_factor * targetFileSizeLevel()) many bytes. // // Dynamically changeable through SetOptions() API int expanded_compaction_factor; // Maximum number of bytes in all source files to be compacted in a // single compaction run. We avoid picking too many files in the // source level so that we do not exceed the total source bytes // for compaction to exceed // (source_compaction_factor * targetFileSizeLevel()) many bytes. // Default:1, i.e. pick maxfilesize amount of data as the source of // a compaction. // // Dynamically changeable through SetOptions() API int source_compaction_factor; // Control maximum bytes of overlaps in grandparent (i.e., level+2) before we // stop building a single file in a level->level+1 compaction. // // Dynamically changeable through SetOptions() API int max_grandparent_overlap_factor; // DEPRECATED -- this options is no longer used // Puts are delayed to options.delayed_write_rate when any level has a // compaction score that exceeds soft_rate_limit. This is ignored when == 0.0. // // Default: 0 (disabled) // // Dynamically changeable through SetOptions() API double soft_rate_limit; // DEPRECATED -- this options is no longer used double hard_rate_limit; // All writes will be slowed down to at least delayed_write_rate if estimated // bytes needed to be compaction exceed this threshold. // // Default: 0 (disabled) uint64_t soft_pending_compaction_bytes_limit; // All writes are stopped if estimated bytes needed to be compaction exceed // this threshold. // // Default: 0 (disabled) uint64_t hard_pending_compaction_bytes_limit; // DEPRECATED -- this options is no longer used unsigned int rate_limit_delay_max_milliseconds; // size of one block in arena memory allocation. // If <= 0, a proper value is automatically calculated (usually 1/8 of // writer_buffer_size, rounded up to a multiple of 4KB). // // There are two additional restriction of the The specified size: // (1) size should be in the range of [4096, 2 << 30] and // (2) be the multiple of the CPU word (which helps with the memory // alignment). // // We'll automatically check and adjust the size number to make sure it // conforms to the restrictions. // // Default: 0 // // Dynamically changeable through SetOptions() API size_t arena_block_size; // Disable automatic compactions. Manual compactions can still // be issued on this column family // // Dynamically changeable through SetOptions() API bool disable_auto_compactions; // DEPREACTED // Does not have any effect. bool purge_redundant_kvs_while_flush; // The compaction style. Default: kCompactionStyleLevel CompactionStyle compaction_style; // If level compaction_style = kCompactionStyleLevel, for each level, // which files are prioritized to be picked to compact. // Default: kCompactionPriByCompensatedSize CompactionPri compaction_pri; // If true, compaction will verify checksum on every read that happens // as part of compaction // // Default: true // // Dynamically changeable through SetOptions() API bool verify_checksums_in_compaction; // The options needed to support Universal Style compactions CompactionOptionsUniversal compaction_options_universal; // The options for FIFO compaction style CompactionOptionsFIFO compaction_options_fifo; // Use KeyMayExist API to filter deletes when this is true. // If KeyMayExist returns false, i.e. the key definitely does not exist, then // the delete is a noop. KeyMayExist only incurs in-memory look up. // This optimization avoids writing the delete to storage when appropriate. // // Default: false // // Dynamically changeable through SetOptions() API bool filter_deletes; // An iteration->Next() sequentially skips over keys with the same // user-key unless this option is set. This number specifies the number // of keys (with the same userkey) that will be sequentially // skipped before a reseek is issued. // // Default: 8 // // Dynamically changeable through SetOptions() API uint64_t max_sequential_skip_in_iterations; // This is a factory that provides MemTableRep objects. // Default: a factory that provides a skip-list-based implementation of // MemTableRep. std::shared_ptr memtable_factory; // This is a factory that provides TableFactory objects. // Default: a block-based table factory that provides a default // implementation of TableBuilder and TableReader with default // BlockBasedTableOptions. std::shared_ptr table_factory; // Block-based table related options are moved to BlockBasedTableOptions. // Related options that were originally here but now moved include: // no_block_cache // block_cache // block_cache_compressed // block_size // block_size_deviation // block_restart_interval // filter_policy // whole_key_filtering // If you'd like to customize some of these options, you will need to // use NewBlockBasedTableFactory() to construct a new table factory. // This option allows user to to collect their own interested statistics of // the tables. // Default: empty vector -- no user-defined statistics collection will be // performed. typedef std::vector> TablePropertiesCollectorFactories; TablePropertiesCollectorFactories table_properties_collector_factories; // Allows thread-safe inplace updates. If this is true, there is no way to // achieve point-in-time consistency using snapshot or iterator (assuming // concurrent updates). Hence iterator and multi-get will return results // which are not consistent as of any point-in-time. // If inplace_callback function is not set, // Put(key, new_value) will update inplace the existing_value iff // * key exists in current memtable // * new sizeof(new_value) <= sizeof(existing_value) // * existing_value for that key is a put i.e. kTypeValue // If inplace_callback function is set, check doc for inplace_callback. // Default: false. bool inplace_update_support; // Number of locks used for inplace update // Default: 10000, if inplace_update_support = true, else 0. // // Dynamically changeable through SetOptions() API size_t inplace_update_num_locks; // existing_value - pointer to previous value (from both memtable and sst). // nullptr if key doesn't exist // existing_value_size - pointer to size of existing_value). // nullptr if key doesn't exist // delta_value - Delta value to be merged with the existing_value. // Stored in transaction logs. // merged_value - Set when delta is applied on the previous value. // Applicable only when inplace_update_support is true, // this callback function is called at the time of updating the memtable // as part of a Put operation, lets say Put(key, delta_value). It allows the // 'delta_value' specified as part of the Put operation to be merged with // an 'existing_value' of the key in the database. // If the merged value is smaller in size that the 'existing_value', // then this function can update the 'existing_value' buffer inplace and // the corresponding 'existing_value'_size pointer, if it wishes to. // The callback should return UpdateStatus::UPDATED_INPLACE. // In this case. (In this case, the snapshot-semantics of the rocksdb // Iterator is not atomic anymore). // If the merged value is larger in size than the 'existing_value' or the // application does not wish to modify the 'existing_value' buffer inplace, // then the merged value should be returned via *merge_value. It is set by // merging the 'existing_value' and the Put 'delta_value'. The callback should // return UpdateStatus::UPDATED in this case. This merged value will be added // to the memtable. // If merging fails or the application does not wish to take any action, // then the callback should return UpdateStatus::UPDATE_FAILED. // Please remember that the original call from the application is Put(key, // delta_value). So the transaction log (if enabled) will still contain (key, // delta_value). The 'merged_value' is not stored in the transaction log. // Hence the inplace_callback function should be consistent across db reopens. // Default: nullptr UpdateStatus (*inplace_callback)(char* existing_value, uint32_t* existing_value_size, Slice delta_value, std::string* merged_value); // if prefix_extractor is set and bloom_bits is not 0, create prefix bloom // for memtable // // Dynamically changeable through SetOptions() API uint32_t memtable_prefix_bloom_bits; // number of hash probes per key // // Dynamically changeable through SetOptions() API uint32_t memtable_prefix_bloom_probes; // Page size for huge page TLB for bloom in memtable. If <=0, not allocate // from huge page TLB but from malloc. // Need to reserve huge pages for it to be allocated. For example: // sysctl -w vm.nr_hugepages=20 // See linux doc Documentation/vm/hugetlbpage.txt // // Dynamically changeable through SetOptions() API size_t memtable_prefix_bloom_huge_page_tlb_size; // Control locality of bloom filter probes to improve cache miss rate. // This option only applies to memtable prefix bloom and plaintable // prefix bloom. It essentially limits every bloom checking to one cache line. // This optimization is turned off when set to 0, and positive number to turn // it on. // Default: 0 uint32_t bloom_locality; // Maximum number of successive merge operations on a key in the memtable. // // When a merge operation is added to the memtable and the maximum number of // successive merges is reached, the value of the key will be calculated and // inserted into the memtable instead of the merge operation. This will // ensure that there are never more than max_successive_merges merge // operations in the memtable. // // Default: 0 (disabled) // // Dynamically changeable through SetOptions() API size_t max_successive_merges; // The number of partial merge operands to accumulate before partial // merge will be performed. Partial merge will not be called // if the list of values to merge is less than min_partial_merge_operands. // // If min_partial_merge_operands < 2, then it will be treated as 2. // // Default: 2 uint32_t min_partial_merge_operands; // This flag specifies that the implementation should optimize the filters // mainly for cases where keys are found rather than also optimize for keys // missed. This would be used in cases where the application knows that // there are very few misses or the performance in the case of misses is not // important. // // For now, this flag allows us to not store filters for the last level i.e // the largest level which contains data of the LSM store. For keys which // are hits, the filters in this level are not useful because we will search // for the data anyway. NOTE: the filters in other levels are still useful // even for key hit because they tell us whether to look in that level or go // to the higher level. // // Default: false bool optimize_filters_for_hits; // After writing every SST file, reopen it and read all the keys. // Default: false bool paranoid_file_checks; // Measure IO stats in compactions, if true. // Default: false bool compaction_measure_io_stats; // Create ColumnFamilyOptions with default values for all fields ColumnFamilyOptions(); // Create ColumnFamilyOptions from Options explicit ColumnFamilyOptions(const Options& options); void Dump(Logger* log) const; }; struct DBOptions { // Some functions that make it easier to optimize RocksDB #ifndef ROCKSDB_LITE // By default, RocksDB uses only one background thread for flush and // compaction. Calling this function will set it up such that total of // `total_threads` is used. Good value for `total_threads` is the number of // cores. You almost definitely want to call this function if your system is // bottlenecked by RocksDB. DBOptions* IncreaseParallelism(int total_threads = 16); #endif // ROCKSDB_LITE // If true, the database will be created if it is missing. // Default: false bool create_if_missing; // If true, missing column families will be automatically created. // Default: false bool create_missing_column_families; // If true, an error is raised if the database already exists. // Default: false bool error_if_exists; // If true, RocksDB will aggressively check consistency of the data. // Also, if any of the writes to the database fails (Put, Delete, Merge, // Write), the database will switch to read-only mode and fail all other // Write operations. // In most cases you want this to be set to true. // Default: true bool paranoid_checks; // Use the specified object to interact with the environment, // e.g. to read/write files, schedule background work, etc. // Default: Env::Default() Env* env; // Use to control write rate of flush and compaction. Flush has higher // priority than compaction. Rate limiting is disabled if nullptr. // If rate limiter is enabled, bytes_per_sync is set to 1MB by default. // Default: nullptr std::shared_ptr rate_limiter; // Use to track SST files and control their file deletion rate, can be used // among multiple RocksDB instances, sst_file_manager only track and throttle // deletes of SST files in first db_path (db_name if db_paths is empty), other // files and other db_paths wont be tracked or affected by sst_file_manager. // Default: nullptr std::shared_ptr sst_file_manager; // Any internal progress/error information generated by the db will // be written to info_log if it is non-nullptr, or to a file stored // in the same directory as the DB contents if info_log is nullptr. // Default: nullptr std::shared_ptr info_log; InfoLogLevel info_log_level; // Number of open files that can be used by the DB. You may need to // increase this if your database has a large working set. Value -1 means // files opened are always kept open. You can estimate number of files based // on target_file_size_base and target_file_size_multiplier for level-based // compaction. For universal-style compaction, you can usually set it to -1. // Default: 5000 or ulimit value of max open files (whichever is smaller) int max_open_files; // If max_open_files is -1, DB will open all files on DB::Open(). You can // use this option to increase the number of threads used to open the files. // Default: 1 int max_file_opening_threads; // Once write-ahead logs exceed this size, we will start forcing the flush of // column families whose memtables are backed by the oldest live WAL file // (i.e. the ones that are causing all the space amplification). If set to 0 // (default), we will dynamically choose the WAL size limit to be // [sum of all write_buffer_size * max_write_buffer_number] * 4 // Default: 0 uint64_t max_total_wal_size; // If non-null, then we should collect metrics about database operations // Statistics objects should not be shared between DB instances as // it does not use any locks to prevent concurrent updates. std::shared_ptr statistics; // If true, then the contents of manifest and data files are not synced // to stable storage. Their contents remain in the OS buffers till the // OS decides to flush them. This option is good for bulk-loading // of data. Once the bulk-loading is complete, please issue a // sync to the OS to flush all dirty buffesrs to stable storage. // Default: false bool disableDataSync; // If true, then every store to stable storage will issue a fsync. // If false, then every store to stable storage will issue a fdatasync. // This parameter should be set to true while storing data to // filesystem like ext3 that can lose files after a reboot. // Default: false bool use_fsync; // A list of paths where SST files can be put into, with its target size. // Newer data is placed into paths specified earlier in the vector while // older data gradually moves to paths specified later in the vector. // // For example, you have a flash device with 10GB allocated for the DB, // as well as a hard drive of 2TB, you should config it to be: // [{"/flash_path", 10GB}, {"/hard_drive", 2TB}] // // The system will try to guarantee data under each path is close to but // not larger than the target size. But current and future file sizes used // by determining where to place a file are based on best-effort estimation, // which means there is a chance that the actual size under the directory // is slightly more than target size under some workloads. User should give // some buffer room for those cases. // // If none of the paths has sufficient room to place a file, the file will // be placed to the last path anyway, despite to the target size. // // Placing newer data to earlier paths is also best-efforts. User should // expect user files to be placed in higher levels in some extreme cases. // // If left empty, only one path will be used, which is db_name passed when // opening the DB. // Default: empty std::vector db_paths; // This specifies the info LOG dir. // If it is empty, the log files will be in the same dir as data. // If it is non empty, the log files will be in the specified dir, // and the db data dir's absolute path will be used as the log file // name's prefix. std::string db_log_dir; // This specifies the absolute dir path for write-ahead logs (WAL). // If it is empty, the log files will be in the same dir as data, // dbname is used as the data dir by default // If it is non empty, the log files will be in kept the specified dir. // When destroying the db, // all log files in wal_dir and the dir itself is deleted std::string wal_dir; // The periodicity when obsolete files get deleted. The default // value is 6 hours. The files that get out of scope by compaction // process will still get automatically delete on every compaction, // regardless of this setting uint64_t delete_obsolete_files_period_micros; // Suggested number of concurrent background compaction jobs, submitted to // the default LOW priority thread pool. // // Default: max_background_compactions int base_background_compactions; // Maximum number of concurrent background compaction jobs, submitted to // the default LOW priority thread pool. // We first try to schedule compactions based on // `base_background_compactions`. If the compaction cannot catch up , we // will increase number of compaction threads up to // `max_background_compactions`. // // If you're increasing this, also consider increasing number of threads in // LOW priority thread pool. For more information, see // Env::SetBackgroundThreads // Default: 1 int max_background_compactions; // This value represents the maximum number of threads that will // concurrently perform a compaction job by breaking it into multiple, // smaller ones that are run simultaneously. // Default: 1 (i.e. no subcompactions) uint32_t max_subcompactions; // Maximum number of concurrent background memtable flush jobs, submitted to // the HIGH priority thread pool. // // By default, all background jobs (major compaction and memtable flush) go // to the LOW priority pool. If this option is set to a positive number, // memtable flush jobs will be submitted to the HIGH priority pool. // It is important when the same Env is shared by multiple db instances. // Without a separate pool, long running major compaction jobs could // potentially block memtable flush jobs of other db instances, leading to // unnecessary Put stalls. // // If you're increasing this, also consider increasing number of threads in // HIGH priority thread pool. For more information, see // Env::SetBackgroundThreads // Default: 1 int max_background_flushes; // Specify the maximal size of the info log file. If the log file // is larger than `max_log_file_size`, a new info log file will // be created. // If max_log_file_size == 0, all logs will be written to one // log file. size_t max_log_file_size; // Time for the info log file to roll (in seconds). // If specified with non-zero value, log file will be rolled // if it has been active longer than `log_file_time_to_roll`. // Default: 0 (disabled) size_t log_file_time_to_roll; // Maximal info log files to be kept. // Default: 1000 size_t keep_log_file_num; // Recycle log files. // If non-zero, we will reuse previously written log files for new // logs, overwriting the old data. The value indicates how many // such files we will keep around at any point in time for later // use. This is more efficient because the blocks are already // allocated and fdatasync does not need to update the inode after // each write. // Default: 0 size_t recycle_log_file_num; // manifest file is rolled over on reaching this limit. // The older manifest file be deleted. // The default value is MAX_INT so that roll-over does not take place. uint64_t max_manifest_file_size; // Number of shards used for table cache. int table_cache_numshardbits; // DEPRECATED // int table_cache_remove_scan_count_limit; // The following two fields affect how archived logs will be deleted. // 1. If both set to 0, logs will be deleted asap and will not get into // the archive. // 2. If WAL_ttl_seconds is 0 and WAL_size_limit_MB is not 0, // WAL files will be checked every 10 min and if total size is greater // then WAL_size_limit_MB, they will be deleted starting with the // earliest until size_limit is met. All empty files will be deleted. // 3. If WAL_ttl_seconds is not 0 and WAL_size_limit_MB is 0, then // WAL files will be checked every WAL_ttl_secondsi / 2 and those that // are older than WAL_ttl_seconds will be deleted. // 4. If both are not 0, WAL files will be checked every 10 min and both // checks will be performed with ttl being first. uint64_t WAL_ttl_seconds; uint64_t WAL_size_limit_MB; // Number of bytes to preallocate (via fallocate) the manifest // files. Default is 4mb, which is reasonable to reduce random IO // as well as prevent overallocation for mounts that preallocate // large amounts of data (such as xfs's allocsize option). size_t manifest_preallocation_size; // Data being read from file storage may be buffered in the OS // Default: true bool allow_os_buffer; // Allow the OS to mmap file for reading sst tables. Default: false bool allow_mmap_reads; // Allow the OS to mmap file for writing. // DB::SyncWAL() only works if this is set to false. // Default: false bool allow_mmap_writes; // If false, fallocate() calls are bypassed bool allow_fallocate; // Disable child process inherit open files. Default: true bool is_fd_close_on_exec; // DEPRECATED -- this options is no longer used bool skip_log_error_on_recovery; // if not zero, dump rocksdb.stats to LOG every stats_dump_period_sec // Default: 600 (10 min) unsigned int stats_dump_period_sec; // If set true, will hint the underlying file system that the file // access pattern is random, when a sst file is opened. // Default: true bool advise_random_on_open; // Amount of data to build up in memtables across all column // families before writing to disk. // // This is distinct from write_buffer_size, which enforces a limit // for a single memtable. // // This feature is disabled by default. Specify a non-zero value // to enable it. // // Default: 0 (disabled) size_t db_write_buffer_size; // Specify the file access pattern once a compaction is started. // It will be applied to all input files of a compaction. // Default: NORMAL enum AccessHint { NONE, NORMAL, SEQUENTIAL, WILLNEED }; AccessHint access_hint_on_compaction_start; // If true, always create a new file descriptor and new table reader // for compaction inputs. Turn this parameter on may introduce extra // memory usage in the table reader, if it allocates extra memory // for indexes. This will allow file descriptor prefetch options // to be set for compaction input files and not to impact file // descriptors for the same file used by user queries. // Suggest to enable BlockBasedTableOptions.cache_index_and_filter_blocks // for this mode if using block-based table. // // Default: false bool new_table_reader_for_compaction_inputs; // If non-zero, we perform bigger reads when doing compaction. If you're // running RocksDB on spinning disks, you should set this to at least 2MB. // That way RocksDB's compaction is doing sequential instead of random reads. // // When non-zero, we also force new_table_reader_for_compaction_inputs to // true. // // Default: 0 size_t compaction_readahead_size; // This is a maximum buffer size that is used by WinMmapReadableFile in // unbuffered disk I/O mode. We need to maintain an aligned buffer for // reads. We allow the buffer to grow until the specified value and then // for bigger requests allocate one shot buffers. In unbuffered mode we // always bypass read-ahead buffer at ReadaheadRandomAccessFile // When read-ahead is required we then make use of compaction_readahead_size // value and always try to read ahead. With read-ahead we always // pre-allocate buffer to the size instead of growing it up to a limit. // // This option is currently honored only on Windows // // Default: 1 Mb // // Special value: 0 - means do not maintain per instance buffer. Allocate // per request buffer and avoid locking. size_t random_access_max_buffer_size; // This is the maximum buffer size that is used by WritableFileWriter. // On Windows, we need to maintain an aligned buffer for writes. // We allow the buffer to grow until it's size hits the limit. // // Default: 1024 * 1024 (1 MB) size_t writable_file_max_buffer_size; // Use adaptive mutex, which spins in the user space before resorting // to kernel. This could reduce context switch when the mutex is not // heavily contended. However, if the mutex is hot, we could end up // wasting spin time. // Default: false bool use_adaptive_mutex; // Create DBOptions with default values for all fields DBOptions(); // Create DBOptions from Options explicit DBOptions(const Options& options); void Dump(Logger* log) const; // Allows OS to incrementally sync files to disk while they are being // written, asynchronously, in the background. This operation can be used // to smooth out write I/Os over time. Users shouldn't reply on it for // persistency guarantee. // Issue one request for every bytes_per_sync written. 0 turns it off. // Default: 0 // // You may consider using rate_limiter to regulate write rate to device. // When rate limiter is enabled, it automatically enables bytes_per_sync // to 1MB. // // This option applies to table files uint64_t bytes_per_sync; // Same as bytes_per_sync, but applies to WAL files // Default: 0, turned off uint64_t wal_bytes_per_sync; // A vector of EventListeners which call-back functions will be called // when specific RocksDB event happens. std::vector> listeners; // If true, then the status of the threads involved in this DB will // be tracked and available via GetThreadList() API. // // Default: false bool enable_thread_tracking; // The limited write rate to DB if soft_pending_compaction_bytes_limit or // level0_slowdown_writes_trigger is triggered, or we are writing to the // last mem table allowed and we allow more than 3 mem tables. It is // calculated using size of user write requests before compression. // RocksDB may decide to slow down more if the compaction still // gets behind further. // Unit: byte per second. // // Default: 2MB/s uint64_t delayed_write_rate; // If true, allow multi-writers to update mem tables in parallel. // Only some memtable_factory-s support concurrent writes; currently it // is implemented only for SkipListFactory. Concurrent memtable writes // are not compatible with inplace_update_support or filter_deletes. // It is strongly recommended to set enable_write_thread_adaptive_yield // if you are going to use this feature. // // THIS FEATURE IS NOT STABLE YET. // // Default: false bool allow_concurrent_memtable_write; // If true, threads synchronizing with the write batch group leader will // wait for up to write_thread_max_yield_usec before blocking on a mutex. // This can substantially improve throughput for concurrent workloads, // regardless of whether allow_concurrent_memtable_write is enabled. // // THIS FEATURE IS NOT STABLE YET. // // Default: false bool enable_write_thread_adaptive_yield; // The maximum number of microseconds that a write operation will use // a yielding spin loop to coordinate with other write threads before // blocking on a mutex. (Assuming write_thread_slow_yield_usec is // set properly) increasing this value is likely to increase RocksDB // throughput at the expense of increased CPU usage. // // Default: 100 uint64_t write_thread_max_yield_usec; // The latency in microseconds after which a std::this_thread::yield // call (sched_yield on Linux) is considered to be a signal that // other processes or threads would like to use the current core. // Increasing this makes writer threads more likely to take CPU // by spinning, which will show up as an increase in the number of // involuntary context switches. // // Default: 3 uint64_t write_thread_slow_yield_usec; // If true, then DB::Open() will not update the statistics used to optimize // compaction decision by loading table properties from many files. // Turning off this feature will improve DBOpen time especially in // disk environment. // // Default: false bool skip_stats_update_on_db_open; // Recovery mode to control the consistency while replaying WAL // Default: kTolerateCorruptedTailRecords WALRecoveryMode wal_recovery_mode; // A global cache for table-level rows. // Default: nullptr (disabled) // Not supported in ROCKSDB_LITE mode! std::shared_ptr row_cache; #ifndef ROCKSDB_LITE // A filter object supplied to be invoked while processing write-ahead-logs // (WALs) during recovery. The filter provides a way to inspect log // records, ignoring a particular record or skipping replay. // The filter is invoked at startup and is invoked from a single-thread // currently. const WalFilter* wal_filter; #endif // ROCKSDB_LITE // If true, then DB::Open / CreateColumnFamily / DropColumnFamily // / SetOptions will fail if options file is not detected or properly // persisted. // // DEFAULT: false bool fail_if_options_file_error; }; // Options to control the behavior of a database (passed to DB::Open) struct Options : public DBOptions, public ColumnFamilyOptions { // Create an Options object with default values for all fields. Options() : DBOptions(), ColumnFamilyOptions() {} Options(const DBOptions& db_options, const ColumnFamilyOptions& column_family_options) : DBOptions(db_options), ColumnFamilyOptions(column_family_options) {} void Dump(Logger* log) const; void DumpCFOptions(Logger* log) const; // Set appropriate parameters for bulk loading. // The reason that this is a function that returns "this" instead of a // constructor is to enable chaining of multiple similar calls in the future. // // All data will be in level 0 without any automatic compaction. // It's recommended to manually call CompactRange(NULL, NULL) before reading // from the database, because otherwise the read can be very slow. Options* PrepareForBulkLoad(); }; // // An application can issue a read request (via Get/Iterators) and specify // if that read should process data that ALREADY resides on a specified cache // level. For example, if an application specifies kBlockCacheTier then the // Get call will process data that is already processed in the memtable or // the block cache. It will not page in data from the OS cache or data that // resides in storage. enum ReadTier { kReadAllTier = 0x0, // data in memtable, block cache, OS cache or storage kBlockCacheTier = 0x1, // data in memtable or block cache kPersistedTier = 0x2 // persisted data. When WAL is disabled, this option // will skip data in memtable. // Note that this ReadTier currently only supports // Get and MultiGet and does not support iterators. }; // Options that control read operations struct ReadOptions { // If true, all data read from underlying storage will be // verified against corresponding checksums. // Default: true bool verify_checksums; // Should the "data block"/"index block"/"filter block" read for this // iteration be cached in memory? // Callers may wish to set this field to false for bulk scans. // Default: true bool fill_cache; // If this option is set and memtable implementation allows, Seek // might only return keys with the same prefix as the seek-key // // ! DEPRECATED: prefix_seek is on by default when prefix_extractor // is configured // bool prefix_seek; // If "snapshot" is non-nullptr, read as of the supplied snapshot // (which must belong to the DB that is being read and which must // not have been released). If "snapshot" is nullptr, use an impliicit // snapshot of the state at the beginning of this read operation. // Default: nullptr const Snapshot* snapshot; // If "prefix" is non-nullptr, and ReadOptions is being passed to // db.NewIterator, only return results when the key begins with this // prefix. This field is ignored by other calls (e.g., Get). // Options.prefix_extractor must also be set, and // prefix_extractor.InRange(prefix) must be true. The iterator // returned by NewIterator when this option is set will behave just // as if the underlying store did not contain any non-matching keys, // with two exceptions. Seek() only accepts keys starting with the // prefix, and SeekToLast() is not supported. prefix filter with this // option will sometimes reduce the number of read IOPs. // Default: nullptr // // ! DEPRECATED // const Slice* prefix; // "iterate_upper_bound" defines the extent upto which the forward iterator // can returns entries. Once the bound is reached, Valid() will be false. // "iterate_upper_bound" is exclusive ie the bound value is // not a valid entry. If iterator_extractor is not null, the Seek target // and iterator_upper_bound need to have the same prefix. // This is because ordering is not guaranteed outside of prefix domain. // There is no lower bound on the iterator. If needed, that can be easily // implemented // // Default: nullptr const Slice* iterate_upper_bound; // Specify if this read request should process data that ALREADY // resides on a particular cache. If the required data is not // found at the specified cache, then Status::Incomplete is returned. // Default: kReadAllTier ReadTier read_tier; // Specify to create a tailing iterator -- a special iterator that has a // view of the complete database (i.e. it can also be used to read newly // added data) and is optimized for sequential reads. It will return records // that were inserted into the database after the creation of the iterator. // Default: false // Not supported in ROCKSDB_LITE mode! bool tailing; // Specify to create a managed iterator -- a special iterator that // uses less resources by having the ability to free its underlying // resources on request. // Default: false // Not supported in ROCKSDB_LITE mode! bool managed; // Enable a total order seek regardless of index format (e.g. hash index) // used in the table. Some table format (e.g. plain table) may not support // this option. bool total_order_seek; // Enforce that the iterator only iterates over the same prefix as the seek. // This option is effective only for prefix seeks, i.e. prefix_extractor is // non-null for the column family and total_order_seek is false. Unlike // iterate_upper_bound, prefix_same_as_start only works within a prefix // but in both directions. // Default: false bool prefix_same_as_start; // Keep the blocks loaded by the iterator pinned in memory as long as the // iterator is not deleted, If used when reading from tables created with // BlockBasedTableOptions::use_delta_encoding = false, Iterator::IsKeyPinned() // is guaranteed to return true. // Default: false bool pin_data; ReadOptions(); ReadOptions(bool cksum, bool cache); }; // Options that control write operations struct WriteOptions { // If true, the write will be flushed from the operating system // buffer cache (by calling WritableFile::Sync()) before the write // is considered complete. If this flag is true, writes will be // slower. // // If this flag is false, and the machine crashes, some recent // writes may be lost. Note that if it is just the process that // crashes (i.e., the machine does not reboot), no writes will be // lost even if sync==false. // // In other words, a DB write with sync==false has similar // crash semantics as the "write()" system call. A DB write // with sync==true has similar crash semantics to a "write()" // system call followed by "fdatasync()". // // Default: false bool sync; // If true, writes will not first go to the write ahead log, // and the write may got lost after a crash. bool disableWAL; // The option is deprecated. It's not used anymore. uint64_t timeout_hint_us; // If true and if user is trying to write to column families that don't exist // (they were dropped), ignore the write (don't return an error). If there // are multiple writes in a WriteBatch, other writes will succeed. // Default: false bool ignore_missing_column_families; WriteOptions() : sync(false), disableWAL(false), timeout_hint_us(0), ignore_missing_column_families(false) {} }; // Options that control flush operations struct FlushOptions { // If true, the flush will wait until the flush is done. // Default: true bool wait; FlushOptions() : wait(true) {} }; // Get options based on some guidelines. Now only tune parameter based on // flush/compaction and fill default parameters for other parameters. // total_write_buffer_limit: budget for memory spent for mem tables // read_amplification_threshold: comfortable value of read amplification // write_amplification_threshold: comfortable value of write amplification. // target_db_size: estimated total DB size. extern Options GetOptions(size_t total_write_buffer_limit, int read_amplification_threshold = 8, int write_amplification_threshold = 32, uint64_t target_db_size = 68719476736 /* 64GB */); // Create a Logger from provided DBOptions extern Status CreateLoggerFromOptions(const std::string& dbname, const DBOptions& options, std::shared_ptr* logger); // CompactionOptions are used in CompactFiles() call. struct CompactionOptions { // Compaction output compression type // Default: snappy CompressionType compression; // Compaction will create files of size `output_file_size_limit`. // Default: MAX, which means that compaction will create a single file uint64_t output_file_size_limit; CompactionOptions() : compression(kSnappyCompression), output_file_size_limit(std::numeric_limits::max()) {} }; // For level based compaction, we can configure if we want to skip/force // bottommost level compaction. enum class BottommostLevelCompaction { // Skip bottommost level compaction kSkip, // Only compact bottommost level if there is a compaction filter // This is the default option kIfHaveCompactionFilter, // Always compact bottommost level kForce, }; // CompactRangeOptions is used by CompactRange() call. struct CompactRangeOptions { // If true, no other compaction will run at the same time as this // manual compaction bool exclusive_manual_compaction = true; // If true, compacted files will be moved to the minimum level capable // of holding the data or given level (specified non-negative target_level). bool change_level = false; // If change_level is true and target_level have non-negative value, compacted // files will be moved to target_level. int target_level = -1; // Compaction outputs will be placed in options.db_paths[target_path_id]. // Behavior is undefined if target_path_id is out of range. uint32_t target_path_id = 0; // By default level based compaction will only compact the bottommost level // if there is a compaction filter BottommostLevelCompaction bottommost_level_compaction = BottommostLevelCompaction::kIfHaveCompactionFilter; }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_ rocksdb-4.5.1/include/rocksdb/perf_context.h000066400000000000000000000103351267530634000211030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef STORAGE_ROCKSDB_INCLUDE_PERF_CONTEXT_H #define STORAGE_ROCKSDB_INCLUDE_PERF_CONTEXT_H #include #include #include "rocksdb/perf_level.h" namespace rocksdb { // A thread local context for gathering performance counter efficiently // and transparently. // Use SetPerfLevel(PerfLevel::kEnableTime) to enable time stats. struct PerfContext { void Reset(); // reset all performance counters to zero std::string ToString(bool exclude_zero_counters = false) const; uint64_t user_key_comparison_count; // total number of user key comparisons uint64_t block_cache_hit_count; // total number of block cache hits uint64_t block_read_count; // total number of block reads (with IO) uint64_t block_read_byte; // total number of bytes from block reads uint64_t block_read_time; // total nanos spent on block reads uint64_t block_checksum_time; // total nanos spent on block checksum uint64_t block_decompress_time; // total nanos spent on block decompression // total number of internal keys skipped over during iteration (overwritten or // deleted, to be more specific, hidden by a put or delete of the same key) uint64_t internal_key_skipped_count; // total number of deletes and single deletes skipped over during iteration uint64_t internal_delete_skipped_count; uint64_t get_snapshot_time; // total nanos spent on getting snapshot uint64_t get_from_memtable_time; // total nanos spent on querying memtables uint64_t get_from_memtable_count; // number of mem tables queried // total nanos spent after Get() finds a key uint64_t get_post_process_time; uint64_t get_from_output_files_time; // total nanos reading from output files // total nanos spent on seeking memtable uint64_t seek_on_memtable_time; // number of seeks issued on memtable uint64_t seek_on_memtable_count; // total nanos spent on seeking child iters uint64_t seek_child_seek_time; // number of seek issued in child iterators uint64_t seek_child_seek_count; uint64_t seek_min_heap_time; // total nanos spent on the merge heap // total nanos spent on seeking the internal entries uint64_t seek_internal_seek_time; // total nanos spent on iterating internal entries to find the next user entry uint64_t find_next_user_entry_time; // total nanos spent on writing to WAL uint64_t write_wal_time; // total nanos spent on writing to mem tables uint64_t write_memtable_time; // total nanos spent on delaying write uint64_t write_delay_time; // total nanos spent on writing a record, excluding the above three times uint64_t write_pre_and_post_process_time; uint64_t db_mutex_lock_nanos; // time spent on acquiring DB mutex. // Time spent on waiting with a condition variable created with DB mutex. uint64_t db_condition_wait_nanos; // Time spent on merge operator. uint64_t merge_operator_time_nanos; // Time spent on reading index block from block cache or SST file uint64_t read_index_block_nanos; // Time spent on reading filter block from block cache or SST file uint64_t read_filter_block_nanos; // Time spent on creating data block iterator uint64_t new_table_block_iter_nanos; // Time spent on creating a iterator of an SST file. uint64_t new_table_iterator_nanos; // Time spent on seeking a key in data/index blocks uint64_t block_seek_nanos; // Time spent on finding or creating a table reader uint64_t find_table_nanos; // total number of mem table bloom hits uint64_t bloom_memtable_hit_count; // total number of mem table bloom misses uint64_t bloom_memtable_miss_count; // total number of SST table bloom hits uint64_t bloom_sst_hit_count; // total number of SST table bloom misses uint64_t bloom_sst_miss_count; }; #if defined(NPERF_CONTEXT) || defined(IOS_CROSS_COMPILE) extern PerfContext perf_context; #elif _WIN32 extern __declspec(thread) PerfContext perf_context; #else extern __thread PerfContext perf_context; #endif } #endif rocksdb-4.5.1/include/rocksdb/perf_level.h000066400000000000000000000020631267530634000205250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef INCLUDE_ROCKSDB_PERF_LEVEL_H_ #define INCLUDE_ROCKSDB_PERF_LEVEL_H_ #include #include namespace rocksdb { // How much perf stats to collect. Affects perf_context and iostats_context. enum PerfLevel { kDisable = 0, // disable perf stats kEnableCount = 1, // enable only count stats kEnableTimeExceptForMutex = 2, // Other than count stats, also enable time // stats except for mutexes kEnableTime = 3 // enable count and time stats }; // set the perf stats level for current thread void SetPerfLevel(PerfLevel level); // get current perf stats level for current thread PerfLevel GetPerfLevel(); } // namespace rocksdb #endif // INCLUDE_ROCKSDB_PERF_LEVEL_H_ rocksdb-4.5.1/include/rocksdb/rate_limiter.h000066400000000000000000000055271267530634000210720ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include "rocksdb/env.h" namespace rocksdb { class RateLimiter { public: virtual ~RateLimiter() {} // This API allows user to dynamically change rate limiter's bytes per second. // REQUIRED: bytes_per_second > 0 virtual void SetBytesPerSecond(int64_t bytes_per_second) = 0; // Request for token to write bytes. If this request can not be satisfied, // the call is blocked. Caller is responsible to make sure // bytes <= GetSingleBurstBytes() virtual void Request(const int64_t bytes, const Env::IOPriority pri) = 0; // Max bytes can be granted in a single burst virtual int64_t GetSingleBurstBytes() const = 0; // Total bytes that go though rate limiter virtual int64_t GetTotalBytesThrough( const Env::IOPriority pri = Env::IO_TOTAL) const = 0; // Total # of requests that go though rate limiter virtual int64_t GetTotalRequests( const Env::IOPriority pri = Env::IO_TOTAL) const = 0; }; // Create a RateLimiter object, which can be shared among RocksDB instances to // control write rate of flush and compaction. // @rate_bytes_per_sec: this is the only parameter you want to set most of the // time. It controls the total write rate of compaction and flush in bytes per // second. Currently, RocksDB does not enforce rate limit for anything other // than flush and compaction, e.g. write to WAL. // @refill_period_us: this controls how often tokens are refilled. For example, // when rate_bytes_per_sec is set to 10MB/s and refill_period_us is set to // 100ms, then 1MB is refilled every 100ms internally. Larger value can lead to // burstier writes while smaller value introduces more CPU overhead. // The default should work for most cases. // @fairness: RateLimiter accepts high-pri requests and low-pri requests. // A low-pri request is usually blocked in favor of hi-pri request. Currently, // RocksDB assigns low-pri to request from compaciton and high-pri to request // from flush. Low-pri requests can get blocked if flush requests come in // continuouly. This fairness parameter grants low-pri requests permission by // 1/fairness chance even though high-pri requests exist to avoid starvation. // You should be good by leaving it at default 10. extern RateLimiter* NewGenericRateLimiter( int64_t rate_bytes_per_sec, int64_t refill_period_us = 100 * 1000, int32_t fairness = 10); } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/slice.h000066400000000000000000000110211267530634000174730ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Slice is a simple structure containing a pointer into some external // storage and a size. The user of a Slice must ensure that the slice // is not used after the corresponding external storage has been // deallocated. // // Multiple threads can invoke const methods on a Slice without // external synchronization, but if any of the threads may call a // non-const method, all threads accessing the same Slice must use // external synchronization. #ifndef STORAGE_ROCKSDB_INCLUDE_SLICE_H_ #define STORAGE_ROCKSDB_INCLUDE_SLICE_H_ #include #include #include #include #include namespace rocksdb { class Slice { public: // Create an empty slice. Slice() : data_(""), size_(0) { } // Create a slice that refers to d[0,n-1]. Slice(const char* d, size_t n) : data_(d), size_(n) { } // Create a slice that refers to the contents of "s" /* implicit */ Slice(const std::string& s) : data_(s.data()), size_(s.size()) { } // Create a slice that refers to s[0,strlen(s)-1] /* implicit */ Slice(const char* s) : data_(s), size_(strlen(s)) { } // Create a single slice from SliceParts using buf as storage. // buf must exist as long as the returned Slice exists. Slice(const struct SliceParts& parts, std::string* buf); // Return a pointer to the beginning of the referenced data const char* data() const { return data_; } // Return the length (in bytes) of the referenced data size_t size() const { return size_; } // Return true iff the length of the referenced data is zero bool empty() const { return size_ == 0; } // Return the ith byte in the referenced data. // REQUIRES: n < size() char operator[](size_t n) const { assert(n < size()); return data_[n]; } // Change this slice to refer to an empty array void clear() { data_ = ""; size_ = 0; } // Drop the first "n" bytes from this slice. void remove_prefix(size_t n) { assert(n <= size()); data_ += n; size_ -= n; } void remove_suffix(size_t n) { assert(n <= size()); size_ -= n; } // Return a string that contains the copy of the referenced data. std::string ToString(bool hex = false) const; // Three-way comparison. Returns value: // < 0 iff "*this" < "b", // == 0 iff "*this" == "b", // > 0 iff "*this" > "b" int compare(const Slice& b) const; // Return true iff "x" is a prefix of "*this" bool starts_with(const Slice& x) const { return ((size_ >= x.size_) && (memcmp(data_, x.data_, x.size_) == 0)); } bool ends_with(const Slice& x) const { return ((size_ >= x.size_) && (memcmp(data_ + size_ - x.size_, x.data_, x.size_) == 0)); } // Compare two slices and returns the first byte where they differ size_t difference_offset(const Slice& b) const; // private: make these public for rocksdbjni access const char* data_; size_t size_; // Intentionally copyable }; // A set of Slices that are virtually concatenated together. 'parts' points // to an array of Slices. The number of elements in the array is 'num_parts'. struct SliceParts { SliceParts(const Slice* _parts, int _num_parts) : parts(_parts), num_parts(_num_parts) { } SliceParts() : parts(nullptr), num_parts(0) {} const Slice* parts; int num_parts; }; inline bool operator==(const Slice& x, const Slice& y) { return ((x.size() == y.size()) && (memcmp(x.data(), y.data(), x.size()) == 0)); } inline bool operator!=(const Slice& x, const Slice& y) { return !(x == y); } inline int Slice::compare(const Slice& b) const { const size_t min_len = (size_ < b.size_) ? size_ : b.size_; int r = memcmp(data_, b.data_, min_len); if (r == 0) { if (size_ < b.size_) r = -1; else if (size_ > b.size_) r = +1; } return r; } inline size_t Slice::difference_offset(const Slice& b) const { size_t off = 0; const size_t len = (size_ < b.size_) ? size_ : b.size_; for (; off < len; off++) { if (data_[off] != b.data_[off]) break; } return off; } } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_SLICE_H_ rocksdb-4.5.1/include/rocksdb/slice_transform.h000066400000000000000000000062441267530634000216010ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2012 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Class for specifying user-defined functions which perform a // transformation on a slice. It is not required that every slice // belong to the domain and/or range of a function. Subclasses should // define InDomain and InRange to determine which slices are in either // of these sets respectively. #ifndef STORAGE_ROCKSDB_INCLUDE_SLICE_TRANSFORM_H_ #define STORAGE_ROCKSDB_INCLUDE_SLICE_TRANSFORM_H_ #include namespace rocksdb { class Slice; class SliceTransform { public: virtual ~SliceTransform() {}; // Return the name of this transformation. virtual const char* Name() const = 0; // transform a src in domain to a dst in the range virtual Slice Transform(const Slice& src) const = 0; // determine whether this is a valid src upon the function applies virtual bool InDomain(const Slice& src) const = 0; // determine whether dst=Transform(src) for some src virtual bool InRange(const Slice& dst) const = 0; // Transform(s)=Transform(`prefix`) for any s with `prefix` as a prefix. // // This function is not used by RocksDB, but for users. If users pass // Options by string to RocksDB, they might not know what prefix extractor // they are using. This function is to help users can determine: // if they want to iterate all keys prefixing `prefix`, whetherit is // safe to use prefix bloom filter and seek to key `prefix`. // If this function returns true, this means a user can Seek() to a prefix // using the bloom filter. Otherwise, user needs to skip the bloom filter // by setting ReadOptions.total_order_seek = true. // // Here is an example: Suppose we implement a slice transform that returns // the first part of the string after spliting it using deimiter ",": // 1. SameResultWhenAppended("abc,") should return true. If aplying prefix // bloom filter using it, all slices matching "abc:.*" will be extracted // to "abc,", so any SST file or memtable containing any of those key // will not be filtered out. // 2. SameResultWhenAppended("abc") should return false. A user will not // guaranteed to see all the keys matching "abc.*" if a user seek to "abc" // against a DB with the same setting. If one SST file only contains // "abcd,e", the file can be filtered out and the key will be invisible. // // i.e., an implementation always returning false is safe. virtual bool SameResultWhenAppended(const Slice& prefix) const { return false; } }; extern const SliceTransform* NewFixedPrefixTransform(size_t prefix_len); extern const SliceTransform* NewCappedPrefixTransform(size_t cap_len); extern const SliceTransform* NewNoopTransform(); } #endif // STORAGE_ROCKSDB_INCLUDE_SLICE_TRANSFORM_H_ rocksdb-4.5.1/include/rocksdb/snapshot.h000066400000000000000000000024221267530634000202400ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include "rocksdb/types.h" namespace rocksdb { class DB; // Abstract handle to particular state of a DB. // A Snapshot is an immutable object and can therefore be safely // accessed from multiple threads without any external synchronization. // // To Create a Snapshot, call DB::GetSnapshot(). // To Destroy a Snapshot, call DB::ReleaseSnapshot(snapshot). class Snapshot { public: // returns Snapshot's sequence number virtual SequenceNumber GetSequenceNumber() const = 0; protected: virtual ~Snapshot(); }; // Simple RAII wrapper class for Snapshot. // Constructing this object will create a snapshot. Destructing will // release the snapshot. class ManagedSnapshot { public: explicit ManagedSnapshot(DB* db); // Instead of creating a snapshot, take ownership of the input snapshot. ManagedSnapshot(DB* db, const Snapshot* _snapshot); ~ManagedSnapshot(); const Snapshot* snapshot(); private: DB* db_; const Snapshot* snapshot_; }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/sst_dump_tool.h000066400000000000000000000007141267530634000212760ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #pragma once namespace rocksdb { class SSTDumpTool { public: int Run(int argc, char** argv); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/sst_file_manager.h000066400000000000000000000050301267530634000217010ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include "rocksdb/status.h" namespace rocksdb { class Env; class Logger; // SstFileManager is used to track SST files in the DB and control there // deletion rate. // All SstFileManager public functions are thread-safe. class SstFileManager { public: virtual ~SstFileManager() {} // Return the total size of all tracked files. // thread-safe virtual uint64_t GetTotalSize() = 0; // Return a map containing all tracked files and there corresponding sizes. // thread-safe virtual std::unordered_map GetTrackedFiles() = 0; // Return delete rate limit in bytes per second. // thread-safe virtual int64_t GetDeleteRateBytesPerSecond() = 0; }; // Create a new SstFileManager that can be shared among multiple RocksDB // instances to track SST file and control there deletion rate. // // @param env: Pointer to Env object, please see "rocksdb/env.h". // @param info_log: If not nullptr, info_log will be used to log errors. // // == Deletion rate limiting specific arguments == // @param trash_dir: Path to the directory where deleted files will be moved // to be deleted in a background thread while applying rate limiting. If this // directory dont exist, it will be created. This directory should not be // used by any other process or any other SstFileManager, Set to "" to // disable deletion rate limiting. // @param rate_bytes_per_sec: How many bytes should be deleted per second, If // this value is set to 1024 (1 Kb / sec) and we deleted a file of size 4 Kb // in 1 second, we will wait for another 3 seconds before we delete other // files, Set to 0 to disable deletion rate limiting. // @param delete_exisitng_trash: If set to true, the newly created // SstFileManager will delete files that already exist in trash_dir. // @param status: If not nullptr, status will contain any errors that happened // during creating the missing trash_dir or deleting existing files in trash. extern SstFileManager* NewSstFileManager( Env* env, std::shared_ptr info_log = nullptr, std::string trash_dir = "", int64_t rate_bytes_per_sec = 0, bool delete_exisitng_trash = true, Status* status = nullptr); } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/sst_file_writer.h000066400000000000000000000054641267530634000216160ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include "rocksdb/env.h" #include "rocksdb/immutable_options.h" #include "rocksdb/types.h" namespace rocksdb { class Comparator; // Table Properties that are specific to tables created by SstFileWriter. struct ExternalSstFilePropertyNames { // value of this property is a fixed int32 number. static const std::string kVersion; }; // ExternalSstFileInfo include information about sst files created // using SstFileWriter struct ExternalSstFileInfo { ExternalSstFileInfo() {} ExternalSstFileInfo(const std::string& _file_path, const std::string& _smallest_key, const std::string& _largest_key, SequenceNumber _sequence_number, uint64_t _file_size, int32_t _num_entries, int32_t _version) : file_path(_file_path), smallest_key(_smallest_key), largest_key(_largest_key), sequence_number(_sequence_number), file_size(_file_size), num_entries(_num_entries), version(_version) {} std::string file_path; // external sst file path std::string smallest_key; // smallest user key in file std::string largest_key; // largest user key in file SequenceNumber sequence_number; // sequence number of all keys in file uint64_t file_size; // file size in bytes uint64_t num_entries; // number of entries in file int32_t version; // file version }; // SstFileWriter is used to create sst files that can be added to database later // All keys in files generated by SstFileWriter will have sequence number = 0 class SstFileWriter { public: SstFileWriter(const EnvOptions& env_options, const ImmutableCFOptions& ioptions, const Comparator* user_comparator); ~SstFileWriter(); // Prepare SstFileWriter to write into file located at "file_path". Status Open(const std::string& file_path); // Add key, value to currently opened file // REQUIRES: key is after any previously added key according to comparator. Status Add(const Slice& user_key, const Slice& value); // Finalize writing to sst file and close file. // // An optional ExternalSstFileInfo pointer can be passed to the function // which will be populated with information about the created sst file Status Finish(ExternalSstFileInfo* file_info = nullptr); private: class SstFileWriterPropertiesCollectorFactory; class SstFileWriterPropertiesCollector; struct Rep; Rep* rep_; }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/statistics.h000066400000000000000000000352411267530634000206000ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef STORAGE_ROCKSDB_INCLUDE_STATISTICS_H_ #define STORAGE_ROCKSDB_INCLUDE_STATISTICS_H_ #include #include #include #include #include #include namespace rocksdb { /** * Keep adding ticker's here. * 1. Any ticker should be added before TICKER_ENUM_MAX. * 2. Add a readable string in TickersNameMap below for the newly added ticker. */ enum Tickers : uint32_t { // total block cache misses // REQUIRES: BLOCK_CACHE_MISS == BLOCK_CACHE_INDEX_MISS + // BLOCK_CACHE_FILTER_MISS + // BLOCK_CACHE_DATA_MISS; BLOCK_CACHE_MISS = 0, // total block cache hit // REQUIRES: BLOCK_CACHE_HIT == BLOCK_CACHE_INDEX_HIT + // BLOCK_CACHE_FILTER_HIT + // BLOCK_CACHE_DATA_HIT; BLOCK_CACHE_HIT, // # of blocks added to block cache. BLOCK_CACHE_ADD, // # of times cache miss when accessing index block from block cache. BLOCK_CACHE_INDEX_MISS, // # of times cache hit when accessing index block from block cache. BLOCK_CACHE_INDEX_HIT, // # of times cache miss when accessing filter block from block cache. BLOCK_CACHE_FILTER_MISS, // # of times cache hit when accessing filter block from block cache. BLOCK_CACHE_FILTER_HIT, // # of times cache miss when accessing data block from block cache. BLOCK_CACHE_DATA_MISS, // # of times cache hit when accessing data block from block cache. BLOCK_CACHE_DATA_HIT, // # of bytes read from cache. BLOCK_CACHE_BYTES_READ, // # of bytes written into cache. BLOCK_CACHE_BYTES_WRITE, // # of times bloom filter has avoided file reads. BLOOM_FILTER_USEFUL, // # of memtable hits. MEMTABLE_HIT, // # of memtable misses. MEMTABLE_MISS, // # of Get() queries served by L0 GET_HIT_L0, // # of Get() queries served by L1 GET_HIT_L1, // # of Get() queries served by L2 and up GET_HIT_L2_AND_UP, /** * COMPACTION_KEY_DROP_* count the reasons for key drop during compaction * There are 3 reasons currently. */ COMPACTION_KEY_DROP_NEWER_ENTRY, // key was written with a newer value. COMPACTION_KEY_DROP_OBSOLETE, // The key is obsolete. COMPACTION_KEY_DROP_USER, // user compaction function has dropped the key. // Number of keys written to the database via the Put and Write call's NUMBER_KEYS_WRITTEN, // Number of Keys read, NUMBER_KEYS_READ, // Number keys updated, if inplace update is enabled NUMBER_KEYS_UPDATED, // The number of uncompressed bytes issued by DB::Put(), DB::Delete(), // DB::Merge(), and DB::Write(). BYTES_WRITTEN, // The number of uncompressed bytes read from DB::Get(). It could be // either from memtables, cache, or table files. // For the number of logical bytes read from DB::MultiGet(), // please use NUMBER_MULTIGET_BYTES_READ. BYTES_READ, // The number of calls to seek/next/prev NUMBER_DB_SEEK, NUMBER_DB_NEXT, NUMBER_DB_PREV, // The number of calls to seek/next/prev that returned data NUMBER_DB_SEEK_FOUND, NUMBER_DB_NEXT_FOUND, NUMBER_DB_PREV_FOUND, // The number of uncompressed bytes read from an iterator. // Includes size of key and value. ITER_BYTES_READ, NO_FILE_CLOSES, NO_FILE_OPENS, NO_FILE_ERRORS, // DEPRECATED Time system had to wait to do LO-L1 compactions STALL_L0_SLOWDOWN_MICROS, // DEPRECATED Time system had to wait to move memtable to L1. STALL_MEMTABLE_COMPACTION_MICROS, // DEPRECATED write throttle because of too many files in L0 STALL_L0_NUM_FILES_MICROS, // Writer has to wait for compaction or flush to finish. STALL_MICROS, // The wait time for db mutex. // Disabled by default. To enable it set stats level to kAll DB_MUTEX_WAIT_MICROS, RATE_LIMIT_DELAY_MILLIS, NO_ITERATORS, // number of iterators currently open // Number of MultiGet calls, keys read, and bytes read NUMBER_MULTIGET_CALLS, NUMBER_MULTIGET_KEYS_READ, NUMBER_MULTIGET_BYTES_READ, // Number of deletes records that were not required to be // written to storage because key does not exist NUMBER_FILTERED_DELETES, NUMBER_MERGE_FAILURES, SEQUENCE_NUMBER, // number of times bloom was checked before creating iterator on a // file, and the number of times the check was useful in avoiding // iterator creation (and thus likely IOPs). BLOOM_FILTER_PREFIX_CHECKED, BLOOM_FILTER_PREFIX_USEFUL, // Number of times we had to reseek inside an iteration to skip // over large number of keys with same userkey. NUMBER_OF_RESEEKS_IN_ITERATION, // Record the number of calls to GetUpadtesSince. Useful to keep track of // transaction log iterator refreshes GET_UPDATES_SINCE_CALLS, BLOCK_CACHE_COMPRESSED_MISS, // miss in the compressed block cache BLOCK_CACHE_COMPRESSED_HIT, // hit in the compressed block cache WAL_FILE_SYNCED, // Number of times WAL sync is done WAL_FILE_BYTES, // Number of bytes written to WAL // Writes can be processed by requesting thread or by the thread at the // head of the writers queue. WRITE_DONE_BY_SELF, WRITE_DONE_BY_OTHER, // Equivalent to writes done for others WRITE_TIMEDOUT, // Number of writes ending up with timed-out. WRITE_WITH_WAL, // Number of Write calls that request WAL COMPACT_READ_BYTES, // Bytes read during compaction COMPACT_WRITE_BYTES, // Bytes written during compaction FLUSH_WRITE_BYTES, // Bytes written during flush // Number of table's properties loaded directly from file, without creating // table reader object. NUMBER_DIRECT_LOAD_TABLE_PROPERTIES, NUMBER_SUPERVERSION_ACQUIRES, NUMBER_SUPERVERSION_RELEASES, NUMBER_SUPERVERSION_CLEANUPS, NUMBER_BLOCK_NOT_COMPRESSED, MERGE_OPERATION_TOTAL_TIME, FILTER_OPERATION_TOTAL_TIME, // Row cache. ROW_CACHE_HIT, ROW_CACHE_MISS, TICKER_ENUM_MAX }; // The order of items listed in Tickers should be the same as // the order listed in TickersNameMap const std::vector> TickersNameMap = { {BLOCK_CACHE_MISS, "rocksdb.block.cache.miss"}, {BLOCK_CACHE_HIT, "rocksdb.block.cache.hit"}, {BLOCK_CACHE_ADD, "rocksdb.block.cache.add"}, {BLOCK_CACHE_INDEX_MISS, "rocksdb.block.cache.index.miss"}, {BLOCK_CACHE_INDEX_HIT, "rocksdb.block.cache.index.hit"}, {BLOCK_CACHE_FILTER_MISS, "rocksdb.block.cache.filter.miss"}, {BLOCK_CACHE_FILTER_HIT, "rocksdb.block.cache.filter.hit"}, {BLOCK_CACHE_DATA_MISS, "rocksdb.block.cache.data.miss"}, {BLOCK_CACHE_DATA_HIT, "rocksdb.block.cache.data.hit"}, {BLOCK_CACHE_BYTES_READ, "rocksdb.block.cache.bytes.read"}, {BLOCK_CACHE_BYTES_WRITE, "rocksdb.block.cache.bytes.write"}, {BLOOM_FILTER_USEFUL, "rocksdb.bloom.filter.useful"}, {MEMTABLE_HIT, "rocksdb.memtable.hit"}, {MEMTABLE_MISS, "rocksdb.memtable.miss"}, {GET_HIT_L0, "rocksdb.l0.hit"}, {GET_HIT_L1, "rocksdb.l1.hit"}, {GET_HIT_L2_AND_UP, "rocksdb.l2andup.hit"}, {COMPACTION_KEY_DROP_NEWER_ENTRY, "rocksdb.compaction.key.drop.new"}, {COMPACTION_KEY_DROP_OBSOLETE, "rocksdb.compaction.key.drop.obsolete"}, {COMPACTION_KEY_DROP_USER, "rocksdb.compaction.key.drop.user"}, {NUMBER_KEYS_WRITTEN, "rocksdb.number.keys.written"}, {NUMBER_KEYS_READ, "rocksdb.number.keys.read"}, {NUMBER_KEYS_UPDATED, "rocksdb.number.keys.updated"}, {BYTES_WRITTEN, "rocksdb.bytes.written"}, {BYTES_READ, "rocksdb.bytes.read"}, {NUMBER_DB_SEEK, "rocksdb.number.db.seek"}, {NUMBER_DB_NEXT, "rocksdb.number.db.next"}, {NUMBER_DB_PREV, "rocksdb.number.db.prev"}, {NUMBER_DB_SEEK_FOUND, "rocksdb.number.db.seek.found"}, {NUMBER_DB_NEXT_FOUND, "rocksdb.number.db.next.found"}, {NUMBER_DB_PREV_FOUND, "rocksdb.number.db.prev.found"}, {ITER_BYTES_READ, "rocksdb.db.iter.bytes.read"}, {NO_FILE_CLOSES, "rocksdb.no.file.closes"}, {NO_FILE_OPENS, "rocksdb.no.file.opens"}, {NO_FILE_ERRORS, "rocksdb.no.file.errors"}, {STALL_L0_SLOWDOWN_MICROS, "rocksdb.l0.slowdown.micros"}, {STALL_MEMTABLE_COMPACTION_MICROS, "rocksdb.memtable.compaction.micros"}, {STALL_L0_NUM_FILES_MICROS, "rocksdb.l0.num.files.stall.micros"}, {STALL_MICROS, "rocksdb.stall.micros"}, {DB_MUTEX_WAIT_MICROS, "rocksdb.db.mutex.wait.micros"}, {RATE_LIMIT_DELAY_MILLIS, "rocksdb.rate.limit.delay.millis"}, {NO_ITERATORS, "rocksdb.num.iterators"}, {NUMBER_MULTIGET_CALLS, "rocksdb.number.multiget.get"}, {NUMBER_MULTIGET_KEYS_READ, "rocksdb.number.multiget.keys.read"}, {NUMBER_MULTIGET_BYTES_READ, "rocksdb.number.multiget.bytes.read"}, {NUMBER_FILTERED_DELETES, "rocksdb.number.deletes.filtered"}, {NUMBER_MERGE_FAILURES, "rocksdb.number.merge.failures"}, {SEQUENCE_NUMBER, "rocksdb.sequence.number"}, {BLOOM_FILTER_PREFIX_CHECKED, "rocksdb.bloom.filter.prefix.checked"}, {BLOOM_FILTER_PREFIX_USEFUL, "rocksdb.bloom.filter.prefix.useful"}, {NUMBER_OF_RESEEKS_IN_ITERATION, "rocksdb.number.reseeks.iteration"}, {GET_UPDATES_SINCE_CALLS, "rocksdb.getupdatessince.calls"}, {BLOCK_CACHE_COMPRESSED_MISS, "rocksdb.block.cachecompressed.miss"}, {BLOCK_CACHE_COMPRESSED_HIT, "rocksdb.block.cachecompressed.hit"}, {WAL_FILE_SYNCED, "rocksdb.wal.synced"}, {WAL_FILE_BYTES, "rocksdb.wal.bytes"}, {WRITE_DONE_BY_SELF, "rocksdb.write.self"}, {WRITE_DONE_BY_OTHER, "rocksdb.write.other"}, {WRITE_WITH_WAL, "rocksdb.write.wal"}, {FLUSH_WRITE_BYTES, "rocksdb.flush.write.bytes"}, {COMPACT_READ_BYTES, "rocksdb.compact.read.bytes"}, {COMPACT_WRITE_BYTES, "rocksdb.compact.write.bytes"}, {NUMBER_DIRECT_LOAD_TABLE_PROPERTIES, "rocksdb.number.direct.load.table.properties"}, {NUMBER_SUPERVERSION_ACQUIRES, "rocksdb.number.superversion_acquires"}, {NUMBER_SUPERVERSION_RELEASES, "rocksdb.number.superversion_releases"}, {NUMBER_SUPERVERSION_CLEANUPS, "rocksdb.number.superversion_cleanups"}, {NUMBER_BLOCK_NOT_COMPRESSED, "rocksdb.number.block.not_compressed"}, {MERGE_OPERATION_TOTAL_TIME, "rocksdb.merge.operation.time.nanos"}, {FILTER_OPERATION_TOTAL_TIME, "rocksdb.filter.operation.time.nanos"}, {ROW_CACHE_HIT, "rocksdb.row.cache.hit"}, {ROW_CACHE_MISS, "rocksdb.row.cache.miss"}, }; /** * Keep adding histogram's here. * Any histogram whould have value less than HISTOGRAM_ENUM_MAX * Add a new Histogram by assigning it the current value of HISTOGRAM_ENUM_MAX * Add a string representation in HistogramsNameMap below * And increment HISTOGRAM_ENUM_MAX */ enum Histograms : uint32_t { DB_GET = 0, DB_WRITE, COMPACTION_TIME, SUBCOMPACTION_SETUP_TIME, TABLE_SYNC_MICROS, COMPACTION_OUTFILE_SYNC_MICROS, WAL_FILE_SYNC_MICROS, MANIFEST_FILE_SYNC_MICROS, // TIME SPENT IN IO DURING TABLE OPEN TABLE_OPEN_IO_MICROS, DB_MULTIGET, READ_BLOCK_COMPACTION_MICROS, READ_BLOCK_GET_MICROS, WRITE_RAW_BLOCK_MICROS, STALL_L0_SLOWDOWN_COUNT, STALL_MEMTABLE_COMPACTION_COUNT, STALL_L0_NUM_FILES_COUNT, HARD_RATE_LIMIT_DELAY_COUNT, SOFT_RATE_LIMIT_DELAY_COUNT, NUM_FILES_IN_SINGLE_COMPACTION, DB_SEEK, WRITE_STALL, SST_READ_MICROS, // The number of subcompactions actually scheduled during a compaction NUM_SUBCOMPACTIONS_SCHEDULED, // Value size distribution in each operation BYTES_PER_READ, BYTES_PER_WRITE, BYTES_PER_MULTIGET, HISTOGRAM_ENUM_MAX, // TODO(ldemailly): enforce HistogramsNameMap match }; const std::vector> HistogramsNameMap = { {DB_GET, "rocksdb.db.get.micros"}, {DB_WRITE, "rocksdb.db.write.micros"}, {COMPACTION_TIME, "rocksdb.compaction.times.micros"}, {SUBCOMPACTION_SETUP_TIME, "rocksdb.subcompaction.setup.times.micros"}, {TABLE_SYNC_MICROS, "rocksdb.table.sync.micros"}, {COMPACTION_OUTFILE_SYNC_MICROS, "rocksdb.compaction.outfile.sync.micros"}, {WAL_FILE_SYNC_MICROS, "rocksdb.wal.file.sync.micros"}, {MANIFEST_FILE_SYNC_MICROS, "rocksdb.manifest.file.sync.micros"}, {TABLE_OPEN_IO_MICROS, "rocksdb.table.open.io.micros"}, {DB_MULTIGET, "rocksdb.db.multiget.micros"}, {READ_BLOCK_COMPACTION_MICROS, "rocksdb.read.block.compaction.micros"}, {READ_BLOCK_GET_MICROS, "rocksdb.read.block.get.micros"}, {WRITE_RAW_BLOCK_MICROS, "rocksdb.write.raw.block.micros"}, {STALL_L0_SLOWDOWN_COUNT, "rocksdb.l0.slowdown.count"}, {STALL_MEMTABLE_COMPACTION_COUNT, "rocksdb.memtable.compaction.count"}, {STALL_L0_NUM_FILES_COUNT, "rocksdb.num.files.stall.count"}, {HARD_RATE_LIMIT_DELAY_COUNT, "rocksdb.hard.rate.limit.delay.count"}, {SOFT_RATE_LIMIT_DELAY_COUNT, "rocksdb.soft.rate.limit.delay.count"}, {NUM_FILES_IN_SINGLE_COMPACTION, "rocksdb.numfiles.in.singlecompaction"}, {DB_SEEK, "rocksdb.db.seek.micros"}, {WRITE_STALL, "rocksdb.db.write.stall"}, {SST_READ_MICROS, "rocksdb.sst.read.micros"}, {NUM_SUBCOMPACTIONS_SCHEDULED, "rocksdb.num.subcompactions.scheduled"}, {BYTES_PER_READ, "rocksdb.bytes.per.read"}, {BYTES_PER_WRITE, "rocksdb.bytes.per.write"}, {BYTES_PER_MULTIGET, "rocksdb.bytes.per.multiget"}, }; struct HistogramData { double median; double percentile95; double percentile99; double average; double standard_deviation; }; enum StatsLevel { // Collect all stats except the counters requiring to get time inside the // mutex lock. kExceptTimeForMutex, // Collect all stats, including measuring duration of mutex operations. // If getting time is expensive on the platform to run, it can // reduce scalability to more threads, especialy for writes. kAll, }; // Analyze the performance of a db class Statistics { public: virtual ~Statistics() {} virtual uint64_t getTickerCount(uint32_t tickerType) const = 0; virtual void histogramData(uint32_t type, HistogramData* const data) const = 0; virtual std::string getHistogramString(uint32_t type) const { return ""; } virtual void recordTick(uint32_t tickerType, uint64_t count = 0) = 0; virtual void setTickerCount(uint32_t tickerType, uint64_t count) = 0; virtual void measureTime(uint32_t histogramType, uint64_t time) = 0; // String representation of the statistic object. virtual std::string ToString() const { // Do nothing by default return std::string("ToString(): not implemented"); } // Override this function to disable particular histogram collection virtual bool HistEnabledForType(uint32_t type) const { return type < HISTOGRAM_ENUM_MAX; } StatsLevel stats_level_ = kExceptTimeForMutex; }; // Create a concrete DBStatistics object std::shared_ptr CreateDBStatistics(); } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_STATISTICS_H_ rocksdb-4.5.1/include/rocksdb/status.h000066400000000000000000000211341267530634000177250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // A Status encapsulates the result of an operation. It may indicate success, // or it may indicate an error with an associated error message. // // Multiple threads can invoke const methods on a Status without // external synchronization, but if any of the threads may call a // non-const method, all threads accessing the same Status must use // external synchronization. #ifndef STORAGE_ROCKSDB_INCLUDE_STATUS_H_ #define STORAGE_ROCKSDB_INCLUDE_STATUS_H_ #include #include "rocksdb/slice.h" namespace rocksdb { class Status { public: // Create a success status. Status() : code_(kOk), subcode_(kNone), state_(nullptr) {} ~Status() { delete[] state_; } // Copy the specified status. Status(const Status& s); Status& operator=(const Status& s); Status(Status&& s) #if !(defined _MSC_VER) || ((defined _MSC_VER) && (_MSC_VER >= 1900)) noexcept #endif ; Status& operator=(Status&& s) #if !(defined _MSC_VER) || ((defined _MSC_VER) && (_MSC_VER >= 1900)) noexcept #endif ; bool operator==(const Status& rhs) const; bool operator!=(const Status& rhs) const; enum Code { kOk = 0, kNotFound = 1, kCorruption = 2, kNotSupported = 3, kInvalidArgument = 4, kIOError = 5, kMergeInProgress = 6, kIncomplete = 7, kShutdownInProgress = 8, kTimedOut = 9, kAborted = 10, kBusy = 11, kExpired = 12, kTryAgain = 13, }; Code code() const { return code_; } enum SubCode { kNone = 0, kMutexTimeout = 1, kLockTimeout = 2, kLockLimit = 3, kMaxSubCode }; SubCode subcode() const { return subcode_; } // Return a success status. static Status OK() { return Status(); } // Return error status of an appropriate type. static Status NotFound(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kNotFound, msg, msg2); } // Fast path for not found without malloc; static Status NotFound(SubCode msg = kNone) { return Status(kNotFound, msg); } static Status Corruption(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kCorruption, msg, msg2); } static Status Corruption(SubCode msg = kNone) { return Status(kCorruption, msg); } static Status NotSupported(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kNotSupported, msg, msg2); } static Status NotSupported(SubCode msg = kNone) { return Status(kNotSupported, msg); } static Status InvalidArgument(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kInvalidArgument, msg, msg2); } static Status InvalidArgument(SubCode msg = kNone) { return Status(kInvalidArgument, msg); } static Status IOError(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kIOError, msg, msg2); } static Status IOError(SubCode msg = kNone) { return Status(kIOError, msg); } static Status MergeInProgress(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kMergeInProgress, msg, msg2); } static Status MergeInProgress(SubCode msg = kNone) { return Status(kMergeInProgress, msg); } static Status Incomplete(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kIncomplete, msg, msg2); } static Status Incomplete(SubCode msg = kNone) { return Status(kIncomplete, msg); } static Status ShutdownInProgress(SubCode msg = kNone) { return Status(kShutdownInProgress, msg); } static Status ShutdownInProgress(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kShutdownInProgress, msg, msg2); } static Status Aborted(SubCode msg = kNone) { return Status(kAborted, msg); } static Status Aborted(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kAborted, msg, msg2); } static Status Busy(SubCode msg = kNone) { return Status(kBusy, msg); } static Status Busy(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kBusy, msg, msg2); } static Status TimedOut(SubCode msg = kNone) { return Status(kTimedOut, msg); } static Status TimedOut(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kTimedOut, msg, msg2); } static Status Expired(SubCode msg = kNone) { return Status(kExpired, msg); } static Status Expired(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kExpired, msg, msg2); } static Status TryAgain(SubCode msg = kNone) { return Status(kTryAgain, msg); } static Status TryAgain(const Slice& msg, const Slice& msg2 = Slice()) { return Status(kTryAgain, msg, msg2); } // Returns true iff the status indicates success. bool ok() const { return code() == kOk; } // Returns true iff the status indicates a NotFound error. bool IsNotFound() const { return code() == kNotFound; } // Returns true iff the status indicates a Corruption error. bool IsCorruption() const { return code() == kCorruption; } // Returns true iff the status indicates a NotSupported error. bool IsNotSupported() const { return code() == kNotSupported; } // Returns true iff the status indicates an InvalidArgument error. bool IsInvalidArgument() const { return code() == kInvalidArgument; } // Returns true iff the status indicates an IOError. bool IsIOError() const { return code() == kIOError; } // Returns true iff the status indicates an MergeInProgress. bool IsMergeInProgress() const { return code() == kMergeInProgress; } // Returns true iff the status indicates Incomplete bool IsIncomplete() const { return code() == kIncomplete; } // Returns true iff the status indicates Shutdown In progress bool IsShutdownInProgress() const { return code() == kShutdownInProgress; } bool IsTimedOut() const { return code() == kTimedOut; } bool IsAborted() const { return code() == kAborted; } // Returns true iff the status indicates that a resource is Busy and // temporarily could not be acquired. bool IsBusy() const { return code() == kBusy; } // Returns true iff the status indicated that the operation has Expired. bool IsExpired() const { return code() == kExpired; } // Returns true iff the status indicates a TryAgain error. // This usually means that the operation failed, but may succeed if // re-attempted. bool IsTryAgain() const { return code() == kTryAgain; } // Return a string representation of this status suitable for printing. // Returns the string "OK" for success. std::string ToString() const; private: // A nullptr state_ (which is always the case for OK) means the message // is empty. // of the following form: // state_[0..3] == length of message // state_[4..] == message Code code_; SubCode subcode_; const char* state_; static const char* msgs[static_cast(kMaxSubCode)]; explicit Status(Code _code, SubCode _subcode = kNone) : code_(_code), subcode_(_subcode), state_(nullptr) {} Status(Code _code, const Slice& msg, const Slice& msg2); static const char* CopyState(const char* s); }; inline Status::Status(const Status& s) : code_(s.code_), subcode_(s.subcode_) { state_ = (s.state_ == nullptr) ? nullptr : CopyState(s.state_); } inline Status& Status::operator=(const Status& s) { // The following condition catches both aliasing (when this == &s), // and the common case where both s and *this are ok. if(this != &s) { code_ = s.code_; subcode_ = s.subcode_; delete[] state_; state_ = (s.state_ == nullptr) ? nullptr : CopyState(s.state_); } return *this; } inline Status::Status(Status&& s) #if !(defined _MSC_VER) || ((defined _MSC_VER) && (_MSC_VER >= 1900)) noexcept #endif : Status() { *this = std::move(s); } inline Status& Status::operator=(Status&& s) #if !(defined _MSC_VER) || ((defined _MSC_VER) && (_MSC_VER >= 1900)) noexcept #endif { if(this != &s) { code_ = std::move(s.code_); s.code_ = kOk; subcode_ = std::move(s.subcode_); s.subcode_ = kNone; delete [] state_; state_ = nullptr; std::swap(state_, s.state_); } return *this; } inline bool Status::operator==(const Status& rhs) const { return (code_ == rhs.code_); } inline bool Status::operator!=(const Status& rhs) const { return !(*this == rhs); } } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_STATUS_H_ rocksdb-4.5.1/include/rocksdb/table.h000066400000000000000000000472241267530634000175010ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Currently we support two types of tables: plain table and block-based table. // 1. Block-based table: this is the default table type that we inherited from // LevelDB, which was designed for storing data in hard disk or flash // device. // 2. Plain table: it is one of RocksDB's SST file format optimized // for low query latency on pure-memory or really low-latency media. // // A tutorial of rocksdb table formats is available here: // https://github.com/facebook/rocksdb/wiki/A-Tutorial-of-RocksDB-SST-formats // // Example code is also available // https://github.com/facebook/rocksdb/wiki/A-Tutorial-of-RocksDB-SST-formats#wiki-examples #pragma once #include #include #include #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "rocksdb/immutable_options.h" #include "rocksdb/status.h" namespace rocksdb { // -- Block-based Table class FlushBlockPolicyFactory; class RandomAccessFile; struct TableReaderOptions; struct TableBuilderOptions; class TableBuilder; class TableReader; class WritableFileWriter; struct EnvOptions; struct Options; using std::unique_ptr; enum ChecksumType : char { kNoChecksum = 0x0, // not yet supported. Will fail kCRC32c = 0x1, kxxHash = 0x2, }; // For advanced user only struct BlockBasedTableOptions { // @flush_block_policy_factory creates the instances of flush block policy. // which provides a configurable way to determine when to flush a block in // the block based tables. If not set, table builder will use the default // block flush policy, which cut blocks by block size (please refer to // `FlushBlockBySizePolicy`). std::shared_ptr flush_block_policy_factory; // TODO(kailiu) Temporarily disable this feature by making the default value // to be false. // // Indicating if we'd put index/filter blocks to the block cache. // If not specified, each "table reader" object will pre-load index/filter // block during table initialization. bool cache_index_and_filter_blocks = false; // The index type that will be used for this table. enum IndexType : char { // A space efficient index block that is optimized for // binary-search-based index. kBinarySearch, // The hash index, if enabled, will do the hash lookup when // `Options.prefix_extractor` is provided. kHashSearch, }; IndexType index_type = kBinarySearch; // Influence the behavior when kHashSearch is used. // if false, stores a precise prefix to block range mapping // if true, does not store prefix and allows prefix hash collision // (less memory consumption) bool hash_index_allow_collision = true; // Use the specified checksum type. Newly created table files will be // protected with this checksum type. Old table files will still be readable, // even though they have different checksum type. ChecksumType checksum = kCRC32c; // Disable block cache. If this is set to true, // then no block cache should be used, and the block_cache should // point to a nullptr object. bool no_block_cache = false; // If non-NULL use the specified cache for blocks. // If NULL, rocksdb will automatically create and use an 8MB internal cache. std::shared_ptr block_cache = nullptr; // If non-NULL use the specified cache for compressed blocks. // If NULL, rocksdb will not use a compressed block cache. std::shared_ptr block_cache_compressed = nullptr; // Approximate size of user data packed per block. Note that the // block size specified here corresponds to uncompressed data. The // actual size of the unit read from disk may be smaller if // compression is enabled. This parameter can be changed dynamically. size_t block_size = 4 * 1024; // This is used to close a block before it reaches the configured // 'block_size'. If the percentage of free space in the current block is less // than this specified number and adding a new record to the block will // exceed the configured block size, then this block will be closed and the // new record will be written to the next block. int block_size_deviation = 10; // Number of keys between restart points for delta encoding of keys. // This parameter can be changed dynamically. Most clients should // leave this parameter alone. The minimum value allowed is 1. Any smaller // value will be silently overwritten with 1. int block_restart_interval = 16; // Same as block_restart_interval but used for the index block. int index_block_restart_interval = 1; // Use delta encoding to compress keys in blocks. // Iterator::PinData() requires this option to be disabled. // // Default: true bool use_delta_encoding = true; // If non-nullptr, use the specified filter policy to reduce disk reads. // Many applications will benefit from passing the result of // NewBloomFilterPolicy() here. std::shared_ptr filter_policy = nullptr; // If true, place whole keys in the filter (not just prefixes). // This must generally be true for gets to be efficient. bool whole_key_filtering = true; // If true, block will not be explicitly flushed to disk during building // a SstTable. Instead, buffer in WritableFileWriter will take // care of the flushing when it is full. // // On Windows, this option helps a lot when unbuffered I/O // (allow_os_buffer = false) is used, since it avoids small // unbuffered disk write. // // User may also adjust writable_file_max_buffer_size to optimize disk I/O // size. // // Default: false bool skip_table_builder_flush = false; // We currently have three versions: // 0 -- This version is currently written out by all RocksDB's versions by // default. Can be read by really old RocksDB's. Doesn't support changing // checksum (default is CRC32). // 1 -- Can be read by RocksDB's versions since 3.0. Supports non-default // checksum, like xxHash. It is written by RocksDB when // BlockBasedTableOptions::checksum is something other than kCRC32c. (version // 0 is silently upconverted) // 2 -- Can be read by RocksDB's versions since 3.10. Changes the way we // encode compressed blocks with LZ4, BZip2 and Zlib compression. If you // don't plan to run RocksDB before version 3.10, you should probably use // this. // This option only affects newly written tables. When reading exising tables, // the information about version is read from the footer. uint32_t format_version = 0; }; // Table Properties that are specific to block-based table properties. struct BlockBasedTablePropertyNames { // value of this propertis is a fixed int32 number. static const std::string kIndexType; // value is "1" for true and "0" for false. static const std::string kWholeKeyFiltering; // value is "1" for true and "0" for false. static const std::string kPrefixFiltering; }; // Create default block based table factory. extern TableFactory* NewBlockBasedTableFactory( const BlockBasedTableOptions& table_options = BlockBasedTableOptions()); #ifndef ROCKSDB_LITE enum EncodingType : char { // Always write full keys without any special encoding. kPlain, // Find opportunity to write the same prefix once for multiple rows. // In some cases, when a key follows a previous key with the same prefix, // instead of writing out the full key, it just writes out the size of the // shared prefix, as well as other bytes, to save some bytes. // // When using this option, the user is required to use the same prefix // extractor to make sure the same prefix will be extracted from the same key. // The Name() value of the prefix extractor will be stored in the file. When // reopening the file, the name of the options.prefix_extractor given will be // bitwise compared to the prefix extractors stored in the file. An error // will be returned if the two don't match. kPrefix, }; // Table Properties that are specific to plain table properties. struct PlainTablePropertyNames { static const std::string kPrefixExtractorName; static const std::string kEncodingType; static const std::string kBloomVersion; static const std::string kNumBloomBlocks; }; const uint32_t kPlainTableVariableLength = 0; struct PlainTableOptions { // @user_key_len: plain table has optimization for fix-sized keys, which can // be specified via user_key_len. Alternatively, you can pass // `kPlainTableVariableLength` if your keys have variable // lengths. uint32_t user_key_len = kPlainTableVariableLength; // @bloom_bits_per_key: the number of bits used for bloom filer per prefix. // You may disable it by passing a zero. int bloom_bits_per_key = 10; // @hash_table_ratio: the desired utilization of the hash table used for // prefix hashing. // hash_table_ratio = number of prefixes / #buckets in the // hash table double hash_table_ratio = 0.75; // @index_sparseness: inside each prefix, need to build one index record for // how many keys for binary search inside each hash bucket. // For encoding type kPrefix, the value will be used when // writing to determine an interval to rewrite the full // key. It will also be used as a suggestion and satisfied // when possible. size_t index_sparseness = 16; // @huge_page_tlb_size: if <=0, allocate hash indexes and blooms from malloc. // Otherwise from huge page TLB. The user needs to // reserve huge pages for it to be allocated, like: // sysctl -w vm.nr_hugepages=20 // See linux doc Documentation/vm/hugetlbpage.txt size_t huge_page_tlb_size = 0; // @encoding_type: how to encode the keys. See enum EncodingType above for // the choices. The value will determine how to encode keys // when writing to a new SST file. This value will be stored // inside the SST file which will be used when reading from // the file, which makes it possible for users to choose // different encoding type when reopening a DB. Files with // different encoding types can co-exist in the same DB and // can be read. EncodingType encoding_type = kPlain; // @full_scan_mode: mode for reading the whole file one record by one without // using the index. bool full_scan_mode = false; // @store_index_in_file: compute plain table index and bloom filter during // file building and store it in file. When reading // file, index will be mmaped instead of recomputation. bool store_index_in_file = false; }; // -- Plain Table with prefix-only seek // For this factory, you need to set Options.prefix_extrator properly to make it // work. Look-up will starts with prefix hash lookup for key prefix. Inside the // hash bucket found, a binary search is executed for hash conflicts. Finally, // a linear search is used. extern TableFactory* NewPlainTableFactory(const PlainTableOptions& options = PlainTableOptions()); struct CuckooTablePropertyNames { // The key that is used to fill empty buckets. static const std::string kEmptyKey; // Fixed length of value. static const std::string kValueLength; // Number of hash functions used in Cuckoo Hash. static const std::string kNumHashFunc; // It denotes the number of buckets in a Cuckoo Block. Given a key and a // particular hash function, a Cuckoo Block is a set of consecutive buckets, // where starting bucket id is given by the hash function on the key. In case // of a collision during inserting the key, the builder tries to insert the // key in other locations of the cuckoo block before using the next hash // function. This reduces cache miss during read operation in case of // collision. static const std::string kCuckooBlockSize; // Size of the hash table. Use this number to compute the modulo of hash // function. The actual number of buckets will be kMaxHashTableSize + // kCuckooBlockSize - 1. The last kCuckooBlockSize-1 buckets are used to // accommodate the Cuckoo Block from end of hash table, due to cache friendly // implementation. static const std::string kHashTableSize; // Denotes if the key sorted in the file is Internal Key (if false) // or User Key only (if true). static const std::string kIsLastLevel; // Indicate if using identity function for the first hash function. static const std::string kIdentityAsFirstHash; // Indicate if using module or bit and to calculate hash value static const std::string kUseModuleHash; // Fixed user key length static const std::string kUserKeyLength; }; struct CuckooTableOptions { // Determines the utilization of hash tables. Smaller values // result in larger hash tables with fewer collisions. double hash_table_ratio = 0.9; // A property used by builder to determine the depth to go to // to search for a path to displace elements in case of // collision. See Builder.MakeSpaceForKey method. Higher // values result in more efficient hash tables with fewer // lookups but take more time to build. uint32_t max_search_depth = 100; // In case of collision while inserting, the builder // attempts to insert in the next cuckoo_block_size // locations before skipping over to the next Cuckoo hash // function. This makes lookups more cache friendly in case // of collisions. uint32_t cuckoo_block_size = 5; // If this option is enabled, user key is treated as uint64_t and its value // is used as hash value directly. This option changes builder's behavior. // Reader ignore this option and behave according to what specified in table // property. bool identity_as_first_hash = false; // If this option is set to true, module is used during hash calculation. // This often yields better space efficiency at the cost of performance. // If this optino is set to false, # of entries in table is constrained to be // power of two, and bit and is used to calculate hash, which is faster in // general. bool use_module_hash = true; }; // Cuckoo Table Factory for SST table format using Cache Friendly Cuckoo Hashing extern TableFactory* NewCuckooTableFactory( const CuckooTableOptions& table_options = CuckooTableOptions()); #endif // ROCKSDB_LITE class RandomAccessFileReader; // A base class for table factories. class TableFactory { public: virtual ~TableFactory() {} // The type of the table. // // The client of this package should switch to a new name whenever // the table format implementation changes. // // Names starting with "rocksdb." are reserved and should not be used // by any clients of this package. virtual const char* Name() const = 0; // Returns a Table object table that can fetch data from file specified // in parameter file. It's the caller's responsibility to make sure // file is in the correct format. // // NewTableReader() is called in three places: // (1) TableCache::FindTable() calls the function when table cache miss // and cache the table object returned. // (2) SstFileReader (for SST Dump) opens the table and dump the table // contents using the interator of the table. // (3) DBImpl::AddFile() calls this function to read the contents of // the sst file it's attempting to add // // table_reader_options is a TableReaderOptions which contain all the // needed parameters and configuration to open the table. // file is a file handler to handle the file for the table. // file_size is the physical file size of the file. // table_reader is the output table reader. virtual Status NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader) const = 0; // Return a table builder to write to a file for this table type. // // It is called in several places: // (1) When flushing memtable to a level-0 output file, it creates a table // builder (In DBImpl::WriteLevel0Table(), by calling BuildTable()) // (2) During compaction, it gets the builder for writing compaction output // files in DBImpl::OpenCompactionOutputFile(). // (3) When recovering from transaction logs, it creates a table builder to // write to a level-0 output file (In DBImpl::WriteLevel0TableForRecovery, // by calling BuildTable()) // (4) When running Repairer, it creates a table builder to convert logs to // SST files (In Repairer::ConvertLogToTable() by calling BuildTable()) // // ImmutableCFOptions is a subset of Options that can not be altered. // Multiple configured can be acceseed from there, including and not limited // to compression options. file is a handle of a writable file. // It is the caller's responsibility to keep the file open and close the file // after closing the table builder. compression_type is the compression type // to use in this table. virtual TableBuilder* NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const = 0; // Sanitizes the specified DB Options and ColumnFamilyOptions. // // If the function cannot find a way to sanitize the input DB Options, // a non-ok Status will be returned. virtual Status SanitizeOptions( const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const = 0; // Return a string that contains printable format of table configurations. // RocksDB prints configurations at DB Open(). virtual std::string GetPrintableTableOptions() const = 0; // Returns the raw pointer of the table options that is used by this // TableFactory, or nullptr if this function is not supported. // Since the return value is a raw pointer, the TableFactory owns the // pointer and the caller should not delete the pointer. // // In certan case, it is desirable to alter the underlying options when the // TableFactory is not used by any open DB by casting the returned pointer // to the right class. For instance, if BlockBasedTableFactory is used, // then the pointer can be casted to BlockBasedTableOptions. // // Note that changing the underlying TableFactory options while the // TableFactory is currently used by any open DB is undefined behavior. // Developers should use DB::SetOption() instead to dynamically change // options while the DB is open. virtual void* GetOptions() { return nullptr; } }; #ifndef ROCKSDB_LITE // Create a special table factory that can open either of the supported // table formats, based on setting inside the SST files. It should be used to // convert a DB from one table format to another. // @table_factory_to_write: the table factory used when writing to new files. // @block_based_table_factory: block based table factory to use. If NULL, use // a default one. // @plain_table_factory: plain table factory to use. If NULL, use a default one. // @cuckoo_table_factory: cuckoo table factory to use. If NULL, use a default one. extern TableFactory* NewAdaptiveTableFactory( std::shared_ptr table_factory_to_write = nullptr, std::shared_ptr block_based_table_factory = nullptr, std::shared_ptr plain_table_factory = nullptr, std::shared_ptr cuckoo_table_factory = nullptr); #endif // ROCKSDB_LITE } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/table_properties.h000066400000000000000000000140421267530634000217450ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include "rocksdb/status.h" #include "rocksdb/types.h" namespace rocksdb { // -- Table Properties // Other than basic table properties, each table may also have the user // collected properties. // The value of the user-collected properties are encoded as raw bytes -- // users have to interprete these values by themselves. // Note: To do prefix seek/scan in `UserCollectedProperties`, you can do // something similar to: // // UserCollectedProperties props = ...; // for (auto pos = props.lower_bound(prefix); // pos != props.end() && pos->first.compare(0, prefix.size(), prefix) == 0; // ++pos) { // ... // } typedef std::map UserCollectedProperties; // TableProperties contains a bunch of read-only properties of its associated // table. struct TableProperties { public: // the total size of all data blocks. uint64_t data_size = 0; // the size of index block. uint64_t index_size = 0; // the size of filter block. uint64_t filter_size = 0; // total raw key size uint64_t raw_key_size = 0; // total raw value size uint64_t raw_value_size = 0; // the number of blocks in this table uint64_t num_data_blocks = 0; // the number of entries in this table uint64_t num_entries = 0; // format version, reserved for backward compatibility uint64_t format_version = 0; // If 0, key is variable length. Otherwise number of bytes for each key. uint64_t fixed_key_len = 0; // The name of the filter policy used in this table. // If no filter policy is used, `filter_policy_name` will be an empty string. std::string filter_policy_name; // user collected properties UserCollectedProperties user_collected_properties; UserCollectedProperties readable_properties; // convert this object to a human readable form // @prop_delim: delimiter for each property. std::string ToString(const std::string& prop_delim = "; ", const std::string& kv_delim = "=") const; // Aggregate the numerical member variables of the specified // TableProperties. void Add(const TableProperties& tp); }; // table properties' human-readable names in the property block. struct TablePropertiesNames { static const std::string kDataSize; static const std::string kIndexSize; static const std::string kFilterSize; static const std::string kRawKeySize; static const std::string kRawValueSize; static const std::string kNumDataBlocks; static const std::string kNumEntries; static const std::string kFormatVersion; static const std::string kFixedKeyLen; static const std::string kFilterPolicy; }; extern const std::string kPropertiesBlock; enum EntryType { kEntryPut, kEntryDelete, kEntrySingleDelete, kEntryMerge, kEntryOther, }; // `TablePropertiesCollector` provides the mechanism for users to collect // their own properties that they are interested in. This class is essentially // a collection of callback functions that will be invoked during table // building. It is construced with TablePropertiesCollectorFactory. The methods // don't need to be thread-safe, as we will create exactly one // TablePropertiesCollector object per table and then call it sequentially class TablePropertiesCollector { public: virtual ~TablePropertiesCollector() {} // DEPRECATE User defined collector should implement AddUserKey(), though // this old function still works for backward compatible reason. // Add() will be called when a new key/value pair is inserted into the table. // @params key the user key that is inserted into the table. // @params value the value that is inserted into the table. virtual Status Add(const Slice& /*key*/, const Slice& /*value*/) { return Status::InvalidArgument( "TablePropertiesCollector::Add() deprecated."); } // AddUserKey() will be called when a new key/value pair is inserted into the // table. // @params key the user key that is inserted into the table. // @params value the value that is inserted into the table. virtual Status AddUserKey(const Slice& key, const Slice& value, EntryType /*type*/, SequenceNumber /*seq*/, uint64_t /*file_size*/) { // For backwards-compatibility. return Add(key, value); } // Finish() will be called when a table has already been built and is ready // for writing the properties block. // @params properties User will add their collected statistics to // `properties`. virtual Status Finish(UserCollectedProperties* properties) = 0; // Return the human-readable properties, where the key is property name and // the value is the human-readable form of value. virtual UserCollectedProperties GetReadableProperties() const = 0; // The name of the properties collector can be used for debugging purpose. virtual const char* Name() const = 0; // EXPERIMENTAL Return whether the output file should be further compacted virtual bool NeedCompact() const { return false; } }; // Constructs TablePropertiesCollector. Internals create a new // TablePropertiesCollector for each new table class TablePropertiesCollectorFactory { public: struct Context { uint32_t column_family_id; static const uint32_t kUnknownColumnFamily; }; virtual ~TablePropertiesCollectorFactory() {} // has to be thread-safe virtual TablePropertiesCollector* CreateTablePropertiesCollector( TablePropertiesCollectorFactory::Context context) = 0; // The name of the properties collector can be used for debugging purpose. virtual const char* Name() const = 0; }; // Extra properties // Below is a list of non-basic properties that are collected by database // itself. Especially some properties regarding to the internal keys (which // is unknown to `table`). extern uint64_t GetDeletedKeys(const UserCollectedProperties& props); } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/thread_status.h000066400000000000000000000144011267530634000212530ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file defines the structures for exposing run-time status of any // rocksdb-related thread. Such run-time status can be obtained via // GetThreadList() API. // // Note that all thread-status features are still under-development, and // thus APIs and class definitions might subject to change at this point. // Will remove this comment once the APIs have been finalized. #pragma once #include #include #include #include #include #include #ifndef ROCKSDB_USING_THREAD_STATUS #define ROCKSDB_USING_THREAD_STATUS \ !defined(ROCKSDB_LITE) && \ !defined(NROCKSDB_THREAD_STATUS) && \ !defined(OS_MACOSX) && \ !defined(IOS_CROSS_COMPILE) #endif namespace rocksdb { // TODO(yhchiang): remove this function once c++14 is available // as std::max will be able to cover this. // Current MS compiler does not support constexpr template struct constexpr_max { static const int result = (A > B) ? A : B; }; // A structure that describes the current status of a thread. // The status of active threads can be fetched using // rocksdb::GetThreadList(). struct ThreadStatus { // The type of a thread. enum ThreadType : int { HIGH_PRIORITY = 0, // RocksDB BG thread in high-pri thread pool LOW_PRIORITY, // RocksDB BG thread in low-pri thread pool USER, // User thread (Non-RocksDB BG thread) NUM_THREAD_TYPES }; // The type used to refer to a thread operation. // A thread operation describes high-level action of a thread. // Examples include compaction and flush. enum OperationType : int { OP_UNKNOWN = 0, OP_COMPACTION, OP_FLUSH, NUM_OP_TYPES }; enum OperationStage : int { STAGE_UNKNOWN = 0, STAGE_FLUSH_RUN, STAGE_FLUSH_WRITE_L0, STAGE_COMPACTION_PREPARE, STAGE_COMPACTION_RUN, STAGE_COMPACTION_PROCESS_KV, STAGE_COMPACTION_INSTALL, STAGE_COMPACTION_SYNC_FILE, STAGE_PICK_MEMTABLES_TO_FLUSH, STAGE_MEMTABLE_ROLLBACK, STAGE_MEMTABLE_INSTALL_FLUSH_RESULTS, NUM_OP_STAGES }; enum CompactionPropertyType : int { COMPACTION_JOB_ID = 0, COMPACTION_INPUT_OUTPUT_LEVEL, COMPACTION_PROP_FLAGS, COMPACTION_TOTAL_INPUT_BYTES, COMPACTION_BYTES_READ, COMPACTION_BYTES_WRITTEN, NUM_COMPACTION_PROPERTIES }; enum FlushPropertyType : int { FLUSH_JOB_ID = 0, FLUSH_BYTES_MEMTABLES, FLUSH_BYTES_WRITTEN, NUM_FLUSH_PROPERTIES }; // The maximum number of properties of an operation. // This number should be set to the biggest NUM_XXX_PROPERTIES. static const int kNumOperationProperties = constexpr_max::result; // The type used to refer to a thread state. // A state describes lower-level action of a thread // such as reading / writing a file or waiting for a mutex. enum StateType : int { STATE_UNKNOWN = 0, STATE_MUTEX_WAIT = 1, NUM_STATE_TYPES }; ThreadStatus(const uint64_t _id, const ThreadType _thread_type, const std::string& _db_name, const std::string& _cf_name, const OperationType _operation_type, const uint64_t _op_elapsed_micros, const OperationStage _operation_stage, const uint64_t _op_props[], const StateType _state_type) : thread_id(_id), thread_type(_thread_type), db_name(_db_name), cf_name(_cf_name), operation_type(_operation_type), op_elapsed_micros(_op_elapsed_micros), operation_stage(_operation_stage), state_type(_state_type) { for (int i = 0; i < kNumOperationProperties; ++i) { op_properties[i] = _op_props[i]; } } // An unique ID for the thread. const uint64_t thread_id; // The type of the thread, it could be HIGH_PRIORITY, // LOW_PRIORITY, and USER const ThreadType thread_type; // The name of the DB instance where the thread is currently // involved with. It would be set to empty string if the thread // does not involve in any DB operation. const std::string db_name; // The name of the column family where the thread is currently // It would be set to empty string if the thread does not involve // in any column family. const std::string cf_name; // The operation (high-level action) that the current thread is involved. const OperationType operation_type; // The elapsed time in micros of the current thread operation. const uint64_t op_elapsed_micros; // An integer showing the current stage where the thread is involved // in the current operation. const OperationStage operation_stage; // A list of properties that describe some details about the current // operation. Same field in op_properties[] might have different // meanings for different operations. uint64_t op_properties[kNumOperationProperties]; // The state (lower-level action) that the current thread is involved. const StateType state_type; // The followings are a set of utility functions for interpreting // the information of ThreadStatus static const std::string& GetThreadTypeName(ThreadType thread_type); // Obtain the name of an operation given its type. static const std::string& GetOperationName(OperationType op_type); static const std::string MicrosToString(uint64_t op_elapsed_time); // Obtain a human-readable string describing the specified operation stage. static const std::string& GetOperationStageName( OperationStage stage); // Obtain the name of the "i"th operation property of the // specified operation. static const std::string& GetOperationPropertyName( OperationType op_type, int i); // Translate the "i"th property of the specified operation given // a property value. static std::map InterpretOperationProperties( OperationType op_type, const uint64_t* op_properties); // Obtain the name of a state given its type. static const std::string& GetStateName(StateType state_type); }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/transaction_log.h000066400000000000000000000076221267530634000215760ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef STORAGE_ROCKSDB_INCLUDE_TRANSACTION_LOG_ITERATOR_H_ #define STORAGE_ROCKSDB_INCLUDE_TRANSACTION_LOG_ITERATOR_H_ #include "rocksdb/status.h" #include "rocksdb/types.h" #include "rocksdb/write_batch.h" #include #include namespace rocksdb { class LogFile; typedef std::vector> VectorLogPtr; enum WalFileType { /* Indicates that WAL file is in archive directory. WAL files are moved from * the main db directory to archive directory once they are not live and stay * there until cleaned up. Files are cleaned depending on archive size * (Options::WAL_size_limit_MB) and time since last cleaning * (Options::WAL_ttl_seconds). */ kArchivedLogFile = 0, /* Indicates that WAL file is live and resides in the main db directory */ kAliveLogFile = 1 } ; class LogFile { public: LogFile() {} virtual ~LogFile() {} // Returns log file's pathname relative to the main db dir // Eg. For a live-log-file = /000003.log // For an archived-log-file = /archive/000003.log virtual std::string PathName() const = 0; // Primary identifier for log file. // This is directly proportional to creation time of the log file virtual uint64_t LogNumber() const = 0; // Log file can be either alive or archived virtual WalFileType Type() const = 0; // Starting sequence number of writebatch written in this log file virtual SequenceNumber StartSequence() const = 0; // Size of log file on disk in Bytes virtual uint64_t SizeFileBytes() const = 0; }; struct BatchResult { SequenceNumber sequence = 0; std::unique_ptr writeBatchPtr; // Add empty __ctor and __dtor for the rule of five // However, preserve the original semantics and prohibit copying // as the unique_ptr member does not copy. BatchResult() {} ~BatchResult() {} BatchResult(const BatchResult&) = delete; BatchResult& operator=(const BatchResult&) = delete; BatchResult(BatchResult&& bResult) : sequence(std::move(bResult.sequence)), writeBatchPtr(std::move(bResult.writeBatchPtr)) {} BatchResult& operator=(BatchResult&& bResult) { sequence = std::move(bResult.sequence); writeBatchPtr = std::move(bResult.writeBatchPtr); return *this; } }; // A TransactionLogIterator is used to iterate over the transactions in a db. // One run of the iterator is continuous, i.e. the iterator will stop at the // beginning of any gap in sequences class TransactionLogIterator { public: TransactionLogIterator() {} virtual ~TransactionLogIterator() {} // An iterator is either positioned at a WriteBatch or not valid. // This method returns true if the iterator is valid. // Can read data from a valid iterator. virtual bool Valid() = 0; // Moves the iterator to the next WriteBatch. // REQUIRES: Valid() to be true. virtual void Next() = 0; // Returns ok if the iterator is valid. // Returns the Error when something has gone wrong. virtual Status status() = 0; // If valid return's the current write_batch and the sequence number of the // earliest transaction contained in the batch. // ONLY use if Valid() is true and status() is OK. virtual BatchResult GetBatch() = 0; // The read options for TransactionLogIterator. struct ReadOptions { // If true, all data read from underlying storage will be // verified against corresponding checksums. // Default: true bool verify_checksums_; ReadOptions() : verify_checksums_(true) {} explicit ReadOptions(bool verify_checksums) : verify_checksums_(verify_checksums) {} }; }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_TRANSACTION_LOG_ITERATOR_H_ rocksdb-4.5.1/include/rocksdb/types.h000066400000000000000000000011361267530634000175460ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef STORAGE_ROCKSDB_INCLUDE_TYPES_H_ #define STORAGE_ROCKSDB_INCLUDE_TYPES_H_ #include namespace rocksdb { // Define all public custom types here. // Represents a sequence number in a WAL file. typedef uint64_t SequenceNumber; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_TYPES_H_ rocksdb-4.5.1/include/rocksdb/universal_compaction.h000066400000000000000000000067551267530634000226420ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef STORAGE_ROCKSDB_UNIVERSAL_COMPACTION_OPTIONS_H #define STORAGE_ROCKSDB_UNIVERSAL_COMPACTION_OPTIONS_H #include #include #include namespace rocksdb { // // Algorithm used to make a compaction request stop picking new files // into a single compaction run // enum CompactionStopStyle { kCompactionStopStyleSimilarSize, // pick files of similar size kCompactionStopStyleTotalSize // total size of picked files > next file }; class CompactionOptionsUniversal { public: // Percentage flexibilty while comparing file size. If the candidate file(s) // size is 1% smaller than the next file's size, then include next file into // this candidate set. // Default: 1 unsigned int size_ratio; // The minimum number of files in a single compaction run. Default: 2 unsigned int min_merge_width; // The maximum number of files in a single compaction run. Default: UINT_MAX unsigned int max_merge_width; // The size amplification is defined as the amount (in percentage) of // additional storage needed to store a single byte of data in the database. // For example, a size amplification of 2% means that a database that // contains 100 bytes of user-data may occupy upto 102 bytes of // physical storage. By this definition, a fully compacted database has // a size amplification of 0%. Rocksdb uses the following heuristic // to calculate size amplification: it assumes that all files excluding // the earliest file contribute to the size amplification. // Default: 200, which means that a 100 byte database could require upto // 300 bytes of storage. unsigned int max_size_amplification_percent; // If this option is set to be -1 (the default value), all the output files // will follow compression type specified. // // If this option is not negative, we will try to make sure compressed // size is just above this value. In normal cases, at least this percentage // of data will be compressed. // When we are compacting to a new file, here is the criteria whether // it needs to be compressed: assuming here are the list of files sorted // by generation time: // A1...An B1...Bm C1...Ct // where A1 is the newest and Ct is the oldest, and we are going to compact // B1...Bm, we calculate the total size of all the files as total_size, as // well as the total size of C1...Ct as total_C, the compaction output file // will be compressed iff // total_C / total_size < this percentage // Default: -1 int compression_size_percent; // The algorithm used to stop picking files into a single compaction run // Default: kCompactionStopStyleTotalSize CompactionStopStyle stop_style; // Option to optimize the universal multi level compaction by enabling // trivial move for non overlapping files. // Default: false bool allow_trivial_move; // Default set of parameters CompactionOptionsUniversal() : size_ratio(1), min_merge_width(2), max_merge_width(UINT_MAX), max_size_amplification_percent(200), compression_size_percent(-1), stop_style(kCompactionStopStyleTotalSize), allow_trivial_move(false) {} }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_UNIVERSAL_COMPACTION_OPTIONS_H rocksdb-4.5.1/include/rocksdb/utilities/000077500000000000000000000000001267530634000202435ustar00rootroot00000000000000rocksdb-4.5.1/include/rocksdb/utilities/backupable_db.h000066400000000000000000000312641267530634000231600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include "rocksdb/utilities/stackable_db.h" #include "rocksdb/env.h" #include "rocksdb/status.h" namespace rocksdb { struct BackupableDBOptions { // Where to keep the backup files. Has to be different than dbname_ // Best to set this to dbname_ + "/backups" // Required std::string backup_dir; // Backup Env object. It will be used for backup file I/O. If it's // nullptr, backups will be written out using DBs Env. If it's // non-nullptr, backup's I/O will be performed using this object. // If you want to have backups on HDFS, use HDFS Env here! // Default: nullptr Env* backup_env; // If share_table_files == true, backup will assume that table files with // same name have the same contents. This enables incremental backups and // avoids unnecessary data copies. // If share_table_files == false, each backup will be on its own and will // not share any data with other backups. // default: true bool share_table_files; // Backup info and error messages will be written to info_log // if non-nullptr. // Default: nullptr Logger* info_log; // If sync == true, we can guarantee you'll get consistent backup even // on a machine crash/reboot. Backup process is slower with sync enabled. // If sync == false, we don't guarantee anything on machine reboot. However, // chances are some of the backups are consistent. // Default: true bool sync; // If true, it will delete whatever backups there are already // Default: false bool destroy_old_data; // If false, we won't backup log files. This option can be useful for backing // up in-memory databases where log file are persisted, but table files are in // memory. // Default: true bool backup_log_files; // Max bytes that can be transferred in a second during backup. // If 0, go as fast as you can // Default: 0 uint64_t backup_rate_limit; // Max bytes that can be transferred in a second during restore. // If 0, go as fast as you can // Default: 0 uint64_t restore_rate_limit; // Only used if share_table_files is set to true. If true, will consider that // backups can come from different databases, hence a sst is not uniquely // identifed by its name, but by the triple (file name, crc32, file length) // Default: false // Note: this is an experimental option, and you'll need to set it manually // *turn it on only if you know what you're doing* bool share_files_with_checksum; // Try to use the file size in file name instead of getting size from HDFS, // if the file is generated with options.share_files_with_checksum = true. // This is a temporary solution to reduce the backupable Db open latency when // There are too many sst files. Will remove the option after we have a // permanent solution. // Default: false bool use_file_size_in_file_name; // Up to this many background threads will copy files for CreateNewBackup() // and RestoreDBFromBackup() // Default: 1 int max_background_operations; // During backup user can get callback every time next // callback_trigger_interval_size bytes being copied. // Default: 4194304 uint64_t callback_trigger_interval_size; void Dump(Logger* logger) const; explicit BackupableDBOptions( const std::string& _backup_dir, Env* _backup_env = nullptr, bool _share_table_files = true, Logger* _info_log = nullptr, bool _sync = true, bool _destroy_old_data = false, bool _backup_log_files = true, uint64_t _backup_rate_limit = 0, uint64_t _restore_rate_limit = 0, int _max_background_operations = 1, uint64_t _callback_trigger_interval_size = 4 * 1024 * 1024) : backup_dir(_backup_dir), backup_env(_backup_env), share_table_files(_share_table_files), info_log(_info_log), sync(_sync), destroy_old_data(_destroy_old_data), backup_log_files(_backup_log_files), backup_rate_limit(_backup_rate_limit), restore_rate_limit(_restore_rate_limit), share_files_with_checksum(false), use_file_size_in_file_name(false), max_background_operations(_max_background_operations), callback_trigger_interval_size(_callback_trigger_interval_size) { assert(share_table_files || !share_files_with_checksum); } }; struct RestoreOptions { // If true, restore won't overwrite the existing log files in wal_dir. It will // also move all log files from archive directory to wal_dir. Use this option // in combination with BackupableDBOptions::backup_log_files = false for // persisting in-memory databases. // Default: false bool keep_log_files; explicit RestoreOptions(bool _keep_log_files = false) : keep_log_files(_keep_log_files) {} }; typedef uint32_t BackupID; struct BackupInfo { BackupID backup_id; int64_t timestamp; uint64_t size; uint32_t number_files; BackupInfo() {} BackupInfo(BackupID _backup_id, int64_t _timestamp, uint64_t _size, uint32_t _number_files) : backup_id(_backup_id), timestamp(_timestamp), size(_size), number_files(_number_files) {} }; class BackupStatistics { public: BackupStatistics() { number_success_backup = 0; number_fail_backup = 0; } BackupStatistics(uint32_t _number_success_backup, uint32_t _number_fail_backup) : number_success_backup(_number_success_backup), number_fail_backup(_number_fail_backup) {} ~BackupStatistics() {} void IncrementNumberSuccessBackup(); void IncrementNumberFailBackup(); uint32_t GetNumberSuccessBackup() const; uint32_t GetNumberFailBackup() const; std::string ToString() const; private: uint32_t number_success_backup; uint32_t number_fail_backup; }; class BackupEngineReadOnly { public: virtual ~BackupEngineReadOnly() {} static Status Open(Env* db_env, const BackupableDBOptions& options, BackupEngineReadOnly** backup_engine_ptr); // You can GetBackupInfo safely, even with other BackupEngine performing // backups on the same directory virtual void GetBackupInfo(std::vector* backup_info) = 0; virtual void GetCorruptedBackups( std::vector* corrupt_backup_ids) = 0; // Restoring DB from backup is NOT safe when there is another BackupEngine // running that might call DeleteBackup() or PurgeOldBackups(). It is caller's // responsibility to synchronize the operation, i.e. don't delete the backup // when you're restoring from it virtual Status RestoreDBFromBackup( BackupID backup_id, const std::string& db_dir, const std::string& wal_dir, const RestoreOptions& restore_options = RestoreOptions()) = 0; virtual Status RestoreDBFromLatestBackup( const std::string& db_dir, const std::string& wal_dir, const RestoreOptions& restore_options = RestoreOptions()) = 0; // checks that each file exists and that the size of the file matches our // expectations. it does not check file checksum. // Returns Status::OK() if all checks are good virtual Status VerifyBackup(BackupID backup_id) = 0; }; // Please see the documentation in BackupableDB and RestoreBackupableDB class BackupEngine { public: virtual ~BackupEngine() {} static Status Open(Env* db_env, const BackupableDBOptions& options, BackupEngine** backup_engine_ptr); virtual Status CreateNewBackup( DB* db, bool flush_before_backup = false, std::function progress_callback = []() {}) = 0; virtual Status PurgeOldBackups(uint32_t num_backups_to_keep) = 0; virtual Status DeleteBackup(BackupID backup_id) = 0; virtual void StopBackup() = 0; virtual void GetBackupInfo(std::vector* backup_info) = 0; virtual void GetCorruptedBackups( std::vector* corrupt_backup_ids) = 0; virtual Status RestoreDBFromBackup( BackupID backup_id, const std::string& db_dir, const std::string& wal_dir, const RestoreOptions& restore_options = RestoreOptions()) = 0; virtual Status RestoreDBFromLatestBackup( const std::string& db_dir, const std::string& wal_dir, const RestoreOptions& restore_options = RestoreOptions()) = 0; // checks that each file exists and that the size of the file matches our // expectations. it does not check file checksum. // Returns Status::OK() if all checks are good virtual Status VerifyBackup(BackupID backup_id) = 0; virtual Status GarbageCollect() = 0; }; // Stack your DB with BackupableDB to be able to backup the DB class BackupableDB : public StackableDB { public: // BackupableDBOptions have to be the same as the ones used in a previous // incarnation of the DB // // BackupableDB ownes the pointer `DB* db` now. You should not delete it or // use it after the invocation of BackupableDB BackupableDB(DB* db, const BackupableDBOptions& options); virtual ~BackupableDB(); // Captures the state of the database in the latest backup // NOT a thread safe call Status CreateNewBackup(bool flush_before_backup = false); // Returns info about backups in backup_info void GetBackupInfo(std::vector* backup_info); // Returns info about corrupt backups in corrupt_backups void GetCorruptedBackups(std::vector* corrupt_backup_ids); // deletes old backups, keeping latest num_backups_to_keep alive Status PurgeOldBackups(uint32_t num_backups_to_keep); // deletes a specific backup Status DeleteBackup(BackupID backup_id); // Call this from another thread if you want to stop the backup // that is currently happening. It will return immediatelly, will // not wait for the backup to stop. // The backup will stop ASAP and the call to CreateNewBackup will // return Status::Incomplete(). It will not clean up after itself, but // the state will remain consistent. The state will be cleaned up // next time you create BackupableDB or RestoreBackupableDB. void StopBackup(); // Will delete all the files we don't need anymore // It will do the full scan of the files/ directory and delete all the // files that are not referenced. Status GarbageCollect(); private: BackupEngine* backup_engine_; Status status_; }; // Use this class to access information about backups and restore from them class RestoreBackupableDB { public: RestoreBackupableDB(Env* db_env, const BackupableDBOptions& options); ~RestoreBackupableDB(); // Returns info about backups in backup_info void GetBackupInfo(std::vector* backup_info); // Returns info about corrupt backups in corrupt_backups void GetCorruptedBackups(std::vector* corrupt_backup_ids); // restore from backup with backup_id // IMPORTANT -- if options_.share_table_files == true and you restore DB // from some backup that is not the latest, and you start creating new // backups from the new DB, they will probably fail // // Example: Let's say you have backups 1, 2, 3, 4, 5 and you restore 3. // If you add new data to the DB and try creating a new backup now, the // database will diverge from backups 4 and 5 and the new backup will fail. // If you want to create new backup, you will first have to delete backups 4 // and 5. Status RestoreDBFromBackup(BackupID backup_id, const std::string& db_dir, const std::string& wal_dir, const RestoreOptions& restore_options = RestoreOptions()); // restore from the latest backup Status RestoreDBFromLatestBackup(const std::string& db_dir, const std::string& wal_dir, const RestoreOptions& restore_options = RestoreOptions()); // deletes old backups, keeping latest num_backups_to_keep alive Status PurgeOldBackups(uint32_t num_backups_to_keep); // deletes a specific backup Status DeleteBackup(BackupID backup_id); // Will delete all the files we don't need anymore // It will do the full scan of the files/ directory and delete all the // files that are not referenced. Status GarbageCollect(); private: BackupEngine* backup_engine_; Status status_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/checkpoint.h000066400000000000000000000023501267530634000225430ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // A checkpoint is an openable snapshot of a database at a point in time. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/status.h" namespace rocksdb { class DB; class Checkpoint { public: // Creates a Checkpoint object to be used for creating openable sbapshots static Status Create(DB* db, Checkpoint** checkpoint_ptr); // Builds an openable snapshot of RocksDB on the same disk, which // accepts an output directory on the same disk, and under the directory // (1) hard-linked SST files pointing to existing live SST files // SST files will be copied if output directory is on a different filesystem // (2) a copied manifest files and other files // The directory should not already exist and will be created by this API. // The directory will be an absolute path virtual Status CreateCheckpoint(const std::string& checkpoint_dir); virtual ~Checkpoint() {} }; } // namespace rocksdb #endif // !ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/convenience.h000066400000000000000000000006021267530634000227060ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once // This file was moved to rocksdb/convenience.h" #include "rocksdb/convenience.h" rocksdb-4.5.1/include/rocksdb/utilities/db_ttl.h000066400000000000000000000053521267530634000216710ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/utilities/stackable_db.h" #include "rocksdb/db.h" namespace rocksdb { // Database with TTL support. // // USE-CASES: // This API should be used to open the db when key-values inserted are // meant to be removed from the db in a non-strict 'ttl' amount of time // Therefore, this guarantees that key-values inserted will remain in the // db for >= ttl amount of time and the db will make efforts to remove the // key-values as soon as possible after ttl seconds of their insertion. // // BEHAVIOUR: // TTL is accepted in seconds // (int32_t)Timestamp(creation) is suffixed to values in Put internally // Expired TTL values deleted in compaction only:(Timestamp+ttl=5 // read_only=true opens in the usual read-only mode. Compactions will not be // triggered(neither manual nor automatic), so no expired entries removed // // CONSTRAINTS: // Not specifying/passing or non-positive TTL behaves like TTL = infinity // // !!!WARNING!!!: // Calling DB::Open directly to re-open a db created by this API will get // corrupt values(timestamp suffixed) and no ttl effect will be there // during the second Open, so use this API consistently to open the db // Be careful when passing ttl with a small positive value because the // whole database may be deleted in a small amount of time class DBWithTTL : public StackableDB { public: virtual Status CreateColumnFamilyWithTtl( const ColumnFamilyOptions& options, const std::string& column_family_name, ColumnFamilyHandle** handle, int ttl) = 0; static Status Open(const Options& options, const std::string& dbname, DBWithTTL** dbptr, int32_t ttl = 0, bool read_only = false); static Status Open(const DBOptions& db_options, const std::string& dbname, const std::vector& column_families, std::vector* handles, DBWithTTL** dbptr, std::vector ttls, bool read_only = false); protected: explicit DBWithTTL(DB* db) : StackableDB(db) {} }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/document_db.h000066400000000000000000000137601267530634000227060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/utilities/stackable_db.h" #include "rocksdb/utilities/json_document.h" #include "rocksdb/db.h" namespace rocksdb { // IMPORTANT: DocumentDB is a work in progress. It is unstable and we might // change the API without warning. Talk to RocksDB team before using this in // production ;) // DocumentDB is a layer on top of RocksDB that provides a very simple JSON API. // When creating a DB, you specify a list of indexes you want to keep on your // data. You can insert a JSON document to the DB, which is automatically // indexed. Every document added to the DB needs to have "_id" field which is // automatically indexed and is an unique primary key. All other indexes are // non-unique. // NOTE: field names in the JSON are NOT allowed to start with '$' or // contain '.'. We don't currently enforce that rule, but will start behaving // badly. // Cursor is what you get as a result of executing query. To get all // results from a query, call Next() on a Cursor while Valid() returns true class Cursor { public: Cursor() = default; virtual ~Cursor() {} virtual bool Valid() const = 0; virtual void Next() = 0; // Lifecycle of the returned JSONDocument is until the next Next() call virtual const JSONDocument& document() const = 0; virtual Status status() const = 0; private: // No copying allowed Cursor(const Cursor&); void operator=(const Cursor&); }; struct DocumentDBOptions { int background_threads = 4; uint64_t memtable_size = 128 * 1024 * 1024; // 128 MB uint64_t cache_size = 1 * 1024 * 1024 * 1024; // 1 GB }; // TODO(icanadi) Add `JSONDocument* info` parameter to all calls that can be // used by the caller to get more information about the call execution (number // of dropped records, number of updated records, etc.) class DocumentDB : public StackableDB { public: struct IndexDescriptor { // Currently, you can only define an index on a single field. To specify an // index on a field X, set index description to JSON "{X: 1}" // Currently the value needs to be 1, which means ascending. // In the future, we plan to also support indexes on multiple keys, where // you could mix ascending sorting (1) with descending sorting indexes (-1) JSONDocument* description; std::string name; }; // Open DocumentDB with specified indexes. The list of indexes has to be // complete, i.e. include all indexes present in the DB, except the primary // key index. // Otherwise, Open() will return an error static Status Open(const DocumentDBOptions& options, const std::string& name, const std::vector& indexes, DocumentDB** db, bool read_only = false); explicit DocumentDB(DB* db) : StackableDB(db) {} // Create a new index. It will stop all writes for the duration of the call. // All current documents in the DB are scanned and corresponding index entries // are created virtual Status CreateIndex(const WriteOptions& write_options, const IndexDescriptor& index) = 0; // Drop an index. Client is responsible to make sure that index is not being // used by currently executing queries virtual Status DropIndex(const std::string& name) = 0; // Insert a document to the DB. The document needs to have a primary key "_id" // which can either be a string or an integer. Otherwise the write will fail // with InvalidArgument. virtual Status Insert(const WriteOptions& options, const JSONDocument& document) = 0; // Deletes all documents matching a filter atomically virtual Status Remove(const ReadOptions& read_options, const WriteOptions& write_options, const JSONDocument& query) = 0; // Does this sequence of operations: // 1. Find all documents matching a filter // 2. For all documents, atomically: // 2.1. apply the update operators // 2.2. update the secondary indexes // // Currently only $set update operator is supported. // Syntax is: {$set: {key1: value1, key2: value2, etc...}} // This operator will change a document's key1 field to value1, key2 to // value2, etc. New values will be set even if a document didn't have an entry // for the specified key. // // You can not change a primary key of a document. // // Update example: Update({id: {$gt: 5}, $index: id}, {$set: {enabled: true}}) virtual Status Update(const ReadOptions& read_options, const WriteOptions& write_options, const JSONDocument& filter, const JSONDocument& updates) = 0; // query has to be an array in which every element is an operator. Currently // only $filter operator is supported. Syntax of $filter operator is: // {$filter: {key1: condition1, key2: condition2, etc.}} where conditions can // be either: // 1) a single value in which case the condition is equality condition, or // 2) a defined operators, like {$gt: 4}, which will match all documents that // have key greater than 4. // // Supported operators are: // 1) $gt -- greater than // 2) $gte -- greater than or equal // 3) $lt -- less than // 4) $lte -- less than or equal // If you want the filter to use an index, you need to specify it like this: // {$filter: {...(conditions)..., $index: index_name}} // // Example query: // * [{$filter: {name: John, age: {$gte: 18}, $index: age}}] // will return all Johns whose age is greater or equal to 18 and it will use // index "age" to satisfy the query. virtual Cursor* Query(const ReadOptions& read_options, const JSONDocument& query) = 0; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/env_mirror.h000066400000000000000000000125631267530634000226050ustar00rootroot00000000000000// Copyright (c) 2015, Red Hat, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // MirrorEnv is an Env implementation that mirrors all file-related // operations to two backing Env's (provided at construction time). // Writes are mirrored. For read operations, we do the read from both // backends and assert that the results match. // // This is useful when implementing a new Env and ensuring that the // semantics and behavior are correct (in that they match that of an // existing, stable Env, like the default POSIX one). #ifndef ROCKSDB_LITE #ifndef STORAGE_ROCKSDB_INCLUDE_UTILITIES_ENVMIRROR_H_ #define STORAGE_ROCKSDB_INCLUDE_UTLIITIES_ENVMIRROR_H_ #include #include #include #include "rocksdb/env.h" namespace rocksdb { class SequentialFileMirror; class RandomAccessFileMirror; class WritableFileMirror; class EnvMirror : public EnvWrapper { Env* a_, *b_; public: EnvMirror(Env* a, Env* b) : EnvWrapper(a), a_(a), b_(b) {} Status NewSequentialFile(const std::string& f, unique_ptr* r, const EnvOptions& options) override; Status NewRandomAccessFile(const std::string& f, unique_ptr* r, const EnvOptions& options) override; Status NewWritableFile(const std::string& f, unique_ptr* r, const EnvOptions& options) override; Status ReuseWritableFile(const std::string& fname, const std::string& old_fname, unique_ptr* r, const EnvOptions& options) override; virtual Status NewDirectory(const std::string& name, unique_ptr* result) override { unique_ptr br; Status as = a_->NewDirectory(name, result); Status bs = b_->NewDirectory(name, &br); assert(as == bs); return as; } Status FileExists(const std::string& f) override { Status as = a_->FileExists(f); Status bs = b_->FileExists(f); assert(as == bs); return as; } Status GetChildren(const std::string& dir, std::vector* r) override { std::vector ar, br; Status as = a_->GetChildren(dir, &ar); Status bs = b_->GetChildren(dir, &br); assert(as == bs); std::sort(ar.begin(), ar.end()); std::sort(br.begin(), br.end()); if (!as.ok() || ar != br) { assert(0 == "getchildren results don't match"); } *r = ar; return as; } Status DeleteFile(const std::string& f) override { Status as = a_->DeleteFile(f); Status bs = b_->DeleteFile(f); assert(as == bs); return as; } Status CreateDir(const std::string& d) override { Status as = a_->CreateDir(d); Status bs = b_->CreateDir(d); assert(as == bs); return as; } Status CreateDirIfMissing(const std::string& d) override { Status as = a_->CreateDirIfMissing(d); Status bs = b_->CreateDirIfMissing(d); assert(as == bs); return as; } Status DeleteDir(const std::string& d) override { Status as = a_->DeleteDir(d); Status bs = b_->DeleteDir(d); assert(as == bs); return as; } Status GetFileSize(const std::string& f, uint64_t* s) override { uint64_t asize, bsize; Status as = a_->GetFileSize(f, &asize); Status bs = b_->GetFileSize(f, &bsize); assert(as == bs); assert(!as.ok() || asize == bsize); *s = asize; return as; } Status GetFileModificationTime(const std::string& fname, uint64_t* file_mtime) override { uint64_t amtime, bmtime; Status as = a_->GetFileModificationTime(fname, &amtime); Status bs = b_->GetFileModificationTime(fname, &bmtime); assert(as == bs); assert(!as.ok() || amtime - bmtime < 10000 || bmtime - amtime < 10000); *file_mtime = amtime; return as; } Status RenameFile(const std::string& s, const std::string& t) override { Status as = a_->RenameFile(s, t); Status bs = b_->RenameFile(s, t); assert(as == bs); return as; } Status LinkFile(const std::string& s, const std::string& t) override { Status as = a_->LinkFile(s, t); Status bs = b_->LinkFile(s, t); assert(as == bs); return as; } class FileLockMirror : public FileLock { public: FileLock* a_, *b_; FileLockMirror(FileLock* a, FileLock* b) : a_(a), b_(b) {} }; Status LockFile(const std::string& f, FileLock** l) override { FileLock* al, *bl; Status as = a_->LockFile(f, &al); Status bs = b_->LockFile(f, &bl); assert(as == bs); if (as.ok()) *l = new FileLockMirror(al, bl); return as; } Status UnlockFile(FileLock* l) override { FileLockMirror* ml = static_cast(l); Status as = a_->UnlockFile(ml->a_); Status bs = b_->UnlockFile(ml->b_); assert(as == bs); return as; } }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_UTILITIES_ENVMIRROR_H_ #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/flashcache.h000066400000000000000000000020371267530634000224770ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include "rocksdb/env.h" namespace rocksdb { // This API is experimental. We will mark it stable once we run it in production // for a while. // NewFlashcacheAwareEnv() creates and Env that blacklists all background // threads (used for flush and compaction) from using flashcache to cache their // reads. Reads from compaction thread don't need to be cached because they are // going to be soon made obsolete (due to nature of compaction) // Usually you would pass Env::Default() as base. // cachedev_fd is a file descriptor of the flashcache device. Caller has to // open flashcache device before calling this API. extern std::unique_ptr NewFlashcacheAwareEnv( Env* base, const int cachedev_fd); } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/utilities/geo_db.h000066400000000000000000000066611267530634000216440ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #ifndef ROCKSDB_LITE #pragma once #include #include #include "rocksdb/utilities/stackable_db.h" #include "rocksdb/status.h" namespace rocksdb { // // Configurable options needed for setting up a Geo database // struct GeoDBOptions { // Backup info and error messages will be written to info_log // if non-nullptr. // Default: nullptr Logger* info_log; explicit GeoDBOptions(Logger* _info_log = nullptr):info_log(_info_log) { } }; // // A position in the earth's geoid // class GeoPosition { public: double latitude; double longitude; explicit GeoPosition(double la = 0, double lo = 0) : latitude(la), longitude(lo) { } }; // // Description of an object on the Geoid. It is located by a GPS location, // and is identified by the id. The value associated with this object is // an opaque string 'value'. Different objects identified by unique id's // can have the same gps-location associated with them. // class GeoObject { public: GeoPosition position; std::string id; std::string value; GeoObject() {} GeoObject(const GeoPosition& pos, const std::string& i, const std::string& val) : position(pos), id(i), value(val) { } }; class GeoIterator { public: GeoIterator() = default; virtual ~GeoIterator() {} virtual void Next() = 0; virtual bool Valid() const = 0; virtual const GeoObject& geo_object() = 0; virtual Status status() const = 0; }; // // Stack your DB with GeoDB to be able to get geo-spatial support // class GeoDB : public StackableDB { public: // GeoDBOptions have to be the same as the ones used in a previous // incarnation of the DB // // GeoDB owns the pointer `DB* db` now. You should not delete it or // use it after the invocation of GeoDB // GeoDB(DB* db, const GeoDBOptions& options) : StackableDB(db) {} GeoDB(DB* db, const GeoDBOptions& options) : StackableDB(db) {} virtual ~GeoDB() {} // Insert a new object into the location database. The object is // uniquely identified by the id. If an object with the same id already // exists in the db, then the old one is overwritten by the new // object being inserted here. virtual Status Insert(const GeoObject& object) = 0; // Retrieve the value of the object located at the specified GPS // location and is identified by the 'id'. virtual Status GetByPosition(const GeoPosition& pos, const Slice& id, std::string* value) = 0; // Retrieve the value of the object identified by the 'id'. This method // could be potentially slower than GetByPosition virtual Status GetById(const Slice& id, GeoObject* object) = 0; // Delete the specified object virtual Status Remove(const Slice& id) = 0; // Returns an iterator for the items within a circular radius from the // specified gps location. If 'number_of_values' is specified, // then the iterator is capped to that number of objects. // The radius is specified in 'meters'. virtual GeoIterator* SearchRadial(const GeoPosition& pos, double radius, int number_of_values = INT_MAX) = 0; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/info_log_finder.h000066400000000000000000000011231267530634000235340ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "rocksdb/db.h" #include "rocksdb/options.h" namespace rocksdb { // This function can be used to list the Information logs, // given the db pointer. Status GetInfoLogList(DB* db, std::vector* info_log_list); } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/utilities/json_document.h000066400000000000000000000113771267530634000232740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include #include #include #include #include "rocksdb/slice.h" // We use JSONDocument for DocumentDB API // Implementation inspired by folly::dynamic, rapidjson and fbson namespace fbson { class FbsonValue; class ObjectVal; template class FbsonWriterT; class FbsonOutStream; typedef FbsonWriterT FbsonWriter; } // namespace fbson namespace rocksdb { // NOTE: none of this is thread-safe class JSONDocument { public: // return nullptr on parse failure static JSONDocument* ParseJSON(const char* json); enum Type { kNull, kArray, kBool, kDouble, kInt64, kObject, kString, }; /* implicit */ JSONDocument(); // null /* implicit */ JSONDocument(bool b); /* implicit */ JSONDocument(double d); /* implicit */ JSONDocument(int8_t i); /* implicit */ JSONDocument(int16_t i); /* implicit */ JSONDocument(int32_t i); /* implicit */ JSONDocument(int64_t i); /* implicit */ JSONDocument(const std::string& s); /* implicit */ JSONDocument(const char* s); // constructs JSONDocument of specific type with default value explicit JSONDocument(Type _type); JSONDocument(const JSONDocument& json_document); JSONDocument(JSONDocument&& json_document); Type type() const; // REQUIRES: IsObject() bool Contains(const std::string& key) const; // REQUIRES: IsObject() // Returns non-owner object JSONDocument operator[](const std::string& key) const; // REQUIRES: IsArray() == true || IsObject() == true size_t Count() const; // REQUIRES: IsArray() // Returns non-owner object JSONDocument operator[](size_t i) const; JSONDocument& operator=(JSONDocument jsonDocument); bool IsNull() const; bool IsArray() const; bool IsBool() const; bool IsDouble() const; bool IsInt64() const; bool IsObject() const; bool IsString() const; // REQUIRES: IsBool() == true bool GetBool() const; // REQUIRES: IsDouble() == true double GetDouble() const; // REQUIRES: IsInt64() == true int64_t GetInt64() const; // REQUIRES: IsString() == true std::string GetString() const; bool operator==(const JSONDocument& rhs) const; bool operator!=(const JSONDocument& rhs) const; JSONDocument Copy() const; bool IsOwner() const; std::string DebugString() const; private: class ItemsIteratorGenerator; public: // REQUIRES: IsObject() ItemsIteratorGenerator Items() const; // appends serialized object to dst void Serialize(std::string* dst) const; // returns nullptr if Slice doesn't represent valid serialized JSONDocument static JSONDocument* Deserialize(const Slice& src); private: friend class JSONDocumentBuilder; JSONDocument(fbson::FbsonValue* val, bool makeCopy); void InitFromValue(const fbson::FbsonValue* val); // iteration on objects class const_item_iterator { private: class Impl; public: typedef std::pair value_type; explicit const_item_iterator(Impl* impl); const_item_iterator(const_item_iterator&&); const_item_iterator& operator++(); bool operator!=(const const_item_iterator& other); value_type operator*(); ~const_item_iterator(); private: friend class ItemsIteratorGenerator; std::unique_ptr it_; }; class ItemsIteratorGenerator { public: explicit ItemsIteratorGenerator(const fbson::ObjectVal& object); const_item_iterator begin() const; const_item_iterator end() const; private: const fbson::ObjectVal& object_; }; std::unique_ptr data_; mutable fbson::FbsonValue* value_; // Our serialization format's first byte specifies the encoding version. That // way, we can easily change our format while providing backwards // compatibility. This constant specifies the current version of the // serialization format static const char kSerializationFormatVersion; }; class JSONDocumentBuilder { public: JSONDocumentBuilder(); explicit JSONDocumentBuilder(fbson::FbsonOutStream* out); void Reset(); bool WriteStartArray(); bool WriteEndArray(); bool WriteStartObject(); bool WriteEndObject(); bool WriteKeyValue(const std::string& key, const JSONDocument& value); bool WriteJSONDocument(const JSONDocument& value); JSONDocument GetJSONDocument(); ~JSONDocumentBuilder(); private: std::unique_ptr writer_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/leveldb_options.h000066400000000000000000000121731267530634000236100ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include namespace rocksdb { class Cache; class Comparator; class Env; class FilterPolicy; class Logger; struct Options; class Snapshot; enum CompressionType : char; // Options to control the behavior of a database (passed to // DB::Open). A LevelDBOptions object can be initialized as though // it were a LevelDB Options object, and then it can be converted into // a RocksDB Options object. struct LevelDBOptions { // ------------------- // Parameters that affect behavior // Comparator used to define the order of keys in the table. // Default: a comparator that uses lexicographic byte-wise ordering // // REQUIRES: The client must ensure that the comparator supplied // here has the same name and orders keys *exactly* the same as the // comparator provided to previous open calls on the same DB. const Comparator* comparator; // If true, the database will be created if it is missing. // Default: false bool create_if_missing; // If true, an error is raised if the database already exists. // Default: false bool error_if_exists; // If true, the implementation will do aggressive checking of the // data it is processing and will stop early if it detects any // errors. This may have unforeseen ramifications: for example, a // corruption of one DB entry may cause a large number of entries to // become unreadable or for the entire DB to become unopenable. // Default: false bool paranoid_checks; // Use the specified object to interact with the environment, // e.g. to read/write files, schedule background work, etc. // Default: Env::Default() Env* env; // Any internal progress/error information generated by the db will // be written to info_log if it is non-NULL, or to a file stored // in the same directory as the DB contents if info_log is NULL. // Default: NULL Logger* info_log; // ------------------- // Parameters that affect performance // Amount of data to build up in memory (backed by an unsorted log // on disk) before converting to a sorted on-disk file. // // Larger values increase performance, especially during bulk loads. // Up to two write buffers may be held in memory at the same time, // so you may wish to adjust this parameter to control memory usage. // Also, a larger write buffer will result in a longer recovery time // the next time the database is opened. // // Default: 4MB size_t write_buffer_size; // Number of open files that can be used by the DB. You may need to // increase this if your database has a large working set (budget // one open file per 2MB of working set). // // Default: 1000 int max_open_files; // Control over blocks (user data is stored in a set of blocks, and // a block is the unit of reading from disk). // If non-NULL, use the specified cache for blocks. // If NULL, leveldb will automatically create and use an 8MB internal cache. // Default: NULL Cache* block_cache; // Approximate size of user data packed per block. Note that the // block size specified here corresponds to uncompressed data. The // actual size of the unit read from disk may be smaller if // compression is enabled. This parameter can be changed dynamically. // // Default: 4K size_t block_size; // Number of keys between restart points for delta encoding of keys. // This parameter can be changed dynamically. Most clients should // leave this parameter alone. // // Default: 16 int block_restart_interval; // Compress blocks using the specified compression algorithm. This // parameter can be changed dynamically. // // Default: kSnappyCompression, which gives lightweight but fast // compression. // // Typical speeds of kSnappyCompression on an Intel(R) Core(TM)2 2.4GHz: // ~200-500MB/s compression // ~400-800MB/s decompression // Note that these speeds are significantly faster than most // persistent storage speeds, and therefore it is typically never // worth switching to kNoCompression. Even if the input data is // incompressible, the kSnappyCompression implementation will // efficiently detect that and will switch to uncompressed mode. CompressionType compression; // If non-NULL, use the specified filter policy to reduce disk reads. // Many applications will benefit from passing the result of // NewBloomFilterPolicy() here. // // Default: NULL const FilterPolicy* filter_policy; // Create a LevelDBOptions object with default values for all fields. LevelDBOptions(); }; // Converts a LevelDBOptions object into a RocksDB Options object. Options ConvertOptions(const LevelDBOptions& leveldb_options); } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/utilities/memory_util.h000066400000000000000000000031421267530634000227610ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #pragma once #include #include #include #include #include "rocksdb/cache.h" #include "rocksdb/db.h" namespace rocksdb { // Returns the current memory usage of the specified DB instances. class MemoryUtil { public: enum UsageType : int { // Memory usage of all the mem-tables. kMemTableTotal = 0, // Memory usage of those un-flushed mem-tables. kMemTableUnFlushed = 1, // Memory usage of all the table readers. kTableReadersTotal = 2, // Memory usage by Cache. kCacheTotal = 3, kNumUsageTypes = 4 }; // Returns the approximate memory usage of different types in the input // list of DBs and Cache set. For instance, in the output map // usage_by_type, usage_by_type[kMemTableTotal] will store the memory // usage of all the mem-tables from all the input rocksdb instances. // // Note that for memory usage inside Cache class, we will // only report the usage of the input "cache_set" without // including those Cache usage inside the input list "dbs" // of DBs. static Status GetApproximateMemoryUsageByType( const std::vector& dbs, const std::unordered_set cache_set, std::map* usage_by_type); }; } // namespace rocksdb #endif // !ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/optimistic_transaction_db.h000066400000000000000000000043731267530634000256610ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/comparator.h" #include "rocksdb/db.h" namespace rocksdb { class Transaction; // Database with Transaction support. // // See optimistic_transaction.h and examples/transaction_example.cc // Options to use when starting an Optimistic Transaction struct OptimisticTransactionOptions { // Setting set_snapshot=true is the same as calling SetSnapshot(). bool set_snapshot = false; // Should be set if the DB has a non-default comparator. // See comment in WriteBatchWithIndex constructor. const Comparator* cmp = BytewiseComparator(); }; class OptimisticTransactionDB { public: // Open an OptimisticTransactionDB similar to DB::Open(). static Status Open(const Options& options, const std::string& dbname, OptimisticTransactionDB** dbptr); static Status Open(const DBOptions& db_options, const std::string& dbname, const std::vector& column_families, std::vector* handles, OptimisticTransactionDB** dbptr); virtual ~OptimisticTransactionDB() {} // Starts a new Transaction. Passing set_snapshot=true has the same effect // as calling SetSnapshot(). // // Caller should delete the returned transaction after calling // Commit() or Rollback(). virtual Transaction* BeginTransaction( const WriteOptions& write_options, const OptimisticTransactionOptions& txn_options = OptimisticTransactionOptions()) = 0; // Return the underlying Database that was opened virtual DB* GetBaseDB() = 0; protected: // To Create an OptimisticTransactionDB, call Open() explicit OptimisticTransactionDB(DB* db) {} OptimisticTransactionDB() {} private: // No copying allowed OptimisticTransactionDB(const OptimisticTransactionDB&); void operator=(const OptimisticTransactionDB&); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/options_util.h000066400000000000000000000061771267530634000231570ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // This file contains utility functions for RocksDB Options. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "rocksdb/status.h" namespace rocksdb { // Constructs the DBOptions and ColumnFamilyDescriptors by loading the // latest RocksDB options file stored in the specified rocksdb database. // // Note that the all the pointer options (except table_factory, which will // be described in more details below) will be initialized with the default // values. Developers can further initialize them after this function call. // Below is an example list of pointer options which will be initialized // // * env // * memtable_factory // * compaction_filter_factory // * prefix_extractor // * comparator // * merge_operator // * compaction_filter // // For table_factory, this function further supports deserializing // BlockBasedTableFactory and its BlockBasedTableOptions except the // pointer options of BlockBasedTableOptions (flush_block_policy_factory, // block_cache, and block_cache_compressed), which will be initialized with // default values. Developers can further specify these three options by // casting the return value of TableFactoroy::GetOptions() to // BlockBasedTableOptions and making necessary changes. // // examples/options_file_example.cc demonstrates how to use this function // to open a RocksDB instance. // // @see LoadOptionsFromFile Status LoadLatestOptions(const std::string& dbpath, Env* env, DBOptions* db_options, std::vector* cf_descs); // Similar to LoadLatestOptions, this function constructs the DBOptions // and ColumnFamilyDescriptors based on the specified RocksDB Options file. // // @see LoadLatestOptions Status LoadOptionsFromFile(const std::string& options_file_name, Env* env, DBOptions* db_options, std::vector* cf_descs); // Returns the latest options file name under the specified db path. Status GetLatestOptionsFileName(const std::string& dbpath, Env* env, std::string* options_file_name); // Returns Status::OK if the input DBOptions and ColumnFamilyDescriptors // are compatible with the latest options stored in the specified DB path. // // If the return status is non-ok, it means the specified RocksDB instance // might not be correctly opened with the input set of options. Currently, // changing one of the following options will fail the compatibility check: // // * comparator // * prefix_extractor // * table_factory // * merge_operator Status CheckOptionsCompatibility( const std::string& dbpath, Env* env, const DBOptions& db_options, const std::vector& cf_descs); } // namespace rocksdb #endif // !ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/spatial_db.h000066400000000000000000000214751267530634000225270ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/db.h" #include "rocksdb/slice.h" #include "rocksdb/utilities/stackable_db.h" namespace rocksdb { namespace spatial { // NOTE: SpatialDB is experimental and we might change its API without warning. // Please talk to us before developing against SpatialDB API. // // SpatialDB is a support for spatial indexes built on top of RocksDB. // When creating a new SpatialDB, clients specifies a list of spatial indexes to // build on their data. Each spatial index is defined by the area and // granularity. If you're storing map data, different spatial index // granularities can be used for different zoom levels. // // Each element inserted into SpatialDB has: // * a bounding box, which determines how will the element be indexed // * string blob, which will usually be WKB representation of the polygon // (http://en.wikipedia.org/wiki/Well-known_text) // * feature set, which is a map of key-value pairs, where value can be null, // int, double, bool, string // * a list of indexes to insert the element in // // Each query is executed on a single spatial index. Query guarantees that it // will return all elements intersecting the specified bounding box, but it // might also return some extra non-intersecting elements. // Variant is a class that can be many things: null, bool, int, double or string // It is used to store different value types in FeatureSet (see below) struct Variant { // Don't change the values here, they are persisted on disk enum Type { kNull = 0x0, kBool = 0x1, kInt = 0x2, kDouble = 0x3, kString = 0x4, }; Variant() : type_(kNull) {} /* implicit */ Variant(bool b) : type_(kBool) { data_.b = b; } /* implicit */ Variant(uint64_t i) : type_(kInt) { data_.i = i; } /* implicit */ Variant(double d) : type_(kDouble) { data_.d = d; } /* implicit */ Variant(const std::string& s) : type_(kString) { new (&data_.s) std::string(s); } Variant(const Variant& v) : type_(v.type_) { Init(v, data_); } Variant& operator=(const Variant& v); Variant(Variant&& rhs) : type_(kNull) { *this = std::move(rhs); } Variant& operator=(Variant&& v); ~Variant() { Destroy(type_, data_); } Type type() const { return type_; } bool get_bool() const { return data_.b; } uint64_t get_int() const { return data_.i; } double get_double() const { return data_.d; } const std::string& get_string() const { return *GetStringPtr(data_); } bool operator==(const Variant& other) const; bool operator!=(const Variant& other) const { return !(*this == other); } private: Type type_; union Data { bool b; uint64_t i; double d; // Current version of MS compiler not C++11 compliant so can not put // std::string // however, even then we still need the rest of the maintenance. char s[sizeof(std::string)]; } data_; // Avoid type_punned aliasing problem static std::string* GetStringPtr(Data& d) { void* p = d.s; return reinterpret_cast(p); } static const std::string* GetStringPtr(const Data& d) { const void* p = d.s; return reinterpret_cast(p); } static void Init(const Variant&, Data&); static void Destroy(Type t, Data& d) { if (t == kString) { using std::string; GetStringPtr(d)->~string(); } } }; // FeatureSet is a map of key-value pairs. One feature set is associated with // each element in SpatialDB. It can be used to add rich data about the element. class FeatureSet { private: typedef std::unordered_map map; public: class iterator { public: /* implicit */ iterator(const map::const_iterator itr) : itr_(itr) {} iterator& operator++() { ++itr_; return *this; } bool operator!=(const iterator& other) { return itr_ != other.itr_; } bool operator==(const iterator& other) { return itr_ == other.itr_; } map::value_type operator*() { return *itr_; } private: map::const_iterator itr_; }; FeatureSet() = default; FeatureSet* Set(const std::string& key, const Variant& value); bool Contains(const std::string& key) const; // REQUIRES: Contains(key) const Variant& Get(const std::string& key) const; iterator Find(const std::string& key) const; iterator begin() const { return map_.begin(); } iterator end() const { return map_.end(); } void Clear(); size_t Size() const { return map_.size(); } void Serialize(std::string* output) const; // REQUIRED: empty FeatureSet bool Deserialize(const Slice& input); std::string DebugString() const; private: map map_; }; // BoundingBox is a helper structure for defining rectangles representing // bounding boxes of spatial elements. template struct BoundingBox { T min_x, min_y, max_x, max_y; BoundingBox() = default; BoundingBox(T _min_x, T _min_y, T _max_x, T _max_y) : min_x(_min_x), min_y(_min_y), max_x(_max_x), max_y(_max_y) {} bool Intersects(const BoundingBox& a) const { return !(min_x > a.max_x || min_y > a.max_y || a.min_x > max_x || a.min_y > max_y); } }; struct SpatialDBOptions { uint64_t cache_size = 1 * 1024 * 1024 * 1024LL; // 1GB int num_threads = 16; bool bulk_load = true; }; // Cursor is used to return data from the query to the client. To get all the // data from the query, just call Next() while Valid() is true class Cursor { public: Cursor() = default; virtual ~Cursor() {} virtual bool Valid() const = 0; // REQUIRES: Valid() virtual void Next() = 0; // Lifetime of the underlying storage until the next call to Next() // REQUIRES: Valid() virtual const Slice blob() = 0; // Lifetime of the underlying storage until the next call to Next() // REQUIRES: Valid() virtual const FeatureSet& feature_set() = 0; virtual Status status() const = 0; private: // No copying allowed Cursor(const Cursor&); void operator=(const Cursor&); }; // SpatialIndexOptions defines a spatial index that will be built on the data struct SpatialIndexOptions { // Spatial indexes are referenced by names std::string name; // An area that is indexed. If the element is not intersecting with spatial // index's bbox, it will not be inserted into the index BoundingBox bbox; // tile_bits control the granularity of the spatial index. Each dimension of // the bbox will be split into (1 << tile_bits) tiles, so there will be a // total of (1 << tile_bits)^2 tiles. It is recommended to configure a size of // each tile to be approximately the size of the query on that spatial index uint32_t tile_bits; SpatialIndexOptions() {} SpatialIndexOptions(const std::string& _name, const BoundingBox& _bbox, uint32_t _tile_bits) : name(_name), bbox(_bbox), tile_bits(_tile_bits) {} }; class SpatialDB : public StackableDB { public: // Creates the SpatialDB with specified list of indexes. // REQUIRED: db doesn't exist static Status Create(const SpatialDBOptions& options, const std::string& name, const std::vector& spatial_indexes); // Open the existing SpatialDB. The resulting db object will be returned // through db parameter. // REQUIRED: db was created using SpatialDB::Create static Status Open(const SpatialDBOptions& options, const std::string& name, SpatialDB** db, bool read_only = false); explicit SpatialDB(DB* db) : StackableDB(db) {} // Insert the element into the DB. Element will be inserted into specified // spatial_indexes, based on specified bbox. // REQUIRES: spatial_indexes.size() > 0 virtual Status Insert(const WriteOptions& write_options, const BoundingBox& bbox, const Slice& blob, const FeatureSet& feature_set, const std::vector& spatial_indexes) = 0; // Calling Compact() after inserting a bunch of elements should speed up // reading. This is especially useful if you use SpatialDBOptions::bulk_load // Num threads determines how many threads we'll use for compactions. Setting // this to bigger number will use more IO and CPU, but finish faster virtual Status Compact(int num_threads = 1) = 0; // Query the specified spatial_index. Query will return all elements that // intersect bbox, but it may also return some extra elements. virtual Cursor* Query(const ReadOptions& read_options, const BoundingBox& bbox, const std::string& spatial_index) = 0; }; } // namespace spatial } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/stackable_db.h000066400000000000000000000241031267530634000230120ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include "rocksdb/db.h" #ifdef _WIN32 // Windows API macro interference #undef DeleteFile #endif namespace rocksdb { // This class contains APIs to stack rocksdb wrappers.Eg. Stack TTL over base d class StackableDB : public DB { public: // StackableDB is the owner of db now! explicit StackableDB(DB* db) : db_(db) {} ~StackableDB() { delete db_; } virtual DB* GetBaseDB() { return db_; } virtual DB* GetRootDB() override { return db_->GetRootDB(); } virtual Status CreateColumnFamily(const ColumnFamilyOptions& options, const std::string& column_family_name, ColumnFamilyHandle** handle) override { return db_->CreateColumnFamily(options, column_family_name, handle); } virtual Status DropColumnFamily(ColumnFamilyHandle* column_family) override { return db_->DropColumnFamily(column_family); } using DB::Put; virtual Status Put(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& val) override { return db_->Put(options, column_family, key, val); } using DB::Get; virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) override { return db_->Get(options, column_family, key, value); } using DB::MultiGet; virtual std::vector MultiGet( const ReadOptions& options, const std::vector& column_family, const std::vector& keys, std::vector* values) override { return db_->MultiGet(options, column_family, keys, values); } using DB::AddFile; virtual Status AddFile(ColumnFamilyHandle* column_family, const ExternalSstFileInfo* file_info, bool move_file) override { return db_->AddFile(column_family, file_info, move_file); } virtual Status AddFile(ColumnFamilyHandle* column_family, const std::string& file_path, bool move_file) override { return db_->AddFile(column_family, file_path, move_file); } using DB::KeyMayExist; virtual bool KeyMayExist(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value, bool* value_found = nullptr) override { return db_->KeyMayExist(options, column_family, key, value, value_found); } using DB::Delete; virtual Status Delete(const WriteOptions& wopts, ColumnFamilyHandle* column_family, const Slice& key) override { return db_->Delete(wopts, column_family, key); } using DB::SingleDelete; virtual Status SingleDelete(const WriteOptions& wopts, ColumnFamilyHandle* column_family, const Slice& key) override { return db_->SingleDelete(wopts, column_family, key); } using DB::Merge; virtual Status Merge(const WriteOptions& options, ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override { return db_->Merge(options, column_family, key, value); } virtual Status Write(const WriteOptions& opts, WriteBatch* updates) override { return db_->Write(opts, updates); } using DB::NewIterator; virtual Iterator* NewIterator(const ReadOptions& opts, ColumnFamilyHandle* column_family) override { return db_->NewIterator(opts, column_family); } virtual Status NewIterators( const ReadOptions& options, const std::vector& column_families, std::vector* iterators) override { return db_->NewIterators(options, column_families, iterators); } virtual const Snapshot* GetSnapshot() override { return db_->GetSnapshot(); } virtual void ReleaseSnapshot(const Snapshot* snapshot) override { return db_->ReleaseSnapshot(snapshot); } using DB::GetProperty; virtual bool GetProperty(ColumnFamilyHandle* column_family, const Slice& property, std::string* value) override { return db_->GetProperty(column_family, property, value); } using DB::GetIntProperty; virtual bool GetIntProperty(ColumnFamilyHandle* column_family, const Slice& property, uint64_t* value) override { return db_->GetIntProperty(column_family, property, value); } using DB::GetAggregatedIntProperty; virtual bool GetAggregatedIntProperty(const Slice& property, uint64_t* value) override { return db_->GetAggregatedIntProperty(property, value); } using DB::GetApproximateSizes; virtual void GetApproximateSizes(ColumnFamilyHandle* column_family, const Range* r, int n, uint64_t* sizes, bool include_memtable = false) override { return db_->GetApproximateSizes(column_family, r, n, sizes, include_memtable); } using DB::CompactRange; virtual Status CompactRange(const CompactRangeOptions& options, ColumnFamilyHandle* column_family, const Slice* begin, const Slice* end) override { return db_->CompactRange(options, column_family, begin, end); } using DB::CompactFiles; virtual Status CompactFiles( const CompactionOptions& compact_options, ColumnFamilyHandle* column_family, const std::vector& input_file_names, const int output_level, const int output_path_id = -1) override { return db_->CompactFiles( compact_options, column_family, input_file_names, output_level, output_path_id); } virtual Status PauseBackgroundWork() override { return db_->PauseBackgroundWork(); } virtual Status ContinueBackgroundWork() override { return db_->ContinueBackgroundWork(); } virtual Status EnableAutoCompaction( const std::vector& column_family_handles) override { return db_->EnableAutoCompaction(column_family_handles); } using DB::NumberLevels; virtual int NumberLevels(ColumnFamilyHandle* column_family) override { return db_->NumberLevels(column_family); } using DB::MaxMemCompactionLevel; virtual int MaxMemCompactionLevel(ColumnFamilyHandle* column_family) override { return db_->MaxMemCompactionLevel(column_family); } using DB::Level0StopWriteTrigger; virtual int Level0StopWriteTrigger(ColumnFamilyHandle* column_family) override { return db_->Level0StopWriteTrigger(column_family); } virtual const std::string& GetName() const override { return db_->GetName(); } virtual Env* GetEnv() const override { return db_->GetEnv(); } using DB::GetOptions; virtual const Options& GetOptions(ColumnFamilyHandle* column_family) const override { return db_->GetOptions(column_family); } using DB::GetDBOptions; virtual const DBOptions& GetDBOptions() const override { return db_->GetDBOptions(); } using DB::Flush; virtual Status Flush(const FlushOptions& fopts, ColumnFamilyHandle* column_family) override { return db_->Flush(fopts, column_family); } virtual Status SyncWAL() override { return db_->SyncWAL(); } #ifndef ROCKSDB_LITE virtual Status DisableFileDeletions() override { return db_->DisableFileDeletions(); } virtual Status EnableFileDeletions(bool force) override { return db_->EnableFileDeletions(force); } virtual void GetLiveFilesMetaData( std::vector* metadata) override { db_->GetLiveFilesMetaData(metadata); } virtual void GetColumnFamilyMetaData( ColumnFamilyHandle *column_family, ColumnFamilyMetaData* cf_meta) override { db_->GetColumnFamilyMetaData(column_family, cf_meta); } #endif // ROCKSDB_LITE virtual Status GetLiveFiles(std::vector& vec, uint64_t* mfs, bool flush_memtable = true) override { return db_->GetLiveFiles(vec, mfs, flush_memtable); } virtual SequenceNumber GetLatestSequenceNumber() const override { return db_->GetLatestSequenceNumber(); } virtual Status GetSortedWalFiles(VectorLogPtr& files) override { return db_->GetSortedWalFiles(files); } virtual Status DeleteFile(std::string name) override { return db_->DeleteFile(name); } virtual Status GetDbIdentity(std::string& identity) const override { return db_->GetDbIdentity(identity); } using DB::SetOptions; virtual Status SetOptions(ColumnFamilyHandle* column_family_handle, const std::unordered_map& new_options) override { return db_->SetOptions(column_family_handle, new_options); } using DB::GetPropertiesOfAllTables; virtual Status GetPropertiesOfAllTables( ColumnFamilyHandle* column_family, TablePropertiesCollection* props) override { return db_->GetPropertiesOfAllTables(column_family, props); } using DB::GetPropertiesOfTablesInRange; virtual Status GetPropertiesOfTablesInRange( ColumnFamilyHandle* column_family, const Range* range, std::size_t n, TablePropertiesCollection* props) override { return db_->GetPropertiesOfTablesInRange(column_family, range, n, props); } virtual Status GetUpdatesSince( SequenceNumber seq_number, unique_ptr* iter, const TransactionLogIterator::ReadOptions& read_options) override { return db_->GetUpdatesSince(seq_number, iter, read_options); } virtual ColumnFamilyHandle* DefaultColumnFamily() const override { return db_->DefaultColumnFamily(); } protected: DB* db_; }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/utilities/table_properties_collectors.h000066400000000000000000000021021267530634000262030ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/table_properties.h" namespace rocksdb { // Creates a factory of a table property collector that marks a SST // file as need-compaction when it observe at least "D" deletion // entries in any "N" consecutive entires. // // @param sliding_window_size "N". Note that this number will be // round up to the smallest multiple of 128 that is no less // than the specified size. // @param deletion_trigger "D". Note that even when "N" is changed, // the specified number for "D" will not be changed. extern std::shared_ptr NewCompactOnDeletionCollectorFactory( size_t sliding_window_size, size_t deletion_trigger); } // namespace rocksdb #endif // !ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/transaction.h000066400000000000000000000375751267530634000227620ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/status.h" namespace rocksdb { class Iterator; class TransactionDB; class WriteBatchWithIndex; // Provides notification to the caller of SetSnapshotOnNextOperation when // the actual snapshot gets created class TransactionNotifier { public: virtual ~TransactionNotifier() {} // Implement this method to receive notification when a snapshot is // requested via SetSnapshotOnNextOperation. virtual void SnapshotCreated(const Snapshot* newSnapshot) = 0; }; // Provides BEGIN/COMMIT/ROLLBACK transactions. // // To use transactions, you must first create either an OptimisticTransactionDB // or a TransactionDB. See examples/[optimistic_]transaction_example.cc for // more information. // // To create a transaction, use [Optimistic]TransactionDB::BeginTransaction(). // // It is up to the caller to synchronize access to this object. // // See examples/transaction_example.cc for some simple examples. // // TODO(agiardullo): Not yet implemented // -PerfContext statistics // -Support for using Transactions with DBWithTTL class Transaction { public: virtual ~Transaction() {} // If a transaction has a snapshot set, the transaction will ensure that // any keys successfully written(or fetched via GetForUpdate()) have not // been modified outside of this transaction since the time the snapshot was // set. // If a snapshot has not been set, the transaction guarantees that keys have // not been modified since the time each key was first written (or fetched via // GetForUpdate()). // // Using SetSnapshot() will provide stricter isolation guarantees at the // expense of potentially more transaction failures due to conflicts with // other writes. // // Calling SetSnapshot() has no effect on keys written before this function // has been called. // // SetSnapshot() may be called multiple times if you would like to change // the snapshot used for different operations in this transaction. // // Calling SetSnapshot will not affect the version of Data returned by Get() // methods. See Transaction::Get() for more details. virtual void SetSnapshot() = 0; // Similar to SetSnapshot(), but will not change the current snapshot // until Put/Merge/Delete/GetForUpdate/MultigetForUpdate is called. // By calling this function, the transaction will essentially call // SetSnapshot() for you right before performing the next write/GetForUpdate. // // Calling SetSnapshotOnNextOperation() will not affect what snapshot is // returned by GetSnapshot() until the next write/GetForUpdate is executed. // // When the snapshot is created the notifier's SnapshotCreated method will // be called so that the caller can get access to the snapshot. // // This is an optimization to reduce the likelihood of conflicts that // could occur in between the time SetSnapshot() is called and the first // write/GetForUpdate operation. Eg, this prevents the following // race-condition: // // txn1->SetSnapshot(); // txn2->Put("A", ...); // txn2->Commit(); // txn1->GetForUpdate(opts, "A", ...); // FAIL! virtual void SetSnapshotOnNextOperation( std::shared_ptr notifier = nullptr) = 0; // Returns the Snapshot created by the last call to SetSnapshot(). // // REQUIRED: The returned Snapshot is only valid up until the next time // SetSnapshot()/SetSnapshotOnNextSavePoint() is called, ClearSnapshot() // is called, or the Transaction is deleted. virtual const Snapshot* GetSnapshot() const = 0; // Clears the current snapshot (i.e. no snapshot will be 'set') // // This removes any snapshot that currently exists or is set to be created // on the next update operation (SetSnapshotOnNextOperation). // // Calling ClearSnapshot() has no effect on keys written before this function // has been called. // // If a reference to a snapshot was retrieved via GetSnapshot(), it will no // longer be valid and should be discarded after a call to ClearSnapshot(). virtual void ClearSnapshot() = 0; // Write all batched keys to the db atomically. // // Returns OK on success. // // May return any error status that could be returned by DB:Write(). // // If this transaction was created by an OptimisticTransactionDB(), // Status::Busy() may be returned if the transaction could not guarantee // that there are no write conflicts. Status::TryAgain() may be returned // if the memtable history size is not large enough // (See max_write_buffer_number_to_maintain). // // If this transaction was created by a TransactionDB(), Status::Expired() // may be returned if this transaction has lived for longer than // TransactionOptions.expiration. virtual Status Commit() = 0; // Discard all batched writes in this transaction. virtual void Rollback() = 0; // Records the state of the transaction for future calls to // RollbackToSavePoint(). May be called multiple times to set multiple save // points. virtual void SetSavePoint() = 0; // Undo all operations in this transaction (Put, Merge, Delete, PutLogData) // since the most recent call to SetSavePoint() and removes the most recent // SetSavePoint(). // If there is no previous call to SetSavePoint(), returns Status::NotFound() virtual Status RollbackToSavePoint() = 0; // This function is similar to DB::Get() except it will also read pending // changes in this transaction. Currently, this function will return // Status::MergeInProgress if the most recent write to the queried key in // this batch is a Merge. // // If read_options.snapshot is not set, the current version of the key will // be read. Calling SetSnapshot() does not affect the version of the data // returned. // // Note that setting read_options.snapshot will affect what is read from the // DB but will NOT change which keys are read from this transaction (the keys // in this transaction do not yet belong to any snapshot and will be fetched // regardless). virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) = 0; virtual Status Get(const ReadOptions& options, const Slice& key, std::string* value) = 0; virtual std::vector MultiGet( const ReadOptions& options, const std::vector& column_family, const std::vector& keys, std::vector* values) = 0; virtual std::vector MultiGet(const ReadOptions& options, const std::vector& keys, std::vector* values) = 0; // Read this key and ensure that this transaction will only // be able to be committed if this key is not written outside this // transaction after it has first been read (or after the snapshot if a // snapshot is set in this transaction). The transaction behavior is the // same regardless of whether the key exists or not. // // Note: Currently, this function will return Status::MergeInProgress // if the most recent write to the queried key in this batch is a Merge. // // The values returned by this function are similar to Transaction::Get(). // If value==nullptr, then this function will not read any data, but will // still ensure that this key cannot be written to by outside of this // transaction. // // If this transaction was created by an OptimisticTransaction, GetForUpdate() // could cause commit() to fail. Otherwise, it could return any error // that could be returned by DB::Get(). // // If this transaction was created by a TransactionDB, it can return // Status::OK() on success, // Status::Busy() if there is a write conflict, // Status::TimedOut() if a lock could not be acquired, // Status::TryAgain() if the memtable history size is not large enough // (See max_write_buffer_number_to_maintain) // Status::MergeInProgress() if merge operations cannot be resolved. // or other errors if this key could not be read. virtual Status GetForUpdate(const ReadOptions& options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value) = 0; virtual Status GetForUpdate(const ReadOptions& options, const Slice& key, std::string* value) = 0; virtual std::vector MultiGetForUpdate( const ReadOptions& options, const std::vector& column_family, const std::vector& keys, std::vector* values) = 0; virtual std::vector MultiGetForUpdate( const ReadOptions& options, const std::vector& keys, std::vector* values) = 0; // Returns an iterator that will iterate on all keys in the default // column family including both keys in the DB and uncommitted keys in this // transaction. // // Setting read_options.snapshot will affect what is read from the // DB but will NOT change which keys are read from this transaction (the keys // in this transaction do not yet belong to any snapshot and will be fetched // regardless). // // Caller is responsible for deleting the returned Iterator. // // The returned iterator is only valid until Commit(), Rollback(), or // RollbackToSavePoint() is called. virtual Iterator* GetIterator(const ReadOptions& read_options) = 0; virtual Iterator* GetIterator(const ReadOptions& read_options, ColumnFamilyHandle* column_family) = 0; // Put, Merge, Delete, and SingleDelete behave similarly to the corresponding // functions in WriteBatch, but will also do conflict checking on the // keys being written. // // If this Transaction was created on an OptimisticTransactionDB, these // functions should always return Status::OK(). // // If this Transaction was created on a TransactionDB, the status returned // can be: // Status::OK() on success, // Status::Busy() if there is a write conflict, // Status::TimedOut() if a lock could not be acquired, // Status::TryAgain() if the memtable history size is not large enough // (See max_write_buffer_number_to_maintain) // or other errors on unexpected failures. virtual Status Put(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual Status Put(const Slice& key, const Slice& value) = 0; virtual Status Put(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) = 0; virtual Status Put(const SliceParts& key, const SliceParts& value) = 0; virtual Status Merge(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual Status Merge(const Slice& key, const Slice& value) = 0; virtual Status Delete(ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual Status Delete(const Slice& key) = 0; virtual Status Delete(ColumnFamilyHandle* column_family, const SliceParts& key) = 0; virtual Status Delete(const SliceParts& key) = 0; virtual Status SingleDelete(ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual Status SingleDelete(const Slice& key) = 0; virtual Status SingleDelete(ColumnFamilyHandle* column_family, const SliceParts& key) = 0; virtual Status SingleDelete(const SliceParts& key) = 0; // PutUntracked() will write a Put to the batch of operations to be committed // in this transaction. This write will only happen if this transaction // gets committed successfully. But unlike Transaction::Put(), // no conflict checking will be done for this key. // // If this Transaction was created on a TransactionDB, this function will // still acquire locks necessary to make sure this write doesn't cause // conflicts in other transactions and may return Status::Busy(). virtual Status PutUntracked(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual Status PutUntracked(const Slice& key, const Slice& value) = 0; virtual Status PutUntracked(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) = 0; virtual Status PutUntracked(const SliceParts& key, const SliceParts& value) = 0; virtual Status MergeUntracked(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual Status MergeUntracked(const Slice& key, const Slice& value) = 0; virtual Status DeleteUntracked(ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual Status DeleteUntracked(const Slice& key) = 0; virtual Status DeleteUntracked(ColumnFamilyHandle* column_family, const SliceParts& key) = 0; virtual Status DeleteUntracked(const SliceParts& key) = 0; // Similar to WriteBatch::PutLogData virtual void PutLogData(const Slice& blob) = 0; // By default, all Put/Merge/Delete operations will be indexed in the // transaction so that Get/GetForUpdate/GetIterator can search for these // keys. // // If the caller does not want to fetch the keys about to be written, // they may want to avoid indexing as a performance optimization. // Calling DisableIndexing() will turn off indexing for all future // Put/Merge/Delete operations until EnableIndexing() is called. // // If a key is Put/Merge/Deleted after DisableIndexing is called and then // is fetched via Get/GetForUpdate/GetIterator, the result of the fetch is // undefined. virtual void DisableIndexing() = 0; virtual void EnableIndexing() = 0; // Returns the number of distinct Keys being tracked by this transaction. // If this transaction was created by a TransactinDB, this is the number of // keys that are currently locked by this transaction. // If this transaction was created by an OptimisticTransactionDB, this is the // number of keys that need to be checked for conflicts at commit time. virtual uint64_t GetNumKeys() const = 0; // Returns the number of Puts/Deletes/Merges that have been applied to this // transaction so far. virtual uint64_t GetNumPuts() const = 0; virtual uint64_t GetNumDeletes() const = 0; virtual uint64_t GetNumMerges() const = 0; // Returns the elapsed time in milliseconds since this Transaction began. virtual uint64_t GetElapsedTime() const = 0; // Fetch the underlying write batch that contains all pending changes to be // committed. // // Note: You should not write or delete anything from the batch directly and // should only use the the functions in the Transaction class to // write to this transaction. virtual WriteBatchWithIndex* GetWriteBatch() = 0; // Change the value of TransactionOptions.lock_timeout (in milliseconds) for // this transaction. // Has no effect on OptimisticTransactions. virtual void SetLockTimeout(int64_t timeout) = 0; // Return the WriteOptions that will be used during Commit() virtual const WriteOptions* GetWriteOptions() = 0; // Reset the WriteOptions that will be used during Commit(). virtual void SetWriteOptions(const WriteOptions& write_options) = 0; protected: explicit Transaction(const TransactionDB* db) {} Transaction() {} private: // No copying allowed Transaction(const Transaction&); void operator=(const Transaction&); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/transaction_db.h000066400000000000000000000120461267530634000234110ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/comparator.h" #include "rocksdb/db.h" #include "rocksdb/utilities/stackable_db.h" #include "rocksdb/utilities/transaction.h" // Database with Transaction support. // // See transaction.h and examples/transaction_example.cc namespace rocksdb { class TransactionDBMutexFactory; struct TransactionDBOptions { // Specifies the maximum number of keys that can be locked at the same time // per column family. // If the number of locked keys is greater than max_num_locks, transaction // writes (or GetForUpdate) will return an error. // If this value is not positive, no limit will be enforced. int64_t max_num_locks = -1; // Increasing this value will increase the concurrency by dividing the lock // table (per column family) into more sub-tables, each with their own // separate // mutex. size_t num_stripes = 16; // If positive, specifies the default wait timeout in milliseconds when // a transaction attempts to lock a key if not specified by // TransactionOptions::lock_timeout. // // If 0, no waiting is done if a lock cannot instantly be acquired. // If negative, there is no timeout. Not using a timeout is not recommended // as it can lead to deadlocks. Currently, there is no deadlock-detection to // recover // from a deadlock. int64_t transaction_lock_timeout = 1000; // 1 second // If positive, specifies the wait timeout in milliseconds when writing a key // OUTSIDE of a transaction (ie by calling DB::Put(),Merge(),Delete(),Write() // directly). // If 0, no waiting is done if a lock cannot instantly be acquired. // If negative, there is no timeout and will block indefinitely when acquiring // a lock. // // Not using a a timeout can lead to deadlocks. Currently, there // is no deadlock-detection to recover from a deadlock. While DB writes // cannot deadlock with other DB writes, they can deadlock with a transaction. // A negative timeout should only be used if all transactions have an small // expiration set. int64_t default_lock_timeout = 1000; // 1 second // If set, the TransactionDB will use this implemenation of a mutex and // condition variable for all transaction locking instead of the default // mutex/condvar implementation. std::shared_ptr custom_mutex_factory; }; struct TransactionOptions { // Setting set_snapshot=true is the same as calling // Transaction::SetSnapshot(). bool set_snapshot = false; // TODO(agiardullo): TransactionDB does not yet support comparators that allow // two non-equal keys to be equivalent. Ie, cmp->Compare(a,b) should only // return 0 if // a.compare(b) returns 0. // If positive, specifies the wait timeout in milliseconds when // a transaction attempts to lock a key. // // If 0, no waiting is done if a lock cannot instantly be acquired. // If negative, TransactionDBOptions::transaction_lock_timeout will be used. int64_t lock_timeout = -1; // Expiration duration in milliseconds. If non-negative, transactions that // last longer than this many milliseconds will fail to commit. If not set, // a forgotten transaction that is never committed, rolled back, or deleted // will never relinquish any locks it holds. This could prevent keys from // being // written by other writers. // // TODO(agiardullo): Improve performance of checking expiration time. int64_t expiration = -1; }; class TransactionDB : public StackableDB { public: // Open a TransactionDB similar to DB::Open(). static Status Open(const Options& options, const TransactionDBOptions& txn_db_options, const std::string& dbname, TransactionDB** dbptr); static Status Open(const DBOptions& db_options, const TransactionDBOptions& txn_db_options, const std::string& dbname, const std::vector& column_families, std::vector* handles, TransactionDB** dbptr); virtual ~TransactionDB() {} // Starts a new Transaction. Passing set_snapshot=true has the same effect // as calling Transaction::SetSnapshot(). // // Caller should delete the returned transaction after calling // Transaction::Commit() or Transaction::Rollback(). virtual Transaction* BeginTransaction( const WriteOptions& write_options, const TransactionOptions& txn_options = TransactionOptions()) = 0; protected: // To Create an TransactionDB, call Open() explicit TransactionDB(DB* db) : StackableDB(db) {} private: // No copying allowed TransactionDB(const TransactionDB&); void operator=(const TransactionDB&); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/transaction_db_mutex.h000066400000000000000000000062701267530634000246350ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/status.h" namespace rocksdb { // TransactionDBMutex and TransactionDBCondVar APIs allows applications to // implement custom mutexes and condition variables to be used by a // TransactionDB when locking keys. // // To open a TransactionDB with a custom TransactionDBMutexFactory, set // TransactionDBOptions.custom_mutex_factory. class TransactionDBMutex { public: virtual ~TransactionDBMutex() {} // Attempt to acquire lock. Return OK on success, or other Status on failure. // If returned status is OK, TransactionDB will eventually call UnLock(). virtual Status Lock() = 0; // Attempt to acquire lock. If timeout is non-negative, operation may be // failed after this many microseconds. // Returns OK on success, // TimedOut if timed out, // or other Status on failure. // If returned status is OK, TransactionDB will eventually call UnLock(). virtual Status TryLockFor(int64_t timeout_time) = 0; // Unlock Mutex that was successfully locked by Lock() or TryLockUntil() virtual void UnLock() = 0; }; class TransactionDBCondVar { public: virtual ~TransactionDBCondVar() {} // Block current thread until condition variable is notified by a call to // Notify() or NotifyAll(). Wait() will be called with mutex locked. // Returns OK if notified. // Returns non-OK if TransactionDB should stop waiting and fail the operation. // May return OK spuriously even if not notified. virtual Status Wait(std::shared_ptr mutex) = 0; // Block current thread until condition variable is notified by a call to // Notify() or NotifyAll(), or if the timeout is reached. // Wait() will be called with mutex locked. // // If timeout is non-negative, operation should be failed after this many // microseconds. // If implementing a custom version of this class, the implementation may // choose to ignore the timeout. // // Returns OK if notified. // Returns TimedOut if timeout is reached. // Returns other status if TransactionDB should otherwis stop waiting and // fail the operation. // May return OK spuriously even if not notified. virtual Status WaitFor(std::shared_ptr mutex, int64_t timeout_time) = 0; // If any threads are waiting on *this, unblock at least one of the // waiting threads. virtual void Notify() = 0; // Unblocks all threads waiting on *this. virtual void NotifyAll() = 0; }; // Factory class that can allocate mutexes and condition variables. class TransactionDBMutexFactory { public: // Create a TransactionDBMutex object. virtual std::shared_ptr AllocateMutex() = 0; // Create a TransactionDBCondVar object. virtual std::shared_ptr AllocateCondVar() = 0; virtual ~TransactionDBMutexFactory() {} }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/utility_db.h000066400000000000000000000022321267530634000225630ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/utilities/stackable_db.h" #include "rocksdb/utilities/db_ttl.h" #include "rocksdb/db.h" namespace rocksdb { // Please don't use this class. It's deprecated class UtilityDB { public: // This function is here only for backwards compatibility. Please use the // functions defined in DBWithTTl (rocksdb/utilities/db_ttl.h) // (deprecated) #if defined(__GNUC__) || defined(__clang__) __attribute__((deprecated)) #elif _WIN32 __declspec(deprecated) #endif static Status OpenTtlDB(const Options& options, const std::string& name, StackableDB** dbptr, int32_t ttl = 0, bool read_only = false); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/utilities/write_batch_with_index.h000066400000000000000000000164561267530634000251450ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // A WriteBatchWithIndex with a binary searchable index built for all the keys // inserted. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/comparator.h" #include "rocksdb/iterator.h" #include "rocksdb/slice.h" #include "rocksdb/status.h" #include "rocksdb/write_batch.h" #include "rocksdb/write_batch_base.h" namespace rocksdb { class ColumnFamilyHandle; class Comparator; class DB; struct ReadOptions; struct DBOptions; enum WriteType { kPutRecord, kMergeRecord, kDeleteRecord, kSingleDeleteRecord, kLogDataRecord }; // an entry for Put, Merge, Delete, or SingleDelete entry for write batches. // Used in WBWIIterator. struct WriteEntry { WriteType type; Slice key; Slice value; }; // Iterator of one column family out of a WriteBatchWithIndex. class WBWIIterator { public: virtual ~WBWIIterator() {} virtual bool Valid() const = 0; virtual void SeekToFirst() = 0; virtual void SeekToLast() = 0; virtual void Seek(const Slice& key) = 0; virtual void Next() = 0; virtual void Prev() = 0; // the return WriteEntry is only valid until the next mutation of // WriteBatchWithIndex virtual WriteEntry Entry() const = 0; virtual Status status() const = 0; }; // A WriteBatchWithIndex with a binary searchable index built for all the keys // inserted. // In Put(), Merge() Delete(), or SingleDelete(), the same function of the // wrapped will be called. At the same time, indexes will be built. // By calling GetWriteBatch(), a user will get the WriteBatch for the data // they inserted, which can be used for DB::Write(). // A user can call NewIterator() to create an iterator. class WriteBatchWithIndex : public WriteBatchBase { public: // backup_index_comparator: the backup comparator used to compare keys // within the same column family, if column family is not given in the // interface, or we can't find a column family from the column family handle // passed in, backup_index_comparator will be used for the column family. // reserved_bytes: reserved bytes in underlying WriteBatch // overwrite_key: if true, overwrite the key in the index when inserting // the same key as previously, so iterator will never // show two entries with the same key. explicit WriteBatchWithIndex( const Comparator* backup_index_comparator = BytewiseComparator(), size_t reserved_bytes = 0, bool overwrite_key = false); virtual ~WriteBatchWithIndex(); using WriteBatchBase::Put; void Put(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override; void Put(const Slice& key, const Slice& value) override; using WriteBatchBase::Merge; void Merge(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override; void Merge(const Slice& key, const Slice& value) override; using WriteBatchBase::Delete; void Delete(ColumnFamilyHandle* column_family, const Slice& key) override; void Delete(const Slice& key) override; using WriteBatchBase::SingleDelete; void SingleDelete(ColumnFamilyHandle* column_family, const Slice& key) override; void SingleDelete(const Slice& key) override; using WriteBatchBase::PutLogData; void PutLogData(const Slice& blob) override; using WriteBatchBase::Clear; void Clear() override; using WriteBatchBase::GetWriteBatch; WriteBatch* GetWriteBatch() override; // Create an iterator of a column family. User can call iterator.Seek() to // search to the next entry of or after a key. Keys will be iterated in the // order given by index_comparator. For multiple updates on the same key, // each update will be returned as a separate entry, in the order of update // time. // // The returned iterator should be deleted by the caller. WBWIIterator* NewIterator(ColumnFamilyHandle* column_family); // Create an iterator of the default column family. WBWIIterator* NewIterator(); // Will create a new Iterator that will use WBWIIterator as a delta and // base_iterator as base. // // This function is only supported if the WriteBatchWithIndex was // constructed with overwrite_key=true. // // The returned iterator should be deleted by the caller. // The base_iterator is now 'owned' by the returned iterator. Deleting the // returned iterator will also delete the base_iterator. Iterator* NewIteratorWithBase(ColumnFamilyHandle* column_family, Iterator* base_iterator); // default column family Iterator* NewIteratorWithBase(Iterator* base_iterator); // Similar to DB::Get() but will only read the key from this batch. // If the batch does not have enough data to resolve Merge operations, // MergeInProgress status may be returned. Status GetFromBatch(ColumnFamilyHandle* column_family, const DBOptions& options, const Slice& key, std::string* value); // Similar to previous function but does not require a column_family. // Note: An InvalidArgument status will be returned if there are any Merge // operators for this key. Use previous method instead. Status GetFromBatch(const DBOptions& options, const Slice& key, std::string* value) { return GetFromBatch(nullptr, options, key, value); } // Similar to DB::Get() but will also read writes from this batch. // // This function will query both this batch and the DB and then merge // the results using the DB's merge operator (if the batch contains any // merge requests). // // Setting read_options.snapshot will affect what is read from the DB // but will NOT change which keys are read from the batch (the keys in // this batch do not yet belong to any snapshot and will be fetched // regardless). Status GetFromBatchAndDB(DB* db, const ReadOptions& read_options, const Slice& key, std::string* value); Status GetFromBatchAndDB(DB* db, const ReadOptions& read_options, ColumnFamilyHandle* column_family, const Slice& key, std::string* value); // Records the state of the batch for future calls to RollbackToSavePoint(). // May be called multiple times to set multiple save points. void SetSavePoint() override; // Remove all entries in this batch (Put, Merge, Delete, SingleDelete, // PutLogData) since the most recent call to SetSavePoint() and removes the // most recent save point. // If there is no previous call to SetSavePoint(), behaves the same as // Clear(). // // Calling RollbackToSavePoint invalidates any open iterators on this batch. // // Returns Status::OK() on success, // Status::NotFound() if no previous call to SetSavePoint(), // or other Status on corruption. Status RollbackToSavePoint() override; private: struct Rep; Rep* rep; }; } // namespace rocksdb #endif // !ROCKSDB_LITE rocksdb-4.5.1/include/rocksdb/version.h000066400000000000000000000012341267530634000200660ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #define ROCKSDB_MAJOR 4 #define ROCKSDB_MINOR 5 #define ROCKSDB_PATCH 1 // Do not use these. We made the mistake of declaring macros starting with // double underscore. Now we have to live with our choice. We'll deprecate these // at some point #define __ROCKSDB_MAJOR__ ROCKSDB_MAJOR #define __ROCKSDB_MINOR__ ROCKSDB_MINOR #define __ROCKSDB_PATCH__ ROCKSDB_PATCH rocksdb-4.5.1/include/rocksdb/wal_filter.h000066400000000000000000000053521267530634000205360ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once namespace rocksdb { class WriteBatch; // WALFilter allows an application to inspect write-ahead-log (WAL) // records or modify their processing on recovery. // Please see the details below. class WalFilter { public: enum class WalProcessingOption { // Continue processing as usual kContinueProcessing = 0, // Ignore the current record but continue processing of log(s) kIgnoreCurrentRecord = 1, // Stop replay of logs and discard logs // Logs won't be replayed on subsequent recovery kStopReplay = 2, // Corrupted record detected by filter kCorruptedRecord = 3, // Marker for enum count kWalProcessingOptionMax = 4 }; virtual ~WalFilter() {} // LogRecord is invoked for each log record encountered for all the logs // during replay on logs on recovery. This method can be used to: // * inspect the record (using the batch parameter) // * ignoring current record // (by returning WalProcessingOption::kIgnoreCurrentRecord) // * reporting corrupted record // (by returning WalProcessingOption::kCorruptedRecord) // * stop log replay // (by returning kStop replay) - please note that this implies // discarding the logs from current record onwards. // // @params batch batch encountered in the log during recovery // @params new_batch new_batch to populate if filter wants to change // the batch (for example to filter some records out, // or alter some records). // Please note that the new batch MUST NOT contain // more records than original, else recovery would // be failed. // @params batch_changed Whether batch was changed by the filter. // It must be set to true if new_batch was populated, // else new_batch has no effect. // @returns Processing option for the current record. // Please see WalProcessingOption enum above for // details. virtual WalProcessingOption LogRecord(const WriteBatch& batch, WriteBatch* new_batch, bool* batch_changed) const = 0; // Returns a name that identifies this WAL filter. // The name will be printed to LOG file on start up for diagnosis. virtual const char* Name() const = 0; }; } // namespace rocksdb rocksdb-4.5.1/include/rocksdb/write_batch.h000066400000000000000000000217421267530634000207020ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // WriteBatch holds a collection of updates to apply atomically to a DB. // // The updates are applied in the order in which they are added // to the WriteBatch. For example, the value of "key" will be "v3" // after the following batch is written: // // batch.Put("key", "v1"); // batch.Delete("key"); // batch.Put("key", "v2"); // batch.Put("key", "v3"); // // Multiple threads can invoke const methods on a WriteBatch without // external synchronization, but if any of the threads may call a // non-const method, all threads accessing the same WriteBatch must use // external synchronization. #ifndef STORAGE_ROCKSDB_INCLUDE_WRITE_BATCH_H_ #define STORAGE_ROCKSDB_INCLUDE_WRITE_BATCH_H_ #include #include #include #include #include "rocksdb/status.h" #include "rocksdb/write_batch_base.h" namespace rocksdb { class Slice; class ColumnFamilyHandle; struct SavePoints; struct SliceParts; class WriteBatch : public WriteBatchBase { public: explicit WriteBatch(size_t reserved_bytes = 0); ~WriteBatch(); using WriteBatchBase::Put; // Store the mapping "key->value" in the database. void Put(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override; void Put(const Slice& key, const Slice& value) override { Put(nullptr, key, value); } // Variant of Put() that gathers output like writev(2). The key and value // that will be written to the database are concatentations of arrays of // slices. void Put(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) override; void Put(const SliceParts& key, const SliceParts& value) override { Put(nullptr, key, value); } using WriteBatchBase::Delete; // If the database contains a mapping for "key", erase it. Else do nothing. void Delete(ColumnFamilyHandle* column_family, const Slice& key) override; void Delete(const Slice& key) override { Delete(nullptr, key); } // variant that takes SliceParts void Delete(ColumnFamilyHandle* column_family, const SliceParts& key) override; void Delete(const SliceParts& key) override { Delete(nullptr, key); } using WriteBatchBase::SingleDelete; // WriteBatch implementation of DB::SingleDelete(). See db.h. void SingleDelete(ColumnFamilyHandle* column_family, const Slice& key) override; void SingleDelete(const Slice& key) override { SingleDelete(nullptr, key); } // variant that takes SliceParts void SingleDelete(ColumnFamilyHandle* column_family, const SliceParts& key) override; void SingleDelete(const SliceParts& key) override { SingleDelete(nullptr, key); } using WriteBatchBase::Merge; // Merge "value" with the existing value of "key" in the database. // "key->merge(existing, value)" void Merge(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) override; void Merge(const Slice& key, const Slice& value) override { Merge(nullptr, key, value); } // variant that takes SliceParts void Merge(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value) override; void Merge(const SliceParts& key, const SliceParts& value) override { Merge(nullptr, key, value); } using WriteBatchBase::PutLogData; // Append a blob of arbitrary size to the records in this batch. The blob will // be stored in the transaction log but not in any other file. In particular, // it will not be persisted to the SST files. When iterating over this // WriteBatch, WriteBatch::Handler::LogData will be called with the contents // of the blob as it is encountered. Blobs, puts, deletes, and merges will be // encountered in the same order in thich they were inserted. The blob will // NOT consume sequence number(s) and will NOT increase the count of the batch // // Example application: add timestamps to the transaction log for use in // replication. void PutLogData(const Slice& blob) override; using WriteBatchBase::Clear; // Clear all updates buffered in this batch. void Clear() override; // Records the state of the batch for future calls to RollbackToSavePoint(). // May be called multiple times to set multiple save points. void SetSavePoint() override; // Remove all entries in this batch (Put, Merge, Delete, PutLogData) since the // most recent call to SetSavePoint() and removes the most recent save point. // If there is no previous call to SetSavePoint(), Status::NotFound() // will be returned. // Oterwise returns Status::OK(). Status RollbackToSavePoint() override; // Support for iterating over the contents of a batch. class Handler { public: virtual ~Handler(); // default implementation will just call Put without column family for // backwards compatibility. If the column family is not default, // the function is noop virtual Status PutCF(uint32_t column_family_id, const Slice& key, const Slice& value) { if (column_family_id == 0) { // Put() historically doesn't return status. We didn't want to be // backwards incompatible so we didn't change the return status // (this is a public API). We do an ordinary get and return Status::OK() Put(key, value); return Status::OK(); } return Status::InvalidArgument( "non-default column family and PutCF not implemented"); } virtual void Put(const Slice& /*key*/, const Slice& /*value*/) {} virtual Status DeleteCF(uint32_t column_family_id, const Slice& key) { if (column_family_id == 0) { Delete(key); return Status::OK(); } return Status::InvalidArgument( "non-default column family and DeleteCF not implemented"); } virtual void Delete(const Slice& /*key*/) {} virtual Status SingleDeleteCF(uint32_t column_family_id, const Slice& key) { if (column_family_id == 0) { SingleDelete(key); return Status::OK(); } return Status::InvalidArgument( "non-default column family and SingleDeleteCF not implemented"); } virtual void SingleDelete(const Slice& /*key*/) {} // Merge and LogData are not pure virtual. Otherwise, we would break // existing clients of Handler on a source code level. The default // implementation of Merge does nothing. virtual Status MergeCF(uint32_t column_family_id, const Slice& key, const Slice& value) { if (column_family_id == 0) { Merge(key, value); return Status::OK(); } return Status::InvalidArgument( "non-default column family and MergeCF not implemented"); } virtual void Merge(const Slice& /*key*/, const Slice& /*value*/) {} // The default implementation of LogData does nothing. virtual void LogData(const Slice& blob); // Continue is called by WriteBatch::Iterate. If it returns false, // iteration is halted. Otherwise, it continues iterating. The default // implementation always returns true. virtual bool Continue(); }; Status Iterate(Handler* handler) const; // Retrieve the serialized version of this batch. const std::string& Data() const { return rep_; } // Retrieve data size of the batch. size_t GetDataSize() const { return rep_.size(); } // Returns the number of updates in the batch int Count() const; // Returns true if PutCF will be called during Iterate bool HasPut() const; // Returns true if DeleteCF will be called during Iterate bool HasDelete() const; // Returns true if SingleDeleteCF will be called during Iterate bool HasSingleDelete() const; // Returns trie if MergeCF will be called during Iterate bool HasMerge() const; using WriteBatchBase::GetWriteBatch; WriteBatch* GetWriteBatch() override { return this; } // Constructor with a serialized string object explicit WriteBatch(const std::string& rep); WriteBatch(const WriteBatch& src); WriteBatch(WriteBatch&& src); WriteBatch& operator=(const WriteBatch& src); WriteBatch& operator=(WriteBatch&& src); private: friend class WriteBatchInternal; SavePoints* save_points_; // For HasXYZ. Mutable to allow lazy computation of results mutable std::atomic content_flags_; // Performs deferred computation of content_flags if necessary uint32_t ComputeContentFlags() const; protected: std::string rep_; // See comment in write_batch.cc for the format of rep_ // Intentionally copyable }; } // namespace rocksdb #endif // STORAGE_ROCKSDB_INCLUDE_WRITE_BATCH_H_ rocksdb-4.5.1/include/rocksdb/write_batch_base.h000066400000000000000000000103431267530634000216670ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once namespace rocksdb { class Slice; class Status; class ColumnFamilyHandle; class WriteBatch; struct SliceParts; // Abstract base class that defines the basic interface for a write batch. // See WriteBatch for a basic implementation and WrithBatchWithIndex for an // indexed implemenation. class WriteBatchBase { public: virtual ~WriteBatchBase() {} // Store the mapping "key->value" in the database. virtual void Put(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual void Put(const Slice& key, const Slice& value) = 0; // Variant of Put() that gathers output like writev(2). The key and value // that will be written to the database are concatentations of arrays of // slices. virtual void Put(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value); virtual void Put(const SliceParts& key, const SliceParts& value); // Merge "value" with the existing value of "key" in the database. // "key->merge(existing, value)" virtual void Merge(ColumnFamilyHandle* column_family, const Slice& key, const Slice& value) = 0; virtual void Merge(const Slice& key, const Slice& value) = 0; // variant that takes SliceParts virtual void Merge(ColumnFamilyHandle* column_family, const SliceParts& key, const SliceParts& value); virtual void Merge(const SliceParts& key, const SliceParts& value); // If the database contains a mapping for "key", erase it. Else do nothing. virtual void Delete(ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual void Delete(const Slice& key) = 0; // variant that takes SliceParts virtual void Delete(ColumnFamilyHandle* column_family, const SliceParts& key); virtual void Delete(const SliceParts& key); // If the database contains a mapping for "key", erase it. Expects that the // key was not overwritten. Else do nothing. virtual void SingleDelete(ColumnFamilyHandle* column_family, const Slice& key) = 0; virtual void SingleDelete(const Slice& key) = 0; // variant that takes SliceParts virtual void SingleDelete(ColumnFamilyHandle* column_family, const SliceParts& key); virtual void SingleDelete(const SliceParts& key); // Append a blob of arbitrary size to the records in this batch. The blob will // be stored in the transaction log but not in any other file. In particular, // it will not be persisted to the SST files. When iterating over this // WriteBatch, WriteBatch::Handler::LogData will be called with the contents // of the blob as it is encountered. Blobs, puts, deletes, and merges will be // encountered in the same order in thich they were inserted. The blob will // NOT consume sequence number(s) and will NOT increase the count of the batch // // Example application: add timestamps to the transaction log for use in // replication. virtual void PutLogData(const Slice& blob) = 0; // Clear all updates buffered in this batch. virtual void Clear() = 0; // Covert this batch into a WriteBatch. This is an abstracted way of // converting any WriteBatchBase(eg WriteBatchWithIndex) into a basic // WriteBatch. virtual WriteBatch* GetWriteBatch() = 0; // Records the state of the batch for future calls to RollbackToSavePoint(). // May be called multiple times to set multiple save points. virtual void SetSavePoint() = 0; // Remove all entries in this batch (Put, Merge, Delete, PutLogData) since the // most recent call to SetSavePoint() and removes the most recent save point. // If there is no previous call to SetSavePoint(), behaves the same as // Clear(). virtual Status RollbackToSavePoint() = 0; }; } // namespace rocksdb rocksdb-4.5.1/java/000077500000000000000000000000001267530634000140775ustar00rootroot00000000000000rocksdb-4.5.1/java/CMakeLists.txt000066400000000000000000000145351267530634000166470ustar00rootroot00000000000000set(JNI_NATIVE_SOURCES rocksjni/backupenginejni.cc rocksjni/backupablejni.cc rocksjni/checkpoint.cc rocksjni/columnfamilyhandle.cc rocksjni/compaction_filter.cc rocksjni/comparator.cc rocksjni/comparatorjnicallback.cc rocksjni/env.cc rocksjni/filter.cc rocksjni/iterator.cc rocksjni/loggerjnicallback.cc rocksjni/memtablejni.cc rocksjni/merge_operator.cc rocksjni/options.cc rocksjni/ratelimiterjni.cc rocksjni/remove_emptyvalue_compactionfilterjni.cc rocksjni/restorejni.cc rocksjni/rocksjni.cc rocksjni/slice.cc rocksjni/snapshot.cc rocksjni/statistics.cc rocksjni/table.cc rocksjni/transaction_log.cc rocksjni/ttl.cc rocksjni/write_batch.cc rocksjni/writebatchhandlerjnicallback.cc rocksjni/write_batch_with_index.cc rocksjni/write_batch_test.cc ) set(NATIVE_JAVA_CLASSES org.rocksdb.AbstractCompactionFilter org.rocksdb.AbstractComparator org.rocksdb.AbstractSlice org.rocksdb.BackupEngine org.rocksdb.BackupableDB org.rocksdb.BackupableDBOptions org.rocksdb.BlockBasedTableConfig org.rocksdb.BloomFilter org.rocksdb.Checkpoint org.rocksdb.ColumnFamilyHandle org.rocksdb.ColumnFamilyOptions org.rocksdb.Comparator org.rocksdb.ComparatorOptions org.rocksdb.DBOptions org.rocksdb.DirectComparator org.rocksdb.DirectSlice org.rocksdb.Env org.rocksdb.FlushOptions org.rocksdb.Filter org.rocksdb.GenericRateLimiterConfig org.rocksdb.HashLinkedListMemTableConfig org.rocksdb.HashSkipListMemTableConfig org.rocksdb.Logger org.rocksdb.MergeOperator org.rocksdb.Options org.rocksdb.PlainTableConfig org.rocksdb.ReadOptions org.rocksdb.RemoveEmptyValueCompactionFilter org.rocksdb.RestoreBackupableDB org.rocksdb.RestoreOptions org.rocksdb.RocksDB org.rocksdb.RocksEnv org.rocksdb.RocksIterator org.rocksdb.RocksMemEnv org.rocksdb.SkipListMemTableConfig org.rocksdb.Slice org.rocksdb.Statistics org.rocksdb.TransactionLogIterator org.rocksdb.TtlDB org.rocksdb.VectorMemTableConfig org.rocksdb.Snapshot org.rocksdb.StringAppendOperator org.rocksdb.WriteBatch org.rocksdb.WriteBatch.Handler org.rocksdb.WriteOptions org.rocksdb.WriteBatchWithIndex org.rocksdb.WBWIRocksIterator org.rocksdb.WriteBatchTest org.rocksdb.WriteBatchTestInternalHelper ) include_directories($ENV{JAVA_HOME}/include) include_directories($ENV{JAVA_HOME}/include/win32) include_directories(${PROJECT_SOURCE_DIR}/java) set(JAVA_TEST_LIBDIR ${PROJECT_SOURCE_DIR}/java/test-libs) set(JAVA_TMP_JAR ${JAVA_TEST_LIBDIR}/tmp.jar) set(JAVA_JUNIT_JAR ${JAVA_TEST_LIBDIR}/junit-4.12.jar) set(JAVA_HAMCR_JAR ${JAVA_TEST_LIBDIR}/hamcrest-core-1.3.jar) set(JAVA_MOCKITO_JAR ${JAVA_TEST_LIBDIR}/mockito-all-1.10.19.jar) set(JAVA_CGLIB_JAR ${JAVA_TEST_LIBDIR}/cglib-2.2.2.jar) set(JAVA_ASSERTJ_JAR ${JAVA_TEST_LIBDIR}/assertj-core-1.7.1.jar) set(JAVA_TESTCLASSPATH "${JAVA_JUNIT_JAR}\;${JAVA_HAMCR_JAR}\;${JAVA_MOCKITO_JAR}\;${JAVA_CGLIB_JAR}\;${JAVA_ASSERTJ_JAR}") if(NOT EXISTS ${PROJECT_SOURCE_DIR}/java/classes) execute_process(COMMAND mkdir ${PROJECT_SOURCE_DIR}/java/classes) endif() if(NOT EXISTS ${JAVA_TEST_LIBDIR}) execute_process(COMMAND mkdir ${JAVA_TEST_LIBDIR}) endif() if(NOT EXISTS ${JAVA_JUNIT_JAR}) message("Downloading ${JAVA_JUNIT_JAR}") file(DOWNLOAD http://search.maven.org/remotecontent?filepath=junit/junit/4.12/junit-4.12.jar ${JAVA_TMP_JAR} STATUS downloadStatus) list(GET downloadStatus 0 error_code) if(NOT error_code EQUAL 0) message(FATAL_ERROR "Failed downloading ${JAVA_JUNIT_JAR}") endif() file(RENAME ${JAVA_TMP_JAR} ${JAVA_JUNIT_JAR}) endif() if(NOT EXISTS ${JAVA_HAMCR_JAR}) message("Downloading ${JAVA_HAMCR_JAR}") file(DOWNLOAD http://search.maven.org/remotecontent?filepath=org/hamcrest/hamcrest-core/1.3/hamcrest-core-1.3.jar ${JAVA_TMP_JAR} STATUS downloadStatus) list(GET downloadStatus 0 error_code) if(NOT error_code EQUAL 0) message(FATAL_ERROR "Failed downloading ${JAVA_HAMCR_JAR}") endif() file(RENAME ${JAVA_TMP_JAR} ${JAVA_HAMCR_JAR}) endif() if(NOT EXISTS ${JAVA_MOCKITO_JAR}) message("Downloading ${JAVA_MOCKITO_JAR}") file(DOWNLOAD http://search.maven.org/remotecontent?filepath=org/mockito/mockito-all/1.10.19/mockito-all-1.10.19.jar ${JAVA_TMP_JAR} STATUS downloadStatus) list(GET downloadStatus 0 error_code) if(NOT error_code EQUAL 0) message(FATAL_ERROR "Failed downloading ${JAVA_MOCKITO_JAR}") endif() file(RENAME ${JAVA_TMP_JAR} ${JAVA_MOCKITO_JAR}) endif() if(NOT EXISTS ${JAVA_CGLIB_JAR}) message("Downloading ${JAVA_CGLIB_JAR}") file(DOWNLOAD http://search.maven.org/remotecontent?filepath=cglib/cglib/2.2.2/cglib-2.2.2.jar ${JAVA_TMP_JAR} STATUS downloadStatus) list(GET downloadStatus 0 error_code) if(NOT error_code EQUAL 0) message(FATAL_ERROR "Failed downloading ${JAVA_CGLIB_JAR}") endif() file(RENAME ${JAVA_TMP_JAR} ${JAVA_CGLIB_JAR}) endif() if(NOT EXISTS ${JAVA_ASSERTJ_JAR}) message("Downloading ${JAVA_ASSERTJ_JAR}") file(DOWNLOAD http://central.maven.org/maven2/org/assertj/assertj-core/1.7.1/assertj-core-1.7.1.jar ${JAVA_TMP_JAR} STATUS downloadStatus) list(GET downloadStatus 0 error_code) if(NOT error_code EQUAL 0) message(FATAL_ERROR "Failed downloading ${JAVA_ASSERTJ_JAR}") endif() file(RENAME ${JAVA_TMP_JAR} ${JAVA_ASSERTJ_JAR}) endif() execute_process(COMMAND javac -cp ${JAVA_TESTCLASSPATH} -d ${PROJECT_SOURCE_DIR}/java/classes ${PROJECT_SOURCE_DIR}/java/src/main/java/org/rocksdb/util/*.java ${PROJECT_SOURCE_DIR}/java/src/main/java/org/rocksdb/*.java ${PROJECT_SOURCE_DIR}/java/src/test/java/org/rocksdb/*.java) execute_process(COMMAND javah -cp ${PROJECT_SOURCE_DIR}/java/classes -d ${PROJECT_SOURCE_DIR}/java/include -jni ${NATIVE_JAVA_CLASSES}) add_library(rocksdbjni${ARTIFACT_SUFFIX} SHARED ${JNI_NATIVE_SOURCES}) set_target_properties(rocksdbjni${ARTIFACT_SUFFIX} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/rocksdbjni${ARTIFACT_SUFFIX}.pdb") target_link_libraries(rocksdbjni${ARTIFACT_SUFFIX} rocksdblib${ARTIFACT_SUFFIX} ${LIBS}) rocksdb-4.5.1/java/HISTORY-JAVA.md000066400000000000000000000062341267530634000163060ustar00rootroot00000000000000# RocksJava Change Log ## 3.13 (8/4/2015) ### New Features * Exposed BackupEngine API. * Added CappedPrefixExtractor support. To use such extractor, simply call useCappedPrefixExtractor in either Options or ColumnFamilyOptions. * Added RemoveEmptyValueCompactionFilter. ## 3.10.0 (3/24/2015) ### New Features * Added compression per level API. * MemEnv is now available in RocksJava via RocksMemEnv class. * lz4 compression is now included in rocksjava static library when running `make rocksdbjavastatic`. ### Public API Changes * Overflowing a size_t when setting rocksdb options now throws an IllegalArgumentException, which removes the necessity for a developer to catch these Exceptions explicitly. * The set and get functions for tableCacheRemoveScanCountLimit are deprecated. ## By 01/31/2015 ### New Features * WriteBatchWithIndex support. * Iterator support for WriteBatch and WriteBatchWithIndex * GetUpdatesSince support. * Snapshots carry now information about the related sequence number. * TTL DB support. ## By 11/14/2014 ### New Features * Full support for Column Family. * Slice and Comparator support. * Default merge operator support. * RateLimiter support. ## By 06/15/2014 ### New Features * Added basic Java binding for rocksdb::Env such that multiple RocksDB can share the same thread pool and environment. * Added RestoreBackupableDB ## By 05/30/2014 ### Internal Framework Improvement * Added disOwnNativeHandle to RocksObject, which allows a RocksObject to give-up the ownership of its native handle. This method is useful when sharing and transferring the ownership of RocksDB C++ resources. ## By 05/15/2014 ### New Features * Added RocksObject --- the base class of all RocksDB classes which holds some RocksDB resources in the C++ side. * Use environmental variable JAVA_HOME in Makefile for RocksJava ### Public API changes * Renamed org.rocksdb.Iterator to org.rocksdb.RocksIterator to avoid potential confliction with Java built-in Iterator. ## By 04/30/2014 ### New Features * Added Java binding for MultiGet. * Added static method RocksDB.loadLibrary(), which loads necessary library files. * Added Java bindings for 60+ rocksdb::Options. * Added Java binding for BloomFilter. * Added Java binding for ReadOptions. * Added Java binding for memtables. * Added Java binding for sst formats. * Added Java binding for RocksDB Iterator which enables sequential scan operation. * Added Java binding for Statistics * Added Java binding for BackupableDB. ### DB Benchmark * Added filluniquerandom, readseq benchmark. * 70+ command-line options. * Enabled BloomFilter configuration. ## By 04/15/2014 ### New Features * Added Java binding for WriteOptions. * Added Java binding for WriteBatch, which enables batch-write. * Added Java binding for rocksdb::Options. * Added Java binding for block cache. * Added Java version DB Benchmark. ### DB Benchmark * Added readwhilewriting benchmark. ### Internal Framework Improvement * Avoid a potential byte-array-copy between c++ and Java in RocksDB.get. * Added SizeUnit in org.rocksdb.util to store consts like KB and GB. ### 03/28/2014 * RocksJava project started. * Added Java binding for RocksDB, which supports Open, Close, Get and Put. rocksdb-4.5.1/java/Makefile000066400000000000000000000172551267530634000155510ustar00rootroot00000000000000NATIVE_JAVA_CLASSES = org.rocksdb.AbstractCompactionFilter\ org.rocksdb.AbstractComparator\ org.rocksdb.AbstractSlice\ org.rocksdb.BackupEngine\ org.rocksdb.BackupableDB\ org.rocksdb.BackupableDBOptions\ org.rocksdb.BlockBasedTableConfig\ org.rocksdb.BloomFilter\ org.rocksdb.Checkpoint\ org.rocksdb.ColumnFamilyHandle\ org.rocksdb.ColumnFamilyOptions\ org.rocksdb.Comparator\ org.rocksdb.ComparatorOptions\ org.rocksdb.DBOptions\ org.rocksdb.DirectComparator\ org.rocksdb.DirectSlice\ org.rocksdb.Env\ org.rocksdb.FlushOptions\ org.rocksdb.Filter\ org.rocksdb.GenericRateLimiterConfig\ org.rocksdb.HashLinkedListMemTableConfig\ org.rocksdb.HashSkipListMemTableConfig\ org.rocksdb.Logger\ org.rocksdb.MergeOperator\ org.rocksdb.Options\ org.rocksdb.PlainTableConfig\ org.rocksdb.ReadOptions\ org.rocksdb.RemoveEmptyValueCompactionFilter\ org.rocksdb.RestoreBackupableDB\ org.rocksdb.RestoreOptions\ org.rocksdb.RocksDB\ org.rocksdb.RocksEnv\ org.rocksdb.RocksIterator\ org.rocksdb.RocksMemEnv\ org.rocksdb.SkipListMemTableConfig\ org.rocksdb.Slice\ org.rocksdb.Statistics\ org.rocksdb.TransactionLogIterator\ org.rocksdb.TtlDB\ org.rocksdb.VectorMemTableConfig\ org.rocksdb.Snapshot\ org.rocksdb.StringAppendOperator\ org.rocksdb.WriteBatch\ org.rocksdb.WriteBatch.Handler\ org.rocksdb.WriteOptions\ org.rocksdb.WriteBatchWithIndex\ org.rocksdb.WBWIRocksIterator NATIVE_JAVA_TEST_CLASSES = org.rocksdb.WriteBatchTest\ org.rocksdb.WriteBatchTestInternalHelper ROCKSDB_MAJOR = $(shell egrep "ROCKSDB_MAJOR.[0-9]" ../include/rocksdb/version.h | cut -d ' ' -f 3) ROCKSDB_MINOR = $(shell egrep "ROCKSDB_MINOR.[0-9]" ../include/rocksdb/version.h | cut -d ' ' -f 3) ROCKSDB_PATCH = $(shell egrep "ROCKSDB_PATCH.[0-9]" ../include/rocksdb/version.h | cut -d ' ' -f 3) NATIVE_INCLUDE = ./include ARCH := $(shell getconf LONG_BIT) ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-linux$(ARCH).jar ifeq ($(PLATFORM), OS_MACOSX) ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-osx.jar endif JAVA_TESTS = org.rocksdb.BackupableDBOptionsTest\ org.rocksdb.BackupEngineTest\ org.rocksdb.BackupableDBTest\ org.rocksdb.BlockBasedTableConfigTest\ org.rocksdb.CheckPointTest\ org.rocksdb.ColumnFamilyOptionsTest\ org.rocksdb.ColumnFamilyTest\ org.rocksdb.ComparatorOptionsTest\ org.rocksdb.ComparatorTest\ org.rocksdb.CompressionOptionsTest\ org.rocksdb.DBOptionsTest\ org.rocksdb.DirectComparatorTest\ org.rocksdb.DirectSliceTest\ org.rocksdb.util.EnvironmentTest\ org.rocksdb.FilterTest\ org.rocksdb.FlushTest\ org.rocksdb.InfoLogLevelTest\ org.rocksdb.KeyMayExistTest\ org.rocksdb.LoggerTest\ org.rocksdb.MemTableTest\ org.rocksdb.MergeTest\ org.rocksdb.MixedOptionsTest\ org.rocksdb.NativeLibraryLoaderTest\ org.rocksdb.OptionsTest\ org.rocksdb.PlainTableConfigTest\ org.rocksdb.ReadOnlyTest\ org.rocksdb.ReadOptionsTest\ org.rocksdb.RocksDBTest\ org.rocksdb.RocksEnvTest\ org.rocksdb.RocksIteratorTest\ org.rocksdb.RocksMemEnvTest\ org.rocksdb.util.SizeUnitTest\ org.rocksdb.SliceTest\ org.rocksdb.SnapshotTest\ org.rocksdb.TransactionLogIteratorTest\ org.rocksdb.TtlDBTest\ org.rocksdb.StatisticsCollectorTest\ org.rocksdb.WriteBatchHandlerTest\ org.rocksdb.WriteBatchTest\ org.rocksdb.WriteBatchThreadedTest\ org.rocksdb.WriteOptionsTest\ org.rocksdb.WriteBatchWithIndexTest MAIN_SRC = src/main/java TEST_SRC = src/test/java OUTPUT = target MAIN_CLASSES = $(OUTPUT)/classes TEST_CLASSES = $(OUTPUT)/test-classes JAVADOC = $(OUTPUT)/apidocs BENCHMARK_MAIN_SRC = benchmark/src/main/java BENCHMARK_OUTPUT = benchmark/target BENCHMARK_MAIN_CLASSES = $(BENCHMARK_OUTPUT)/classes SAMPLES_MAIN_SRC = samples/src/main/java SAMPLES_OUTPUT = samples/target SAMPLES_MAIN_CLASSES = $(SAMPLES_OUTPUT)/classes JAVA_TEST_LIBDIR = test-libs JAVA_JUNIT_JAR = $(JAVA_TEST_LIBDIR)/junit-4.12.jar JAVA_HAMCR_JAR = $(JAVA_TEST_LIBDIR)/hamcrest-core-1.3.jar JAVA_MOCKITO_JAR = $(JAVA_TEST_LIBDIR)/mockito-all-1.10.19.jar JAVA_CGLIB_JAR = $(JAVA_TEST_LIBDIR)/cglib-2.2.2.jar JAVA_ASSERTJ_JAR = $(JAVA_TEST_LIBDIR)/assertj-core-1.7.1.jar JAVA_TESTCLASSPATH = $(JAVA_JUNIT_JAR):$(JAVA_HAMCR_JAR):$(JAVA_MOCKITO_JAR):$(JAVA_CGLIB_JAR):$(JAVA_ASSERTJ_JAR) MVN_LOCAL = ~/.m2/repository clean: $(AM_V_at)rm -rf include/* $(AM_V_at)rm -rf test-libs/ $(AM_V_at)rm -rf $(OUTPUT) $(AM_V_at)rm -rf $(BENCHMARK_OUTPUT) $(AM_V_at)rm -rf $(SAMPLES_OUTPUT) javadocs: $(AM_V_GEN)mkdir -p $(JAVADOC) $(AM_V_at)javadoc -d $(JAVADOC) -sourcepath $(MAIN_SRC) -subpackages org javalib: java java_test javadocs java: $(AM_V_GEN)mkdir -p $(MAIN_CLASSES) $(AM_V_at)javac -d $(MAIN_CLASSES)\ $(MAIN_SRC)/org/rocksdb/util/*.java\ $(MAIN_SRC)/org/rocksdb/*.java $(AM_V_at)@cp ../HISTORY.md ./HISTORY-CPP.md $(AM_V_at)@rm -f ./HISTORY-CPP.md $(AM_V_at)javah -cp $(MAIN_CLASSES) -d $(NATIVE_INCLUDE) -jni $(NATIVE_JAVA_CLASSES) sample: java $(AM_V_GEN)mkdir -p $(SAMPLES_MAIN_CLASSES) $(AM_V_at)javac -cp $(MAIN_CLASSES) -d $(SAMPLES_MAIN_CLASSES) $(SAMPLES_MAIN_SRC)/RocksDBSample.java $(AM_V_at)@rm -rf /tmp/rocksdbjni $(AM_V_at)@rm -rf /tmp/rocksdbjni_not_found java -ea -Xcheck:jni -Djava.library.path=target -cp $(MAIN_CLASSES):$(SAMPLES_MAIN_CLASSES) RocksDBSample /tmp/rocksdbjni $(AM_V_at)@rm -rf /tmp/rocksdbjni $(AM_V_at)@rm -rf /tmp/rocksdbjni_not_found column_family_sample: java $(AM_V_GEN)mkdir -p $(SAMPLES_MAIN_CLASSES) $(AM_V_at)javac -cp $(MAIN_CLASSES) -d $(SAMPLES_MAIN_CLASSES) $(SAMPLES_MAIN_SRC)/RocksDBColumnFamilySample.java $(AM_V_at)@rm -rf /tmp/rocksdbjni java -ea -Xcheck:jni -Djava.library.path=target -cp $(MAIN_CLASSES):$(SAMPLES_MAIN_CLASSES) RocksDBColumnFamilySample /tmp/rocksdbjni $(AM_V_at)@rm -rf /tmp/rocksdbjni resolve_test_deps: test -s "$(JAVA_TEST_LIBDIR)" || mkdir -p "$(JAVA_TEST_LIBDIR)" test -s "$(JAVA_JUNIT_JAR)" || cp $(MVN_LOCAL)/junit/junit/4.12/junit-4.12.jar $(JAVA_TEST_LIBDIR) || curl -k -L -o $(JAVA_JUNIT_JAR) http://search.maven.org/remotecontent?filepath=junit/junit/4.12/junit-4.12.jar test -s "$(JAVA_HAMCR_JAR)" || cp $(MVN_LOCAL)/org/hamcrest/hamcrest-core/1.3/hamcrest-core-1.3.jar $(JAVA_TEST_LIBDIR) || curl -k -L -o $(JAVA_HAMCR_JAR) http://search.maven.org/remotecontent?filepath=org/hamcrest/hamcrest-core/1.3/hamcrest-core-1.3.jar test -s "$(JAVA_MOCKITO_JAR)" || cp $(MVN_LOCAL)/org/mockito/mockito-all/1.10.19/mockito-all-1.10.19.jar $(JAVA_TEST_LIBDIR) || curl -k -L -o "$(JAVA_MOCKITO_JAR)" http://search.maven.org/remotecontent?filepath=org/mockito/mockito-all/1.10.19/mockito-all-1.10.19.jar test -s "$(JAVA_CGLIB_JAR)" || cp $(MVN_LOCAL)/cglib/cglib/2.2.2/cglib-2.2.2.jar $(JAVA_TEST_LIBDIR) || curl -k -L -o "$(JAVA_CGLIB_JAR)" http://search.maven.org/remotecontent?filepath=cglib/cglib/2.2.2/cglib-2.2.2.jar test -s "$(JAVA_ASSERTJ_JAR)" || cp $(MVN_LOCAL)/org/assertj/assertj-core/1.7.1/assertj-core-1.7.1.jar $(JAVA_TEST_LIBDIR) || curl -k -L -o "$(JAVA_ASSERTJ_JAR)" http://central.maven.org/maven2/org/assertj/assertj-core/1.7.1/assertj-core-1.7.1.jar java_test: resolve_test_deps $(AM_V_GEN)mkdir -p $(TEST_CLASSES) $(AM_V_at)javac -cp $(MAIN_CLASSES):$(JAVA_TESTCLASSPATH) -d $(TEST_CLASSES)\ $(TEST_SRC)/org/rocksdb/test/*.java\ $(TEST_SRC)/org/rocksdb/util/*.java\ $(TEST_SRC)/org/rocksdb/*.java $(AM_V_at)javah -cp $(MAIN_CLASSES):$(TEST_CLASSES) -d $(NATIVE_INCLUDE) -jni $(NATIVE_JAVA_TEST_CLASSES) test: java resolve_test_deps java_test java -ea -Xcheck:jni -Djava.library.path=target -cp "$(MAIN_CLASSES):$(TEST_CLASSES):$(JAVA_TESTCLASSPATH):target/*" org.rocksdb.test.RocksJunitRunner $(JAVA_TESTS) db_bench: java $(AM_V_GEN)mkdir -p $(BENCHMARK_MAIN_CLASSES) $(AM_V_at)javac -cp $(MAIN_CLASSES) -d $(BENCHMARK_MAIN_CLASSES) $(BENCHMARK_MAIN_SRC)/org/rocksdb/benchmark/*.java rocksdb-4.5.1/java/RELEASE.md000066400000000000000000000044631267530634000155100ustar00rootroot00000000000000## Cross-building RocksDB can be built as a single self contained cross-platform JAR. The cross-platform jar can be usd on any 64-bit OSX system, 32-bit Linux system, or 64-bit Linux system. Building a cross-platform JAR requires: * [Vagrant](https://www.vagrantup.com/) * [Virtualbox](https://www.virtualbox.org/) * A Mac OSX machine that can compile RocksDB. * Java 7 set as JAVA_HOME. Once you have these items, run this make command from RocksDB's root source directory: make jclean clean rocksdbjavastaticrelease This command will build RocksDB natively on OSX, and will then spin up two Vagrant Virtualbox Ubuntu images to build RocksDB for both 32-bit and 64-bit Linux. You can find all native binaries and JARs in the java/target directory upon completion: librocksdbjni-linux32.so librocksdbjni-linux64.so librocksdbjni-osx.jnilib rocksdbjni-3.5.0-javadoc.jar rocksdbjni-3.5.0-linux32.jar rocksdbjni-3.5.0-linux64.jar rocksdbjni-3.5.0-osx.jar rocksdbjni-3.5.0-sources.jar rocksdbjni-3.5.0.jar ## Maven publication Set ~/.m2/settings.xml to contain: sonatype-nexus-staging your-sonatype-jira-username your-sonatype-jira-password From RocksDB's root directory, first build the Java static JARs: make jclean clean rocksdbjavastaticpublish This command will [stage the JAR artifacts on the Sonatype staging repository](http://central.sonatype.org/pages/manual-staging-bundle-creation-and-deployment.html). To release the staged artifacts. 1. Go to [https://oss.sonatype.org/#stagingRepositories](https://oss.sonatype.org/#stagingRepositories) and search for "rocksdb" in the upper right hand search box. 2. Select the rocksdb staging repository, and inspect its contents. 3. If all is well, follow [these steps](https://oss.sonatype.org/#stagingRepositories) to close the repository and release it. After the release has occurred, the artifacts will be synced to Maven central within 24-48 hours. rocksdb-4.5.1/java/benchmark/000077500000000000000000000000001267530634000160315ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/000077500000000000000000000000001267530634000166205ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/main/000077500000000000000000000000001267530634000175445ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/main/java/000077500000000000000000000000001267530634000204655ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/main/java/org/000077500000000000000000000000001267530634000212545ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/main/java/org/rocksdb/000077500000000000000000000000001267530634000227035ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/main/java/org/rocksdb/benchmark/000077500000000000000000000000001267530634000246355ustar00rootroot00000000000000rocksdb-4.5.1/java/benchmark/src/main/java/org/rocksdb/benchmark/DbBenchmark.java000066400000000000000000001602421267530634000276450ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. /** * Copyright (C) 2011 the original author or authors. * See the notice.md file distributed with this work for additional * information regarding copyright ownership. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package org.rocksdb.benchmark; import java.lang.Runnable; import java.lang.Math; import java.io.File; import java.nio.ByteBuffer; import java.util.Collection; import java.util.Date; import java.util.EnumMap; import java.util.List; import java.util.Map; import java.util.Random; import java.util.concurrent.TimeUnit; import java.util.Arrays; import java.util.ArrayList; import java.util.concurrent.Callable; import java.util.concurrent.Executors; import java.util.concurrent.ExecutorService; import java.util.concurrent.Future; import java.util.concurrent.TimeUnit; import org.rocksdb.*; import org.rocksdb.RocksMemEnv; import org.rocksdb.util.SizeUnit; class Stats { int id_; long start_; long finish_; double seconds_; long done_; long found_; long lastOpTime_; long nextReport_; long bytes_; StringBuilder message_; boolean excludeFromMerge_; // TODO(yhchiang): use the following arguments: // (Long)Flag.stats_interval // (Integer)Flag.stats_per_interval Stats(int id) { id_ = id; nextReport_ = 100; done_ = 0; bytes_ = 0; seconds_ = 0; start_ = System.nanoTime(); lastOpTime_ = start_; finish_ = start_; found_ = 0; message_ = new StringBuilder(""); excludeFromMerge_ = false; } void merge(final Stats other) { if (other.excludeFromMerge_) { return; } done_ += other.done_; found_ += other.found_; bytes_ += other.bytes_; seconds_ += other.seconds_; if (other.start_ < start_) start_ = other.start_; if (other.finish_ > finish_) finish_ = other.finish_; // Just keep the messages from one thread if (message_.length() == 0) { message_ = other.message_; } } void stop() { finish_ = System.nanoTime(); seconds_ = (double) (finish_ - start_) * 1e-9; } void addMessage(String msg) { if (message_.length() > 0) { message_.append(" "); } message_.append(msg); } void setId(int id) { id_ = id; } void setExcludeFromMerge() { excludeFromMerge_ = true; } void finishedSingleOp(int bytes) { done_++; lastOpTime_ = System.nanoTime(); bytes_ += bytes; if (done_ >= nextReport_) { if (nextReport_ < 1000) { nextReport_ += 100; } else if (nextReport_ < 5000) { nextReport_ += 500; } else if (nextReport_ < 10000) { nextReport_ += 1000; } else if (nextReport_ < 50000) { nextReport_ += 5000; } else if (nextReport_ < 100000) { nextReport_ += 10000; } else if (nextReport_ < 500000) { nextReport_ += 50000; } else { nextReport_ += 100000; } System.err.printf("... Task %s finished %d ops%30s\r", id_, done_, ""); } } void report(String name) { // Pretend at least one op was done in case we are running a benchmark // that does not call FinishedSingleOp(). if (done_ < 1) done_ = 1; StringBuilder extra = new StringBuilder(""); if (bytes_ > 0) { // Rate is computed on actual elapsed time, not the sum of per-thread // elapsed times. double elapsed = (finish_ - start_) * 1e-9; extra.append(String.format("%6.1f MB/s", (bytes_ / 1048576.0) / elapsed)); } extra.append(message_.toString()); double elapsed = (finish_ - start_); double throughput = (double) done_ / (elapsed * 1e-9); System.out.format("%-12s : %11.3f micros/op %d ops/sec;%s%s\n", name, (elapsed * 1e-6) / done_, (long) throughput, (extra.length() == 0 ? "" : " "), extra.toString()); } } public class DbBenchmark { enum Order { SEQUENTIAL, RANDOM } enum DBState { FRESH, EXISTING } static { RocksDB.loadLibrary(); } abstract class BenchmarkTask implements Callable { // TODO(yhchiang): use (Integer)Flag.perf_level. public BenchmarkTask( int tid, long randSeed, long numEntries, long keyRange) { tid_ = tid; rand_ = new Random(randSeed + tid * 1000); numEntries_ = numEntries; keyRange_ = keyRange; stats_ = new Stats(tid); } @Override public Stats call() throws RocksDBException { stats_.start_ = System.nanoTime(); runTask(); stats_.finish_ = System.nanoTime(); return stats_; } abstract protected void runTask() throws RocksDBException; protected int tid_; protected Random rand_; protected long numEntries_; protected long keyRange_; protected Stats stats_; protected void getFixedKey(byte[] key, long sn) { generateKeyFromLong(key, sn); } protected void getRandomKey(byte[] key, long range) { generateKeyFromLong(key, Math.abs(rand_.nextLong() % range)); } } abstract class WriteTask extends BenchmarkTask { public WriteTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch) { super(tid, randSeed, numEntries, keyRange); writeOpt_ = writeOpt; entriesPerBatch_ = entriesPerBatch; maxWritesPerSecond_ = -1; } public WriteTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch, long maxWritesPerSecond) { super(tid, randSeed, numEntries, keyRange); writeOpt_ = writeOpt; entriesPerBatch_ = entriesPerBatch; maxWritesPerSecond_ = maxWritesPerSecond; } @Override public void runTask() throws RocksDBException { if (numEntries_ != DbBenchmark.this.num_) { stats_.message_.append(String.format(" (%d ops)", numEntries_)); } byte[] key = new byte[keySize_]; byte[] value = new byte[valueSize_]; try { if (entriesPerBatch_ == 1) { for (long i = 0; i < numEntries_; ++i) { getKey(key, i, keyRange_); DbBenchmark.this.gen_.generate(value); db_.put(writeOpt_, key, value); stats_.finishedSingleOp(keySize_ + valueSize_); writeRateControl(i); if (isFinished()) { return; } } } else { for (long i = 0; i < numEntries_; i += entriesPerBatch_) { WriteBatch batch = new WriteBatch(); for (long j = 0; j < entriesPerBatch_; j++) { getKey(key, i + j, keyRange_); DbBenchmark.this.gen_.generate(value); batch.put(key, value); stats_.finishedSingleOp(keySize_ + valueSize_); } db_.write(writeOpt_, batch); batch.dispose(); writeRateControl(i); if (isFinished()) { return; } } } } catch (InterruptedException e) { // thread has been terminated. } } protected void writeRateControl(long writeCount) throws InterruptedException { if (maxWritesPerSecond_ <= 0) return; long minInterval = writeCount * TimeUnit.SECONDS.toNanos(1) / maxWritesPerSecond_; long interval = System.nanoTime() - stats_.start_; if (minInterval - interval > TimeUnit.MILLISECONDS.toNanos(1)) { TimeUnit.NANOSECONDS.sleep(minInterval - interval); } } abstract protected void getKey(byte[] key, long id, long range); protected WriteOptions writeOpt_; protected long entriesPerBatch_; protected long maxWritesPerSecond_; } class WriteSequentialTask extends WriteTask { public WriteSequentialTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch) { super(tid, randSeed, numEntries, keyRange, writeOpt, entriesPerBatch); } public WriteSequentialTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch, long maxWritesPerSecond) { super(tid, randSeed, numEntries, keyRange, writeOpt, entriesPerBatch, maxWritesPerSecond); } @Override protected void getKey(byte[] key, long id, long range) { getFixedKey(key, id); } } class WriteRandomTask extends WriteTask { public WriteRandomTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch) { super(tid, randSeed, numEntries, keyRange, writeOpt, entriesPerBatch); } public WriteRandomTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch, long maxWritesPerSecond) { super(tid, randSeed, numEntries, keyRange, writeOpt, entriesPerBatch, maxWritesPerSecond); } @Override protected void getKey(byte[] key, long id, long range) { getRandomKey(key, range); } } class WriteUniqueRandomTask extends WriteTask { static final int MAX_BUFFER_SIZE = 10000000; public WriteUniqueRandomTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch) { super(tid, randSeed, numEntries, keyRange, writeOpt, entriesPerBatch); initRandomKeySequence(); } public WriteUniqueRandomTask( int tid, long randSeed, long numEntries, long keyRange, WriteOptions writeOpt, long entriesPerBatch, long maxWritesPerSecond) { super(tid, randSeed, numEntries, keyRange, writeOpt, entriesPerBatch, maxWritesPerSecond); initRandomKeySequence(); } @Override protected void getKey(byte[] key, long id, long range) { generateKeyFromLong(key, nextUniqueRandom()); } protected void initRandomKeySequence() { bufferSize_ = MAX_BUFFER_SIZE; if (bufferSize_ > keyRange_) { bufferSize_ = (int) keyRange_; } currentKeyCount_ = bufferSize_; keyBuffer_ = new long[MAX_BUFFER_SIZE]; for (int k = 0; k < bufferSize_; ++k) { keyBuffer_[k] = k; } } /** * Semi-randomly return the next unique key. It is guaranteed to be * fully random if keyRange_ <= MAX_BUFFER_SIZE. */ long nextUniqueRandom() { if (bufferSize_ == 0) { System.err.println("bufferSize_ == 0."); return 0; } int r = rand_.nextInt(bufferSize_); // randomly pick one from the keyBuffer long randKey = keyBuffer_[r]; if (currentKeyCount_ < keyRange_) { // if we have not yet inserted all keys, insert next new key to [r]. keyBuffer_[r] = currentKeyCount_++; } else { // move the last element to [r] and decrease the size by 1. keyBuffer_[r] = keyBuffer_[--bufferSize_]; } return randKey; } int bufferSize_; long currentKeyCount_; long[] keyBuffer_; } class ReadRandomTask extends BenchmarkTask { public ReadRandomTask( int tid, long randSeed, long numEntries, long keyRange) { super(tid, randSeed, numEntries, keyRange); } @Override public void runTask() throws RocksDBException { byte[] key = new byte[keySize_]; byte[] value = new byte[valueSize_]; for (long i = 0; i < numEntries_; i++) { getRandomKey(key, keyRange_); int len = db_.get(key, value); if (len != RocksDB.NOT_FOUND) { stats_.found_++; stats_.finishedSingleOp(keySize_ + valueSize_); } else { stats_.finishedSingleOp(keySize_); } if (isFinished()) { return; } } } } class ReadSequentialTask extends BenchmarkTask { public ReadSequentialTask( int tid, long randSeed, long numEntries, long keyRange) { super(tid, randSeed, numEntries, keyRange); } @Override public void runTask() throws RocksDBException { RocksIterator iter = db_.newIterator(); long i; for (iter.seekToFirst(), i = 0; iter.isValid() && i < numEntries_; iter.next(), ++i) { stats_.found_++; stats_.finishedSingleOp(iter.key().length + iter.value().length); if (isFinished()) { iter.dispose(); return; } } iter.dispose(); } } public DbBenchmark(Map flags) throws Exception { benchmarks_ = (List) flags.get(Flag.benchmarks); num_ = (Integer) flags.get(Flag.num); threadNum_ = (Integer) flags.get(Flag.threads); reads_ = (Integer) (flags.get(Flag.reads) == null ? flags.get(Flag.num) : flags.get(Flag.reads)); keySize_ = (Integer) flags.get(Flag.key_size); valueSize_ = (Integer) flags.get(Flag.value_size); compressionRatio_ = (Double) flags.get(Flag.compression_ratio); useExisting_ = (Boolean) flags.get(Flag.use_existing_db); randSeed_ = (Long) flags.get(Flag.seed); databaseDir_ = (String) flags.get(Flag.db); writesPerSeconds_ = (Integer) flags.get(Flag.writes_per_second); memtable_ = (String) flags.get(Flag.memtablerep); maxWriteBufferNumber_ = (Integer) flags.get(Flag.max_write_buffer_number); prefixSize_ = (Integer) flags.get(Flag.prefix_size); keysPerPrefix_ = (Integer) flags.get(Flag.keys_per_prefix); hashBucketCount_ = (Long) flags.get(Flag.hash_bucket_count); usePlainTable_ = (Boolean) flags.get(Flag.use_plain_table); useMemenv_ = (Boolean) flags.get(Flag.use_mem_env); flags_ = flags; finishLock_ = new Object(); // options.setPrefixSize((Integer)flags_.get(Flag.prefix_size)); // options.setKeysPerPrefix((Long)flags_.get(Flag.keys_per_prefix)); compressionType_ = (String) flags.get(Flag.compression_type); compression_ = CompressionType.NO_COMPRESSION; try { if (compressionType_!=null) { final CompressionType compressionType = CompressionType.getCompressionType(compressionType_); if (compressionType != null && compressionType != CompressionType.NO_COMPRESSION) { System.loadLibrary(compressionType.getLibraryName()); } } } catch (UnsatisfiedLinkError e) { System.err.format("Unable to load %s library:%s%n" + "No compression is used.%n", compressionType_, e.toString()); compressionType_ = "none"; } gen_ = new RandomGenerator(randSeed_, compressionRatio_); } private void prepareReadOptions(ReadOptions options) { options.setVerifyChecksums((Boolean)flags_.get(Flag.verify_checksum)); options.setTailing((Boolean)flags_.get(Flag.use_tailing_iterator)); } private void prepareWriteOptions(WriteOptions options) { options.setSync((Boolean)flags_.get(Flag.sync)); options.setDisableWAL((Boolean)flags_.get(Flag.disable_wal)); } private void prepareOptions(Options options) throws RocksDBException { if (!useExisting_) { options.setCreateIfMissing(true); } else { options.setCreateIfMissing(false); } if (useMemenv_) { options.setEnv(new RocksMemEnv()); } switch (memtable_) { case "skip_list": options.setMemTableConfig(new SkipListMemTableConfig()); break; case "vector": options.setMemTableConfig(new VectorMemTableConfig()); break; case "hash_linkedlist": options.setMemTableConfig( new HashLinkedListMemTableConfig() .setBucketCount(hashBucketCount_)); options.useFixedLengthPrefixExtractor(prefixSize_); break; case "hash_skiplist": case "prefix_hash": options.setMemTableConfig( new HashSkipListMemTableConfig() .setBucketCount(hashBucketCount_)); options.useFixedLengthPrefixExtractor(prefixSize_); break; default: System.err.format( "unable to detect the specified memtable, " + "use the default memtable factory %s%n", options.memTableFactoryName()); break; } if (usePlainTable_) { options.setTableFormatConfig( new PlainTableConfig().setKeySize(keySize_)); } else { BlockBasedTableConfig table_options = new BlockBasedTableConfig(); table_options.setBlockSize((Long)flags_.get(Flag.block_size)) .setBlockCacheSize((Long)flags_.get(Flag.cache_size)) .setCacheNumShardBits( (Integer)flags_.get(Flag.cache_numshardbits)); options.setTableFormatConfig(table_options); } options.setWriteBufferSize( (Long)flags_.get(Flag.write_buffer_size)); options.setMaxWriteBufferNumber( (Integer)flags_.get(Flag.max_write_buffer_number)); options.setMaxBackgroundCompactions( (Integer)flags_.get(Flag.max_background_compactions)); options.getEnv().setBackgroundThreads( (Integer)flags_.get(Flag.max_background_compactions)); options.setMaxBackgroundFlushes( (Integer)flags_.get(Flag.max_background_flushes)); options.setMaxOpenFiles( (Integer)flags_.get(Flag.open_files)); options.setDisableDataSync( (Boolean)flags_.get(Flag.disable_data_sync)); options.setUseFsync( (Boolean)flags_.get(Flag.use_fsync)); options.setWalDir( (String)flags_.get(Flag.wal_dir)); options.setDeleteObsoleteFilesPeriodMicros( (Integer)flags_.get(Flag.delete_obsolete_files_period_micros)); options.setTableCacheNumshardbits( (Integer)flags_.get(Flag.table_cache_numshardbits)); options.setAllowMmapReads( (Boolean)flags_.get(Flag.mmap_read)); options.setAllowMmapWrites( (Boolean)flags_.get(Flag.mmap_write)); options.setAdviseRandomOnOpen( (Boolean)flags_.get(Flag.advise_random_on_open)); options.setUseAdaptiveMutex( (Boolean)flags_.get(Flag.use_adaptive_mutex)); options.setBytesPerSync( (Long)flags_.get(Flag.bytes_per_sync)); options.setBloomLocality( (Integer)flags_.get(Flag.bloom_locality)); options.setMinWriteBufferNumberToMerge( (Integer)flags_.get(Flag.min_write_buffer_number_to_merge)); options.setMemtablePrefixBloomBits( (Integer)flags_.get(Flag.memtable_bloom_bits)); options.setNumLevels( (Integer)flags_.get(Flag.num_levels)); options.setTargetFileSizeBase( (Integer)flags_.get(Flag.target_file_size_base)); options.setTargetFileSizeMultiplier( (Integer)flags_.get(Flag.target_file_size_multiplier)); options.setMaxBytesForLevelBase( (Integer)flags_.get(Flag.max_bytes_for_level_base)); options.setMaxBytesForLevelMultiplier( (Integer)flags_.get(Flag.max_bytes_for_level_multiplier)); options.setLevelZeroStopWritesTrigger( (Integer)flags_.get(Flag.level0_stop_writes_trigger)); options.setLevelZeroSlowdownWritesTrigger( (Integer)flags_.get(Flag.level0_slowdown_writes_trigger)); options.setLevelZeroFileNumCompactionTrigger( (Integer)flags_.get(Flag.level0_file_num_compaction_trigger)); options.setSoftRateLimit( (Double)flags_.get(Flag.soft_rate_limit)); options.setHardRateLimit( (Double)flags_.get(Flag.hard_rate_limit)); options.setRateLimitDelayMaxMilliseconds( (Integer)flags_.get(Flag.rate_limit_delay_max_milliseconds)); options.setMaxGrandparentOverlapFactor( (Integer)flags_.get(Flag.max_grandparent_overlap_factor)); options.setDisableAutoCompactions( (Boolean)flags_.get(Flag.disable_auto_compactions)); options.setSourceCompactionFactor( (Integer)flags_.get(Flag.source_compaction_factor)); options.setFilterDeletes( (Boolean)flags_.get(Flag.filter_deletes)); options.setMaxSuccessiveMerges( (Integer)flags_.get(Flag.max_successive_merges)); options.setWalTtlSeconds((Long)flags_.get(Flag.wal_ttl_seconds)); options.setWalSizeLimitMB((Long)flags_.get(Flag.wal_size_limit_MB)); /* TODO(yhchiang): enable the following parameters options.setCompressionType((String)flags_.get(Flag.compression_type)); options.setCompressionLevel((Integer)flags_.get(Flag.compression_level)); options.setMinLevelToCompress((Integer)flags_.get(Flag.min_level_to_compress)); options.setHdfs((String)flags_.get(Flag.hdfs)); // env options.setStatistics((Boolean)flags_.get(Flag.statistics)); options.setUniversalSizeRatio( (Integer)flags_.get(Flag.universal_size_ratio)); options.setUniversalMinMergeWidth( (Integer)flags_.get(Flag.universal_min_merge_width)); options.setUniversalMaxMergeWidth( (Integer)flags_.get(Flag.universal_max_merge_width)); options.setUniversalMaxSizeAmplificationPercent( (Integer)flags_.get(Flag.universal_max_size_amplification_percent)); options.setUniversalCompressionSizePercent( (Integer)flags_.get(Flag.universal_compression_size_percent)); // TODO(yhchiang): add RocksDB.openForReadOnly() to enable Flag.readonly // TODO(yhchiang): enable Flag.merge_operator by switch options.setAccessHintOnCompactionStart( (String)flags_.get(Flag.compaction_fadvice)); // available values of fadvice are "NONE", "NORMAL", "SEQUENTIAL", "WILLNEED" for fadvice */ } private void run() throws RocksDBException { if (!useExisting_) { destroyDb(); } Options options = new Options(); prepareOptions(options); open(options); printHeader(options); for (String benchmark : benchmarks_) { List> tasks = new ArrayList>(); List> bgTasks = new ArrayList>(); WriteOptions writeOpt = new WriteOptions(); prepareWriteOptions(writeOpt); ReadOptions readOpt = new ReadOptions(); prepareReadOptions(readOpt); int currentTaskId = 0; boolean known = true; switch (benchmark) { case "fillseq": tasks.add(new WriteSequentialTask( currentTaskId++, randSeed_, num_, num_, writeOpt, 1)); break; case "fillbatch": tasks.add(new WriteRandomTask( currentTaskId++, randSeed_, num_ / 1000, num_, writeOpt, 1000)); break; case "fillrandom": tasks.add(new WriteRandomTask( currentTaskId++, randSeed_, num_, num_, writeOpt, 1)); break; case "filluniquerandom": tasks.add(new WriteUniqueRandomTask( currentTaskId++, randSeed_, num_, num_, writeOpt, 1)); break; case "fillsync": writeOpt.setSync(true); tasks.add(new WriteRandomTask( currentTaskId++, randSeed_, num_ / 1000, num_ / 1000, writeOpt, 1)); break; case "readseq": for (int t = 0; t < threadNum_; ++t) { tasks.add(new ReadSequentialTask( currentTaskId++, randSeed_, reads_ / threadNum_, num_)); } break; case "readrandom": for (int t = 0; t < threadNum_; ++t) { tasks.add(new ReadRandomTask( currentTaskId++, randSeed_, reads_ / threadNum_, num_)); } break; case "readwhilewriting": WriteTask writeTask = new WriteRandomTask( -1, randSeed_, Long.MAX_VALUE, num_, writeOpt, 1, writesPerSeconds_); writeTask.stats_.setExcludeFromMerge(); bgTasks.add(writeTask); for (int t = 0; t < threadNum_; ++t) { tasks.add(new ReadRandomTask( currentTaskId++, randSeed_, reads_ / threadNum_, num_)); } break; case "readhot": for (int t = 0; t < threadNum_; ++t) { tasks.add(new ReadRandomTask( currentTaskId++, randSeed_, reads_ / threadNum_, num_ / 100)); } break; case "delete": destroyDb(); open(options); break; default: known = false; System.err.println("Unknown benchmark: " + benchmark); break; } if (known) { ExecutorService executor = Executors.newCachedThreadPool(); ExecutorService bgExecutor = Executors.newCachedThreadPool(); try { // measure only the main executor time List> bgResults = new ArrayList>(); for (Callable bgTask : bgTasks) { bgResults.add(bgExecutor.submit(bgTask)); } start(); List> results = executor.invokeAll(tasks); executor.shutdown(); boolean finished = executor.awaitTermination(10, TimeUnit.SECONDS); if (!finished) { System.out.format( "Benchmark %s was not finished before timeout.", benchmark); executor.shutdownNow(); } setFinished(true); bgExecutor.shutdown(); finished = bgExecutor.awaitTermination(10, TimeUnit.SECONDS); if (!finished) { System.out.format( "Benchmark %s was not finished before timeout.", benchmark); bgExecutor.shutdownNow(); } stop(benchmark, results, currentTaskId); } catch (InterruptedException e) { System.err.println(e); } } writeOpt.dispose(); readOpt.dispose(); } options.dispose(); db_.close(); } private void printHeader(Options options) { int kKeySize = 16; System.out.printf("Keys: %d bytes each\n", kKeySize); System.out.printf("Values: %d bytes each (%d bytes after compression)\n", valueSize_, (int) (valueSize_ * compressionRatio_ + 0.5)); System.out.printf("Entries: %d\n", num_); System.out.printf("RawSize: %.1f MB (estimated)\n", ((double)(kKeySize + valueSize_) * num_) / SizeUnit.MB); System.out.printf("FileSize: %.1f MB (estimated)\n", (((kKeySize + valueSize_ * compressionRatio_) * num_) / SizeUnit.MB)); System.out.format("Memtable Factory: %s%n", options.memTableFactoryName()); System.out.format("Prefix: %d bytes%n", prefixSize_); System.out.format("Compression: %s%n", compressionType_); printWarnings(); System.out.printf("------------------------------------------------\n"); } void printWarnings() { boolean assertsEnabled = false; assert assertsEnabled = true; // Intentional side effect!!! if (assertsEnabled) { System.out.printf( "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); } } private void open(Options options) throws RocksDBException { db_ = RocksDB.open(options, databaseDir_); } private void start() { setFinished(false); startTime_ = System.nanoTime(); } private void stop( String benchmark, List> results, int concurrentThreads) { long endTime = System.nanoTime(); double elapsedSeconds = 1.0d * (endTime - startTime_) / TimeUnit.SECONDS.toNanos(1); Stats stats = new Stats(-1); int taskFinishedCount = 0; for (Future result : results) { if (result.isDone()) { try { Stats taskStats = result.get(3, TimeUnit.SECONDS); if (!result.isCancelled()) { taskFinishedCount++; } stats.merge(taskStats); } catch (Exception e) { // then it's not successful, the output will indicate this } } } String extra = ""; if (benchmark.indexOf("read") >= 0) { extra = String.format(" %d / %d found; ", stats.found_, stats.done_); } else { extra = String.format(" %d ops done; ", stats.done_); } System.out.printf( "%-16s : %11.5f micros/op; %6.1f MB/s;%s %d / %d task(s) finished.\n", benchmark, elapsedSeconds / stats.done_ * 1e6, (stats.bytes_ / 1048576.0) / elapsedSeconds, extra, taskFinishedCount, concurrentThreads); } public void generateKeyFromLong(byte[] slice, long n) { assert(n >= 0); int startPos = 0; if (keysPerPrefix_ > 0) { long numPrefix = (num_ + keysPerPrefix_ - 1) / keysPerPrefix_; long prefix = n % numPrefix; int bytesToFill = Math.min(prefixSize_, 8); for (int i = 0; i < bytesToFill; ++i) { slice[i] = (byte) (prefix % 256); prefix /= 256; } for (int i = 8; i < bytesToFill; ++i) { slice[i] = '0'; } startPos = bytesToFill; } for (int i = slice.length - 1; i >= startPos; --i) { slice[i] = (byte) ('0' + (n % 10)); n /= 10; } } private void destroyDb() { if (db_ != null) { db_.close(); } // TODO(yhchiang): develop our own FileUtil // FileUtil.deleteDir(databaseDir_); } private void printStats() { } static void printHelp() { System.out.println("usage:"); for (Flag flag : Flag.values()) { System.out.format(" --%s%n\t%s%n", flag.name(), flag.desc()); if (flag.getDefaultValue() != null) { System.out.format("\tDEFAULT: %s%n", flag.getDefaultValue().toString()); } } } public static void main(String[] args) throws Exception { Map flags = new EnumMap(Flag.class); for (Flag flag : Flag.values()) { if (flag.getDefaultValue() != null) { flags.put(flag, flag.getDefaultValue()); } } for (String arg : args) { boolean valid = false; if (arg.equals("--help") || arg.equals("-h")) { printHelp(); System.exit(0); } if (arg.startsWith("--")) { try { String[] parts = arg.substring(2).split("="); if (parts.length >= 1) { Flag key = Flag.valueOf(parts[0]); if (key != null) { Object value = null; if (parts.length >= 2) { value = key.parseValue(parts[1]); } flags.put(key, value); valid = true; } } } catch (Exception e) { } } if (!valid) { System.err.println("Invalid argument " + arg); System.exit(1); } } new DbBenchmark(flags).run(); } private enum Flag { benchmarks( Arrays.asList( "fillseq", "readrandom", "fillrandom"), "Comma-separated list of operations to run in the specified order\n" + "\tActual benchmarks:\n" + "\t\tfillseq -- write N values in sequential key order in async mode.\n" + "\t\tfillrandom -- write N values in random key order in async mode.\n" + "\t\tfillbatch -- write N/1000 batch where each batch has 1000 values\n" + "\t\t in random key order in sync mode.\n" + "\t\tfillsync -- write N/100 values in random key order in sync mode.\n" + "\t\tfill100K -- write N/1000 100K values in random order in async mode.\n" + "\t\treadseq -- read N times sequentially.\n" + "\t\treadrandom -- read N times in random order.\n" + "\t\treadhot -- read N times in random order from 1% section of DB.\n" + "\t\treadwhilewriting -- measure the read performance of multiple readers\n" + "\t\t with a bg single writer. The write rate of the bg\n" + "\t\t is capped by --writes_per_second.\n" + "\tMeta Operations:\n" + "\t\tdelete -- delete DB") { @Override public Object parseValue(String value) { return new ArrayList(Arrays.asList(value.split(","))); } }, compression_ratio(0.5d, "Arrange to generate values that shrink to this fraction of\n" + "\ttheir original size after compression.") { @Override public Object parseValue(String value) { return Double.parseDouble(value); } }, use_existing_db(false, "If true, do not destroy the existing database. If you set this\n" + "\tflag and also specify a benchmark that wants a fresh database,\n" + "\tthat benchmark will fail.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, num(1000000, "Number of key/values to place in database.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, threads(1, "Number of concurrent threads to run.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, reads(null, "Number of read operations to do. If negative, do --nums reads.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, key_size(16, "The size of each key in bytes.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, value_size(100, "The size of each value in bytes.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, write_buffer_size(4 * SizeUnit.MB, "Number of bytes to buffer in memtable before compacting\n" + "\t(initialized to default value by 'main'.)") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, max_write_buffer_number(2, "The number of in-memory memtables. Each memtable is of size\n" + "\twrite_buffer_size.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, prefix_size(0, "Controls the prefix size for HashSkipList, HashLinkedList,\n" + "\tand plain table.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, keys_per_prefix(0, "Controls the average number of keys generated\n" + "\tper prefix, 0 means no special handling of the prefix,\n" + "\ti.e. use the prefix comes with the generated random number.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, memtablerep("skip_list", "The memtable format. Available options are\n" + "\tskip_list,\n" + "\tvector,\n" + "\thash_linkedlist,\n" + "\thash_skiplist (prefix_hash.)") { @Override public Object parseValue(String value) { return value; } }, hash_bucket_count(SizeUnit.MB, "The number of hash buckets used in the hash-bucket-based\n" + "\tmemtables. Memtables that currently support this argument are\n" + "\thash_linkedlist and hash_skiplist.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, writes_per_second(10000, "The write-rate of the background writer used in the\n" + "\t`readwhilewriting` benchmark. Non-positive number indicates\n" + "\tusing an unbounded write-rate in `readwhilewriting` benchmark.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, use_plain_table(false, "Use plain-table sst format.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, cache_size(-1L, "Number of bytes to use as a cache of uncompressed data.\n" + "\tNegative means use default settings.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, seed(0L, "Seed base for random number generators.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, num_levels(7, "The total number of levels.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, numdistinct(1000, "Number of distinct keys to use. Used in RandomWithVerify to\n" + "\tread/write on fewer keys so that gets are more likely to find the\n" + "\tkey and puts are more likely to update the same key.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, merge_keys(-1, "Number of distinct keys to use for MergeRandom and\n" + "\tReadRandomMergeRandom.\n" + "\tIf negative, there will be FLAGS_num keys.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, bloom_locality(0,"Control bloom filter probes locality.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, duration(0,"Time in seconds for the random-ops tests to run.\n" + "\tWhen 0 then num & reads determine the test duration.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, num_multi_db(0, "Number of DBs used in the benchmark. 0 means single DB.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, histogram(false,"Print histogram of operation timings.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, min_write_buffer_number_to_merge( defaultOptions_.minWriteBufferNumberToMerge(), "The minimum number of write buffers that will be merged together\n" + "\tbefore writing to storage. This is cheap because it is an\n" + "\tin-memory merge. If this feature is not enabled, then all these\n" + "\twrite buffers are flushed to L0 as separate files and this\n" + "\tincreases read amplification because a get request has to check\n" + "\tin all of these files. Also, an in-memory merge may result in\n" + "\twriting less data to storage if there are duplicate records\n" + "\tin each of these individual write buffers.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, max_background_compactions( defaultOptions_.maxBackgroundCompactions(), "The maximum number of concurrent background compactions\n" + "\tthat can occur in parallel.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, max_background_flushes( defaultOptions_.maxBackgroundFlushes(), "The maximum number of concurrent background flushes\n" + "\tthat can occur in parallel.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, /* TODO(yhchiang): enable the following compaction_style((int32_t) defaultOptions_.compactionStyle(), "style of compaction: level-based vs universal.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } },*/ universal_size_ratio(0, "Percentage flexibility while comparing file size\n" + "\t(for universal compaction only).") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, universal_min_merge_width(0,"The minimum number of files in a\n" + "\tsingle compaction run (for universal compaction only).") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, universal_max_merge_width(0,"The max number of files to compact\n" + "\tin universal style compaction.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, universal_max_size_amplification_percent(0, "The max size amplification for universal style compaction.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, universal_compression_size_percent(-1, "The percentage of the database to compress for universal\n" + "\tcompaction. -1 means compress everything.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, block_size(defaultBlockBasedTableOptions_.blockSize(), "Number of bytes in a block.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, compressed_cache_size(-1, "Number of bytes to use as a cache of compressed data.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, open_files(defaultOptions_.maxOpenFiles(), "Maximum number of files to keep open at the same time\n" + "\t(use default if == 0)") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, bloom_bits(-1,"Bloom filter bits per key. Negative means\n" + "\tuse default settings.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, memtable_bloom_bits(0,"Bloom filter bits per key for memtable.\n" + "\tNegative means no bloom filter.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, cache_numshardbits(-1,"Number of shards for the block cache\n" + "\tis 2 ** cache_numshardbits. Negative means use default settings.\n" + "\tThis is applied only if FLAGS_cache_size is non-negative.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, verify_checksum(false,"Verify checksum for every block read\n" + "\tfrom storage.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, statistics(false,"Database statistics.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, writes(-1,"Number of write operations to do. If negative, do\n" + "\t--num reads.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, sync(false,"Sync all writes to disk.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, disable_data_sync(false,"If true, do not wait until data is\n" + "\tsynced to disk.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, use_fsync(false,"If true, issue fsync instead of fdatasync.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, disable_wal(false,"If true, do not write WAL for write.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, wal_dir("", "If not empty, use the given dir for WAL.") { @Override public Object parseValue(String value) { return value; } }, target_file_size_base(2 * 1048576,"Target file size at level-1") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, target_file_size_multiplier(1, "A multiplier to compute target level-N file size (N >= 2)") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, max_bytes_for_level_base(10 * 1048576, "Max bytes for level-1") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, max_bytes_for_level_multiplier(10, "A multiplier to compute max bytes for level-N (N >= 2)") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, level0_stop_writes_trigger(12,"Number of files in level-0\n" + "\tthat will trigger put stop.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, level0_slowdown_writes_trigger(8,"Number of files in level-0\n" + "\tthat will slow down writes.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, level0_file_num_compaction_trigger(4,"Number of files in level-0\n" + "\twhen compactions start.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, readwritepercent(90,"Ratio of reads to reads/writes (expressed\n" + "\tas percentage) for the ReadRandomWriteRandom workload. The\n" + "\tdefault value 90 means 90% operations out of all reads and writes\n" + "\toperations are reads. In other words, 9 gets for every 1 put.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, mergereadpercent(70,"Ratio of merges to merges&reads (expressed\n" + "\tas percentage) for the ReadRandomMergeRandom workload. The\n" + "\tdefault value 70 means 70% out of all read and merge operations\n" + "\tare merges. In other words, 7 merges for every 3 gets.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, deletepercent(2,"Percentage of deletes out of reads/writes/\n" + "\tdeletes (used in RandomWithVerify only). RandomWithVerify\n" + "\tcalculates writepercent as (100 - FLAGS_readwritepercent -\n" + "\tdeletepercent), so deletepercent must be smaller than (100 -\n" + "\tFLAGS_readwritepercent)") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, delete_obsolete_files_period_micros(0,"Option to delete\n" + "\tobsolete files periodically. 0 means that obsolete files are\n" + "\tdeleted after every compaction run.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, compression_type("snappy", "Algorithm used to compress the database.") { @Override public Object parseValue(String value) { return value; } }, compression_level(-1, "Compression level. For zlib this should be -1 for the\n" + "\tdefault level, or between 0 and 9.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, min_level_to_compress(-1,"If non-negative, compression starts\n" + "\tfrom this level. Levels with number < min_level_to_compress are\n" + "\tnot compressed. Otherwise, apply compression_type to\n" + "\tall levels.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, table_cache_numshardbits(4,"") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, stats_interval(0,"Stats are reported every N operations when\n" + "\tthis is greater than zero. When 0 the interval grows over time.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, stats_per_interval(0,"Reports additional stats per interval when\n" + "\tthis is greater than 0.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, perf_level(0,"Level of perf collection.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, soft_rate_limit(0.0,"") { @Override public Object parseValue(String value) { return Double.parseDouble(value); } }, hard_rate_limit(0.0,"When not equal to 0 this make threads\n" + "\tsleep at each stats reporting interval until the compaction\n" + "\tscore for all levels is less than or equal to this value.") { @Override public Object parseValue(String value) { return Double.parseDouble(value); } }, rate_limit_delay_max_milliseconds(1000, "When hard_rate_limit is set then this is the max time a put will\n" + "\tbe stalled.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, max_grandparent_overlap_factor(10,"Control maximum bytes of\n" + "\toverlaps in grandparent (i.e., level+2) before we stop building a\n" + "\tsingle file in a level->level+1 compaction.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, readonly(false,"Run read only benchmarks.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, disable_auto_compactions(false,"Do not auto trigger compactions.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, source_compaction_factor(1,"Cap the size of data in level-K for\n" + "\ta compaction run that compacts Level-K with Level-(K+1) (for\n" + "\tK >= 1)") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, wal_ttl_seconds(0L,"Set the TTL for the WAL Files in seconds.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, wal_size_limit_MB(0L,"Set the size limit for the WAL Files\n" + "\tin MB.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, /* TODO(yhchiang): enable the following bufferedio(rocksdb::EnvOptions().use_os_buffer, "Allow buffered io using OS buffers.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, */ mmap_read(false, "Allow reads to occur via mmap-ing files.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, mmap_write(false, "Allow writes to occur via mmap-ing files.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, advise_random_on_open(defaultOptions_.adviseRandomOnOpen(), "Advise random access on table file open.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, compaction_fadvice("NORMAL", "Access pattern advice when a file is compacted.") { @Override public Object parseValue(String value) { return value; } }, use_tailing_iterator(false, "Use tailing iterator to access a series of keys instead of get.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, use_adaptive_mutex(defaultOptions_.useAdaptiveMutex(), "Use adaptive mutex.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, bytes_per_sync(defaultOptions_.bytesPerSync(), "Allows OS to incrementally sync files to disk while they are\n" + "\tbeing written, in the background. Issue one request for every\n" + "\tbytes_per_sync written. 0 turns it off.") { @Override public Object parseValue(String value) { return Long.parseLong(value); } }, filter_deletes(false," On true, deletes use bloom-filter and drop\n" + "\tthe delete if key not present.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }, max_successive_merges(0,"Maximum number of successive merge\n" + "\toperations on a key in the memtable.") { @Override public Object parseValue(String value) { return Integer.parseInt(value); } }, db("/tmp/rocksdbjni-bench", "Use the db with the following name.") { @Override public Object parseValue(String value) { return value; } }, use_mem_env(false, "Use RocksMemEnv instead of default filesystem based\n" + "environment.") { @Override public Object parseValue(String value) { return parseBoolean(value); } }; private Flag(Object defaultValue, String desc) { defaultValue_ = defaultValue; desc_ = desc; } public Object getDefaultValue() { return defaultValue_; } public String desc() { return desc_; } public boolean parseBoolean(String value) { if (value.equals("1")) { return true; } else if (value.equals("0")) { return false; } return Boolean.parseBoolean(value); } protected abstract Object parseValue(String value); private final Object defaultValue_; private final String desc_; } private static class RandomGenerator { private final byte[] data_; private int dataLength_; private int position_; private double compressionRatio_; Random rand_; private RandomGenerator(long seed, double compressionRatio) { // We use a limited amount of data over and over again and ensure // that it is larger than the compression window (32KB), and also byte[] value = new byte[100]; // large enough to serve all typical value sizes we want to write. rand_ = new Random(seed); dataLength_ = value.length * 10000; data_ = new byte[dataLength_]; compressionRatio_ = compressionRatio; int pos = 0; while (pos < dataLength_) { compressibleBytes(value); System.arraycopy(value, 0, data_, pos, Math.min(value.length, dataLength_ - pos)); pos += value.length; } } private void compressibleBytes(byte[] value) { int baseLength = value.length; if (compressionRatio_ < 1.0d) { baseLength = (int) (compressionRatio_ * value.length + 0.5); } if (baseLength <= 0) { baseLength = 1; } int pos; for (pos = 0; pos < baseLength; ++pos) { value[pos] = (byte) (' ' + rand_.nextInt(95)); // ' ' .. '~' } while (pos < value.length) { System.arraycopy(value, 0, value, pos, Math.min(baseLength, value.length - pos)); pos += baseLength; } } private void generate(byte[] value) { if (position_ + value.length > data_.length) { position_ = 0; assert(value.length <= data_.length); } position_ += value.length; System.arraycopy(data_, position_ - value.length, value, 0, value.length); } } boolean isFinished() { synchronized(finishLock_) { return isFinished_; } } void setFinished(boolean flag) { synchronized(finishLock_) { isFinished_ = flag; } } RocksDB db_; final List benchmarks_; final int num_; final int reads_; final int keySize_; final int valueSize_; final int threadNum_; final int writesPerSeconds_; final long randSeed_; final boolean useExisting_; final String databaseDir_; double compressionRatio_; RandomGenerator gen_; long startTime_; // env boolean useMemenv_; // memtable related final int maxWriteBufferNumber_; final int prefixSize_; final int keysPerPrefix_; final String memtable_; final long hashBucketCount_; // sst format related boolean usePlainTable_; Object finishLock_; boolean isFinished_; Map flags_; // as the scope of a static member equals to the scope of the problem, // we let its c++ pointer to be disposed in its finalizer. static Options defaultOptions_ = new Options(); static BlockBasedTableConfig defaultBlockBasedTableOptions_ = new BlockBasedTableConfig(); String compressionType_; CompressionType compression_; } rocksdb-4.5.1/java/crossbuild/000077500000000000000000000000001267530634000162505ustar00rootroot00000000000000rocksdb-4.5.1/java/crossbuild/Vagrantfile000066400000000000000000000012741267530634000204410ustar00rootroot00000000000000# -*- mode: ruby -*- # vi: set ft=ruby : # Vagrantfile API/syntax version. Don't touch unless you know what you're doing! VAGRANTFILE_API_VERSION = "2" Vagrant.configure(VAGRANTFILE_API_VERSION) do |config| config.vm.define "linux32" do |linux32| linux32.vm.box = "hansode/centos-5.6-i386" end config.vm.define "linux64" do |linux64| linux64.vm.box = "hansode/centos-5.6-x86_64" end config.vm.provider "virtualbox" do |v| v.memory = 2048 v.cpus = 4 end config.vm.provision :shell, path: "build-linux-centos.sh" config.vm.synced_folder "../target", "/rocksdb-build" config.vm.synced_folder "../..", "/rocksdb", type: "rsync" config.vm.boot_timeout = 1200 end rocksdb-4.5.1/java/crossbuild/build-linux-centos.sh000077500000000000000000000022021267530634000223300ustar00rootroot00000000000000#!/usr/bin/env bash # install all required packages for rocksdb that are available through yum ARCH=$(uname -i) sudo yum -y install openssl java-1.7.0-openjdk-devel.$ARCH # install gcc/g++ 4.8.2 via CERN (http://linux.web.cern.ch/linux/devtoolset/) sudo wget -O /etc/yum.repos.d/slc5-devtoolset.repo http://linuxsoft.cern.ch/cern/devtoolset/slc5-devtoolset.repo sudo wget -O /etc/pki/rpm-gpg/RPM-GPG-KEY-cern http://ftp.mirrorservice.org/sites/ftp.scientificlinux.org/linux/scientific/51/i386/RPM-GPG-KEYs/RPM-GPG-KEY-cern sudo yum -y install devtoolset-2 wget http://gflags.googlecode.com/files/gflags-2.0-no-svn-files.tar.gz tar xvfz gflags-2.0-no-svn-files.tar.gz; cd gflags-2.0; scl enable devtoolset-2 ./configure; scl enable devtoolset-2 make; sudo make install export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib # set java home so we can build rocksdb jars export JAVA_HOME=/usr/lib/jvm/java-1.7.0 # build rocksdb cd /rocksdb scl enable devtoolset-2 'make jclean clean' scl enable devtoolset-2 'PORTABLE=1 make rocksdbjavastatic' cp /rocksdb/java/target/librocksdbjni-* /rocksdb-build cp /rocksdb/java/target/rocksdbjni-* /rocksdb-build rocksdb-4.5.1/java/crossbuild/build-linux.sh000077500000000000000000000007321267530634000210450ustar00rootroot00000000000000#!/usr/bin/env bash # install all required packages for rocksdb sudo apt-get update sudo apt-get -y install git make gcc g++ libgflags-dev libsnappy-dev zlib1g-dev libbz2-dev default-jdk # set java home so we can build rocksdb jars export JAVA_HOME=$(echo /usr/lib/jvm/java-7-openjdk*) cd /rocksdb make jclean clean make -j 4 rocksdbjavastatic cp /rocksdb/java/target/librocksdbjni-* /rocksdb-build cp /rocksdb/java/target/rocksdbjni-* /rocksdb-build sudo shutdown -h now rocksdb-4.5.1/java/jdb_bench.sh000077500000000000000000000005141267530634000163340ustar00rootroot00000000000000PLATFORM=64 if [ `getconf LONG_BIT` != "64" ] then PLATFORM=32 fi ROCKS_JAR=`find target -name rocksdbjni*.jar` echo "Running benchmark in $PLATFORM-Bit mode." java -server -d$PLATFORM -XX:NewSize=4m -XX:+AggressiveOpts -Djava.library.path=target -cp "${ROCKS_JAR}:benchmark/target/classes" org.rocksdb.benchmark.DbBenchmark $@ rocksdb-4.5.1/java/rocksjni.pom000066400000000000000000000140301267530634000164340ustar00rootroot00000000000000 4.0.0 RocksDB JNI http://rocksdb.org/ org.rocksdb rocksdbjni - RocksDB fat jar that contains .so files for linux32 and linux64, and jnilib files for Mac OSX. Apache License 2.0 http://www.apache.org/licenses/LICENSE-2.0.html repo scm:git:git://github.com/dropwizard/metrics.git scm:git:git@github.com:dropwizard/metrics.git http://github.com/dropwizard/metrics/ HEAD Facebook help@facebook.com America/New_York architect 1.7 1.7 UTF-8 org.apache.maven.plugins maven-compiler-plugin 3.2 ${project.build.source} ${project.build.target} ${project.build.sourceEncoding} org.apache.maven.plugins maven-surefire-plugin 2.18.1 ${argLine} -ea -Xcheck:jni -Djava.library.path=${project.build.directory} false false ${project.build.directory}/* org.jacoco jacoco-maven-plugin 0.7.2.201409121644 prepare-agent report prepare-package report org.codehaus.gmaven groovy-maven-plugin 2.0 process-classes execute Xenu String fileContents = new File("${project.basedir}/../include/rocksdb/version.h").getText('UTF-8') matcher = (fileContents =~ /(?s).*ROCKSDB_MAJOR ([0-9]+).*?/) String major_version = matcher.getAt(0).getAt(1) matcher = (fileContents =~ /(?s).*ROCKSDB_MINOR ([0-9]+).*?/) String minor_version = matcher.getAt(0).getAt(1) matcher = (fileContents =~ /(?s).*ROCKSDB_PATCH ([0-9]+).*?/) String patch_version = matcher.getAt(0).getAt(1) String version = String.format('%s.%s.%s', major_version, minor_version, patch_version) // Set version to be used in pom.properties project.version = version // Set version to be set as jar name project.build.finalName = project.artifactId + "-" + version junit junit 4.12 test org.assertj assertj-core 1.7.1 test org.mockito mockito-all 1.10.19 test rocksdb-4.5.1/java/rocksjni/000077500000000000000000000000001267530634000157215ustar00rootroot00000000000000rocksdb-4.5.1/java/rocksjni/backupablejni.cc000066400000000000000000000234311267530634000210250ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::BackupableDB and rocksdb::BackupableDBOptions methods // from Java side. #include #include #include #include #include #include "include/org_rocksdb_BackupableDB.h" #include "include/org_rocksdb_BackupableDBOptions.h" #include "rocksjni/portal.h" #include "rocksdb/utilities/backupable_db.h" /* * Class: org_rocksdb_BackupableDB * Method: open * Signature: (JJ)V */ void Java_org_rocksdb_BackupableDB_open( JNIEnv* env, jobject jbdb, jlong jdb_handle, jlong jopt_handle) { auto db = reinterpret_cast(jdb_handle); auto opt = reinterpret_cast(jopt_handle); auto bdb = new rocksdb::BackupableDB(db, *opt); // as BackupableDB extends RocksDB on the java side, we can reuse // the RocksDB portal here. rocksdb::RocksDBJni::setHandle(env, jbdb, bdb); } /* * Class: org_rocksdb_BackupableDB * Method: createNewBackup * Signature: (JZ)V */ void Java_org_rocksdb_BackupableDB_createNewBackup( JNIEnv* env, jobject jbdb, jlong jhandle, jboolean jflag) { rocksdb::Status s = reinterpret_cast(jhandle)->CreateNewBackup(jflag); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_BackupableDB * Method: purgeOldBackups * Signature: (JI)V */ void Java_org_rocksdb_BackupableDB_purgeOldBackups( JNIEnv* env, jobject jbdb, jlong jhandle, jint jnumBackupsToKeep) { rocksdb::Status s = reinterpret_cast(jhandle)-> PurgeOldBackups(jnumBackupsToKeep); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_BackupableDB * Method: deleteBackup0 * Signature: (JI)V */ void Java_org_rocksdb_BackupableDB_deleteBackup0(JNIEnv* env, jobject jobj, jlong jhandle, jint jbackup_id) { auto rdb = reinterpret_cast(jhandle); rocksdb::Status s = rdb->DeleteBackup(jbackup_id); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_BackupableDB * Method: getBackupInfo * Signature: (J)Ljava/util/List; */ jobject Java_org_rocksdb_BackupableDB_getBackupInfo( JNIEnv* env, jobject jbdb, jlong jhandle) { std::vector backup_infos; reinterpret_cast(jhandle)-> GetBackupInfo(&backup_infos); return rocksdb::BackupInfoListJni::getBackupInfo(env, backup_infos); } /* * Class: org_rocksdb_BackupableDB * Method: getCorruptedBackups * Signature: (J)[I; */ jintArray Java_org_rocksdb_BackupableDB_getCorruptedBackups( JNIEnv* env, jobject jbdb, jlong jhandle) { std::vector backup_ids; reinterpret_cast(jhandle)-> GetCorruptedBackups(&backup_ids); // store backupids in int array std::vector int_backup_ids(backup_ids.begin(), backup_ids.end()); // Store ints in java array jintArray ret_backup_ids; // Its ok to loose precision here (64->32) jsize ret_backup_ids_size = static_cast(backup_ids.size()); ret_backup_ids = env->NewIntArray(ret_backup_ids_size); env->SetIntArrayRegion(ret_backup_ids, 0, ret_backup_ids_size, int_backup_ids.data()); return ret_backup_ids; } /* * Class: org_rocksdb_BackupableDB * Method: garbageCollect * Signature: (J)V */ void Java_org_rocksdb_BackupableDB_garbageCollect(JNIEnv* env, jobject jobj, jlong jhandle) { auto db = reinterpret_cast(jhandle); rocksdb::Status s = db->GarbageCollect(); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /////////////////////////////////////////////////////////////////////////// // BackupDBOptions /* * Class: org_rocksdb_BackupableDBOptions * Method: newBackupableDBOptions * Signature: (Ljava/lang/String;)V */ void Java_org_rocksdb_BackupableDBOptions_newBackupableDBOptions( JNIEnv* env, jobject jobj, jstring jpath) { const char* cpath = env->GetStringUTFChars(jpath, 0); auto bopt = new rocksdb::BackupableDBOptions(cpath); env->ReleaseStringUTFChars(jpath, cpath); rocksdb::BackupableDBOptionsJni::setHandle(env, jobj, bopt); } /* * Class: org_rocksdb_BackupableDBOptions * Method: backupDir * Signature: (J)Ljava/lang/String; */ jstring Java_org_rocksdb_BackupableDBOptions_backupDir( JNIEnv* env, jobject jopt, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return env->NewStringUTF(bopt->backup_dir.c_str()); } /* * Class: org_rocksdb_BackupableDBOptions * Method: setShareTableFiles * Signature: (JZ)V */ void Java_org_rocksdb_BackupableDBOptions_setShareTableFiles( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { auto bopt = reinterpret_cast(jhandle); bopt->share_table_files = flag; } /* * Class: org_rocksdb_BackupableDBOptions * Method: shareTableFiles * Signature: (J)Z */ jboolean Java_org_rocksdb_BackupableDBOptions_shareTableFiles( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->share_table_files; } /* * Class: org_rocksdb_BackupableDBOptions * Method: setSync * Signature: (JZ)V */ void Java_org_rocksdb_BackupableDBOptions_setSync( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { auto bopt = reinterpret_cast(jhandle); bopt->sync = flag; } /* * Class: org_rocksdb_BackupableDBOptions * Method: sync * Signature: (J)Z */ jboolean Java_org_rocksdb_BackupableDBOptions_sync( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->sync; } /* * Class: org_rocksdb_BackupableDBOptions * Method: setDestroyOldData * Signature: (JZ)V */ void Java_org_rocksdb_BackupableDBOptions_setDestroyOldData( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { auto bopt = reinterpret_cast(jhandle); bopt->destroy_old_data = flag; } /* * Class: org_rocksdb_BackupableDBOptions * Method: destroyOldData * Signature: (J)Z */ jboolean Java_org_rocksdb_BackupableDBOptions_destroyOldData( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->destroy_old_data; } /* * Class: org_rocksdb_BackupableDBOptions * Method: setBackupLogFiles * Signature: (JZ)V */ void Java_org_rocksdb_BackupableDBOptions_setBackupLogFiles( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { auto bopt = reinterpret_cast(jhandle); bopt->backup_log_files = flag; } /* * Class: org_rocksdb_BackupableDBOptions * Method: backupLogFiles * Signature: (J)Z */ jboolean Java_org_rocksdb_BackupableDBOptions_backupLogFiles( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->backup_log_files; } /* * Class: org_rocksdb_BackupableDBOptions * Method: setBackupRateLimit * Signature: (JJ)V */ void Java_org_rocksdb_BackupableDBOptions_setBackupRateLimit( JNIEnv* env, jobject jobj, jlong jhandle, jlong jbackup_rate_limit) { auto bopt = reinterpret_cast(jhandle); bopt->backup_rate_limit = jbackup_rate_limit; } /* * Class: org_rocksdb_BackupableDBOptions * Method: backupRateLimit * Signature: (J)J */ jlong Java_org_rocksdb_BackupableDBOptions_backupRateLimit( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->backup_rate_limit; } /* * Class: org_rocksdb_BackupableDBOptions * Method: setRestoreRateLimit * Signature: (JJ)V */ void Java_org_rocksdb_BackupableDBOptions_setRestoreRateLimit( JNIEnv* env, jobject jobj, jlong jhandle, jlong jrestore_rate_limit) { auto bopt = reinterpret_cast(jhandle); bopt->restore_rate_limit = jrestore_rate_limit; } /* * Class: org_rocksdb_BackupableDBOptions * Method: restoreRateLimit * Signature: (J)J */ jlong Java_org_rocksdb_BackupableDBOptions_restoreRateLimit( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->restore_rate_limit; } /* * Class: org_rocksdb_BackupableDBOptions * Method: setShareFilesWithChecksum * Signature: (JZ)V */ void Java_org_rocksdb_BackupableDBOptions_setShareFilesWithChecksum( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { auto bopt = reinterpret_cast(jhandle); bopt->share_files_with_checksum = flag; } /* * Class: org_rocksdb_BackupableDBOptions * Method: shareFilesWithChecksum * Signature: (J)Z */ jboolean Java_org_rocksdb_BackupableDBOptions_shareFilesWithChecksum( JNIEnv* env, jobject jobj, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); return bopt->share_files_with_checksum; } /* * Class: org_rocksdb_BackupableDBOptions * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_BackupableDBOptions_disposeInternal( JNIEnv* env, jobject jopt, jlong jhandle) { auto bopt = reinterpret_cast(jhandle); assert(bopt); delete bopt; rocksdb::BackupableDBOptionsJni::setHandle(env, jopt, nullptr); } rocksdb-4.5.1/java/rocksjni/backupenginejni.cc000066400000000000000000000151431267530634000213700ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling C++ rocksdb::BackupEngine methods from the Java side. #include #include #include "include/org_rocksdb_BackupEngine.h" #include "rocksdb/utilities/backupable_db.h" #include "rocksjni/portal.h" /* * Class: org_rocksdb_BackupEngine * Method: open * Signature: (JJ)V */ void Java_org_rocksdb_BackupEngine_open( JNIEnv* env, jobject jbe, jlong env_handle, jlong backupable_db_options_handle) { auto* rocks_env = reinterpret_cast(env_handle); auto* backupable_db_options = reinterpret_cast( backupable_db_options_handle); rocksdb::BackupEngine* backup_engine; auto status = rocksdb::BackupEngine::Open(rocks_env, *backupable_db_options, &backup_engine); if (status.ok()) { rocksdb::BackupEngineJni::setHandle(env, jbe, backup_engine); return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: createNewBackup * Signature: (JJZ)V */ void Java_org_rocksdb_BackupEngine_createNewBackup( JNIEnv* env, jobject jbe, jlong jbe_handle, jlong db_handle, jboolean jflush_before_backup) { auto* db = reinterpret_cast(db_handle); auto* backup_engine = reinterpret_cast(jbe_handle); auto status = backup_engine->CreateNewBackup(db, static_cast(jflush_before_backup)); if (status.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: getBackupInfo * Signature: (J)Ljava/util/List; */ jobject Java_org_rocksdb_BackupEngine_getBackupInfo( JNIEnv* env, jobject jbe, jlong jbe_handle) { auto* backup_engine = reinterpret_cast(jbe_handle); std::vector backup_infos; backup_engine->GetBackupInfo(&backup_infos); return rocksdb::BackupInfoListJni::getBackupInfo(env, backup_infos); } /* * Class: org_rocksdb_BackupEngine * Method: getCorruptedBackups * Signature: (J)[I */ jintArray Java_org_rocksdb_BackupEngine_getCorruptedBackups( JNIEnv* env, jobject jbe, jlong jbe_handle) { auto* backup_engine = reinterpret_cast(jbe_handle); std::vector backup_ids; backup_engine->GetCorruptedBackups(&backup_ids); // store backupids in int array std::vector int_backup_ids(backup_ids.begin(), backup_ids.end()); // Store ints in java array jintArray ret_backup_ids; // Its ok to loose precision here (64->32) jsize ret_backup_ids_size = static_cast(backup_ids.size()); ret_backup_ids = env->NewIntArray(ret_backup_ids_size); env->SetIntArrayRegion(ret_backup_ids, 0, ret_backup_ids_size, int_backup_ids.data()); return ret_backup_ids; } /* * Class: org_rocksdb_BackupEngine * Method: garbageCollect * Signature: (J)V */ void Java_org_rocksdb_BackupEngine_garbageCollect( JNIEnv* env, jobject jbe, jlong jbe_handle) { auto* backup_engine = reinterpret_cast(jbe_handle); auto status = backup_engine->GarbageCollect(); if (status.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: purgeOldBackups * Signature: (JI)V */ void Java_org_rocksdb_BackupEngine_purgeOldBackups( JNIEnv* env, jobject jbe, jlong jbe_handle, jint jnum_backups_to_keep) { auto* backup_engine = reinterpret_cast(jbe_handle); auto status = backup_engine-> PurgeOldBackups(static_cast(jnum_backups_to_keep)); if (status.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: deleteBackup * Signature: (JI)V */ void Java_org_rocksdb_BackupEngine_deleteBackup( JNIEnv* env, jobject jbe, jlong jbe_handle, jint jbackup_id) { auto* backup_engine = reinterpret_cast(jbe_handle); auto status = backup_engine->DeleteBackup(static_cast(jbackup_id)); if (status.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: restoreDbFromBackup * Signature: (JILjava/lang/String;Ljava/lang/String;J)V */ void Java_org_rocksdb_BackupEngine_restoreDbFromBackup( JNIEnv* env, jobject jbe, jlong jbe_handle, jint jbackup_id, jstring jdb_dir, jstring jwal_dir, jlong jrestore_options_handle) { auto* backup_engine = reinterpret_cast(jbe_handle); const char* db_dir = env->GetStringUTFChars(jdb_dir, 0); const char* wal_dir = env->GetStringUTFChars(jwal_dir, 0); auto* restore_options = reinterpret_cast(jrestore_options_handle); auto status = backup_engine->RestoreDBFromBackup( static_cast(jbackup_id), db_dir, wal_dir, *restore_options); env->ReleaseStringUTFChars(jwal_dir, wal_dir); env->ReleaseStringUTFChars(jdb_dir, db_dir); if (status.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: restoreDbFromLatestBackup * Signature: (JLjava/lang/String;Ljava/lang/String;J)V */ void Java_org_rocksdb_BackupEngine_restoreDbFromLatestBackup( JNIEnv* env, jobject jbe, jlong jbe_handle, jstring jdb_dir, jstring jwal_dir, jlong jrestore_options_handle) { auto* backup_engine = reinterpret_cast(jbe_handle); const char* db_dir = env->GetStringUTFChars(jdb_dir, 0); const char* wal_dir = env->GetStringUTFChars(jwal_dir, 0); auto* restore_options = reinterpret_cast(jrestore_options_handle); auto status = backup_engine->RestoreDBFromLatestBackup(db_dir, wal_dir, *restore_options); env->ReleaseStringUTFChars(jwal_dir, wal_dir); env->ReleaseStringUTFChars(jdb_dir, db_dir); if (status.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, status); } /* * Class: org_rocksdb_BackupEngine * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_BackupEngine_disposeInternal( JNIEnv* env, jobject jbe, jlong jbe_handle) { delete reinterpret_cast(jbe_handle); } rocksdb-4.5.1/java/rocksjni/checkpoint.cc000066400000000000000000000036401267530634000203620ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::Checkpoint methods from Java side. #include #include #include #include #include "include/org_rocksdb_Checkpoint.h" #include "rocksjni/portal.h" #include "rocksdb/db.h" #include "rocksdb/utilities/checkpoint.h" /* * Class: org_rocksdb_Checkpoint * Method: newCheckpoint * Signature: (J)J */ jlong Java_org_rocksdb_Checkpoint_newCheckpoint(JNIEnv* env, jclass jclazz, jlong jdb_handle) { auto db = reinterpret_cast(jdb_handle); rocksdb::Checkpoint* checkpoint; rocksdb::Checkpoint::Create(db, &checkpoint); return reinterpret_cast(checkpoint); } /* * Class: org_rocksdb_Checkpoint * Method: dispose * Signature: (J)V */ void Java_org_rocksdb_Checkpoint_disposeInternal(JNIEnv* env, jobject jobj, jlong jhandle) { auto checkpoint = reinterpret_cast(jhandle); assert(checkpoint); delete checkpoint; } /* * Class: org_rocksdb_Checkpoint * Method: createCheckpoint * Signature: (JLjava/lang/String;)V */ void Java_org_rocksdb_Checkpoint_createCheckpoint( JNIEnv* env, jobject jobj, jlong jcheckpoint_handle, jstring jcheckpoint_path) { auto checkpoint = reinterpret_cast( jcheckpoint_handle); const char* checkpoint_path = env->GetStringUTFChars( jcheckpoint_path, 0); rocksdb::Status s = checkpoint->CreateCheckpoint( checkpoint_path); env->ReleaseStringUTFChars(jcheckpoint_path, checkpoint_path); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } rocksdb-4.5.1/java/rocksjni/columnfamilyhandle.cc000066400000000000000000000015171267530634000221070ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::Iterator methods from Java side. #include #include #include #include "include/org_rocksdb_ColumnFamilyHandle.h" #include "rocksjni/portal.h" /* * Class: org_rocksdb_ColumnFamilyHandle * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_ColumnFamilyHandle_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { auto it = reinterpret_cast(handle); delete it; } rocksdb-4.5.1/java/rocksjni/compaction_filter.cc000066400000000000000000000014451267530634000217350ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for // rocksdb::CompactionFilter. #include #include "rocksdb/compaction_filter.h" // /* * Class: org_rocksdb_AbstractCompactionFilter * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_AbstractCompactionFilter_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } // rocksdb-4.5.1/java/rocksjni/comparator.cc000066400000000000000000000042441267530634000204030ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for // rocksdb::Comparator. #include #include #include #include #include #include "include/org_rocksdb_AbstractComparator.h" #include "include/org_rocksdb_Comparator.h" #include "include/org_rocksdb_DirectComparator.h" #include "rocksjni/comparatorjnicallback.h" #include "rocksjni/portal.h" // /* * Class: org_rocksdb_Comparator * Method: createNewComparator0 * Signature: ()V */ void Java_org_rocksdb_Comparator_createNewComparator0( JNIEnv* env, jobject jobj, jlong copt_handle) { const rocksdb::ComparatorJniCallbackOptions* copt = reinterpret_cast(copt_handle); const rocksdb::ComparatorJniCallback* c = new rocksdb::ComparatorJniCallback(env, jobj, copt); rocksdb::AbstractComparatorJni::setHandle(env, jobj, c); } // // use_adaptive_mutex)), mtx_findShortestSeparator(new port::Mutex(copt->use_adaptive_mutex)) { // Note: Comparator methods may be accessed by multiple threads, // so we ref the jvm not the env const jint rs __attribute__((unused)) = env->GetJavaVM(&m_jvm); assert(rs == JNI_OK); // Note: we want to access the Java Comparator instance // across multiple method calls, so we create a global ref m_jComparator = env->NewGlobalRef(jComparator); // Note: The name of a Comparator will not change during it's lifetime, // so we cache it in a global var jmethodID jNameMethodId = AbstractComparatorJni::getNameMethodId(env); jstring jsName = (jstring)env->CallObjectMethod(m_jComparator, jNameMethodId); m_name = JniUtil::copyString(env, jsName); // also releases jsName m_jCompareMethodId = AbstractComparatorJni::getCompareMethodId(env); m_jFindShortestSeparatorMethodId = AbstractComparatorJni::getFindShortestSeparatorMethodId(env); m_jFindShortSuccessorMethodId = AbstractComparatorJni::getFindShortSuccessorMethodId(env); } /** * Attach/Get a JNIEnv for the current native thread */ JNIEnv* BaseComparatorJniCallback::getJniEnv() const { JNIEnv *env; jint rs __attribute__((unused)) = m_jvm->AttachCurrentThread(reinterpret_cast(&env), NULL); assert(rs == JNI_OK); return env; } const char* BaseComparatorJniCallback::Name() const { return m_name.c_str(); } int BaseComparatorJniCallback::Compare(const Slice& a, const Slice& b) const { JNIEnv* m_env = getJniEnv(); // TODO(adamretter): slice objects can potentially be cached using thread // local variables to avoid locking. Could make this configurable depending on // performance. mtx_compare->Lock(); AbstractSliceJni::setHandle(m_env, m_jSliceA, &a); AbstractSliceJni::setHandle(m_env, m_jSliceB, &b); jint result = m_env->CallIntMethod(m_jComparator, m_jCompareMethodId, m_jSliceA, m_jSliceB); mtx_compare->Unlock(); m_jvm->DetachCurrentThread(); return result; } void BaseComparatorJniCallback::FindShortestSeparator( std::string* start, const Slice& limit) const { if (start == nullptr) { return; } JNIEnv* m_env = getJniEnv(); const char* startUtf = start->c_str(); jstring jsStart = m_env->NewStringUTF(startUtf); // TODO(adamretter): slice object can potentially be cached using thread local // variable to avoid locking. Could make this configurable depending on // performance. mtx_findShortestSeparator->Lock(); AbstractSliceJni::setHandle(m_env, m_jSliceLimit, &limit); jstring jsResultStart = (jstring)m_env->CallObjectMethod(m_jComparator, m_jFindShortestSeparatorMethodId, jsStart, m_jSliceLimit); mtx_findShortestSeparator->Unlock(); m_env->DeleteLocalRef(jsStart); if (jsResultStart != nullptr) { // update start with result *start = JniUtil::copyString(m_env, jsResultStart); // also releases jsResultStart } m_jvm->DetachCurrentThread(); } void BaseComparatorJniCallback::FindShortSuccessor(std::string* key) const { if (key == nullptr) { return; } JNIEnv* m_env = getJniEnv(); const char* keyUtf = key->c_str(); jstring jsKey = m_env->NewStringUTF(keyUtf); jstring jsResultKey = (jstring)m_env->CallObjectMethod(m_jComparator, m_jFindShortSuccessorMethodId, jsKey); m_env->DeleteLocalRef(jsKey); if (jsResultKey != nullptr) { // updates key with result, also releases jsResultKey. *key = JniUtil::copyString(m_env, jsResultKey); } m_jvm->DetachCurrentThread(); } BaseComparatorJniCallback::~BaseComparatorJniCallback() { JNIEnv* m_env = getJniEnv(); m_env->DeleteGlobalRef(m_jComparator); // Note: do not need to explicitly detach, as this function is effectively // called from the Java class's disposeInternal method, and so already // has an attached thread, getJniEnv above is just a no-op Attach to get // the env jvm->DetachCurrentThread(); } ComparatorJniCallback::ComparatorJniCallback( JNIEnv* env, jobject jComparator, const ComparatorJniCallbackOptions* copt) : BaseComparatorJniCallback(env, jComparator, copt) { m_jSliceA = env->NewGlobalRef(SliceJni::construct0(env)); m_jSliceB = env->NewGlobalRef(SliceJni::construct0(env)); m_jSliceLimit = env->NewGlobalRef(SliceJni::construct0(env)); } ComparatorJniCallback::~ComparatorJniCallback() { JNIEnv* m_env = getJniEnv(); m_env->DeleteGlobalRef(m_jSliceA); m_env->DeleteGlobalRef(m_jSliceB); m_env->DeleteGlobalRef(m_jSliceLimit); } DirectComparatorJniCallback::DirectComparatorJniCallback( JNIEnv* env, jobject jComparator, const ComparatorJniCallbackOptions* copt) : BaseComparatorJniCallback(env, jComparator, copt) { m_jSliceA = env->NewGlobalRef(DirectSliceJni::construct0(env)); m_jSliceB = env->NewGlobalRef(DirectSliceJni::construct0(env)); m_jSliceLimit = env->NewGlobalRef(DirectSliceJni::construct0(env)); } DirectComparatorJniCallback::~DirectComparatorJniCallback() { JNIEnv* m_env = getJniEnv(); m_env->DeleteGlobalRef(m_jSliceA); m_env->DeleteGlobalRef(m_jSliceB); m_env->DeleteGlobalRef(m_jSliceLimit); } } // namespace rocksdb rocksdb-4.5.1/java/rocksjni/comparatorjnicallback.h000066400000000000000000000063471267530634000224310ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the callback "bridge" between Java and C++ for // rocksdb::Comparator and rocksdb::DirectComparator. #ifndef JAVA_ROCKSJNI_COMPARATORJNICALLBACK_H_ #define JAVA_ROCKSJNI_COMPARATORJNICALLBACK_H_ #include #include #include "rocksdb/comparator.h" #include "rocksdb/slice.h" #include "port/port.h" namespace rocksdb { struct ComparatorJniCallbackOptions { // Use adaptive mutex, which spins in the user space before resorting // to kernel. This could reduce context switch when the mutex is not // heavily contended. However, if the mutex is hot, we could end up // wasting spin time. // Default: false bool use_adaptive_mutex; ComparatorJniCallbackOptions() : use_adaptive_mutex(false) { } }; /** * This class acts as a bridge between C++ * and Java. The methods in this class will be * called back from the RocksDB storage engine (C++) * we then callback to the appropriate Java method * this enables Comparators to be implemented in Java. * * The design of this Comparator caches the Java Slice * objects that are used in the compare and findShortestSeparator * method callbacks. Instead of creating new objects for each callback * of those functions, by reuse via setHandle we are a lot * faster; Unfortunately this means that we have to * introduce independent locking in regions of each of those methods * via the mutexs mtx_compare and mtx_findShortestSeparator respectively */ class BaseComparatorJniCallback : public Comparator { public: BaseComparatorJniCallback( JNIEnv* env, jobject jComparator, const ComparatorJniCallbackOptions* copt); virtual ~BaseComparatorJniCallback(); virtual const char* Name() const; virtual int Compare(const Slice& a, const Slice& b) const; virtual void FindShortestSeparator( std::string* start, const Slice& limit) const; virtual void FindShortSuccessor(std::string* key) const; private: // used for synchronisation in compare method port::Mutex* mtx_compare; // used for synchronisation in findShortestSeparator method port::Mutex* mtx_findShortestSeparator; JavaVM* m_jvm; jobject m_jComparator; std::string m_name; jmethodID m_jCompareMethodId; jmethodID m_jFindShortestSeparatorMethodId; jmethodID m_jFindShortSuccessorMethodId; protected: JNIEnv* getJniEnv() const; jobject m_jSliceA; jobject m_jSliceB; jobject m_jSliceLimit; }; class ComparatorJniCallback : public BaseComparatorJniCallback { public: ComparatorJniCallback( JNIEnv* env, jobject jComparator, const ComparatorJniCallbackOptions* copt); ~ComparatorJniCallback(); }; class DirectComparatorJniCallback : public BaseComparatorJniCallback { public: DirectComparatorJniCallback( JNIEnv* env, jobject jComparator, const ComparatorJniCallbackOptions* copt); ~DirectComparatorJniCallback(); }; } // namespace rocksdb #endif // JAVA_ROCKSJNI_COMPARATORJNICALLBACK_H_ rocksdb-4.5.1/java/rocksjni/env.cc000066400000000000000000000045631267530634000170300ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::Env methods from Java side. #include "include/org_rocksdb_Env.h" #include "include/org_rocksdb_RocksEnv.h" #include "include/org_rocksdb_RocksMemEnv.h" #include "rocksdb/env.h" /* * Class: org_rocksdb_Env * Method: getDefaultEnvInternal * Signature: ()J */ jlong Java_org_rocksdb_Env_getDefaultEnvInternal( JNIEnv* env, jclass jclazz) { return reinterpret_cast(rocksdb::Env::Default()); } /* * Class: org_rocksdb_Env * Method: setBackgroundThreads * Signature: (JII)V */ void Java_org_rocksdb_Env_setBackgroundThreads( JNIEnv* env, jobject jobj, jlong jhandle, jint num, jint priority) { auto* rocks_env = reinterpret_cast(jhandle); switch (priority) { case org_rocksdb_Env_FLUSH_POOL: rocks_env->SetBackgroundThreads(num, rocksdb::Env::Priority::LOW); break; case org_rocksdb_Env_COMPACTION_POOL: rocks_env->SetBackgroundThreads(num, rocksdb::Env::Priority::HIGH); break; } } /* * Class: org_rocksdb_sEnv * Method: getThreadPoolQueueLen * Signature: (JI)I */ jint Java_org_rocksdb_Env_getThreadPoolQueueLen( JNIEnv* env, jobject jobj, jlong jhandle, jint pool_id) { auto* rocks_env = reinterpret_cast(jhandle); switch (pool_id) { case org_rocksdb_RocksEnv_FLUSH_POOL: return rocks_env->GetThreadPoolQueueLen(rocksdb::Env::Priority::LOW); case org_rocksdb_RocksEnv_COMPACTION_POOL: return rocks_env->GetThreadPoolQueueLen(rocksdb::Env::Priority::HIGH); } return 0; } /* * Class: org_rocksdb_RocksMemEnv * Method: createMemEnv * Signature: ()J */ jlong Java_org_rocksdb_RocksMemEnv_createMemEnv( JNIEnv* env, jclass jclazz) { return reinterpret_cast(rocksdb::NewMemEnv( rocksdb::Env::Default())); } /* * Class: org_rocksdb_RocksMemEnv * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_RocksMemEnv_disposeInternal( JNIEnv* env, jobject jobj, jlong jhandle) { delete reinterpret_cast(jhandle); } rocksdb-4.5.1/java/rocksjni/filter.cc000066400000000000000000000030221267530634000175120ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for // rocksdb::FilterPolicy. #include #include #include #include #include "include/org_rocksdb_Filter.h" #include "include/org_rocksdb_BloomFilter.h" #include "rocksjni/portal.h" #include "rocksdb/filter_policy.h" /* * Class: org_rocksdb_BloomFilter * Method: createBloomFilter * Signature: (IZ)V */ void Java_org_rocksdb_BloomFilter_createNewBloomFilter( JNIEnv* env, jobject jobj, jint bits_per_key, jboolean use_block_base_builder) { rocksdb::FilterPolicy* fp = const_cast( rocksdb::NewBloomFilterPolicy(bits_per_key, use_block_base_builder)); std::shared_ptr *pFilterPolicy = new std::shared_ptr; *pFilterPolicy = std::shared_ptr(fp); rocksdb::FilterJni::setHandle(env, jobj, pFilterPolicy); } /* * Class: org_rocksdb_Filter * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_Filter_disposeInternal( JNIEnv* env, jobject jobj, jlong jhandle) { std::shared_ptr *handle = reinterpret_cast *>(jhandle); handle->reset(); } rocksdb-4.5.1/java/rocksjni/iterator.cc000066400000000000000000000077121267530634000200700ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::Iterator methods from Java side. #include #include #include #include "include/org_rocksdb_RocksIterator.h" #include "rocksjni/portal.h" #include "rocksdb/iterator.h" /* * Class: org_rocksdb_RocksIterator * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_RocksIterator_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { auto it = reinterpret_cast(handle); delete it; } /* * Class: org_rocksdb_RocksIterator * Method: isValid0 * Signature: (J)Z */ jboolean Java_org_rocksdb_RocksIterator_isValid0( JNIEnv* env, jobject jobj, jlong handle) { return reinterpret_cast(handle)->Valid(); } /* * Class: org_rocksdb_RocksIterator * Method: seekToFirst0 * Signature: (J)V */ void Java_org_rocksdb_RocksIterator_seekToFirst0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->SeekToFirst(); } /* * Class: org_rocksdb_RocksIterator * Method: seekToLast0 * Signature: (J)V */ void Java_org_rocksdb_RocksIterator_seekToLast0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->SeekToLast(); } /* * Class: org_rocksdb_RocksIterator * Method: next0 * Signature: (J)V */ void Java_org_rocksdb_RocksIterator_next0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->Next(); } /* * Class: org_rocksdb_RocksIterator * Method: prev0 * Signature: (J)V */ void Java_org_rocksdb_RocksIterator_prev0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->Prev(); } /* * Class: org_rocksdb_RocksIterator * Method: seek0 * Signature: (J[BI)V */ void Java_org_rocksdb_RocksIterator_seek0( JNIEnv* env, jobject jobj, jlong handle, jbyteArray jtarget, jint jtarget_len) { auto it = reinterpret_cast(handle); jbyte* target = env->GetByteArrayElements(jtarget, 0); rocksdb::Slice target_slice( reinterpret_cast(target), jtarget_len); it->Seek(target_slice); env->ReleaseByteArrayElements(jtarget, target, JNI_ABORT); } /* * Class: org_rocksdb_RocksIterator * Method: status0 * Signature: (J)V */ void Java_org_rocksdb_RocksIterator_status0( JNIEnv* env, jobject jobj, jlong handle) { auto it = reinterpret_cast(handle); rocksdb::Status s = it->status(); if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksIterator * Method: key0 * Signature: (J)[B */ jbyteArray Java_org_rocksdb_RocksIterator_key0( JNIEnv* env, jobject jobj, jlong handle) { auto it = reinterpret_cast(handle); rocksdb::Slice key_slice = it->key(); jbyteArray jkey = env->NewByteArray(static_cast(key_slice.size())); env->SetByteArrayRegion(jkey, 0, static_cast(key_slice.size()), reinterpret_cast(key_slice.data())); return jkey; } /* * Class: org_rocksdb_RocksIterator * Method: value0 * Signature: (J)[B */ jbyteArray Java_org_rocksdb_RocksIterator_value0( JNIEnv* env, jobject jobj, jlong handle) { auto it = reinterpret_cast(handle); rocksdb::Slice value_slice = it->value(); jbyteArray jkeyValue = env->NewByteArray(static_cast(value_slice.size())); env->SetByteArrayRegion(jkeyValue, 0, static_cast(value_slice.size()), reinterpret_cast(value_slice.data())); return jkeyValue; } rocksdb-4.5.1/java/rocksjni/loggerjnicallback.cc000066400000000000000000000136121267530634000216700ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the callback "bridge" between Java and C++ for // rocksdb::Logger. #include "include/org_rocksdb_Logger.h" #include "rocksjni/loggerjnicallback.h" #include "rocksjni/portal.h" namespace rocksdb { LoggerJniCallback::LoggerJniCallback( JNIEnv* env, jobject jlogger) { const jint rs __attribute__((unused)) = env->GetJavaVM(&m_jvm); assert(rs == JNI_OK); // Note: we want to access the Java Logger instance // across multiple method calls, so we create a global ref m_jLogger = env->NewGlobalRef(jlogger); m_jLogMethodId = LoggerJni::getLogMethodId(env); } /** * Get JNIEnv for current native thread */ JNIEnv* LoggerJniCallback::getJniEnv() const { JNIEnv *env; jint rs __attribute__((unused)) = m_jvm->AttachCurrentThread(reinterpret_cast(&env), NULL); assert(rs == JNI_OK); return env; } void LoggerJniCallback::Logv(const char* format, va_list ap) { // We implement this method because it is virtual but we don't // use it because we need to know about the log level. } void LoggerJniCallback::Logv(const InfoLogLevel log_level, const char* format, va_list ap) { if (GetInfoLogLevel() <= log_level) { JNIEnv* env = getJniEnv(); // determine InfoLogLevel java enum instance jobject jlog_level; switch (log_level) { case rocksdb::InfoLogLevel::DEBUG_LEVEL: jlog_level = InfoLogLevelJni::DEBUG_LEVEL(env); break; case rocksdb::InfoLogLevel::INFO_LEVEL: jlog_level = InfoLogLevelJni::INFO_LEVEL(env); break; case rocksdb::InfoLogLevel::ERROR_LEVEL: jlog_level = InfoLogLevelJni::ERROR_LEVEL(env); break; case rocksdb::InfoLogLevel::FATAL_LEVEL: jlog_level = InfoLogLevelJni::FATAL_LEVEL(env); break; default: jlog_level = InfoLogLevelJni::FATAL_LEVEL(env); break; } // We try twice: the first time with a fixed-size stack allocated buffer, // and the second time with a much larger dynamically allocated buffer. char buffer[500]; for (int iter = 0; iter < 2; iter++) { char* base; int bufsize; if (iter == 0) { bufsize = sizeof(buffer); base = buffer; } else { bufsize = 30000; base = new char[bufsize]; } char* p = base; char* limit = base + bufsize; // Print the message if (p < limit) { va_list backup_ap; va_copy(backup_ap, ap); p += vsnprintf(p, limit - p, format, backup_ap); va_end(backup_ap); } // Truncate to available space if necessary if (p >= limit) { if (iter == 0) { continue; // Try again with larger buffer } else { p = limit - 1; } } assert(p < limit); *p++ = '\0'; // pass java string to callback handler env->CallVoidMethod( m_jLogger, m_jLogMethodId, jlog_level, env->NewStringUTF(base)); if (base != buffer) { delete[] base; } break; } m_jvm->DetachCurrentThread(); } } LoggerJniCallback::~LoggerJniCallback() { JNIEnv* env = getJniEnv(); env->DeleteGlobalRef(m_jLogger); m_jvm->DetachCurrentThread(); } } // namespace rocksdb /* * Class: org_rocksdb_Logger * Method: createNewLoggerOptions * Signature: (J)V */ void Java_org_rocksdb_Logger_createNewLoggerOptions( JNIEnv* env, jobject jobj, jlong joptions) { rocksdb::LoggerJniCallback* c = new rocksdb::LoggerJniCallback(env, jobj); // set log level c->SetInfoLogLevel(reinterpret_cast (joptions)->info_log_level); std::shared_ptr *pLoggerJniCallback = new std::shared_ptr; *pLoggerJniCallback = std::shared_ptr(c); rocksdb::LoggerJni::setHandle(env, jobj, pLoggerJniCallback); } /* * Class: org_rocksdb_Logger * Method: createNewLoggerDbOptions * Signature: (J)V */ void Java_org_rocksdb_Logger_createNewLoggerDbOptions( JNIEnv* env, jobject jobj, jlong jdb_options) { rocksdb::LoggerJniCallback* c = new rocksdb::LoggerJniCallback(env, jobj); // set log level c->SetInfoLogLevel(reinterpret_cast (jdb_options)->info_log_level); std::shared_ptr *pLoggerJniCallback = new std::shared_ptr; *pLoggerJniCallback = std::shared_ptr(c); rocksdb::LoggerJni::setHandle(env, jobj, pLoggerJniCallback); } /* * Class: org_rocksdb_Logger * Method: setInfoLogLevel * Signature: (JB)V */ void Java_org_rocksdb_Logger_setInfoLogLevel( JNIEnv* env, jobject jobj, jlong jhandle, jbyte jlog_level) { std::shared_ptr *handle = reinterpret_cast *>(jhandle); (*handle)->SetInfoLogLevel(static_cast(jlog_level)); } /* * Class: org_rocksdb_Logger * Method: infoLogLevel * Signature: (J)B */ jbyte Java_org_rocksdb_Logger_infoLogLevel( JNIEnv* env, jobject jobj, jlong jhandle) { std::shared_ptr *handle = reinterpret_cast *>(jhandle); return static_cast((*handle)->GetInfoLogLevel()); } /* * Class: org_rocksdb_Logger * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_Logger_disposeInternal( JNIEnv* env, jobject jobj, jlong jhandle) { std::shared_ptr *handle = reinterpret_cast *>(jhandle); handle->reset(); } rocksdb-4.5.1/java/rocksjni/loggerjnicallback.h000066400000000000000000000026451267530634000215360ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the callback "bridge" between Java and C++ for // rocksdb::Logger #ifndef JAVA_ROCKSJNI_LOGGERJNICALLBACK_H_ #define JAVA_ROCKSJNI_LOGGERJNICALLBACK_H_ #include #include #include "port/port.h" #include "rocksdb/env.h" namespace rocksdb { class LoggerJniCallback : public Logger { public: LoggerJniCallback(JNIEnv* env, jobject jLogger); virtual ~LoggerJniCallback(); using Logger::SetInfoLogLevel; using Logger::GetInfoLogLevel; // Write an entry to the log file with the specified format. virtual void Logv(const char* format, va_list ap); // Write an entry to the log file with the specified log level // and format. Any log with level under the internal log level // of *this (see @SetInfoLogLevel and @GetInfoLogLevel) will not be // printed. virtual void Logv(const InfoLogLevel log_level, const char* format, va_list ap); protected: JNIEnv* getJniEnv() const; private: JavaVM* m_jvm; jobject m_jLogger; jmethodID m_jLogMethodId; }; } // namespace rocksdb #endif // JAVA_ROCKSJNI_LOGGERJNICALLBACK_H_ rocksdb-4.5.1/java/rocksjni/memtablejni.cc000066400000000000000000000066511267530634000205270ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for MemTables. #include "rocksjni/portal.h" #include "include/org_rocksdb_HashSkipListMemTableConfig.h" #include "include/org_rocksdb_HashLinkedListMemTableConfig.h" #include "include/org_rocksdb_VectorMemTableConfig.h" #include "include/org_rocksdb_SkipListMemTableConfig.h" #include "rocksdb/memtablerep.h" /* * Class: org_rocksdb_HashSkipListMemTableConfig * Method: newMemTableFactoryHandle * Signature: (JII)J */ jlong Java_org_rocksdb_HashSkipListMemTableConfig_newMemTableFactoryHandle( JNIEnv* env, jobject jobj, jlong jbucket_count, jint jheight, jint jbranching_factor) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jbucket_count); if (s.ok()) { return reinterpret_cast(rocksdb::NewHashSkipListRepFactory( static_cast(jbucket_count), static_cast(jheight), static_cast(jbranching_factor))); } rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); return 0; } /* * Class: org_rocksdb_HashLinkedListMemTableConfig * Method: newMemTableFactoryHandle * Signature: (JJIZI)J */ jlong Java_org_rocksdb_HashLinkedListMemTableConfig_newMemTableFactoryHandle( JNIEnv* env, jobject jobj, jlong jbucket_count, jlong jhuge_page_tlb_size, jint jbucket_entries_logging_threshold, jboolean jif_log_bucket_dist_when_flash, jint jthreshold_use_skiplist) { rocksdb::Status statusBucketCount = rocksdb::check_if_jlong_fits_size_t(jbucket_count); rocksdb::Status statusHugePageTlb = rocksdb::check_if_jlong_fits_size_t(jhuge_page_tlb_size); if (statusBucketCount.ok() && statusHugePageTlb.ok()) { return reinterpret_cast(rocksdb::NewHashLinkListRepFactory( static_cast(jbucket_count), static_cast(jhuge_page_tlb_size), static_cast(jbucket_entries_logging_threshold), static_cast(jif_log_bucket_dist_when_flash), static_cast(jthreshold_use_skiplist))); } rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, !statusBucketCount.ok()?statusBucketCount:statusHugePageTlb); return 0; } /* * Class: org_rocksdb_VectorMemTableConfig * Method: newMemTableFactoryHandle * Signature: (J)J */ jlong Java_org_rocksdb_VectorMemTableConfig_newMemTableFactoryHandle( JNIEnv* env, jobject jobj, jlong jreserved_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jreserved_size); if (s.ok()) { return reinterpret_cast(new rocksdb::VectorRepFactory( static_cast(jreserved_size))); } rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); return 0; } /* * Class: org_rocksdb_SkipListMemTableConfig * Method: newMemTableFactoryHandle0 * Signature: (J)J */ jlong Java_org_rocksdb_SkipListMemTableConfig_newMemTableFactoryHandle0( JNIEnv* env, jobject jobj, jlong jlookahead) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jlookahead); if (s.ok()) { return reinterpret_cast(new rocksdb::SkipListFactory( static_cast(jlookahead))); } rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); return 0; } rocksdb-4.5.1/java/rocksjni/merge_operator.cc000066400000000000000000000023421267530634000212430ustar00rootroot00000000000000// Copyright (c) 2014, Vlad Balan (vlad.gm@gmail.com). All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ // for rocksdb::MergeOperator. #include #include #include #include #include #include "include/org_rocksdb_StringAppendOperator.h" #include "rocksjni/portal.h" #include "rocksdb/db.h" #include "rocksdb/options.h" #include "rocksdb/statistics.h" #include "rocksdb/memtablerep.h" #include "rocksdb/table.h" #include "rocksdb/slice_transform.h" #include "rocksdb/merge_operator.h" #include "utilities/merge_operators.h" /* * Class: org_rocksdb_StringAppendOperator * Method: newMergeOperatorHandle * Signature: ()J */ jlong Java_org_rocksdb_StringAppendOperator_newMergeOperatorHandleImpl (JNIEnv* env, jobject jobj) { std::shared_ptr *op = new std::shared_ptr(); *op = rocksdb::MergeOperators::CreateFromStringId("stringappend"); return reinterpret_cast(op); } rocksdb-4.5.1/java/rocksjni/options.cc000066400000000000000000003635511267530634000177400ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for rocksdb::Options. #include #include #include #include #include "include/org_rocksdb_Options.h" #include "include/org_rocksdb_DBOptions.h" #include "include/org_rocksdb_ColumnFamilyOptions.h" #include "include/org_rocksdb_WriteOptions.h" #include "include/org_rocksdb_ReadOptions.h" #include "include/org_rocksdb_ComparatorOptions.h" #include "include/org_rocksdb_FlushOptions.h" #include "rocksjni/comparatorjnicallback.h" #include "rocksjni/portal.h" #include "rocksdb/db.h" #include "rocksdb/options.h" #include "rocksdb/statistics.h" #include "rocksdb/memtablerep.h" #include "rocksdb/table.h" #include "rocksdb/slice_transform.h" #include "rocksdb/rate_limiter.h" #include "rocksdb/comparator.h" #include "rocksdb/convenience.h" #include "rocksdb/merge_operator.h" #include "utilities/merge_operators.h" /* * Class: org_rocksdb_Options * Method: newOptions * Signature: ()V */ void Java_org_rocksdb_Options_newOptions__(JNIEnv* env, jobject jobj) { rocksdb::Options* op = new rocksdb::Options(); rocksdb::OptionsJni::setHandle(env, jobj, op); } /* * Class: org_rocksdb_Options * Method: newOptions * Signature: (JJ)V */ void Java_org_rocksdb_Options_newOptions__JJ(JNIEnv* env, jobject jobj, jlong jdboptions, jlong jcfoptions) { auto dbOpt = reinterpret_cast(jdboptions); auto cfOpt = reinterpret_cast( jcfoptions); rocksdb::Options* op = new rocksdb::Options(*dbOpt, *cfOpt); rocksdb::OptionsJni::setHandle(env, jobj, op); } /* * Class: org_rocksdb_Options * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_Options_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } /* * Class: org_rocksdb_Options * Method: setIncreaseParallelism * Signature: (JI)V */ void Java_org_rocksdb_Options_setIncreaseParallelism( JNIEnv * evnv, jobject jobj, jlong jhandle, jint totalThreads) { reinterpret_cast (jhandle)->IncreaseParallelism(static_cast(totalThreads)); } /* * Class: org_rocksdb_Options * Method: setCreateIfMissing * Signature: (JZ)V */ void Java_org_rocksdb_Options_setCreateIfMissing( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { reinterpret_cast(jhandle)->create_if_missing = flag; } /* * Class: org_rocksdb_Options * Method: createIfMissing * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_createIfMissing( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->create_if_missing; } /* * Class: org_rocksdb_Options * Method: setCreateMissingColumnFamilies * Signature: (JZ)V */ void Java_org_rocksdb_Options_setCreateMissingColumnFamilies( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { reinterpret_cast (jhandle)->create_missing_column_families = flag; } /* * Class: org_rocksdb_Options * Method: createMissingColumnFamilies * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_createMissingColumnFamilies( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast (jhandle)->create_missing_column_families; } /* * Class: org_rocksdb_Options * Method: setComparatorHandle * Signature: (JI)V */ void Java_org_rocksdb_Options_setComparatorHandle__JI( JNIEnv* env, jobject jobj, jlong jhandle, jint builtinComparator) { switch (builtinComparator) { case 1: reinterpret_cast(jhandle)->comparator = rocksdb::ReverseBytewiseComparator(); break; default: reinterpret_cast(jhandle)->comparator = rocksdb::BytewiseComparator(); break; } } /* * Class: org_rocksdb_Options * Method: setComparatorHandle * Signature: (JJ)V */ void Java_org_rocksdb_Options_setComparatorHandle__JJ( JNIEnv* env, jobject jobj, jlong jopt_handle, jlong jcomparator_handle) { reinterpret_cast(jopt_handle)->comparator = reinterpret_cast(jcomparator_handle); } /* * Class: org_rocksdb_Options * Method: setMergeOperatorName * Signature: (JJjava/lang/String)V */ void Java_org_rocksdb_Options_setMergeOperatorName( JNIEnv* env, jobject jobj, jlong jhandle, jstring jop_name) { auto options = reinterpret_cast(jhandle); const char* op_name = env->GetStringUTFChars(jop_name, 0); options->merge_operator = rocksdb::MergeOperators::CreateFromStringId( op_name); env->ReleaseStringUTFChars(jop_name, op_name); } /* * Class: org_rocksdb_Options * Method: setMergeOperator * Signature: (JJjava/lang/String)V */ void Java_org_rocksdb_Options_setMergeOperator( JNIEnv* env, jobject jobj, jlong jhandle, jlong mergeOperatorHandle) { reinterpret_cast(jhandle)->merge_operator = *(reinterpret_cast*> (mergeOperatorHandle)); } /* * Class: org_rocksdb_Options * Method: setWriteBufferSize * Signature: (JJ)I */ void Java_org_rocksdb_Options_setWriteBufferSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_buffer_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jwrite_buffer_size); if (s.ok()) { reinterpret_cast(jhandle)->write_buffer_size = jwrite_buffer_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: writeBufferSize * Signature: (J)J */ jlong Java_org_rocksdb_Options_writeBufferSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->write_buffer_size; } /* * Class: org_rocksdb_Options * Method: setMaxWriteBufferNumber * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxWriteBufferNumber( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_write_buffer_number) { reinterpret_cast(jhandle)->max_write_buffer_number = jmax_write_buffer_number; } /* * Class: org_rocksdb_Options * Method: createStatistics * Signature: (J)V */ void Java_org_rocksdb_Options_createStatistics( JNIEnv* env, jobject jobj, jlong jOptHandle) { reinterpret_cast(jOptHandle)->statistics = rocksdb::CreateDBStatistics(); } /* * Class: org_rocksdb_Options * Method: statisticsPtr * Signature: (J)J */ jlong Java_org_rocksdb_Options_statisticsPtr( JNIEnv* env, jobject jobj, jlong jOptHandle) { auto st = reinterpret_cast(jOptHandle)->statistics.get(); return reinterpret_cast(st); } /* * Class: org_rocksdb_Options * Method: maxWriteBufferNumber * Signature: (J)I */ jint Java_org_rocksdb_Options_maxWriteBufferNumber( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_write_buffer_number; } /* * Class: org_rocksdb_Options * Method: errorIfExists * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_errorIfExists( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->error_if_exists; } /* * Class: org_rocksdb_Options * Method: setErrorIfExists * Signature: (JZ)V */ void Java_org_rocksdb_Options_setErrorIfExists( JNIEnv* env, jobject jobj, jlong jhandle, jboolean error_if_exists) { reinterpret_cast(jhandle)->error_if_exists = static_cast(error_if_exists); } /* * Class: org_rocksdb_Options * Method: paranoidChecks * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_paranoidChecks( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->paranoid_checks; } /* * Class: org_rocksdb_Options * Method: setParanoidChecks * Signature: (JZ)V */ void Java_org_rocksdb_Options_setParanoidChecks( JNIEnv* env, jobject jobj, jlong jhandle, jboolean paranoid_checks) { reinterpret_cast(jhandle)->paranoid_checks = static_cast(paranoid_checks); } /* * Class: org_rocksdb_Options * Method: setEnv * Signature: (JJ)V */ void Java_org_rocksdb_Options_setEnv( JNIEnv* env, jobject jobj, jlong jhandle, jlong jenv) { reinterpret_cast(jhandle)->env = reinterpret_cast(jenv); } /* * Class: org_rocksdb_Options * Method: setMaxTotalWalSize * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxTotalWalSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_total_wal_size) { reinterpret_cast(jhandle)->max_total_wal_size = static_cast(jmax_total_wal_size); } /* * Class: org_rocksdb_Options * Method: maxTotalWalSize * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxTotalWalSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> max_total_wal_size; } /* * Class: org_rocksdb_Options * Method: maxOpenFiles * Signature: (J)I */ jint Java_org_rocksdb_Options_maxOpenFiles( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_open_files; } /* * Class: org_rocksdb_Options * Method: setMaxOpenFiles * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxOpenFiles( JNIEnv* env, jobject jobj, jlong jhandle, jint max_open_files) { reinterpret_cast(jhandle)->max_open_files = static_cast(max_open_files); } /* * Class: org_rocksdb_Options * Method: disableDataSync * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_disableDataSync( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->disableDataSync; } /* * Class: org_rocksdb_Options * Method: setDisableDataSync * Signature: (JZ)V */ void Java_org_rocksdb_Options_setDisableDataSync( JNIEnv* env, jobject jobj, jlong jhandle, jboolean disableDataSync) { reinterpret_cast(jhandle)->disableDataSync = static_cast(disableDataSync); } /* * Class: org_rocksdb_Options * Method: useFsync * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_useFsync( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->use_fsync; } /* * Class: org_rocksdb_Options * Method: setUseFsync * Signature: (JZ)V */ void Java_org_rocksdb_Options_setUseFsync( JNIEnv* env, jobject jobj, jlong jhandle, jboolean use_fsync) { reinterpret_cast(jhandle)->use_fsync = static_cast(use_fsync); } /* * Class: org_rocksdb_Options * Method: dbLogDir * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_Options_dbLogDir( JNIEnv* env, jobject jobj, jlong jhandle) { return env->NewStringUTF( reinterpret_cast(jhandle)->db_log_dir.c_str()); } /* * Class: org_rocksdb_Options * Method: setDbLogDir * Signature: (JLjava/lang/String)V */ void Java_org_rocksdb_Options_setDbLogDir( JNIEnv* env, jobject jobj, jlong jhandle, jstring jdb_log_dir) { const char* log_dir = env->GetStringUTFChars(jdb_log_dir, 0); reinterpret_cast(jhandle)->db_log_dir.assign(log_dir); env->ReleaseStringUTFChars(jdb_log_dir, log_dir); } /* * Class: org_rocksdb_Options * Method: walDir * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_Options_walDir( JNIEnv* env, jobject jobj, jlong jhandle) { return env->NewStringUTF( reinterpret_cast(jhandle)->wal_dir.c_str()); } /* * Class: org_rocksdb_Options * Method: setWalDir * Signature: (JLjava/lang/String)V */ void Java_org_rocksdb_Options_setWalDir( JNIEnv* env, jobject jobj, jlong jhandle, jstring jwal_dir) { const char* wal_dir = env->GetStringUTFChars(jwal_dir, 0); reinterpret_cast(jhandle)->wal_dir.assign(wal_dir); env->ReleaseStringUTFChars(jwal_dir, wal_dir); } /* * Class: org_rocksdb_Options * Method: deleteObsoleteFilesPeriodMicros * Signature: (J)J */ jlong Java_org_rocksdb_Options_deleteObsoleteFilesPeriodMicros( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->delete_obsolete_files_period_micros; } /* * Class: org_rocksdb_Options * Method: setDeleteObsoleteFilesPeriodMicros * Signature: (JJ)V */ void Java_org_rocksdb_Options_setDeleteObsoleteFilesPeriodMicros( JNIEnv* env, jobject jobj, jlong jhandle, jlong micros) { reinterpret_cast(jhandle) ->delete_obsolete_files_period_micros = static_cast(micros); } /* * Class: org_rocksdb_Options * Method: maxBackgroundCompactions * Signature: (J)I */ jint Java_org_rocksdb_Options_maxBackgroundCompactions( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_background_compactions; } /* * Class: org_rocksdb_Options * Method: setMaxBackgroundCompactions * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxBackgroundCompactions( JNIEnv* env, jobject jobj, jlong jhandle, jint max) { reinterpret_cast(jhandle) ->max_background_compactions = static_cast(max); } /* * Class: org_rocksdb_Options * Method: maxBackgroundFlushes * Signature: (J)I */ jint Java_org_rocksdb_Options_maxBackgroundFlushes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_background_flushes; } /* * Class: org_rocksdb_Options * Method: setMaxBackgroundFlushes * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxBackgroundFlushes( JNIEnv* env, jobject jobj, jlong jhandle, jint max_background_flushes) { reinterpret_cast(jhandle)->max_background_flushes = static_cast(max_background_flushes); } /* * Class: org_rocksdb_Options * Method: maxLogFileSize * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxLogFileSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_log_file_size; } /* * Class: org_rocksdb_Options * Method: setMaxLogFileSize * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxLogFileSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong max_log_file_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(max_log_file_size); if (s.ok()) { reinterpret_cast(jhandle)->max_log_file_size = max_log_file_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: logFileTimeToRoll * Signature: (J)J */ jlong Java_org_rocksdb_Options_logFileTimeToRoll( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->log_file_time_to_roll; } /* * Class: org_rocksdb_Options * Method: setLogFileTimeToRoll * Signature: (JJ)V */ void Java_org_rocksdb_Options_setLogFileTimeToRoll( JNIEnv* env, jobject jobj, jlong jhandle, jlong log_file_time_to_roll) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t( log_file_time_to_roll); if (s.ok()) { reinterpret_cast(jhandle)->log_file_time_to_roll = log_file_time_to_roll; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: keepLogFileNum * Signature: (J)J */ jlong Java_org_rocksdb_Options_keepLogFileNum( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->keep_log_file_num; } /* * Class: org_rocksdb_Options * Method: setKeepLogFileNum * Signature: (JJ)V */ void Java_org_rocksdb_Options_setKeepLogFileNum( JNIEnv* env, jobject jobj, jlong jhandle, jlong keep_log_file_num) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(keep_log_file_num); if (s.ok()) { reinterpret_cast(jhandle)->keep_log_file_num = keep_log_file_num; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: recycleLogFiles * Signature: (J)J */ jlong Java_org_rocksdb_Options_recycleLogFileNum(JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->recycle_log_file_num; } /* * Class: org_rocksdb_Options * Method: setRecycleLogFiles * Signature: (JJ)V */ void Java_org_rocksdb_Options_setRecycleLogFiles(JNIEnv* env, jobject jobj, jlong jhandle, jlong recycle_log_file_num) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(recycle_log_file_num); if (s.ok()) { reinterpret_cast(jhandle)->recycle_log_file_num = recycle_log_file_num; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: maxManifestFileSize * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxManifestFileSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_manifest_file_size; } /* * Method: memTableFactoryName * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_Options_memTableFactoryName( JNIEnv* env, jobject jobj, jlong jhandle) { auto opt = reinterpret_cast(jhandle); rocksdb::MemTableRepFactory* tf = opt->memtable_factory.get(); // Should never be nullptr. // Default memtable factory is SkipListFactory assert(tf); // temporarly fix for the historical typo if (strcmp(tf->Name(), "HashLinkListRepFactory") == 0) { return env->NewStringUTF("HashLinkedListRepFactory"); } return env->NewStringUTF(tf->Name()); } /* * Class: org_rocksdb_Options * Method: setMaxManifestFileSize * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxManifestFileSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong max_manifest_file_size) { reinterpret_cast(jhandle)->max_manifest_file_size = static_cast(max_manifest_file_size); } /* * Method: setMemTableFactory * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMemTableFactory( JNIEnv* env, jobject jobj, jlong jhandle, jlong jfactory_handle) { reinterpret_cast(jhandle)->memtable_factory.reset( reinterpret_cast(jfactory_handle)); } /* * Class: org_rocksdb_Options * Method: setRateLimiter * Signature: (JJ)V */ void Java_org_rocksdb_Options_setRateLimiter( JNIEnv* env, jobject jobj, jlong jhandle, jlong jrate_limiter_handle) { reinterpret_cast(jhandle)->rate_limiter.reset( reinterpret_cast(jrate_limiter_handle)); } /* * Class: org_rocksdb_Options * Method: setLogger * Signature: (JJ)V */ void Java_org_rocksdb_Options_setLogger( JNIEnv* env, jobject jobj, jlong jhandle, jlong jlogger_handle) { std::shared_ptr *pLogger = reinterpret_cast *>( jlogger_handle); reinterpret_cast(jhandle)->info_log = *pLogger; } /* * Class: org_rocksdb_Options * Method: setInfoLogLevel * Signature: (JB)V */ void Java_org_rocksdb_Options_setInfoLogLevel( JNIEnv* env, jobject jobj, jlong jhandle, jbyte jlog_level) { reinterpret_cast(jhandle)->info_log_level = static_cast(jlog_level); } /* * Class: org_rocksdb_Options * Method: infoLogLevel * Signature: (J)B */ jbyte Java_org_rocksdb_Options_infoLogLevel( JNIEnv* env, jobject jobj, jlong jhandle) { return static_cast( reinterpret_cast(jhandle)->info_log_level); } /* * Class: org_rocksdb_Options * Method: tableCacheNumshardbits * Signature: (J)I */ jint Java_org_rocksdb_Options_tableCacheNumshardbits( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->table_cache_numshardbits; } /* * Class: org_rocksdb_Options * Method: setTableCacheNumshardbits * Signature: (JI)V */ void Java_org_rocksdb_Options_setTableCacheNumshardbits( JNIEnv* env, jobject jobj, jlong jhandle, jint table_cache_numshardbits) { reinterpret_cast(jhandle)->table_cache_numshardbits = static_cast(table_cache_numshardbits); } /* * Method: useFixedLengthPrefixExtractor * Signature: (JI)V */ void Java_org_rocksdb_Options_useFixedLengthPrefixExtractor( JNIEnv* env, jobject jobj, jlong jhandle, jint jprefix_length) { reinterpret_cast(jhandle)->prefix_extractor.reset( rocksdb::NewFixedPrefixTransform( static_cast(jprefix_length))); } /* * Method: useCappedPrefixExtractor * Signature: (JI)V */ void Java_org_rocksdb_Options_useCappedPrefixExtractor( JNIEnv* env, jobject jobj, jlong jhandle, jint jprefix_length) { reinterpret_cast(jhandle)->prefix_extractor.reset( rocksdb::NewCappedPrefixTransform( static_cast(jprefix_length))); } /* * Class: org_rocksdb_Options * Method: walTtlSeconds * Signature: (J)J */ jlong Java_org_rocksdb_Options_walTtlSeconds( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->WAL_ttl_seconds; } /* * Class: org_rocksdb_Options * Method: setWalTtlSeconds * Signature: (JJ)V */ void Java_org_rocksdb_Options_setWalTtlSeconds( JNIEnv* env, jobject jobj, jlong jhandle, jlong WAL_ttl_seconds) { reinterpret_cast(jhandle)->WAL_ttl_seconds = static_cast(WAL_ttl_seconds); } /* * Class: org_rocksdb_Options * Method: walTtlSeconds * Signature: (J)J */ jlong Java_org_rocksdb_Options_walSizeLimitMB( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->WAL_size_limit_MB; } /* * Class: org_rocksdb_Options * Method: setWalSizeLimitMB * Signature: (JJ)V */ void Java_org_rocksdb_Options_setWalSizeLimitMB( JNIEnv* env, jobject jobj, jlong jhandle, jlong WAL_size_limit_MB) { reinterpret_cast(jhandle)->WAL_size_limit_MB = static_cast(WAL_size_limit_MB); } /* * Class: org_rocksdb_Options * Method: manifestPreallocationSize * Signature: (J)J */ jlong Java_org_rocksdb_Options_manifestPreallocationSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->manifest_preallocation_size; } /* * Class: org_rocksdb_Options * Method: setManifestPreallocationSize * Signature: (JJ)V */ void Java_org_rocksdb_Options_setManifestPreallocationSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong preallocation_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(preallocation_size); if (s.ok()) { reinterpret_cast(jhandle)->manifest_preallocation_size = preallocation_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: allowOsBuffer * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_allowOsBuffer( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->allow_os_buffer; } /* * Class: org_rocksdb_Options * Method: setAllowOsBuffer * Signature: (JZ)V */ void Java_org_rocksdb_Options_setAllowOsBuffer( JNIEnv* env, jobject jobj, jlong jhandle, jboolean allow_os_buffer) { reinterpret_cast(jhandle)->allow_os_buffer = static_cast(allow_os_buffer); } /* * Method: setTableFactory * Signature: (JJ)V */ void Java_org_rocksdb_Options_setTableFactory( JNIEnv* env, jobject jobj, jlong jhandle, jlong jfactory_handle) { reinterpret_cast(jhandle)->table_factory.reset( reinterpret_cast(jfactory_handle)); } /* * Class: org_rocksdb_Options * Method: allowMmapReads * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_allowMmapReads( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->allow_mmap_reads; } /* * Class: org_rocksdb_Options * Method: setAllowMmapReads * Signature: (JZ)V */ void Java_org_rocksdb_Options_setAllowMmapReads( JNIEnv* env, jobject jobj, jlong jhandle, jboolean allow_mmap_reads) { reinterpret_cast(jhandle)->allow_mmap_reads = static_cast(allow_mmap_reads); } /* * Class: org_rocksdb_Options * Method: allowMmapWrites * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_allowMmapWrites( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->allow_mmap_writes; } /* * Class: org_rocksdb_Options * Method: setAllowMmapWrites * Signature: (JZ)V */ void Java_org_rocksdb_Options_setAllowMmapWrites( JNIEnv* env, jobject jobj, jlong jhandle, jboolean allow_mmap_writes) { reinterpret_cast(jhandle)->allow_mmap_writes = static_cast(allow_mmap_writes); } /* * Class: org_rocksdb_Options * Method: isFdCloseOnExec * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_isFdCloseOnExec( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->is_fd_close_on_exec; } /* * Class: org_rocksdb_Options * Method: setIsFdCloseOnExec * Signature: (JZ)V */ void Java_org_rocksdb_Options_setIsFdCloseOnExec( JNIEnv* env, jobject jobj, jlong jhandle, jboolean is_fd_close_on_exec) { reinterpret_cast(jhandle)->is_fd_close_on_exec = static_cast(is_fd_close_on_exec); } /* * Class: org_rocksdb_Options * Method: statsDumpPeriodSec * Signature: (J)I */ jint Java_org_rocksdb_Options_statsDumpPeriodSec( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->stats_dump_period_sec; } /* * Class: org_rocksdb_Options * Method: setStatsDumpPeriodSec * Signature: (JI)V */ void Java_org_rocksdb_Options_setStatsDumpPeriodSec( JNIEnv* env, jobject jobj, jlong jhandle, jint stats_dump_period_sec) { reinterpret_cast(jhandle)->stats_dump_period_sec = static_cast(stats_dump_period_sec); } /* * Class: org_rocksdb_Options * Method: adviseRandomOnOpen * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_adviseRandomOnOpen( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->advise_random_on_open; } /* * Class: org_rocksdb_Options * Method: setAdviseRandomOnOpen * Signature: (JZ)V */ void Java_org_rocksdb_Options_setAdviseRandomOnOpen( JNIEnv* env, jobject jobj, jlong jhandle, jboolean advise_random_on_open) { reinterpret_cast(jhandle)->advise_random_on_open = static_cast(advise_random_on_open); } /* * Class: org_rocksdb_Options * Method: useAdaptiveMutex * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_useAdaptiveMutex( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->use_adaptive_mutex; } /* * Class: org_rocksdb_Options * Method: setUseAdaptiveMutex * Signature: (JZ)V */ void Java_org_rocksdb_Options_setUseAdaptiveMutex( JNIEnv* env, jobject jobj, jlong jhandle, jboolean use_adaptive_mutex) { reinterpret_cast(jhandle)->use_adaptive_mutex = static_cast(use_adaptive_mutex); } /* * Class: org_rocksdb_Options * Method: bytesPerSync * Signature: (J)J */ jlong Java_org_rocksdb_Options_bytesPerSync( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->bytes_per_sync; } /* * Class: org_rocksdb_Options * Method: setBytesPerSync * Signature: (JJ)V */ void Java_org_rocksdb_Options_setBytesPerSync( JNIEnv* env, jobject jobj, jlong jhandle, jlong bytes_per_sync) { reinterpret_cast(jhandle)->bytes_per_sync = static_cast(bytes_per_sync); } /* * Method: tableFactoryName * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_Options_tableFactoryName( JNIEnv* env, jobject jobj, jlong jhandle) { auto opt = reinterpret_cast(jhandle); rocksdb::TableFactory* tf = opt->table_factory.get(); // Should never be nullptr. // Default memtable factory is SkipListFactory assert(tf); return env->NewStringUTF(tf->Name()); } /* * Class: org_rocksdb_Options * Method: minWriteBufferNumberToMerge * Signature: (J)I */ jint Java_org_rocksdb_Options_minWriteBufferNumberToMerge( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->min_write_buffer_number_to_merge; } /* * Class: org_rocksdb_Options * Method: setMinWriteBufferNumberToMerge * Signature: (JI)V */ void Java_org_rocksdb_Options_setMinWriteBufferNumberToMerge( JNIEnv* env, jobject jobj, jlong jhandle, jint jmin_write_buffer_number_to_merge) { reinterpret_cast( jhandle)->min_write_buffer_number_to_merge = static_cast(jmin_write_buffer_number_to_merge); } /* * Class: org_rocksdb_Options * Method: maxWriteBufferNumberToMaintain * Signature: (J)I */ jint Java_org_rocksdb_Options_maxWriteBufferNumberToMaintain(JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->max_write_buffer_number_to_maintain; } /* * Class: org_rocksdb_Options * Method: setMaxWriteBufferNumberToMaintain * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxWriteBufferNumberToMaintain( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_write_buffer_number_to_maintain) { reinterpret_cast(jhandle) ->max_write_buffer_number_to_maintain = static_cast(jmax_write_buffer_number_to_maintain); } /* * Class: org_rocksdb_Options * Method: setCompressionType * Signature: (JB)V */ void Java_org_rocksdb_Options_setCompressionType( JNIEnv* env, jobject jobj, jlong jhandle, jbyte compression) { reinterpret_cast(jhandle)->compression = static_cast(compression); } /* * Class: org_rocksdb_Options * Method: compressionType * Signature: (J)B */ jbyte Java_org_rocksdb_Options_compressionType( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->compression; } /* * Helper method to convert a Java list to a CompressionType * vector. */ std::vector rocksdb_compression_vector_helper( JNIEnv* env, jobject jcompressionLevels) { std::vector compressionLevels; // iterate over compressionLevels jobject iteratorObj = env->CallObjectMethod( jcompressionLevels, rocksdb::ListJni::getIteratorMethod(env)); while (env->CallBooleanMethod( iteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { // get compression jobject jcompression_obj = env->CallObjectMethod(iteratorObj, rocksdb::ListJni::getNextMethod(env)); jbyte jcompression = env->CallByteMethod(jcompression_obj, rocksdb::ByteJni::getByteValueMethod(env)); compressionLevels.push_back(static_cast( jcompression)); } return compressionLevels; } /* * Helper method to convert a CompressionType vector to a Java * List. */ jobject rocksdb_compression_list_helper(JNIEnv* env, std::vector compressionLevels) { jclass jListClazz = env->FindClass("java/util/ArrayList"); jmethodID midList = rocksdb::ListJni::getArrayListConstructorMethodId( env, jListClazz); jobject jcompressionLevels = env->NewObject(jListClazz, midList, compressionLevels.size()); // insert in java list for (std::vector::size_type i = 0; i != compressionLevels.size(); i++) { jclass jByteClazz = env->FindClass("java/lang/Byte"); jmethodID midByte = env->GetMethodID(jByteClazz, "", "(B)V"); jobject obj = env->NewObject(jByteClazz, midByte, compressionLevels[i]); env->CallBooleanMethod(jcompressionLevels, rocksdb::ListJni::getListAddMethodId(env), obj); } return jcompressionLevels; } /* * Class: org_rocksdb_Options * Method: setCompressionPerLevel * Signature: (JLjava/util/List;)V */ void Java_org_rocksdb_Options_setCompressionPerLevel( JNIEnv* env, jobject jobj, jlong jhandle, jobject jcompressionLevels) { auto* options = reinterpret_cast(jhandle); std::vector compressionLevels = rocksdb_compression_vector_helper(env, jcompressionLevels); options->compression_per_level = compressionLevels; } /* * Class: org_rocksdb_Options * Method: compressionPerLevel * Signature: (J)Ljava/util/List; */ jobject Java_org_rocksdb_Options_compressionPerLevel( JNIEnv* env, jobject jobj, jlong jhandle) { auto* options = reinterpret_cast(jhandle); return rocksdb_compression_list_helper(env, options->compression_per_level); } /* * Class: org_rocksdb_Options * Method: setCompactionStyle * Signature: (JB)V */ void Java_org_rocksdb_Options_setCompactionStyle( JNIEnv* env, jobject jobj, jlong jhandle, jbyte compaction_style) { reinterpret_cast(jhandle)->compaction_style = static_cast(compaction_style); } /* * Class: org_rocksdb_Options * Method: compactionStyle * Signature: (J)B */ jbyte Java_org_rocksdb_Options_compactionStyle( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->compaction_style; } /* * Class: org_rocksdb_Options * Method: setMaxTableFilesSizeFIFO * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxTableFilesSizeFIFO( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_table_files_size) { reinterpret_cast(jhandle)->compaction_options_fifo.max_table_files_size = static_cast(jmax_table_files_size); } /* * Class: org_rocksdb_Options * Method: maxTableFilesSizeFIFO * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxTableFilesSizeFIFO( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->compaction_options_fifo.max_table_files_size; } /* * Class: org_rocksdb_Options * Method: numLevels * Signature: (J)I */ jint Java_org_rocksdb_Options_numLevels( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->num_levels; } /* * Class: org_rocksdb_Options * Method: setNumLevels * Signature: (JI)V */ void Java_org_rocksdb_Options_setNumLevels( JNIEnv* env, jobject jobj, jlong jhandle, jint jnum_levels) { reinterpret_cast(jhandle)->num_levels = static_cast(jnum_levels); } /* * Class: org_rocksdb_Options * Method: levelZeroFileNumCompactionTrigger * Signature: (J)I */ jint Java_org_rocksdb_Options_levelZeroFileNumCompactionTrigger( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level0_file_num_compaction_trigger; } /* * Class: org_rocksdb_Options * Method: setLevelZeroFileNumCompactionTrigger * Signature: (JI)V */ void Java_org_rocksdb_Options_setLevelZeroFileNumCompactionTrigger( JNIEnv* env, jobject jobj, jlong jhandle, jint jlevel0_file_num_compaction_trigger) { reinterpret_cast( jhandle)->level0_file_num_compaction_trigger = static_cast(jlevel0_file_num_compaction_trigger); } /* * Class: org_rocksdb_Options * Method: levelZeroSlowdownWritesTrigger * Signature: (J)I */ jint Java_org_rocksdb_Options_levelZeroSlowdownWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level0_slowdown_writes_trigger; } /* * Class: org_rocksdb_Options * Method: setLevelSlowdownWritesTrigger * Signature: (JI)V */ void Java_org_rocksdb_Options_setLevelZeroSlowdownWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle, jint jlevel0_slowdown_writes_trigger) { reinterpret_cast( jhandle)->level0_slowdown_writes_trigger = static_cast(jlevel0_slowdown_writes_trigger); } /* * Class: org_rocksdb_Options * Method: levelZeroStopWritesTrigger * Signature: (J)I */ jint Java_org_rocksdb_Options_levelZeroStopWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level0_stop_writes_trigger; } /* * Class: org_rocksdb_Options * Method: setLevelStopWritesTrigger * Signature: (JI)V */ void Java_org_rocksdb_Options_setLevelZeroStopWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle, jint jlevel0_stop_writes_trigger) { reinterpret_cast(jhandle)->level0_stop_writes_trigger = static_cast(jlevel0_stop_writes_trigger); } /* * Class: org_rocksdb_Options * Method: targetFileSizeBase * Signature: (J)J */ jlong Java_org_rocksdb_Options_targetFileSizeBase( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->target_file_size_base; } /* * Class: org_rocksdb_Options * Method: setTargetFileSizeBase * Signature: (JJ)V */ void Java_org_rocksdb_Options_setTargetFileSizeBase( JNIEnv* env, jobject jobj, jlong jhandle, jlong jtarget_file_size_base) { reinterpret_cast(jhandle)->target_file_size_base = static_cast(jtarget_file_size_base); } /* * Class: org_rocksdb_Options * Method: targetFileSizeMultiplier * Signature: (J)I */ jint Java_org_rocksdb_Options_targetFileSizeMultiplier( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->target_file_size_multiplier; } /* * Class: org_rocksdb_Options * Method: setTargetFileSizeMultiplier * Signature: (JI)V */ void Java_org_rocksdb_Options_setTargetFileSizeMultiplier( JNIEnv* env, jobject jobj, jlong jhandle, jint jtarget_file_size_multiplier) { reinterpret_cast( jhandle)->target_file_size_multiplier = static_cast(jtarget_file_size_multiplier); } /* * Class: org_rocksdb_Options * Method: maxBytesForLevelBase * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxBytesForLevelBase( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_bytes_for_level_base; } /* * Class: org_rocksdb_Options * Method: setMaxBytesForLevelBase * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxBytesForLevelBase( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_bytes_for_level_base) { reinterpret_cast( jhandle)->max_bytes_for_level_base = static_cast(jmax_bytes_for_level_base); } /* * Class: org_rocksdb_Options * Method: levelCompactionDynamicLevelBytes * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_levelCompactionDynamicLevelBytes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level_compaction_dynamic_level_bytes; } /* * Class: org_rocksdb_Options * Method: setLevelCompactionDynamicLevelBytes * Signature: (JZ)V */ void Java_org_rocksdb_Options_setLevelCompactionDynamicLevelBytes( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jenable_dynamic_level_bytes) { reinterpret_cast( jhandle)->level_compaction_dynamic_level_bytes = (jenable_dynamic_level_bytes); } /* * Class: org_rocksdb_Options * Method: maxBytesForLevelMultiplier * Signature: (J)I */ jint Java_org_rocksdb_Options_maxBytesForLevelMultiplier( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_bytes_for_level_multiplier; } /* * Class: org_rocksdb_Options * Method: setMaxBytesForLevelMultiplier * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxBytesForLevelMultiplier( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_bytes_for_level_multiplier) { reinterpret_cast( jhandle)->max_bytes_for_level_multiplier = static_cast(jmax_bytes_for_level_multiplier); } /* * Class: org_rocksdb_Options * Method: expandedCompactionFactor * Signature: (J)I */ jint Java_org_rocksdb_Options_expandedCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->expanded_compaction_factor; } /* * Class: org_rocksdb_Options * Method: setExpandedCompactionFactor * Signature: (JI)V */ void Java_org_rocksdb_Options_setExpandedCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle, jint jexpanded_compaction_factor) { reinterpret_cast( jhandle)->expanded_compaction_factor = static_cast(jexpanded_compaction_factor); } /* * Class: org_rocksdb_Options * Method: sourceCompactionFactor * Signature: (J)I */ jint Java_org_rocksdb_Options_sourceCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->source_compaction_factor; } /* * Class: org_rocksdb_Options * Method: setSourceCompactionFactor * Signature: (JI)V */ void Java_org_rocksdb_Options_setSourceCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle, jint jsource_compaction_factor) { reinterpret_cast( jhandle)->source_compaction_factor = static_cast(jsource_compaction_factor); } /* * Class: org_rocksdb_Options * Method: maxGrandparentOverlapFactor * Signature: (J)I */ jint Java_org_rocksdb_Options_maxGrandparentOverlapFactor( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_grandparent_overlap_factor; } /* * Class: org_rocksdb_Options * Method: setMaxGrandparentOverlapFactor * Signature: (JI)V */ void Java_org_rocksdb_Options_setMaxGrandparentOverlapFactor( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_grandparent_overlap_factor) { reinterpret_cast( jhandle)->max_grandparent_overlap_factor = static_cast(jmax_grandparent_overlap_factor); } /* * Class: org_rocksdb_Options * Method: softRateLimit * Signature: (J)D */ jdouble Java_org_rocksdb_Options_softRateLimit( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->soft_rate_limit; } /* * Class: org_rocksdb_Options * Method: setSoftRateLimit * Signature: (JD)V */ void Java_org_rocksdb_Options_setSoftRateLimit( JNIEnv* env, jobject jobj, jlong jhandle, jdouble jsoft_rate_limit) { reinterpret_cast(jhandle)->soft_rate_limit = static_cast(jsoft_rate_limit); } /* * Class: org_rocksdb_Options * Method: hardRateLimit * Signature: (J)D */ jdouble Java_org_rocksdb_Options_hardRateLimit( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->hard_rate_limit; } /* * Class: org_rocksdb_Options * Method: setHardRateLimit * Signature: (JD)V */ void Java_org_rocksdb_Options_setHardRateLimit( JNIEnv* env, jobject jobj, jlong jhandle, jdouble jhard_rate_limit) { reinterpret_cast(jhandle)->hard_rate_limit = static_cast(jhard_rate_limit); } /* * Class: org_rocksdb_Options * Method: rateLimitDelayMaxMilliseconds * Signature: (J)I */ jint Java_org_rocksdb_Options_rateLimitDelayMaxMilliseconds( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->rate_limit_delay_max_milliseconds; } /* * Class: org_rocksdb_Options * Method: setRateLimitDelayMaxMilliseconds * Signature: (JI)V */ void Java_org_rocksdb_Options_setRateLimitDelayMaxMilliseconds( JNIEnv* env, jobject jobj, jlong jhandle, jint jrate_limit_delay_max_milliseconds) { reinterpret_cast( jhandle)->rate_limit_delay_max_milliseconds = static_cast(jrate_limit_delay_max_milliseconds); } /* * Class: org_rocksdb_Options * Method: arenaBlockSize * Signature: (J)J */ jlong Java_org_rocksdb_Options_arenaBlockSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->arena_block_size; } /* * Class: org_rocksdb_Options * Method: setArenaBlockSize * Signature: (JJ)V */ void Java_org_rocksdb_Options_setArenaBlockSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong jarena_block_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jarena_block_size); if (s.ok()) { reinterpret_cast(jhandle)->arena_block_size = jarena_block_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: disableAutoCompactions * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_disableAutoCompactions( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->disable_auto_compactions; } /* * Class: org_rocksdb_Options * Method: setDisableAutoCompactions * Signature: (JZ)V */ void Java_org_rocksdb_Options_setDisableAutoCompactions( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jdisable_auto_compactions) { reinterpret_cast( jhandle)->disable_auto_compactions = static_cast(jdisable_auto_compactions); } /* * Class: org_rocksdb_Options * Method: purgeRedundantKvsWhileFlush * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_purgeRedundantKvsWhileFlush( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->purge_redundant_kvs_while_flush; } /* * Class: org_rocksdb_Options * Method: setPurgeRedundantKvsWhileFlush * Signature: (JZ)V */ void Java_org_rocksdb_Options_setPurgeRedundantKvsWhileFlush( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jpurge_redundant_kvs_while_flush) { reinterpret_cast( jhandle)->purge_redundant_kvs_while_flush = static_cast(jpurge_redundant_kvs_while_flush); } /* * Class: org_rocksdb_Options * Method: verifyChecksumsInCompaction * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_verifyChecksumsInCompaction( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->verify_checksums_in_compaction; } /* * Class: org_rocksdb_Options * Method: setVerifyChecksumsInCompaction * Signature: (JZ)V */ void Java_org_rocksdb_Options_setVerifyChecksumsInCompaction( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jverify_checksums_in_compaction) { reinterpret_cast( jhandle)->verify_checksums_in_compaction = static_cast(jverify_checksums_in_compaction); } /* * Class: org_rocksdb_Options * Method: filterDeletes * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_filterDeletes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->filter_deletes; } /* * Class: org_rocksdb_Options * Method: setFilterDeletes * Signature: (JZ)V */ void Java_org_rocksdb_Options_setFilterDeletes( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jfilter_deletes) { reinterpret_cast(jhandle)->filter_deletes = static_cast(jfilter_deletes); } /* * Class: org_rocksdb_Options * Method: maxSequentialSkipInIterations * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxSequentialSkipInIterations( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_sequential_skip_in_iterations; } /* * Class: org_rocksdb_Options * Method: setMaxSequentialSkipInIterations * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxSequentialSkipInIterations( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_sequential_skip_in_iterations) { reinterpret_cast( jhandle)->max_sequential_skip_in_iterations = static_cast(jmax_sequential_skip_in_iterations); } /* * Class: org_rocksdb_Options * Method: inplaceUpdateSupport * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_inplaceUpdateSupport( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->inplace_update_support; } /* * Class: org_rocksdb_Options * Method: setInplaceUpdateSupport * Signature: (JZ)V */ void Java_org_rocksdb_Options_setInplaceUpdateSupport( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jinplace_update_support) { reinterpret_cast( jhandle)->inplace_update_support = static_cast(jinplace_update_support); } /* * Class: org_rocksdb_Options * Method: inplaceUpdateNumLocks * Signature: (J)J */ jlong Java_org_rocksdb_Options_inplaceUpdateNumLocks( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->inplace_update_num_locks; } /* * Class: org_rocksdb_Options * Method: setInplaceUpdateNumLocks * Signature: (JJ)V */ void Java_org_rocksdb_Options_setInplaceUpdateNumLocks( JNIEnv* env, jobject jobj, jlong jhandle, jlong jinplace_update_num_locks) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t( jinplace_update_num_locks); if (s.ok()) { reinterpret_cast(jhandle)->inplace_update_num_locks = jinplace_update_num_locks; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: memtablePrefixBloomBits * Signature: (J)I */ jint Java_org_rocksdb_Options_memtablePrefixBloomBits( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->memtable_prefix_bloom_bits; } /* * Class: org_rocksdb_Options * Method: setMemtablePrefixBloomBits * Signature: (JI)V */ void Java_org_rocksdb_Options_setMemtablePrefixBloomBits( JNIEnv* env, jobject jobj, jlong jhandle, jint jmemtable_prefix_bloom_bits) { reinterpret_cast( jhandle)->memtable_prefix_bloom_bits = static_cast(jmemtable_prefix_bloom_bits); } /* * Class: org_rocksdb_Options * Method: memtablePrefixBloomProbes * Signature: (J)I */ jint Java_org_rocksdb_Options_memtablePrefixBloomProbes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->memtable_prefix_bloom_probes; } /* * Class: org_rocksdb_Options * Method: setMemtablePrefixBloomProbes * Signature: (JI)V */ void Java_org_rocksdb_Options_setMemtablePrefixBloomProbes( JNIEnv* env, jobject jobj, jlong jhandle, jint jmemtable_prefix_bloom_probes) { reinterpret_cast( jhandle)->memtable_prefix_bloom_probes = static_cast(jmemtable_prefix_bloom_probes); } /* * Class: org_rocksdb_Options * Method: bloomLocality * Signature: (J)I */ jint Java_org_rocksdb_Options_bloomLocality( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->bloom_locality; } /* * Class: org_rocksdb_Options * Method: setBloomLocality * Signature: (JI)V */ void Java_org_rocksdb_Options_setBloomLocality( JNIEnv* env, jobject jobj, jlong jhandle, jint jbloom_locality) { reinterpret_cast(jhandle)->bloom_locality = static_cast(jbloom_locality); } /* * Class: org_rocksdb_Options * Method: maxSuccessiveMerges * Signature: (J)J */ jlong Java_org_rocksdb_Options_maxSuccessiveMerges( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_successive_merges; } /* * Class: org_rocksdb_Options * Method: setMaxSuccessiveMerges * Signature: (JJ)V */ void Java_org_rocksdb_Options_setMaxSuccessiveMerges( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_successive_merges) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t( jmax_successive_merges); if (s.ok()) { reinterpret_cast(jhandle)->max_successive_merges = jmax_successive_merges; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_Options * Method: minPartialMergeOperands * Signature: (J)I */ jint Java_org_rocksdb_Options_minPartialMergeOperands( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->min_partial_merge_operands; } /* * Class: org_rocksdb_Options * Method: setMinPartialMergeOperands * Signature: (JI)V */ void Java_org_rocksdb_Options_setMinPartialMergeOperands( JNIEnv* env, jobject jobj, jlong jhandle, jint jmin_partial_merge_operands) { reinterpret_cast( jhandle)->min_partial_merge_operands = static_cast(jmin_partial_merge_operands); } /* * Class: org_rocksdb_Options * Method: optimizeFiltersForHits * Signature: (J)Z */ jboolean Java_org_rocksdb_Options_optimizeFiltersForHits( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->optimize_filters_for_hits; } /* * Class: org_rocksdb_Options * Method: setOptimizeFiltersForHits * Signature: (JZ)V */ void Java_org_rocksdb_Options_setOptimizeFiltersForHits( JNIEnv* env, jobject jobj, jlong jhandle, jboolean joptimize_filters_for_hits) { reinterpret_cast( jhandle)->optimize_filters_for_hits = static_cast(joptimize_filters_for_hits); } /* * Method: optimizeForPointLookup * Signature: (JJ)V */ void Java_org_rocksdb_Options_optimizeForPointLookup( JNIEnv* env, jobject jobj, jlong jhandle, jlong block_cache_size_mb) { reinterpret_cast(jhandle)-> OptimizeForPointLookup(block_cache_size_mb); } /* * Method: optimizeLevelStyleCompaction * Signature: (JJ)V */ void Java_org_rocksdb_Options_optimizeLevelStyleCompaction( JNIEnv* env, jobject jobj, jlong jhandle, jlong memtable_memory_budget) { reinterpret_cast(jhandle)-> OptimizeLevelStyleCompaction(memtable_memory_budget); } /* * Class: org_rocksdb_Options * Method: optimizeUniversalStyleCompaction * Signature: (JJ)V */ void Java_org_rocksdb_Options_optimizeUniversalStyleCompaction( JNIEnv* env, jobject jobj, jlong jhandle, jlong memtable_memory_budget) { reinterpret_cast(jhandle)-> OptimizeUniversalStyleCompaction(memtable_memory_budget); } /* * Class: org_rocksdb_Options * Method: prepareForBulkLoad * Signature: (J)V */ void Java_org_rocksdb_Options_prepareForBulkLoad( JNIEnv* env, jobject jobj, jlong jhandle) { reinterpret_cast(jhandle)-> PrepareForBulkLoad(); } ////////////////////////////////////////////////////////////////////////////// // rocksdb::ColumnFamilyOptions /* * Class: org_rocksdb_ColumnFamilyOptions * Method: newColumnFamilyOptions * Signature: ()V */ void Java_org_rocksdb_ColumnFamilyOptions_newColumnFamilyOptions( JNIEnv* env, jobject jobj) { rocksdb::ColumnFamilyOptions* op = new rocksdb::ColumnFamilyOptions(); rocksdb::ColumnFamilyOptionsJni::setHandle(env, jobj, op); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: getColumnFamilyOptionsFromProps * Signature: (Ljava/util/String;)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_getColumnFamilyOptionsFromProps( JNIEnv* env, jclass jclazz, jstring jopt_string) { jlong ret_value = 0; rocksdb::ColumnFamilyOptions* cf_options = new rocksdb::ColumnFamilyOptions(); const char* opt_string = env->GetStringUTFChars(jopt_string, 0); rocksdb::Status status = rocksdb::GetColumnFamilyOptionsFromString( rocksdb::ColumnFamilyOptions(), opt_string, cf_options); env->ReleaseStringUTFChars(jopt_string, opt_string); // Check if ColumnFamilyOptions creation was possible. if (status.ok()) { ret_value = reinterpret_cast(cf_options); } else { // if operation failed the ColumnFamilyOptions need to be deleted // again to prevent a memory leak. delete cf_options; } return ret_value; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_ColumnFamilyOptions_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: optimizeForPointLookup * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_optimizeForPointLookup( JNIEnv* env, jobject jobj, jlong jhandle, jlong block_cache_size_mb) { reinterpret_cast(jhandle)-> OptimizeForPointLookup(block_cache_size_mb); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: optimizeLevelStyleCompaction * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_optimizeLevelStyleCompaction( JNIEnv* env, jobject jobj, jlong jhandle, jlong memtable_memory_budget) { reinterpret_cast(jhandle)-> OptimizeLevelStyleCompaction(memtable_memory_budget); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: optimizeUniversalStyleCompaction * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_optimizeUniversalStyleCompaction( JNIEnv* env, jobject jobj, jlong jhandle, jlong memtable_memory_budget) { reinterpret_cast(jhandle)-> OptimizeUniversalStyleCompaction(memtable_memory_budget); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setComparatorHandle * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setComparatorHandle__JI( JNIEnv* env, jobject jobj, jlong jhandle, jint builtinComparator) { switch (builtinComparator) { case 1: reinterpret_cast(jhandle)->comparator = rocksdb::ReverseBytewiseComparator(); break; default: reinterpret_cast(jhandle)->comparator = rocksdb::BytewiseComparator(); break; } } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setComparatorHandle * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setComparatorHandle__JJ( JNIEnv* env, jobject jobj, jlong jopt_handle, jlong jcomparator_handle) { reinterpret_cast(jopt_handle)->comparator = reinterpret_cast(jcomparator_handle); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMergeOperatorName * Signature: (JJjava/lang/String)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMergeOperatorName( JNIEnv* env, jobject jobj, jlong jhandle, jstring jop_name) { auto options = reinterpret_cast(jhandle); const char* op_name = env->GetStringUTFChars(jop_name, 0); options->merge_operator = rocksdb::MergeOperators::CreateFromStringId( op_name); env->ReleaseStringUTFChars(jop_name, op_name); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMergeOperator * Signature: (JJjava/lang/String)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMergeOperator( JNIEnv* env, jobject jobj, jlong jhandle, jlong mergeOperatorHandle) { reinterpret_cast(jhandle)->merge_operator = *(reinterpret_cast*> (mergeOperatorHandle)); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setCompactionFilterHandle * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setCompactionFilterHandle__JJ( JNIEnv* env, jobject jobj, jlong jopt_handle, jlong jcompactionfilter_handle) { reinterpret_cast(jopt_handle)-> compaction_filter = reinterpret_cast (jcompactionfilter_handle); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setWriteBufferSize * Signature: (JJ)I */ void Java_org_rocksdb_ColumnFamilyOptions_setWriteBufferSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_buffer_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jwrite_buffer_size); if (s.ok()) { reinterpret_cast(jhandle)-> write_buffer_size = jwrite_buffer_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: writeBufferSize * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_writeBufferSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> write_buffer_size; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxWriteBufferNumber * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxWriteBufferNumber( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_write_buffer_number) { reinterpret_cast(jhandle)-> max_write_buffer_number = jmax_write_buffer_number; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxWriteBufferNumber * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_maxWriteBufferNumber( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> max_write_buffer_number; } /* * Method: setMemTableFactory * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMemTableFactory( JNIEnv* env, jobject jobj, jlong jhandle, jlong jfactory_handle) { reinterpret_cast(jhandle)-> memtable_factory.reset( reinterpret_cast(jfactory_handle)); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: memTableFactoryName * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_ColumnFamilyOptions_memTableFactoryName( JNIEnv* env, jobject jobj, jlong jhandle) { auto opt = reinterpret_cast(jhandle); rocksdb::MemTableRepFactory* tf = opt->memtable_factory.get(); // Should never be nullptr. // Default memtable factory is SkipListFactory assert(tf); // temporarly fix for the historical typo if (strcmp(tf->Name(), "HashLinkListRepFactory") == 0) { return env->NewStringUTF("HashLinkedListRepFactory"); } return env->NewStringUTF(tf->Name()); } /* * Method: useFixedLengthPrefixExtractor * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_useFixedLengthPrefixExtractor( JNIEnv* env, jobject jobj, jlong jhandle, jint jprefix_length) { reinterpret_cast(jhandle)-> prefix_extractor.reset(rocksdb::NewFixedPrefixTransform( static_cast(jprefix_length))); } /* * Method: useCappedPrefixExtractor * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_useCappedPrefixExtractor( JNIEnv* env, jobject jobj, jlong jhandle, jint jprefix_length) { reinterpret_cast(jhandle)-> prefix_extractor.reset(rocksdb::NewCappedPrefixTransform( static_cast(jprefix_length))); } /* * Method: setTableFactory * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setTableFactory( JNIEnv* env, jobject jobj, jlong jhandle, jlong jfactory_handle) { reinterpret_cast(jhandle)-> table_factory.reset(reinterpret_cast( jfactory_handle)); } /* * Method: tableFactoryName * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_ColumnFamilyOptions_tableFactoryName( JNIEnv* env, jobject jobj, jlong jhandle) { auto opt = reinterpret_cast(jhandle); rocksdb::TableFactory* tf = opt->table_factory.get(); // Should never be nullptr. // Default memtable factory is SkipListFactory assert(tf); return env->NewStringUTF(tf->Name()); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: minWriteBufferNumberToMerge * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_minWriteBufferNumberToMerge( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->min_write_buffer_number_to_merge; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMinWriteBufferNumberToMerge * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMinWriteBufferNumberToMerge( JNIEnv* env, jobject jobj, jlong jhandle, jint jmin_write_buffer_number_to_merge) { reinterpret_cast( jhandle)->min_write_buffer_number_to_merge = static_cast(jmin_write_buffer_number_to_merge); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxWriteBufferNumberToMaintain * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_maxWriteBufferNumberToMaintain( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->max_write_buffer_number_to_maintain; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxWriteBufferNumberToMaintain * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxWriteBufferNumberToMaintain( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_write_buffer_number_to_maintain) { reinterpret_cast(jhandle) ->max_write_buffer_number_to_maintain = static_cast(jmax_write_buffer_number_to_maintain); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setCompressionType * Signature: (JB)V */ void Java_org_rocksdb_ColumnFamilyOptions_setCompressionType( JNIEnv* env, jobject jobj, jlong jhandle, jbyte compression) { reinterpret_cast(jhandle)-> compression = static_cast(compression); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: compressionType * Signature: (J)B */ jbyte Java_org_rocksdb_ColumnFamilyOptions_compressionType( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> compression; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setCompressionPerLevel * Signature: (JLjava/util/List;)V */ void Java_org_rocksdb_ColumnFamilyOptions_setCompressionPerLevel( JNIEnv* env, jobject jobj, jlong jhandle, jobject jcompressionLevels) { auto* options = reinterpret_cast(jhandle); std::vector compressionLevels = rocksdb_compression_vector_helper(env, jcompressionLevels); options->compression_per_level = compressionLevels; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: compressionPerLevel * Signature: (J)Ljava/util/List; */ jobject Java_org_rocksdb_ColumnFamilyOptions_compressionPerLevel( JNIEnv* env, jobject jobj, jlong jhandle) { auto* options = reinterpret_cast(jhandle); return rocksdb_compression_list_helper(env, options->compression_per_level); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setCompactionStyle * Signature: (JB)V */ void Java_org_rocksdb_ColumnFamilyOptions_setCompactionStyle( JNIEnv* env, jobject jobj, jlong jhandle, jbyte compaction_style) { reinterpret_cast(jhandle)->compaction_style = static_cast(compaction_style); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: compactionStyle * Signature: (J)B */ jbyte Java_org_rocksdb_ColumnFamilyOptions_compactionStyle( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast (jhandle)->compaction_style; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxTableFilesSizeFIFO * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxTableFilesSizeFIFO( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_table_files_size) { reinterpret_cast(jhandle)->compaction_options_fifo.max_table_files_size = static_cast(jmax_table_files_size); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxTableFilesSizeFIFO * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_maxTableFilesSizeFIFO( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->compaction_options_fifo.max_table_files_size; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: numLevels * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_numLevels( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->num_levels; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setNumLevels * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setNumLevels( JNIEnv* env, jobject jobj, jlong jhandle, jint jnum_levels) { reinterpret_cast(jhandle)->num_levels = static_cast(jnum_levels); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: levelZeroFileNumCompactionTrigger * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_levelZeroFileNumCompactionTrigger( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level0_file_num_compaction_trigger; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setLevelZeroFileNumCompactionTrigger * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setLevelZeroFileNumCompactionTrigger( JNIEnv* env, jobject jobj, jlong jhandle, jint jlevel0_file_num_compaction_trigger) { reinterpret_cast( jhandle)->level0_file_num_compaction_trigger = static_cast(jlevel0_file_num_compaction_trigger); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: levelZeroSlowdownWritesTrigger * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_levelZeroSlowdownWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level0_slowdown_writes_trigger; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setLevelSlowdownWritesTrigger * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setLevelZeroSlowdownWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle, jint jlevel0_slowdown_writes_trigger) { reinterpret_cast( jhandle)->level0_slowdown_writes_trigger = static_cast(jlevel0_slowdown_writes_trigger); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: levelZeroStopWritesTrigger * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_levelZeroStopWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level0_stop_writes_trigger; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setLevelStopWritesTrigger * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setLevelZeroStopWritesTrigger( JNIEnv* env, jobject jobj, jlong jhandle, jint jlevel0_stop_writes_trigger) { reinterpret_cast(jhandle)-> level0_stop_writes_trigger = static_cast( jlevel0_stop_writes_trigger); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxMemCompactionLevel * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_maxMemCompactionLevel( JNIEnv* env, jobject jobj, jlong jhandle) { return 0; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxMemCompactionLevel * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxMemCompactionLevel( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_mem_compaction_level) {} /* * Class: org_rocksdb_ColumnFamilyOptions * Method: targetFileSizeBase * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_targetFileSizeBase( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> target_file_size_base; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setTargetFileSizeBase * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setTargetFileSizeBase( JNIEnv* env, jobject jobj, jlong jhandle, jlong jtarget_file_size_base) { reinterpret_cast(jhandle)-> target_file_size_base = static_cast(jtarget_file_size_base); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: targetFileSizeMultiplier * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_targetFileSizeMultiplier( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->target_file_size_multiplier; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setTargetFileSizeMultiplier * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setTargetFileSizeMultiplier( JNIEnv* env, jobject jobj, jlong jhandle, jint jtarget_file_size_multiplier) { reinterpret_cast( jhandle)->target_file_size_multiplier = static_cast(jtarget_file_size_multiplier); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxBytesForLevelBase * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_maxBytesForLevelBase( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_bytes_for_level_base; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxBytesForLevelBase * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxBytesForLevelBase( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_bytes_for_level_base) { reinterpret_cast( jhandle)->max_bytes_for_level_base = static_cast(jmax_bytes_for_level_base); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: levelCompactionDynamicLevelBytes * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_levelCompactionDynamicLevelBytes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->level_compaction_dynamic_level_bytes; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setLevelCompactionDynamicLevelBytes * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setLevelCompactionDynamicLevelBytes( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jenable_dynamic_level_bytes) { reinterpret_cast( jhandle)->level_compaction_dynamic_level_bytes = (jenable_dynamic_level_bytes); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxBytesForLevelMultiplier * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_maxBytesForLevelMultiplier( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_bytes_for_level_multiplier; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxBytesForLevelMultiplier * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxBytesForLevelMultiplier( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_bytes_for_level_multiplier) { reinterpret_cast( jhandle)->max_bytes_for_level_multiplier = static_cast(jmax_bytes_for_level_multiplier); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: expandedCompactionFactor * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_expandedCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->expanded_compaction_factor; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setExpandedCompactionFactor * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setExpandedCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle, jint jexpanded_compaction_factor) { reinterpret_cast( jhandle)->expanded_compaction_factor = static_cast(jexpanded_compaction_factor); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: sourceCompactionFactor * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_sourceCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->source_compaction_factor; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setSourceCompactionFactor * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setSourceCompactionFactor( JNIEnv* env, jobject jobj, jlong jhandle, jint jsource_compaction_factor) { reinterpret_cast( jhandle)->source_compaction_factor = static_cast(jsource_compaction_factor); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxGrandparentOverlapFactor * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_maxGrandparentOverlapFactor( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_grandparent_overlap_factor; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxGrandparentOverlapFactor * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxGrandparentOverlapFactor( JNIEnv* env, jobject jobj, jlong jhandle, jint jmax_grandparent_overlap_factor) { reinterpret_cast( jhandle)->max_grandparent_overlap_factor = static_cast(jmax_grandparent_overlap_factor); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: softRateLimit * Signature: (J)D */ jdouble Java_org_rocksdb_ColumnFamilyOptions_softRateLimit( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> soft_rate_limit; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setSoftRateLimit * Signature: (JD)V */ void Java_org_rocksdb_ColumnFamilyOptions_setSoftRateLimit( JNIEnv* env, jobject jobj, jlong jhandle, jdouble jsoft_rate_limit) { reinterpret_cast(jhandle)->soft_rate_limit = static_cast(jsoft_rate_limit); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: hardRateLimit * Signature: (J)D */ jdouble Java_org_rocksdb_ColumnFamilyOptions_hardRateLimit( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> hard_rate_limit; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setHardRateLimit * Signature: (JD)V */ void Java_org_rocksdb_ColumnFamilyOptions_setHardRateLimit( JNIEnv* env, jobject jobj, jlong jhandle, jdouble jhard_rate_limit) { reinterpret_cast(jhandle)->hard_rate_limit = static_cast(jhard_rate_limit); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: rateLimitDelayMaxMilliseconds * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_rateLimitDelayMaxMilliseconds( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->rate_limit_delay_max_milliseconds; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setRateLimitDelayMaxMilliseconds * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setRateLimitDelayMaxMilliseconds( JNIEnv* env, jobject jobj, jlong jhandle, jint jrate_limit_delay_max_milliseconds) { reinterpret_cast( jhandle)->rate_limit_delay_max_milliseconds = static_cast(jrate_limit_delay_max_milliseconds); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: arenaBlockSize * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_arenaBlockSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> arena_block_size; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setArenaBlockSize * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setArenaBlockSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong jarena_block_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(jarena_block_size); if (s.ok()) { reinterpret_cast(jhandle)-> arena_block_size = jarena_block_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: disableAutoCompactions * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_disableAutoCompactions( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->disable_auto_compactions; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setDisableAutoCompactions * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setDisableAutoCompactions( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jdisable_auto_compactions) { reinterpret_cast( jhandle)->disable_auto_compactions = static_cast(jdisable_auto_compactions); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: purgeRedundantKvsWhileFlush * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_purgeRedundantKvsWhileFlush( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->purge_redundant_kvs_while_flush; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setPurgeRedundantKvsWhileFlush * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setPurgeRedundantKvsWhileFlush( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jpurge_redundant_kvs_while_flush) { reinterpret_cast( jhandle)->purge_redundant_kvs_while_flush = static_cast(jpurge_redundant_kvs_while_flush); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: verifyChecksumsInCompaction * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_verifyChecksumsInCompaction( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->verify_checksums_in_compaction; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setVerifyChecksumsInCompaction * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setVerifyChecksumsInCompaction( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jverify_checksums_in_compaction) { reinterpret_cast( jhandle)->verify_checksums_in_compaction = static_cast(jverify_checksums_in_compaction); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: filterDeletes * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_filterDeletes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> filter_deletes; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setFilterDeletes * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setFilterDeletes( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jfilter_deletes) { reinterpret_cast(jhandle)->filter_deletes = static_cast(jfilter_deletes); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxSequentialSkipInIterations * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_maxSequentialSkipInIterations( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_sequential_skip_in_iterations; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxSequentialSkipInIterations * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxSequentialSkipInIterations( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_sequential_skip_in_iterations) { reinterpret_cast( jhandle)->max_sequential_skip_in_iterations = static_cast(jmax_sequential_skip_in_iterations); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: inplaceUpdateSupport * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_inplaceUpdateSupport( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->inplace_update_support; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setInplaceUpdateSupport * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setInplaceUpdateSupport( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jinplace_update_support) { reinterpret_cast( jhandle)->inplace_update_support = static_cast(jinplace_update_support); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: inplaceUpdateNumLocks * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_inplaceUpdateNumLocks( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->inplace_update_num_locks; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setInplaceUpdateNumLocks * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setInplaceUpdateNumLocks( JNIEnv* env, jobject jobj, jlong jhandle, jlong jinplace_update_num_locks) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t( jinplace_update_num_locks); if (s.ok()) { reinterpret_cast(jhandle)-> inplace_update_num_locks = jinplace_update_num_locks; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: memtablePrefixBloomBits * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_memtablePrefixBloomBits( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->memtable_prefix_bloom_bits; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMemtablePrefixBloomBits * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMemtablePrefixBloomBits( JNIEnv* env, jobject jobj, jlong jhandle, jint jmemtable_prefix_bloom_bits) { reinterpret_cast( jhandle)->memtable_prefix_bloom_bits = static_cast(jmemtable_prefix_bloom_bits); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: memtablePrefixBloomProbes * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_memtablePrefixBloomProbes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->memtable_prefix_bloom_probes; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMemtablePrefixBloomProbes * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMemtablePrefixBloomProbes( JNIEnv* env, jobject jobj, jlong jhandle, jint jmemtable_prefix_bloom_probes) { reinterpret_cast( jhandle)->memtable_prefix_bloom_probes = static_cast(jmemtable_prefix_bloom_probes); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: bloomLocality * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_bloomLocality( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> bloom_locality; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setBloomLocality * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setBloomLocality( JNIEnv* env, jobject jobj, jlong jhandle, jint jbloom_locality) { reinterpret_cast(jhandle)->bloom_locality = static_cast(jbloom_locality); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: maxSuccessiveMerges * Signature: (J)J */ jlong Java_org_rocksdb_ColumnFamilyOptions_maxSuccessiveMerges( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> max_successive_merges; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMaxSuccessiveMerges * Signature: (JJ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMaxSuccessiveMerges( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_successive_merges) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t( jmax_successive_merges); if (s.ok()) { reinterpret_cast(jhandle)-> max_successive_merges = jmax_successive_merges; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: minPartialMergeOperands * Signature: (J)I */ jint Java_org_rocksdb_ColumnFamilyOptions_minPartialMergeOperands( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->min_partial_merge_operands; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setMinPartialMergeOperands * Signature: (JI)V */ void Java_org_rocksdb_ColumnFamilyOptions_setMinPartialMergeOperands( JNIEnv* env, jobject jobj, jlong jhandle, jint jmin_partial_merge_operands) { reinterpret_cast( jhandle)->min_partial_merge_operands = static_cast(jmin_partial_merge_operands); } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: optimizeFiltersForHits * Signature: (J)Z */ jboolean Java_org_rocksdb_ColumnFamilyOptions_optimizeFiltersForHits( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->optimize_filters_for_hits; } /* * Class: org_rocksdb_ColumnFamilyOptions * Method: setOptimizeFiltersForHits * Signature: (JZ)V */ void Java_org_rocksdb_ColumnFamilyOptions_setOptimizeFiltersForHits( JNIEnv* env, jobject jobj, jlong jhandle, jboolean joptimize_filters_for_hits) { reinterpret_cast( jhandle)->optimize_filters_for_hits = static_cast(joptimize_filters_for_hits); } ///////////////////////////////////////////////////////////////////// // rocksdb::DBOptions /* * Class: org_rocksdb_DBOptions * Method: newDBOptions * Signature: ()V */ void Java_org_rocksdb_DBOptions_newDBOptions(JNIEnv* env, jobject jobj) { rocksdb::DBOptions* dbop = new rocksdb::DBOptions(); rocksdb::DBOptionsJni::setHandle(env, jobj, dbop); } /* * Class: org_rocksdb_DBOptions * Method: getDBOptionsFromProps * Signature: (Ljava/util/String;)J */ jlong Java_org_rocksdb_DBOptions_getDBOptionsFromProps( JNIEnv* env, jclass jclazz, jstring jopt_string) { jlong ret_value = 0; rocksdb::DBOptions* db_options = new rocksdb::DBOptions(); const char* opt_string = env->GetStringUTFChars(jopt_string, 0); rocksdb::Status status = rocksdb::GetDBOptionsFromString( rocksdb::DBOptions(), opt_string, db_options); env->ReleaseStringUTFChars(jopt_string, opt_string); // Check if DBOptions creation was possible. if (status.ok()) { ret_value = reinterpret_cast(db_options); } else { // if operation failed the DBOptions need to be deleted // again to prevent a memory leak. delete db_options; } return ret_value; } /* * Class: org_rocksdb_DBOptions * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_DBOptions_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } /* * Class: org_rocksdb_DBOptions * Method: setIncreaseParallelism * Signature: (JI)V */ void Java_org_rocksdb_DBOptions_setIncreaseParallelism( JNIEnv * env, jobject jobj, jlong jhandle, jint totalThreads) { reinterpret_cast (jhandle)->IncreaseParallelism(static_cast(totalThreads)); } /* * Class: org_rocksdb_DBOptions * Method: setCreateIfMissing * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setCreateIfMissing( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { reinterpret_cast(jhandle)-> create_if_missing = flag; } /* * Class: org_rocksdb_DBOptions * Method: createIfMissing * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_createIfMissing( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->create_if_missing; } /* * Class: org_rocksdb_DBOptions * Method: setCreateMissingColumnFamilies * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setCreateMissingColumnFamilies( JNIEnv* env, jobject jobj, jlong jhandle, jboolean flag) { reinterpret_cast (jhandle)->create_missing_column_families = flag; } /* * Class: org_rocksdb_DBOptions * Method: createMissingColumnFamilies * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_createMissingColumnFamilies( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast (jhandle)->create_missing_column_families; } /* * Class: org_rocksdb_DBOptions * Method: setErrorIfExists * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setErrorIfExists( JNIEnv* env, jobject jobj, jlong jhandle, jboolean error_if_exists) { reinterpret_cast(jhandle)->error_if_exists = static_cast(error_if_exists); } /* * Class: org_rocksdb_DBOptions * Method: errorIfExists * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_errorIfExists( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->error_if_exists; } /* * Class: org_rocksdb_DBOptions * Method: setParanoidChecks * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setParanoidChecks( JNIEnv* env, jobject jobj, jlong jhandle, jboolean paranoid_checks) { reinterpret_cast(jhandle)->paranoid_checks = static_cast(paranoid_checks); } /* * Class: org_rocksdb_DBOptions * Method: paranoidChecks * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_paranoidChecks( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->paranoid_checks; } /* * Class: org_rocksdb_DBOptions * Method: setRateLimiter * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setRateLimiter( JNIEnv* env, jobject jobj, jlong jhandle, jlong jrate_limiter_handle) { reinterpret_cast(jhandle)->rate_limiter.reset( reinterpret_cast(jrate_limiter_handle)); } /* * Class: org_rocksdb_DBOptions * Method: setLogger * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setLogger( JNIEnv* env, jobject jobj, jlong jhandle, jlong jlogger_handle) { std::shared_ptr *pLogger = reinterpret_cast *>( jlogger_handle); reinterpret_cast(jhandle)->info_log = *pLogger; } /* * Class: org_rocksdb_DBOptions * Method: setInfoLogLevel * Signature: (JB)V */ void Java_org_rocksdb_DBOptions_setInfoLogLevel( JNIEnv* env, jobject jobj, jlong jhandle, jbyte jlog_level) { reinterpret_cast(jhandle)->info_log_level = static_cast(jlog_level); } /* * Class: org_rocksdb_DBOptions * Method: infoLogLevel * Signature: (J)B */ jbyte Java_org_rocksdb_DBOptions_infoLogLevel( JNIEnv* env, jobject jobj, jlong jhandle) { return static_cast( reinterpret_cast(jhandle)->info_log_level); } /* * Class: org_rocksdb_DBOptions * Method: setMaxTotalWalSize * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setMaxTotalWalSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_total_wal_size) { reinterpret_cast(jhandle)->max_total_wal_size = static_cast(jmax_total_wal_size); } /* * Class: org_rocksdb_DBOptions * Method: maxTotalWalSize * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_maxTotalWalSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> max_total_wal_size; } /* * Class: org_rocksdb_DBOptions * Method: setMaxOpenFiles * Signature: (JI)V */ void Java_org_rocksdb_DBOptions_setMaxOpenFiles( JNIEnv* env, jobject jobj, jlong jhandle, jint max_open_files) { reinterpret_cast(jhandle)->max_open_files = static_cast(max_open_files); } /* * Class: org_rocksdb_DBOptions * Method: maxOpenFiles * Signature: (J)I */ jint Java_org_rocksdb_DBOptions_maxOpenFiles( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_open_files; } /* * Class: org_rocksdb_DBOptions * Method: createStatistics * Signature: (J)V */ void Java_org_rocksdb_DBOptions_createStatistics( JNIEnv* env, jobject jobj, jlong jOptHandle) { reinterpret_cast(jOptHandle)->statistics = rocksdb::CreateDBStatistics(); } /* * Class: org_rocksdb_DBOptions * Method: statisticsPtr * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_statisticsPtr( JNIEnv* env, jobject jobj, jlong jOptHandle) { auto st = reinterpret_cast(jOptHandle)-> statistics.get(); return reinterpret_cast(st); } /* * Class: org_rocksdb_DBOptions * Method: setDisableDataSync * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setDisableDataSync( JNIEnv* env, jobject jobj, jlong jhandle, jboolean disableDataSync) { reinterpret_cast(jhandle)->disableDataSync = static_cast(disableDataSync); } /* * Class: org_rocksdb_DBOptions * Method: disableDataSync * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_disableDataSync( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->disableDataSync; } /* * Class: org_rocksdb_DBOptions * Method: setUseFsync * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setUseFsync( JNIEnv* env, jobject jobj, jlong jhandle, jboolean use_fsync) { reinterpret_cast(jhandle)->use_fsync = static_cast(use_fsync); } /* * Class: org_rocksdb_DBOptions * Method: useFsync * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_useFsync( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->use_fsync; } /* * Class: org_rocksdb_DBOptions * Method: setDbLogDir * Signature: (JLjava/lang/String)V */ void Java_org_rocksdb_DBOptions_setDbLogDir( JNIEnv* env, jobject jobj, jlong jhandle, jstring jdb_log_dir) { const char* log_dir = env->GetStringUTFChars(jdb_log_dir, 0); reinterpret_cast(jhandle)->db_log_dir.assign(log_dir); env->ReleaseStringUTFChars(jdb_log_dir, log_dir); } /* * Class: org_rocksdb_DBOptions * Method: dbLogDir * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_DBOptions_dbLogDir( JNIEnv* env, jobject jobj, jlong jhandle) { return env->NewStringUTF( reinterpret_cast(jhandle)->db_log_dir.c_str()); } /* * Class: org_rocksdb_DBOptions * Method: setWalDir * Signature: (JLjava/lang/String)V */ void Java_org_rocksdb_DBOptions_setWalDir( JNIEnv* env, jobject jobj, jlong jhandle, jstring jwal_dir) { const char* wal_dir = env->GetStringUTFChars(jwal_dir, 0); reinterpret_cast(jhandle)->wal_dir.assign(wal_dir); env->ReleaseStringUTFChars(jwal_dir, wal_dir); } /* * Class: org_rocksdb_DBOptions * Method: walDir * Signature: (J)Ljava/lang/String */ jstring Java_org_rocksdb_DBOptions_walDir( JNIEnv* env, jobject jobj, jlong jhandle) { return env->NewStringUTF( reinterpret_cast(jhandle)->wal_dir.c_str()); } /* * Class: org_rocksdb_DBOptions * Method: setDeleteObsoleteFilesPeriodMicros * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setDeleteObsoleteFilesPeriodMicros( JNIEnv* env, jobject jobj, jlong jhandle, jlong micros) { reinterpret_cast(jhandle) ->delete_obsolete_files_period_micros = static_cast(micros); } /* * Class: org_rocksdb_DBOptions * Method: deleteObsoleteFilesPeriodMicros * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_deleteObsoleteFilesPeriodMicros( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->delete_obsolete_files_period_micros; } /* * Class: org_rocksdb_DBOptions * Method: setMaxBackgroundCompactions * Signature: (JI)V */ void Java_org_rocksdb_DBOptions_setMaxBackgroundCompactions( JNIEnv* env, jobject jobj, jlong jhandle, jint max) { reinterpret_cast(jhandle) ->max_background_compactions = static_cast(max); } /* * Class: org_rocksdb_DBOptions * Method: maxBackgroundCompactions * Signature: (J)I */ jint Java_org_rocksdb_DBOptions_maxBackgroundCompactions( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->max_background_compactions; } /* * Class: org_rocksdb_DBOptions * Method: setMaxBackgroundFlushes * Signature: (JI)V */ void Java_org_rocksdb_DBOptions_setMaxBackgroundFlushes( JNIEnv* env, jobject jobj, jlong jhandle, jint max_background_flushes) { reinterpret_cast(jhandle)->max_background_flushes = static_cast(max_background_flushes); } /* * Class: org_rocksdb_DBOptions * Method: maxBackgroundFlushes * Signature: (J)I */ jint Java_org_rocksdb_DBOptions_maxBackgroundFlushes( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> max_background_flushes; } /* * Class: org_rocksdb_DBOptions * Method: setMaxLogFileSize * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setMaxLogFileSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong max_log_file_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(max_log_file_size); if (s.ok()) { reinterpret_cast(jhandle)->max_log_file_size = max_log_file_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_DBOptions * Method: maxLogFileSize * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_maxLogFileSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->max_log_file_size; } /* * Class: org_rocksdb_DBOptions * Method: setLogFileTimeToRoll * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setLogFileTimeToRoll( JNIEnv* env, jobject jobj, jlong jhandle, jlong log_file_time_to_roll) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t( log_file_time_to_roll); if (s.ok()) { reinterpret_cast(jhandle)->log_file_time_to_roll = log_file_time_to_roll; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_DBOptions * Method: logFileTimeToRoll * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_logFileTimeToRoll( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->log_file_time_to_roll; } /* * Class: org_rocksdb_DBOptions * Method: setKeepLogFileNum * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setKeepLogFileNum( JNIEnv* env, jobject jobj, jlong jhandle, jlong keep_log_file_num) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(keep_log_file_num); if (s.ok()) { reinterpret_cast(jhandle)->keep_log_file_num = keep_log_file_num; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_DBOptions * Method: keepLogFileNum * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_keepLogFileNum( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->keep_log_file_num; } /* * Class: org_rocksdb_DBOptions * Method: setRecycleLogFiles * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setRecycleLogFileNum( JNIEnv* env, jobject jobj, jlong jhandle, jlong recycle_log_file_num) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(recycle_log_file_num); if (s.ok()) { reinterpret_cast(jhandle)->recycle_log_file_num = recycle_log_file_num; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_DBOptions * Method: recycleLogFiles * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_recycleLogFileNum(JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->recycle_log_file_num; } /* * Class: org_rocksdb_DBOptions * Method: setMaxManifestFileSize * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setMaxManifestFileSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong max_manifest_file_size) { reinterpret_cast(jhandle)->max_manifest_file_size = static_cast(max_manifest_file_size); } /* * Class: org_rocksdb_DBOptions * Method: maxManifestFileSize * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_maxManifestFileSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> max_manifest_file_size; } /* * Class: org_rocksdb_DBOptions * Method: setTableCacheNumshardbits * Signature: (JI)V */ void Java_org_rocksdb_DBOptions_setTableCacheNumshardbits( JNIEnv* env, jobject jobj, jlong jhandle, jint table_cache_numshardbits) { reinterpret_cast(jhandle)->table_cache_numshardbits = static_cast(table_cache_numshardbits); } /* * Class: org_rocksdb_DBOptions * Method: tableCacheNumshardbits * Signature: (J)I */ jint Java_org_rocksdb_DBOptions_tableCacheNumshardbits( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)-> table_cache_numshardbits; } /* * Class: org_rocksdb_DBOptions * Method: setWalTtlSeconds * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setWalTtlSeconds( JNIEnv* env, jobject jobj, jlong jhandle, jlong WAL_ttl_seconds) { reinterpret_cast(jhandle)->WAL_ttl_seconds = static_cast(WAL_ttl_seconds); } /* * Class: org_rocksdb_DBOptions * Method: walTtlSeconds * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_walTtlSeconds( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->WAL_ttl_seconds; } /* * Class: org_rocksdb_DBOptions * Method: setWalSizeLimitMB * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setWalSizeLimitMB( JNIEnv* env, jobject jobj, jlong jhandle, jlong WAL_size_limit_MB) { reinterpret_cast(jhandle)->WAL_size_limit_MB = static_cast(WAL_size_limit_MB); } /* * Class: org_rocksdb_DBOptions * Method: walTtlSeconds * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_walSizeLimitMB( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->WAL_size_limit_MB; } /* * Class: org_rocksdb_DBOptions * Method: setManifestPreallocationSize * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setManifestPreallocationSize( JNIEnv* env, jobject jobj, jlong jhandle, jlong preallocation_size) { rocksdb::Status s = rocksdb::check_if_jlong_fits_size_t(preallocation_size); if (s.ok()) { reinterpret_cast(jhandle)-> manifest_preallocation_size = preallocation_size; } else { rocksdb::IllegalArgumentExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_DBOptions * Method: manifestPreallocationSize * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_manifestPreallocationSize( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->manifest_preallocation_size; } /* * Class: org_rocksdb_DBOptions * Method: setAllowOsBuffer * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setAllowOsBuffer( JNIEnv* env, jobject jobj, jlong jhandle, jboolean allow_os_buffer) { reinterpret_cast(jhandle)->allow_os_buffer = static_cast(allow_os_buffer); } /* * Class: org_rocksdb_DBOptions * Method: allowOsBuffer * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_allowOsBuffer( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->allow_os_buffer; } /* * Class: org_rocksdb_DBOptions * Method: setAllowMmapReads * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setAllowMmapReads( JNIEnv* env, jobject jobj, jlong jhandle, jboolean allow_mmap_reads) { reinterpret_cast(jhandle)->allow_mmap_reads = static_cast(allow_mmap_reads); } /* * Class: org_rocksdb_DBOptions * Method: allowMmapReads * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_allowMmapReads( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->allow_mmap_reads; } /* * Class: org_rocksdb_DBOptions * Method: setAllowMmapWrites * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setAllowMmapWrites( JNIEnv* env, jobject jobj, jlong jhandle, jboolean allow_mmap_writes) { reinterpret_cast(jhandle)->allow_mmap_writes = static_cast(allow_mmap_writes); } /* * Class: org_rocksdb_DBOptions * Method: allowMmapWrites * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_allowMmapWrites( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->allow_mmap_writes; } /* * Class: org_rocksdb_DBOptions * Method: setIsFdCloseOnExec * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setIsFdCloseOnExec( JNIEnv* env, jobject jobj, jlong jhandle, jboolean is_fd_close_on_exec) { reinterpret_cast(jhandle)->is_fd_close_on_exec = static_cast(is_fd_close_on_exec); } /* * Class: org_rocksdb_DBOptions * Method: isFdCloseOnExec * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_isFdCloseOnExec( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->is_fd_close_on_exec; } /* * Class: org_rocksdb_DBOptions * Method: setStatsDumpPeriodSec * Signature: (JI)V */ void Java_org_rocksdb_DBOptions_setStatsDumpPeriodSec( JNIEnv* env, jobject jobj, jlong jhandle, jint stats_dump_period_sec) { reinterpret_cast(jhandle)->stats_dump_period_sec = static_cast(stats_dump_period_sec); } /* * Class: org_rocksdb_DBOptions * Method: statsDumpPeriodSec * Signature: (J)I */ jint Java_org_rocksdb_DBOptions_statsDumpPeriodSec( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->stats_dump_period_sec; } /* * Class: org_rocksdb_DBOptions * Method: setAdviseRandomOnOpen * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setAdviseRandomOnOpen( JNIEnv* env, jobject jobj, jlong jhandle, jboolean advise_random_on_open) { reinterpret_cast(jhandle)->advise_random_on_open = static_cast(advise_random_on_open); } /* * Class: org_rocksdb_DBOptions * Method: adviseRandomOnOpen * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_adviseRandomOnOpen( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->advise_random_on_open; } /* * Class: org_rocksdb_DBOptions * Method: setUseAdaptiveMutex * Signature: (JZ)V */ void Java_org_rocksdb_DBOptions_setUseAdaptiveMutex( JNIEnv* env, jobject jobj, jlong jhandle, jboolean use_adaptive_mutex) { reinterpret_cast(jhandle)->use_adaptive_mutex = static_cast(use_adaptive_mutex); } /* * Class: org_rocksdb_DBOptions * Method: useAdaptiveMutex * Signature: (J)Z */ jboolean Java_org_rocksdb_DBOptions_useAdaptiveMutex( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->use_adaptive_mutex; } /* * Class: org_rocksdb_DBOptions * Method: setBytesPerSync * Signature: (JJ)V */ void Java_org_rocksdb_DBOptions_setBytesPerSync( JNIEnv* env, jobject jobj, jlong jhandle, jlong bytes_per_sync) { reinterpret_cast(jhandle)->bytes_per_sync = static_cast(bytes_per_sync); } /* * Class: org_rocksdb_DBOptions * Method: bytesPerSync * Signature: (J)J */ jlong Java_org_rocksdb_DBOptions_bytesPerSync( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->bytes_per_sync; } ////////////////////////////////////////////////////////////////////////////// // rocksdb::WriteOptions /* * Class: org_rocksdb_WriteOptions * Method: newWriteOptions * Signature: ()V */ void Java_org_rocksdb_WriteOptions_newWriteOptions( JNIEnv* env, jobject jwrite_options) { rocksdb::WriteOptions* op = new rocksdb::WriteOptions(); rocksdb::WriteOptionsJni::setHandle(env, jwrite_options, op); } /* * Class: org_rocksdb_WriteOptions * Method: disposeInternal * Signature: ()V */ void Java_org_rocksdb_WriteOptions_disposeInternal( JNIEnv* env, jobject jwrite_options, jlong jhandle) { auto write_options = reinterpret_cast(jhandle); delete write_options; rocksdb::WriteOptionsJni::setHandle(env, jwrite_options, nullptr); } /* * Class: org_rocksdb_WriteOptions * Method: setSync * Signature: (JZ)V */ void Java_org_rocksdb_WriteOptions_setSync( JNIEnv* env, jobject jwrite_options, jlong jhandle, jboolean jflag) { reinterpret_cast(jhandle)->sync = jflag; } /* * Class: org_rocksdb_WriteOptions * Method: sync * Signature: (J)Z */ jboolean Java_org_rocksdb_WriteOptions_sync( JNIEnv* env, jobject jwrite_options, jlong jhandle) { return reinterpret_cast(jhandle)->sync; } /* * Class: org_rocksdb_WriteOptions * Method: setDisableWAL * Signature: (JZ)V */ void Java_org_rocksdb_WriteOptions_setDisableWAL( JNIEnv* env, jobject jwrite_options, jlong jhandle, jboolean jflag) { reinterpret_cast(jhandle)->disableWAL = jflag; } /* * Class: org_rocksdb_WriteOptions * Method: disableWAL * Signature: (J)Z */ jboolean Java_org_rocksdb_WriteOptions_disableWAL( JNIEnv* env, jobject jwrite_options, jlong jhandle) { return reinterpret_cast(jhandle)->disableWAL; } ///////////////////////////////////////////////////////////////////// // rocksdb::ReadOptions /* * Class: org_rocksdb_ReadOptions * Method: newReadOptions * Signature: ()V */ void Java_org_rocksdb_ReadOptions_newReadOptions( JNIEnv* env, jobject jobj) { auto read_opt = new rocksdb::ReadOptions(); rocksdb::ReadOptionsJni::setHandle(env, jobj, read_opt); } /* * Class: org_rocksdb_ReadOptions * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_ReadOptions_disposeInternal( JNIEnv* env, jobject jobj, jlong jhandle) { delete reinterpret_cast(jhandle); rocksdb::ReadOptionsJni::setHandle(env, jobj, nullptr); } /* * Class: org_rocksdb_ReadOptions * Method: setVerifyChecksums * Signature: (JZ)V */ void Java_org_rocksdb_ReadOptions_setVerifyChecksums( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jverify_checksums) { reinterpret_cast(jhandle)->verify_checksums = static_cast(jverify_checksums); } /* * Class: org_rocksdb_ReadOptions * Method: verifyChecksums * Signature: (J)Z */ jboolean Java_org_rocksdb_ReadOptions_verifyChecksums( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast( jhandle)->verify_checksums; } /* * Class: org_rocksdb_ReadOptions * Method: setFillCache * Signature: (JZ)V */ void Java_org_rocksdb_ReadOptions_setFillCache( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jfill_cache) { reinterpret_cast(jhandle)->fill_cache = static_cast(jfill_cache); } /* * Class: org_rocksdb_ReadOptions * Method: fillCache * Signature: (J)Z */ jboolean Java_org_rocksdb_ReadOptions_fillCache( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->fill_cache; } /* * Class: org_rocksdb_ReadOptions * Method: setTailing * Signature: (JZ)V */ void Java_org_rocksdb_ReadOptions_setTailing( JNIEnv* env, jobject jobj, jlong jhandle, jboolean jtailing) { reinterpret_cast(jhandle)->tailing = static_cast(jtailing); } /* * Class: org_rocksdb_ReadOptions * Method: tailing * Signature: (J)Z */ jboolean Java_org_rocksdb_ReadOptions_tailing( JNIEnv* env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle)->tailing; } /* * Class: org_rocksdb_ReadOptions * Method: setSnapshot * Signature: (JJ)V */ void Java_org_rocksdb_ReadOptions_setSnapshot( JNIEnv* env, jobject jobj, jlong jhandle, jlong jsnapshot) { reinterpret_cast(jhandle)->snapshot = reinterpret_cast(jsnapshot); } /* * Class: org_rocksdb_ReadOptions * Method: snapshot * Signature: (J)J */ jlong Java_org_rocksdb_ReadOptions_snapshot( JNIEnv* env, jobject jobj, jlong jhandle) { auto& snapshot = reinterpret_cast(jhandle)->snapshot; return reinterpret_cast(snapshot); } ///////////////////////////////////////////////////////////////////// // rocksdb::ComparatorOptions /* * Class: org_rocksdb_ComparatorOptions * Method: newComparatorOptions * Signature: ()V */ void Java_org_rocksdb_ComparatorOptions_newComparatorOptions( JNIEnv* env, jobject jobj) { auto comparator_opt = new rocksdb::ComparatorJniCallbackOptions(); rocksdb::ComparatorOptionsJni::setHandle(env, jobj, comparator_opt); } /* * Class: org_rocksdb_ComparatorOptions * Method: useAdaptiveMutex * Signature: (J)Z */ jboolean Java_org_rocksdb_ComparatorOptions_useAdaptiveMutex( JNIEnv * env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->use_adaptive_mutex; } /* * Class: org_rocksdb_ComparatorOptions * Method: setUseAdaptiveMutex * Signature: (JZ)V */ void Java_org_rocksdb_ComparatorOptions_setUseAdaptiveMutex( JNIEnv * env, jobject jobj, jlong jhandle, jboolean juse_adaptive_mutex) { reinterpret_cast(jhandle) ->use_adaptive_mutex = static_cast(juse_adaptive_mutex); } /* * Class: org_rocksdb_ComparatorOptions * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_ComparatorOptions_disposeInternal( JNIEnv * env, jobject jobj, jlong jhandle) { delete reinterpret_cast(jhandle); rocksdb::ComparatorOptionsJni::setHandle(env, jobj, nullptr); } ///////////////////////////////////////////////////////////////////// // rocksdb::FlushOptions /* * Class: org_rocksdb_FlushOptions * Method: newFlushOptions * Signature: ()V */ void Java_org_rocksdb_FlushOptions_newFlushOptions( JNIEnv* env, jobject jobj) { auto flush_opt = new rocksdb::FlushOptions(); rocksdb::FlushOptionsJni::setHandle(env, jobj, flush_opt); } /* * Class: org_rocksdb_FlushOptions * Method: setWaitForFlush * Signature: (JZ)V */ void Java_org_rocksdb_FlushOptions_setWaitForFlush( JNIEnv * env, jobject jobj, jlong jhandle, jboolean jwait) { reinterpret_cast(jhandle) ->wait = static_cast(jwait); } /* * Class: org_rocksdb_FlushOptions * Method: waitForFlush * Signature: (J)Z */ jboolean Java_org_rocksdb_FlushOptions_waitForFlush( JNIEnv * env, jobject jobj, jlong jhandle) { return reinterpret_cast(jhandle) ->wait; } /* * Class: org_rocksdb_FlushOptions * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_FlushOptions_disposeInternal( JNIEnv * env, jobject jobj, jlong jhandle) { delete reinterpret_cast(jhandle); rocksdb::FlushOptionsJni::setHandle(env, jobj, nullptr); } rocksdb-4.5.1/java/rocksjni/portal.h000066400000000000000000000624131267530634000174010ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // This file is designed for caching those frequently used IDs and provide // efficient portal (i.e, a set of static functions) to access java code // from c++. #ifndef JAVA_ROCKSJNI_PORTAL_H_ #define JAVA_ROCKSJNI_PORTAL_H_ #include #include #include #include #include "rocksdb/db.h" #include "rocksdb/filter_policy.h" #include "rocksdb/status.h" #include "rocksdb/utilities/backupable_db.h" #include "rocksdb/utilities/write_batch_with_index.h" #include "rocksjni/comparatorjnicallback.h" #include "rocksjni/loggerjnicallback.h" #include "rocksjni/writebatchhandlerjnicallback.h" // Remove macro on windows #ifdef DELETE #undef DELETE #endif namespace rocksdb { // Detect if jlong overflows size_t inline Status check_if_jlong_fits_size_t(const jlong& jvalue) { Status s = Status::OK(); if (static_cast(jvalue) > std::numeric_limits::max()) { s = Status::InvalidArgument(Slice("jlong overflows 32 bit value.")); } return s; } // Native class template template class RocksDBNativeClass { public: // Get the java class id static jclass getJClass(JNIEnv* env, const char* jclazz_name) { jclass jclazz = env->FindClass(jclazz_name); assert(jclazz != nullptr); return jclazz; } // Get the field id of the member variable to store // the ptr static jfieldID getHandleFieldID(JNIEnv* env) { static jfieldID fid = env->GetFieldID( DERIVED::getJClass(env), "nativeHandle_", "J"); assert(fid != nullptr); return fid; } // Get the pointer from Java static PTR getHandle(JNIEnv* env, jobject jobj) { return reinterpret_cast( env->GetLongField(jobj, getHandleFieldID(env))); } // Pass the pointer to the java side. static void setHandle(JNIEnv* env, jobject jdb, PTR ptr) { env->SetLongField( jdb, getHandleFieldID(env), reinterpret_cast(ptr)); } }; // Java Exception template template class RocksDBJavaException { public: // Get the java class id static jclass getJClass(JNIEnv* env, const char* jclazz_name) { jclass jclazz = env->FindClass(jclazz_name); assert(jclazz != nullptr); return jclazz; } // Create and throw a java exception by converting the input // Status. // // In case s.ok() is true, then this function will not throw any // exception. static void ThrowNew(JNIEnv* env, Status s) { if (s.ok()) { return; } jstring msg = env->NewStringUTF(s.ToString().c_str()); // get the constructor id of org.rocksdb.RocksDBException static jmethodID mid = env->GetMethodID( DERIVED::getJClass(env), "", "(Ljava/lang/String;)V"); assert(mid != nullptr); env->Throw((jthrowable)env->NewObject(DERIVED::getJClass(env), mid, msg)); } }; // The portal class for org.rocksdb.RocksDB class RocksDBJni : public RocksDBNativeClass { public: // Get the java class id of org.rocksdb.RocksDB. static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/RocksDB"); } }; // The portal class for org.rocksdb.RocksDBException class RocksDBExceptionJni : public RocksDBJavaException { public: // Get the java class id of java.lang.IllegalArgumentException static jclass getJClass(JNIEnv* env) { return RocksDBJavaException::getJClass(env, "org/rocksdb/RocksDBException"); } }; // The portal class for java.lang.IllegalArgumentException class IllegalArgumentExceptionJni : public RocksDBJavaException { public: // Get the java class id of java.lang.IllegalArgumentException static jclass getJClass(JNIEnv* env) { return RocksDBJavaException::getJClass(env, "java/lang/IllegalArgumentException"); } }; // The portal class for org.rocksdb.Options class OptionsJni : public RocksDBNativeClass< rocksdb::Options*, OptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/Options"); } }; // The portal class for org.rocksdb.DBOptions class DBOptionsJni : public RocksDBNativeClass< rocksdb::DBOptions*, DBOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/DBOptions"); } }; class ColumnFamilyDescriptorJni { public: // Get the java class id of org.rocksdb.ColumnFamilyDescriptor static jclass getColumnFamilyDescriptorClass(JNIEnv* env) { jclass jclazz = env->FindClass("org/rocksdb/ColumnFamilyDescriptor"); assert(jclazz != nullptr); return jclazz; } // Get the java method id of columnFamilyName static jmethodID getColumnFamilyNameMethod(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getColumnFamilyDescriptorClass(env), "columnFamilyName", "()[B"); assert(mid != nullptr); return mid; } // Get the java method id of columnFamilyOptions static jmethodID getColumnFamilyOptionsMethod(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getColumnFamilyDescriptorClass(env), "columnFamilyOptions", "()Lorg/rocksdb/ColumnFamilyOptions;"); assert(mid != nullptr); return mid; } }; // The portal class for org.rocksdb.ColumnFamilyOptions class ColumnFamilyOptionsJni : public RocksDBNativeClass< rocksdb::ColumnFamilyOptions*, ColumnFamilyOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/ColumnFamilyOptions"); } }; // The portal class for org.rocksdb.WriteOptions class WriteOptionsJni : public RocksDBNativeClass< rocksdb::WriteOptions*, WriteOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/WriteOptions"); } }; // The portal class for org.rocksdb.ReadOptions class ReadOptionsJni : public RocksDBNativeClass< rocksdb::ReadOptions*, ReadOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/ReadOptions"); } }; // The portal class for org.rocksdb.ReadOptions class WriteBatchJni : public RocksDBNativeClass< rocksdb::WriteBatch*, WriteBatchJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/WriteBatch"); } }; // The portal class for org.rocksdb.WriteBatch.Handler class WriteBatchHandlerJni : public RocksDBNativeClass< const rocksdb::WriteBatchHandlerJniCallback*, WriteBatchHandlerJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/WriteBatch$Handler"); } // Get the java method `put` of org.rocksdb.WriteBatch.Handler. static jmethodID getPutMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "put", "([B[B)V"); assert(mid != nullptr); return mid; } // Get the java method `merge` of org.rocksdb.WriteBatch.Handler. static jmethodID getMergeMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "merge", "([B[B)V"); assert(mid != nullptr); return mid; } // Get the java method `delete` of org.rocksdb.WriteBatch.Handler. static jmethodID getDeleteMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "delete", "([B)V"); assert(mid != nullptr); return mid; } // Get the java method `logData` of org.rocksdb.WriteBatch.Handler. static jmethodID getLogDataMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "logData", "([B)V"); assert(mid != nullptr); return mid; } // Get the java method `shouldContinue` of org.rocksdb.WriteBatch.Handler. static jmethodID getContinueMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "shouldContinue", "()Z"); assert(mid != nullptr); return mid; } }; // The portal class for org.rocksdb.WriteBatchWithIndex class WriteBatchWithIndexJni : public RocksDBNativeClass< rocksdb::WriteBatchWithIndex*, WriteBatchWithIndexJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/WriteBatch"); } }; class HistogramDataJni { public: static jmethodID getConstructorMethodId(JNIEnv* env, jclass jclazz) { static jmethodID mid = env->GetMethodID(jclazz, "", "(DDDDD)V"); assert(mid != nullptr); return mid; } }; // The portal class for org.rocksdb.WriteBatchWithIndex class BackupableDBOptionsJni : public RocksDBNativeClass< rocksdb::BackupableDBOptions*, BackupableDBOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/BackupableDBOptions"); } }; class BackupEngineJni : public RocksDBNativeClass< rocksdb::BackupEngine*, BackupEngineJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/BackupEngine"); } }; // The portal class for org.rocksdb.RocksIterator class IteratorJni : public RocksDBNativeClass< rocksdb::Iterator*, IteratorJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/RocksIterator"); } }; // The portal class for org.rocksdb.Filter class FilterJni : public RocksDBNativeClass< std::shared_ptr*, FilterJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/Filter"); } }; // The portal class for org.rocksdb.ColumnFamilyHandle class ColumnFamilyHandleJni : public RocksDBNativeClass< rocksdb::ColumnFamilyHandle*, ColumnFamilyHandleJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/ColumnFamilyHandle"); } }; // The portal class for org.rocksdb.FlushOptions class FlushOptionsJni : public RocksDBNativeClass< rocksdb::FlushOptions*, FlushOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/FlushOptions"); } }; // The portal class for org.rocksdb.ComparatorOptions class ComparatorOptionsJni : public RocksDBNativeClass< rocksdb::ComparatorJniCallbackOptions*, ComparatorOptionsJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/ComparatorOptions"); } }; // The portal class for org.rocksdb.AbstractComparator class AbstractComparatorJni : public RocksDBNativeClass< const rocksdb::BaseComparatorJniCallback*, AbstractComparatorJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/AbstractComparator"); } // Get the java method `name` of org.rocksdb.Comparator. static jmethodID getNameMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "name", "()Ljava/lang/String;"); assert(mid != nullptr); return mid; } // Get the java method `compare` of org.rocksdb.Comparator. static jmethodID getCompareMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID(getJClass(env), "compare", "(Lorg/rocksdb/AbstractSlice;Lorg/rocksdb/AbstractSlice;)I"); assert(mid != nullptr); return mid; } // Get the java method `findShortestSeparator` of org.rocksdb.Comparator. static jmethodID getFindShortestSeparatorMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID(getJClass(env), "findShortestSeparator", "(Ljava/lang/String;Lorg/rocksdb/AbstractSlice;)Ljava/lang/String;"); assert(mid != nullptr); return mid; } // Get the java method `findShortSuccessor` of org.rocksdb.Comparator. static jmethodID getFindShortSuccessorMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID(getJClass(env), "findShortSuccessor", "(Ljava/lang/String;)Ljava/lang/String;"); assert(mid != nullptr); return mid; } }; // The portal class for org.rocksdb.AbstractSlice class AbstractSliceJni : public RocksDBNativeClass< const rocksdb::Slice*, AbstractSliceJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/AbstractSlice"); } }; class SliceJni { public: // Get the java class id of org.rocksdb.Slice. static jclass getJClass(JNIEnv* env) { jclass jclazz = env->FindClass("org/rocksdb/Slice"); assert(jclazz != nullptr); return jclazz; } static jobject construct0(JNIEnv* env) { static jmethodID mid = env->GetMethodID(getJClass(env), "", "()V"); assert(mid != nullptr); return env->NewObject(getJClass(env), mid); } }; class DirectSliceJni { public: // Get the java class id of org.rocksdb.DirectSlice. static jclass getJClass(JNIEnv* env) { jclass jclazz = env->FindClass("org/rocksdb/DirectSlice"); assert(jclazz != nullptr); return jclazz; } static jobject construct0(JNIEnv* env) { static jmethodID mid = env->GetMethodID(getJClass(env), "", "()V"); assert(mid != nullptr); return env->NewObject(getJClass(env), mid); } }; class ListJni { public: // Get the java class id of java.util.List. static jclass getListClass(JNIEnv* env) { jclass jclazz = env->FindClass("java/util/List"); assert(jclazz != nullptr); return jclazz; } // Get the java class id of java.util.ArrayList. static jclass getArrayListClass(JNIEnv* env) { jclass jclazz = env->FindClass("java/util/ArrayList"); assert(jclazz != nullptr); return jclazz; } // Get the java class id of java.util.Iterator. static jclass getIteratorClass(JNIEnv* env) { jclass jclazz = env->FindClass("java/util/Iterator"); assert(jclazz != nullptr); return jclazz; } // Get the java method id of java.util.List.iterator(). static jmethodID getIteratorMethod(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getListClass(env), "iterator", "()Ljava/util/Iterator;"); assert(mid != nullptr); return mid; } // Get the java method id of java.util.Iterator.hasNext(). static jmethodID getHasNextMethod(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getIteratorClass(env), "hasNext", "()Z"); assert(mid != nullptr); return mid; } // Get the java method id of java.util.Iterator.next(). static jmethodID getNextMethod(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getIteratorClass(env), "next", "()Ljava/lang/Object;"); assert(mid != nullptr); return mid; } // Get the java method id of arrayList constructor. static jmethodID getArrayListConstructorMethodId(JNIEnv* env, jclass jclazz) { static jmethodID mid = env->GetMethodID( jclazz, "", "(I)V"); assert(mid != nullptr); return mid; } // Get the java method id of java.util.List.add(). static jmethodID getListAddMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getListClass(env), "add", "(Ljava/lang/Object;)Z"); assert(mid != nullptr); return mid; } }; class ByteJni { public: // Get the java class id of java.lang.Byte. static jclass getByteClass(JNIEnv* env) { jclass jclazz = env->FindClass("java/lang/Byte"); assert(jclazz != nullptr); return jclazz; } // Get the java method id of java.lang.Byte.byteValue. static jmethodID getByteValueMethod(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getByteClass(env), "byteValue", "()B"); assert(mid != nullptr); return mid; } }; class BackupInfoJni { public: // Get the java class id of org.rocksdb.BackupInfo. static jclass getJClass(JNIEnv* env) { jclass jclazz = env->FindClass("org/rocksdb/BackupInfo"); assert(jclazz != nullptr); return jclazz; } static jobject construct0(JNIEnv* env, uint32_t backup_id, int64_t timestamp, uint64_t size, uint32_t number_files) { static jmethodID mid = env->GetMethodID(getJClass(env), "", "(IJJI)V"); assert(mid != nullptr); return env->NewObject(getJClass(env), mid, backup_id, timestamp, size, number_files); } }; class BackupInfoListJni { public: static jobject getBackupInfo(JNIEnv* env, std::vector backup_infos) { jclass jclazz = env->FindClass("java/util/ArrayList"); jmethodID mid = rocksdb::ListJni::getArrayListConstructorMethodId( env, jclazz); jobject jbackup_info_handle_list = env->NewObject(jclazz, mid, backup_infos.size()); // insert in java list for (std::vector::size_type i = 0; i != backup_infos.size(); i++) { rocksdb::BackupInfo backup_info = backup_infos[i]; jobject obj = rocksdb::BackupInfoJni::construct0(env, backup_info.backup_id, backup_info.timestamp, backup_info.size, backup_info.number_files); env->CallBooleanMethod(jbackup_info_handle_list, rocksdb::ListJni::getListAddMethodId(env), obj); } return jbackup_info_handle_list; } }; class WBWIRocksIteratorJni { public: // Get the java class id of org.rocksdb.WBWIRocksIterator. static jclass getJClass(JNIEnv* env) { static jclass jclazz = env->FindClass("org/rocksdb/WBWIRocksIterator"); assert(jclazz != nullptr); return jclazz; } static jfieldID getWriteEntryField(JNIEnv* env) { static jfieldID fid = env->GetFieldID(getJClass(env), "entry", "Lorg/rocksdb/WBWIRocksIterator$WriteEntry;"); assert(fid != nullptr); return fid; } static jobject getWriteEntry(JNIEnv* env, jobject jwbwi_rocks_iterator) { jobject jwe = env->GetObjectField(jwbwi_rocks_iterator, getWriteEntryField(env)); assert(jwe != nullptr); return jwe; } }; class WriteTypeJni { public: // Get the PUT enum field of org.rocksdb.WBWIRocksIterator.WriteType static jobject PUT(JNIEnv* env) { return getEnum(env, "PUT"); } // Get the MERGE enum field of org.rocksdb.WBWIRocksIterator.WriteType static jobject MERGE(JNIEnv* env) { return getEnum(env, "MERGE"); } // Get the DELETE enum field of org.rocksdb.WBWIRocksIterator.WriteType static jobject DELETE(JNIEnv* env) { return getEnum(env, "DELETE"); } // Get the LOG enum field of org.rocksdb.WBWIRocksIterator.WriteType static jobject LOG(JNIEnv* env) { return getEnum(env, "LOG"); } private: // Get the java class id of org.rocksdb.WBWIRocksIterator.WriteType. static jclass getJClass(JNIEnv* env) { jclass jclazz = env->FindClass("org/rocksdb/WBWIRocksIterator$WriteType"); assert(jclazz != nullptr); return jclazz; } // Get an enum field of org.rocksdb.WBWIRocksIterator.WriteType static jobject getEnum(JNIEnv* env, const char name[]) { jclass jclazz = getJClass(env); jfieldID jfid = env->GetStaticFieldID(jclazz, name, "Lorg/rocksdb/WBWIRocksIterator$WriteType;"); assert(jfid != nullptr); return env->GetStaticObjectField(jclazz, jfid); } }; class WriteEntryJni { public: // Get the java class id of org.rocksdb.WBWIRocksIterator.WriteEntry. static jclass getJClass(JNIEnv* env) { static jclass jclazz = env->FindClass("org/rocksdb/WBWIRocksIterator$WriteEntry"); assert(jclazz != nullptr); return jclazz; } static void setWriteType(JNIEnv* env, jobject jwrite_entry, WriteType write_type) { jobject jwrite_type; switch (write_type) { case kPutRecord: jwrite_type = WriteTypeJni::PUT(env); break; case kMergeRecord: jwrite_type = WriteTypeJni::MERGE(env); break; case kDeleteRecord: jwrite_type = WriteTypeJni::DELETE(env); break; case kLogDataRecord: jwrite_type = WriteTypeJni::LOG(env); break; default: jwrite_type = nullptr; } assert(jwrite_type != nullptr); env->SetObjectField(jwrite_entry, getWriteTypeField(env), jwrite_type); } static void setKey(JNIEnv* env, jobject jwrite_entry, const rocksdb::Slice* slice) { jobject jkey = env->GetObjectField(jwrite_entry, getKeyField(env)); AbstractSliceJni::setHandle(env, jkey, slice); } static void setValue(JNIEnv* env, jobject jwrite_entry, const rocksdb::Slice* slice) { jobject jvalue = env->GetObjectField(jwrite_entry, getValueField(env)); AbstractSliceJni::setHandle(env, jvalue, slice); } private: static jfieldID getWriteTypeField(JNIEnv* env) { static jfieldID fid = env->GetFieldID( getJClass(env), "type", "Lorg/rocksdb/WBWIRocksIterator$WriteType;"); assert(fid != nullptr); return fid; } static jfieldID getKeyField(JNIEnv* env) { static jfieldID fid = env->GetFieldID( getJClass(env), "key", "Lorg/rocksdb/DirectSlice;"); assert(fid != nullptr); return fid; } static jfieldID getValueField(JNIEnv* env) { static jfieldID fid = env->GetFieldID( getJClass(env), "value", "Lorg/rocksdb/DirectSlice;"); assert(fid != nullptr); return fid; } }; class InfoLogLevelJni { public: // Get the DEBUG_LEVEL enum field of org.rocksdb.InfoLogLevel static jobject DEBUG_LEVEL(JNIEnv* env) { return getEnum(env, "DEBUG_LEVEL"); } // Get the INFO_LEVEL enum field of org.rocksdb.InfoLogLevel static jobject INFO_LEVEL(JNIEnv* env) { return getEnum(env, "INFO_LEVEL"); } // Get the WARN_LEVEL enum field of org.rocksdb.InfoLogLevel static jobject WARN_LEVEL(JNIEnv* env) { return getEnum(env, "WARN_LEVEL"); } // Get the ERROR_LEVEL enum field of org.rocksdb.InfoLogLevel static jobject ERROR_LEVEL(JNIEnv* env) { return getEnum(env, "ERROR_LEVEL"); } // Get the FATAL_LEVEL enum field of org.rocksdb.InfoLogLevel static jobject FATAL_LEVEL(JNIEnv* env) { return getEnum(env, "FATAL_LEVEL"); } private: // Get the java class id of org.rocksdb.WBWIRocksIterator.WriteType. static jclass getJClass(JNIEnv* env) { jclass jclazz = env->FindClass("org/rocksdb/InfoLogLevel"); assert(jclazz != nullptr); return jclazz; } // Get an enum field of org.rocksdb.WBWIRocksIterator.WriteType static jobject getEnum(JNIEnv* env, const char name[]) { jclass jclazz = getJClass(env); jfieldID jfid = env->GetStaticFieldID(jclazz, name, "Lorg/rocksdb/InfoLogLevel;"); assert(jfid != nullptr); return env->GetStaticObjectField(jclazz, jfid); } }; // The portal class for org.rocksdb.Logger class LoggerJni : public RocksDBNativeClass< std::shared_ptr*, LoggerJni> { public: static jclass getJClass(JNIEnv* env) { return RocksDBNativeClass::getJClass(env, "org/rocksdb/Logger"); } // Get the java method `name` of org.rocksdb.Logger. static jmethodID getLogMethodId(JNIEnv* env) { static jmethodID mid = env->GetMethodID( getJClass(env), "log", "(Lorg/rocksdb/InfoLogLevel;Ljava/lang/String;)V"); assert(mid != nullptr); return mid; } }; class JniUtil { public: /* * Copies a jstring to a std::string * and releases the original jstring */ static std::string copyString(JNIEnv* env, jstring js) { const char *utf = env->GetStringUTFChars(js, NULL); std::string name(utf); env->ReleaseStringUTFChars(js, utf); return name; } /* * Helper for operations on a key and value * for example WriteBatch->Put * * TODO(AR) could be extended to cover returning rocksdb::Status * from `op` and used for RocksDB->Put etc. */ static void kv_op( std::function op, JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { jbyte* key = env->GetByteArrayElements(jkey, nullptr); jbyte* value = env->GetByteArrayElements(jentry_value, nullptr); rocksdb::Slice key_slice(reinterpret_cast(key), jkey_len); rocksdb::Slice value_slice(reinterpret_cast(value), jentry_value_len); op(key_slice, value_slice); env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); env->ReleaseByteArrayElements(jentry_value, value, JNI_ABORT); } /* * Helper for operations on a key * for example WriteBatch->Delete * * TODO(AR) could be extended to cover returning rocksdb::Status * from `op` and used for RocksDB->Delete etc. */ static void k_op( std::function op, JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len) { jbyte* key = env->GetByteArrayElements(jkey, nullptr); rocksdb::Slice key_slice(reinterpret_cast(key), jkey_len); op(key_slice); env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); } }; } // namespace rocksdb #endif // JAVA_ROCKSJNI_PORTAL_H_ rocksdb-4.5.1/java/rocksjni/ratelimiterjni.cc000066400000000000000000000017441267530634000212600ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for RateLimiter. #include "rocksjni/portal.h" #include "include/org_rocksdb_GenericRateLimiterConfig.h" #include "rocksdb/rate_limiter.h" /* * Class: org_rocksdb_GenericRateLimiterConfig * Method: newRateLimiterHandle * Signature: (JJI)J */ jlong Java_org_rocksdb_GenericRateLimiterConfig_newRateLimiterHandle( JNIEnv* env, jobject jobj, jlong jrate_bytes_per_second, jlong jrefill_period_micros, jint jfairness) { return reinterpret_cast(rocksdb::NewGenericRateLimiter( static_cast(jrate_bytes_per_second), static_cast(jrefill_period_micros), static_cast(jfairness))); } rocksdb-4.5.1/java/rocksjni/remove_emptyvalue_compactionfilterjni.cc000066400000000000000000000022031267530634000261200ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include "include/org_rocksdb_RemoveEmptyValueCompactionFilter.h" #include "utilities/compaction_filters/remove_emptyvalue_compactionfilter.h" /* * Class: org_rocksdb_RemoveEmptyValueCompactionFilter * Method: createNewRemoveEmptyValueCompactionFilter0 * Signature: ()V */ void Java_org_rocksdb_RemoveEmptyValueCompactionFilter_createNewRemoveEmptyValueCompactionFilter0( JNIEnv* env, jobject jobj) { const rocksdb::RemoveEmptyValueCompactionFilter* compaction_filter = new rocksdb::RemoveEmptyValueCompactionFilter(); // set the native handle to our native compaction filter static jclass jclazz = env->FindClass("org/rocksdb/RemoveEmptyValueCompactionFilter"); static jfieldID fid = env->GetFieldID(jclazz, "nativeHandle_", "J"); env->SetLongField(jobj, fid, reinterpret_cast(compaction_filter)); } rocksdb-4.5.1/java/rocksjni/restorejni.cc000066400000000000000000000143011267530634000204130ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::RestoreBackupableDB and rocksdb::RestoreOptions methods // from Java side. #include #include #include #include #include "include/org_rocksdb_RestoreOptions.h" #include "include/org_rocksdb_RestoreBackupableDB.h" #include "rocksjni/portal.h" #include "rocksdb/utilities/backupable_db.h" /* * Class: org_rocksdb_RestoreOptions * Method: newRestoreOptions * Signature: (Z)J */ jlong Java_org_rocksdb_RestoreOptions_newRestoreOptions(JNIEnv* env, jobject jobj, jboolean keep_log_files) { auto ropt = new rocksdb::RestoreOptions(keep_log_files); return reinterpret_cast(ropt); } /* * Class: org_rocksdb_RestoreOptions * Method: dispose * Signature: (J)V */ void Java_org_rocksdb_RestoreOptions_dispose(JNIEnv* env, jobject jobj, jlong jhandle) { auto ropt = reinterpret_cast(jhandle); assert(ropt); delete ropt; } /* * Class: org_rocksdb_RestoreBackupableDB * Method: newRestoreBackupableDB * Signature: (J)J */ jlong Java_org_rocksdb_RestoreBackupableDB_newRestoreBackupableDB(JNIEnv* env, jobject jobj, jlong jopt_handle) { auto opt = reinterpret_cast(jopt_handle); auto rdb = new rocksdb::RestoreBackupableDB(rocksdb::Env::Default(), *opt); return reinterpret_cast(rdb); } /* * Class: org_rocksdb_RestoreBackupableDB * Method: restoreDBFromBackup0 * Signature: (JJLjava/lang/String;Ljava/lang/String;J)V */ void Java_org_rocksdb_RestoreBackupableDB_restoreDBFromBackup0(JNIEnv* env, jobject jobj, jlong jhandle, jlong jbackup_id, jstring jdb_dir, jstring jwal_dir, jlong jopt_handle) { auto opt = reinterpret_cast(jopt_handle); const char* cdb_dir = env->GetStringUTFChars(jdb_dir, 0); const char* cwal_dir = env->GetStringUTFChars(jwal_dir, 0); auto rdb = reinterpret_cast(jhandle); rocksdb::Status s = rdb->RestoreDBFromBackup( static_cast(jbackup_id), cdb_dir, cwal_dir, *opt); env->ReleaseStringUTFChars(jdb_dir, cdb_dir); env->ReleaseStringUTFChars(jwal_dir, cwal_dir); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RestoreBackupableDB * Method: restoreDBFromLatestBackup0 * Signature: (JLjava/lang/String;Ljava/lang/String;J)V */ void Java_org_rocksdb_RestoreBackupableDB_restoreDBFromLatestBackup0( JNIEnv* env, jobject jobj, jlong jhandle, jstring jdb_dir, jstring jwal_dir, jlong jopt_handle) { auto opt = reinterpret_cast(jopt_handle); const char* cdb_dir = env->GetStringUTFChars(jdb_dir, 0); const char* cwal_dir = env->GetStringUTFChars(jwal_dir, 0); auto rdb = reinterpret_cast(jhandle); rocksdb::Status s = rdb->RestoreDBFromLatestBackup(cdb_dir, cwal_dir, *opt); env->ReleaseStringUTFChars(jdb_dir, cdb_dir); env->ReleaseStringUTFChars(jwal_dir, cwal_dir); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RestoreBackupableDB * Method: purgeOldBackups0 * Signature: (JI)V */ void Java_org_rocksdb_RestoreBackupableDB_purgeOldBackups0(JNIEnv* env, jobject jobj, jlong jhandle, jint jnum_backups_to_keep) { auto rdb = reinterpret_cast(jhandle); rocksdb::Status s = rdb->PurgeOldBackups(jnum_backups_to_keep); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RestoreBackupableDB * Method: deleteBackup0 * Signature: (JI)V */ void Java_org_rocksdb_RestoreBackupableDB_deleteBackup0(JNIEnv* env, jobject jobj, jlong jhandle, jint jbackup_id) { auto rdb = reinterpret_cast(jhandle); rocksdb::Status s = rdb->DeleteBackup(jbackup_id); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RestoreBackupableDB * Method: getBackupInfo * Signature: (J)Ljava/util/List; */ jobject Java_org_rocksdb_RestoreBackupableDB_getBackupInfo( JNIEnv* env, jobject jbdb, jlong jhandle) { std::vector backup_infos; reinterpret_cast(jhandle)-> GetBackupInfo(&backup_infos); return rocksdb::BackupInfoListJni::getBackupInfo(env, backup_infos); } /* * Class: org_rocksdb_RestoreBackupableDB * Method: getCorruptedBackups * Signature: (J)[I; */ jintArray Java_org_rocksdb_RestoreBackupableDB_getCorruptedBackups( JNIEnv* env, jobject jbdb, jlong jhandle) { std::vector backup_ids; reinterpret_cast(jhandle)-> GetCorruptedBackups(&backup_ids); // store backupids in int array std::vector int_backup_ids(backup_ids.begin(), backup_ids.end()); // Store ints in java array jintArray ret_backup_ids; // Its ok to loose precision here (64->32) jsize ret_backup_ids_size = static_cast(backup_ids.size()); ret_backup_ids = env->NewIntArray(ret_backup_ids_size); env->SetIntArrayRegion(ret_backup_ids, 0, ret_backup_ids_size, int_backup_ids.data()); return ret_backup_ids; } /* * Class: org_rocksdb_RestoreBackupableDB * Method: garbageCollect * Signature: (J)V */ void Java_org_rocksdb_RestoreBackupableDB_garbageCollect( JNIEnv* env, jobject jobj, jlong jhandle) { auto db = reinterpret_cast( jhandle); rocksdb::Status s = db->GarbageCollect(); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RestoreBackupableDB * Method: dispose * Signature: (J)V */ void Java_org_rocksdb_RestoreBackupableDB_dispose(JNIEnv* env, jobject jobj, jlong jhandle) { auto ropt = reinterpret_cast(jhandle); assert(ropt); delete ropt; } rocksdb-4.5.1/java/rocksjni/rocksjni.cc000066400000000000000000001610431267530634000200570ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::DB methods from Java side. #include #include #include #include #include #include #include #include "include/org_rocksdb_RocksDB.h" #include "rocksdb/db.h" #include "rocksdb/cache.h" #include "rocksdb/types.h" #include "rocksjni/portal.h" #ifdef min #undef min #endif ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Open /* * Class: org_rocksdb_RocksDB * Method: open * Signature: (JLjava/lang/String;)V */ void Java_org_rocksdb_RocksDB_open__JLjava_lang_String_2( JNIEnv* env, jobject jdb, jlong jopt_handle, jstring jdb_path) { auto opt = reinterpret_cast(jopt_handle); rocksdb::DB* db = nullptr; const char* db_path = env->GetStringUTFChars(jdb_path, 0); rocksdb::Status s = rocksdb::DB::Open(*opt, db_path, &db); env->ReleaseStringUTFChars(jdb_path, db_path); if (s.ok()) { rocksdb::RocksDBJni::setHandle(env, jdb, db); return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksDB * Method: openROnly * Signature: (JLjava/lang/String;)V */ void Java_org_rocksdb_RocksDB_openROnly__JLjava_lang_String_2( JNIEnv* env, jobject jdb, jlong jopt_handle, jstring jdb_path) { auto opt = reinterpret_cast(jopt_handle); rocksdb::DB* db = nullptr; const char* db_path = env->GetStringUTFChars(jdb_path, 0); rocksdb::Status s = rocksdb::DB::OpenForReadOnly(*opt, db_path, &db); env->ReleaseStringUTFChars(jdb_path, db_path); if (s.ok()) { rocksdb::RocksDBJni::setHandle(env, jdb, db); return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksDB * Method: openROnly * Signature: (JLjava/lang/String;Ljava/util/List;I)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_openROnly__JLjava_lang_String_2Ljava_util_List_2I( JNIEnv* env, jobject jdb, jlong jopt_handle, jstring jdb_path, jobject jcfdesc_list, jint jcfdesc_count) { auto opt = reinterpret_cast(jopt_handle); rocksdb::DB* db = nullptr; const char* db_path = env->GetStringUTFChars(jdb_path, 0); std::vector cfnames_to_free; std::vector jcfnames_for_free; std::vector column_families; std::vector handles; // get iterator for ColumnFamilyDescriptors jobject iteratorObj = env->CallObjectMethod( jcfdesc_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over ColumnFamilyDescriptors while (env->CallBooleanMethod( iteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { // get ColumnFamilyDescriptor jobject jcf_descriptor = env->CallObjectMethod(iteratorObj, rocksdb::ListJni::getNextMethod(env)); // get ColumnFamilyName jbyteArray cf_name_in_byte_array = static_cast( env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyNameMethod( env))); // get CF Options jobject jcf_opt_obj = env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyOptionsMethod( env)); rocksdb::ColumnFamilyOptions* cfOptions = rocksdb::ColumnFamilyOptionsJni::getHandle(env, jcf_opt_obj); jbyte* cfname = env->GetByteArrayElements(cf_name_in_byte_array, 0); const int len = env->GetArrayLength(cf_name_in_byte_array); // free allocated cfnames after call to open cfnames_to_free.push_back(cfname); jcfnames_for_free.push_back(cf_name_in_byte_array); column_families.push_back(rocksdb::ColumnFamilyDescriptor( std::string(reinterpret_cast(cfname), len), *cfOptions)); } rocksdb::Status s = rocksdb::DB::OpenForReadOnly(*opt, db_path, column_families, &handles, &db); env->ReleaseStringUTFChars(jdb_path, db_path); // free jbyte allocations for (std::vector::size_type i = 0; i != cfnames_to_free.size(); i++) { // free cfnames env->ReleaseByteArrayElements(jcfnames_for_free[i], cfnames_to_free[i], 0); } // check if open operation was successful if (s.ok()) { rocksdb::RocksDBJni::setHandle(env, jdb, db); jclass jListClazz = env->FindClass("java/util/ArrayList"); jmethodID midList = rocksdb::ListJni::getArrayListConstructorMethodId( env, jListClazz); jobject jcfhandle_list = env->NewObject(jListClazz, midList, handles.size()); // insert in java list for (std::vector::size_type i = 0; i != handles.size(); i++) { // jlong must be converted to Long due to collections restrictions jclass jLongClazz = env->FindClass("java/lang/Long"); jmethodID midLong = env->GetMethodID(jLongClazz, "", "(J)V"); jobject obj = env->NewObject(jLongClazz, midLong, reinterpret_cast(handles[i])); env->CallBooleanMethod(jcfhandle_list, rocksdb::ListJni::getListAddMethodId(env), obj); } return jcfhandle_list; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return nullptr; } /* * Class: org_rocksdb_RocksDB * Method: open * Signature: (JLjava/lang/String;Ljava/util/List;I)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_open__JLjava_lang_String_2Ljava_util_List_2I( JNIEnv* env, jobject jdb, jlong jopt_handle, jstring jdb_path, jobject jcfdesc_list, jint jcfdesc_count) { auto opt = reinterpret_cast(jopt_handle); rocksdb::DB* db = nullptr; const char* db_path = env->GetStringUTFChars(jdb_path, 0); std::vector cfnames_to_free; std::vector jcfnames_for_free; std::vector column_families; std::vector handles; // get iterator for ColumnFamilyDescriptors jobject iteratorObj = env->CallObjectMethod( jcfdesc_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over ColumnFamilyDescriptors while (env->CallBooleanMethod( iteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { // get ColumnFamilyDescriptor jobject jcf_descriptor = env->CallObjectMethod(iteratorObj, rocksdb::ListJni::getNextMethod(env)); // get ColumnFamilyName jbyteArray cf_name_in_byte_array = static_cast( env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyNameMethod( env))); // get CF Options jobject jcf_opt_obj = env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyOptionsMethod( env)); rocksdb::ColumnFamilyOptions* cfOptions = rocksdb::ColumnFamilyOptionsJni::getHandle(env, jcf_opt_obj); jbyte* cfname = env->GetByteArrayElements(cf_name_in_byte_array, 0); const int len = env->GetArrayLength(cf_name_in_byte_array); // free allocated cfnames after call to open cfnames_to_free.push_back(cfname); jcfnames_for_free.push_back(cf_name_in_byte_array); column_families.push_back(rocksdb::ColumnFamilyDescriptor( std::string(reinterpret_cast(cfname), len), *cfOptions)); } rocksdb::Status s = rocksdb::DB::Open(*opt, db_path, column_families, &handles, &db); env->ReleaseStringUTFChars(jdb_path, db_path); // free jbyte allocations for (std::vector::size_type i = 0; i != cfnames_to_free.size(); i++) { // free cfnames env->ReleaseByteArrayElements(jcfnames_for_free[i], cfnames_to_free[i], 0); } // check if open operation was successful if (s.ok()) { rocksdb::RocksDBJni::setHandle(env, jdb, db); jclass jListClazz = env->FindClass("java/util/ArrayList"); jmethodID midList = rocksdb::ListJni::getArrayListConstructorMethodId( env, jListClazz); jobject jcfhandle_list = env->NewObject(jListClazz, midList, handles.size()); // insert in java list for (std::vector::size_type i = 0; i != handles.size(); i++) { // jlong must be converted to Long due to collections restrictions jclass jLongClazz = env->FindClass("java/lang/Long"); jmethodID midLong = env->GetMethodID(jLongClazz, "", "(J)V"); jobject obj = env->NewObject(jLongClazz, midLong, reinterpret_cast(handles[i])); env->CallBooleanMethod(jcfhandle_list, rocksdb::ListJni::getListAddMethodId(env), obj); } return jcfhandle_list; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return nullptr; } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::ListColumnFamilies /* * Class: org_rocksdb_RocksDB * Method: listColumnFamilies * Signature: (JLjava/lang/String;)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_listColumnFamilies( JNIEnv* env, jclass jclazz, jlong jopt_handle, jstring jdb_path) { std::vector column_family_names; auto opt = reinterpret_cast(jopt_handle); const char* db_path = env->GetStringUTFChars(jdb_path, 0); jobject jvalue_list = nullptr; rocksdb::Status s = rocksdb::DB::ListColumnFamilies(*opt, db_path, &column_family_names); env->ReleaseStringUTFChars(jdb_path, db_path); if (s.ok()) { // Don't reuse class pointer jclass jListClazz = env->FindClass("java/util/ArrayList"); jmethodID mid = rocksdb::ListJni::getArrayListConstructorMethodId(env, jListClazz); jvalue_list = env->NewObject(jListClazz, mid, column_family_names.size()); for (std::vector::size_type i = 0; i < column_family_names.size(); i++) { jbyteArray jcf_value = env->NewByteArray(static_cast(column_family_names[i].size())); env->SetByteArrayRegion( jcf_value, 0, static_cast(column_family_names[i].size()), reinterpret_cast(column_family_names[i].data())); env->CallBooleanMethod(jvalue_list, rocksdb::ListJni::getListAddMethodId(env), jcf_value); } } return jvalue_list; } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Put void rocksdb_put_helper( JNIEnv* env, rocksdb::DB* db, const rocksdb::WriteOptions& write_options, rocksdb::ColumnFamilyHandle* cf_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { jbyte* key = env->GetByteArrayElements(jkey, 0); jbyte* value = env->GetByteArrayElements(jentry_value, 0); rocksdb::Slice key_slice(reinterpret_cast(key), jkey_len); rocksdb::Slice value_slice(reinterpret_cast(value), jentry_value_len); rocksdb::Status s; if (cf_handle != nullptr) { s = db->Put(write_options, cf_handle, key_slice, value_slice); } else { // backwards compatibility s = db->Put(write_options, key_slice, value_slice); } // trigger java unref on key and value. // by passing JNI_ABORT, it will simply release the reference without // copying the result back to the java byte array. env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); env->ReleaseByteArrayElements(jentry_value, value, JNI_ABORT); if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksDB * Method: put * Signature: (J[BI[BI)V */ void Java_org_rocksdb_RocksDB_put__J_3BI_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto db = reinterpret_cast(jdb_handle); static const rocksdb::WriteOptions default_write_options = rocksdb::WriteOptions(); rocksdb_put_helper(env, db, default_write_options, nullptr, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_RocksDB * Method: put * Signature: (J[BI[BIJ)V */ void Java_org_rocksdb_RocksDB_put__J_3BI_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); static const rocksdb::WriteOptions default_write_options = rocksdb::WriteOptions(); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { rocksdb_put_helper(env, db, default_write_options, cf_handle, jkey, jkey_len, jentry_value, jentry_value_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } } /* * Class: org_rocksdb_RocksDB * Method: put * Signature: (JJ[BI[BI)V */ void Java_org_rocksdb_RocksDB_put__JJ_3BI_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jwrite_options_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto db = reinterpret_cast(jdb_handle); auto write_options = reinterpret_cast( jwrite_options_handle); rocksdb_put_helper(env, db, *write_options, nullptr, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_RocksDB * Method: put * Signature: (JJ[BI[BIJ)V */ void Java_org_rocksdb_RocksDB_put__JJ_3BI_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jwrite_options_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); auto write_options = reinterpret_cast( jwrite_options_handle); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { rocksdb_put_helper(env, db, *write_options, cf_handle, jkey, jkey_len, jentry_value, jentry_value_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Write /* * Class: org_rocksdb_RocksDB * Method: write0 * Signature: (JJ)V */ void Java_org_rocksdb_RocksDB_write0( JNIEnv* env, jobject jdb, jlong jwrite_options_handle, jlong jwb_handle) { rocksdb::DB* db = rocksdb::RocksDBJni::getHandle(env, jdb); auto* write_options = reinterpret_cast( jwrite_options_handle); auto* wb = reinterpret_cast(jwb_handle); rocksdb::Status s = db->Write(*write_options, wb); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RocksDB * Method: write1 * Signature: (JJ)V */ void Java_org_rocksdb_RocksDB_write1( JNIEnv* env, jobject jdb, jlong jwrite_options_handle, jlong jwbwi_handle) { rocksdb::DB* db = rocksdb::RocksDBJni::getHandle(env, jdb); auto* write_options = reinterpret_cast( jwrite_options_handle); auto* wbwi = reinterpret_cast(jwbwi_handle); auto* wb = wbwi->GetWriteBatch(); rocksdb::Status s = db->Write(*write_options, wb); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::KeyMayExist jboolean key_may_exist_helper(JNIEnv* env, rocksdb::DB* db, const rocksdb::ReadOptions& read_opt, rocksdb::ColumnFamilyHandle* cf_handle, jbyteArray jkey, jint jkey_len, jobject jstring_buffer) { std::string value; bool value_found = false; jboolean isCopy; jbyte* key = env->GetByteArrayElements(jkey, &isCopy); rocksdb::Slice key_slice(reinterpret_cast(key), jkey_len); bool keyMayExist; if (cf_handle != nullptr) { keyMayExist = db->KeyMayExist(read_opt, cf_handle, key_slice, &value, &value_found); } else { keyMayExist = db->KeyMayExist(read_opt, key_slice, &value, &value_found); } if (value_found && !value.empty()) { jclass clazz = env->GetObjectClass(jstring_buffer); jmethodID mid = env->GetMethodID(clazz, "append", "(Ljava/lang/String;)Ljava/lang/StringBuffer;"); jstring new_value_str = env->NewStringUTF(value.c_str()); env->CallObjectMethod(jstring_buffer, mid, new_value_str); } env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); return static_cast(keyMayExist); } /* * Class: org_rocksdb_RocksDB * Method: keyMayExist * Signature: ([BILjava/lang/StringBuffer;)Z */ jboolean Java_org_rocksdb_RocksDB_keyMayExist___3BILjava_lang_StringBuffer_2( JNIEnv* env, jobject jdb, jbyteArray jkey, jint jkey_len, jobject jstring_buffer) { rocksdb::DB* db = rocksdb::RocksDBJni::getHandle(env, jdb); return key_may_exist_helper(env, db, rocksdb::ReadOptions(), nullptr, jkey, jkey_len, jstring_buffer); } /* * Class: org_rocksdb_RocksDB * Method: keyMayExist * Signature: ([BIJLjava/lang/StringBuffer;)Z */ jboolean Java_org_rocksdb_RocksDB_keyMayExist___3BIJLjava_lang_StringBuffer_2( JNIEnv* env, jobject jdb, jbyteArray jkey, jint jkey_len, jlong jcf_handle, jobject jstring_buffer) { rocksdb::DB* db = rocksdb::RocksDBJni::getHandle(env, jdb); auto cf_handle = reinterpret_cast( jcf_handle); if (cf_handle != nullptr) { return key_may_exist_helper(env, db, rocksdb::ReadOptions(), cf_handle, jkey, jkey_len, jstring_buffer); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } return true; } /* * Class: org_rocksdb_RocksDB * Method: keyMayExist * Signature: (J[BILjava/lang/StringBuffer;)Z */ jboolean Java_org_rocksdb_RocksDB_keyMayExist__J_3BILjava_lang_StringBuffer_2( JNIEnv* env, jobject jdb, jlong jread_options_handle, jbyteArray jkey, jint jkey_len, jobject jstring_buffer) { rocksdb::DB* db = rocksdb::RocksDBJni::getHandle(env, jdb); auto& read_options = *reinterpret_cast( jread_options_handle); return key_may_exist_helper(env, db, read_options, nullptr, jkey, jkey_len, jstring_buffer); } /* * Class: org_rocksdb_RocksDB * Method: keyMayExist * Signature: (J[BIJLjava/lang/StringBuffer;)Z */ jboolean Java_org_rocksdb_RocksDB_keyMayExist__J_3BIJLjava_lang_StringBuffer_2( JNIEnv* env, jobject jdb, jlong jread_options_handle, jbyteArray jkey, jint jkey_len, jlong jcf_handle, jobject jstring_buffer) { rocksdb::DB* db = rocksdb::RocksDBJni::getHandle(env, jdb); auto& read_options = *reinterpret_cast( jread_options_handle); auto cf_handle = reinterpret_cast( jcf_handle); if (cf_handle != nullptr) { return key_may_exist_helper(env, db, read_options, cf_handle, jkey, jkey_len, jstring_buffer); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } return true; } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Get jbyteArray rocksdb_get_helper( JNIEnv* env, rocksdb::DB* db, const rocksdb::ReadOptions& read_opt, rocksdb::ColumnFamilyHandle* column_family_handle, jbyteArray jkey, jint jkey_len) { jboolean isCopy; jbyte* key = env->GetByteArrayElements(jkey, &isCopy); rocksdb::Slice key_slice( reinterpret_cast(key), jkey_len); std::string value; rocksdb::Status s; if (column_family_handle != nullptr) { s = db->Get(read_opt, column_family_handle, key_slice, &value); } else { // backwards compatibility s = db->Get(read_opt, key_slice, &value); } // trigger java unref on key. // by passing JNI_ABORT, it will simply release the reference without // copying the result back to the java byte array. env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); if (s.IsNotFound()) { return nullptr; } if (s.ok()) { jbyteArray jret_value = env->NewByteArray(static_cast(value.size())); env->SetByteArrayRegion(jret_value, 0, static_cast(value.size()), reinterpret_cast(value.c_str())); return jret_value; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return nullptr; } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (J[BI)[B */ jbyteArray Java_org_rocksdb_RocksDB_get__J_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len) { return rocksdb_get_helper(env, reinterpret_cast(jdb_handle), rocksdb::ReadOptions(), nullptr, jkey, jkey_len); } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (J[BIJ)[B */ jbyteArray Java_org_rocksdb_RocksDB_get__J_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jlong jcf_handle) { auto db_handle = reinterpret_cast(jdb_handle); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { return rocksdb_get_helper(env, db_handle, rocksdb::ReadOptions(), cf_handle, jkey, jkey_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); // will never be evaluated return env->NewByteArray(0); } } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (JJ[BI)[B */ jbyteArray Java_org_rocksdb_RocksDB_get__JJ_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jropt_handle, jbyteArray jkey, jint jkey_len) { return rocksdb_get_helper(env, reinterpret_cast(jdb_handle), *reinterpret_cast(jropt_handle), nullptr, jkey, jkey_len); } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (JJ[BIJ)[B */ jbyteArray Java_org_rocksdb_RocksDB_get__JJ_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jropt_handle, jbyteArray jkey, jint jkey_len, jlong jcf_handle) { auto db_handle = reinterpret_cast(jdb_handle); auto& ro_opt = *reinterpret_cast(jropt_handle); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { return rocksdb_get_helper(env, db_handle, ro_opt, cf_handle, jkey, jkey_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); // will never be evaluated return env->NewByteArray(0); } } jint rocksdb_get_helper( JNIEnv* env, rocksdb::DB* db, const rocksdb::ReadOptions& read_options, rocksdb::ColumnFamilyHandle* column_family_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { static const int kNotFound = -1; static const int kStatusError = -2; jbyte* key = env->GetByteArrayElements(jkey, 0); rocksdb::Slice key_slice( reinterpret_cast(key), jkey_len); // TODO(yhchiang): we might save one memory allocation here by adding // a DB::Get() function which takes preallocated jbyte* as input. std::string cvalue; rocksdb::Status s; if (column_family_handle != nullptr) { s = db->Get(read_options, column_family_handle, key_slice, &cvalue); } else { // backwards compatibility s = db->Get(read_options, key_slice, &cvalue); } // trigger java unref on key. // by passing JNI_ABORT, it will simply release the reference without // copying the result back to the java byte array. env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); if (s.IsNotFound()) { return kNotFound; } else if (!s.ok()) { // Here since we are throwing a Java exception from c++ side. // As a result, c++ does not know calling this function will in fact // throwing an exception. As a result, the execution flow will // not stop here, and codes after this throw will still be // executed. rocksdb::RocksDBExceptionJni::ThrowNew(env, s); // Return a dummy const value to avoid compilation error, although // java side might not have a chance to get the return value :) return kStatusError; } jint cvalue_len = static_cast(cvalue.size()); jint length = std::min(jentry_value_len, cvalue_len); env->SetByteArrayRegion( jentry_value, 0, length, reinterpret_cast(cvalue.c_str())); return cvalue_len; } // cf multi get jobject multi_get_helper(JNIEnv* env, jobject jdb, rocksdb::DB* db, const rocksdb::ReadOptions& rOpt, jobject jkey_list, jint jkeys_count, jobject jcfhandle_list) { std::vector keys; std::vector keys_to_free; std::vector cf_handles; if (jcfhandle_list != nullptr) { // get cf iterator jobject cfIteratorObj = env->CallObjectMethod( jcfhandle_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over keys and convert java byte array to slice while (env->CallBooleanMethod( cfIteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { jobject jobj = (jbyteArray) env->CallObjectMethod( cfIteratorObj, rocksdb::ListJni::getNextMethod(env)); rocksdb::ColumnFamilyHandle* cfHandle = rocksdb::ColumnFamilyHandleJni::getHandle(env, jobj); cf_handles.push_back(cfHandle); } } // Process key list // get iterator jobject iteratorObj = env->CallObjectMethod( jkey_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over keys and convert java byte array to slice while (env->CallBooleanMethod( iteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { jbyteArray jkey = (jbyteArray) env->CallObjectMethod( iteratorObj, rocksdb::ListJni::getNextMethod(env)); jint key_length = env->GetArrayLength(jkey); jbyte* key = new jbyte[key_length]; env->GetByteArrayRegion(jkey, 0, key_length, key); // store allocated jbyte to free it after multiGet call keys_to_free.push_back(key); rocksdb::Slice key_slice( reinterpret_cast(key), key_length); keys.push_back(key_slice); } std::vector values; std::vector s; if (cf_handles.size() == 0) { s = db->MultiGet(rOpt, keys, &values); } else { s = db->MultiGet(rOpt, cf_handles, keys, &values); } // Don't reuse class pointer jclass jclazz = env->FindClass("java/util/ArrayList"); jmethodID mid = rocksdb::ListJni::getArrayListConstructorMethodId( env, jclazz); jobject jvalue_list = env->NewObject(jclazz, mid, jkeys_count); // insert in java list for (std::vector::size_type i = 0; i != s.size(); i++) { if (s[i].ok()) { jbyteArray jentry_value = env->NewByteArray(static_cast(values[i].size())); env->SetByteArrayRegion( jentry_value, 0, static_cast(values[i].size()), reinterpret_cast(values[i].c_str())); env->CallBooleanMethod( jvalue_list, rocksdb::ListJni::getListAddMethodId(env), jentry_value); } else { env->CallBooleanMethod( jvalue_list, rocksdb::ListJni::getListAddMethodId(env), nullptr); } } // free up allocated byte arrays for (std::vector::size_type i = 0; i != keys_to_free.size(); i++) { delete[] keys_to_free[i]; } keys_to_free.clear(); return jvalue_list; } /* * Class: org_rocksdb_RocksDB * Method: multiGet * Signature: (JLjava/util/List;I)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_multiGet__JLjava_util_List_2I( JNIEnv* env, jobject jdb, jlong jdb_handle, jobject jkey_list, jint jkeys_count) { return multi_get_helper(env, jdb, reinterpret_cast(jdb_handle), rocksdb::ReadOptions(), jkey_list, jkeys_count, nullptr); } /* * Class: org_rocksdb_RocksDB * Method: multiGet * Signature: (JLjava/util/List;ILjava/util/List;)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_multiGet__JLjava_util_List_2ILjava_util_List_2( JNIEnv* env, jobject jdb, jlong jdb_handle, jobject jkey_list, jint jkeys_count, jobject jcfhandle_list) { return multi_get_helper(env, jdb, reinterpret_cast(jdb_handle), rocksdb::ReadOptions(), jkey_list, jkeys_count, jcfhandle_list); } /* * Class: org_rocksdb_RocksDB * Method: multiGet * Signature: (JJLjava/util/List;I)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_multiGet__JJLjava_util_List_2I( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jropt_handle, jobject jkey_list, jint jkeys_count) { return multi_get_helper(env, jdb, reinterpret_cast(jdb_handle), *reinterpret_cast(jropt_handle), jkey_list, jkeys_count, nullptr); } /* * Class: org_rocksdb_RocksDB * Method: multiGet * Signature: (JJLjava/util/List;ILjava/util/List;)Ljava/util/List; */ jobject Java_org_rocksdb_RocksDB_multiGet__JJLjava_util_List_2ILjava_util_List_2( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jropt_handle, jobject jkey_list, jint jkeys_count, jobject jcfhandle_list) { return multi_get_helper(env, jdb, reinterpret_cast(jdb_handle), *reinterpret_cast(jropt_handle), jkey_list, jkeys_count, jcfhandle_list); } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (J[BI[BI)I */ jint Java_org_rocksdb_RocksDB_get__J_3BI_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { return rocksdb_get_helper(env, reinterpret_cast(jdb_handle), rocksdb::ReadOptions(), nullptr, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (J[BI[BIJ)I */ jint Java_org_rocksdb_RocksDB_get__J_3BI_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto db_handle = reinterpret_cast(jdb_handle); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { return rocksdb_get_helper(env, db_handle, rocksdb::ReadOptions(), cf_handle, jkey, jkey_len, jentry_value, jentry_value_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); // will never be evaluated return 0; } } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (JJ[BI[BI)I */ jint Java_org_rocksdb_RocksDB_get__JJ_3BI_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jropt_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { return rocksdb_get_helper(env, reinterpret_cast(jdb_handle), *reinterpret_cast(jropt_handle), nullptr, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_RocksDB * Method: get * Signature: (JJ[BI[BIJ)I */ jint Java_org_rocksdb_RocksDB_get__JJ_3BI_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jropt_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto db_handle = reinterpret_cast(jdb_handle); auto& ro_opt = *reinterpret_cast(jropt_handle); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { return rocksdb_get_helper(env, db_handle, ro_opt, cf_handle, jkey, jkey_len, jentry_value, jentry_value_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); // will never be evaluated return 0; } } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Delete() void rocksdb_remove_helper( JNIEnv* env, rocksdb::DB* db, const rocksdb::WriteOptions& write_options, rocksdb::ColumnFamilyHandle* cf_handle, jbyteArray jkey, jint jkey_len) { jbyte* key = env->GetByteArrayElements(jkey, 0); rocksdb::Slice key_slice(reinterpret_cast(key), jkey_len); rocksdb::Status s; if (cf_handle != nullptr) { s = db->Delete(write_options, cf_handle, key_slice); } else { // backwards compatibility s = db->Delete(write_options, key_slice); } // trigger java unref on key and value. // by passing JNI_ABORT, it will simply release the reference without // copying the result back to the java byte array. env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } return; } /* * Class: org_rocksdb_RocksDB * Method: remove * Signature: (J[BI)V */ void Java_org_rocksdb_RocksDB_remove__J_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len) { auto db = reinterpret_cast(jdb_handle); static const rocksdb::WriteOptions default_write_options = rocksdb::WriteOptions(); rocksdb_remove_helper(env, db, default_write_options, nullptr, jkey, jkey_len); } /* * Class: org_rocksdb_RocksDB * Method: remove * Signature: (J[BIJ)V */ void Java_org_rocksdb_RocksDB_remove__J_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); static const rocksdb::WriteOptions default_write_options = rocksdb::WriteOptions(); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { rocksdb_remove_helper(env, db, default_write_options, cf_handle, jkey, jkey_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } } /* * Class: org_rocksdb_RocksDB * Method: remove * Signature: (JJ[BIJ)V */ void Java_org_rocksdb_RocksDB_remove__JJ_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jwrite_options, jbyteArray jkey, jint jkey_len) { auto db = reinterpret_cast(jdb_handle); auto write_options = reinterpret_cast(jwrite_options); rocksdb_remove_helper(env, db, *write_options, nullptr, jkey, jkey_len); } /* * Class: org_rocksdb_RocksDB * Method: remove * Signature: (JJ[BIJ)V */ void Java_org_rocksdb_RocksDB_remove__JJ_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jwrite_options, jbyteArray jkey, jint jkey_len, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); auto write_options = reinterpret_cast(jwrite_options); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { rocksdb_remove_helper(env, db, *write_options, cf_handle, jkey, jkey_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Merge void rocksdb_merge_helper( JNIEnv* env, rocksdb::DB* db, const rocksdb::WriteOptions& write_options, rocksdb::ColumnFamilyHandle* cf_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { jbyte* key = env->GetByteArrayElements(jkey, 0); jbyte* value = env->GetByteArrayElements(jentry_value, 0); rocksdb::Slice key_slice(reinterpret_cast(key), jkey_len); rocksdb::Slice value_slice(reinterpret_cast(value), jentry_value_len); rocksdb::Status s; if (cf_handle != nullptr) { s = db->Merge(write_options, cf_handle, key_slice, value_slice); } else { s = db->Merge(write_options, key_slice, value_slice); } // trigger java unref on key and value. // by passing JNI_ABORT, it will simply release the reference without // copying the result back to the java byte array. env->ReleaseByteArrayElements(jkey, key, JNI_ABORT); env->ReleaseByteArrayElements(jentry_value, value, JNI_ABORT); if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksDB * Method: merge * Signature: (J[BI[BI)V */ void Java_org_rocksdb_RocksDB_merge__J_3BI_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto db = reinterpret_cast(jdb_handle); static const rocksdb::WriteOptions default_write_options = rocksdb::WriteOptions(); rocksdb_merge_helper(env, db, default_write_options, nullptr, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_RocksDB * Method: merge * Signature: (J[BI[BIJ)V */ void Java_org_rocksdb_RocksDB_merge__J_3BI_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); static const rocksdb::WriteOptions default_write_options = rocksdb::WriteOptions(); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { rocksdb_merge_helper(env, db, default_write_options, cf_handle, jkey, jkey_len, jentry_value, jentry_value_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } } /* * Class: org_rocksdb_RocksDB * Method: merge * Signature: (JJ[BI[BI)V */ void Java_org_rocksdb_RocksDB_merge__JJ_3BI_3BI( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jwrite_options_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto db = reinterpret_cast(jdb_handle); auto write_options = reinterpret_cast( jwrite_options_handle); rocksdb_merge_helper(env, db, *write_options, nullptr, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_RocksDB * Method: merge * Signature: (JJ[BI[BIJ)V */ void Java_org_rocksdb_RocksDB_merge__JJ_3BI_3BIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jwrite_options_handle, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); auto write_options = reinterpret_cast( jwrite_options_handle); auto cf_handle = reinterpret_cast(jcf_handle); if (cf_handle != nullptr) { rocksdb_merge_helper(env, db, *write_options, cf_handle, jkey, jkey_len, jentry_value, jentry_value_len); } else { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::InvalidArgument("Invalid ColumnFamilyHandle.")); } } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::~DB() /* * Class: org_rocksdb_RocksDB * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_RocksDB_disposeInternal( JNIEnv* env, jobject java_db, jlong jhandle) { delete reinterpret_cast(jhandle); } jlong rocksdb_iterator_helper( rocksdb::DB* db, rocksdb::ReadOptions read_options, rocksdb::ColumnFamilyHandle* cf_handle) { rocksdb::Iterator* iterator = nullptr; if (cf_handle != nullptr) { iterator = db->NewIterator(read_options, cf_handle); } else { iterator = db->NewIterator(read_options); } return reinterpret_cast(iterator); } /* * Class: org_rocksdb_RocksDB * Method: iterator * Signature: (J)J */ jlong Java_org_rocksdb_RocksDB_iterator__J( JNIEnv* env, jobject jdb, jlong db_handle) { auto db = reinterpret_cast(db_handle); return rocksdb_iterator_helper(db, rocksdb::ReadOptions(), nullptr); } /* * Class: org_rocksdb_RocksDB * Method: iterator * Signature: (JJ)J */ jlong Java_org_rocksdb_RocksDB_iterator__JJ( JNIEnv* env, jobject jdb, jlong db_handle, jlong jread_options_handle) { auto db = reinterpret_cast(db_handle); auto& read_options = *reinterpret_cast( jread_options_handle); return rocksdb_iterator_helper(db, read_options, nullptr); } /* * Class: org_rocksdb_RocksDB * Method: iteratorCF * Signature: (JJ)J */ jlong Java_org_rocksdb_RocksDB_iteratorCF__JJ( JNIEnv* env, jobject jdb, jlong db_handle, jlong jcf_handle) { auto db = reinterpret_cast(db_handle); auto cf_handle = reinterpret_cast(jcf_handle); return rocksdb_iterator_helper(db, rocksdb::ReadOptions(), cf_handle); } /* * Class: org_rocksdb_RocksDB * Method: iteratorCF * Signature: (JJJ)J */ jlong Java_org_rocksdb_RocksDB_iteratorCF__JJJ( JNIEnv* env, jobject jdb, jlong db_handle, jlong jcf_handle, jlong jread_options_handle) { auto db = reinterpret_cast(db_handle); auto cf_handle = reinterpret_cast(jcf_handle); auto& read_options = *reinterpret_cast( jread_options_handle); return rocksdb_iterator_helper(db, read_options, cf_handle); } /* * Class: org_rocksdb_RocksDB * Method: iterators * Signature: (JLjava/util/List;J)[J */ jlongArray Java_org_rocksdb_RocksDB_iterators( JNIEnv* env, jobject jdb, jlong db_handle, jobject jcfhandle_list, jlong jread_options_handle) { auto db = reinterpret_cast(db_handle); auto& read_options = *reinterpret_cast( jread_options_handle); std::vector cf_handles; std::vector iterators; if (jcfhandle_list != nullptr) { // get cf iterator jobject cfIteratorObj = env->CallObjectMethod( jcfhandle_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over keys and convert java byte array to slice while (env->CallBooleanMethod( cfIteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { jobject jobj = (jbyteArray) env->CallObjectMethod( cfIteratorObj, rocksdb::ListJni::getNextMethod(env)); rocksdb::ColumnFamilyHandle* cfHandle = rocksdb::ColumnFamilyHandleJni::getHandle(env, jobj); cf_handles.push_back(cfHandle); } } rocksdb::Status s = db->NewIterators(read_options, cf_handles, &iterators); if (s.ok()) { jlongArray jLongArray = env->NewLongArray(static_cast(iterators.size())); for (std::vector::size_type i = 0; i < iterators.size(); i++) { env->SetLongArrayRegion(jLongArray, static_cast(i), 1, reinterpret_cast(&iterators[i])); } return jLongArray; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return env->NewLongArray(0); } /* * Class: org_rocksdb_RocksDB * Method: getDefaultColumnFamily * Signature: (J)J */ jlong Java_org_rocksdb_RocksDB_getDefaultColumnFamily( JNIEnv* env, jobject jobj, jlong jdb_handle) { auto* db_handle = reinterpret_cast(jdb_handle); auto* cf_handle = db_handle->DefaultColumnFamily(); return reinterpret_cast(cf_handle); } /* * Class: org_rocksdb_RocksDB * Method: createColumnFamily * Signature: (JLorg/rocksdb/ColumnFamilyDescriptor;)J; */ jlong Java_org_rocksdb_RocksDB_createColumnFamily( JNIEnv* env, jobject jdb, jlong jdb_handle, jobject jcf_descriptor) { rocksdb::ColumnFamilyHandle* handle; auto db_handle = reinterpret_cast(jdb_handle); // get ColumnFamilyName jbyteArray byteArray = static_cast(env->CallObjectMethod( jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyNameMethod( env))); // get CF Options jobject jcf_opt_obj = env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyOptionsMethod( env)); rocksdb::ColumnFamilyOptions* cfOptions = rocksdb::ColumnFamilyOptionsJni::getHandle(env, jcf_opt_obj); jbyte* cfname = env->GetByteArrayElements(byteArray, 0); const int len = env->GetArrayLength(byteArray); rocksdb::Status s = db_handle->CreateColumnFamily( *cfOptions, std::string(reinterpret_cast(cfname), len), &handle); env->ReleaseByteArrayElements(byteArray, cfname, 0); if (s.ok()) { return reinterpret_cast(handle); } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return 0; } /* * Class: org_rocksdb_RocksDB * Method: dropColumnFamily * Signature: (JJ)V; */ void Java_org_rocksdb_RocksDB_dropColumnFamily( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jcf_handle) { auto cf_handle = reinterpret_cast(jcf_handle); auto db_handle = reinterpret_cast(jdb_handle); rocksdb::Status s = db_handle->DropColumnFamily(cf_handle); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Method: getSnapshot * Signature: (J)J */ jlong Java_org_rocksdb_RocksDB_getSnapshot( JNIEnv* env, jobject jdb, jlong db_handle) { auto db = reinterpret_cast(db_handle); const rocksdb::Snapshot* snapshot = db->GetSnapshot(); return reinterpret_cast(snapshot); } /* * Method: releaseSnapshot * Signature: (JJ)V */ void Java_org_rocksdb_RocksDB_releaseSnapshot( JNIEnv* env, jobject jdb, jlong db_handle, jlong snapshot_handle) { auto db = reinterpret_cast(db_handle); auto snapshot = reinterpret_cast(snapshot_handle); db->ReleaseSnapshot(snapshot); } /* * Class: org_rocksdb_RocksDB * Method: getProperty0 * Signature: (JLjava/lang/String;I)Ljava/lang/String; */ jstring Java_org_rocksdb_RocksDB_getProperty0__JLjava_lang_String_2I( JNIEnv* env, jobject jdb, jlong db_handle, jstring jproperty, jint jproperty_len) { auto db = reinterpret_cast(db_handle); const char* property = env->GetStringUTFChars(jproperty, 0); rocksdb::Slice property_slice(property, jproperty_len); std::string property_value; bool retCode = db->GetProperty(property_slice, &property_value); env->ReleaseStringUTFChars(jproperty, property); if (!retCode) { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::NotFound()); } return env->NewStringUTF(property_value.data()); } /* * Class: org_rocksdb_RocksDB * Method: getProperty0 * Signature: (JJLjava/lang/String;I)Ljava/lang/String; */ jstring Java_org_rocksdb_RocksDB_getProperty0__JJLjava_lang_String_2I( JNIEnv* env, jobject jdb, jlong db_handle, jlong jcf_handle, jstring jproperty, jint jproperty_len) { auto db = reinterpret_cast(db_handle); auto cf_handle = reinterpret_cast(jcf_handle); const char* property = env->GetStringUTFChars(jproperty, 0); rocksdb::Slice property_slice(property, jproperty_len); std::string property_value; bool retCode = db->GetProperty(cf_handle, property_slice, &property_value); env->ReleaseStringUTFChars(jproperty, property); if (!retCode) { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::NotFound()); } return env->NewStringUTF(property_value.data()); } /* * Class: org_rocksdb_RocksDB * Method: getLongProperty * Signature: (JLjava/lang/String;I)L; */ jlong Java_org_rocksdb_RocksDB_getLongProperty__JLjava_lang_String_2I( JNIEnv* env, jobject jdb, jlong db_handle, jstring jproperty, jint jproperty_len) { auto db = reinterpret_cast(db_handle); const char* property = env->GetStringUTFChars(jproperty, 0); rocksdb::Slice property_slice(property, jproperty_len); uint64_t property_value = 0; bool retCode = db->GetIntProperty(property_slice, &property_value); env->ReleaseStringUTFChars(jproperty, property); if (!retCode) { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::NotFound()); } return property_value; } /* * Class: org_rocksdb_RocksDB * Method: getLongProperty * Signature: (JJLjava/lang/String;I)L; */ jlong Java_org_rocksdb_RocksDB_getLongProperty__JJLjava_lang_String_2I( JNIEnv* env, jobject jdb, jlong db_handle, jlong jcf_handle, jstring jproperty, jint jproperty_len) { auto db = reinterpret_cast(db_handle); auto cf_handle = reinterpret_cast(jcf_handle); const char* property = env->GetStringUTFChars(jproperty, 0); rocksdb::Slice property_slice(property, jproperty_len); uint64_t property_value; bool retCode = db->GetIntProperty(cf_handle, property_slice, &property_value); env->ReleaseStringUTFChars(jproperty, property); if (!retCode) { rocksdb::RocksDBExceptionJni::ThrowNew(env, rocksdb::Status::NotFound()); } return property_value; } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::Flush void rocksdb_flush_helper( JNIEnv* env, rocksdb::DB* db, const rocksdb::FlushOptions& flush_options, rocksdb::ColumnFamilyHandle* column_family_handle) { rocksdb::Status s; if (column_family_handle != nullptr) { s = db->Flush(flush_options, column_family_handle); } else { s = db->Flush(flush_options); } if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RocksDB * Method: flush * Signature: (JJ)V */ void Java_org_rocksdb_RocksDB_flush__JJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jflush_options) { auto db = reinterpret_cast(jdb_handle); auto flush_options = reinterpret_cast(jflush_options); rocksdb_flush_helper(env, db, *flush_options, nullptr); } /* * Class: org_rocksdb_RocksDB * Method: flush * Signature: (JJJ)V */ void Java_org_rocksdb_RocksDB_flush__JJJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jflush_options, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); auto flush_options = reinterpret_cast(jflush_options); auto cf_handle = reinterpret_cast(jcf_handle); rocksdb_flush_helper(env, db, *flush_options, cf_handle); } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::CompactRange - Full void rocksdb_compactrange_helper(JNIEnv* env, rocksdb::DB* db, rocksdb::ColumnFamilyHandle* cf_handle, jboolean jreduce_level, jint jtarget_level, jint jtarget_path_id) { rocksdb::Status s; rocksdb::CompactRangeOptions compact_options; compact_options.change_level = jreduce_level; compact_options.target_level = jtarget_level; compact_options.target_path_id = static_cast(jtarget_path_id); if (cf_handle != nullptr) { s = db->CompactRange(compact_options, cf_handle, nullptr, nullptr); } else { // backwards compatibility s = db->CompactRange(compact_options, nullptr, nullptr); } if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksDB * Method: compactRange0 * Signature: (JZII)V */ void Java_org_rocksdb_RocksDB_compactRange0__JZII(JNIEnv* env, jobject jdb, jlong jdb_handle, jboolean jreduce_level, jint jtarget_level, jint jtarget_path_id) { auto db = reinterpret_cast(jdb_handle); rocksdb_compactrange_helper(env, db, nullptr, jreduce_level, jtarget_level, jtarget_path_id); } /* * Class: org_rocksdb_RocksDB * Method: compactRange * Signature: (JZIIJ)V */ void Java_org_rocksdb_RocksDB_compactRange__JZIIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jboolean jreduce_level, jint jtarget_level, jint jtarget_path_id, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); auto cf_handle = reinterpret_cast(jcf_handle); rocksdb_compactrange_helper(env, db, cf_handle, jreduce_level, jtarget_level, jtarget_path_id); } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::CompactRange - Range void rocksdb_compactrange_helper(JNIEnv* env, rocksdb::DB* db, rocksdb::ColumnFamilyHandle* cf_handle, jbyteArray jbegin, jint jbegin_len, jbyteArray jend, jint jend_len, jboolean jreduce_level, jint jtarget_level, jint jtarget_path_id) { jbyte* begin = env->GetByteArrayElements(jbegin, 0); jbyte* end = env->GetByteArrayElements(jend, 0); const rocksdb::Slice begin_slice(reinterpret_cast(begin), jbegin_len); const rocksdb::Slice end_slice(reinterpret_cast(end), jend_len); rocksdb::Status s; rocksdb::CompactRangeOptions compact_options; compact_options.change_level = jreduce_level; compact_options.target_level = jtarget_level; compact_options.target_path_id = static_cast(jtarget_path_id); if (cf_handle != nullptr) { s = db->CompactRange(compact_options, cf_handle, &begin_slice, &end_slice); } else { // backwards compatibility s = db->CompactRange(compact_options, &begin_slice, &end_slice); } env->ReleaseByteArrayElements(jbegin, begin, JNI_ABORT); env->ReleaseByteArrayElements(jend, end, JNI_ABORT); if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_RocksDB * Method: compactRange0 * Signature: (J[BI[BIZII)V */ void Java_org_rocksdb_RocksDB_compactRange0__J_3BI_3BIZII(JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jbegin, jint jbegin_len, jbyteArray jend, jint jend_len, jboolean jreduce_level, jint jtarget_level, jint jtarget_path_id) { auto db = reinterpret_cast(jdb_handle); rocksdb_compactrange_helper(env, db, nullptr, jbegin, jbegin_len, jend, jend_len, jreduce_level, jtarget_level, jtarget_path_id); } /* * Class: org_rocksdb_RocksDB * Method: compactRange * Signature: (JJ[BI[BIZII)V */ void Java_org_rocksdb_RocksDB_compactRange__J_3BI_3BIZIIJ( JNIEnv* env, jobject jdb, jlong jdb_handle, jbyteArray jbegin, jint jbegin_len, jbyteArray jend, jint jend_len, jboolean jreduce_level, jint jtarget_level, jint jtarget_path_id, jlong jcf_handle) { auto db = reinterpret_cast(jdb_handle); auto cf_handle = reinterpret_cast(jcf_handle); rocksdb_compactrange_helper(env, db, cf_handle, jbegin, jbegin_len, jend, jend_len, jreduce_level, jtarget_level, jtarget_path_id); } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::GetLatestSequenceNumber /* * Class: org_rocksdb_RocksDB * Method: getLatestSequenceNumber * Signature: (J)V */ jlong Java_org_rocksdb_RocksDB_getLatestSequenceNumber(JNIEnv* env, jobject jdb, jlong jdb_handle) { auto* db = reinterpret_cast(jdb_handle); return db->GetLatestSequenceNumber(); } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB enable/disable file deletions /* * Class: org_rocksdb_RocksDB * Method: enableFileDeletions * Signature: (J)V */ void Java_org_rocksdb_RocksDB_disableFileDeletions(JNIEnv* env, jobject jdb, jlong jdb_handle) { auto* db = reinterpret_cast(jdb_handle); rocksdb::Status s = db->DisableFileDeletions(); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_RocksDB * Method: enableFileDeletions * Signature: (JZ)V */ void Java_org_rocksdb_RocksDB_enableFileDeletions(JNIEnv* env, jobject jdb, jlong jdb_handle, jboolean jforce) { auto* db = reinterpret_cast(jdb_handle); rocksdb::Status s = db->EnableFileDeletions(jforce); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } ////////////////////////////////////////////////////////////////////////////// // rocksdb::DB::GetUpdatesSince /* * Class: org_rocksdb_RocksDB * Method: getUpdatesSince * Signature: (JJ)J */ jlong Java_org_rocksdb_RocksDB_getUpdatesSince(JNIEnv* env, jobject jdb, jlong jdb_handle, jlong jsequence_number) { auto* db = reinterpret_cast(jdb_handle); rocksdb::SequenceNumber sequence_number = static_cast(jsequence_number); std::unique_ptr iter; rocksdb::Status s = db->GetUpdatesSince(sequence_number, &iter); if (s.ok()) { return reinterpret_cast(iter.release()); } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return 0; } rocksdb-4.5.1/java/rocksjni/slice.cc000066400000000000000000000162201267530634000173300ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for // rocksdb::Slice. #include #include #include #include #include "include/org_rocksdb_AbstractSlice.h" #include "include/org_rocksdb_Slice.h" #include "include/org_rocksdb_DirectSlice.h" #include "rocksdb/slice.h" #include "rocksjni/portal.h" // /* * Class: org_rocksdb_Slice * Method: createNewSlice0 * Signature: ([BI)V */ void Java_org_rocksdb_Slice_createNewSlice0( JNIEnv * env, jobject jobj, jbyteArray data, jint offset) { const jsize dataSize = env->GetArrayLength(data); const int len = dataSize - offset; jbyte* ptrData = new jbyte[len]; env->GetByteArrayRegion(data, offset, len, ptrData); const auto* slice = new rocksdb::Slice((const char*)ptrData, len); rocksdb::AbstractSliceJni::setHandle(env, jobj, slice); } /* * Class: org_rocksdb_Slice * Method: createNewSlice1 * Signature: ([B)V */ void Java_org_rocksdb_Slice_createNewSlice1( JNIEnv * env, jobject jobj, jbyteArray data) { const int len = env->GetArrayLength(data) + 1; jboolean isCopy; jbyte* ptrData = env->GetByteArrayElements(data, &isCopy); char* buf = new char[len]; memcpy(buf, ptrData, len - 1); buf[len-1]='\0'; const auto* slice = new rocksdb::Slice(buf, len - 1); rocksdb::AbstractSliceJni::setHandle(env, jobj, slice); env->ReleaseByteArrayElements(data, ptrData, JNI_ABORT); // NOTE: buf will be deleted in the org.rocksdb.Slice#dispose method } /* * Class: org_rocksdb_Slice * Method: data0 * Signature: (J)[B */ jbyteArray Java_org_rocksdb_Slice_data0( JNIEnv* env, jobject jobj, jlong handle) { const auto* slice = reinterpret_cast(handle); const int len = static_cast(slice->size()); const jbyteArray data = env->NewByteArray(len); env->SetByteArrayRegion(data, 0, len, reinterpret_cast(slice->data())); return data; } /* * Class: org_rocksdb_Slice * Method: disposeInternalBuf * Signature: (J)V */ void Java_org_rocksdb_Slice_disposeInternalBuf( JNIEnv * env, jobject jobj, jlong handle) { const auto* slice = reinterpret_cast(handle); delete [] slice->data_; } // // #include #include #include "include/org_rocksdb_Snapshot.h" #include "rocksdb/db.h" #include "rocksjni/portal.h" /* * Class: org_rocksdb_Snapshot * Method: getSequenceNumber * Signature: (J)J */ jlong Java_org_rocksdb_Snapshot_getSequenceNumber(JNIEnv* env, jobject jobj, jlong jsnapshot_handle) { auto* snapshot = reinterpret_cast( jsnapshot_handle); return snapshot->GetSequenceNumber(); } rocksdb-4.5.1/java/rocksjni/statistics.cc000066400000000000000000000032621267530634000204250ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::Statistics methods from Java side. #include #include #include #include "include/org_rocksdb_Statistics.h" #include "rocksjni/portal.h" #include "rocksdb/statistics.h" /* * Class: org_rocksdb_Statistics * Method: getTickerCount0 * Signature: (IJ)J */ jlong Java_org_rocksdb_Statistics_getTickerCount0( JNIEnv* env, jobject jobj, jint tickerType, jlong handle) { auto st = reinterpret_cast(handle); assert(st != nullptr); return st->getTickerCount(static_cast(tickerType)); } /* * Class: org_rocksdb_Statistics * Method: geHistogramData0 * Signature: (IJ)Lorg/rocksdb/HistogramData; */ jobject Java_org_rocksdb_Statistics_geHistogramData0( JNIEnv* env, jobject jobj, jint histogramType, jlong handle) { auto st = reinterpret_cast(handle); assert(st != nullptr); rocksdb::HistogramData data; st->histogramData(static_cast(histogramType), &data); // Don't reuse class pointer jclass jclazz = env->FindClass("org/rocksdb/HistogramData"); jmethodID mid = rocksdb::HistogramDataJni::getConstructorMethodId( env, jclazz); return env->NewObject(jclazz, mid, data.median, data.percentile95, data.percentile99, data.average, data.standard_deviation); } rocksdb-4.5.1/java/rocksjni/table.cc000066400000000000000000000073551267530634000173310ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ for rocksdb::Options. #include #include "include/org_rocksdb_PlainTableConfig.h" #include "include/org_rocksdb_BlockBasedTableConfig.h" #include "rocksdb/table.h" #include "rocksdb/cache.h" #include "rocksdb/filter_policy.h" /* * Class: org_rocksdb_PlainTableConfig * Method: newTableFactoryHandle * Signature: (IIDIIBZZ)J */ jlong Java_org_rocksdb_PlainTableConfig_newTableFactoryHandle( JNIEnv* env, jobject jobj, jint jkey_size, jint jbloom_bits_per_key, jdouble jhash_table_ratio, jint jindex_sparseness, jint jhuge_page_tlb_size, jbyte jencoding_type, jboolean jfull_scan_mode, jboolean jstore_index_in_file) { rocksdb::PlainTableOptions options = rocksdb::PlainTableOptions(); options.user_key_len = jkey_size; options.bloom_bits_per_key = jbloom_bits_per_key; options.hash_table_ratio = jhash_table_ratio; options.index_sparseness = jindex_sparseness; options.huge_page_tlb_size = jhuge_page_tlb_size; options.encoding_type = static_cast( jencoding_type); options.full_scan_mode = jfull_scan_mode; options.store_index_in_file = jstore_index_in_file; return reinterpret_cast(rocksdb::NewPlainTableFactory(options)); } /* * Class: org_rocksdb_BlockBasedTableConfig * Method: newTableFactoryHandle * Signature: (ZJIJIIZIZZJIBBI)J */ jlong Java_org_rocksdb_BlockBasedTableConfig_newTableFactoryHandle( JNIEnv* env, jobject jobj, jboolean no_block_cache, jlong block_cache_size, jint block_cache_num_shardbits, jlong block_size, jint block_size_deviation, jint block_restart_interval, jboolean whole_key_filtering, jlong jfilterPolicy, jboolean cache_index_and_filter_blocks, jboolean hash_index_allow_collision, jlong block_cache_compressed_size, jint block_cache_compressd_num_shard_bits, jbyte jchecksum_type, jbyte jindex_type, jint jformat_version) { rocksdb::BlockBasedTableOptions options; options.no_block_cache = no_block_cache; if (!no_block_cache && block_cache_size > 0) { if (block_cache_num_shardbits > 0) { options.block_cache = rocksdb::NewLRUCache(block_cache_size, block_cache_num_shardbits); } else { options.block_cache = rocksdb::NewLRUCache(block_cache_size); } } options.block_size = block_size; options.block_size_deviation = block_size_deviation; options.block_restart_interval = block_restart_interval; options.whole_key_filtering = whole_key_filtering; if (jfilterPolicy > 0) { std::shared_ptr *pFilterPolicy = reinterpret_cast *>( jfilterPolicy); options.filter_policy = *pFilterPolicy; } options.cache_index_and_filter_blocks = cache_index_and_filter_blocks; options.hash_index_allow_collision = hash_index_allow_collision; if (block_cache_compressed_size > 0) { if (block_cache_compressd_num_shard_bits > 0) { options.block_cache = rocksdb::NewLRUCache(block_cache_compressed_size, block_cache_compressd_num_shard_bits); } else { options.block_cache = rocksdb::NewLRUCache(block_cache_compressed_size); } } options.checksum = static_cast(jchecksum_type); options.index_type = static_cast< rocksdb::BlockBasedTableOptions::IndexType>(jindex_type); options.format_version = jformat_version; return reinterpret_cast(rocksdb::NewBlockBasedTableFactory(options)); } rocksdb-4.5.1/java/rocksjni/transaction_log.cc000066400000000000000000000047461267530634000214310ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::Iterator methods from Java side. #include #include #include #include "include/org_rocksdb_TransactionLogIterator.h" #include "rocksdb/transaction_log.h" #include "rocksjni/portal.h" /* * Class: org_rocksdb_TransactionLogIterator * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_TransactionLogIterator_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } /* * Class: org_rocksdb_TransactionLogIterator * Method: isValid * Signature: (J)Z */ jboolean Java_org_rocksdb_TransactionLogIterator_isValid( JNIEnv* env, jobject jobj, jlong handle) { return reinterpret_cast(handle)->Valid(); } /* * Class: org_rocksdb_TransactionLogIterator * Method: next * Signature: (J)V */ void Java_org_rocksdb_TransactionLogIterator_next( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->Next(); } /* * Class: org_rocksdb_TransactionLogIterator * Method: status * Signature: (J)V */ void Java_org_rocksdb_TransactionLogIterator_status( JNIEnv* env, jobject jobj, jlong handle) { rocksdb::Status s = reinterpret_cast< rocksdb::TransactionLogIterator*>(handle)->status(); if (!s.ok()) { rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } } /* * Class: org_rocksdb_TransactionLogIterator * Method: getBatch * Signature: (J)Lorg/rocksdb/TransactionLogIterator$BatchResult */ jobject Java_org_rocksdb_TransactionLogIterator_getBatch( JNIEnv* env, jobject jobj, jlong handle) { rocksdb::BatchResult batch_result = reinterpret_cast(handle)->GetBatch(); jclass jclazz = env->FindClass( "org/rocksdb/TransactionLogIterator$BatchResult"); assert(jclazz != nullptr); jmethodID mid = env->GetMethodID( jclazz, "", "(Lorg/rocksdb/TransactionLogIterator;JJ)V"); assert(mid != nullptr); return env->NewObject(jclazz, mid, jobj, batch_result.sequence, batch_result.writeBatchPtr.release()); } rocksdb-4.5.1/java/rocksjni/ttl.cc000066400000000000000000000154221267530634000170370ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::TtlDB methods. // from Java side. #include #include #include #include #include #include "include/org_rocksdb_TtlDB.h" #include "rocksdb/utilities/db_ttl.h" #include "rocksjni/portal.h" /* * Class: org_rocksdb_TtlDB * Method: open * Signature: (JLjava/lang/String;IZ)V */ void Java_org_rocksdb_TtlDB_open(JNIEnv* env, jobject jttldb, jlong joptions_handle, jstring jdb_path, jint jttl, jboolean jread_only) { auto* opt = reinterpret_cast(joptions_handle); rocksdb::DBWithTTL* db = nullptr; const char* db_path = env->GetStringUTFChars(jdb_path, 0); rocksdb::Status s = rocksdb::DBWithTTL::Open(*opt, db_path, &db, jttl, jread_only); env->ReleaseStringUTFChars(jdb_path, db_path); // as TTLDB extends RocksDB on the java side, we can reuse // the RocksDB portal here. if (s.ok()) { rocksdb::RocksDBJni::setHandle(env, jttldb, db); return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_TtlDB * Method: openCF * Signature: (JLjava/lang/String;Ljava/util/List; * ILjava/util/List;Z)Ljava/util/List; */ jobject Java_org_rocksdb_TtlDB_openCF( JNIEnv* env, jobject jdb, jlong jopt_handle, jstring jdb_path, jobject jcfdesc_list, jint jcfdesc_count, jobject jttl_list, jboolean jread_only) { auto* opt = reinterpret_cast(jopt_handle); rocksdb::DBWithTTL* db = nullptr; const char* db_path = env->GetStringUTFChars(jdb_path, 0); std::vector cfnames_to_free; std::vector jcfnames_for_free; std::vector column_families; std::vector ttl_values; std::vector handles; // get iterator for ColumnFamilyDescriptors jobject iteratorObj = env->CallObjectMethod( jcfdesc_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over ColumnFamilyDescriptors while (env->CallBooleanMethod( iteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { // get ColumnFamilyDescriptor jobject jcf_descriptor = env->CallObjectMethod(iteratorObj, rocksdb::ListJni::getNextMethod(env)); // get ColumnFamilyName jbyteArray byteArray = static_cast(env->CallObjectMethod( jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyNameMethod( env))); // get CF Options jobject jcf_opt_obj = env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyOptionsMethod( env)); rocksdb::ColumnFamilyOptions* cfOptions = rocksdb::ColumnFamilyOptionsJni::getHandle(env, jcf_opt_obj); jbyte* cfname = env->GetByteArrayElements(byteArray, 0); const int len = env->GetArrayLength(byteArray); // free allocated cfnames after call to open cfnames_to_free.push_back(cfname); jcfnames_for_free.push_back(byteArray); column_families.push_back(rocksdb::ColumnFamilyDescriptor( std::string(reinterpret_cast(cfname), len), *cfOptions)); } // get iterator for TTL values iteratorObj = env->CallObjectMethod( jttl_list, rocksdb::ListJni::getIteratorMethod(env)); // iterate over TTL values while (env->CallBooleanMethod( iteratorObj, rocksdb::ListJni::getHasNextMethod(env)) == JNI_TRUE) { // get TTL object jobject jttl_object = env->CallObjectMethod(iteratorObj, rocksdb::ListJni::getNextMethod(env)); // get Integer value jclass jIntClazz = env->FindClass("java/lang/Integer"); jmethodID getVal = env->GetMethodID(jIntClazz, "intValue", "()I"); ttl_values.push_back(env->CallIntMethod(jttl_object, getVal)); } rocksdb::Status s = rocksdb::DBWithTTL::Open(*opt, db_path, column_families, &handles, &db, ttl_values, jread_only); env->ReleaseStringUTFChars(jdb_path, db_path); // free jbyte allocations for (std::vector::size_type i = 0; i != cfnames_to_free.size(); i++) { // free cfnames env->ReleaseByteArrayElements(jcfnames_for_free[i], cfnames_to_free[i], 0); } // check if open operation was successful if (s.ok()) { rocksdb::RocksDBJni::setHandle(env, jdb, db); jclass jListClazz = env->FindClass("java/util/ArrayList"); jmethodID midList = rocksdb::ListJni::getArrayListConstructorMethodId( env, jListClazz); jobject jcfhandle_list = env->NewObject(jListClazz, midList, handles.size()); // insert in java list for (std::vector::size_type i = 0; i != handles.size(); i++) { // jlong must be converted to Long due to collections restrictions jclass jLongClazz = env->FindClass("java/lang/Long"); jmethodID midLong = env->GetMethodID(jLongClazz, "", "(J)V"); jobject obj = env->NewObject(jLongClazz, midLong, reinterpret_cast(handles[i])); env->CallBooleanMethod(jcfhandle_list, rocksdb::ListJni::getListAddMethodId(env), obj); } return jcfhandle_list; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return nullptr; } /* * Class: org_rocksdb_TtlDB * Method: createColumnFamilyWithTtl * Signature: (JLorg/rocksdb/ColumnFamilyDescriptor;I)J; */ jlong Java_org_rocksdb_TtlDB_createColumnFamilyWithTtl( JNIEnv* env, jobject jobj, jlong jdb_handle, jobject jcf_descriptor, jint jttl) { rocksdb::ColumnFamilyHandle* handle; auto* db_handle = reinterpret_cast(jdb_handle); // get ColumnFamilyName jbyteArray byteArray = static_cast(env->CallObjectMethod( jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyNameMethod( env))); // get CF Options jobject jcf_opt_obj = env->CallObjectMethod(jcf_descriptor, rocksdb::ColumnFamilyDescriptorJni::getColumnFamilyOptionsMethod( env)); rocksdb::ColumnFamilyOptions* cfOptions = rocksdb::ColumnFamilyOptionsJni::getHandle(env, jcf_opt_obj); jbyte* cfname = env->GetByteArrayElements(byteArray, 0); const int len = env->GetArrayLength(byteArray); rocksdb::Status s = db_handle->CreateColumnFamilyWithTtl( *cfOptions, std::string(reinterpret_cast(cfname), len), &handle, jttl); env->ReleaseByteArrayElements(byteArray, cfname, 0); if (s.ok()) { return reinterpret_cast(handle); } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); return 0; } rocksdb-4.5.1/java/rocksjni/write_batch.cc000066400000000000000000000160201267530634000205220ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::WriteBatch methods from Java side. #include #include "include/org_rocksdb_WriteBatch.h" #include "include/org_rocksdb_WriteBatch_Handler.h" #include "rocksjni/portal.h" #include "rocksjni/writebatchhandlerjnicallback.h" #include "rocksdb/db.h" #include "rocksdb/immutable_options.h" #include "db/memtable.h" #include "rocksdb/write_batch.h" #include "rocksdb/status.h" #include "db/write_batch_internal.h" #include "db/writebuffer.h" #include "rocksdb/env.h" #include "rocksdb/memtablerep.h" #include "table/scoped_arena_iterator.h" #include "util/logging.h" #include "util/testharness.h" /* * Class: org_rocksdb_WriteBatch * Method: newWriteBatch * Signature: (I)V */ void Java_org_rocksdb_WriteBatch_newWriteBatch( JNIEnv* env, jobject jobj, jint jreserved_bytes) { rocksdb::WriteBatch* wb = new rocksdb::WriteBatch( static_cast(jreserved_bytes)); rocksdb::WriteBatchJni::setHandle(env, jobj, wb); } /* * Class: org_rocksdb_WriteBatch * Method: count0 * Signature: ()I */ jint Java_org_rocksdb_WriteBatch_count0(JNIEnv* env, jobject jobj) { rocksdb::WriteBatch* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); return static_cast(wb->Count()); } /* * Class: org_rocksdb_WriteBatch * Method: clear0 * Signature: ()V */ void Java_org_rocksdb_WriteBatch_clear0(JNIEnv* env, jobject jobj) { rocksdb::WriteBatch* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); wb->Clear(); } /* * Class: org_rocksdb_WriteBatch * Method: put * Signature: ([BI[BI)V */ void Java_org_rocksdb_WriteBatch_put___3BI_3BI( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto put = [&wb] (rocksdb::Slice key, rocksdb::Slice value) { wb->Put(key, value); }; rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatch * Method: put * Signature: ([BI[BIJ)V */ void Java_org_rocksdb_WriteBatch_put___3BI_3BIJ( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto* cf_handle = reinterpret_cast(jcf_handle); assert(cf_handle != nullptr); auto put = [&wb, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) { wb->Put(cf_handle, key, value); }; rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatch * Method: merge * Signature: ([BI[BI)V */ void Java_org_rocksdb_WriteBatch_merge___3BI_3BI( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto merge = [&wb] (rocksdb::Slice key, rocksdb::Slice value) { wb->Merge(key, value); }; rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatch * Method: merge * Signature: ([BI[BIJ)V */ void Java_org_rocksdb_WriteBatch_merge___3BI_3BIJ( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto* cf_handle = reinterpret_cast(jcf_handle); assert(cf_handle != nullptr); auto merge = [&wb, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) { wb->Merge(cf_handle, key, value); }; rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatch * Method: remove * Signature: ([BI)V */ void Java_org_rocksdb_WriteBatch_remove___3BI( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto remove = [&wb] (rocksdb::Slice key) { wb->Delete(key); }; rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len); } /* * Class: org_rocksdb_WriteBatch * Method: remove * Signature: ([BIJ)V */ void Java_org_rocksdb_WriteBatch_remove___3BIJ( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jlong jcf_handle) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto* cf_handle = reinterpret_cast(jcf_handle); assert(cf_handle != nullptr); auto remove = [&wb, &cf_handle] (rocksdb::Slice key) { wb->Delete(cf_handle, key); }; rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len); } /* * Class: org_rocksdb_WriteBatch * Method: putLogData * Signature: ([BI)V */ void Java_org_rocksdb_WriteBatch_putLogData( JNIEnv* env, jobject jobj, jbyteArray jblob, jint jblob_len) { auto* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); auto putLogData = [&wb] (rocksdb::Slice blob) { wb->PutLogData(blob); }; rocksdb::JniUtil::k_op(putLogData, env, jobj, jblob, jblob_len); } /* * Class: org_rocksdb_WriteBatch * Method: iterate * Signature: (J)V */ void Java_org_rocksdb_WriteBatch_iterate( JNIEnv* env , jobject jobj, jlong handlerHandle) { rocksdb::WriteBatch* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); rocksdb::Status s = wb->Iterate( reinterpret_cast(handlerHandle)); if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_WriteBatch * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_WriteBatch_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } /* * Class: org_rocksdb_WriteBatch_Handler * Method: createNewHandler0 * Signature: ()V */ void Java_org_rocksdb_WriteBatch_00024Handler_createNewHandler0( JNIEnv* env, jobject jobj) { const rocksdb::WriteBatchHandlerJniCallback* h = new rocksdb::WriteBatchHandlerJniCallback(env, jobj); rocksdb::WriteBatchHandlerJni::setHandle(env, jobj, h); } /* * Class: org_rocksdb_WriteBatch_Handler * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_WriteBatch_00024Handler_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { delete reinterpret_cast(handle); } rocksdb-4.5.1/java/rocksjni/write_batch_test.cc000066400000000000000000000122131267530634000215610ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::WriteBatch methods testing from Java side. #include #include "db/memtable.h" #include "db/write_batch_internal.h" #include "db/writebuffer.h" #include "include/org_rocksdb_WriteBatch.h" #include "include/org_rocksdb_WriteBatch_Handler.h" #include "include/org_rocksdb_WriteBatchTest.h" #include "include/org_rocksdb_WriteBatchTestInternalHelper.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/immutable_options.h" #include "rocksdb/memtablerep.h" #include "rocksdb/status.h" #include "rocksdb/write_batch.h" #include "rocksjni/portal.h" #include "table/scoped_arena_iterator.h" #include "util/logging.h" #include "util/testharness.h" /* * Class: org_rocksdb_WriteBatchTest * Method: getContents * Signature: (Lorg/rocksdb/WriteBatch;)[B */ jbyteArray Java_org_rocksdb_WriteBatchTest_getContents( JNIEnv* env, jclass jclazz, jobject jobj) { rocksdb::WriteBatch* b = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(b != nullptr); // todo: Currently the following code is directly copied from // db/write_bench_test.cc. It could be implemented in java once // all the necessary components can be accessed via jni api. rocksdb::InternalKeyComparator cmp(rocksdb::BytewiseComparator()); auto factory = std::make_shared(); rocksdb::Options options; rocksdb::WriteBuffer wb(options.db_write_buffer_size); options.memtable_factory = factory; rocksdb::MemTable* mem = new rocksdb::MemTable( cmp, rocksdb::ImmutableCFOptions(options), rocksdb::MutableCFOptions(options, rocksdb::ImmutableCFOptions(options)), &wb, rocksdb::kMaxSequenceNumber); mem->Ref(); std::string state; rocksdb::ColumnFamilyMemTablesDefault cf_mems_default(mem); rocksdb::Status s = rocksdb::WriteBatchInternal::InsertInto(b, &cf_mems_default, nullptr); int count = 0; rocksdb::Arena arena; rocksdb::ScopedArenaIterator iter(mem->NewIterator( rocksdb::ReadOptions(), &arena)); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { rocksdb::ParsedInternalKey ikey; memset(reinterpret_cast(&ikey), 0, sizeof(ikey)); bool parsed = rocksdb::ParseInternalKey(iter->key(), &ikey); assert(parsed); switch (ikey.type) { case rocksdb::kTypeValue: state.append("Put("); state.append(ikey.user_key.ToString()); state.append(", "); state.append(iter->value().ToString()); state.append(")"); count++; break; case rocksdb::kTypeMerge: state.append("Merge("); state.append(ikey.user_key.ToString()); state.append(", "); state.append(iter->value().ToString()); state.append(")"); count++; break; case rocksdb::kTypeDeletion: state.append("Delete("); state.append(ikey.user_key.ToString()); state.append(")"); count++; break; default: assert(false); break; } state.append("@"); state.append(rocksdb::NumberToString(ikey.sequence)); } if (!s.ok()) { state.append(s.ToString()); } else if (count != rocksdb::WriteBatchInternal::Count(b)) { state.append("CountMismatch()"); } delete mem->Unref(); jbyteArray jstate = env->NewByteArray(static_cast(state.size())); env->SetByteArrayRegion(jstate, 0, static_cast(state.size()), reinterpret_cast(state.c_str())); return jstate; } /* * Class: org_rocksdb_WriteBatchTestInternalHelper * Method: setSequence * Signature: (Lorg/rocksdb/WriteBatch;J)V */ void Java_org_rocksdb_WriteBatchTestInternalHelper_setSequence( JNIEnv* env, jclass jclazz, jobject jobj, jlong jsn) { rocksdb::WriteBatch* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); rocksdb::WriteBatchInternal::SetSequence( wb, static_cast(jsn)); } /* * Class: org_rocksdb_WriteBatchTestInternalHelper * Method: sequence * Signature: (Lorg/rocksdb/WriteBatch;)J */ jlong Java_org_rocksdb_WriteBatchTestInternalHelper_sequence( JNIEnv* env, jclass jclazz, jobject jobj) { rocksdb::WriteBatch* wb = rocksdb::WriteBatchJni::getHandle(env, jobj); assert(wb != nullptr); return static_cast(rocksdb::WriteBatchInternal::Sequence(wb)); } /* * Class: org_rocksdb_WriteBatchTestInternalHelper * Method: append * Signature: (Lorg/rocksdb/WriteBatch;Lorg/rocksdb/WriteBatch;)V */ void Java_org_rocksdb_WriteBatchTestInternalHelper_append( JNIEnv* env, jclass jclazz, jobject jwb1, jobject jwb2) { rocksdb::WriteBatch* wb1 = rocksdb::WriteBatchJni::getHandle(env, jwb1); assert(wb1 != nullptr); rocksdb::WriteBatch* wb2 = rocksdb::WriteBatchJni::getHandle(env, jwb2); assert(wb2 != nullptr); rocksdb::WriteBatchInternal::Append(wb1, wb2); } rocksdb-4.5.1/java/rocksjni/write_batch_with_index.cc000066400000000000000000000301631267530634000227500ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the "bridge" between Java and C++ and enables // calling c++ rocksdb::WriteBatchWithIndex methods from Java side. #include "include/org_rocksdb_WBWIRocksIterator.h" #include "include/org_rocksdb_WriteBatchWithIndex.h" #include "rocksdb/comparator.h" #include "rocksdb/utilities/write_batch_with_index.h" #include "rocksjni/portal.h" /* * Class: org_rocksdb_WriteBatchWithIndex * Method: newWriteBatchWithIndex * Signature: ()V */ void Java_org_rocksdb_WriteBatchWithIndex_newWriteBatchWithIndex__( JNIEnv* env, jobject jobj) { rocksdb::WriteBatchWithIndex* wbwi = new rocksdb::WriteBatchWithIndex(); rocksdb::WriteBatchWithIndexJni::setHandle(env, jobj, wbwi); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: newWriteBatchWithIndex * Signature: (Z)V */ void Java_org_rocksdb_WriteBatchWithIndex_newWriteBatchWithIndex__Z( JNIEnv* env, jobject jobj, jboolean joverwrite_key) { rocksdb::WriteBatchWithIndex* wbwi = new rocksdb::WriteBatchWithIndex(rocksdb::BytewiseComparator(), 0, static_cast(joverwrite_key)); rocksdb::WriteBatchWithIndexJni::setHandle(env, jobj, wbwi); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: newWriteBatchWithIndex * Signature: (JIZ)V */ void Java_org_rocksdb_WriteBatchWithIndex_newWriteBatchWithIndex__JIZ( JNIEnv* env, jobject jobj, jlong jfallback_index_comparator_handle, jint jreserved_bytes, jboolean joverwrite_key) { rocksdb::WriteBatchWithIndex* wbwi = new rocksdb::WriteBatchWithIndex( reinterpret_cast(jfallback_index_comparator_handle), static_cast(jreserved_bytes), static_cast(joverwrite_key)); rocksdb::WriteBatchWithIndexJni::setHandle(env, jobj, wbwi); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: count * Signature: ()I */ jint Java_org_rocksdb_WriteBatchWithIndex_count0( JNIEnv* env, jobject jobj) { rocksdb::WriteBatchWithIndex* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); return static_cast(wbwi->GetWriteBatch()->Count()); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: put * Signature: ([BI[BI)V */ void Java_org_rocksdb_WriteBatchWithIndex_put___3BI_3BI( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto put = [&wbwi] (rocksdb::Slice key, rocksdb::Slice value) { wbwi->Put(key, value); }; rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: put * Signature: ([BI[BIJ)V */ void Java_org_rocksdb_WriteBatchWithIndex_put___3BI_3BIJ( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto* cf_handle = reinterpret_cast(jcf_handle); assert(cf_handle != nullptr); auto put = [&wbwi, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) { wbwi->Put(cf_handle, key, value); }; rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: merge * Signature: ([BI[BI)V */ void Java_org_rocksdb_WriteBatchWithIndex_merge___3BI_3BI( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto merge = [&wbwi] (rocksdb::Slice key, rocksdb::Slice value) { wbwi->Merge(key, value); }; rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: merge * Signature: ([BI[BIJ)V */ void Java_org_rocksdb_WriteBatchWithIndex_merge___3BI_3BIJ( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jbyteArray jentry_value, jint jentry_value_len, jlong jcf_handle) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto* cf_handle = reinterpret_cast(jcf_handle); assert(cf_handle != nullptr); auto merge = [&wbwi, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) { wbwi->Merge(cf_handle, key, value); }; rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value, jentry_value_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: remove * Signature: ([BI)V */ void Java_org_rocksdb_WriteBatchWithIndex_remove___3BI( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto remove = [&wbwi] (rocksdb::Slice key) { wbwi->Delete(key); }; rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: remove * Signature: ([BIJ)V */ void Java_org_rocksdb_WriteBatchWithIndex_remove___3BIJ( JNIEnv* env, jobject jobj, jbyteArray jkey, jint jkey_len, jlong jcf_handle) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto* cf_handle = reinterpret_cast(jcf_handle); assert(cf_handle != nullptr); auto remove = [&wbwi, &cf_handle] (rocksdb::Slice key) { wbwi->Delete(cf_handle, key); }; rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: putLogData * Signature: ([BI)V */ void Java_org_rocksdb_WriteBatchWithIndex_putLogData( JNIEnv* env, jobject jobj, jbyteArray jblob, jint jblob_len) { auto* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); auto putLogData = [&wbwi] (rocksdb::Slice blob) { wbwi->PutLogData(blob); }; rocksdb::JniUtil::k_op(putLogData, env, jobj, jblob, jblob_len); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: clear * Signature: ()V */ void Java_org_rocksdb_WriteBatchWithIndex_clear0( JNIEnv* env, jobject jobj) { rocksdb::WriteBatchWithIndex* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); assert(wbwi != nullptr); wbwi->GetWriteBatch()->Clear(); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: iterator0 * Signature: ()J */ jlong Java_org_rocksdb_WriteBatchWithIndex_iterator0( JNIEnv* env, jobject jobj) { rocksdb::WriteBatchWithIndex* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); rocksdb::WBWIIterator* wbwi_iterator = wbwi->NewIterator(); return reinterpret_cast(wbwi_iterator); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: iterator1 * Signature: (J)J */ jlong Java_org_rocksdb_WriteBatchWithIndex_iterator1( JNIEnv* env, jobject jobj, jlong jcf_handle) { rocksdb::WriteBatchWithIndex* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); auto* cf_handle = reinterpret_cast(jcf_handle); rocksdb::WBWIIterator* wbwi_iterator = wbwi->NewIterator(cf_handle); return reinterpret_cast(wbwi_iterator); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: iteratorWithBase * Signature: (JJ)J */ jlong Java_org_rocksdb_WriteBatchWithIndex_iteratorWithBase( JNIEnv* env, jobject jobj, jlong jcf_handle, jlong jbi_handle) { rocksdb::WriteBatchWithIndex* wbwi = rocksdb::WriteBatchWithIndexJni::getHandle(env, jobj); auto* cf_handle = reinterpret_cast(jcf_handle); auto* base_iterator = reinterpret_cast(jbi_handle); auto* iterator = wbwi->NewIteratorWithBase(cf_handle, base_iterator); return reinterpret_cast(iterator); } /* * Class: org_rocksdb_WriteBatchWithIndex * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_WriteBatchWithIndex_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { auto* wbwi = reinterpret_cast(handle); delete wbwi; } /* WBWIRocksIterator below */ /* * Class: org_rocksdb_WBWIRocksIterator * Method: disposeInternal * Signature: (J)V */ void Java_org_rocksdb_WBWIRocksIterator_disposeInternal( JNIEnv* env, jobject jobj, jlong handle) { auto* it = reinterpret_cast(handle); delete it; } /* * Class: org_rocksdb_WBWIRocksIterator * Method: isValid0 * Signature: (J)Z */ jboolean Java_org_rocksdb_WBWIRocksIterator_isValid0( JNIEnv* env, jobject jobj, jlong handle) { return reinterpret_cast(handle)->Valid(); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: seekToFirst0 * Signature: (J)V */ void Java_org_rocksdb_WBWIRocksIterator_seekToFirst0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->SeekToFirst(); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: seekToLast0 * Signature: (J)V */ void Java_org_rocksdb_WBWIRocksIterator_seekToLast0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->SeekToLast(); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: next0 * Signature: (J)V */ void Java_org_rocksdb_WBWIRocksIterator_next0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->Next(); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: prev0 * Signature: (J)V */ void Java_org_rocksdb_WBWIRocksIterator_prev0( JNIEnv* env, jobject jobj, jlong handle) { reinterpret_cast(handle)->Prev(); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: seek0 * Signature: (J[BI)V */ void Java_org_rocksdb_WBWIRocksIterator_seek0( JNIEnv* env, jobject jobj, jlong handle, jbyteArray jtarget, jint jtarget_len) { auto* it = reinterpret_cast(handle); jbyte* target = env->GetByteArrayElements(jtarget, 0); rocksdb::Slice target_slice( reinterpret_cast(target), jtarget_len); it->Seek(target_slice); env->ReleaseByteArrayElements(jtarget, target, JNI_ABORT); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: status0 * Signature: (J)V */ void Java_org_rocksdb_WBWIRocksIterator_status0( JNIEnv* env, jobject jobj, jlong handle) { auto* it = reinterpret_cast(handle); rocksdb::Status s = it->status(); if (s.ok()) { return; } rocksdb::RocksDBExceptionJni::ThrowNew(env, s); } /* * Class: org_rocksdb_WBWIRocksIterator * Method: entry1 * Signature: (JLorg/rocksdb/WBWIRocksIterator/WriteEntry;)V */ void Java_org_rocksdb_WBWIRocksIterator_entry1( JNIEnv* env, jobject jobj, jlong handle, jobject jwrite_entry) { auto* it = reinterpret_cast(handle); const rocksdb::WriteEntry& we = it->Entry(); jobject jwe = rocksdb::WBWIRocksIteratorJni::getWriteEntry(env, jobj); rocksdb::WriteEntryJni::setWriteType(env, jwe, we.type); char* buf = new char[we.key.size()]; memcpy(buf, we.key.data(), we.key.size()); auto* key_slice = new rocksdb::Slice(buf, we.key.size()); rocksdb::WriteEntryJni::setKey(env, jwe, key_slice); if (we.type == rocksdb::kDeleteRecord || we.type == rocksdb::kLogDataRecord) { // set native handle of value slice to null if no value available rocksdb::WriteEntryJni::setValue(env, jwe, nullptr); } else { char* value_buf = new char[we.value.size()]; memcpy(value_buf, we.value.data(), we.value.size()); auto* value_slice = new rocksdb::Slice(value_buf, we.value.size()); rocksdb::WriteEntryJni::setValue(env, jwe, value_slice); } } rocksdb-4.5.1/java/rocksjni/writebatchhandlerjnicallback.cc000066400000000000000000000061211267530634000241000ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the callback "bridge" between Java and C++ for // rocksdb::Comparator. #include "rocksjni/writebatchhandlerjnicallback.h" #include "rocksjni/portal.h" namespace rocksdb { WriteBatchHandlerJniCallback::WriteBatchHandlerJniCallback( JNIEnv* env, jobject jWriteBatchHandler) : m_env(env) { // Note: we want to access the Java WriteBatchHandler instance // across multiple method calls, so we create a global ref m_jWriteBatchHandler = env->NewGlobalRef(jWriteBatchHandler); m_jPutMethodId = WriteBatchHandlerJni::getPutMethodId(env); m_jMergeMethodId = WriteBatchHandlerJni::getMergeMethodId(env); m_jDeleteMethodId = WriteBatchHandlerJni::getDeleteMethodId(env); m_jLogDataMethodId = WriteBatchHandlerJni::getLogDataMethodId(env); m_jContinueMethodId = WriteBatchHandlerJni::getContinueMethodId(env); } void WriteBatchHandlerJniCallback::Put(const Slice& key, const Slice& value) { const jbyteArray j_key = sliceToJArray(key); const jbyteArray j_value = sliceToJArray(value); m_env->CallVoidMethod( m_jWriteBatchHandler, m_jPutMethodId, j_key, j_value); m_env->DeleteLocalRef(j_value); m_env->DeleteLocalRef(j_key); } void WriteBatchHandlerJniCallback::Merge(const Slice& key, const Slice& value) { const jbyteArray j_key = sliceToJArray(key); const jbyteArray j_value = sliceToJArray(value); m_env->CallVoidMethod( m_jWriteBatchHandler, m_jMergeMethodId, j_key, j_value); m_env->DeleteLocalRef(j_value); m_env->DeleteLocalRef(j_key); } void WriteBatchHandlerJniCallback::Delete(const Slice& key) { const jbyteArray j_key = sliceToJArray(key); m_env->CallVoidMethod( m_jWriteBatchHandler, m_jDeleteMethodId, j_key); m_env->DeleteLocalRef(j_key); } void WriteBatchHandlerJniCallback::LogData(const Slice& blob) { const jbyteArray j_blob = sliceToJArray(blob); m_env->CallVoidMethod( m_jWriteBatchHandler, m_jLogDataMethodId, j_blob); m_env->DeleteLocalRef(j_blob); } bool WriteBatchHandlerJniCallback::Continue() { jboolean jContinue = m_env->CallBooleanMethod( m_jWriteBatchHandler, m_jContinueMethodId); return static_cast(jContinue == JNI_TRUE); } /* * Creates a Java Byte Array from the data in a Slice * * When calling this function * you must remember to call env->DeleteLocalRef * on the result after you have finished with it */ jbyteArray WriteBatchHandlerJniCallback::sliceToJArray(const Slice& s) { jbyteArray ja = m_env->NewByteArray(static_cast(s.size())); m_env->SetByteArrayRegion( ja, 0, static_cast(s.size()), reinterpret_cast(s.data())); return ja; } WriteBatchHandlerJniCallback::~WriteBatchHandlerJniCallback() { m_env->DeleteGlobalRef(m_jWriteBatchHandler); } } // namespace rocksdb rocksdb-4.5.1/java/rocksjni/writebatchhandlerjnicallback.h000066400000000000000000000030621267530634000237430ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // This file implements the callback "bridge" between Java and C++ for // rocksdb::WriteBatch::Handler. #ifndef JAVA_ROCKSJNI_WRITEBATCHHANDLERJNICALLBACK_H_ #define JAVA_ROCKSJNI_WRITEBATCHHANDLERJNICALLBACK_H_ #include #include "rocksdb/write_batch.h" namespace rocksdb { /** * This class acts as a bridge between C++ * and Java. The methods in this class will be * called back from the RocksDB storage engine (C++) * which calls the appropriate Java method. * This enables Write Batch Handlers to be implemented in Java. */ class WriteBatchHandlerJniCallback : public WriteBatch::Handler { public: WriteBatchHandlerJniCallback( JNIEnv* env, jobject jWriteBackHandler); ~WriteBatchHandlerJniCallback(); void Put(const Slice& key, const Slice& value); void Merge(const Slice& key, const Slice& value); void Delete(const Slice& key); void LogData(const Slice& blob); bool Continue(); private: JNIEnv* m_env; jobject m_jWriteBatchHandler; jbyteArray sliceToJArray(const Slice& s); jmethodID m_jPutMethodId; jmethodID m_jMergeMethodId; jmethodID m_jDeleteMethodId; jmethodID m_jLogDataMethodId; jmethodID m_jContinueMethodId; }; } // namespace rocksdb #endif // JAVA_ROCKSJNI_WRITEBATCHHANDLERJNICALLBACK_H_ rocksdb-4.5.1/java/samples/000077500000000000000000000000001267530634000155435ustar00rootroot00000000000000rocksdb-4.5.1/java/samples/src/000077500000000000000000000000001267530634000163325ustar00rootroot00000000000000rocksdb-4.5.1/java/samples/src/main/000077500000000000000000000000001267530634000172565ustar00rootroot00000000000000rocksdb-4.5.1/java/samples/src/main/java/000077500000000000000000000000001267530634000201775ustar00rootroot00000000000000rocksdb-4.5.1/java/samples/src/main/java/RocksDBColumnFamilySample.java000066400000000000000000000056631267530634000260250ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. import org.rocksdb.*; import java.util.ArrayList; import java.util.List; public class RocksDBColumnFamilySample { static { RocksDB.loadLibrary(); } public static void main(String[] args) throws RocksDBException { if (args.length < 1) { System.out.println( "usage: RocksDBColumnFamilySample db_path"); return; } String db_path = args[0]; System.out.println("RocksDBColumnFamilySample"); RocksDB db = null; Options options = null; ColumnFamilyHandle columnFamilyHandle = null; WriteBatch wb = null; try { options = new Options().setCreateIfMissing(true); db = RocksDB.open(options, db_path); assert(db != null); // create column family columnFamilyHandle = db.createColumnFamily( new ColumnFamilyDescriptor("new_cf".getBytes(), new ColumnFamilyOptions())); assert(columnFamilyHandle != null); } finally { if (columnFamilyHandle != null) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); db = null; } if (options != null) { options.dispose(); } } // open DB with two column families List columnFamilyDescriptors = new ArrayList<>(); // have to open default column family columnFamilyDescriptors.add(new ColumnFamilyDescriptor( RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); // open the new one, too columnFamilyDescriptors.add(new ColumnFamilyDescriptor( "new_cf".getBytes(), new ColumnFamilyOptions())); List columnFamilyHandles = new ArrayList<>(); try { db = RocksDB.open(new DBOptions(), db_path, columnFamilyDescriptors, columnFamilyHandles); assert(db != null); // put and get from non-default column family db.put(columnFamilyHandles.get(0), new WriteOptions(), "key".getBytes(), "value".getBytes()); String value = new String(db.get(columnFamilyHandles.get(0), "key".getBytes())); // atomic write wb = new WriteBatch(); wb.put(columnFamilyHandles.get(0), "key2".getBytes(), "value2".getBytes()); wb.put(columnFamilyHandles.get(1), "key3".getBytes(), "value3".getBytes()); wb.remove(columnFamilyHandles.get(0), "key".getBytes()); db.write(new WriteOptions(), wb); // drop column family db.dropColumnFamily(columnFamilyHandles.get(1)); } finally { for (ColumnFamilyHandle handle : columnFamilyHandles){ handle.dispose(); } if (db != null) { db.close(); } if (wb != null) { wb.dispose(); } } } } rocksdb-4.5.1/java/samples/src/main/java/RocksDBSample.java000066400000000000000000000253621267530634000235030ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. import java.lang.IllegalArgumentException; import java.util.Arrays; import java.util.List; import java.util.Map; import java.util.ArrayList; import org.rocksdb.*; import org.rocksdb.util.SizeUnit; import java.io.IOException; public class RocksDBSample { static { RocksDB.loadLibrary(); } public static void main(String[] args) { if (args.length < 1) { System.out.println("usage: RocksDBSample db_path"); return; } String db_path = args[0]; String db_path_not_found = db_path + "_not_found"; System.out.println("RocksDBSample"); RocksDB db = null; Options options = new Options(); try { db = RocksDB.open(options, db_path_not_found); assert(false); } catch (RocksDBException e) { System.out.format("caught the expceted exception -- %s\n", e); assert(db == null); } try { options.setCreateIfMissing(true) .createStatistics() .setWriteBufferSize(8 * SizeUnit.KB) .setMaxWriteBufferNumber(3) .setMaxBackgroundCompactions(10) .setCompressionType(CompressionType.SNAPPY_COMPRESSION) .setCompactionStyle(CompactionStyle.UNIVERSAL); } catch (IllegalArgumentException e) { assert(false); } Statistics stats = options.statisticsPtr(); assert(options.createIfMissing() == true); assert(options.writeBufferSize() == 8 * SizeUnit.KB); assert(options.maxWriteBufferNumber() == 3); assert(options.maxBackgroundCompactions() == 10); assert(options.compressionType() == CompressionType.SNAPPY_COMPRESSION); assert(options.compactionStyle() == CompactionStyle.UNIVERSAL); assert(options.memTableFactoryName().equals("SkipListFactory")); options.setMemTableConfig( new HashSkipListMemTableConfig() .setHeight(4) .setBranchingFactor(4) .setBucketCount(2000000)); assert(options.memTableFactoryName().equals("HashSkipListRepFactory")); options.setMemTableConfig( new HashLinkedListMemTableConfig() .setBucketCount(100000)); assert(options.memTableFactoryName().equals("HashLinkedListRepFactory")); options.setMemTableConfig( new VectorMemTableConfig().setReservedSize(10000)); assert(options.memTableFactoryName().equals("VectorRepFactory")); options.setMemTableConfig(new SkipListMemTableConfig()); assert(options.memTableFactoryName().equals("SkipListFactory")); options.setTableFormatConfig(new PlainTableConfig()); // Plain-Table requires mmap read options.setAllowMmapReads(true); assert(options.tableFactoryName().equals("PlainTable")); options.setRateLimiterConfig(new GenericRateLimiterConfig(10000000, 10000, 10)); options.setRateLimiterConfig(new GenericRateLimiterConfig(10000000)); Filter bloomFilter = new BloomFilter(10); BlockBasedTableConfig table_options = new BlockBasedTableConfig(); table_options.setBlockCacheSize(64 * SizeUnit.KB) .setFilter(bloomFilter) .setCacheNumShardBits(6) .setBlockSizeDeviation(5) .setBlockRestartInterval(10) .setCacheIndexAndFilterBlocks(true) .setHashIndexAllowCollision(false) .setBlockCacheCompressedSize(64 * SizeUnit.KB) .setBlockCacheCompressedNumShardBits(10); assert(table_options.blockCacheSize() == 64 * SizeUnit.KB); assert(table_options.cacheNumShardBits() == 6); assert(table_options.blockSizeDeviation() == 5); assert(table_options.blockRestartInterval() == 10); assert(table_options.cacheIndexAndFilterBlocks() == true); assert(table_options.hashIndexAllowCollision() == false); assert(table_options.blockCacheCompressedSize() == 64 * SizeUnit.KB); assert(table_options.blockCacheCompressedNumShardBits() == 10); options.setTableFormatConfig(table_options); assert(options.tableFactoryName().equals("BlockBasedTable")); try { db = RocksDB.open(options, db_path); db.put("hello".getBytes(), "world".getBytes()); byte[] value = db.get("hello".getBytes()); assert("world".equals(new String(value))); String str = db.getProperty("rocksdb.stats"); assert(str != null && !str.equals("")); } catch (RocksDBException e) { System.out.format("[ERROR] caught the unexpceted exception -- %s\n", e); assert(db == null); assert(false); } // be sure to release the c++ pointer db.close(); ReadOptions readOptions = new ReadOptions(); readOptions.setFillCache(false); try { db = RocksDB.open(options, db_path); db.put("hello".getBytes(), "world".getBytes()); byte[] value = db.get("hello".getBytes()); System.out.format("Get('hello') = %s\n", new String(value)); for (int i = 1; i <= 9; ++i) { for (int j = 1; j <= 9; ++j) { db.put(String.format("%dx%d", i, j).getBytes(), String.format("%d", i * j).getBytes()); } } for (int i = 1; i <= 9; ++i) { for (int j = 1; j <= 9; ++j) { System.out.format("%s ", new String(db.get( String.format("%dx%d", i, j).getBytes()))); } System.out.println(""); } // write batch test WriteOptions writeOpt = new WriteOptions(); for (int i = 10; i <= 19; ++i) { WriteBatch batch = new WriteBatch(); for (int j = 10; j <= 19; ++j) { batch.put(String.format("%dx%d", i, j).getBytes(), String.format("%d", i * j).getBytes()); } db.write(writeOpt, batch); batch.dispose(); } for (int i = 10; i <= 19; ++i) { for (int j = 10; j <= 19; ++j) { assert(new String( db.get(String.format("%dx%d", i, j).getBytes())).equals( String.format("%d", i * j))); System.out.format("%s ", new String(db.get( String.format("%dx%d", i, j).getBytes()))); } System.out.println(""); } writeOpt.dispose(); value = db.get("1x1".getBytes()); assert(value != null); value = db.get("world".getBytes()); assert(value == null); value = db.get(readOptions, "world".getBytes()); assert(value == null); byte[] testKey = "asdf".getBytes(); byte[] testValue = "asdfghjkl;'?> insufficientArray.length); len = db.get("asdfjkl;".getBytes(), enoughArray); assert(len == RocksDB.NOT_FOUND); len = db.get(testKey, enoughArray); assert(len == testValue.length); len = db.get(readOptions, testKey, insufficientArray); assert(len > insufficientArray.length); len = db.get(readOptions, "asdfjkl;".getBytes(), enoughArray); assert(len == RocksDB.NOT_FOUND); len = db.get(readOptions, testKey, enoughArray); assert(len == testValue.length); db.remove(testKey); len = db.get(testKey, enoughArray); assert(len == RocksDB.NOT_FOUND); // repeat the test with WriteOptions WriteOptions writeOpts = new WriteOptions(); writeOpts.setSync(true); writeOpts.setDisableWAL(true); db.put(writeOpts, testKey, testValue); len = db.get(testKey, enoughArray); assert(len == testValue.length); assert(new String(testValue).equals( new String(enoughArray, 0, len))); writeOpts.dispose(); try { for (TickerType statsType : TickerType.values()) { stats.getTickerCount(statsType); } System.out.println("getTickerCount() passed."); } catch (Exception e) { System.out.println("Failed in call to getTickerCount()"); assert(false); //Should never reach here. } try { for (HistogramType histogramType : HistogramType.values()) { HistogramData data = stats.geHistogramData(histogramType); } System.out.println("geHistogramData() passed."); } catch (Exception e) { System.out.println("Failed in call to geHistogramData()"); assert(false); //Should never reach here. } RocksIterator iterator = db.newIterator(); boolean seekToFirstPassed = false; for (iterator.seekToFirst(); iterator.isValid(); iterator.next()) { iterator.status(); assert(iterator.key() != null); assert(iterator.value() != null); seekToFirstPassed = true; } if(seekToFirstPassed) { System.out.println("iterator seekToFirst tests passed."); } boolean seekToLastPassed = false; for (iterator.seekToLast(); iterator.isValid(); iterator.prev()) { iterator.status(); assert(iterator.key() != null); assert(iterator.value() != null); seekToLastPassed = true; } if(seekToLastPassed) { System.out.println("iterator seekToLastPassed tests passed."); } iterator.seekToFirst(); iterator.seek(iterator.key()); assert(iterator.key() != null); assert(iterator.value() != null); System.out.println("iterator seek test passed."); iterator.dispose(); System.out.println("iterator tests passed."); iterator = db.newIterator(); List keys = new ArrayList(); for (iterator.seekToLast(); iterator.isValid(); iterator.prev()) { keys.add(iterator.key()); } iterator.dispose(); Map values = db.multiGet(keys); assert(values.size() == keys.size()); for(byte[] value1 : values.values()) { assert(value1 != null); } values = db.multiGet(new ReadOptions(), keys); assert(values.size() == keys.size()); for(byte[] value1 : values.values()) { assert(value1 != null); } } catch (RocksDBException e) { System.err.println(e); } if (db != null) { db.close(); } // be sure to dispose c++ pointers options.dispose(); readOptions.dispose(); } } rocksdb-4.5.1/java/src/000077500000000000000000000000001267530634000146665ustar00rootroot00000000000000rocksdb-4.5.1/java/src/main/000077500000000000000000000000001267530634000156125ustar00rootroot00000000000000rocksdb-4.5.1/java/src/main/java/000077500000000000000000000000001267530634000165335ustar00rootroot00000000000000rocksdb-4.5.1/java/src/main/java/org/000077500000000000000000000000001267530634000173225ustar00rootroot00000000000000rocksdb-4.5.1/java/src/main/java/org/rocksdb/000077500000000000000000000000001267530634000207515ustar00rootroot00000000000000rocksdb-4.5.1/java/src/main/java/org/rocksdb/AbstractCompactionFilter.java000066400000000000000000000017711267530634000265500ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * A CompactionFilter allows an application to modify/delete a key-value at * the time of compaction. * * At present we just permit an overriding Java class to wrap a C++ implementation */ public abstract class AbstractCompactionFilter> extends RocksObject { /** * Deletes underlying C++ comparator pointer. * * Note that this function should be called only after all * RocksDB instances referencing the comparator are closed. * Otherwise an undefined behavior will occur. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/AbstractComparator.java000066400000000000000000000057641267530634000254230ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Comparators are used by RocksDB to determine * the ordering of keys. * * This class is package private, implementers * should extend either of the public abstract classes: * @see org.rocksdb.Comparator * @see org.rocksdb.DirectComparator */ public abstract class AbstractComparator> extends RocksObject { /** * The name of the comparator. Used to check for comparator * mismatches (i.e., a DB created with one comparator is * accessed using a different comparator). * * A new name should be used whenever * the comparator implementation changes in a way that will cause * the relative ordering of any two keys to change. * * Names starting with "rocksdb." are reserved and should not be used. * * @return The name of this comparator implementation */ public abstract String name(); /** * Three-way key comparison * * @param a Slice access to first key * @param b Slice access to second key * * @return Should return either: * 1) < 0 if "a" < "b" * 2) == 0 if "a" == "b" * 3) > 0 if "a" > "b" */ public abstract int compare(final T a, final T b); /** *

Used to reduce the space requirements * for internal data structures like index blocks.

* *

If start < limit, you may return a new start which is a * shorter string in [start, limit).

* *

Simple comparator implementations may return null if they * wish to use start unchanged. i.e., an implementation of * this method that does nothing is correct.

* * @param start String * @param limit of type T * * @return a shorter start, or null */ public String findShortestSeparator(final String start, final T limit) { return null; } /** *

Used to reduce the space requirements * for internal data structures like index blocks.

* *

You may return a new short key (key1) where * key1 ≥ key.

* *

Simple comparator implementations may return null if they * wish to leave the key unchanged. i.e., an implementation of * this method that does nothing is correct.

* * @param key String * * @return a shorter key, or null */ public String findShortSuccessor(final String key) { return null; } /** * Deletes underlying C++ comparator pointer. * * Note that this function should be called only after all * RocksDB instances referencing the comparator are closed. * Otherwise an undefined behavior will occur. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/AbstractRocksIterator.java000066400000000000000000000061331267530634000260760ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Base class implementation for Rocks Iterators * in the Java API * *

Multiple threads can invoke const methods on an RocksIterator without * external synchronization, but if any of the threads may call a * non-const method, all threads accessing the same RocksIterator must use * external synchronization.

* * @param

The type of the Parent Object from which the Rocks Iterator was * created. This is used by disposeInternal to avoid double-free * issues with the underlying C++ object. * @see org.rocksdb.RocksObject */ public abstract class AbstractRocksIterator

extends RocksObject implements RocksIteratorInterface { final P parent_; protected AbstractRocksIterator(final P parent, final long nativeHandle) { super(); nativeHandle_ = nativeHandle; // parent must point to a valid RocksDB instance. assert (parent != null); // RocksIterator must hold a reference to the related parent instance // to guarantee that while a GC cycle starts RocksIterator instances // are freed prior to parent instances. parent_ = parent; } @Override public boolean isValid() { assert (isInitialized()); return isValid0(nativeHandle_); } @Override public void seekToFirst() { assert (isInitialized()); seekToFirst0(nativeHandle_); } @Override public void seekToLast() { assert (isInitialized()); seekToLast0(nativeHandle_); } @Override public void seek(byte[] target) { assert (isInitialized()); seek0(nativeHandle_, target, target.length); } @Override public void next() { assert (isInitialized()); next0(nativeHandle_); } @Override public void prev() { assert (isInitialized()); prev0(nativeHandle_); } @Override public void status() throws RocksDBException { assert (isInitialized()); status0(nativeHandle_); } /** *

Deletes underlying C++ iterator pointer.

* *

Note: the underlying handle can only be safely deleted if the parent * instance related to a certain RocksIterator is still valid and initialized. * Therefore {@code disposeInternal()} checks if the parent is initialized * before freeing the native handle.

*/ @Override protected void disposeInternal() { synchronized (parent_) { assert (isInitialized()); if (parent_.isInitialized()) { disposeInternal(nativeHandle_); } } } abstract void disposeInternal(long handle); abstract boolean isValid0(long handle); abstract void seekToFirst0(long handle); abstract void seekToLast0(long handle); abstract void next0(long handle); abstract void prev0(long handle); abstract void seek0(long handle, byte[] target, int targetLen); abstract void status0(long handle) throws RocksDBException; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/AbstractSlice.java000066400000000000000000000113041267530634000243360ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Slices are used by RocksDB to provide * efficient access to keys and values. * * This class is package private, implementers * should extend either of the public abstract classes: * @see org.rocksdb.Slice * @see org.rocksdb.DirectSlice * * Regards the lifecycle of Java Slices in RocksDB: * At present when you configure a Comparator from Java, it creates an * instance of a C++ BaseComparatorJniCallback subclass and * passes that to RocksDB as the comparator. That subclass of * BaseComparatorJniCallback creates the Java * @see org.rocksdb.AbstractSlice subclass Objects. When you dispose * the Java @see org.rocksdb.AbstractComparator subclass, it disposes the * C++ BaseComparatorJniCallback subclass, which in turn destroys the * Java @see org.rocksdb.AbstractSlice subclass Objects. */ abstract class AbstractSlice extends RocksObject { /** * Returns the data of the slice. * * @return The slice data. Note, the type of access is * determined by the subclass * @see org.rocksdb.AbstractSlice#data0(long) */ public T data() { assert (isInitialized()); return data0(nativeHandle_); } /** * Access to the data is provided by the * subtype as it needs to handle the * generic typing. * * @param handle The address of the underlying * native object. * * @return Java typed access to the data. */ protected abstract T data0(long handle); /** * Return the length (in bytes) of the data. * * @return The length in bytes. */ public int size() { assert (isInitialized()); return size0(nativeHandle_); } /** * Return true if the length of the * data is zero. * * @return true if there is no data, false otherwise. */ public boolean empty() { assert (isInitialized()); return empty0(nativeHandle_); } /** * Creates a string representation of the data * * @param hex When true, the representation * will be encoded in hexadecimal. * * @return The string representation of the data. */ public String toString(final boolean hex) { assert (isInitialized()); return toString0(nativeHandle_, hex); } @Override public String toString() { return toString(false); } /** * Three-way key comparison * * @param other A slice to compare against * * @return Should return either: * 1) < 0 if this < other * 2) == 0 if this == other * 3) > 0 if this > other */ public int compare(final AbstractSlice other) { assert (other != null); assert (isInitialized()); return compare0(nativeHandle_, other.nativeHandle_); } @Override public int hashCode() { return toString().hashCode(); } /** * If other is a slice object, then * we defer to {@link #compare(AbstractSlice) compare} * to check equality, otherwise we return false. * * @param other Object to test for equality * * @return true when {@code this.compare(other) == 0}, * false otherwise. */ @Override public boolean equals(final Object other) { if (other != null && other instanceof AbstractSlice) { return compare((AbstractSlice)other) == 0; } else { return false; } } /** * Determines whether this slice starts with * another slice * * @param prefix Another slice which may of may not * be a prefix of this slice. * * @return true when this slice starts with the * {@code prefix} slice */ public boolean startsWith(final AbstractSlice prefix) { if (prefix != null) { assert (isInitialized()); return startsWith0(nativeHandle_, prefix.nativeHandle_); } else { return false; } } /** * Deletes underlying C++ slice pointer. * Note that this function should be called only after all * RocksDB instances referencing the slice are closed. * Otherwise an undefined behavior will occur. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } protected native void createNewSliceFromString(String str); private native int size0(long handle); private native boolean empty0(long handle); private native String toString0(long handle, boolean hex); private native int compare0(long handle, long otherHandle); private native boolean startsWith0(long handle, long otherHandle); private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/AbstractWriteBatch.java000066400000000000000000000046561267530634000253470ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public abstract class AbstractWriteBatch extends RocksObject implements WriteBatchInterface { @Override public int count() { assert (isInitialized()); return count0(); } @Override public void put(byte[] key, byte[] value) { assert (isInitialized()); put(key, key.length, value, value.length); } @Override public void put(ColumnFamilyHandle columnFamilyHandle, byte[] key, byte[] value) { assert (isInitialized()); put(key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } @Override public void merge(byte[] key, byte[] value) { assert (isInitialized()); merge(key, key.length, value, value.length); } @Override public void merge(ColumnFamilyHandle columnFamilyHandle, byte[] key, byte[] value) { assert (isInitialized()); merge(key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } @Override public void remove(byte[] key) { assert (isInitialized()); remove(key, key.length); } @Override public void remove(ColumnFamilyHandle columnFamilyHandle, byte[] key) { assert (isInitialized()); remove(key, key.length, columnFamilyHandle.nativeHandle_); } @Override public void putLogData(byte[] blob) { assert (isInitialized()); putLogData(blob, blob.length); } @Override public void clear() { assert (isInitialized()); clear0(); } /** * Delete the c++ side pointer. */ @Override protected void disposeInternal() { assert (isInitialized()); disposeInternal(nativeHandle_); } abstract void disposeInternal(long handle); abstract int count0(); abstract void put(byte[] key, int keyLen, byte[] value, int valueLen); abstract void put(byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle); abstract void merge(byte[] key, int keyLen, byte[] value, int valueLen); abstract void merge(byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle); abstract void remove(byte[] key, int keyLen); abstract void remove(byte[] key, int keyLen, long cfHandle); abstract void putLogData(byte[] blob, int blobLen); abstract void clear0(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BackupEngine.java000066400000000000000000000171401267530634000241520ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; /** * BackupEngine allows you to backup * and restore the database * * Be aware, that `new BackupEngine` takes time proportional to the amount * of backups. So if you have a slow filesystem to backup (like HDFS) * and you have a lot of backups then restoring can take some time. * That's why we recommend to limit the number of backups. * Also we recommend to keep BackupEngine alive and not to recreate it every * time you need to do a backup. */ public class BackupEngine extends RocksObject implements AutoCloseable { protected BackupEngine() { super(); } /** * Opens a new Backup Engine * * @param env The environment that the backup engine should operate within * @param options Any options for the backup engine * * @return A new BackupEngine instance */ public static BackupEngine open(final Env env, final BackupableDBOptions options) throws RocksDBException { final BackupEngine be = new BackupEngine(); be.open(env.nativeHandle_, options.nativeHandle_); return be; } /** * Captures the state of the database in the latest backup * * Just a convenience for {@link #createNewBackup(RocksDB, boolean)} with * the flushBeforeBackup parameter set to false * * @param db The database to backup * * Note - This method is not thread safe */ public void createNewBackup(final RocksDB db) throws RocksDBException { createNewBackup(db, false); } /** * Captures the state of the database in the latest backup * * @param db The database to backup * @param flushBeforeBackup When true, the Backup Engine will first issue a * memtable flush and only then copy the DB files to * the backup directory. Doing so will prevent log * files from being copied to the backup directory * (since flush will delete them). * When false, the Backup Engine will not issue a * flush before starting the backup. In that case, * the backup will also include log files * corresponding to live memtables. The backup will * always be consistent with the current state of the * database regardless of the flushBeforeBackup * parameter. * * Note - This method is not thread safe */ public void createNewBackup( final RocksDB db, final boolean flushBeforeBackup) throws RocksDBException { assert (isInitialized()); createNewBackup(nativeHandle_, db.nativeHandle_, flushBeforeBackup); } /** * Gets information about the available * backups * * @return A list of information about each available backup */ public List getBackupInfo() { assert (isInitialized()); return getBackupInfo(nativeHandle_); } /** *

Returns a list of corrupted backup ids. If there * is no corrupted backup the method will return an * empty list.

* * @return array of backup ids as int ids. */ public int[] getCorruptedBackups() { assert(isInitialized()); return getCorruptedBackups(nativeHandle_); } /** *

Will delete all the files we don't need anymore. It will * do the full scan of the files/ directory and delete all the * files that are not referenced.

* * @throws RocksDBException thrown if error happens in underlying * native library. */ public void garbageCollect() throws RocksDBException { assert(isInitialized()); garbageCollect(nativeHandle_); } /** * Deletes old backups, keeping just the latest numBackupsToKeep * * @param numBackupsToKeep The latest n backups to keep */ public void purgeOldBackups( final int numBackupsToKeep) throws RocksDBException { assert (isInitialized()); purgeOldBackups(nativeHandle_, numBackupsToKeep); } /** * Deletes a backup * * @param backupId The id of the backup to delete */ public void deleteBackup(final int backupId) throws RocksDBException { assert (isInitialized()); deleteBackup(nativeHandle_, backupId); } /** * Restore the database from a backup * * IMPORTANT: if options.share_table_files == true and you restore the DB * from some backup that is not the latest, and you start creating new * backups from the new DB, they will probably fail! * * Example: Let's say you have backups 1, 2, 3, 4, 5 and you restore 3. * If you add new data to the DB and try creating a new backup now, the * database will diverge from backups 4 and 5 and the new backup will fail. * If you want to create new backup, you will first have to delete backups 4 * and 5. * * @param backupId The id of the backup to restore * @param dbDir The directory to restore the backup to, i.e. where your * database is * @param walDir The location of the log files for your database, * often the same as dbDir * @param restoreOptions Options for controlling the restore */ public void restoreDbFromBackup( final int backupId, final String dbDir, final String walDir, final RestoreOptions restoreOptions) throws RocksDBException { assert (isInitialized()); restoreDbFromBackup(nativeHandle_, backupId, dbDir, walDir, restoreOptions.nativeHandle_); } /** * Restore the database from the latest backup * * @param dbDir The directory to restore the backup to, i.e. where your database is * @param walDir The location of the log files for your database, often the same as dbDir * @param restoreOptions Options for controlling the restore */ public void restoreDbFromLatestBackup( final String dbDir, final String walDir, final RestoreOptions restoreOptions) throws RocksDBException { assert (isInitialized()); restoreDbFromLatestBackup(nativeHandle_, dbDir, walDir, restoreOptions.nativeHandle_); } /** * Close the Backup Engine */ @Override public void close() throws RocksDBException { dispose(); } @Override protected void disposeInternal() { assert (isInitialized()); disposeInternal(nativeHandle_); } private native void open(final long env, final long backupableDbOptions) throws RocksDBException; private native void createNewBackup(final long handle, final long dbHandle, final boolean flushBeforeBackup) throws RocksDBException; private native List getBackupInfo(final long handle); private native int[] getCorruptedBackups(final long handle); private native void garbageCollect(final long handle) throws RocksDBException; private native void purgeOldBackups(final long handle, final int numBackupsToKeep) throws RocksDBException; private native void deleteBackup(final long handle, final int backupId) throws RocksDBException; private native void restoreDbFromBackup(final long handle, final int backupId, final String dbDir, final String walDir, final long restoreOptionsHandle) throws RocksDBException; private native void restoreDbFromLatestBackup(final long handle, final String dbDir, final String walDir, final long restoreOptionsHandle) throws RocksDBException; private native void disposeInternal(final long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BackupInfo.java000066400000000000000000000030041267530634000236320ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Instances of this class describe a Backup made by * {@link org.rocksdb.BackupableDB}. */ public class BackupInfo { /** * Package private constructor used to create instances * of BackupInfo by {@link org.rocksdb.BackupableDB} and * {@link org.rocksdb.RestoreBackupableDB}. * * @param backupId id of backup * @param timestamp timestamp of backup * @param size size of backup * @param numberFiles number of files related to this backup. */ BackupInfo(final int backupId, final long timestamp, final long size, final int numberFiles) { backupId_ = backupId; timestamp_ = timestamp; size_ = size; numberFiles_ = numberFiles; } /** * * @return the backup id. */ public int backupId() { return backupId_; } /** * * @return the timestamp of the backup. */ public long timestamp() { return timestamp_; } /** * * @return the size of the backup */ public long size() { return size_; } /** * * @return the number of files of this backup. */ public int numberFiles() { return numberFiles_; } private int backupId_; private long timestamp_; private long size_; private int numberFiles_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BackupableDB.java000066400000000000000000000122041267530634000240520ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; /** *

A subclass of RocksDB which supports * backup-related operations.

* * @see org.rocksdb.BackupableDBOptions */ public class BackupableDB extends RocksDB { /** *

Open a {@code BackupableDB} under the specified path. * Note that the backup path should be set properly in the * input BackupableDBOptions.

* * @param opt {@link org.rocksdb.Options} to set for the database. * @param bopt {@link org.rocksdb.BackupableDBOptions} to use. * @param db_path Path to store data to. The path for storing the backup should be * specified in the {@link org.rocksdb.BackupableDBOptions}. * * @return {@link BackupableDB} reference to the opened database. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public static BackupableDB open( final Options opt, final BackupableDBOptions bopt, final String db_path) throws RocksDBException { RocksDB db = RocksDB.open(opt, db_path); BackupableDB bdb = new BackupableDB(); bdb.open(db.nativeHandle_, bopt.nativeHandle_); // Prevent the RocksDB object from attempting to delete // the underly C++ DB object. db.disOwnNativeHandle(); return bdb; } /** *

Captures the state of the database in the latest backup. * Note that this function is not thread-safe.

* * @param flushBeforeBackup if true, then all data will be flushed * before creating backup. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void createNewBackup(final boolean flushBeforeBackup) throws RocksDBException { assert(isInitialized()); createNewBackup(nativeHandle_, flushBeforeBackup); } /** *

Deletes old backups, keeping latest numBackupsToKeep alive.

* * @param numBackupsToKeep Number of latest backups to keep. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void purgeOldBackups(final int numBackupsToKeep) throws RocksDBException { assert(isInitialized()); purgeOldBackups(nativeHandle_, numBackupsToKeep); } /** *

Deletes a specific backup.

* * @param backupId of backup to delete. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void deleteBackup(final int backupId) throws RocksDBException { assert(isInitialized()); deleteBackup0(nativeHandle_, backupId); } /** *

Returns a list of {@link BackupInfo} instances, which describe * already made backups.

* * @return List of {@link BackupInfo} instances. */ public List getBackupInfos() { assert(isInitialized()); return getBackupInfo(nativeHandle_); } /** *

Returns a list of corrupted backup ids. If there * is no corrupted backup the method will return an * empty list.

* * @return array of backup ids as int ids. */ public int[] getCorruptedBackups() { assert(isInitialized()); return getCorruptedBackups(nativeHandle_); } /** *

Will delete all the files we don't need anymore. It will * do the full scan of the files/ directory and delete all the * files that are not referenced.

* * @throws RocksDBException thrown if error happens in underlying * native library. */ public void garbageCollect() throws RocksDBException { assert(isInitialized()); garbageCollect(nativeHandle_); } /** *

Close the BackupableDB instance and release resource.

* *

Internally, {@link BackupableDB} owns the {@code rocksdb::DB} * pointer to its associated {@link org.rocksdb.RocksDB}. * The release of that RocksDB pointer is handled in the destructor * of the c++ {@code rocksdb::BackupableDB} and should be transparent * to Java developers.

*/ @Override public synchronized void close() { if (isInitialized()) { super.close(); } } /** *

A protected construction that will be used in the static * factory method {@link #open(Options, BackupableDBOptions, String)}. *

*/ protected BackupableDB() { super(); } @Override protected void finalize() throws Throwable { close(); super.finalize(); } protected native void open(long rocksDBHandle, long backupDBOptionsHandle); protected native void createNewBackup(long handle, boolean flag) throws RocksDBException; protected native void purgeOldBackups(long handle, int numBackupsToKeep) throws RocksDBException; private native void deleteBackup0(long nativeHandle, int backupId) throws RocksDBException; protected native List getBackupInfo(long handle); private native int[] getCorruptedBackups(long handle); private native void garbageCollect(long handle) throws RocksDBException; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BackupableDBOptions.java000066400000000000000000000214571267530634000254400ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.io.File; import java.nio.file.Path; /** *

BackupableDBOptions to control the behavior of a backupable database. * It will be used during the creation of a {@link org.rocksdb.BackupableDB}. *

*

Note that dispose() must be called before an Options instance * become out-of-scope to release the allocated memory in c++.

* * @see org.rocksdb.BackupableDB */ public class BackupableDBOptions extends RocksObject { /** *

BackupableDBOptions constructor.

* * @param path Where to keep the backup files. Has to be different than db name. * Best to set this to {@code db name_ + "/backups"} * @throws java.lang.IllegalArgumentException if illegal path is used. */ public BackupableDBOptions(final String path) { super(); File backupPath = path == null ? null : new File(path); if (backupPath == null || !backupPath.isDirectory() || !backupPath.canWrite()) { throw new IllegalArgumentException("Illegal path provided."); } newBackupableDBOptions(path); } /** *

Returns the path to the BackupableDB directory.

* * @return the path to the BackupableDB directory. */ public String backupDir() { assert(isInitialized()); return backupDir(nativeHandle_); } /** *

Share table files between backups.

* * @param shareTableFiles If {@code share_table_files == true}, backup will assume * that table files with same name have the same contents. This enables incremental * backups and avoids unnecessary data copies. If {@code share_table_files == false}, * each backup will be on its own and will not share any data with other backups. * *

Default: true

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setShareTableFiles(final boolean shareTableFiles) { assert(isInitialized()); setShareTableFiles(nativeHandle_, shareTableFiles); return this; } /** *

Share table files between backups.

* * @return boolean value indicating if SST files will be shared between * backups. */ public boolean shareTableFiles() { assert(isInitialized()); return shareTableFiles(nativeHandle_); } /** *

Set synchronous backups.

* * @param sync If {@code sync == true}, we can guarantee you'll get consistent backup * even on a machine crash/reboot. Backup process is slower with sync enabled. * If {@code sync == false}, we don't guarantee anything on machine reboot. * However,chances are some of the backups are consistent. * *

Default: true

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setSync(final boolean sync) { assert(isInitialized()); setSync(nativeHandle_, sync); return this; } /** *

Are synchronous backups activated.

* * @return boolean value if synchronous backups are configured. */ public boolean sync() { assert(isInitialized()); return sync(nativeHandle_); } /** *

Set if old data will be destroyed.

* * @param destroyOldData If true, it will delete whatever backups there are already. * *

Default: false

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setDestroyOldData(final boolean destroyOldData) { assert(isInitialized()); setDestroyOldData(nativeHandle_, destroyOldData); return this; } /** *

Returns if old data will be destroyed will performing new backups.

* * @return boolean value indicating if old data will be destroyed. */ public boolean destroyOldData() { assert(isInitialized()); return destroyOldData(nativeHandle_); } /** *

Set if log files shall be persisted.

* * @param backupLogFiles If false, we won't backup log files. This option can be * useful for backing up in-memory databases where log file are persisted,but table * files are in memory. * *

Default: true

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setBackupLogFiles(final boolean backupLogFiles) { assert(isInitialized()); setBackupLogFiles(nativeHandle_, backupLogFiles); return this; } /** *

Return information if log files shall be persisted.

* * @return boolean value indicating if log files will be persisted. */ public boolean backupLogFiles() { assert(isInitialized()); return backupLogFiles(nativeHandle_); } /** *

Set backup rate limit.

* * @param backupRateLimit Max bytes that can be transferred in a second during backup. * If 0 or negative, then go as fast as you can. * *

Default: 0

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setBackupRateLimit(long backupRateLimit) { assert(isInitialized()); backupRateLimit = (backupRateLimit <= 0) ? 0 : backupRateLimit; setBackupRateLimit(nativeHandle_, backupRateLimit); return this; } /** *

Return backup rate limit which described the max bytes that can be transferred in a * second during backup.

* * @return numerical value describing the backup transfer limit in bytes per second. */ public long backupRateLimit() { assert(isInitialized()); return backupRateLimit(nativeHandle_); } /** *

Set restore rate limit.

* * @param restoreRateLimit Max bytes that can be transferred in a second during restore. * If 0 or negative, then go as fast as you can. * *

Default: 0

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setRestoreRateLimit(long restoreRateLimit) { assert(isInitialized()); restoreRateLimit = (restoreRateLimit <= 0) ? 0 : restoreRateLimit; setRestoreRateLimit(nativeHandle_, restoreRateLimit); return this; } /** *

Return restore rate limit which described the max bytes that can be transferred in a * second during restore.

* * @return numerical value describing the restore transfer limit in bytes per second. */ public long restoreRateLimit() { assert(isInitialized()); return restoreRateLimit(nativeHandle_); } /** *

Only used if share_table_files is set to true. If true, will consider that * backups can come from different databases, hence a sst is not uniquely * identified by its name, but by the triple (file name, crc32, file length)

* * @param shareFilesWithChecksum boolean value indicating if SST files are stored * using the triple (file name, crc32, file length) and not its name. * *

Note: this is an experimental option, and you'll need to set it manually * turn it on only if you know what you're doing*

* *

Default: false

* * @return instance of current BackupableDBOptions. */ public BackupableDBOptions setShareFilesWithChecksum( final boolean shareFilesWithChecksum) { assert(isInitialized()); setShareFilesWithChecksum(nativeHandle_, shareFilesWithChecksum); return this; } /** *

Return of share files with checksum is active.

* * @return boolean value indicating if share files with checksum * is active. */ public boolean shareFilesWithChecksum() { assert(isInitialized()); return shareFilesWithChecksum(nativeHandle_); } /** * Release the memory allocated for the current instance * in the c++ side. */ @Override protected void disposeInternal() { disposeInternal(nativeHandle_); } private native void newBackupableDBOptions(String path); private native String backupDir(long handle); private native void setShareTableFiles(long handle, boolean flag); private native boolean shareTableFiles(long handle); private native void setSync(long handle, boolean flag); private native boolean sync(long handle); private native void setDestroyOldData(long handle, boolean flag); private native boolean destroyOldData(long handle); private native void setBackupLogFiles(long handle, boolean flag); private native boolean backupLogFiles(long handle); private native void setBackupRateLimit(long handle, long rateLimit); private native long backupRateLimit(long handle); private native void setRestoreRateLimit(long handle, long rateLimit); private native long restoreRateLimit(long handle); private native void setShareFilesWithChecksum(long handle, boolean flag); private native boolean shareFilesWithChecksum(long handle); private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BlockBasedTableConfig.java000066400000000000000000000323071267530634000257100ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * The config for plain table sst format. * * BlockBasedTable is a RocksDB's default SST file format. */ public class BlockBasedTableConfig extends TableFormatConfig { public BlockBasedTableConfig() { noBlockCache_ = false; blockCacheSize_ = 8 * 1024 * 1024; blockCacheNumShardBits_ = 0; blockSize_ = 4 * 1024; blockSizeDeviation_ = 10; blockRestartInterval_ = 16; wholeKeyFiltering_ = true; filter_ = null; cacheIndexAndFilterBlocks_ = false; hashIndexAllowCollision_ = true; blockCacheCompressedSize_ = 0; blockCacheCompressedNumShardBits_ = 0; checksumType_ = ChecksumType.kCRC32c; indexType_ = IndexType.kBinarySearch; formatVersion_ = 0; } /** * Disable block cache. If this is set to true, * then no block cache should be used, and the block_cache should * point to a {@code nullptr} object. * Default: false * * @param noBlockCache if use block cache * @return the reference to the current config. */ public BlockBasedTableConfig setNoBlockCache(final boolean noBlockCache) { noBlockCache_ = noBlockCache; return this; } /** * @return if block cache is disabled */ public boolean noBlockCache() { return noBlockCache_; } /** * Set the amount of cache in bytes that will be used by RocksDB. * If cacheSize is non-positive, then cache will not be used. * DEFAULT: 8M * * @param blockCacheSize block cache size in bytes * @return the reference to the current config. */ public BlockBasedTableConfig setBlockCacheSize(final long blockCacheSize) { blockCacheSize_ = blockCacheSize; return this; } /** * @return block cache size in bytes */ public long blockCacheSize() { return blockCacheSize_; } /** * Controls the number of shards for the block cache. * This is applied only if cacheSize is set to non-negative. * * @param blockCacheNumShardBits the number of shard bits. The resulting * number of shards would be 2 ^ numShardBits. Any negative * number means use default settings." * @return the reference to the current option. */ public BlockBasedTableConfig setCacheNumShardBits( final int blockCacheNumShardBits) { blockCacheNumShardBits_ = blockCacheNumShardBits; return this; } /** * Returns the number of shard bits used in the block cache. * The resulting number of shards would be 2 ^ (returned value). * Any negative number means use default settings. * * @return the number of shard bits used in the block cache. */ public int cacheNumShardBits() { return blockCacheNumShardBits_; } /** * Approximate size of user data packed per block. Note that the * block size specified here corresponds to uncompressed data. The * actual size of the unit read from disk may be smaller if * compression is enabled. This parameter can be changed dynamically. * Default: 4K * * @param blockSize block size in bytes * @return the reference to the current config. */ public BlockBasedTableConfig setBlockSize(final long blockSize) { blockSize_ = blockSize; return this; } /** * @return block size in bytes */ public long blockSize() { return blockSize_; } /** * This is used to close a block before it reaches the configured * 'block_size'. If the percentage of free space in the current block is less * than this specified number and adding a new record to the block will * exceed the configured block size, then this block will be closed and the * new record will be written to the next block. * Default is 10. * * @param blockSizeDeviation the deviation to block size allowed * @return the reference to the current config. */ public BlockBasedTableConfig setBlockSizeDeviation( final int blockSizeDeviation) { blockSizeDeviation_ = blockSizeDeviation; return this; } /** * @return the hash table ratio. */ public int blockSizeDeviation() { return blockSizeDeviation_; } /** * Set block restart interval * * @param restartInterval block restart interval. * @return the reference to the current config. */ public BlockBasedTableConfig setBlockRestartInterval( final int restartInterval) { blockRestartInterval_ = restartInterval; return this; } /** * @return block restart interval */ public int blockRestartInterval() { return blockRestartInterval_; } /** * If true, place whole keys in the filter (not just prefixes). * This must generally be true for gets to be efficient. * Default: true * * @param wholeKeyFiltering if enable whole key filtering * @return the reference to the current config. */ public BlockBasedTableConfig setWholeKeyFiltering( final boolean wholeKeyFiltering) { wholeKeyFiltering_ = wholeKeyFiltering; return this; } /** * @return if whole key filtering is enabled */ public boolean wholeKeyFiltering() { return wholeKeyFiltering_; } /** * Use the specified filter policy to reduce disk reads. * * {@link org.rocksdb.Filter} should not be disposed before options instances * using this filter is disposed. If {@link Filter#dispose()} function is not * called, then filter object will be GC'd automatically. * * {@link org.rocksdb.Filter} instance can be re-used in multiple options * instances. * * @param filter {@link org.rocksdb.Filter} Filter Policy java instance. * @return the reference to the current config. */ public BlockBasedTableConfig setFilter( final Filter filter) { filter_ = filter; return this; } /** * Indicating if we'd put index/filter blocks to the block cache. If not specified, each "table reader" object will pre-load index/filter block during table initialization. * * @return if index and filter blocks should be put in block cache. */ public boolean cacheIndexAndFilterBlocks() { return cacheIndexAndFilterBlocks_; } /** * Indicating if we'd put index/filter blocks to the block cache. If not specified, each "table reader" object will pre-load index/filter block during table initialization. * * @param cacheIndexAndFilterBlocks and filter blocks should be put in block cache. * @return the reference to the current config. */ public BlockBasedTableConfig setCacheIndexAndFilterBlocks( final boolean cacheIndexAndFilterBlocks) { cacheIndexAndFilterBlocks_ = cacheIndexAndFilterBlocks; return this; } /** * Influence the behavior when kHashSearch is used. if false, stores a precise prefix to block range mapping if true, does not store prefix and allows prefix hash collision (less memory consumption) * * @return if hash collisions should be allowed. */ public boolean hashIndexAllowCollision() { return hashIndexAllowCollision_; } /** * Influence the behavior when kHashSearch is used. if false, stores a precise prefix to block range mapping if true, does not store prefix and allows prefix hash collision (less memory consumption) * * @param hashIndexAllowCollision points out if hash collisions should be allowed. * @return the reference to the current config. */ public BlockBasedTableConfig setHashIndexAllowCollision( final boolean hashIndexAllowCollision) { hashIndexAllowCollision_ = hashIndexAllowCollision; return this; } /** * Size of compressed block cache. If 0, then block_cache_compressed is set * to null. * * @return size of compressed block cache. */ public long blockCacheCompressedSize() { return blockCacheCompressedSize_; } /** * Size of compressed block cache. If 0, then block_cache_compressed is set * to null. * * @param blockCacheCompressedSize of compressed block cache. * @return the reference to the current config. */ public BlockBasedTableConfig setBlockCacheCompressedSize( final long blockCacheCompressedSize) { blockCacheCompressedSize_ = blockCacheCompressedSize; return this; } /** * Controls the number of shards for the block compressed cache. * This is applied only if blockCompressedCacheSize is set to non-negative. * * @return numShardBits the number of shard bits. The resulting * number of shards would be 2 ^ numShardBits. Any negative * number means use default settings. */ public int blockCacheCompressedNumShardBits() { return blockCacheCompressedNumShardBits_; } /** * Controls the number of shards for the block compressed cache. * This is applied only if blockCompressedCacheSize is set to non-negative. * * @param blockCacheCompressedNumShardBits the number of shard bits. The resulting * number of shards would be 2 ^ numShardBits. Any negative * number means use default settings." * @return the reference to the current option. */ public BlockBasedTableConfig setBlockCacheCompressedNumShardBits( final int blockCacheCompressedNumShardBits) { blockCacheCompressedNumShardBits_ = blockCacheCompressedNumShardBits; return this; } /** * Sets the checksum type to be used with this table. * * @param checksumType {@link org.rocksdb.ChecksumType} value. * @return the reference to the current option. */ public BlockBasedTableConfig setChecksumType( final ChecksumType checksumType) { checksumType_ = checksumType; return this; } /** * * @return the currently set checksum type */ public ChecksumType checksumType() { return checksumType_; } /** * Sets the index type to used with this table. * * @param indexType {@link org.rocksdb.IndexType} value * @return the reference to the current option. */ public BlockBasedTableConfig setIndexType( final IndexType indexType) { indexType_ = indexType; return this; } /** * * @return the currently set index type */ public IndexType indexType() { return indexType_; } /** *

We currently have three versions:

* *
    *
  • 0 - This version is currently written * out by all RocksDB's versions by default. Can be read by really old * RocksDB's. Doesn't support changing checksum (default is CRC32).
    • *
  • 1 - Can be read by RocksDB's versions since 3.0. * Supports non-default checksum, like xxHash. It is written by RocksDB when * BlockBasedTableOptions::checksum is something other than kCRC32c. (version * 0 is silently upconverted)
    • *
  • 2 - Can be read by RocksDB's versions since 3.10. * Changes the way we encode compressed blocks with LZ4, BZip2 and Zlib * compression. If you don't plan to run RocksDB before version 3.10, * you should probably use this.
    • *
*

This option only affects newly written tables. When reading existing * tables, the information about version is read from the footer.

* * @param formatVersion integer representing the version to be used. * @return the reference to the current option. */ public BlockBasedTableConfig setFormatVersion( final int formatVersion) { assert(formatVersion >= 0 && formatVersion <= 2); formatVersion_ = formatVersion; return this; } /** * * @return the currently configured format version. * See also: {@link #setFormatVersion(int)}. */ public int formatVersion() { return formatVersion_; } @Override protected long newTableFactoryHandle() { long filterHandle = 0; if (filter_ != null) { filterHandle = filter_.nativeHandle_; } return newTableFactoryHandle(noBlockCache_, blockCacheSize_, blockCacheNumShardBits_, blockSize_, blockSizeDeviation_, blockRestartInterval_, wholeKeyFiltering_, filterHandle, cacheIndexAndFilterBlocks_, hashIndexAllowCollision_, blockCacheCompressedSize_, blockCacheCompressedNumShardBits_, checksumType_.getValue(), indexType_.getValue(), formatVersion_); } private native long newTableFactoryHandle( boolean noBlockCache, long blockCacheSize, int blockCacheNumShardBits, long blockSize, int blockSizeDeviation, int blockRestartInterval, boolean wholeKeyFiltering, long filterPolicyHandle, boolean cacheIndexAndFilterBlocks, boolean hashIndexAllowCollision, long blockCacheCompressedSize, int blockCacheCompressedNumShardBits, byte checkSumType, byte indexType, int formatVersion); private boolean cacheIndexAndFilterBlocks_; private IndexType indexType_; private boolean hashIndexAllowCollision_; private ChecksumType checksumType_; private boolean noBlockCache_; private long blockSize_; private long blockCacheSize_; private int blockCacheNumShardBits_; private long blockCacheCompressedSize_; private int blockCacheCompressedNumShardBits_; private int blockSizeDeviation_; private int blockRestartInterval_; private Filter filter_; private boolean wholeKeyFiltering_; private int formatVersion_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BloomFilter.java000066400000000000000000000055421267530634000240400ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Bloom filter policy that uses a bloom filter with approximately * the specified number of bits per key. * *

* Note: if you are using a custom comparator that ignores some parts * of the keys being compared, you must not use this {@code BloomFilter} * and must provide your own FilterPolicy that also ignores the * corresponding parts of the keys. For example, if the comparator * ignores trailing spaces, it would be incorrect to use a * FilterPolicy (like {@code BloomFilter}) that does not ignore * trailing spaces in keys.

*/ public class BloomFilter extends Filter { private static final int DEFAULT_BITS_PER_KEY = 10; private static final boolean DEFAULT_MODE = true; private final int bitsPerKey_; private final boolean useBlockBasedMode_; /** * BloomFilter constructor * *

* Callers must delete the result after any database that is using the * result has been closed.

*/ public BloomFilter() { this(DEFAULT_BITS_PER_KEY, DEFAULT_MODE); } /** * BloomFilter constructor * *

* bits_per_key: bits per key in bloom filter. A good value for bits_per_key * is 10, which yields a filter with ~ 1% false positive rate. *

*

* Callers must delete the result after any database that is using the * result has been closed.

* * @param bitsPerKey number of bits to use */ public BloomFilter(final int bitsPerKey) { this(bitsPerKey, DEFAULT_MODE); } /** * BloomFilter constructor * *

* bits_per_key: bits per key in bloom filter. A good value for bits_per_key * is 10, which yields a filter with ~ 1% false positive rate. *

default bits_per_key: 10

* *

use_block_based_builder: use block based filter rather than full filter. * If you want to builder full filter, it needs to be set to false. *

*

default mode: block based filter

*

* Callers must delete the result after any database that is using the * result has been closed.

* * @param bitsPerKey number of bits to use * @param useBlockBasedMode use block based mode or full filter mode */ public BloomFilter(final int bitsPerKey, final boolean useBlockBasedMode) { super(); bitsPerKey_ = bitsPerKey; useBlockBasedMode_ = useBlockBasedMode; createNewFilter(); } @Override protected final void createNewFilter() { createNewBloomFilter(bitsPerKey_, useBlockBasedMode_); } private native void createNewBloomFilter(int bitsKeyKey, boolean useBlockBasedMode); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/BuiltinComparator.java000066400000000000000000000011711267530634000252520ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Builtin RocksDB comparators * *
    *
  1. BYTEWISE_COMPARATOR - Sorts all keys in ascending bytewise * order.
    2. *
  2. REVERSE_BYTEWISE_COMPARATOR - Sorts all keys in descending bytewise * order
    3. *
*/ public enum BuiltinComparator { BYTEWISE_COMPARATOR, REVERSE_BYTEWISE_COMPARATOR } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Checkpoint.java000066400000000000000000000044141267530634000237060ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Provides Checkpoint functionality. Checkpoints * provide persistent snapshots of RocksDB databases. */ public class Checkpoint extends RocksObject { /** * Creates a Checkpoint object to be used for creating open-able * snapshots. * * @param db {@link RocksDB} instance. * @return a Checkpoint instance. * * @throws java.lang.IllegalArgumentException if {@link RocksDB} * instance is null. * @throws java.lang.IllegalStateException if {@link RocksDB} * instance is not initialized. */ public static Checkpoint create(final RocksDB db) { if (db == null) { throw new IllegalArgumentException( "RocksDB instance shall not be null."); } else if (!db.isInitialized()) { throw new IllegalStateException( "RocksDB instance must be initialized."); } Checkpoint checkpoint = new Checkpoint(db); return checkpoint; } /** *

Builds an open-able snapshot of RocksDB on the same disk, which * accepts an output directory on the same disk, and under the directory * (1) hard-linked SST files pointing to existing live SST files * (2) a copied manifest files and other files

* * @param checkpointPath path to the folder where the snapshot is going * to be stored. * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void createCheckpoint(final String checkpointPath) throws RocksDBException { createCheckpoint(nativeHandle_, checkpointPath); } @Override protected void disposeInternal() { disposeInternal(nativeHandle_); } private Checkpoint(RocksDB db) { super(); nativeHandle_ = newCheckpoint(db.nativeHandle_); db_ = db; } private RocksDB db_; private static native long newCheckpoint(long dbHandle); private native void disposeInternal(long handle); private native void createCheckpoint(long handle, String checkpointPath) throws RocksDBException; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ChecksumType.java000066400000000000000000000014611267530634000242220ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Checksum types used in conjunction with BlockBasedTable. */ public enum ChecksumType { /** * Not implemented yet. */ kNoChecksum((byte) 0), /** * CRC32 Checksum */ kCRC32c((byte) 1), /** * XX Hash */ kxxHash((byte) 2); /** * Returns the byte value of the enumerations value * * @return byte representation */ public byte getValue() { return value_; } private ChecksumType(byte value) { value_ = value; } private final byte value_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ColumnFamilyDescriptor.java000066400000000000000000000032061267530634000262530ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

Describes a column family with a * name and respective Options.

*/ public class ColumnFamilyDescriptor { /** *

Creates a new Column Family using a name and default * options,

* * @param columnFamilyName name of column family. * @since 3.10.0 */ public ColumnFamilyDescriptor(final byte[] columnFamilyName) { this(columnFamilyName, new ColumnFamilyOptions()); } /** *

Creates a new Column Family using a name and custom * options.

* * @param columnFamilyName name of column family. * @param columnFamilyOptions options to be used with * column family. * @since 3.10.0 */ public ColumnFamilyDescriptor(final byte[] columnFamilyName, final ColumnFamilyOptions columnFamilyOptions) { columnFamilyName_ = columnFamilyName; columnFamilyOptions_ = columnFamilyOptions; } /** * Retrieve name of column family. * * @return column family name. * @since 3.10.0 */ public byte[] columnFamilyName() { return columnFamilyName_; } /** * Retrieve assigned options instance. * * @return Options instance assigned to this instance. */ public ColumnFamilyOptions columnFamilyOptions() { return columnFamilyOptions_; } private final byte[] columnFamilyName_; private final ColumnFamilyOptions columnFamilyOptions_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ColumnFamilyHandle.java000066400000000000000000000030231267530634000253250ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * ColumnFamilyHandle class to hold handles to underlying rocksdb * ColumnFamily Pointers. */ public class ColumnFamilyHandle extends RocksObject { ColumnFamilyHandle(final RocksDB rocksDB, final long nativeHandle) { super(); nativeHandle_ = nativeHandle; // rocksDB must point to a valid RocksDB instance; assert(rocksDB != null); // ColumnFamilyHandle must hold a reference to the related RocksDB instance // to guarantee that while a GC cycle starts ColumnFamilyHandle instances // are freed prior to RocksDB instances. rocksDB_ = rocksDB; } /** *

Deletes underlying C++ iterator pointer.

* *

Note: the underlying handle can only be safely deleted if the RocksDB * instance related to a certain ColumnFamilyHandle is still valid and initialized. * Therefore {@code disposeInternal()} checks if the RocksDB is initialized * before freeing the native handle.

*/ @Override protected void disposeInternal() { synchronized (rocksDB_) { assert (isInitialized()); if (rocksDB_.isInitialized()) { disposeInternal(nativeHandle_); } } } private native void disposeInternal(long handle); private final RocksDB rocksDB_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ColumnFamilyOptions.java000066400000000000000000000623031267530634000255730ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.ArrayList; import java.util.List; import java.util.Properties; /** * ColumnFamilyOptions to control the behavior of a database. It will be used * during the creation of a {@link org.rocksdb.RocksDB} (i.e., RocksDB.open()). * * If {@link #dispose()} function is not called, then it will be GC'd automatically * and native resources will be released as part of the process. */ public class ColumnFamilyOptions extends RocksObject implements ColumnFamilyOptionsInterface { static { RocksDB.loadLibrary(); } /** * Construct ColumnFamilyOptions. * * This constructor will create (by allocating a block of memory) * an {@code rocksdb::DBOptions} in the c++ side. */ public ColumnFamilyOptions() { super(); newColumnFamilyOptions(); } /** *

Method to get a options instance by using pre-configured * property values. If one or many values are undefined in * the context of RocksDB the method will return a null * value.

* *

Note: Property keys can be derived from * getter methods within the options class. Example: the method * {@code writeBufferSize()} has a property key: * {@code write_buffer_size}.

* * @param properties {@link java.util.Properties} instance. * * @return {@link org.rocksdb.ColumnFamilyOptions instance} * or null. * * @throws java.lang.IllegalArgumentException if null or empty * {@link Properties} instance is passed to the method call. */ public static ColumnFamilyOptions getColumnFamilyOptionsFromProps( final Properties properties) { if (properties == null || properties.size() == 0) { throw new IllegalArgumentException( "Properties value must contain at least one value."); } ColumnFamilyOptions columnFamilyOptions = null; StringBuilder stringBuilder = new StringBuilder(); for (final String name : properties.stringPropertyNames()){ stringBuilder.append(name); stringBuilder.append("="); stringBuilder.append(properties.getProperty(name)); stringBuilder.append(";"); } long handle = getColumnFamilyOptionsFromProps( stringBuilder.toString()); if (handle != 0){ columnFamilyOptions = new ColumnFamilyOptions(handle); } return columnFamilyOptions; } @Override public ColumnFamilyOptions optimizeForPointLookup( final long blockCacheSizeMb) { optimizeForPointLookup(nativeHandle_, blockCacheSizeMb); return this; } @Override public ColumnFamilyOptions optimizeLevelStyleCompaction() { optimizeLevelStyleCompaction(nativeHandle_, DEFAULT_COMPACTION_MEMTABLE_MEMORY_BUDGET); return this; } @Override public ColumnFamilyOptions optimizeLevelStyleCompaction( final long memtableMemoryBudget) { optimizeLevelStyleCompaction(nativeHandle_, memtableMemoryBudget); return this; } @Override public ColumnFamilyOptions optimizeUniversalStyleCompaction() { optimizeUniversalStyleCompaction(nativeHandle_, DEFAULT_COMPACTION_MEMTABLE_MEMORY_BUDGET); return this; } @Override public ColumnFamilyOptions optimizeUniversalStyleCompaction( final long memtableMemoryBudget) { optimizeUniversalStyleCompaction(nativeHandle_, memtableMemoryBudget); return this; } @Override public ColumnFamilyOptions setComparator(final BuiltinComparator builtinComparator) { assert(isInitialized()); setComparatorHandle(nativeHandle_, builtinComparator.ordinal()); return this; } @Override public ColumnFamilyOptions setComparator( final AbstractComparator> comparator) { assert (isInitialized()); setComparatorHandle(nativeHandle_, comparator.nativeHandle_); comparator_ = comparator; return this; } @Override public ColumnFamilyOptions setMergeOperatorName(final String name) { assert (isInitialized()); if (name == null) { throw new IllegalArgumentException( "Merge operator name must not be null."); } setMergeOperatorName(nativeHandle_, name); return this; } @Override public ColumnFamilyOptions setMergeOperator(final MergeOperator mergeOperator) { setMergeOperator(nativeHandle_, mergeOperator.newMergeOperatorHandle()); return this; } public ColumnFamilyOptions setCompactionFilter( final AbstractCompactionFilter> compactionFilter) { setCompactionFilterHandle(nativeHandle_, compactionFilter.nativeHandle_); compactionFilter_ = compactionFilter; return this; } @Override public ColumnFamilyOptions setWriteBufferSize(final long writeBufferSize) { assert(isInitialized()); setWriteBufferSize(nativeHandle_, writeBufferSize); return this; } @Override public long writeBufferSize() { assert(isInitialized()); return writeBufferSize(nativeHandle_); } @Override public ColumnFamilyOptions setMaxWriteBufferNumber( final int maxWriteBufferNumber) { assert(isInitialized()); setMaxWriteBufferNumber(nativeHandle_, maxWriteBufferNumber); return this; } @Override public int maxWriteBufferNumber() { assert(isInitialized()); return maxWriteBufferNumber(nativeHandle_); } @Override public ColumnFamilyOptions setMinWriteBufferNumberToMerge( final int minWriteBufferNumberToMerge) { setMinWriteBufferNumberToMerge(nativeHandle_, minWriteBufferNumberToMerge); return this; } @Override public int minWriteBufferNumberToMerge() { return minWriteBufferNumberToMerge(nativeHandle_); } @Override public ColumnFamilyOptions useFixedLengthPrefixExtractor(final int n) { assert(isInitialized()); useFixedLengthPrefixExtractor(nativeHandle_, n); return this; } @Override public ColumnFamilyOptions useCappedPrefixExtractor(final int n) { assert(isInitialized()); useCappedPrefixExtractor(nativeHandle_, n); return this; } @Override public ColumnFamilyOptions setCompressionType(final CompressionType compressionType) { setCompressionType(nativeHandle_, compressionType.getValue()); return this; } @Override public CompressionType compressionType() { return CompressionType.values()[compressionType(nativeHandle_)]; } @Override public ColumnFamilyOptions setCompressionPerLevel( final List compressionLevels) { final List byteCompressionTypes = new ArrayList<>( compressionLevels.size()); for (final CompressionType compressionLevel : compressionLevels) { byteCompressionTypes.add(compressionLevel.getValue()); } setCompressionPerLevel(nativeHandle_, byteCompressionTypes); return this; } @Override public List compressionPerLevel() { final List byteCompressionTypes = compressionPerLevel(nativeHandle_); final List compressionLevels = new ArrayList<>(); for (final Byte byteCompressionType : byteCompressionTypes) { compressionLevels.add(CompressionType.getCompressionType( byteCompressionType)); } return compressionLevels; } @Override public ColumnFamilyOptions setNumLevels(final int numLevels) { setNumLevels(nativeHandle_, numLevels); return this; } @Override public int numLevels() { return numLevels(nativeHandle_); } @Override public ColumnFamilyOptions setLevelZeroFileNumCompactionTrigger( final int numFiles) { setLevelZeroFileNumCompactionTrigger( nativeHandle_, numFiles); return this; } @Override public int levelZeroFileNumCompactionTrigger() { return levelZeroFileNumCompactionTrigger(nativeHandle_); } @Override public ColumnFamilyOptions setLevelZeroSlowdownWritesTrigger( final int numFiles) { setLevelZeroSlowdownWritesTrigger(nativeHandle_, numFiles); return this; } @Override public int levelZeroSlowdownWritesTrigger() { return levelZeroSlowdownWritesTrigger(nativeHandle_); } @Override public ColumnFamilyOptions setLevelZeroStopWritesTrigger(final int numFiles) { setLevelZeroStopWritesTrigger(nativeHandle_, numFiles); return this; } @Override public int levelZeroStopWritesTrigger() { return levelZeroStopWritesTrigger(nativeHandle_); } @Override public ColumnFamilyOptions setMaxMemCompactionLevel( final int maxMemCompactionLevel) { return this; } @Override public int maxMemCompactionLevel() { return 0; } @Override public ColumnFamilyOptions setTargetFileSizeBase( final long targetFileSizeBase) { setTargetFileSizeBase(nativeHandle_, targetFileSizeBase); return this; } @Override public long targetFileSizeBase() { return targetFileSizeBase(nativeHandle_); } @Override public ColumnFamilyOptions setTargetFileSizeMultiplier( final int multiplier) { setTargetFileSizeMultiplier(nativeHandle_, multiplier); return this; } @Override public int targetFileSizeMultiplier() { return targetFileSizeMultiplier(nativeHandle_); } @Override public ColumnFamilyOptions setMaxBytesForLevelBase( final long maxBytesForLevelBase) { setMaxBytesForLevelBase(nativeHandle_, maxBytesForLevelBase); return this; } @Override public long maxBytesForLevelBase() { return maxBytesForLevelBase(nativeHandle_); } @Override public ColumnFamilyOptions setLevelCompactionDynamicLevelBytes( final boolean enableLevelCompactionDynamicLevelBytes) { setLevelCompactionDynamicLevelBytes(nativeHandle_, enableLevelCompactionDynamicLevelBytes); return this; } @Override public boolean levelCompactionDynamicLevelBytes() { return levelCompactionDynamicLevelBytes(nativeHandle_); } @Override public ColumnFamilyOptions setMaxBytesForLevelMultiplier( final int multiplier) { setMaxBytesForLevelMultiplier(nativeHandle_, multiplier); return this; } @Override public int maxBytesForLevelMultiplier() { return maxBytesForLevelMultiplier(nativeHandle_); } @Override public ColumnFamilyOptions setExpandedCompactionFactor( final int expandedCompactionFactor) { setExpandedCompactionFactor(nativeHandle_, expandedCompactionFactor); return this; } @Override public int expandedCompactionFactor() { return expandedCompactionFactor(nativeHandle_); } @Override public ColumnFamilyOptions setSourceCompactionFactor( final int sourceCompactionFactor) { setSourceCompactionFactor(nativeHandle_, sourceCompactionFactor); return this; } @Override public int sourceCompactionFactor() { return sourceCompactionFactor(nativeHandle_); } @Override public ColumnFamilyOptions setMaxGrandparentOverlapFactor( final int maxGrandparentOverlapFactor) { setMaxGrandparentOverlapFactor(nativeHandle_, maxGrandparentOverlapFactor); return this; } @Override public int maxGrandparentOverlapFactor() { return maxGrandparentOverlapFactor(nativeHandle_); } @Override public ColumnFamilyOptions setSoftRateLimit( final double softRateLimit) { setSoftRateLimit(nativeHandle_, softRateLimit); return this; } @Override public double softRateLimit() { return softRateLimit(nativeHandle_); } @Override public ColumnFamilyOptions setHardRateLimit( final double hardRateLimit) { setHardRateLimit(nativeHandle_, hardRateLimit); return this; } @Override public double hardRateLimit() { return hardRateLimit(nativeHandle_); } @Override public ColumnFamilyOptions setRateLimitDelayMaxMilliseconds( final int rateLimitDelayMaxMilliseconds) { setRateLimitDelayMaxMilliseconds( nativeHandle_, rateLimitDelayMaxMilliseconds); return this; } @Override public int rateLimitDelayMaxMilliseconds() { return rateLimitDelayMaxMilliseconds(nativeHandle_); } @Override public ColumnFamilyOptions setArenaBlockSize( final long arenaBlockSize) { setArenaBlockSize(nativeHandle_, arenaBlockSize); return this; } @Override public long arenaBlockSize() { return arenaBlockSize(nativeHandle_); } @Override public ColumnFamilyOptions setDisableAutoCompactions( final boolean disableAutoCompactions) { setDisableAutoCompactions(nativeHandle_, disableAutoCompactions); return this; } @Override public boolean disableAutoCompactions() { return disableAutoCompactions(nativeHandle_); } @Override public ColumnFamilyOptions setPurgeRedundantKvsWhileFlush( final boolean purgeRedundantKvsWhileFlush) { setPurgeRedundantKvsWhileFlush( nativeHandle_, purgeRedundantKvsWhileFlush); return this; } @Override public boolean purgeRedundantKvsWhileFlush() { return purgeRedundantKvsWhileFlush(nativeHandle_); } @Override public ColumnFamilyOptions setCompactionStyle( final CompactionStyle compactionStyle) { setCompactionStyle(nativeHandle_, compactionStyle.getValue()); return this; } @Override public CompactionStyle compactionStyle() { return CompactionStyle.values()[compactionStyle(nativeHandle_)]; } @Override public ColumnFamilyOptions setMaxTableFilesSizeFIFO( final long maxTableFilesSize) { assert(maxTableFilesSize > 0); // unsigned native type assert(isInitialized()); setMaxTableFilesSizeFIFO(nativeHandle_, maxTableFilesSize); return this; } @Override public long maxTableFilesSizeFIFO() { return maxTableFilesSizeFIFO(nativeHandle_); } @Override public ColumnFamilyOptions setVerifyChecksumsInCompaction( final boolean verifyChecksumsInCompaction) { setVerifyChecksumsInCompaction( nativeHandle_, verifyChecksumsInCompaction); return this; } @Override public boolean verifyChecksumsInCompaction() { return verifyChecksumsInCompaction(nativeHandle_); } @Override public ColumnFamilyOptions setFilterDeletes( final boolean filterDeletes) { setFilterDeletes(nativeHandle_, filterDeletes); return this; } @Override public boolean filterDeletes() { return filterDeletes(nativeHandle_); } @Override public ColumnFamilyOptions setMaxSequentialSkipInIterations( final long maxSequentialSkipInIterations) { setMaxSequentialSkipInIterations(nativeHandle_, maxSequentialSkipInIterations); return this; } @Override public long maxSequentialSkipInIterations() { return maxSequentialSkipInIterations(nativeHandle_); } @Override public ColumnFamilyOptions setMemTableConfig( final MemTableConfig config) { memTableConfig_ = config; setMemTableFactory(nativeHandle_, config.newMemTableFactoryHandle()); return this; } @Override public String memTableFactoryName() { assert(isInitialized()); return memTableFactoryName(nativeHandle_); } @Override public ColumnFamilyOptions setTableFormatConfig( final TableFormatConfig config) { tableFormatConfig_ = config; setTableFactory(nativeHandle_, config.newTableFactoryHandle()); return this; } @Override public String tableFactoryName() { assert(isInitialized()); return tableFactoryName(nativeHandle_); } @Override public ColumnFamilyOptions setInplaceUpdateSupport( final boolean inplaceUpdateSupport) { setInplaceUpdateSupport(nativeHandle_, inplaceUpdateSupport); return this; } @Override public boolean inplaceUpdateSupport() { return inplaceUpdateSupport(nativeHandle_); } @Override public ColumnFamilyOptions setInplaceUpdateNumLocks( final long inplaceUpdateNumLocks) { setInplaceUpdateNumLocks(nativeHandle_, inplaceUpdateNumLocks); return this; } @Override public long inplaceUpdateNumLocks() { return inplaceUpdateNumLocks(nativeHandle_); } @Override public ColumnFamilyOptions setMemtablePrefixBloomBits( final int memtablePrefixBloomBits) { setMemtablePrefixBloomBits(nativeHandle_, memtablePrefixBloomBits); return this; } @Override public int memtablePrefixBloomBits() { return memtablePrefixBloomBits(nativeHandle_); } @Override public ColumnFamilyOptions setMemtablePrefixBloomProbes( final int memtablePrefixBloomProbes) { setMemtablePrefixBloomProbes(nativeHandle_, memtablePrefixBloomProbes); return this; } @Override public int memtablePrefixBloomProbes() { return memtablePrefixBloomProbes(nativeHandle_); } @Override public ColumnFamilyOptions setBloomLocality(int bloomLocality) { setBloomLocality(nativeHandle_, bloomLocality); return this; } @Override public int bloomLocality() { return bloomLocality(nativeHandle_); } @Override public ColumnFamilyOptions setMaxSuccessiveMerges( final long maxSuccessiveMerges) { setMaxSuccessiveMerges(nativeHandle_, maxSuccessiveMerges); return this; } @Override public long maxSuccessiveMerges() { return maxSuccessiveMerges(nativeHandle_); } @Override public ColumnFamilyOptions setMinPartialMergeOperands( final int minPartialMergeOperands) { setMinPartialMergeOperands(nativeHandle_, minPartialMergeOperands); return this; } @Override public int minPartialMergeOperands() { return minPartialMergeOperands(nativeHandle_); } @Override public ColumnFamilyOptions setOptimizeFiltersForHits( final boolean optimizeFiltersForHits) { setOptimizeFiltersForHits(nativeHandle_, optimizeFiltersForHits); return this; } @Override public boolean optimizeFiltersForHits() { return optimizeFiltersForHits(nativeHandle_); } /** * Release the memory allocated for the current instance * in the c++ side. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } /** *

Private constructor to be used by * {@link #getColumnFamilyOptionsFromProps(java.util.Properties)}

* * @param handle native handle to ColumnFamilyOptions instance. */ private ColumnFamilyOptions(final long handle) { super(); nativeHandle_ = handle; } private static native long getColumnFamilyOptionsFromProps( String optString); private native void newColumnFamilyOptions(); private native void disposeInternal(long handle); private native void optimizeForPointLookup(long handle, long blockCacheSizeMb); private native void optimizeLevelStyleCompaction(long handle, long memtableMemoryBudget); private native void optimizeUniversalStyleCompaction(long handle, long memtableMemoryBudget); private native void setComparatorHandle(long handle, int builtinComparator); private native void setComparatorHandle(long optHandle, long comparatorHandle); private native void setMergeOperatorName( long handle, String name); private native void setMergeOperator( long handle, long mergeOperatorHandle); private native void setCompactionFilterHandle(long handle, long compactionFilterHandle); private native void setWriteBufferSize(long handle, long writeBufferSize) throws IllegalArgumentException; private native long writeBufferSize(long handle); private native void setMaxWriteBufferNumber( long handle, int maxWriteBufferNumber); private native int maxWriteBufferNumber(long handle); private native void setMinWriteBufferNumberToMerge( long handle, int minWriteBufferNumberToMerge); private native int minWriteBufferNumberToMerge(long handle); private native void setCompressionType(long handle, byte compressionType); private native byte compressionType(long handle); private native void setCompressionPerLevel(long handle, List compressionLevels); private native List compressionPerLevel(long handle); private native void useFixedLengthPrefixExtractor( long handle, int prefixLength); private native void useCappedPrefixExtractor( long handle, int prefixLength); private native void setNumLevels( long handle, int numLevels); private native int numLevels(long handle); private native void setLevelZeroFileNumCompactionTrigger( long handle, int numFiles); private native int levelZeroFileNumCompactionTrigger(long handle); private native void setLevelZeroSlowdownWritesTrigger( long handle, int numFiles); private native int levelZeroSlowdownWritesTrigger(long handle); private native void setLevelZeroStopWritesTrigger( long handle, int numFiles); private native int levelZeroStopWritesTrigger(long handle); private native void setTargetFileSizeBase( long handle, long targetFileSizeBase); private native long targetFileSizeBase(long handle); private native void setTargetFileSizeMultiplier( long handle, int multiplier); private native int targetFileSizeMultiplier(long handle); private native void setMaxBytesForLevelBase( long handle, long maxBytesForLevelBase); private native long maxBytesForLevelBase(long handle); private native void setLevelCompactionDynamicLevelBytes( long handle, boolean enableLevelCompactionDynamicLevelBytes); private native boolean levelCompactionDynamicLevelBytes( long handle); private native void setMaxBytesForLevelMultiplier( long handle, int multiplier); private native int maxBytesForLevelMultiplier(long handle); private native void setExpandedCompactionFactor( long handle, int expandedCompactionFactor); private native int expandedCompactionFactor(long handle); private native void setSourceCompactionFactor( long handle, int sourceCompactionFactor); private native int sourceCompactionFactor(long handle); private native void setMaxGrandparentOverlapFactor( long handle, int maxGrandparentOverlapFactor); private native int maxGrandparentOverlapFactor(long handle); private native void setSoftRateLimit( long handle, double softRateLimit); private native double softRateLimit(long handle); private native void setHardRateLimit( long handle, double hardRateLimit); private native double hardRateLimit(long handle); private native void setRateLimitDelayMaxMilliseconds( long handle, int rateLimitDelayMaxMilliseconds); private native int rateLimitDelayMaxMilliseconds(long handle); private native void setArenaBlockSize( long handle, long arenaBlockSize) throws IllegalArgumentException; private native long arenaBlockSize(long handle); private native void setDisableAutoCompactions( long handle, boolean disableAutoCompactions); private native boolean disableAutoCompactions(long handle); private native void setCompactionStyle(long handle, byte compactionStyle); private native byte compactionStyle(long handle); private native void setMaxTableFilesSizeFIFO( long handle, long max_table_files_size); private native long maxTableFilesSizeFIFO(long handle); private native void setPurgeRedundantKvsWhileFlush( long handle, boolean purgeRedundantKvsWhileFlush); private native boolean purgeRedundantKvsWhileFlush(long handle); private native void setVerifyChecksumsInCompaction( long handle, boolean verifyChecksumsInCompaction); private native boolean verifyChecksumsInCompaction(long handle); private native void setFilterDeletes( long handle, boolean filterDeletes); private native boolean filterDeletes(long handle); private native void setMaxSequentialSkipInIterations( long handle, long maxSequentialSkipInIterations); private native long maxSequentialSkipInIterations(long handle); private native void setMemTableFactory(long handle, long factoryHandle); private native String memTableFactoryName(long handle); private native void setTableFactory(long handle, long factoryHandle); private native String tableFactoryName(long handle); private native void setInplaceUpdateSupport( long handle, boolean inplaceUpdateSupport); private native boolean inplaceUpdateSupport(long handle); private native void setInplaceUpdateNumLocks( long handle, long inplaceUpdateNumLocks) throws IllegalArgumentException; private native long inplaceUpdateNumLocks(long handle); private native void setMemtablePrefixBloomBits( long handle, int memtablePrefixBloomBits); private native int memtablePrefixBloomBits(long handle); private native void setMemtablePrefixBloomProbes( long handle, int memtablePrefixBloomProbes); private native int memtablePrefixBloomProbes(long handle); private native void setBloomLocality( long handle, int bloomLocality); private native int bloomLocality(long handle); private native void setMaxSuccessiveMerges( long handle, long maxSuccessiveMerges) throws IllegalArgumentException; private native long maxSuccessiveMerges(long handle); private native void setMinPartialMergeOperands( long handle, int minPartialMergeOperands); private native int minPartialMergeOperands(long handle); private native void setOptimizeFiltersForHits(long handle, boolean optimizeFiltersForHits); private native boolean optimizeFiltersForHits(long handle); MemTableConfig memTableConfig_; TableFormatConfig tableFormatConfig_; AbstractComparator> comparator_; AbstractCompactionFilter> compactionFilter_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ColumnFamilyOptionsInterface.java000066400000000000000000001244041267530634000274150ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; public interface ColumnFamilyOptionsInterface { /** * Use this if you don't need to keep the data sorted, i.e. you'll never use * an iterator, only Put() and Get() API calls * * @param blockCacheSizeMb Block cache size in MB * @return the instance of the current Object. */ Object optimizeForPointLookup(long blockCacheSizeMb); /** *

Default values for some parameters in ColumnFamilyOptions are not * optimized for heavy workloads and big datasets, which means you might * observe write stalls under some conditions. As a starting point for tuning * RocksDB options, use the following for level style compaction.

* *

Make sure to also call IncreaseParallelism(), which will provide the * biggest performance gains.

*

Note: we might use more memory than memtable_memory_budget during high * write rate period

* * @return the instance of the current Object. */ Object optimizeLevelStyleCompaction(); /** *

Default values for some parameters in ColumnFamilyOptions are not * optimized for heavy workloads and big datasets, which means you might * observe write stalls under some conditions. As a starting point for tuning * RocksDB options, use the following for level style compaction.

* *

Make sure to also call IncreaseParallelism(), which will provide the * biggest performance gains.

*

Note: we might use more memory than memtable_memory_budget during high * write rate period

* * @param memtableMemoryBudget memory budget in bytes * @return the instance of the current Object. */ Object optimizeLevelStyleCompaction(long memtableMemoryBudget); /** *

Default values for some parameters in ColumnFamilyOptions are not * optimized for heavy workloads and big datasets, which means you might * observe write stalls under some conditions. As a starting point for tuning * RocksDB options, use the following for universal style compaction.

* *

Universal style compaction is focused on reducing Write Amplification * Factor for big data sets, but increases Space Amplification.

* *

Make sure to also call IncreaseParallelism(), which will provide the * biggest performance gains.

* *

Note: we might use more memory than memtable_memory_budget during high * write rate period

* * @return the instance of the current Object. */ Object optimizeUniversalStyleCompaction(); /** *

Default values for some parameters in ColumnFamilyOptions are not * optimized for heavy workloads and big datasets, which means you might * observe write stalls under some conditions. As a starting point for tuning * RocksDB options, use the following for universal style compaction.

* *

Universal style compaction is focused on reducing Write Amplification * Factor for big data sets, but increases Space Amplification.

* *

Make sure to also call IncreaseParallelism(), which will provide the * biggest performance gains.

* *

Note: we might use more memory than memtable_memory_budget during high * write rate period

* * @param memtableMemoryBudget memory budget in bytes * @return the instance of the current Object. */ Object optimizeUniversalStyleCompaction(long memtableMemoryBudget); /** * Set {@link BuiltinComparator} to be used with RocksDB. * * Note: Comparator can be set once upon database creation. * * Default: BytewiseComparator. * @param builtinComparator a {@link BuiltinComparator} type. * @return the instance of the current Object. */ Object setComparator(BuiltinComparator builtinComparator); /** * Use the specified comparator for key ordering. * * Comparator should not be disposed before options instances using this comparator is * disposed. If dispose() function is not called, then comparator object will be * GC'd automatically. * * Comparator instance can be re-used in multiple options instances. * * @param comparator java instance. * @return the instance of the current Object. */ Object setComparator(AbstractComparator> comparator); /** *

Set the merge operator to be used for merging two merge operands * of the same key. The merge function is invoked during * compaction and at lookup time, if multiple key/value pairs belonging * to the same key are found in the database.

* * @param name the name of the merge function, as defined by * the MergeOperators factory (see utilities/MergeOperators.h) * The merge function is specified by name and must be one of the * standard merge operators provided by RocksDB. The available * operators are "put", "uint64add", "stringappend" and "stringappendtest". * @return the instance of the current Object. */ Object setMergeOperatorName(String name); /** *

Set the merge operator to be used for merging two different key/value * pairs that share the same key. The merge function is invoked during * compaction and at lookup time, if multiple key/value pairs belonging * to the same key are found in the database.

* * @param mergeOperator {@link MergeOperator} instance. * @return the instance of the current Object. */ Object setMergeOperator(MergeOperator mergeOperator); /** * Amount of data to build up in memory (backed by an unsorted log * on disk) before converting to a sorted on-disk file. * * Larger values increase performance, especially during bulk loads. * Up to {@code max_write_buffer_number} write buffers may be held in memory * at the same time, so you may wish to adjust this parameter * to control memory usage. * * Also, a larger write buffer will result in a longer recovery time * the next time the database is opened. * * Default: 4MB * @param writeBufferSize the size of write buffer. * @return the instance of the current Object. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setWriteBufferSize(long writeBufferSize); /** * Return size of write buffer size. * * @return size of write buffer. * @see #setWriteBufferSize(long) */ long writeBufferSize(); /** * The maximum number of write buffers that are built up in memory. * The default is 2, so that when 1 write buffer is being flushed to * storage, new writes can continue to the other write buffer. * Default: 2 * * @param maxWriteBufferNumber maximum number of write buffers. * @return the instance of the current Object. */ Object setMaxWriteBufferNumber( int maxWriteBufferNumber); /** * Returns maximum number of write buffers. * * @return maximum number of write buffers. * @see #setMaxWriteBufferNumber(int) */ int maxWriteBufferNumber(); /** * The minimum number of write buffers that will be merged together * before writing to storage. If set to 1, then * all write buffers are flushed to L0 as individual files and this increases * read amplification because a get request has to check in all of these * files. Also, an in-memory merge may result in writing lesser * data to storage if there are duplicate records in each of these * individual write buffers. Default: 1 * * @param minWriteBufferNumberToMerge the minimum number of write buffers * that will be merged together. * @return the reference to the current option. */ Object setMinWriteBufferNumberToMerge( int minWriteBufferNumberToMerge); /** * The minimum number of write buffers that will be merged together * before writing to storage. If set to 1, then * all write buffers are flushed to L0 as individual files and this increases * read amplification because a get request has to check in all of these * files. Also, an in-memory merge may result in writing lesser * data to storage if there are duplicate records in each of these * individual write buffers. Default: 1 * * @return the minimum number of write buffers that will be merged together. */ int minWriteBufferNumberToMerge(); /** * This prefix-extractor uses the first n bytes of a key as its prefix. * * In some hash-based memtable representation such as HashLinkedList * and HashSkipList, prefixes are used to partition the keys into * several buckets. Prefix extractor is used to specify how to * extract the prefix given a key. * * @param n use the first n bytes of a key as its prefix. * @return the reference to the current option. */ Object useFixedLengthPrefixExtractor(int n); /** * Same as fixed length prefix extractor, except that when slice is * shorter than the fixed length, it will use the full key. * * @param n use the first n bytes of a key as its prefix. * @return the reference to the current option. */ Object useCappedPrefixExtractor(int n); /** * Compress blocks using the specified compression algorithm. This * parameter can be changed dynamically. * * Default: SNAPPY_COMPRESSION, which gives lightweight but fast compression. * * @param compressionType Compression Type. * @return the reference to the current option. */ Object setCompressionType(CompressionType compressionType); /** * Compress blocks using the specified compression algorithm. This * parameter can be changed dynamically. * * Default: SNAPPY_COMPRESSION, which gives lightweight but fast compression. * * @return Compression type. */ CompressionType compressionType(); /** *

Different levels can have different compression * policies. There are cases where most lower levels * would like to use quick compression algorithms while * the higher levels (which have more data) use * compression algorithms that have better compression * but could be slower. This array, if non-empty, should * have an entry for each level of the database; * these override the value specified in the previous * field 'compression'.

* * NOTICE *

If {@code level_compaction_dynamic_level_bytes=true}, * {@code compression_per_level[0]} still determines {@code L0}, * but other elements of the array are based on base level * (the level {@code L0} files are merged to), and may not * match the level users see from info log for metadata. *

*

If {@code L0} files are merged to {@code level - n}, * then, for {@code i>0}, {@code compression_per_level[i]} * determines compaction type for level {@code n+i-1}.

* * Example *

For example, if we have 5 levels, and we determine to * merge {@code L0} data to {@code L4} (which means {@code L1..L3} * will be empty), then the new files go to {@code L4} uses * compression type {@code compression_per_level[1]}.

* *

If now {@code L0} is merged to {@code L2}. Data goes to * {@code L2} will be compressed according to * {@code compression_per_level[1]}, {@code L3} using * {@code compression_per_level[2]}and {@code L4} using * {@code compression_per_level[3]}. Compaction for each * level can change when data grows.

* *

Default: empty

* * @param compressionLevels list of * {@link org.rocksdb.CompressionType} instances. * * @return the reference to the current option. */ Object setCompressionPerLevel( List compressionLevels); /** *

Return the currently set {@link org.rocksdb.CompressionType} * per instances.

* *

See: {@link #setCompressionPerLevel(java.util.List)}

* * @return list of {@link org.rocksdb.CompressionType} * instances. */ List compressionPerLevel(); /** * Set the number of levels for this database * If level-styled compaction is used, then this number determines * the total number of levels. * * @param numLevels the number of levels. * @return the reference to the current option. */ Object setNumLevels(int numLevels); /** * If level-styled compaction is used, then this number determines * the total number of levels. * * @return the number of levels. */ int numLevels(); /** * Number of files to trigger level-0 compaction. A value < 0 means that * level-0 compaction will not be triggered by number of files at all. * Default: 4 * * @param numFiles the number of files in level-0 to trigger compaction. * @return the reference to the current option. */ Object setLevelZeroFileNumCompactionTrigger( int numFiles); /** * The number of files in level 0 to trigger compaction from level-0 to * level-1. A value < 0 means that level-0 compaction will not be * triggered by number of files at all. * Default: 4 * * @return the number of files in level 0 to trigger compaction. */ int levelZeroFileNumCompactionTrigger(); /** * Soft limit on number of level-0 files. We start slowing down writes at this * point. A value < 0 means that no writing slow down will be triggered by * number of files in level-0. * * @param numFiles soft limit on number of level-0 files. * @return the reference to the current option. */ Object setLevelZeroSlowdownWritesTrigger( int numFiles); /** * Soft limit on the number of level-0 files. We start slowing down writes * at this point. A value < 0 means that no writing slow down will be * triggered by number of files in level-0. * * @return the soft limit on the number of level-0 files. */ int levelZeroSlowdownWritesTrigger(); /** * Maximum number of level-0 files. We stop writes at this point. * * @param numFiles the hard limit of the number of level-0 files. * @return the reference to the current option. */ Object setLevelZeroStopWritesTrigger(int numFiles); /** * Maximum number of level-0 files. We stop writes at this point. * * @return the hard limit of the number of level-0 file. */ int levelZeroStopWritesTrigger(); /** * This does nothing anymore. Deprecated. * * @param maxMemCompactionLevel Unused. * * @return the reference to the current option. */ @Deprecated Object setMaxMemCompactionLevel( int maxMemCompactionLevel); /** * This does nothing anymore. Deprecated. * * @return Always returns 0. */ @Deprecated int maxMemCompactionLevel(); /** * The target file size for compaction. * This targetFileSizeBase determines a level-1 file size. * Target file size for level L can be calculated by * targetFileSizeBase * (targetFileSizeMultiplier ^ (L-1)) * For example, if targetFileSizeBase is 2MB and * target_file_size_multiplier is 10, then each file on level-1 will * be 2MB, and each file on level 2 will be 20MB, * and each file on level-3 will be 200MB. * by default targetFileSizeBase is 2MB. * * @param targetFileSizeBase the target size of a level-0 file. * @return the reference to the current option. * * @see #setTargetFileSizeMultiplier(int) */ Object setTargetFileSizeBase(long targetFileSizeBase); /** * The target file size for compaction. * This targetFileSizeBase determines a level-1 file size. * Target file size for level L can be calculated by * targetFileSizeBase * (targetFileSizeMultiplier ^ (L-1)) * For example, if targetFileSizeBase is 2MB and * target_file_size_multiplier is 10, then each file on level-1 will * be 2MB, and each file on level 2 will be 20MB, * and each file on level-3 will be 200MB. * by default targetFileSizeBase is 2MB. * * @return the target size of a level-0 file. * * @see #targetFileSizeMultiplier() */ long targetFileSizeBase(); /** * targetFileSizeMultiplier defines the size ratio between a * level-L file and level-(L+1) file. * By default target_file_size_multiplier is 1, meaning * files in different levels have the same target. * * @param multiplier the size ratio between a level-(L+1) file * and level-L file. * @return the reference to the current option. */ Object setTargetFileSizeMultiplier(int multiplier); /** * targetFileSizeMultiplier defines the size ratio between a * level-(L+1) file and level-L file. * By default targetFileSizeMultiplier is 1, meaning * files in different levels have the same target. * * @return the size ratio between a level-(L+1) file and level-L file. */ int targetFileSizeMultiplier(); /** * The upper-bound of the total size of level-1 files in bytes. * Maximum number of bytes for level L can be calculated as * (maxBytesForLevelBase) * (maxBytesForLevelMultiplier ^ (L-1)) * For example, if maxBytesForLevelBase is 20MB, and if * max_bytes_for_level_multiplier is 10, total data size for level-1 * will be 20MB, total file size for level-2 will be 200MB, * and total file size for level-3 will be 2GB. * by default 'maxBytesForLevelBase' is 10MB. * * @param maxBytesForLevelBase maximum bytes for level base. * * @return the reference to the current option. * @see #setMaxBytesForLevelMultiplier(int) */ Object setMaxBytesForLevelBase( long maxBytesForLevelBase); /** * The upper-bound of the total size of level-1 files in bytes. * Maximum number of bytes for level L can be calculated as * (maxBytesForLevelBase) * (maxBytesForLevelMultiplier ^ (L-1)) * For example, if maxBytesForLevelBase is 20MB, and if * max_bytes_for_level_multiplier is 10, total data size for level-1 * will be 20MB, total file size for level-2 will be 200MB, * and total file size for level-3 will be 2GB. * by default 'maxBytesForLevelBase' is 10MB. * * @return the upper-bound of the total size of level-1 files * in bytes. * @see #maxBytesForLevelMultiplier() */ long maxBytesForLevelBase(); /** *

If {@code true}, RocksDB will pick target size of each level * dynamically. We will pick a base level b >= 1. L0 will be * directly merged into level b, instead of always into level 1. * Level 1 to b-1 need to be empty. We try to pick b and its target * size so that

* *
    *
  1. target size is in the range of * (max_bytes_for_level_base / max_bytes_for_level_multiplier, * max_bytes_for_level_base]
    2. *
  2. target size of the last level (level num_levels-1) equals to extra size * of the level.
    3. *
* *

At the same time max_bytes_for_level_multiplier and * max_bytes_for_level_multiplier_additional are still satisfied.

* *

With this option on, from an empty DB, we make last level the base * level, which means merging L0 data into the last level, until it exceeds * max_bytes_for_level_base. And then we make the second last level to be * base level, to start to merge L0 data to second last level, with its * target size to be {@code 1/max_bytes_for_level_multiplier} of the last * levels extra size. After the data accumulates more so that we need to * move the base level to the third last one, and so on.

* *

Example

*

For example, assume {@code max_bytes_for_level_multiplier=10}, * {@code num_levels=6}, and {@code max_bytes_for_level_base=10MB}.

* *

Target sizes of level 1 to 5 starts with:

* {@code [- - - - 10MB]} *

with base level is level. Target sizes of level 1 to 4 are not applicable * because they will not be used. * Until the size of Level 5 grows to more than 10MB, say 11MB, we make * base target to level 4 and now the targets looks like:

* {@code [- - - 1.1MB 11MB]} *

While data are accumulated, size targets are tuned based on actual data * of level 5. When level 5 has 50MB of data, the target is like:

* {@code [- - - 5MB 50MB]} *

Until level 5's actual size is more than 100MB, say 101MB. Now if we * keep level 4 to be the base level, its target size needs to be 10.1MB, * which doesn't satisfy the target size range. So now we make level 3 * the target size and the target sizes of the levels look like:

* {@code [- - 1.01MB 10.1MB 101MB]} *

In the same way, while level 5 further grows, all levels' targets grow, * like

* {@code [- - 5MB 50MB 500MB]} *

Until level 5 exceeds 1000MB and becomes 1001MB, we make level 2 the * base level and make levels' target sizes like this:

* {@code [- 1.001MB 10.01MB 100.1MB 1001MB]} *

and go on...

* *

By doing it, we give {@code max_bytes_for_level_multiplier} a priority * against {@code max_bytes_for_level_base}, for a more predictable LSM tree * shape. It is useful to limit worse case space amplification.

* *

{@code max_bytes_for_level_multiplier_additional} is ignored with * this flag on.

* *

Turning this feature on or off for an existing DB can cause unexpected * LSM tree structure so it's not recommended.

* *

Caution: this option is experimental

* *

Default: false

* * @param enableLevelCompactionDynamicLevelBytes boolean value indicating * if {@code LevelCompactionDynamicLevelBytes} shall be enabled. * @return the reference to the current option. */ Object setLevelCompactionDynamicLevelBytes( boolean enableLevelCompactionDynamicLevelBytes); /** *

Return if {@code LevelCompactionDynamicLevelBytes} is enabled. *

* *

For further information see * {@link #setLevelCompactionDynamicLevelBytes(boolean)}

* * @return boolean value indicating if * {@code levelCompactionDynamicLevelBytes} is enabled. */ boolean levelCompactionDynamicLevelBytes(); /** * The ratio between the total size of level-(L+1) files and the total * size of level-L files for all L. * DEFAULT: 10 * * @param multiplier the ratio between the total size of level-(L+1) * files and the total size of level-L files for all L. * @return the reference to the current option. * @see #setMaxBytesForLevelBase(long) */ Object setMaxBytesForLevelMultiplier(int multiplier); /** * The ratio between the total size of level-(L+1) files and the total * size of level-L files for all L. * DEFAULT: 10 * * @return the ratio between the total size of level-(L+1) files and * the total size of level-L files for all L. * @see #maxBytesForLevelBase() */ int maxBytesForLevelMultiplier(); /** * Maximum number of bytes in all compacted files. We avoid expanding * the lower level file set of a compaction if it would make the * total compaction cover more than * (expanded_compaction_factor * targetFileSizeLevel()) many bytes. * * @param expandedCompactionFactor the maximum number of bytes in all * compacted files. * @return the reference to the current option. * @see #setSourceCompactionFactor(int) */ Object setExpandedCompactionFactor(int expandedCompactionFactor); /** * Maximum number of bytes in all compacted files. We avoid expanding * the lower level file set of a compaction if it would make the * total compaction cover more than * (expanded_compaction_factor * targetFileSizeLevel()) many bytes. * * @return the maximum number of bytes in all compacted files. * @see #sourceCompactionFactor() */ int expandedCompactionFactor(); /** * Maximum number of bytes in all source files to be compacted in a * single compaction run. We avoid picking too many files in the * source level so that we do not exceed the total source bytes * for compaction to exceed * (source_compaction_factor * targetFileSizeLevel()) many bytes. * Default:1, i.e. pick maxfilesize amount of data as the source of * a compaction. * * @param sourceCompactionFactor the maximum number of bytes in all * source files to be compacted in a single compaction run. * @return the reference to the current option. * @see #setExpandedCompactionFactor(int) */ Object setSourceCompactionFactor(int sourceCompactionFactor); /** * Maximum number of bytes in all source files to be compacted in a * single compaction run. We avoid picking too many files in the * source level so that we do not exceed the total source bytes * for compaction to exceed * (source_compaction_factor * targetFileSizeLevel()) many bytes. * Default:1, i.e. pick maxfilesize amount of data as the source of * a compaction. * * @return the maximum number of bytes in all source files to be compactedo. * @see #expandedCompactionFactor() */ int sourceCompactionFactor(); /** * Control maximum bytes of overlaps in grandparent (i.e., level+2) before we * stop building a single file in a level->level+1 compaction. * * @param maxGrandparentOverlapFactor maximum bytes of overlaps in * "grandparent" level. * @return the reference to the current option. */ Object setMaxGrandparentOverlapFactor( int maxGrandparentOverlapFactor); /** * Control maximum bytes of overlaps in grandparent (i.e., level+2) before we * stop building a single file in a level->level+1 compaction. * * @return maximum bytes of overlaps in "grandparent" level. */ int maxGrandparentOverlapFactor(); /** * Puts are delayed 0-1 ms when any level has a compaction score that exceeds * soft_rate_limit. This is ignored when == 0.0. * CONSTRAINT: soft_rate_limit ≤ hard_rate_limit. If this constraint does not * hold, RocksDB will set soft_rate_limit = hard_rate_limit * Default: 0 (disabled) * * @param softRateLimit the soft-rate-limit of a compaction score * for put delay. * @return the reference to the current option. */ Object setSoftRateLimit(double softRateLimit); /** * Puts are delayed 0-1 ms when any level has a compaction score that exceeds * soft_rate_limit. This is ignored when == 0.0. * CONSTRAINT: soft_rate_limit ≤ hard_rate_limit. If this constraint does not * hold, RocksDB will set soft_rate_limit = hard_rate_limit * Default: 0 (disabled) * * @return soft-rate-limit for put delay. */ double softRateLimit(); /** * Puts are delayed 1ms at a time when any level has a compaction score that * exceeds hard_rate_limit. This is ignored when ≤ 1.0. * Default: 0 (disabled) * * @param hardRateLimit the hard-rate-limit of a compaction score for put * delay. * @return the reference to the current option. */ Object setHardRateLimit(double hardRateLimit); /** * Puts are delayed 1ms at a time when any level has a compaction score that * exceeds hard_rate_limit. This is ignored when ≤ 1.0. * Default: 0 (disabled) * * @return the hard-rate-limit of a compaction score for put delay. */ double hardRateLimit(); /** * The maximum time interval a put will be stalled when hard_rate_limit * is enforced. If 0, then there is no limit. * Default: 1000 * * @param rateLimitDelayMaxMilliseconds the maximum time interval a put * will be stalled. * @return the reference to the current option. */ Object setRateLimitDelayMaxMilliseconds( int rateLimitDelayMaxMilliseconds); /** * The maximum time interval a put will be stalled when hard_rate_limit * is enforced. If 0, then there is no limit. * Default: 1000 * * @return the maximum time interval a put will be stalled when * hard_rate_limit is enforced. */ int rateLimitDelayMaxMilliseconds(); /** * The size of one block in arena memory allocation. * If ≤ 0, a proper value is automatically calculated (usually 1/10 of * writer_buffer_size). * * There are two additional restriction of the The specified size: * (1) size should be in the range of [4096, 2 << 30] and * (2) be the multiple of the CPU word (which helps with the memory * alignment). * * We'll automatically check and adjust the size number to make sure it * conforms to the restrictions. * Default: 0 * * @param arenaBlockSize the size of an arena block * @return the reference to the current option. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setArenaBlockSize(long arenaBlockSize); /** * The size of one block in arena memory allocation. * If ≤ 0, a proper value is automatically calculated (usually 1/10 of * writer_buffer_size). * * There are two additional restriction of the The specified size: * (1) size should be in the range of [4096, 2 << 30] and * (2) be the multiple of the CPU word (which helps with the memory * alignment). * * We'll automatically check and adjust the size number to make sure it * conforms to the restrictions. * Default: 0 * * @return the size of an arena block */ long arenaBlockSize(); /** * Disable automatic compactions. Manual compactions can still * be issued on this column family * * @param disableAutoCompactions true if auto-compactions are disabled. * @return the reference to the current option. */ Object setDisableAutoCompactions(boolean disableAutoCompactions); /** * Disable automatic compactions. Manual compactions can still * be issued on this column family * * @return true if auto-compactions are disabled. */ boolean disableAutoCompactions(); /** * Purge duplicate/deleted keys when a memtable is flushed to storage. * Default: true * * @param purgeRedundantKvsWhileFlush true if purging keys is disabled. * @return the reference to the current option. */ Object setPurgeRedundantKvsWhileFlush( boolean purgeRedundantKvsWhileFlush); /** * Purge duplicate/deleted keys when a memtable is flushed to storage. * Default: true * * @return true if purging keys is disabled. */ boolean purgeRedundantKvsWhileFlush(); /** * Set compaction style for DB. * * Default: LEVEL. * * @param compactionStyle Compaction style. * @return the reference to the current option. */ Object setCompactionStyle(CompactionStyle compactionStyle); /** * Compaction style for DB. * * @return Compaction style. */ CompactionStyle compactionStyle(); /** * FIFO compaction option. * The oldest table file will be deleted * once the sum of table files reaches this size. * The default value is 1GB (1 * 1024 * 1024 * 1024). * * @param maxTableFilesSize the size limit of the total sum of table files. * @return the instance of the current Object. */ Object setMaxTableFilesSizeFIFO(long maxTableFilesSize); /** * FIFO compaction option. * The oldest table file will be deleted * once the sum of table files reaches this size. * The default value is 1GB (1 * 1024 * 1024 * 1024). * * @return the size limit of the total sum of table files. */ long maxTableFilesSizeFIFO(); /** * If true, compaction will verify checksum on every read that happens * as part of compaction * Default: true * * @param verifyChecksumsInCompaction true if compaction verifies * checksum on every read. * @return the reference to the current option. */ Object setVerifyChecksumsInCompaction( boolean verifyChecksumsInCompaction); /** * If true, compaction will verify checksum on every read that happens * as part of compaction * Default: true * * @return true if compaction verifies checksum on every read. */ boolean verifyChecksumsInCompaction(); /** * Use KeyMayExist API to filter deletes when this is true. * If KeyMayExist returns false, i.e. the key definitely does not exist, then * the delete is a noop. KeyMayExist only incurs in-memory look up. * This optimization avoids writing the delete to storage when appropriate. * Default: false * * @param filterDeletes true if filter-deletes behavior is on. * @return the reference to the current option. */ Object setFilterDeletes(boolean filterDeletes); /** * Use KeyMayExist API to filter deletes when this is true. * If KeyMayExist returns false, i.e. the key definitely does not exist, then * the delete is a noop. KeyMayExist only incurs in-memory look up. * This optimization avoids writing the delete to storage when appropriate. * Default: false * * @return true if filter-deletes behavior is on. */ boolean filterDeletes(); /** * An iteration->Next() sequentially skips over keys with the same * user-key unless this option is set. This number specifies the number * of keys (with the same userkey) that will be sequentially * skipped before a reseek is issued. * Default: 8 * * @param maxSequentialSkipInIterations the number of keys could * be skipped in a iteration. * @return the reference to the current option. */ Object setMaxSequentialSkipInIterations(long maxSequentialSkipInIterations); /** * An iteration->Next() sequentially skips over keys with the same * user-key unless this option is set. This number specifies the number * of keys (with the same userkey) that will be sequentially * skipped before a reseek is issued. * Default: 8 * * @return the number of keys could be skipped in a iteration. */ long maxSequentialSkipInIterations(); /** * Set the config for mem-table. * * @param config the mem-table config. * @return the instance of the current Object. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setMemTableConfig(MemTableConfig config); /** * Returns the name of the current mem table representation. * Memtable format can be set using setTableFormatConfig. * * @return the name of the currently-used memtable factory. * @see #setTableFormatConfig(org.rocksdb.TableFormatConfig) */ String memTableFactoryName(); /** * Set the config for table format. * * @param config the table format config. * @return the reference of the current Options. */ Object setTableFormatConfig(TableFormatConfig config); /** * @return the name of the currently used table factory. */ String tableFactoryName(); /** * Allows thread-safe inplace updates. * If inplace_callback function is not set, * Put(key, new_value) will update inplace the existing_value iff * * key exists in current memtable * * new sizeof(new_value) ≤ sizeof(existing_value) * * existing_value for that key is a put i.e. kTypeValue * If inplace_callback function is set, check doc for inplace_callback. * Default: false. * * @param inplaceUpdateSupport true if thread-safe inplace updates * are allowed. * @return the reference to the current option. */ Object setInplaceUpdateSupport(boolean inplaceUpdateSupport); /** * Allows thread-safe inplace updates. * If inplace_callback function is not set, * Put(key, new_value) will update inplace the existing_value iff * * key exists in current memtable * * new sizeof(new_value) ≤ sizeof(existing_value) * * existing_value for that key is a put i.e. kTypeValue * If inplace_callback function is set, check doc for inplace_callback. * Default: false. * * @return true if thread-safe inplace updates are allowed. */ boolean inplaceUpdateSupport(); /** * Number of locks used for inplace update * Default: 10000, if inplace_update_support = true, else 0. * * @param inplaceUpdateNumLocks the number of locks used for * inplace updates. * @return the reference to the current option. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setInplaceUpdateNumLocks(long inplaceUpdateNumLocks); /** * Number of locks used for inplace update * Default: 10000, if inplace_update_support = true, else 0. * * @return the number of locks used for inplace update. */ long inplaceUpdateNumLocks(); /** * Sets the number of bits used in the prefix bloom filter. * * This value will be used only when a prefix-extractor is specified. * * @param memtablePrefixBloomBits the number of bits used in the * prefix bloom filter. * @return the reference to the current option. */ Object setMemtablePrefixBloomBits(int memtablePrefixBloomBits); /** * Returns the number of bits used in the prefix bloom filter. * * This value will be used only when a prefix-extractor is specified. * * @return the number of bloom-bits. * @see #useFixedLengthPrefixExtractor(int) */ int memtablePrefixBloomBits(); /** * The number of hash probes per key used in the mem-table. * * @param memtablePrefixBloomProbes the number of hash probes per key. * @return the reference to the current option. */ Object setMemtablePrefixBloomProbes(int memtablePrefixBloomProbes); /** * The number of hash probes per key used in the mem-table. * * @return the number of hash probes per key. */ int memtablePrefixBloomProbes(); /** * Control locality of bloom filter probes to improve cache miss rate. * This option only applies to memtable prefix bloom and plaintable * prefix bloom. It essentially limits the max number of cache lines each * bloom filter check can touch. * This optimization is turned off when set to 0. The number should never * be greater than number of probes. This option can boost performance * for in-memory workload but should use with care since it can cause * higher false positive rate. * Default: 0 * * @param bloomLocality the level of locality of bloom-filter probes. * @return the reference to the current option. */ Object setBloomLocality(int bloomLocality); /** * Control locality of bloom filter probes to improve cache miss rate. * This option only applies to memtable prefix bloom and plaintable * prefix bloom. It essentially limits the max number of cache lines each * bloom filter check can touch. * This optimization is turned off when set to 0. The number should never * be greater than number of probes. This option can boost performance * for in-memory workload but should use with care since it can cause * higher false positive rate. * Default: 0 * * @return the level of locality of bloom-filter probes. * @see #setMemtablePrefixBloomProbes(int) */ int bloomLocality(); /** * Maximum number of successive merge operations on a key in the memtable. * * When a merge operation is added to the memtable and the maximum number of * successive merges is reached, the value of the key will be calculated and * inserted into the memtable instead of the merge operation. This will * ensure that there are never more than max_successive_merges merge * operations in the memtable. * * Default: 0 (disabled) * * @param maxSuccessiveMerges the maximum number of successive merges. * @return the reference to the current option. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setMaxSuccessiveMerges(long maxSuccessiveMerges); /** * Maximum number of successive merge operations on a key in the memtable. * * When a merge operation is added to the memtable and the maximum number of * successive merges is reached, the value of the key will be calculated and * inserted into the memtable instead of the merge operation. This will * ensure that there are never more than max_successive_merges merge * operations in the memtable. * * Default: 0 (disabled) * * @return the maximum number of successive merges. */ long maxSuccessiveMerges(); /** * The number of partial merge operands to accumulate before partial * merge will be performed. Partial merge will not be called * if the list of values to merge is less than min_partial_merge_operands. * * If min_partial_merge_operands < 2, then it will be treated as 2. * * Default: 2 * * @param minPartialMergeOperands min partial merge operands * @return the reference to the current option. */ Object setMinPartialMergeOperands(int minPartialMergeOperands); /** * The number of partial merge operands to accumulate before partial * merge will be performed. Partial merge will not be called * if the list of values to merge is less than min_partial_merge_operands. * * If min_partial_merge_operands < 2, then it will be treated as 2. * * Default: 2 * * @return min partial merge operands */ int minPartialMergeOperands(); /** *

This flag specifies that the implementation should optimize the filters * mainly for cases where keys are found rather than also optimize for keys * missed. This would be used in cases where the application knows that * there are very few misses or the performance in the case of misses is not * important.

* *

For now, this flag allows us to not store filters for the last level i.e * the largest level which contains data of the LSM store. For keys which * are hits, the filters in this level are not useful because we will search * for the data anyway.

* *

NOTE: the filters in other levels are still useful * even for key hit because they tell us whether to look in that level or go * to the higher level.

* *

Default: false

* * @param optimizeFiltersForHits boolean value indicating if this flag is set. * @return the reference to the current option. */ Object setOptimizeFiltersForHits(boolean optimizeFiltersForHits); /** *

Returns the current state of the {@code optimize_filters_for_hits} * setting.

* * @return boolean value indicating if the flag * {@code optimize_filters_for_hits} was set. */ boolean optimizeFiltersForHits(); /** * Default memtable memory budget used with the following methods: * *
    *
  1. {@link #optimizeLevelStyleCompaction()}
    2. *
  2. {@link #optimizeUniversalStyleCompaction()}
    3. *
*/ long DEFAULT_COMPACTION_MEMTABLE_MEMORY_BUDGET = 512 * 1024 * 1024; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/CompactionStyle.java000066400000000000000000000031311267530634000247270ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Enum CompactionStyle * * RocksDB supports different styles of compaction. Available * compaction styles can be chosen using this enumeration. * *
    *
  1. LEVEL - Level based Compaction style
    2. *
  2. UNIVERSAL - Universal Compaction Style is a * compaction style, targeting the use cases requiring lower write * amplification, trading off read amplification and space * amplification.
    3. *
  3. FIFO - FIFO compaction style is the simplest * compaction strategy. It is suited for keeping event log data with * very low overhead (query log for example). It periodically deletes * the old data, so it's basically a TTL compaction style.
    4. *
* * @see * Universal Compaction * @see * FIFO Compaction */ public enum CompactionStyle { LEVEL((byte) 0), UNIVERSAL((byte) 1), FIFO((byte) 2); private final byte value_; private CompactionStyle(byte value) { value_ = value; } /** * Returns the byte value of the enumerations value * * @return byte representation */ public byte getValue() { return value_; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Comparator.java000066400000000000000000000015501267530634000237240ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Base class for comparators which will receive * byte[] based access via org.rocksdb.Slice in their * compare method implementation. * * byte[] based slices perform better when small keys * are involved. When using larger keys consider * using @see org.rocksdb.DirectComparator */ public abstract class Comparator extends AbstractComparator { public Comparator(final ComparatorOptions copt) { super(); createNewComparator0(copt.nativeHandle_); } private native void createNewComparator0(final long comparatorOptionsHandle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ComparatorOptions.java000066400000000000000000000034001267530634000252740ustar00rootroot00000000000000package org.rocksdb; /** * This class controls the behaviour * of Java implementations of * AbstractComparator * * Note that dispose() must be called before a ComparatorOptions * instance becomes out-of-scope to release the allocated memory in C++. */ public class ComparatorOptions extends RocksObject { public ComparatorOptions() { super(); newComparatorOptions(); } /** * Use adaptive mutex, which spins in the user space before resorting * to kernel. This could reduce context switch when the mutex is not * heavily contended. However, if the mutex is hot, we could end up * wasting spin time. * Default: false * * @return true if adaptive mutex is used. */ public boolean useAdaptiveMutex() { assert(isInitialized()); return useAdaptiveMutex(nativeHandle_); } /** * Use adaptive mutex, which spins in the user space before resorting * to kernel. This could reduce context switch when the mutex is not * heavily contended. However, if the mutex is hot, we could end up * wasting spin time. * Default: false * * @param useAdaptiveMutex true if adaptive mutex is used. * @return the reference to the current comparator options. */ public ComparatorOptions setUseAdaptiveMutex(final boolean useAdaptiveMutex) { assert (isInitialized()); setUseAdaptiveMutex(nativeHandle_, useAdaptiveMutex); return this; } @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } private native void newComparatorOptions(); private native boolean useAdaptiveMutex(final long handle); private native void setUseAdaptiveMutex(final long handle, final boolean useAdaptiveMutex); private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/CompressionType.java000066400000000000000000000053601267530634000247630ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Enum CompressionType * *

DB contents are stored in a set of blocks, each of which holds a * sequence of key,value pairs. Each block may be compressed before * being stored in a file. The following enum describes which * compression method (if any) is used to compress a block.

*/ public enum CompressionType { NO_COMPRESSION((byte) 0, null), SNAPPY_COMPRESSION((byte) 1, "snappy"), ZLIB_COMPRESSION((byte) 2, "z"), BZLIB2_COMPRESSION((byte) 3, "bzip2"), LZ4_COMPRESSION((byte) 4, "lz4"), LZ4HC_COMPRESSION((byte) 5, "lz4hc"); /** *

Get the CompressionType enumeration value by * passing the library name to this method.

* *

If library cannot be found the enumeration * value {@code NO_COMPRESSION} will be returned.

* * @param libraryName compression library name. * * @return CompressionType instance. */ public static CompressionType getCompressionType(String libraryName) { if (libraryName != null) { for (CompressionType compressionType : CompressionType.values()) { if (compressionType.getLibraryName() != null && compressionType.getLibraryName().equals(libraryName)) { return compressionType; } } } return CompressionType.NO_COMPRESSION; } /** *

Get the CompressionType enumeration value by * passing the byte identifier to this method.

* *

If library cannot be found the enumeration * value {@code NO_COMPRESSION} will be returned.

* * @param byteIdentifier of CompressionType. * * @return CompressionType instance. */ public static CompressionType getCompressionType(byte byteIdentifier) { for (CompressionType compressionType : CompressionType.values()) { if (compressionType.getValue() == byteIdentifier) { return compressionType; } } return CompressionType.NO_COMPRESSION; } /** *

Returns the byte value of the enumerations value.

* * @return byte representation */ public byte getValue() { return value_; } /** *

Returns the library name of the compression type * identified by the enumeration value.

* * @return library name */ public String getLibraryName() { return libraryName_; } private CompressionType(byte value, final String libraryName) { value_ = value; libraryName_ = libraryName; } private final byte value_; private final String libraryName_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/DBOptions.java000066400000000000000000000437511267530634000234670ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.Properties; /** * DBOptions to control the behavior of a database. It will be used * during the creation of a {@link org.rocksdb.RocksDB} (i.e., RocksDB.open()). * * If {@link #dispose()} function is not called, then it will be GC'd automatically * and native resources will be released as part of the process. */ public class DBOptions extends RocksObject implements DBOptionsInterface { static { RocksDB.loadLibrary(); } /** * Construct DBOptions. * * This constructor will create (by allocating a block of memory) * an {@code rocksdb::DBOptions} in the c++ side. */ public DBOptions() { super(); numShardBits_ = DEFAULT_NUM_SHARD_BITS; newDBOptions(); } /** *

Method to get a options instance by using pre-configured * property values. If one or many values are undefined in * the context of RocksDB the method will return a null * value.

* *

Note: Property keys can be derived from * getter methods within the options class. Example: the method * {@code allowMmapReads()} has a property key: * {@code allow_mmap_reads}.

* * @param properties {@link java.util.Properties} instance. * * @return {@link org.rocksdb.DBOptions instance} * or null. * * @throws java.lang.IllegalArgumentException if null or empty * {@link java.util.Properties} instance is passed to the method call. */ public static DBOptions getDBOptionsFromProps( final Properties properties) { if (properties == null || properties.size() == 0) { throw new IllegalArgumentException( "Properties value must contain at least one value."); } DBOptions dbOptions = null; StringBuilder stringBuilder = new StringBuilder(); for (final String name : properties.stringPropertyNames()){ stringBuilder.append(name); stringBuilder.append("="); stringBuilder.append(properties.getProperty(name)); stringBuilder.append(";"); } long handle = getDBOptionsFromProps( stringBuilder.toString()); if (handle != 0){ dbOptions = new DBOptions(handle); } return dbOptions; } @Override public DBOptions setIncreaseParallelism( final int totalThreads) { assert (isInitialized()); setIncreaseParallelism(nativeHandle_, totalThreads); return this; } @Override public DBOptions setCreateIfMissing(final boolean flag) { assert(isInitialized()); setCreateIfMissing(nativeHandle_, flag); return this; } @Override public boolean createIfMissing() { assert(isInitialized()); return createIfMissing(nativeHandle_); } @Override public DBOptions setCreateMissingColumnFamilies( final boolean flag) { assert(isInitialized()); setCreateMissingColumnFamilies(nativeHandle_, flag); return this; } @Override public boolean createMissingColumnFamilies() { assert(isInitialized()); return createMissingColumnFamilies(nativeHandle_); } @Override public DBOptions setErrorIfExists( final boolean errorIfExists) { assert(isInitialized()); setErrorIfExists(nativeHandle_, errorIfExists); return this; } @Override public boolean errorIfExists() { assert(isInitialized()); return errorIfExists(nativeHandle_); } @Override public DBOptions setParanoidChecks( final boolean paranoidChecks) { assert(isInitialized()); setParanoidChecks(nativeHandle_, paranoidChecks); return this; } @Override public boolean paranoidChecks() { assert(isInitialized()); return paranoidChecks(nativeHandle_); } @Override public DBOptions setRateLimiterConfig( final RateLimiterConfig config) { assert(isInitialized()); rateLimiterConfig_ = config; setRateLimiter(nativeHandle_, config.newRateLimiterHandle()); return this; } @Override public DBOptions setLogger(final Logger logger) { assert(isInitialized()); setLogger(nativeHandle_, logger.nativeHandle_); return this; } @Override public DBOptions setInfoLogLevel( final InfoLogLevel infoLogLevel) { assert(isInitialized()); setInfoLogLevel(nativeHandle_, infoLogLevel.getValue()); return this; } @Override public InfoLogLevel infoLogLevel() { assert(isInitialized()); return InfoLogLevel.getInfoLogLevel( infoLogLevel(nativeHandle_)); } @Override public DBOptions setMaxOpenFiles( final int maxOpenFiles) { assert(isInitialized()); setMaxOpenFiles(nativeHandle_, maxOpenFiles); return this; } @Override public int maxOpenFiles() { assert(isInitialized()); return maxOpenFiles(nativeHandle_); } @Override public DBOptions setMaxTotalWalSize( final long maxTotalWalSize) { assert(isInitialized()); setMaxTotalWalSize(nativeHandle_, maxTotalWalSize); return this; } @Override public long maxTotalWalSize() { assert(isInitialized()); return maxTotalWalSize(nativeHandle_); } @Override public DBOptions createStatistics() { assert(isInitialized()); createStatistics(nativeHandle_); return this; } @Override public Statistics statisticsPtr() { assert(isInitialized()); long statsPtr = statisticsPtr(nativeHandle_); if(statsPtr == 0) { createStatistics(); statsPtr = statisticsPtr(nativeHandle_); } return new Statistics(statsPtr); } @Override public DBOptions setDisableDataSync( final boolean disableDataSync) { assert(isInitialized()); setDisableDataSync(nativeHandle_, disableDataSync); return this; } @Override public boolean disableDataSync() { assert(isInitialized()); return disableDataSync(nativeHandle_); } @Override public DBOptions setUseFsync( final boolean useFsync) { assert(isInitialized()); setUseFsync(nativeHandle_, useFsync); return this; } @Override public boolean useFsync() { assert(isInitialized()); return useFsync(nativeHandle_); } @Override public DBOptions setDbLogDir( final String dbLogDir) { assert(isInitialized()); setDbLogDir(nativeHandle_, dbLogDir); return this; } @Override public String dbLogDir() { assert(isInitialized()); return dbLogDir(nativeHandle_); } @Override public DBOptions setWalDir( final String walDir) { assert(isInitialized()); setWalDir(nativeHandle_, walDir); return this; } @Override public String walDir() { assert(isInitialized()); return walDir(nativeHandle_); } @Override public DBOptions setDeleteObsoleteFilesPeriodMicros( final long micros) { assert(isInitialized()); setDeleteObsoleteFilesPeriodMicros(nativeHandle_, micros); return this; } @Override public long deleteObsoleteFilesPeriodMicros() { assert(isInitialized()); return deleteObsoleteFilesPeriodMicros(nativeHandle_); } @Override public DBOptions setMaxBackgroundCompactions( final int maxBackgroundCompactions) { assert(isInitialized()); setMaxBackgroundCompactions(nativeHandle_, maxBackgroundCompactions); return this; } @Override public int maxBackgroundCompactions() { assert(isInitialized()); return maxBackgroundCompactions(nativeHandle_); } @Override public DBOptions setMaxBackgroundFlushes( final int maxBackgroundFlushes) { assert(isInitialized()); setMaxBackgroundFlushes(nativeHandle_, maxBackgroundFlushes); return this; } @Override public int maxBackgroundFlushes() { assert(isInitialized()); return maxBackgroundFlushes(nativeHandle_); } @Override public DBOptions setMaxLogFileSize( final long maxLogFileSize) { assert(isInitialized()); setMaxLogFileSize(nativeHandle_, maxLogFileSize); return this; } @Override public long maxLogFileSize() { assert(isInitialized()); return maxLogFileSize(nativeHandle_); } @Override public DBOptions setLogFileTimeToRoll( final long logFileTimeToRoll) { assert(isInitialized()); setLogFileTimeToRoll(nativeHandle_, logFileTimeToRoll); return this; } @Override public long logFileTimeToRoll() { assert(isInitialized()); return logFileTimeToRoll(nativeHandle_); } @Override public DBOptions setKeepLogFileNum( final long keepLogFileNum) { assert(isInitialized()); setKeepLogFileNum(nativeHandle_, keepLogFileNum); return this; } @Override public long keepLogFileNum() { assert(isInitialized()); return keepLogFileNum(nativeHandle_); } @Override public DBOptions setMaxManifestFileSize( final long maxManifestFileSize) { assert(isInitialized()); setMaxManifestFileSize(nativeHandle_, maxManifestFileSize); return this; } @Override public long maxManifestFileSize() { assert(isInitialized()); return maxManifestFileSize(nativeHandle_); } @Override public DBOptions setTableCacheNumshardbits( final int tableCacheNumshardbits) { assert(isInitialized()); setTableCacheNumshardbits(nativeHandle_, tableCacheNumshardbits); return this; } @Override public int tableCacheNumshardbits() { assert(isInitialized()); return tableCacheNumshardbits(nativeHandle_); } @Override public DBOptions setWalTtlSeconds( final long walTtlSeconds) { assert(isInitialized()); setWalTtlSeconds(nativeHandle_, walTtlSeconds); return this; } @Override public long walTtlSeconds() { assert(isInitialized()); return walTtlSeconds(nativeHandle_); } @Override public DBOptions setWalSizeLimitMB( final long sizeLimitMB) { assert(isInitialized()); setWalSizeLimitMB(nativeHandle_, sizeLimitMB); return this; } @Override public long walSizeLimitMB() { assert(isInitialized()); return walSizeLimitMB(nativeHandle_); } @Override public DBOptions setManifestPreallocationSize( final long size) { assert(isInitialized()); setManifestPreallocationSize(nativeHandle_, size); return this; } @Override public long manifestPreallocationSize() { assert(isInitialized()); return manifestPreallocationSize(nativeHandle_); } @Override public DBOptions setAllowOsBuffer( final boolean allowOsBuffer) { assert(isInitialized()); setAllowOsBuffer(nativeHandle_, allowOsBuffer); return this; } @Override public boolean allowOsBuffer() { assert(isInitialized()); return allowOsBuffer(nativeHandle_); } @Override public DBOptions setAllowMmapReads( final boolean allowMmapReads) { assert(isInitialized()); setAllowMmapReads(nativeHandle_, allowMmapReads); return this; } @Override public boolean allowMmapReads() { assert(isInitialized()); return allowMmapReads(nativeHandle_); } @Override public DBOptions setAllowMmapWrites( final boolean allowMmapWrites) { assert(isInitialized()); setAllowMmapWrites(nativeHandle_, allowMmapWrites); return this; } @Override public boolean allowMmapWrites() { assert(isInitialized()); return allowMmapWrites(nativeHandle_); } @Override public DBOptions setIsFdCloseOnExec( final boolean isFdCloseOnExec) { assert(isInitialized()); setIsFdCloseOnExec(nativeHandle_, isFdCloseOnExec); return this; } @Override public boolean isFdCloseOnExec() { assert(isInitialized()); return isFdCloseOnExec(nativeHandle_); } @Override public DBOptions setStatsDumpPeriodSec( final int statsDumpPeriodSec) { assert(isInitialized()); setStatsDumpPeriodSec(nativeHandle_, statsDumpPeriodSec); return this; } @Override public int statsDumpPeriodSec() { assert(isInitialized()); return statsDumpPeriodSec(nativeHandle_); } @Override public DBOptions setAdviseRandomOnOpen( final boolean adviseRandomOnOpen) { assert(isInitialized()); setAdviseRandomOnOpen(nativeHandle_, adviseRandomOnOpen); return this; } @Override public boolean adviseRandomOnOpen() { return adviseRandomOnOpen(nativeHandle_); } @Override public DBOptions setUseAdaptiveMutex( final boolean useAdaptiveMutex) { assert(isInitialized()); setUseAdaptiveMutex(nativeHandle_, useAdaptiveMutex); return this; } @Override public boolean useAdaptiveMutex() { assert(isInitialized()); return useAdaptiveMutex(nativeHandle_); } @Override public DBOptions setBytesPerSync( final long bytesPerSync) { assert(isInitialized()); setBytesPerSync(nativeHandle_, bytesPerSync); return this; } @Override public long bytesPerSync() { return bytesPerSync(nativeHandle_); } /** * Release the memory allocated for the current instance * in the c++ side. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } static final int DEFAULT_NUM_SHARD_BITS = -1; /** *

Private constructor to be used by * {@link #getDBOptionsFromProps(java.util.Properties)}

* * @param handle native handle to DBOptions instance. */ private DBOptions(final long handle) { super(); nativeHandle_ = handle; } private static native long getDBOptionsFromProps( String optString); private native void newDBOptions(); private native void disposeInternal(long handle); private native void setIncreaseParallelism(long handle, int totalThreads); private native void setCreateIfMissing(long handle, boolean flag); private native boolean createIfMissing(long handle); private native void setCreateMissingColumnFamilies( long handle, boolean flag); private native boolean createMissingColumnFamilies(long handle); private native void setErrorIfExists(long handle, boolean errorIfExists); private native boolean errorIfExists(long handle); private native void setParanoidChecks( long handle, boolean paranoidChecks); private native boolean paranoidChecks(long handle); private native void setRateLimiter(long handle, long rateLimiterHandle); private native void setLogger(long handle, long loggerHandle); private native void setInfoLogLevel(long handle, byte logLevel); private native byte infoLogLevel(long handle); private native void setMaxOpenFiles(long handle, int maxOpenFiles); private native int maxOpenFiles(long handle); private native void setMaxTotalWalSize(long handle, long maxTotalWalSize); private native long maxTotalWalSize(long handle); private native void createStatistics(long optHandle); private native long statisticsPtr(long optHandle); private native void setDisableDataSync(long handle, boolean disableDataSync); private native boolean disableDataSync(long handle); private native boolean useFsync(long handle); private native void setUseFsync(long handle, boolean useFsync); private native void setDbLogDir(long handle, String dbLogDir); private native String dbLogDir(long handle); private native void setWalDir(long handle, String walDir); private native String walDir(long handle); private native void setDeleteObsoleteFilesPeriodMicros( long handle, long micros); private native long deleteObsoleteFilesPeriodMicros(long handle); private native void setMaxBackgroundCompactions( long handle, int maxBackgroundCompactions); private native int maxBackgroundCompactions(long handle); private native void setMaxBackgroundFlushes( long handle, int maxBackgroundFlushes); private native int maxBackgroundFlushes(long handle); private native void setMaxLogFileSize(long handle, long maxLogFileSize) throws IllegalArgumentException; private native long maxLogFileSize(long handle); private native void setLogFileTimeToRoll( long handle, long logFileTimeToRoll) throws IllegalArgumentException; private native long logFileTimeToRoll(long handle); private native void setKeepLogFileNum(long handle, long keepLogFileNum) throws IllegalArgumentException; private native long keepLogFileNum(long handle); private native void setMaxManifestFileSize( long handle, long maxManifestFileSize); private native long maxManifestFileSize(long handle); private native void setTableCacheNumshardbits( long handle, int tableCacheNumshardbits); private native int tableCacheNumshardbits(long handle); private native void setWalTtlSeconds(long handle, long walTtlSeconds); private native long walTtlSeconds(long handle); private native void setWalSizeLimitMB(long handle, long sizeLimitMB); private native long walSizeLimitMB(long handle); private native void setManifestPreallocationSize( long handle, long size) throws IllegalArgumentException; private native long manifestPreallocationSize(long handle); private native void setAllowOsBuffer( long handle, boolean allowOsBuffer); private native boolean allowOsBuffer(long handle); private native void setAllowMmapReads( long handle, boolean allowMmapReads); private native boolean allowMmapReads(long handle); private native void setAllowMmapWrites( long handle, boolean allowMmapWrites); private native boolean allowMmapWrites(long handle); private native void setIsFdCloseOnExec( long handle, boolean isFdCloseOnExec); private native boolean isFdCloseOnExec(long handle); private native void setStatsDumpPeriodSec( long handle, int statsDumpPeriodSec); private native int statsDumpPeriodSec(long handle); private native void setAdviseRandomOnOpen( long handle, boolean adviseRandomOnOpen); private native boolean adviseRandomOnOpen(long handle); private native void setUseAdaptiveMutex( long handle, boolean useAdaptiveMutex); private native boolean useAdaptiveMutex(long handle); private native void setBytesPerSync( long handle, long bytesPerSync); private native long bytesPerSync(long handle); int numShardBits_; RateLimiterConfig rateLimiterConfig_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/DBOptionsInterface.java000066400000000000000000000651361267530634000253110ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public interface DBOptionsInterface { /** *

By default, RocksDB uses only one background thread for flush and * compaction. Calling this function will set it up such that total of * `total_threads` is used.

* *

You almost definitely want to call this function if your system is * bottlenecked by RocksDB.

* * @param totalThreads The total number of threads to be used by RocksDB. * A good value is the number of cores. * * @return the instance of the current Options */ Object setIncreaseParallelism(int totalThreads); /** * If this value is set to true, then the database will be created * if it is missing during {@code RocksDB.open()}. * Default: false * * @param flag a flag indicating whether to create a database the * specified database in {@link RocksDB#open(org.rocksdb.Options, String)} operation * is missing. * @return the instance of the current Options * @see RocksDB#open(org.rocksdb.Options, String) */ Object setCreateIfMissing(boolean flag); /** * Return true if the create_if_missing flag is set to true. * If true, the database will be created if it is missing. * * @return true if the createIfMissing option is set to true. * @see #setCreateIfMissing(boolean) */ boolean createIfMissing(); /** *

If true, missing column families will be automatically created

* *

Default: false

* * @param flag a flag indicating if missing column families shall be * created automatically. * @return true if missing column families shall be created automatically * on open. */ Object setCreateMissingColumnFamilies(boolean flag); /** * Return true if the create_missing_column_families flag is set * to true. If true column families be created if missing. * * @return true if the createMissingColumnFamilies is set to * true. * @see #setCreateMissingColumnFamilies(boolean) */ boolean createMissingColumnFamilies(); /** * If true, an error will be thrown during RocksDB.open() if the * database already exists. * Default: false * * @param errorIfExists if true, an exception will be thrown * during {@code RocksDB.open()} if the database already exists. * @return the reference to the current option. * @see RocksDB#open(org.rocksdb.Options, String) */ Object setErrorIfExists(boolean errorIfExists); /** * If true, an error will be thrown during RocksDB.open() if the * database already exists. * * @return if true, an error is raised when the specified database * already exists before open. */ boolean errorIfExists(); /** * If true, the implementation will do aggressive checking of the * data it is processing and will stop early if it detects any * errors. This may have unforeseen ramifications: for example, a * corruption of one DB entry may cause a large number of entries to * become unreadable or for the entire DB to become unopenable. * If any of the writes to the database fails (Put, Delete, Merge, Write), * the database will switch to read-only mode and fail all other * Write operations. * Default: true * * @param paranoidChecks a flag to indicate whether paranoid-check * is on. * @return the reference to the current option. */ Object setParanoidChecks(boolean paranoidChecks); /** * If true, the implementation will do aggressive checking of the * data it is processing and will stop early if it detects any * errors. This may have unforeseen ramifications: for example, a * corruption of one DB entry may cause a large number of entries to * become unreadable or for the entire DB to become unopenable. * If any of the writes to the database fails (Put, Delete, Merge, Write), * the database will switch to read-only mode and fail all other * Write operations. * * @return a boolean indicating whether paranoid-check is on. */ boolean paranoidChecks(); /** * Use to control write rate of flush and compaction. Flush has higher * priority than compaction. Rate limiting is disabled if nullptr. * Default: nullptr * * @param config rate limiter config. * @return the instance of the current Object. */ Object setRateLimiterConfig(RateLimiterConfig config); /** *

Any internal progress/error information generated by * the db will be written to the Logger if it is non-nullptr, * or to a file stored in the same directory as the DB * contents if info_log is nullptr.

* *

Default: nullptr

* * @param logger {@link Logger} instance. * @return the instance of the current Object. */ Object setLogger(Logger logger); /** *

Sets the RocksDB log level. Default level is INFO

* * @param infoLogLevel log level to set. * @return the instance of the current Object. */ Object setInfoLogLevel(InfoLogLevel infoLogLevel); /** *

Returns currently set log level.

* @return {@link org.rocksdb.InfoLogLevel} instance. */ InfoLogLevel infoLogLevel(); /** * Number of open files that can be used by the DB. You may need to * increase this if your database has a large working set. Value -1 means * files opened are always kept open. You can estimate number of files based * on {@code target_file_size_base} and {@code target_file_size_multiplier} * for level-based compaction. For universal-style compaction, you can usually * set it to -1. * Default: 5000 * * @param maxOpenFiles the maximum number of open files. * @return the instance of the current Object. */ Object setMaxOpenFiles(int maxOpenFiles); /** * Number of open files that can be used by the DB. You may need to * increase this if your database has a large working set. Value -1 means * files opened are always kept open. You can estimate number of files based * on {@code target_file_size_base} and {@code target_file_size_multiplier} * for level-based compaction. For universal-style compaction, you can usually * set it to -1. * * @return the maximum number of open files. */ int maxOpenFiles(); /** *

Once write-ahead logs exceed this size, we will start forcing the * flush of column families whose memtables are backed by the oldest live * WAL file (i.e. the ones that are causing all the space amplification). *

*

If set to 0 (default), we will dynamically choose the WAL size limit to * be [sum of all write_buffer_size * max_write_buffer_number] * 2

*

Default: 0

* * @param maxTotalWalSize max total wal size. * @return the instance of the current Object. */ Object setMaxTotalWalSize(long maxTotalWalSize); /** *

Returns the max total wal size. Once write-ahead logs exceed this size, * we will start forcing the flush of column families whose memtables are * backed by the oldest live WAL file (i.e. the ones that are causing all * the space amplification).

* *

If set to 0 (default), we will dynamically choose the WAL size limit * to be [sum of all write_buffer_size * max_write_buffer_number] * 2 *

* * @return max total wal size */ long maxTotalWalSize(); /** *

Creates statistics object which collects metrics about database operations. * Statistics objects should not be shared between DB instances as * it does not use any locks to prevent concurrent updates.

* * @return the instance of the current Object. * @see RocksDB#open(org.rocksdb.Options, String) */ Object createStatistics(); /** *

Returns statistics object. Calls {@link #createStatistics()} if * C++ returns {@code nullptr} for statistics.

* * @return the instance of the statistics object. * @see #createStatistics() */ Statistics statisticsPtr(); /** *

If true, then the contents of manifest and data files are * not synced to stable storage. Their contents remain in the * OS buffers till theOS decides to flush them.

* *

This option is good for bulk-loading of data.

* *

Once the bulk-loading is complete, please issue a sync to * the OS to flush all dirty buffers to stable storage.

* *

Default: false

* * @param disableDataSync a boolean flag to specify whether to * disable data sync. * @return the instance of the current Object. */ Object setDisableDataSync(boolean disableDataSync); /** * If true, then the contents of data files are not synced * to stable storage. Their contents remain in the OS buffers till the * OS decides to flush them. This option is good for bulk-loading * of data. Once the bulk-loading is complete, please issue a * sync to the OS to flush all dirty buffers to stable storage. * * @return if true, then data-sync is disabled. */ boolean disableDataSync(); /** *

If true, then every store to stable storage will issue a fsync.

*

If false, then every store to stable storage will issue a fdatasync. * This parameter should be set to true while storing data to * filesystem like ext3 that can lose files after a reboot.

*

Default: false

* * @param useFsync a boolean flag to specify whether to use fsync * @return the instance of the current Object. */ Object setUseFsync(boolean useFsync); /** *

If true, then every store to stable storage will issue a fsync.

*

If false, then every store to stable storage will issue a fdatasync. * This parameter should be set to true while storing data to * filesystem like ext3 that can lose files after a reboot.

* * @return boolean value indicating if fsync is used. */ boolean useFsync(); /** * This specifies the info LOG dir. * If it is empty, the log files will be in the same dir as data. * If it is non empty, the log files will be in the specified dir, * and the db data dir's absolute path will be used as the log file * name's prefix. * * @param dbLogDir the path to the info log directory * @return the instance of the current Object. */ Object setDbLogDir(String dbLogDir); /** * Returns the directory of info log. * * If it is empty, the log files will be in the same dir as data. * If it is non empty, the log files will be in the specified dir, * and the db data dir's absolute path will be used as the log file * name's prefix. * * @return the path to the info log directory */ String dbLogDir(); /** * This specifies the absolute dir path for write-ahead logs (WAL). * If it is empty, the log files will be in the same dir as data, * dbname is used as the data dir by default * If it is non empty, the log files will be in kept the specified dir. * When destroying the db, * all log files in wal_dir and the dir itself is deleted * * @param walDir the path to the write-ahead-log directory. * @return the instance of the current Object. */ Object setWalDir(String walDir); /** * Returns the path to the write-ahead-logs (WAL) directory. * * If it is empty, the log files will be in the same dir as data, * dbname is used as the data dir by default * If it is non empty, the log files will be in kept the specified dir. * When destroying the db, * all log files in wal_dir and the dir itself is deleted * * @return the path to the write-ahead-logs (WAL) directory. */ String walDir(); /** * The periodicity when obsolete files get deleted. The default * value is 6 hours. The files that get out of scope by compaction * process will still get automatically delete on every compaction, * regardless of this setting * * @param micros the time interval in micros * @return the instance of the current Object. */ Object setDeleteObsoleteFilesPeriodMicros(long micros); /** * The periodicity when obsolete files get deleted. The default * value is 6 hours. The files that get out of scope by compaction * process will still get automatically delete on every compaction, * regardless of this setting * * @return the time interval in micros when obsolete files will be deleted. */ long deleteObsoleteFilesPeriodMicros(); /** * Specifies the maximum number of concurrent background compaction jobs, * submitted to the default LOW priority thread pool. * If you're increasing this, also consider increasing number of threads in * LOW priority thread pool. For more information, see * Default: 1 * * @param maxBackgroundCompactions the maximum number of background * compaction jobs. * @return the instance of the current Object. * * @see RocksEnv#setBackgroundThreads(int) * @see RocksEnv#setBackgroundThreads(int, int) * @see #maxBackgroundFlushes() */ Object setMaxBackgroundCompactions(int maxBackgroundCompactions); /** * Returns the maximum number of concurrent background compaction jobs, * submitted to the default LOW priority thread pool. * When increasing this number, we may also want to consider increasing * number of threads in LOW priority thread pool. * Default: 1 * * @return the maximum number of concurrent background compaction jobs. * @see RocksEnv#setBackgroundThreads(int) * @see RocksEnv#setBackgroundThreads(int, int) */ int maxBackgroundCompactions(); /** * Specifies the maximum number of concurrent background flush jobs. * If you're increasing this, also consider increasing number of threads in * HIGH priority thread pool. For more information, see * Default: 1 * * @param maxBackgroundFlushes number of max concurrent flush jobs * @return the instance of the current Object. * * @see RocksEnv#setBackgroundThreads(int) * @see RocksEnv#setBackgroundThreads(int, int) * @see #maxBackgroundCompactions() */ Object setMaxBackgroundFlushes(int maxBackgroundFlushes); /** * Returns the maximum number of concurrent background flush jobs. * If you're increasing this, also consider increasing number of threads in * HIGH priority thread pool. For more information, see * Default: 1 * * @return the maximum number of concurrent background flush jobs. * @see RocksEnv#setBackgroundThreads(int) * @see RocksEnv#setBackgroundThreads(int, int) */ int maxBackgroundFlushes(); /** * Specifies the maximum size of a info log file. If the current log file * is larger than `max_log_file_size`, a new info log file will * be created. * If 0, all logs will be written to one log file. * * @param maxLogFileSize the maximum size of a info log file. * @return the instance of the current Object. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setMaxLogFileSize(long maxLogFileSize); /** * Returns the maximum size of a info log file. If the current log file * is larger than this size, a new info log file will be created. * If 0, all logs will be written to one log file. * * @return the maximum size of the info log file. */ long maxLogFileSize(); /** * Specifies the time interval for the info log file to roll (in seconds). * If specified with non-zero value, log file will be rolled * if it has been active longer than `log_file_time_to_roll`. * Default: 0 (disabled) * * @param logFileTimeToRoll the time interval in seconds. * @return the instance of the current Object. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setLogFileTimeToRoll(long logFileTimeToRoll); /** * Returns the time interval for the info log file to roll (in seconds). * If specified with non-zero value, log file will be rolled * if it has been active longer than `log_file_time_to_roll`. * Default: 0 (disabled) * * @return the time interval in seconds. */ long logFileTimeToRoll(); /** * Specifies the maximum number of info log files to be kept. * Default: 1000 * * @param keepLogFileNum the maximum number of info log files to be kept. * @return the instance of the current Object. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setKeepLogFileNum(long keepLogFileNum); /** * Returns the maximum number of info log files to be kept. * Default: 1000 * * @return the maximum number of info log files to be kept. */ long keepLogFileNum(); /** * Manifest file is rolled over on reaching this limit. * The older manifest file be deleted. * The default value is MAX_INT so that roll-over does not take place. * * @param maxManifestFileSize the size limit of a manifest file. * @return the instance of the current Object. */ Object setMaxManifestFileSize(long maxManifestFileSize); /** * Manifest file is rolled over on reaching this limit. * The older manifest file be deleted. * The default value is MAX_INT so that roll-over does not take place. * * @return the size limit of a manifest file. */ long maxManifestFileSize(); /** * Number of shards used for table cache. * * @param tableCacheNumshardbits the number of chards * @return the instance of the current Object. */ Object setTableCacheNumshardbits(int tableCacheNumshardbits); /** * Number of shards used for table cache. * * @return the number of shards used for table cache. */ int tableCacheNumshardbits(); /** * {@link #walTtlSeconds()} and {@link #walSizeLimitMB()} affect how archived logs * will be deleted. *
    *
  1. If both set to 0, logs will be deleted asap and will not get into * the archive.
    2. *
  2. If WAL_ttl_seconds is 0 and WAL_size_limit_MB is not 0, * WAL files will be checked every 10 min and if total size is greater * then WAL_size_limit_MB, they will be deleted starting with the * earliest until size_limit is met. All empty files will be deleted.
    3. *
  3. If WAL_ttl_seconds is not 0 and WAL_size_limit_MB is 0, then * WAL files will be checked every WAL_ttl_secondsi / 2 and those that * are older than WAL_ttl_seconds will be deleted.
    4. *
  4. If both are not 0, WAL files will be checked every 10 min and both * checks will be performed with ttl being first.
    5. *
* * @param walTtlSeconds the ttl seconds * @return the instance of the current Object. * @see #setWalSizeLimitMB(long) */ Object setWalTtlSeconds(long walTtlSeconds); /** * WalTtlSeconds() and walSizeLimitMB() affect how archived logs * will be deleted. *
    *
  1. If both set to 0, logs will be deleted asap and will not get into * the archive.
    2. *
  2. If WAL_ttl_seconds is 0 and WAL_size_limit_MB is not 0, * WAL files will be checked every 10 min and if total size is greater * then WAL_size_limit_MB, they will be deleted starting with the * earliest until size_limit is met. All empty files will be deleted.
    3. *
  3. If WAL_ttl_seconds is not 0 and WAL_size_limit_MB is 0, then * WAL files will be checked every WAL_ttl_secondsi / 2 and those that * are older than WAL_ttl_seconds will be deleted.
    4. *
  4. If both are not 0, WAL files will be checked every 10 min and both * checks will be performed with ttl being first.
    5. *
* * @return the wal-ttl seconds * @see #walSizeLimitMB() */ long walTtlSeconds(); /** * WalTtlSeconds() and walSizeLimitMB() affect how archived logs * will be deleted. *
    *
  1. If both set to 0, logs will be deleted asap and will not get into * the archive.
    2. *
  2. If WAL_ttl_seconds is 0 and WAL_size_limit_MB is not 0, * WAL files will be checked every 10 min and if total size is greater * then WAL_size_limit_MB, they will be deleted starting with the * earliest until size_limit is met. All empty files will be deleted.
    3. *
  3. If WAL_ttl_seconds is not 0 and WAL_size_limit_MB is 0, then * WAL files will be checked every WAL_ttl_secondsi / 2 and those that * are older than WAL_ttl_seconds will be deleted.
    4. *
  4. If both are not 0, WAL files will be checked every 10 min and both * checks will be performed with ttl being first.
    5. *
* * @param sizeLimitMB size limit in mega-bytes. * @return the instance of the current Object. * @see #setWalSizeLimitMB(long) */ Object setWalSizeLimitMB(long sizeLimitMB); /** * {@link #walTtlSeconds()} and {@code #walSizeLimitMB()} affect how archived logs * will be deleted. *
    *
  1. If both set to 0, logs will be deleted asap and will not get into * the archive.
    2. *
  2. If WAL_ttl_seconds is 0 and WAL_size_limit_MB is not 0, * WAL files will be checked every 10 min and if total size is greater * then WAL_size_limit_MB, they will be deleted starting with the * earliest until size_limit is met. All empty files will be deleted.
    3. *
  3. If WAL_ttl_seconds is not 0 and WAL_size_limit_MB is 0, then * WAL files will be checked every WAL_ttl_seconds i / 2 and those that * are older than WAL_ttl_seconds will be deleted.
    4. *
  4. If both are not 0, WAL files will be checked every 10 min and both * checks will be performed with ttl being first.
    5. *
* @return size limit in mega-bytes. * @see #walSizeLimitMB() */ long walSizeLimitMB(); /** * Number of bytes to preallocate (via fallocate) the manifest * files. Default is 4mb, which is reasonable to reduce random IO * as well as prevent overallocation for mounts that preallocate * large amounts of data (such as xfs's allocsize option). * * @param size the size in byte * @return the instance of the current Object. * @throws java.lang.IllegalArgumentException thrown on 32-Bit platforms * while overflowing the underlying platform specific value. */ Object setManifestPreallocationSize(long size); /** * Number of bytes to preallocate (via fallocate) the manifest * files. Default is 4mb, which is reasonable to reduce random IO * as well as prevent overallocation for mounts that preallocate * large amounts of data (such as xfs's allocsize option). * * @return size in bytes. */ long manifestPreallocationSize(); /** * Data being read from file storage may be buffered in the OS * Default: true * * @param allowOsBuffer if true, then OS buffering is allowed. * @return the instance of the current Object. */ Object setAllowOsBuffer(boolean allowOsBuffer); /** * Data being read from file storage may be buffered in the OS * Default: true * * @return if true, then OS buffering is allowed. */ boolean allowOsBuffer(); /** * Allow the OS to mmap file for reading sst tables. * Default: false * * @param allowMmapReads true if mmap reads are allowed. * @return the instance of the current Object. */ Object setAllowMmapReads(boolean allowMmapReads); /** * Allow the OS to mmap file for reading sst tables. * Default: false * * @return true if mmap reads are allowed. */ boolean allowMmapReads(); /** * Allow the OS to mmap file for writing. Default: false * * @param allowMmapWrites true if mmap writes are allowd. * @return the instance of the current Object. */ Object setAllowMmapWrites(boolean allowMmapWrites); /** * Allow the OS to mmap file for writing. Default: false * * @return true if mmap writes are allowed. */ boolean allowMmapWrites(); /** * Disable child process inherit open files. Default: true * * @param isFdCloseOnExec true if child process inheriting open * files is disabled. * @return the instance of the current Object. */ Object setIsFdCloseOnExec(boolean isFdCloseOnExec); /** * Disable child process inherit open files. Default: true * * @return true if child process inheriting open files is disabled. */ boolean isFdCloseOnExec(); /** * if not zero, dump rocksdb.stats to LOG every stats_dump_period_sec * Default: 3600 (1 hour) * * @param statsDumpPeriodSec time interval in seconds. * @return the instance of the current Object. */ Object setStatsDumpPeriodSec(int statsDumpPeriodSec); /** * If not zero, dump rocksdb.stats to LOG every stats_dump_period_sec * Default: 3600 (1 hour) * * @return time interval in seconds. */ int statsDumpPeriodSec(); /** * If set true, will hint the underlying file system that the file * access pattern is random, when a sst file is opened. * Default: true * * @param adviseRandomOnOpen true if hinting random access is on. * @return the instance of the current Object. */ Object setAdviseRandomOnOpen(boolean adviseRandomOnOpen); /** * If set true, will hint the underlying file system that the file * access pattern is random, when a sst file is opened. * Default: true * * @return true if hinting random access is on. */ boolean adviseRandomOnOpen(); /** * Use adaptive mutex, which spins in the user space before resorting * to kernel. This could reduce context switch when the mutex is not * heavily contended. However, if the mutex is hot, we could end up * wasting spin time. * Default: false * * @param useAdaptiveMutex true if adaptive mutex is used. * @return the instance of the current Object. */ Object setUseAdaptiveMutex(boolean useAdaptiveMutex); /** * Use adaptive mutex, which spins in the user space before resorting * to kernel. This could reduce context switch when the mutex is not * heavily contended. However, if the mutex is hot, we could end up * wasting spin time. * Default: false * * @return true if adaptive mutex is used. */ boolean useAdaptiveMutex(); /** * Allows OS to incrementally sync files to disk while they are being * written, asynchronously, in the background. * Issue one request for every bytes_per_sync written. 0 turns it off. * Default: 0 * * @param bytesPerSync size in bytes * @return the instance of the current Object. */ Object setBytesPerSync(long bytesPerSync); /** * Allows OS to incrementally sync files to disk while they are being * written, asynchronously, in the background. * Issue one request for every bytes_per_sync written. 0 turns it off. * Default: 0 * * @return size in bytes */ long bytesPerSync(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/DirectComparator.java000066400000000000000000000016171267530634000250630ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Base class for comparators which will receive * ByteBuffer based access via org.rocksdb.DirectSlice * in their compare method implementation. * * ByteBuffer based slices perform better when large keys * are involved. When using smaller keys consider * using @see org.rocksdb.Comparator */ public abstract class DirectComparator extends AbstractComparator { public DirectComparator(final ComparatorOptions copt) { super(); createNewDirectComparator0(copt.nativeHandle_); } private native void createNewDirectComparator0(final long comparatorOptionsHandle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/DirectSlice.java000066400000000000000000000065701267530634000240160ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.nio.ByteBuffer; /** * Base class for slices which will receive direct * ByteBuffer based access to the underlying data. * * ByteBuffer backed slices typically perform better with * larger keys and values. When using smaller keys and * values consider using @see org.rocksdb.Slice */ public class DirectSlice extends AbstractSlice { //TODO(AR) only needed by WriteBatchWithIndexTest until JDK8 public final static DirectSlice NONE = new DirectSlice(); /** * Called from JNI to construct a new Java DirectSlice * without an underlying C++ object set * at creation time. * * Note: You should be aware that * {@see org.rocksdb.RocksObject#disOwnNativeHandle()} is intentionally * called from the default DirectSlice constructor, and that it is marked as * package-private. This is so that developers cannot construct their own default * DirectSlice objects (at present). As developers cannot construct their own * DirectSlice objects through this, they are not creating underlying C++ * DirectSlice objects, and so there is nothing to free (dispose) from Java. */ DirectSlice() { super(); disOwnNativeHandle(); } /** * Constructs a slice * where the data is taken from * a String. * * @param str The string */ public DirectSlice(final String str) { super(); createNewSliceFromString(str); } /** * Constructs a slice where the data is * read from the provided * ByteBuffer up to a certain length * * @param data The buffer containing the data * @param length The length of the data to use for the slice */ public DirectSlice(final ByteBuffer data, final int length) { super(); assert(data.isDirect()); createNewDirectSlice0(data, length); } /** * Constructs a slice where the data is * read from the provided * ByteBuffer * * @param data The bugger containing the data */ public DirectSlice(final ByteBuffer data) { super(); assert(data.isDirect()); createNewDirectSlice1(data); } /** * Retrieves the byte at a specific offset * from the underlying data * * @param offset The (zero-based) offset of the byte to retrieve * * @return the requested byte */ public byte get(int offset) { assert (isInitialized()); return get0(nativeHandle_, offset); } /** * Clears the backing slice */ public void clear() { assert (isInitialized()); clear0(nativeHandle_); } /** * Drops the specified {@code n} * number of bytes from the start * of the backing slice * * @param n The number of bytes to drop */ public void removePrefix(final int n) { assert (isInitialized()); removePrefix0(nativeHandle_, n); } private native void createNewDirectSlice0(ByteBuffer data, int length); private native void createNewDirectSlice1(ByteBuffer data); @Override protected final native ByteBuffer data0(long handle); private native byte get0(long handle, int offset); private native void clear0(long handle); private native void removePrefix0(long handle, int length); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/EncodingType.java000066400000000000000000000035031267530634000242050ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * EncodingType * *

The value will determine how to encode keys * when writing to a new SST file.

* *

This value will be stored * inside the SST file which will be used when reading from * the file, which makes it possible for users to choose * different encoding type when reopening a DB. Files with * different encoding types can co-exist in the same DB and * can be read.

*/ public enum EncodingType { /** * Always write full keys without any special encoding. */ kPlain((byte) 0), /** *

Find opportunity to write the same prefix once for multiple rows. * In some cases, when a key follows a previous key with the same prefix, * instead of writing out the full key, it just writes out the size of the * shared prefix, as well as other bytes, to save some bytes.

* *

When using this option, the user is required to use the same prefix * extractor to make sure the same prefix will be extracted from the same key. * The Name() value of the prefix extractor will be stored in the file. When * reopening the file, the name of the options.prefix_extractor given will be * bitwise compared to the prefix extractors stored in the file. An error * will be returned if the two don't match.

*/ kPrefix((byte) 1); /** * Returns the byte value of the enumerations value * * @return byte representation */ public byte getValue() { return value_; } private EncodingType(byte value) { value_ = value; } private final byte value_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Env.java000066400000000000000000000053031267530634000223450ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Base class for all Env implementations in RocksDB. */ public abstract class Env extends RocksObject { public static final int FLUSH_POOL = 0; public static final int COMPACTION_POOL = 1; /** *

Returns the default environment suitable for the current operating * system.

* *

The result of {@code getDefault()} is a singleton whose ownership * belongs to rocksdb c++. As a result, the returned RocksEnv will not * have the ownership of its c++ resource, and calling its dispose() * will be no-op.

* * @return the default {@link org.rocksdb.RocksEnv} instance. */ public static Env getDefault() { return default_env_; } /** *

Sets the number of background worker threads of the flush pool * for this environment.

*

Default number: 1

* * @param num the number of threads * * @return current {@link RocksEnv} instance. */ public Env setBackgroundThreads(final int num) { return setBackgroundThreads(num, FLUSH_POOL); } /** *

Sets the number of background worker threads of the specified thread * pool for this environment.

* * @param num the number of threads * @param poolID the id to specified a thread pool. Should be either * FLUSH_POOL or COMPACTION_POOL. * *

Default number: 1

* @return current {@link RocksEnv} instance. */ public Env setBackgroundThreads(final int num, final int poolID) { setBackgroundThreads(nativeHandle_, num, poolID); return this; } /** *

Returns the length of the queue associated with the specified * thread pool.

* * @param poolID the id to specified a thread pool. Should be either * FLUSH_POOL or COMPACTION_POOL. * * @return the thread pool queue length. */ public int getThreadPoolQueueLen(final int poolID) { return getThreadPoolQueueLen(nativeHandle_, poolID); } protected Env() { super(); } static { default_env_ = new RocksEnv(getDefaultEnvInternal()); } /** *

The static default Env. The ownership of its native handle * belongs to rocksdb c++ and is not able to be released on the Java * side.

*/ static Env default_env_; private static native long getDefaultEnvInternal(); private native void setBackgroundThreads( long handle, int num, int priority); private native int getThreadPoolQueueLen(long handle, int poolID); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Filter.java000066400000000000000000000020741267530634000230440ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Filters are stored in rocksdb and are consulted automatically * by rocksdb to decide whether or not to read some * information from disk. In many cases, a filter can cut down the * number of disk seeks form a handful to a single disk seek per * DB::Get() call. */ public abstract class Filter extends RocksObject { protected abstract void createNewFilter(); /** * Deletes underlying C++ filter pointer. * * Note that this function should be called only after all * RocksDB instances referencing the filter are closed. * Otherwise an undefined behavior will occur. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/FlushOptions.java000066400000000000000000000024211267530634000242500ustar00rootroot00000000000000package org.rocksdb; /** * FlushOptions to be passed to flush operations of * {@link org.rocksdb.RocksDB}. */ public class FlushOptions extends RocksObject { /** * Construct a new instance of FlushOptions. */ public FlushOptions(){ super(); newFlushOptions(); } /** * Set if the flush operation shall block until it terminates. * * @param waitForFlush boolean value indicating if the flush * operations waits for termination of the flush process. * * @return instance of current FlushOptions. */ public FlushOptions setWaitForFlush(final boolean waitForFlush) { assert(isInitialized()); setWaitForFlush(nativeHandle_, waitForFlush); return this; } /** * Wait for flush to finished. * * @return boolean value indicating if the flush operation * waits for termination of the flush process. */ public boolean waitForFlush() { assert(isInitialized()); return waitForFlush(nativeHandle_); } @Override protected void disposeInternal() { disposeInternal(nativeHandle_); } private native void newFlushOptions(); private native void disposeInternal(long handle); private native void setWaitForFlush(long handle, boolean wait); private native boolean waitForFlush(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/GenericRateLimiterConfig.java000066400000000000000000000060061267530634000264620ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Config for rate limiter, which is used to control write rate of flush and * compaction. * * @see RateLimiterConfig */ public class GenericRateLimiterConfig extends RateLimiterConfig { private static final long DEFAULT_REFILL_PERIOD_MICROS = (100 * 1000); private static final int DEFAULT_FAIRNESS = 10; /** * GenericRateLimiterConfig constructor * * @param rateBytesPerSecond this is the only parameter you want to set * most of the time. It controls the total write rate of compaction * and flush in bytes per second. Currently, RocksDB does not enforce * rate limit for anything other than flush and compaction, e.g. write to WAL. * @param refillPeriodMicros this controls how often tokens are refilled. For example, * when rate_bytes_per_sec is set to 10MB/s and refill_period_us is set to * 100ms, then 1MB is refilled every 100ms internally. Larger value can lead to * burstier writes while smaller value introduces more CPU overhead. * The default should work for most cases. * @param fairness RateLimiter accepts high-pri requests and low-pri requests. * A low-pri request is usually blocked in favor of hi-pri request. Currently, * RocksDB assigns low-pri to request from compaction and high-pri to request * from flush. Low-pri requests can get blocked if flush requests come in * continuously. This fairness parameter grants low-pri requests permission by * fairness chance even though high-pri requests exist to avoid starvation. * You should be good by leaving it at default 10. */ public GenericRateLimiterConfig(final long rateBytesPerSecond, final long refillPeriodMicros, final int fairness) { rateBytesPerSecond_ = rateBytesPerSecond; refillPeriodMicros_ = refillPeriodMicros; fairness_ = fairness; } /** * GenericRateLimiterConfig constructor * * @param rateBytesPerSecond this is the only parameter you want to set * most of the time. It controls the total write rate of compaction * and flush in bytes per second. Currently, RocksDB does not enforce * rate limit for anything other than flush and compaction, e.g. write to WAL. */ public GenericRateLimiterConfig(final long rateBytesPerSecond) { this(rateBytesPerSecond, DEFAULT_REFILL_PERIOD_MICROS, DEFAULT_FAIRNESS); } @Override protected long newRateLimiterHandle() { return newRateLimiterHandle(rateBytesPerSecond_, refillPeriodMicros_, fairness_); } private native long newRateLimiterHandle(long rateBytesPerSecond, long refillPeriodMicros, int fairness); private final long rateBytesPerSecond_; private final long refillPeriodMicros_; private final int fairness_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/HashLinkedListMemTableConfig.java000066400000000000000000000126211267530634000272210ustar00rootroot00000000000000package org.rocksdb; /** * The config for hash linked list memtable representation * Such memtable contains a fix-sized array of buckets, where * each bucket points to a sorted singly-linked * list (or null if the bucket is empty). * * Note that since this mem-table representation relies on the * key prefix, it is required to invoke one of the usePrefixExtractor * functions to specify how to extract key prefix given a key. * If proper prefix-extractor is not set, then RocksDB will * use the default memtable representation (SkipList) instead * and post a warning in the LOG. */ public class HashLinkedListMemTableConfig extends MemTableConfig { public static final long DEFAULT_BUCKET_COUNT = 50000; public static final long DEFAULT_HUGE_PAGE_TLB_SIZE = 0; public static final int DEFAULT_BUCKET_ENTRIES_LOG_THRES = 4096; public static final boolean DEFAULT_IF_LOG_BUCKET_DIST_WHEN_FLUSH = true; public static final int DEFAUL_THRESHOLD_USE_SKIPLIST = 256; /** * HashLinkedListMemTableConfig constructor */ public HashLinkedListMemTableConfig() { bucketCount_ = DEFAULT_BUCKET_COUNT; hugePageTlbSize_ = DEFAULT_HUGE_PAGE_TLB_SIZE; bucketEntriesLoggingThreshold_ = DEFAULT_BUCKET_ENTRIES_LOG_THRES; ifLogBucketDistWhenFlush_ = DEFAULT_IF_LOG_BUCKET_DIST_WHEN_FLUSH; thresholdUseSkiplist_ = DEFAUL_THRESHOLD_USE_SKIPLIST; } /** * Set the number of buckets in the fixed-size array used * in the hash linked-list mem-table. * * @param count the number of hash buckets. * @return the reference to the current HashLinkedListMemTableConfig. */ public HashLinkedListMemTableConfig setBucketCount( final long count) { bucketCount_ = count; return this; } /** * Returns the number of buckets that will be used in the memtable * created based on this config. * * @return the number of buckets */ public long bucketCount() { return bucketCount_; } /** *

Set the size of huge tlb or allocate the hashtable bytes from * malloc if {@code size <= 0}.

* *

The user needs to reserve huge pages for it to be allocated, * like: {@code sysctl -w vm.nr_hugepages=20}

* *

See linux documentation/vm/hugetlbpage.txt

* * @param size if set to {@code <= 0} hashtable bytes from malloc * @return the reference to the current HashLinkedListMemTableConfig. */ public HashLinkedListMemTableConfig setHugePageTlbSize( final long size) { hugePageTlbSize_ = size; return this; } /** * Returns the size value of hugePageTlbSize. * * @return the hugePageTlbSize. */ public long hugePageTlbSize() { return hugePageTlbSize_; } /** * If number of entries in one bucket exceeds that setting, log * about it. * * @param threshold - number of entries in a single bucket before * logging starts. * @return the reference to the current HashLinkedListMemTableConfig. */ public HashLinkedListMemTableConfig setBucketEntriesLoggingThreshold(final int threshold) { bucketEntriesLoggingThreshold_ = threshold; return this; } /** * Returns the maximum number of entries in one bucket before * logging starts. * * @return maximum number of entries in one bucket before logging * starts. */ public int bucketEntriesLoggingThreshold() { return bucketEntriesLoggingThreshold_; } /** * If true the distrubition of number of entries will be logged. * * @param logDistribution - boolean parameter indicating if number * of entry distribution shall be logged. * @return the reference to the current HashLinkedListMemTableConfig. */ public HashLinkedListMemTableConfig setIfLogBucketDistWhenFlush(final boolean logDistribution) { ifLogBucketDistWhenFlush_ = logDistribution; return this; } /** * Returns information about logging the distribution of * number of entries on flush. * * @return if distrubtion of number of entries shall be logged. */ public boolean ifLogBucketDistWhenFlush() { return ifLogBucketDistWhenFlush_; } /** * Set maximum number of entries in one bucket. Exceeding this val * leads to a switch from LinkedList to SkipList. * * @param threshold maximum number of entries before SkipList is * used. * @return the reference to the current HashLinkedListMemTableConfig. */ public HashLinkedListMemTableConfig setThresholdUseSkiplist(final int threshold) { thresholdUseSkiplist_ = threshold; return this; } /** * Returns entries per bucket threshold before LinkedList is * replaced by SkipList usage for that bucket. * * @return entries per bucket threshold before SkipList is used. */ public int thresholdUseSkiplist() { return thresholdUseSkiplist_; } @Override protected long newMemTableFactoryHandle() { return newMemTableFactoryHandle(bucketCount_, hugePageTlbSize_, bucketEntriesLoggingThreshold_, ifLogBucketDistWhenFlush_, thresholdUseSkiplist_); } private native long newMemTableFactoryHandle(long bucketCount, long hugePageTlbSize, int bucketEntriesLoggingThreshold, boolean ifLogBucketDistWhenFlush, int thresholdUseSkiplist) throws IllegalArgumentException; private long bucketCount_; private long hugePageTlbSize_; private int bucketEntriesLoggingThreshold_; private boolean ifLogBucketDistWhenFlush_; private int thresholdUseSkiplist_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/HashSkipListMemTableConfig.java000066400000000000000000000056601267530634000267260ustar00rootroot00000000000000package org.rocksdb; /** * The config for hash skip-list mem-table representation. * Such mem-table representation contains a fix-sized array of * buckets, where each bucket points to a skiplist (or null if the * bucket is empty). * * Note that since this mem-table representation relies on the * key prefix, it is required to invoke one of the usePrefixExtractor * functions to specify how to extract key prefix given a key. * If proper prefix-extractor is not set, then RocksDB will * use the default memtable representation (SkipList) instead * and post a warning in the LOG. */ public class HashSkipListMemTableConfig extends MemTableConfig { public static final int DEFAULT_BUCKET_COUNT = 1000000; public static final int DEFAULT_BRANCHING_FACTOR = 4; public static final int DEFAULT_HEIGHT = 4; /** * HashSkipListMemTableConfig constructor */ public HashSkipListMemTableConfig() { bucketCount_ = DEFAULT_BUCKET_COUNT; branchingFactor_ = DEFAULT_BRANCHING_FACTOR; height_ = DEFAULT_HEIGHT; } /** * Set the number of hash buckets used in the hash skiplist memtable. * Default = 1000000. * * @param count the number of hash buckets used in the hash * skiplist memtable. * @return the reference to the current HashSkipListMemTableConfig. */ public HashSkipListMemTableConfig setBucketCount( final long count) { bucketCount_ = count; return this; } /** * @return the number of hash buckets */ public long bucketCount() { return bucketCount_; } /** * Set the height of the skip list. Default = 4. * * @param height height to set. * * @return the reference to the current HashSkipListMemTableConfig. */ public HashSkipListMemTableConfig setHeight(final int height) { height_ = height; return this; } /** * @return the height of the skip list. */ public int height() { return height_; } /** * Set the branching factor used in the hash skip-list memtable. * This factor controls the probabilistic size ratio between adjacent * links in the skip list. * * @param bf the probabilistic size ratio between adjacent link * lists in the skip list. * @return the reference to the current HashSkipListMemTableConfig. */ public HashSkipListMemTableConfig setBranchingFactor( final int bf) { branchingFactor_ = bf; return this; } /** * @return branching factor, the probabilistic size ratio between * adjacent links in the skip list. */ public int branchingFactor() { return branchingFactor_; } @Override protected long newMemTableFactoryHandle() { return newMemTableFactoryHandle( bucketCount_, height_, branchingFactor_); } private native long newMemTableFactoryHandle( long bucketCount, int height, int branchingFactor) throws IllegalArgumentException; private long bucketCount_; private int branchingFactor_; private int height_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/HistogramData.java000066400000000000000000000022321267530634000243420ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public class HistogramData { private final double median_; private final double percentile95_; private final double percentile99_; private final double average_; private final double standardDeviation_; public HistogramData(final double median, final double percentile95, final double percentile99, final double average, final double standardDeviation) { median_ = median; percentile95_ = percentile95; percentile99_ = percentile99; average_ = average; standardDeviation_ = standardDeviation; } public double getMedian() { return median_; } public double getPercentile95() { return percentile95_; } public double getPercentile99() { return percentile99_; } public double getAverage() { return average_; } public double getStandardDeviation() { return standardDeviation_; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/HistogramType.java000066400000000000000000000020631267530634000244140ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public enum HistogramType { DB_GET(0), DB_WRITE(1), COMPACTION_TIME(2), TABLE_SYNC_MICROS(3), COMPACTION_OUTFILE_SYNC_MICROS(4), WAL_FILE_SYNC_MICROS(5), MANIFEST_FILE_SYNC_MICROS(6), // TIME SPENT IN IO DURING TABLE OPEN TABLE_OPEN_IO_MICROS(7), DB_MULTIGET(8), READ_BLOCK_COMPACTION_MICROS(9), READ_BLOCK_GET_MICROS(10), WRITE_RAW_BLOCK_MICROS(11), STALL_L0_SLOWDOWN_COUNT(12), STALL_MEMTABLE_COMPACTION_COUNT(13), STALL_L0_NUM_FILES_COUNT(14), HARD_RATE_LIMIT_DELAY_COUNT(15), SOFT_RATE_LIMIT_DELAY_COUNT(16), NUM_FILES_IN_SINGLE_COMPACTION(17), DB_SEEK(18), WRITE_STALL(19); private final int value_; private HistogramType(int value) { value_ = value; } public int getValue() { return value_; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/IndexType.java000066400000000000000000000016321267530634000235270ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * IndexType used in conjunction with BlockBasedTable. */ public enum IndexType { /** * A space efficient index block that is optimized for * binary-search-based index. */ kBinarySearch((byte) 0), /** * The hash index, if enabled, will do the hash lookup when * {@code Options.prefix_extractor} is provided. */ kHashSearch((byte) 1); /** * Returns the byte value of the enumerations value * * @return byte representation */ public byte getValue() { return value_; } private IndexType(byte value) { value_ = value; } private final byte value_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/InfoLogLevel.java000066400000000000000000000020361267530634000241420ustar00rootroot00000000000000package org.rocksdb; /** * RocksDB log levels. */ public enum InfoLogLevel { DEBUG_LEVEL((byte)0), INFO_LEVEL((byte)1), WARN_LEVEL((byte)2), ERROR_LEVEL((byte)3), FATAL_LEVEL((byte)4), NUM_INFO_LOG_LEVELS((byte)5); private final byte value_; private InfoLogLevel(byte value) { value_ = value; } /** * Returns the byte value of the enumerations value * * @return byte representation */ public byte getValue() { return value_; } /** * Get InfoLogLevel by byte value. * * @param value byte representation of InfoLogLevel. * * @return {@link org.rocksdb.InfoLogLevel} instance or null. * @throws java.lang.IllegalArgumentException if an invalid * value is provided. */ public static InfoLogLevel getInfoLogLevel(byte value) { for (InfoLogLevel infoLogLevel : InfoLogLevel.values()) { if (infoLogLevel.getValue() == value){ return infoLogLevel; } } throw new IllegalArgumentException( "Illegal value provided for InfoLogLevel."); } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Logger.java000066400000000000000000000066471267530634000230500ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

This class provides a custom logger functionality * in Java which wraps {@code RocksDB} logging facilities. *

* *

Using this class RocksDB can log with common * Java logging APIs like Log4j or Slf4j without keeping * database logs in the filesystem.

* * Performance *

There are certain performance penalties using a Java * {@code Logger} implementation within production code. *

* *

* A log level can be set using {@link org.rocksdb.Options} or * {@link Logger#setInfoLogLevel(InfoLogLevel)}. The set log level * influences the underlying native code. Each log message is * checked against the set log level and if the log level is more * verbose as the set log level, native allocations will be made * and data structures are allocated. *

* *

Every log message which will be emitted by native code will * trigger expensive native to Java transitions. So the preferred * setting for production use is either * {@link org.rocksdb.InfoLogLevel#ERROR_LEVEL} or * {@link org.rocksdb.InfoLogLevel#FATAL_LEVEL}. *

*/ public abstract class Logger extends RocksObject { /** *

AbstractLogger constructor.

* *

Important: the log level set within * the {@link org.rocksdb.Options} instance will be used as * maximum log level of RocksDB.

* * @param options {@link org.rocksdb.Options} instance. */ public Logger(final Options options) { createNewLoggerOptions(options.nativeHandle_); } /** *

AbstractLogger constructor.

* *

Important: the log level set within * the {@link org.rocksdb.DBOptions} instance will be used * as maximum log level of RocksDB.

* * @param dboptions {@link org.rocksdb.DBOptions} instance. */ public Logger(final DBOptions dboptions) { createNewLoggerDbOptions(dboptions.nativeHandle_); } /** * Set {@link org.rocksdb.InfoLogLevel} to AbstractLogger. * * @param infoLogLevel {@link org.rocksdb.InfoLogLevel} instance. */ public void setInfoLogLevel(final InfoLogLevel infoLogLevel) { setInfoLogLevel(nativeHandle_, infoLogLevel.getValue()); } /** * Return the loggers log level. * * @return {@link org.rocksdb.InfoLogLevel} instance. */ public InfoLogLevel infoLogLevel() { return InfoLogLevel.getInfoLogLevel( infoLogLevel(nativeHandle_)); } protected abstract void log(InfoLogLevel infoLogLevel, String logMsg); /** * Deletes underlying C++ slice pointer. * Note that this function should be called only after all * RocksDB instances referencing the slice are closed. * Otherwise an undefined behavior will occur. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } protected native void createNewLoggerOptions( long options); protected native void createNewLoggerDbOptions( long dbOptions); protected native void setInfoLogLevel(long handle, byte infoLogLevel); protected native byte infoLogLevel(long handle); private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/MemTableConfig.java000066400000000000000000000022061267530634000244300ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * MemTableConfig is used to config the internal mem-table of a RocksDB. * It is required for each memtable to have one such sub-class to allow * Java developers to use it. * * To make a RocksDB to use a specific MemTable format, its associated * MemTableConfig should be properly set and passed into Options * via Options.setMemTableFactory() and open the db using that Options. * * @see Options */ public abstract class MemTableConfig { /** * This function should only be called by Options.setMemTableConfig(), * which will create a c++ shared-pointer to the c++ MemTableRepFactory * that associated with the Java MemTableConfig. * * @see Options#setMemTableConfig(MemTableConfig) * * @return native handle address to native memory table instance. */ abstract protected long newMemTableFactoryHandle(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/MergeOperator.java000066400000000000000000000010531267530634000243660ustar00rootroot00000000000000// Copyright (c) 2014, Vlad Balan (vlad.gm@gmail.com). All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * MergeOperator holds an operator to be applied when compacting * two merge operands held under the same key in order to obtain a single * value. */ public interface MergeOperator { long newMergeOperatorHandle(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/NativeLibraryLoader.java000066400000000000000000000075751267530634000255340ustar00rootroot00000000000000package org.rocksdb; import java.io.*; import java.nio.file.Files; import java.nio.file.StandardCopyOption; import org.rocksdb.util.Environment; /** * This class is used to load the RocksDB shared library from within the jar. * The shared library is extracted to a temp folder and loaded from there. */ public class NativeLibraryLoader { //singleton private static final NativeLibraryLoader instance = new NativeLibraryLoader(); private static boolean initialized = false; private static final String sharedLibraryName = Environment.getSharedLibraryName("rocksdb"); private static final String jniLibraryName = Environment.getJniLibraryName("rocksdb"); private static final String jniLibraryFileName = Environment.getJniLibraryFileName("rocksdb"); private static final String tempFilePrefix = "librocksdbjni"; private static final String tempFileSuffix = Environment.getJniLibraryExtension(); /** * Get a reference to the NativeLibraryLoader * * @return The NativeLibraryLoader */ public static NativeLibraryLoader getInstance() { return instance; } /** * Firstly attempts to load the library from java.library.path, * if that fails then it falls back to extracting * the library from the classpath * {@link org.rocksdb.NativeLibraryLoader#loadLibraryFromJar(java.lang.String)} * * @param tmpDir A temporary directory to use * to copy the native library to when loading from the classpath. * If null, or the empty string, we rely on Java's * {@link java.io.File#createTempFile(String, String)} * function to provide a temporary location. * The temporary file will be registered for deletion * on exit. * * @throws java.io.IOException if a filesystem operation fails. */ public synchronized void loadLibrary(final String tmpDir) throws IOException { try { System.loadLibrary(sharedLibraryName); } catch(final UnsatisfiedLinkError ule1) { try { System.loadLibrary(jniLibraryName); } catch(final UnsatisfiedLinkError ule2) { loadLibraryFromJar(tmpDir); } } } /** * Attempts to extract the native RocksDB library * from the classpath and load it * * @param tmpDir A temporary directory to use * to copy the native library to. If null, * or the empty string, we rely on Java's * {@link java.io.File#createTempFile(String, String)} * function to provide a temporary location. * The temporary file will be registered for deletion * on exit. * * @throws java.io.IOException if a filesystem operation fails. */ void loadLibraryFromJar(final String tmpDir) throws IOException { if (!initialized) { System.load(loadLibraryFromJarToTemp(tmpDir).getAbsolutePath()); initialized = true; } } File loadLibraryFromJarToTemp(final String tmpDir) throws IOException { final File temp; if (tmpDir == null || tmpDir.isEmpty()) { temp = File.createTempFile(tempFilePrefix, tempFileSuffix); } else { temp = new File(tmpDir, jniLibraryFileName); if (!temp.createNewFile()) { throw new RuntimeException("File: " + temp.getAbsolutePath() + " could not be created."); } } if (!temp.exists()) { throw new RuntimeException("File " + temp.getAbsolutePath() + " does not exist."); } else { temp.deleteOnExit(); } // attempt to copy the library from the Jar file to the temp destination try (final InputStream is = getClass().getClassLoader(). getResourceAsStream(jniLibraryFileName)) { if (is == null) { throw new RuntimeException(jniLibraryFileName + " was not found inside JAR."); } else { Files.copy(is, temp.toPath(), StandardCopyOption.REPLACE_EXISTING); } } return temp; } /** * Private constructor to disallow instantiation */ private NativeLibraryLoader() { } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Options.java000066400000000000000000001151451267530634000232560ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.ArrayList; import java.util.List; /** * Options to control the behavior of a database. It will be used * during the creation of a {@link org.rocksdb.RocksDB} (i.e., RocksDB.open()). * * If {@link #dispose()} function is not called, then it will be GC'd automatically * and native resources will be released as part of the process. */ public class Options extends RocksObject implements DBOptionsInterface, ColumnFamilyOptionsInterface { static { RocksDB.loadLibrary(); } /** * Construct options for opening a RocksDB. * * This constructor will create (by allocating a block of memory) * an {@code rocksdb::Options} in the c++ side. */ public Options() { super(); newOptions(); env_ = Env.getDefault(); } /** * Construct options for opening a RocksDB. Reusing database options * and column family options. * * @param dbOptions {@link org.rocksdb.DBOptions} instance * @param columnFamilyOptions {@link org.rocksdb.ColumnFamilyOptions} * instance */ public Options(final DBOptions dbOptions, final ColumnFamilyOptions columnFamilyOptions) { super(); newOptions(dbOptions.nativeHandle_, columnFamilyOptions.nativeHandle_); env_ = Env.getDefault(); } @Override public Options setIncreaseParallelism(final int totalThreads) { assert(isInitialized()); setIncreaseParallelism(nativeHandle_, totalThreads); return this; } @Override public Options setCreateIfMissing(final boolean flag) { assert(isInitialized()); setCreateIfMissing(nativeHandle_, flag); return this; } @Override public Options setCreateMissingColumnFamilies(final boolean flag) { assert(isInitialized()); setCreateMissingColumnFamilies(nativeHandle_, flag); return this; } /** * Use the specified object to interact with the environment, * e.g. to read/write files, schedule background work, etc. * Default: {@link Env#getDefault()} * * @param env {@link Env} instance. * @return the instance of the current Options. */ public Options setEnv(final Env env) { assert(isInitialized()); setEnv(nativeHandle_, env.nativeHandle_); env_ = env; return this; } /** * Returns the set RocksEnv instance. * * @return {@link RocksEnv} instance set in the Options. */ public Env getEnv() { return env_; } /** *

Set appropriate parameters for bulk loading. * The reason that this is a function that returns "this" instead of a * constructor is to enable chaining of multiple similar calls in the future. *

* *

All data will be in level 0 without any automatic compaction. * It's recommended to manually call CompactRange(NULL, NULL) before reading * from the database, because otherwise the read can be very slow.

* * @return the instance of the current Options. */ public Options prepareForBulkLoad() { prepareForBulkLoad(nativeHandle_); return this; } @Override public boolean createIfMissing() { assert(isInitialized()); return createIfMissing(nativeHandle_); } @Override public boolean createMissingColumnFamilies() { assert(isInitialized()); return createMissingColumnFamilies(nativeHandle_); } @Override public Options optimizeForPointLookup( long blockCacheSizeMb) { optimizeForPointLookup(nativeHandle_, blockCacheSizeMb); return this; } @Override public Options optimizeLevelStyleCompaction() { optimizeLevelStyleCompaction(nativeHandle_, DEFAULT_COMPACTION_MEMTABLE_MEMORY_BUDGET); return this; } @Override public Options optimizeLevelStyleCompaction( long memtableMemoryBudget) { optimizeLevelStyleCompaction(nativeHandle_, memtableMemoryBudget); return this; } @Override public Options optimizeUniversalStyleCompaction() { optimizeUniversalStyleCompaction(nativeHandle_, DEFAULT_COMPACTION_MEMTABLE_MEMORY_BUDGET); return this; } @Override public Options optimizeUniversalStyleCompaction( final long memtableMemoryBudget) { optimizeUniversalStyleCompaction(nativeHandle_, memtableMemoryBudget); return this; } @Override public Options setComparator(final BuiltinComparator builtinComparator) { assert(isInitialized()); setComparatorHandle(nativeHandle_, builtinComparator.ordinal()); return this; } @Override public Options setComparator( final AbstractComparator> comparator) { assert (isInitialized()); setComparatorHandle(nativeHandle_, comparator.nativeHandle_); comparator_ = comparator; return this; } @Override public Options setMergeOperatorName(final String name) { assert (isInitialized()); if (name == null) { throw new IllegalArgumentException( "Merge operator name must not be null."); } setMergeOperatorName(nativeHandle_, name); return this; } @Override public Options setMergeOperator(final MergeOperator mergeOperator) { setMergeOperator(nativeHandle_, mergeOperator.newMergeOperatorHandle()); return this; } @Override public Options setWriteBufferSize(final long writeBufferSize) { assert(isInitialized()); setWriteBufferSize(nativeHandle_, writeBufferSize); return this; } @Override public long writeBufferSize() { assert(isInitialized()); return writeBufferSize(nativeHandle_); } @Override public Options setMaxWriteBufferNumber(final int maxWriteBufferNumber) { assert(isInitialized()); setMaxWriteBufferNumber(nativeHandle_, maxWriteBufferNumber); return this; } @Override public int maxWriteBufferNumber() { assert(isInitialized()); return maxWriteBufferNumber(nativeHandle_); } @Override public boolean errorIfExists() { assert(isInitialized()); return errorIfExists(nativeHandle_); } @Override public Options setErrorIfExists(final boolean errorIfExists) { assert(isInitialized()); setErrorIfExists(nativeHandle_, errorIfExists); return this; } @Override public boolean paranoidChecks() { assert(isInitialized()); return paranoidChecks(nativeHandle_); } @Override public Options setParanoidChecks(final boolean paranoidChecks) { assert(isInitialized()); setParanoidChecks(nativeHandle_, paranoidChecks); return this; } @Override public int maxOpenFiles() { assert(isInitialized()); return maxOpenFiles(nativeHandle_); } @Override public Options setMaxTotalWalSize(final long maxTotalWalSize) { assert(isInitialized()); setMaxTotalWalSize(nativeHandle_, maxTotalWalSize); return this; } @Override public long maxTotalWalSize() { assert(isInitialized()); return maxTotalWalSize(nativeHandle_); } @Override public Options setMaxOpenFiles(final int maxOpenFiles) { assert(isInitialized()); setMaxOpenFiles(nativeHandle_, maxOpenFiles); return this; } @Override public boolean disableDataSync() { assert(isInitialized()); return disableDataSync(nativeHandle_); } @Override public Options setDisableDataSync(final boolean disableDataSync) { assert(isInitialized()); setDisableDataSync(nativeHandle_, disableDataSync); return this; } @Override public boolean useFsync() { assert(isInitialized()); return useFsync(nativeHandle_); } @Override public Options setUseFsync(final boolean useFsync) { assert(isInitialized()); setUseFsync(nativeHandle_, useFsync); return this; } @Override public String dbLogDir() { assert(isInitialized()); return dbLogDir(nativeHandle_); } @Override public Options setDbLogDir(final String dbLogDir) { assert(isInitialized()); setDbLogDir(nativeHandle_, dbLogDir); return this; } @Override public String walDir() { assert(isInitialized()); return walDir(nativeHandle_); } @Override public Options setWalDir(final String walDir) { assert(isInitialized()); setWalDir(nativeHandle_, walDir); return this; } @Override public long deleteObsoleteFilesPeriodMicros() { assert(isInitialized()); return deleteObsoleteFilesPeriodMicros(nativeHandle_); } @Override public Options setDeleteObsoleteFilesPeriodMicros( final long micros) { assert(isInitialized()); setDeleteObsoleteFilesPeriodMicros(nativeHandle_, micros); return this; } @Override public int maxBackgroundCompactions() { assert(isInitialized()); return maxBackgroundCompactions(nativeHandle_); } @Override public Options createStatistics() { assert(isInitialized()); createStatistics(nativeHandle_); return this; } @Override public Statistics statisticsPtr() { assert(isInitialized()); long statsPtr = statisticsPtr(nativeHandle_); if(statsPtr == 0) { createStatistics(); statsPtr = statisticsPtr(nativeHandle_); } return new Statistics(statsPtr); } @Override public Options setMaxBackgroundCompactions( final int maxBackgroundCompactions) { assert(isInitialized()); setMaxBackgroundCompactions(nativeHandle_, maxBackgroundCompactions); return this; } @Override public int maxBackgroundFlushes() { assert(isInitialized()); return maxBackgroundFlushes(nativeHandle_); } @Override public Options setMaxBackgroundFlushes( final int maxBackgroundFlushes) { assert(isInitialized()); setMaxBackgroundFlushes(nativeHandle_, maxBackgroundFlushes); return this; } @Override public long maxLogFileSize() { assert(isInitialized()); return maxLogFileSize(nativeHandle_); } @Override public Options setMaxLogFileSize(final long maxLogFileSize) { assert(isInitialized()); setMaxLogFileSize(nativeHandle_, maxLogFileSize); return this; } @Override public long logFileTimeToRoll() { assert(isInitialized()); return logFileTimeToRoll(nativeHandle_); } @Override public Options setLogFileTimeToRoll(final long logFileTimeToRoll) { assert(isInitialized()); setLogFileTimeToRoll(nativeHandle_, logFileTimeToRoll); return this; } @Override public long keepLogFileNum() { assert(isInitialized()); return keepLogFileNum(nativeHandle_); } @Override public Options setKeepLogFileNum(final long keepLogFileNum) { assert(isInitialized()); setKeepLogFileNum(nativeHandle_, keepLogFileNum); return this; } @Override public long maxManifestFileSize() { assert(isInitialized()); return maxManifestFileSize(nativeHandle_); } @Override public Options setMaxManifestFileSize( final long maxManifestFileSize) { assert(isInitialized()); setMaxManifestFileSize(nativeHandle_, maxManifestFileSize); return this; } @Override public Options setMaxTableFilesSizeFIFO( final long maxTableFilesSize) { assert(maxTableFilesSize > 0); // unsigned native type assert(isInitialized()); setMaxTableFilesSizeFIFO(nativeHandle_, maxTableFilesSize); return this; } @Override public long maxTableFilesSizeFIFO() { return maxTableFilesSizeFIFO(nativeHandle_); } @Override public int tableCacheNumshardbits() { assert(isInitialized()); return tableCacheNumshardbits(nativeHandle_); } @Override public Options setTableCacheNumshardbits( final int tableCacheNumshardbits) { assert(isInitialized()); setTableCacheNumshardbits(nativeHandle_, tableCacheNumshardbits); return this; } @Override public long walTtlSeconds() { assert(isInitialized()); return walTtlSeconds(nativeHandle_); } @Override public Options setWalTtlSeconds(final long walTtlSeconds) { assert(isInitialized()); setWalTtlSeconds(nativeHandle_, walTtlSeconds); return this; } @Override public long walSizeLimitMB() { assert(isInitialized()); return walSizeLimitMB(nativeHandle_); } @Override public Options setWalSizeLimitMB(final long sizeLimitMB) { assert(isInitialized()); setWalSizeLimitMB(nativeHandle_, sizeLimitMB); return this; } @Override public long manifestPreallocationSize() { assert(isInitialized()); return manifestPreallocationSize(nativeHandle_); } @Override public Options setManifestPreallocationSize(final long size) { assert(isInitialized()); setManifestPreallocationSize(nativeHandle_, size); return this; } @Override public boolean allowOsBuffer() { assert(isInitialized()); return allowOsBuffer(nativeHandle_); } @Override public Options setAllowOsBuffer(final boolean allowOsBuffer) { assert(isInitialized()); setAllowOsBuffer(nativeHandle_, allowOsBuffer); return this; } @Override public boolean allowMmapReads() { assert(isInitialized()); return allowMmapReads(nativeHandle_); } @Override public Options setAllowMmapReads(final boolean allowMmapReads) { assert(isInitialized()); setAllowMmapReads(nativeHandle_, allowMmapReads); return this; } @Override public boolean allowMmapWrites() { assert(isInitialized()); return allowMmapWrites(nativeHandle_); } @Override public Options setAllowMmapWrites(final boolean allowMmapWrites) { assert(isInitialized()); setAllowMmapWrites(nativeHandle_, allowMmapWrites); return this; } @Override public boolean isFdCloseOnExec() { assert(isInitialized()); return isFdCloseOnExec(nativeHandle_); } @Override public Options setIsFdCloseOnExec(final boolean isFdCloseOnExec) { assert(isInitialized()); setIsFdCloseOnExec(nativeHandle_, isFdCloseOnExec); return this; } @Override public int statsDumpPeriodSec() { assert(isInitialized()); return statsDumpPeriodSec(nativeHandle_); } @Override public Options setStatsDumpPeriodSec(final int statsDumpPeriodSec) { assert(isInitialized()); setStatsDumpPeriodSec(nativeHandle_, statsDumpPeriodSec); return this; } @Override public boolean adviseRandomOnOpen() { return adviseRandomOnOpen(nativeHandle_); } @Override public Options setAdviseRandomOnOpen(final boolean adviseRandomOnOpen) { assert(isInitialized()); setAdviseRandomOnOpen(nativeHandle_, adviseRandomOnOpen); return this; } @Override public boolean useAdaptiveMutex() { assert(isInitialized()); return useAdaptiveMutex(nativeHandle_); } @Override public Options setUseAdaptiveMutex(final boolean useAdaptiveMutex) { assert(isInitialized()); setUseAdaptiveMutex(nativeHandle_, useAdaptiveMutex); return this; } @Override public long bytesPerSync() { return bytesPerSync(nativeHandle_); } @Override public Options setBytesPerSync(final long bytesPerSync) { assert(isInitialized()); setBytesPerSync(nativeHandle_, bytesPerSync); return this; } @Override public Options setMemTableConfig(final MemTableConfig config) { memTableConfig_ = config; setMemTableFactory(nativeHandle_, config.newMemTableFactoryHandle()); return this; } @Override public Options setRateLimiterConfig(final RateLimiterConfig config) { rateLimiterConfig_ = config; setRateLimiter(nativeHandle_, config.newRateLimiterHandle()); return this; } @Override public Options setLogger(final Logger logger) { assert(isInitialized()); setLogger(nativeHandle_, logger.nativeHandle_); return this; } @Override public Options setInfoLogLevel(final InfoLogLevel infoLogLevel) { assert(isInitialized()); setInfoLogLevel(nativeHandle_, infoLogLevel.getValue()); return this; } @Override public InfoLogLevel infoLogLevel() { assert(isInitialized()); return InfoLogLevel.getInfoLogLevel( infoLogLevel(nativeHandle_)); } @Override public String memTableFactoryName() { assert(isInitialized()); return memTableFactoryName(nativeHandle_); } @Override public Options setTableFormatConfig(final TableFormatConfig config) { tableFormatConfig_ = config; setTableFactory(nativeHandle_, config.newTableFactoryHandle()); return this; } @Override public String tableFactoryName() { assert(isInitialized()); return tableFactoryName(nativeHandle_); } @Override public Options useFixedLengthPrefixExtractor(final int n) { assert(isInitialized()); useFixedLengthPrefixExtractor(nativeHandle_, n); return this; } @Override public Options useCappedPrefixExtractor(final int n) { assert(isInitialized()); useCappedPrefixExtractor(nativeHandle_, n); return this; } @Override public CompressionType compressionType() { return CompressionType.values()[compressionType(nativeHandle_)]; } @Override public Options setCompressionPerLevel(final List compressionLevels) { final List byteCompressionTypes = new ArrayList<>( compressionLevels.size()); for (final CompressionType compressionLevel : compressionLevels) { byteCompressionTypes.add(compressionLevel.getValue()); } setCompressionPerLevel(nativeHandle_, byteCompressionTypes); return this; } @Override public List compressionPerLevel() { final List byteCompressionTypes = compressionPerLevel(nativeHandle_); final List compressionLevels = new ArrayList<>(); for (final Byte byteCompressionType : byteCompressionTypes) { compressionLevels.add(CompressionType.getCompressionType( byteCompressionType)); } return compressionLevels; } @Override public Options setCompressionType(CompressionType compressionType) { setCompressionType(nativeHandle_, compressionType.getValue()); return this; } @Override public CompactionStyle compactionStyle() { return CompactionStyle.values()[compactionStyle(nativeHandle_)]; } @Override public Options setCompactionStyle( final CompactionStyle compactionStyle) { setCompactionStyle(nativeHandle_, compactionStyle.getValue()); return this; } @Override public int numLevels() { return numLevels(nativeHandle_); } @Override public Options setNumLevels(int numLevels) { setNumLevels(nativeHandle_, numLevels); return this; } @Override public int levelZeroFileNumCompactionTrigger() { return levelZeroFileNumCompactionTrigger(nativeHandle_); } @Override public Options setLevelZeroFileNumCompactionTrigger( final int numFiles) { setLevelZeroFileNumCompactionTrigger( nativeHandle_, numFiles); return this; } @Override public int levelZeroSlowdownWritesTrigger() { return levelZeroSlowdownWritesTrigger(nativeHandle_); } @Override public Options setLevelZeroSlowdownWritesTrigger( final int numFiles) { setLevelZeroSlowdownWritesTrigger(nativeHandle_, numFiles); return this; } @Override public int levelZeroStopWritesTrigger() { return levelZeroStopWritesTrigger(nativeHandle_); } @Override public Options setLevelZeroStopWritesTrigger( final int numFiles) { setLevelZeroStopWritesTrigger(nativeHandle_, numFiles); return this; } @Override public int maxMemCompactionLevel() { return 0; } @Override public Options setMaxMemCompactionLevel( final int maxMemCompactionLevel) { return this; } @Override public long targetFileSizeBase() { return targetFileSizeBase(nativeHandle_); } @Override public Options setTargetFileSizeBase(long targetFileSizeBase) { setTargetFileSizeBase(nativeHandle_, targetFileSizeBase); return this; } @Override public int targetFileSizeMultiplier() { return targetFileSizeMultiplier(nativeHandle_); } @Override public Options setTargetFileSizeMultiplier(int multiplier) { setTargetFileSizeMultiplier(nativeHandle_, multiplier); return this; } @Override public Options setMaxBytesForLevelBase(final long maxBytesForLevelBase) { setMaxBytesForLevelBase(nativeHandle_, maxBytesForLevelBase); return this; } @Override public long maxBytesForLevelBase() { return maxBytesForLevelBase(nativeHandle_); } @Override public Options setLevelCompactionDynamicLevelBytes( final boolean enableLevelCompactionDynamicLevelBytes) { setLevelCompactionDynamicLevelBytes(nativeHandle_, enableLevelCompactionDynamicLevelBytes); return this; } @Override public boolean levelCompactionDynamicLevelBytes() { return levelCompactionDynamicLevelBytes(nativeHandle_); } @Override public int maxBytesForLevelMultiplier() { return maxBytesForLevelMultiplier(nativeHandle_); } @Override public Options setMaxBytesForLevelMultiplier(final int multiplier) { setMaxBytesForLevelMultiplier(nativeHandle_, multiplier); return this; } @Override public int expandedCompactionFactor() { return expandedCompactionFactor(nativeHandle_); } @Override public Options setExpandedCompactionFactor( final int expandedCompactionFactor) { setExpandedCompactionFactor(nativeHandle_, expandedCompactionFactor); return this; } @Override public int sourceCompactionFactor() { return sourceCompactionFactor(nativeHandle_); } @Override public Options setSourceCompactionFactor(int sourceCompactionFactor) { setSourceCompactionFactor(nativeHandle_, sourceCompactionFactor); return this; } @Override public int maxGrandparentOverlapFactor() { return maxGrandparentOverlapFactor(nativeHandle_); } @Override public Options setMaxGrandparentOverlapFactor( final int maxGrandparentOverlapFactor) { setMaxGrandparentOverlapFactor(nativeHandle_, maxGrandparentOverlapFactor); return this; } @Override public double softRateLimit() { return softRateLimit(nativeHandle_); } @Override public Options setSoftRateLimit(final double softRateLimit) { setSoftRateLimit(nativeHandle_, softRateLimit); return this; } @Override public double hardRateLimit() { return hardRateLimit(nativeHandle_); } @Override public Options setHardRateLimit(double hardRateLimit) { setHardRateLimit(nativeHandle_, hardRateLimit); return this; } @Override public int rateLimitDelayMaxMilliseconds() { return rateLimitDelayMaxMilliseconds(nativeHandle_); } @Override public Options setRateLimitDelayMaxMilliseconds( final int rateLimitDelayMaxMilliseconds) { setRateLimitDelayMaxMilliseconds( nativeHandle_, rateLimitDelayMaxMilliseconds); return this; } @Override public long arenaBlockSize() { return arenaBlockSize(nativeHandle_); } @Override public Options setArenaBlockSize(final long arenaBlockSize) { setArenaBlockSize(nativeHandle_, arenaBlockSize); return this; } @Override public boolean disableAutoCompactions() { return disableAutoCompactions(nativeHandle_); } @Override public Options setDisableAutoCompactions( final boolean disableAutoCompactions) { setDisableAutoCompactions(nativeHandle_, disableAutoCompactions); return this; } @Override public boolean purgeRedundantKvsWhileFlush() { return purgeRedundantKvsWhileFlush(nativeHandle_); } @Override public Options setPurgeRedundantKvsWhileFlush( final boolean purgeRedundantKvsWhileFlush) { setPurgeRedundantKvsWhileFlush( nativeHandle_, purgeRedundantKvsWhileFlush); return this; } @Override public boolean verifyChecksumsInCompaction() { return verifyChecksumsInCompaction(nativeHandle_); } @Override public Options setVerifyChecksumsInCompaction( final boolean verifyChecksumsInCompaction) { setVerifyChecksumsInCompaction( nativeHandle_, verifyChecksumsInCompaction); return this; } @Override public boolean filterDeletes() { return filterDeletes(nativeHandle_); } @Override public Options setFilterDeletes( final boolean filterDeletes) { setFilterDeletes(nativeHandle_, filterDeletes); return this; } @Override public long maxSequentialSkipInIterations() { return maxSequentialSkipInIterations(nativeHandle_); } @Override public Options setMaxSequentialSkipInIterations( final long maxSequentialSkipInIterations) { setMaxSequentialSkipInIterations(nativeHandle_, maxSequentialSkipInIterations); return this; } @Override public boolean inplaceUpdateSupport() { return inplaceUpdateSupport(nativeHandle_); } @Override public Options setInplaceUpdateSupport( final boolean inplaceUpdateSupport) { setInplaceUpdateSupport(nativeHandle_, inplaceUpdateSupport); return this; } @Override public long inplaceUpdateNumLocks() { return inplaceUpdateNumLocks(nativeHandle_); } @Override public Options setInplaceUpdateNumLocks( final long inplaceUpdateNumLocks) { setInplaceUpdateNumLocks(nativeHandle_, inplaceUpdateNumLocks); return this; } @Override public int memtablePrefixBloomBits() { return memtablePrefixBloomBits(nativeHandle_); } @Override public Options setMemtablePrefixBloomBits( final int memtablePrefixBloomBits) { setMemtablePrefixBloomBits(nativeHandle_, memtablePrefixBloomBits); return this; } @Override public int memtablePrefixBloomProbes() { return memtablePrefixBloomProbes(nativeHandle_); } @Override public Options setMemtablePrefixBloomProbes( final int memtablePrefixBloomProbes) { setMemtablePrefixBloomProbes(nativeHandle_, memtablePrefixBloomProbes); return this; } @Override public int bloomLocality() { return bloomLocality(nativeHandle_); } @Override public Options setBloomLocality(final int bloomLocality) { setBloomLocality(nativeHandle_, bloomLocality); return this; } @Override public long maxSuccessiveMerges() { return maxSuccessiveMerges(nativeHandle_); } @Override public Options setMaxSuccessiveMerges(long maxSuccessiveMerges) { setMaxSuccessiveMerges(nativeHandle_, maxSuccessiveMerges); return this; } @Override public int minWriteBufferNumberToMerge() { return minWriteBufferNumberToMerge(nativeHandle_); } @Override public Options setMinWriteBufferNumberToMerge( final int minWriteBufferNumberToMerge) { setMinWriteBufferNumberToMerge(nativeHandle_, minWriteBufferNumberToMerge); return this; } @Override public int minPartialMergeOperands() { return minPartialMergeOperands(nativeHandle_); } @Override public Options setMinPartialMergeOperands( final int minPartialMergeOperands) { setMinPartialMergeOperands(nativeHandle_, minPartialMergeOperands); return this; } @Override public Options setOptimizeFiltersForHits( final boolean optimizeFiltersForHits) { setOptimizeFiltersForHits(nativeHandle_, optimizeFiltersForHits); return this; } @Override public boolean optimizeFiltersForHits() { return optimizeFiltersForHits(nativeHandle_); } /** * Release the memory allocated for the current instance * in the c++ side. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } private native void newOptions(); private native void newOptions(long dbOptHandle, long cfOptHandle); private native void disposeInternal(long handle); private native void setEnv(long optHandle, long envHandle); private native void prepareForBulkLoad(long handle); // DB native handles private native void setIncreaseParallelism(long handle, int totalThreads); private native void setCreateIfMissing(long handle, boolean flag); private native boolean createIfMissing(long handle); private native void setCreateMissingColumnFamilies( long handle, boolean flag); private native boolean createMissingColumnFamilies(long handle); private native void setErrorIfExists(long handle, boolean errorIfExists); private native boolean errorIfExists(long handle); private native void setParanoidChecks( long handle, boolean paranoidChecks); private native boolean paranoidChecks(long handle); private native void setRateLimiter(long handle, long rateLimiterHandle); private native void setLogger(long handle, long loggerHandle); private native void setInfoLogLevel(long handle, byte logLevel); private native byte infoLogLevel(long handle); private native void setMaxOpenFiles(long handle, int maxOpenFiles); private native int maxOpenFiles(long handle); private native void setMaxTotalWalSize(long handle, long maxTotalWalSize); private native long maxTotalWalSize(long handle); private native void createStatistics(long optHandle); private native long statisticsPtr(long optHandle); private native void setDisableDataSync(long handle, boolean disableDataSync); private native boolean disableDataSync(long handle); private native boolean useFsync(long handle); private native void setUseFsync(long handle, boolean useFsync); private native void setDbLogDir(long handle, String dbLogDir); private native String dbLogDir(long handle); private native void setWalDir(long handle, String walDir); private native String walDir(long handle); private native void setDeleteObsoleteFilesPeriodMicros( long handle, long micros); private native long deleteObsoleteFilesPeriodMicros(long handle); private native void setMaxBackgroundCompactions( long handle, int maxBackgroundCompactions); private native int maxBackgroundCompactions(long handle); private native void setMaxBackgroundFlushes( long handle, int maxBackgroundFlushes); private native int maxBackgroundFlushes(long handle); private native void setMaxLogFileSize(long handle, long maxLogFileSize) throws IllegalArgumentException; private native long maxLogFileSize(long handle); private native void setLogFileTimeToRoll( long handle, long logFileTimeToRoll) throws IllegalArgumentException; private native long logFileTimeToRoll(long handle); private native void setKeepLogFileNum(long handle, long keepLogFileNum) throws IllegalArgumentException; private native long keepLogFileNum(long handle); private native void setMaxManifestFileSize( long handle, long maxManifestFileSize); private native long maxManifestFileSize(long handle); private native void setMaxTableFilesSizeFIFO( long handle, long maxTableFilesSize); private native long maxTableFilesSizeFIFO(long handle); private native void setTableCacheNumshardbits( long handle, int tableCacheNumshardbits); private native int tableCacheNumshardbits(long handle); private native void setWalTtlSeconds(long handle, long walTtlSeconds); private native long walTtlSeconds(long handle); private native void setWalSizeLimitMB(long handle, long sizeLimitMB); private native long walSizeLimitMB(long handle); private native void setManifestPreallocationSize( long handle, long size) throws IllegalArgumentException; private native long manifestPreallocationSize(long handle); private native void setAllowOsBuffer( long handle, boolean allowOsBuffer); private native boolean allowOsBuffer(long handle); private native void setAllowMmapReads( long handle, boolean allowMmapReads); private native boolean allowMmapReads(long handle); private native void setAllowMmapWrites( long handle, boolean allowMmapWrites); private native boolean allowMmapWrites(long handle); private native void setIsFdCloseOnExec( long handle, boolean isFdCloseOnExec); private native boolean isFdCloseOnExec(long handle); private native void setStatsDumpPeriodSec( long handle, int statsDumpPeriodSec); private native int statsDumpPeriodSec(long handle); private native void setAdviseRandomOnOpen( long handle, boolean adviseRandomOnOpen); private native boolean adviseRandomOnOpen(long handle); private native void setUseAdaptiveMutex( long handle, boolean useAdaptiveMutex); private native boolean useAdaptiveMutex(long handle); private native void setBytesPerSync( long handle, long bytesPerSync); private native long bytesPerSync(long handle); // CF native handles private native void optimizeForPointLookup(long handle, long blockCacheSizeMb); private native void optimizeLevelStyleCompaction(long handle, long memtableMemoryBudget); private native void optimizeUniversalStyleCompaction(long handle, long memtableMemoryBudget); private native void setComparatorHandle(long handle, int builtinComparator); private native void setComparatorHandle(long optHandle, long comparatorHandle); private native void setMergeOperatorName( long handle, String name); private native void setMergeOperator( long handle, long mergeOperatorHandle); private native void setWriteBufferSize(long handle, long writeBufferSize) throws IllegalArgumentException; private native long writeBufferSize(long handle); private native void setMaxWriteBufferNumber( long handle, int maxWriteBufferNumber); private native int maxWriteBufferNumber(long handle); private native void setMinWriteBufferNumberToMerge( long handle, int minWriteBufferNumberToMerge); private native int minWriteBufferNumberToMerge(long handle); private native void setCompressionType(long handle, byte compressionType); private native byte compressionType(long handle); private native void setCompressionPerLevel(long handle, List compressionLevels); private native List compressionPerLevel(long handle); private native void useFixedLengthPrefixExtractor( long handle, int prefixLength); private native void useCappedPrefixExtractor( long handle, int prefixLength); private native void setNumLevels( long handle, int numLevels); private native int numLevels(long handle); private native void setLevelZeroFileNumCompactionTrigger( long handle, int numFiles); private native int levelZeroFileNumCompactionTrigger(long handle); private native void setLevelZeroSlowdownWritesTrigger( long handle, int numFiles); private native int levelZeroSlowdownWritesTrigger(long handle); private native void setLevelZeroStopWritesTrigger( long handle, int numFiles); private native int levelZeroStopWritesTrigger(long handle); private native void setTargetFileSizeBase( long handle, long targetFileSizeBase); private native long targetFileSizeBase(long handle); private native void setTargetFileSizeMultiplier( long handle, int multiplier); private native int targetFileSizeMultiplier(long handle); private native void setMaxBytesForLevelBase( long handle, long maxBytesForLevelBase); private native long maxBytesForLevelBase(long handle); private native void setLevelCompactionDynamicLevelBytes( long handle, boolean enableLevelCompactionDynamicLevelBytes); private native boolean levelCompactionDynamicLevelBytes( long handle); private native void setMaxBytesForLevelMultiplier( long handle, int multiplier); private native int maxBytesForLevelMultiplier(long handle); private native void setExpandedCompactionFactor( long handle, int expandedCompactionFactor); private native int expandedCompactionFactor(long handle); private native void setSourceCompactionFactor( long handle, int sourceCompactionFactor); private native int sourceCompactionFactor(long handle); private native void setMaxGrandparentOverlapFactor( long handle, int maxGrandparentOverlapFactor); private native int maxGrandparentOverlapFactor(long handle); private native void setSoftRateLimit( long handle, double softRateLimit); private native double softRateLimit(long handle); private native void setHardRateLimit( long handle, double hardRateLimit); private native double hardRateLimit(long handle); private native void setRateLimitDelayMaxMilliseconds( long handle, int rateLimitDelayMaxMilliseconds); private native int rateLimitDelayMaxMilliseconds(long handle); private native void setArenaBlockSize( long handle, long arenaBlockSize) throws IllegalArgumentException; private native long arenaBlockSize(long handle); private native void setDisableAutoCompactions( long handle, boolean disableAutoCompactions); private native boolean disableAutoCompactions(long handle); private native void setCompactionStyle(long handle, byte compactionStyle); private native byte compactionStyle(long handle); private native void setPurgeRedundantKvsWhileFlush( long handle, boolean purgeRedundantKvsWhileFlush); private native boolean purgeRedundantKvsWhileFlush(long handle); private native void setVerifyChecksumsInCompaction( long handle, boolean verifyChecksumsInCompaction); private native boolean verifyChecksumsInCompaction(long handle); private native void setFilterDeletes( long handle, boolean filterDeletes); private native boolean filterDeletes(long handle); private native void setMaxSequentialSkipInIterations( long handle, long maxSequentialSkipInIterations); private native long maxSequentialSkipInIterations(long handle); private native void setMemTableFactory(long handle, long factoryHandle); private native String memTableFactoryName(long handle); private native void setTableFactory(long handle, long factoryHandle); private native String tableFactoryName(long handle); private native void setInplaceUpdateSupport( long handle, boolean inplaceUpdateSupport); private native boolean inplaceUpdateSupport(long handle); private native void setInplaceUpdateNumLocks( long handle, long inplaceUpdateNumLocks) throws IllegalArgumentException; private native long inplaceUpdateNumLocks(long handle); private native void setMemtablePrefixBloomBits( long handle, int memtablePrefixBloomBits); private native int memtablePrefixBloomBits(long handle); private native void setMemtablePrefixBloomProbes( long handle, int memtablePrefixBloomProbes); private native int memtablePrefixBloomProbes(long handle); private native void setBloomLocality( long handle, int bloomLocality); private native int bloomLocality(long handle); private native void setMaxSuccessiveMerges( long handle, long maxSuccessiveMerges) throws IllegalArgumentException; private native long maxSuccessiveMerges(long handle); private native void setMinPartialMergeOperands( long handle, int minPartialMergeOperands); private native int minPartialMergeOperands(long handle); private native void setOptimizeFiltersForHits(long handle, boolean optimizeFiltersForHits); private native boolean optimizeFiltersForHits(long handle); // instance variables Env env_; MemTableConfig memTableConfig_; TableFormatConfig tableFormatConfig_; RateLimiterConfig rateLimiterConfig_; AbstractComparator> comparator_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/PlainTableConfig.java000066400000000000000000000167061267530634000247670ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * The config for plain table sst format. * *

PlainTable is a RocksDB's SST file format optimized for low query * latency on pure-memory or really low-latency media.

* *

It also support prefix hash feature.

*/ public class PlainTableConfig extends TableFormatConfig { public static final int VARIABLE_LENGTH = 0; public static final int DEFAULT_BLOOM_BITS_PER_KEY = 10; public static final double DEFAULT_HASH_TABLE_RATIO = 0.75; public static final int DEFAULT_INDEX_SPARSENESS = 16; public static final int DEFAULT_HUGE_TLB_SIZE = 0; public static final EncodingType DEFAULT_ENCODING_TYPE = EncodingType.kPlain; public static final boolean DEFAULT_FULL_SCAN_MODE = false; public static final boolean DEFAULT_STORE_INDEX_IN_FILE = false; public PlainTableConfig() { keySize_ = VARIABLE_LENGTH; bloomBitsPerKey_ = DEFAULT_BLOOM_BITS_PER_KEY; hashTableRatio_ = DEFAULT_HASH_TABLE_RATIO; indexSparseness_ = DEFAULT_INDEX_SPARSENESS; hugePageTlbSize_ = DEFAULT_HUGE_TLB_SIZE; encodingType_ = DEFAULT_ENCODING_TYPE; fullScanMode_ = DEFAULT_FULL_SCAN_MODE; storeIndexInFile_ = DEFAULT_STORE_INDEX_IN_FILE; } /** *

Set the length of the user key. If it is set to be * VARIABLE_LENGTH, then it indicates the user keys are * of variable length.

* *

Otherwise,all the keys need to have the same length * in byte.

* *

DEFAULT: VARIABLE_LENGTH

* * @param keySize the length of the user key. * @return the reference to the current config. */ public PlainTableConfig setKeySize(int keySize) { keySize_ = keySize; return this; } /** * @return the specified size of the user key. If VARIABLE_LENGTH, * then it indicates variable-length key. */ public int keySize() { return keySize_; } /** * Set the number of bits per key used by the internal bloom filter * in the plain table sst format. * * @param bitsPerKey the number of bits per key for bloom filer. * @return the reference to the current config. */ public PlainTableConfig setBloomBitsPerKey(int bitsPerKey) { bloomBitsPerKey_ = bitsPerKey; return this; } /** * @return the number of bits per key used for the bloom filter. */ public int bloomBitsPerKey() { return bloomBitsPerKey_; } /** * hashTableRatio is the desired utilization of the hash table used * for prefix hashing. The ideal ratio would be the number of * prefixes / the number of hash buckets. If this value is set to * zero, then hash table will not be used. * * @param ratio the hash table ratio. * @return the reference to the current config. */ public PlainTableConfig setHashTableRatio(double ratio) { hashTableRatio_ = ratio; return this; } /** * @return the hash table ratio. */ public double hashTableRatio() { return hashTableRatio_; } /** * Index sparseness determines the index interval for keys inside the * same prefix. This number is equal to the maximum number of linear * search required after hash and binary search. If it's set to 0, * then each key will be indexed. * * @param sparseness the index sparseness. * @return the reference to the current config. */ public PlainTableConfig setIndexSparseness(int sparseness) { indexSparseness_ = sparseness; return this; } /** * @return the index sparseness. */ public long indexSparseness() { return indexSparseness_; } /** *

huge_page_tlb_size: if ≤0, allocate hash indexes and blooms * from malloc otherwise from huge page TLB.

* *

The user needs to reserve huge pages for it to be allocated, * like: {@code sysctl -w vm.nr_hugepages=20}

* *

See linux doc Documentation/vm/hugetlbpage.txt

* * @param hugePageTlbSize huge page tlb size * @return the reference to the current config. */ public PlainTableConfig setHugePageTlbSize(int hugePageTlbSize) { this.hugePageTlbSize_ = hugePageTlbSize; return this; } /** * Returns the value for huge page tlb size * * @return hugePageTlbSize */ public int hugePageTlbSize() { return hugePageTlbSize_; } /** * Sets the encoding type. * *

This setting determines how to encode * the keys. See enum {@link EncodingType} for * the choices.

* *

The value will determine how to encode keys * when writing to a new SST file. This value will be stored * inside the SST file which will be used when reading from * the file, which makes it possible for users to choose * different encoding type when reopening a DB. Files with * different encoding types can co-exist in the same DB and * can be read.

* * @param encodingType {@link org.rocksdb.EncodingType} value. * @return the reference to the current config. */ public PlainTableConfig setEncodingType(EncodingType encodingType) { this.encodingType_ = encodingType; return this; } /** * Returns the active EncodingType * * @return currently set encoding type */ public EncodingType encodingType() { return encodingType_; } /** * Set full scan mode, if true the whole file will be read * one record by one without using the index. * * @param fullScanMode boolean value indicating if full * scan mode shall be enabled. * @return the reference to the current config. */ public PlainTableConfig setFullScanMode(boolean fullScanMode) { this.fullScanMode_ = fullScanMode; return this; } /** * Return if full scan mode is active * @return boolean value indicating if the full scan mode is * enabled. */ public boolean fullScanMode() { return fullScanMode_; } /** *

If set to true: compute plain table index and bloom * filter during file building and store it in file. * When reading file, index will be mmaped instead * of doing recomputation.

* * @param storeIndexInFile value indicating if index shall * be stored in a file * @return the reference to the current config. */ public PlainTableConfig setStoreIndexInFile(boolean storeIndexInFile) { this.storeIndexInFile_ = storeIndexInFile; return this; } /** * Return a boolean value indicating if index shall be stored * in a file. * * @return currently set value for store index in file. */ public boolean storeIndexInFile() { return storeIndexInFile_; } @Override protected long newTableFactoryHandle() { return newTableFactoryHandle(keySize_, bloomBitsPerKey_, hashTableRatio_, indexSparseness_, hugePageTlbSize_, encodingType_.getValue(), fullScanMode_, storeIndexInFile_); } private native long newTableFactoryHandle( int keySize, int bloomBitsPerKey, double hashTableRatio, int indexSparseness, int hugePageTlbSize, byte encodingType, boolean fullScanMode, boolean storeIndexInFile); private int keySize_; private int bloomBitsPerKey_; private double hashTableRatio_; private int indexSparseness_; private int hugePageTlbSize_; private EncodingType encodingType_; private boolean fullScanMode_; private boolean storeIndexInFile_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RateLimiterConfig.java000066400000000000000000000015631267530634000251700ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Config for rate limiter, which is used to control write rate of flush and * compaction. */ public abstract class RateLimiterConfig { /** * This function should only be called by * {@link org.rocksdb.DBOptions#setRateLimiter(long, long)}, which will * create a c++ shared-pointer to the c++ {@code RateLimiter} that is associated * with a Java RateLimiterConfig. * * @see org.rocksdb.DBOptions#setRateLimiter(long, long) * * @return native handle address to rate limiter instance. */ abstract protected long newRateLimiterHandle(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/ReadOptions.java000066400000000000000000000122061267530634000240440ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * The class that controls the get behavior. * * Note that dispose() must be called before an Options instance * become out-of-scope to release the allocated memory in c++. */ public class ReadOptions extends RocksObject { public ReadOptions() { super(); newReadOptions(); } private native void newReadOptions(); /** * If true, all data read from underlying storage will be * verified against corresponding checksums. * Default: true * * @return true if checksum verification is on. */ public boolean verifyChecksums() { assert(isInitialized()); return verifyChecksums(nativeHandle_); } private native boolean verifyChecksums(long handle); /** * If true, all data read from underlying storage will be * verified against corresponding checksums. * Default: true * * @param verifyChecksums if true, then checksum verification * will be performed on every read. * @return the reference to the current ReadOptions. */ public ReadOptions setVerifyChecksums( final boolean verifyChecksums) { assert(isInitialized()); setVerifyChecksums(nativeHandle_, verifyChecksums); return this; } private native void setVerifyChecksums( long handle, boolean verifyChecksums); // TODO(yhchiang): this option seems to be block-based table only. // move this to a better place? /** * Fill the cache when loading the block-based sst formated db. * Callers may wish to set this field to false for bulk scans. * Default: true * * @return true if the fill-cache behavior is on. */ public boolean fillCache() { assert(isInitialized()); return fillCache(nativeHandle_); } private native boolean fillCache(long handle); /** * Fill the cache when loading the block-based sst formatted db. * Callers may wish to set this field to false for bulk scans. * Default: true * * @param fillCache if true, then fill-cache behavior will be * performed. * @return the reference to the current ReadOptions. */ public ReadOptions setFillCache(final boolean fillCache) { assert(isInitialized()); setFillCache(nativeHandle_, fillCache); return this; } private native void setFillCache( long handle, boolean fillCache); /** *

If "snapshot" is non-nullptr, read as of the supplied snapshot * (which must belong to the DB that is being read and which must * not have been released). If "snapshot" is nullptr, use an implicit * snapshot of the state at the beginning of this read operation.

*

Default: null

* * @param snapshot {@link Snapshot} instance * @return the reference to the current ReadOptions. */ public ReadOptions setSnapshot(final Snapshot snapshot) { assert(isInitialized()); if (snapshot != null) { setSnapshot(nativeHandle_, snapshot.nativeHandle_); } else { setSnapshot(nativeHandle_, 0l); } return this; } private native void setSnapshot(long handle, long snapshotHandle); /** * Returns the currently assigned Snapshot instance. * * @return the Snapshot assigned to this instance. If no Snapshot * is assigned null. */ public Snapshot snapshot() { assert(isInitialized()); long snapshotHandle = snapshot(nativeHandle_); if (snapshotHandle != 0) { return new Snapshot(snapshotHandle); } return null; } private native long snapshot(long handle); /** * Specify to create a tailing iterator -- a special iterator that has a * view of the complete database (i.e. it can also be used to read newly * added data) and is optimized for sequential reads. It will return records * that were inserted into the database after the creation of the iterator. * Default: false * * Not supported in {@code ROCKSDB_LITE} mode! * * @return true if tailing iterator is enabled. */ public boolean tailing() { assert(isInitialized()); return tailing(nativeHandle_); } private native boolean tailing(long handle); /** * Specify to create a tailing iterator -- a special iterator that has a * view of the complete database (i.e. it can also be used to read newly * added data) and is optimized for sequential reads. It will return records * that were inserted into the database after the creation of the iterator. * Default: false * Not supported in ROCKSDB_LITE mode! * * @param tailing if true, then tailing iterator will be enabled. * @return the reference to the current ReadOptions. */ public ReadOptions setTailing(final boolean tailing) { assert(isInitialized()); setTailing(nativeHandle_, tailing); return this; } private native void setTailing( long handle, boolean tailing); @Override protected void disposeInternal() { disposeInternal(nativeHandle_); } private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RemoveEmptyValueCompactionFilter.java000066400000000000000000000012331267530634000302470ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Just a Java wrapper around EmptyValueCompactionFilter implemented in C++ */ public class RemoveEmptyValueCompactionFilter extends AbstractCompactionFilter { public RemoveEmptyValueCompactionFilter() { super(); createNewRemoveEmptyValueCompactionFilter0(); } private native void createNewRemoveEmptyValueCompactionFilter0(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RestoreBackupableDB.java000066400000000000000000000130061267530634000254170ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; /** *

This class is used to access information about backups and * restore from them.

* *

Note: {@code dispose()} must be called before this instance * become out-of-scope to release the allocated * memory in c++.

* */ public class RestoreBackupableDB extends RocksObject { /** *

Construct new estoreBackupableDB instance.

* * @param options {@link org.rocksdb.BackupableDBOptions} instance */ public RestoreBackupableDB(final BackupableDBOptions options) { super(); nativeHandle_ = newRestoreBackupableDB(options.nativeHandle_); } /** *

Restore from backup with backup_id.

* *

Important: If options_.share_table_files == true * and you restore DB from some backup that is not the latest, and you * start creating new backups from the new DB, they will probably * fail.

* *

Example: Let's say you have backups 1, 2, 3, 4, 5 * and you restore 3. If you add new data to the DB and try creating a new * backup now, the database will diverge from backups 4 and 5 and the new * backup will fail. If you want to create new backup, you will first have * to delete backups 4 and 5.

* * @param backupId id pointing to backup * @param dbDir database directory to restore to * @param walDir directory where wal files are located * @param restoreOptions {@link org.rocksdb.RestoreOptions} instance. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void restoreDBFromBackup(final long backupId, final String dbDir, final String walDir, final RestoreOptions restoreOptions) throws RocksDBException { assert(isInitialized()); restoreDBFromBackup0(nativeHandle_, backupId, dbDir, walDir, restoreOptions.nativeHandle_); } /** *

Restore from the latest backup.

* * @param dbDir database directory to restore to * @param walDir directory where wal files are located * @param restoreOptions {@link org.rocksdb.RestoreOptions} instance * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void restoreDBFromLatestBackup(final String dbDir, final String walDir, final RestoreOptions restoreOptions) throws RocksDBException { assert(isInitialized()); restoreDBFromLatestBackup0(nativeHandle_, dbDir, walDir, restoreOptions.nativeHandle_); } /** *

Deletes old backups, keeping latest numBackupsToKeep alive.

* * @param numBackupsToKeep of latest backups to keep * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void purgeOldBackups(final int numBackupsToKeep) throws RocksDBException { assert(isInitialized()); purgeOldBackups0(nativeHandle_, numBackupsToKeep); } /** *

Deletes a specific backup.

* * @param backupId of backup to delete. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void deleteBackup(final int backupId) throws RocksDBException { assert(isInitialized()); deleteBackup0(nativeHandle_, backupId); } /** *

Returns a list of {@link BackupInfo} instances, which describe * already made backups.

* * @return List of {@link BackupInfo} instances. */ public List getBackupInfos() { assert(isInitialized()); return getBackupInfo(nativeHandle_); } /** *

Returns a list of corrupted backup ids. If there * is no corrupted backup the method will return an * empty list.

* * @return array of backup ids as int ids. */ public int[] getCorruptedBackups() { assert(isInitialized()); return getCorruptedBackups(nativeHandle_); } /** *

Will delete all the files we don't need anymore. It will * do the full scan of the files/ directory and delete all the * files that are not referenced.

* * @throws RocksDBException thrown if error happens in underlying * native library. */ public void garbageCollect() throws RocksDBException { assert(isInitialized()); garbageCollect(nativeHandle_); } /** *

Release the memory allocated for the current instance * in the c++ side.

*/ @Override public synchronized void disposeInternal() { dispose(nativeHandle_); } private native long newRestoreBackupableDB(long options); private native void restoreDBFromBackup0(long nativeHandle, long backupId, String dbDir, String walDir, long restoreOptions) throws RocksDBException; private native void restoreDBFromLatestBackup0(long nativeHandle, String dbDir, String walDir, long restoreOptions) throws RocksDBException; private native void purgeOldBackups0(long nativeHandle, int numBackupsToKeep) throws RocksDBException; private native void deleteBackup0(long nativeHandle, int backupId) throws RocksDBException; private native List getBackupInfo(long handle); private native int[] getCorruptedBackups(long handle); private native void garbageCollect(long handle) throws RocksDBException; private native void dispose(long nativeHandle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RestoreOptions.java000066400000000000000000000025201267530634000246120ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * RestoreOptions to control the behavior of restore. * * Note that dispose() must be called before this instance become out-of-scope * to release the allocated memory in c++. * */ public class RestoreOptions extends RocksObject { /** * Constructor * * @param keepLogFiles If true, restore won't overwrite the existing log files in wal_dir. It * will also move all log files from archive directory to wal_dir. Use this * option in combination with BackupableDBOptions::backup_log_files = false * for persisting in-memory databases. * Default: false */ public RestoreOptions(final boolean keepLogFiles) { super(); nativeHandle_ = newRestoreOptions(keepLogFiles); } /** * Release the memory allocated for the current instance * in the c++ side. */ @Override public synchronized void disposeInternal() { assert(isInitialized()); dispose(nativeHandle_); } private native long newRestoreOptions(boolean keepLogFiles); private native void dispose(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksDB.java000066400000000000000000002115041267530634000231060ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.*; import java.io.IOException; import org.rocksdb.util.Environment; /** * A RocksDB is a persistent ordered map from keys to values. It is safe for * concurrent access from multiple threads without any external synchronization. * All methods of this class could potentially throw RocksDBException, which * indicates sth wrong at the RocksDB library side and the call failed. */ public class RocksDB extends RocksObject { public static final byte[] DEFAULT_COLUMN_FAMILY = "default".getBytes(); public static final int NOT_FOUND = -1; static { RocksDB.loadLibrary(); } /** * Loads the necessary library files. * Calling this method twice will have no effect. * By default the method extracts the shared library for loading at * java.io.tmpdir, however, you can override this temporary location by * setting the environment variable ROCKSDB_SHAREDLIB_DIR. */ public static synchronized void loadLibrary() { String tmpDir = System.getenv("ROCKSDB_SHAREDLIB_DIR"); // loading possibly necessary libraries. for (CompressionType compressionType : CompressionType.values()) { try { if (compressionType.getLibraryName() != null) { System.loadLibrary(compressionType.getLibraryName()); } } catch (UnsatisfiedLinkError e) { // since it may be optional, we ignore its loading failure here. } } try { NativeLibraryLoader.getInstance().loadLibrary(tmpDir); } catch (IOException e) { throw new RuntimeException("Unable to load the RocksDB shared library" + e); } } /** * Tries to load the necessary library files from the given list of * directories. * * @param paths a list of strings where each describes a directory * of a library. */ public static synchronized void loadLibrary(final List paths) { for (CompressionType compressionType : CompressionType.values()) { if (compressionType.equals(CompressionType.NO_COMPRESSION)) { continue; } for (String path : paths) { try { System.load(path + "/" + Environment.getSharedLibraryFileName( compressionType.getLibraryName())); break; } catch (UnsatisfiedLinkError e) { // since they are optional, we ignore loading fails. } } } boolean success = false; UnsatisfiedLinkError err = null; for (String path : paths) { try { System.load(path + "/" + Environment.getJniLibraryFileName("rocksdbjni")); success = true; break; } catch (UnsatisfiedLinkError e) { err = e; } } if (!success) { throw err; } } /** * The factory constructor of RocksDB that opens a RocksDB instance given * the path to the database using the default options w/ createIfMissing * set to true. * * @param path the path to the rocksdb. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. * @see Options#setCreateIfMissing(boolean) */ public static RocksDB open(final String path) throws RocksDBException { // This allows to use the rocksjni default Options instead of // the c++ one. Options options = new Options(); options.setCreateIfMissing(true); return open(options, path); } /** * The factory constructor of RocksDB that opens a RocksDB instance given * the path to the database using the specified options and db path and a list * of column family names. *

* If opened in read write mode every existing column family name must be passed * within the list to this method.

*

* If opened in read-only mode only a subset of existing column families must * be passed to this method.

*

* Options instance *should* not be disposed before all DBs using this options * instance have been closed. If user doesn't call options dispose explicitly, * then this options instance will be GC'd automatically

*

* ColumnFamily handles are disposed when the RocksDB instance is disposed. *

* * @param path the path to the rocksdb. * @param columnFamilyDescriptors list of column family descriptors * @param columnFamilyHandles will be filled with ColumnFamilyHandle instances * on open. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. * @see DBOptions#setCreateIfMissing(boolean) */ public static RocksDB open(final String path, final List columnFamilyDescriptors, final List columnFamilyHandles) throws RocksDBException { // This allows to use the rocksjni default Options instead of // the c++ one. DBOptions options = new DBOptions(); return open(options, path, columnFamilyDescriptors, columnFamilyHandles); } /** * The factory constructor of RocksDB that opens a RocksDB instance given * the path to the database using the specified options and db path. * *

* Options instance *should* not be disposed before all DBs using this options * instance have been closed. If user doesn't call options dispose explicitly, * then this options instance will be GC'd automatically.

*

* Options instance can be re-used to open multiple DBs if DB statistics is * not used. If DB statistics are required, then its recommended to open DB * with new Options instance as underlying native statistics instance does not * use any locks to prevent concurrent updates.

* * @param options {@link org.rocksdb.Options} instance. * @param path the path to the rocksdb. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. * * @see Options#setCreateIfMissing(boolean) */ public static RocksDB open(final Options options, final String path) throws RocksDBException { // when non-default Options is used, keeping an Options reference // in RocksDB can prevent Java to GC during the life-time of // the currently-created RocksDB. RocksDB db = new RocksDB(); db.open(options.nativeHandle_, path); db.storeOptionsInstance(options); return db; } /** * The factory constructor of RocksDB that opens a RocksDB instance given * the path to the database using the specified options and db path and a list * of column family names. *

* If opened in read write mode every existing column family name must be passed * within the list to this method.

*

* If opened in read-only mode only a subset of existing column families must * be passed to this method.

*

* Options instance *should* not be disposed before all DBs using this options * instance have been closed. If user doesn't call options dispose explicitly, * then this options instance will be GC'd automatically.

*

* Options instance can be re-used to open multiple DBs if DB statistics is * not used. If DB statistics are required, then its recommended to open DB * with new Options instance as underlying native statistics instance does not * use any locks to prevent concurrent updates.

*

* ColumnFamily handles are disposed when the RocksDB instance is disposed.

* * @param options {@link org.rocksdb.DBOptions} instance. * @param path the path to the rocksdb. * @param columnFamilyDescriptors list of column family descriptors * @param columnFamilyHandles will be filled with ColumnFamilyHandle instances * on open. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. * * @see DBOptions#setCreateIfMissing(boolean) */ public static RocksDB open(final DBOptions options, final String path, final List columnFamilyDescriptors, final List columnFamilyHandles) throws RocksDBException { RocksDB db = new RocksDB(); List cfReferences = db.open(options.nativeHandle_, path, columnFamilyDescriptors, columnFamilyDescriptors.size()); for (int i = 0; i < columnFamilyDescriptors.size(); i++) { columnFamilyHandles.add(new ColumnFamilyHandle(db, cfReferences.get(i))); } db.storeOptionsInstance(options); return db; } /** * The factory constructor of RocksDB that opens a RocksDB instance in * Read-Only mode given the path to the database using the default * options. * * @param path the path to the RocksDB. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public static RocksDB openReadOnly(final String path) throws RocksDBException { // This allows to use the rocksjni default Options instead of // the c++ one. Options options = new Options(); return openReadOnly(options, path); } /** * The factory constructor of RocksDB that opens a RocksDB instance in * Read-Only mode given the path to the database using the default * options. * * @param path the path to the RocksDB. * @param columnFamilyDescriptors list of column family descriptors * @param columnFamilyHandles will be filled with ColumnFamilyHandle instances * on open. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public static RocksDB openReadOnly(final String path, final List columnFamilyDescriptors, final List columnFamilyHandles) throws RocksDBException { // This allows to use the rocksjni default Options instead of // the c++ one. DBOptions options = new DBOptions(); return openReadOnly(options, path, columnFamilyDescriptors, columnFamilyHandles); } /** * The factory constructor of RocksDB that opens a RocksDB instance in * Read-Only mode given the path to the database using the specified * options and db path. * * Options instance *should* not be disposed before all DBs using this options * instance have been closed. If user doesn't call options dispose explicitly, * then this options instance will be GC'd automatically. * * @param options {@link Options} instance. * @param path the path to the RocksDB. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public static RocksDB openReadOnly(final Options options, final String path) throws RocksDBException { // when non-default Options is used, keeping an Options reference // in RocksDB can prevent Java to GC during the life-time of // the currently-created RocksDB. RocksDB db = new RocksDB(); db.openROnly(options.nativeHandle_, path); db.storeOptionsInstance(options); return db; } /** * The factory constructor of RocksDB that opens a RocksDB instance in * Read-Only mode given the path to the database using the specified * options and db path. * *

This open method allows to open RocksDB using a subset of available * column families

*

Options instance *should* not be disposed before all DBs using this * options instance have been closed. If user doesn't call options dispose * explicitly,then this options instance will be GC'd automatically.

* * @param options {@link DBOptions} instance. * @param path the path to the RocksDB. * @param columnFamilyDescriptors list of column family descriptors * @param columnFamilyHandles will be filled with ColumnFamilyHandle instances * on open. * @return a {@link RocksDB} instance on success, null if the specified * {@link RocksDB} can not be opened. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public static RocksDB openReadOnly(final DBOptions options, final String path, final List columnFamilyDescriptors, final List columnFamilyHandles) throws RocksDBException { // when non-default Options is used, keeping an Options reference // in RocksDB can prevent Java to GC during the life-time of // the currently-created RocksDB. RocksDB db = new RocksDB(); List cfReferences = db.openROnly(options.nativeHandle_, path, columnFamilyDescriptors, columnFamilyDescriptors.size()); for (int i=0; i listColumnFamilies(final Options options, final String path) throws RocksDBException { return RocksDB.listColumnFamilies(options.nativeHandle_, path); } private void storeOptionsInstance(DBOptionsInterface options) { options_ = options; } @Override protected void disposeInternal() { synchronized (this) { assert (isInitialized()); disposeInternal(nativeHandle_); } } /** * Close the RocksDB instance. * This function is equivalent to dispose(). */ public void close() { dispose(); } /** * Set the database entry for "key" to "value". * * @param key the specified key to be inserted. * @param value the value associated with the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void put(final byte[] key, final byte[] value) throws RocksDBException { put(nativeHandle_, key, key.length, value, value.length); } /** * Set the database entry for "key" to "value" in the specified * column family. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param key the specified key to be inserted. * @param value the value associated with the specified key. * * throws IllegalArgumentException if column family is not present * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void put(final ColumnFamilyHandle columnFamilyHandle, final byte[] key, final byte[] value) throws RocksDBException { put(nativeHandle_, key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } /** * Set the database entry for "key" to "value". * * @param writeOpts {@link org.rocksdb.WriteOptions} instance. * @param key the specified key to be inserted. * @param value the value associated with the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void put(final WriteOptions writeOpts, final byte[] key, final byte[] value) throws RocksDBException { put(nativeHandle_, writeOpts.nativeHandle_, key, key.length, value, value.length); } /** * Set the database entry for "key" to "value" for the specified * column family. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param writeOpts {@link org.rocksdb.WriteOptions} instance. * @param key the specified key to be inserted. * @param value the value associated with the specified key. * * throws IllegalArgumentException if column family is not present * * @throws RocksDBException thrown if error happens in underlying * native library. * @see IllegalArgumentException */ public void put(final ColumnFamilyHandle columnFamilyHandle, final WriteOptions writeOpts, final byte[] key, final byte[] value) throws RocksDBException { put(nativeHandle_, writeOpts.nativeHandle_, key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } /** * If the key definitely does not exist in the database, then this method * returns false, else true. * * This check is potentially lighter-weight than invoking DB::Get(). One way * to make this lighter weight is to avoid doing any IOs. * * @param key byte array of a key to search for * @param value StringBuffer instance which is a out parameter if a value is * found in block-cache. * @return boolean value indicating if key does not exist or might exist. */ public boolean keyMayExist(final byte[] key, final StringBuffer value){ return keyMayExist(key, key.length, value); } /** * If the key definitely does not exist in the database, then this method * returns false, else true. * * This check is potentially lighter-weight than invoking DB::Get(). One way * to make this lighter weight is to avoid doing any IOs. * * @param columnFamilyHandle {@link ColumnFamilyHandle} instance * @param key byte array of a key to search for * @param value StringBuffer instance which is a out parameter if a value is * found in block-cache. * @return boolean value indicating if key does not exist or might exist. */ public boolean keyMayExist(final ColumnFamilyHandle columnFamilyHandle, final byte[] key, final StringBuffer value){ return keyMayExist(key, key.length, columnFamilyHandle.nativeHandle_, value); } /** * If the key definitely does not exist in the database, then this method * returns false, else true. * * This check is potentially lighter-weight than invoking DB::Get(). One way * to make this lighter weight is to avoid doing any IOs. * * @param readOptions {@link ReadOptions} instance * @param key byte array of a key to search for * @param value StringBuffer instance which is a out parameter if a value is * found in block-cache. * @return boolean value indicating if key does not exist or might exist. */ public boolean keyMayExist(final ReadOptions readOptions, final byte[] key, final StringBuffer value){ return keyMayExist(readOptions.nativeHandle_, key, key.length, value); } /** * If the key definitely does not exist in the database, then this method * returns false, else true. * * This check is potentially lighter-weight than invoking DB::Get(). One way * to make this lighter weight is to avoid doing any IOs. * * @param readOptions {@link ReadOptions} instance * @param columnFamilyHandle {@link ColumnFamilyHandle} instance * @param key byte array of a key to search for * @param value StringBuffer instance which is a out parameter if a value is * found in block-cache. * @return boolean value indicating if key does not exist or might exist. */ public boolean keyMayExist(final ReadOptions readOptions, final ColumnFamilyHandle columnFamilyHandle, final byte[] key, final StringBuffer value){ return keyMayExist(readOptions.nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_, value); } /** * Apply the specified updates to the database. * * @param writeOpts WriteOptions instance * @param updates WriteBatch instance * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void write(final WriteOptions writeOpts, final WriteBatch updates) throws RocksDBException { write0(writeOpts.nativeHandle_, updates.nativeHandle_); } /** * Apply the specified updates to the database. * * @param writeOpts WriteOptions instance * @param updates WriteBatchWithIndex instance * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void write(final WriteOptions writeOpts, final WriteBatchWithIndex updates) throws RocksDBException { write1(writeOpts.nativeHandle_, updates.nativeHandle_); } /** * Add merge operand for key/value pair. * * @param key the specified key to be merged. * @param value the value to be merged with the current value for * the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void merge(final byte[] key, final byte[] value) throws RocksDBException { merge(nativeHandle_, key, key.length, value, value.length); } /** * Add merge operand for key/value pair in a ColumnFamily. * * @param columnFamilyHandle {@link ColumnFamilyHandle} instance * @param key the specified key to be merged. * @param value the value to be merged with the current value for * the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void merge(final ColumnFamilyHandle columnFamilyHandle, final byte[] key, final byte[] value) throws RocksDBException { merge(nativeHandle_, key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } /** * Add merge operand for key/value pair. * * @param writeOpts {@link WriteOptions} for this write. * @param key the specified key to be merged. * @param value the value to be merged with the current value for * the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void merge(final WriteOptions writeOpts, final byte[] key, final byte[] value) throws RocksDBException { merge(nativeHandle_, writeOpts.nativeHandle_, key, key.length, value, value.length); } /** * Add merge operand for key/value pair. * * @param columnFamilyHandle {@link ColumnFamilyHandle} instance * @param writeOpts {@link WriteOptions} for this write. * @param key the specified key to be merged. * @param value the value to be merged with the current value for * the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void merge(final ColumnFamilyHandle columnFamilyHandle, final WriteOptions writeOpts, final byte[] key, final byte[] value) throws RocksDBException { merge(nativeHandle_, writeOpts.nativeHandle_, key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } /** * Get the value associated with the specified key within column family* * @param key the key to retrieve the value. * @param value the out-value to receive the retrieved value. * @return The size of the actual value that matches the specified * {@code key} in byte. If the return value is greater than the * length of {@code value}, then it indicates that the size of the * input buffer {@code value} is insufficient and partial result will * be returned. RocksDB.NOT_FOUND will be returned if the value not * found. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public int get(final byte[] key, final byte[] value) throws RocksDBException { return get(nativeHandle_, key, key.length, value, value.length); } /** * Get the value associated with the specified key within column family. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param key the key to retrieve the value. * @param value the out-value to receive the retrieved value. * @return The size of the actual value that matches the specified * {@code key} in byte. If the return value is greater than the * length of {@code value}, then it indicates that the size of the * input buffer {@code value} is insufficient and partial result will * be returned. RocksDB.NOT_FOUND will be returned if the value not * found. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public int get(final ColumnFamilyHandle columnFamilyHandle, final byte[] key, final byte[] value) throws RocksDBException, IllegalArgumentException { return get(nativeHandle_, key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } /** * Get the value associated with the specified key. * * @param opt {@link org.rocksdb.ReadOptions} instance. * @param key the key to retrieve the value. * @param value the out-value to receive the retrieved value. * @return The size of the actual value that matches the specified * {@code key} in byte. If the return value is greater than the * length of {@code value}, then it indicates that the size of the * input buffer {@code value} is insufficient and partial result will * be returned. RocksDB.NOT_FOUND will be returned if the value not * found. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public int get(final ReadOptions opt, final byte[] key, final byte[] value) throws RocksDBException { return get(nativeHandle_, opt.nativeHandle_, key, key.length, value, value.length); } /** * Get the value associated with the specified key within column family. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param opt {@link org.rocksdb.ReadOptions} instance. * @param key the key to retrieve the value. * @param value the out-value to receive the retrieved value. * @return The size of the actual value that matches the specified * {@code key} in byte. If the return value is greater than the * length of {@code value}, then it indicates that the size of the * input buffer {@code value} is insufficient and partial result will * be returned. RocksDB.NOT_FOUND will be returned if the value not * found. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public int get(final ColumnFamilyHandle columnFamilyHandle, final ReadOptions opt, final byte[] key, final byte[] value) throws RocksDBException { return get(nativeHandle_, opt.nativeHandle_, key, key.length, value, value.length, columnFamilyHandle.nativeHandle_); } /** * The simplified version of get which returns a new byte array storing * the value associated with the specified input key if any. null will be * returned if the specified key is not found. * * @param key the key retrieve the value. * @return a byte array storing the value associated with the input key if * any. null if it does not find the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public byte[] get(final byte[] key) throws RocksDBException { return get(nativeHandle_, key, key.length); } /** * The simplified version of get which returns a new byte array storing * the value associated with the specified input key if any. null will be * returned if the specified key is not found. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param key the key retrieve the value. * @return a byte array storing the value associated with the input key if * any. null if it does not find the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public byte[] get(final ColumnFamilyHandle columnFamilyHandle, final byte[] key) throws RocksDBException { return get(nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_); } /** * The simplified version of get which returns a new byte array storing * the value associated with the specified input key if any. null will be * returned if the specified key is not found. * * @param key the key retrieve the value. * @param opt Read options. * @return a byte array storing the value associated with the input key if * any. null if it does not find the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public byte[] get(final ReadOptions opt, final byte[] key) throws RocksDBException { return get(nativeHandle_, opt.nativeHandle_, key, key.length); } /** * The simplified version of get which returns a new byte array storing * the value associated with the specified input key if any. null will be * returned if the specified key is not found. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param key the key retrieve the value. * @param opt Read options. * @return a byte array storing the value associated with the input key if * any. null if it does not find the specified key. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public byte[] get(final ColumnFamilyHandle columnFamilyHandle, final ReadOptions opt, final byte[] key) throws RocksDBException { return get(nativeHandle_, opt.nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_); } /** * Returns a map of keys for which values were found in DB. * * @param keys List of keys for which values need to be retrieved. * @return Map where key of map is the key passed by user and value for map * entry is the corresponding value in DB. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public Map multiGet(final List keys) throws RocksDBException { assert(keys.size() != 0); List values = multiGet( nativeHandle_, keys, keys.size()); Map keyValueMap = new HashMap<>(); for(int i = 0; i < values.size(); i++) { if(values.get(i) == null) { continue; } keyValueMap.put(keys.get(i), values.get(i)); } return keyValueMap; } /** * Returns a map of keys for which values were found in DB. *

* Note: Every key needs to have a related column family name in * {@code columnFamilyHandleList}. *

* * @param columnFamilyHandleList {@link java.util.List} containing * {@link org.rocksdb.ColumnFamilyHandle} instances. * @param keys List of keys for which values need to be retrieved. * @return Map where key of map is the key passed by user and value for map * entry is the corresponding value in DB. * * @throws RocksDBException thrown if error happens in underlying * native library. * @throws IllegalArgumentException thrown if the size of passed keys is not * equal to the amount of passed column family handles. */ public Map multiGet(final List columnFamilyHandleList, final List keys) throws RocksDBException, IllegalArgumentException { assert(keys.size() != 0); // Check if key size equals cfList size. If not a exception must be // thrown. If not a Segmentation fault happens. if (keys.size()!=columnFamilyHandleList.size()) { throw new IllegalArgumentException( "For each key there must be a ColumnFamilyHandle."); } List values = multiGet(nativeHandle_, keys, keys.size(), columnFamilyHandleList); Map keyValueMap = new HashMap<>(); for(int i = 0; i < values.size(); i++) { if (values.get(i) == null) { continue; } keyValueMap.put(keys.get(i), values.get(i)); } return keyValueMap; } /** * Returns a map of keys for which values were found in DB. * * @param opt Read options. * @param keys of keys for which values need to be retrieved. * @return Map where key of map is the key passed by user and value for map * entry is the corresponding value in DB. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public Map multiGet(final ReadOptions opt, final List keys) throws RocksDBException { assert(keys.size() != 0); List values = multiGet( nativeHandle_, opt.nativeHandle_, keys, keys.size()); Map keyValueMap = new HashMap<>(); for(int i = 0; i < values.size(); i++) { if(values.get(i) == null) { continue; } keyValueMap.put(keys.get(i), values.get(i)); } return keyValueMap; } /** * Returns a map of keys for which values were found in DB. *

* Note: Every key needs to have a related column family name in * {@code columnFamilyHandleList}. *

* * @param opt Read options. * @param columnFamilyHandleList {@link java.util.List} containing * {@link org.rocksdb.ColumnFamilyHandle} instances. * @param keys of keys for which values need to be retrieved. * @return Map where key of map is the key passed by user and value for map * entry is the corresponding value in DB. * * @throws RocksDBException thrown if error happens in underlying * native library. * @throws IllegalArgumentException thrown if the size of passed keys is not * equal to the amount of passed column family handles. */ public Map multiGet(final ReadOptions opt, final List columnFamilyHandleList, final List keys) throws RocksDBException { assert(keys.size() != 0); // Check if key size equals cfList size. If not a exception must be // thrown. If not a Segmentation fault happens. if (keys.size()!=columnFamilyHandleList.size()){ throw new IllegalArgumentException( "For each key there must be a ColumnFamilyHandle."); } List values = multiGet(nativeHandle_, opt.nativeHandle_, keys, keys.size(), columnFamilyHandleList); Map keyValueMap = new HashMap<>(); for(int i = 0; i < values.size(); i++) { if(values.get(i) == null) { continue; } keyValueMap.put(keys.get(i), values.get(i)); } return keyValueMap; } /** * Remove the database entry (if any) for "key". Returns OK on * success, and a non-OK status on error. It is not an error if "key" * did not exist in the database. * * @param key Key to delete within database * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void remove(final byte[] key) throws RocksDBException { remove(nativeHandle_, key, key.length); } /** * Remove the database entry (if any) for "key". Returns OK on * success, and a non-OK status on error. It is not an error if "key" * did not exist in the database. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param key Key to delete within database * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void remove(final ColumnFamilyHandle columnFamilyHandle, final byte[] key) throws RocksDBException { remove(nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_); } /** * Remove the database entry (if any) for "key". Returns OK on * success, and a non-OK status on error. It is not an error if "key" * did not exist in the database. * * @param writeOpt WriteOptions to be used with delete operation * @param key Key to delete within database * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void remove(final WriteOptions writeOpt, final byte[] key) throws RocksDBException { remove(nativeHandle_, writeOpt.nativeHandle_, key, key.length); } /** * Remove the database entry (if any) for "key". Returns OK on * success, and a non-OK status on error. It is not an error if "key" * did not exist in the database. * * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param writeOpt WriteOptions to be used with delete operation * @param key Key to delete within database * * @throws RocksDBException thrown if error happens in underlying * native library. */ public void remove(final ColumnFamilyHandle columnFamilyHandle, final WriteOptions writeOpt, final byte[] key) throws RocksDBException { remove(nativeHandle_, writeOpt.nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_); } /** * DB implements can export properties about their state * via this method on a per column family level. * *

If {@code property} is a valid property understood by this DB * implementation, fills {@code value} with its current value and * returns true. Otherwise returns false.

* *

Valid property names include: *

    *
  • "rocksdb.num-files-at-level<N>" - return the number of files at level <N>, * where <N> is an ASCII representation of a level number (e.g. "0").
    • *
  • "rocksdb.stats" - returns a multi-line string that describes statistics * about the internal operation of the DB.
    • *
  • "rocksdb.sstables" - returns a multi-line string that describes all * of the sstables that make up the db contents.
    • *
* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param property to be fetched. See above for examples * @return property value * * @throws RocksDBException thrown if error happens in underlying * native library. */ public String getProperty(final ColumnFamilyHandle columnFamilyHandle, final String property) throws RocksDBException { return getProperty0(nativeHandle_, columnFamilyHandle.nativeHandle_, property, property.length()); } /** * DB implementations can export properties about their state * via this method. If "property" is a valid property understood by this * DB implementation, fills "*value" with its current value and returns * true. Otherwise returns false. * *

Valid property names include: *

    *
  • "rocksdb.num-files-at-level<N>" - return the number of files at level <N>, * where <N> is an ASCII representation of a level number (e.g. "0").
    • *
  • "rocksdb.stats" - returns a multi-line string that describes statistics * about the internal operation of the DB.
    • *
  • "rocksdb.sstables" - returns a multi-line string that describes all * of the sstables that make up the db contents.
    • *
* * @param property to be fetched. See above for examples * @return property value * * @throws RocksDBException thrown if error happens in underlying * native library. */ public String getProperty(final String property) throws RocksDBException { return getProperty0(nativeHandle_, property, property.length()); } /** *

Similar to GetProperty(), but only works for a subset of properties whose * return value is a numerical value. Return the value as long.

* *

Note: As the returned property is of type * {@code uint64_t} on C++ side the returning value can be negative * because Java supports in Java 7 only signed long values.

* *

Java 7: To mitigate the problem of the non * existent unsigned long tpye, values should be encapsulated using * {@link java.math.BigInteger} to reflect the correct value. The correct * behavior is guaranteed if {@code 2^64} is added to negative values.

* *

Java 8: In Java 8 the value should be treated as * unsigned long using provided methods of type {@link Long}.

* * @param property to be fetched. * * @return numerical property value. * * @throws RocksDBException if an error happens in the underlying native code. */ public long getLongProperty(final String property) throws RocksDBException { return getLongProperty(nativeHandle_, property, property.length()); } /** *

Similar to GetProperty(), but only works for a subset of properties whose * return value is a numerical value. Return the value as long.

* *

Note: As the returned property is of type * {@code uint64_t} on C++ side the returning value can be negative * because Java supports in Java 7 only signed long values.

* *

Java 7: To mitigate the problem of the non * existent unsigned long tpye, values should be encapsulated using * {@link java.math.BigInteger} to reflect the correct value. The correct * behavior is guaranteed if {@code 2^64} is added to negative values.

* *

Java 8: In Java 8 the value should be treated as * unsigned long using provided methods of type {@link Long}.

* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param property to be fetched. * * @return numerical property value * * @throws RocksDBException if an error happens in the underlying native code. */ public long getLongProperty(final ColumnFamilyHandle columnFamilyHandle, final String property) throws RocksDBException { return getLongProperty(nativeHandle_, columnFamilyHandle.nativeHandle_, property, property.length()); } /** *

Return a heap-allocated iterator over the contents of the * database. The result of newIterator() is initially invalid * (caller must call one of the Seek methods on the iterator * before using it).

* *

Caller should close the iterator when it is no longer needed. * The returned iterator should be closed before this db is closed. *

* * @return instance of iterator object. */ public RocksIterator newIterator() { return new RocksIterator(this, iterator(nativeHandle_)); } /** *

Return a heap-allocated iterator over the contents of the * database. The result of newIterator() is initially invalid * (caller must call one of the Seek methods on the iterator * before using it).

* *

Caller should close the iterator when it is no longer needed. * The returned iterator should be closed before this db is closed. *

* * @param readOptions {@link ReadOptions} instance. * @return instance of iterator object. */ public RocksIterator newIterator(final ReadOptions readOptions) { return new RocksIterator(this, iterator(nativeHandle_, readOptions.nativeHandle_)); } /** *

Return a handle to the current DB state. Iterators created with * this handle will all observe a stable snapshot of the current DB * state. The caller must call ReleaseSnapshot(result) when the * snapshot is no longer needed.

* *

nullptr will be returned if the DB fails to take a snapshot or does * not support snapshot.

* * @return Snapshot {@link Snapshot} instance */ public Snapshot getSnapshot() { long snapshotHandle = getSnapshot(nativeHandle_); if (snapshotHandle != 0) { return new Snapshot(snapshotHandle); } return null; } /** * Release a previously acquired snapshot. The caller must not * use "snapshot" after this call. * * @param snapshot {@link Snapshot} instance */ public void releaseSnapshot(final Snapshot snapshot) { if (snapshot != null) { releaseSnapshot(nativeHandle_, snapshot.nativeHandle_); } } /** *

Return a heap-allocated iterator over the contents of the * database. The result of newIterator() is initially invalid * (caller must call one of the Seek methods on the iterator * before using it).

* *

Caller should close the iterator when it is no longer needed. * The returned iterator should be closed before this db is closed. *

* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @return instance of iterator object. */ public RocksIterator newIterator(final ColumnFamilyHandle columnFamilyHandle) { return new RocksIterator(this, iteratorCF(nativeHandle_, columnFamilyHandle.nativeHandle_)); } /** *

Return a heap-allocated iterator over the contents of the * database. The result of newIterator() is initially invalid * (caller must call one of the Seek methods on the iterator * before using it).

* *

Caller should close the iterator when it is no longer needed. * The returned iterator should be closed before this db is closed. *

* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param readOptions {@link ReadOptions} instance. * @return instance of iterator object. */ public RocksIterator newIterator(final ColumnFamilyHandle columnFamilyHandle, final ReadOptions readOptions) { return new RocksIterator(this, iteratorCF(nativeHandle_, columnFamilyHandle.nativeHandle_, readOptions.nativeHandle_)); } /** * Returns iterators from a consistent database state across multiple * column families. Iterators are heap allocated and need to be deleted * before the db is deleted * * @param columnFamilyHandleList {@link java.util.List} containing * {@link org.rocksdb.ColumnFamilyHandle} instances. * @return {@link java.util.List} containing {@link org.rocksdb.RocksIterator} * instances * * @throws RocksDBException thrown if error happens in underlying * native library. */ public List newIterators( final List columnFamilyHandleList) throws RocksDBException { return newIterators(columnFamilyHandleList, new ReadOptions()); } /** * Returns iterators from a consistent database state across multiple * column families. Iterators are heap allocated and need to be deleted * before the db is deleted * * @param columnFamilyHandleList {@link java.util.List} containing * {@link org.rocksdb.ColumnFamilyHandle} instances. * @param readOptions {@link ReadOptions} instance. * @return {@link java.util.List} containing {@link org.rocksdb.RocksIterator} * instances * * @throws RocksDBException thrown if error happens in underlying * native library. */ public List newIterators( final List columnFamilyHandleList, final ReadOptions readOptions) throws RocksDBException { List iterators = new ArrayList<>(columnFamilyHandleList.size()); long[] iteratorRefs = iterators(nativeHandle_, columnFamilyHandleList, readOptions.nativeHandle_); for (int i=0; iFlush all memory table data.

* *

Note: it must be ensured that the FlushOptions instance * is not GC'ed before this method finishes. If the wait parameter is * set to false, flush processing is asynchronous.

* * @param flushOptions {@link org.rocksdb.FlushOptions} instance. * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void flush(final FlushOptions flushOptions) throws RocksDBException { flush(nativeHandle_, flushOptions.nativeHandle_); } /** *

Flush all memory table data.

* *

Note: it must be ensured that the FlushOptions instance * is not GC'ed before this method finishes. If the wait parameter is * set to false, flush processing is asynchronous.

* * @param flushOptions {@link org.rocksdb.FlushOptions} instance. * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} instance. * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void flush(final FlushOptions flushOptions, final ColumnFamilyHandle columnFamilyHandle) throws RocksDBException { flush(nativeHandle_, flushOptions.nativeHandle_, columnFamilyHandle.nativeHandle_); } /** *

Range compaction of database.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

See also

*
    *
  • {@link #compactRange(boolean, int, int)}
    • *
  • {@link #compactRange(byte[], byte[])}
    • *
  • {@link #compactRange(byte[], byte[], boolean, int, int)}
    • *
* * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange() throws RocksDBException { compactRange0(nativeHandle_, false, -1, 0); } /** *

Range compaction of database.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

See also

*
    *
  • {@link #compactRange()}
    • *
  • {@link #compactRange(boolean, int, int)}
    • *
  • {@link #compactRange(byte[], byte[], boolean, int, int)}
    • *
* * @param begin start of key range (included in range) * @param end end of key range (excluded from range) * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final byte[] begin, final byte[] end) throws RocksDBException { compactRange0(nativeHandle_, begin, begin.length, end, end.length, false, -1, 0); } /** *

Range compaction of database.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

Compaction outputs should be placed in options.db_paths * [target_path_id]. Behavior is undefined if target_path_id is * out of range.

* *

See also

*
    *
  • {@link #compactRange()}
    • *
  • {@link #compactRange(byte[], byte[])}
    • *
  • {@link #compactRange(byte[], byte[], boolean, int, int)}
    • *
* * @param reduce_level reduce level after compaction * @param target_level target level to compact to * @param target_path_id the target path id of output path * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final boolean reduce_level, final int target_level, final int target_path_id) throws RocksDBException { compactRange0(nativeHandle_, reduce_level, target_level, target_path_id); } /** *

Range compaction of database.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

Compaction outputs should be placed in options.db_paths * [target_path_id]. Behavior is undefined if target_path_id is * out of range.

* *

See also

*
    *
  • {@link #compactRange()}
    • *
  • {@link #compactRange(boolean, int, int)}
    • *
  • {@link #compactRange(byte[], byte[])}
    • *
* * @param begin start of key range (included in range) * @param end end of key range (excluded from range) * @param reduce_level reduce level after compaction * @param target_level target level to compact to * @param target_path_id the target path id of output path * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final byte[] begin, final byte[] end, final boolean reduce_level, final int target_level, final int target_path_id) throws RocksDBException { compactRange0(nativeHandle_, begin, begin.length, end, end.length, reduce_level, target_level, target_path_id); } /** *

Range compaction of column family.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

See also

*
    *
  • * {@link #compactRange(ColumnFamilyHandle, boolean, int, int)} *
    • *
  • * {@link #compactRange(ColumnFamilyHandle, byte[], byte[])} *
    • *
  • * {@link #compactRange(ColumnFamilyHandle, byte[], byte[], * boolean, int, int)} *
    • *
* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance. * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final ColumnFamilyHandle columnFamilyHandle) throws RocksDBException { compactRange(nativeHandle_, false, -1, 0, columnFamilyHandle.nativeHandle_); } /** *

Range compaction of column family.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

See also

*
    *
  • {@link #compactRange(ColumnFamilyHandle)}
    • *
  • * {@link #compactRange(ColumnFamilyHandle, boolean, int, int)} *
    • *
  • * {@link #compactRange(ColumnFamilyHandle, byte[], byte[], * boolean, int, int)} *
    • *
* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance. * @param begin start of key range (included in range) * @param end end of key range (excluded from range) * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final ColumnFamilyHandle columnFamilyHandle, final byte[] begin, final byte[] end) throws RocksDBException { compactRange(nativeHandle_, begin, begin.length, end, end.length, false, -1, 0, columnFamilyHandle.nativeHandle_); } /** *

Range compaction of column family.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

Compaction outputs should be placed in options.db_paths * [target_path_id]. Behavior is undefined if target_path_id is * out of range.

* *

See also

*
    *
  • {@link #compactRange(ColumnFamilyHandle)}
    • *
  • * {@link #compactRange(ColumnFamilyHandle, byte[], byte[])} *
    • *
  • * {@link #compactRange(ColumnFamilyHandle, byte[], byte[], * boolean, int, int)} *
    • *
* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance. * @param reduce_level reduce level after compaction * @param target_level target level to compact to * @param target_path_id the target path id of output path * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final ColumnFamilyHandle columnFamilyHandle, final boolean reduce_level, final int target_level, final int target_path_id) throws RocksDBException { compactRange(nativeHandle_, reduce_level, target_level, target_path_id, columnFamilyHandle.nativeHandle_); } /** *

Range compaction of column family.

*

Note: After the database has been compacted, * all data will have been pushed down to the last level containing * any data.

* *

Compaction outputs should be placed in options.db_paths * [target_path_id]. Behavior is undefined if target_path_id is * out of range.

* *

See also

*
    *
  • {@link #compactRange(ColumnFamilyHandle)}
    • *
  • * {@link #compactRange(ColumnFamilyHandle, boolean, int, int)} *
    • *
  • * {@link #compactRange(ColumnFamilyHandle, byte[], byte[])} *
    • *
* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance. * @param begin start of key range (included in range) * @param end end of key range (excluded from range) * @param reduce_level reduce level after compaction * @param target_level target level to compact to * @param target_path_id the target path id of output path * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public void compactRange(final ColumnFamilyHandle columnFamilyHandle, final byte[] begin, final byte[] end, final boolean reduce_level, final int target_level, final int target_path_id) throws RocksDBException { compactRange(nativeHandle_, begin, begin.length, end, end.length, reduce_level, target_level, target_path_id, columnFamilyHandle.nativeHandle_); } /** *

The sequence number of the most recent transaction.

* * @return sequence number of the most * recent transaction. */ public long getLatestSequenceNumber() { return getLatestSequenceNumber(nativeHandle_); } /** *

Prevent file deletions. Compactions will continue to occur, * but no obsolete files will be deleted. Calling this multiple * times have the same effect as calling it once.

* * @throws RocksDBException thrown if operation was not performed * successfully. */ public void disableFileDeletions() throws RocksDBException { disableFileDeletions(nativeHandle_); } /** *

Allow compactions to delete obsolete files. * If force == true, the call to EnableFileDeletions() * will guarantee that file deletions are enabled after * the call, even if DisableFileDeletions() was called * multiple times before.

* *

If force == false, EnableFileDeletions will only * enable file deletion after it's been called at least * as many times as DisableFileDeletions(), enabling * the two methods to be called by two threads * concurrently without synchronization * -- i.e., file deletions will be enabled only after both * threads call EnableFileDeletions()

* * @param force boolean value described above. * * @throws RocksDBException thrown if operation was not performed * successfully. */ public void enableFileDeletions(final boolean force) throws RocksDBException { enableFileDeletions(nativeHandle_, force); } /** *

Returns an iterator that is positioned at a write-batch containing * seq_number. If the sequence number is non existent, it returns an iterator * at the first available seq_no after the requested seq_no.

* *

Must set WAL_ttl_seconds or WAL_size_limit_MB to large values to * use this api, else the WAL files will get * cleared aggressively and the iterator might keep getting invalid before * an update is read.

* * @param sequenceNumber sequence number offset * * @return {@link org.rocksdb.TransactionLogIterator} instance. * * @throws org.rocksdb.RocksDBException if iterator cannot be retrieved * from native-side. */ public TransactionLogIterator getUpdatesSince(final long sequenceNumber) throws RocksDBException { return new TransactionLogIterator( getUpdatesSince(nativeHandle_, sequenceNumber)); } /** * Private constructor. */ protected RocksDB() { super(); } // native methods protected native void open( long optionsHandle, String path) throws RocksDBException; protected native List open(long optionsHandle, String path, List columnFamilyDescriptors, int columnFamilyDescriptorsLength) throws RocksDBException; protected native static List listColumnFamilies( long optionsHandle, String path) throws RocksDBException; protected native void openROnly( long optionsHandle, String path) throws RocksDBException; protected native List openROnly( long optionsHandle, String path, List columnFamilyDescriptors, int columnFamilyDescriptorsLength) throws RocksDBException; protected native void put( long handle, byte[] key, int keyLen, byte[] value, int valueLen) throws RocksDBException; protected native void put( long handle, byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle) throws RocksDBException; protected native void put( long handle, long writeOptHandle, byte[] key, int keyLen, byte[] value, int valueLen) throws RocksDBException; protected native void put( long handle, long writeOptHandle, byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle) throws RocksDBException; protected native void write0( long writeOptHandle, long wbHandle) throws RocksDBException; protected native void write1( long writeOptHandle, long wbwiHandle) throws RocksDBException; protected native boolean keyMayExist(byte[] key, int keyLen, StringBuffer stringBuffer); protected native boolean keyMayExist(byte[] key, int keyLen, long cfHandle, StringBuffer stringBuffer); protected native boolean keyMayExist(long optionsHandle, byte[] key, int keyLen, StringBuffer stringBuffer); protected native boolean keyMayExist(long optionsHandle, byte[] key, int keyLen, long cfHandle, StringBuffer stringBuffer); protected native void merge( long handle, byte[] key, int keyLen, byte[] value, int valueLen) throws RocksDBException; protected native void merge( long handle, byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle) throws RocksDBException; protected native void merge( long handle, long writeOptHandle, byte[] key, int keyLen, byte[] value, int valueLen) throws RocksDBException; protected native void merge( long handle, long writeOptHandle, byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle) throws RocksDBException; protected native int get( long handle, byte[] key, int keyLen, byte[] value, int valueLen) throws RocksDBException; protected native int get( long handle, byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle) throws RocksDBException; protected native int get( long handle, long readOptHandle, byte[] key, int keyLen, byte[] value, int valueLen) throws RocksDBException; protected native int get( long handle, long readOptHandle, byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle) throws RocksDBException; protected native List multiGet( long dbHandle, List keys, int keysCount); protected native List multiGet( long dbHandle, List keys, int keysCount, List cfHandles); protected native List multiGet( long dbHandle, long rOptHandle, List keys, int keysCount); protected native List multiGet( long dbHandle, long rOptHandle, List keys, int keysCount, List cfHandles); protected native byte[] get( long handle, byte[] key, int keyLen) throws RocksDBException; protected native byte[] get( long handle, byte[] key, int keyLen, long cfHandle) throws RocksDBException; protected native byte[] get( long handle, long readOptHandle, byte[] key, int keyLen) throws RocksDBException; protected native byte[] get( long handle, long readOptHandle, byte[] key, int keyLen, long cfHandle) throws RocksDBException; protected native void remove( long handle, byte[] key, int keyLen) throws RocksDBException; protected native void remove( long handle, byte[] key, int keyLen, long cfHandle) throws RocksDBException; protected native void remove( long handle, long writeOptHandle, byte[] key, int keyLen) throws RocksDBException; protected native void remove( long handle, long writeOptHandle, byte[] key, int keyLen, long cfHandle) throws RocksDBException; protected native String getProperty0(long nativeHandle, String property, int propertyLength) throws RocksDBException; protected native String getProperty0(long nativeHandle, long cfHandle, String property, int propertyLength) throws RocksDBException; protected native long getLongProperty(long nativeHandle, String property, int propertyLength) throws RocksDBException; protected native long getLongProperty(long nativeHandle, long cfHandle, String property, int propertyLength) throws RocksDBException; protected native long iterator(long handle); protected native long iterator(long handle, long readOptHandle); protected native long iteratorCF(long handle, long cfHandle); protected native long iteratorCF(long handle, long cfHandle, long readOptHandle); protected native long[] iterators(long handle, List columnFamilyNames, long readOptHandle) throws RocksDBException; protected native long getSnapshot(long nativeHandle); protected native void releaseSnapshot( long nativeHandle, long snapshotHandle); private native void disposeInternal(long handle); private native long getDefaultColumnFamily(long handle); private native long createColumnFamily(long handle, ColumnFamilyDescriptor columnFamilyDescriptor) throws RocksDBException; private native void dropColumnFamily(long handle, long cfHandle) throws RocksDBException; private native void flush(long handle, long flushOptHandle) throws RocksDBException; private native void flush(long handle, long flushOptHandle, long cfHandle) throws RocksDBException; private native void compactRange0(long handle, boolean reduce_level, int target_level, int target_path_id) throws RocksDBException; private native void compactRange0(long handle, byte[] begin, int beginLen, byte[] end, int endLen, boolean reduce_level, int target_level, int target_path_id) throws RocksDBException; private native void compactRange(long handle, boolean reduce_level, int target_level, int target_path_id, long cfHandle) throws RocksDBException; private native void compactRange(long handle, byte[] begin, int beginLen, byte[] end, int endLen, boolean reduce_level, int target_level, int target_path_id, long cfHandle) throws RocksDBException; private native long getLatestSequenceNumber(long handle); private native void disableFileDeletions(long handle) throws RocksDBException; private native void enableFileDeletions(long handle, boolean force) throws RocksDBException; private native long getUpdatesSince(long handle, long sequenceNumber) throws RocksDBException; protected DBOptionsInterface options_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksDBException.java000066400000000000000000000013271267530634000247650ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * A RocksDBException encapsulates the error of an operation. This exception * type is used to describe an internal error from the c++ rocksdb library. */ public class RocksDBException extends Exception { /** * The private construct used by a set of public static factory method. * * @param msg the specified error message. */ public RocksDBException(final String msg) { super(msg); } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksEnv.java000066400000000000000000000027331267530634000233530ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

A RocksEnv is an interface used by the rocksdb implementation to access * operating system functionality like the filesystem etc.

* *

All Env implementations are safe for concurrent access from * multiple threads without any external synchronization.

*/ public class RocksEnv extends Env { /** *

Package-private constructor that uses the specified native handle * to construct a RocksEnv.

* *

Note that the ownership of the input handle * belongs to the caller, and the newly created RocksEnv will not take * the ownership of the input handle. As a result, calling * {@code dispose()} of the created RocksEnv will be no-op.

*/ RocksEnv(final long handle) { super(); nativeHandle_ = handle; disOwnNativeHandle(); } /** *

The helper function of {@link #dispose()} which all subclasses of * {@link RocksObject} must implement to release their associated C++ * resource.

* *

Note: this class is used to use the default * RocksEnv with RocksJava. The default env allocation is managed * by C++.

*/ @Override protected void disposeInternal() { } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksIterator.java000066400000000000000000000044071267530634000244140ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

An iterator that yields a sequence of key/value pairs from a source. * Multiple implementations are provided by this library. * In particular, iterators are provided * to access the contents of a Table or a DB.

* *

Multiple threads can invoke const methods on an RocksIterator without * external synchronization, but if any of the threads may call a * non-const method, all threads accessing the same RocksIterator must use * external synchronization.

* * @see org.rocksdb.RocksObject */ public class RocksIterator extends AbstractRocksIterator { protected RocksIterator(RocksDB rocksDB, long nativeHandle) { super(rocksDB, nativeHandle); } /** *

Return the key for the current entry. The underlying storage for * the returned slice is valid only until the next modification of * the iterator.

* *

REQUIRES: {@link #isValid()}

* * @return key for the current entry. */ public byte[] key() { assert(isInitialized()); return key0(nativeHandle_); } /** *

Return the value for the current entry. The underlying storage for * the returned slice is valid only until the next modification of * the iterator.

* *

REQUIRES: !AtEnd() && !AtStart()

* @return value for the current entry. */ public byte[] value() { assert(isInitialized()); return value0(nativeHandle_); } @Override final native void disposeInternal(long handle); @Override final native boolean isValid0(long handle); @Override final native void seekToFirst0(long handle); @Override final native void seekToLast0(long handle); @Override final native void next0(long handle); @Override final native void prev0(long handle); @Override final native void seek0(long handle, byte[] target, int targetLen); @Override final native void status0(long handle) throws RocksDBException; private native byte[] key0(long handle); private native byte[] value0(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksIteratorInterface.java000066400000000000000000000051241267530634000262320ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

Defines the interface for an Iterator which provides * access to data one entry at a time. Multiple implementations * are provided by this library. In particular, iterators are provided * to access the contents of a DB and Write Batch.

* *

Multiple threads can invoke const methods on an RocksIterator without * external synchronization, but if any of the threads may call a * non-const method, all threads accessing the same RocksIterator must use * external synchronization.

* * @see org.rocksdb.RocksObject */ public interface RocksIteratorInterface { /** *

An iterator is either positioned at an entry, or * not valid. This method returns true if the iterator is valid.

* * @return true if iterator is valid. */ boolean isValid(); /** *

Position at the first entry in the source. The iterator is Valid() * after this call if the source is not empty.

*/ void seekToFirst(); /** *

Position at the last entry in the source. The iterator is * valid after this call if the source is not empty.

*/ void seekToLast(); /** *

Position at the first entry in the source whose key is that or * past target.

* *

The iterator is valid after this call if the source contains * a key that comes at or past target.

* * @param target byte array describing a key or a * key prefix to seek for. */ void seek(byte[] target); /** *

Moves to the next entry in the source. After this call, Valid() is * true if the iterator was not positioned at the last entry in the source.

* *

REQUIRES: {@link #isValid()}

*/ void next(); /** *

Moves to the previous entry in the source. After this call, Valid() is * true if the iterator was not positioned at the first entry in source.

* *

REQUIRES: {@link #isValid()}

*/ void prev(); /** *

If an error has occurred, return it. Else return an ok status. * If non-blocking IO is requested and this operation cannot be * satisfied without doing some IO, then this returns Status::Incomplete().

* * @throws RocksDBException thrown if error happens in underlying * native library. */ void status() throws RocksDBException; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksMemEnv.java000066400000000000000000000016771267530634000240200ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * RocksDB memory environment. */ public class RocksMemEnv extends Env { /** *

Creates a new RocksDB environment that stores its data * in memory and delegates all non-file-storage tasks to * base_env. The caller must delete the result when it is * no longer needed.

* *

{@code *base_env} must remain live while the result is in use.

*/ public RocksMemEnv() { super(); nativeHandle_ = createMemEnv(); } @Override protected void disposeInternal() { disposeInternal(nativeHandle_); } private static native long createMemEnv(); private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/RocksObject.java000066400000000000000000000102521267530634000240240ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * RocksObject is the base-class of all RocksDB classes that has a pointer to * some c++ {@code rocksdb} object. * *

* RocksObject has {@code dispose()} function, which releases its associated c++ * resource.

*

* This function can be either called manually, or being called automatically * during the regular Java GC process. However, since Java may wrongly assume a * RocksObject only contains a long member variable and think it is small in size, * Java may give {@code RocksObject} low priority in the GC process. For this, it is * suggested to call {@code dispose()} manually. However, it is safe to let * {@code RocksObject} go out-of-scope without manually calling {@code dispose()} * as {@code dispose()} will be called in the finalizer during the * regular GC process.

*/ public abstract class RocksObject { protected RocksObject() { nativeHandle_ = 0; owningHandle_ = true; } /** * Release the c++ object manually pointed by the native handle. *

* Note that {@code dispose()} will also be called during the GC process * if it was not called before its {@code RocksObject} went out-of-scope. * However, since Java may wrongly wrongly assume those objects are * small in that they seems to only hold a long variable. As a result, * they might have low priority in the GC process. To prevent this, * it is suggested to call {@code dispose()} manually. *

*

* Note that once an instance of {@code RocksObject} has been disposed, * calling its function will lead undefined behavior. *

*/ public final synchronized void dispose() { if (isOwningNativeHandle() && isInitialized()) { disposeInternal(); } nativeHandle_ = 0; disOwnNativeHandle(); } /** * The helper function of {@code dispose()} which all subclasses of * {@code RocksObject} must implement to release their associated * C++ resource. */ protected abstract void disposeInternal(); /** * Revoke ownership of the native object. *

* This will prevent the object from attempting to delete the underlying * native object in its finalizer. This must be used when another object * takes over ownership of the native object or both will attempt to delete * the underlying object when garbage collected. *

* When {@code disOwnNativeHandle()} is called, {@code dispose()} will simply set * {@code nativeHandle_} to 0 without releasing its associated C++ resource. * As a result, incorrectly use this function may cause memory leak, and this * function call will not affect the return value of {@code isInitialized()}. *

* @see #dispose() * @see #isInitialized() */ protected void disOwnNativeHandle() { owningHandle_ = false; } /** * Returns true if the current {@code RocksObject} is responsible to release * its native handle. * * @return true if the current {@code RocksObject} is responsible to release * its native handle. * * @see #disOwnNativeHandle() * @see #dispose() */ protected boolean isOwningNativeHandle() { return owningHandle_; } /** * Returns true if the associated native handle has been initialized. * * @return true if the associated native handle has been initialized. * * @see #dispose() */ protected boolean isInitialized() { return (nativeHandle_ != 0); } /** * Simply calls {@code dispose()} and release its c++ resource if it has not * yet released. */ @Override protected void finalize() throws Throwable { dispose(); super.finalize(); } /** * A long variable holding c++ pointer pointing to some RocksDB C++ object. */ protected long nativeHandle_; /** * A flag indicating whether the current {@code RocksObject} is responsible to * release the c++ object stored in its {@code nativeHandle_}. */ private boolean owningHandle_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/SkipListMemTableConfig.java000066400000000000000000000023221267530634000261120ustar00rootroot00000000000000package org.rocksdb; /** * The config for skip-list memtable representation. */ public class SkipListMemTableConfig extends MemTableConfig { public static final long DEFAULT_LOOKAHEAD = 0; /** * SkipListMemTableConfig constructor */ public SkipListMemTableConfig() { lookahead_ = DEFAULT_LOOKAHEAD; } /** * Sets lookahead for SkipList * * @param lookahead If non-zero, each iterator's seek operation * will start the search from the previously visited record * (doing at most 'lookahead' steps). This is an * optimization for the access pattern including many * seeks with consecutive keys. * @return the current instance of SkipListMemTableConfig */ public SkipListMemTableConfig setLookahead(final long lookahead) { lookahead_ = lookahead; return this; } /** * Returns the currently set lookahead value. * * @return lookahead value */ public long lookahead() { return lookahead_; } @Override protected long newMemTableFactoryHandle() { return newMemTableFactoryHandle0(lookahead_); } private native long newMemTableFactoryHandle0(long lookahead) throws IllegalArgumentException; private long lookahead_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Slice.java000066400000000000000000000051721267530634000226600ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

Base class for slices which will receive * byte[] based access to the underlying data.

* *

byte[] backed slices typically perform better with * small keys and values. When using larger keys and * values consider using {@link org.rocksdb.DirectSlice}

*/ public class Slice extends AbstractSlice { /** *

Called from JNI to construct a new Java Slice * without an underlying C++ object set * at creation time.

* *

Note: You should be aware that * {@see org.rocksdb.RocksObject#disOwnNativeHandle()} is intentionally * called from the default Slice constructor, and that it is marked as * private. This is so that developers cannot construct their own default * Slice objects (at present). As developers cannot construct their own * Slice objects through this, they are not creating underlying C++ Slice * objects, and so there is nothing to free (dispose) from Java.

*/ private Slice() { super(); disOwnNativeHandle(); } /** *

Constructs a slice where the data is taken from * a String.

* * @param str String value. */ public Slice(final String str) { super(); createNewSliceFromString(str); } /** *

Constructs a slice where the data is a copy of * the byte array from a specific offset.

* * @param data byte array. * @param offset offset within the byte array. */ public Slice(final byte[] data, final int offset) { super(); createNewSlice0(data, offset); } /** *

Constructs a slice where the data is a copy of * the byte array.

* * @param data byte array. */ public Slice(final byte[] data) { super(); createNewSlice1(data); } /** *

Deletes underlying C++ slice pointer * and any buffered data.

* *

* Note that this function should be called only after all * RocksDB instances referencing the slice are closed. * Otherwise an undefined behavior will occur.

*/ @Override protected void disposeInternal() { disposeInternalBuf(nativeHandle_); super.disposeInternal(); } @Override protected final native byte[] data0(long handle); private native void createNewSlice0(byte[] data, int length); private native void createNewSlice1(byte[] data); private native void disposeInternalBuf(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Snapshot.java000066400000000000000000000017001267530634000234110ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Snapshot of database */ public class Snapshot extends RocksObject { Snapshot(final long nativeHandle) { super(); nativeHandle_ = nativeHandle; } /** * Return the associated sequence number; * * @return the associated sequence number of * this snapshot. */ public long getSequenceNumber() { assert(isInitialized()); return getSequenceNumber(nativeHandle_); } /** * Dont release C++ Snapshot pointer. The pointer * to the snapshot is released by the database * instance. */ @Override protected void disposeInternal() { } private native long getSequenceNumber(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/Statistics.java000066400000000000000000000021661267530634000237530ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Statistics to analyze the performance of a db. Pointer for statistics object * is managed by Options class. */ public class Statistics { private final long statsHandle_; public Statistics(final long statsHandle) { statsHandle_ = statsHandle; } public long getTickerCount(TickerType tickerType) { assert(isInitialized()); return getTickerCount0(tickerType.getValue(), statsHandle_); } public HistogramData geHistogramData(final HistogramType histogramType) { assert(isInitialized()); return geHistogramData0( histogramType.getValue(), statsHandle_); } private boolean isInitialized() { return (statsHandle_ != 0); } private native long getTickerCount0(int tickerType, long handle); private native HistogramData geHistogramData0(int histogramType, long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/StatisticsCollector.java000066400000000000000000000074651267530634000256310ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; import java.util.concurrent.Executors; import java.util.concurrent.ExecutorService; import java.util.concurrent.TimeUnit; /** *

Helper class to collect DB statistics periodically at a period specified in * constructor. Callback function (provided in constructor) is called with * every statistics collection.

* *

Caller should call start() to start statistics collection. Shutdown() should * be called to stop stats collection and should be called before statistics ( * provided in constructor) reference has been disposed.

*/ public class StatisticsCollector { private final List _statsCollectorInputList; private final ExecutorService _executorService; private final int _statsCollectionInterval; private volatile boolean _isRunning = true; /** * Constructor for statistics collector. * * @param statsCollectorInputList List of statistics collector input. * @param statsCollectionIntervalInMilliSeconds Statistics collection time * period (specified in milliseconds). */ public StatisticsCollector( final List statsCollectorInputList, final int statsCollectionIntervalInMilliSeconds) { _statsCollectorInputList = statsCollectorInputList; _statsCollectionInterval = statsCollectionIntervalInMilliSeconds; _executorService = Executors.newSingleThreadExecutor(); } public void start() { _executorService.submit(collectStatistics()); } /** * Shuts down statistics collector. * * @param shutdownTimeout Time in milli-seconds to wait for shutdown before * killing the collection process. * @throws java.lang.InterruptedException thrown if Threads are interrupted. */ public void shutDown(final int shutdownTimeout) throws InterruptedException { _isRunning = false; _executorService.shutdownNow(); // Wait for collectStatistics runnable to finish so that disposal of // statistics does not cause any exceptions to be thrown. _executorService.awaitTermination(shutdownTimeout, TimeUnit.MILLISECONDS); } private Runnable collectStatistics() { return new Runnable() { @Override public void run() { while (_isRunning) { try { if(Thread.currentThread().isInterrupted()) { break; } for(StatsCollectorInput statsCollectorInput : _statsCollectorInputList) { Statistics statistics = statsCollectorInput.getStatistics(); StatisticsCollectorCallback statsCallback = statsCollectorInput.getCallback(); // Collect ticker data for(TickerType ticker : TickerType.values()) { long tickerValue = statistics.getTickerCount(ticker); statsCallback.tickerCallback(ticker, tickerValue); } // Collect histogram data for(HistogramType histogramType : HistogramType.values()) { HistogramData histogramData = statistics.geHistogramData(histogramType); statsCallback.histogramCallback(histogramType, histogramData); } Thread.sleep(_statsCollectionInterval); } } catch (InterruptedException e) { Thread.currentThread().interrupt(); break; } catch (Exception e) { throw new RuntimeException("Error while calculating statistics", e); } } } }; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/StatisticsCollectorCallback.java000066400000000000000000000021771267530634000272410ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Callback interface provided to StatisticsCollector. * * Thread safety: * StatisticsCollector doesn't make any guarantees about thread safety. * If the same reference of StatisticsCollectorCallback is passed to multiple * StatisticsCollector references, then its the responsibility of the * user to make StatisticsCollectorCallback's implementation thread-safe. * */ public interface StatisticsCollectorCallback { /** * Callback function to get ticker values. * @param tickerType Ticker type. * @param tickerCount Value of ticker type. */ void tickerCallback(TickerType tickerType, long tickerCount); /** * Callback function to get histogram values. * @param histType Histogram type. * @param histData Histogram data. */ void histogramCallback(HistogramType histType, HistogramData histData); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/StatsCollectorInput.java000066400000000000000000000020611267530634000256000ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Contains all information necessary to collect statistics from one instance * of DB statistics. */ public class StatsCollectorInput { private final Statistics _statistics; private final StatisticsCollectorCallback _statsCallback; /** * Constructor for StatsCollectorInput. * * @param statistics Reference of DB statistics. * @param statsCallback Reference of statistics callback interface. */ public StatsCollectorInput(final Statistics statistics, final StatisticsCollectorCallback statsCallback) { _statistics = statistics; _statsCallback = statsCallback; } public Statistics getStatistics() { return _statistics; } public StatisticsCollectorCallback getCallback() { return _statsCallback; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/StringAppendOperator.java000066400000000000000000000011731267530634000257300ustar00rootroot00000000000000// Copyright (c) 2014, Vlad Balan (vlad.gm@gmail.com). All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * StringAppendOperator is a merge operator that concatenates * two strings. */ public class StringAppendOperator implements MergeOperator { @Override public long newMergeOperatorHandle() { return newMergeOperatorHandleImpl(); } private native long newMergeOperatorHandleImpl(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/TableFormatConfig.java000066400000000000000000000017311267530634000251440ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * TableFormatConfig is used to config the internal Table format of a RocksDB. * To make a RocksDB to use a specific Table format, its associated * TableFormatConfig should be properly set and passed into Options via * Options.setTableFormatConfig() and open the db using that Options. */ public abstract class TableFormatConfig { /** *

This function should only be called by Options.setTableFormatConfig(), * which will create a c++ shared-pointer to the c++ TableFactory * that associated with the Java TableFormatConfig.

* * @return native handle address to native table instance. */ abstract protected long newTableFactoryHandle(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/TickerType.java000066400000000000000000000120011267530634000236710ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public enum TickerType { // total block cache misses // REQUIRES: BLOCK_CACHE_MISS == BLOCK_CACHE_INDEX_MISS + // BLOCK_CACHE_FILTER_MISS + // BLOCK_CACHE_DATA_MISS; BLOCK_CACHE_MISS(0), // total block cache hit // REQUIRES: BLOCK_CACHE_HIT == BLOCK_CACHE_INDEX_HIT + // BLOCK_CACHE_FILTER_HIT + // BLOCK_CACHE_DATA_HIT; BLOCK_CACHE_HIT(1), // # of blocks added to block cache. BLOCK_CACHE_ADD(2), // # of times cache miss when accessing index block from block cache. BLOCK_CACHE_INDEX_MISS(3), // # of times cache hit when accessing index block from block cache. BLOCK_CACHE_INDEX_HIT(4), // # of times cache miss when accessing filter block from block cache. BLOCK_CACHE_FILTER_MISS(5), // # of times cache hit when accessing filter block from block cache. BLOCK_CACHE_FILTER_HIT(6), // # of times cache miss when accessing data block from block cache. BLOCK_CACHE_DATA_MISS(7), // # of times cache hit when accessing data block from block cache. BLOCK_CACHE_DATA_HIT(8), // # of times bloom filter has avoided file reads. BLOOM_FILTER_USEFUL(9), // # of memtable hits. MEMTABLE_HIT(10), // # of memtable misses. MEMTABLE_MISS(11), // # of Get() queries served by L0 GET_HIT_L0(12), // # of Get() queries served by L1 GET_HIT_L1(13), // # of Get() queries served by L2 and up GET_HIT_L2_AND_UP(14), /** * COMPACTION_KEY_DROP_* count the reasons for key drop during compaction * There are 3 reasons currently. */ COMPACTION_KEY_DROP_NEWER_ENTRY(15), // key was written with a newer value. COMPACTION_KEY_DROP_OBSOLETE(16), // The key is obsolete. COMPACTION_KEY_DROP_USER(17), // user compaction function has dropped the key. // Number of keys written to the database via the Put and Write call's NUMBER_KEYS_WRITTEN(18), // Number of Keys read, NUMBER_KEYS_READ(19), // Number keys updated, if inplace update is enabled NUMBER_KEYS_UPDATED(20), // Bytes written / read BYTES_WRITTEN(21), BYTES_READ(22), NO_FILE_CLOSES(23), NO_FILE_OPENS(24), NO_FILE_ERRORS(25), // Time system had to wait to do LO-L1 compactions STALL_L0_SLOWDOWN_MICROS(26), // Time system had to wait to move memtable to L1. STALL_MEMTABLE_COMPACTION_MICROS(27), // write throttle because of too many files in L0 STALL_L0_NUM_FILES_MICROS(28), // Writer has to wait for compaction or flush to finish. STALL_MICROS(29), // The wait time for db mutex. DB_MUTEX_WAIT_MICROS(30), RATE_LIMIT_DELAY_MILLIS(31), NO_ITERATORS(32), // number of iterators currently open // Number of MultiGet calls, keys read, and bytes read NUMBER_MULTIGET_CALLS(33), NUMBER_MULTIGET_KEYS_READ(34), NUMBER_MULTIGET_BYTES_READ(35), // Number of deletes records that were not required to be // written to storage because key does not exist NUMBER_FILTERED_DELETES(36), NUMBER_MERGE_FAILURES(37), SEQUENCE_NUMBER(38), // number of times bloom was checked before creating iterator on a // file, and the number of times the check was useful in avoiding // iterator creation (and thus likely IOPs). BLOOM_FILTER_PREFIX_CHECKED(39), BLOOM_FILTER_PREFIX_USEFUL(40), // Number of times we had to reseek inside an iteration to skip // over large number of keys with same userkey. NUMBER_OF_RESEEKS_IN_ITERATION(41), // Record the number of calls to GetUpadtesSince. Useful to keep track of // transaction log iterator refreshes GET_UPDATES_SINCE_CALLS(42), BLOCK_CACHE_COMPRESSED_MISS(43), // miss in the compressed block cache BLOCK_CACHE_COMPRESSED_HIT(44), // hit in the compressed block cache WAL_FILE_SYNCED(45), // Number of times WAL sync is done WAL_FILE_BYTES(46), // Number of bytes written to WAL // Writes can be processed by requesting thread or by the thread at the // head of the writers queue. WRITE_DONE_BY_SELF(47), WRITE_DONE_BY_OTHER(48), WRITE_TIMEDOUT(49), // Number of writes ending up with timed-out. WRITE_WITH_WAL(50), // Number of Write calls that request WAL COMPACT_READ_BYTES(51), // Bytes read during compaction COMPACT_WRITE_BYTES(52), // Bytes written during compaction FLUSH_WRITE_BYTES(53), // Bytes written during flush // Number of table's properties loaded directly from file, without creating // table reader object. NUMBER_DIRECT_LOAD_TABLE_PROPERTIES(54), NUMBER_SUPERVERSION_ACQUIRES(55), NUMBER_SUPERVERSION_RELEASES(56), NUMBER_SUPERVERSION_CLEANUPS(57), NUMBER_BLOCK_NOT_COMPRESSED(58); private final int value_; private TickerType(int value) { value_ = value; } public int getValue() { return value_; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/TransactionLogIterator.java000066400000000000000000000061301267530634000262550ustar00rootroot00000000000000package org.rocksdb; /** *

A TransactionLogIterator is used to iterate over the transactions in a db. * One run of the iterator is continuous, i.e. the iterator will stop at the * beginning of any gap in sequences.

*/ public class TransactionLogIterator extends RocksObject { /** *

An iterator is either positioned at a WriteBatch * or not valid. This method returns true if the iterator * is valid. Can read data from a valid iterator.

* * @return true if iterator position is valid. */ public boolean isValid() { return isValid(nativeHandle_); } /** *

Moves the iterator to the next WriteBatch. * REQUIRES: Valid() to be true.

*/ public void next() { next(nativeHandle_); } /** *

Throws RocksDBException if something went wrong.

* * @throws org.rocksdb.RocksDBException if something went * wrong in the underlying C++ code. */ public void status() throws RocksDBException { status(nativeHandle_); } /** *

If iterator position is valid, return the current * write_batch and the sequence number of the earliest * transaction contained in the batch.

* *

ONLY use if Valid() is true and status() is OK.

* * @return {@link org.rocksdb.TransactionLogIterator.BatchResult} * instance. */ public BatchResult getBatch() { assert(isValid()); return getBatch(nativeHandle_); } /** *

TransactionLogIterator constructor.

* * @param nativeHandle address to native address. */ TransactionLogIterator(final long nativeHandle) { super(); nativeHandle_ = nativeHandle; } @Override protected void disposeInternal() { disposeInternal(nativeHandle_); } /** *

BatchResult represents a data structure returned * by a TransactionLogIterator containing a sequence * number and a {@link WriteBatch} instance.

*/ public final class BatchResult { /** *

Constructor of BatchResult class.

* * @param sequenceNumber related to this BatchResult instance. * @param nativeHandle to {@link org.rocksdb.WriteBatch} * native instance. */ public BatchResult(final long sequenceNumber, final long nativeHandle) { sequenceNumber_ = sequenceNumber; writeBatch_ = new WriteBatch(nativeHandle); } /** *

Return sequence number related to this BatchResult.

* * @return Sequence number. */ public long sequenceNumber() { return sequenceNumber_; } /** *

Return contained {@link org.rocksdb.WriteBatch} * instance

* * @return {@link org.rocksdb.WriteBatch} instance. */ public WriteBatch writeBatch() { return writeBatch_; } private final long sequenceNumber_; private final WriteBatch writeBatch_; } private native void disposeInternal(long handle); private native boolean isValid(long handle); private native void next(long handle); private native void status(long handle) throws RocksDBException; private native BatchResult getBatch(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/TtlDB.java000066400000000000000000000157221267530634000225740ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; /** * Database with TTL support. * *

Use case

*

This API should be used to open the db when key-values inserted are * meant to be removed from the db in a non-strict 'ttl' amount of time * Therefore, this guarantees that key-values inserted will remain in the * db for >= ttl amount of time and the db will make efforts to remove the * key-values as soon as possible after ttl seconds of their insertion. *

* *

Behaviour

*

TTL is accepted in seconds * (int32_t)Timestamp(creation) is suffixed to values in Put internally * Expired TTL values deleted in compaction only:(Timestamp+ttl<time_now) * Get/Iterator may return expired entries(compaction not run on them yet) * Different TTL may be used during different Opens *

* *

Example

*
    *
  • Open1 at t=0 with ttl=4 and insert k1,k2, close at t=2
    • *
  • Open2 at t=3 with ttl=5. Now k1,k2 should be deleted at t>=5
    • *
* *

* read_only=true opens in the usual read-only mode. Compactions will not be * triggered(neither manual nor automatic), so no expired entries removed *

* *

Constraints

*

Not specifying/passing or non-positive TTL behaves * like TTL = infinity

* *

!!!WARNING!!!

*

Calling DB::Open directly to re-open a db created by this API will get * corrupt values(timestamp suffixed) and no ttl effect will be there * during the second Open, so use this API consistently to open the db * Be careful when passing ttl with a small positive value because the * whole database may be deleted in a small amount of time.

*/ public class TtlDB extends RocksDB { /** *

Opens a TtlDB.

* *

Database is opened in read-write mode without default TTL.

* * @param options {@link org.rocksdb.Options} instance. * @param db_path path to database. * * @return TtlDB instance. * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public static TtlDB open(final Options options, final String db_path) throws RocksDBException { return open(options, db_path, 0, false); } /** *

Opens a TtlDB.

* * @param options {@link org.rocksdb.Options} instance. * @param db_path path to database. * @param ttl time to live for new entries. * @param readOnly boolean value indicating if database if db is * opened read-only. * * @return TtlDB instance. * * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ public static TtlDB open(final Options options, final String db_path, final int ttl, final boolean readOnly) throws RocksDBException { TtlDB ttldb = new TtlDB(); ttldb.open(options.nativeHandle_, db_path, ttl, readOnly); return ttldb; } /** *

Opens a TtlDB.

* * @param options {@link org.rocksdb.Options} instance. * @param db_path path to database. * @param columnFamilyDescriptors list of column family descriptors * @param columnFamilyHandles will be filled with ColumnFamilyHandle instances * on open. * @param ttlValues time to live values per column family handle * @param readOnly boolean value indicating if database if db is * opened read-only. * * @return TtlDB instance. * * @throws RocksDBException thrown if an error occurs within the native * part of the library. * @throws java.lang.IllegalArgumentException when there is not a ttl value * per given column family handle. */ public static TtlDB open(final DBOptions options, final String db_path, final List columnFamilyDescriptors, final List columnFamilyHandles, final List ttlValues, final boolean readOnly) throws RocksDBException { if (columnFamilyDescriptors.size() != ttlValues.size()) { throw new IllegalArgumentException("There must be a ttl value per column" + "family handle."); } TtlDB ttlDB = new TtlDB(); List cfReferences = ttlDB.openCF(options.nativeHandle_, db_path, columnFamilyDescriptors, columnFamilyDescriptors.size(), ttlValues, readOnly); for (int i=0; iCreates a new ttl based column family with a name defined * in given ColumnFamilyDescriptor and allocates a * ColumnFamilyHandle within an internal structure.

* *

The ColumnFamilyHandle is automatically disposed with DB * disposal.

* * @param columnFamilyDescriptor column family to be created. * @param ttl TTL to set for this column family. * * @return {@link org.rocksdb.ColumnFamilyHandle} instance. * * @throws RocksDBException thrown if error happens in underlying * native library. */ public ColumnFamilyHandle createColumnFamilyWithTtl( final ColumnFamilyDescriptor columnFamilyDescriptor, final int ttl) throws RocksDBException { assert(isInitialized()); return new ColumnFamilyHandle(this, createColumnFamilyWithTtl(nativeHandle_, columnFamilyDescriptor, ttl)); } /** *

Close the TtlDB instance and release resource.

* *

Internally, TtlDB owns the {@code rocksdb::DB} pointer * to its associated {@link org.rocksdb.RocksDB}. The release * of that RocksDB pointer is handled in the destructor of the * c++ {@code rocksdb::TtlDB} and should be transparent to * Java developers.

*/ @Override public synchronized void close() { if (isInitialized()) { super.close(); } } /** *

A protected constructor that will be used in the static * factory method * {@link #open(Options, String, int, boolean)} * and * {@link #open(DBOptions, String, java.util.List, java.util.List, * java.util.List, boolean)}. *

*/ protected TtlDB() { super(); } @Override protected void finalize() throws Throwable { close(); super.finalize(); } private native void open(long optionsHandle, String db_path, int ttl, boolean readOnly) throws RocksDBException; private native List openCF(long optionsHandle, String db_path, List columnFamilyDescriptors, int columnFamilyDescriptorsLength, List ttlValues, boolean readOnly) throws RocksDBException; private native long createColumnFamilyWithTtl(long handle, ColumnFamilyDescriptor columnFamilyDescriptor, int ttl) throws RocksDBException; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/VectorMemTableConfig.java000066400000000000000000000021561267530634000256170ustar00rootroot00000000000000package org.rocksdb; /** * The config for vector memtable representation. */ public class VectorMemTableConfig extends MemTableConfig { public static final int DEFAULT_RESERVED_SIZE = 0; /** * VectorMemTableConfig constructor */ public VectorMemTableConfig() { reservedSize_ = DEFAULT_RESERVED_SIZE; } /** * Set the initial size of the vector that will be used * by the memtable created based on this config. * * @param size the initial size of the vector. * @return the reference to the current config. */ public VectorMemTableConfig setReservedSize(final int size) { reservedSize_ = size; return this; } /** * Returns the initial size of the vector used by the memtable * created based on this config. * * @return the initial size of the vector. */ public int reservedSize() { return reservedSize_; } @Override protected long newMemTableFactoryHandle() { return newMemTableFactoryHandle(reservedSize_); } private native long newMemTableFactoryHandle(long reservedSize) throws IllegalArgumentException; private int reservedSize_; } rocksdb-4.5.1/java/src/main/java/org/rocksdb/WBWIRocksIterator.java000066400000000000000000000103371267530634000251040ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public class WBWIRocksIterator extends AbstractRocksIterator { private final WriteEntry entry = new WriteEntry(); protected WBWIRocksIterator(final WriteBatchWithIndex wbwi, final long nativeHandle) { super(wbwi, nativeHandle); } /** * Get the current entry * * The WriteEntry is only valid * until the iterator is repositioned. * If you want to keep the WriteEntry across iterator * movements, you must make a copy of its data! * * @return The WriteEntry of the current entry */ public WriteEntry entry() { assert(isInitialized()); assert(entry != null); entry1(nativeHandle_, entry); return entry; } @Override final native void disposeInternal(long handle); @Override final native boolean isValid0(long handle); @Override final native void seekToFirst0(long handle); @Override final native void seekToLast0(long handle); @Override final native void next0(long handle); @Override final native void prev0(long handle); @Override final native void seek0(long handle, byte[] target, int targetLen); @Override final native void status0(long handle) throws RocksDBException; private native void entry1(long handle, WriteEntry entry); /** * Enumeration of the Write operation * that created the record in the Write Batch */ public enum WriteType { PUT, MERGE, DELETE, LOG } /** * Represents an entry returned by * {@link org.rocksdb.WBWIRocksIterator#entry()} * * It is worth noting that a WriteEntry with * the type {@link org.rocksdb.WBWIRocksIterator.WriteType#DELETE} * or {@link org.rocksdb.WBWIRocksIterator.WriteType#LOG} * will not have a value. */ public static class WriteEntry { WriteType type = null; final DirectSlice key; final DirectSlice value; /** * Intentionally private as this * should only be instantiated in * this manner by the outer WBWIRocksIterator * class; The class members are then modified * by calling {@link org.rocksdb.WBWIRocksIterator#entry()} */ private WriteEntry() { key = new DirectSlice(); value = new DirectSlice(); } public WriteEntry(WriteType type, DirectSlice key, DirectSlice value) { this.type = type; this.key = key; this.value = value; } /** * Returns the type of the Write Entry * * @return the WriteType of the WriteEntry */ public WriteType getType() { return type; } /** * Returns the key of the Write Entry * * @return The slice containing the key * of the WriteEntry */ public DirectSlice getKey() { return key; } /** * Returns the value of the Write Entry * * @return The slice containing the value of * the WriteEntry or null if the WriteEntry has * no value */ public DirectSlice getValue() { if(!value.isInitialized()) { return null; //TODO(AR) migrate to JDK8 java.util.Optional#empty() } else { return value; } } /** * Generates a hash code for the Write Entry. NOTE: The hash code is based * on the string representation of the key, so it may not work correctly * with exotic custom comparators. * * @return The hash code for the Write Entry */ @Override public int hashCode() { return (key == null) ? 0 : key.hashCode(); } @Override public boolean equals(Object other) { if(other == null) { return false; } else if (this == other) { return true; } else if(other instanceof WriteEntry) { final WriteEntry otherWriteEntry = (WriteEntry)other; return type.equals(otherWriteEntry.type) && key.equals(otherWriteEntry.key) && (value.isInitialized() ? value.equals(otherWriteEntry.value) : !otherWriteEntry.value.isInitialized()); } else { return false; } } } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/WriteBatch.java000066400000000000000000000077641267530634000236660ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * WriteBatch holds a collection of updates to apply atomically to a DB. * * The updates are applied in the order in which they are added * to the WriteBatch. For example, the value of "key" will be "v3" * after the following batch is written: * * batch.put("key", "v1"); * batch.remove("key"); * batch.put("key", "v2"); * batch.put("key", "v3"); * * Multiple threads can invoke const methods on a WriteBatch without * external synchronization, but if any of the threads may call a * non-const method, all threads accessing the same WriteBatch must use * external synchronization. */ public class WriteBatch extends AbstractWriteBatch { /** * Constructs a WriteBatch instance. */ public WriteBatch() { super(); newWriteBatch(0); } /** * Constructs a WriteBatch instance with a given size. * * @param reserved_bytes reserved size for WriteBatch */ public WriteBatch(final int reserved_bytes) { nativeHandle_ = 0; newWriteBatch(reserved_bytes); } /** * Support for iterating over the contents of a batch. * * @param handler A handler that is called back for each * update present in the batch * * @throws RocksDBException If we cannot iterate over the batch */ public void iterate(final Handler handler) throws RocksDBException { iterate(handler.nativeHandle_); } /** *

Private WriteBatch constructor which is used to construct * WriteBatch instances from C++ side. As the reference to this * object is also managed from C++ side the handle will be disowned.

* * @param nativeHandle address of native instance. */ WriteBatch(final long nativeHandle) { super(); disOwnNativeHandle(); nativeHandle_ = nativeHandle; } @Override final native void disposeInternal(long handle); @Override final native int count0(); @Override final native void put(byte[] key, int keyLen, byte[] value, int valueLen); @Override final native void put(byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle); @Override final native void merge(byte[] key, int keyLen, byte[] value, int valueLen); @Override final native void merge(byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle); @Override final native void remove(byte[] key, int keyLen); @Override final native void remove(byte[] key, int keyLen, long cfHandle); @Override final native void putLogData(byte[] blob, int blobLen); @Override final native void clear0(); private native void newWriteBatch(int reserved_bytes); private native void iterate(long handlerHandle) throws RocksDBException; /** * Handler callback for iterating over the contents of a batch. */ public static abstract class Handler extends RocksObject { public Handler() { super(); createNewHandler0(); } public abstract void put(byte[] key, byte[] value); public abstract void merge(byte[] key, byte[] value); public abstract void delete(byte[] key); public abstract void logData(byte[] blob); /** * shouldContinue is called by the underlying iterator * WriteBatch::Iterate. If it returns false, * iteration is halted. Otherwise, it continues * iterating. The default implementation always * returns true. * * @return boolean value indicating if the * iteration is halted. */ public boolean shouldContinue() { return true; } /** * Deletes underlying C++ handler pointer. */ @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } private native void createNewHandler0(); private native void disposeInternal(long handle); } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/WriteBatchInterface.java000066400000000000000000000065331267530634000255000ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** *

Defines the interface for a Write Batch which * holds a collection of updates to apply atomically to a DB.

*/ public interface WriteBatchInterface { /** * Returns the number of updates in the batch. * * @return number of items in WriteBatch */ int count(); /** *

Store the mapping "key->value" in the database.

* * @param key the specified key to be inserted. * @param value the value associated with the specified key. */ void put(byte[] key, byte[] value); /** *

Store the mapping "key->value" within given column * family.

* * @param columnFamilyHandle {@link org.rocksdb.ColumnFamilyHandle} * instance * @param key the specified key to be inserted. * @param value the value associated with the specified key. */ void put(ColumnFamilyHandle columnFamilyHandle, byte[] key, byte[] value); /** *

Merge "value" with the existing value of "key" in the database. * "key->merge(existing, value)"

* * @param key the specified key to be merged. * @param value the value to be merged with the current value for * the specified key. */ void merge(byte[] key, byte[] value); /** *

Merge "value" with the existing value of "key" in given column family. * "key->merge(existing, value)"

* * @param columnFamilyHandle {@link ColumnFamilyHandle} instance * @param key the specified key to be merged. * @param value the value to be merged with the current value for * the specified key. */ void merge(ColumnFamilyHandle columnFamilyHandle, byte[] key, byte[] value); /** *

If the database contains a mapping for "key", erase it. Else do nothing.

* * @param key Key to delete within database */ void remove(byte[] key); /** *

If column family contains a mapping for "key", erase it. Else do nothing.

* * @param columnFamilyHandle {@link ColumnFamilyHandle} instance * @param key Key to delete within database */ void remove(ColumnFamilyHandle columnFamilyHandle, byte[] key); /** * Append a blob of arbitrary size to the records in this batch. The blob will * be stored in the transaction log but not in any other file. In particular, * it will not be persisted to the SST files. When iterating over this * WriteBatch, WriteBatch::Handler::LogData will be called with the contents * of the blob as it is encountered. Blobs, puts, deletes, and merges will be * encountered in the same order in thich they were inserted. The blob will * NOT consume sequence number(s) and will NOT increase the count of the batch * * Example application: add timestamps to the transaction log for use in * replication. * * @param blob binary object to be inserted */ void putLogData(byte[] blob); /** * Clear all updates buffered in this batch */ void clear(); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/WriteBatchWithIndex.java000066400000000000000000000146301267530634000255000ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Similar to {@link org.rocksdb.WriteBatch} but with a binary searchable * index built for all the keys inserted. * * Calling put, merge, remove or putLogData calls the same function * as with {@link org.rocksdb.WriteBatch} whilst also building an index. * * A user can call {@link org.rocksdb.WriteBatchWithIndex#newIterator() }to create an iterator * over the write batch or * {@link org.rocksdb.WriteBatchWithIndex#newIteratorWithBase(org.rocksdb.RocksIterator)} to * get an iterator for the database with Read-Your-Own-Writes like capability */ public class WriteBatchWithIndex extends AbstractWriteBatch { /** * Creates a WriteBatchWithIndex where no bytes * are reserved up-front, bytewise comparison is * used for fallback key comparisons, * and duplicate keys operations are retained */ public WriteBatchWithIndex() { super(); newWriteBatchWithIndex(); } /** * Creates a WriteBatchWithIndex where no bytes * are reserved up-front, bytewise comparison is * used for fallback key comparisons, and duplicate key * assignment is determined by the constructor argument * * @param overwriteKey if true, overwrite the key in the index when * inserting a duplicate key, in this way an iterator will never * show two entries with the same key. */ public WriteBatchWithIndex(final boolean overwriteKey) { super(); newWriteBatchWithIndex(overwriteKey); } /** * Creates a WriteBatchWithIndex * * @param fallbackIndexComparator We fallback to this comparator * to compare keys within a column family if we cannot determine * the column family and so look up it's comparator. * * @param reservedBytes reserved bytes in underlying WriteBatch * * @param overwriteKey if true, overwrite the key in the index when * inserting a duplicate key, in this way an iterator will never * show two entries with the same key. */ public WriteBatchWithIndex(final AbstractComparator> fallbackIndexComparator, final int reservedBytes, final boolean overwriteKey) { super(); newWriteBatchWithIndex(fallbackIndexComparator.nativeHandle_, reservedBytes, overwriteKey); } /** * Create an iterator of a column family. User can call * {@link org.rocksdb.RocksIteratorInterface#seek(byte[])} to * search to the next entry of or after a key. Keys will be iterated in the * order given by index_comparator. For multiple updates on the same key, * each update will be returned as a separate entry, in the order of update * time. * * @param columnFamilyHandle The column family to iterate over * @return An iterator for the Write Batch contents, restricted to the column family */ public WBWIRocksIterator newIterator(final ColumnFamilyHandle columnFamilyHandle) { return new WBWIRocksIterator(this, iterator1(columnFamilyHandle.nativeHandle_)); } /** * Create an iterator of the default column family. User can call * {@link org.rocksdb.RocksIteratorInterface#seek(byte[])} to * search to the next entry of or after a key. Keys will be iterated in the * order given by index_comparator. For multiple updates on the same key, * each update will be returned as a separate entry, in the order of update * time. * * @return An iterator for the Write Batch contents */ public WBWIRocksIterator newIterator() { return new WBWIRocksIterator(this, iterator0()); } /** * Provides Read-Your-Own-Writes like functionality by * creating a new Iterator that will use {@link org.rocksdb.WBWIRocksIterator} * as a delta and baseIterator as a base * * @param columnFamilyHandle The column family to iterate over * @param baseIterator The base iterator, e.g. {@link org.rocksdb.RocksDB#newIterator()} * @return An iterator which shows a view comprised of both the database point-in-time * from baseIterator and modifications made in this write batch. */ public RocksIterator newIteratorWithBase(final ColumnFamilyHandle columnFamilyHandle, final RocksIterator baseIterator) { RocksIterator iterator = new RocksIterator( baseIterator.parent_, iteratorWithBase(columnFamilyHandle.nativeHandle_, baseIterator.nativeHandle_)); //when the iterator is deleted it will also delete the baseIterator baseIterator.disOwnNativeHandle(); return iterator; } /** * Provides Read-Your-Own-Writes like functionality by * creating a new Iterator that will use {@link org.rocksdb.WBWIRocksIterator} * as a delta and baseIterator as a base. Operates on the default column family. * * @param baseIterator The base iterator, e.g. {@link org.rocksdb.RocksDB#newIterator()} * @return An iterator which shows a view comprised of both the database point-in-time * from baseIterator and modifications made in this write batch. */ public RocksIterator newIteratorWithBase(final RocksIterator baseIterator) { return newIteratorWithBase(baseIterator.parent_.getDefaultColumnFamily(), baseIterator); } @Override final native void disposeInternal(long handle); @Override final native int count0(); @Override final native void put(byte[] key, int keyLen, byte[] value, int valueLen); @Override final native void put(byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle); @Override final native void merge(byte[] key, int keyLen, byte[] value, int valueLen); @Override final native void merge(byte[] key, int keyLen, byte[] value, int valueLen, long cfHandle); @Override final native void remove(byte[] key, int keyLen); @Override final native void remove(byte[] key, int keyLen, long cfHandle); @Override final native void putLogData(byte[] blob, int blobLen); @Override final native void clear0(); private native void newWriteBatchWithIndex(); private native void newWriteBatchWithIndex(boolean overwriteKey); private native void newWriteBatchWithIndex(long fallbackIndexComparatorHandle, int reservedBytes, boolean overwriteKey); private native long iterator0(); private native long iterator1(long cfHandle); private native long iteratorWithBase(long baseIteratorHandle, long cfHandle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/WriteOptions.java000066400000000000000000000067631267530634000242760ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Options that control write operations. * * Note that developers should call WriteOptions.dispose() to release the * c++ side memory before a WriteOptions instance runs out of scope. */ public class WriteOptions extends RocksObject { /** * Construct WriteOptions instance. */ public WriteOptions() { super(); newWriteOptions(); } @Override protected void disposeInternal() { assert(isInitialized()); disposeInternal(nativeHandle_); } /** * If true, the write will be flushed from the operating system * buffer cache (by calling WritableFile::Sync()) before the write * is considered complete. If this flag is true, writes will be * slower. * * If this flag is false, and the machine crashes, some recent * writes may be lost. Note that if it is just the process that * crashes (i.e., the machine does not reboot), no writes will be * lost even if sync==false. * * In other words, a DB write with sync==false has similar * crash semantics as the "write()" system call. A DB write * with sync==true has similar crash semantics to a "write()" * system call followed by "fdatasync()". * * Default: false * * @param flag a boolean flag to indicate whether a write * should be synchronized. * @return the instance of the current WriteOptions. */ public WriteOptions setSync(final boolean flag) { setSync(nativeHandle_, flag); return this; } /** * If true, the write will be flushed from the operating system * buffer cache (by calling WritableFile::Sync()) before the write * is considered complete. If this flag is true, writes will be * slower. * * If this flag is false, and the machine crashes, some recent * writes may be lost. Note that if it is just the process that * crashes (i.e., the machine does not reboot), no writes will be * lost even if sync==false. * * In other words, a DB write with sync==false has similar * crash semantics as the "write()" system call. A DB write * with sync==true has similar crash semantics to a "write()" * system call followed by "fdatasync()". * * @return boolean value indicating if sync is active. */ public boolean sync() { return sync(nativeHandle_); } /** * If true, writes will not first go to the write ahead log, * and the write may got lost after a crash. * * @param flag a boolean flag to specify whether to disable * write-ahead-log on writes. * @return the instance of the current WriteOptions. */ public WriteOptions setDisableWAL(final boolean flag) { setDisableWAL(nativeHandle_, flag); return this; } /** * If true, writes will not first go to the write ahead log, * and the write may got lost after a crash. * * @return boolean value indicating if WAL is disabled. */ public boolean disableWAL() { return disableWAL(nativeHandle_); } private native void newWriteOptions(); private native void setSync(long handle, boolean flag); private native boolean sync(long handle); private native void setDisableWAL(long handle, boolean flag); private native boolean disableWAL(long handle); private native void disposeInternal(long handle); } rocksdb-4.5.1/java/src/main/java/org/rocksdb/util/000077500000000000000000000000001267530634000217265ustar00rootroot00000000000000rocksdb-4.5.1/java/src/main/java/org/rocksdb/util/Environment.java000066400000000000000000000040341267530634000250760ustar00rootroot00000000000000package org.rocksdb.util; public class Environment { private static String OS = System.getProperty("os.name").toLowerCase(); private static String ARCH = System.getProperty("os.arch").toLowerCase(); public static boolean isWindows() { return (OS.contains("win")); } public static boolean isMac() { return (OS.contains("mac")); } public static boolean isUnix() { return (OS.contains("nix") || OS.contains("nux") || OS.contains("aix")); } public static boolean isSolaris() { return OS.contains("sunos"); } public static boolean is64Bit() { return (ARCH.indexOf("64") > 0); } public static String getSharedLibraryName(final String name) { return name + "jni"; } public static String getSharedLibraryFileName(final String name) { return appendLibOsSuffix("lib" + getSharedLibraryName(name), true); } public static String getJniLibraryName(final String name) { if (isUnix()) { final String arch = (is64Bit()) ? "64" : "32"; return String.format("%sjni-linux%s", name, arch); } else if (isMac()) { return String.format("%sjni-osx", name); } else if (isSolaris()) { return String.format("%sjni-solaris%d", name, is64Bit() ? 64 : 32); } else if (isWindows() && is64Bit()) { return String.format("%sjni-win64", name); } throw new UnsupportedOperationException(); } public static String getJniLibraryFileName(final String name) { return appendLibOsSuffix("lib" + getJniLibraryName(name), false); } private static String appendLibOsSuffix(final String libraryFileName, final boolean shared) { if (isUnix() || isSolaris()) { return libraryFileName + ".so"; } else if (isMac()) { return libraryFileName + (shared ? ".dylib" : ".jnilib"); } else if (isWindows()) { return libraryFileName + ".dll"; } throw new UnsupportedOperationException(); } public static String getJniLibraryExtension() { if (isWindows()) { return ".dll"; } return (isMac()) ? ".jnilib" : ".so"; } } rocksdb-4.5.1/java/src/main/java/org/rocksdb/util/SizeUnit.java000066400000000000000000000010721267530634000243430ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb.util; public class SizeUnit { public static final long KB = 1024L; public static final long MB = KB * KB; public static final long GB = KB * MB; public static final long TB = KB * GB; public static final long PB = KB * TB; private SizeUnit() {} } rocksdb-4.5.1/java/src/test/000077500000000000000000000000001267530634000156455ustar00rootroot00000000000000rocksdb-4.5.1/java/src/test/java/000077500000000000000000000000001267530634000165665ustar00rootroot00000000000000rocksdb-4.5.1/java/src/test/java/org/000077500000000000000000000000001267530634000173555ustar00rootroot00000000000000rocksdb-4.5.1/java/src/test/java/org/rocksdb/000077500000000000000000000000001267530634000210045ustar00rootroot00000000000000rocksdb-4.5.1/java/src/test/java/org/rocksdb/AbstractComparatorTest.java000066400000000000000000000144541267530634000263120ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.io.IOException; import java.nio.file.*; import java.util.ArrayList; import java.util.List; import java.util.Random; import static org.assertj.core.api.Assertions.assertThat; import static org.rocksdb.Types.byteToInt; import static org.rocksdb.Types.intToByte; /** * Abstract tests for both Comparator and DirectComparator */ public abstract class AbstractComparatorTest { /** * Get a comparator which will expect Integer keys * and determine an ascending order * * @return An integer ascending order key comparator */ public abstract AbstractComparator getAscendingIntKeyComparator(); /** * Test which stores random keys into the database * using an @see getAscendingIntKeyComparator * it then checks that these keys are read back in * ascending order * * @param db_path A path where we can store database * files temporarily * * @throws java.io.IOException if IO error happens. */ public void testRoundtrip(final Path db_path) throws IOException, RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options(); opt.setCreateIfMissing(true); opt.setComparator(getAscendingIntKeyComparator()); // store 10,000 random integer keys final int ITERATIONS = 10000; db = RocksDB.open(opt, db_path.toString()); final Random random = new Random(); for (int i = 0; i < ITERATIONS; i++) { final byte key[] = intToByte(random.nextInt()); if (i > 0 && db.get(key) != null) { // does key already exist (avoid duplicates) i--; // generate a different key } else { db.put(key, "value".getBytes()); } } db.close(); // re-open db and read from start to end // integer keys should be in ascending // order as defined by SimpleIntComparator db = RocksDB.open(opt, db_path.toString()); final RocksIterator it = db.newIterator(); it.seekToFirst(); int lastKey = Integer.MIN_VALUE; int count = 0; for (it.seekToFirst(); it.isValid(); it.next()) { final int thisKey = byteToInt(it.key()); assertThat(thisKey).isGreaterThan(lastKey); lastKey = thisKey; count++; } it.dispose(); db.close(); assertThat(count).isEqualTo(ITERATIONS); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } /** * Test which stores random keys into a column family * in the database * using an @see getAscendingIntKeyComparator * it then checks that these keys are read back in * ascending order * * @param db_path A path where we can store database * files temporarily * * @throws java.io.IOException if IO error happens. */ public void testRoundtripCf(final Path db_path) throws IOException, RocksDBException { DBOptions opt = null; RocksDB db = null; List cfDescriptors = new ArrayList<>(); cfDescriptors.add(new ColumnFamilyDescriptor( RocksDB.DEFAULT_COLUMN_FAMILY)); cfDescriptors.add(new ColumnFamilyDescriptor("new_cf".getBytes(), new ColumnFamilyOptions().setComparator( getAscendingIntKeyComparator()))); List cfHandles = new ArrayList<>(); try { opt = new DBOptions(). setCreateIfMissing(true). setCreateMissingColumnFamilies(true); // store 10,000 random integer keys final int ITERATIONS = 10000; db = RocksDB.open(opt, db_path.toString(), cfDescriptors, cfHandles); assertThat(cfDescriptors.size()).isEqualTo(2); assertThat(cfHandles.size()).isEqualTo(2); final Random random = new Random(); for (int i = 0; i < ITERATIONS; i++) { final byte key[] = intToByte(random.nextInt()); if (i > 0 && db.get(cfHandles.get(1), key) != null) { // does key already exist (avoid duplicates) i--; // generate a different key } else { db.put(cfHandles.get(1), key, "value".getBytes()); } } for (ColumnFamilyHandle handle : cfHandles) { handle.dispose(); } cfHandles.clear(); db.close(); // re-open db and read from start to end // integer keys should be in ascending // order as defined by SimpleIntComparator db = RocksDB.open(opt, db_path.toString(), cfDescriptors, cfHandles); assertThat(cfDescriptors.size()).isEqualTo(2); assertThat(cfHandles.size()).isEqualTo(2); final RocksIterator it = db.newIterator(cfHandles.get(1)); it.seekToFirst(); int lastKey = Integer.MIN_VALUE; int count = 0; for (it.seekToFirst(); it.isValid(); it.next()) { final int thisKey = byteToInt(it.key()); assertThat(thisKey).isGreaterThan(lastKey); lastKey = thisKey; count++; } it.dispose(); for (ColumnFamilyHandle handle : cfHandles) { handle.dispose(); } cfHandles.clear(); db.close(); assertThat(count).isEqualTo(ITERATIONS); } finally { for (ColumnFamilyHandle handle : cfHandles) { handle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } /** * Compares integer keys * so that they are in ascending order * * @param a 4-bytes representing an integer key * @param b 4-bytes representing an integer key * * @return negative if a < b, 0 if a == b, positive otherwise */ protected final int compareIntKeys(final byte[] a, final byte[] b) { final int iA = byteToInt(a); final int iB = byteToInt(b); // protect against int key calculation overflow final double diff = (double)iA - iB; final int result; if (diff < Integer.MIN_VALUE) { result = Integer.MIN_VALUE; } else if(diff > Integer.MAX_VALUE) { result = Integer.MAX_VALUE; } else { result = (int)diff; } return result; } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/BackupEngineTest.java000066400000000000000000000230601267530634000250430ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.List; import static org.assertj.core.api.Assertions.assertThat; public class BackupEngineTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Rule public TemporaryFolder backupFolder = new TemporaryFolder(); @Test public void backupDb() throws RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options().setCreateIfMissing(true); // Open empty database. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(db); // Create two backups BackupableDBOptions bopt = null; try { bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); try(final BackupEngine be = BackupEngine.open(opt.getEnv(), bopt)) { be.createNewBackup(db, false); be.createNewBackup(db, true); verifyNumberOfValidBackups(be, 2); } } finally { if(bopt != null) { bopt.dispose(); } } } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void deleteBackup() throws RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options().setCreateIfMissing(true); // Open empty database. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(db); // Create two backups BackupableDBOptions bopt = null; try { bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); try(final BackupEngine be = BackupEngine.open(opt.getEnv(), bopt)) { be.createNewBackup(db, false); be.createNewBackup(db, true); final List backupInfo = verifyNumberOfValidBackups(be, 2); // Delete the first backup be.deleteBackup(backupInfo.get(0).backupId()); final List newBackupInfo = verifyNumberOfValidBackups(be, 1); // The second backup must remain. assertThat(newBackupInfo.get(0).backupId()). isEqualTo(backupInfo.get(1).backupId()); } } finally { if(bopt != null) { bopt.dispose(); } } } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void purgeOldBackups() throws RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options().setCreateIfMissing(true); // Open empty database. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(db); // Create four backups BackupableDBOptions bopt = null; try { bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); try(final BackupEngine be = BackupEngine.open(opt.getEnv(), bopt)) { be.createNewBackup(db, false); be.createNewBackup(db, true); be.createNewBackup(db, true); be.createNewBackup(db, true); final List backupInfo = verifyNumberOfValidBackups(be, 4); // Delete everything except the latest backup be.purgeOldBackups(1); final List newBackupInfo = verifyNumberOfValidBackups(be, 1); // The latest backup must remain. assertThat(newBackupInfo.get(0).backupId()). isEqualTo(backupInfo.get(3).backupId()); } } finally { if(bopt != null) { bopt.dispose(); } } } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void restoreLatestBackup() throws RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options().setCreateIfMissing(true); // Open empty database. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(db); BackupableDBOptions bopt = null; try { bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); try (final BackupEngine be = BackupEngine.open(opt.getEnv(), bopt)) { be.createNewBackup(db, true); verifyNumberOfValidBackups(be, 1); db.put("key1".getBytes(), "valueV2".getBytes()); db.put("key2".getBytes(), "valueV2".getBytes()); be.createNewBackup(db, true); verifyNumberOfValidBackups(be, 2); db.put("key1".getBytes(), "valueV3".getBytes()); db.put("key2".getBytes(), "valueV3".getBytes()); assertThat(new String(db.get("key1".getBytes()))).endsWith("V3"); assertThat(new String(db.get("key2".getBytes()))).endsWith("V3"); db.close(); verifyNumberOfValidBackups(be, 2); // restore db from latest backup be.restoreDbFromLatestBackup(dbFolder.getRoot().getAbsolutePath(), dbFolder.getRoot().getAbsolutePath(), new RestoreOptions(false)); // Open database again. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Values must have suffix V2 because of restoring latest backup. assertThat(new String(db.get("key1".getBytes()))).endsWith("V2"); assertThat(new String(db.get("key2".getBytes()))).endsWith("V2"); } } finally { if(bopt != null) { bopt.dispose(); } } } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void restoreFromBackup() throws RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options().setCreateIfMissing(true); // Open empty database. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(db); BackupableDBOptions bopt = null; try { bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); try (final BackupEngine be = BackupEngine.open(opt.getEnv(), bopt)) { be.createNewBackup(db, true); verifyNumberOfValidBackups(be, 1); db.put("key1".getBytes(), "valueV2".getBytes()); db.put("key2".getBytes(), "valueV2".getBytes()); be.createNewBackup(db, true); verifyNumberOfValidBackups(be, 2); db.put("key1".getBytes(), "valueV3".getBytes()); db.put("key2".getBytes(), "valueV3".getBytes()); assertThat(new String(db.get("key1".getBytes()))).endsWith("V3"); assertThat(new String(db.get("key2".getBytes()))).endsWith("V3"); //close the database db.close(); //restore the backup List backupInfo = verifyNumberOfValidBackups(be, 2); // restore db from first backup be.restoreDbFromBackup(backupInfo.get(0).backupId(), dbFolder.getRoot().getAbsolutePath(), dbFolder.getRoot().getAbsolutePath(), new RestoreOptions(false)); // Open database again. db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // Values must have suffix V2 because of restoring latest backup. assertThat(new String(db.get("key1".getBytes()))).endsWith("V1"); assertThat(new String(db.get("key2".getBytes()))).endsWith("V1"); } } finally { if(bopt != null) { bopt.dispose(); } } } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } /** * Verify backups. * * @param be {@link BackupEngine} instance. * @param expectedNumberOfBackups numerical value * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ private List verifyNumberOfValidBackups(final BackupEngine be, final int expectedNumberOfBackups) throws RocksDBException { // Verify that backups exist assertThat(be.getCorruptedBackups().length). isEqualTo(0); be.garbageCollect(); final List backupInfo = be.getBackupInfo(); assertThat(backupInfo.size()). isEqualTo(expectedNumberOfBackups); return backupInfo; } /** * Fill database with some test values. * * @param db {@link RocksDB} instance. * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ private void prepareDatabase(final RocksDB db) throws RocksDBException { db.put("key1".getBytes(), "valueV1".getBytes()); db.put("key2".getBytes(), "valueV1".getBytes()); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/BackupableDBOptionsTest.java000066400000000000000000000177171267530634000263370ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import static org.assertj.core.api.Assertions.assertThat; import java.util.Random; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.ExpectedException; public class BackupableDBOptionsTest { private final static String ARBITRARY_PATH = System.getProperty("java.io.tmpdir"); @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public ExpectedException exception = ExpectedException.none(); public static final Random rand = PlatformRandomHelper. getPlatformSpecificRandomFactory(); @Test public void backupDir() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); assertThat(backupableDBOptions.backupDir()). isEqualTo(ARBITRARY_PATH); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void shareTableFiles() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); boolean value = rand.nextBoolean(); backupableDBOptions.setShareTableFiles(value); assertThat(backupableDBOptions.shareTableFiles()). isEqualTo(value); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void sync() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); boolean value = rand.nextBoolean(); backupableDBOptions.setSync(value); assertThat(backupableDBOptions.sync()).isEqualTo(value); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void destroyOldData() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); boolean value = rand.nextBoolean(); backupableDBOptions.setDestroyOldData(value); assertThat(backupableDBOptions.destroyOldData()). isEqualTo(value); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void backupLogFiles() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); boolean value = rand.nextBoolean(); backupableDBOptions.setBackupLogFiles(value); assertThat(backupableDBOptions.backupLogFiles()). isEqualTo(value); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void backupRateLimit() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); long value = Math.abs(rand.nextLong()); backupableDBOptions.setBackupRateLimit(value); assertThat(backupableDBOptions.backupRateLimit()). isEqualTo(value); // negative will be mapped to 0 backupableDBOptions.setBackupRateLimit(-1); assertThat(backupableDBOptions.backupRateLimit()). isEqualTo(0); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void restoreRateLimit() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); long value = Math.abs(rand.nextLong()); backupableDBOptions.setRestoreRateLimit(value); assertThat(backupableDBOptions.restoreRateLimit()). isEqualTo(value); // negative will be mapped to 0 backupableDBOptions.setRestoreRateLimit(-1); assertThat(backupableDBOptions.restoreRateLimit()). isEqualTo(0); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void shareFilesWithChecksum() { BackupableDBOptions backupableDBOptions = null; try { backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); boolean value = rand.nextBoolean(); backupableDBOptions.setShareFilesWithChecksum(value); assertThat(backupableDBOptions.shareFilesWithChecksum()). isEqualTo(value); } finally { if (backupableDBOptions != null) { backupableDBOptions.dispose(); } } } @Test public void failBackupDirIsNull() { exception.expect(IllegalArgumentException.class); new BackupableDBOptions(null); } @Test public void failBackupDirIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.backupDir(); } @Test public void failSetShareTableFilesIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setShareTableFiles(true); } @Test public void failShareTableFilesIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.shareTableFiles(); } @Test public void failSetSyncIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setSync(true); } @Test public void failSyncIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.sync(); } @Test public void failSetDestroyOldDataIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setDestroyOldData(true); } @Test public void failDestroyOldDataIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.destroyOldData(); } @Test public void failSetBackupLogFilesIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setBackupLogFiles(true); } @Test public void failBackupLogFilesIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.backupLogFiles(); } @Test public void failSetBackupRateLimitIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setBackupRateLimit(1); } @Test public void failBackupRateLimitIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.backupRateLimit(); } @Test public void failSetRestoreRateLimitIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setRestoreRateLimit(1); } @Test public void failRestoreRateLimitIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.restoreRateLimit(); } @Test public void failSetShareFilesWithChecksumIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.setShareFilesWithChecksum(true); } @Test public void failShareFilesWithChecksumIfDisposed(){ BackupableDBOptions options = setupUninitializedBackupableDBOptions( exception); options.shareFilesWithChecksum(); } private BackupableDBOptions setupUninitializedBackupableDBOptions( ExpectedException exception) { BackupableDBOptions backupableDBOptions = new BackupableDBOptions(ARBITRARY_PATH); backupableDBOptions.dispose(); exception.expect(AssertionError.class); return backupableDBOptions; } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/BackupableDBTest.java000066400000000000000000000327301267530634000247530ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.List; import static org.assertj.core.api.Assertions.assertThat; public class BackupableDBTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Rule public TemporaryFolder backupFolder = new TemporaryFolder(); @Test public void backupDb() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); // Create two backups bdb.createNewBackup(false); bdb.createNewBackup(true); verifyNumberOfValidBackups(bdb, 2); } finally { if (bdb != null) { bdb.close(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } @Test public void deleteBackup() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); // Create two backups bdb.createNewBackup(false); bdb.createNewBackup(true); List backupInfo = verifyNumberOfValidBackups(bdb, 2); // Delete the first backup bdb.deleteBackup(backupInfo.get(0).backupId()); List newBackupInfo = verifyNumberOfValidBackups(bdb, 1); // The second backup must remain. assertThat(newBackupInfo.get(0).backupId()). isEqualTo(backupInfo.get(1).backupId()); } finally { if (bdb != null) { bdb.close(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } @Test public void deleteBackupWithRestoreBackupableDB() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; RestoreBackupableDB rdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); // Create two backups bdb.createNewBackup(false); bdb.createNewBackup(true); List backupInfo = verifyNumberOfValidBackups(bdb, 2); // init RestoreBackupableDB rdb = new RestoreBackupableDB(bopt); // Delete the first backup rdb.deleteBackup(backupInfo.get(0).backupId()); // Fetch backup info using RestoreBackupableDB List newBackupInfo = verifyNumberOfValidBackups(rdb, 1); // The second backup must remain. assertThat(newBackupInfo.get(0).backupId()). isEqualTo(backupInfo.get(1).backupId()); } finally { if (bdb != null) { bdb.close(); } if (rdb != null) { rdb.dispose(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } @Test public void purgeOldBackups() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); // Create two backups bdb.createNewBackup(false); bdb.createNewBackup(true); bdb.createNewBackup(true); bdb.createNewBackup(true); List backupInfo = verifyNumberOfValidBackups(bdb, 4); // Delete everything except the latest backup bdb.purgeOldBackups(1); List newBackupInfo = verifyNumberOfValidBackups(bdb, 1); // The latest backup must remain. assertThat(newBackupInfo.get(0).backupId()). isEqualTo(backupInfo.get(3).backupId()); } finally { if (bdb != null) { bdb.close(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } @Test public void purgeOldBackupsWithRestoreBackupableDb() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; RestoreBackupableDB rdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); // Create two backups bdb.createNewBackup(false); bdb.createNewBackup(true); bdb.createNewBackup(true); bdb.createNewBackup(true); List infos = verifyNumberOfValidBackups(bdb, 4); assertThat(infos.get(1).size()). isEqualTo(infos.get(2).size()); assertThat(infos.get(1).numberFiles()). isEqualTo(infos.get(2).numberFiles()); long maxTimeBeforePurge = Long.MIN_VALUE; for (BackupInfo backupInfo : infos) { if (maxTimeBeforePurge < backupInfo.timestamp()) { maxTimeBeforePurge = backupInfo.timestamp(); } } // init RestoreBackupableDB rdb = new RestoreBackupableDB(bopt); // the same number of backups must // exist using RestoreBackupableDB. verifyNumberOfValidBackups(rdb, 4); rdb.purgeOldBackups(1); infos = verifyNumberOfValidBackups(rdb, 1); assertThat(infos.get(0).timestamp()). isEqualTo(maxTimeBeforePurge); } finally { if (bdb != null) { bdb.close(); } if (rdb != null) { rdb.dispose(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } @Test public void restoreLatestBackup() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; RestoreBackupableDB rdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); bdb.createNewBackup(true); verifyNumberOfValidBackups(bdb, 1); bdb.put("key1".getBytes(), "valueV2".getBytes()); bdb.put("key2".getBytes(), "valueV2".getBytes()); bdb.createNewBackup(true); verifyNumberOfValidBackups(bdb, 2); bdb.put("key1".getBytes(), "valueV3".getBytes()); bdb.put("key2".getBytes(), "valueV3".getBytes()); assertThat(new String(bdb.get("key1".getBytes()))).endsWith("V3"); assertThat(new String(bdb.get("key2".getBytes()))).endsWith("V3"); bdb.close(); // init RestoreBackupableDB rdb = new RestoreBackupableDB(bopt); verifyNumberOfValidBackups(rdb, 2); // restore db from latest backup rdb.restoreDBFromLatestBackup(dbFolder.getRoot().getAbsolutePath(), dbFolder.getRoot().getAbsolutePath(), new RestoreOptions(false)); // Open database again. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Values must have suffix V2 because of restoring latest backup. assertThat(new String(bdb.get("key1".getBytes()))).endsWith("V2"); assertThat(new String(bdb.get("key2".getBytes()))).endsWith("V2"); } finally { if (bdb != null) { bdb.close(); } if (rdb != null) { rdb.dispose(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } @Test public void restoreFromBackup() throws RocksDBException { Options opt = null; BackupableDBOptions bopt = null; BackupableDB bdb = null; RestoreBackupableDB rdb = null; try { opt = new Options().setCreateIfMissing(true); bopt = new BackupableDBOptions( backupFolder.getRoot().getAbsolutePath()); assertThat(bopt.backupDir()).isEqualTo( backupFolder.getRoot().getAbsolutePath()); // Open empty database. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Fill database with some test values prepareDatabase(bdb); bdb.createNewBackup(true); verifyNumberOfValidBackups(bdb, 1); bdb.put("key1".getBytes(), "valueV2".getBytes()); bdb.put("key2".getBytes(), "valueV2".getBytes()); bdb.createNewBackup(true); verifyNumberOfValidBackups(bdb, 2); bdb.put("key1".getBytes(), "valueV3".getBytes()); bdb.put("key2".getBytes(), "valueV3".getBytes()); assertThat(new String(bdb.get("key1".getBytes()))).endsWith("V3"); assertThat(new String(bdb.get("key2".getBytes()))).endsWith("V3"); bdb.close(); // init RestoreBackupableDB rdb = new RestoreBackupableDB(bopt); List backupInfo = verifyNumberOfValidBackups(rdb, 2); // restore db from first backup rdb.restoreDBFromBackup(backupInfo.get(0).backupId(), dbFolder.getRoot().getAbsolutePath(), dbFolder.getRoot().getAbsolutePath(), new RestoreOptions(false)); // Open database again. bdb = BackupableDB.open(opt, bopt, dbFolder.getRoot().getAbsolutePath()); // Values must have suffix V2 because of restoring latest backup. assertThat(new String(bdb.get("key1".getBytes()))).endsWith("V1"); assertThat(new String(bdb.get("key2".getBytes()))).endsWith("V1"); } finally { if (bdb != null) { bdb.close(); } if (rdb != null) { rdb.dispose(); } if (bopt != null) { bopt.dispose(); } if (opt != null) { opt.dispose(); } } } /** * Verify backups. * * @param bdb {@link BackupableDB} instance. * @param expectedNumberOfBackups numerical value * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ private List verifyNumberOfValidBackups(BackupableDB bdb, int expectedNumberOfBackups) throws RocksDBException { // Verify that backups exist assertThat(bdb.getCorruptedBackups().length). isEqualTo(0); bdb.garbageCollect(); List backupInfo = bdb.getBackupInfos(); assertThat(backupInfo.size()). isEqualTo(expectedNumberOfBackups); return backupInfo; } /** * Verify backups. * * @param rdb {@link RestoreBackupableDB} instance. * @param expectedNumberOfBackups numerical value * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ private List verifyNumberOfValidBackups( RestoreBackupableDB rdb, int expectedNumberOfBackups) throws RocksDBException { // Verify that backups exist assertThat(rdb.getCorruptedBackups().length). isEqualTo(0); rdb.garbageCollect(); List backupInfo = rdb.getBackupInfos(); assertThat(backupInfo.size()). isEqualTo(expectedNumberOfBackups); return backupInfo; } /** * Fill database with some test values. * * @param db {@link RocksDB} instance. * @throws RocksDBException thrown if an error occurs within the native * part of the library. */ private void prepareDatabase(RocksDB db) throws RocksDBException { db.put("key1".getBytes(), "valueV1".getBytes()); db.put("key2".getBytes(), "valueV1".getBytes()); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/BlockBasedTableConfigTest.java000066400000000000000000000136731267530634000266100ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class BlockBasedTableConfigTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void noBlockCache() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setNoBlockCache(true); assertThat(blockBasedTableConfig.noBlockCache()).isTrue(); } @Test public void blockCacheSize() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setBlockCacheSize(8 * 1024); assertThat(blockBasedTableConfig.blockCacheSize()). isEqualTo(8 * 1024); } @Test public void blockSizeDeviation() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setBlockSizeDeviation(12); assertThat(blockBasedTableConfig.blockSizeDeviation()). isEqualTo(12); } @Test public void blockRestartInterval() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setBlockRestartInterval(15); assertThat(blockBasedTableConfig.blockRestartInterval()). isEqualTo(15); } @Test public void wholeKeyFiltering() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setWholeKeyFiltering(false); assertThat(blockBasedTableConfig.wholeKeyFiltering()). isFalse(); } @Test public void cacheIndexAndFilterBlocks() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setCacheIndexAndFilterBlocks(true); assertThat(blockBasedTableConfig.cacheIndexAndFilterBlocks()). isTrue(); } @Test public void hashIndexAllowCollision() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setHashIndexAllowCollision(false); assertThat(blockBasedTableConfig.hashIndexAllowCollision()). isFalse(); } @Test public void blockCacheCompressedSize() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setBlockCacheCompressedSize(40); assertThat(blockBasedTableConfig.blockCacheCompressedSize()). isEqualTo(40); } @Test public void checksumType() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); assertThat(ChecksumType.values().length).isEqualTo(3); assertThat(ChecksumType.valueOf("kxxHash")). isEqualTo(ChecksumType.kxxHash); blockBasedTableConfig.setChecksumType(ChecksumType.kNoChecksum); blockBasedTableConfig.setChecksumType(ChecksumType.kxxHash); assertThat(blockBasedTableConfig.checksumType().equals( ChecksumType.kxxHash)); } @Test public void indexType() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); assertThat(IndexType.values().length).isEqualTo(2); blockBasedTableConfig.setIndexType(IndexType.kHashSearch); assertThat(blockBasedTableConfig.indexType().equals( IndexType.kHashSearch)); assertThat(IndexType.valueOf("kBinarySearch")).isNotNull(); blockBasedTableConfig.setIndexType(IndexType.valueOf("kBinarySearch")); assertThat(blockBasedTableConfig.indexType().equals( IndexType.kBinarySearch)); } @Test public void blockCacheCompressedNumShardBits() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setBlockCacheCompressedNumShardBits(4); assertThat(blockBasedTableConfig.blockCacheCompressedNumShardBits()). isEqualTo(4); } @Test public void cacheNumShardBits() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setCacheNumShardBits(5); assertThat(blockBasedTableConfig.cacheNumShardBits()). isEqualTo(5); } @Test public void blockSize() { BlockBasedTableConfig blockBasedTableConfig = new BlockBasedTableConfig(); blockBasedTableConfig.setBlockSize(10); assertThat(blockBasedTableConfig.blockSize()).isEqualTo(10); } @Test public void blockBasedTableWithFilter() { Options options = null; try { options = new Options(); options.setTableFormatConfig( new BlockBasedTableConfig().setFilter( new BloomFilter(10))); assertThat(options.tableFactoryName()). isEqualTo("BlockBasedTable"); } finally { if (options != null) { options.dispose(); } } } @Test public void blockBasedTableWithoutFilter() { Options options = null; try { options = new Options(); options.setTableFormatConfig( new BlockBasedTableConfig().setFilter(null)); assertThat(options.tableFactoryName()). isEqualTo("BlockBasedTable"); } finally { if (options != null) { options.dispose(); } } } @Test public void blockBasedTableFormatVersion() { BlockBasedTableConfig config = new BlockBasedTableConfig(); for (int version=0; version<=2; version++) { config.setFormatVersion(version); assertThat(config.formatVersion()).isEqualTo(version); } } @Test(expected = AssertionError.class) public void blockBasedTableFormatVersionFailNegative() { BlockBasedTableConfig config = new BlockBasedTableConfig(); config.setFormatVersion(-1); } @Test(expected = AssertionError.class) public void blockBasedTableFormatVersionFailIllegalVersion() { BlockBasedTableConfig config = new BlockBasedTableConfig(); config.setFormatVersion(3); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/CheckPointTest.java000066400000000000000000000053401267530634000245400ustar00rootroot00000000000000package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class CheckPointTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Rule public TemporaryFolder checkpointFolder = new TemporaryFolder(); @Test public void checkPoint() throws RocksDBException { RocksDB db = null; Options options = null; Checkpoint checkpoint = null; try { options = new Options(). setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); checkpoint = Checkpoint.create(db); checkpoint.createCheckpoint(checkpointFolder. getRoot().getAbsolutePath() + "/snapshot1"); db.put("key2".getBytes(), "value2".getBytes()); checkpoint.createCheckpoint(checkpointFolder. getRoot().getAbsolutePath() + "/snapshot2"); db.close(); db = RocksDB.open(options, checkpointFolder.getRoot().getAbsolutePath() + "/snapshot1"); assertThat(new String(db.get("key".getBytes()))). isEqualTo("value"); assertThat(db.get("key2".getBytes())).isNull(); db.close(); db = RocksDB.open(options, checkpointFolder.getRoot().getAbsolutePath() + "/snapshot2"); assertThat(new String(db.get("key".getBytes()))). isEqualTo("value"); assertThat(new String(db.get("key2".getBytes()))). isEqualTo("value2"); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (checkpoint != null) { checkpoint.dispose(); } } } @Test(expected = IllegalArgumentException.class) public void failIfDbIsNull() { Checkpoint.create(null); } @Test(expected = IllegalStateException.class) public void failIfDbNotInitialized() throws RocksDBException { RocksDB db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); db.dispose(); Checkpoint.create(db); } @Test(expected = RocksDBException.class) public void failWithIllegalPath() throws RocksDBException { RocksDB db = null; Checkpoint checkpoint = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); checkpoint = Checkpoint.create(db); checkpoint.createCheckpoint("/Z:///:\\C:\\TZ/-"); } finally { if (db != null) { db.close(); } if (checkpoint != null) { checkpoint.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/ColumnFamilyOptionsTest.java000066400000000000000000000471201267530634000264660ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import java.util.ArrayList; import java.util.List; import java.util.Properties; import java.util.Random; import static org.assertj.core.api.Assertions.assertThat; public class ColumnFamilyOptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); public static final Random rand = PlatformRandomHelper. getPlatformSpecificRandomFactory(); @Test public void getColumnFamilyOptionsFromProps() { ColumnFamilyOptions opt = null; try { // setup sample properties Properties properties = new Properties(); properties.put("write_buffer_size", "112"); properties.put("max_write_buffer_number", "13"); opt = ColumnFamilyOptions. getColumnFamilyOptionsFromProps(properties); assertThat(opt).isNotNull(); assertThat(String.valueOf(opt.writeBufferSize())). isEqualTo(properties.get("write_buffer_size")); assertThat(String.valueOf(opt.maxWriteBufferNumber())). isEqualTo(properties.get("max_write_buffer_number")); } finally { if (opt != null) { opt.dispose(); } } } @Test public void failColumnFamilyOptionsFromPropsWithIllegalValue() { ColumnFamilyOptions opt = null; try { // setup sample properties Properties properties = new Properties(); properties.put("tomato", "1024"); properties.put("burger", "2"); opt = ColumnFamilyOptions. getColumnFamilyOptionsFromProps(properties); assertThat(opt).isNull(); } finally { if (opt != null) { opt.dispose(); } } } @Test(expected = IllegalArgumentException.class) public void failColumnFamilyOptionsFromPropsWithNullValue() { ColumnFamilyOptions.getColumnFamilyOptionsFromProps(null); } @Test(expected = IllegalArgumentException.class) public void failColumnFamilyOptionsFromPropsWithEmptyProps() { ColumnFamilyOptions.getColumnFamilyOptionsFromProps( new Properties()); } @Test public void writeBufferSize() throws RocksDBException { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); opt.setWriteBufferSize(longValue); assertThat(opt.writeBufferSize()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxWriteBufferNumber() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setMaxWriteBufferNumber(intValue); assertThat(opt.maxWriteBufferNumber()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void minWriteBufferNumberToMerge() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setMinWriteBufferNumberToMerge(intValue); assertThat(opt.minWriteBufferNumberToMerge()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void numLevels() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setNumLevels(intValue); assertThat(opt.numLevels()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelZeroFileNumCompactionTrigger() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setLevelZeroFileNumCompactionTrigger(intValue); assertThat(opt.levelZeroFileNumCompactionTrigger()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelZeroSlowdownWritesTrigger() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setLevelZeroSlowdownWritesTrigger(intValue); assertThat(opt.levelZeroSlowdownWritesTrigger()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelZeroStopWritesTrigger() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setLevelZeroStopWritesTrigger(intValue); assertThat(opt.levelZeroStopWritesTrigger()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void targetFileSizeBase() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); opt.setTargetFileSizeBase(longValue); assertThat(opt.targetFileSizeBase()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void targetFileSizeMultiplier() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setTargetFileSizeMultiplier(intValue); assertThat(opt.targetFileSizeMultiplier()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBytesForLevelBase() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); opt.setMaxBytesForLevelBase(longValue); assertThat(opt.maxBytesForLevelBase()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelCompactionDynamicLevelBytes() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); final boolean boolValue = rand.nextBoolean(); opt.setLevelCompactionDynamicLevelBytes(boolValue); assertThat(opt.levelCompactionDynamicLevelBytes()) .isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBytesForLevelMultiplier() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setMaxBytesForLevelMultiplier(intValue); assertThat(opt.maxBytesForLevelMultiplier()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void expandedCompactionFactor() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setExpandedCompactionFactor(intValue); assertThat(opt.expandedCompactionFactor()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void sourceCompactionFactor() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setSourceCompactionFactor(intValue); assertThat(opt.sourceCompactionFactor()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxGrandparentOverlapFactor() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setMaxGrandparentOverlapFactor(intValue); assertThat(opt.maxGrandparentOverlapFactor()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void softRateLimit() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); double doubleValue = rand.nextDouble(); opt.setSoftRateLimit(doubleValue); assertThat(opt.softRateLimit()).isEqualTo(doubleValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void hardRateLimit() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); double doubleValue = rand.nextDouble(); opt.setHardRateLimit(doubleValue); assertThat(opt.hardRateLimit()).isEqualTo(doubleValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void rateLimitDelayMaxMilliseconds() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setRateLimitDelayMaxMilliseconds(intValue); assertThat(opt.rateLimitDelayMaxMilliseconds()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void arenaBlockSize() throws RocksDBException { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); opt.setArenaBlockSize(longValue); assertThat(opt.arenaBlockSize()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void disableAutoCompactions() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); boolean boolValue = rand.nextBoolean(); opt.setDisableAutoCompactions(boolValue); assertThat(opt.disableAutoCompactions()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void purgeRedundantKvsWhileFlush() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); boolean boolValue = rand.nextBoolean(); opt.setPurgeRedundantKvsWhileFlush(boolValue); assertThat(opt.purgeRedundantKvsWhileFlush()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void verifyChecksumsInCompaction() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); boolean boolValue = rand.nextBoolean(); opt.setVerifyChecksumsInCompaction(boolValue); assertThat(opt.verifyChecksumsInCompaction()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void filterDeletes() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); boolean boolValue = rand.nextBoolean(); opt.setFilterDeletes(boolValue); assertThat(opt.filterDeletes()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxSequentialSkipInIterations() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); opt.setMaxSequentialSkipInIterations(longValue); assertThat(opt.maxSequentialSkipInIterations()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void inplaceUpdateSupport() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); boolean boolValue = rand.nextBoolean(); opt.setInplaceUpdateSupport(boolValue); assertThat(opt.inplaceUpdateSupport()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void inplaceUpdateNumLocks() throws RocksDBException { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); opt.setInplaceUpdateNumLocks(longValue); assertThat(opt.inplaceUpdateNumLocks()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void memtablePrefixBloomBits() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); int intValue = rand.nextInt(); opt.setMemtablePrefixBloomBits(intValue); assertThat(opt.memtablePrefixBloomBits()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void memtablePrefixBloomProbes() { ColumnFamilyOptions opt = null; try { int intValue = rand.nextInt(); opt = new ColumnFamilyOptions(); opt.setMemtablePrefixBloomProbes(intValue); assertThat(opt.memtablePrefixBloomProbes()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void bloomLocality() { ColumnFamilyOptions opt = null; try { int intValue = rand.nextInt(); opt = new ColumnFamilyOptions(); opt.setBloomLocality(intValue); assertThat(opt.bloomLocality()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxSuccessiveMerges() throws RocksDBException { ColumnFamilyOptions opt = null; try { long longValue = rand.nextLong(); opt = new ColumnFamilyOptions(); opt.setMaxSuccessiveMerges(longValue); assertThat(opt.maxSuccessiveMerges()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void minPartialMergeOperands() { ColumnFamilyOptions opt = null; try { int intValue = rand.nextInt(); opt = new ColumnFamilyOptions(); opt.setMinPartialMergeOperands(intValue); assertThat(opt.minPartialMergeOperands()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void optimizeFiltersForHits() { ColumnFamilyOptions opt = null; try { boolean aBoolean = rand.nextBoolean(); opt = new ColumnFamilyOptions(); opt.setOptimizeFiltersForHits(aBoolean); assertThat(opt.optimizeFiltersForHits()).isEqualTo(aBoolean); } finally { if (opt != null) { opt.dispose(); } } } @Test public void memTable() throws RocksDBException { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); opt.setMemTableConfig(new HashLinkedListMemTableConfig()); assertThat(opt.memTableFactoryName()). isEqualTo("HashLinkedListRepFactory"); } finally { if (opt != null) { opt.dispose(); } } } @Test public void comparator() throws RocksDBException { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); opt.setComparator(BuiltinComparator.BYTEWISE_COMPARATOR); } finally { if (opt != null) { opt.dispose(); } } } @Test public void linkageOfPrepMethods() { ColumnFamilyOptions options = null; try { options = new ColumnFamilyOptions(); options.optimizeUniversalStyleCompaction(); options.optimizeUniversalStyleCompaction(4000); options.optimizeLevelStyleCompaction(); options.optimizeLevelStyleCompaction(3000); options.optimizeForPointLookup(10); } finally { if (options != null) { options.dispose(); } } } @Test public void shouldSetTestPrefixExtractor() { ColumnFamilyOptions options = null; try { options = new ColumnFamilyOptions(); options.useFixedLengthPrefixExtractor(100); options.useFixedLengthPrefixExtractor(10); } finally { if (options != null) { options.dispose(); } } } @Test public void shouldSetTestCappedPrefixExtractor() { ColumnFamilyOptions options = null; try { options = new ColumnFamilyOptions(); options.useCappedPrefixExtractor(100); options.useCappedPrefixExtractor(10); } finally { if (options != null) { options.dispose(); } } } @Test public void compressionTypes() { ColumnFamilyOptions columnFamilyOptions = null; try { columnFamilyOptions = new ColumnFamilyOptions(); for (CompressionType compressionType : CompressionType.values()) { columnFamilyOptions.setCompressionType(compressionType); assertThat(columnFamilyOptions.compressionType()). isEqualTo(compressionType); assertThat(CompressionType.valueOf("NO_COMPRESSION")). isEqualTo(CompressionType.NO_COMPRESSION); } } finally { if (columnFamilyOptions != null) { columnFamilyOptions.dispose(); } } } @Test public void compressionPerLevel() { ColumnFamilyOptions columnFamilyOptions = null; try { columnFamilyOptions = new ColumnFamilyOptions(); assertThat(columnFamilyOptions.compressionPerLevel()).isEmpty(); List compressionTypeList = new ArrayList<>(); for (int i=0; i < columnFamilyOptions.numLevels(); i++) { compressionTypeList.add(CompressionType.NO_COMPRESSION); } columnFamilyOptions.setCompressionPerLevel(compressionTypeList); compressionTypeList = columnFamilyOptions.compressionPerLevel(); for (CompressionType compressionType : compressionTypeList) { assertThat(compressionType).isEqualTo( CompressionType.NO_COMPRESSION); } } finally { if (columnFamilyOptions != null) { columnFamilyOptions.dispose(); } } } @Test public void differentCompressionsPerLevel() { ColumnFamilyOptions columnFamilyOptions = null; try { columnFamilyOptions = new ColumnFamilyOptions(); columnFamilyOptions.setNumLevels(3); assertThat(columnFamilyOptions.compressionPerLevel()).isEmpty(); List compressionTypeList = new ArrayList<>(); compressionTypeList.add(CompressionType.BZLIB2_COMPRESSION); compressionTypeList.add(CompressionType.SNAPPY_COMPRESSION); compressionTypeList.add(CompressionType.LZ4_COMPRESSION); columnFamilyOptions.setCompressionPerLevel(compressionTypeList); compressionTypeList = columnFamilyOptions.compressionPerLevel(); assertThat(compressionTypeList.size()).isEqualTo(3); assertThat(compressionTypeList). containsExactly( CompressionType.BZLIB2_COMPRESSION, CompressionType.SNAPPY_COMPRESSION, CompressionType.LZ4_COMPRESSION); } finally { if (columnFamilyOptions != null) { columnFamilyOptions.dispose(); } } } @Test public void compactionStyles() { ColumnFamilyOptions ColumnFamilyOptions = null; try { ColumnFamilyOptions = new ColumnFamilyOptions(); for (CompactionStyle compactionStyle : CompactionStyle.values()) { ColumnFamilyOptions.setCompactionStyle(compactionStyle); assertThat(ColumnFamilyOptions.compactionStyle()). isEqualTo(compactionStyle); assertThat(CompactionStyle.valueOf("FIFO")). isEqualTo(CompactionStyle.FIFO); } } finally { if (ColumnFamilyOptions != null) { ColumnFamilyOptions.dispose(); } } } @Test public void maxTableFilesSizeFIFO() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); long longValue = rand.nextLong(); // Size has to be positive longValue = (longValue < 0) ? -longValue : longValue; longValue = (longValue == 0) ? longValue + 1 : longValue; opt.setMaxTableFilesSizeFIFO(longValue); assertThat(opt.maxTableFilesSizeFIFO()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/ColumnFamilyTest.java000066400000000000000000000620511267530634000251120ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.*; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class ColumnFamilyTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void listColumnFamilies() throws RocksDBException { RocksDB db = null; Options options = null; try { options = new Options(); options.setCreateIfMissing(true); DBOptions dbOptions = new DBOptions(); dbOptions.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); // Test listColumnFamilies List columnFamilyNames; columnFamilyNames = RocksDB.listColumnFamilies(options, dbFolder.getRoot().getAbsolutePath()); assertThat(columnFamilyNames).isNotNull(); assertThat(columnFamilyNames.size()).isGreaterThan(0); assertThat(columnFamilyNames.size()).isEqualTo(1); assertThat(new String(columnFamilyNames.get(0))).isEqualTo("default"); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void defaultColumnFamily() throws RocksDBException { RocksDB db = null; Options options = null; ColumnFamilyHandle cfh; try { options = new Options().setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); cfh = db.getDefaultColumnFamily(); assertThat(cfh).isNotNull(); final byte[] key = "key".getBytes(); final byte[] value = "value".getBytes(); db.put(cfh, key, value); final byte[] actualValue = db.get(cfh, key); assertThat(cfh).isNotNull(); assertThat(actualValue).isEqualTo(value); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void createColumnFamily() throws RocksDBException { RocksDB db = null; Options options = null; ColumnFamilyHandle columnFamilyHandle = null; try { options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); columnFamilyHandle = db.createColumnFamily( new ColumnFamilyDescriptor("new_cf".getBytes(), new ColumnFamilyOptions())); List columnFamilyNames; columnFamilyNames = RocksDB.listColumnFamilies(options, dbFolder.getRoot().getAbsolutePath()); assertThat(columnFamilyNames).isNotNull(); assertThat(columnFamilyNames.size()).isGreaterThan(0); assertThat(columnFamilyNames.size()).isEqualTo(2); assertThat(new String(columnFamilyNames.get(0))).isEqualTo("default"); assertThat(new String(columnFamilyNames.get(1))).isEqualTo("new_cf"); } finally { if (columnFamilyHandle != null) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void openWithColumnFamilies() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); options.setCreateMissingColumnFamilies(true); // Test open database with column family names cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); assertThat(columnFamilyHandleList.size()).isEqualTo(2); db.put("dfkey1".getBytes(), "dfvalue".getBytes()); db.put(columnFamilyHandleList.get(0), "dfkey2".getBytes(), "dfvalue".getBytes()); db.put(columnFamilyHandleList.get(1), "newcfkey1".getBytes(), "newcfvalue".getBytes()); String retVal = new String(db.get(columnFamilyHandleList.get(1), "newcfkey1".getBytes())); assertThat(retVal).isEqualTo("newcfvalue"); assertThat((db.get(columnFamilyHandleList.get(1), "dfkey1".getBytes()))).isNull(); db.remove(columnFamilyHandleList.get(1), "newcfkey1".getBytes()); assertThat((db.get(columnFamilyHandleList.get(1), "newcfkey1".getBytes()))).isNull(); db.remove(columnFamilyHandleList.get(0), new WriteOptions(), "dfkey2".getBytes()); assertThat(db.get(columnFamilyHandleList.get(0), new ReadOptions(), "dfkey2".getBytes())).isNull(); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void getWithOutValueAndCf() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfDescriptors = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); options.setCreateMissingColumnFamilies(true); // Test open database with column family names cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfDescriptors, columnFamilyHandleList); db.put(columnFamilyHandleList.get(0), new WriteOptions(), "key1".getBytes(), "value".getBytes()); db.put("key2".getBytes(), "12345678".getBytes()); byte[] outValue = new byte[5]; // not found value int getResult = db.get("keyNotFound".getBytes(), outValue); assertThat(getResult).isEqualTo(RocksDB.NOT_FOUND); // found value which fits in outValue getResult = db.get(columnFamilyHandleList.get(0), "key1".getBytes(), outValue); assertThat(getResult).isNotEqualTo(RocksDB.NOT_FOUND); assertThat(outValue).isEqualTo("value".getBytes()); // found value which fits partially getResult = db.get(columnFamilyHandleList.get(0), new ReadOptions(), "key2".getBytes(), outValue); assertThat(getResult).isNotEqualTo(RocksDB.NOT_FOUND); assertThat(outValue).isEqualTo("12345".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void createWriteDropColumnFamily() throws RocksDBException { RocksDB db = null; DBOptions opt = null; ColumnFamilyHandle tmpColumnFamilyHandle = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { opt = new DBOptions(); opt.setCreateIfMissing(true); opt.setCreateMissingColumnFamilies(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); tmpColumnFamilyHandle = db.createColumnFamily( new ColumnFamilyDescriptor("tmpCF".getBytes(), new ColumnFamilyOptions())); db.put(tmpColumnFamilyHandle, "key".getBytes(), "value".getBytes()); db.dropColumnFamily(tmpColumnFamilyHandle); tmpColumnFamilyHandle.dispose(); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (tmpColumnFamilyHandle != null) { tmpColumnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void writeBatch() throws RocksDBException { RocksDB db = null; DBOptions opt = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { opt = new DBOptions(); opt.setCreateIfMissing(true); opt.setCreateMissingColumnFamilies(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions().setMergeOperator(new StringAppendOperator()))); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); WriteBatch writeBatch = new WriteBatch(); WriteOptions writeOpt = new WriteOptions(); writeBatch.put("key".getBytes(), "value".getBytes()); writeBatch.put(db.getDefaultColumnFamily(), "mergeKey".getBytes(), "merge".getBytes()); writeBatch.merge(db.getDefaultColumnFamily(), "mergeKey".getBytes(), "merge".getBytes()); writeBatch.put(columnFamilyHandleList.get(1), "newcfkey".getBytes(), "value".getBytes()); writeBatch.put(columnFamilyHandleList.get(1), "newcfkey2".getBytes(), "value2".getBytes()); writeBatch.remove("xyz".getBytes()); writeBatch.remove(columnFamilyHandleList.get(1), "xyz".getBytes()); db.write(writeOpt, writeBatch); writeBatch.dispose(); assertThat(db.get(columnFamilyHandleList.get(1), "xyz".getBytes()) == null); assertThat(new String(db.get(columnFamilyHandleList.get(1), "newcfkey".getBytes()))).isEqualTo("value"); assertThat(new String(db.get(columnFamilyHandleList.get(1), "newcfkey2".getBytes()))).isEqualTo("value2"); assertThat(new String(db.get("key".getBytes()))).isEqualTo("value"); // check if key is merged assertThat(new String(db.get(db.getDefaultColumnFamily(), "mergeKey".getBytes()))).isEqualTo("merge,merge"); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void iteratorOnColumnFamily() throws RocksDBException { RocksDB db = null; DBOptions options = null; RocksIterator rocksIterator = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); options.setCreateMissingColumnFamilies(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); db.put(columnFamilyHandleList.get(1), "newcfkey".getBytes(), "value".getBytes()); db.put(columnFamilyHandleList.get(1), "newcfkey2".getBytes(), "value2".getBytes()); rocksIterator = db.newIterator( columnFamilyHandleList.get(1)); rocksIterator.seekToFirst(); Map refMap = new HashMap<>(); refMap.put("newcfkey", "value"); refMap.put("newcfkey2", "value2"); int i = 0; while (rocksIterator.isValid()) { i++; assertThat(refMap.get(new String(rocksIterator.key()))). isEqualTo(new String(rocksIterator.value())); rocksIterator.next(); } assertThat(i).isEqualTo(2); rocksIterator.dispose(); } finally { if (rocksIterator != null) { rocksIterator.dispose(); } for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void multiGet() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfDescriptors = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); options.setCreateMissingColumnFamilies(true); cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfDescriptors.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfDescriptors, columnFamilyHandleList); db.put(columnFamilyHandleList.get(0), "key".getBytes(), "value".getBytes()); db.put(columnFamilyHandleList.get(1), "newcfkey".getBytes(), "value".getBytes()); List keys = new ArrayList<>(); keys.add("key".getBytes()); keys.add("newcfkey".getBytes()); Map retValues = db.multiGet(columnFamilyHandleList, keys); assertThat(retValues.size()).isEqualTo(2); assertThat(new String(retValues.get(keys.get(0)))) .isEqualTo("value"); assertThat(new String(retValues.get(keys.get(1)))) .isEqualTo("value"); retValues = db.multiGet(new ReadOptions(), columnFamilyHandleList, keys); assertThat(retValues.size()).isEqualTo(2); assertThat(new String(retValues.get(keys.get(0)))) .isEqualTo("value"); assertThat(new String(retValues.get(keys.get(1)))) .isEqualTo("value"); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void properties() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); options.setCreateMissingColumnFamilies(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); assertThat(db.getProperty("rocksdb.estimate-num-keys")). isNotNull(); assertThat(db.getLongProperty(columnFamilyHandleList.get(0), "rocksdb.estimate-num-keys")).isGreaterThanOrEqualTo(0); assertThat(db.getProperty("rocksdb.stats")).isNotNull(); assertThat(db.getProperty(columnFamilyHandleList.get(0), "rocksdb.sstables")).isNotNull(); assertThat(db.getProperty(columnFamilyHandleList.get(1), "rocksdb.estimate-num-keys")).isNotNull(); assertThat(db.getProperty(columnFamilyHandleList.get(1), "rocksdb.stats")).isNotNull(); assertThat(db.getProperty(columnFamilyHandleList.get(1), "rocksdb.sstables")).isNotNull(); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void iterators() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); List iterators = null; try { options = new DBOptions(); options.setCreateIfMissing(true); options.setCreateMissingColumnFamilies(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); iterators = db.newIterators(columnFamilyHandleList); assertThat(iterators.size()).isEqualTo(2); RocksIterator iter = iterators.get(0); iter.seekToFirst(); Map defRefMap = new HashMap<>(); defRefMap.put("dfkey1", "dfvalue"); defRefMap.put("key", "value"); while (iter.isValid()) { assertThat(defRefMap.get(new String(iter.key()))). isEqualTo(new String(iter.value())); iter.next(); } // iterate over new_cf key/value pairs Map cfRefMap = new HashMap<>(); cfRefMap.put("newcfkey", "value"); cfRefMap.put("newcfkey2", "value2"); iter = iterators.get(1); iter.seekToFirst(); while (iter.isValid()) { assertThat(cfRefMap.get(new String(iter.key()))). isEqualTo(new String(iter.value())); iter.next(); } } finally { if (iterators != null) { for (RocksIterator rocksIterator : iterators) { rocksIterator.dispose(); } } for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failPutDisposedCF() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); db.dropColumnFamily(columnFamilyHandleList.get(1)); db.put(columnFamilyHandleList.get(1), "key".getBytes(), "value".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failRemoveDisposedCF() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); db.dropColumnFamily(columnFamilyHandleList.get(1)); db.remove(columnFamilyHandleList.get(1), "key".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failGetDisposedCF() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); db.dropColumnFamily(columnFamilyHandleList.get(1)); db.get(columnFamilyHandleList.get(1), "key".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failMultiGetWithoutCorrectNumberOfCF() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList); List keys = new ArrayList<>(); keys.add("key".getBytes()); keys.add("newcfkey".getBytes()); List cfCustomList = new ArrayList<>(); db.multiGet(cfCustomList, keys); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void testByteCreateFolumnFamily() throws RocksDBException { RocksDB db = null; Options options = null; ColumnFamilyHandle cf1 = null, cf2 = null, cf3 = null; try { options = new Options().setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); byte[] b0 = new byte[] { (byte)0x00 }; byte[] b1 = new byte[] { (byte)0x01 }; byte[] b2 = new byte[] { (byte)0x02 }; cf1 = db.createColumnFamily(new ColumnFamilyDescriptor(b0)); cf2 = db.createColumnFamily(new ColumnFamilyDescriptor(b1)); List families = RocksDB.listColumnFamilies(options, dbFolder.getRoot().getAbsolutePath()); assertThat(families).contains("default".getBytes(), b0, b1); cf3 = db.createColumnFamily(new ColumnFamilyDescriptor(b2)); } finally { if (cf1 != null) { cf1.dispose(); } if (cf2 != null) { cf2.dispose(); } if (cf3 != null) { cf3.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void testCFNamesWithZeroBytes() throws RocksDBException { RocksDB db = null; Options options = null; ColumnFamilyHandle cf1 = null, cf2 = null; try { options = new Options().setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); byte[] b0 = new byte[] { 0, 0 }; byte[] b1 = new byte[] { 0, 1 }; cf1 = db.createColumnFamily(new ColumnFamilyDescriptor(b0)); cf2 = db.createColumnFamily(new ColumnFamilyDescriptor(b1)); List families = RocksDB.listColumnFamilies(options, dbFolder.getRoot().getAbsolutePath()); assertThat(families).contains("default".getBytes(), b0, b1); } finally { if (cf1 != null) { cf1.dispose(); } if (cf2 != null) { cf2.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void testCFNameSimplifiedChinese() throws RocksDBException { RocksDB db = null; Options options = null; ColumnFamilyHandle columnFamilyHandle = null; try { options = new Options().setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); final String simplifiedChinese = "\u7b80\u4f53\u5b57"; columnFamilyHandle = db.createColumnFamily( new ColumnFamilyDescriptor(simplifiedChinese.getBytes())); List families = RocksDB.listColumnFamilies(options, dbFolder.getRoot().getAbsolutePath()); assertThat(families).contains("default".getBytes(), simplifiedChinese.getBytes()); } finally { if (columnFamilyHandle != null) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/ComparatorOptionsTest.java000066400000000000000000000016761267530634000262040ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class ComparatorOptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void comparatorOptions() { final ComparatorOptions copt = new ComparatorOptions(); assertThat(copt).isNotNull(); { // UseAdaptiveMutex test copt.setUseAdaptiveMutex(true); assertThat(copt.useAdaptiveMutex()).isTrue(); copt.setUseAdaptiveMutex(false); assertThat(copt.useAdaptiveMutex()).isFalse(); } copt.dispose(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/ComparatorTest.java000066400000000000000000000164531267530634000246270ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.io.IOException; import java.nio.file.FileSystems; import static org.assertj.core.api.Assertions.assertThat; public class ComparatorTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void javaComparator() throws IOException, RocksDBException { final AbstractComparatorTest comparatorTest = new AbstractComparatorTest() { @Override public AbstractComparator getAscendingIntKeyComparator() { return new Comparator(new ComparatorOptions()) { @Override public String name() { return "test.AscendingIntKeyComparator"; } @Override public int compare(final Slice a, final Slice b) { return compareIntKeys(a.data(), b.data()); } }; } }; // test the round-tripability of keys written and read with the Comparator comparatorTest.testRoundtrip(FileSystems.getDefault().getPath( dbFolder.getRoot().getAbsolutePath())); } @Test public void javaComparatorCf() throws IOException, RocksDBException { final AbstractComparatorTest comparatorTest = new AbstractComparatorTest() { @Override public AbstractComparator getAscendingIntKeyComparator() { return new Comparator(new ComparatorOptions()) { @Override public String name() { return "test.AscendingIntKeyComparator"; } @Override public int compare(final Slice a, final Slice b) { return compareIntKeys(a.data(), b.data()); } }; } }; // test the round-tripability of keys written and read with the Comparator comparatorTest.testRoundtripCf(FileSystems.getDefault().getPath( dbFolder.getRoot().getAbsolutePath())); } @Test public void builtinForwardComparator() throws RocksDBException { Options options = null; RocksDB rocksDB = null; RocksIterator rocksIterator = null; try { options = new Options(); options.setCreateIfMissing(true); options.setComparator(BuiltinComparator.BYTEWISE_COMPARATOR); rocksDB = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); rocksDB.put("abc1".getBytes(), "abc1".getBytes()); rocksDB.put("abc2".getBytes(), "abc2".getBytes()); rocksDB.put("abc3".getBytes(), "abc3".getBytes()); rocksIterator = rocksDB.newIterator(); // Iterate over keys using a iterator rocksIterator.seekToFirst(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc1".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc1".getBytes()); rocksIterator.next(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc2".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc2".getBytes()); rocksIterator.next(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc3".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc3".getBytes()); rocksIterator.next(); assertThat(rocksIterator.isValid()).isFalse(); // Get last one rocksIterator.seekToLast(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc3".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc3".getBytes()); // Seek for abc rocksIterator.seek("abc".getBytes()); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc1".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc1".getBytes()); } finally { if (rocksIterator != null) { rocksIterator.dispose(); } if (rocksDB != null) { rocksDB.close(); } if (options != null) { options.dispose(); } } } @Test public void builtinReverseComparator() throws RocksDBException { Options options = null; RocksDB rocksDB = null; RocksIterator rocksIterator = null; try { options = new Options(); options.setCreateIfMissing(true); options.setComparator( BuiltinComparator.REVERSE_BYTEWISE_COMPARATOR); rocksDB = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); rocksDB.put("abc1".getBytes(), "abc1".getBytes()); rocksDB.put("abc2".getBytes(), "abc2".getBytes()); rocksDB.put("abc3".getBytes(), "abc3".getBytes()); rocksIterator = rocksDB.newIterator(); // Iterate over keys using a iterator rocksIterator.seekToFirst(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc3".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc3".getBytes()); rocksIterator.next(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc2".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc2".getBytes()); rocksIterator.next(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc1".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc1".getBytes()); rocksIterator.next(); assertThat(rocksIterator.isValid()).isFalse(); // Get last one rocksIterator.seekToLast(); assertThat(rocksIterator.isValid()).isTrue(); assertThat(rocksIterator.key()).isEqualTo( "abc1".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc1".getBytes()); // Will be invalid because abc is after abc1 rocksIterator.seek("abc".getBytes()); assertThat(rocksIterator.isValid()).isFalse(); // Will be abc3 because the next one after abc999 // is abc3 rocksIterator.seek("abc999".getBytes()); assertThat(rocksIterator.key()).isEqualTo( "abc3".getBytes()); assertThat(rocksIterator.value()).isEqualTo( "abc3".getBytes()); } finally { if (rocksIterator != null) { rocksIterator.dispose(); } if (rocksDB != null) { rocksDB.close(); } if (options != null) { options.dispose(); } } } @Test public void builtinComparatorEnum(){ assertThat(BuiltinComparator.BYTEWISE_COMPARATOR.ordinal()) .isEqualTo(0); assertThat( BuiltinComparator.REVERSE_BYTEWISE_COMPARATOR.ordinal()) .isEqualTo(1); assertThat(BuiltinComparator.values().length).isEqualTo(2); assertThat(BuiltinComparator.valueOf("BYTEWISE_COMPARATOR")). isEqualTo(BuiltinComparator.BYTEWISE_COMPARATOR); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/CompressionOptionsTest.java000066400000000000000000000012151267530634000263630ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.Test; public class CompressionOptionsTest { @Test public void getCompressionType() { for (CompressionType compressionType : CompressionType.values()) { String libraryName = compressionType.getLibraryName(); compressionType.equals(CompressionType.getCompressionType( libraryName)); } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/DBOptionsTest.java000066400000000000000000000322251267530634000243540ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import java.util.Properties; import java.util.Random; import static org.assertj.core.api.Assertions.assertThat; public class DBOptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); public static final Random rand = PlatformRandomHelper. getPlatformSpecificRandomFactory(); @Test public void getDBOptionsFromProps() { DBOptions opt = null; try { // setup sample properties Properties properties = new Properties(); properties.put("allow_mmap_reads", "true"); properties.put("bytes_per_sync", "13"); opt = DBOptions.getDBOptionsFromProps(properties); assertThat(opt).isNotNull(); assertThat(String.valueOf(opt.allowMmapReads())). isEqualTo(properties.get("allow_mmap_reads")); assertThat(String.valueOf(opt.bytesPerSync())). isEqualTo(properties.get("bytes_per_sync")); } finally { if (opt != null) { opt.dispose(); } } } @Test public void failDBOptionsFromPropsWithIllegalValue() { DBOptions opt = null; try { // setup sample properties Properties properties = new Properties(); properties.put("tomato", "1024"); properties.put("burger", "2"); opt = DBOptions. getDBOptionsFromProps(properties); assertThat(opt).isNull(); } finally { if (opt != null) { opt.dispose(); } } } @Test(expected = IllegalArgumentException.class) public void failDBOptionsFromPropsWithNullValue() { DBOptions.getDBOptionsFromProps(null); } @Test(expected = IllegalArgumentException.class) public void failDBOptionsFromPropsWithEmptyProps() { DBOptions.getDBOptionsFromProps( new Properties()); } @Test public void setIncreaseParallelism() { DBOptions opt = null; try { opt = new DBOptions(); final int threads = Runtime.getRuntime().availableProcessors() * 2; opt.setIncreaseParallelism(threads); } finally { if (opt != null) { opt.dispose(); } } } @Test public void createIfMissing() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setCreateIfMissing(boolValue); assertThat(opt.createIfMissing()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void createMissingColumnFamilies() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setCreateMissingColumnFamilies(boolValue); assertThat(opt.createMissingColumnFamilies()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void errorIfExists() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setErrorIfExists(boolValue); assertThat(opt.errorIfExists()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void paranoidChecks() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setParanoidChecks(boolValue); assertThat(opt.paranoidChecks()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxTotalWalSize() { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setMaxTotalWalSize(longValue); assertThat(opt.maxTotalWalSize()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxOpenFiles() { DBOptions opt = null; try { opt = new DBOptions(); int intValue = rand.nextInt(); opt.setMaxOpenFiles(intValue); assertThat(opt.maxOpenFiles()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void disableDataSync() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setDisableDataSync(boolValue); assertThat(opt.disableDataSync()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void useFsync() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setUseFsync(boolValue); assertThat(opt.useFsync()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void dbLogDir() { DBOptions opt = null; try { opt = new DBOptions(); String str = "path/to/DbLogDir"; opt.setDbLogDir(str); assertThat(opt.dbLogDir()).isEqualTo(str); } finally { if (opt != null) { opt.dispose(); } } } @Test public void walDir() { DBOptions opt = null; try { opt = new DBOptions(); String str = "path/to/WalDir"; opt.setWalDir(str); assertThat(opt.walDir()).isEqualTo(str); } finally { if (opt != null) { opt.dispose(); } } } @Test public void deleteObsoleteFilesPeriodMicros() { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setDeleteObsoleteFilesPeriodMicros(longValue); assertThat(opt.deleteObsoleteFilesPeriodMicros()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBackgroundCompactions() { DBOptions opt = null; try { opt = new DBOptions(); int intValue = rand.nextInt(); opt.setMaxBackgroundCompactions(intValue); assertThat(opt.maxBackgroundCompactions()). isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBackgroundFlushes() { DBOptions opt = null; try { opt = new DBOptions(); int intValue = rand.nextInt(); opt.setMaxBackgroundFlushes(intValue); assertThat(opt.maxBackgroundFlushes()). isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxLogFileSize() throws RocksDBException { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setMaxLogFileSize(longValue); assertThat(opt.maxLogFileSize()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void logFileTimeToRoll() throws RocksDBException { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setLogFileTimeToRoll(longValue); assertThat(opt.logFileTimeToRoll()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void keepLogFileNum() throws RocksDBException { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setKeepLogFileNum(longValue); assertThat(opt.keepLogFileNum()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxManifestFileSize() { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setMaxManifestFileSize(longValue); assertThat(opt.maxManifestFileSize()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void tableCacheNumshardbits() { DBOptions opt = null; try { opt = new DBOptions(); int intValue = rand.nextInt(); opt.setTableCacheNumshardbits(intValue); assertThat(opt.tableCacheNumshardbits()). isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void walSizeLimitMB() { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setWalSizeLimitMB(longValue); assertThat(opt.walSizeLimitMB()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void walTtlSeconds() { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setWalTtlSeconds(longValue); assertThat(opt.walTtlSeconds()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void manifestPreallocationSize() throws RocksDBException { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setManifestPreallocationSize(longValue); assertThat(opt.manifestPreallocationSize()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void allowOsBuffer() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setAllowOsBuffer(boolValue); assertThat(opt.allowOsBuffer()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void allowMmapReads() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setAllowMmapReads(boolValue); assertThat(opt.allowMmapReads()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void allowMmapWrites() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setAllowMmapWrites(boolValue); assertThat(opt.allowMmapWrites()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void isFdCloseOnExec() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setIsFdCloseOnExec(boolValue); assertThat(opt.isFdCloseOnExec()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void statsDumpPeriodSec() { DBOptions opt = null; try { opt = new DBOptions(); int intValue = rand.nextInt(); opt.setStatsDumpPeriodSec(intValue); assertThat(opt.statsDumpPeriodSec()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void adviseRandomOnOpen() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setAdviseRandomOnOpen(boolValue); assertThat(opt.adviseRandomOnOpen()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void useAdaptiveMutex() { DBOptions opt = null; try { opt = new DBOptions(); boolean boolValue = rand.nextBoolean(); opt.setUseAdaptiveMutex(boolValue); assertThat(opt.useAdaptiveMutex()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void bytesPerSync() { DBOptions opt = null; try { opt = new DBOptions(); long longValue = rand.nextLong(); opt.setBytesPerSync(longValue); assertThat(opt.bytesPerSync()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void rateLimiterConfig() { DBOptions options = null; DBOptions anotherOptions = null; try { options = new DBOptions(); RateLimiterConfig rateLimiterConfig = new GenericRateLimiterConfig(1000, 100 * 1000, 1); options.setRateLimiterConfig(rateLimiterConfig); // Test with parameter initialization anotherOptions = new DBOptions(); anotherOptions.setRateLimiterConfig( new GenericRateLimiterConfig(1000)); } finally { if (options != null) { options.dispose(); } if (anotherOptions != null) { anotherOptions.dispose(); } } } @Test public void statistics() { DBOptions options = new DBOptions(); Statistics statistics = options.createStatistics(). statisticsPtr(); assertThat(statistics).isNotNull(); DBOptions anotherOptions = new DBOptions(); statistics = anotherOptions.statisticsPtr(); assertThat(statistics).isNotNull(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/DirectComparatorTest.java000066400000000000000000000031501267530634000257500ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.io.IOException; import java.nio.file.FileSystems; public class DirectComparatorTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void directComparator() throws IOException, RocksDBException { final AbstractComparatorTest comparatorTest = new AbstractComparatorTest() { @Override public AbstractComparator getAscendingIntKeyComparator() { return new DirectComparator(new ComparatorOptions()) { @Override public String name() { return "test.AscendingIntKeyDirectComparator"; } @Override public int compare(final DirectSlice a, final DirectSlice b) { final byte ax[] = new byte[4], bx[] = new byte[4]; a.data().get(ax); b.data().get(bx); return compareIntKeys(ax, bx); } }; } }; // test the round-tripability of keys written and read with the DirectComparator comparatorTest.testRoundtrip(FileSystems.getDefault().getPath( dbFolder.getRoot().getAbsolutePath())); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/DirectSliceTest.java000066400000000000000000000057601267530634000247110ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import java.nio.ByteBuffer; import static org.assertj.core.api.Assertions.assertThat; public class DirectSliceTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void directSlice() { DirectSlice directSlice = null; DirectSlice otherSlice = null; try { directSlice = new DirectSlice("abc"); otherSlice = new DirectSlice("abc"); assertThat(directSlice.toString()).isEqualTo("abc"); // clear first slice directSlice.clear(); assertThat(directSlice.toString()).isEmpty(); // get first char in otherslice assertThat(otherSlice.get(0)).isEqualTo("a".getBytes()[0]); // remove prefix otherSlice.removePrefix(1); assertThat(otherSlice.toString()).isEqualTo("bc"); } finally { if (directSlice != null) { directSlice.dispose(); } if (otherSlice != null) { otherSlice.dispose(); } } } @Test public void directSliceWithByteBuffer() { DirectSlice directSlice = null; try { byte[] data = "Some text".getBytes(); ByteBuffer buffer = ByteBuffer.allocateDirect(data.length + 1); buffer.put(data); buffer.put(data.length, (byte)0); directSlice = new DirectSlice(buffer); assertThat(directSlice.toString()).isEqualTo("Some text"); } finally { if (directSlice != null) { directSlice.dispose(); } } } @Test public void directSliceWithByteBufferAndLength() { DirectSlice directSlice = null; try { byte[] data = "Some text".getBytes(); ByteBuffer buffer = ByteBuffer.allocateDirect(data.length); buffer.put(data); directSlice = new DirectSlice(buffer, 4); assertThat(directSlice.toString()).isEqualTo("Some"); } finally { if (directSlice != null) { directSlice.dispose(); } } } @Test(expected = AssertionError.class) public void directSliceInitWithoutDirectAllocation() { DirectSlice directSlice = null; try { byte[] data = "Some text".getBytes(); ByteBuffer buffer = ByteBuffer.wrap(data); directSlice = new DirectSlice(buffer); } finally { if (directSlice != null) { directSlice.dispose(); } } } @Test(expected = AssertionError.class) public void directSlicePrefixInitWithoutDirectAllocation() { DirectSlice directSlice = null; try { byte[] data = "Some text".getBytes(); ByteBuffer buffer = ByteBuffer.wrap(data); directSlice = new DirectSlice(buffer, 4); } finally { if (directSlice != null) { directSlice.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/FilterTest.java000066400000000000000000000026261267530634000237420ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; public class FilterTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void filter() { Options options = null; try { options = new Options(); // test table config options.setTableFormatConfig(new BlockBasedTableConfig(). setFilter(new BloomFilter())); options.dispose(); System.gc(); System.runFinalization(); // new Bloom filter options = new Options(); BlockBasedTableConfig blockConfig = new BlockBasedTableConfig(); blockConfig.setFilter(new BloomFilter()); options.setTableFormatConfig(blockConfig); BloomFilter bloomFilter = new BloomFilter(10); blockConfig.setFilter(bloomFilter); options.setTableFormatConfig(blockConfig); System.gc(); System.runFinalization(); blockConfig.setFilter(new BloomFilter(10, false)); options.setTableFormatConfig(blockConfig); } finally { if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/FlushTest.java000066400000000000000000000040311267530634000235660ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class FlushTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void flush() throws RocksDBException { RocksDB db = null; Options options = null; WriteOptions wOpt = null; FlushOptions flushOptions = null; try { options = new Options(); // Setup options options.setCreateIfMissing(true); options.setMaxWriteBufferNumber(10); options.setMinWriteBufferNumberToMerge(10); wOpt = new WriteOptions(); flushOptions = new FlushOptions(); flushOptions.setWaitForFlush(true); assertThat(flushOptions.waitForFlush()).isTrue(); wOpt.setDisableWAL(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put(wOpt, "key1".getBytes(), "value1".getBytes()); db.put(wOpt, "key2".getBytes(), "value2".getBytes()); db.put(wOpt, "key3".getBytes(), "value3".getBytes()); db.put(wOpt, "key4".getBytes(), "value4".getBytes()); assertThat(db.getProperty("rocksdb.num-entries-active-mem-table")).isEqualTo("4"); db.flush(flushOptions); assertThat(db.getProperty("rocksdb.num-entries-active-mem-table")). isEqualTo("0"); } finally { if (flushOptions != null) { flushOptions.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (wOpt != null) { wOpt.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/InfoLogLevelTest.java000066400000000000000000000075201267530634000250400ustar00rootroot00000000000000package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import org.rocksdb.util.Environment; import java.io.IOException; import static java.nio.file.Files.readAllBytes; import static java.nio.file.Paths.get; import static org.assertj.core.api.Assertions.assertThat; public class InfoLogLevelTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void testInfoLogLevel() throws RocksDBException, IOException { RocksDB db = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); assertThat(getLogContentsWithoutHeader()).isNotEmpty(); } finally { if (db != null) { db.close(); } } } @Test public void testFatalLogLevel() throws RocksDBException, IOException { RocksDB db = null; Options options = null; try { options = new Options(). setCreateIfMissing(true). setInfoLogLevel(InfoLogLevel.FATAL_LEVEL); assertThat(options.infoLogLevel()). isEqualTo(InfoLogLevel.FATAL_LEVEL); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); // As InfoLogLevel is set to FATAL_LEVEL, here we expect the log // content to be empty. assertThat(getLogContentsWithoutHeader()).isEmpty(); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void testFatalLogLevelWithDBOptions() throws RocksDBException, IOException { RocksDB db = null; Options options = null; DBOptions dbOptions = null; try { dbOptions = new DBOptions(). setInfoLogLevel(InfoLogLevel.FATAL_LEVEL); options = new Options(dbOptions, new ColumnFamilyOptions()). setCreateIfMissing(true); assertThat(dbOptions.infoLogLevel()). isEqualTo(InfoLogLevel.FATAL_LEVEL); assertThat(options.infoLogLevel()). isEqualTo(InfoLogLevel.FATAL_LEVEL); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); assertThat(getLogContentsWithoutHeader()).isEmpty(); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (dbOptions != null) { dbOptions.dispose(); } } } @Test(expected = IllegalArgumentException.class) public void failIfIllegalByteValueProvided() { InfoLogLevel.getInfoLogLevel((byte)-1); } @Test public void valueOf() { assertThat(InfoLogLevel.valueOf("DEBUG_LEVEL")). isEqualTo(InfoLogLevel.DEBUG_LEVEL); } /** * Read LOG file contents into String. * * @return LOG file contents as String. * @throws IOException if file is not found. */ private String getLogContentsWithoutHeader() throws IOException { final String separator = Environment.isWindows() ? "\n" : System.getProperty("line.separator"); final String[] lines = new String(readAllBytes(get( dbFolder.getRoot().getAbsolutePath()+ "/LOG"))).split(separator); int first_non_header = lines.length; // Identify the last line of the header for (int i = lines.length - 1; i >= 0; --i) { if (lines[i].indexOf("Options.") >= 0 && lines[i].indexOf(':') >= 0) { first_non_header = i + 1; break; } } StringBuilder builder = new StringBuilder(); for (int i = first_non_header; i < lines.length; ++i) { builder.append(lines[i]).append(separator); } return builder.toString(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/KeyMayExistTest.java000066400000000000000000000061411267530634000247250ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.ArrayList; import java.util.List; import static org.assertj.core.api.Assertions.assertThat; public class KeyMayExistTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void keyMayExist() throws RocksDBException { RocksDB db = null; DBOptions options = null; List cfDescriptors = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { options = new DBOptions(); options.setCreateIfMissing(true) .setCreateMissingColumnFamilies(true); // open database using cf names cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfDescriptors.add(new ColumnFamilyDescriptor("new_cf".getBytes())); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfDescriptors, columnFamilyHandleList); assertThat(columnFamilyHandleList.size()). isEqualTo(2); db.put("key".getBytes(), "value".getBytes()); // Test without column family StringBuffer retValue = new StringBuffer(); boolean exists = db.keyMayExist("key".getBytes(), retValue); assertThat(exists).isTrue(); assertThat(retValue.toString()). isEqualTo("value"); // Test without column family but with readOptions retValue = new StringBuffer(); exists = db.keyMayExist(new ReadOptions(), "key".getBytes(), retValue); assertThat(exists).isTrue(); assertThat(retValue.toString()). isEqualTo("value"); // Test with column family retValue = new StringBuffer(); exists = db.keyMayExist(columnFamilyHandleList.get(0), "key".getBytes(), retValue); assertThat(exists).isTrue(); assertThat(retValue.toString()). isEqualTo("value"); // Test with column family and readOptions retValue = new StringBuffer(); exists = db.keyMayExist(new ReadOptions(), columnFamilyHandleList.get(0), "key".getBytes(), retValue); assertThat(exists).isTrue(); assertThat(retValue.toString()). isEqualTo("value"); // KeyMayExist in CF1 must return false assertThat(db.keyMayExist(columnFamilyHandleList.get(1), "key".getBytes(), retValue)).isFalse(); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/LoggerTest.java000066400000000000000000000144311267530634000237310ustar00rootroot00000000000000package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.ArrayList; import java.util.List; import java.util.concurrent.atomic.AtomicInteger; import static org.assertj.core.api.Assertions.assertThat; public class LoggerTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); private AtomicInteger logMessageCounter = new AtomicInteger(); @Test public void customLogger() throws RocksDBException { RocksDB db = null; logMessageCounter.set(0); try { // Setup options final Options options = new Options(). setInfoLogLevel(InfoLogLevel.DEBUG_LEVEL). setCreateIfMissing(true); // Create new logger with max log level passed by options Logger logger = new Logger(options) { @Override protected void log(InfoLogLevel infoLogLevel, String logMsg) { assertThat(logMsg).isNotNull(); assertThat(logMsg.length()).isGreaterThan(0); logMessageCounter.incrementAndGet(); } }; // Set custom logger to options options.setLogger(logger); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); // there should be more than zero received log messages in // debug level. assertThat(logMessageCounter.get()).isGreaterThan(0); } finally { if (db != null) { db.close(); } } logMessageCounter.set(0); } @Test public void fatalLogger() throws RocksDBException { RocksDB db = null; logMessageCounter.set(0); try { // Setup options final Options options = new Options(). setInfoLogLevel(InfoLogLevel.FATAL_LEVEL). setCreateIfMissing(true); // Create new logger with max log level passed by options Logger logger = new Logger(options) { @Override protected void log(InfoLogLevel infoLogLevel, String logMsg) { assertThat(logMsg).isNotNull(); assertThat(logMsg.length()).isGreaterThan(0); logMessageCounter.incrementAndGet(); } }; // Set custom logger to options options.setLogger(logger); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); // there should be zero messages // using fatal level as log level. assertThat(logMessageCounter.get()).isEqualTo(0); } finally { if (db != null) { db.close(); } } logMessageCounter.set(0); } @Test public void dbOptionsLogger() throws RocksDBException { RocksDB db = null; Logger logger = null; List cfHandles = new ArrayList<>(); List cfDescriptors = new ArrayList<>(); cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); logMessageCounter.set(0); try { // Setup options final DBOptions options = new DBOptions(). setInfoLogLevel(InfoLogLevel.FATAL_LEVEL). setCreateIfMissing(true); // Create new logger with max log level passed by options logger = new Logger(options) { @Override protected void log(InfoLogLevel infoLogLevel, String logMsg) { assertThat(logMsg).isNotNull(); assertThat(logMsg.length()).isGreaterThan(0); logMessageCounter.incrementAndGet(); } }; // Set custom logger to options options.setLogger(logger); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath(), cfDescriptors, cfHandles); // there should be zero messages // using fatal level as log level. assertThat(logMessageCounter.get()).isEqualTo(0); logMessageCounter.set(0); } finally { for (ColumnFamilyHandle columnFamilyHandle : cfHandles) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (logger != null) { logger.dispose(); } } } @Test public void setInfoLogLevel() { Logger logger = null; try { // Setup options final Options options = new Options(). setInfoLogLevel(InfoLogLevel.FATAL_LEVEL). setCreateIfMissing(true); // Create new logger with max log level passed by options logger = new Logger(options) { @Override protected void log(InfoLogLevel infoLogLevel, String logMsg) { assertThat(logMsg).isNotNull(); assertThat(logMsg.length()).isGreaterThan(0); logMessageCounter.incrementAndGet(); } }; assertThat(logger.infoLogLevel()). isEqualTo(InfoLogLevel.FATAL_LEVEL); logger.setInfoLogLevel(InfoLogLevel.DEBUG_LEVEL); assertThat(logger.infoLogLevel()). isEqualTo(InfoLogLevel.DEBUG_LEVEL); } finally { if (logger != null) { logger.dispose(); } } } @Test public void changeLogLevelAtRuntime() throws RocksDBException { RocksDB db = null; logMessageCounter.set(0); try { // Setup options final Options options = new Options(). setInfoLogLevel(InfoLogLevel.FATAL_LEVEL). setCreateIfMissing(true); // Create new logger with max log level passed by options Logger logger = new Logger(options) { @Override protected void log(InfoLogLevel infoLogLevel, String logMsg) { assertThat(logMsg).isNotNull(); assertThat(logMsg.length()).isGreaterThan(0); logMessageCounter.incrementAndGet(); } }; // Set custom logger to options options.setLogger(logger); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); // there should be zero messages // using fatal level as log level. assertThat(logMessageCounter.get()).isEqualTo(0); // change log level to debug level logger.setInfoLogLevel(InfoLogLevel.DEBUG_LEVEL); db.put("key".getBytes(), "value".getBytes()); db.flush(new FlushOptions().setWaitForFlush(true)); // messages shall be received due to previous actions. assertThat(logMessageCounter.get()).isNotEqualTo(0); } finally { if (db != null) { db.close(); } } logMessageCounter.set(0); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/MemTableTest.java000066400000000000000000000106151267530634000242000ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class MemTableTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void hashSkipListMemTable() throws RocksDBException { Options options = null; try { options = new Options(); // Test HashSkipListMemTableConfig HashSkipListMemTableConfig memTableConfig = new HashSkipListMemTableConfig(); assertThat(memTableConfig.bucketCount()). isEqualTo(1000000); memTableConfig.setBucketCount(2000000); assertThat(memTableConfig.bucketCount()). isEqualTo(2000000); assertThat(memTableConfig.height()). isEqualTo(4); memTableConfig.setHeight(5); assertThat(memTableConfig.height()). isEqualTo(5); assertThat(memTableConfig.branchingFactor()). isEqualTo(4); memTableConfig.setBranchingFactor(6); assertThat(memTableConfig.branchingFactor()). isEqualTo(6); options.setMemTableConfig(memTableConfig); } finally { if (options != null) { options.dispose(); } } } @Test public void skipListMemTable() throws RocksDBException { Options options = null; try { options = new Options(); SkipListMemTableConfig skipMemTableConfig = new SkipListMemTableConfig(); assertThat(skipMemTableConfig.lookahead()). isEqualTo(0); skipMemTableConfig.setLookahead(20); assertThat(skipMemTableConfig.lookahead()). isEqualTo(20); options.setMemTableConfig(skipMemTableConfig); options.dispose(); } finally { if (options != null) { options.dispose(); } } } @Test public void hashLinkedListMemTable() throws RocksDBException { Options options = null; try { options = new Options(); HashLinkedListMemTableConfig hashLinkedListMemTableConfig = new HashLinkedListMemTableConfig(); assertThat(hashLinkedListMemTableConfig.bucketCount()). isEqualTo(50000); hashLinkedListMemTableConfig.setBucketCount(100000); assertThat(hashLinkedListMemTableConfig.bucketCount()). isEqualTo(100000); assertThat(hashLinkedListMemTableConfig.hugePageTlbSize()). isEqualTo(0); hashLinkedListMemTableConfig.setHugePageTlbSize(1); assertThat(hashLinkedListMemTableConfig.hugePageTlbSize()). isEqualTo(1); assertThat(hashLinkedListMemTableConfig. bucketEntriesLoggingThreshold()). isEqualTo(4096); hashLinkedListMemTableConfig. setBucketEntriesLoggingThreshold(200); assertThat(hashLinkedListMemTableConfig. bucketEntriesLoggingThreshold()). isEqualTo(200); assertThat(hashLinkedListMemTableConfig. ifLogBucketDistWhenFlush()).isTrue(); hashLinkedListMemTableConfig. setIfLogBucketDistWhenFlush(false); assertThat(hashLinkedListMemTableConfig. ifLogBucketDistWhenFlush()).isFalse(); assertThat(hashLinkedListMemTableConfig. thresholdUseSkiplist()). isEqualTo(256); hashLinkedListMemTableConfig.setThresholdUseSkiplist(29); assertThat(hashLinkedListMemTableConfig. thresholdUseSkiplist()). isEqualTo(29); options.setMemTableConfig(hashLinkedListMemTableConfig); } finally { if (options != null) { options.dispose(); } } } @Test public void vectorMemTable() throws RocksDBException { Options options = null; try { options = new Options(); VectorMemTableConfig vectorMemTableConfig = new VectorMemTableConfig(); assertThat(vectorMemTableConfig.reservedSize()). isEqualTo(0); vectorMemTableConfig.setReservedSize(123); assertThat(vectorMemTableConfig.reservedSize()). isEqualTo(123); options.setMemTableConfig(vectorMemTableConfig); options.dispose(); } finally { if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/MergeTest.java000066400000000000000000000214371267530634000235550ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.List; import java.util.ArrayList; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class MergeTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void stringOption() throws InterruptedException, RocksDBException { RocksDB db = null; Options opt = null; try { String db_path_string = dbFolder.getRoot().getAbsolutePath(); opt = new Options(); opt.setCreateIfMissing(true); opt.setMergeOperatorName("stringappend"); db = RocksDB.open(opt, db_path_string); // writing aa under key db.put("key".getBytes(), "aa".getBytes()); // merge bb under key db.merge("key".getBytes(), "bb".getBytes()); byte[] value = db.get("key".getBytes()); String strValue = new String(value); assertThat(strValue).isEqualTo("aa,bb"); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void cFStringOption() throws InterruptedException, RocksDBException { RocksDB db = null; DBOptions opt = null; List columnFamilyHandleList = new ArrayList<>(); try { String db_path_string = dbFolder.getRoot().getAbsolutePath(); opt = new DBOptions(); opt.setCreateIfMissing(true); opt.setCreateMissingColumnFamilies(true); List cfDescriptors = new ArrayList<>(); cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions().setMergeOperatorName( "stringappend"))); cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions().setMergeOperatorName( "stringappend"))); db = RocksDB.open(opt, db_path_string, cfDescriptors, columnFamilyHandleList); // writing aa under key db.put(columnFamilyHandleList.get(1), "cfkey".getBytes(), "aa".getBytes()); // merge bb under key db.merge(columnFamilyHandleList.get(1), "cfkey".getBytes(), "bb".getBytes()); byte[] value = db.get(columnFamilyHandleList.get(1), "cfkey".getBytes()); String strValue = new String(value); assertThat(strValue).isEqualTo("aa,bb"); } finally { for (ColumnFamilyHandle handle : columnFamilyHandleList) { handle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void operatorOption() throws InterruptedException, RocksDBException { RocksDB db = null; Options opt = null; try { String db_path_string = dbFolder.getRoot().getAbsolutePath(); opt = new Options(); opt.setCreateIfMissing(true); StringAppendOperator stringAppendOperator = new StringAppendOperator(); opt.setMergeOperator(stringAppendOperator); db = RocksDB.open(opt, db_path_string); // Writing aa under key db.put("key".getBytes(), "aa".getBytes()); // Writing bb under key db.merge("key".getBytes(), "bb".getBytes()); byte[] value = db.get("key".getBytes()); String strValue = new String(value); assertThat(strValue).isEqualTo("aa,bb"); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void cFOperatorOption() throws InterruptedException, RocksDBException { RocksDB db = null; DBOptions opt = null; ColumnFamilyHandle cfHandle = null; List cfDescriptors = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); try { String db_path_string = dbFolder.getRoot().getAbsolutePath(); opt = new DBOptions(); opt.setCreateIfMissing(true); opt.setCreateMissingColumnFamilies(true); StringAppendOperator stringAppendOperator = new StringAppendOperator(); cfDescriptors.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions().setMergeOperator( stringAppendOperator))); cfDescriptors.add(new ColumnFamilyDescriptor("new_cf".getBytes(), new ColumnFamilyOptions().setMergeOperator( stringAppendOperator))); db = RocksDB.open(opt, db_path_string, cfDescriptors, columnFamilyHandleList); // writing aa under key db.put(columnFamilyHandleList.get(1), "cfkey".getBytes(), "aa".getBytes()); // merge bb under key db.merge(columnFamilyHandleList.get(1), "cfkey".getBytes(), "bb".getBytes()); byte[] value = db.get(columnFamilyHandleList.get(1), "cfkey".getBytes()); String strValue = new String(value); // Test also with createColumnFamily cfHandle = db.createColumnFamily( new ColumnFamilyDescriptor("new_cf2".getBytes(), new ColumnFamilyOptions().setMergeOperator(stringAppendOperator))); // writing xx under cfkey2 db.put(cfHandle, "cfkey2".getBytes(), "xx".getBytes()); // merge yy under cfkey2 db.merge(cfHandle, new WriteOptions(), "cfkey2".getBytes(), "yy".getBytes()); value = db.get(cfHandle, "cfkey2".getBytes()); String strValueTmpCf = new String(value); assertThat(strValue).isEqualTo("aa,bb"); assertThat(strValueTmpCf).isEqualTo("xx,yy"); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (cfHandle != null) { cfHandle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void operatorGcBehaviour() throws RocksDBException { Options opt = null; RocksDB db = null; try { String db_path_string = dbFolder.getRoot().getAbsolutePath(); opt = new Options(); opt.setCreateIfMissing(true); StringAppendOperator stringAppendOperator = new StringAppendOperator(); opt.setMergeOperator(stringAppendOperator); db = RocksDB.open(opt, db_path_string); db.close(); opt.dispose(); System.gc(); System.runFinalization(); // test reuse opt = new Options(); opt.setMergeOperator(stringAppendOperator); db = RocksDB.open(opt, db_path_string); db.close(); opt.dispose(); System.gc(); System.runFinalization(); // test param init opt = new Options(); opt.setMergeOperator(new StringAppendOperator()); db = RocksDB.open(opt, db_path_string); db.close(); opt.dispose(); System.gc(); System.runFinalization(); // test replace one with another merge operator instance opt = new Options(); opt.setMergeOperator(stringAppendOperator); StringAppendOperator newStringAppendOperator = new StringAppendOperator(); opt.setMergeOperator(newStringAppendOperator); db = RocksDB.open(opt, db_path_string); db.close(); opt.dispose(); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void emptyStringInSetMergeOperatorByName() { Options opt = null; ColumnFamilyOptions cOpt = null; try { opt = new Options(); cOpt = new ColumnFamilyOptions(); opt.setMergeOperatorName(""); cOpt.setMergeOperatorName(""); } finally { if (opt != null) { opt.dispose(); } if (cOpt != null) { cOpt.dispose(); } } } @Test(expected = IllegalArgumentException.class) public void nullStringInSetMergeOperatorByNameOptions() { Options opt = null; try { opt = new Options(); opt.setMergeOperatorName(null); } finally { if (opt != null) { opt.dispose(); } } } @Test(expected = IllegalArgumentException.class) public void nullStringInSetMergeOperatorByNameColumnFamilyOptions() { ColumnFamilyOptions opt = null; try { opt = new ColumnFamilyOptions(); opt.setMergeOperatorName(null); } finally { if (opt != null) { opt.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/MixedOptionsTest.java000066400000000000000000000037241267530634000251370ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class MixedOptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void mixedOptionsTest(){ // Set a table factory and check the names ColumnFamilyOptions cfOptions = new ColumnFamilyOptions(); cfOptions.setTableFormatConfig(new BlockBasedTableConfig(). setFilter(new BloomFilter())); assertThat(cfOptions.tableFactoryName()).isEqualTo( "BlockBasedTable"); cfOptions.setTableFormatConfig(new PlainTableConfig()); assertThat(cfOptions.tableFactoryName()).isEqualTo("PlainTable"); // Initialize a dbOptions object from cf options and // db options DBOptions dbOptions = new DBOptions(); Options options = new Options(dbOptions, cfOptions); assertThat(options.tableFactoryName()).isEqualTo("PlainTable"); // Free instances options.dispose(); options = null; cfOptions.dispose(); cfOptions = null; dbOptions.dispose(); dbOptions = null; System.gc(); System.runFinalization(); // Test Optimize for statements cfOptions = new ColumnFamilyOptions(); cfOptions.optimizeUniversalStyleCompaction(); cfOptions.optimizeLevelStyleCompaction(); cfOptions.optimizeForPointLookup(1024); options = new Options(); options.optimizeLevelStyleCompaction(); options.optimizeLevelStyleCompaction(400); options.optimizeUniversalStyleCompaction(); options.optimizeUniversalStyleCompaction(400); options.optimizeForPointLookup(1024); options.prepareForBulkLoad(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/NativeLibraryLoaderTest.java000066400000000000000000000020451267530634000264120ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import org.rocksdb.util.Environment; import java.io.IOException; import java.nio.file.*; import static org.assertj.core.api.Assertions.assertThat; public class NativeLibraryLoaderTest { @Rule public TemporaryFolder temporaryFolder = new TemporaryFolder(); @Test public void tempFolder() throws IOException { NativeLibraryLoader.getInstance().loadLibraryFromJarToTemp( temporaryFolder.getRoot().getAbsolutePath()); Path path = Paths.get(temporaryFolder.getRoot().getAbsolutePath(), Environment.getJniLibraryFileName("rocksdb")); assertThat(Files.exists(path)).isTrue(); assertThat(Files.isReadable(path)).isTrue(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/OptionsTest.java000066400000000000000000000705311267530634000241500ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.ArrayList; import java.util.List; import java.util.Random; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class OptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); public static final Random rand = PlatformRandomHelper. getPlatformSpecificRandomFactory(); @Test public void setIncreaseParallelism() { Options opt = null; try { opt = new Options(); final int threads = Runtime.getRuntime().availableProcessors() * 2; opt.setIncreaseParallelism(threads); } finally { if (opt != null) { opt.dispose(); } } } @Test public void writeBufferSize() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setWriteBufferSize(longValue); assertThat(opt.writeBufferSize()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxWriteBufferNumber() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMaxWriteBufferNumber(intValue); assertThat(opt.maxWriteBufferNumber()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void minWriteBufferNumberToMerge() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMinWriteBufferNumberToMerge(intValue); assertThat(opt.minWriteBufferNumberToMerge()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void numLevels() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setNumLevels(intValue); assertThat(opt.numLevels()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelZeroFileNumCompactionTrigger() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setLevelZeroFileNumCompactionTrigger(intValue); assertThat(opt.levelZeroFileNumCompactionTrigger()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelZeroSlowdownWritesTrigger() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setLevelZeroSlowdownWritesTrigger(intValue); assertThat(opt.levelZeroSlowdownWritesTrigger()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelZeroStopWritesTrigger() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setLevelZeroStopWritesTrigger(intValue); assertThat(opt.levelZeroStopWritesTrigger()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void targetFileSizeBase() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setTargetFileSizeBase(longValue); assertThat(opt.targetFileSizeBase()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void targetFileSizeMultiplier() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setTargetFileSizeMultiplier(intValue); assertThat(opt.targetFileSizeMultiplier()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBytesForLevelBase() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setMaxBytesForLevelBase(longValue); assertThat(opt.maxBytesForLevelBase()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void levelCompactionDynamicLevelBytes() { Options opt = null; try { opt = new Options(); final boolean boolValue = rand.nextBoolean(); opt.setLevelCompactionDynamicLevelBytes(boolValue); assertThat(opt.levelCompactionDynamicLevelBytes()) .isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBytesForLevelMultiplier() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMaxBytesForLevelMultiplier(intValue); assertThat(opt.maxBytesForLevelMultiplier()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void expandedCompactionFactor() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setExpandedCompactionFactor(intValue); assertThat(opt.expandedCompactionFactor()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void sourceCompactionFactor() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setSourceCompactionFactor(intValue); assertThat(opt.sourceCompactionFactor()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxGrandparentOverlapFactor() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMaxGrandparentOverlapFactor(intValue); assertThat(opt.maxGrandparentOverlapFactor()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void softRateLimit() { Options opt = null; try { opt = new Options(); double doubleValue = rand.nextDouble(); opt.setSoftRateLimit(doubleValue); assertThat(opt.softRateLimit()).isEqualTo(doubleValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void hardRateLimit() { Options opt = null; try { opt = new Options(); double doubleValue = rand.nextDouble(); opt.setHardRateLimit(doubleValue); assertThat(opt.hardRateLimit()).isEqualTo(doubleValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void rateLimitDelayMaxMilliseconds() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setRateLimitDelayMaxMilliseconds(intValue); assertThat(opt.rateLimitDelayMaxMilliseconds()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void arenaBlockSize() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setArenaBlockSize(longValue); assertThat(opt.arenaBlockSize()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void disableAutoCompactions() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setDisableAutoCompactions(boolValue); assertThat(opt.disableAutoCompactions()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void purgeRedundantKvsWhileFlush() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setPurgeRedundantKvsWhileFlush(boolValue); assertThat(opt.purgeRedundantKvsWhileFlush()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void verifyChecksumsInCompaction() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setVerifyChecksumsInCompaction(boolValue); assertThat(opt.verifyChecksumsInCompaction()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void filterDeletes() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setFilterDeletes(boolValue); assertThat(opt.filterDeletes()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxSequentialSkipInIterations() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setMaxSequentialSkipInIterations(longValue); assertThat(opt.maxSequentialSkipInIterations()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void inplaceUpdateSupport() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setInplaceUpdateSupport(boolValue); assertThat(opt.inplaceUpdateSupport()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void inplaceUpdateNumLocks() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setInplaceUpdateNumLocks(longValue); assertThat(opt.inplaceUpdateNumLocks()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void memtablePrefixBloomBits() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMemtablePrefixBloomBits(intValue); assertThat(opt.memtablePrefixBloomBits()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void memtablePrefixBloomProbes() { Options opt = null; try { int intValue = rand.nextInt(); opt = new Options(); opt.setMemtablePrefixBloomProbes(intValue); assertThat(opt.memtablePrefixBloomProbes()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void bloomLocality() { Options opt = null; try { int intValue = rand.nextInt(); opt = new Options(); opt.setBloomLocality(intValue); assertThat(opt.bloomLocality()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxSuccessiveMerges() throws RocksDBException { Options opt = null; try { long longValue = rand.nextLong(); opt = new Options(); opt.setMaxSuccessiveMerges(longValue); assertThat(opt.maxSuccessiveMerges()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void minPartialMergeOperands() { Options opt = null; try { int intValue = rand.nextInt(); opt = new Options(); opt.setMinPartialMergeOperands(intValue); assertThat(opt.minPartialMergeOperands()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void optimizeFiltersForHits() { Options opt = null; try { boolean aBoolean = rand.nextBoolean(); opt = new Options(); opt.setOptimizeFiltersForHits(aBoolean); assertThat(opt.optimizeFiltersForHits()).isEqualTo(aBoolean); } finally { if (opt != null) { opt.dispose(); } } } @Test public void createIfMissing() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setCreateIfMissing(boolValue); assertThat(opt.createIfMissing()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void createMissingColumnFamilies() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setCreateMissingColumnFamilies(boolValue); assertThat(opt.createMissingColumnFamilies()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void errorIfExists() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setErrorIfExists(boolValue); assertThat(opt.errorIfExists()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void paranoidChecks() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setParanoidChecks(boolValue); assertThat(opt.paranoidChecks()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxTotalWalSize() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setMaxTotalWalSize(longValue); assertThat(opt.maxTotalWalSize()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxOpenFiles() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMaxOpenFiles(intValue); assertThat(opt.maxOpenFiles()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void disableDataSync() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setDisableDataSync(boolValue); assertThat(opt.disableDataSync()). isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void useFsync() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setUseFsync(boolValue); assertThat(opt.useFsync()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void dbLogDir() { Options opt = null; try { opt = new Options(); String str = "path/to/DbLogDir"; opt.setDbLogDir(str); assertThat(opt.dbLogDir()).isEqualTo(str); } finally { if (opt != null) { opt.dispose(); } } } @Test public void walDir() { Options opt = null; try { opt = new Options(); String str = "path/to/WalDir"; opt.setWalDir(str); assertThat(opt.walDir()).isEqualTo(str); } finally { if (opt != null) { opt.dispose(); } } } @Test public void deleteObsoleteFilesPeriodMicros() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setDeleteObsoleteFilesPeriodMicros(longValue); assertThat(opt.deleteObsoleteFilesPeriodMicros()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBackgroundCompactions() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMaxBackgroundCompactions(intValue); assertThat(opt.maxBackgroundCompactions()). isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxBackgroundFlushes() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setMaxBackgroundFlushes(intValue); assertThat(opt.maxBackgroundFlushes()). isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxLogFileSize() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setMaxLogFileSize(longValue); assertThat(opt.maxLogFileSize()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void logFileTimeToRoll() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setLogFileTimeToRoll(longValue); assertThat(opt.logFileTimeToRoll()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void keepLogFileNum() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setKeepLogFileNum(longValue); assertThat(opt.keepLogFileNum()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void maxManifestFileSize() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setMaxManifestFileSize(longValue); assertThat(opt.maxManifestFileSize()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void tableCacheNumshardbits() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setTableCacheNumshardbits(intValue); assertThat(opt.tableCacheNumshardbits()). isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void walSizeLimitMB() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setWalSizeLimitMB(longValue); assertThat(opt.walSizeLimitMB()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void walTtlSeconds() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setWalTtlSeconds(longValue); assertThat(opt.walTtlSeconds()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void manifestPreallocationSize() throws RocksDBException { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setManifestPreallocationSize(longValue); assertThat(opt.manifestPreallocationSize()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void allowOsBuffer() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setAllowOsBuffer(boolValue); assertThat(opt.allowOsBuffer()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void allowMmapReads() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setAllowMmapReads(boolValue); assertThat(opt.allowMmapReads()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void allowMmapWrites() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setAllowMmapWrites(boolValue); assertThat(opt.allowMmapWrites()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void isFdCloseOnExec() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setIsFdCloseOnExec(boolValue); assertThat(opt.isFdCloseOnExec()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void statsDumpPeriodSec() { Options opt = null; try { opt = new Options(); int intValue = rand.nextInt(); opt.setStatsDumpPeriodSec(intValue); assertThat(opt.statsDumpPeriodSec()).isEqualTo(intValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void adviseRandomOnOpen() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setAdviseRandomOnOpen(boolValue); assertThat(opt.adviseRandomOnOpen()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void useAdaptiveMutex() { Options opt = null; try { opt = new Options(); boolean boolValue = rand.nextBoolean(); opt.setUseAdaptiveMutex(boolValue); assertThat(opt.useAdaptiveMutex()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void bytesPerSync() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); opt.setBytesPerSync(longValue); assertThat(opt.bytesPerSync()).isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void env() { Options options = null; try { options = new Options(); Env env = Env.getDefault(); options.setEnv(env); assertThat(options.getEnv()).isSameAs(env); } finally { if (options != null) { options.dispose(); } } } @Test public void linkageOfPrepMethods() { Options options = null; try { options = new Options(); options.optimizeUniversalStyleCompaction(); options.optimizeUniversalStyleCompaction(4000); options.optimizeLevelStyleCompaction(); options.optimizeLevelStyleCompaction(3000); options.optimizeForPointLookup(10); options.prepareForBulkLoad(); } finally { if (options != null) { options.dispose(); } } } @Test public void compressionTypes() { Options options = null; try { options = new Options(); for (CompressionType compressionType : CompressionType.values()) { options.setCompressionType(compressionType); assertThat(options.compressionType()). isEqualTo(compressionType); assertThat(CompressionType.valueOf("NO_COMPRESSION")). isEqualTo(CompressionType.NO_COMPRESSION); } } finally { if (options != null) { options.dispose(); } } } @Test public void compressionPerLevel() { ColumnFamilyOptions columnFamilyOptions = null; try { columnFamilyOptions = new ColumnFamilyOptions(); assertThat(columnFamilyOptions.compressionPerLevel()).isEmpty(); List compressionTypeList = new ArrayList<>(); for (int i=0; i < columnFamilyOptions.numLevels(); i++) { compressionTypeList.add(CompressionType.NO_COMPRESSION); } columnFamilyOptions.setCompressionPerLevel(compressionTypeList); compressionTypeList = columnFamilyOptions.compressionPerLevel(); for (final CompressionType compressionType : compressionTypeList) { assertThat(compressionType).isEqualTo( CompressionType.NO_COMPRESSION); } } finally { if (columnFamilyOptions != null) { columnFamilyOptions.dispose(); } } } @Test public void differentCompressionsPerLevel() { ColumnFamilyOptions columnFamilyOptions = null; try { columnFamilyOptions = new ColumnFamilyOptions(); columnFamilyOptions.setNumLevels(3); assertThat(columnFamilyOptions.compressionPerLevel()).isEmpty(); List compressionTypeList = new ArrayList<>(); compressionTypeList.add(CompressionType.BZLIB2_COMPRESSION); compressionTypeList.add(CompressionType.SNAPPY_COMPRESSION); compressionTypeList.add(CompressionType.LZ4_COMPRESSION); columnFamilyOptions.setCompressionPerLevel(compressionTypeList); compressionTypeList = columnFamilyOptions.compressionPerLevel(); assertThat(compressionTypeList.size()).isEqualTo(3); assertThat(compressionTypeList). containsExactly( CompressionType.BZLIB2_COMPRESSION, CompressionType.SNAPPY_COMPRESSION, CompressionType.LZ4_COMPRESSION); } finally { if (columnFamilyOptions != null) { columnFamilyOptions.dispose(); } } } @Test public void compactionStyles() { Options options = null; try { options = new Options(); for (CompactionStyle compactionStyle : CompactionStyle.values()) { options.setCompactionStyle(compactionStyle); assertThat(options.compactionStyle()). isEqualTo(compactionStyle); assertThat(CompactionStyle.valueOf("FIFO")). isEqualTo(CompactionStyle.FIFO); } } finally { if (options != null) { options.dispose(); } } } @Test public void maxTableFilesSizeFIFO() { Options opt = null; try { opt = new Options(); long longValue = rand.nextLong(); // Size has to be positive longValue = (longValue < 0) ? -longValue : longValue; longValue = (longValue == 0) ? longValue + 1 : longValue; opt.setMaxTableFilesSizeFIFO(longValue); assertThat(opt.maxTableFilesSizeFIFO()). isEqualTo(longValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void rateLimiterConfig() { Options options = null; Options anotherOptions = null; RateLimiterConfig rateLimiterConfig; try { options = new Options(); rateLimiterConfig = new GenericRateLimiterConfig(1000, 100 * 1000, 1); options.setRateLimiterConfig(rateLimiterConfig); // Test with parameter initialization anotherOptions = new Options(); anotherOptions.setRateLimiterConfig( new GenericRateLimiterConfig(1000)); } finally { if (options != null) { options.dispose(); } if (anotherOptions != null) { anotherOptions.dispose(); } } } @Test public void shouldSetTestPrefixExtractor() { Options options = null; try { options = new Options(); options.useFixedLengthPrefixExtractor(100); options.useFixedLengthPrefixExtractor(10); } finally { if (options != null) { options.dispose(); } } } @Test public void shouldSetTestCappedPrefixExtractor() { Options options = null; try { options = new Options(); options.useCappedPrefixExtractor(100); options.useCappedPrefixExtractor(10); } finally { if (options != null) { options.dispose(); } } } @Test public void shouldTestMemTableFactoryName() throws RocksDBException { Options options = null; try { options = new Options(); options.setMemTableConfig(new VectorMemTableConfig()); assertThat(options.memTableFactoryName()). isEqualTo("VectorRepFactory"); options.setMemTableConfig( new HashLinkedListMemTableConfig()); assertThat(options.memTableFactoryName()). isEqualTo("HashLinkedListRepFactory"); } finally { if (options != null) { options.dispose(); } } } @Test public void statistics() { Options options = null; Options anotherOptions = null; try { options = new Options(); Statistics statistics = options.createStatistics(). statisticsPtr(); assertThat(statistics).isNotNull(); anotherOptions = new Options(); statistics = anotherOptions.statisticsPtr(); assertThat(statistics).isNotNull(); } finally { if (options != null) { options.dispose(); } if (anotherOptions != null) { anotherOptions.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/PlainTableConfigTest.java000066400000000000000000000053161267530634000256550ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class PlainTableConfigTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void keySize() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setKeySize(5); assertThat(plainTableConfig.keySize()). isEqualTo(5); } @Test public void bloomBitsPerKey() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setBloomBitsPerKey(11); assertThat(plainTableConfig.bloomBitsPerKey()). isEqualTo(11); } @Test public void hashTableRatio() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setHashTableRatio(0.95); assertThat(plainTableConfig.hashTableRatio()). isEqualTo(0.95); } @Test public void indexSparseness() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setIndexSparseness(18); assertThat(plainTableConfig.indexSparseness()). isEqualTo(18); } @Test public void hugePageTlbSize() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setHugePageTlbSize(1); assertThat(plainTableConfig.hugePageTlbSize()). isEqualTo(1); } @Test public void encodingType() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setEncodingType(EncodingType.kPrefix); assertThat(plainTableConfig.encodingType()).isEqualTo( EncodingType.kPrefix); } @Test public void fullScanMode() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setFullScanMode(true); assertThat(plainTableConfig.fullScanMode()).isTrue(); } @Test public void storeIndexInFile() { PlainTableConfig plainTableConfig = new PlainTableConfig(); plainTableConfig.setStoreIndexInFile(true); assertThat(plainTableConfig.storeIndexInFile()). isTrue(); } @Test public void plainTableConfig() { Options opt = null; try { opt = new Options(); PlainTableConfig plainTableConfig = new PlainTableConfig(); opt.setTableFormatConfig(plainTableConfig); assertThat(opt.tableFactoryName()).isEqualTo("PlainTable"); } finally { if (opt != null) { opt.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/PlatformRandomHelper.java000066400000000000000000000032521267530634000257360ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.Random; /** * Helper class to get the appropriate Random class instance dependent * on the current platform architecture (32bit vs 64bit) */ public class PlatformRandomHelper { /** * Determine if OS is 32-Bit/64-Bit * * @return boolean value indicating if operating system is 64 Bit. */ public static boolean isOs64Bit(){ boolean is64Bit; if (System.getProperty("os.name").contains("Windows")) { is64Bit = (System.getenv("ProgramFiles(x86)") != null); } else { is64Bit = (System.getProperty("os.arch").contains("64")); } return is64Bit; } /** * Factory to get a platform specific Random instance * * @return {@link java.util.Random} instance. */ public static Random getPlatformSpecificRandomFactory(){ if (isOs64Bit()) { return new Random(); } return new Random32Bit(); } /** * Random32Bit is a class which overrides {@code nextLong} to * provide random numbers which fit in size_t. This workaround * is necessary because there is no unsigned_int < Java 8 */ private static class Random32Bit extends Random { @Override public long nextLong(){ return this.nextInt(Integer.MAX_VALUE); } } /** * Utility class constructor */ private PlatformRandomHelper() { } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/ReadOnlyTest.java000066400000000000000000000257731267530634000242420ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.ArrayList; import java.util.List; import static org.assertj.core.api.Assertions.assertThat; public class ReadOnlyTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void readOnlyOpen() throws RocksDBException { RocksDB db = null; RocksDB db2 = null; RocksDB db3 = null; Options options = null; List columnFamilyHandleList = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); List readOnlyColumnFamilyHandleList2 = new ArrayList<>(); try { options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); db2 = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath()); assertThat("value"). isEqualTo(new String(db2.get("key".getBytes()))); db.close(); db2.close(); List cfDescriptors = new ArrayList<>(); cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); db = RocksDB.open( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, columnFamilyHandleList); columnFamilyHandleList.add(db.createColumnFamily( new ColumnFamilyDescriptor("new_cf".getBytes(), new ColumnFamilyOptions()))); columnFamilyHandleList.add(db.createColumnFamily( new ColumnFamilyDescriptor("new_cf2".getBytes(), new ColumnFamilyOptions()))); db.put(columnFamilyHandleList.get(2), "key2".getBytes(), "value2".getBytes()); db2 = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); assertThat(db2.get("key2".getBytes())).isNull(); assertThat(db2.get(readOnlyColumnFamilyHandleList.get(0), "key2".getBytes())). isNull(); cfDescriptors.clear(); cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); cfDescriptors.add( new ColumnFamilyDescriptor("new_cf2".getBytes(), new ColumnFamilyOptions())); db3 = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList2); assertThat(new String(db3.get(readOnlyColumnFamilyHandleList2.get(1), "key2".getBytes()))).isEqualTo("value2"); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db2 != null) { db2.close(); } for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList2) { columnFamilyHandle.dispose(); } if (db3 != null) { db3.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failToWriteInReadOnly() throws RocksDBException { RocksDB db = null; RocksDB rDb = null; Options options = null; List cfDescriptors = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); try { cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.close(); rDb = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); // test that put fails in readonly mode rDb.put("key".getBytes(), "value".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (rDb != null) { rDb.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failToCFWriteInReadOnly() throws RocksDBException { RocksDB db = null; RocksDB rDb = null; Options options = null; List cfDescriptors = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); try { cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.close(); rDb = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); rDb.put(readOnlyColumnFamilyHandleList.get(0), "key".getBytes(), "value".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (rDb != null) { rDb.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failToRemoveInReadOnly() throws RocksDBException { RocksDB db = null; RocksDB rDb = null; Options options = null; List cfDescriptors = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); try { cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.close(); rDb = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); rDb.remove("key".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (rDb != null) { rDb.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failToCFRemoveInReadOnly() throws RocksDBException { RocksDB db = null; RocksDB rDb = null; Options options = null; List cfDescriptors = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); try { cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.close(); rDb = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); rDb.remove(readOnlyColumnFamilyHandleList.get(0), "key".getBytes()); } finally { for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (rDb != null) { rDb.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failToWriteBatchReadOnly() throws RocksDBException { RocksDB db = null; RocksDB rDb = null; Options options = null; List cfDescriptors = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); try { cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.close(); rDb = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); WriteBatch wb = new WriteBatch(); wb.put("key".getBytes(), "value".getBytes()); rDb.write(new WriteOptions(), wb); } finally { for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (rDb != null) { rDb.close(); } if (options != null) { options.dispose(); } } } @Test(expected = RocksDBException.class) public void failToCFWriteBatchReadOnly() throws RocksDBException { RocksDB db = null; RocksDB rDb = null; Options options = null; WriteBatch wb = null; List cfDescriptors = new ArrayList<>(); List readOnlyColumnFamilyHandleList = new ArrayList<>(); try { cfDescriptors.add( new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY, new ColumnFamilyOptions())); options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.close(); rDb = RocksDB.openReadOnly( dbFolder.getRoot().getAbsolutePath(), cfDescriptors, readOnlyColumnFamilyHandleList); wb = new WriteBatch(); wb.put(readOnlyColumnFamilyHandleList.get(0), "key".getBytes(), "value".getBytes()); rDb.write(new WriteOptions(), wb); } finally { for (ColumnFamilyHandle columnFamilyHandle : readOnlyColumnFamilyHandleList) { columnFamilyHandle.dispose(); } if (db != null) { db.close(); } if (rDb != null) { rDb.close(); } if (options != null) { options.dispose(); } if (wb != null) { wb.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/ReadOptionsTest.java000066400000000000000000000071631267530634000247450ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.Random; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.ExpectedException; import static org.assertj.core.api.Assertions.assertThat; public class ReadOptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public ExpectedException exception = ExpectedException.none(); @Test public void verifyChecksum(){ ReadOptions opt = null; try { opt = new ReadOptions(); Random rand = new Random(); boolean boolValue = rand.nextBoolean(); opt.setVerifyChecksums(boolValue); assertThat(opt.verifyChecksums()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void fillCache(){ ReadOptions opt = null; try { opt = new ReadOptions(); Random rand = new Random(); boolean boolValue = rand.nextBoolean(); opt.setFillCache(boolValue); assertThat(opt.fillCache()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void tailing(){ ReadOptions opt = null; try { opt = new ReadOptions(); Random rand = new Random(); boolean boolValue = rand.nextBoolean(); opt.setTailing(boolValue); assertThat(opt.tailing()).isEqualTo(boolValue); } finally { if (opt != null) { opt.dispose(); } } } @Test public void snapshot(){ ReadOptions opt = null; try { opt = new ReadOptions(); opt.setSnapshot(null); assertThat(opt.snapshot()).isNull(); } finally { if (opt != null) { opt.dispose(); } } } @Test public void failSetVerifyChecksumUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.setVerifyChecksums(true); } @Test public void failVerifyChecksumUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.verifyChecksums(); } @Test public void failSetFillCacheUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.setFillCache(true); } @Test public void failFillCacheUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.fillCache(); } @Test public void failSetTailingUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.setTailing(true); } @Test public void failTailingUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.tailing(); } @Test public void failSetSnapshotUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.setSnapshot(null); } @Test public void failSnapshotUninitialized(){ ReadOptions readOptions = setupUninitializedReadOptions( exception); readOptions.snapshot(); } private ReadOptions setupUninitializedReadOptions( ExpectedException exception) { ReadOptions readOptions = new ReadOptions(); readOptions.dispose(); exception.expect(AssertionError.class); return readOptions; } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/RocksDBTest.java000066400000000000000000000644451267530634000240130ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.ArrayList; import java.util.List; import java.util.Map; import java.util.Random; import static org.assertj.core.api.Assertions.assertThat; public class RocksDBTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); public static final Random rand = PlatformRandomHelper. getPlatformSpecificRandomFactory(); @Test public void open() throws RocksDBException { RocksDB db = null; Options opt = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); db.close(); opt = new Options(); opt.setCreateIfMissing(true); db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void put() throws RocksDBException { RocksDB db = null; WriteOptions opt = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value".getBytes()); opt = new WriteOptions(); db.put(opt, "key2".getBytes(), "12345678".getBytes()); assertThat(db.get("key1".getBytes())).isEqualTo( "value".getBytes()); assertThat(db.get("key2".getBytes())).isEqualTo( "12345678".getBytes()); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void write() throws RocksDBException { RocksDB db = null; Options options = null; WriteBatch wb1 = null; WriteBatch wb2 = null; WriteOptions opts = null; try { options = new Options(). setMergeOperator(new StringAppendOperator()). setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); opts = new WriteOptions(); wb1 = new WriteBatch(); wb1.put("key1".getBytes(), "aa".getBytes()); wb1.merge("key1".getBytes(), "bb".getBytes()); wb2 = new WriteBatch(); wb2.put("key2".getBytes(), "xx".getBytes()); wb2.merge("key2".getBytes(), "yy".getBytes()); db.write(opts, wb1); db.write(opts, wb2); assertThat(db.get("key1".getBytes())).isEqualTo( "aa,bb".getBytes()); assertThat(db.get("key2".getBytes())).isEqualTo( "xx,yy".getBytes()); } finally { if (db != null) { db.close(); } if (wb1 != null) { wb1.dispose(); } if (wb2 != null) { wb2.dispose(); } if (options != null) { options.dispose(); } if (opts != null) { opts.dispose(); } } } @Test public void getWithOutValue() throws RocksDBException { RocksDB db = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value".getBytes()); db.put("key2".getBytes(), "12345678".getBytes()); byte[] outValue = new byte[5]; // not found value int getResult = db.get("keyNotFound".getBytes(), outValue); assertThat(getResult).isEqualTo(RocksDB.NOT_FOUND); // found value which fits in outValue getResult = db.get("key1".getBytes(), outValue); assertThat(getResult).isNotEqualTo(RocksDB.NOT_FOUND); assertThat(outValue).isEqualTo("value".getBytes()); // found value which fits partially getResult = db.get("key2".getBytes(), outValue); assertThat(getResult).isNotEqualTo(RocksDB.NOT_FOUND); assertThat(outValue).isEqualTo("12345".getBytes()); } finally { if (db != null) { db.close(); } } } @Test public void getWithOutValueReadOptions() throws RocksDBException { RocksDB db = null; ReadOptions rOpt = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); rOpt = new ReadOptions(); db.put("key1".getBytes(), "value".getBytes()); db.put("key2".getBytes(), "12345678".getBytes()); byte[] outValue = new byte[5]; // not found value int getResult = db.get(rOpt, "keyNotFound".getBytes(), outValue); assertThat(getResult).isEqualTo(RocksDB.NOT_FOUND); // found value which fits in outValue getResult = db.get(rOpt, "key1".getBytes(), outValue); assertThat(getResult).isNotEqualTo(RocksDB.NOT_FOUND); assertThat(outValue).isEqualTo("value".getBytes()); // found value which fits partially getResult = db.get(rOpt, "key2".getBytes(), outValue); assertThat(getResult).isNotEqualTo(RocksDB.NOT_FOUND); assertThat(outValue).isEqualTo("12345".getBytes()); } finally { if (db != null) { db.close(); } if (rOpt != null) { rOpt.dispose(); } } } @Test public void multiGet() throws RocksDBException { RocksDB db = null; ReadOptions rOpt = null; try { db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); rOpt = new ReadOptions(); db.put("key1".getBytes(), "value".getBytes()); db.put("key2".getBytes(), "12345678".getBytes()); List lookupKeys = new ArrayList() {{ add("key1".getBytes()); add("key2".getBytes()); }}; Map results = db.multiGet(lookupKeys); assertThat(results).isNotNull(); assertThat(results.values()).isNotNull(); assertThat(results.values()). contains("value".getBytes(), "12345678".getBytes()); // test same method with ReadOptions results = db.multiGet(rOpt, lookupKeys); assertThat(results).isNotNull(); assertThat(results.values()).isNotNull(); assertThat(results.values()). contains("value".getBytes(), "12345678".getBytes()); // remove existing key lookupKeys.remove("key2".getBytes()); // add non existing key lookupKeys.add("key3".getBytes()); results = db.multiGet(lookupKeys); assertThat(results).isNotNull(); assertThat(results.values()).isNotNull(); assertThat(results.values()). contains("value".getBytes()); // test same call with readOptions results = db.multiGet(rOpt, lookupKeys); assertThat(results).isNotNull(); assertThat(results.values()).isNotNull(); assertThat(results.values()). contains("value".getBytes()); } finally { if (db != null) { db.close(); } if (rOpt != null) { rOpt.dispose(); } } } @Test public void merge() throws RocksDBException { RocksDB db = null; Options opt = null; WriteOptions wOpt; try { opt = new Options(). setCreateIfMissing(true). setMergeOperator(new StringAppendOperator()); wOpt = new WriteOptions(); db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value".getBytes()); assertThat(db.get("key1".getBytes())).isEqualTo( "value".getBytes()); // merge key1 with another value portion db.merge("key1".getBytes(), "value2".getBytes()); assertThat(db.get("key1".getBytes())).isEqualTo( "value,value2".getBytes()); // merge key1 with another value portion db.merge(wOpt, "key1".getBytes(), "value3".getBytes()); assertThat(db.get("key1".getBytes())).isEqualTo( "value,value2,value3".getBytes()); // merge on non existent key shall insert the value db.merge(wOpt, "key2".getBytes(), "xxxx".getBytes()); assertThat(db.get("key2".getBytes())).isEqualTo( "xxxx".getBytes()); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void remove() throws RocksDBException { RocksDB db = null; WriteOptions wOpt; try { wOpt = new WriteOptions(); db = RocksDB.open(dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value".getBytes()); db.put("key2".getBytes(), "12345678".getBytes()); assertThat(db.get("key1".getBytes())).isEqualTo( "value".getBytes()); assertThat(db.get("key2".getBytes())).isEqualTo( "12345678".getBytes()); db.remove("key1".getBytes()); db.remove(wOpt, "key2".getBytes()); assertThat(db.get("key1".getBytes())).isNull(); assertThat(db.get("key2".getBytes())).isNull(); } finally { if (db != null) { db.close(); } } } @Test public void getIntProperty() throws RocksDBException { RocksDB db = null; Options options = null; WriteOptions wOpt = null; try { options = new Options(); wOpt = new WriteOptions(); // Setup options options.setCreateIfMissing(true); options.setMaxWriteBufferNumber(10); options.setMinWriteBufferNumberToMerge(10); wOpt.setDisableWAL(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put(wOpt, "key1".getBytes(), "value1".getBytes()); db.put(wOpt, "key2".getBytes(), "value2".getBytes()); db.put(wOpt, "key3".getBytes(), "value3".getBytes()); db.put(wOpt, "key4".getBytes(), "value4".getBytes()); assertThat(db.getLongProperty("rocksdb.num-entries-active-mem-table")).isGreaterThan(0); assertThat(db.getLongProperty("rocksdb.cur-size-active-mem-table")).isGreaterThan(0); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (wOpt != null) { wOpt.dispose(); } } } @Test public void fullCompactRange() throws RocksDBException { RocksDB db = null; Options opt = null; try { opt = new Options(). setCreateIfMissing(true). setDisableAutoCompactions(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(4). setWriteBufferSize(100<<10). setLevelZeroFileNumCompactionTrigger(3). setTargetFileSizeBase(200 << 10). setTargetFileSizeMultiplier(1). setMaxBytesForLevelBase(500 << 10). setMaxBytesForLevelMultiplier(1). setDisableAutoCompactions(false); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // fill database with key/value pairs byte[] b = new byte[10000]; for (int i = 0; i < 200; i++) { rand.nextBytes(b); db.put((String.valueOf(i)).getBytes(), b); } db.compactRange(); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void fullCompactRangeColumnFamily() throws RocksDBException { RocksDB db = null; DBOptions opt = null; List columnFamilyHandles = new ArrayList<>(); try { opt = new DBOptions(). setCreateIfMissing(true). setCreateMissingColumnFamilies(true); List columnFamilyDescriptors = new ArrayList<>(); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( RocksDB.DEFAULT_COLUMN_FAMILY)); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( "new_cf".getBytes(), new ColumnFamilyOptions(). setDisableAutoCompactions(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(4). setWriteBufferSize(100 << 10). setLevelZeroFileNumCompactionTrigger(3). setTargetFileSizeBase(200 << 10). setTargetFileSizeMultiplier(1). setMaxBytesForLevelBase(500 << 10). setMaxBytesForLevelMultiplier(1). setDisableAutoCompactions(false))); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath(), columnFamilyDescriptors, columnFamilyHandles); // fill database with key/value pairs byte[] b = new byte[10000]; for (int i = 0; i < 200; i++) { rand.nextBytes(b); db.put(columnFamilyHandles.get(1), String.valueOf(i).getBytes(), b); } db.compactRange(columnFamilyHandles.get(1)); } finally { for (ColumnFamilyHandle handle : columnFamilyHandles) { handle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void compactRangeWithKeys() throws RocksDBException { RocksDB db = null; Options opt = null; try { opt = new Options(). setCreateIfMissing(true). setDisableAutoCompactions(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(4). setWriteBufferSize(100<<10). setLevelZeroFileNumCompactionTrigger(3). setTargetFileSizeBase(200 << 10). setTargetFileSizeMultiplier(1). setMaxBytesForLevelBase(500 << 10). setMaxBytesForLevelMultiplier(1). setDisableAutoCompactions(false); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // fill database with key/value pairs byte[] b = new byte[10000]; for (int i = 0; i < 200; i++) { rand.nextBytes(b); db.put((String.valueOf(i)).getBytes(), b); } db.compactRange("0".getBytes(), "201".getBytes()); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void compactRangeWithKeysReduce() throws RocksDBException { RocksDB db = null; Options opt = null; try { opt = new Options(). setCreateIfMissing(true). setDisableAutoCompactions(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(4). setWriteBufferSize(100<<10). setLevelZeroFileNumCompactionTrigger(3). setTargetFileSizeBase(200 << 10). setTargetFileSizeMultiplier(1). setMaxBytesForLevelBase(500 << 10). setMaxBytesForLevelMultiplier(1). setDisableAutoCompactions(false); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // fill database with key/value pairs byte[] b = new byte[10000]; for (int i = 0; i < 200; i++) { rand.nextBytes(b); db.put((String.valueOf(i)).getBytes(), b); } db.flush(new FlushOptions().setWaitForFlush(true)); db.compactRange("0".getBytes(), "201".getBytes(), true, -1, 0); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void compactRangeWithKeysColumnFamily() throws RocksDBException { RocksDB db = null; DBOptions opt = null; List columnFamilyHandles = new ArrayList<>(); try { opt = new DBOptions(). setCreateIfMissing(true). setCreateMissingColumnFamilies(true); List columnFamilyDescriptors = new ArrayList<>(); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( RocksDB.DEFAULT_COLUMN_FAMILY)); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( "new_cf".getBytes(), new ColumnFamilyOptions(). setDisableAutoCompactions(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(4). setWriteBufferSize(100<<10). setLevelZeroFileNumCompactionTrigger(3). setTargetFileSizeBase(200 << 10). setTargetFileSizeMultiplier(1). setMaxBytesForLevelBase(500 << 10). setMaxBytesForLevelMultiplier(1). setDisableAutoCompactions(false))); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath(), columnFamilyDescriptors, columnFamilyHandles); // fill database with key/value pairs byte[] b = new byte[10000]; for (int i = 0; i < 200; i++) { rand.nextBytes(b); db.put(columnFamilyHandles.get(1), String.valueOf(i).getBytes(), b); } db.compactRange(columnFamilyHandles.get(1), "0".getBytes(), "201".getBytes()); } finally { for (ColumnFamilyHandle handle : columnFamilyHandles) { handle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void compactRangeWithKeysReduceColumnFamily() throws RocksDBException { RocksDB db = null; DBOptions opt = null; List columnFamilyHandles = new ArrayList<>(); try { opt = new DBOptions(). setCreateIfMissing(true). setCreateMissingColumnFamilies(true); List columnFamilyDescriptors = new ArrayList<>(); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( RocksDB.DEFAULT_COLUMN_FAMILY)); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( "new_cf".getBytes(), new ColumnFamilyOptions(). setDisableAutoCompactions(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(4). setWriteBufferSize(100<<10). setLevelZeroFileNumCompactionTrigger(3). setTargetFileSizeBase(200 << 10). setTargetFileSizeMultiplier(1). setMaxBytesForLevelBase(500 << 10). setMaxBytesForLevelMultiplier(1). setDisableAutoCompactions(false))); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath(), columnFamilyDescriptors, columnFamilyHandles); // fill database with key/value pairs byte[] b = new byte[10000]; for (int i = 0; i < 200; i++) { rand.nextBytes(b); db.put(columnFamilyHandles.get(1), String.valueOf(i).getBytes(), b); } db.compactRange(columnFamilyHandles.get(1), "0".getBytes(), "201".getBytes(), true, -1, 0); } finally { for (ColumnFamilyHandle handle : columnFamilyHandles) { handle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void compactRangeToLevel() throws RocksDBException, InterruptedException { RocksDB db = null; Options opt = null; try { final int NUM_KEYS_PER_L0_FILE = 100; final int KEY_SIZE = 20; final int VALUE_SIZE = 300; final int L0_FILE_SIZE = NUM_KEYS_PER_L0_FILE * (KEY_SIZE + VALUE_SIZE); final int NUM_L0_FILES = 10; final int TEST_SCALE = 5; final int KEY_INTERVAL = 100; opt = new Options(). setCreateIfMissing(true). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(5). // a slightly bigger write buffer than L0 file // so that we can ensure manual flush always // go before background flush happens. setWriteBufferSize(L0_FILE_SIZE * 2). // Disable auto L0 -> L1 compaction setLevelZeroFileNumCompactionTrigger(20). setTargetFileSizeBase(L0_FILE_SIZE * 100). setTargetFileSizeMultiplier(1). // To disable auto compaction setMaxBytesForLevelBase(NUM_L0_FILES * L0_FILE_SIZE * 100). setMaxBytesForLevelMultiplier(2). setDisableAutoCompactions(true); db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); // fill database with key/value pairs byte[] value = new byte[VALUE_SIZE]; int int_key = 0; for (int round = 0; round < 5; ++round) { int initial_key = int_key; for (int f = 1; f <= NUM_L0_FILES; ++f) { for (int i = 0; i < NUM_KEYS_PER_L0_FILE; ++i) { int_key += KEY_INTERVAL; rand.nextBytes(value); db.put(String.format("%020d", int_key).getBytes(), value); } db.flush(new FlushOptions().setWaitForFlush(true)); // Make sure we do create one more L0 files. assertThat( db.getProperty("rocksdb.num-files-at-level0")). isEqualTo("" + f); } // Compact all L0 files we just created db.compactRange( String.format("%020d", initial_key).getBytes(), String.format("%020d", int_key - 1).getBytes()); // Making sure there isn't any L0 files. assertThat( db.getProperty("rocksdb.num-files-at-level0")). isEqualTo("0"); // Making sure there are some L1 files. // Here we only use != 0 instead of a specific number // as we don't want the test make any assumption on // how compaction works. assertThat( db.getProperty("rocksdb.num-files-at-level1")). isNotEqualTo("0"); // Because we only compacted those keys we issued // in this round, there shouldn't be any L1 -> L2 // compaction. So we expect zero L2 files here. assertThat( db.getProperty("rocksdb.num-files-at-level2")). isEqualTo("0"); } } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void compactRangeToLevelColumnFamily() throws RocksDBException { RocksDB db = null; DBOptions opt = null; List columnFamilyHandles = new ArrayList<>(); try { final int NUM_KEYS_PER_L0_FILE = 100; final int KEY_SIZE = 20; final int VALUE_SIZE = 300; final int L0_FILE_SIZE = NUM_KEYS_PER_L0_FILE * (KEY_SIZE + VALUE_SIZE); final int NUM_L0_FILES = 10; final int TEST_SCALE = 5; final int KEY_INTERVAL = 100; opt = new DBOptions(). setCreateIfMissing(true). setCreateMissingColumnFamilies(true); List columnFamilyDescriptors = new ArrayList<>(); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( RocksDB.DEFAULT_COLUMN_FAMILY)); columnFamilyDescriptors.add(new ColumnFamilyDescriptor( "new_cf".getBytes(), new ColumnFamilyOptions(). setCompactionStyle(CompactionStyle.LEVEL). setNumLevels(5). // a slightly bigger write buffer than L0 file // so that we can ensure manual flush always // go before background flush happens. setWriteBufferSize(L0_FILE_SIZE * 2). // Disable auto L0 -> L1 compaction setLevelZeroFileNumCompactionTrigger(20). setTargetFileSizeBase(L0_FILE_SIZE * 100). setTargetFileSizeMultiplier(1). // To disable auto compaction setMaxBytesForLevelBase(NUM_L0_FILES * L0_FILE_SIZE * 100). setMaxBytesForLevelMultiplier(2). setDisableAutoCompactions(true))); // open database db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath(), columnFamilyDescriptors, columnFamilyHandles); // fill database with key/value pairs byte[] value = new byte[VALUE_SIZE]; int int_key = 0; for (int round = 0; round < 5; ++round) { int initial_key = int_key; for (int f = 1; f <= NUM_L0_FILES; ++f) { for (int i = 0; i < NUM_KEYS_PER_L0_FILE; ++i) { int_key += KEY_INTERVAL; rand.nextBytes(value); db.put(columnFamilyHandles.get(1), String.format("%020d", int_key).getBytes(), value); } db.flush(new FlushOptions().setWaitForFlush(true), columnFamilyHandles.get(1)); // Make sure we do create one more L0 files. assertThat( db.getProperty(columnFamilyHandles.get(1), "rocksdb.num-files-at-level0")). isEqualTo("" + f); } // Compact all L0 files we just created db.compactRange( columnFamilyHandles.get(1), String.format("%020d", initial_key).getBytes(), String.format("%020d", int_key - 1).getBytes()); // Making sure there isn't any L0 files. assertThat( db.getProperty(columnFamilyHandles.get(1), "rocksdb.num-files-at-level0")). isEqualTo("0"); // Making sure there are some L1 files. // Here we only use != 0 instead of a specific number // as we don't want the test make any assumption on // how compaction works. assertThat( db.getProperty(columnFamilyHandles.get(1), "rocksdb.num-files-at-level1")). isNotEqualTo("0"); // Because we only compacted those keys we issued // in this round, there shouldn't be any L1 -> L2 // compaction. So we expect zero L2 files here. assertThat( db.getProperty(columnFamilyHandles.get(1), "rocksdb.num-files-at-level2")). isEqualTo("0"); } } finally { for (ColumnFamilyHandle handle : columnFamilyHandles) { handle.dispose(); } if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } @Test public void enableDisableFileDeletions() throws RocksDBException { RocksDB db = null; Options options = null; try { options = new Options().setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.disableFileDeletions(); db.enableFileDeletions(false); db.disableFileDeletions(); db.enableFileDeletions(true); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/RocksEnvTest.java000066400000000000000000000026421267530634000242450ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class RocksEnvTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void rocksEnv(){ Env rocksEnv = RocksEnv.getDefault(); rocksEnv.setBackgroundThreads(5); // default rocksenv will always return zero for flush pool // no matter what was set via setBackgroundThreads assertThat(rocksEnv.getThreadPoolQueueLen(RocksEnv.FLUSH_POOL)). isEqualTo(0); rocksEnv.setBackgroundThreads(5, RocksEnv.FLUSH_POOL); // default rocksenv will always return zero for flush pool // no matter what was set via setBackgroundThreads assertThat(rocksEnv.getThreadPoolQueueLen(RocksEnv.FLUSH_POOL)). isEqualTo(0); rocksEnv.setBackgroundThreads(5, RocksEnv.COMPACTION_POOL); // default rocksenv will always return zero for compaction pool // no matter what was set via setBackgroundThreads assertThat(rocksEnv.getThreadPoolQueueLen(RocksEnv.COMPACTION_POOL)). isEqualTo(0); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/RocksIteratorTest.java000066400000000000000000000045151267530634000253070ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class RocksIteratorTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void rocksIterator() throws RocksDBException { RocksDB db = null; Options options = null; RocksIterator iterator = null; try { options = new Options(); options.setCreateIfMissing(true) .setCreateMissingColumnFamilies(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value1".getBytes()); db.put("key2".getBytes(), "value2".getBytes()); iterator = db.newIterator(); iterator.seekToFirst(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key1".getBytes()); assertThat(iterator.value()).isEqualTo("value1".getBytes()); iterator.next(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key2".getBytes()); assertThat(iterator.value()).isEqualTo("value2".getBytes()); iterator.next(); assertThat(iterator.isValid()).isFalse(); iterator.seekToLast(); iterator.prev(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key1".getBytes()); assertThat(iterator.value()).isEqualTo("value1".getBytes()); iterator.seekToFirst(); iterator.seekToLast(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key2".getBytes()); assertThat(iterator.value()).isEqualTo("value2".getBytes()); iterator.status(); } finally { if (iterator != null) { iterator.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/RocksMemEnvTest.java000066400000000000000000000117571267530634000247130ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class RocksMemEnvTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void memEnvFillAndReopen() throws RocksDBException { final byte[][] keys = { "aaa".getBytes(), "bbb".getBytes(), "ccc".getBytes() }; final byte[][] values = { "foo".getBytes(), "bar".getBytes(), "baz".getBytes() }; Env env = null; Options options = null; RocksDB db = null; FlushOptions flushOptions = null; try { env = new RocksMemEnv(); options = new Options(). setCreateIfMissing(true). setEnv(env); flushOptions = new FlushOptions(). setWaitForFlush(true); db = RocksDB.open(options, "dir/db"); // write key/value pairs using MemEnv for (int i=0; i < keys.length; i++) { db.put(keys[i], values[i]); } // read key/value pairs using MemEnv for (int i=0; i < keys.length; i++) { assertThat(db.get(keys[i])).isEqualTo(values[i]); } // Check iterator access RocksIterator iterator = db.newIterator(); iterator.seekToFirst(); for (int i=0; i < keys.length; i++) { assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo(keys[i]); assertThat(iterator.value()).isEqualTo(values[i]); iterator.next(); } // reached end of database assertThat(iterator.isValid()).isFalse(); iterator.dispose(); // flush db.flush(flushOptions); // read key/value pairs after flush using MemEnv for (int i=0; i < keys.length; i++) { assertThat(db.get(keys[i])).isEqualTo(values[i]); } db.close(); options.setCreateIfMissing(false); // After reopen the values shall still be in the mem env. // as long as the env is not freed. db = RocksDB.open(options, "dir/db"); // read key/value pairs using MemEnv for (int i=0; i < keys.length; i++) { assertThat(db.get(keys[i])).isEqualTo(values[i]); } } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (flushOptions != null) { flushOptions.dispose(); } if (env != null) { env.dispose(); } } } @Test public void multipleDatabaseInstances() throws RocksDBException { // db - keys final byte[][] keys = { "aaa".getBytes(), "bbb".getBytes(), "ccc".getBytes() }; // otherDb - keys final byte[][] otherKeys = { "111".getBytes(), "222".getBytes(), "333".getBytes() }; // values final byte[][] values = { "foo".getBytes(), "bar".getBytes(), "baz".getBytes() }; Env env = null; Options options = null; RocksDB db = null, otherDb = null; try { env = new RocksMemEnv(); options = new Options(). setCreateIfMissing(true). setEnv(env); db = RocksDB.open(options, "dir/db"); otherDb = RocksDB.open(options, "dir/otherDb"); // write key/value pairs using MemEnv // to db and to otherDb. for (int i=0; i < keys.length; i++) { db.put(keys[i], values[i]); otherDb.put(otherKeys[i], values[i]); } // verify key/value pairs after flush using MemEnv for (int i=0; i < keys.length; i++) { // verify db assertThat(db.get(otherKeys[i])).isNull(); assertThat(db.get(keys[i])).isEqualTo(values[i]); // verify otherDb assertThat(otherDb.get(keys[i])).isNull(); assertThat(otherDb.get(otherKeys[i])).isEqualTo(values[i]); } } finally { if (db != null) { db.close(); } if (otherDb != null) { otherDb.close(); } if (options != null) { options.dispose(); } if (env != null) { env.dispose(); } } } @Test(expected = RocksDBException.class) public void createIfMissingFalse() throws RocksDBException { Env env = null; Options options = null; RocksDB db = null; try { env = new RocksMemEnv(); options = new Options(). setCreateIfMissing(false). setEnv(env); // shall throw an exception because db dir does not // exist. db = RocksDB.open(options, "db/dir"); } finally { if (options != null) { options.dispose(); } if (env != null) { env.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/RocksMemoryResource.java000066400000000000000000000005161267530634000256330ustar00rootroot00000000000000package org.rocksdb; import org.junit.rules.ExternalResource; /** * Resource to trigger garbage collection after each test * run. */ public class RocksMemoryResource extends ExternalResource { static { RocksDB.loadLibrary(); } @Override protected void after() { System.gc(); System.runFinalization(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/SliceTest.java000066400000000000000000000051371267530634000235540ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class SliceTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void slice() { Slice slice = null; Slice otherSlice = null; Slice thirdSlice = null; try { slice = new Slice("testSlice"); assertThat(slice.empty()).isFalse(); assertThat(slice.size()).isEqualTo(9); assertThat(slice.data()).isEqualTo("testSlice".getBytes()); otherSlice = new Slice("otherSlice".getBytes()); assertThat(otherSlice.data()).isEqualTo("otherSlice".getBytes()); thirdSlice = new Slice("otherSlice".getBytes(), 5); assertThat(thirdSlice.data()).isEqualTo("Slice".getBytes()); } finally { if (slice != null) { slice.dispose(); } if (otherSlice != null) { otherSlice.dispose(); } if (thirdSlice != null) { thirdSlice.dispose(); } } } @Test public void sliceEquals() { Slice slice = null; Slice slice2 = null; try { slice = new Slice("abc"); slice2 = new Slice("abc"); assertThat(slice.equals(slice2)).isTrue(); assertThat(slice.hashCode() == slice2.hashCode()).isTrue(); } finally { if (slice != null) { slice.dispose(); } if (slice2 != null) { slice2.dispose(); } } } @Test public void sliceStartWith() { Slice slice = null; Slice match = null; Slice noMatch = null; try { slice = new Slice("matchpoint"); match = new Slice("mat"); noMatch = new Slice("nomatch"); //assertThat(slice.startsWith(match)).isTrue(); assertThat(slice.startsWith(noMatch)).isFalse(); } finally { if (slice != null) { slice.dispose(); } if (match != null) { match.dispose(); } if (noMatch != null) { noMatch.dispose(); } } } @Test public void sliceToString() { Slice slice = null; try { slice = new Slice("stringTest"); assertThat(slice.toString()).isEqualTo("stringTest"); assertThat(slice.toString(true)).isNotEqualTo(""); } finally { if (slice != null) { slice.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/SnapshotTest.java000066400000000000000000000161031267530634000243070ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class SnapshotTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void snapshots() throws RocksDBException { RocksDB db = null; Options options = null; ReadOptions readOptions = null; try { options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); // Get new Snapshot of database Snapshot snapshot = db.getSnapshot(); assertThat(snapshot.getSequenceNumber()).isGreaterThan(0); assertThat(snapshot.getSequenceNumber()).isEqualTo(1); readOptions = new ReadOptions(); // set snapshot in ReadOptions readOptions.setSnapshot(snapshot); // retrieve key value pair assertThat(new String(db.get("key".getBytes()))). isEqualTo("value"); // retrieve key value pair created before // the snapshot was made assertThat(new String(db.get(readOptions, "key".getBytes()))).isEqualTo("value"); // add new key/value pair db.put("newkey".getBytes(), "newvalue".getBytes()); // using no snapshot the latest db entries // will be taken into account assertThat(new String(db.get("newkey".getBytes()))). isEqualTo("newvalue"); // snapshopot was created before newkey assertThat(db.get(readOptions, "newkey".getBytes())). isNull(); // Retrieve snapshot from read options Snapshot sameSnapshot = readOptions.snapshot(); readOptions.setSnapshot(sameSnapshot); // results must be the same with new Snapshot // instance using the same native pointer assertThat(new String(db.get(readOptions, "key".getBytes()))).isEqualTo("value"); // update key value pair to newvalue db.put("key".getBytes(), "newvalue".getBytes()); // read with previously created snapshot will // read previous version of key value pair assertThat(new String(db.get(readOptions, "key".getBytes()))).isEqualTo("value"); // read for newkey using the snapshot must be // null assertThat(db.get(readOptions, "newkey".getBytes())). isNull(); // setting null to snapshot in ReadOptions leads // to no Snapshot being used. readOptions.setSnapshot(null); assertThat(new String(db.get(readOptions, "newkey".getBytes()))).isEqualTo("newvalue"); // release Snapshot db.releaseSnapshot(snapshot); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (readOptions != null) { readOptions.dispose(); } } } @Test public void iteratorWithSnapshot() throws RocksDBException { RocksDB db = null; Options options = null; ReadOptions readOptions = null; RocksIterator iterator = null; RocksIterator snapshotIterator = null; try { options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); // Get new Snapshot of database Snapshot snapshot = db.getSnapshot(); readOptions = new ReadOptions(); // set snapshot in ReadOptions readOptions.setSnapshot(snapshot); db.put("key2".getBytes(), "value2".getBytes()); // iterate over current state of db iterator = db.newIterator(); iterator.seekToFirst(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key".getBytes()); iterator.next(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key2".getBytes()); iterator.next(); assertThat(iterator.isValid()).isFalse(); // iterate using a snapshot snapshotIterator = db.newIterator(readOptions); snapshotIterator.seekToFirst(); assertThat(snapshotIterator.isValid()).isTrue(); assertThat(snapshotIterator.key()).isEqualTo("key".getBytes()); snapshotIterator.next(); assertThat(snapshotIterator.isValid()).isFalse(); // release Snapshot db.releaseSnapshot(snapshot); } finally { if (iterator != null) { iterator.dispose(); } if (snapshotIterator != null) { snapshotIterator.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (readOptions != null) { readOptions.dispose(); } } } @Test public void iteratorWithSnapshotOnColumnFamily() throws RocksDBException { RocksDB db = null; Options options = null; ReadOptions readOptions = null; RocksIterator iterator = null; RocksIterator snapshotIterator = null; try { options = new Options(); options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key".getBytes(), "value".getBytes()); // Get new Snapshot of database Snapshot snapshot = db.getSnapshot(); readOptions = new ReadOptions(); // set snapshot in ReadOptions readOptions.setSnapshot(snapshot); db.put("key2".getBytes(), "value2".getBytes()); // iterate over current state of column family iterator = db.newIterator(db.getDefaultColumnFamily()); iterator.seekToFirst(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key".getBytes()); iterator.next(); assertThat(iterator.isValid()).isTrue(); assertThat(iterator.key()).isEqualTo("key2".getBytes()); iterator.next(); assertThat(iterator.isValid()).isFalse(); // iterate using a snapshot on default column family snapshotIterator = db.newIterator(db.getDefaultColumnFamily(), readOptions); snapshotIterator.seekToFirst(); assertThat(snapshotIterator.isValid()).isTrue(); assertThat(snapshotIterator.key()).isEqualTo("key".getBytes()); snapshotIterator.next(); assertThat(snapshotIterator.isValid()).isFalse(); // release Snapshot db.releaseSnapshot(snapshot); } finally { if (iterator != null) { iterator.dispose(); } if (snapshotIterator != null) { snapshotIterator.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } if (readOptions != null) { readOptions.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/StatisticsCollectorTest.java000066400000000000000000000032601267530634000265110ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.util.Collections; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class StatisticsCollectorTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void statisticsCollector() throws InterruptedException, RocksDBException { Options opt = null; RocksDB db = null; try { opt = new Options().createStatistics().setCreateIfMissing(true); Statistics stats = opt.statisticsPtr(); db = RocksDB.open(opt, dbFolder.getRoot().getAbsolutePath()); StatsCallbackMock callback = new StatsCallbackMock(); StatsCollectorInput statsInput = new StatsCollectorInput(stats, callback); StatisticsCollector statsCollector = new StatisticsCollector( Collections.singletonList(statsInput), 100); statsCollector.start(); Thread.sleep(1000); assertThat(callback.tickerCallbackCount).isGreaterThan(0); assertThat(callback.histCallbackCount).isGreaterThan(0); statsCollector.shutDown(1000); } finally { if (db != null) { db.close(); } if (opt != null) { opt.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/StatsCallbackMock.java000066400000000000000000000012511267530634000251730ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; public class StatsCallbackMock implements StatisticsCollectorCallback { public int tickerCallbackCount = 0; public int histCallbackCount = 0; public void tickerCallback(TickerType tickerType, long tickerCount) { tickerCallbackCount++; } public void histogramCallback(HistogramType histType, HistogramData histData) { histCallbackCount++; } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/TransactionLogIteratorTest.java000066400000000000000000000127011267530634000271510ustar00rootroot00000000000000package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import static org.assertj.core.api.Assertions.assertThat; public class TransactionLogIteratorTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void transactionLogIterator() throws RocksDBException { RocksDB db = null; Options options = null; TransactionLogIterator transactionLogIterator = null; try { options = new Options(). setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); transactionLogIterator = db.getUpdatesSince(0); } finally { if (transactionLogIterator != null) { transactionLogIterator.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void getBatch() throws RocksDBException { final int numberOfPuts = 5; RocksDB db = null; Options options = null; ColumnFamilyHandle cfHandle = null; TransactionLogIterator transactionLogIterator = null; try { options = new Options(). setCreateIfMissing(true). setWalTtlSeconds(1000). setWalSizeLimitMB(10); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); for (int i = 0; i < numberOfPuts; i++){ db.put(String.valueOf(i).getBytes(), String.valueOf(i).getBytes()); } db.flush(new FlushOptions().setWaitForFlush(true)); // the latest sequence number is 5 because 5 puts // were written beforehand assertThat(db.getLatestSequenceNumber()). isEqualTo(numberOfPuts); // insert 5 writes into a cf cfHandle = db.createColumnFamily( new ColumnFamilyDescriptor("new_cf".getBytes())); for (int i = 0; i < numberOfPuts; i++){ db.put(cfHandle, String.valueOf(i).getBytes(), String.valueOf(i).getBytes()); } // the latest sequence number is 10 because // (5 + 5) puts were written beforehand assertThat(db.getLatestSequenceNumber()). isEqualTo(numberOfPuts + numberOfPuts); // Get updates since the beginning transactionLogIterator = db.getUpdatesSince(0); assertThat(transactionLogIterator.isValid()).isTrue(); transactionLogIterator.status(); // The first sequence number is 1 TransactionLogIterator.BatchResult batchResult = transactionLogIterator.getBatch(); assertThat(batchResult.sequenceNumber()).isEqualTo(1); } finally { if (transactionLogIterator != null) { transactionLogIterator.dispose(); } if (cfHandle != null) { cfHandle.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void transactionLogIteratorStallAtLastRecord() throws RocksDBException { RocksDB db = null; Options options = null; TransactionLogIterator transactionLogIterator = null; try { options = new Options(). setCreateIfMissing(true). setWalTtlSeconds(1000). setWalSizeLimitMB(10); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value1".getBytes()); // Get updates since the beginning transactionLogIterator = db.getUpdatesSince(0); transactionLogIterator.status(); assertThat(transactionLogIterator.isValid()).isTrue(); transactionLogIterator.next(); assertThat(transactionLogIterator.isValid()).isFalse(); transactionLogIterator.status(); db.put("key2".getBytes(), "value2".getBytes()); transactionLogIterator.next(); transactionLogIterator.status(); assertThat(transactionLogIterator.isValid()).isTrue(); } finally { if (transactionLogIterator != null) { transactionLogIterator.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void transactionLogIteratorCheckAfterRestart() throws RocksDBException { final int numberOfKeys = 2; RocksDB db = null; Options options = null; TransactionLogIterator transactionLogIterator = null; try { options = new Options(). setCreateIfMissing(true). setWalTtlSeconds(1000). setWalSizeLimitMB(10); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); db.put("key1".getBytes(), "value1".getBytes()); db.put("key2".getBytes(), "value2".getBytes()); db.flush(new FlushOptions().setWaitForFlush(true)); // reopen db.close(); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); assertThat(db.getLatestSequenceNumber()).isEqualTo(numberOfKeys); transactionLogIterator = db.getUpdatesSince(0); for (int i = 0; i < numberOfKeys; i++) { transactionLogIterator.status(); assertThat(transactionLogIterator.isValid()).isTrue(); transactionLogIterator.next(); } } finally { if (transactionLogIterator != null) { transactionLogIterator.dispose(); } if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/TtlDBTest.java000066400000000000000000000120131267530634000234550ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.util.ArrayList; import java.util.List; import java.util.concurrent.TimeUnit; import static org.assertj.core.api.Assertions.assertThat; public class TtlDBTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void ttlDBOpen() throws RocksDBException, InterruptedException { Options options = null; TtlDB ttlDB = null; try { options = new Options(). setCreateIfMissing(true). setMaxGrandparentOverlapFactor(0); ttlDB = TtlDB.open(options, dbFolder.getRoot().getAbsolutePath()); ttlDB.put("key".getBytes(), "value".getBytes()); assertThat(ttlDB.get("key".getBytes())). isEqualTo("value".getBytes()); assertThat(ttlDB.get("key".getBytes())).isNotNull(); } finally { if (ttlDB != null) { ttlDB.close(); } if (options != null) { options.dispose(); } } } @Test public void ttlDBOpenWithTtl() throws RocksDBException, InterruptedException { Options options = null; TtlDB ttlDB = null; try { options = new Options(). setCreateIfMissing(true). setMaxGrandparentOverlapFactor(0); ttlDB = TtlDB.open(options, dbFolder.getRoot().getAbsolutePath(), 1, false); ttlDB.put("key".getBytes(), "value".getBytes()); assertThat(ttlDB.get("key".getBytes())). isEqualTo("value".getBytes()); TimeUnit.SECONDS.sleep(2); ttlDB.compactRange(); assertThat(ttlDB.get("key".getBytes())).isNull(); } finally { if (ttlDB != null) { ttlDB.close(); } if (options != null) { options.dispose(); } } } @Test public void ttlDbOpenWithColumnFamilies() throws RocksDBException, InterruptedException { DBOptions dbOptions = null; TtlDB ttlDB = null; List cfNames = new ArrayList<>(); List columnFamilyHandleList = new ArrayList<>(); cfNames.add(new ColumnFamilyDescriptor(RocksDB.DEFAULT_COLUMN_FAMILY)); cfNames.add(new ColumnFamilyDescriptor("new_cf".getBytes())); List ttlValues = new ArrayList<>(); // Default column family with infinite lifetime ttlValues.add(0); // new column family with 1 second ttl ttlValues.add(1); try { dbOptions = new DBOptions(). setCreateMissingColumnFamilies(true). setCreateIfMissing(true); ttlDB = TtlDB.open(dbOptions, dbFolder.getRoot().getAbsolutePath(), cfNames, columnFamilyHandleList, ttlValues, false); ttlDB.put("key".getBytes(), "value".getBytes()); assertThat(ttlDB.get("key".getBytes())). isEqualTo("value".getBytes()); ttlDB.put(columnFamilyHandleList.get(1), "key".getBytes(), "value".getBytes()); assertThat(ttlDB.get(columnFamilyHandleList.get(1), "key".getBytes())).isEqualTo("value".getBytes()); TimeUnit.SECONDS.sleep(2); ttlDB.compactRange(); ttlDB.compactRange(columnFamilyHandleList.get(1)); assertThat(ttlDB.get("key".getBytes())).isNotNull(); assertThat(ttlDB.get(columnFamilyHandleList.get(1), "key".getBytes())).isNull(); } finally { for (ColumnFamilyHandle columnFamilyHandle : columnFamilyHandleList) { columnFamilyHandle.dispose(); } if (ttlDB != null) { ttlDB.close(); } if (dbOptions != null) { dbOptions.dispose(); } } } @Test public void createTtlColumnFamily() throws RocksDBException, InterruptedException { Options options = null; TtlDB ttlDB = null; ColumnFamilyHandle columnFamilyHandle = null; try { options = new Options().setCreateIfMissing(true); ttlDB = TtlDB.open(options, dbFolder.getRoot().getAbsolutePath()); columnFamilyHandle = ttlDB.createColumnFamilyWithTtl( new ColumnFamilyDescriptor("new_cf".getBytes()), 1); ttlDB.put(columnFamilyHandle, "key".getBytes(), "value".getBytes()); assertThat(ttlDB.get(columnFamilyHandle, "key".getBytes())). isEqualTo("value".getBytes()); TimeUnit.SECONDS.sleep(2); ttlDB.compactRange(columnFamilyHandle); assertThat(ttlDB.get(columnFamilyHandle, "key".getBytes())).isNull(); } finally { if (columnFamilyHandle != null) { columnFamilyHandle.dispose(); } if (ttlDB != null) { ttlDB.close(); } if (options != null) { options.dispose(); } } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/Types.java000066400000000000000000000020411267530634000227500ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; /** * Simple type conversion methods * for use in tests */ public class Types { /** * Convert first 4 bytes of a byte array to an int * * @param data The byte array * * @return An integer */ public static int byteToInt(final byte data[]) { return (data[0] & 0xff) | ((data[1] & 0xff) << 8) | ((data[2] & 0xff) << 16) | ((data[3] & 0xff) << 24); } /** * Convert an int to 4 bytes * * @param v The int * * @return A byte array containing 4 bytes */ public static byte[] intToByte(final int v) { return new byte[] { (byte)((v >>> 0) & 0xff), (byte)((v >>> 8) & 0xff), (byte)((v >>> 16) & 0xff), (byte)((v >>> 24) & 0xff) }; } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/WriteBatchHandlerTest.java000066400000000000000000000117531267530634000260500ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import java.io.IOException; import java.util.ArrayList; import java.util.Arrays; import java.util.List; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class WriteBatchHandlerTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void writeBatchHandler() throws IOException, RocksDBException { WriteBatch batch = null; CapturingWriteBatchHandler handler = null; try { // setup test data final List>> testEvents = new ArrayList<>(); testEvents.add(new Tuple<>(Action.DELETE, new Tuple("k0".getBytes(), null))); testEvents.add(new Tuple<>(Action.PUT, new Tuple<>("k1".getBytes(), "v1".getBytes()))); testEvents.add(new Tuple<>(Action.PUT, new Tuple<>("k2".getBytes(), "v2".getBytes()))); testEvents.add(new Tuple<>(Action.PUT, new Tuple<>("k3".getBytes(), "v3".getBytes()))); testEvents.add(new Tuple<>(Action.LOG, new Tuple(null, "log1".getBytes()))); testEvents.add(new Tuple<>(Action.MERGE, new Tuple<>("k2".getBytes(), "v22".getBytes()))); testEvents.add(new Tuple<>(Action.DELETE, new Tuple("k3".getBytes(), null))); // load test data to the write batch batch = new WriteBatch(); for (final Tuple> testEvent : testEvents) { final Tuple data = testEvent.value; switch (testEvent.key) { case PUT: batch.put(data.key, data.value); break; case MERGE: batch.merge(data.key, data.value); break; case DELETE: batch.remove(data.key); break; case LOG: batch.putLogData(data.value); break; } } // attempt to read test data back from the WriteBatch by iterating with a handler handler = new CapturingWriteBatchHandler(); batch.iterate(handler); // compare the results to the test data final List>> actualEvents = handler.getEvents(); assertThat(testEvents.size()).isSameAs(actualEvents.size()); for (int i = 0; i < testEvents.size(); i++) { assertThat(equals(testEvents.get(i), actualEvents.get(i))).isTrue(); } } finally { if (handler != null) { handler.dispose(); } if (batch != null) { batch.dispose(); } } } private static boolean equals(final Tuple> expected, final Tuple> actual) { if (!expected.key.equals(actual.key)) { return false; } final Tuple expectedData = expected.value; final Tuple actualData = actual.value; return equals(expectedData.key, actualData.key) && equals(expectedData.value, actualData.value); } private static boolean equals(byte[] expected, byte[] actual) { if (expected != null) { return Arrays.equals(expected, actual); } else { return actual == null; } } private static class Tuple { public final K key; public final V value; public Tuple(final K key, final V value) { this.key = key; this.value = value; } } /** * Enumeration of Write Batch * event actions */ private enum Action { PUT, MERGE, DELETE, LOG } /** * A simple WriteBatch Handler which adds a record * of each event that it receives to a list */ private static class CapturingWriteBatchHandler extends WriteBatch.Handler { private final List>> events = new ArrayList<>(); /** * Returns a copy of the current events list * * @return a list of the events which have happened upto now */ public List>> getEvents() { return new ArrayList<>(events); } @Override public void put(final byte[] key, final byte[] value) { events.add(new Tuple<>(Action.PUT, new Tuple<>(key, value))); } @Override public void merge(final byte[] key, final byte[] value) { events.add(new Tuple<>(Action.MERGE, new Tuple<>(key, value))); } @Override public void delete(final byte[] key) { events.add(new Tuple<>(Action.DELETE, new Tuple(key, null))); } @Override public void logData(final byte[] blob) { events.add(new Tuple<>(Action.LOG, new Tuple(null, blob))); } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/WriteBatchTest.java000066400000000000000000000104741267530634000245510ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.io.UnsupportedEncodingException; import static org.assertj.core.api.Assertions.assertThat; /** * This class mimics the db/write_batch_test.cc * in the c++ rocksdb library. * * Not ported yet: * * Continue(); * PutGatherSlices(); */ public class WriteBatchTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void emptyWriteBatch() { WriteBatch batch = new WriteBatch(); assertThat(batch.count()).isEqualTo(0); } @Test public void multipleBatchOperations() throws UnsupportedEncodingException { WriteBatch batch = new WriteBatch(); batch.put("foo".getBytes("US-ASCII"), "bar".getBytes("US-ASCII")); batch.remove("box".getBytes("US-ASCII")); batch.put("baz".getBytes("US-ASCII"), "boo".getBytes("US-ASCII")); WriteBatchTestInternalHelper.setSequence(batch, 100); assertThat(WriteBatchTestInternalHelper.sequence(batch)). isNotNull(). isEqualTo(100); assertThat(batch.count()).isEqualTo(3); assertThat(new String(getContents(batch), "US-ASCII")). isEqualTo("Put(baz, boo)@102" + "Delete(box)@101" + "Put(foo, bar)@100"); } @Test public void testAppendOperation() throws UnsupportedEncodingException { WriteBatch b1 = new WriteBatch(); WriteBatch b2 = new WriteBatch(); WriteBatchTestInternalHelper.setSequence(b1, 200); WriteBatchTestInternalHelper.setSequence(b2, 300); WriteBatchTestInternalHelper.append(b1, b2); assertThat(getContents(b1).length).isEqualTo(0); assertThat(b1.count()).isEqualTo(0); b2.put("a".getBytes("US-ASCII"), "va".getBytes("US-ASCII")); WriteBatchTestInternalHelper.append(b1, b2); assertThat("Put(a, va)@200".equals(new String(getContents(b1), "US-ASCII"))); assertThat(b1.count()).isEqualTo(1); b2.clear(); b2.put("b".getBytes("US-ASCII"), "vb".getBytes("US-ASCII")); WriteBatchTestInternalHelper.append(b1, b2); assertThat(("Put(a, va)@200" + "Put(b, vb)@201") .equals(new String(getContents(b1), "US-ASCII"))); assertThat(b1.count()).isEqualTo(2); b2.remove("foo".getBytes("US-ASCII")); WriteBatchTestInternalHelper.append(b1, b2); assertThat(("Put(a, va)@200" + "Put(b, vb)@202" + "Put(b, vb)@201" + "Delete(foo)@203") .equals(new String(getContents(b1), "US-ASCII"))); assertThat(b1.count()).isEqualTo(4); } @Test public void blobOperation() throws UnsupportedEncodingException { WriteBatch batch = new WriteBatch(); batch.put("k1".getBytes("US-ASCII"), "v1".getBytes("US-ASCII")); batch.put("k2".getBytes("US-ASCII"), "v2".getBytes("US-ASCII")); batch.put("k3".getBytes("US-ASCII"), "v3".getBytes("US-ASCII")); batch.putLogData("blob1".getBytes("US-ASCII")); batch.remove("k2".getBytes("US-ASCII")); batch.putLogData("blob2".getBytes("US-ASCII")); batch.merge("foo".getBytes("US-ASCII"), "bar".getBytes("US-ASCII")); assertThat(batch.count()).isEqualTo(5); assertThat(("Merge(foo, bar)@4" + "Put(k1, v1)@0" + "Delete(k2)@3" + "Put(k2, v2)@1" + "Put(k3, v3)@2") .equals(new String(getContents(batch), "US-ASCII"))); } static native byte[] getContents(WriteBatch batch); } /** * Package-private class which provides java api to access * c++ WriteBatchInternal. */ class WriteBatchTestInternalHelper { static native void setSequence(WriteBatch batch, long sn); static native long sequence(WriteBatch batch); static native void append(WriteBatch b1, WriteBatch b2); } rocksdb-4.5.1/java/src/test/java/org/rocksdb/WriteBatchThreadedTest.java000066400000000000000000000057111267530634000262100ustar00rootroot00000000000000// Copyright (c) 2016, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.After; import org.junit.Before; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import org.junit.runner.RunWith; import org.junit.runners.Parameterized; import org.junit.runners.Parameterized.Parameter; import org.junit.runners.Parameterized.Parameters; import java.nio.ByteBuffer; import java.util.*; import java.util.concurrent.*; @RunWith(Parameterized.class) public class WriteBatchThreadedTest { @Parameters(name = "WriteBatchThreadedTest(threadCount={0})") public static Iterable data() { return Arrays.asList(new Integer[]{1, 10, 50, 100}); } @Parameter public int threadCount; @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); RocksDB db; @Before public void setUp() throws Exception { RocksDB.loadLibrary(); final Options options = new Options() .setCreateIfMissing(true) .setIncreaseParallelism(32); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); assert (db != null); } @After public void tearDown() throws Exception { if (db != null) { db.close(); } } @Test public void threadedWrites() throws InterruptedException, ExecutionException { final List> callables = new ArrayList<>(); for (int i = 0; i < 100; i++) { final int offset = i * 100; callables.add(new Callable() { @Override public Void call() throws RocksDBException { final WriteBatch wb = new WriteBatch(); for (int i = offset; i < offset + 100; i++) { wb.put(ByteBuffer.allocate(4).putInt(i).array(), "parallel rocks test".getBytes()); } db.write(new WriteOptions(), wb); return null; } }); } //submit the callables final ExecutorService executorService = Executors.newFixedThreadPool(threadCount); try { final ExecutorCompletionService completionService = new ExecutorCompletionService<>(executorService); final Set> futures = new HashSet<>(); for (final Callable callable : callables) { futures.add(completionService.submit(callable)); } while (futures.size() > 0) { final Future future = completionService.take(); futures.remove(future); try { future.get(); } catch (final ExecutionException e) { for (final Future f : futures) { f.cancel(true); } throw e; } } } finally { executorService.shutdown(); executorService.awaitTermination(10, TimeUnit.SECONDS); } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/WriteBatchWithIndexTest.java000066400000000000000000000171101267530634000263670ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. package org.rocksdb; import org.junit.ClassRule; import org.junit.Rule; import org.junit.Test; import org.junit.rules.TemporaryFolder; import java.nio.ByteBuffer; import java.util.ArrayDeque; import java.util.Deque; import static org.assertj.core.api.Assertions.assertThat; public class WriteBatchWithIndexTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Rule public TemporaryFolder dbFolder = new TemporaryFolder(); @Test public void readYourOwnWrites() throws RocksDBException { RocksDB db = null; Options options = null; try { options = new Options(); // Setup options options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); final byte[] k1 = "key1".getBytes(); final byte[] v1 = "value1".getBytes(); final byte[] k2 = "key2".getBytes(); final byte[] v2 = "value2".getBytes(); db.put(k1, v1); db.put(k2, v2); final WriteBatchWithIndex wbwi = new WriteBatchWithIndex(true); RocksIterator base = null; RocksIterator it = null; try { base = db.newIterator(); it = wbwi.newIteratorWithBase(base); it.seek(k1); assertThat(it.isValid()).isTrue(); assertThat(it.key()).isEqualTo(k1); assertThat(it.value()).isEqualTo(v1); it.seek(k2); assertThat(it.isValid()).isTrue(); assertThat(it.key()).isEqualTo(k2); assertThat(it.value()).isEqualTo(v2); //put data to the write batch and make sure we can read it. final byte[] k3 = "key3".getBytes(); final byte[] v3 = "value3".getBytes(); wbwi.put(k3, v3); it.seek(k3); assertThat(it.isValid()).isTrue(); assertThat(it.key()).isEqualTo(k3); assertThat(it.value()).isEqualTo(v3); //update k2 in the write batch and check the value final byte[] v2Other = "otherValue2".getBytes(); wbwi.put(k2, v2Other); it.seek(k2); assertThat(it.isValid()).isTrue(); assertThat(it.key()).isEqualTo(k2); assertThat(it.value()).isEqualTo(v2Other); //remove k1 and make sure we can read back the write wbwi.remove(k1); it.seek(k1); assertThat(it.key()).isNotEqualTo(k1); //reinsert k1 and make sure we see the new value final byte[] v1Other = "otherValue1".getBytes(); wbwi.put(k1, v1Other); it.seek(k1); assertThat(it.isValid()).isTrue(); assertThat(it.key()).isEqualTo(k1); assertThat(it.value()).isEqualTo(v1Other); } finally { if (it != null) { it.dispose(); } if (base != null) { base.dispose(); } } } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void write_writeBatchWithIndex() throws RocksDBException { RocksDB db = null; Options options = null; try { options = new Options(); // Setup options options.setCreateIfMissing(true); db = RocksDB.open(options, dbFolder.getRoot().getAbsolutePath()); final byte[] k1 = "key1".getBytes(); final byte[] v1 = "value1".getBytes(); final byte[] k2 = "key2".getBytes(); final byte[] v2 = "value2".getBytes(); WriteBatchWithIndex wbwi = null; try { wbwi = new WriteBatchWithIndex(); wbwi.put(k1, v1); wbwi.put(k2, v2); db.write(new WriteOptions(), wbwi); } finally { if(wbwi != null) { wbwi.dispose(); } } assertThat(db.get(k1)).isEqualTo(v1); assertThat(db.get(k2)).isEqualTo(v2); } finally { if (db != null) { db.close(); } if (options != null) { options.dispose(); } } } @Test public void iterator() throws RocksDBException { final WriteBatchWithIndex wbwi = new WriteBatchWithIndex(true); final String k1 = "key1"; final String v1 = "value1"; final String k2 = "key2"; final String v2 = "value2"; final String k3 = "key3"; final String v3 = "value3"; final byte[] k1b = k1.getBytes(); final byte[] v1b = v1.getBytes(); final byte[] k2b = k2.getBytes(); final byte[] v2b = v2.getBytes(); final byte[] k3b = k3.getBytes(); final byte[] v3b = v3.getBytes(); //add put records wbwi.put(k1b, v1b); wbwi.put(k2b, v2b); wbwi.put(k3b, v3b); //add a deletion record final String k4 = "key4"; final byte[] k4b = k4.getBytes(); wbwi.remove(k4b); WBWIRocksIterator.WriteEntry[] expected = { new WBWIRocksIterator.WriteEntry(WBWIRocksIterator.WriteType.PUT, new DirectSlice(k1), new DirectSlice(v1)), new WBWIRocksIterator.WriteEntry(WBWIRocksIterator.WriteType.PUT, new DirectSlice(k2), new DirectSlice(v2)), new WBWIRocksIterator.WriteEntry(WBWIRocksIterator.WriteType.PUT, new DirectSlice(k3), new DirectSlice(v3)), new WBWIRocksIterator.WriteEntry(WBWIRocksIterator.WriteType.DELETE, new DirectSlice(k4), DirectSlice.NONE) }; WBWIRocksIterator it = null; try { it = wbwi.newIterator(); //direct access - seek to key offsets final int[] testOffsets = {2, 0, 1, 3}; for(int i = 0; i < testOffsets.length; i++) { final int testOffset = testOffsets[i]; final byte[] key = toArray(expected[testOffset].getKey().data()); it.seek(key); assertThat(it.isValid()).isTrue(); assertThat(it.entry().equals(expected[testOffset])).isTrue(); } //forward iterative access int i = 0; for(it.seekToFirst(); it.isValid(); it.next()) { assertThat(it.entry().equals(expected[i++])).isTrue(); } //reverse iterative access i = expected.length - 1; for(it.seekToLast(); it.isValid(); it.prev()) { assertThat(it.entry().equals(expected[i--])).isTrue(); } } finally { if(it != null) { it.dispose(); } } } @Test public void zeroByteTests() { final WriteBatchWithIndex wbwi = new WriteBatchWithIndex(true); byte[] zeroByteValue = new byte[] { 0, 0 }; //add zero byte value wbwi.put(zeroByteValue, zeroByteValue); ByteBuffer buffer = ByteBuffer.allocateDirect(zeroByteValue.length); buffer.put(zeroByteValue); WBWIRocksIterator.WriteEntry[] expected = { new WBWIRocksIterator.WriteEntry(WBWIRocksIterator.WriteType.PUT, new DirectSlice(buffer, zeroByteValue.length), new DirectSlice(buffer, zeroByteValue.length)) }; WBWIRocksIterator it = null; try { it = wbwi.newIterator(); it.seekToFirst(); assertThat(it.entry().equals(expected[0])).isTrue(); assertThat(it.entry().hashCode() == expected[0].hashCode()).isTrue(); } finally { if(it != null) { it.dispose(); } } } private byte[] toArray(final ByteBuffer buf) { final byte[] ary = new byte[buf.remaining()]; buf.get(ary); return ary; } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/WriteOptionsTest.java000066400000000000000000000017521267530634000251620ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb; import org.junit.ClassRule; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class WriteOptionsTest { @ClassRule public static final RocksMemoryResource rocksMemoryResource = new RocksMemoryResource(); @Test public void writeOptions(){ WriteOptions writeOptions = new WriteOptions(); writeOptions.setDisableWAL(true); assertThat(writeOptions.disableWAL()).isTrue(); writeOptions.setDisableWAL(false); assertThat(writeOptions.disableWAL()).isFalse(); writeOptions.setSync(true); assertThat(writeOptions.sync()).isTrue(); writeOptions.setSync(false); assertThat(writeOptions.sync()).isFalse(); } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/test/000077500000000000000000000000001267530634000217635ustar00rootroot00000000000000rocksdb-4.5.1/java/src/test/java/org/rocksdb/test/RocksJunitRunner.java000066400000000000000000000036071267530634000261210ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb.test; import org.junit.internal.JUnitSystem; import org.junit.internal.RealSystem; import org.junit.internal.TextListener; import org.junit.runner.Description; import org.junit.runner.JUnitCore; import org.junit.runner.Result; import java.util.ArrayList; import java.util.List; /** * Custom Junit Runner to print also Test classes * and executed methods to command prompt. */ public class RocksJunitRunner { /** * Listener which overrides default functionality * to print class and method to system out. */ static class RocksJunitListener extends TextListener { /** * RocksJunitListener constructor * * @param system JUnitSystem */ public RocksJunitListener(JUnitSystem system) { super(system); } @Override public void testStarted(Description description) { System.out.format("Run: %s testing now -> %s \n", description.getClassName(), description.getMethodName()); } } /** * Main method to execute tests * * @param args Test classes as String names */ public static void main(String[] args){ JUnitCore runner = new JUnitCore(); final JUnitSystem system = new RealSystem(); runner.addListener(new RocksJunitListener(system)); try { List> classes = new ArrayList<>(); for (String arg : args) { classes.add(Class.forName(arg)); } final Result result = runner.run(classes.toArray(new Class[1])); if(!result.wasSuccessful()) { System.exit(-1); } } catch (ClassNotFoundException e) { e.printStackTrace(); } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/util/000077500000000000000000000000001267530634000217615ustar00rootroot00000000000000rocksdb-4.5.1/java/src/test/java/org/rocksdb/util/EnvironmentTest.java000066400000000000000000000144521267530634000257760ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb.util; import org.junit.AfterClass; import org.junit.BeforeClass; import org.junit.Test; import java.lang.reflect.Field; import java.lang.reflect.Modifier; import static org.assertj.core.api.Assertions.assertThat; public class EnvironmentTest { private final static String ARCH_FIELD_NAME = "ARCH"; private final static String OS_FIELD_NAME = "OS"; private static String INITIAL_OS; private static String INITIAL_ARCH; @BeforeClass public static void saveState() { INITIAL_ARCH = getEnvironmentClassField(ARCH_FIELD_NAME); INITIAL_OS = getEnvironmentClassField(OS_FIELD_NAME); } @Test public void mac32() { setEnvironmentClassFields("mac", "32"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".jnilib"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-osx.jnilib"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.dylib"); } @Test public void mac64() { setEnvironmentClassFields("mac", "64"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".jnilib"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-osx.jnilib"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.dylib"); } @Test public void nix32() { // Linux setEnvironmentClassFields("Linux", "32"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".so"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-linux32.so"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.so"); // UNIX setEnvironmentClassFields("Unix", "32"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".so"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-linux32.so"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.so"); // AIX setEnvironmentClassFields("aix", "32"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".so"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-linux32.so"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.so"); } @Test public void nix64() { setEnvironmentClassFields("Linux", "x64"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".so"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-linux64.so"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.so"); // UNIX setEnvironmentClassFields("Unix", "x64"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".so"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-linux64.so"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.so"); // AIX setEnvironmentClassFields("aix", "x64"); assertThat(Environment.isWindows()).isFalse(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".so"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-linux64.so"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.so"); } @Test public void detectWindows(){ setEnvironmentClassFields("win", "x64"); assertThat(Environment.isWindows()).isTrue(); } @Test public void win64() { setEnvironmentClassFields("win", "x64"); assertThat(Environment.isWindows()).isTrue(); assertThat(Environment.getJniLibraryExtension()). isEqualTo(".dll"); assertThat(Environment.getJniLibraryFileName("rocksdb")). isEqualTo("librocksdbjni-win64.dll"); assertThat(Environment.getSharedLibraryFileName("rocksdb")). isEqualTo("librocksdbjni.dll"); } @Test(expected = UnsupportedOperationException.class) public void win32(){ setEnvironmentClassFields("win", "32"); Environment.getJniLibraryFileName("rocksdb"); } private void setEnvironmentClassFields(String osName, String osArch) { setEnvironmentClassField(OS_FIELD_NAME, osName); setEnvironmentClassField(ARCH_FIELD_NAME, osArch); } @AfterClass public static void restoreState() { setEnvironmentClassField(OS_FIELD_NAME, INITIAL_OS); setEnvironmentClassField(ARCH_FIELD_NAME, INITIAL_ARCH); } private static String getEnvironmentClassField(String fieldName) { final Field field; try { field = Environment.class.getDeclaredField(fieldName); field.setAccessible(true); final Field modifiersField = Field.class.getDeclaredField("modifiers"); modifiersField.setAccessible(true); modifiersField.setInt(field, field.getModifiers() & ~Modifier.FINAL); return (String)field.get(null); } catch (NoSuchFieldException | IllegalAccessException e) { throw new RuntimeException(e); } } private static void setEnvironmentClassField(String fieldName, String value) { final Field field; try { field = Environment.class.getDeclaredField(fieldName); field.setAccessible(true); final Field modifiersField = Field.class.getDeclaredField("modifiers"); modifiersField.setAccessible(true); modifiersField.setInt(field, field.getModifiers() & ~Modifier.FINAL); field.set(null, value); } catch (NoSuchFieldException | IllegalAccessException e) { throw new RuntimeException(e); } } } rocksdb-4.5.1/java/src/test/java/org/rocksdb/util/SizeUnitTest.java000066400000000000000000000016031267530634000252360ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. package org.rocksdb.util; import org.junit.Test; import static org.assertj.core.api.Assertions.assertThat; public class SizeUnitTest { public static final long COMPUTATION_UNIT = 1024L; @Test public void sizeUnit() { assertThat(SizeUnit.KB).isEqualTo(COMPUTATION_UNIT); assertThat(SizeUnit.MB).isEqualTo( SizeUnit.KB * COMPUTATION_UNIT); assertThat(SizeUnit.GB).isEqualTo( SizeUnit.MB * COMPUTATION_UNIT); assertThat(SizeUnit.TB).isEqualTo( SizeUnit.GB * COMPUTATION_UNIT); assertThat(SizeUnit.PB).isEqualTo( SizeUnit.TB * COMPUTATION_UNIT); } } rocksdb-4.5.1/memtable/000077500000000000000000000000001267530634000147445ustar00rootroot00000000000000rocksdb-4.5.1/memtable/hash_cuckoo_rep.cc000066400000000000000000000607511267530634000204200ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #ifndef ROCKSDB_LITE #include "memtable/hash_cuckoo_rep.h" #include #include #include #include #include #include #include #include "db/memtable.h" #include "db/skiplist.h" #include "memtable/stl_wrappers.h" #include "port/port.h" #include "rocksdb/memtablerep.h" #include "util/murmurhash.h" namespace rocksdb { namespace { // the default maximum size of the cuckoo path searching queue static const int kCuckooPathMaxSearchSteps = 100; struct CuckooStep { static const int kNullStep = -1; // the bucket id in the cuckoo array. int bucket_id_; // index of cuckoo-step array that points to its previous step, // -1 if it the beginning step. int prev_step_id_; // the depth of the current step. unsigned int depth_; CuckooStep() : bucket_id_(-1), prev_step_id_(kNullStep), depth_(1) {} // MSVC does not support = default yet CuckooStep(CuckooStep&& o) ROCKSDB_NOEXCEPT { *this = std::move(o); } CuckooStep& operator=(CuckooStep&& rhs) { bucket_id_ = std::move(rhs.bucket_id_); prev_step_id_ = std::move(rhs.prev_step_id_); depth_ = std::move(rhs.depth_); return *this; } CuckooStep(const CuckooStep&) = delete; CuckooStep& operator=(const CuckooStep&) = delete; CuckooStep(int bucket_id, int prev_step_id, int depth) : bucket_id_(bucket_id), prev_step_id_(prev_step_id), depth_(depth) {} }; class HashCuckooRep : public MemTableRep { public: explicit HashCuckooRep(const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const size_t bucket_count, const unsigned int hash_func_count, const size_t approximate_entry_size) : MemTableRep(allocator), compare_(compare), allocator_(allocator), bucket_count_(bucket_count), approximate_entry_size_(approximate_entry_size), cuckoo_path_max_depth_(kDefaultCuckooPathMaxDepth), occupied_count_(0), hash_function_count_(hash_func_count), backup_table_(nullptr) { char* mem = reinterpret_cast( allocator_->Allocate(sizeof(std::atomic) * bucket_count_)); cuckoo_array_ = new (mem) std::atomic[bucket_count_]; for (unsigned int bid = 0; bid < bucket_count_; ++bid) { cuckoo_array_[bid].store(nullptr, std::memory_order_relaxed); } cuckoo_path_ = reinterpret_cast( allocator_->Allocate(sizeof(int) * (cuckoo_path_max_depth_ + 1))); is_nearly_full_ = false; } // return false, indicating HashCuckooRep does not support merge operator. virtual bool IsMergeOperatorSupported() const override { return false; } // return false, indicating HashCuckooRep does not support snapshot. virtual bool IsSnapshotSupported() const override { return false; } // Returns true iff an entry that compares equal to key is in the collection. virtual bool Contains(const char* internal_key) const override; virtual ~HashCuckooRep() override {} // Insert the specified key (internal_key) into the mem-table. Assertion // fails if // the current mem-table already contains the specified key. virtual void Insert(KeyHandle handle) override; // This function returns bucket_count_ * approximate_entry_size_ when any // of the followings happen to disallow further write operations: // 1. when the fullness reaches kMaxFullnes. // 2. when the backup_table_ is used. // // otherwise, this function will always return 0. virtual size_t ApproximateMemoryUsage() override { if (is_nearly_full_) { return bucket_count_ * approximate_entry_size_; } return 0; } virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) override; class Iterator : public MemTableRep::Iterator { std::shared_ptr> bucket_; std::vector::const_iterator mutable cit_; const KeyComparator& compare_; std::string tmp_; // For passing to EncodeKey bool mutable sorted_; void DoSort() const; public: explicit Iterator(std::shared_ptr> bucket, const KeyComparator& compare); // Initialize an iterator over the specified collection. // The returned iterator is not valid. // explicit Iterator(const MemTableRep* collection); virtual ~Iterator() override{}; // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const override; // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const override; // Advances to the next position. // REQUIRES: Valid() virtual void Next() override; // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() override; // Advance to the first entry with a key >= target virtual void Seek(const Slice& user_key, const char* memtable_key) override; // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() override; // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() override; }; struct CuckooStepBuffer { CuckooStepBuffer() : write_index_(0), read_index_(0) {} ~CuckooStepBuffer() {} int write_index_; int read_index_; CuckooStep steps_[kCuckooPathMaxSearchSteps]; CuckooStep& NextWriteBuffer() { return steps_[write_index_++]; } inline const CuckooStep& ReadNext() { return steps_[read_index_++]; } inline bool HasNewWrite() { return write_index_ > read_index_; } inline void reset() { write_index_ = 0; read_index_ = 0; } inline bool IsFull() { return write_index_ >= kCuckooPathMaxSearchSteps; } // returns the number of steps that has been read inline int ReadCount() { return read_index_; } // returns the number of steps that has been written to the buffer. inline int WriteCount() { return write_index_; } }; private: const MemTableRep::KeyComparator& compare_; // the pointer to Allocator to allocate memory, immutable after construction. MemTableAllocator* const allocator_; // the number of hash bucket in the hash table. const size_t bucket_count_; // approximate size of each entry const size_t approximate_entry_size_; // the maxinum depth of the cuckoo path. const unsigned int cuckoo_path_max_depth_; // the current number of entries in cuckoo_array_ which has been occupied. size_t occupied_count_; // the current number of hash functions used in the cuckoo hash. unsigned int hash_function_count_; // the backup MemTableRep to handle the case where cuckoo hash cannot find // a vacant bucket for inserting the key of a put request. std::shared_ptr backup_table_; // the array to store pointers, pointing to the actual data. std::atomic* cuckoo_array_; // a buffer to store cuckoo path int* cuckoo_path_; // a boolean flag indicating whether the fullness of bucket array // reaches the point to make the current memtable immutable. bool is_nearly_full_; // the default maximum depth of the cuckoo path. static const unsigned int kDefaultCuckooPathMaxDepth = 10; CuckooStepBuffer step_buffer_; // returns the bucket id assogied to the input slice based on the unsigned int GetHash(const Slice& slice, const int hash_func_id) const { // the seeds used in the Murmur hash to produce different hash functions. static const int kMurmurHashSeeds[HashCuckooRepFactory::kMaxHashCount] = { 545609244, 1769731426, 763324157, 13099088, 592422103, 1899789565, 248369300, 1984183468, 1613664382, 1491157517}; return static_cast( MurmurHash(slice.data(), static_cast(slice.size()), kMurmurHashSeeds[hash_func_id]) % bucket_count_); } // A cuckoo path is a sequence of bucket ids, where each id points to a // location of cuckoo_array_. This path describes the displacement sequence // of entries in order to store the desired data specified by the input user // key. The path starts from one of the locations associated with the // specified user key and ends at a vacant space in the cuckoo array. This // function will update the cuckoo_path. // // @return true if it found a cuckoo path. bool FindCuckooPath(const char* internal_key, const Slice& user_key, int* cuckoo_path, size_t* cuckoo_path_length, int initial_hash_id = 0); // Perform quick insert by checking whether there is a vacant bucket in one // of the possible locations of the input key. If so, then the function will // return true and the key will be stored in that vacant bucket. // // This function is a helper function of FindCuckooPath that discovers the // first possible steps of a cuckoo path. It begins by first computing // the possible locations of the input keys (and stores them in bucket_ids.) // Then, if one of its possible locations is vacant, then the input key will // be stored in that vacant space and the function will return true. // Otherwise, the function will return false indicating a complete search // of cuckoo-path is needed. bool QuickInsert(const char* internal_key, const Slice& user_key, int bucket_ids[], const int initial_hash_id); // Returns the pointer to the internal iterator to the buckets where buckets // are sorted according to the user specified KeyComparator. Note that // any insert after this function call may affect the sorted nature of // the returned iterator. virtual MemTableRep::Iterator* GetIterator(Arena* arena) override { std::vector compact_buckets; for (unsigned int bid = 0; bid < bucket_count_; ++bid) { const char* bucket = cuckoo_array_[bid].load(std::memory_order_relaxed); if (bucket != nullptr) { compact_buckets.push_back(bucket); } } MemTableRep* backup_table = backup_table_.get(); if (backup_table != nullptr) { std::unique_ptr iter(backup_table->GetIterator()); for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { compact_buckets.push_back(iter->key()); } } if (arena == nullptr) { return new Iterator( std::shared_ptr>( new std::vector(std::move(compact_buckets))), compare_); } else { auto mem = arena->AllocateAligned(sizeof(Iterator)); return new (mem) Iterator( std::shared_ptr>( new std::vector(std::move(compact_buckets))), compare_); } } }; void HashCuckooRep::Get(const LookupKey& key, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) { Slice user_key = key.user_key(); for (unsigned int hid = 0; hid < hash_function_count_; ++hid) { const char* bucket = cuckoo_array_[GetHash(user_key, hid)].load(std::memory_order_acquire); if (bucket != nullptr) { Slice bucket_user_key = UserKey(bucket); if (user_key == bucket_user_key) { callback_func(callback_args, bucket); break; } } else { // as Put() always stores at the vacant bucket located by the // hash function with the smallest possible id, when we first // find a vacant bucket in Get(), that means a miss. break; } } MemTableRep* backup_table = backup_table_.get(); if (backup_table != nullptr) { backup_table->Get(key, callback_args, callback_func); } } void HashCuckooRep::Insert(KeyHandle handle) { static const float kMaxFullness = 0.90f; auto* key = static_cast(handle); int initial_hash_id = 0; size_t cuckoo_path_length = 0; auto user_key = UserKey(key); // find cuckoo path if (FindCuckooPath(key, user_key, cuckoo_path_, &cuckoo_path_length, initial_hash_id) == false) { // if true, then we can't find a vacant bucket for this key even we // have used up all the hash functions. Then use a backup memtable to // store such key, which will further make this mem-table become // immutable. if (backup_table_.get() == nullptr) { VectorRepFactory factory(10); backup_table_.reset( factory.CreateMemTableRep(compare_, allocator_, nullptr, nullptr)); is_nearly_full_ = true; } backup_table_->Insert(key); return; } // when reaching this point, means the insert can be done successfully. occupied_count_++; if (occupied_count_ >= bucket_count_ * kMaxFullness) { is_nearly_full_ = true; } // perform kickout process if the length of cuckoo path > 1. if (cuckoo_path_length == 0) return; // the cuckoo path stores the kickout path in reverse order. // so the kickout or displacement is actually performed // in reverse order, which avoids false-negatives on read // by moving each key involved in the cuckoo path to the new // location before replacing it. for (size_t i = 1; i < cuckoo_path_length; ++i) { int kicked_out_bid = cuckoo_path_[i - 1]; int current_bid = cuckoo_path_[i]; // since we only allow one writer at a time, it is safe to do relaxed read. cuckoo_array_[kicked_out_bid] .store(cuckoo_array_[current_bid].load(std::memory_order_relaxed), std::memory_order_release); } int insert_key_bid = cuckoo_path_[cuckoo_path_length - 1]; cuckoo_array_[insert_key_bid].store(key, std::memory_order_release); } bool HashCuckooRep::Contains(const char* internal_key) const { auto user_key = UserKey(internal_key); for (unsigned int hid = 0; hid < hash_function_count_; ++hid) { const char* stored_key = cuckoo_array_[GetHash(user_key, hid)].load(std::memory_order_acquire); if (stored_key != nullptr) { if (compare_(internal_key, stored_key) == 0) { return true; } } } return false; } bool HashCuckooRep::QuickInsert(const char* internal_key, const Slice& user_key, int bucket_ids[], const int initial_hash_id) { int cuckoo_bucket_id = -1; // Below does the followings: // 0. Calculate all possible locations of the input key. // 1. Check if there is a bucket having same user_key as the input does. // 2. If there exists such bucket, then replace this bucket by the newly // insert data and return. This step also performs duplication check. // 3. If no such bucket exists but exists a vacant bucket, then insert the // input data into it. // 4. If step 1 to 3 all fail, then return false. for (unsigned int hid = initial_hash_id; hid < hash_function_count_; ++hid) { bucket_ids[hid] = GetHash(user_key, hid); // since only one PUT is allowed at a time, and this is part of the PUT // operation, so we can safely perform relaxed load. const char* stored_key = cuckoo_array_[bucket_ids[hid]].load(std::memory_order_relaxed); if (stored_key == nullptr) { if (cuckoo_bucket_id == -1) { cuckoo_bucket_id = bucket_ids[hid]; } } else { const auto bucket_user_key = UserKey(stored_key); if (bucket_user_key.compare(user_key) == 0) { cuckoo_bucket_id = bucket_ids[hid]; break; } } } if (cuckoo_bucket_id != -1) { cuckoo_array_[cuckoo_bucket_id].store(const_cast(internal_key), std::memory_order_release); return true; } return false; } // Perform pre-check and find the shortest cuckoo path. A cuckoo path // is a displacement sequence for inserting the specified input key. // // @return true if it successfully found a vacant space or cuckoo-path. // If the return value is true but the length of cuckoo_path is zero, // then it indicates that a vacant bucket or an bucket with matched user // key with the input is found, and a quick insertion is done. bool HashCuckooRep::FindCuckooPath(const char* internal_key, const Slice& user_key, int* cuckoo_path, size_t* cuckoo_path_length, const int initial_hash_id) { int bucket_ids[HashCuckooRepFactory::kMaxHashCount]; *cuckoo_path_length = 0; if (QuickInsert(internal_key, user_key, bucket_ids, initial_hash_id)) { return true; } // If this step is reached, then it means: // 1. no vacant bucket in any of the possible locations of the input key. // 2. none of the possible locations of the input key has the same user // key as the input `internal_key`. // the front and back indices for the step_queue_ step_buffer_.reset(); for (unsigned int hid = initial_hash_id; hid < hash_function_count_; ++hid) { /// CuckooStep& current_step = step_queue_[front_pos++]; CuckooStep& current_step = step_buffer_.NextWriteBuffer(); current_step.bucket_id_ = bucket_ids[hid]; current_step.prev_step_id_ = CuckooStep::kNullStep; current_step.depth_ = 1; } while (step_buffer_.HasNewWrite()) { int step_id = step_buffer_.read_index_; const CuckooStep& step = step_buffer_.ReadNext(); // Since it's a BFS process, then the first step with its depth deeper // than the maximum allowed depth indicates all the remaining steps // in the step buffer queue will all exceed the maximum depth. // Return false immediately indicating we can't find a vacant bucket // for the input key before the maximum allowed depth. if (step.depth_ >= cuckoo_path_max_depth_) { return false; } // again, we can perform no barrier load safely here as the current // thread is the only writer. Slice bucket_user_key = UserKey(cuckoo_array_[step.bucket_id_].load(std::memory_order_relaxed)); if (step.prev_step_id_ != CuckooStep::kNullStep) { if (bucket_user_key == user_key) { // then there is a loop in the current path, stop discovering this path. continue; } } // if the current bucket stores at its nth location, then we only consider // its mth location where m > n. This property makes sure that all reads // will not miss if we do have data associated to the query key. // // The n and m in the above statement is the start_hid and hid in the code. unsigned int start_hid = hash_function_count_; for (unsigned int hid = 0; hid < hash_function_count_; ++hid) { bucket_ids[hid] = GetHash(bucket_user_key, hid); if (step.bucket_id_ == bucket_ids[hid]) { start_hid = hid; } } // must found a bucket which is its current "home". assert(start_hid != hash_function_count_); // explore all possible next steps from the current step. for (unsigned int hid = start_hid + 1; hid < hash_function_count_; ++hid) { CuckooStep& next_step = step_buffer_.NextWriteBuffer(); next_step.bucket_id_ = bucket_ids[hid]; next_step.prev_step_id_ = step_id; next_step.depth_ = step.depth_ + 1; // once a vacant bucket is found, trace back all its previous steps // to generate a cuckoo path. if (cuckoo_array_[next_step.bucket_id_].load(std::memory_order_relaxed) == nullptr) { // store the last step in the cuckoo path. Note that cuckoo_path // stores steps in reverse order. This allows us to move keys along // the cuckoo path by storing each key to the new place first before // removing it from the old place. This property ensures reads will // not missed due to moving keys along the cuckoo path. cuckoo_path[(*cuckoo_path_length)++] = next_step.bucket_id_; int depth; for (depth = step.depth_; depth > 0 && step_id != CuckooStep::kNullStep; depth--) { const CuckooStep& prev_step = step_buffer_.steps_[step_id]; cuckoo_path[(*cuckoo_path_length)++] = prev_step.bucket_id_; step_id = prev_step.prev_step_id_; } assert(depth == 0 && step_id == CuckooStep::kNullStep); return true; } if (step_buffer_.IsFull()) { // if true, then it reaches maxinum number of cuckoo search steps. return false; } } } // tried all possible paths but still not unable to find a cuckoo path // which path leads to a vacant bucket. return false; } HashCuckooRep::Iterator::Iterator( std::shared_ptr> bucket, const KeyComparator& compare) : bucket_(bucket), cit_(bucket_->end()), compare_(compare), sorted_(false) {} void HashCuckooRep::Iterator::DoSort() const { if (!sorted_) { std::sort(bucket_->begin(), bucket_->end(), stl_wrappers::Compare(compare_)); cit_ = bucket_->begin(); sorted_ = true; } } // Returns true iff the iterator is positioned at a valid node. bool HashCuckooRep::Iterator::Valid() const { DoSort(); return cit_ != bucket_->end(); } // Returns the key at the current position. // REQUIRES: Valid() const char* HashCuckooRep::Iterator::key() const { assert(Valid()); return *cit_; } // Advances to the next position. // REQUIRES: Valid() void HashCuckooRep::Iterator::Next() { assert(Valid()); if (cit_ == bucket_->end()) { return; } ++cit_; } // Advances to the previous position. // REQUIRES: Valid() void HashCuckooRep::Iterator::Prev() { assert(Valid()); if (cit_ == bucket_->begin()) { // If you try to go back from the first element, the iterator should be // invalidated. So we set it to past-the-end. This means that you can // treat the container circularly. cit_ = bucket_->end(); } else { --cit_; } } // Advance to the first entry with a key >= target void HashCuckooRep::Iterator::Seek(const Slice& user_key, const char* memtable_key) { DoSort(); // Do binary search to find first value not less than the target const char* encoded_key = (memtable_key != nullptr) ? memtable_key : EncodeKey(&tmp_, user_key); cit_ = std::equal_range(bucket_->begin(), bucket_->end(), encoded_key, [this](const char* a, const char* b) { return compare_(a, b) < 0; }).first; } // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. void HashCuckooRep::Iterator::SeekToFirst() { DoSort(); cit_ = bucket_->begin(); } // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. void HashCuckooRep::Iterator::SeekToLast() { DoSort(); cit_ = bucket_->end(); if (bucket_->size() != 0) { --cit_; } } } // anom namespace MemTableRep* HashCuckooRepFactory::CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) { // The estimated average fullness. The write performance of any close hash // degrades as the fullness of the mem-table increases. Setting kFullness // to a value around 0.7 can better avoid write performance degradation while // keeping efficient memory usage. static const float kFullness = 0.7f; size_t pointer_size = sizeof(std::atomic); assert(write_buffer_size_ >= (average_data_size_ + pointer_size)); size_t bucket_count = static_cast( (write_buffer_size_ / (average_data_size_ + pointer_size)) / kFullness + 1); unsigned int hash_function_count = hash_function_count_; if (hash_function_count < 2) { hash_function_count = 2; } if (hash_function_count > kMaxHashCount) { hash_function_count = kMaxHashCount; } return new HashCuckooRep(compare, allocator, bucket_count, hash_function_count, static_cast( (average_data_size_ + pointer_size) / kFullness) ); } MemTableRepFactory* NewHashCuckooRepFactory(size_t write_buffer_size, size_t average_data_size, unsigned int hash_function_count) { return new HashCuckooRepFactory(write_buffer_size, average_data_size, hash_function_count); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/memtable/hash_cuckoo_rep.h000066400000000000000000000031351267530634000202530ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include "port/port.h" #include "rocksdb/slice_transform.h" #include "rocksdb/memtablerep.h" namespace rocksdb { class HashCuckooRepFactory : public MemTableRepFactory { public: // maxinum number of hash functions used in the cuckoo hash. static const unsigned int kMaxHashCount = 10; explicit HashCuckooRepFactory(size_t write_buffer_size, size_t average_data_size, unsigned int hash_function_count) : write_buffer_size_(write_buffer_size), average_data_size_(average_data_size), hash_function_count_(hash_function_count) {} virtual ~HashCuckooRepFactory() {} virtual MemTableRep* CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) override; virtual const char* Name() const override { return "HashCuckooRepFactory"; } private: size_t write_buffer_size_; size_t average_data_size_; const unsigned int hash_function_count_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/memtable/hash_linklist_rep.cc000066400000000000000000000703511267530634000207630ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #ifndef ROCKSDB_LITE #include "memtable/hash_linklist_rep.h" #include #include #include "rocksdb/memtablerep.h" #include "util/arena.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "port/port.h" #include "util/histogram.h" #include "util/murmurhash.h" #include "db/memtable.h" #include "db/skiplist.h" namespace rocksdb { namespace { typedef const char* Key; typedef SkipList MemtableSkipList; typedef std::atomic Pointer; // A data structure used as the header of a link list of a hash bucket. struct BucketHeader { Pointer next; std::atomic num_entries; explicit BucketHeader(void* n, uint32_t count) : next(n), num_entries(count) {} bool IsSkipListBucket() { return next.load(std::memory_order_relaxed) == this; } uint32_t GetNumEntries() const { return num_entries.load(std::memory_order_relaxed); } // REQUIRES: called from single-threaded Insert() void IncNumEntries() { // Only one thread can do write at one time. No need to do atomic // incremental. Update it with relaxed load and store. num_entries.store(GetNumEntries() + 1, std::memory_order_relaxed); } }; // A data structure used as the header of a skip list of a hash bucket. struct SkipListBucketHeader { BucketHeader Counting_header; MemtableSkipList skip_list; explicit SkipListBucketHeader(const MemTableRep::KeyComparator& cmp, MemTableAllocator* allocator, uint32_t count) : Counting_header(this, // Pointing to itself to indicate header type. count), skip_list(cmp, allocator) {} }; struct Node { // Accessors/mutators for links. Wrapped in methods so we can // add the appropriate barriers as necessary. Node* Next() { // Use an 'acquire load' so that we observe a fully initialized // version of the returned Node. return next_.load(std::memory_order_acquire); } void SetNext(Node* x) { // Use a 'release store' so that anybody who reads through this // pointer observes a fully initialized version of the inserted node. next_.store(x, std::memory_order_release); } // No-barrier variants that can be safely used in a few locations. Node* NoBarrier_Next() { return next_.load(std::memory_order_relaxed); } void NoBarrier_SetNext(Node* x) { next_.store(x, std::memory_order_relaxed); } // Needed for placement new below which is fine Node() {} private: std::atomic next_; // Prohibit copying due to the below Node(const Node&) = delete; Node& operator=(const Node&) = delete; public: char key[1]; }; // Memory structure of the mem table: // It is a hash table, each bucket points to one entry, a linked list or a // skip list. In order to track total number of records in a bucket to determine // whether should switch to skip list, a header is added just to indicate // number of entries in the bucket. // // // +-----> NULL Case 1. Empty bucket // | // | // | +---> +-------+ // | | | Next +--> NULL // | | +-------+ // +-----+ | | | | Case 2. One Entry in bucket. // | +-+ | | Data | next pointer points to // +-----+ | | | NULL. All other cases // | | | | | next pointer is not NULL. // +-----+ | +-------+ // | +---+ // +-----+ +-> +-------+ +> +-------+ +-> +-------+ // | | | | Next +--+ | Next +--+ | Next +-->NULL // +-----+ | +-------+ +-------+ +-------+ // | +-----+ | Count | | | | | // +-----+ +-------+ | Data | | Data | // | | | | | | // +-----+ Case 3. | | | | // | | A header +-------+ +-------+ // +-----+ points to // | | a linked list. Count indicates total number // +-----+ of rows in this bucket. // | | // +-----+ +-> +-------+ <--+ // | | | | Next +----+ // +-----+ | +-------+ Case 4. A header points to a skip // | +----+ | Count | list and next pointer points to // +-----+ +-------+ itself, to distinguish case 3 or 4. // | | | | Count still is kept to indicates total // +-----+ | Skip +--> of entries in the bucket for debugging // | | | List | Data purpose. // | | | +--> // +-----+ | | // | | +-------+ // +-----+ // // We don't have data race when changing cases because: // (1) When changing from case 2->3, we create a new bucket header, put the // single node there first without changing the original node, and do a // release store when changing the bucket pointer. In that case, a reader // who sees a stale value of the bucket pointer will read this node, while // a reader sees the correct value because of the release store. // (2) When changing case 3->4, a new header is created with skip list points // to the data, before doing an acquire store to change the bucket pointer. // The old header and nodes are never changed, so any reader sees any // of those existing pointers will guarantee to be able to iterate to the // end of the linked list. // (3) Header's next pointer in case 3 might change, but they are never equal // to itself, so no matter a reader sees any stale or newer value, it will // be able to correctly distinguish case 3 and 4. // // The reason that we use case 2 is we want to make the format to be efficient // when the utilization of buckets is relatively low. If we use case 3 for // single entry bucket, we will need to waste 12 bytes for every entry, // which can be significant decrease of memory utilization. class HashLinkListRep : public MemTableRep { public: HashLinkListRep(const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, size_t bucket_size, uint32_t threshold_use_skiplist, size_t huge_page_tlb_size, Logger* logger, int bucket_entries_logging_threshold, bool if_log_bucket_dist_when_flash); virtual KeyHandle Allocate(const size_t len, char** buf) override; virtual void Insert(KeyHandle handle) override; virtual bool Contains(const char* key) const override; virtual size_t ApproximateMemoryUsage() override; virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) override; virtual ~HashLinkListRep(); virtual MemTableRep::Iterator* GetIterator(Arena* arena = nullptr) override; virtual MemTableRep::Iterator* GetDynamicPrefixIterator( Arena* arena = nullptr) override; private: friend class DynamicIterator; size_t bucket_size_; // Maps slices (which are transformed user keys) to buckets of keys sharing // the same transform. Pointer* buckets_; const uint32_t threshold_use_skiplist_; // The user-supplied transform whose domain is the user keys. const SliceTransform* transform_; const MemTableRep::KeyComparator& compare_; Logger* logger_; int bucket_entries_logging_threshold_; bool if_log_bucket_dist_when_flash_; bool LinkListContains(Node* head, const Slice& key) const; SkipListBucketHeader* GetSkipListBucketHeader(Pointer* first_next_pointer) const; Node* GetLinkListFirstNode(Pointer* first_next_pointer) const; Slice GetPrefix(const Slice& internal_key) const { return transform_->Transform(ExtractUserKey(internal_key)); } size_t GetHash(const Slice& slice) const { return MurmurHash(slice.data(), static_cast(slice.size()), 0) % bucket_size_; } Pointer* GetBucket(size_t i) const { return static_cast(buckets_[i].load(std::memory_order_acquire)); } Pointer* GetBucket(const Slice& slice) const { return GetBucket(GetHash(slice)); } bool Equal(const Slice& a, const Key& b) const { return (compare_(b, a) == 0); } bool Equal(const Key& a, const Key& b) const { return (compare_(a, b) == 0); } bool KeyIsAfterNode(const Slice& internal_key, const Node* n) const { // nullptr n is considered infinite return (n != nullptr) && (compare_(n->key, internal_key) < 0); } bool KeyIsAfterNode(const Key& key, const Node* n) const { // nullptr n is considered infinite return (n != nullptr) && (compare_(n->key, key) < 0); } Node* FindGreaterOrEqualInBucket(Node* head, const Slice& key) const; class FullListIterator : public MemTableRep::Iterator { public: explicit FullListIterator(MemtableSkipList* list, Allocator* allocator) : iter_(list), full_list_(list), allocator_(allocator) {} virtual ~FullListIterator() { } // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const override { return iter_.Valid(); } // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const override { assert(Valid()); return iter_.key(); } // Advances to the next position. // REQUIRES: Valid() virtual void Next() override { assert(Valid()); iter_.Next(); } // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() override { assert(Valid()); iter_.Prev(); } // Advance to the first entry with a key >= target virtual void Seek(const Slice& internal_key, const char* memtable_key) override { const char* encoded_key = (memtable_key != nullptr) ? memtable_key : EncodeKey(&tmp_, internal_key); iter_.Seek(encoded_key); } // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() override { iter_.SeekToFirst(); } // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() override { iter_.SeekToLast(); } private: MemtableSkipList::Iterator iter_; // To destruct with the iterator. std::unique_ptr full_list_; std::unique_ptr allocator_; std::string tmp_; // For passing to EncodeKey }; class LinkListIterator : public MemTableRep::Iterator { public: explicit LinkListIterator(const HashLinkListRep* const hash_link_list_rep, Node* head) : hash_link_list_rep_(hash_link_list_rep), head_(head), node_(nullptr) {} virtual ~LinkListIterator() {} // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const override { return node_ != nullptr; } // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const override { assert(Valid()); return node_->key; } // Advances to the next position. // REQUIRES: Valid() virtual void Next() override { assert(Valid()); node_ = node_->Next(); } // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() override { // Prefix iterator does not support total order. // We simply set the iterator to invalid state Reset(nullptr); } // Advance to the first entry with a key >= target virtual void Seek(const Slice& internal_key, const char* memtable_key) override { node_ = hash_link_list_rep_->FindGreaterOrEqualInBucket(head_, internal_key); } // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() override { // Prefix iterator does not support total order. // We simply set the iterator to invalid state Reset(nullptr); } // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() override { // Prefix iterator does not support total order. // We simply set the iterator to invalid state Reset(nullptr); } protected: void Reset(Node* head) { head_ = head; node_ = nullptr; } private: friend class HashLinkListRep; const HashLinkListRep* const hash_link_list_rep_; Node* head_; Node* node_; virtual void SeekToHead() { node_ = head_; } }; class DynamicIterator : public HashLinkListRep::LinkListIterator { public: explicit DynamicIterator(HashLinkListRep& memtable_rep) : HashLinkListRep::LinkListIterator(&memtable_rep, nullptr), memtable_rep_(memtable_rep) {} // Advance to the first entry with a key >= target virtual void Seek(const Slice& k, const char* memtable_key) override { auto transformed = memtable_rep_.GetPrefix(k); auto* bucket = memtable_rep_.GetBucket(transformed); SkipListBucketHeader* skip_list_header = memtable_rep_.GetSkipListBucketHeader(bucket); if (skip_list_header != nullptr) { // The bucket is organized as a skip list if (!skip_list_iter_) { skip_list_iter_.reset( new MemtableSkipList::Iterator(&skip_list_header->skip_list)); } else { skip_list_iter_->SetList(&skip_list_header->skip_list); } if (memtable_key != nullptr) { skip_list_iter_->Seek(memtable_key); } else { IterKey encoded_key; encoded_key.EncodeLengthPrefixedKey(k); skip_list_iter_->Seek(encoded_key.GetKey().data()); } } else { // The bucket is organized as a linked list skip_list_iter_.reset(); Reset(memtable_rep_.GetLinkListFirstNode(bucket)); HashLinkListRep::LinkListIterator::Seek(k, memtable_key); } } virtual bool Valid() const override { if (skip_list_iter_) { return skip_list_iter_->Valid(); } return HashLinkListRep::LinkListIterator::Valid(); } virtual const char* key() const override { if (skip_list_iter_) { return skip_list_iter_->key(); } return HashLinkListRep::LinkListIterator::key(); } virtual void Next() override { if (skip_list_iter_) { skip_list_iter_->Next(); } else { HashLinkListRep::LinkListIterator::Next(); } } private: // the underlying memtable const HashLinkListRep& memtable_rep_; std::unique_ptr skip_list_iter_; }; class EmptyIterator : public MemTableRep::Iterator { // This is used when there wasn't a bucket. It is cheaper than // instantiating an empty bucket over which to iterate. public: EmptyIterator() { } virtual bool Valid() const override { return false; } virtual const char* key() const override { assert(false); return nullptr; } virtual void Next() override {} virtual void Prev() override {} virtual void Seek(const Slice& user_key, const char* memtable_key) override {} virtual void SeekToFirst() override {} virtual void SeekToLast() override {} private: }; }; HashLinkListRep::HashLinkListRep(const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, size_t bucket_size, uint32_t threshold_use_skiplist, size_t huge_page_tlb_size, Logger* logger, int bucket_entries_logging_threshold, bool if_log_bucket_dist_when_flash) : MemTableRep(allocator), bucket_size_(bucket_size), // Threshold to use skip list doesn't make sense if less than 3, so we // force it to be minimum of 3 to simplify implementation. threshold_use_skiplist_(std::max(threshold_use_skiplist, 3U)), transform_(transform), compare_(compare), logger_(logger), bucket_entries_logging_threshold_(bucket_entries_logging_threshold), if_log_bucket_dist_when_flash_(if_log_bucket_dist_when_flash) { char* mem = allocator_->AllocateAligned(sizeof(Pointer) * bucket_size, huge_page_tlb_size, logger); buckets_ = new (mem) Pointer[bucket_size]; for (size_t i = 0; i < bucket_size_; ++i) { buckets_[i].store(nullptr, std::memory_order_relaxed); } } HashLinkListRep::~HashLinkListRep() { } KeyHandle HashLinkListRep::Allocate(const size_t len, char** buf) { char* mem = allocator_->AllocateAligned(sizeof(Node) + len); Node* x = new (mem) Node(); *buf = x->key; return static_cast(x); } SkipListBucketHeader* HashLinkListRep::GetSkipListBucketHeader( Pointer* first_next_pointer) const { if (first_next_pointer == nullptr) { return nullptr; } if (first_next_pointer->load(std::memory_order_relaxed) == nullptr) { // Single entry bucket return nullptr; } // Counting header BucketHeader* header = reinterpret_cast(first_next_pointer); if (header->IsSkipListBucket()) { assert(header->GetNumEntries() > threshold_use_skiplist_); auto* skip_list_bucket_header = reinterpret_cast(header); assert(skip_list_bucket_header->Counting_header.next.load( std::memory_order_relaxed) == header); return skip_list_bucket_header; } assert(header->GetNumEntries() <= threshold_use_skiplist_); return nullptr; } Node* HashLinkListRep::GetLinkListFirstNode(Pointer* first_next_pointer) const { if (first_next_pointer == nullptr) { return nullptr; } if (first_next_pointer->load(std::memory_order_relaxed) == nullptr) { // Single entry bucket return reinterpret_cast(first_next_pointer); } // Counting header BucketHeader* header = reinterpret_cast(first_next_pointer); if (!header->IsSkipListBucket()) { assert(header->GetNumEntries() <= threshold_use_skiplist_); return reinterpret_cast( header->next.load(std::memory_order_acquire)); } assert(header->GetNumEntries() > threshold_use_skiplist_); return nullptr; } void HashLinkListRep::Insert(KeyHandle handle) { Node* x = static_cast(handle); assert(!Contains(x->key)); Slice internal_key = GetLengthPrefixedSlice(x->key); auto transformed = GetPrefix(internal_key); auto& bucket = buckets_[GetHash(transformed)]; Pointer* first_next_pointer = static_cast(bucket.load(std::memory_order_relaxed)); if (first_next_pointer == nullptr) { // Case 1. empty bucket // NoBarrier_SetNext() suffices since we will add a barrier when // we publish a pointer to "x" in prev[i]. x->NoBarrier_SetNext(nullptr); bucket.store(x, std::memory_order_release); return; } BucketHeader* header = nullptr; if (first_next_pointer->load(std::memory_order_relaxed) == nullptr) { // Case 2. only one entry in the bucket // Need to convert to a Counting bucket and turn to case 4. Node* first = reinterpret_cast(first_next_pointer); // Need to add a bucket header. // We have to first convert it to a bucket with header before inserting // the new node. Otherwise, we might need to change next pointer of first. // In that case, a reader might sees the next pointer is NULL and wrongly // think the node is a bucket header. auto* mem = allocator_->AllocateAligned(sizeof(BucketHeader)); header = new (mem) BucketHeader(first, 1); bucket.store(header, std::memory_order_release); } else { header = reinterpret_cast(first_next_pointer); if (header->IsSkipListBucket()) { // Case 4. Bucket is already a skip list assert(header->GetNumEntries() > threshold_use_skiplist_); auto* skip_list_bucket_header = reinterpret_cast(header); // Only one thread can execute Insert() at one time. No need to do atomic // incremental. skip_list_bucket_header->Counting_header.IncNumEntries(); skip_list_bucket_header->skip_list.Insert(x->key); return; } } if (bucket_entries_logging_threshold_ > 0 && header->GetNumEntries() == static_cast(bucket_entries_logging_threshold_)) { Info(logger_, "HashLinkedList bucket %" ROCKSDB_PRIszt " has more than %d " "entries. Key to insert: %s", GetHash(transformed), header->GetNumEntries(), GetLengthPrefixedSlice(x->key).ToString(true).c_str()); } if (header->GetNumEntries() == threshold_use_skiplist_) { // Case 3. number of entries reaches the threshold so need to convert to // skip list. LinkListIterator bucket_iter( this, reinterpret_cast( first_next_pointer->load(std::memory_order_relaxed))); auto mem = allocator_->AllocateAligned(sizeof(SkipListBucketHeader)); SkipListBucketHeader* new_skip_list_header = new (mem) SkipListBucketHeader(compare_, allocator_, header->GetNumEntries() + 1); auto& skip_list = new_skip_list_header->skip_list; // Add all current entries to the skip list for (bucket_iter.SeekToHead(); bucket_iter.Valid(); bucket_iter.Next()) { skip_list.Insert(bucket_iter.key()); } // insert the new entry skip_list.Insert(x->key); // Set the bucket bucket.store(new_skip_list_header, std::memory_order_release); } else { // Case 5. Need to insert to the sorted linked list without changing the // header. Node* first = reinterpret_cast(header->next.load(std::memory_order_relaxed)); assert(first != nullptr); // Advance counter unless the bucket needs to be advanced to skip list. // In that case, we need to make sure the previous count never exceeds // threshold_use_skiplist_ to avoid readers to cast to wrong format. header->IncNumEntries(); Node* cur = first; Node* prev = nullptr; while (true) { if (cur == nullptr) { break; } Node* next = cur->Next(); // Make sure the lists are sorted. // If x points to head_ or next points nullptr, it is trivially satisfied. assert((cur == first) || (next == nullptr) || KeyIsAfterNode(next->key, cur)); if (KeyIsAfterNode(internal_key, cur)) { // Keep searching in this list prev = cur; cur = next; } else { break; } } // Our data structure does not allow duplicate insertion assert(cur == nullptr || !Equal(x->key, cur->key)); // NoBarrier_SetNext() suffices since we will add a barrier when // we publish a pointer to "x" in prev[i]. x->NoBarrier_SetNext(cur); if (prev) { prev->SetNext(x); } else { header->next.store(static_cast(x), std::memory_order_release); } } } bool HashLinkListRep::Contains(const char* key) const { Slice internal_key = GetLengthPrefixedSlice(key); auto transformed = GetPrefix(internal_key); auto bucket = GetBucket(transformed); if (bucket == nullptr) { return false; } SkipListBucketHeader* skip_list_header = GetSkipListBucketHeader(bucket); if (skip_list_header != nullptr) { return skip_list_header->skip_list.Contains(key); } else { return LinkListContains(GetLinkListFirstNode(bucket), internal_key); } } size_t HashLinkListRep::ApproximateMemoryUsage() { // Memory is always allocated from the allocator. return 0; } void HashLinkListRep::Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) { auto transformed = transform_->Transform(k.user_key()); auto bucket = GetBucket(transformed); auto* skip_list_header = GetSkipListBucketHeader(bucket); if (skip_list_header != nullptr) { // Is a skip list MemtableSkipList::Iterator iter(&skip_list_header->skip_list); for (iter.Seek(k.memtable_key().data()); iter.Valid() && callback_func(callback_args, iter.key()); iter.Next()) { } } else { auto* link_list_head = GetLinkListFirstNode(bucket); if (link_list_head != nullptr) { LinkListIterator iter(this, link_list_head); for (iter.Seek(k.internal_key(), nullptr); iter.Valid() && callback_func(callback_args, iter.key()); iter.Next()) { } } } } MemTableRep::Iterator* HashLinkListRep::GetIterator(Arena* alloc_arena) { // allocate a new arena of similar size to the one currently in use Arena* new_arena = new Arena(allocator_->BlockSize()); auto list = new MemtableSkipList(compare_, new_arena); HistogramImpl keys_per_bucket_hist; for (size_t i = 0; i < bucket_size_; ++i) { int count = 0; auto* bucket = GetBucket(i); if (bucket != nullptr) { auto* skip_list_header = GetSkipListBucketHeader(bucket); if (skip_list_header != nullptr) { // Is a skip list MemtableSkipList::Iterator itr(&skip_list_header->skip_list); for (itr.SeekToFirst(); itr.Valid(); itr.Next()) { list->Insert(itr.key()); count++; } } else { auto* link_list_head = GetLinkListFirstNode(bucket); if (link_list_head != nullptr) { LinkListIterator itr(this, link_list_head); for (itr.SeekToHead(); itr.Valid(); itr.Next()) { list->Insert(itr.key()); count++; } } } } if (if_log_bucket_dist_when_flash_) { keys_per_bucket_hist.Add(count); } } if (if_log_bucket_dist_when_flash_ && logger_ != nullptr) { Info(logger_, "hashLinkedList Entry distribution among buckets: %s", keys_per_bucket_hist.ToString().c_str()); } if (alloc_arena == nullptr) { return new FullListIterator(list, new_arena); } else { auto mem = alloc_arena->AllocateAligned(sizeof(FullListIterator)); return new (mem) FullListIterator(list, new_arena); } } MemTableRep::Iterator* HashLinkListRep::GetDynamicPrefixIterator( Arena* alloc_arena) { if (alloc_arena == nullptr) { return new DynamicIterator(*this); } else { auto mem = alloc_arena->AllocateAligned(sizeof(DynamicIterator)); return new (mem) DynamicIterator(*this); } } bool HashLinkListRep::LinkListContains(Node* head, const Slice& user_key) const { Node* x = FindGreaterOrEqualInBucket(head, user_key); return (x != nullptr && Equal(user_key, x->key)); } Node* HashLinkListRep::FindGreaterOrEqualInBucket(Node* head, const Slice& key) const { Node* x = head; while (true) { if (x == nullptr) { return x; } Node* next = x->Next(); // Make sure the lists are sorted. // If x points to head_ or next points nullptr, it is trivially satisfied. assert((x == head) || (next == nullptr) || KeyIsAfterNode(next->key, x)); if (KeyIsAfterNode(key, x)) { // Keep searching in this list x = next; } else { break; } } return x; } } // anon namespace MemTableRep* HashLinkListRepFactory::CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) { return new HashLinkListRep(compare, allocator, transform, bucket_count_, threshold_use_skiplist_, huge_page_tlb_size_, logger, bucket_entries_logging_threshold_, if_log_bucket_dist_when_flash_); } MemTableRepFactory* NewHashLinkListRepFactory( size_t bucket_count, size_t huge_page_tlb_size, int bucket_entries_logging_threshold, bool if_log_bucket_dist_when_flash, uint32_t threshold_use_skiplist) { return new HashLinkListRepFactory( bucket_count, threshold_use_skiplist, huge_page_tlb_size, bucket_entries_logging_threshold, if_log_bucket_dist_when_flash); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/memtable/hash_linklist_rep.h000066400000000000000000000035051267530634000206220ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include "rocksdb/slice_transform.h" #include "rocksdb/memtablerep.h" namespace rocksdb { class HashLinkListRepFactory : public MemTableRepFactory { public: explicit HashLinkListRepFactory(size_t bucket_count, uint32_t threshold_use_skiplist, size_t huge_page_tlb_size, int bucket_entries_logging_threshold, bool if_log_bucket_dist_when_flash) : bucket_count_(bucket_count), threshold_use_skiplist_(threshold_use_skiplist), huge_page_tlb_size_(huge_page_tlb_size), bucket_entries_logging_threshold_(bucket_entries_logging_threshold), if_log_bucket_dist_when_flash_(if_log_bucket_dist_when_flash) {} virtual ~HashLinkListRepFactory() {} virtual MemTableRep* CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) override; virtual const char* Name() const override { return "HashLinkListRepFactory"; } private: const size_t bucket_count_; const uint32_t threshold_use_skiplist_; const size_t huge_page_tlb_size_; int bucket_entries_logging_threshold_; bool if_log_bucket_dist_when_flash_; }; } #endif // ROCKSDB_LITE rocksdb-4.5.1/memtable/hash_skiplist_rep.cc000066400000000000000000000255101267530634000207710ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #ifndef ROCKSDB_LITE #include "memtable/hash_skiplist_rep.h" #include #include "rocksdb/memtablerep.h" #include "util/arena.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "port/port.h" #include "util/murmurhash.h" #include "db/memtable.h" #include "db/skiplist.h" namespace rocksdb { namespace { class HashSkipListRep : public MemTableRep { public: HashSkipListRep(const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, size_t bucket_size, int32_t skiplist_height, int32_t skiplist_branching_factor); virtual void Insert(KeyHandle handle) override; virtual bool Contains(const char* key) const override; virtual size_t ApproximateMemoryUsage() override; virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) override; virtual ~HashSkipListRep(); virtual MemTableRep::Iterator* GetIterator(Arena* arena = nullptr) override; virtual MemTableRep::Iterator* GetDynamicPrefixIterator( Arena* arena = nullptr) override; private: friend class DynamicIterator; typedef SkipList Bucket; size_t bucket_size_; const int32_t skiplist_height_; const int32_t skiplist_branching_factor_; // Maps slices (which are transformed user keys) to buckets of keys sharing // the same transform. std::atomic* buckets_; // The user-supplied transform whose domain is the user keys. const SliceTransform* transform_; const MemTableRep::KeyComparator& compare_; // immutable after construction MemTableAllocator* const allocator_; inline size_t GetHash(const Slice& slice) const { return MurmurHash(slice.data(), static_cast(slice.size()), 0) % bucket_size_; } inline Bucket* GetBucket(size_t i) const { return buckets_[i].load(std::memory_order_acquire); } inline Bucket* GetBucket(const Slice& slice) const { return GetBucket(GetHash(slice)); } // Get a bucket from buckets_. If the bucket hasn't been initialized yet, // initialize it before returning. Bucket* GetInitializedBucket(const Slice& transformed); class Iterator : public MemTableRep::Iterator { public: explicit Iterator(Bucket* list, bool own_list = true, Arena* arena = nullptr) : list_(list), iter_(list), own_list_(own_list), arena_(arena) {} virtual ~Iterator() { // if we own the list, we should also delete it if (own_list_) { assert(list_ != nullptr); delete list_; } } // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const override { return list_ != nullptr && iter_.Valid(); } // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const override { assert(Valid()); return iter_.key(); } // Advances to the next position. // REQUIRES: Valid() virtual void Next() override { assert(Valid()); iter_.Next(); } // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() override { assert(Valid()); iter_.Prev(); } // Advance to the first entry with a key >= target virtual void Seek(const Slice& internal_key, const char* memtable_key) override { if (list_ != nullptr) { const char* encoded_key = (memtable_key != nullptr) ? memtable_key : EncodeKey(&tmp_, internal_key); iter_.Seek(encoded_key); } } // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() override { if (list_ != nullptr) { iter_.SeekToFirst(); } } // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() override { if (list_ != nullptr) { iter_.SeekToLast(); } } protected: void Reset(Bucket* list) { if (own_list_) { assert(list_ != nullptr); delete list_; } list_ = list; iter_.SetList(list); own_list_ = false; } private: // if list_ is nullptr, we should NEVER call any methods on iter_ // if list_ is nullptr, this Iterator is not Valid() Bucket* list_; Bucket::Iterator iter_; // here we track if we own list_. If we own it, we are also // responsible for it's cleaning. This is a poor man's shared_ptr bool own_list_; std::unique_ptr arena_; std::string tmp_; // For passing to EncodeKey }; class DynamicIterator : public HashSkipListRep::Iterator { public: explicit DynamicIterator(const HashSkipListRep& memtable_rep) : HashSkipListRep::Iterator(nullptr, false), memtable_rep_(memtable_rep) {} // Advance to the first entry with a key >= target virtual void Seek(const Slice& k, const char* memtable_key) override { auto transformed = memtable_rep_.transform_->Transform(ExtractUserKey(k)); Reset(memtable_rep_.GetBucket(transformed)); HashSkipListRep::Iterator::Seek(k, memtable_key); } // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() override { // Prefix iterator does not support total order. // We simply set the iterator to invalid state Reset(nullptr); } // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() override { // Prefix iterator does not support total order. // We simply set the iterator to invalid state Reset(nullptr); } private: // the underlying memtable const HashSkipListRep& memtable_rep_; }; class EmptyIterator : public MemTableRep::Iterator { // This is used when there wasn't a bucket. It is cheaper than // instantiating an empty bucket over which to iterate. public: EmptyIterator() { } virtual bool Valid() const override { return false; } virtual const char* key() const override { assert(false); return nullptr; } virtual void Next() override {} virtual void Prev() override {} virtual void Seek(const Slice& internal_key, const char* memtable_key) override {} virtual void SeekToFirst() override {} virtual void SeekToLast() override {} private: }; }; HashSkipListRep::HashSkipListRep(const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, size_t bucket_size, int32_t skiplist_height, int32_t skiplist_branching_factor) : MemTableRep(allocator), bucket_size_(bucket_size), skiplist_height_(skiplist_height), skiplist_branching_factor_(skiplist_branching_factor), transform_(transform), compare_(compare), allocator_(allocator) { auto mem = allocator->AllocateAligned( sizeof(std::atomic) * bucket_size); buckets_ = new (mem) std::atomic[bucket_size]; for (size_t i = 0; i < bucket_size_; ++i) { buckets_[i].store(nullptr, std::memory_order_relaxed); } } HashSkipListRep::~HashSkipListRep() { } HashSkipListRep::Bucket* HashSkipListRep::GetInitializedBucket( const Slice& transformed) { size_t hash = GetHash(transformed); auto bucket = GetBucket(hash); if (bucket == nullptr) { auto addr = allocator_->AllocateAligned(sizeof(Bucket)); bucket = new (addr) Bucket(compare_, allocator_, skiplist_height_, skiplist_branching_factor_); buckets_[hash].store(bucket, std::memory_order_release); } return bucket; } void HashSkipListRep::Insert(KeyHandle handle) { auto* key = static_cast(handle); assert(!Contains(key)); auto transformed = transform_->Transform(UserKey(key)); auto bucket = GetInitializedBucket(transformed); bucket->Insert(key); } bool HashSkipListRep::Contains(const char* key) const { auto transformed = transform_->Transform(UserKey(key)); auto bucket = GetBucket(transformed); if (bucket == nullptr) { return false; } return bucket->Contains(key); } size_t HashSkipListRep::ApproximateMemoryUsage() { return 0; } void HashSkipListRep::Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) { auto transformed = transform_->Transform(k.user_key()); auto bucket = GetBucket(transformed); if (bucket != nullptr) { Bucket::Iterator iter(bucket); for (iter.Seek(k.memtable_key().data()); iter.Valid() && callback_func(callback_args, iter.key()); iter.Next()) { } } } MemTableRep::Iterator* HashSkipListRep::GetIterator(Arena* arena) { // allocate a new arena of similar size to the one currently in use Arena* new_arena = new Arena(allocator_->BlockSize()); auto list = new Bucket(compare_, new_arena); for (size_t i = 0; i < bucket_size_; ++i) { auto bucket = GetBucket(i); if (bucket != nullptr) { Bucket::Iterator itr(bucket); for (itr.SeekToFirst(); itr.Valid(); itr.Next()) { list->Insert(itr.key()); } } } if (arena == nullptr) { return new Iterator(list, true, new_arena); } else { auto mem = arena->AllocateAligned(sizeof(Iterator)); return new (mem) Iterator(list, true, new_arena); } } MemTableRep::Iterator* HashSkipListRep::GetDynamicPrefixIterator(Arena* arena) { if (arena == nullptr) { return new DynamicIterator(*this); } else { auto mem = arena->AllocateAligned(sizeof(DynamicIterator)); return new (mem) DynamicIterator(*this); } } } // anon namespace MemTableRep* HashSkipListRepFactory::CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) { return new HashSkipListRep(compare, allocator, transform, bucket_count_, skiplist_height_, skiplist_branching_factor_); } MemTableRepFactory* NewHashSkipListRepFactory( size_t bucket_count, int32_t skiplist_height, int32_t skiplist_branching_factor) { return new HashSkipListRepFactory(bucket_count, skiplist_height, skiplist_branching_factor); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/memtable/hash_skiplist_rep.h000066400000000000000000000026351267530634000206360ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include "rocksdb/slice_transform.h" #include "rocksdb/memtablerep.h" namespace rocksdb { class HashSkipListRepFactory : public MemTableRepFactory { public: explicit HashSkipListRepFactory( size_t bucket_count, int32_t skiplist_height, int32_t skiplist_branching_factor) : bucket_count_(bucket_count), skiplist_height_(skiplist_height), skiplist_branching_factor_(skiplist_branching_factor) { } virtual ~HashSkipListRepFactory() {} virtual MemTableRep* CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) override; virtual const char* Name() const override { return "HashSkipListRepFactory"; } private: const size_t bucket_count_; const int32_t skiplist_height_; const int32_t skiplist_branching_factor_; }; } #endif // ROCKSDB_LITE rocksdb-4.5.1/memtable/skiplistrep.cc000066400000000000000000000173511267530634000176330ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include "db/inlineskiplist.h" #include "db/memtable.h" #include "rocksdb/memtablerep.h" #include "util/arena.h" namespace rocksdb { namespace { class SkipListRep : public MemTableRep { InlineSkipList skip_list_; const MemTableRep::KeyComparator& cmp_; const SliceTransform* transform_; const size_t lookahead_; friend class LookaheadIterator; public: explicit SkipListRep(const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, const size_t lookahead) : MemTableRep(allocator), skip_list_(compare, allocator), cmp_(compare), transform_(transform), lookahead_(lookahead) { } virtual KeyHandle Allocate(const size_t len, char** buf) override { *buf = skip_list_.AllocateKey(len); return static_cast(*buf); } // Insert key into the list. // REQUIRES: nothing that compares equal to key is currently in the list. virtual void Insert(KeyHandle handle) override { skip_list_.Insert(static_cast(handle)); } virtual void InsertConcurrently(KeyHandle handle) override { skip_list_.InsertConcurrently(static_cast(handle)); } // Returns true iff an entry that compares equal to key is in the list. virtual bool Contains(const char* key) const override { return skip_list_.Contains(key); } virtual size_t ApproximateMemoryUsage() override { // All memory is allocated through allocator; nothing to report here return 0; } virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) override { SkipListRep::Iterator iter(&skip_list_); Slice dummy_slice; for (iter.Seek(dummy_slice, k.memtable_key().data()); iter.Valid() && callback_func(callback_args, iter.key()); iter.Next()) { } } uint64_t ApproximateNumEntries(const Slice& start_ikey, const Slice& end_ikey) override { std::string tmp; uint64_t start_count = skip_list_.EstimateCount(EncodeKey(&tmp, start_ikey)); uint64_t end_count = skip_list_.EstimateCount(EncodeKey(&tmp, end_ikey)); return (end_count >= start_count) ? (end_count - start_count) : 0; } virtual ~SkipListRep() override { } // Iteration over the contents of a skip list class Iterator : public MemTableRep::Iterator { InlineSkipList::Iterator iter_; public: // Initialize an iterator over the specified list. // The returned iterator is not valid. explicit Iterator( const InlineSkipList* list) : iter_(list) {} virtual ~Iterator() override { } // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const override { return iter_.Valid(); } // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const override { return iter_.key(); } // Advances to the next position. // REQUIRES: Valid() virtual void Next() override { iter_.Next(); } // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() override { iter_.Prev(); } // Advance to the first entry with a key >= target virtual void Seek(const Slice& user_key, const char* memtable_key) override { if (memtable_key != nullptr) { iter_.Seek(memtable_key); } else { iter_.Seek(EncodeKey(&tmp_, user_key)); } } // Position at the first entry in list. // Final state of iterator is Valid() iff list is not empty. virtual void SeekToFirst() override { iter_.SeekToFirst(); } // Position at the last entry in list. // Final state of iterator is Valid() iff list is not empty. virtual void SeekToLast() override { iter_.SeekToLast(); } protected: std::string tmp_; // For passing to EncodeKey }; // Iterator over the contents of a skip list which also keeps track of the // previously visited node. In Seek(), it examines a few nodes after it // first, falling back to O(log n) search from the head of the list only if // the target key hasn't been found. class LookaheadIterator : public MemTableRep::Iterator { public: explicit LookaheadIterator(const SkipListRep& rep) : rep_(rep), iter_(&rep_.skip_list_), prev_(iter_) {} virtual ~LookaheadIterator() override {} virtual bool Valid() const override { return iter_.Valid(); } virtual const char *key() const override { assert(Valid()); return iter_.key(); } virtual void Next() override { assert(Valid()); bool advance_prev = true; if (prev_.Valid()) { auto k1 = rep_.UserKey(prev_.key()); auto k2 = rep_.UserKey(iter_.key()); if (k1.compare(k2) == 0) { // same user key, don't move prev_ advance_prev = false; } else if (rep_.transform_) { // only advance prev_ if it has the same prefix as iter_ auto t1 = rep_.transform_->Transform(k1); auto t2 = rep_.transform_->Transform(k2); advance_prev = t1.compare(t2) == 0; } } if (advance_prev) { prev_ = iter_; } iter_.Next(); } virtual void Prev() override { assert(Valid()); iter_.Prev(); prev_ = iter_; } virtual void Seek(const Slice& internal_key, const char *memtable_key) override { const char *encoded_key = (memtable_key != nullptr) ? memtable_key : EncodeKey(&tmp_, internal_key); if (prev_.Valid() && rep_.cmp_(encoded_key, prev_.key()) >= 0) { // prev_.key() is smaller or equal to our target key; do a quick // linear search (at most lookahead_ steps) starting from prev_ iter_ = prev_; size_t cur = 0; while (cur++ <= rep_.lookahead_ && iter_.Valid()) { if (rep_.cmp_(encoded_key, iter_.key()) <= 0) { return; } Next(); } } iter_.Seek(encoded_key); prev_ = iter_; } virtual void SeekToFirst() override { iter_.SeekToFirst(); prev_ = iter_; } virtual void SeekToLast() override { iter_.SeekToLast(); prev_ = iter_; } protected: std::string tmp_; // For passing to EncodeKey private: const SkipListRep& rep_; InlineSkipList::Iterator iter_; InlineSkipList::Iterator prev_; }; virtual MemTableRep::Iterator* GetIterator(Arena* arena = nullptr) override { if (lookahead_ > 0) { void *mem = arena ? arena->AllocateAligned(sizeof(SkipListRep::LookaheadIterator)) : operator new(sizeof(SkipListRep::LookaheadIterator)); return new (mem) SkipListRep::LookaheadIterator(*this); } else { void *mem = arena ? arena->AllocateAligned(sizeof(SkipListRep::Iterator)) : operator new(sizeof(SkipListRep::Iterator)); return new (mem) SkipListRep::Iterator(&skip_list_); } } }; } MemTableRep* SkipListFactory::CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform* transform, Logger* logger) { return new SkipListRep(compare, allocator, transform, lookahead_); } } // namespace rocksdb rocksdb-4.5.1/memtable/stl_wrappers.h000066400000000000000000000016231267530634000176440ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "rocksdb/comparator.h" #include "rocksdb/memtablerep.h" #include "rocksdb/slice.h" #include "util/coding.h" #include "util/murmurhash.h" namespace rocksdb { namespace stl_wrappers { class Base { protected: const MemTableRep::KeyComparator& compare_; explicit Base(const MemTableRep::KeyComparator& compare) : compare_(compare) {} }; struct Compare : private Base { explicit Compare(const MemTableRep::KeyComparator& compare) : Base(compare) {} inline bool operator()(const char* a, const char* b) const { return compare_(a, b) < 0; } }; } } rocksdb-4.5.1/memtable/vectorrep.cc000066400000000000000000000212121267530634000172620ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #ifndef ROCKSDB_LITE #include "rocksdb/memtablerep.h" #include #include #include #include #include #include "util/arena.h" #include "db/memtable.h" #include "memtable/stl_wrappers.h" #include "port/port.h" #include "util/mutexlock.h" namespace rocksdb { namespace { using namespace stl_wrappers; class VectorRep : public MemTableRep { public: VectorRep(const KeyComparator& compare, MemTableAllocator* allocator, size_t count); // Insert key into the collection. (The caller will pack key and value into a // single buffer and pass that in as the parameter to Insert) // REQUIRES: nothing that compares equal to key is currently in the // collection. virtual void Insert(KeyHandle handle) override; // Returns true iff an entry that compares equal to key is in the collection. virtual bool Contains(const char* key) const override; virtual void MarkReadOnly() override; virtual size_t ApproximateMemoryUsage() override; virtual void Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) override; virtual ~VectorRep() override { } class Iterator : public MemTableRep::Iterator { class VectorRep* vrep_; std::shared_ptr> bucket_; std::vector::const_iterator mutable cit_; const KeyComparator& compare_; std::string tmp_; // For passing to EncodeKey bool mutable sorted_; void DoSort() const; public: explicit Iterator(class VectorRep* vrep, std::shared_ptr> bucket, const KeyComparator& compare); // Initialize an iterator over the specified collection. // The returned iterator is not valid. // explicit Iterator(const MemTableRep* collection); virtual ~Iterator() override { }; // Returns true iff the iterator is positioned at a valid node. virtual bool Valid() const override; // Returns the key at the current position. // REQUIRES: Valid() virtual const char* key() const override; // Advances to the next position. // REQUIRES: Valid() virtual void Next() override; // Advances to the previous position. // REQUIRES: Valid() virtual void Prev() override; // Advance to the first entry with a key >= target virtual void Seek(const Slice& user_key, const char* memtable_key) override; // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToFirst() override; // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. virtual void SeekToLast() override; }; // Return an iterator over the keys in this representation. virtual MemTableRep::Iterator* GetIterator(Arena* arena) override; private: friend class Iterator; typedef std::vector Bucket; std::shared_ptr bucket_; mutable port::RWMutex rwlock_; bool immutable_; bool sorted_; const KeyComparator& compare_; }; void VectorRep::Insert(KeyHandle handle) { auto* key = static_cast(handle); WriteLock l(&rwlock_); assert(!immutable_); bucket_->push_back(key); } // Returns true iff an entry that compares equal to key is in the collection. bool VectorRep::Contains(const char* key) const { ReadLock l(&rwlock_); return std::find(bucket_->begin(), bucket_->end(), key) != bucket_->end(); } void VectorRep::MarkReadOnly() { WriteLock l(&rwlock_); immutable_ = true; } size_t VectorRep::ApproximateMemoryUsage() { return sizeof(bucket_) + sizeof(*bucket_) + bucket_->size() * sizeof( std::remove_reference::type::value_type ); } VectorRep::VectorRep(const KeyComparator& compare, MemTableAllocator* allocator, size_t count) : MemTableRep(allocator), bucket_(new Bucket()), immutable_(false), sorted_(false), compare_(compare) { bucket_.get()->reserve(count); } VectorRep::Iterator::Iterator(class VectorRep* vrep, std::shared_ptr> bucket, const KeyComparator& compare) : vrep_(vrep), bucket_(bucket), cit_(bucket_->end()), compare_(compare), sorted_(false) { } void VectorRep::Iterator::DoSort() const { // vrep is non-null means that we are working on an immutable memtable if (!sorted_ && vrep_ != nullptr) { WriteLock l(&vrep_->rwlock_); if (!vrep_->sorted_) { std::sort(bucket_->begin(), bucket_->end(), Compare(compare_)); cit_ = bucket_->begin(); vrep_->sorted_ = true; } sorted_ = true; } if (!sorted_) { std::sort(bucket_->begin(), bucket_->end(), Compare(compare_)); cit_ = bucket_->begin(); sorted_ = true; } assert(sorted_); assert(vrep_ == nullptr || vrep_->sorted_); } // Returns true iff the iterator is positioned at a valid node. bool VectorRep::Iterator::Valid() const { DoSort(); return cit_ != bucket_->end(); } // Returns the key at the current position. // REQUIRES: Valid() const char* VectorRep::Iterator::key() const { assert(sorted_); return *cit_; } // Advances to the next position. // REQUIRES: Valid() void VectorRep::Iterator::Next() { assert(sorted_); if (cit_ == bucket_->end()) { return; } ++cit_; } // Advances to the previous position. // REQUIRES: Valid() void VectorRep::Iterator::Prev() { assert(sorted_); if (cit_ == bucket_->begin()) { // If you try to go back from the first element, the iterator should be // invalidated. So we set it to past-the-end. This means that you can // treat the container circularly. cit_ = bucket_->end(); } else { --cit_; } } // Advance to the first entry with a key >= target void VectorRep::Iterator::Seek(const Slice& user_key, const char* memtable_key) { DoSort(); // Do binary search to find first value not less than the target const char* encoded_key = (memtable_key != nullptr) ? memtable_key : EncodeKey(&tmp_, user_key); cit_ = std::equal_range(bucket_->begin(), bucket_->end(), encoded_key, [this] (const char* a, const char* b) { return compare_(a, b) < 0; }).first; } // Position at the first entry in collection. // Final state of iterator is Valid() iff collection is not empty. void VectorRep::Iterator::SeekToFirst() { DoSort(); cit_ = bucket_->begin(); } // Position at the last entry in collection. // Final state of iterator is Valid() iff collection is not empty. void VectorRep::Iterator::SeekToLast() { DoSort(); cit_ = bucket_->end(); if (bucket_->size() != 0) { --cit_; } } void VectorRep::Get(const LookupKey& k, void* callback_args, bool (*callback_func)(void* arg, const char* entry)) { rwlock_.ReadLock(); VectorRep* vector_rep; std::shared_ptr bucket; if (immutable_) { vector_rep = this; } else { vector_rep = nullptr; bucket.reset(new Bucket(*bucket_)); // make a copy } VectorRep::Iterator iter(vector_rep, immutable_ ? bucket_ : bucket, compare_); rwlock_.ReadUnlock(); for (iter.Seek(k.user_key(), k.memtable_key().data()); iter.Valid() && callback_func(callback_args, iter.key()); iter.Next()) { } } MemTableRep::Iterator* VectorRep::GetIterator(Arena* arena) { char* mem = nullptr; if (arena != nullptr) { mem = arena->AllocateAligned(sizeof(Iterator)); } ReadLock l(&rwlock_); // Do not sort here. The sorting would be done the first time // a Seek is performed on the iterator. if (immutable_) { if (arena == nullptr) { return new Iterator(this, bucket_, compare_); } else { return new (mem) Iterator(this, bucket_, compare_); } } else { std::shared_ptr tmp; tmp.reset(new Bucket(*bucket_)); // make a copy if (arena == nullptr) { return new Iterator(nullptr, tmp, compare_); } else { return new (mem) Iterator(nullptr, tmp, compare_); } } } } // anon namespace MemTableRep* VectorRepFactory::CreateMemTableRep( const MemTableRep::KeyComparator& compare, MemTableAllocator* allocator, const SliceTransform*, Logger* logger) { return new VectorRep(compare, allocator, count_); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/port/000077500000000000000000000000001267530634000141425ustar00rootroot00000000000000rocksdb-4.5.1/port/README000066400000000000000000000006251267530634000150250ustar00rootroot00000000000000This directory contains interfaces and implementations that isolate the rest of the package from platform details. Code in the rest of the package includes "port.h" from this directory. "port.h" in turn includes a platform specific "port_.h" file that provides the platform specific implementation. See port_posix.h for an example of what must be provided in a platform specific header file. rocksdb-4.5.1/port/dirent.h000066400000000000000000000022271267530634000156030ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // See port_example.h for documentation for the following types/functions. #ifndef STORAGE_LEVELDB_PORT_DIRENT_H_ #define STORAGE_LEVELDB_PORT_DIRENT_H_ #ifdef ROCKSDB_PLATFORM_POSIX #include #include #elif defined(OS_WIN) namespace rocksdb { namespace port { struct dirent { char d_name[_MAX_PATH]; /* filename */ }; struct DIR; DIR* opendir(const char* name); dirent* readdir(DIR* dirp); int closedir(DIR* dirp); } // namespace port using port::dirent; using port::DIR; using port::opendir; using port::readdir; using port::closedir; } // namespace rocksdb #endif // OS_WIN #endif // STORAGE_LEVELDB_PORT_DIRENT_H_ rocksdb-4.5.1/port/likely.h000066400000000000000000000014141267530634000156040ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef PORT_LIKELY_H_ #define PORT_LIKELY_H_ #if defined(__GNUC__) && __GNUC__ >= 4 #define LIKELY(x) (__builtin_expect((x), 1)) #define UNLIKELY(x) (__builtin_expect((x), 0)) #else #define LIKELY(x) (x) #define UNLIKELY(x) (x) #endif #endif // PORT_LIKELY_H_ rocksdb-4.5.1/port/port.h000066400000000000000000000015401267530634000152770ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include // Include the appropriate platform specific file below. If you are // porting to a new platform, see "port_example.h" for documentation // of what the new port_.h file must provide. #if defined(ROCKSDB_PLATFORM_POSIX) #include "port/port_posix.h" #elif defined(OS_WIN) #include "port/win/port_win.h" #endif rocksdb-4.5.1/port/port_example.h000066400000000000000000000072611267530634000170200ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // This file contains the specification, but not the implementations, // of the types/operations/etc. that should be defined by a platform // specific port_.h file. Use this file as a reference for // how to port this package to a new platform. #ifndef STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ #define STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ namespace rocksdb { namespace port { // TODO(jorlow): Many of these belong more in the environment class rather than // here. We should try moving them and see if it affects perf. // The following boolean constant must be true on a little-endian machine // and false otherwise. static const bool kLittleEndian = true /* or some other expression */; // ------------------ Threading ------------------- // A Mutex represents an exclusive lock. class Mutex { public: Mutex(); ~Mutex(); // Lock the mutex. Waits until other lockers have exited. // Will deadlock if the mutex is already locked by this thread. void Lock(); // Unlock the mutex. // REQUIRES: This mutex was locked by this thread. void Unlock(); // Optionally crash if this thread does not hold this mutex. // The implementation must be fast, especially if NDEBUG is // defined. The implementation is allowed to skip all checks. void AssertHeld(); }; class CondVar { public: explicit CondVar(Mutex* mu); ~CondVar(); // Atomically release *mu and block on this condition variable until // either a call to SignalAll(), or a call to Signal() that picks // this thread to wakeup. // REQUIRES: this thread holds *mu void Wait(); // If there are some threads waiting, wake up at least one of them. void Signal(); // Wake up all waiting threads. void SignallAll(); }; // Thread-safe initialization. // Used as follows: // static port::OnceType init_control = LEVELDB_ONCE_INIT; // static void Initializer() { ... do something ...; } // ... // port::InitOnce(&init_control, &Initializer); typedef intptr_t OnceType; #define LEVELDB_ONCE_INIT 0 extern void InitOnce(port::OnceType*, void (*initializer)()); // ------------------ Compression ------------------- // Store the snappy compression of "input[0,input_length-1]" in *output. // Returns false if snappy is not supported by this port. extern bool Snappy_Compress(const char* input, size_t input_length, std::string* output); // If input[0,input_length-1] looks like a valid snappy compressed // buffer, store the size of the uncompressed data in *result and // return true. Else return false. extern bool Snappy_GetUncompressedLength(const char* input, size_t length, size_t* result); // Attempt to snappy uncompress input[0,input_length-1] into *output. // Returns true if successful, false if the input is invalid lightweight // compressed data. // // REQUIRES: at least the first "n" bytes of output[] must be writable // where "n" is the result of a successful call to // Snappy_GetUncompressedLength. extern bool Snappy_Uncompress(const char* input_data, size_t input_length, char* output); } // namespace port } // namespace rocksdb #endif // STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ rocksdb-4.5.1/port/port_posix.cc000066400000000000000000000112701267530634000166600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "port/port_posix.h" #include #if defined(__i386__) || defined(__x86_64__) #include #endif #include #include #include #include #include #include #include #include #include "util/logging.h" namespace rocksdb { namespace port { static int PthreadCall(const char* label, int result) { if (result != 0 && result != ETIMEDOUT) { fprintf(stderr, "pthread %s: %s\n", label, strerror(result)); abort(); } return result; } Mutex::Mutex(bool adaptive) { #ifdef OS_LINUX if (!adaptive) { PthreadCall("init mutex", pthread_mutex_init(&mu_, nullptr)); } else { pthread_mutexattr_t mutex_attr; PthreadCall("init mutex attr", pthread_mutexattr_init(&mutex_attr)); PthreadCall("set mutex attr", pthread_mutexattr_settype(&mutex_attr, PTHREAD_MUTEX_ADAPTIVE_NP)); PthreadCall("init mutex", pthread_mutex_init(&mu_, &mutex_attr)); PthreadCall("destroy mutex attr", pthread_mutexattr_destroy(&mutex_attr)); } #else // ignore adaptive for non-linux platform PthreadCall("init mutex", pthread_mutex_init(&mu_, nullptr)); #endif // OS_LINUX } Mutex::~Mutex() { PthreadCall("destroy mutex", pthread_mutex_destroy(&mu_)); } void Mutex::Lock() { PthreadCall("lock", pthread_mutex_lock(&mu_)); #ifndef NDEBUG locked_ = true; #endif } void Mutex::Unlock() { #ifndef NDEBUG locked_ = false; #endif PthreadCall("unlock", pthread_mutex_unlock(&mu_)); } void Mutex::AssertHeld() { #ifndef NDEBUG assert(locked_); #endif } CondVar::CondVar(Mutex* mu) : mu_(mu) { PthreadCall("init cv", pthread_cond_init(&cv_, nullptr)); } CondVar::~CondVar() { PthreadCall("destroy cv", pthread_cond_destroy(&cv_)); } void CondVar::Wait() { #ifndef NDEBUG mu_->locked_ = false; #endif PthreadCall("wait", pthread_cond_wait(&cv_, &mu_->mu_)); #ifndef NDEBUG mu_->locked_ = true; #endif } bool CondVar::TimedWait(uint64_t abs_time_us) { struct timespec ts; ts.tv_sec = static_cast(abs_time_us / 1000000); ts.tv_nsec = static_cast((abs_time_us % 1000000) * 1000); #ifndef NDEBUG mu_->locked_ = false; #endif int err = pthread_cond_timedwait(&cv_, &mu_->mu_, &ts); #ifndef NDEBUG mu_->locked_ = true; #endif if (err == ETIMEDOUT) { return true; } if (err != 0) { PthreadCall("timedwait", err); } return false; } void CondVar::Signal() { PthreadCall("signal", pthread_cond_signal(&cv_)); } void CondVar::SignalAll() { PthreadCall("broadcast", pthread_cond_broadcast(&cv_)); } RWMutex::RWMutex() { PthreadCall("init mutex", pthread_rwlock_init(&mu_, nullptr)); } RWMutex::~RWMutex() { PthreadCall("destroy mutex", pthread_rwlock_destroy(&mu_)); } void RWMutex::ReadLock() { PthreadCall("read lock", pthread_rwlock_rdlock(&mu_)); } void RWMutex::WriteLock() { PthreadCall("write lock", pthread_rwlock_wrlock(&mu_)); } void RWMutex::ReadUnlock() { PthreadCall("read unlock", pthread_rwlock_unlock(&mu_)); } void RWMutex::WriteUnlock() { PthreadCall("write unlock", pthread_rwlock_unlock(&mu_)); } int PhysicalCoreID() { #if defined(__i386__) || defined(__x86_64__) // if you ever find that this function is hot on Linux, you can go from // ~200 nanos to ~20 nanos by adding the machinery to use __vdso_getcpu unsigned eax, ebx = 0, ecx, edx; __get_cpuid(1, &eax, &ebx, &ecx, &edx); return ebx >> 24; #else // getcpu or sched_getcpu could work here return -1; #endif } void InitOnce(OnceType* once, void (*initializer)()) { PthreadCall("once", pthread_once(once, initializer)); } void Crash(const std::string& srcfile, int srcline) { fprintf(stdout, "Crashing at %s:%d\n", srcfile.c_str(), srcline); fflush(stdout); kill(getpid(), SIGTERM); } int GetMaxOpenFiles() { #if defined(RLIMIT_NOFILE) struct rlimit no_files_limit; if (getrlimit(RLIMIT_NOFILE, &no_files_limit) != 0) { return -1; } // protect against overflow if (no_files_limit.rlim_cur >= std::numeric_limits::max()) { return std::numeric_limits::max(); } return static_cast(no_files_limit.rlim_cur); #endif return -1; } } // namespace port } // namespace rocksdb rocksdb-4.5.1/port/port_posix.h000066400000000000000000000110141267530634000165160ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // See port_example.h for documentation for the following types/functions. #pragma once // size_t printf formatting named in the manner of C99 standard formatting // strings such as PRIu64 // in fact, we could use that one #define ROCKSDB_PRIszt "zu" #define __declspec(S) #define ROCKSDB_NOEXCEPT noexcept #undef PLATFORM_IS_LITTLE_ENDIAN #if defined(OS_MACOSX) #include #if defined(__DARWIN_LITTLE_ENDIAN) && defined(__DARWIN_BYTE_ORDER) #define PLATFORM_IS_LITTLE_ENDIAN \ (__DARWIN_BYTE_ORDER == __DARWIN_LITTLE_ENDIAN) #endif #elif defined(OS_SOLARIS) #include #ifdef _LITTLE_ENDIAN #define PLATFORM_IS_LITTLE_ENDIAN true #else #define PLATFORM_IS_LITTLE_ENDIAN false #endif #elif defined(OS_FREEBSD) || defined(OS_OPENBSD) || defined(OS_NETBSD) || \ defined(OS_DRAGONFLYBSD) || defined(OS_ANDROID) #include #include #define PLATFORM_IS_LITTLE_ENDIAN (_BYTE_ORDER == _LITTLE_ENDIAN) #else #include #endif #include #include #include #include #include #ifndef PLATFORM_IS_LITTLE_ENDIAN #define PLATFORM_IS_LITTLE_ENDIAN (__BYTE_ORDER == __LITTLE_ENDIAN) #endif #if defined(OS_MACOSX) || defined(OS_SOLARIS) || defined(OS_FREEBSD) ||\ defined(OS_NETBSD) || defined(OS_OPENBSD) || defined(OS_DRAGONFLYBSD) ||\ defined(OS_ANDROID) || defined(CYGWIN) // Use fread/fwrite/fflush on platforms without _unlocked variants #define fread_unlocked fread #define fwrite_unlocked fwrite #define fflush_unlocked fflush #endif #if defined(OS_MACOSX) || defined(OS_FREEBSD) ||\ defined(OS_OPENBSD) || defined(OS_DRAGONFLYBSD) // Use fsync() on platforms without fdatasync() #define fdatasync fsync #endif #if defined(OS_ANDROID) && __ANDROID_API__ < 9 // fdatasync() was only introduced in API level 9 on Android. Use fsync() // when targeting older platforms. #define fdatasync fsync #endif namespace rocksdb { namespace port { // For use at db/file_indexer.h kLevelMaxIndex const int kMaxInt32 = std::numeric_limits::max(); const uint64_t kMaxUint64 = std::numeric_limits::max(); const size_t kMaxSizet = std::numeric_limits::max(); static const bool kLittleEndian = PLATFORM_IS_LITTLE_ENDIAN; #undef PLATFORM_IS_LITTLE_ENDIAN class CondVar; class Mutex { public: /* implicit */ Mutex(bool adaptive = false); ~Mutex(); void Lock(); void Unlock(); // this will assert if the mutex is not locked // it does NOT verify that mutex is held by a calling thread void AssertHeld(); private: friend class CondVar; pthread_mutex_t mu_; #ifndef NDEBUG bool locked_; #endif // No copying Mutex(const Mutex&); void operator=(const Mutex&); }; class RWMutex { public: RWMutex(); ~RWMutex(); void ReadLock(); void WriteLock(); void ReadUnlock(); void WriteUnlock(); void AssertHeld() { } private: pthread_rwlock_t mu_; // the underlying platform mutex // No copying allowed RWMutex(const RWMutex&); void operator=(const RWMutex&); }; class CondVar { public: explicit CondVar(Mutex* mu); ~CondVar(); void Wait(); // Timed condition wait. Returns true if timeout occurred. bool TimedWait(uint64_t abs_time_us); void Signal(); void SignalAll(); private: pthread_cond_t cv_; Mutex* mu_; }; static inline void AsmVolatilePause() { #if defined(__i386__) || defined(__x86_64__) asm volatile("pause"); #elif defined(__aarch64__) asm volatile("wfe"); #elif defined(__powerpc64__) asm volatile("or 27,27,27"); #endif // it's okay for other platforms to be no-ops } // Returns -1 if not available on this platform extern int PhysicalCoreID(); typedef pthread_once_t OnceType; #define LEVELDB_ONCE_INIT PTHREAD_ONCE_INIT extern void InitOnce(OnceType* once, void (*initializer)()); #define CACHE_LINE_SIZE 64U #define PREFETCH(addr, rw, locality) __builtin_prefetch(addr, rw, locality) extern void Crash(const std::string& srcfile, int srcline); extern int GetMaxOpenFiles(); } // namespace port } // namespace rocksdb rocksdb-4.5.1/port/stack_trace.cc000066400000000000000000000066501267530634000167430ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include "port/stack_trace.h" #if defined(ROCKSDB_LITE) || !(defined(OS_LINUX) || defined(OS_MACOSX)) || \ defined(CYGWIN) // noop namespace rocksdb { namespace port { void InstallStackTraceHandler() {} void PrintStack(int first_frames_to_skip) {} } // namespace port } // namespace rocksdb #else #include #include #include #include #include #include #include namespace rocksdb { namespace port { namespace { #ifdef OS_LINUX const char* GetExecutableName() { static char name[1024]; char link[1024]; snprintf(link, sizeof(link), "/proc/%d/exe", getpid()); auto read = readlink(link, name, sizeof(name) - 1); if (-1 == read) { return nullptr; } else { name[read] = 0; return name; } } void PrintStackTraceLine(const char* symbol, void* frame) { static const char* executable = GetExecutableName(); if (symbol) { fprintf(stderr, "%s ", symbol); } if (executable) { // out source to addr2line, for the address translation const int kLineMax = 256; char cmd[kLineMax]; snprintf(cmd, kLineMax, "addr2line %p -e %s -f -C 2>&1", frame, executable); auto f = popen(cmd, "r"); if (f) { char line[kLineMax]; while (fgets(line, sizeof(line), f)) { line[strlen(line) - 1] = 0; // remove newline fprintf(stderr, "%s\t", line); } pclose(f); } } else { fprintf(stderr, " %p", frame); } fprintf(stderr, "\n"); } #elif defined(OS_MACOSX) void PrintStackTraceLine(const char* symbol, void* frame) { static int pid = getpid(); // out source to atos, for the address translation const int kLineMax = 256; char cmd[kLineMax]; snprintf(cmd, kLineMax, "xcrun atos %p -p %d 2>&1", frame, pid); auto f = popen(cmd, "r"); if (f) { char line[kLineMax]; while (fgets(line, sizeof(line), f)) { line[strlen(line) - 1] = 0; // remove newline fprintf(stderr, "%s\t", line); } pclose(f); } else if (symbol) { fprintf(stderr, "%s ", symbol); } fprintf(stderr, "\n"); } #endif } // namespace void PrintStack(int first_frames_to_skip) { const int kMaxFrames = 100; void* frames[kMaxFrames]; auto num_frames = backtrace(frames, kMaxFrames); auto symbols = backtrace_symbols(frames, num_frames); for (int i = first_frames_to_skip; i < num_frames; ++i) { fprintf(stderr, "#%-2d ", i - first_frames_to_skip); PrintStackTraceLine((symbols != nullptr) ? symbols[i] : nullptr, frames[i]); } } static void StackTraceHandler(int sig) { // reset to default handler signal(sig, SIG_DFL); fprintf(stderr, "Received signal %d (%s)\n", sig, strsignal(sig)); // skip the top three signal handler related frames PrintStack(3); // re-signal to default handler (so we still get core dump if needed...) raise(sig); } void InstallStackTraceHandler() { // just use the plain old signal as it's simple and sufficient // for this use case signal(SIGILL, StackTraceHandler); signal(SIGSEGV, StackTraceHandler); signal(SIGBUS, StackTraceHandler); signal(SIGABRT, StackTraceHandler); } } // namespace port } // namespace rocksdb #endif rocksdb-4.5.1/port/stack_trace.h000066400000000000000000000012461267530634000166010ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #pragma once namespace rocksdb { namespace port { // Install a signal handler to print callstack on the following signals: // SIGILL SIGSEGV SIGBUS SIGABRT // Currently supports linux only. No-op otherwise. void InstallStackTraceHandler(); // Prints stack, skips skip_first_frames frames void PrintStack(int first_frames_to_skip = 0); } // namespace port } // namespace rocksdb rocksdb-4.5.1/port/sys_time.h000066400000000000000000000024161267530634000161520ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // This file is a portable substitute for sys/time.h which does not exist on // Windows #ifndef STORAGE_LEVELDB_PORT_SYS_TIME_H_ #define STORAGE_LEVELDB_PORT_SYS_TIME_H_ #if defined(OS_WIN) && defined(_MSC_VER) #include namespace rocksdb { namespace port { // Avoid including winsock2.h for this definition typedef struct timeval { long tv_sec; long tv_usec; } timeval; void gettimeofday(struct timeval* tv, struct timezone* tz); inline struct tm* localtime_r(const time_t* timep, struct tm* result) { errno_t ret = localtime_s(result, timep); return (ret == 0) ? result : NULL; } } using port::timeval; using port::gettimeofday; using port::localtime_r; } #else #include #include #endif #endif // STORAGE_LEVELDB_PORT_SYS_TIME_H_ rocksdb-4.5.1/port/util_logger.h000066400000000000000000000017151267530634000166330ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef STORAGE_LEVELDB_PORT_UTIL_LOGGER_H_ #define STORAGE_LEVELDB_PORT_UTIL_LOGGER_H_ // Include the appropriate platform specific file below. If you are // porting to a new platform, see "port_example.h" for documentation // of what the new port_.h file must provide. #if defined(ROCKSDB_PLATFORM_POSIX) #include "util/posix_logger.h" #elif defined(OS_WIN) #include "port/win/win_logger.h" #endif #endif // STORAGE_LEVELDB_PORT_UTIL_LOGGER_H_ rocksdb-4.5.1/port/win/000077500000000000000000000000001267530634000147375ustar00rootroot00000000000000rocksdb-4.5.1/port/win/env_win.cc000066400000000000000000002065121267530634000167210ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include #include #include #include #include #include #include #include #include #include #include "rocksdb/env.h" #include "rocksdb/slice.h" #include "port/port.h" #include "port/dirent.h" #include "port/win/win_logger.h" #include "util/random.h" #include "util/iostats_context_imp.h" #include "util/rate_limiter.h" #include "util/sync_point.h" #include "util/aligned_buffer.h" #include "util/thread_status_updater.h" #include "util/thread_status_util.h" #include // For UUID generation #include namespace rocksdb { std::string GetWindowsErrSz(DWORD err) { LPSTR lpMsgBuf; FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS, NULL, err, 0, // Default language reinterpret_cast(&lpMsgBuf), 0, NULL); std::string Err = lpMsgBuf; LocalFree(lpMsgBuf); return Err; } namespace { const size_t c_OneMB = (1 << 20); ThreadStatusUpdater* CreateThreadStatusUpdater() { return new ThreadStatusUpdater(); } inline Status IOErrorFromWindowsError(const std::string& context, DWORD err) { return Status::IOError(context, GetWindowsErrSz(err)); } inline Status IOErrorFromLastWindowsError(const std::string& context) { return IOErrorFromWindowsError(context, GetLastError()); } inline Status IOError(const std::string& context, int err_number) { return Status::IOError(context, strerror(err_number)); } // TODO(sdong): temp logging. Need to help debugging. Remove it when // the feature is proved to be stable. inline void PrintThreadInfo(size_t thread_id, size_t terminatingId) { fprintf(stdout, "Bg thread %Iu terminates %Iu\n", thread_id, terminatingId); } // returns the ID of the current process inline int current_process_id() { return _getpid(); } // RAII helpers for HANDLEs const auto CloseHandleFunc = [](HANDLE h) { ::CloseHandle(h); }; typedef std::unique_ptr UniqueCloseHandlePtr; // We preserve the original name of this interface to denote the original idea // behind it. // All reads happen by a specified offset and pwrite interface does not change // the position of the file pointer. Judging from the man page and errno it does // execute // lseek atomically to return the position of the file back where it was. // WriteFile() does not // have this capability. Therefore, for both pread and pwrite the pointer is // advanced to the next position // which is fine for writes because they are (should be) sequential. // Because all the reads/writes happen by the specified offset, the caller in // theory should not // rely on the current file offset. SSIZE_T pwrite(HANDLE hFile, const char* src, size_t numBytes, uint64_t offset) { assert(numBytes <= std::numeric_limits::max()); OVERLAPPED overlapped = {0}; ULARGE_INTEGER offsetUnion; offsetUnion.QuadPart = offset; overlapped.Offset = offsetUnion.LowPart; overlapped.OffsetHigh = offsetUnion.HighPart; SSIZE_T result = 0; unsigned long bytesWritten = 0; if (FALSE == WriteFile(hFile, src, static_cast(numBytes), &bytesWritten, &overlapped)) { result = -1; } else { result = bytesWritten; } return result; } // See comments for pwrite above SSIZE_T pread(HANDLE hFile, char* src, size_t numBytes, uint64_t offset) { assert(numBytes <= std::numeric_limits::max()); OVERLAPPED overlapped = {0}; ULARGE_INTEGER offsetUnion; offsetUnion.QuadPart = offset; overlapped.Offset = offsetUnion.LowPart; overlapped.OffsetHigh = offsetUnion.HighPart; SSIZE_T result = 0; unsigned long bytesRead = 0; if (FALSE == ReadFile(hFile, src, static_cast(numBytes), &bytesRead, &overlapped)) { return -1; } else { result = bytesRead; } return result; } // Note the below two do not set errno because they are used only here in this // file // on a Windows handle and, therefore, not necessary. Translating GetLastError() // to errno // is a sad business inline int fsync(HANDLE hFile) { if (!FlushFileBuffers(hFile)) { return -1; } return 0; } // SetFileInformationByHandle() is capable of fast pre-allocates. // However, this does not change the file end position unless the file is // truncated and the pre-allocated space is not considered filled with zeros. inline Status fallocate(const std::string& filename, HANDLE hFile, uint64_t to_size) { Status status; FILE_ALLOCATION_INFO alloc_info; alloc_info.AllocationSize.QuadPart = to_size; if (!SetFileInformationByHandle(hFile, FileAllocationInfo, &alloc_info, sizeof(FILE_ALLOCATION_INFO))) { auto lastError = GetLastError(); status = IOErrorFromWindowsError( "Failed to pre-allocate space: " + filename, lastError); } return status; } inline Status ftruncate(const std::string& filename, HANDLE hFile, uint64_t toSize) { Status status; FILE_END_OF_FILE_INFO end_of_file; end_of_file.EndOfFile.QuadPart = toSize; if (!SetFileInformationByHandle(hFile, FileEndOfFileInfo, &end_of_file, sizeof(FILE_END_OF_FILE_INFO))) { auto lastError = GetLastError(); status = IOErrorFromWindowsError("Failed to Set end of file: " + filename, lastError); } return status; } // mmap() based random-access class WinMmapReadableFile : public RandomAccessFile { const std::string fileName_; HANDLE hFile_; HANDLE hMap_; const void* mapped_region_; const size_t length_; public: // mapped_region_[0,length-1] contains the mmapped contents of the file. WinMmapReadableFile(const std::string& fileName, HANDLE hFile, HANDLE hMap, const void* mapped_region, size_t length) : fileName_(fileName), hFile_(hFile), hMap_(hMap), mapped_region_(mapped_region), length_(length) {} ~WinMmapReadableFile() { BOOL ret = ::UnmapViewOfFile(mapped_region_); assert(ret); ret = ::CloseHandle(hMap_); assert(ret); ret = ::CloseHandle(hFile_); assert(ret); } virtual Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const override { Status s; if (offset > length_) { *result = Slice(); return IOError(fileName_, EINVAL); } else if (offset + n > length_) { n = length_ - offset; } *result = Slice(reinterpret_cast(mapped_region_) + offset, n); return s; } virtual Status InvalidateCache(size_t offset, size_t length) override { return Status::OK(); } }; // We preallocate up to an extra megabyte and use memcpy to append new // data to the file. This is safe since we either properly close the // file before reading from it, or for log files, the reading code // knows enough to skip zero suffixes. class WinMmapFile : public WritableFile { private: const std::string filename_; HANDLE hFile_; HANDLE hMap_; const size_t page_size_; // We flush the mapping view in page_size // increments. We may decide if this is a memory // page size or SSD page size const size_t allocation_granularity_; // View must start at such a granularity size_t mapping_size_; // We want file mapping to be of a specific size // because then the file is expandable size_t view_size_; // How much memory to map into a view at a time char* mapped_begin_; // Must begin at the file offset that is aligned with // allocation_granularity_ char* mapped_end_; char* dst_; // Where to write next (in range [mapped_begin_,mapped_end_]) char* last_sync_; // Where have we synced up to uint64_t file_offset_; // Offset of mapped_begin_ in file // Do we have unsynced writes? bool pending_sync_; // Can only truncate or reserve to a sector size aligned if // used on files that are opened with Unbuffered I/O Status TruncateFile(uint64_t toSize) { return ftruncate(filename_, hFile_, toSize); } // Can only truncate or reserve to a sector size aligned if // used on files that are opened with Unbuffered I/O // Normally it does not present a problem since in memory mapped files // we do not disable buffering Status ReserveFileSpace(uint64_t toSize) { IOSTATS_TIMER_GUARD(allocate_nanos); return fallocate(filename_, hFile_, toSize); } Status UnmapCurrentRegion() { Status status; if (mapped_begin_ != nullptr) { if (!::UnmapViewOfFile(mapped_begin_)) { status = IOErrorFromWindowsError( "Failed to unmap file view: " + filename_, GetLastError()); } // UnmapView automatically sends data to disk but not the metadata // which is good and provides some equivalent of fdatasync() on Linux // therefore, we donot need separate flag for metadata pending_sync_ = false; mapped_begin_ = nullptr; mapped_end_ = nullptr; dst_ = nullptr; last_sync_ = nullptr; // Move on to the next portion of the file file_offset_ += view_size_; // Increase the amount we map the next time, but capped at 1MB view_size_ *= 2; view_size_ = std::min(view_size_, c_OneMB); } return status; } Status MapNewRegion() { Status status; assert(mapped_begin_ == nullptr); size_t minMappingSize = file_offset_ + view_size_; // Check if we need to create a new mapping since we want to write beyond // the current one // If the mapping view is now too short // CreateFileMapping will extend the size of the file automatically if the // mapping size is greater than // the current length of the file, which reserves the space and makes // writing faster, except, windows can not map an empty file. // Thus the first time around we must actually extend the file ourselves if (hMap_ == NULL || minMappingSize > mapping_size_) { if (NULL == hMap_) { // Creating mapping for the first time so reserve the space on disk status = ReserveFileSpace(minMappingSize); if (!status.ok()) { return status; } } if (hMap_) { // Unmap the previous one BOOL ret = ::CloseHandle(hMap_); assert(ret); hMap_ = NULL; } // Calculate the new mapping size which will hopefully reserve space for // several consecutive sliding views // Query preallocation block size if set size_t preallocationBlockSize = 0; size_t lastAllocatedBlockSize = 0; // Not used GetPreallocationStatus(&preallocationBlockSize, &lastAllocatedBlockSize); if (preallocationBlockSize) { preallocationBlockSize = Roundup(preallocationBlockSize, allocation_granularity_); } else { preallocationBlockSize = 2 * view_size_; } mapping_size_ += preallocationBlockSize; ULARGE_INTEGER mappingSize; mappingSize.QuadPart = mapping_size_; hMap_ = CreateFileMappingA( hFile_, NULL, // Security attributes PAGE_READWRITE, // There is not a write only mode for mapping mappingSize.HighPart, // Enable mapping the whole file but the actual // amount mapped is determined by MapViewOfFile mappingSize.LowPart, NULL); // Mapping name if (NULL == hMap_) { return IOErrorFromWindowsError( "WindowsMmapFile failed to create file mapping for: " + filename_, GetLastError()); } } ULARGE_INTEGER offset; offset.QuadPart = file_offset_; // View must begin at the granularity aligned offset mapped_begin_ = reinterpret_cast( MapViewOfFileEx(hMap_, FILE_MAP_WRITE, offset.HighPart, offset.LowPart, view_size_, NULL)); if (!mapped_begin_) { status = IOErrorFromWindowsError( "WindowsMmapFile failed to map file view: " + filename_, GetLastError()); } else { mapped_end_ = mapped_begin_ + view_size_; dst_ = mapped_begin_; last_sync_ = mapped_begin_; pending_sync_ = false; } return status; } public: WinMmapFile(const std::string& fname, HANDLE hFile, size_t page_size, size_t allocation_granularity, const EnvOptions& options) : filename_(fname), hFile_(hFile), hMap_(NULL), page_size_(page_size), allocation_granularity_(allocation_granularity), mapping_size_(0), view_size_(0), mapped_begin_(nullptr), mapped_end_(nullptr), dst_(nullptr), last_sync_(nullptr), file_offset_(0), pending_sync_(false) { // Allocation granularity must be obtained from GetSystemInfo() and must be // a power of two. assert(allocation_granularity > 0); assert((allocation_granularity & (allocation_granularity - 1)) == 0); assert(page_size > 0); assert((page_size & (page_size - 1)) == 0); // Only for memory mapped writes assert(options.use_mmap_writes); // Make sure buffering is not disabled. It is ignored for mapping // purposes but also imposes restriction on moving file position // it is not a problem so much with reserving space since it is probably a // factor // of allocation_granularity but we also want to truncate the file in // Close() at // arbitrary position so we do not have to feel this with zeros. assert(options.use_os_buffer); // View size must be both the multiple of allocation_granularity AND the // page size if ((allocation_granularity_ % page_size_) == 0) { view_size_ = 2 * allocation_granularity; } else if ((page_size_ % allocation_granularity_) == 0) { view_size_ = 2 * page_size_; } else { // we can multiply them together assert(false); } } ~WinMmapFile() { if (hFile_) { this->Close(); } } virtual Status Append(const Slice& data) override { const char* src = data.data(); size_t left = data.size(); while (left > 0) { assert(mapped_begin_ <= dst_); size_t avail = mapped_end_ - dst_; if (avail == 0) { Status s = UnmapCurrentRegion(); if (s.ok()) { s = MapNewRegion(); } if (!s.ok()) { return s; } } size_t n = std::min(left, avail); memcpy(dst_, src, n); dst_ += n; src += n; left -= n; pending_sync_ = true; } return Status::OK(); } // Means Close() will properly take care of truncate // and it does not need any additional information virtual Status Truncate(uint64_t size) override { return Status::OK(); } virtual Status Close() override { Status s; assert(NULL != hFile_); // We truncate to the precise size so no // uninitialized data at the end. SetEndOfFile // which we use does not write zeros and it is good. uint64_t targetSize = GetFileSize(); s = UnmapCurrentRegion(); if (NULL != hMap_) { BOOL ret = ::CloseHandle(hMap_); if (!ret && s.ok()) { auto lastError = GetLastError(); s = IOErrorFromWindowsError( "Failed to Close mapping for file: " + filename_, lastError); } hMap_ = NULL; } TruncateFile(targetSize); BOOL ret = ::CloseHandle(hFile_); hFile_ = NULL; if (!ret && s.ok()) { auto lastError = GetLastError(); s = IOErrorFromWindowsError( "Failed to close file map handle: " + filename_, lastError); } return s; } virtual Status Flush() override { return Status::OK(); } // Flush only data virtual Status Sync() override { Status s; // Some writes occurred since last sync if (pending_sync_) { assert(mapped_begin_); assert(dst_); assert(dst_ > mapped_begin_); assert(dst_ < mapped_end_); size_t page_begin = TruncateToPageBoundary(page_size_, last_sync_ - mapped_begin_); size_t page_end = TruncateToPageBoundary(page_size_, dst_ - mapped_begin_ - 1); last_sync_ = dst_; // Flush only the amount of that is a multiple of pages if (!::FlushViewOfFile(mapped_begin_ + page_begin, (page_end - page_begin) + page_size_)) { s = IOErrorFromWindowsError("Failed to FlushViewOfFile: " + filename_, GetLastError()); } pending_sync_ = false; } return s; } /** * Flush data as well as metadata to stable storage. */ virtual Status Fsync() override { Status s; // Flush metadata if pending const bool pending = pending_sync_; s = Sync(); // Flush metadata if (s.ok() && pending) { if (!::FlushFileBuffers(hFile_)) { s = IOErrorFromWindowsError("Failed to FlushFileBuffers: " + filename_, GetLastError()); } } return s; } /** * Get the size of valid data in the file. This will not match the * size that is returned from the filesystem because we use mmap * to extend file by map_size every time. */ virtual uint64_t GetFileSize() override { size_t used = dst_ - mapped_begin_; return file_offset_ + used; } virtual Status InvalidateCache(size_t offset, size_t length) override { return Status::OK(); } virtual Status Allocate(uint64_t offset, uint64_t len) override { return Status::OK(); } }; class WinSequentialFile : public SequentialFile { private: const std::string filename_; HANDLE file_; // There is no equivalent of advising away buffered pages as in posix. // To implement this flag we would need to do unbuffered reads which // will need to be aligned (not sure there is a guarantee that the buffer // passed in is aligned). // Hence we currently ignore this flag. It is used only in a few cases // which should not be perf critical. // If perf evaluation finds this to be a problem, we can look into // implementing this. bool use_os_buffer_; public: WinSequentialFile(const std::string& fname, HANDLE f, const EnvOptions& options) : filename_(fname), file_(f), use_os_buffer_(options.use_os_buffer) {} virtual ~WinSequentialFile() { assert(file_ != INVALID_HANDLE_VALUE); CloseHandle(file_); } virtual Status Read(size_t n, Slice* result, char* scratch) override { Status s; size_t r = 0; // Windows ReadFile API accepts a DWORD. // While it is possible to read in a loop if n is > UINT_MAX // it is a highly unlikely case. if (n > UINT_MAX) { return IOErrorFromWindowsError(filename_, ERROR_INVALID_PARAMETER); } DWORD bytesToRead = static_cast(n); //cast is safe due to the check above DWORD bytesRead = 0; BOOL ret = ReadFile(file_, scratch, bytesToRead, &bytesRead, NULL); if (ret == TRUE) { r = bytesRead; } else { return IOErrorFromWindowsError(filename_, GetLastError()); } *result = Slice(scratch, r); return s; } virtual Status Skip(uint64_t n) override { // Can't handle more than signed max as SetFilePointerEx accepts a signed 64-bit // integer. As such it is a highly unlikley case to have n so large. if (n > _I64_MAX) { return IOErrorFromWindowsError(filename_, ERROR_INVALID_PARAMETER); } LARGE_INTEGER li; li.QuadPart = static_cast(n); //cast is safe due to the check above BOOL ret = SetFilePointerEx(file_, li, NULL, FILE_CURRENT); if (ret == FALSE) { return IOErrorFromWindowsError(filename_, GetLastError()); } return Status::OK(); } virtual Status InvalidateCache(size_t offset, size_t length) override { return Status::OK(); } }; // pread() based random-access class WinRandomAccessFile : public RandomAccessFile { const std::string filename_; HANDLE hFile_; const bool use_os_buffer_; bool read_ahead_; const size_t compaction_readahead_size_; const size_t random_access_max_buffer_size_; mutable std::mutex buffer_mut_; mutable AlignedBuffer buffer_; mutable uint64_t buffered_start_; // file offset set that is currently buffered /* * The function reads a requested amount of bytes into the specified aligned * buffer Upon success the function sets the length of the buffer to the * amount of bytes actually read even though it might be less than actually * requested. It then copies the amount of bytes requested by the user (left) * to the user supplied buffer (dest) and reduces left by the amount of bytes * copied to the user buffer * * @user_offset [in] - offset on disk where the read was requested by the user * @first_page_start [in] - actual page aligned disk offset that we want to * read from * @bytes_to_read [in] - total amount of bytes that will be read from disk * which is generally greater or equal to the amount * that the user has requested due to the * either alignment requirements or read_ahead in * effect. * @left [in/out] total amount of bytes that needs to be copied to the user * buffer. It is reduced by the amount of bytes that actually * copied * @buffer - buffer to use * @dest - user supplied buffer */ SSIZE_T ReadIntoBuffer(uint64_t user_offset, uint64_t first_page_start, size_t bytes_to_read, size_t& left, AlignedBuffer& buffer, char* dest) const { assert(buffer.CurrentSize() == 0); assert(buffer.Capacity() >= bytes_to_read); SSIZE_T read = pread(hFile_, buffer.Destination(), bytes_to_read, first_page_start); if (read > 0) { buffer.Size(read); // Let's figure out how much we read from the users standpoint if ((first_page_start + buffer.CurrentSize()) > user_offset) { assert(first_page_start <= user_offset); size_t buffer_offset = user_offset - first_page_start; read = buffer.Read(dest, buffer_offset, left); } else { read = 0; } left -= read; } return read; } SSIZE_T ReadIntoOneShotBuffer(uint64_t user_offset, uint64_t first_page_start, size_t bytes_to_read, size_t& left, char* dest) const { AlignedBuffer bigBuffer; bigBuffer.Alignment(buffer_.Alignment()); bigBuffer.AllocateNewBuffer(bytes_to_read); return ReadIntoBuffer(user_offset, first_page_start, bytes_to_read, left, bigBuffer, dest); } SSIZE_T ReadIntoInstanceBuffer(uint64_t user_offset, uint64_t first_page_start, size_t bytes_to_read, size_t& left, char* dest) const { SSIZE_T read = ReadIntoBuffer(user_offset, first_page_start, bytes_to_read, left, buffer_, dest); if (read > 0) { buffered_start_ = first_page_start; } return read; } void CalculateReadParameters(uint64_t offset, size_t bytes_requested, size_t& actual_bytes_toread, uint64_t& first_page_start) const { const size_t alignment = buffer_.Alignment(); first_page_start = TruncateToPageBoundary(alignment, offset); const uint64_t last_page_start = TruncateToPageBoundary(alignment, offset + bytes_requested - 1); actual_bytes_toread = (last_page_start - first_page_start) + alignment; } public: WinRandomAccessFile(const std::string& fname, HANDLE hFile, size_t alignment, const EnvOptions& options) : filename_(fname), hFile_(hFile), use_os_buffer_(options.use_os_buffer), read_ahead_(false), compaction_readahead_size_(options.compaction_readahead_size), random_access_max_buffer_size_(options.random_access_max_buffer_size), buffer_(), buffered_start_(0) { assert(!options.use_mmap_reads); // Unbuffered access, use internal buffer for reads if (!use_os_buffer_) { // Do not allocate the buffer either until the first request or // until there is a call to allocate a read-ahead buffer buffer_.Alignment(alignment); } } virtual ~WinRandomAccessFile() { if (hFile_ != NULL && hFile_ != INVALID_HANDLE_VALUE) { ::CloseHandle(hFile_); } } virtual void EnableReadAhead() override { this->Hint(SEQUENTIAL); } virtual Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const override { Status s; SSIZE_T r = -1; size_t left = n; char* dest = scratch; if (n == 0) { *result = Slice(scratch, 0); return s; } // When in unbuffered mode we need to do the following changes: // - use our own aligned buffer // - always read at the offset of that is a multiple of alignment if (!use_os_buffer_) { uint64_t first_page_start = 0; size_t actual_bytes_toread = 0; size_t bytes_requested = left; if (!read_ahead_ && random_access_max_buffer_size_ == 0) { CalculateReadParameters(offset, bytes_requested, actual_bytes_toread, first_page_start); assert(actual_bytes_toread > 0); r = ReadIntoOneShotBuffer(offset, first_page_start, actual_bytes_toread, left, dest); } else { std::unique_lock lock(buffer_mut_); // Let's see if at least some of the requested data is already // in the buffer if (offset >= buffered_start_ && offset < (buffered_start_ + buffer_.CurrentSize())) { size_t buffer_offset = offset - buffered_start_; r = buffer_.Read(dest, buffer_offset, left); assert(r >= 0); left -= size_t(r); offset += r; dest += r; } // Still some left or none was buffered if (left > 0) { // Figure out the start/end offset for reading and amount to read bytes_requested = left; if (read_ahead_ && bytes_requested < compaction_readahead_size_) { bytes_requested = compaction_readahead_size_; } CalculateReadParameters(offset, bytes_requested, actual_bytes_toread, first_page_start); assert(actual_bytes_toread > 0); if (buffer_.Capacity() < actual_bytes_toread) { // If we are in read-ahead mode or the requested size // exceeds max buffer size then use one-shot // big buffer otherwise reallocate main buffer if (read_ahead_ || (actual_bytes_toread > random_access_max_buffer_size_)) { // Unlock the mutex since we are not using instance buffer lock.unlock(); r = ReadIntoOneShotBuffer(offset, first_page_start, actual_bytes_toread, left, dest); } else { buffer_.AllocateNewBuffer(actual_bytes_toread); r = ReadIntoInstanceBuffer(offset, first_page_start, actual_bytes_toread, left, dest); } } else { buffer_.Clear(); r = ReadIntoInstanceBuffer(offset, first_page_start, actual_bytes_toread, left, dest); } } } } else { r = pread(hFile_, scratch, left, offset); if (r > 0) { left -= r; } } *result = Slice(scratch, (r < 0) ? 0 : n - left); if (r < 0) { s = IOErrorFromLastWindowsError(filename_); } return s; } virtual bool ShouldForwardRawRequest() const override { return true; } virtual void Hint(AccessPattern pattern) override { if (pattern == SEQUENTIAL && !use_os_buffer_ && compaction_readahead_size_ > 0) { std::lock_guard lg(buffer_mut_); if (!read_ahead_) { read_ahead_ = true; // This would allocate read-ahead size + 2 alignments // - one for memory alignment which added implicitly by AlignedBuffer // - We add one more alignment because we will read one alignment more // from disk buffer_.AllocateNewBuffer(compaction_readahead_size_ + buffer_.Alignment()); } } } virtual Status InvalidateCache(size_t offset, size_t length) override { return Status::OK(); } }; // This is a sequential write class. It has been mimicked (as others) after // the original Posix class. We add support for unbuffered I/O on windows as // well // we utilize the original buffer as an alignment buffer to write directly to // file with no buffering. // No buffering requires that the provided buffer is aligned to the physical // sector size (SSD page size) and // that all SetFilePointer() operations to occur with such an alignment. // We thus always write in sector/page size increments to the drive and leave // the tail for the next write OR for Close() at which point we pad with zeros. // No padding is required for // buffered access. class WinWritableFile : public WritableFile { private: const std::string filename_; HANDLE hFile_; const bool use_os_buffer_; // Used to indicate unbuffered access, the file const uint64_t alignment_; // must be opened as unbuffered if false uint64_t filesize_; // How much data is actually written disk uint64_t reservedsize_; // how far we have reserved space public: WinWritableFile(const std::string& fname, HANDLE hFile, size_t alignment, size_t capacity, const EnvOptions& options) : filename_(fname), hFile_(hFile), use_os_buffer_(options.use_os_buffer), alignment_(alignment), filesize_(0), reservedsize_(0) { assert(!options.use_mmap_writes); } ~WinWritableFile() { if (NULL != hFile_ && INVALID_HANDLE_VALUE != hFile_) { WinWritableFile::Close(); } } // Indicates if the class makes use of unbuffered I/O virtual bool UseOSBuffer() const override { return use_os_buffer_; } virtual size_t GetRequiredBufferAlignment() const override { return alignment_; } virtual Status Append(const Slice& data) override { // Used for buffered access ONLY assert(use_os_buffer_); assert(data.size() < std::numeric_limits::max()); Status s; DWORD bytesWritten = 0; if (!WriteFile(hFile_, data.data(), static_cast(data.size()), &bytesWritten, NULL)) { auto lastError = GetLastError(); s = IOErrorFromWindowsError( "Failed to WriteFile: " + filename_, lastError); } else { assert(size_t(bytesWritten) == data.size()); filesize_ += data.size(); } return s; } virtual Status PositionedAppend(const Slice& data, uint64_t offset) override { Status s; SSIZE_T ret = pwrite(hFile_, data.data(), data.size(), offset); // Error break if (ret < 0) { auto lastError = GetLastError(); s = IOErrorFromWindowsError( "Failed to pwrite for: " + filename_, lastError); } else { // With positional write it is not clear at all // if this actually extends the filesize assert(size_t(ret) == data.size()); filesize_ += data.size(); } return s; } // Need to implement this so the file is truncated correctly // when buffered and unbuffered mode virtual Status Truncate(uint64_t size) override { Status s = ftruncate(filename_, hFile_, size); if (s.ok()) { filesize_ = size; } return s; } virtual Status Close() override { Status s; assert(INVALID_HANDLE_VALUE != hFile_); if (fsync(hFile_) < 0) { auto lastError = GetLastError(); s = IOErrorFromWindowsError("fsync failed at Close() for: " + filename_, lastError); } if (FALSE == ::CloseHandle(hFile_)) { auto lastError = GetLastError(); s = IOErrorFromWindowsError("CloseHandle failed for: " + filename_, lastError); } hFile_ = INVALID_HANDLE_VALUE; return s; } // write out the cached data to the OS cache // This is now taken care of the WritableFileWriter virtual Status Flush() override { return Status::OK(); } virtual Status Sync() override { Status s; // Calls flush buffers if (fsync(hFile_) < 0) { auto lastError = GetLastError(); s = IOErrorFromWindowsError("fsync failed at Sync() for: " + filename_, lastError); } return s; } virtual Status Fsync() override { return Sync(); } virtual uint64_t GetFileSize() override { // Double accounting now here with WritableFileWriter // and this size will be wrong when unbuffered access is used // but tests implement their own writable files and do not use WritableFileWrapper // so we need to squeeze a square peg through // a round hole here. return filesize_; } virtual Status Allocate(uint64_t offset, uint64_t len) override { Status status; TEST_KILL_RANDOM("WinWritableFile::Allocate", rocksdb_kill_odds); // Make sure that we reserve an aligned amount of space // since the reservation block size is driven outside so we want // to check if we are ok with reservation here size_t spaceToReserve = Roundup(offset + len, alignment_); // Nothing to do if (spaceToReserve <= reservedsize_) { return status; } IOSTATS_TIMER_GUARD(allocate_nanos); status = fallocate(filename_, hFile_, spaceToReserve); if (status.ok()) { reservedsize_ = spaceToReserve; } return status; } }; class WinDirectory : public Directory { public: WinDirectory() {} virtual Status Fsync() override { return Status::OK(); } }; class WinFileLock : public FileLock { public: explicit WinFileLock(HANDLE hFile) : hFile_(hFile) { assert(hFile != NULL); assert(hFile != INVALID_HANDLE_VALUE); } ~WinFileLock() { BOOL ret = ::CloseHandle(hFile_); assert(ret); } private: HANDLE hFile_; }; namespace { void WinthreadCall(const char* label, std::error_code result) { if (0 != result.value()) { fprintf(stderr, "pthread %s: %s\n", label, strerror(result.value())); abort(); } } } typedef VOID(WINAPI * FnGetSystemTimePreciseAsFileTime)(LPFILETIME); class WinEnv : public Env { public: WinEnv(); virtual ~WinEnv() { for (auto& th : threads_to_join_) { th.join(); } threads_to_join_.clear(); for (auto& thpool : thread_pools_) { thpool.JoinAllThreads(); } // All threads must be joined before the deletion of // thread_status_updater_. delete thread_status_updater_; } virtual Status DeleteFile(const std::string& fname) override { Status result; if (_unlink(fname.c_str())) { result = IOError("Failed to delete: " + fname, errno); } return result; } Status GetCurrentTime(int64_t* unix_time) override { time_t time = std::time(nullptr); if (time == (time_t)(-1)) { return Status::NotSupported("Failed to get time"); } *unix_time = time; return Status::OK(); } virtual Status NewSequentialFile(const std::string& fname, std::unique_ptr* result, const EnvOptions& options) override { Status s; result->reset(); // Corruption test needs to rename and delete files of these kind // while they are still open with another handle. For that reason we // allow share_write and delete(allows rename). HANDLE hFile = INVALID_HANDLE_VALUE; { IOSTATS_TIMER_GUARD(open_nanos); hFile = CreateFileA( fname.c_str(), GENERIC_READ, FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, OPEN_EXISTING, // Original fopen mode is "rb" FILE_ATTRIBUTE_NORMAL, NULL); } if (INVALID_HANDLE_VALUE == hFile) { auto lastError = GetLastError(); s = IOErrorFromWindowsError("Failed to open NewSequentialFile" + fname, lastError); } else { result->reset(new WinSequentialFile(fname, hFile, options)); } return s; } virtual Status NewRandomAccessFile(const std::string& fname, std::unique_ptr* result, const EnvOptions& options) override { result->reset(); Status s; // Open the file for read-only random access // Random access is to disable read-ahead as the system reads too much data DWORD fileFlags = FILE_ATTRIBUTE_READONLY; if (!options.use_os_buffer && !options.use_mmap_reads) { fileFlags |= FILE_FLAG_NO_BUFFERING; } else { fileFlags |= FILE_FLAG_RANDOM_ACCESS; } /// Shared access is necessary for corruption test to pass // almost all tests would work with a possible exception of fault_injection HANDLE hFile = 0; { IOSTATS_TIMER_GUARD(open_nanos); hFile = CreateFileA(fname.c_str(), GENERIC_READ, FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, OPEN_EXISTING, fileFlags, NULL); } if (INVALID_HANDLE_VALUE == hFile) { auto lastError = GetLastError(); return IOErrorFromWindowsError( "NewRandomAccessFile failed to Create/Open: " + fname, lastError); } UniqueCloseHandlePtr fileGuard(hFile, CloseHandleFunc); // CAUTION! This will map the entire file into the process address space if (options.use_mmap_reads && sizeof(void*) >= 8) { // Use mmap when virtual address-space is plentiful. uint64_t fileSize; s = GetFileSize(fname, &fileSize); if (s.ok()) { // Will not map empty files if (fileSize == 0) { return IOError( "NewRandomAccessFile failed to map empty file: " + fname, EINVAL); } HANDLE hMap = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, // Whole file at its present length 0, NULL); // Mapping name if (!hMap) { auto lastError = GetLastError(); return IOErrorFromWindowsError( "Failed to create file mapping for NewRandomAccessFile: " + fname, lastError); } UniqueCloseHandlePtr mapGuard(hMap, CloseHandleFunc); const void* mapped_region = MapViewOfFileEx(hMap, FILE_MAP_READ, 0, // High DWORD of access start 0, // Low DWORD fileSize, NULL); // Let the OS choose the mapping if (!mapped_region) { auto lastError = GetLastError(); return IOErrorFromWindowsError( "Failed to MapViewOfFile for NewRandomAccessFile: " + fname, lastError); } result->reset(new WinMmapReadableFile(fname, hFile, hMap, mapped_region, fileSize)); mapGuard.release(); fileGuard.release(); } } else { result->reset(new WinRandomAccessFile(fname, hFile, page_size_, options)); fileGuard.release(); } return s; } virtual Status NewWritableFile(const std::string& fname, std::unique_ptr* result, const EnvOptions& options) override { const size_t c_BufferCapacity = 64 * 1024; EnvOptions local_options(options); result->reset(); Status s; DWORD fileFlags = FILE_ATTRIBUTE_NORMAL; if (!local_options.use_os_buffer && !local_options.use_mmap_writes) { fileFlags = FILE_FLAG_NO_BUFFERING; } // Desired access. We are want to write only here but if we want to memory // map // the file then there is no write only mode so we have to create it // Read/Write // However, MapViewOfFile specifies only Write only DWORD desired_access = GENERIC_WRITE; DWORD shared_mode = FILE_SHARE_READ; if (local_options.use_mmap_writes) { desired_access |= GENERIC_READ; } else { // Adding this solely for tests to pass (fault_injection_test, // wal_manager_test). shared_mode |= (FILE_SHARE_WRITE | FILE_SHARE_DELETE); } HANDLE hFile = 0; { IOSTATS_TIMER_GUARD(open_nanos); hFile = CreateFileA( fname.c_str(), desired_access, // Access desired shared_mode, NULL, // Security attributes CREATE_ALWAYS, // Posix env says O_CREAT | O_RDWR | O_TRUNC fileFlags, // Flags NULL); // Template File } if (INVALID_HANDLE_VALUE == hFile) { auto lastError = GetLastError(); return IOErrorFromWindowsError( "Failed to create a NewWriteableFile: " + fname, lastError); } if (options.use_mmap_writes) { // We usually do not use mmmapping on SSD and thus we pass memory // page_size result->reset(new WinMmapFile(fname, hFile, page_size_, allocation_granularity_, local_options)); } else { // Here we want the buffer allocation to be aligned by the SSD page size // and to be a multiple of it result->reset(new WinWritableFile(fname, hFile, page_size_, c_BufferCapacity, local_options)); } return s; } virtual Status NewDirectory(const std::string& name, std::unique_ptr* result) override { Status s; // Must be nullptr on failure result->reset(); // Must fail if directory does not exist if (!DirExists(name)) { s = IOError("Directory does not exist: " + name, EEXIST); } else { IOSTATS_TIMER_GUARD(open_nanos); result->reset(new WinDirectory); } return s; } virtual Status FileExists(const std::string& fname) override { // F_OK == 0 const int F_OK_ = 0; return _access(fname.c_str(), F_OK_) == 0 ? Status::OK() : Status::NotFound(); } virtual Status GetChildren(const std::string& dir, std::vector* result) override { std::vector output; Status status; auto CloseDir = [](DIR* p) { closedir(p); }; std::unique_ptr dirp(opendir(dir.c_str()), CloseDir); if (!dirp) { status = IOError(dir, errno); } else { if (result->capacity() > 0) { output.reserve(result->capacity()); } struct dirent* ent = readdir(dirp.get()); while (ent) { output.push_back(ent->d_name); ent = readdir(dirp.get()); } } output.swap(*result); return status; } virtual Status CreateDir(const std::string& name) override { Status result; if (_mkdir(name.c_str()) != 0) { auto code = errno; result = IOError("Failed to create dir: " + name, code); } return result; } virtual Status CreateDirIfMissing(const std::string& name) override { Status result; if (DirExists(name)) { return result; } if (_mkdir(name.c_str()) != 0) { if (errno == EEXIST) { result = Status::IOError("`" + name + "' exists but is not a directory"); } else { auto code = errno; result = IOError("Failed to create dir: " + name, code); } } return result; } virtual Status DeleteDir(const std::string& name) override { Status result; if (_rmdir(name.c_str()) != 0) { auto code = errno; result = IOError("Failed to remove dir: " + name, code); } return result; } virtual Status GetFileSize(const std::string& fname, uint64_t* size) override { Status s; WIN32_FILE_ATTRIBUTE_DATA attrs; if (GetFileAttributesExA(fname.c_str(), GetFileExInfoStandard, &attrs)) { ULARGE_INTEGER file_size; file_size.HighPart = attrs.nFileSizeHigh; file_size.LowPart = attrs.nFileSizeLow; *size = file_size.QuadPart; } else { auto lastError = GetLastError(); s = IOErrorFromWindowsError("Can not get size for: " + fname, lastError); } return s; } static inline uint64_t FileTimeToUnixTime(const FILETIME& ftTime) { const uint64_t c_FileTimePerSecond = 10000000U; // UNIX epoch starts on 1970-01-01T00:00:00Z // Windows FILETIME starts on 1601-01-01T00:00:00Z // Therefore, we need to subtract the below number of seconds from // the seconds that we obtain from FILETIME with an obvious loss of // precision const uint64_t c_SecondBeforeUnixEpoch = 11644473600U; ULARGE_INTEGER li; li.HighPart = ftTime.dwHighDateTime; li.LowPart = ftTime.dwLowDateTime; uint64_t result = (li.QuadPart / c_FileTimePerSecond) - c_SecondBeforeUnixEpoch; return result; } virtual Status GetFileModificationTime(const std::string& fname, uint64_t* file_mtime) override { Status s; WIN32_FILE_ATTRIBUTE_DATA attrs; if (GetFileAttributesExA(fname.c_str(), GetFileExInfoStandard, &attrs)) { *file_mtime = FileTimeToUnixTime(attrs.ftLastWriteTime); } else { auto lastError = GetLastError(); s = IOErrorFromWindowsError( "Can not get file modification time for: " + fname, lastError); *file_mtime = 0; } return s; } virtual Status RenameFile(const std::string& src, const std::string& target) override { Status result; // rename() is not capable of replacing the existing file as on Linux // so use OS API directly if (!MoveFileExA(src.c_str(), target.c_str(), MOVEFILE_REPLACE_EXISTING)) { DWORD lastError = GetLastError(); std::string text("Failed to rename: "); text.append(src).append(" to: ").append(target); result = IOErrorFromWindowsError(text, lastError); } return result; } virtual Status LinkFile(const std::string& src, const std::string& target) override { Status result; if (!CreateHardLinkA(target.c_str(), src.c_str(), NULL)) { DWORD lastError = GetLastError(); std::string text("Failed to link: "); text.append(src).append(" to: ").append(target); result = IOErrorFromWindowsError(text, lastError); } return result; } virtual Status LockFile(const std::string& lockFname, FileLock** lock) override { assert(lock != nullptr); *lock = NULL; Status result; // No-sharing, this is a LOCK file const DWORD ExclusiveAccessON = 0; // Obtain exclusive access to the LOCK file // Previously, instead of NORMAL attr we set DELETE on close and that worked // well except with fault_injection test that insists on deleting it. HANDLE hFile = 0; { IOSTATS_TIMER_GUARD(open_nanos); hFile = CreateFileA(lockFname.c_str(), (GENERIC_READ | GENERIC_WRITE), ExclusiveAccessON, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); } if (INVALID_HANDLE_VALUE == hFile) { auto lastError = GetLastError(); result = IOErrorFromWindowsError( "Failed to create lock file: " + lockFname, lastError); } else { *lock = new WinFileLock(hFile); } return result; } virtual Status UnlockFile(FileLock* lock) override { Status result; assert(lock != nullptr); delete lock; return result; } virtual void Schedule(void (*function)(void*), void* arg, Priority pri = LOW, void* tag = nullptr, void (*unschedFunction)(void* arg) = 0) override; virtual int UnSchedule(void* arg, Priority pri) override; virtual void StartThread(void (*function)(void* arg), void* arg) override; virtual void WaitForJoin() override; virtual unsigned int GetThreadPoolQueueLen(Priority pri = LOW) const override; virtual Status GetTestDirectory(std::string* result) override { std::string output; const char* env = getenv("TEST_TMPDIR"); if (env && env[0] != '\0') { output = env; CreateDir(output); } else { env = getenv("TMP"); if (env && env[0] != '\0') { output = env; } else { output = "c:\\tmp"; } CreateDir(output); } output.append("\\testrocksdb-"); output.append(std::to_string(_getpid())); CreateDir(output); output.swap(*result); return Status::OK(); } virtual Status GetThreadList( std::vector* thread_list) override { assert(thread_status_updater_); return thread_status_updater_->GetThreadList(thread_list); } static uint64_t gettid() { uint64_t thread_id = GetCurrentThreadId(); return thread_id; } virtual uint64_t GetThreadID() const override { return gettid(); } virtual Status NewLogger(const std::string& fname, std::shared_ptr* result) override { Status s; result->reset(); HANDLE hFile = 0; { IOSTATS_TIMER_GUARD(open_nanos); hFile = CreateFileA( fname.c_str(), GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_DELETE, // In RocksDb log files are // renamed and deleted before // they are closed. This enables // doing so. NULL, CREATE_ALWAYS, // Original fopen mode is "w" FILE_ATTRIBUTE_NORMAL, NULL); } if (INVALID_HANDLE_VALUE == hFile) { auto lastError = GetLastError(); s = IOErrorFromWindowsError("Failed to open LogFile" + fname, lastError); } else { { // With log files we want to set the true creation time as of now // because the system // for some reason caches the attributes of the previous file that just // been renamed from // this name so auto_roll_logger_test fails FILETIME ft; GetSystemTimeAsFileTime(&ft); // Set creation, last access and last write time to the same value SetFileTime(hFile, &ft, &ft, &ft); } result->reset(new WinLogger(&WinEnv::gettid, this, hFile)); } return s; } virtual uint64_t NowMicros() override { if (GetSystemTimePreciseAsFileTime_ != NULL) { // all std::chrono clocks on windows proved to return // values that may repeat that is not good enough for some uses. const int64_t c_UnixEpochStartTicks = 116444736000000000i64; const int64_t c_FtToMicroSec = 10; // This interface needs to return system time and not // just any microseconds because it is often used as an argument // to TimedWait() on condition variable FILETIME ftSystemTime; GetSystemTimePreciseAsFileTime_(&ftSystemTime); LARGE_INTEGER li; li.LowPart = ftSystemTime.dwLowDateTime; li.HighPart = ftSystemTime.dwHighDateTime; // Subtract unix epoch start li.QuadPart -= c_UnixEpochStartTicks; // Convert to microsecs li.QuadPart /= c_FtToMicroSec; return li.QuadPart; } using namespace std::chrono; return duration_cast(system_clock::now().time_since_epoch()).count(); } virtual uint64_t NowNanos() override { // all std::chrono clocks on windows have the same resolution that is only // good enough for microseconds but not nanoseconds // On Windows 8 and Windows 2012 Server // GetSystemTimePreciseAsFileTime(¤t_time) can be used LARGE_INTEGER li; QueryPerformanceCounter(&li); // Convert to nanoseconds first to avoid loss of precision // and divide by frequency li.QuadPart *= std::nano::den; li.QuadPart /= perf_counter_frequency_; return li.QuadPart; } virtual void SleepForMicroseconds(int micros) override { std::this_thread::sleep_for(std::chrono::microseconds(micros)); } virtual Status GetHostName(char* name, uint64_t len) override { Status s; DWORD nSize = static_cast(std::min(len, std::numeric_limits::max())); if (!::GetComputerNameA(name, &nSize)) { auto lastError = GetLastError(); s = IOErrorFromWindowsError("GetHostName", lastError); } else { name[nSize] = 0; } return s; } virtual Status GetCurrTime(int64_t* unix_time) { Status s; time_t ret = time(nullptr); if (ret == (time_t)-1) { *unix_time = 0; s = IOError("GetCurrTime", errno); } else { *unix_time = (int64_t)ret; } return s; } virtual Status GetAbsolutePath(const std::string& db_path, std::string* output_path) override { // Check if we already have an absolute path // that starts with non dot and has a semicolon in it if ((!db_path.empty() && (db_path[0] == '/' || db_path[0] == '\\')) || (db_path.size() > 2 && db_path[0] != '.' && ((db_path[1] == ':' && db_path[2] == '\\') || (db_path[1] == ':' && db_path[2] == '/')))) { *output_path = db_path; return Status::OK(); } std::string result; result.resize(_MAX_PATH); char* ret = _getcwd(&result[0], _MAX_PATH); if (ret == nullptr) { return Status::IOError("Failed to get current working directory", strerror(errno)); } result.resize(strlen(result.data())); result.swap(*output_path); return Status::OK(); } // Allow increasing the number of worker threads. virtual void SetBackgroundThreads(int num, Priority pri) override { assert(pri >= Priority::LOW && pri <= Priority::HIGH); thread_pools_[pri].SetBackgroundThreads(num); } virtual void IncBackgroundThreadsIfNeeded(int num, Priority pri) override { assert(pri >= Priority::LOW && pri <= Priority::HIGH); thread_pools_[pri].IncBackgroundThreadsIfNeeded(num); } virtual std::string TimeToString(uint64_t secondsSince1970) override { std::string result; const time_t seconds = secondsSince1970; const int maxsize = 64; struct tm t; errno_t ret = localtime_s(&t, &seconds); if (ret) { result = std::to_string(seconds); } else { result.resize(maxsize); char* p = &result[0]; int len = snprintf(p, maxsize, "%04d/%02d/%02d-%02d:%02d:%02d ", t.tm_year + 1900, t.tm_mon + 1, t.tm_mday, t.tm_hour, t.tm_min, t.tm_sec); assert(len > 0); result.resize(len); } return result; } EnvOptions OptimizeForLogWrite(const EnvOptions& env_options, const DBOptions& db_options) const override { EnvOptions optimized = env_options; optimized.use_mmap_writes = false; optimized.bytes_per_sync = db_options.wal_bytes_per_sync; optimized.use_os_buffer = true; // This is because we flush only whole pages on unbuffered io and // the last records are not guaranteed to be flushed. // TODO(icanadi) it's faster if fallocate_with_keep_size is false, but it // breaks TransactionLogIteratorStallAtLastRecord unit test. Fix the unit // test and make this false optimized.fallocate_with_keep_size = true; return optimized; } EnvOptions OptimizeForManifestWrite( const EnvOptions& env_options) const override { EnvOptions optimized = env_options; optimized.use_mmap_writes = false; optimized.use_os_buffer = true; optimized.fallocate_with_keep_size = true; return optimized; } private: // Returns true iff the named directory exists and is a directory. virtual bool DirExists(const std::string& dname) { WIN32_FILE_ATTRIBUTE_DATA attrs; if (GetFileAttributesExA(dname.c_str(), GetFileExInfoStandard, &attrs)) { return 0 != (attrs.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY); } return false; } bool SupportsFastAllocate(const std::string& /* path */) { return false; } class ThreadPool { public: ThreadPool() : total_threads_limit_(1), bgthreads_(0), queue_(), queue_len_(0U), exit_all_threads_(false), low_io_priority_(false), env_(nullptr) {} ~ThreadPool() { assert(bgthreads_.size() == 0U); } void JoinAllThreads() { { std::lock_guard lock(mu_); assert(!exit_all_threads_); exit_all_threads_ = true; bgsignal_.notify_all(); } for (std::thread& th : bgthreads_) { th.join(); } // Subject to assert in the __dtor bgthreads_.clear(); } void SetHostEnv(Env* env) { env_ = env; } // Return true if there is at least one thread needs to terminate. bool HasExcessiveThread() const { return bgthreads_.size() > total_threads_limit_; } // Return true iff the current thread is the excessive thread to terminate. // Always terminate the running thread that is added last, even if there are // more than one thread to terminate. bool IsLastExcessiveThread(size_t thread_id) const { return HasExcessiveThread() && thread_id == bgthreads_.size() - 1; } // Is one of the threads to terminate. bool IsExcessiveThread(size_t thread_id) const { return thread_id >= total_threads_limit_; } // Return the thread priority. // This would allow its member-thread to know its priority. Env::Priority GetThreadPriority() { return priority_; } // Set the thread priority. void SetThreadPriority(Env::Priority priority) { priority_ = priority; } void BGThread(size_t thread_id) { while (true) { // Wait until there is an item that is ready to run std::unique_lock uniqueLock(mu_); // Stop waiting if the thread needs to do work or needs to terminate. while (!exit_all_threads_ && !IsLastExcessiveThread(thread_id) && (queue_.empty() || IsExcessiveThread(thread_id))) { bgsignal_.wait(uniqueLock); } if (exit_all_threads_) { // mechanism to let BG threads exit safely uniqueLock.unlock(); break; } if (IsLastExcessiveThread(thread_id)) { // Current thread is the last generated one and is excessive. // We always terminate excessive thread in the reverse order of // generation time. std::thread& terminating_thread = bgthreads_.back(); auto tid = terminating_thread.get_id(); // Ensure that that this thread is ours assert(tid == std::this_thread::get_id()); terminating_thread.detach(); bgthreads_.pop_back(); if (HasExcessiveThread()) { // There is still at least more excessive thread to terminate. WakeUpAllThreads(); } uniqueLock.unlock(); PrintThreadInfo(thread_id, gettid()); break; } void (*function)(void*) = queue_.front().function; void* arg = queue_.front().arg; queue_.pop_front(); queue_len_.store(queue_.size(), std::memory_order_relaxed); uniqueLock.unlock(); (*function)(arg); } } // Helper struct for passing arguments when creating threads. struct BGThreadMetadata { ThreadPool* thread_pool_; size_t thread_id_; // Thread count in the thread. BGThreadMetadata(ThreadPool* thread_pool, size_t thread_id) : thread_pool_(thread_pool), thread_id_(thread_id) {} }; static void* BGThreadWrapper(void* arg) { std::unique_ptr meta( reinterpret_cast(arg)); size_t thread_id = meta->thread_id_; ThreadPool* tp = meta->thread_pool_; #if ROCKSDB_USING_THREAD_STATUS // for thread-status ThreadStatusUtil::RegisterThread( tp->env_, (tp->GetThreadPriority() == Env::Priority::HIGH ? ThreadStatus::HIGH_PRIORITY : ThreadStatus::LOW_PRIORITY)); #endif tp->BGThread(thread_id); #if ROCKSDB_USING_THREAD_STATUS ThreadStatusUtil::UnregisterThread(); #endif return nullptr; } void WakeUpAllThreads() { bgsignal_.notify_all(); } void SetBackgroundThreadsInternal(size_t num, bool allow_reduce) { std::lock_guard lg(mu_); if (exit_all_threads_) { return; } if (num > total_threads_limit_ || (num < total_threads_limit_ && allow_reduce)) { total_threads_limit_ = std::max(size_t(1), num); WakeUpAllThreads(); StartBGThreads(); } assert(total_threads_limit_ > 0); } void IncBackgroundThreadsIfNeeded(int num) { SetBackgroundThreadsInternal(num, false); } void SetBackgroundThreads(int num) { SetBackgroundThreadsInternal(num, true); } void StartBGThreads() { // Start background thread if necessary while (bgthreads_.size() < total_threads_limit_) { std::thread p_t(&ThreadPool::BGThreadWrapper, new BGThreadMetadata(this, bgthreads_.size())); bgthreads_.push_back(std::move(p_t)); } } void Schedule(void (*function)(void* arg1), void* arg, void* tag, void (*unschedFunction)(void* arg)) { std::lock_guard lg(mu_); if (exit_all_threads_) { return; } StartBGThreads(); // Add to priority queue queue_.push_back(BGItem()); queue_.back().function = function; queue_.back().arg = arg; queue_.back().tag = tag; queue_.back().unschedFunction = unschedFunction; queue_len_.store(queue_.size(), std::memory_order_relaxed); if (!HasExcessiveThread()) { // Wake up at least one waiting thread. bgsignal_.notify_one(); } else { // Need to wake up all threads to make sure the one woken // up is not the one to terminate. WakeUpAllThreads(); } } int UnSchedule(void* arg) { int count = 0; std::lock_guard lg(mu_); // Remove from priority queue BGQueue::iterator it = queue_.begin(); while (it != queue_.end()) { if (arg == (*it).tag) { void (*unschedFunction)(void*) = (*it).unschedFunction; void* arg1 = (*it).arg; if (unschedFunction != nullptr) { (*unschedFunction)(arg1); } it = queue_.erase(it); count++; } else { ++it; } } queue_len_.store(queue_.size(), std::memory_order_relaxed); return count; } unsigned int GetQueueLen() const { return static_cast( queue_len_.load(std::memory_order_relaxed)); } private: // Entry per Schedule() call struct BGItem { void* arg; void (*function)(void*); void* tag; void (*unschedFunction)(void*); }; typedef std::deque BGQueue; std::mutex mu_; std::condition_variable bgsignal_; size_t total_threads_limit_; std::vector bgthreads_; BGQueue queue_; std::atomic_size_t queue_len_; // Queue length. Used for stats reporting bool exit_all_threads_; bool low_io_priority_; Env::Priority priority_; Env* env_; }; bool checkedDiskForMmap_; bool forceMmapOff; // do we override Env options? size_t page_size_; size_t allocation_granularity_; uint64_t perf_counter_frequency_; std::vector thread_pools_; mutable std::mutex mu_; std::vector threads_to_join_; FnGetSystemTimePreciseAsFileTime GetSystemTimePreciseAsFileTime_; }; WinEnv::WinEnv() : checkedDiskForMmap_(false), forceMmapOff(false), page_size_(4 * 1012), allocation_granularity_(page_size_), perf_counter_frequency_(0), thread_pools_(Priority::TOTAL), GetSystemTimePreciseAsFileTime_(NULL) { HMODULE module = GetModuleHandle("kernel32.dll"); if (module != NULL) { GetSystemTimePreciseAsFileTime_ = (FnGetSystemTimePreciseAsFileTime)GetProcAddress( module, "GetSystemTimePreciseAsFileTime"); } SYSTEM_INFO sinfo; GetSystemInfo(&sinfo); page_size_ = sinfo.dwPageSize; allocation_granularity_ = sinfo.dwAllocationGranularity; { LARGE_INTEGER qpf; BOOL ret = QueryPerformanceFrequency(&qpf); assert(ret == TRUE); perf_counter_frequency_ = qpf.QuadPart; } for (int pool_id = 0; pool_id < Env::Priority::TOTAL; ++pool_id) { thread_pools_[pool_id].SetThreadPriority( static_cast(pool_id)); // This allows later initializing the thread-local-env of each thread. thread_pools_[pool_id].SetHostEnv(this); } // Protected member of the base class thread_status_updater_ = CreateThreadStatusUpdater(); } void WinEnv::Schedule(void (*function)(void*), void* arg, Priority pri, void* tag, void (*unschedFunction)(void* arg)) { assert(pri >= Priority::LOW && pri <= Priority::HIGH); thread_pools_[pri].Schedule(function, arg, tag, unschedFunction); } int WinEnv::UnSchedule(void* arg, Priority pri) { return thread_pools_[pri].UnSchedule(arg); } unsigned int WinEnv::GetThreadPoolQueueLen(Priority pri) const { assert(pri >= Priority::LOW && pri <= Priority::HIGH); return thread_pools_[pri].GetQueueLen(); } namespace { struct StartThreadState { void (*user_function)(void*); void* arg; }; } static void* StartThreadWrapper(void* arg) { std::unique_ptr state( reinterpret_cast(arg)); state->user_function(state->arg); return nullptr; } void WinEnv::StartThread(void (*function)(void* arg), void* arg) { StartThreadState* state = new StartThreadState; state->user_function = function; state->arg = arg; try { std::thread th(&StartThreadWrapper, state); std::lock_guard lg(mu_); threads_to_join_.push_back(std::move(th)); } catch (const std::system_error& ex) { WinthreadCall("start thread", ex.code()); } } void WinEnv::WaitForJoin() { for (auto& th : threads_to_join_) { th.join(); } threads_to_join_.clear(); } } // namespace std::string Env::GenerateUniqueId() { std::string result; UUID uuid; UuidCreateSequential(&uuid); RPC_CSTR rpc_str; auto status = UuidToStringA(&uuid, &rpc_str); assert(status == RPC_S_OK); result = reinterpret_cast(rpc_str); status = RpcStringFreeA(&rpc_str); assert(status == RPC_S_OK); return result; } // We choose to create this on the heap and using std::once for the following // reasons // 1) Currently available MS compiler does not implement atomic C++11 // initialization of // function local statics // 2) We choose not to destroy the env because joining the threads from the // system loader // which destroys the statics (same as from DLLMain) creates a system loader // dead-lock. // in this manner any remaining threads are terminated OK. namespace { std::once_flag winenv_once_flag; Env* envptr; }; Env* Env::Default() { std::call_once(winenv_once_flag, []() { envptr = new WinEnv(); }); return envptr; } } // namespace rocksdb rocksdb-4.5.1/port/win/port_win.cc000066400000000000000000000156521267530634000171200ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #if !defined(OS_WIN) && !defined(WIN32) && !defined(_WIN32) #error Windows Specific Code #endif #include "port/win/port_win.h" #include #include "port/dirent.h" #include "port/sys_time.h" #include #include #include #include #include #include #include #include "util/logging.h" namespace rocksdb { namespace port { void gettimeofday(struct timeval* tv, struct timezone* /* tz */) { using namespace std::chrono; microseconds usNow( duration_cast(system_clock::now().time_since_epoch())); seconds secNow(duration_cast(usNow)); tv->tv_sec = static_cast(secNow.count()); tv->tv_usec = static_cast(usNow.count() - duration_cast(secNow).count()); } Mutex::~Mutex() {} CondVar::~CondVar() {} void CondVar::Wait() { // Caller must ensure that mutex is held prior to calling this method std::unique_lock lk(mu_->getLock(), std::adopt_lock); #ifndef NDEBUG mu_->locked_ = false; #endif cv_.wait(lk); #ifndef NDEBUG mu_->locked_ = true; #endif // Release ownership of the lock as we don't want it to be unlocked when // it goes out of scope (as we adopted the lock and didn't lock it ourselves) lk.release(); } bool CondVar::TimedWait(uint64_t abs_time_us) { using namespace std::chrono; // MSVC++ library implements wait_until in terms of wait_for so // we need to convert absolute wait into relative wait. microseconds usAbsTime(abs_time_us); microseconds usNow( duration_cast(system_clock::now().time_since_epoch())); microseconds relTimeUs = (usAbsTime > usNow) ? (usAbsTime - usNow) : microseconds::zero(); // Caller must ensure that mutex is held prior to calling this method std::unique_lock lk(mu_->getLock(), std::adopt_lock); #ifndef NDEBUG mu_->locked_ = false; #endif std::cv_status cvStatus = cv_.wait_for(lk, relTimeUs); #ifndef NDEBUG mu_->locked_ = true; #endif // Release ownership of the lock as we don't want it to be unlocked when // it goes out of scope (as we adopted the lock and didn't lock it ourselves) lk.release(); if (cvStatus == std::cv_status::timeout) { return true; } return false; } void CondVar::Signal() { cv_.notify_one(); } void CondVar::SignalAll() { cv_.notify_all(); } int PhysicalCoreID() { return GetCurrentProcessorNumber(); } void InitOnce(OnceType* once, void (*initializer)()) { std::call_once(once->flag_, initializer); } // Private structure, exposed only by pointer struct DIR { intptr_t handle_; bool firstread_; struct __finddata64_t data_; dirent entry_; DIR() : handle_(-1), firstread_(true) {} DIR(const DIR&) = delete; DIR& operator=(const DIR&) = delete; ~DIR() { if (-1 != handle_) { _findclose(handle_); } } }; DIR* opendir(const char* name) { if (!name || *name == 0) { errno = ENOENT; return nullptr; } std::string pattern(name); pattern.append("\\").append("*"); std::unique_ptr dir(new DIR); dir->handle_ = _findfirst64(pattern.c_str(), &dir->data_); if (dir->handle_ == -1) { return nullptr; } strncpy_s(dir->entry_.d_name, dir->data_.name, strlen(dir->data_.name)); return dir.release(); } struct dirent* readdir(DIR* dirp) { if (!dirp || dirp->handle_ == -1) { errno = EBADF; return nullptr; } if (dirp->firstread_) { dirp->firstread_ = false; return &dirp->entry_; } auto ret = _findnext64(dirp->handle_, &dirp->data_); if (ret != 0) { return nullptr; } strncpy_s(dirp->entry_.d_name, dirp->data_.name, strlen(dirp->data_.name)); return &dirp->entry_; } int closedir(DIR* dirp) { delete dirp; return 0; } int truncate(const char* path, int64_t len) { if (path == nullptr) { errno = EFAULT; return -1; } if (len < 0) { errno = EINVAL; return -1; } HANDLE hFile = CreateFile(path, GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, // Security attrs OPEN_EXISTING, // Truncate existing file only FILE_ATTRIBUTE_NORMAL, NULL); if (INVALID_HANDLE_VALUE == hFile) { auto lastError = GetLastError(); if (lastError == ERROR_FILE_NOT_FOUND) { errno = ENOENT; } else if (lastError == ERROR_ACCESS_DENIED) { errno = EACCES; } else { errno = EIO; } return -1; } int result = 0; FILE_END_OF_FILE_INFO end_of_file; end_of_file.EndOfFile.QuadPart = len; if (!SetFileInformationByHandle(hFile, FileEndOfFileInfo, &end_of_file, sizeof(FILE_END_OF_FILE_INFO))) { errno = EIO; result = -1; } CloseHandle(hFile); return result; } void Crash(const std::string& srcfile, int srcline) { fprintf(stdout, "Crashing at %s:%d\n", srcfile.c_str(), srcline); fflush(stdout); abort(); } int GetMaxOpenFiles() { return -1; } } // namespace port } // namespace rocksdb #ifdef JEMALLOC #include "jemalloc/jemalloc.h" namespace rocksdb { namespace port { __declspec(noinline) void WINAPI InitializeJemalloc() { je_init(); atexit(je_uninit); } } // port } // rocksdb extern "C" { #ifdef _WIN64 #pragma comment(linker, "/INCLUDE:p_rocksdb_init_jemalloc") typedef void(WINAPI* CRT_Startup_Routine)(void); // .CRT section is merged with .rdata on x64 so it must be constant data. // must be of external linkage // We put this into XCT since we want to run this earlier than C++ static // constructors // which are placed into XCU #pragma const_seg(".CRT$XCT") extern const CRT_Startup_Routine p_rocksdb_init_jemalloc; const CRT_Startup_Routine p_rocksdb_init_jemalloc = rocksdb::port::InitializeJemalloc; #pragma const_seg() #else // _WIN64 // x86 untested #pragma comment(linker, "/INCLUDE:_p_rocksdb_init_jemalloc") #pragma section(".CRT$XCT", read) JEMALLOC_SECTION(".CRT$XCT") JEMALLOC_ATTR(used) static const void( WINAPI* p_rocksdb_init_jemalloc)(void) = rocksdb::port::InitializeJemalloc; #endif // _WIN64 } // extern "C" // Global operators to be replaced by a linker void* operator new(size_t size) { void* p = je_malloc(size); if (!p) { throw std::bad_alloc(); } return p; } void* operator new[](size_t size) { void* p = je_malloc(size); if (!p) { throw std::bad_alloc(); } return p; } void operator delete(void* p) { je_free(p); } void operator delete[](void* p) { je_free(p); } #endif // JEMALLOC rocksdb-4.5.1/port/win/port_win.h000066400000000000000000000145231267530634000167560ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // See port_example.h for documentation for the following types/functions. #ifndef STORAGE_LEVELDB_PORT_PORT_WIN_H_ #define STORAGE_LEVELDB_PORT_PORT_WIN_H_ // Always want minimum headers #ifndef WIN32_LEAN_AND_MEAN #define WIN32_LEAN_AND_MEAN #endif // Assume that for everywhere #undef PLATFORM_IS_LITTLE_ENDIAN #define PLATFORM_IS_LITTLE_ENDIAN true #include #include #include #include #include #include #include #include "rocksdb/options.h" #undef min #undef max #undef DeleteFile #undef GetCurrentTime #ifndef strcasecmp #define strcasecmp _stricmp #endif #undef GetCurrentTime #undef DeleteFile typedef SSIZE_T ssize_t; // size_t printf formatting named in the manner of C99 standard formatting // strings such as PRIu64 // in fact, we could use that one #ifndef ROCKSDB_PRIszt #define ROCKSDB_PRIszt "Iu" #endif #define __attribute__(A) #ifdef ZLIB #include #endif #ifdef BZIP2 #include #endif #if defined(LZ4) #include #include #endif #ifdef SNAPPY #include #endif // Thread local storage on Linux // There is thread_local in C++11 #ifndef __thread #define __thread __declspec(thread) #endif #ifndef PLATFORM_IS_LITTLE_ENDIAN #define PLATFORM_IS_LITTLE_ENDIAN (__BYTE_ORDER == __LITTLE_ENDIAN) #endif namespace rocksdb { #define PREFETCH(addr, rw, locality) std::string GetWindowsErrSz(DWORD err); namespace port { // VS 15 #if (defined _MSC_VER) && (_MSC_VER >= 1900) #define ROCKSDB_NOEXCEPT noexcept // For use at db/file_indexer.h kLevelMaxIndex const int kMaxInt32 = std::numeric_limits::max(); const uint64_t kMaxUint64 = std::numeric_limits::max(); const size_t kMaxSizet = std::numeric_limits::max(); #else //_MSC_VER // VS 15 has snprintf #define snprintf _snprintf #define ROCKSDB_NOEXCEPT // std::numeric_limits::max() is not constexpr just yet // therefore, use the same limits // For use at db/file_indexer.h kLevelMaxIndex const int kMaxInt32 = INT32_MAX; const uint64_t kMaxUint64 = UINT64_MAX; #ifdef _WIN64 const size_t kMaxSizet = UINT64_MAX; #else const size_t kMaxSizet = UINT_MAX; #endif #endif //_MSC_VER const bool kLittleEndian = true; class CondVar; class Mutex { public: /* implicit */ Mutex(bool adaptive = false) #ifndef NDEBUG : locked_(false) #endif { } ~Mutex(); void Lock() { mutex_.lock(); #ifndef NDEBUG locked_ = true; #endif } void Unlock() { #ifndef NDEBUG locked_ = false; #endif mutex_.unlock(); } // this will assert if the mutex is not locked // it does NOT verify that mutex is held by a calling thread void AssertHeld() { #ifndef NDEBUG assert(locked_); #endif } // Mutex is move only with lock ownership transfer Mutex(const Mutex&) = delete; void operator=(const Mutex&) = delete; private: friend class CondVar; std::mutex& getLock() { return mutex_; } std::mutex mutex_; #ifndef NDEBUG bool locked_; #endif }; class RWMutex { public: RWMutex() { InitializeSRWLock(&srwLock_); } void ReadLock() { AcquireSRWLockShared(&srwLock_); } void WriteLock() { AcquireSRWLockExclusive(&srwLock_); } void ReadUnlock() { ReleaseSRWLockShared(&srwLock_); } void WriteUnlock() { ReleaseSRWLockExclusive(&srwLock_); } // Empty as in POSIX void AssertHeld() {} private: SRWLOCK srwLock_; // No copying allowed RWMutex(const RWMutex&); void operator=(const RWMutex&); }; class CondVar { public: explicit CondVar(Mutex* mu) : mu_(mu) { } ~CondVar(); void Wait(); bool TimedWait(uint64_t expiration_time); void Signal(); void SignalAll(); // Condition var is not copy/move constructible CondVar(const CondVar&) = delete; CondVar& operator=(const CondVar&) = delete; CondVar(CondVar&&) = delete; CondVar& operator=(CondVar&&) = delete; private: std::condition_variable cv_; Mutex* mu_; }; // OnceInit type helps emulate // Posix semantics with initialization // adopted in the project struct OnceType { struct Init {}; OnceType() {} OnceType(const Init&) {} OnceType(const OnceType&) = delete; OnceType& operator=(const OnceType&) = delete; std::once_flag flag_; }; #define LEVELDB_ONCE_INIT port::OnceType::Init() extern void InitOnce(OnceType* once, void (*initializer)()); #define CACHE_LINE_SIZE 64U static inline void AsmVolatilePause() { #if defined(_M_IX86) || defined(_M_X64) YieldProcessor(); #endif // it would be nice to get "wfe" on ARM here } extern int PhysicalCoreID(); // For Thread Local Storage abstraction typedef DWORD pthread_key_t; inline int pthread_key_create(pthread_key_t* key, void (*destructor)(void*)) { // Not used (void)destructor; pthread_key_t k = TlsAlloc(); if (TLS_OUT_OF_INDEXES == k) { return ENOMEM; } *key = k; return 0; } inline int pthread_key_delete(pthread_key_t key) { if (!TlsFree(key)) { return EINVAL; } return 0; } inline int pthread_setspecific(pthread_key_t key, const void* value) { if (!TlsSetValue(key, const_cast(value))) { return ENOMEM; } return 0; } inline void* pthread_getspecific(pthread_key_t key) { void* result = TlsGetValue(key); if (!result) { if (GetLastError() != ERROR_SUCCESS) { errno = EINVAL; } else { errno = NOERROR; } } return result; } // UNIX equiv although errno numbers will be off // using C-runtime to implement. Note, this does not // feel space with zeros in case the file is extended. int truncate(const char* path, int64_t length); void Crash(const std::string& srcfile, int srcline); extern int GetMaxOpenFiles(); } // namespace port using port::pthread_key_t; using port::pthread_key_create; using port::pthread_key_delete; using port::pthread_setspecific; using port::pthread_getspecific; using port::truncate; } // namespace rocksdb #endif // STORAGE_LEVELDB_PORT_PORT_WIN_H_ rocksdb-4.5.1/port/win/win_logger.cc000066400000000000000000000106121267530634000174020ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Logger implementation that can be shared by all environments // where enough posix functionality is available. #include "port/win/win_logger.h" #include #include #include #include #include #include "rocksdb/env.h" #include "port/sys_time.h" #include "util/iostats_context_imp.h" namespace rocksdb { WinLogger::WinLogger(uint64_t (*gettid)(), Env* env, HANDLE file, const InfoLogLevel log_level) : Logger(log_level), gettid_(gettid), log_size_(0), last_flush_micros_(0), env_(env), flush_pending_(false), file_(file) {} void WinLogger::DebugWriter(const char* str, int len) { DWORD bytesWritten = 0; BOOL ret = WriteFile(file_, str, len, &bytesWritten, NULL); if (ret == FALSE) { std::string errSz = GetWindowsErrSz(GetLastError()); fprintf(stderr, errSz.c_str()); } } WinLogger::~WinLogger() { close(); } void WinLogger::close() { CloseHandle(file_); } void WinLogger::Flush() { if (flush_pending_) { flush_pending_ = false; // With Windows API writes go to OS buffers directly so no fflush needed // unlike with C runtime API. We don't flush all the way to disk // for perf reasons. } last_flush_micros_ = env_->NowMicros(); } void WinLogger::Logv(const char* format, va_list ap) { IOSTATS_TIMER_GUARD(logger_nanos); const uint64_t thread_id = (*gettid_)(); // We try twice: the first time with a fixed-size stack allocated buffer, // and the second time with a much larger dynamically allocated buffer. char buffer[500]; std::unique_ptr largeBuffer; for (int iter = 0; iter < 2; ++iter) { char* base; int bufsize; if (iter == 0) { bufsize = sizeof(buffer); base = buffer; } else { bufsize = 30000; largeBuffer.reset(new char[bufsize]); base = largeBuffer.get(); } char* p = base; char* limit = base + bufsize; struct timeval now_tv; gettimeofday(&now_tv, nullptr); const time_t seconds = now_tv.tv_sec; struct tm t; localtime_s(&t, &seconds); p += snprintf(p, limit - p, "%04d/%02d/%02d-%02d:%02d:%02d.%06d %llx ", t.tm_year + 1900, t.tm_mon + 1, t.tm_mday, t.tm_hour, t.tm_min, t.tm_sec, static_cast(now_tv.tv_usec), static_cast(thread_id)); // Print the message if (p < limit) { va_list backup_ap; va_copy(backup_ap, ap); int done = vsnprintf(p, limit - p, format, backup_ap); if (done > 0) { p += done; } else { continue; } va_end(backup_ap); } // Truncate to available space if necessary if (p >= limit) { if (iter == 0) { continue; // Try again with larger buffer } else { p = limit - 1; } } // Add newline if necessary if (p == base || p[-1] != '\n') { *p++ = '\n'; } assert(p <= limit); const size_t write_size = p - base; DWORD bytesWritten = 0; BOOL ret = WriteFile(file_, base, static_cast(write_size), &bytesWritten, NULL); if (ret == FALSE) { std::string errSz = GetWindowsErrSz(GetLastError()); fprintf(stderr, errSz.c_str()); } flush_pending_ = true; assert((bytesWritten == write_size) || (ret == FALSE)); if (bytesWritten > 0) { log_size_ += write_size; } uint64_t now_micros = static_cast(now_tv.tv_sec) * 1000000 + now_tv.tv_usec; if (now_micros - last_flush_micros_ >= flush_every_seconds_ * 1000000) { flush_pending_ = false; // With Windows API writes go to OS buffers directly so no fflush needed // unlike with C runtime API. We don't flush all the way to disk // for perf reasons. last_flush_micros_ = now_micros; } break; } } size_t WinLogger::GetLogFileSize() const { return log_size_; } } // namespace rocksdb rocksdb-4.5.1/port/win/win_logger.h000066400000000000000000000030641267530634000172470ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Logger implementation that can be shared by all environments // where enough posix functionality is available. #pragma once #include #include "rocksdb/env.h" #include #include namespace rocksdb { class Env; const int kDebugLogChunkSize = 128 * 1024; class WinLogger : public rocksdb::Logger { public: WinLogger(uint64_t (*gettid)(), Env* env, HANDLE file, const InfoLogLevel log_level = InfoLogLevel::ERROR_LEVEL); virtual ~WinLogger(); WinLogger(const WinLogger&) = delete; WinLogger& operator=(const WinLogger&) = delete; void close(); void Flush() override; void Logv(const char* format, va_list ap) override; size_t GetLogFileSize() const override; void DebugWriter(const char* str, int len); private: HANDLE file_; uint64_t (*gettid_)(); // Return the thread id for the current thread std::atomic_size_t log_size_; std::atomic_uint_fast64_t last_flush_micros_; Env* env_; bool flush_pending_; const static uint64_t flush_every_seconds_ = 5; }; } // namespace rocksdb rocksdb-4.5.1/src.mk000066400000000000000000000516541267530634000143110ustar00rootroot00000000000000# These are the sources from which librocksdb.a is built: LIB_SOURCES = \ db/auto_roll_logger.cc \ db/builder.cc \ db/c.cc \ db/column_family.cc \ db/compacted_db_impl.cc \ db/compaction.cc \ db/compaction_iterator.cc \ db/compaction_job.cc \ db/compaction_picker.cc \ db/convenience.cc \ db/db_filesnapshot.cc \ db/dbformat.cc \ db/db_impl.cc \ db/db_impl_debug.cc \ db/db_impl_readonly.cc \ db/db_impl_experimental.cc \ db/db_info_dumper.cc \ db/db_iter.cc \ db/experimental.cc \ db/event_helpers.cc \ db/file_indexer.cc \ db/filename.cc \ db/flush_job.cc \ db/flush_scheduler.cc \ db/forward_iterator.cc \ db/internal_stats.cc \ db/log_reader.cc \ db/log_writer.cc \ db/managed_iterator.cc \ db/memtable_allocator.cc \ db/memtable.cc \ db/memtable_list.cc \ db/merge_helper.cc \ db/merge_operator.cc \ db/repair.cc \ db/slice.cc \ db/snapshot_impl.cc \ db/table_cache.cc \ db/table_properties_collector.cc \ db/transaction_log_impl.cc \ db/version_builder.cc \ db/version_edit.cc \ db/version_set.cc \ db/wal_manager.cc \ db/write_batch.cc \ db/write_batch_base.cc \ db/write_controller.cc \ db/write_thread.cc \ db/xfunc_test_points.cc \ memtable/hash_cuckoo_rep.cc \ memtable/hash_linklist_rep.cc \ memtable/hash_skiplist_rep.cc \ memtable/skiplistrep.cc \ memtable/vectorrep.cc \ port/stack_trace.cc \ port/port_posix.cc \ table/adaptive_table_factory.cc \ table/block_based_filter_block.cc \ table/block_based_table_builder.cc \ table/block_based_table_factory.cc \ table/block_based_table_reader.cc \ table/block_builder.cc \ table/block.cc \ table/block_hash_index.cc \ table/block_prefix_index.cc \ table/bloom_block.cc \ table/cuckoo_table_builder.cc \ table/cuckoo_table_factory.cc \ table/cuckoo_table_reader.cc \ table/flush_block_policy.cc \ table/format.cc \ table/full_filter_block.cc \ table/get_context.cc \ table/iterator.cc \ table/merger.cc \ table/meta_blocks.cc \ table/sst_file_writer.cc \ table/plain_table_builder.cc \ table/plain_table_factory.cc \ table/plain_table_index.cc \ table/plain_table_key_coding.cc \ table/plain_table_reader.cc \ table/table_properties.cc \ table/two_level_iterator.cc \ tools/dump/db_dump_tool.cc \ util/arena.cc \ util/bloom.cc \ util/build_version.cc \ util/cache.cc \ util/coding.cc \ util/comparator.cc \ util/compaction_job_stats_impl.cc \ util/concurrent_arena.cc \ util/crc32c.cc \ util/delete_scheduler.cc \ util/dynamic_bloom.cc \ util/env.cc \ util/env_hdfs.cc \ util/env_posix.cc \ util/io_posix.cc \ util/thread_posix.cc \ util/sst_file_manager_impl.cc \ util/file_util.cc \ util/file_reader_writer.cc \ util/filter_policy.cc \ util/hash.cc \ util/histogram.cc \ util/instrumented_mutex.cc \ util/iostats_context.cc \ utilities/backupable/backupable_db.cc \ utilities/convenience/info_log_finder.cc \ utilities/checkpoint/checkpoint.cc \ utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc \ utilities/document/document_db.cc \ utilities/document/json_document_builder.cc \ utilities/document/json_document.cc \ utilities/env_mirror.cc \ utilities/flashcache/flashcache.cc \ utilities/geodb/geodb_impl.cc \ utilities/leveldb_options/leveldb_options.cc \ utilities/memory/memory_util.cc \ utilities/merge_operators/put.cc \ utilities/merge_operators/string_append/stringappend2.cc \ utilities/merge_operators/string_append/stringappend.cc \ utilities/merge_operators/uint64add.cc \ utilities/options/options_util.cc \ utilities/redis/redis_lists.cc \ utilities/spatialdb/spatial_db.cc \ utilities/table_properties_collectors/compact_on_deletion_collector.cc \ utilities/transactions/optimistic_transaction_impl.cc \ utilities/transactions/optimistic_transaction_db_impl.cc \ utilities/transactions/transaction_base.cc \ utilities/transactions/transaction_db_impl.cc \ utilities/transactions/transaction_db_mutex_impl.cc \ utilities/transactions/transaction_lock_mgr.cc \ utilities/transactions/transaction_impl.cc \ utilities/transactions/transaction_util.cc \ utilities/ttl/db_ttl_impl.cc \ utilities/write_batch_with_index/write_batch_with_index.cc \ utilities/write_batch_with_index/write_batch_with_index_internal.cc \ util/event_logger.cc \ util/log_buffer.cc \ util/logging.cc \ util/memenv.cc \ util/murmurhash.cc \ util/mutable_cf_options.cc \ util/options.cc \ util/options_builder.cc \ util/options_helper.cc \ util/options_parser.cc \ util/options_sanity_check.cc \ util/perf_context.cc \ util/perf_level.cc \ util/random.cc \ util/rate_limiter.cc \ util/slice.cc \ util/statistics.cc \ util/status.cc \ util/status_message.cc \ util/string_util.cc \ util/sync_point.cc \ util/thread_local.cc \ util/thread_status_impl.cc \ util/thread_status_updater.cc \ util/thread_status_updater_debug.cc \ util/thread_status_util.cc \ util/thread_status_util_debug.cc \ util/xfunc.cc \ util/xxhash.cc \ TOOL_SOURCES = \ tools/ldb_cmd.cc \ tools/ldb_tool.cc \ tools/sst_dump_tool.cc \ MOCK_SOURCES = \ table/mock_table.cc \ util/mock_env.cc TEST_BENCH_SOURCES = \ third-party/gtest-1.7.0/fused-src/gtest/gtest-all.cc \ db/auto_roll_logger_test.cc \ db/column_family_test.cc \ db/compaction_job_test.cc \ db/compaction_job_stats_test.cc \ db/compaction_picker_test.cc \ db/comparator_db_test.cc \ db/corruption_test.cc \ db/cuckoo_table_db_test.cc \ db/db_bench.cc \ db/dbformat_test.cc \ db/db_iter_test.cc \ db/db_test.cc \ db/db_compaction_filter_test.cc \ db/db_compaction_test.cc \ db/db_dynamic_level_test.cc \ db/db_inplace_update_test.cc \ db/db_log_iter_test.cc \ db/db_universal_compaction_test.cc \ db/db_tailing_iter_test.cc \ db/db_wal_test.cc \ db/db_table_properties_test.cc \ db/deletefile_test.cc \ db/fault_injection_test.cc \ db/file_indexer_test.cc \ db/filename_test.cc \ db/flush_job_test.cc \ db/inlineskiplist_test.cc \ db/listener_test.cc \ db/log_test.cc \ db/manual_compaction_test.cc \ db/memtablerep_bench.cc \ db/merge_test.cc \ db/options_file_test.cc \ db/perf_context_test.cc \ db/plain_table_db_test.cc \ db/prefix_test.cc \ db/skiplist_test.cc \ db/table_properties_collector_test.cc \ db/db_test_util.cc \ db/version_builder_test.cc \ db/version_edit_test.cc \ db/version_set_test.cc \ db/wal_manager_test.cc \ db/write_batch_test.cc \ db/write_controller_test.cc \ db/write_callback_test.cc \ table/block_based_filter_block_test.cc \ table/block_hash_index_test.cc \ table/block_test.cc \ table/cuckoo_table_builder_test.cc \ table/cuckoo_table_reader_test.cc \ table/full_filter_block_test.cc \ table/merger_test.cc \ table/table_reader_bench.cc \ table/table_test.cc \ tools/db_sanity_test.cc \ tools/ldb_cmd_test.cc \ tools/reduce_levels_test.cc \ tools/sst_dump_test.cc \ util/arena_test.cc \ util/autovector_test.cc \ util/bloom_test.cc \ util/cache_bench.cc \ util/cache_test.cc \ util/coding_test.cc \ util/crc32c_test.cc \ util/dynamic_bloom_test.cc \ util/env_test.cc \ util/filelock_test.cc \ util/histogram_test.cc \ utilities/backupable/backupable_db_test.cc \ utilities/checkpoint/checkpoint_test.cc \ utilities/document/document_db_test.cc \ utilities/document/json_document_test.cc \ utilities/geodb/geodb_test.cc \ utilities/memory/memory_test.cc \ utilities/merge_operators/string_append/stringappend_test.cc \ utilities/options/options_util_test.cc \ utilities/redis/redis_lists_test.cc \ utilities/spatialdb/spatial_db_test.cc \ utilities/table_properties_collectors/compact_on_deletion_collector_test.cc \ utilities/transactions/optimistic_transaction_test.cc \ utilities/transactions/transaction_test.cc \ utilities/ttl/ttl_test.cc \ utilities/write_batch_with_index/write_batch_with_index_test.cc \ util/log_write_bench.cc \ util/memenv_test.cc \ util/mock_env_test.cc \ util/options_test.cc \ util/event_logger_test.cc \ util/rate_limiter_test.cc \ util/slice_transform_test.cc \ util/testharness.cc \ util/testutil.cc \ util/thread_list_test.cc \ util/thread_local_test.cc JNI_NATIVE_SOURCES = \ java/rocksjni/backupenginejni.cc \ java/rocksjni/backupablejni.cc \ java/rocksjni/checkpoint.cc \ java/rocksjni/columnfamilyhandle.cc \ java/rocksjni/compaction_filter.cc \ java/rocksjni/comparator.cc \ java/rocksjni/comparatorjnicallback.cc \ java/rocksjni/env.cc \ java/rocksjni/filter.cc \ java/rocksjni/iterator.cc \ java/rocksjni/loggerjnicallback.cc \ java/rocksjni/memtablejni.cc \ java/rocksjni/merge_operator.cc \ java/rocksjni/options.cc \ java/rocksjni/ratelimiterjni.cc \ java/rocksjni/remove_emptyvalue_compactionfilterjni.cc \ java/rocksjni/restorejni.cc \ java/rocksjni/rocksjni.cc \ java/rocksjni/slice.cc \ java/rocksjni/snapshot.cc \ java/rocksjni/statistics.cc \ java/rocksjni/table.cc \ java/rocksjni/transaction_log.cc \ java/rocksjni/ttl.cc \ java/rocksjni/write_batch.cc \ java/rocksjni/writebatchhandlerjnicallback.cc \ java/rocksjni/write_batch_test.cc \ java/rocksjni/write_batch_with_index.cc # Currently, we do not generate dependencies for # java/rocksjni/write_batch_test.cc, because its dependent, # java/include/org_rocksdb_WriteBatch.h is generated. # TODO/FIXME: fix the above. Otherwise, the current rules would fail: # java/rocksjni/write_batch_test.cc:13:44: fatal error: include/org_rocksdb_WriteBatch.h: No such file or directory # #include "include/org_rocksdb_WriteBatch.h" # These are the xfunc tests run : XFUNC_TESTS = \ "managed_new" \ "managed_xftest_dropold" \ "managed_xftest_release" \ "inplace_lock_test" \ "transaction" rocksdb-4.5.1/table/000077500000000000000000000000001267530634000142455ustar00rootroot00000000000000rocksdb-4.5.1/table/adaptive_table_factory.cc000066400000000000000000000107351267530634000212550ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "table/adaptive_table_factory.h" #include "table/table_builder.h" #include "table/format.h" #include "port/port.h" namespace rocksdb { AdaptiveTableFactory::AdaptiveTableFactory( std::shared_ptr table_factory_to_write, std::shared_ptr block_based_table_factory, std::shared_ptr plain_table_factory, std::shared_ptr cuckoo_table_factory) : table_factory_to_write_(table_factory_to_write), block_based_table_factory_(block_based_table_factory), plain_table_factory_(plain_table_factory), cuckoo_table_factory_(cuckoo_table_factory) { if (!table_factory_to_write_) { table_factory_to_write_ = block_based_table_factory_; } if (!plain_table_factory_) { plain_table_factory_.reset(NewPlainTableFactory()); } if (!block_based_table_factory_) { block_based_table_factory_.reset(NewBlockBasedTableFactory()); } if (!cuckoo_table_factory_) { cuckoo_table_factory_.reset(NewCuckooTableFactory()); } } extern const uint64_t kPlainTableMagicNumber; extern const uint64_t kLegacyPlainTableMagicNumber; extern const uint64_t kBlockBasedTableMagicNumber; extern const uint64_t kLegacyBlockBasedTableMagicNumber; extern const uint64_t kCuckooTableMagicNumber; Status AdaptiveTableFactory::NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table) const { Footer footer; auto s = ReadFooterFromFile(file.get(), file_size, &footer); if (!s.ok()) { return s; } if (footer.table_magic_number() == kPlainTableMagicNumber || footer.table_magic_number() == kLegacyPlainTableMagicNumber) { return plain_table_factory_->NewTableReader( table_reader_options, std::move(file), file_size, table); } else if (footer.table_magic_number() == kBlockBasedTableMagicNumber || footer.table_magic_number() == kLegacyBlockBasedTableMagicNumber) { return block_based_table_factory_->NewTableReader( table_reader_options, std::move(file), file_size, table); } else if (footer.table_magic_number() == kCuckooTableMagicNumber) { return cuckoo_table_factory_->NewTableReader( table_reader_options, std::move(file), file_size, table); } else { return Status::NotSupported("Unidentified table format"); } } TableBuilder* AdaptiveTableFactory::NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const { return table_factory_to_write_->NewTableBuilder(table_builder_options, column_family_id, file); } std::string AdaptiveTableFactory::GetPrintableTableOptions() const { std::string ret; ret.reserve(20000); const int kBufferSize = 200; char buffer[kBufferSize]; if (!table_factory_to_write_) { snprintf(buffer, kBufferSize, " write factory (%s) options:\n%s\n", table_factory_to_write_->Name(), table_factory_to_write_->GetPrintableTableOptions().c_str()); ret.append(buffer); } if (!plain_table_factory_) { snprintf(buffer, kBufferSize, " %s options:\n%s\n", plain_table_factory_->Name(), plain_table_factory_->GetPrintableTableOptions().c_str()); ret.append(buffer); } if (!block_based_table_factory_) { snprintf(buffer, kBufferSize, " %s options:\n%s\n", block_based_table_factory_->Name(), block_based_table_factory_->GetPrintableTableOptions().c_str()); ret.append(buffer); } if (!cuckoo_table_factory_) { snprintf(buffer, kBufferSize, " %s options:\n%s\n", cuckoo_table_factory_->Name(), cuckoo_table_factory_->GetPrintableTableOptions().c_str()); ret.append(buffer); } return ret; } extern TableFactory* NewAdaptiveTableFactory( std::shared_ptr table_factory_to_write, std::shared_ptr block_based_table_factory, std::shared_ptr plain_table_factory, std::shared_ptr cuckoo_table_factory) { return new AdaptiveTableFactory(table_factory_to_write, block_based_table_factory, plain_table_factory, cuckoo_table_factory); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/adaptive_table_factory.h000066400000000000000000000035361267530634000211200ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/options.h" #include "rocksdb/table.h" namespace rocksdb { struct EnvOptions; using std::unique_ptr; class Status; class RandomAccessFile; class WritableFile; class Table; class TableBuilder; class AdaptiveTableFactory : public TableFactory { public: ~AdaptiveTableFactory() {} explicit AdaptiveTableFactory( std::shared_ptr table_factory_to_write, std::shared_ptr block_based_table_factory, std::shared_ptr plain_table_factory, std::shared_ptr cuckoo_table_factory); const char* Name() const override { return "AdaptiveTableFactory"; } Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table) const override; TableBuilder* NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const override; // Sanitizes the specified DB Options. Status SanitizeOptions(const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const override { return Status::OK(); } std::string GetPrintableTableOptions() const override; private: std::shared_ptr table_factory_to_write_; std::shared_ptr block_based_table_factory_; std::shared_ptr plain_table_factory_; std::shared_ptr cuckoo_table_factory_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/block.cc000066400000000000000000000270441267530634000156550ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // Decodes the blocks generated by block_builder.cc. #include "table/block.h" #include #include #include #include #include "rocksdb/comparator.h" #include "table/format.h" #include "table/block_hash_index.h" #include "table/block_prefix_index.h" #include "util/coding.h" #include "util/logging.h" #include "util/perf_context_imp.h" namespace rocksdb { // Helper routine: decode the next block entry starting at "p", // storing the number of shared key bytes, non_shared key bytes, // and the length of the value in "*shared", "*non_shared", and // "*value_length", respectively. Will not derefence past "limit". // // If any errors are detected, returns nullptr. Otherwise, returns a // pointer to the key delta (just past the three decoded values). static inline const char* DecodeEntry(const char* p, const char* limit, uint32_t* shared, uint32_t* non_shared, uint32_t* value_length) { if (limit - p < 3) return nullptr; *shared = reinterpret_cast(p)[0]; *non_shared = reinterpret_cast(p)[1]; *value_length = reinterpret_cast(p)[2]; if ((*shared | *non_shared | *value_length) < 128) { // Fast path: all three values are encoded in one byte each p += 3; } else { if ((p = GetVarint32Ptr(p, limit, shared)) == nullptr) return nullptr; if ((p = GetVarint32Ptr(p, limit, non_shared)) == nullptr) return nullptr; if ((p = GetVarint32Ptr(p, limit, value_length)) == nullptr) return nullptr; } if (static_cast(limit - p) < (*non_shared + *value_length)) { return nullptr; } return p; } void BlockIter::Next() { assert(Valid()); ParseNextKey(); } void BlockIter::Prev() { assert(Valid()); // Scan backwards to a restart point before current_ const uint32_t original = current_; while (GetRestartPoint(restart_index_) >= original) { if (restart_index_ == 0) { // No more entries current_ = restarts_; restart_index_ = num_restarts_; return; } restart_index_--; } SeekToRestartPoint(restart_index_); do { // Loop until end of current entry hits the start of original entry } while (ParseNextKey() && NextEntryOffset() < original); } void BlockIter::Seek(const Slice& target) { PERF_TIMER_GUARD(block_seek_nanos); if (data_ == nullptr) { // Not init yet return; } uint32_t index = 0; bool ok = false; if (prefix_index_) { ok = PrefixSeek(target, &index); } else { ok = hash_index_ ? HashSeek(target, &index) : BinarySeek(target, 0, num_restarts_ - 1, &index); } if (!ok) { return; } SeekToRestartPoint(index); // Linear search (within restart block) for first key >= target while (true) { if (!ParseNextKey() || Compare(key_.GetKey(), target) >= 0) { return; } } } void BlockIter::SeekToFirst() { if (data_ == nullptr) { // Not init yet return; } SeekToRestartPoint(0); ParseNextKey(); } void BlockIter::SeekToLast() { if (data_ == nullptr) { // Not init yet return; } SeekToRestartPoint(num_restarts_ - 1); while (ParseNextKey() && NextEntryOffset() < restarts_) { // Keep skipping } } void BlockIter::CorruptionError() { current_ = restarts_; restart_index_ = num_restarts_; status_ = Status::Corruption("bad entry in block"); key_.Clear(); value_.clear(); } bool BlockIter::ParseNextKey() { current_ = NextEntryOffset(); const char* p = data_ + current_; const char* limit = data_ + restarts_; // Restarts come right after data if (p >= limit) { // No more entries to return. Mark as invalid. current_ = restarts_; restart_index_ = num_restarts_; return false; } // Decode next entry uint32_t shared, non_shared, value_length; p = DecodeEntry(p, limit, &shared, &non_shared, &value_length); if (p == nullptr || key_.Size() < shared) { CorruptionError(); return false; } else { if (shared == 0) { // If this key dont share any bytes with prev key then we dont need // to decode it and can use it's address in the block directly. key_.SetKey(Slice(p, non_shared), false /* copy */); } else { // This key share `shared` bytes with prev key, we need to decode it key_.TrimAppend(shared, p, non_shared); } value_ = Slice(p + non_shared, value_length); while (restart_index_ + 1 < num_restarts_ && GetRestartPoint(restart_index_ + 1) < current_) { ++restart_index_; } return true; } } // Binary search in restart array to find the first restart point // with a key >= target (TODO: this comment is inaccurate) bool BlockIter::BinarySeek(const Slice& target, uint32_t left, uint32_t right, uint32_t* index) { assert(left <= right); while (left < right) { uint32_t mid = (left + right + 1) / 2; uint32_t region_offset = GetRestartPoint(mid); uint32_t shared, non_shared, value_length; const char* key_ptr = DecodeEntry(data_ + region_offset, data_ + restarts_, &shared, &non_shared, &value_length); if (key_ptr == nullptr || (shared != 0)) { CorruptionError(); return false; } Slice mid_key(key_ptr, non_shared); int cmp = Compare(mid_key, target); if (cmp < 0) { // Key at "mid" is smaller than "target". Therefore all // blocks before "mid" are uninteresting. left = mid; } else if (cmp > 0) { // Key at "mid" is >= "target". Therefore all blocks at or // after "mid" are uninteresting. right = mid - 1; } else { left = right = mid; } } *index = left; return true; } // Compare target key and the block key of the block of `block_index`. // Return -1 if error. int BlockIter::CompareBlockKey(uint32_t block_index, const Slice& target) { uint32_t region_offset = GetRestartPoint(block_index); uint32_t shared, non_shared, value_length; const char* key_ptr = DecodeEntry(data_ + region_offset, data_ + restarts_, &shared, &non_shared, &value_length); if (key_ptr == nullptr || (shared != 0)) { CorruptionError(); return 1; // Return target is smaller } Slice block_key(key_ptr, non_shared); return Compare(block_key, target); } // Binary search in block_ids to find the first block // with a key >= target bool BlockIter::BinaryBlockIndexSeek(const Slice& target, uint32_t* block_ids, uint32_t left, uint32_t right, uint32_t* index) { assert(left <= right); uint32_t left_bound = left; while (left <= right) { uint32_t mid = (left + right) / 2; int cmp = CompareBlockKey(block_ids[mid], target); if (!status_.ok()) { return false; } if (cmp < 0) { // Key at "target" is larger than "mid". Therefore all // blocks before or at "mid" are uninteresting. left = mid + 1; } else { // Key at "target" is <= "mid". Therefore all blocks // after "mid" are uninteresting. // If there is only one block left, we found it. if (left == right) break; right = mid; } } if (left == right) { // In one of the two following cases: // (1) left is the first one of block_ids // (2) there is a gap of blocks between block of `left` and `left-1`. // we can further distinguish the case of key in the block or key not // existing, by comparing the target key and the key of the previous // block to the left of the block found. if (block_ids[left] > 0 && (left == left_bound || block_ids[left - 1] != block_ids[left] - 1) && CompareBlockKey(block_ids[left] - 1, target) > 0) { current_ = restarts_; return false; } *index = block_ids[left]; return true; } else { assert(left > right); // Mark iterator invalid current_ = restarts_; return false; } } bool BlockIter::HashSeek(const Slice& target, uint32_t* index) { assert(hash_index_); auto restart_index = hash_index_->GetRestartIndex(target); if (restart_index == nullptr) { current_ = restarts_; return false; } // the elements in restart_array[index : index + num_blocks] // are all with same prefix. We'll do binary search in that small range. auto left = restart_index->first_index; auto right = restart_index->first_index + restart_index->num_blocks - 1; return BinarySeek(target, left, right, index); } bool BlockIter::PrefixSeek(const Slice& target, uint32_t* index) { assert(prefix_index_); uint32_t* block_ids = nullptr; uint32_t num_blocks = prefix_index_->GetBlocks(target, &block_ids); if (num_blocks == 0) { current_ = restarts_; return false; } else { return BinaryBlockIndexSeek(target, block_ids, 0, num_blocks - 1, index); } } uint32_t Block::NumRestarts() const { assert(size_ >= 2*sizeof(uint32_t)); return DecodeFixed32(data_ + size_ - sizeof(uint32_t)); } Block::Block(BlockContents&& contents) : contents_(std::move(contents)), data_(contents_.data.data()), size_(contents_.data.size()) { if (size_ < sizeof(uint32_t)) { size_ = 0; // Error marker } else { restart_offset_ = static_cast(size_) - (1 + NumRestarts()) * sizeof(uint32_t); if (restart_offset_ > size_ - sizeof(uint32_t)) { // The size is too small for NumRestarts() and therefore // restart_offset_ wrapped around. size_ = 0; } } } InternalIterator* Block::NewIterator(const Comparator* cmp, BlockIter* iter, bool total_order_seek) { if (size_ < 2*sizeof(uint32_t)) { if (iter != nullptr) { iter->SetStatus(Status::Corruption("bad block contents")); return iter; } else { return NewErrorInternalIterator(Status::Corruption("bad block contents")); } } const uint32_t num_restarts = NumRestarts(); if (num_restarts == 0) { if (iter != nullptr) { iter->SetStatus(Status::OK()); return iter; } else { return NewEmptyInternalIterator(); } } else { BlockHashIndex* hash_index_ptr = total_order_seek ? nullptr : hash_index_.get(); BlockPrefixIndex* prefix_index_ptr = total_order_seek ? nullptr : prefix_index_.get(); if (iter != nullptr) { iter->Initialize(cmp, data_, restart_offset_, num_restarts, hash_index_ptr, prefix_index_ptr); } else { iter = new BlockIter(cmp, data_, restart_offset_, num_restarts, hash_index_ptr, prefix_index_ptr); } } return iter; } void Block::SetBlockHashIndex(BlockHashIndex* hash_index) { hash_index_.reset(hash_index); } void Block::SetBlockPrefixIndex(BlockPrefixIndex* prefix_index) { prefix_index_.reset(prefix_index); } size_t Block::ApproximateMemoryUsage() const { size_t usage = usable_size(); if (hash_index_) { usage += hash_index_->ApproximateMemoryUsage(); } if (prefix_index_) { usage += prefix_index_->ApproximateMemoryUsage(); } return usage; } } // namespace rocksdb rocksdb-4.5.1/table/block.h000066400000000000000000000155341267530634000155200ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #ifdef ROCKSDB_MALLOC_USABLE_SIZE #include #endif #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "db/dbformat.h" #include "table/block_prefix_index.h" #include "table/block_hash_index.h" #include "table/internal_iterator.h" #include "format.h" namespace rocksdb { struct BlockContents; class Comparator; class BlockIter; class BlockHashIndex; class BlockPrefixIndex; class Block { public: // Initialize the block with the specified contents. explicit Block(BlockContents&& contents); ~Block() = default; size_t size() const { return size_; } const char* data() const { return data_; } bool cachable() const { return contents_.cachable; } size_t usable_size() const { #ifdef ROCKSDB_MALLOC_USABLE_SIZE if (contents_.allocation.get() != nullptr) { return malloc_usable_size(contents_.allocation.get()); } #endif // ROCKSDB_MALLOC_USABLE_SIZE return size_; } uint32_t NumRestarts() const; CompressionType compression_type() const { return contents_.compression_type; } // If hash index lookup is enabled and `use_hash_index` is true. This block // will do hash lookup for the key prefix. // // NOTE: for the hash based lookup, if a key prefix doesn't match any key, // the iterator will simply be set as "invalid", rather than returning // the key that is just pass the target key. // // If iter is null, return new Iterator // If iter is not null, update this one and return it as Iterator* // // If total_order_seek is true, hash_index_ and prefix_index_ are ignored. // This option only applies for index block. For data block, hash_index_ // and prefix_index_ are null, so this option does not matter. InternalIterator* NewIterator(const Comparator* comparator, BlockIter* iter = nullptr, bool total_order_seek = true); void SetBlockHashIndex(BlockHashIndex* hash_index); void SetBlockPrefixIndex(BlockPrefixIndex* prefix_index); // Report an approximation of how much memory has been used. size_t ApproximateMemoryUsage() const; private: BlockContents contents_; const char* data_; // contents_.data.data() size_t size_; // contents_.data.size() uint32_t restart_offset_; // Offset in data_ of restart array std::unique_ptr hash_index_; std::unique_ptr prefix_index_; // No copying allowed Block(const Block&); void operator=(const Block&); }; class BlockIter : public InternalIterator { public: BlockIter() : comparator_(nullptr), data_(nullptr), restarts_(0), num_restarts_(0), current_(0), restart_index_(0), status_(Status::OK()), hash_index_(nullptr), prefix_index_(nullptr) {} BlockIter(const Comparator* comparator, const char* data, uint32_t restarts, uint32_t num_restarts, BlockHashIndex* hash_index, BlockPrefixIndex* prefix_index) : BlockIter() { Initialize(comparator, data, restarts, num_restarts, hash_index, prefix_index); } void Initialize(const Comparator* comparator, const char* data, uint32_t restarts, uint32_t num_restarts, BlockHashIndex* hash_index, BlockPrefixIndex* prefix_index) { assert(data_ == nullptr); // Ensure it is called only once assert(num_restarts > 0); // Ensure the param is valid comparator_ = comparator; data_ = data; restarts_ = restarts; num_restarts_ = num_restarts; current_ = restarts_; restart_index_ = num_restarts_; hash_index_ = hash_index; prefix_index_ = prefix_index; } void SetStatus(Status s) { status_ = s; } virtual bool Valid() const override { return current_ < restarts_; } virtual Status status() const override { return status_; } virtual Slice key() const override { assert(Valid()); return key_.GetKey(); } virtual Slice value() const override { assert(Valid()); return value_; } virtual void Next() override; virtual void Prev() override; virtual void Seek(const Slice& target) override; virtual void SeekToFirst() override; virtual void SeekToLast() override; virtual Status PinData() override { // block data is always pinned. return Status::OK(); } virtual Status ReleasePinnedData() override { // block data is always pinned. return Status::OK(); } virtual bool IsKeyPinned() const override { return key_.IsKeyPinned(); } private: const Comparator* comparator_; const char* data_; // underlying block contents uint32_t restarts_; // Offset of restart array (list of fixed32) uint32_t num_restarts_; // Number of uint32_t entries in restart array // current_ is offset in data_ of current entry. >= restarts_ if !Valid uint32_t current_; uint32_t restart_index_; // Index of restart block in which current_ falls IterKey key_; Slice value_; Status status_; BlockHashIndex* hash_index_; BlockPrefixIndex* prefix_index_; inline int Compare(const Slice& a, const Slice& b) const { return comparator_->Compare(a, b); } // Return the offset in data_ just past the end of the current entry. inline uint32_t NextEntryOffset() const { // NOTE: We don't support files bigger than 2GB return static_cast((value_.data() + value_.size()) - data_); } uint32_t GetRestartPoint(uint32_t index) { assert(index < num_restarts_); return DecodeFixed32(data_ + restarts_ + index * sizeof(uint32_t)); } void SeekToRestartPoint(uint32_t index) { key_.Clear(); restart_index_ = index; // current_ will be fixed by ParseNextKey(); // ParseNextKey() starts at the end of value_, so set value_ accordingly uint32_t offset = GetRestartPoint(index); value_ = Slice(data_ + offset, 0); } void CorruptionError(); bool ParseNextKey(); bool BinarySeek(const Slice& target, uint32_t left, uint32_t right, uint32_t* index); int CompareBlockKey(uint32_t block_index, const Slice& target); bool BinaryBlockIndexSeek(const Slice& target, uint32_t* block_ids, uint32_t left, uint32_t right, uint32_t* index); bool HashSeek(const Slice& target, uint32_t* index); bool PrefixSeek(const Slice& target, uint32_t* index); }; } // namespace rocksdb rocksdb-4.5.1/table/block_based_filter_block.cc000066400000000000000000000210031267530634000215170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2012 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/block_based_filter_block.h" #include #include "db/dbformat.h" #include "rocksdb/filter_policy.h" #include "util/coding.h" #include "util/perf_context_imp.h" #include "util/string_util.h" namespace rocksdb { namespace { bool SamePrefix(const SliceTransform* prefix_extractor, const Slice& key1, const Slice& key2) { if (!prefix_extractor->InDomain(key1) && !prefix_extractor->InDomain(key2)) { return true; } else if (!prefix_extractor->InDomain(key1) || !prefix_extractor->InDomain(key2)) { return false; } else { return (prefix_extractor->Transform(key1) == prefix_extractor->Transform(key2)); } } void AppendItem(std::string* props, const std::string& key, const std::string& value) { char cspace = ' '; std::string value_str(""); size_t i = 0; const size_t dataLength = 64; const size_t tabLength = 2; const size_t offLength = 16; value_str.append(&value[i], std::min(size_t(dataLength), value.size())); i += dataLength; while (i < value.size()) { value_str.append("\n"); value_str.append(offLength, cspace); value_str.append(&value[i], std::min(size_t(dataLength), value.size() - i)); i += dataLength; } std::string result(""); if (key.size() < (offLength - tabLength)) result.append(size_t((offLength - tabLength)) - key.size(), cspace); result.append(key); props->append(result + ": " + value_str + "\n"); } template void AppendItem(std::string* props, const TKey& key, const std::string& value) { std::string key_str = rocksdb::ToString(key); AppendItem(props, key_str, value); } } // namespace // See doc/table_format.txt for an explanation of the filter block format. // Generate new filter every 2KB of data static const size_t kFilterBaseLg = 11; static const size_t kFilterBase = 1 << kFilterBaseLg; BlockBasedFilterBlockBuilder::BlockBasedFilterBlockBuilder( const SliceTransform* prefix_extractor, const BlockBasedTableOptions& table_opt) : policy_(table_opt.filter_policy.get()), prefix_extractor_(prefix_extractor), whole_key_filtering_(table_opt.whole_key_filtering) { assert(policy_); } void BlockBasedFilterBlockBuilder::StartBlock(uint64_t block_offset) { uint64_t filter_index = (block_offset / kFilterBase); assert(filter_index >= filter_offsets_.size()); while (filter_index > filter_offsets_.size()) { GenerateFilter(); } } void BlockBasedFilterBlockBuilder::Add(const Slice& key) { added_to_start_ = 0; if (whole_key_filtering_) { AddKey(key); added_to_start_ = 1; } if (prefix_extractor_ && prefix_extractor_->InDomain(key)) { AddPrefix(key); } } // Add key to filter if needed inline void BlockBasedFilterBlockBuilder::AddKey(const Slice& key) { start_.push_back(entries_.size()); entries_.append(key.data(), key.size()); } // Add prefix to filter if needed inline void BlockBasedFilterBlockBuilder::AddPrefix(const Slice& key) { // get slice for most recently added entry Slice prev; if (start_.size() > added_to_start_) { size_t prev_start = start_[start_.size() - 1 - added_to_start_]; const char* base = entries_.data() + prev_start; size_t length = entries_.size() - prev_start; prev = Slice(base, length); } // this assumes prefix(prefix(key)) == prefix(key), as the last // entry in entries_ may be either a key or prefix, and we use // prefix(last entry) to get the prefix of the last key. if (prev.size() == 0 || !SamePrefix(prefix_extractor_, key, prev)) { Slice prefix = prefix_extractor_->Transform(key); start_.push_back(entries_.size()); entries_.append(prefix.data(), prefix.size()); } } Slice BlockBasedFilterBlockBuilder::Finish() { if (!start_.empty()) { GenerateFilter(); } // Append array of per-filter offsets const uint32_t array_offset = static_cast(result_.size()); for (size_t i = 0; i < filter_offsets_.size(); i++) { PutFixed32(&result_, filter_offsets_[i]); } PutFixed32(&result_, array_offset); result_.push_back(kFilterBaseLg); // Save encoding parameter in result return Slice(result_); } void BlockBasedFilterBlockBuilder::GenerateFilter() { const size_t num_entries = start_.size(); if (num_entries == 0) { // Fast path if there are no keys for this filter filter_offsets_.push_back(static_cast(result_.size())); return; } // Make list of keys from flattened key structure start_.push_back(entries_.size()); // Simplify length computation tmp_entries_.resize(num_entries); for (size_t i = 0; i < num_entries; i++) { const char* base = entries_.data() + start_[i]; size_t length = start_[i + 1] - start_[i]; tmp_entries_[i] = Slice(base, length); } // Generate filter for current set of keys and append to result_. filter_offsets_.push_back(static_cast(result_.size())); policy_->CreateFilter(&tmp_entries_[0], static_cast(num_entries), &result_); tmp_entries_.clear(); entries_.clear(); start_.clear(); } BlockBasedFilterBlockReader::BlockBasedFilterBlockReader( const SliceTransform* prefix_extractor, const BlockBasedTableOptions& table_opt, bool whole_key_filtering, BlockContents&& contents) : policy_(table_opt.filter_policy.get()), prefix_extractor_(prefix_extractor), whole_key_filtering_(whole_key_filtering), data_(nullptr), offset_(nullptr), num_(0), base_lg_(0), contents_(std::move(contents)) { assert(policy_); size_t n = contents_.data.size(); if (n < 5) return; // 1 byte for base_lg_ and 4 for start of offset array base_lg_ = contents_.data[n - 1]; uint32_t last_word = DecodeFixed32(contents_.data.data() + n - 5); if (last_word > n - 5) return; data_ = contents_.data.data(); offset_ = data_ + last_word; num_ = (n - 5 - last_word) / 4; } bool BlockBasedFilterBlockReader::KeyMayMatch(const Slice& key, uint64_t block_offset) { assert(block_offset != kNotValid); if (!whole_key_filtering_) { return true; } return MayMatch(key, block_offset); } bool BlockBasedFilterBlockReader::PrefixMayMatch(const Slice& prefix, uint64_t block_offset) { assert(block_offset != kNotValid); if (!prefix_extractor_) { return true; } return MayMatch(prefix, block_offset); } bool BlockBasedFilterBlockReader::MayMatch(const Slice& entry, uint64_t block_offset) { uint64_t index = block_offset >> base_lg_; if (index < num_) { uint32_t start = DecodeFixed32(offset_ + index * 4); uint32_t limit = DecodeFixed32(offset_ + index * 4 + 4); if (start <= limit && limit <= (uint32_t)(offset_ - data_)) { Slice filter = Slice(data_ + start, limit - start); bool const may_match = policy_->KeyMayMatch(entry, filter); if (may_match) { PERF_COUNTER_ADD(bloom_sst_hit_count, 1); return true; } else { PERF_COUNTER_ADD(bloom_sst_miss_count, 1); return false; } } else if (start == limit) { // Empty filters do not match any entries return false; } } return true; // Errors are treated as potential matches } size_t BlockBasedFilterBlockReader::ApproximateMemoryUsage() const { return num_ * 4 + 5 + (offset_ - data_); } std::string BlockBasedFilterBlockReader::ToString() const { std::string result, filter_meta; result.reserve(1024); std::string s_bo("Block offset"), s_hd("Hex dump"), s_fb("# filter blocks"); AppendItem(&result, s_fb, rocksdb::ToString(num_)); AppendItem(&result, s_bo, s_hd); for (size_t index = 0; index < num_; index++) { uint32_t start = DecodeFixed32(offset_ + index * 4); uint32_t limit = DecodeFixed32(offset_ + index * 4 + 4); if (start != limit) { result.append(" filter block # " + rocksdb::ToString(index + 1) + "\n"); Slice filter = Slice(data_ + start, limit - start); AppendItem(&result, start, filter.ToString(true)); } } return result; } } // namespace rocksdb rocksdb-4.5.1/table/block_based_filter_block.h000066400000000000000000000101651267530634000213700ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2012 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // A filter block is stored near the end of a Table file. It contains // filters (e.g., bloom filters) for all data blocks in the table combined // into a single filter block. #pragma once #include #include #include #include #include #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "table/filter_block.h" #include "util/hash.h" namespace rocksdb { // A BlockBasedFilterBlockBuilder is used to construct all of the filters for a // particular Table. It generates a single string which is stored as // a special block in the Table. // // The sequence of calls to BlockBasedFilterBlockBuilder must match the regexp: // (StartBlock Add*)* Finish class BlockBasedFilterBlockBuilder : public FilterBlockBuilder { public: BlockBasedFilterBlockBuilder(const SliceTransform* prefix_extractor, const BlockBasedTableOptions& table_opt); virtual bool IsBlockBased() override { return true; } virtual void StartBlock(uint64_t block_offset) override; virtual void Add(const Slice& key) override; virtual Slice Finish() override; private: void AddKey(const Slice& key); void AddPrefix(const Slice& key); void GenerateFilter(); // important: all of these might point to invalid addresses // at the time of destruction of this filter block. destructor // should NOT dereference them. const FilterPolicy* policy_; const SliceTransform* prefix_extractor_; bool whole_key_filtering_; std::string entries_; // Flattened entry contents std::vector start_; // Starting index in entries_ of each entry uint32_t added_to_start_; // To indicate if key is added std::string result_; // Filter data computed so far std::vector tmp_entries_; // policy_->CreateFilter() argument std::vector filter_offsets_; // No copying allowed BlockBasedFilterBlockBuilder(const BlockBasedFilterBlockBuilder&); void operator=(const BlockBasedFilterBlockBuilder&); }; // A FilterBlockReader is used to parse filter from SST table. // KeyMayMatch and PrefixMayMatch would trigger filter checking class BlockBasedFilterBlockReader : public FilterBlockReader { public: // REQUIRES: "contents" and *policy must stay live while *this is live. BlockBasedFilterBlockReader(const SliceTransform* prefix_extractor, const BlockBasedTableOptions& table_opt, bool whole_key_filtering, BlockContents&& contents); virtual bool IsBlockBased() override { return true; } virtual bool KeyMayMatch(const Slice& key, uint64_t block_offset = kNotValid) override; virtual bool PrefixMayMatch(const Slice& prefix, uint64_t block_offset = kNotValid) override; virtual size_t ApproximateMemoryUsage() const override; // convert this object to a human readable form std::string ToString() const override; private: const FilterPolicy* policy_; const SliceTransform* prefix_extractor_; bool whole_key_filtering_; const char* data_; // Pointer to filter data (at block-start) const char* offset_; // Pointer to beginning of offset array (at block-end) size_t num_; // Number of entries in offset array size_t base_lg_; // Encoding parameter (see kFilterBaseLg in .cc file) BlockContents contents_; bool MayMatch(const Slice& entry, uint64_t block_offset); // No copying allowed BlockBasedFilterBlockReader(const BlockBasedFilterBlockReader&); void operator=(const BlockBasedFilterBlockReader&); }; } // namespace rocksdb rocksdb-4.5.1/table/block_based_filter_block_test.cc000066400000000000000000000175711267530634000225750ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2012 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/block_based_filter_block.h" #include "rocksdb/filter_policy.h" #include "util/coding.h" #include "util/hash.h" #include "util/logging.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { // For testing: emit an array with one hash value per key class TestHashFilter : public FilterPolicy { public: virtual const char* Name() const override { return "TestHashFilter"; } virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const override { for (int i = 0; i < n; i++) { uint32_t h = Hash(keys[i].data(), keys[i].size(), 1); PutFixed32(dst, h); } } virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const override { uint32_t h = Hash(key.data(), key.size(), 1); for (unsigned int i = 0; i + 4 <= filter.size(); i += 4) { if (h == DecodeFixed32(filter.data() + i)) { return true; } } return false; } }; class FilterBlockTest : public testing::Test { public: TestHashFilter policy_; BlockBasedTableOptions table_options_; FilterBlockTest() { table_options_.filter_policy.reset(new TestHashFilter()); } }; TEST_F(FilterBlockTest, EmptyBuilder) { BlockBasedFilterBlockBuilder builder(nullptr, table_options_); BlockContents block(builder.Finish(), false, kNoCompression); ASSERT_EQ("\\x00\\x00\\x00\\x00\\x0b", EscapeString(block.data)); BlockBasedFilterBlockReader reader(nullptr, table_options_, true, std::move(block)); ASSERT_TRUE(reader.KeyMayMatch("foo", 0)); ASSERT_TRUE(reader.KeyMayMatch("foo", 100000)); } TEST_F(FilterBlockTest, SingleChunk) { BlockBasedFilterBlockBuilder builder(nullptr, table_options_); builder.StartBlock(100); builder.Add("foo"); builder.Add("bar"); builder.Add("box"); builder.StartBlock(200); builder.Add("box"); builder.StartBlock(300); builder.Add("hello"); BlockContents block(builder.Finish(), false, kNoCompression); BlockBasedFilterBlockReader reader(nullptr, table_options_, true, std::move(block)); ASSERT_TRUE(reader.KeyMayMatch("foo", 100)); ASSERT_TRUE(reader.KeyMayMatch("bar", 100)); ASSERT_TRUE(reader.KeyMayMatch("box", 100)); ASSERT_TRUE(reader.KeyMayMatch("hello", 100)); ASSERT_TRUE(reader.KeyMayMatch("foo", 100)); ASSERT_TRUE(!reader.KeyMayMatch("missing", 100)); ASSERT_TRUE(!reader.KeyMayMatch("other", 100)); } TEST_F(FilterBlockTest, MultiChunk) { BlockBasedFilterBlockBuilder builder(nullptr, table_options_); // First filter builder.StartBlock(0); builder.Add("foo"); builder.StartBlock(2000); builder.Add("bar"); // Second filter builder.StartBlock(3100); builder.Add("box"); // Third filter is empty // Last filter builder.StartBlock(9000); builder.Add("box"); builder.Add("hello"); BlockContents block(builder.Finish(), false, kNoCompression); BlockBasedFilterBlockReader reader(nullptr, table_options_, true, std::move(block)); // Check first filter ASSERT_TRUE(reader.KeyMayMatch("foo", 0)); ASSERT_TRUE(reader.KeyMayMatch("bar", 2000)); ASSERT_TRUE(!reader.KeyMayMatch("box", 0)); ASSERT_TRUE(!reader.KeyMayMatch("hello", 0)); // Check second filter ASSERT_TRUE(reader.KeyMayMatch("box", 3100)); ASSERT_TRUE(!reader.KeyMayMatch("foo", 3100)); ASSERT_TRUE(!reader.KeyMayMatch("bar", 3100)); ASSERT_TRUE(!reader.KeyMayMatch("hello", 3100)); // Check third filter (empty) ASSERT_TRUE(!reader.KeyMayMatch("foo", 4100)); ASSERT_TRUE(!reader.KeyMayMatch("bar", 4100)); ASSERT_TRUE(!reader.KeyMayMatch("box", 4100)); ASSERT_TRUE(!reader.KeyMayMatch("hello", 4100)); // Check last filter ASSERT_TRUE(reader.KeyMayMatch("box", 9000)); ASSERT_TRUE(reader.KeyMayMatch("hello", 9000)); ASSERT_TRUE(!reader.KeyMayMatch("foo", 9000)); ASSERT_TRUE(!reader.KeyMayMatch("bar", 9000)); } // Test for block based filter block // use new interface in FilterPolicy to create filter builder/reader class BlockBasedFilterBlockTest : public testing::Test { public: BlockBasedTableOptions table_options_; BlockBasedFilterBlockTest() { table_options_.filter_policy.reset(NewBloomFilterPolicy(10)); } ~BlockBasedFilterBlockTest() {} }; TEST_F(BlockBasedFilterBlockTest, BlockBasedEmptyBuilder) { FilterBlockBuilder* builder = new BlockBasedFilterBlockBuilder( nullptr, table_options_); BlockContents block(builder->Finish(), false, kNoCompression); ASSERT_EQ("\\x00\\x00\\x00\\x00\\x0b", EscapeString(block.data)); FilterBlockReader* reader = new BlockBasedFilterBlockReader( nullptr, table_options_, true, std::move(block)); ASSERT_TRUE(reader->KeyMayMatch("foo", 0)); ASSERT_TRUE(reader->KeyMayMatch("foo", 100000)); delete builder; delete reader; } TEST_F(BlockBasedFilterBlockTest, BlockBasedSingleChunk) { FilterBlockBuilder* builder = new BlockBasedFilterBlockBuilder( nullptr, table_options_); builder->StartBlock(100); builder->Add("foo"); builder->Add("bar"); builder->Add("box"); builder->StartBlock(200); builder->Add("box"); builder->StartBlock(300); builder->Add("hello"); BlockContents block(builder->Finish(), false, kNoCompression); FilterBlockReader* reader = new BlockBasedFilterBlockReader( nullptr, table_options_, true, std::move(block)); ASSERT_TRUE(reader->KeyMayMatch("foo", 100)); ASSERT_TRUE(reader->KeyMayMatch("bar", 100)); ASSERT_TRUE(reader->KeyMayMatch("box", 100)); ASSERT_TRUE(reader->KeyMayMatch("hello", 100)); ASSERT_TRUE(reader->KeyMayMatch("foo", 100)); ASSERT_TRUE(!reader->KeyMayMatch("missing", 100)); ASSERT_TRUE(!reader->KeyMayMatch("other", 100)); delete builder; delete reader; } TEST_F(BlockBasedFilterBlockTest, BlockBasedMultiChunk) { FilterBlockBuilder* builder = new BlockBasedFilterBlockBuilder( nullptr, table_options_); // First filter builder->StartBlock(0); builder->Add("foo"); builder->StartBlock(2000); builder->Add("bar"); // Second filter builder->StartBlock(3100); builder->Add("box"); // Third filter is empty // Last filter builder->StartBlock(9000); builder->Add("box"); builder->Add("hello"); BlockContents block(builder->Finish(), false, kNoCompression); FilterBlockReader* reader = new BlockBasedFilterBlockReader( nullptr, table_options_, true, std::move(block)); // Check first filter ASSERT_TRUE(reader->KeyMayMatch("foo", 0)); ASSERT_TRUE(reader->KeyMayMatch("bar", 2000)); ASSERT_TRUE(!reader->KeyMayMatch("box", 0)); ASSERT_TRUE(!reader->KeyMayMatch("hello", 0)); // Check second filter ASSERT_TRUE(reader->KeyMayMatch("box", 3100)); ASSERT_TRUE(!reader->KeyMayMatch("foo", 3100)); ASSERT_TRUE(!reader->KeyMayMatch("bar", 3100)); ASSERT_TRUE(!reader->KeyMayMatch("hello", 3100)); // Check third filter (empty) ASSERT_TRUE(!reader->KeyMayMatch("foo", 4100)); ASSERT_TRUE(!reader->KeyMayMatch("bar", 4100)); ASSERT_TRUE(!reader->KeyMayMatch("box", 4100)); ASSERT_TRUE(!reader->KeyMayMatch("hello", 4100)); // Check last filter ASSERT_TRUE(reader->KeyMayMatch("box", 9000)); ASSERT_TRUE(reader->KeyMayMatch("hello", 9000)); ASSERT_TRUE(!reader->KeyMayMatch("foo", 9000)); ASSERT_TRUE(!reader->KeyMayMatch("bar", 9000)); delete builder; delete reader; } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/table/block_based_table_builder.cc000066400000000000000000001024021267530634000216600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/block_based_table_builder.h" #include #include #include #include #include #include #include #include #include "db/dbformat.h" #include "rocksdb/cache.h" #include "rocksdb/comparator.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/flush_block_policy.h" #include "rocksdb/table.h" #include "table/block.h" #include "table/block_based_table_reader.h" #include "table/block_builder.h" #include "table/filter_block.h" #include "table/block_based_filter_block.h" #include "table/block_based_table_factory.h" #include "table/full_filter_block.h" #include "table/format.h" #include "table/meta_blocks.h" #include "table/table_builder.h" #include "util/string_util.h" #include "util/coding.h" #include "util/compression.h" #include "util/crc32c.h" #include "util/stop_watch.h" #include "util/xxhash.h" namespace rocksdb { extern const std::string kHashIndexPrefixesBlock; extern const std::string kHashIndexPrefixesMetadataBlock; typedef BlockBasedTableOptions::IndexType IndexType; // The interface for building index. // Instruction for adding a new concrete IndexBuilder: // 1. Create a subclass instantiated from IndexBuilder. // 2. Add a new entry associated with that subclass in TableOptions::IndexType. // 3. Add a create function for the new subclass in CreateIndexBuilder. // Note: we can devise more advanced design to simplify the process for adding // new subclass, which will, on the other hand, increase the code complexity and // catch unwanted attention from readers. Given that we won't add/change // indexes frequently, it makes sense to just embrace a more straightforward // design that just works. class IndexBuilder { public: // Index builder will construct a set of blocks which contain: // 1. One primary index block. // 2. (Optional) a set of metablocks that contains the metadata of the // primary index. struct IndexBlocks { Slice index_block_contents; std::unordered_map meta_blocks; }; explicit IndexBuilder(const Comparator* comparator) : comparator_(comparator) {} virtual ~IndexBuilder() {} // Add a new index entry to index block. // To allow further optimization, we provide `last_key_in_current_block` and // `first_key_in_next_block`, based on which the specific implementation can // determine the best index key to be used for the index block. // @last_key_in_current_block: this parameter maybe overridden with the value // "substitute key". // @first_key_in_next_block: it will be nullptr if the entry being added is // the last one in the table // // REQUIRES: Finish() has not yet been called. virtual void AddIndexEntry(std::string* last_key_in_current_block, const Slice* first_key_in_next_block, const BlockHandle& block_handle) = 0; // This method will be called whenever a key is added. The subclasses may // override OnKeyAdded() if they need to collect additional information. virtual void OnKeyAdded(const Slice& key) {} // Inform the index builder that all entries has been written. Block builder // may therefore perform any operation required for block finalization. // // REQUIRES: Finish() has not yet been called. virtual Status Finish(IndexBlocks* index_blocks) = 0; // Get the estimated size for index block. virtual size_t EstimatedSize() const = 0; protected: const Comparator* comparator_; }; // This index builder builds space-efficient index block. // // Optimizations: // 1. Made block's `block_restart_interval` to be 1, which will avoid linear // search when doing index lookup (can be disabled by setting // index_block_restart_interval). // 2. Shorten the key length for index block. Other than honestly using the // last key in the data block as the index key, we instead find a shortest // substitute key that serves the same function. class ShortenedIndexBuilder : public IndexBuilder { public: explicit ShortenedIndexBuilder(const Comparator* comparator, int index_block_restart_interval) : IndexBuilder(comparator), index_block_builder_(index_block_restart_interval) {} virtual void AddIndexEntry(std::string* last_key_in_current_block, const Slice* first_key_in_next_block, const BlockHandle& block_handle) override { if (first_key_in_next_block != nullptr) { comparator_->FindShortestSeparator(last_key_in_current_block, *first_key_in_next_block); } else { comparator_->FindShortSuccessor(last_key_in_current_block); } std::string handle_encoding; block_handle.EncodeTo(&handle_encoding); index_block_builder_.Add(*last_key_in_current_block, handle_encoding); } virtual Status Finish(IndexBlocks* index_blocks) override { index_blocks->index_block_contents = index_block_builder_.Finish(); return Status::OK(); } virtual size_t EstimatedSize() const override { return index_block_builder_.CurrentSizeEstimate(); } private: BlockBuilder index_block_builder_; }; // HashIndexBuilder contains a binary-searchable primary index and the // metadata for secondary hash index construction. // The metadata for hash index consists two parts: // - a metablock that compactly contains a sequence of prefixes. All prefixes // are stored consectively without any metadata (like, prefix sizes) being // stored, which is kept in the other metablock. // - a metablock contains the metadata of the prefixes, including prefix size, // restart index and number of block it spans. The format looks like: // // +-----------------+---------------------------+---------------------+ <=prefix 1 // | length: 4 bytes | restart interval: 4 bytes | num-blocks: 4 bytes | // +-----------------+---------------------------+---------------------+ <=prefix 2 // | length: 4 bytes | restart interval: 4 bytes | num-blocks: 4 bytes | // +-----------------+---------------------------+---------------------+ // | | // | .... | // | | // +-----------------+---------------------------+---------------------+ <=prefix n // | length: 4 bytes | restart interval: 4 bytes | num-blocks: 4 bytes | // +-----------------+---------------------------+---------------------+ // // The reason of separating these two metablocks is to enable the efficiently // reuse the first metablock during hash index construction without unnecessary // data copy or small heap allocations for prefixes. class HashIndexBuilder : public IndexBuilder { public: explicit HashIndexBuilder(const Comparator* comparator, const SliceTransform* hash_key_extractor, int index_block_restart_interval) : IndexBuilder(comparator), primary_index_builder_(comparator, index_block_restart_interval), hash_key_extractor_(hash_key_extractor) {} virtual void AddIndexEntry(std::string* last_key_in_current_block, const Slice* first_key_in_next_block, const BlockHandle& block_handle) override { ++current_restart_index_; primary_index_builder_.AddIndexEntry(last_key_in_current_block, first_key_in_next_block, block_handle); } virtual void OnKeyAdded(const Slice& key) override { auto key_prefix = hash_key_extractor_->Transform(key); bool is_first_entry = pending_block_num_ == 0; // Keys may share the prefix if (is_first_entry || pending_entry_prefix_ != key_prefix) { if (!is_first_entry) { FlushPendingPrefix(); } // need a hard copy otherwise the underlying data changes all the time. // TODO(kailiu) ToString() is expensive. We may speed up can avoid data // copy. pending_entry_prefix_ = key_prefix.ToString(); pending_block_num_ = 1; pending_entry_index_ = static_cast(current_restart_index_); } else { // entry number increments when keys share the prefix reside in // different data blocks. auto last_restart_index = pending_entry_index_ + pending_block_num_ - 1; assert(last_restart_index <= current_restart_index_); if (last_restart_index != current_restart_index_) { ++pending_block_num_; } } } virtual Status Finish(IndexBlocks* index_blocks) override { FlushPendingPrefix(); primary_index_builder_.Finish(index_blocks); index_blocks->meta_blocks.insert( {kHashIndexPrefixesBlock.c_str(), prefix_block_}); index_blocks->meta_blocks.insert( {kHashIndexPrefixesMetadataBlock.c_str(), prefix_meta_block_}); return Status::OK(); } virtual size_t EstimatedSize() const override { return primary_index_builder_.EstimatedSize() + prefix_block_.size() + prefix_meta_block_.size(); } private: void FlushPendingPrefix() { prefix_block_.append(pending_entry_prefix_.data(), pending_entry_prefix_.size()); PutVarint32(&prefix_meta_block_, static_cast(pending_entry_prefix_.size())); PutVarint32(&prefix_meta_block_, pending_entry_index_); PutVarint32(&prefix_meta_block_, pending_block_num_); } ShortenedIndexBuilder primary_index_builder_; const SliceTransform* hash_key_extractor_; // stores a sequence of prefixes std::string prefix_block_; // stores the metadata of prefixes std::string prefix_meta_block_; // The following 3 variables keeps unflushed prefix and its metadata. // The details of block_num and entry_index can be found in // "block_hash_index.{h,cc}" uint32_t pending_block_num_ = 0; uint32_t pending_entry_index_ = 0; std::string pending_entry_prefix_; uint64_t current_restart_index_ = 0; }; // Without anonymous namespace here, we fail the warning -Wmissing-prototypes namespace { // Create a index builder based on its type. IndexBuilder* CreateIndexBuilder(IndexType type, const Comparator* comparator, const SliceTransform* prefix_extractor, int index_block_restart_interval) { switch (type) { case BlockBasedTableOptions::kBinarySearch: { return new ShortenedIndexBuilder(comparator, index_block_restart_interval); } case BlockBasedTableOptions::kHashSearch: { return new HashIndexBuilder(comparator, prefix_extractor, index_block_restart_interval); } default: { assert(!"Do not recognize the index type "); return nullptr; } } // impossible. assert(false); return nullptr; } // Create a index builder based on its type. FilterBlockBuilder* CreateFilterBlockBuilder(const ImmutableCFOptions& opt, const BlockBasedTableOptions& table_opt) { if (table_opt.filter_policy == nullptr) return nullptr; FilterBitsBuilder* filter_bits_builder = table_opt.filter_policy->GetFilterBitsBuilder(); if (filter_bits_builder == nullptr) { return new BlockBasedFilterBlockBuilder(opt.prefix_extractor, table_opt); } else { return new FullFilterBlockBuilder(opt.prefix_extractor, table_opt.whole_key_filtering, filter_bits_builder); } } bool GoodCompressionRatio(size_t compressed_size, size_t raw_size) { // Check to see if compressed less than 12.5% return compressed_size < raw_size - (raw_size / 8u); } // format_version is the block format as defined in include/rocksdb/table.h Slice CompressBlock(const Slice& raw, const CompressionOptions& compression_options, CompressionType* type, uint32_t format_version, std::string* compressed_output) { if (*type == kNoCompression) { return raw; } // Will return compressed block contents if (1) the compression method is // supported in this platform and (2) the compression rate is "good enough". switch (*type) { case kSnappyCompression: if (Snappy_Compress(compression_options, raw.data(), raw.size(), compressed_output) && GoodCompressionRatio(compressed_output->size(), raw.size())) { return *compressed_output; } break; // fall back to no compression. case kZlibCompression: if (Zlib_Compress( compression_options, GetCompressFormatForVersion(kZlibCompression, format_version), raw.data(), raw.size(), compressed_output) && GoodCompressionRatio(compressed_output->size(), raw.size())) { return *compressed_output; } break; // fall back to no compression. case kBZip2Compression: if (BZip2_Compress( compression_options, GetCompressFormatForVersion(kBZip2Compression, format_version), raw.data(), raw.size(), compressed_output) && GoodCompressionRatio(compressed_output->size(), raw.size())) { return *compressed_output; } break; // fall back to no compression. case kLZ4Compression: if (LZ4_Compress( compression_options, GetCompressFormatForVersion(kLZ4Compression, format_version), raw.data(), raw.size(), compressed_output) && GoodCompressionRatio(compressed_output->size(), raw.size())) { return *compressed_output; } break; // fall back to no compression. case kLZ4HCCompression: if (LZ4HC_Compress( compression_options, GetCompressFormatForVersion(kLZ4HCCompression, format_version), raw.data(), raw.size(), compressed_output) && GoodCompressionRatio(compressed_output->size(), raw.size())) { return *compressed_output; } break; // fall back to no compression. case kZSTDNotFinalCompression: if (ZSTD_Compress(compression_options, raw.data(), raw.size(), compressed_output) && GoodCompressionRatio(compressed_output->size(), raw.size())) { return *compressed_output; } break; // fall back to no compression. default: {} // Do not recognize this compression type } // Compression method is not supported, or not good compression ratio, so just // fall back to uncompressed form. *type = kNoCompression; return raw; } } // namespace // kBlockBasedTableMagicNumber was picked by running // echo rocksdb.table.block_based | sha1sum // and taking the leading 64 bits. // Please note that kBlockBasedTableMagicNumber may also be accessed by other // .cc files // for that reason we declare it extern in the header but to get the space // allocated // it must be not extern in one place. const uint64_t kBlockBasedTableMagicNumber = 0x88e241b785f4cff7ull; // We also support reading and writing legacy block based table format (for // backwards compatibility) const uint64_t kLegacyBlockBasedTableMagicNumber = 0xdb4775248b80fb57ull; // A collector that collects properties of interest to block-based table. // For now this class looks heavy-weight since we only write one additional // property. // But in the foreseeable future, we will add more and more properties that are // specific to block-based table. class BlockBasedTableBuilder::BlockBasedTablePropertiesCollector : public IntTblPropCollector { public: explicit BlockBasedTablePropertiesCollector( BlockBasedTableOptions::IndexType index_type, bool whole_key_filtering, bool prefix_filtering) : index_type_(index_type), whole_key_filtering_(whole_key_filtering), prefix_filtering_(prefix_filtering) {} virtual Status InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) override { // Intentionally left blank. Have no interest in collecting stats for // individual key/value pairs. return Status::OK(); } virtual Status Finish(UserCollectedProperties* properties) override { std::string val; PutFixed32(&val, static_cast(index_type_)); properties->insert({BlockBasedTablePropertyNames::kIndexType, val}); properties->insert({BlockBasedTablePropertyNames::kWholeKeyFiltering, whole_key_filtering_ ? kPropTrue : kPropFalse}); properties->insert({BlockBasedTablePropertyNames::kPrefixFiltering, prefix_filtering_ ? kPropTrue : kPropFalse}); return Status::OK(); } // The name of the properties collector can be used for debugging purpose. virtual const char* Name() const override { return "BlockBasedTablePropertiesCollector"; } virtual UserCollectedProperties GetReadableProperties() const override { // Intentionally left blank. return UserCollectedProperties(); } private: BlockBasedTableOptions::IndexType index_type_; bool whole_key_filtering_; bool prefix_filtering_; }; struct BlockBasedTableBuilder::Rep { const ImmutableCFOptions ioptions; const BlockBasedTableOptions table_options; const InternalKeyComparator& internal_comparator; WritableFileWriter* file; uint64_t offset = 0; Status status; BlockBuilder data_block; InternalKeySliceTransform internal_prefix_transform; std::unique_ptr index_builder; std::string last_key; const CompressionType compression_type; const CompressionOptions compression_opts; TableProperties props; bool closed = false; // Either Finish() or Abandon() has been called. std::unique_ptr filter_block; char compressed_cache_key_prefix[BlockBasedTable::kMaxCacheKeyPrefixSize]; size_t compressed_cache_key_prefix_size; BlockHandle pending_handle; // Handle to add to index block std::string compressed_output; std::unique_ptr flush_block_policy; std::vector> table_properties_collectors; Rep(const ImmutableCFOptions& _ioptions, const BlockBasedTableOptions& table_opt, const InternalKeyComparator& icomparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* f, const CompressionType _compression_type, const CompressionOptions& _compression_opts, const bool skip_filters) : ioptions(_ioptions), table_options(table_opt), internal_comparator(icomparator), file(f), data_block(table_options.block_restart_interval, table_options.use_delta_encoding), internal_prefix_transform(_ioptions.prefix_extractor), index_builder( CreateIndexBuilder(table_options.index_type, &internal_comparator, &this->internal_prefix_transform, table_options.index_block_restart_interval)), compression_type(_compression_type), compression_opts(_compression_opts), filter_block(skip_filters ? nullptr : CreateFilterBlockBuilder( _ioptions, table_options)), flush_block_policy( table_options.flush_block_policy_factory->NewFlushBlockPolicy( table_options, data_block)) { for (auto& collector_factories : *int_tbl_prop_collector_factories) { table_properties_collectors.emplace_back( collector_factories->CreateIntTblPropCollector(column_family_id)); } table_properties_collectors.emplace_back( new BlockBasedTablePropertiesCollector( table_options.index_type, table_options.whole_key_filtering, _ioptions.prefix_extractor != nullptr)); } }; BlockBasedTableBuilder::BlockBasedTableBuilder( const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* file, const CompressionType compression_type, const CompressionOptions& compression_opts, const bool skip_filters) { BlockBasedTableOptions sanitized_table_options(table_options); if (sanitized_table_options.format_version == 0 && sanitized_table_options.checksum != kCRC32c) { Log(InfoLogLevel::WARN_LEVEL, ioptions.info_log, "Silently converting format_version to 1 because checksum is " "non-default"); // silently convert format_version to 1 to keep consistent with current // behavior sanitized_table_options.format_version = 1; } rep_ = new Rep(ioptions, sanitized_table_options, internal_comparator, int_tbl_prop_collector_factories, column_family_id, file, compression_type, compression_opts, skip_filters); if (rep_->filter_block != nullptr) { rep_->filter_block->StartBlock(0); } if (table_options.block_cache_compressed.get() != nullptr) { BlockBasedTable::GenerateCachePrefix( table_options.block_cache_compressed.get(), file->writable_file(), &rep_->compressed_cache_key_prefix[0], &rep_->compressed_cache_key_prefix_size); } } BlockBasedTableBuilder::~BlockBasedTableBuilder() { assert(rep_->closed); // Catch errors where caller forgot to call Finish() delete rep_; } void BlockBasedTableBuilder::Add(const Slice& key, const Slice& value) { Rep* r = rep_; assert(!r->closed); if (!ok()) return; if (r->props.num_entries > 0) { assert(r->internal_comparator.Compare(key, Slice(r->last_key)) > 0); } auto should_flush = r->flush_block_policy->Update(key, value); if (should_flush) { assert(!r->data_block.empty()); Flush(); // Add item to index block. // We do not emit the index entry for a block until we have seen the // first key for the next data block. This allows us to use shorter // keys in the index block. For example, consider a block boundary // between the keys "the quick brown fox" and "the who". We can use // "the r" as the key for the index block entry since it is >= all // entries in the first block and < all entries in subsequent // blocks. if (ok()) { r->index_builder->AddIndexEntry(&r->last_key, &key, r->pending_handle); } } if (r->filter_block != nullptr) { r->filter_block->Add(ExtractUserKey(key)); } r->last_key.assign(key.data(), key.size()); r->data_block.Add(key, value); r->props.num_entries++; r->props.raw_key_size += key.size(); r->props.raw_value_size += value.size(); r->index_builder->OnKeyAdded(key); NotifyCollectTableCollectorsOnAdd(key, value, r->offset, r->table_properties_collectors, r->ioptions.info_log); } void BlockBasedTableBuilder::Flush() { Rep* r = rep_; assert(!r->closed); if (!ok()) return; if (r->data_block.empty()) return; WriteBlock(&r->data_block, &r->pending_handle); if (ok() && !r->table_options.skip_table_builder_flush) { r->status = r->file->Flush(); } if (r->filter_block != nullptr) { r->filter_block->StartBlock(r->offset); } r->props.data_size = r->offset; ++r->props.num_data_blocks; } void BlockBasedTableBuilder::WriteBlock(BlockBuilder* block, BlockHandle* handle) { WriteBlock(block->Finish(), handle); block->Reset(); } void BlockBasedTableBuilder::WriteBlock(const Slice& raw_block_contents, BlockHandle* handle) { // File format contains a sequence of blocks where each block has: // block_data: uint8[n] // type: uint8 // crc: uint32 assert(ok()); Rep* r = rep_; auto type = r->compression_type; Slice block_contents; if (raw_block_contents.size() < kCompressionSizeLimit) { block_contents = CompressBlock(raw_block_contents, r->compression_opts, &type, r->table_options.format_version, &r->compressed_output); } else { RecordTick(r->ioptions.statistics, NUMBER_BLOCK_NOT_COMPRESSED); type = kNoCompression; block_contents = raw_block_contents; } WriteRawBlock(block_contents, type, handle); r->compressed_output.clear(); } void BlockBasedTableBuilder::WriteRawBlock(const Slice& block_contents, CompressionType type, BlockHandle* handle) { Rep* r = rep_; StopWatch sw(r->ioptions.env, r->ioptions.statistics, WRITE_RAW_BLOCK_MICROS); handle->set_offset(r->offset); handle->set_size(block_contents.size()); r->status = r->file->Append(block_contents); if (r->status.ok()) { char trailer[kBlockTrailerSize]; trailer[0] = type; char* trailer_without_type = trailer + 1; switch (r->table_options.checksum) { case kNoChecksum: // we don't support no checksum yet assert(false); // intentional fallthrough in release binary case kCRC32c: { auto crc = crc32c::Value(block_contents.data(), block_contents.size()); crc = crc32c::Extend(crc, trailer, 1); // Extend to cover block type EncodeFixed32(trailer_without_type, crc32c::Mask(crc)); break; } case kxxHash: { void* xxh = XXH32_init(0); XXH32_update(xxh, block_contents.data(), static_cast(block_contents.size())); XXH32_update(xxh, trailer, 1); // Extend to cover block type EncodeFixed32(trailer_without_type, XXH32_digest(xxh)); break; } } r->status = r->file->Append(Slice(trailer, kBlockTrailerSize)); if (r->status.ok()) { r->status = InsertBlockInCache(block_contents, type, handle); } if (r->status.ok()) { r->offset += block_contents.size() + kBlockTrailerSize; } } } Status BlockBasedTableBuilder::status() const { return rep_->status; } static void DeleteCachedBlock(const Slice& key, void* value) { Block* block = reinterpret_cast(value); delete block; } // // Make a copy of the block contents and insert into compressed block cache // Status BlockBasedTableBuilder::InsertBlockInCache(const Slice& block_contents, const CompressionType type, const BlockHandle* handle) { Rep* r = rep_; Cache* block_cache_compressed = r->table_options.block_cache_compressed.get(); if (type != kNoCompression && block_cache_compressed != nullptr) { Cache::Handle* cache_handle = nullptr; size_t size = block_contents.size(); std::unique_ptr ubuf(new char[size + 1]); memcpy(ubuf.get(), block_contents.data(), size); ubuf[size] = type; BlockContents results(std::move(ubuf), size, true, type); Block* block = new Block(std::move(results)); // make cache key by appending the file offset to the cache prefix id char* end = EncodeVarint64( r->compressed_cache_key_prefix + r->compressed_cache_key_prefix_size, handle->offset()); Slice key(r->compressed_cache_key_prefix, static_cast (end - r->compressed_cache_key_prefix)); // Insert into compressed block cache. cache_handle = block_cache_compressed->Insert( key, block, block->usable_size(), &DeleteCachedBlock); block_cache_compressed->Release(cache_handle); // Invalidate OS cache. r->file->InvalidateCache(static_cast(r->offset), size); } return Status::OK(); } Status BlockBasedTableBuilder::Finish() { Rep* r = rep_; bool empty_data_block = r->data_block.empty(); Flush(); assert(!r->closed); r->closed = true; BlockHandle filter_block_handle, metaindex_block_handle, index_block_handle; // Write filter block if (ok() && r->filter_block != nullptr) { auto filter_contents = r->filter_block->Finish(); r->props.filter_size = filter_contents.size(); WriteRawBlock(filter_contents, kNoCompression, &filter_block_handle); } // To make sure properties block is able to keep the accurate size of index // block, we will finish writing all index entries here and flush them // to storage after metaindex block is written. if (ok() && !empty_data_block) { r->index_builder->AddIndexEntry( &r->last_key, nullptr /* no next data block */, r->pending_handle); } IndexBuilder::IndexBlocks index_blocks; auto s = r->index_builder->Finish(&index_blocks); if (!s.ok()) { return s; } // Write meta blocks and metaindex block with the following order. // 1. [meta block: filter] // 2. [other meta blocks] // 3. [meta block: properties] // 4. [metaindex block] // write meta blocks MetaIndexBuilder meta_index_builder; for (const auto& item : index_blocks.meta_blocks) { BlockHandle block_handle; WriteBlock(item.second, &block_handle); meta_index_builder.Add(item.first, block_handle); } if (ok()) { if (r->filter_block != nullptr) { // Add mapping from ".Name" to location // of filter data. std::string key; if (r->filter_block->IsBlockBased()) { key = BlockBasedTable::kFilterBlockPrefix; } else { key = BlockBasedTable::kFullFilterBlockPrefix; } key.append(r->table_options.filter_policy->Name()); meta_index_builder.Add(key, filter_block_handle); } // Write properties block. { PropertyBlockBuilder property_block_builder; r->props.filter_policy_name = r->table_options.filter_policy != nullptr ? r->table_options.filter_policy->Name() : ""; r->props.index_size = r->index_builder->EstimatedSize() + kBlockTrailerSize; // Add basic properties property_block_builder.AddTableProperty(r->props); // Add use collected properties NotifyCollectTableCollectorsOnFinish(r->table_properties_collectors, r->ioptions.info_log, &property_block_builder); BlockHandle properties_block_handle; WriteRawBlock( property_block_builder.Finish(), kNoCompression, &properties_block_handle ); meta_index_builder.Add(kPropertiesBlock, properties_block_handle); } // end of properties block writing } // meta blocks // Write index block if (ok()) { // flush the meta index block WriteRawBlock(meta_index_builder.Finish(), kNoCompression, &metaindex_block_handle); WriteBlock(index_blocks.index_block_contents, &index_block_handle); } // Write footer if (ok()) { // No need to write out new footer if we're using default checksum. // We're writing legacy magic number because we want old versions of RocksDB // be able to read files generated with new release (just in case if // somebody wants to roll back after an upgrade) // TODO(icanadi) at some point in the future, when we're absolutely sure // nobody will roll back to RocksDB 2.x versions, retire the legacy magic // number and always write new table files with new magic number bool legacy = (r->table_options.format_version == 0); // this is guaranteed by BlockBasedTableBuilder's constructor assert(r->table_options.checksum == kCRC32c || r->table_options.format_version != 0); Footer footer(legacy ? kLegacyBlockBasedTableMagicNumber : kBlockBasedTableMagicNumber, r->table_options.format_version); footer.set_metaindex_handle(metaindex_block_handle); footer.set_index_handle(index_block_handle); footer.set_checksum(r->table_options.checksum); std::string footer_encoding; footer.EncodeTo(&footer_encoding); r->status = r->file->Append(footer_encoding); if (r->status.ok()) { r->offset += footer_encoding.size(); } } return r->status; } void BlockBasedTableBuilder::Abandon() { Rep* r = rep_; assert(!r->closed); r->closed = true; } uint64_t BlockBasedTableBuilder::NumEntries() const { return rep_->props.num_entries; } uint64_t BlockBasedTableBuilder::FileSize() const { return rep_->offset; } bool BlockBasedTableBuilder::NeedCompact() const { for (const auto& collector : rep_->table_properties_collectors) { if (collector->NeedCompact()) { return true; } } return false; } TableProperties BlockBasedTableBuilder::GetTableProperties() const { TableProperties ret = rep_->props; for (const auto& collector : rep_->table_properties_collectors) { for (const auto& prop : collector->GetReadableProperties()) { ret.readable_properties.insert(prop); } collector->Finish(&ret.user_collected_properties); } return ret; } const std::string BlockBasedTable::kFilterBlockPrefix = "filter."; const std::string BlockBasedTable::kFullFilterBlockPrefix = "fullfilter."; } // namespace rocksdb rocksdb-4.5.1/table/block_based_table_builder.h000066400000000000000000000105101267530634000215200ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include #include "rocksdb/flush_block_policy.h" #include "rocksdb/options.h" #include "rocksdb/status.h" #include "table/table_builder.h" namespace rocksdb { class BlockBuilder; class BlockHandle; class WritableFile; struct BlockBasedTableOptions; extern const uint64_t kBlockBasedTableMagicNumber; extern const uint64_t kLegacyBlockBasedTableMagicNumber; class BlockBasedTableBuilder : public TableBuilder { public: // Create a builder that will store the contents of the table it is // building in *file. Does not close the file. It is up to the // caller to close the file after calling Finish(). BlockBasedTableBuilder( const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* file, const CompressionType compression_type, const CompressionOptions& compression_opts, const bool skip_filters); // REQUIRES: Either Finish() or Abandon() has been called. ~BlockBasedTableBuilder(); // Add key,value to the table being constructed. // REQUIRES: key is after any previously added key according to comparator. // REQUIRES: Finish(), Abandon() have not been called void Add(const Slice& key, const Slice& value) override; // Return non-ok iff some error has been detected. Status status() const override; // Finish building the table. Stops using the file passed to the // constructor after this function returns. // REQUIRES: Finish(), Abandon() have not been called Status Finish() override; // Indicate that the contents of this builder should be abandoned. Stops // using the file passed to the constructor after this function returns. // If the caller is not going to call Finish(), it must call Abandon() // before destroying this builder. // REQUIRES: Finish(), Abandon() have not been called void Abandon() override; // Number of calls to Add() so far. uint64_t NumEntries() const override; // Size of the file generated so far. If invoked after a successful // Finish() call, returns the size of the final generated file. uint64_t FileSize() const override; bool NeedCompact() const override; // Get table properties TableProperties GetTableProperties() const override; private: bool ok() const { return status().ok(); } // Call block's Finish() method and then write the finalize block contents to // file. void WriteBlock(BlockBuilder* block, BlockHandle* handle); // Directly write block content to the file. void WriteBlock(const Slice& block_contents, BlockHandle* handle); void WriteRawBlock(const Slice& data, CompressionType, BlockHandle* handle); Status InsertBlockInCache(const Slice& block_contents, const CompressionType type, const BlockHandle* handle); struct Rep; class BlockBasedTablePropertiesCollectorFactory; class BlockBasedTablePropertiesCollector; Rep* rep_; // Advanced operation: flush any buffered key/value pairs to file. // Can be used to ensure that two adjacent entries never live in // the same data block. Most clients should not need to use this method. // REQUIRES: Finish(), Abandon() have not been called void Flush(); // Some compression libraries fail when the raw size is bigger than int. If // uncompressed size is bigger than kCompressionSizeLimit, don't compress it const uint64_t kCompressionSizeLimit = std::numeric_limits::max(); // No copying allowed BlockBasedTableBuilder(const BlockBasedTableBuilder&) = delete; void operator=(const BlockBasedTableBuilder&) = delete; }; } // namespace rocksdb rocksdb-4.5.1/table/block_based_table_factory.cc000066400000000000000000000172261267530634000217120ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/block_based_table_factory.h" #include #include #include #include "port/port.h" #include "rocksdb/flush_block_policy.h" #include "rocksdb/cache.h" #include "table/block_based_table_builder.h" #include "table/block_based_table_reader.h" #include "table/format.h" namespace rocksdb { BlockBasedTableFactory::BlockBasedTableFactory( const BlockBasedTableOptions& _table_options) : table_options_(_table_options) { if (table_options_.flush_block_policy_factory == nullptr) { table_options_.flush_block_policy_factory.reset( new FlushBlockBySizePolicyFactory()); } if (table_options_.no_block_cache) { table_options_.block_cache.reset(); } else if (table_options_.block_cache == nullptr) { table_options_.block_cache = NewLRUCache(8 << 20); } if (table_options_.block_size_deviation < 0 || table_options_.block_size_deviation > 100) { table_options_.block_size_deviation = 0; } if (table_options_.block_restart_interval < 1) { table_options_.block_restart_interval = 1; } if (table_options_.index_block_restart_interval < 1) { table_options_.index_block_restart_interval = 1; } } Status BlockBasedTableFactory::NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader) const { return NewTableReader(table_reader_options, std::move(file), file_size, table_reader, /*prefetch_index_and_filter=*/true); } Status BlockBasedTableFactory::NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader, const bool prefetch_enabled) const { return BlockBasedTable::Open( table_reader_options.ioptions, table_reader_options.env_options, table_options_, table_reader_options.internal_comparator, std::move(file), file_size, table_reader, prefetch_enabled, table_reader_options.skip_filters); } TableBuilder* BlockBasedTableFactory::NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const { auto table_builder = new BlockBasedTableBuilder( table_builder_options.ioptions, table_options_, table_builder_options.internal_comparator, table_builder_options.int_tbl_prop_collector_factories, column_family_id, file, table_builder_options.compression_type, table_builder_options.compression_opts, table_builder_options.skip_filters); return table_builder; } Status BlockBasedTableFactory::SanitizeOptions( const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const { if (table_options_.index_type == BlockBasedTableOptions::kHashSearch && cf_opts.prefix_extractor == nullptr) { return Status::InvalidArgument("Hash index is specified for block-based " "table, but prefix_extractor is not given"); } if (table_options_.cache_index_and_filter_blocks && table_options_.no_block_cache) { return Status::InvalidArgument("Enable cache_index_and_filter_blocks, " ", but block cache is disabled"); } if (!BlockBasedTableSupportedVersion(table_options_.format_version)) { return Status::InvalidArgument( "Unsupported BlockBasedTable format_version. Please check " "include/rocksdb/table.h for more info"); } return Status::OK(); } std::string BlockBasedTableFactory::GetPrintableTableOptions() const { std::string ret; ret.reserve(20000); const int kBufferSize = 200; char buffer[kBufferSize]; snprintf(buffer, kBufferSize, " flush_block_policy_factory: %s (%p)\n", table_options_.flush_block_policy_factory->Name(), table_options_.flush_block_policy_factory.get()); ret.append(buffer); snprintf(buffer, kBufferSize, " cache_index_and_filter_blocks: %d\n", table_options_.cache_index_and_filter_blocks); ret.append(buffer); snprintf(buffer, kBufferSize, " index_type: %d\n", table_options_.index_type); ret.append(buffer); snprintf(buffer, kBufferSize, " hash_index_allow_collision: %d\n", table_options_.hash_index_allow_collision); ret.append(buffer); snprintf(buffer, kBufferSize, " checksum: %d\n", table_options_.checksum); ret.append(buffer); snprintf(buffer, kBufferSize, " no_block_cache: %d\n", table_options_.no_block_cache); ret.append(buffer); snprintf(buffer, kBufferSize, " block_cache: %p\n", table_options_.block_cache.get()); ret.append(buffer); if (table_options_.block_cache) { snprintf(buffer, kBufferSize, " block_cache_size: %" ROCKSDB_PRIszt "\n", table_options_.block_cache->GetCapacity()); ret.append(buffer); } snprintf(buffer, kBufferSize, " block_cache_compressed: %p\n", table_options_.block_cache_compressed.get()); ret.append(buffer); if (table_options_.block_cache_compressed) { snprintf(buffer, kBufferSize, " block_cache_compressed_size: %" ROCKSDB_PRIszt "\n", table_options_.block_cache_compressed->GetCapacity()); ret.append(buffer); } snprintf(buffer, kBufferSize, " block_size: %" ROCKSDB_PRIszt "\n", table_options_.block_size); ret.append(buffer); snprintf(buffer, kBufferSize, " block_size_deviation: %d\n", table_options_.block_size_deviation); ret.append(buffer); snprintf(buffer, kBufferSize, " block_restart_interval: %d\n", table_options_.block_restart_interval); ret.append(buffer); snprintf(buffer, kBufferSize, " index_block_restart_interval: %d\n", table_options_.index_block_restart_interval); ret.append(buffer); snprintf(buffer, kBufferSize, " filter_policy: %s\n", table_options_.filter_policy == nullptr ? "nullptr" : table_options_.filter_policy->Name()); ret.append(buffer); snprintf(buffer, kBufferSize, " whole_key_filtering: %d\n", table_options_.whole_key_filtering); ret.append(buffer); snprintf(buffer, kBufferSize, " skip_table_builder_flush: %d\n", table_options_.skip_table_builder_flush); ret.append(buffer); snprintf(buffer, kBufferSize, " format_version: %d\n", table_options_.format_version); ret.append(buffer); return ret; } const BlockBasedTableOptions& BlockBasedTableFactory::table_options() const { return table_options_; } TableFactory* NewBlockBasedTableFactory( const BlockBasedTableOptions& _table_options) { return new BlockBasedTableFactory(_table_options); } const std::string BlockBasedTablePropertyNames::kIndexType = "rocksdb.block.based.table.index.type"; const std::string BlockBasedTablePropertyNames::kWholeKeyFiltering = "rocksdb.block.based.table.whole.key.filtering"; const std::string BlockBasedTablePropertyNames::kPrefixFiltering = "rocksdb.block.based.table.prefix.filtering"; const std::string kHashIndexPrefixesBlock = "rocksdb.hashindex.prefixes"; const std::string kHashIndexPrefixesMetadataBlock = "rocksdb.hashindex.metadata"; const std::string kPropTrue = "1"; const std::string kPropFalse = "0"; } // namespace rocksdb rocksdb-4.5.1/table/block_based_table_factory.h000066400000000000000000000047651267530634000215600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include "rocksdb/flush_block_policy.h" #include "rocksdb/table.h" #include "db/dbformat.h" namespace rocksdb { struct EnvOptions; using std::unique_ptr; class BlockBasedTableBuilder; class BlockBasedTableFactory : public TableFactory { public: explicit BlockBasedTableFactory( const BlockBasedTableOptions& table_options = BlockBasedTableOptions()); ~BlockBasedTableFactory() {} const char* Name() const override { return "BlockBasedTable"; } Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader) const override; // This is a variant of virtual member function NewTableReader function with // added capability to disable pre-fetching of blocks on BlockBasedTable::Open Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader, bool prefetch_index_and_filter) const; TableBuilder* NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const override; // Sanitizes the specified DB Options. Status SanitizeOptions(const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const override; std::string GetPrintableTableOptions() const override; const BlockBasedTableOptions& table_options() const; void* GetOptions() override { return &table_options_; } private: BlockBasedTableOptions table_options_; }; extern const std::string kHashIndexPrefixesBlock; extern const std::string kHashIndexPrefixesMetadataBlock; extern const std::string kPropTrue; extern const std::string kPropFalse; } // namespace rocksdb rocksdb-4.5.1/table/block_based_table_reader.cc000066400000000000000000001661161267530634000215100ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/block_based_table_reader.h" #include #include #include "db/dbformat.h" #include "rocksdb/cache.h" #include "rocksdb/comparator.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "rocksdb/statistics.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "table/block.h" #include "table/filter_block.h" #include "table/block_based_filter_block.h" #include "table/block_based_table_factory.h" #include "table/full_filter_block.h" #include "table/block_hash_index.h" #include "table/block_prefix_index.h" #include "table/format.h" #include "table/internal_iterator.h" #include "table/meta_blocks.h" #include "table/two_level_iterator.h" #include "table/get_context.h" #include "util/coding.h" #include "util/file_reader_writer.h" #include "util/perf_context_imp.h" #include "util/stop_watch.h" #include "util/string_util.h" namespace rocksdb { extern const uint64_t kBlockBasedTableMagicNumber; extern const std::string kHashIndexPrefixesBlock; extern const std::string kHashIndexPrefixesMetadataBlock; using std::unique_ptr; typedef BlockBasedTable::IndexReader IndexReader; namespace { // The longest the prefix of the cache key used to identify blocks can be. // We are using the fact that we know for Posix files the unique ID is three // varints. // For some reason, compiling for iOS complains that this variable is unused const size_t kMaxCacheKeyPrefixSize __attribute__((unused)) = kMaxVarint64Length * 3 + 1; // Read the block identified by "handle" from "file". // The only relevant option is options.verify_checksums for now. // On failure return non-OK. // On success fill *result and return OK - caller owns *result Status ReadBlockFromFile(RandomAccessFileReader* file, const Footer& footer, const ReadOptions& options, const BlockHandle& handle, std::unique_ptr* result, Env* env, bool do_uncompress = true) { BlockContents contents; Status s = ReadBlockContents(file, footer, options, handle, &contents, env, do_uncompress); if (s.ok()) { result->reset(new Block(std::move(contents))); } return s; } // Delete the resource that is held by the iterator. template void DeleteHeldResource(void* arg, void* ignored) { delete reinterpret_cast(arg); } // Delete the entry resided in the cache. template void DeleteCachedEntry(const Slice& key, void* value) { auto entry = reinterpret_cast(value); delete entry; } // Release the cached entry and decrement its ref count. void ReleaseCachedEntry(void* arg, void* h) { Cache* cache = reinterpret_cast(arg); Cache::Handle* handle = reinterpret_cast(h); cache->Release(handle); } Slice GetCacheKey(const char* cache_key_prefix, size_t cache_key_prefix_size, const BlockHandle& handle, char* cache_key) { assert(cache_key != nullptr); assert(cache_key_prefix_size != 0); assert(cache_key_prefix_size <= kMaxCacheKeyPrefixSize); memcpy(cache_key, cache_key_prefix, cache_key_prefix_size); char* end = EncodeVarint64(cache_key + cache_key_prefix_size, handle.offset()); return Slice(cache_key, static_cast(end - cache_key)); } Cache::Handle* GetEntryFromCache(Cache* block_cache, const Slice& key, Tickers block_cache_miss_ticker, Tickers block_cache_hit_ticker, Statistics* statistics) { auto cache_handle = block_cache->Lookup(key); if (cache_handle != nullptr) { PERF_COUNTER_ADD(block_cache_hit_count, 1); // overall cache hit RecordTick(statistics, BLOCK_CACHE_HIT); // total bytes read from cache RecordTick(statistics, BLOCK_CACHE_BYTES_READ, block_cache->GetUsage(cache_handle)); // block-type specific cache hit RecordTick(statistics, block_cache_hit_ticker); } else { // overall cache miss RecordTick(statistics, BLOCK_CACHE_MISS); // block-type specific cache miss RecordTick(statistics, block_cache_miss_ticker); } return cache_handle; } } // namespace // -- IndexReader and its subclasses // IndexReader is the interface that provide the functionality for index access. class BlockBasedTable::IndexReader { public: explicit IndexReader(const Comparator* comparator) : comparator_(comparator) {} virtual ~IndexReader() {} // Create an iterator for index access. // An iter is passed in, if it is not null, update this one and return it // If it is null, create a new Iterator virtual InternalIterator* NewIterator(BlockIter* iter = nullptr, bool total_order_seek = true) = 0; // The size of the index. virtual size_t size() const = 0; // Memory usage of the index block virtual size_t usable_size() const = 0; // Report an approximation of how much memory has been used other than memory // that was allocated in block cache. virtual size_t ApproximateMemoryUsage() const = 0; protected: const Comparator* comparator_; }; // Index that allows binary search lookup for the first key of each block. // This class can be viewed as a thin wrapper for `Block` class which already // supports binary search. class BinarySearchIndexReader : public IndexReader { public: // Read index from the file and create an intance for // `BinarySearchIndexReader`. // On success, index_reader will be populated; otherwise it will remain // unmodified. static Status Create(RandomAccessFileReader* file, const Footer& footer, const BlockHandle& index_handle, Env* env, const Comparator* comparator, IndexReader** index_reader) { std::unique_ptr index_block; auto s = ReadBlockFromFile(file, footer, ReadOptions(), index_handle, &index_block, env); if (s.ok()) { *index_reader = new BinarySearchIndexReader(comparator, std::move(index_block)); } return s; } virtual InternalIterator* NewIterator(BlockIter* iter = nullptr, bool dont_care = true) override { return index_block_->NewIterator(comparator_, iter, true); } virtual size_t size() const override { return index_block_->size(); } virtual size_t usable_size() const override { return index_block_->usable_size(); } virtual size_t ApproximateMemoryUsage() const override { assert(index_block_); return index_block_->ApproximateMemoryUsage(); } private: BinarySearchIndexReader(const Comparator* comparator, std::unique_ptr&& index_block) : IndexReader(comparator), index_block_(std::move(index_block)) { assert(index_block_ != nullptr); } std::unique_ptr index_block_; }; // Index that leverages an internal hash table to quicken the lookup for a given // key. class HashIndexReader : public IndexReader { public: static Status Create(const SliceTransform* hash_key_extractor, const Footer& footer, RandomAccessFileReader* file, Env* env, const Comparator* comparator, const BlockHandle& index_handle, InternalIterator* meta_index_iter, IndexReader** index_reader, bool hash_index_allow_collision) { std::unique_ptr index_block; auto s = ReadBlockFromFile(file, footer, ReadOptions(), index_handle, &index_block, env); if (!s.ok()) { return s; } // Note, failure to create prefix hash index does not need to be a // hard error. We can still fall back to the original binary search index. // So, Create will succeed regardless, from this point on. auto new_index_reader = new HashIndexReader(comparator, std::move(index_block)); *index_reader = new_index_reader; // Get prefixes block BlockHandle prefixes_handle; s = FindMetaBlock(meta_index_iter, kHashIndexPrefixesBlock, &prefixes_handle); if (!s.ok()) { // TODO: log error return Status::OK(); } // Get index metadata block BlockHandle prefixes_meta_handle; s = FindMetaBlock(meta_index_iter, kHashIndexPrefixesMetadataBlock, &prefixes_meta_handle); if (!s.ok()) { // TODO: log error return Status::OK(); } // Read contents for the blocks BlockContents prefixes_contents; s = ReadBlockContents(file, footer, ReadOptions(), prefixes_handle, &prefixes_contents, env, true /* do decompression */); if (!s.ok()) { return s; } BlockContents prefixes_meta_contents; s = ReadBlockContents(file, footer, ReadOptions(), prefixes_meta_handle, &prefixes_meta_contents, env, true /* do decompression */); if (!s.ok()) { // TODO: log error return Status::OK(); } if (!hash_index_allow_collision) { // TODO: deprecate once hash_index_allow_collision proves to be stable. BlockHashIndex* hash_index = nullptr; s = CreateBlockHashIndex(hash_key_extractor, prefixes_contents.data, prefixes_meta_contents.data, &hash_index); // TODO: log error if (s.ok()) { new_index_reader->index_block_->SetBlockHashIndex(hash_index); new_index_reader->OwnPrefixesContents(std::move(prefixes_contents)); } } else { BlockPrefixIndex* prefix_index = nullptr; s = BlockPrefixIndex::Create(hash_key_extractor, prefixes_contents.data, prefixes_meta_contents.data, &prefix_index); // TODO: log error if (s.ok()) { new_index_reader->index_block_->SetBlockPrefixIndex(prefix_index); } } return Status::OK(); } virtual InternalIterator* NewIterator(BlockIter* iter = nullptr, bool total_order_seek = true) override { return index_block_->NewIterator(comparator_, iter, total_order_seek); } virtual size_t size() const override { return index_block_->size(); } virtual size_t usable_size() const override { return index_block_->usable_size(); } virtual size_t ApproximateMemoryUsage() const override { assert(index_block_); return index_block_->ApproximateMemoryUsage() + prefixes_contents_.data.size(); } private: HashIndexReader(const Comparator* comparator, std::unique_ptr&& index_block) : IndexReader(comparator), index_block_(std::move(index_block)) { assert(index_block_ != nullptr); } ~HashIndexReader() { } void OwnPrefixesContents(BlockContents&& prefixes_contents) { prefixes_contents_ = std::move(prefixes_contents); } std::unique_ptr index_block_; BlockContents prefixes_contents_; }; struct BlockBasedTable::Rep { Rep(const ImmutableCFOptions& _ioptions, const EnvOptions& _env_options, const BlockBasedTableOptions& _table_opt, const InternalKeyComparator& _internal_comparator, bool skip_filters) : ioptions(_ioptions), env_options(_env_options), table_options(_table_opt), filter_policy(skip_filters ? nullptr : _table_opt.filter_policy.get()), internal_comparator(_internal_comparator), filter_type(FilterType::kNoFilter), whole_key_filtering(_table_opt.whole_key_filtering), prefix_filtering(true) {} const ImmutableCFOptions& ioptions; const EnvOptions& env_options; const BlockBasedTableOptions& table_options; const FilterPolicy* const filter_policy; const InternalKeyComparator& internal_comparator; Status status; unique_ptr file; char cache_key_prefix[kMaxCacheKeyPrefixSize]; size_t cache_key_prefix_size = 0; char compressed_cache_key_prefix[kMaxCacheKeyPrefixSize]; size_t compressed_cache_key_prefix_size = 0; // Footer contains the fixed table information Footer footer; // index_reader and filter will be populated and used only when // options.block_cache is nullptr; otherwise we will get the index block via // the block cache. unique_ptr index_reader; unique_ptr filter; enum class FilterType { kNoFilter, kFullFilter, kBlockFilter, }; FilterType filter_type; BlockHandle filter_handle; std::shared_ptr table_properties; BlockBasedTableOptions::IndexType index_type; bool hash_index_allow_collision; bool whole_key_filtering; bool prefix_filtering; // TODO(kailiu) It is very ugly to use internal key in table, since table // module should not be relying on db module. However to make things easier // and compatible with existing code, we introduce a wrapper that allows // block to extract prefix without knowing if a key is internal or not. unique_ptr internal_prefix_transform; }; BlockBasedTable::~BlockBasedTable() { delete rep_; } // CachableEntry represents the entries that *may* be fetched from block cache. // field `value` is the item we want to get. // field `cache_handle` is the cache handle to the block cache. If the value // was not read from cache, `cache_handle` will be nullptr. template struct BlockBasedTable::CachableEntry { CachableEntry(TValue* _value, Cache::Handle* _cache_handle) : value(_value), cache_handle(_cache_handle) {} CachableEntry() : CachableEntry(nullptr, nullptr) {} void Release(Cache* cache) { if (cache_handle) { cache->Release(cache_handle); value = nullptr; cache_handle = nullptr; } } TValue* value = nullptr; // if the entry is from the cache, cache_handle will be populated. Cache::Handle* cache_handle = nullptr; }; // Helper function to setup the cache key's prefix for the Table. void BlockBasedTable::SetupCacheKeyPrefix(Rep* rep) { assert(kMaxCacheKeyPrefixSize >= 10); rep->cache_key_prefix_size = 0; rep->compressed_cache_key_prefix_size = 0; if (rep->table_options.block_cache != nullptr) { GenerateCachePrefix(rep->table_options.block_cache.get(), rep->file->file(), &rep->cache_key_prefix[0], &rep->cache_key_prefix_size); } if (rep->table_options.block_cache_compressed != nullptr) { GenerateCachePrefix(rep->table_options.block_cache_compressed.get(), rep->file->file(), &rep->compressed_cache_key_prefix[0], &rep->compressed_cache_key_prefix_size); } } void BlockBasedTable::GenerateCachePrefix(Cache* cc, RandomAccessFile* file, char* buffer, size_t* size) { // generate an id from the file *size = file->GetUniqueId(buffer, kMaxCacheKeyPrefixSize); // If the prefix wasn't generated or was too long, // create one from the cache. if (*size == 0) { char* end = EncodeVarint64(buffer, cc->NewId()); *size = static_cast(end - buffer); } } void BlockBasedTable::GenerateCachePrefix(Cache* cc, WritableFile* file, char* buffer, size_t* size) { // generate an id from the file *size = file->GetUniqueId(buffer, kMaxCacheKeyPrefixSize); // If the prefix wasn't generated or was too long, // create one from the cache. if (*size == 0) { char* end = EncodeVarint64(buffer, cc->NewId()); *size = static_cast(end - buffer); } } namespace { // Return True if table_properties has `user_prop_name` has a `true` value // or it doesn't contain this property (for backward compatible). bool IsFeatureSupported(const TableProperties& table_properties, const std::string& user_prop_name, Logger* info_log) { auto& props = table_properties.user_collected_properties; auto pos = props.find(user_prop_name); // Older version doesn't have this value set. Skip this check. if (pos != props.end()) { if (pos->second == kPropFalse) { return false; } else if (pos->second != kPropTrue) { Log(InfoLogLevel::WARN_LEVEL, info_log, "Property %s has invalidate value %s", user_prop_name.c_str(), pos->second.c_str()); } } return true; } } // namespace Status BlockBasedTable::Open(const ImmutableCFOptions& ioptions, const EnvOptions& env_options, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader, const bool prefetch_index_and_filter, const bool skip_filters) { table_reader->reset(); Footer footer; auto s = ReadFooterFromFile(file.get(), file_size, &footer, kBlockBasedTableMagicNumber); if (!s.ok()) { return s; } if (!BlockBasedTableSupportedVersion(footer.version())) { return Status::Corruption( "Unknown Footer version. Maybe this file was created with newer " "version of RocksDB?"); } // We've successfully read the footer and the index block: we're // ready to serve requests. Rep* rep = new BlockBasedTable::Rep(ioptions, env_options, table_options, internal_comparator, skip_filters); rep->file = std::move(file); rep->footer = footer; rep->index_type = table_options.index_type; rep->hash_index_allow_collision = table_options.hash_index_allow_collision; SetupCacheKeyPrefix(rep); unique_ptr new_table(new BlockBasedTable(rep)); // Read meta index std::unique_ptr meta; std::unique_ptr meta_iter; s = ReadMetaBlock(rep, &meta, &meta_iter); if (!s.ok()) { return s; } // Find filter handle and filter type if (rep->filter_policy) { for (auto prefix : {kFullFilterBlockPrefix, kFilterBlockPrefix}) { std::string filter_block_key = prefix; filter_block_key.append(rep->filter_policy->Name()); if (FindMetaBlock(meta_iter.get(), filter_block_key, &rep->filter_handle) .ok()) { rep->filter_type = (prefix == kFullFilterBlockPrefix) ? Rep::FilterType::kFullFilter : Rep::FilterType::kBlockFilter; break; } } } // Read the properties bool found_properties_block = true; s = SeekToPropertiesBlock(meta_iter.get(), &found_properties_block); if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, rep->ioptions.info_log, "Cannot seek to properties block from file: %s", s.ToString().c_str()); } else if (found_properties_block) { s = meta_iter->status(); TableProperties* table_properties = nullptr; if (s.ok()) { s = ReadProperties(meta_iter->value(), rep->file.get(), rep->footer, rep->ioptions.env, rep->ioptions.info_log, &table_properties); } if (!s.ok()) { Log(InfoLogLevel::WARN_LEVEL, rep->ioptions.info_log, "Encountered error while reading data from properties " "block %s", s.ToString().c_str()); } else { rep->table_properties.reset(table_properties); } } else { Log(InfoLogLevel::ERROR_LEVEL, rep->ioptions.info_log, "Cannot find Properties block from file."); } // Determine whether whole key filtering is supported. if (rep->table_properties) { rep->whole_key_filtering &= IsFeatureSupported(*(rep->table_properties), BlockBasedTablePropertyNames::kWholeKeyFiltering, rep->ioptions.info_log); rep->prefix_filtering &= IsFeatureSupported( *(rep->table_properties), BlockBasedTablePropertyNames::kPrefixFiltering, rep->ioptions.info_log); } if (prefetch_index_and_filter) { // pre-fetching of blocks is turned on // Will use block cache for index/filter blocks access? if (table_options.cache_index_and_filter_blocks) { assert(table_options.block_cache != nullptr); // Hack: Call NewIndexIterator() to implicitly add index to the // block_cache unique_ptr iter( new_table->NewIndexIterator(ReadOptions())); s = iter->status(); if (s.ok()) { // Hack: Call GetFilter() to implicitly add filter to the block_cache auto filter_entry = new_table->GetFilter(); filter_entry.Release(table_options.block_cache.get()); } } else { // If we don't use block cache for index/filter blocks access, we'll // pre-load these blocks, which will kept in member variables in Rep // and with a same life-time as this table object. IndexReader* index_reader = nullptr; s = new_table->CreateIndexReader(&index_reader, meta_iter.get()); if (s.ok()) { rep->index_reader.reset(index_reader); // Set filter block if (rep->filter_policy) { rep->filter.reset(ReadFilter(rep, nullptr)); } } else { delete index_reader; } } } if (s.ok()) { *table_reader = std::move(new_table); } return s; } void BlockBasedTable::SetupForCompaction() { switch (rep_->ioptions.access_hint_on_compaction_start) { case Options::NONE: break; case Options::NORMAL: rep_->file->file()->Hint(RandomAccessFile::NORMAL); break; case Options::SEQUENTIAL: rep_->file->file()->Hint(RandomAccessFile::SEQUENTIAL); break; case Options::WILLNEED: rep_->file->file()->Hint(RandomAccessFile::WILLNEED); break; default: assert(false); } compaction_optimized_ = true; } std::shared_ptr BlockBasedTable::GetTableProperties() const { return rep_->table_properties; } size_t BlockBasedTable::ApproximateMemoryUsage() const { size_t usage = 0; if (rep_->filter) { usage += rep_->filter->ApproximateMemoryUsage(); } if (rep_->index_reader) { usage += rep_->index_reader->ApproximateMemoryUsage(); } return usage; } // Load the meta-block from the file. On success, return the loaded meta block // and its iterator. Status BlockBasedTable::ReadMetaBlock(Rep* rep, std::unique_ptr* meta_block, std::unique_ptr* iter) { // TODO(sanjay): Skip this if footer.metaindex_handle() size indicates // it is an empty block. // TODO: we never really verify check sum for meta index block std::unique_ptr meta; Status s = ReadBlockFromFile( rep->file.get(), rep->footer, ReadOptions(), rep->footer.metaindex_handle(), &meta, rep->ioptions.env); if (!s.ok()) { Log(InfoLogLevel::ERROR_LEVEL, rep->ioptions.info_log, "Encountered error while reading data from properties" " block %s", s.ToString().c_str()); return s; } *meta_block = std::move(meta); // meta block uses bytewise comparator. iter->reset(meta_block->get()->NewIterator(BytewiseComparator())); return Status::OK(); } Status BlockBasedTable::GetDataBlockFromCache( const Slice& block_cache_key, const Slice& compressed_block_cache_key, Cache* block_cache, Cache* block_cache_compressed, Statistics* statistics, const ReadOptions& read_options, BlockBasedTable::CachableEntry* block, uint32_t format_version) { Status s; Block* compressed_block = nullptr; Cache::Handle* block_cache_compressed_handle = nullptr; // Lookup uncompressed cache first if (block_cache != nullptr) { block->cache_handle = GetEntryFromCache(block_cache, block_cache_key, BLOCK_CACHE_DATA_MISS, BLOCK_CACHE_DATA_HIT, statistics); if (block->cache_handle != nullptr) { block->value = reinterpret_cast(block_cache->Value(block->cache_handle)); return s; } } // If not found, search from the compressed block cache. assert(block->cache_handle == nullptr && block->value == nullptr); if (block_cache_compressed == nullptr) { return s; } assert(!compressed_block_cache_key.empty()); block_cache_compressed_handle = block_cache_compressed->Lookup(compressed_block_cache_key); // if we found in the compressed cache, then uncompress and insert into // uncompressed cache if (block_cache_compressed_handle == nullptr) { RecordTick(statistics, BLOCK_CACHE_COMPRESSED_MISS); return s; } // found compressed block RecordTick(statistics, BLOCK_CACHE_COMPRESSED_HIT); compressed_block = reinterpret_cast( block_cache_compressed->Value(block_cache_compressed_handle)); assert(compressed_block->compression_type() != kNoCompression); // Retrieve the uncompressed contents into a new buffer BlockContents contents; s = UncompressBlockContents(compressed_block->data(), compressed_block->size(), &contents, format_version); // Insert uncompressed block into block cache if (s.ok()) { block->value = new Block(std::move(contents)); // uncompressed block assert(block->value->compression_type() == kNoCompression); if (block_cache != nullptr && block->value->cachable() && read_options.fill_cache) { block->cache_handle = block_cache->Insert(block_cache_key, block->value, block->value->usable_size(), &DeleteCachedEntry); assert(reinterpret_cast( block_cache->Value(block->cache_handle)) == block->value); } } // Release hold on compressed cache entry block_cache_compressed->Release(block_cache_compressed_handle); return s; } Status BlockBasedTable::PutDataBlockToCache( const Slice& block_cache_key, const Slice& compressed_block_cache_key, Cache* block_cache, Cache* block_cache_compressed, const ReadOptions& read_options, Statistics* statistics, CachableEntry* block, Block* raw_block, uint32_t format_version) { assert(raw_block->compression_type() == kNoCompression || block_cache_compressed != nullptr); Status s; // Retrieve the uncompressed contents into a new buffer BlockContents contents; if (raw_block->compression_type() != kNoCompression) { s = UncompressBlockContents(raw_block->data(), raw_block->size(), &contents, format_version); } if (!s.ok()) { delete raw_block; return s; } if (raw_block->compression_type() != kNoCompression) { block->value = new Block(std::move(contents)); // uncompressed block } else { block->value = raw_block; raw_block = nullptr; } // Insert compressed block into compressed block cache. // Release the hold on the compressed cache entry immediately. if (block_cache_compressed != nullptr && raw_block != nullptr && raw_block->cachable()) { auto cache_handle = block_cache_compressed->Insert( compressed_block_cache_key, raw_block, raw_block->usable_size(), &DeleteCachedEntry); block_cache_compressed->Release(cache_handle); RecordTick(statistics, BLOCK_CACHE_COMPRESSED_MISS); // Avoid the following code to delete this cached block. raw_block = nullptr; } delete raw_block; // insert into uncompressed block cache assert((block->value->compression_type() == kNoCompression)); if (block_cache != nullptr && block->value->cachable()) { block->cache_handle = block_cache->Insert(block_cache_key, block->value, block->value->usable_size(), &DeleteCachedEntry); RecordTick(statistics, BLOCK_CACHE_ADD); RecordTick(statistics, BLOCK_CACHE_BYTES_WRITE, block->value->usable_size()); assert(reinterpret_cast(block_cache->Value(block->cache_handle)) == block->value); } return s; } FilterBlockReader* BlockBasedTable::ReadFilter(Rep* rep, size_t* filter_size) { // TODO: We might want to unify with ReadBlockFromFile() if we start // requiring checksum verification in Table::Open. if (rep->filter_type == Rep::FilterType::kNoFilter) { return nullptr; } BlockContents block; if (!ReadBlockContents(rep->file.get(), rep->footer, ReadOptions(), rep->filter_handle, &block, rep->ioptions.env, false).ok()) { // Error reading the block return nullptr; } if (filter_size) { *filter_size = block.data.size(); } assert(rep->filter_policy); if (rep->filter_type == Rep::FilterType::kBlockFilter) { return new BlockBasedFilterBlockReader( rep->prefix_filtering ? rep->ioptions.prefix_extractor : nullptr, rep->table_options, rep->whole_key_filtering, std::move(block)); } else if (rep->filter_type == Rep::FilterType::kFullFilter) { auto filter_bits_reader = rep->filter_policy->GetFilterBitsReader(block.data); if (filter_bits_reader != nullptr) { return new FullFilterBlockReader( rep->prefix_filtering ? rep->ioptions.prefix_extractor : nullptr, rep->whole_key_filtering, std::move(block), filter_bits_reader); } } // filter_type is either kNoFilter (exited the function at the first if), // kBlockFilter or kFullFilter. there is no way for the execution to come here assert(false); return nullptr; } BlockBasedTable::CachableEntry BlockBasedTable::GetFilter( bool no_io) const { // If cache_index_and_filter_blocks is false, filter should be pre-populated. // We will return rep_->filter anyway. rep_->filter can be nullptr if filter // read fails at Open() time. We don't want to reload again since it will // most probably fail again. if (!rep_->table_options.cache_index_and_filter_blocks) { return {rep_->filter.get(), nullptr /* cache handle */}; } PERF_TIMER_GUARD(read_filter_block_nanos); Cache* block_cache = rep_->table_options.block_cache.get(); if (rep_->filter_policy == nullptr /* do not use filter */ || block_cache == nullptr /* no block cache at all */) { return {nullptr /* filter */, nullptr /* cache handle */}; } // Fetching from the cache char cache_key[kMaxCacheKeyPrefixSize + kMaxVarint64Length]; auto key = GetCacheKey(rep_->cache_key_prefix, rep_->cache_key_prefix_size, rep_->footer.metaindex_handle(), cache_key); Statistics* statistics = rep_->ioptions.statistics; auto cache_handle = GetEntryFromCache(block_cache, key, BLOCK_CACHE_FILTER_MISS, BLOCK_CACHE_FILTER_HIT, statistics); FilterBlockReader* filter = nullptr; if (cache_handle != nullptr) { filter = reinterpret_cast( block_cache->Value(cache_handle)); } else if (no_io) { // Do not invoke any io. return CachableEntry(); } else { size_t filter_size = 0; filter = ReadFilter(rep_, &filter_size); if (filter != nullptr) { assert(filter_size > 0); cache_handle = block_cache->Insert(key, filter, filter_size, &DeleteCachedEntry); RecordTick(statistics, BLOCK_CACHE_ADD); RecordTick(statistics, BLOCK_CACHE_BYTES_WRITE, filter_size); } } return { filter, cache_handle }; } InternalIterator* BlockBasedTable::NewIndexIterator( const ReadOptions& read_options, BlockIter* input_iter) { // index reader has already been pre-populated. if (rep_->index_reader) { return rep_->index_reader->NewIterator( input_iter, read_options.total_order_seek); } PERF_TIMER_GUARD(read_index_block_nanos); bool no_io = read_options.read_tier == kBlockCacheTier; Cache* block_cache = rep_->table_options.block_cache.get(); char cache_key[kMaxCacheKeyPrefixSize + kMaxVarint64Length]; auto key = GetCacheKey(rep_->cache_key_prefix, rep_->cache_key_prefix_size, rep_->footer.index_handle(), cache_key); Statistics* statistics = rep_->ioptions.statistics; auto cache_handle = GetEntryFromCache(block_cache, key, BLOCK_CACHE_INDEX_MISS, BLOCK_CACHE_INDEX_HIT, statistics); if (cache_handle == nullptr && no_io) { if (input_iter != nullptr) { input_iter->SetStatus(Status::Incomplete("no blocking io")); return input_iter; } else { return NewErrorInternalIterator(Status::Incomplete("no blocking io")); } } IndexReader* index_reader = nullptr; if (cache_handle != nullptr) { index_reader = reinterpret_cast(block_cache->Value(cache_handle)); } else { // Create index reader and put it in the cache. Status s; s = CreateIndexReader(&index_reader); if (!s.ok()) { // make sure if something goes wrong, index_reader shall remain intact. assert(index_reader == nullptr); if (input_iter != nullptr) { input_iter->SetStatus(s); return input_iter; } else { return NewErrorInternalIterator(s); } } cache_handle = block_cache->Insert(key, index_reader, index_reader->usable_size(), &DeleteCachedEntry); RecordTick(statistics, BLOCK_CACHE_ADD); RecordTick(statistics, BLOCK_CACHE_BYTES_WRITE, index_reader->usable_size()); } assert(cache_handle); auto* iter = index_reader->NewIterator( input_iter, read_options.total_order_seek); iter->RegisterCleanup(&ReleaseCachedEntry, block_cache, cache_handle); return iter; } // Convert an index iterator value (i.e., an encoded BlockHandle) // into an iterator over the contents of the corresponding block. // If input_iter is null, new a iterator // If input_iter is not null, update this iter and return it InternalIterator* BlockBasedTable::NewDataBlockIterator( Rep* rep, const ReadOptions& ro, const Slice& index_value, BlockIter* input_iter) { PERF_TIMER_GUARD(new_table_block_iter_nanos); const bool no_io = (ro.read_tier == kBlockCacheTier); Cache* block_cache = rep->table_options.block_cache.get(); Cache* block_cache_compressed = rep->table_options.block_cache_compressed.get(); CachableEntry block; BlockHandle handle; Slice input = index_value; // We intentionally allow extra stuff in index_value so that we // can add more features in the future. Status s = handle.DecodeFrom(&input); if (!s.ok()) { if (input_iter != nullptr) { input_iter->SetStatus(s); return input_iter; } else { return NewErrorInternalIterator(s); } } // If either block cache is enabled, we'll try to read from it. if (block_cache != nullptr || block_cache_compressed != nullptr) { Statistics* statistics = rep->ioptions.statistics; char cache_key[kMaxCacheKeyPrefixSize + kMaxVarint64Length]; char compressed_cache_key[kMaxCacheKeyPrefixSize + kMaxVarint64Length]; Slice key, /* key to the block cache */ ckey /* key to the compressed block cache */; // create key for block cache if (block_cache != nullptr) { key = GetCacheKey(rep->cache_key_prefix, rep->cache_key_prefix_size, handle, cache_key); } if (block_cache_compressed != nullptr) { ckey = GetCacheKey(rep->compressed_cache_key_prefix, rep->compressed_cache_key_prefix_size, handle, compressed_cache_key); } s = GetDataBlockFromCache(key, ckey, block_cache, block_cache_compressed, statistics, ro, &block, rep->table_options.format_version); if (block.value == nullptr && !no_io && ro.fill_cache) { std::unique_ptr raw_block; { StopWatch sw(rep->ioptions.env, statistics, READ_BLOCK_GET_MICROS); s = ReadBlockFromFile(rep->file.get(), rep->footer, ro, handle, &raw_block, rep->ioptions.env, block_cache_compressed == nullptr); } if (s.ok()) { s = PutDataBlockToCache(key, ckey, block_cache, block_cache_compressed, ro, statistics, &block, raw_block.release(), rep->table_options.format_version); } } } // Didn't get any data from block caches. if (block.value == nullptr) { if (no_io) { // Could not read from block_cache and can't do IO if (input_iter != nullptr) { input_iter->SetStatus(Status::Incomplete("no blocking io")); return input_iter; } else { return NewErrorInternalIterator(Status::Incomplete("no blocking io")); } } std::unique_ptr block_value; s = ReadBlockFromFile(rep->file.get(), rep->footer, ro, handle, &block_value, rep->ioptions.env); if (s.ok()) { block.value = block_value.release(); } } InternalIterator* iter; if (block.value != nullptr) { iter = block.value->NewIterator(&rep->internal_comparator, input_iter); if (block.cache_handle != nullptr) { iter->RegisterCleanup(&ReleaseCachedEntry, block_cache, block.cache_handle); } else { iter->RegisterCleanup(&DeleteHeldResource, block.value, nullptr); } } else { if (input_iter != nullptr) { input_iter->SetStatus(s); iter = input_iter; } else { iter = NewErrorInternalIterator(s); } } return iter; } class BlockBasedTable::BlockEntryIteratorState : public TwoLevelIteratorState { public: BlockEntryIteratorState(BlockBasedTable* table, const ReadOptions& read_options, bool skip_filters) : TwoLevelIteratorState(table->rep_->ioptions.prefix_extractor != nullptr), table_(table), read_options_(read_options), skip_filters_(skip_filters) {} InternalIterator* NewSecondaryIterator(const Slice& index_value) override { return NewDataBlockIterator(table_->rep_, read_options_, index_value); } bool PrefixMayMatch(const Slice& internal_key) override { if (read_options_.total_order_seek || skip_filters_) { return true; } return table_->PrefixMayMatch(internal_key); } private: // Don't own table_ BlockBasedTable* table_; const ReadOptions read_options_; bool skip_filters_; }; // This will be broken if the user specifies an unusual implementation // of Options.comparator, or if the user specifies an unusual // definition of prefixes in BlockBasedTableOptions.filter_policy. // In particular, we require the following three properties: // // 1) key.starts_with(prefix(key)) // 2) Compare(prefix(key), key) <= 0. // 3) If Compare(key1, key2) <= 0, then Compare(prefix(key1), prefix(key2)) <= 0 // // Otherwise, this method guarantees no I/O will be incurred. // // REQUIRES: this method shouldn't be called while the DB lock is held. bool BlockBasedTable::PrefixMayMatch(const Slice& internal_key) { if (!rep_->filter_policy) { return true; } assert(rep_->ioptions.prefix_extractor != nullptr); auto user_key = ExtractUserKey(internal_key); if (!rep_->ioptions.prefix_extractor->InDomain(user_key)) { return true; } auto prefix = rep_->ioptions.prefix_extractor->Transform(user_key); InternalKey internal_key_prefix(prefix, kMaxSequenceNumber, kTypeValue); auto internal_prefix = internal_key_prefix.Encode(); bool may_match = true; Status s; // To prevent any io operation in this method, we set `read_tier` to make // sure we always read index or filter only when they have already been // loaded to memory. ReadOptions no_io_read_options; no_io_read_options.read_tier = kBlockCacheTier; // First, try check with full filter auto filter_entry = GetFilter(true /* no io */); FilterBlockReader* filter = filter_entry.value; if (filter != nullptr && !filter->IsBlockBased()) { may_match = filter->PrefixMayMatch(prefix); } // Then, try find it within each block if (may_match) { unique_ptr iiter(NewIndexIterator(no_io_read_options)); iiter->Seek(internal_prefix); if (!iiter->Valid()) { // we're past end of file // if it's incomplete, it means that we avoided I/O // and we're not really sure that we're past the end // of the file may_match = iiter->status().IsIncomplete(); } else if (ExtractUserKey(iiter->key()).starts_with( ExtractUserKey(internal_prefix))) { // we need to check for this subtle case because our only // guarantee is that "the key is a string >= last key in that data // block" according to the doc/table_format.txt spec. // // Suppose iiter->key() starts with the desired prefix; it is not // necessarily the case that the corresponding data block will // contain the prefix, since iiter->key() need not be in the // block. However, the next data block may contain the prefix, so // we return true to play it safe. may_match = true; } else if (filter != nullptr && filter->IsBlockBased()) { // iiter->key() does NOT start with the desired prefix. Because // Seek() finds the first key that is >= the seek target, this // means that iiter->key() > prefix. Thus, any data blocks coming // after the data block corresponding to iiter->key() cannot // possibly contain the key. Thus, the corresponding data block // is the only on could potentially contain the prefix. Slice handle_value = iiter->value(); BlockHandle handle; s = handle.DecodeFrom(&handle_value); assert(s.ok()); may_match = filter->PrefixMayMatch(prefix, handle.offset()); } } Statistics* statistics = rep_->ioptions.statistics; RecordTick(statistics, BLOOM_FILTER_PREFIX_CHECKED); if (!may_match) { RecordTick(statistics, BLOOM_FILTER_PREFIX_USEFUL); } filter_entry.Release(rep_->table_options.block_cache.get()); return may_match; } InternalIterator* BlockBasedTable::NewIterator(const ReadOptions& read_options, Arena* arena, bool skip_filters) { return NewTwoLevelIterator( new BlockEntryIteratorState(this, read_options, skip_filters), NewIndexIterator(read_options), arena); } bool BlockBasedTable::FullFilterKeyMayMatch(FilterBlockReader* filter, const Slice& internal_key) const { if (filter == nullptr || filter->IsBlockBased()) { return true; } Slice user_key = ExtractUserKey(internal_key); if (!filter->KeyMayMatch(user_key)) { return false; } if (rep_->ioptions.prefix_extractor && rep_->ioptions.prefix_extractor->InDomain(user_key) && !filter->PrefixMayMatch( rep_->ioptions.prefix_extractor->Transform(user_key))) { return false; } return true; } Status BlockBasedTable::Get(const ReadOptions& read_options, const Slice& key, GetContext* get_context, bool skip_filters) { Status s; CachableEntry filter_entry; if (!skip_filters) { filter_entry = GetFilter(read_options.read_tier == kBlockCacheTier); } FilterBlockReader* filter = filter_entry.value; // First check the full filter // If full filter not useful, Then go into each block if (!FullFilterKeyMayMatch(filter, key)) { RecordTick(rep_->ioptions.statistics, BLOOM_FILTER_USEFUL); } else { BlockIter iiter; NewIndexIterator(read_options, &iiter); bool done = false; for (iiter.Seek(key); iiter.Valid() && !done; iiter.Next()) { Slice handle_value = iiter.value(); BlockHandle handle; bool not_exist_in_filter = filter != nullptr && filter->IsBlockBased() == true && handle.DecodeFrom(&handle_value).ok() && !filter->KeyMayMatch(ExtractUserKey(key), handle.offset()); if (not_exist_in_filter) { // Not found // TODO: think about interaction with Merge. If a user key cannot // cross one data block, we should be fine. RecordTick(rep_->ioptions.statistics, BLOOM_FILTER_USEFUL); break; } else { BlockIter biter; NewDataBlockIterator(rep_, read_options, iiter.value(), &biter); if (read_options.read_tier == kBlockCacheTier && biter.status().IsIncomplete()) { // couldn't get block from block_cache // Update Saver.state to Found because we are only looking for whether // we can guarantee the key is not there when "no_io" is set get_context->MarkKeyMayExist(); break; } if (!biter.status().ok()) { s = biter.status(); break; } // Call the *saver function on each entry/block until it returns false for (biter.Seek(key); biter.Valid(); biter.Next()) { ParsedInternalKey parsed_key; if (!ParseInternalKey(biter.key(), &parsed_key)) { s = Status::Corruption(Slice()); } if (!get_context->SaveValue(parsed_key, biter.value())) { done = true; break; } } s = biter.status(); } } if (s.ok()) { s = iiter.status(); } } filter_entry.Release(rep_->table_options.block_cache.get()); return s; } Status BlockBasedTable::Prefetch(const Slice* const begin, const Slice* const end) { auto& comparator = rep_->internal_comparator; // pre-condition if (begin && end && comparator.Compare(*begin, *end) > 0) { return Status::InvalidArgument(*begin, *end); } BlockIter iiter; NewIndexIterator(ReadOptions(), &iiter); if (!iiter.status().ok()) { // error opening index iterator return iiter.status(); } // indicates if we are on the last page that need to be pre-fetched bool prefetching_boundary_page = false; for (begin ? iiter.Seek(*begin) : iiter.SeekToFirst(); iiter.Valid(); iiter.Next()) { Slice block_handle = iiter.value(); if (end && comparator.Compare(iiter.key(), *end) >= 0) { if (prefetching_boundary_page) { break; } // The index entry represents the last key in the data block. // We should load this page into memory as well, but no more prefetching_boundary_page = true; } // Load the block specified by the block_handle into the block cache BlockIter biter; NewDataBlockIterator(rep_, ReadOptions(), block_handle, &biter); if (!biter.status().ok()) { // there was an unexpected error while pre-fetching return biter.status(); } } return Status::OK(); } bool BlockBasedTable::TEST_KeyInCache(const ReadOptions& options, const Slice& key) { std::unique_ptr iiter(NewIndexIterator(options)); iiter->Seek(key); assert(iiter->Valid()); CachableEntry block; BlockHandle handle; Slice input = iiter->value(); Status s = handle.DecodeFrom(&input); assert(s.ok()); Cache* block_cache = rep_->table_options.block_cache.get(); assert(block_cache != nullptr); char cache_key_storage[kMaxCacheKeyPrefixSize + kMaxVarint64Length]; Slice cache_key = GetCacheKey(rep_->cache_key_prefix, rep_->cache_key_prefix_size, handle, cache_key_storage); Slice ckey; s = GetDataBlockFromCache(cache_key, ckey, block_cache, nullptr, nullptr, options, &block, rep_->table_options.format_version); assert(s.ok()); bool in_cache = block.value != nullptr; if (in_cache) { ReleaseCachedEntry(block_cache, block.cache_handle); } return in_cache; } // REQUIRES: The following fields of rep_ should have already been populated: // 1. file // 2. index_handle, // 3. options // 4. internal_comparator // 5. index_type Status BlockBasedTable::CreateIndexReader( IndexReader** index_reader, InternalIterator* preloaded_meta_index_iter) { // Some old version of block-based tables don't have index type present in // table properties. If that's the case we can safely use the kBinarySearch. auto index_type_on_file = BlockBasedTableOptions::kBinarySearch; if (rep_->table_properties) { auto& props = rep_->table_properties->user_collected_properties; auto pos = props.find(BlockBasedTablePropertyNames::kIndexType); if (pos != props.end()) { index_type_on_file = static_cast( DecodeFixed32(pos->second.c_str())); } } auto file = rep_->file.get(); auto env = rep_->ioptions.env; auto comparator = &rep_->internal_comparator; const Footer& footer = rep_->footer; if (index_type_on_file == BlockBasedTableOptions::kHashSearch && rep_->ioptions.prefix_extractor == nullptr) { Log(InfoLogLevel::WARN_LEVEL, rep_->ioptions.info_log, "BlockBasedTableOptions::kHashSearch requires " "options.prefix_extractor to be set." " Fall back to binary search index."); index_type_on_file = BlockBasedTableOptions::kBinarySearch; } switch (index_type_on_file) { case BlockBasedTableOptions::kBinarySearch: { return BinarySearchIndexReader::Create( file, footer, footer.index_handle(), env, comparator, index_reader); } case BlockBasedTableOptions::kHashSearch: { std::unique_ptr meta_guard; std::unique_ptr meta_iter_guard; auto meta_index_iter = preloaded_meta_index_iter; if (meta_index_iter == nullptr) { auto s = ReadMetaBlock(rep_, &meta_guard, &meta_iter_guard); if (!s.ok()) { // we simply fall back to binary search in case there is any // problem with prefix hash index loading. Log(InfoLogLevel::WARN_LEVEL, rep_->ioptions.info_log, "Unable to read the metaindex block." " Fall back to binary search index."); return BinarySearchIndexReader::Create( file, footer, footer.index_handle(), env, comparator, index_reader); } meta_index_iter = meta_iter_guard.get(); } // We need to wrap data with internal_prefix_transform to make sure it can // handle prefix correctly. rep_->internal_prefix_transform.reset( new InternalKeySliceTransform(rep_->ioptions.prefix_extractor)); return HashIndexReader::Create( rep_->internal_prefix_transform.get(), footer, file, env, comparator, footer.index_handle(), meta_index_iter, index_reader, rep_->hash_index_allow_collision); } default: { std::string error_message = "Unrecognized index type: " + ToString(rep_->index_type); return Status::InvalidArgument(error_message.c_str()); } } } uint64_t BlockBasedTable::ApproximateOffsetOf(const Slice& key) { unique_ptr index_iter(NewIndexIterator(ReadOptions())); index_iter->Seek(key); uint64_t result; if (index_iter->Valid()) { BlockHandle handle; Slice input = index_iter->value(); Status s = handle.DecodeFrom(&input); if (s.ok()) { result = handle.offset(); } else { // Strange: we can't decode the block handle in the index block. // We'll just return the offset of the metaindex block, which is // close to the whole file size for this case. result = rep_->footer.metaindex_handle().offset(); } } else { // key is past the last key in the file. If table_properties is not // available, approximate the offset by returning the offset of the // metaindex block (which is right near the end of the file). result = 0; if (rep_->table_properties) { result = rep_->table_properties->data_size; } // table_properties is not present in the table. if (result == 0) { result = rep_->footer.metaindex_handle().offset(); } } return result; } bool BlockBasedTable::TEST_filter_block_preloaded() const { return rep_->filter != nullptr; } bool BlockBasedTable::TEST_index_reader_preloaded() const { return rep_->index_reader != nullptr; } Status BlockBasedTable::DumpTable(WritableFile* out_file) { // Output Footer out_file->Append( "Footer Details:\n" "--------------------------------------\n" " "); out_file->Append(rep_->footer.ToString().c_str()); out_file->Append("\n"); // Output MetaIndex out_file->Append( "Metaindex Details:\n" "--------------------------------------\n"); std::unique_ptr meta; std::unique_ptr meta_iter; Status s = ReadMetaBlock(rep_, &meta, &meta_iter); if (s.ok()) { for (meta_iter->SeekToFirst(); meta_iter->Valid(); meta_iter->Next()) { s = meta_iter->status(); if (!s.ok()) { return s; } if (meta_iter->key() == rocksdb::kPropertiesBlock) { out_file->Append(" Properties block handle: "); out_file->Append(meta_iter->value().ToString(true).c_str()); out_file->Append("\n"); } else if (strstr(meta_iter->key().ToString().c_str(), "filter.rocksdb.") != nullptr) { out_file->Append(" Filter block handle: "); out_file->Append(meta_iter->value().ToString(true).c_str()); out_file->Append("\n"); } } out_file->Append("\n"); } else { return s; } // Output TableProperties const rocksdb::TableProperties* table_properties; table_properties = rep_->table_properties.get(); if (table_properties != nullptr) { out_file->Append( "Table Properties:\n" "--------------------------------------\n" " "); out_file->Append(table_properties->ToString("\n ", ": ").c_str()); out_file->Append("\n"); } // Output Filter blocks if (!rep_->filter && !table_properties->filter_policy_name.empty()) { // Support only BloomFilter as off now rocksdb::BlockBasedTableOptions table_options; table_options.filter_policy.reset(rocksdb::NewBloomFilterPolicy(1)); if (table_properties->filter_policy_name.compare( table_options.filter_policy->Name()) == 0) { std::string filter_block_key = kFilterBlockPrefix; filter_block_key.append(table_properties->filter_policy_name); BlockHandle handle; if (FindMetaBlock(meta_iter.get(), filter_block_key, &handle).ok()) { BlockContents block; if (ReadBlockContents(rep_->file.get(), rep_->footer, ReadOptions(), handle, &block, rep_->ioptions.env, false).ok()) { rep_->filter.reset(new BlockBasedFilterBlockReader( rep_->ioptions.prefix_extractor, table_options, table_options.whole_key_filtering, std::move(block))); } } } } if (rep_->filter) { out_file->Append( "Filter Details:\n" "--------------------------------------\n" " "); out_file->Append(rep_->filter->ToString().c_str()); out_file->Append("\n"); } // Output Index block s = DumpIndexBlock(out_file); if (!s.ok()) { return s; } // Output Data blocks s = DumpDataBlocks(out_file); return s; } Status BlockBasedTable::DumpIndexBlock(WritableFile* out_file) { out_file->Append( "Index Details:\n" "--------------------------------------\n"); std::unique_ptr blockhandles_iter( NewIndexIterator(ReadOptions())); Status s = blockhandles_iter->status(); if (!s.ok()) { out_file->Append("Can not read Index Block \n\n"); return s; } out_file->Append(" Block key hex dump: Data block handle\n"); out_file->Append(" Block key ascii\n\n"); for (blockhandles_iter->SeekToFirst(); blockhandles_iter->Valid(); blockhandles_iter->Next()) { s = blockhandles_iter->status(); if (!s.ok()) { break; } Slice key = blockhandles_iter->key(); InternalKey ikey; ikey.DecodeFrom(key); out_file->Append(" HEX "); out_file->Append(ikey.user_key().ToString(true).c_str()); out_file->Append(": "); out_file->Append(blockhandles_iter->value().ToString(true).c_str()); out_file->Append("\n"); std::string str_key = ikey.user_key().ToString(); std::string res_key(""); char cspace = ' '; for (size_t i = 0; i < str_key.size(); i++) { res_key.append(&str_key[i], 1); res_key.append(1, cspace); } out_file->Append(" ASCII "); out_file->Append(res_key.c_str()); out_file->Append("\n ------\n"); } out_file->Append("\n"); return Status::OK(); } Status BlockBasedTable::DumpDataBlocks(WritableFile* out_file) { std::unique_ptr blockhandles_iter( NewIndexIterator(ReadOptions())); Status s = blockhandles_iter->status(); if (!s.ok()) { out_file->Append("Can not read Index Block \n\n"); return s; } size_t block_id = 1; for (blockhandles_iter->SeekToFirst(); blockhandles_iter->Valid(); block_id++, blockhandles_iter->Next()) { s = blockhandles_iter->status(); if (!s.ok()) { break; } out_file->Append("Data Block # "); out_file->Append(rocksdb::ToString(block_id)); out_file->Append(" @ "); out_file->Append(blockhandles_iter->value().ToString(true).c_str()); out_file->Append("\n"); out_file->Append("--------------------------------------\n"); std::unique_ptr datablock_iter; datablock_iter.reset( NewDataBlockIterator(rep_, ReadOptions(), blockhandles_iter->value())); s = datablock_iter->status(); if (!s.ok()) { out_file->Append("Error reading the block - Skipped \n\n"); continue; } for (datablock_iter->SeekToFirst(); datablock_iter->Valid(); datablock_iter->Next()) { s = datablock_iter->status(); if (!s.ok()) { out_file->Append("Error reading the block - Skipped \n"); break; } Slice key = datablock_iter->key(); Slice value = datablock_iter->value(); InternalKey ikey, iValue; ikey.DecodeFrom(key); iValue.DecodeFrom(value); out_file->Append(" HEX "); out_file->Append(ikey.user_key().ToString(true).c_str()); out_file->Append(": "); out_file->Append(iValue.user_key().ToString(true).c_str()); out_file->Append("\n"); std::string str_key = ikey.user_key().ToString(); std::string str_value = iValue.user_key().ToString(); std::string res_key(""), res_value(""); char cspace = ' '; for (size_t i = 0; i < str_key.size(); i++) { res_key.append(&str_key[i], 1); res_key.append(1, cspace); } for (size_t i = 0; i < str_value.size(); i++) { res_value.append(&str_value[i], 1); res_value.append(1, cspace); } out_file->Append(" ASCII "); out_file->Append(res_key.c_str()); out_file->Append(": "); out_file->Append(res_value.c_str()); out_file->Append("\n ------\n"); } out_file->Append("\n"); } return Status::OK(); } } // namespace rocksdb rocksdb-4.5.1/table/block_based_table_reader.h000066400000000000000000000227061267530634000213460ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include "rocksdb/options.h" #include "rocksdb/statistics.h" #include "rocksdb/status.h" #include "rocksdb/table.h" #include "table/table_reader.h" #include "table/table_properties_internal.h" #include "util/coding.h" #include "util/file_reader_writer.h" namespace rocksdb { class Block; class BlockIter; class BlockHandle; class Cache; class FilterBlockReader; class BlockBasedFilterBlockReader; class FullFilterBlockReader; class Footer; class InternalKeyComparator; class Iterator; class RandomAccessFile; class TableCache; class TableReader; class WritableFile; struct BlockBasedTableOptions; struct EnvOptions; struct ReadOptions; class GetContext; class InternalIterator; using std::unique_ptr; // A Table is a sorted map from strings to strings. Tables are // immutable and persistent. A Table may be safely accessed from // multiple threads without external synchronization. class BlockBasedTable : public TableReader { public: static const std::string kFilterBlockPrefix; static const std::string kFullFilterBlockPrefix; // Attempt to open the table that is stored in bytes [0..file_size) // of "file", and read the metadata entries necessary to allow // retrieving data from the table. // // If successful, returns ok and sets "*table_reader" to the newly opened // table. The client should delete "*table_reader" when no longer needed. // If there was an error while initializing the table, sets "*table_reader" // to nullptr and returns a non-ok status. // // @param file must remain live while this Table is in use. // @param prefetch_index_and_filter can be used to disable prefetching of // index and filter blocks at startup // @param skip_filters Disables loading/accessing the filter block. Overrides // prefetch_index_and_filter, so filter will be skipped if both are set. static Status Open(const ImmutableCFOptions& ioptions, const EnvOptions& env_options, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_key_comparator, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader, bool prefetch_index_and_filter = true, bool skip_filters = false); bool PrefixMayMatch(const Slice& internal_key); // Returns a new iterator over the table contents. // The result of NewIterator() is initially invalid (caller must // call one of the Seek methods on the iterator before using it). // @param skip_filters Disables loading/accessing the filter block InternalIterator* NewIterator(const ReadOptions&, Arena* arena = nullptr, bool skip_filters = false) override; // @param skip_filters Disables loading/accessing the filter block Status Get(const ReadOptions& readOptions, const Slice& key, GetContext* get_context, bool skip_filters = false) override; // Pre-fetch the disk blocks that correspond to the key range specified by // (kbegin, kend). The call will return return error status in the event of // IO or iteration error. Status Prefetch(const Slice* begin, const Slice* end) override; // Given a key, return an approximate byte offset in the file where // the data for that key begins (or would begin if the key were // present in the file). The returned value is in terms of file // bytes, and so includes effects like compression of the underlying data. // E.g., the approximate offset of the last key in the table will // be close to the file length. uint64_t ApproximateOffsetOf(const Slice& key) override; // Returns true if the block for the specified key is in cache. // REQUIRES: key is in this table && block cache enabled bool TEST_KeyInCache(const ReadOptions& options, const Slice& key); // Set up the table for Compaction. Might change some parameters with // posix_fadvise void SetupForCompaction() override; std::shared_ptr GetTableProperties() const override; size_t ApproximateMemoryUsage() const override; // convert SST file to a human readable form Status DumpTable(WritableFile* out_file) override; ~BlockBasedTable(); bool TEST_filter_block_preloaded() const; bool TEST_index_reader_preloaded() const; // Implementation of IndexReader will be exposed to internal cc file only. class IndexReader; private: template struct CachableEntry; struct Rep; Rep* rep_; bool compaction_optimized_; class BlockEntryIteratorState; // input_iter: if it is not null, update this one and return it as Iterator static InternalIterator* NewDataBlockIterator( Rep* rep, const ReadOptions& ro, const Slice& index_value, BlockIter* input_iter = nullptr); // For the following two functions: // if `no_io == true`, we will not try to read filter/index from sst file // were they not present in cache yet. CachableEntry GetFilter(bool no_io = false) const; // Get the iterator from the index reader. // If input_iter is not set, return new Iterator // If input_iter is set, update it and return it as Iterator // // Note: ErrorIterator with Status::Incomplete shall be returned if all the // following conditions are met: // 1. We enabled table_options.cache_index_and_filter_blocks. // 2. index is not present in block cache. // 3. We disallowed any io to be performed, that is, read_options == // kBlockCacheTier InternalIterator* NewIndexIterator(const ReadOptions& read_options, BlockIter* input_iter = nullptr); // Read block cache from block caches (if set): block_cache and // block_cache_compressed. // On success, Status::OK with be returned and @block will be populated with // pointer to the block as well as its block handle. static Status GetDataBlockFromCache( const Slice& block_cache_key, const Slice& compressed_block_cache_key, Cache* block_cache, Cache* block_cache_compressed, Statistics* statistics, const ReadOptions& read_options, BlockBasedTable::CachableEntry* block, uint32_t format_version); // Put a raw block (maybe compressed) to the corresponding block caches. // This method will perform decompression against raw_block if needed and then // populate the block caches. // On success, Status::OK will be returned; also @block will be populated with // uncompressed block and its cache handle. // // REQUIRES: raw_block is heap-allocated. PutDataBlockToCache() will be // responsible for releasing its memory if error occurs. static Status PutDataBlockToCache( const Slice& block_cache_key, const Slice& compressed_block_cache_key, Cache* block_cache, Cache* block_cache_compressed, const ReadOptions& read_options, Statistics* statistics, CachableEntry* block, Block* raw_block, uint32_t format_version); // Calls (*handle_result)(arg, ...) repeatedly, starting with the entry found // after a call to Seek(key), until handle_result returns false. // May not make such a call if filter policy says that key is not present. friend class TableCache; friend class BlockBasedTableBuilder; void ReadMeta(const Footer& footer); // Create a index reader based on the index type stored in the table. // Optionally, user can pass a preloaded meta_index_iter for the index that // need to access extra meta blocks for index construction. This parameter // helps avoid re-reading meta index block if caller already created one. Status CreateIndexReader( IndexReader** index_reader, InternalIterator* preloaded_meta_index_iter = nullptr); bool FullFilterKeyMayMatch(FilterBlockReader* filter, const Slice& user_key) const; // Read the meta block from sst. static Status ReadMetaBlock(Rep* rep, std::unique_ptr* meta_block, std::unique_ptr* iter); // Create the filter from the filter block. static FilterBlockReader* ReadFilter(Rep* rep, size_t* filter_size = nullptr); static void SetupCacheKeyPrefix(Rep* rep); explicit BlockBasedTable(Rep* rep) : rep_(rep), compaction_optimized_(false) {} // Generate a cache key prefix from the file static void GenerateCachePrefix(Cache* cc, RandomAccessFile* file, char* buffer, size_t* size); static void GenerateCachePrefix(Cache* cc, WritableFile* file, char* buffer, size_t* size); // The longest prefix of the cache key used to identify blocks. // For Posix files the unique ID is three varints. static const size_t kMaxCacheKeyPrefixSize = kMaxVarint64Length*3+1; // Helper functions for DumpTable() Status DumpIndexBlock(WritableFile* out_file); Status DumpDataBlocks(WritableFile* out_file); // No copying allowed explicit BlockBasedTable(const TableReader&) = delete; void operator=(const TableReader&) = delete; }; } // namespace rocksdb rocksdb-4.5.1/table/block_builder.cc000066400000000000000000000107661267530634000173660ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // BlockBuilder generates blocks where keys are prefix-compressed: // // When we store a key, we drop the prefix shared with the previous // string. This helps reduce the space requirement significantly. // Furthermore, once every K keys, we do not apply the prefix // compression and store the entire key. We call this a "restart // point". The tail end of the block stores the offsets of all of the // restart points, and can be used to do a binary search when looking // for a particular key. Values are stored as-is (without compression) // immediately following the corresponding key. // // An entry for a particular key-value pair has the form: // shared_bytes: varint32 // unshared_bytes: varint32 // value_length: varint32 // key_delta: char[unshared_bytes] // value: char[value_length] // shared_bytes == 0 for restart points. // // The trailer of the block has the form: // restarts: uint32[num_restarts] // num_restarts: uint32 // restarts[i] contains the offset within the block of the ith restart point. #include "table/block_builder.h" #include #include #include "rocksdb/comparator.h" #include "db/dbformat.h" #include "util/coding.h" namespace rocksdb { BlockBuilder::BlockBuilder(int block_restart_interval, bool use_delta_encoding) : block_restart_interval_(block_restart_interval), use_delta_encoding_(use_delta_encoding), restarts_(), counter_(0), finished_(false) { assert(block_restart_interval_ >= 1); restarts_.push_back(0); // First restart point is at offset 0 } void BlockBuilder::Reset() { buffer_.clear(); restarts_.clear(); restarts_.push_back(0); // First restart point is at offset 0 counter_ = 0; finished_ = false; last_key_.clear(); } size_t BlockBuilder::CurrentSizeEstimate() const { return (buffer_.size() + // Raw data buffer restarts_.size() * sizeof(uint32_t) + // Restart array sizeof(uint32_t)); // Restart array length } size_t BlockBuilder::EstimateSizeAfterKV(const Slice& key, const Slice& value) const { size_t estimate = CurrentSizeEstimate(); estimate += key.size() + value.size(); if (counter_ >= block_restart_interval_) { estimate += sizeof(uint32_t); // a new restart entry. } estimate += sizeof(int32_t); // varint for shared prefix length. estimate += VarintLength(key.size()); // varint for key length. estimate += VarintLength(value.size()); // varint for value length. return estimate; } Slice BlockBuilder::Finish() { // Append restart array for (size_t i = 0; i < restarts_.size(); i++) { PutFixed32(&buffer_, restarts_[i]); } PutFixed32(&buffer_, static_cast(restarts_.size())); finished_ = true; return Slice(buffer_); } void BlockBuilder::Add(const Slice& key, const Slice& value) { Slice last_key_piece(last_key_); assert(!finished_); assert(counter_ <= block_restart_interval_); size_t shared = 0; // number of bytes shared with prev key if (counter_ >= block_restart_interval_) { // Restart compression restarts_.push_back(static_cast(buffer_.size())); counter_ = 0; } else if (use_delta_encoding_) { // See how much sharing to do with previous string const size_t min_length = std::min(last_key_piece.size(), key.size()); while ((shared < min_length) && (last_key_piece[shared] == key[shared])) { shared++; } } const size_t non_shared = key.size() - shared; // Add "" to buffer_ PutVarint32(&buffer_, static_cast(shared)); PutVarint32(&buffer_, static_cast(non_shared)); PutVarint32(&buffer_, static_cast(value.size())); // Add string delta to buffer_ followed by value buffer_.append(key.data() + shared, non_shared); buffer_.append(value.data(), value.size()); // Update state last_key_.resize(shared); last_key_.append(key.data() + shared, non_shared); assert(Slice(last_key_) == key); counter_++; } } // namespace rocksdb rocksdb-4.5.1/table/block_builder.h000066400000000000000000000041541267530634000172220ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include "rocksdb/slice.h" namespace rocksdb { class BlockBuilder { public: BlockBuilder(const BlockBuilder&) = delete; void operator=(const BlockBuilder&) = delete; explicit BlockBuilder(int block_restart_interval, bool use_delta_encoding = true); // Reset the contents as if the BlockBuilder was just constructed. void Reset(); // REQUIRES: Finish() has not been callled since the last call to Reset(). // REQUIRES: key is larger than any previously added key void Add(const Slice& key, const Slice& value); // Finish building the block and return a slice that refers to the // block contents. The returned slice will remain valid for the // lifetime of this builder or until Reset() is called. Slice Finish(); // Returns an estimate of the current (uncompressed) size of the block // we are building. size_t CurrentSizeEstimate() const; // Returns an estimated block size after appending key and value. size_t EstimateSizeAfterKV(const Slice& key, const Slice& value) const; // Return true iff no entries have been added since the last Reset() bool empty() const { return buffer_.empty(); } private: const int block_restart_interval_; const bool use_delta_encoding_; std::string buffer_; // Destination buffer std::vector restarts_; // Restart points int counter_; // Number of entries emitted since restart bool finished_; // Has Finish() been called? std::string last_key_; }; } // namespace rocksdb rocksdb-4.5.1/table/block_hash_index.cc000066400000000000000000000121361267530634000200430ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/block_hash_index.h" #include #include "rocksdb/comparator.h" #include "rocksdb/iterator.h" #include "rocksdb/slice_transform.h" #include "table/internal_iterator.h" #include "util/coding.h" namespace rocksdb { Status CreateBlockHashIndex(const SliceTransform* hash_key_extractor, const Slice& prefixes, const Slice& prefix_meta, BlockHashIndex** hash_index) { uint64_t pos = 0; auto meta_pos = prefix_meta; Status s; *hash_index = new BlockHashIndex( hash_key_extractor, false /* external module manages memory space for prefixes */); while (!meta_pos.empty()) { uint32_t prefix_size = 0; uint32_t entry_index = 0; uint32_t num_blocks = 0; if (!GetVarint32(&meta_pos, &prefix_size) || !GetVarint32(&meta_pos, &entry_index) || !GetVarint32(&meta_pos, &num_blocks)) { s = Status::Corruption( "Corrupted prefix meta block: unable to read from it."); break; } Slice prefix(prefixes.data() + pos, prefix_size); (*hash_index)->Add(prefix, entry_index, num_blocks); pos += prefix_size; } if (s.ok() && pos != prefixes.size()) { s = Status::Corruption("Corrupted prefix meta block"); } if (!s.ok()) { delete *hash_index; } return s; } BlockHashIndex* CreateBlockHashIndexOnTheFly( InternalIterator* index_iter, InternalIterator* data_iter, const uint32_t num_restarts, const Comparator* comparator, const SliceTransform* hash_key_extractor) { assert(hash_key_extractor); auto hash_index = new BlockHashIndex( hash_key_extractor, true /* hash_index will copy prefix when Add() is called */); uint32_t current_restart_index = 0; std::string pending_entry_prefix; // pending_block_num == 0 also implies there is no entry inserted at all. uint32_t pending_block_num = 0; uint32_t pending_entry_index = 0; // scan all the entries and create a hash index based on their prefixes. data_iter->SeekToFirst(); for (index_iter->SeekToFirst(); index_iter->Valid() && current_restart_index < num_restarts; index_iter->Next()) { Slice last_key_in_block = index_iter->key(); assert(data_iter->Valid() && data_iter->status().ok()); // scan through all entries within a data block. while (data_iter->Valid() && comparator->Compare(data_iter->key(), last_key_in_block) <= 0) { auto key_prefix = hash_key_extractor->Transform(data_iter->key()); bool is_first_entry = pending_block_num == 0; // Keys may share the prefix if (is_first_entry || pending_entry_prefix != key_prefix) { if (!is_first_entry) { bool succeeded = hash_index->Add( pending_entry_prefix, pending_entry_index, pending_block_num); if (!succeeded) { delete hash_index; return nullptr; } } // update the status. // needs a hard copy otherwise the underlying data changes all the time. pending_entry_prefix = key_prefix.ToString(); pending_block_num = 1; pending_entry_index = current_restart_index; } else { // entry number increments when keys share the prefix reside in // different data blocks. auto last_restart_index = pending_entry_index + pending_block_num - 1; assert(last_restart_index <= current_restart_index); if (last_restart_index != current_restart_index) { ++pending_block_num; } } data_iter->Next(); } ++current_restart_index; } // make sure all entries has been scaned. assert(!index_iter->Valid()); assert(!data_iter->Valid()); if (pending_block_num > 0) { auto succeeded = hash_index->Add(pending_entry_prefix, pending_entry_index, pending_block_num); if (!succeeded) { delete hash_index; return nullptr; } } return hash_index; } bool BlockHashIndex::Add(const Slice& prefix, uint32_t restart_index, uint32_t num_blocks) { auto prefix_to_insert = prefix; if (kOwnPrefixes) { auto prefix_ptr = arena_.Allocate(prefix.size()); // MSVC reports C4996 Function call with parameters that may be // unsafe when using std::copy with a output iterator - pointer memcpy(prefix_ptr, prefix.data(), prefix.size()); prefix_to_insert = Slice(prefix_ptr, prefix.size()); } auto result = restart_indices_.insert( {prefix_to_insert, RestartIndex(restart_index, num_blocks)}); return result.second; } const BlockHashIndex::RestartIndex* BlockHashIndex::GetRestartIndex( const Slice& key) { auto key_prefix = hash_key_extractor_->Transform(key); auto pos = restart_indices_.find(key_prefix); if (pos == restart_indices_.end()) { return nullptr; } return &pos->second; } } // namespace rocksdb rocksdb-4.5.1/table/block_hash_index.h000066400000000000000000000060351267530634000177060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "rocksdb/status.h" #include "util/arena.h" #include "util/murmurhash.h" namespace rocksdb { class Comparator; class InternalIterator; class Slice; class SliceTransform; // Build a hash-based index to speed up the lookup for "index block". // BlockHashIndex accepts a key and, if found, returns its restart index within // that index block. class BlockHashIndex { public: // Represents a restart index in the index block's restart array. struct RestartIndex { explicit RestartIndex(uint32_t _first_index, uint32_t _num_blocks = 1) : first_index(_first_index), num_blocks(_num_blocks) {} // For a given prefix, what is the restart index for the first data block // that contains it. uint32_t first_index = 0; // How many data blocks contains this prefix? uint32_t num_blocks = 1; }; // @params own_prefixes indicate if we should take care the memory space for // the `key_prefix` // passed by Add() explicit BlockHashIndex(const SliceTransform* hash_key_extractor, bool own_prefixes) : hash_key_extractor_(hash_key_extractor), kOwnPrefixes(own_prefixes) {} // Maps a key to its restart first_index. // Returns nullptr if the restart first_index is found const RestartIndex* GetRestartIndex(const Slice& key); bool Add(const Slice& key_prefix, uint32_t restart_index, uint32_t num_blocks); size_t ApproximateMemoryUsage() const { return arena_.ApproximateMemoryUsage(); } private: const SliceTransform* hash_key_extractor_; std::unordered_map restart_indices_; Arena arena_; bool kOwnPrefixes; }; // Create hash index by reading from the metadata blocks. // @params prefixes: a sequence of prefixes. // @params prefix_meta: contains the "metadata" to of the prefixes. Status CreateBlockHashIndex(const SliceTransform* hash_key_extractor, const Slice& prefixes, const Slice& prefix_meta, BlockHashIndex** hash_index); // Create hash index by scanning the entries in index as well as the whole // dataset. // @params index_iter: an iterator with the pointer to the first entry in a // block. // @params data_iter: an iterator that can scan all the entries reside in a // table. // @params num_restarts: used for correctness verification. // @params hash_key_extractor: extract the hashable part of a given key. // On error, nullptr will be returned. BlockHashIndex* CreateBlockHashIndexOnTheFly( InternalIterator* index_iter, InternalIterator* data_iter, const uint32_t num_restarts, const Comparator* comparator, const SliceTransform* hash_key_extractor); } // namespace rocksdb rocksdb-4.5.1/table/block_hash_index_test.cc000066400000000000000000000075761267530634000211160ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include #include "rocksdb/comparator.h" #include "rocksdb/iterator.h" #include "rocksdb/slice_transform.h" #include "table/block_hash_index.h" #include "table/internal_iterator.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { typedef std::map Data; class MapIterator : public InternalIterator { public: explicit MapIterator(const Data& data) : data_(data), pos_(data_.end()) {} virtual bool Valid() const override { return pos_ != data_.end(); } virtual void SeekToFirst() override { pos_ = data_.begin(); } virtual void SeekToLast() override { pos_ = data_.end(); --pos_; } virtual void Seek(const Slice& target) override { pos_ = data_.find(target.ToString()); } virtual void Next() override { ++pos_; } virtual void Prev() override { --pos_; } virtual Slice key() const override { return pos_->first; } virtual Slice value() const override { return pos_->second; } virtual Status status() const override { return Status::OK(); } private: const Data& data_; Data::const_iterator pos_; }; class BlockTest : public testing::Test {}; TEST_F(BlockTest, BasicTest) { const size_t keys_per_block = 4; const size_t prefix_size = 2; std::vector keys = {/* block 1 */ "0101", "0102", "0103", "0201", /* block 2 */ "0202", "0203", "0301", "0401", /* block 3 */ "0501", "0601", "0701", "0801", /* block 4 */ "0802", "0803", "0804", "0805", /* block 5 */ "0806", "0807", "0808", "0809", }; Data data_entries; for (const auto key : keys) { data_entries.insert({key, key}); } Data index_entries; for (size_t i = 3; i < keys.size(); i += keys_per_block) { // simply ignore the value part index_entries.insert({keys[i], ""}); } MapIterator data_iter(data_entries); MapIterator index_iter(index_entries); auto prefix_extractor = NewFixedPrefixTransform(prefix_size); std::unique_ptr block_hash_index(CreateBlockHashIndexOnTheFly( &index_iter, &data_iter, static_cast(index_entries.size()), BytewiseComparator(), prefix_extractor)); std::map expected = { {"01xx", BlockHashIndex::RestartIndex(0, 1)}, {"02yy", BlockHashIndex::RestartIndex(0, 2)}, {"03zz", BlockHashIndex::RestartIndex(1, 1)}, {"04pp", BlockHashIndex::RestartIndex(1, 1)}, {"05ww", BlockHashIndex::RestartIndex(2, 1)}, {"06xx", BlockHashIndex::RestartIndex(2, 1)}, {"07pp", BlockHashIndex::RestartIndex(2, 1)}, {"08xz", BlockHashIndex::RestartIndex(2, 3)}, }; const BlockHashIndex::RestartIndex* index = nullptr; // search existed prefixes for (const auto& item : expected) { index = block_hash_index->GetRestartIndex(item.first); ASSERT_TRUE(index != nullptr); ASSERT_EQ(item.second.first_index, index->first_index); ASSERT_EQ(item.second.num_blocks, index->num_blocks); } // search non exist prefixes ASSERT_TRUE(!block_hash_index->GetRestartIndex("00xx")); ASSERT_TRUE(!block_hash_index->GetRestartIndex("10yy")); ASSERT_TRUE(!block_hash_index->GetRestartIndex("20zz")); delete prefix_extractor; } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/table/block_prefix_index.cc000066400000000000000000000173731267530634000204250ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/block_prefix_index.h" #include #include "rocksdb/comparator.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "util/arena.h" #include "util/coding.h" #include "util/hash.h" namespace rocksdb { inline uint32_t Hash(const Slice& s) { return rocksdb::Hash(s.data(), s.size(), 0); } inline uint32_t PrefixToBucket(const Slice& prefix, uint32_t num_buckets) { return Hash(prefix) % num_buckets; } // The prefix block index is simply a bucket array, with each entry pointing to // the blocks that span the prefixes hashed to this bucket. // // To reduce memory footprint, if there is only one block per bucket, the entry // stores the block id directly. If there are more than one blocks per bucket, // because of hash collision or a single prefix spanning multiple blocks, // the entry points to an array of block ids. The block array is an array of // uint32_t's. The first uint32_t indicates the total number of blocks, followed // by the block ids. // // To differentiate the two cases, the high order bit of the entry indicates // whether it is a 'pointer' into a separate block array. // 0x7FFFFFFF is reserved for empty bucket. const uint32_t kNoneBlock = 0x7FFFFFFF; const uint32_t kBlockArrayMask = 0x80000000; inline bool IsNone(uint32_t block_id) { return block_id == kNoneBlock; } inline bool IsBlockId(uint32_t block_id) { return (block_id & kBlockArrayMask) == 0; } inline uint32_t DecodeIndex(uint32_t block_id) { uint32_t index = block_id ^ kBlockArrayMask; assert(index < kBlockArrayMask); return index; } inline uint32_t EncodeIndex(uint32_t index) { assert(index < kBlockArrayMask); return index | kBlockArrayMask; } // temporary storage for prefix information during index building struct PrefixRecord { Slice prefix; uint32_t start_block; uint32_t end_block; uint32_t num_blocks; PrefixRecord* next; }; class BlockPrefixIndex::Builder { public: explicit Builder(const SliceTransform* internal_prefix_extractor) : internal_prefix_extractor_(internal_prefix_extractor) {} void Add(const Slice& key_prefix, uint32_t start_block, uint32_t num_blocks) { PrefixRecord* record = reinterpret_cast( arena_.AllocateAligned(sizeof(PrefixRecord))); record->prefix = key_prefix; record->start_block = start_block; record->end_block = start_block + num_blocks - 1; record->num_blocks = num_blocks; prefixes_.push_back(record); } BlockPrefixIndex* Finish() { // For now, use roughly 1:1 prefix to bucket ratio. uint32_t num_buckets = static_cast(prefixes_.size()) + 1; // Collect prefix records that hash to the same bucket, into a single // linklist. std::vector prefixes_per_bucket(num_buckets, nullptr); std::vector num_blocks_per_bucket(num_buckets, 0); for (PrefixRecord* current : prefixes_) { uint32_t bucket = PrefixToBucket(current->prefix, num_buckets); // merge the prefix block span if the first block of this prefix is // connected to the last block of the previous prefix. PrefixRecord* prev = prefixes_per_bucket[bucket]; if (prev) { assert(current->start_block >= prev->end_block); auto distance = current->start_block - prev->end_block; if (distance <= 1) { prev->end_block = current->end_block; prev->num_blocks = prev->end_block - prev->start_block + 1; num_blocks_per_bucket[bucket] += (current->num_blocks + distance - 1); continue; } } current->next = prev; prefixes_per_bucket[bucket] = current; num_blocks_per_bucket[bucket] += current->num_blocks; } // Calculate the block array buffer size uint32_t total_block_array_entries = 0; for (uint32_t i = 0; i < num_buckets; i++) { uint32_t num_blocks = num_blocks_per_bucket[i]; if (num_blocks > 1) { total_block_array_entries += (num_blocks + 1); } } // Populate the final prefix block index uint32_t* block_array_buffer = new uint32_t[total_block_array_entries]; uint32_t* buckets = new uint32_t[num_buckets]; uint32_t offset = 0; for (uint32_t i = 0; i < num_buckets; i++) { uint32_t num_blocks = num_blocks_per_bucket[i]; if (num_blocks == 0) { assert(prefixes_per_bucket[i] == nullptr); buckets[i] = kNoneBlock; } else if (num_blocks == 1) { assert(prefixes_per_bucket[i] != nullptr); assert(prefixes_per_bucket[i]->next == nullptr); buckets[i] = prefixes_per_bucket[i]->start_block; } else { assert(prefixes_per_bucket[i] != nullptr); buckets[i] = EncodeIndex(offset); block_array_buffer[offset] = num_blocks; uint32_t* last_block = &block_array_buffer[offset + num_blocks]; auto current = prefixes_per_bucket[i]; // populate block ids from largest to smallest while (current != nullptr) { for (uint32_t iter = 0; iter < current->num_blocks; iter++) { *last_block = current->end_block - iter; last_block--; } current = current->next; } assert(last_block == &block_array_buffer[offset]); offset += (num_blocks + 1); } } assert(offset == total_block_array_entries); return new BlockPrefixIndex(internal_prefix_extractor_, num_buckets, buckets, total_block_array_entries, block_array_buffer); } private: const SliceTransform* internal_prefix_extractor_; std::vector prefixes_; Arena arena_; }; Status BlockPrefixIndex::Create(const SliceTransform* internal_prefix_extractor, const Slice& prefixes, const Slice& prefix_meta, BlockPrefixIndex** prefix_index) { uint64_t pos = 0; auto meta_pos = prefix_meta; Status s; Builder builder(internal_prefix_extractor); while (!meta_pos.empty()) { uint32_t prefix_size = 0; uint32_t entry_index = 0; uint32_t num_blocks = 0; if (!GetVarint32(&meta_pos, &prefix_size) || !GetVarint32(&meta_pos, &entry_index) || !GetVarint32(&meta_pos, &num_blocks)) { s = Status::Corruption( "Corrupted prefix meta block: unable to read from it."); break; } if (pos + prefix_size > prefixes.size()) { s = Status::Corruption( "Corrupted prefix meta block: size inconsistency."); break; } Slice prefix(prefixes.data() + pos, prefix_size); builder.Add(prefix, entry_index, num_blocks); pos += prefix_size; } if (s.ok() && pos != prefixes.size()) { s = Status::Corruption("Corrupted prefix meta block"); } if (s.ok()) { *prefix_index = builder.Finish(); } return s; } uint32_t BlockPrefixIndex::GetBlocks(const Slice& key, uint32_t** blocks) { Slice prefix = internal_prefix_extractor_->Transform(key); uint32_t bucket = PrefixToBucket(prefix, num_buckets_); uint32_t block_id = buckets_[bucket]; if (IsNone(block_id)) { return 0; } else if (IsBlockId(block_id)) { *blocks = &buckets_[bucket]; return 1; } else { uint32_t index = DecodeIndex(block_id); assert(index < num_block_array_buffer_entries_); *blocks = &block_array_buffer_[index+1]; uint32_t num_blocks = block_array_buffer_[index]; assert(num_blocks > 1); assert(index + num_blocks < num_block_array_buffer_entries_); return num_blocks; } } } // namespace rocksdb rocksdb-4.5.1/table/block_prefix_index.h000066400000000000000000000043431267530634000202600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include "rocksdb/status.h" namespace rocksdb { class Comparator; class Iterator; class Slice; class SliceTransform; // Build a hash-based index to speed up the lookup for "index block". // BlockHashIndex accepts a key and, if found, returns its restart index within // that index block. class BlockPrefixIndex { public: // Maps a key to a list of data blocks that could potentially contain // the key, based on the prefix. // Returns the total number of relevant blocks, 0 means the key does // not exist. uint32_t GetBlocks(const Slice& key, uint32_t** blocks); size_t ApproximateMemoryUsage() const { return sizeof(BlockPrefixIndex) + (num_block_array_buffer_entries_ + num_buckets_) * sizeof(uint32_t); } // Create hash index by reading from the metadata blocks. // @params prefixes: a sequence of prefixes. // @params prefix_meta: contains the "metadata" to of the prefixes. static Status Create(const SliceTransform* hash_key_extractor, const Slice& prefixes, const Slice& prefix_meta, BlockPrefixIndex** prefix_index); ~BlockPrefixIndex() { delete[] buckets_; delete[] block_array_buffer_; } private: class Builder; friend Builder; BlockPrefixIndex(const SliceTransform* internal_prefix_extractor, uint32_t num_buckets, uint32_t* buckets, uint32_t num_block_array_buffer_entries, uint32_t* block_array_buffer) : internal_prefix_extractor_(internal_prefix_extractor), num_buckets_(num_buckets), num_block_array_buffer_entries_(num_block_array_buffer_entries), buckets_(buckets), block_array_buffer_(block_array_buffer) {} const SliceTransform* internal_prefix_extractor_; uint32_t num_buckets_; uint32_t num_block_array_buffer_entries_; uint32_t* buckets_; uint32_t* block_array_buffer_; }; } // namespace rocksdb rocksdb-4.5.1/table/block_test.cc000066400000000000000000000166341267530634000167170ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #include #include #include #include "db/dbformat.h" #include "db/memtable.h" #include "db/write_batch_internal.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "rocksdb/table.h" #include "rocksdb/slice_transform.h" #include "table/block.h" #include "table/block_builder.h" #include "table/format.h" #include "table/block_hash_index.h" #include "util/random.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { static std::string RandomString(Random* rnd, int len) { std::string r; test::RandomString(rnd, len, &r); return r; } std::string GenerateKey(int primary_key, int secondary_key, int padding_size, Random *rnd) { char buf[50]; char *p = &buf[0]; snprintf(buf, sizeof(buf), "%6d%4d", primary_key, secondary_key); std::string k(p); if (padding_size) { k += RandomString(rnd, padding_size); } return k; } // Generate random key value pairs. // The generated key will be sorted. You can tune the parameters to generated // different kinds of test key/value pairs for different scenario. void GenerateRandomKVs(std::vector *keys, std::vector *values, const int from, const int len, const int step = 1, const int padding_size = 0, const int keys_share_prefix = 1) { Random rnd(302); // generate different prefix for (int i = from; i < from + len; i += step) { // generating keys that shares the prefix for (int j = 0; j < keys_share_prefix; ++j) { keys->emplace_back(GenerateKey(i, j, padding_size, &rnd)); // 100 bytes values values->emplace_back(RandomString(&rnd, 100)); } } } class BlockTest : public testing::Test {}; // block test TEST_F(BlockTest, SimpleTest) { Random rnd(301); Options options = Options(); std::unique_ptr ic; ic.reset(new test::PlainInternalKeyComparator(options.comparator)); std::vector keys; std::vector values; BlockBuilder builder(16); int num_records = 100000; GenerateRandomKVs(&keys, &values, 0, num_records); // add a bunch of records to a block for (int i = 0; i < num_records; i++) { builder.Add(keys[i], values[i]); } // read serialized contents of the block Slice rawblock = builder.Finish(); // create block reader BlockContents contents; contents.data = rawblock; contents.cachable = false; Block reader(std::move(contents)); // read contents of block sequentially int count = 0; InternalIterator *iter = reader.NewIterator(options.comparator); for (iter->SeekToFirst();iter->Valid(); count++, iter->Next()) { // read kv from block Slice k = iter->key(); Slice v = iter->value(); // compare with lookaside array ASSERT_EQ(k.ToString().compare(keys[count]), 0); ASSERT_EQ(v.ToString().compare(values[count]), 0); } delete iter; // read block contents randomly iter = reader.NewIterator(options.comparator); for (int i = 0; i < num_records; i++) { // find a random key in the lookaside array int index = rnd.Uniform(num_records); Slice k(keys[index]); // search in block for this key iter->Seek(k); ASSERT_TRUE(iter->Valid()); Slice v = iter->value(); ASSERT_EQ(v.ToString().compare(values[index]), 0); } delete iter; } // return the block contents BlockContents GetBlockContents(std::unique_ptr *builder, const std::vector &keys, const std::vector &values, const int prefix_group_size = 1) { builder->reset(new BlockBuilder(1 /* restart interval */)); // Add only half of the keys for (size_t i = 0; i < keys.size(); ++i) { (*builder)->Add(keys[i], values[i]); } Slice rawblock = (*builder)->Finish(); BlockContents contents; contents.data = rawblock; contents.cachable = false; return contents; } void CheckBlockContents(BlockContents contents, const int max_key, const std::vector &keys, const std::vector &values) { const size_t prefix_size = 6; // create block reader BlockContents contents_ref(contents.data, contents.cachable, contents.compression_type); Block reader1(std::move(contents)); Block reader2(std::move(contents_ref)); std::unique_ptr prefix_extractor( NewFixedPrefixTransform(prefix_size)); { auto iter1 = reader1.NewIterator(nullptr); auto iter2 = reader1.NewIterator(nullptr); reader1.SetBlockHashIndex(CreateBlockHashIndexOnTheFly( iter1, iter2, static_cast(keys.size()), BytewiseComparator(), prefix_extractor.get())); delete iter1; delete iter2; } std::unique_ptr hash_iter( reader1.NewIterator(BytewiseComparator(), nullptr, false)); std::unique_ptr regular_iter( reader2.NewIterator(BytewiseComparator())); // Seek existent keys for (size_t i = 0; i < keys.size(); i++) { hash_iter->Seek(keys[i]); ASSERT_OK(hash_iter->status()); ASSERT_TRUE(hash_iter->Valid()); Slice v = hash_iter->value(); ASSERT_EQ(v.ToString().compare(values[i]), 0); } // Seek non-existent keys. // For hash index, if no key with a given prefix is not found, iterator will // simply be set as invalid; whereas the binary search based iterator will // return the one that is closest. for (int i = 1; i < max_key - 1; i += 2) { auto key = GenerateKey(i, 0, 0, nullptr); hash_iter->Seek(key); ASSERT_TRUE(!hash_iter->Valid()); regular_iter->Seek(key); ASSERT_TRUE(regular_iter->Valid()); } } // In this test case, no two key share same prefix. TEST_F(BlockTest, SimpleIndexHash) { const int kMaxKey = 100000; std::vector keys; std::vector values; GenerateRandomKVs(&keys, &values, 0 /* first key id */, kMaxKey /* last key id */, 2 /* step */, 8 /* padding size (8 bytes randomly generated suffix) */); std::unique_ptr builder; auto contents = GetBlockContents(&builder, keys, values); CheckBlockContents(std::move(contents), kMaxKey, keys, values); } TEST_F(BlockTest, IndexHashWithSharedPrefix) { const int kMaxKey = 100000; // for each prefix, there will be 5 keys starts with it. const int kPrefixGroup = 5; std::vector keys; std::vector values; // Generate keys with same prefix. GenerateRandomKVs(&keys, &values, 0, // first key id kMaxKey, // last key id 2, // step 10, // padding size, kPrefixGroup); std::unique_ptr builder; auto contents = GetBlockContents(&builder, keys, values, kPrefixGroup); CheckBlockContents(std::move(contents), kMaxKey, keys, values); } } // namespace rocksdb int main(int argc, char **argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/table/bloom_block.cc000066400000000000000000000013321267530634000170350ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/bloom_block.h" #include #include "rocksdb/slice.h" #include "util/dynamic_bloom.h" namespace rocksdb { void BloomBlockBuilder::AddKeysHashes(const std::vector& keys_hashes) { for (auto hash : keys_hashes) { bloom_.AddHash(hash); } } Slice BloomBlockBuilder::Finish() { return bloom_.GetRawData(); } const std::string BloomBlockBuilder::kBloomBlock = "kBloomBlock"; } // namespace rocksdb rocksdb-4.5.1/table/bloom_block.h000066400000000000000000000020631267530634000167010ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include "util/dynamic_bloom.h" namespace rocksdb { class Logger; class BloomBlockBuilder { public: static const std::string kBloomBlock; explicit BloomBlockBuilder(uint32_t num_probes = 6) : bloom_(num_probes, nullptr) {} void SetTotalBits(Allocator* allocator, uint32_t total_bits, uint32_t locality, size_t huge_page_tlb_size, Logger* logger) { bloom_.SetTotalBits(allocator, total_bits, locality, huge_page_tlb_size, logger); } uint32_t GetNumBlocks() const { return bloom_.GetNumBlocks(); } void AddKeysHashes(const std::vector& keys_hashes); Slice Finish(); private: DynamicBloom bloom_; }; }; // namespace rocksdb rocksdb-4.5.1/table/cuckoo_table_builder.cc000066400000000000000000000460201267530634000207160ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "table/cuckoo_table_builder.h" #include #include #include #include #include #include "db/dbformat.h" #include "rocksdb/env.h" #include "rocksdb/table.h" #include "table/block_builder.h" #include "table/cuckoo_table_factory.h" #include "table/format.h" #include "table/meta_blocks.h" #include "util/autovector.h" #include "util/file_reader_writer.h" #include "util/random.h" #include "util/string_util.h" namespace rocksdb { const std::string CuckooTablePropertyNames::kEmptyKey = "rocksdb.cuckoo.bucket.empty.key"; const std::string CuckooTablePropertyNames::kNumHashFunc = "rocksdb.cuckoo.hash.num"; const std::string CuckooTablePropertyNames::kHashTableSize = "rocksdb.cuckoo.hash.size"; const std::string CuckooTablePropertyNames::kValueLength = "rocksdb.cuckoo.value.length"; const std::string CuckooTablePropertyNames::kIsLastLevel = "rocksdb.cuckoo.file.islastlevel"; const std::string CuckooTablePropertyNames::kCuckooBlockSize = "rocksdb.cuckoo.hash.cuckooblocksize"; const std::string CuckooTablePropertyNames::kIdentityAsFirstHash = "rocksdb.cuckoo.hash.identityfirst"; const std::string CuckooTablePropertyNames::kUseModuleHash = "rocksdb.cuckoo.hash.usemodule"; const std::string CuckooTablePropertyNames::kUserKeyLength = "rocksdb.cuckoo.hash.userkeylength"; // Obtained by running echo rocksdb.table.cuckoo | sha1sum extern const uint64_t kCuckooTableMagicNumber = 0x926789d0c5f17873ull; CuckooTableBuilder::CuckooTableBuilder( WritableFileWriter* file, double max_hash_table_ratio, uint32_t max_num_hash_table, uint32_t max_search_depth, const Comparator* user_comparator, uint32_t cuckoo_block_size, bool use_module_hash, bool identity_as_first_hash, uint64_t (*get_slice_hash)(const Slice&, uint32_t, uint64_t)) : num_hash_func_(2), file_(file), max_hash_table_ratio_(max_hash_table_ratio), max_num_hash_func_(max_num_hash_table), max_search_depth_(max_search_depth), cuckoo_block_size_(std::max(1U, cuckoo_block_size)), hash_table_size_(use_module_hash ? 0 : 2), is_last_level_file_(false), has_seen_first_key_(false), has_seen_first_value_(false), key_size_(0), value_size_(0), num_entries_(0), num_values_(0), ucomp_(user_comparator), use_module_hash_(use_module_hash), identity_as_first_hash_(identity_as_first_hash), get_slice_hash_(get_slice_hash), closed_(false) { // Data is in a huge block. properties_.num_data_blocks = 1; properties_.index_size = 0; properties_.filter_size = 0; } void CuckooTableBuilder::Add(const Slice& key, const Slice& value) { if (num_entries_ >= kMaxVectorIdx - 1) { status_ = Status::NotSupported("Number of keys in a file must be < 2^32-1"); return; } ParsedInternalKey ikey; if (!ParseInternalKey(key, &ikey)) { status_ = Status::Corruption("Unable to parse key into inernal key."); return; } if (ikey.type != kTypeDeletion && ikey.type != kTypeValue) { status_ = Status::NotSupported("Unsupported key type " + ToString(ikey.type)); return; } // Determine if we can ignore the sequence number and value type from // internal keys by looking at sequence number from first key. We assume // that if first key has a zero sequence number, then all the remaining // keys will have zero seq. no. if (!has_seen_first_key_) { is_last_level_file_ = ikey.sequence == 0; has_seen_first_key_ = true; smallest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size()); largest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size()); key_size_ = is_last_level_file_ ? ikey.user_key.size() : key.size(); } if (key_size_ != (is_last_level_file_ ? ikey.user_key.size() : key.size())) { status_ = Status::NotSupported("all keys have to be the same size"); return; } if (ikey.type == kTypeValue) { if (!has_seen_first_value_) { has_seen_first_value_ = true; value_size_ = value.size(); } if (value_size_ != value.size()) { status_ = Status::NotSupported("all values have to be the same size"); return; } if (is_last_level_file_) { kvs_.append(ikey.user_key.data(), ikey.user_key.size()); } else { kvs_.append(key.data(), key.size()); } kvs_.append(value.data(), value.size()); ++num_values_; } else { if (is_last_level_file_) { deleted_keys_.append(ikey.user_key.data(), ikey.user_key.size()); } else { deleted_keys_.append(key.data(), key.size()); } } ++num_entries_; // In order to fill the empty buckets in the hash table, we identify a // key which is not used so far (unused_user_key). We determine this by // maintaining smallest and largest keys inserted so far in bytewise order // and use them to find a key outside this range in Finish() operation. // Note that this strategy is independent of user comparator used here. if (ikey.user_key.compare(smallest_user_key_) < 0) { smallest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size()); } else if (ikey.user_key.compare(largest_user_key_) > 0) { largest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size()); } if (!use_module_hash_) { if (hash_table_size_ < num_entries_ / max_hash_table_ratio_) { hash_table_size_ *= 2; } } } bool CuckooTableBuilder::IsDeletedKey(uint64_t idx) const { assert(closed_); return idx >= num_values_; } Slice CuckooTableBuilder::GetKey(uint64_t idx) const { assert(closed_); if (IsDeletedKey(idx)) { return Slice(&deleted_keys_[(idx - num_values_) * key_size_], key_size_); } return Slice(&kvs_[idx * (key_size_ + value_size_)], key_size_); } Slice CuckooTableBuilder::GetUserKey(uint64_t idx) const { assert(closed_); return is_last_level_file_ ? GetKey(idx) : ExtractUserKey(GetKey(idx)); } Slice CuckooTableBuilder::GetValue(uint64_t idx) const { assert(closed_); if (IsDeletedKey(idx)) { static std::string empty_value(value_size_, 'a'); return Slice(empty_value); } return Slice(&kvs_[idx * (key_size_ + value_size_) + key_size_], value_size_); } Status CuckooTableBuilder::MakeHashTable(std::vector* buckets) { buckets->resize(hash_table_size_ + cuckoo_block_size_ - 1); uint32_t make_space_for_key_call_id = 0; for (uint32_t vector_idx = 0; vector_idx < num_entries_; vector_idx++) { uint64_t bucket_id; bool bucket_found = false; autovector hash_vals; Slice user_key = GetUserKey(vector_idx); for (uint32_t hash_cnt = 0; hash_cnt < num_hash_func_ && !bucket_found; ++hash_cnt) { uint64_t hash_val = CuckooHash(user_key, hash_cnt, use_module_hash_, hash_table_size_, identity_as_first_hash_, get_slice_hash_); // If there is a collision, check next cuckoo_block_size_ locations for // empty locations. While checking, if we reach end of the hash table, // stop searching and proceed for next hash function. for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_; ++block_idx, ++hash_val) { if ((*buckets)[hash_val].vector_idx == kMaxVectorIdx) { bucket_id = hash_val; bucket_found = true; break; } else { if (ucomp_->Compare(user_key, GetUserKey((*buckets)[hash_val].vector_idx)) == 0) { return Status::NotSupported("Same key is being inserted again."); } hash_vals.push_back(hash_val); } } } while (!bucket_found && !MakeSpaceForKey(hash_vals, ++make_space_for_key_call_id, buckets, &bucket_id)) { // Rehash by increashing number of hash tables. if (num_hash_func_ >= max_num_hash_func_) { return Status::NotSupported("Too many collisions. Unable to hash."); } // We don't really need to rehash the entire table because old hashes are // still valid and we only increased the number of hash functions. uint64_t hash_val = CuckooHash(user_key, num_hash_func_, use_module_hash_, hash_table_size_, identity_as_first_hash_, get_slice_hash_); ++num_hash_func_; for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_; ++block_idx, ++hash_val) { if ((*buckets)[hash_val].vector_idx == kMaxVectorIdx) { bucket_found = true; bucket_id = hash_val; break; } else { hash_vals.push_back(hash_val); } } } (*buckets)[bucket_id].vector_idx = vector_idx; } return Status::OK(); } Status CuckooTableBuilder::Finish() { assert(!closed_); closed_ = true; std::vector buckets; Status s; std::string unused_bucket; if (num_entries_ > 0) { // Calculate the real hash size if module hash is enabled. if (use_module_hash_) { hash_table_size_ = static_cast(num_entries_ / max_hash_table_ratio_); } s = MakeHashTable(&buckets); if (!s.ok()) { return s; } // Determine unused_user_key to fill empty buckets. std::string unused_user_key = smallest_user_key_; int curr_pos = static_cast(unused_user_key.size()) - 1; while (curr_pos >= 0) { --unused_user_key[curr_pos]; if (Slice(unused_user_key).compare(smallest_user_key_) < 0) { break; } --curr_pos; } if (curr_pos < 0) { // Try using the largest key to identify an unused key. unused_user_key = largest_user_key_; curr_pos = static_cast(unused_user_key.size()) - 1; while (curr_pos >= 0) { ++unused_user_key[curr_pos]; if (Slice(unused_user_key).compare(largest_user_key_) > 0) { break; } --curr_pos; } } if (curr_pos < 0) { return Status::Corruption("Unable to find unused key"); } if (is_last_level_file_) { unused_bucket = unused_user_key; } else { ParsedInternalKey ikey(unused_user_key, 0, kTypeValue); AppendInternalKey(&unused_bucket, ikey); } } properties_.num_entries = num_entries_; properties_.fixed_key_len = key_size_; properties_.user_collected_properties[ CuckooTablePropertyNames::kValueLength].assign( reinterpret_cast(&value_size_), sizeof(value_size_)); uint64_t bucket_size = key_size_ + value_size_; unused_bucket.resize(bucket_size, 'a'); // Write the table. uint32_t num_added = 0; for (auto& bucket : buckets) { if (bucket.vector_idx == kMaxVectorIdx) { s = file_->Append(Slice(unused_bucket)); } else { ++num_added; s = file_->Append(GetKey(bucket.vector_idx)); if (s.ok()) { if (value_size_ > 0) { s = file_->Append(GetValue(bucket.vector_idx)); } } } if (!s.ok()) { return s; } } assert(num_added == NumEntries()); properties_.raw_key_size = num_added * properties_.fixed_key_len; properties_.raw_value_size = num_added * value_size_; uint64_t offset = buckets.size() * bucket_size; properties_.data_size = offset; unused_bucket.resize(properties_.fixed_key_len); properties_.user_collected_properties[ CuckooTablePropertyNames::kEmptyKey] = unused_bucket; properties_.user_collected_properties[ CuckooTablePropertyNames::kNumHashFunc].assign( reinterpret_cast(&num_hash_func_), sizeof(num_hash_func_)); properties_.user_collected_properties[ CuckooTablePropertyNames::kHashTableSize].assign( reinterpret_cast(&hash_table_size_), sizeof(hash_table_size_)); properties_.user_collected_properties[ CuckooTablePropertyNames::kIsLastLevel].assign( reinterpret_cast(&is_last_level_file_), sizeof(is_last_level_file_)); properties_.user_collected_properties[ CuckooTablePropertyNames::kCuckooBlockSize].assign( reinterpret_cast(&cuckoo_block_size_), sizeof(cuckoo_block_size_)); properties_.user_collected_properties[ CuckooTablePropertyNames::kIdentityAsFirstHash].assign( reinterpret_cast(&identity_as_first_hash_), sizeof(identity_as_first_hash_)); properties_.user_collected_properties[ CuckooTablePropertyNames::kUseModuleHash].assign( reinterpret_cast(&use_module_hash_), sizeof(use_module_hash_)); uint32_t user_key_len = static_cast(smallest_user_key_.size()); properties_.user_collected_properties[ CuckooTablePropertyNames::kUserKeyLength].assign( reinterpret_cast(&user_key_len), sizeof(user_key_len)); // Write meta blocks. MetaIndexBuilder meta_index_builder; PropertyBlockBuilder property_block_builder; property_block_builder.AddTableProperty(properties_); property_block_builder.Add(properties_.user_collected_properties); Slice property_block = property_block_builder.Finish(); BlockHandle property_block_handle; property_block_handle.set_offset(offset); property_block_handle.set_size(property_block.size()); s = file_->Append(property_block); offset += property_block.size(); if (!s.ok()) { return s; } meta_index_builder.Add(kPropertiesBlock, property_block_handle); Slice meta_index_block = meta_index_builder.Finish(); BlockHandle meta_index_block_handle; meta_index_block_handle.set_offset(offset); meta_index_block_handle.set_size(meta_index_block.size()); s = file_->Append(meta_index_block); if (!s.ok()) { return s; } Footer footer(kCuckooTableMagicNumber, 1); footer.set_metaindex_handle(meta_index_block_handle); footer.set_index_handle(BlockHandle::NullBlockHandle()); std::string footer_encoding; footer.EncodeTo(&footer_encoding); s = file_->Append(footer_encoding); return s; } void CuckooTableBuilder::Abandon() { assert(!closed_); closed_ = true; } uint64_t CuckooTableBuilder::NumEntries() const { return num_entries_; } uint64_t CuckooTableBuilder::FileSize() const { if (closed_) { return file_->GetFileSize(); } else if (num_entries_ == 0) { return 0; } if (use_module_hash_) { return static_cast((key_size_ + value_size_) * num_entries_ / max_hash_table_ratio_); } else { // Account for buckets being a power of two. // As elements are added, file size remains constant for a while and // doubles its size. Since compaction algorithm stops adding elements // only after it exceeds the file limit, we account for the extra element // being added here. uint64_t expected_hash_table_size = hash_table_size_; if (expected_hash_table_size < (num_entries_ + 1) / max_hash_table_ratio_) { expected_hash_table_size *= 2; } return (key_size_ + value_size_) * expected_hash_table_size - 1; } } // This method is invoked when there is no place to insert the target key. // It searches for a set of elements that can be moved to accommodate target // key. The search is a BFS graph traversal with first level (hash_vals) // being all the buckets target key could go to. // Then, from each node (curr_node), we find all the buckets that curr_node // could go to. They form the children of curr_node in the tree. // We continue the traversal until we find an empty bucket, in which case, we // move all elements along the path from first level to this empty bucket, to // make space for target key which is inserted at first level (*bucket_id). // If tree depth exceedes max depth, we return false indicating failure. bool CuckooTableBuilder::MakeSpaceForKey( const autovector& hash_vals, const uint32_t make_space_for_key_call_id, std::vector* buckets, uint64_t* bucket_id) { struct CuckooNode { uint64_t bucket_id; uint32_t depth; uint32_t parent_pos; CuckooNode(uint64_t _bucket_id, uint32_t _depth, int _parent_pos) : bucket_id(_bucket_id), depth(_depth), parent_pos(_parent_pos) {} }; // This is BFS search tree that is stored simply as a vector. // Each node stores the index of parent node in the vector. std::vector tree; // We want to identify already visited buckets in the current method call so // that we don't add same buckets again for exploration in the tree. // We do this by maintaining a count of current method call in // make_space_for_key_call_id, which acts as a unique id for this invocation // of the method. We store this number into the nodes that we explore in // current method call. // It is unlikely for the increment operation to overflow because the maximum // no. of times this will be called is <= max_num_hash_func_ + num_entries_. for (uint32_t hash_cnt = 0; hash_cnt < num_hash_func_; ++hash_cnt) { uint64_t bid = hash_vals[hash_cnt]; (*buckets)[bid].make_space_for_key_call_id = make_space_for_key_call_id; tree.push_back(CuckooNode(bid, 0, 0)); } bool null_found = false; uint32_t curr_pos = 0; while (!null_found && curr_pos < tree.size()) { CuckooNode& curr_node = tree[curr_pos]; uint32_t curr_depth = curr_node.depth; if (curr_depth >= max_search_depth_) { break; } CuckooBucket& curr_bucket = (*buckets)[curr_node.bucket_id]; for (uint32_t hash_cnt = 0; hash_cnt < num_hash_func_ && !null_found; ++hash_cnt) { uint64_t child_bucket_id = CuckooHash(GetUserKey(curr_bucket.vector_idx), hash_cnt, use_module_hash_, hash_table_size_, identity_as_first_hash_, get_slice_hash_); // Iterate inside Cuckoo Block. for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_; ++block_idx, ++child_bucket_id) { if ((*buckets)[child_bucket_id].make_space_for_key_call_id == make_space_for_key_call_id) { continue; } (*buckets)[child_bucket_id].make_space_for_key_call_id = make_space_for_key_call_id; tree.push_back(CuckooNode(child_bucket_id, curr_depth + 1, curr_pos)); if ((*buckets)[child_bucket_id].vector_idx == kMaxVectorIdx) { null_found = true; break; } } } ++curr_pos; } if (null_found) { // There is an empty node in tree.back(). Now, traverse the path from this // empty node to top of the tree and at every node in the path, replace // child with the parent. Stop when first level is reached in the tree // (happens when 0 <= bucket_to_replace_pos < num_hash_func_) and return // this location in first level for target key to be inserted. uint32_t bucket_to_replace_pos = static_cast(tree.size()) - 1; while (bucket_to_replace_pos >= num_hash_func_) { CuckooNode& curr_node = tree[bucket_to_replace_pos]; (*buckets)[curr_node.bucket_id] = (*buckets)[tree[curr_node.parent_pos].bucket_id]; bucket_to_replace_pos = curr_node.parent_pos; } *bucket_id = tree[bucket_to_replace_pos].bucket_id; } return null_found; } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/cuckoo_table_builder.h000066400000000000000000000107471267530634000205670ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include #include #include "port/port.h" #include "rocksdb/status.h" #include "table/table_builder.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "util/autovector.h" namespace rocksdb { class CuckooTableBuilder: public TableBuilder { public: CuckooTableBuilder(WritableFileWriter* file, double max_hash_table_ratio, uint32_t max_num_hash_func, uint32_t max_search_depth, const Comparator* user_comparator, uint32_t cuckoo_block_size, bool use_module_hash, bool identity_as_first_hash, uint64_t (*get_slice_hash)(const Slice&, uint32_t, uint64_t)); // REQUIRES: Either Finish() or Abandon() has been called. ~CuckooTableBuilder() {} // Add key,value to the table being constructed. // REQUIRES: key is after any previously added key according to comparator. // REQUIRES: Finish(), Abandon() have not been called void Add(const Slice& key, const Slice& value) override; // Return non-ok iff some error has been detected. Status status() const override { return status_; } // Finish building the table. Stops using the file passed to the // constructor after this function returns. // REQUIRES: Finish(), Abandon() have not been called Status Finish() override; // Indicate that the contents of this builder should be abandoned. Stops // using the file passed to the constructor after this function returns. // If the caller is not going to call Finish(), it must call Abandon() // before destroying this builder. // REQUIRES: Finish(), Abandon() have not been called void Abandon() override; // Number of calls to Add() so far. uint64_t NumEntries() const override; // Size of the file generated so far. If invoked after a successful // Finish() call, returns the size of the final generated file. uint64_t FileSize() const override; TableProperties GetTableProperties() const override { return properties_; } private: struct CuckooBucket { CuckooBucket() : vector_idx(kMaxVectorIdx), make_space_for_key_call_id(0) {} uint32_t vector_idx; // This number will not exceed kvs_.size() + max_num_hash_func_. // We assume number of items is <= 2^32. uint32_t make_space_for_key_call_id; }; static const uint32_t kMaxVectorIdx = port::kMaxInt32; bool MakeSpaceForKey(const autovector& hash_vals, const uint32_t call_id, std::vector* buckets, uint64_t* bucket_id); Status MakeHashTable(std::vector* buckets); inline bool IsDeletedKey(uint64_t idx) const; inline Slice GetKey(uint64_t idx) const; inline Slice GetUserKey(uint64_t idx) const; inline Slice GetValue(uint64_t idx) const; uint32_t num_hash_func_; WritableFileWriter* file_; const double max_hash_table_ratio_; const uint32_t max_num_hash_func_; const uint32_t max_search_depth_; const uint32_t cuckoo_block_size_; uint64_t hash_table_size_; bool is_last_level_file_; bool has_seen_first_key_; bool has_seen_first_value_; uint64_t key_size_; uint64_t value_size_; // A list of fixed-size key-value pairs concatenating into a string. // Use GetKey(), GetUserKey(), and GetValue() to retrieve a specific // key / value given an index std::string kvs_; std::string deleted_keys_; // Number of key-value pairs stored in kvs_ + number of deleted keys uint64_t num_entries_; // Number of keys that contain value (non-deletion op) uint64_t num_values_; Status status_; TableProperties properties_; const Comparator* ucomp_; bool use_module_hash_; bool identity_as_first_hash_; uint64_t (*get_slice_hash_)(const Slice& s, uint32_t index, uint64_t max_num_buckets); std::string largest_user_key_ = ""; std::string smallest_user_key_ = ""; bool closed_; // Either Finish() or Abandon() has been called. // No copying allowed CuckooTableBuilder(const CuckooTableBuilder&) = delete; void operator=(const CuckooTableBuilder&) = delete; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/cuckoo_table_builder_test.cc000066400000000000000000000624201267530634000217570ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include #include #include #include #include "table/meta_blocks.h" #include "table/cuckoo_table_builder.h" #include "util/file_reader_writer.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { extern const uint64_t kCuckooTableMagicNumber; namespace { std::unordered_map> hash_map; uint64_t GetSliceHash(const Slice& s, uint32_t index, uint64_t max_num_buckets) { return hash_map[s.ToString()][index]; } } // namespace class CuckooBuilderTest : public testing::Test { public: CuckooBuilderTest() { env_ = Env::Default(); Options options; options.allow_mmap_reads = true; env_options_ = EnvOptions(options); } void CheckFileContents(const std::vector& keys, const std::vector& values, const std::vector& expected_locations, std::string expected_unused_bucket, uint64_t expected_table_size, uint32_t expected_num_hash_func, bool expected_is_last_level, uint32_t expected_cuckoo_block_size = 1) { // Read file unique_ptr read_file; ASSERT_OK(env_->NewRandomAccessFile(fname, &read_file, env_options_)); uint64_t read_file_size; ASSERT_OK(env_->GetFileSize(fname, &read_file_size)); // Assert Table Properties. TableProperties* props = nullptr; unique_ptr file_reader( new RandomAccessFileReader(std::move(read_file))); ASSERT_OK(ReadTableProperties(file_reader.get(), read_file_size, kCuckooTableMagicNumber, env_, nullptr, &props)); // Check unused bucket. std::string unused_key = props->user_collected_properties[ CuckooTablePropertyNames::kEmptyKey]; ASSERT_EQ(expected_unused_bucket.substr(0, props->fixed_key_len), unused_key); uint32_t value_len_found = *reinterpret_cast(props->user_collected_properties[ CuckooTablePropertyNames::kValueLength].data()); ASSERT_EQ(values.empty() ? 0 : values[0].size(), value_len_found); ASSERT_EQ(props->raw_value_size, values.size()*value_len_found); const uint64_t table_size = *reinterpret_cast(props->user_collected_properties[ CuckooTablePropertyNames::kHashTableSize].data()); ASSERT_EQ(expected_table_size, table_size); const uint32_t num_hash_func_found = *reinterpret_cast(props->user_collected_properties[ CuckooTablePropertyNames::kNumHashFunc].data()); ASSERT_EQ(expected_num_hash_func, num_hash_func_found); const uint32_t cuckoo_block_size = *reinterpret_cast(props->user_collected_properties[ CuckooTablePropertyNames::kCuckooBlockSize].data()); ASSERT_EQ(expected_cuckoo_block_size, cuckoo_block_size); const bool is_last_level_found = *reinterpret_cast(props->user_collected_properties[ CuckooTablePropertyNames::kIsLastLevel].data()); ASSERT_EQ(expected_is_last_level, is_last_level_found); ASSERT_EQ(props->num_entries, keys.size()); ASSERT_EQ(props->fixed_key_len, keys.empty() ? 0 : keys[0].size()); ASSERT_EQ(props->data_size, expected_unused_bucket.size() * (expected_table_size + expected_cuckoo_block_size - 1)); ASSERT_EQ(props->raw_key_size, keys.size()*props->fixed_key_len); delete props; // Check contents of the bucket. std::vector keys_found(keys.size(), false); size_t bucket_size = expected_unused_bucket.size(); for (uint32_t i = 0; i < table_size + cuckoo_block_size - 1; ++i) { Slice read_slice; ASSERT_OK(file_reader->Read(i * bucket_size, bucket_size, &read_slice, nullptr)); size_t key_idx = std::find(expected_locations.begin(), expected_locations.end(), i) - expected_locations.begin(); if (key_idx == keys.size()) { // i is not one of the expected locaitons. Empty bucket. ASSERT_EQ(read_slice.compare(expected_unused_bucket), 0); } else { keys_found[key_idx] = true; ASSERT_EQ(read_slice.compare(keys[key_idx] + values[key_idx]), 0); } } for (auto key_found : keys_found) { // Check that all keys wereReader found. ASSERT_TRUE(key_found); } } std::string GetInternalKey(Slice user_key, bool zero_seqno) { IterKey ikey; ikey.SetInternalKey(user_key, zero_seqno ? 0 : 1000, kTypeValue); return ikey.GetKey().ToString(); } uint64_t NextPowOf2(uint64_t num) { uint64_t n = 2; while (n <= num) { n *= 2; } return n; } uint64_t GetExpectedTableSize(uint64_t num) { return NextPowOf2(static_cast(num / kHashTableRatio)); } Env* env_; EnvOptions env_options_; std::string fname; const double kHashTableRatio = 0.9; }; TEST_F(CuckooBuilderTest, SuccessWithEmptyFile) { unique_ptr writable_file; fname = test::TmpDir() + "/EmptyFile"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, 4, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); ASSERT_EQ(0UL, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); CheckFileContents({}, {}, {}, "", 2, 2, false); } TEST_F(CuckooBuilderTest, WriteSuccessNoCollisionFullKey) { uint32_t num_hash_fun = 4; std::vector user_keys = {"key01", "key02", "key03", "key04"}; std::vector values = {"v01", "v02", "v03", "v04"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1, 2, 3}}, {user_keys[1], {1, 2, 3, 4}}, {user_keys[2], {2, 3, 4, 5}}, {user_keys[3], {3, 4, 5, 6}}}; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 2, 3}; std::vector keys; for (auto& user_key : user_keys) { keys.push_back(GetInternalKey(user_key, false)); } uint64_t expected_table_size = GetExpectedTableSize(keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/NoCollisionFullKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(keys[i]), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = GetInternalKey("key00", true); expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(keys, values, expected_locations, expected_unused_bucket, expected_table_size, 2, false); } TEST_F(CuckooBuilderTest, WriteSuccessWithCollisionFullKey) { uint32_t num_hash_fun = 4; std::vector user_keys = {"key01", "key02", "key03", "key04"}; std::vector values = {"v01", "v02", "v03", "v04"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1, 2, 3}}, {user_keys[1], {0, 1, 2, 3}}, {user_keys[2], {0, 1, 2, 3}}, {user_keys[3], {0, 1, 2, 3}}, }; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 2, 3}; std::vector keys; for (auto& user_key : user_keys) { keys.push_back(GetInternalKey(user_key, false)); } uint64_t expected_table_size = GetExpectedTableSize(keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/WithCollisionFullKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(keys[i]), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = GetInternalKey("key00", true); expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(keys, values, expected_locations, expected_unused_bucket, expected_table_size, 4, false); } TEST_F(CuckooBuilderTest, WriteSuccessWithCollisionAndCuckooBlock) { uint32_t num_hash_fun = 4; std::vector user_keys = {"key01", "key02", "key03", "key04"}; std::vector values = {"v01", "v02", "v03", "v04"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1, 2, 3}}, {user_keys[1], {0, 1, 2, 3}}, {user_keys[2], {0, 1, 2, 3}}, {user_keys[3], {0, 1, 2, 3}}, }; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 2, 3}; std::vector keys; for (auto& user_key : user_keys) { keys.push_back(GetInternalKey(user_key, false)); } uint64_t expected_table_size = GetExpectedTableSize(keys.size()); unique_ptr writable_file; uint32_t cuckoo_block_size = 2; fname = test::TmpDir() + "/WithCollisionFullKey2"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), cuckoo_block_size, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(keys[i]), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = GetInternalKey("key00", true); expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(keys, values, expected_locations, expected_unused_bucket, expected_table_size, 3, false, cuckoo_block_size); } TEST_F(CuckooBuilderTest, WithCollisionPathFullKey) { // Have two hash functions. Insert elements with overlapping hashes. // Finally insert an element with hash value somewhere in the middle // so that it displaces all the elements after that. uint32_t num_hash_fun = 2; std::vector user_keys = {"key01", "key02", "key03", "key04", "key05"}; std::vector values = {"v01", "v02", "v03", "v04", "v05"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1}}, {user_keys[1], {1, 2}}, {user_keys[2], {2, 3}}, {user_keys[3], {3, 4}}, {user_keys[4], {0, 2}}, }; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 3, 4, 2}; std::vector keys; for (auto& user_key : user_keys) { keys.push_back(GetInternalKey(user_key, false)); } uint64_t expected_table_size = GetExpectedTableSize(keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/WithCollisionPathFullKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(keys[i]), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = GetInternalKey("key00", true); expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(keys, values, expected_locations, expected_unused_bucket, expected_table_size, 2, false); } TEST_F(CuckooBuilderTest, WithCollisionPathFullKeyAndCuckooBlock) { uint32_t num_hash_fun = 2; std::vector user_keys = {"key01", "key02", "key03", "key04", "key05"}; std::vector values = {"v01", "v02", "v03", "v04", "v05"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1}}, {user_keys[1], {1, 2}}, {user_keys[2], {3, 4}}, {user_keys[3], {4, 5}}, {user_keys[4], {0, 3}}, }; hash_map = std::move(hm); std::vector expected_locations = {2, 1, 3, 4, 0}; std::vector keys; for (auto& user_key : user_keys) { keys.push_back(GetInternalKey(user_key, false)); } uint64_t expected_table_size = GetExpectedTableSize(keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/WithCollisionPathFullKeyAndCuckooBlock"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 2, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(keys[i]), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = GetInternalKey("key00", true); expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(keys, values, expected_locations, expected_unused_bucket, expected_table_size, 2, false, 2); } TEST_F(CuckooBuilderTest, WriteSuccessNoCollisionUserKey) { uint32_t num_hash_fun = 4; std::vector user_keys = {"key01", "key02", "key03", "key04"}; std::vector values = {"v01", "v02", "v03", "v04"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1, 2, 3}}, {user_keys[1], {1, 2, 3, 4}}, {user_keys[2], {2, 3, 4, 5}}, {user_keys[3], {3, 4, 5, 6}}}; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 2, 3}; uint64_t expected_table_size = GetExpectedTableSize(user_keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/NoCollisionUserKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(GetInternalKey(user_keys[i], true)), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = user_keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = "key00"; expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(user_keys, values, expected_locations, expected_unused_bucket, expected_table_size, 2, true); } TEST_F(CuckooBuilderTest, WriteSuccessWithCollisionUserKey) { uint32_t num_hash_fun = 4; std::vector user_keys = {"key01", "key02", "key03", "key04"}; std::vector values = {"v01", "v02", "v03", "v04"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1, 2, 3}}, {user_keys[1], {0, 1, 2, 3}}, {user_keys[2], {0, 1, 2, 3}}, {user_keys[3], {0, 1, 2, 3}}, }; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 2, 3}; uint64_t expected_table_size = GetExpectedTableSize(user_keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/WithCollisionUserKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(GetInternalKey(user_keys[i], true)), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = user_keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = "key00"; expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(user_keys, values, expected_locations, expected_unused_bucket, expected_table_size, 4, true); } TEST_F(CuckooBuilderTest, WithCollisionPathUserKey) { uint32_t num_hash_fun = 2; std::vector user_keys = {"key01", "key02", "key03", "key04", "key05"}; std::vector values = {"v01", "v02", "v03", "v04", "v05"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1}}, {user_keys[1], {1, 2}}, {user_keys[2], {2, 3}}, {user_keys[3], {3, 4}}, {user_keys[4], {0, 2}}, }; hash_map = std::move(hm); std::vector expected_locations = {0, 1, 3, 4, 2}; uint64_t expected_table_size = GetExpectedTableSize(user_keys.size()); unique_ptr writable_file; fname = test::TmpDir() + "/WithCollisionPathUserKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 2, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(GetInternalKey(user_keys[i], true)), Slice(values[i])); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } size_t bucket_size = user_keys[0].size() + values[0].size(); ASSERT_EQ(expected_table_size * bucket_size - 1, builder.FileSize()); ASSERT_OK(builder.Finish()); ASSERT_OK(file_writer->Close()); ASSERT_LE(expected_table_size * bucket_size, builder.FileSize()); std::string expected_unused_bucket = "key00"; expected_unused_bucket += std::string(values[0].size(), 'a'); CheckFileContents(user_keys, values, expected_locations, expected_unused_bucket, expected_table_size, 2, true); } TEST_F(CuckooBuilderTest, FailWhenCollisionPathTooLong) { // Have two hash functions. Insert elements with overlapping hashes. // Finally try inserting an element with hash value somewhere in the middle // and it should fail because the no. of elements to displace is too high. uint32_t num_hash_fun = 2; std::vector user_keys = {"key01", "key02", "key03", "key04", "key05"}; // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {user_keys[0], {0, 1}}, {user_keys[1], {1, 2}}, {user_keys[2], {2, 3}}, {user_keys[3], {3, 4}}, {user_keys[4], {0, 1}}, }; hash_map = std::move(hm); unique_ptr writable_file; fname = test::TmpDir() + "/WithCollisionPathUserKey"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 2, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t i = 0; i < user_keys.size(); i++) { builder.Add(Slice(GetInternalKey(user_keys[i], false)), Slice("value")); ASSERT_EQ(builder.NumEntries(), i + 1); ASSERT_OK(builder.status()); } ASSERT_TRUE(builder.Finish().IsNotSupported()); ASSERT_OK(file_writer->Close()); } TEST_F(CuckooBuilderTest, FailWhenSameKeyInserted) { // Need to have a temporary variable here as VS compiler does not currently // support operator= with initializer_list as a parameter std::unordered_map> hm = { {"repeatedkey", {0, 1, 2, 3}}}; hash_map = std::move(hm); uint32_t num_hash_fun = 4; std::string user_key = "repeatedkey"; unique_ptr writable_file; fname = test::TmpDir() + "/FailWhenSameKeyInserted"; ASSERT_OK(env_->NewWritableFile(fname, &writable_file, env_options_)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), EnvOptions())); CuckooTableBuilder builder(file_writer.get(), kHashTableRatio, num_hash_fun, 100, BytewiseComparator(), 1, false, false, GetSliceHash); ASSERT_OK(builder.status()); builder.Add(Slice(GetInternalKey(user_key, false)), Slice("value1")); ASSERT_EQ(builder.NumEntries(), 1u); ASSERT_OK(builder.status()); builder.Add(Slice(GetInternalKey(user_key, true)), Slice("value2")); ASSERT_EQ(builder.NumEntries(), 2u); ASSERT_OK(builder.status()); ASSERT_TRUE(builder.Finish().IsNotSupported()); ASSERT_OK(file_writer->Close()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as Cuckoo table is not supported in ROCKSDB_LITE\n"); return 0; } #endif // ROCKSDB_LITE rocksdb-4.5.1/table/cuckoo_table_factory.cc000066400000000000000000000047121267530634000207410ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "table/cuckoo_table_factory.h" #include "db/dbformat.h" #include "table/cuckoo_table_builder.h" #include "table/cuckoo_table_reader.h" namespace rocksdb { Status CuckooTableFactory::NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, std::unique_ptr* table) const { std::unique_ptr new_reader(new CuckooTableReader( table_reader_options.ioptions, std::move(file), file_size, table_reader_options.internal_comparator.user_comparator(), nullptr)); Status s = new_reader->status(); if (s.ok()) { *table = std::move(new_reader); } return s; } TableBuilder* CuckooTableFactory::NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const { // Ignore the skipFIlters flag. Does not apply to this file format // // TODO: change builder to take the option struct return new CuckooTableBuilder( file, table_options_.hash_table_ratio, 64, table_options_.max_search_depth, table_builder_options.internal_comparator.user_comparator(), table_options_.cuckoo_block_size, table_options_.use_module_hash, table_options_.identity_as_first_hash, nullptr); } std::string CuckooTableFactory::GetPrintableTableOptions() const { std::string ret; ret.reserve(2000); const int kBufferSize = 200; char buffer[kBufferSize]; snprintf(buffer, kBufferSize, " hash_table_ratio: %lf\n", table_options_.hash_table_ratio); ret.append(buffer); snprintf(buffer, kBufferSize, " max_search_depth: %u\n", table_options_.max_search_depth); ret.append(buffer); snprintf(buffer, kBufferSize, " cuckoo_block_size: %u\n", table_options_.cuckoo_block_size); ret.append(buffer); snprintf(buffer, kBufferSize, " identity_as_first_hash: %d\n", table_options_.identity_as_first_hash); ret.append(buffer); return ret; } TableFactory* NewCuckooTableFactory(const CuckooTableOptions& table_options) { return new CuckooTableFactory(table_options); } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/cuckoo_table_factory.h000066400000000000000000000053231267530634000206020ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/table.h" #include "util/murmurhash.h" #include "rocksdb/options.h" namespace rocksdb { const uint32_t kCuckooMurmurSeedMultiplier = 816922183; static inline uint64_t CuckooHash( const Slice& user_key, uint32_t hash_cnt, bool use_module_hash, uint64_t table_size_, bool identity_as_first_hash, uint64_t (*get_slice_hash)(const Slice&, uint32_t, uint64_t)) { #if !defined NDEBUG || defined OS_WIN // This part is used only in unit tests but we have to keep it for Windows // build as we run test in both debug and release modes under Windows. if (get_slice_hash != nullptr) { return get_slice_hash(user_key, hash_cnt, table_size_); } #endif uint64_t value = 0; if (hash_cnt == 0 && identity_as_first_hash) { value = (*reinterpret_cast(user_key.data())); } else { value = MurmurHash(user_key.data(), static_cast(user_key.size()), kCuckooMurmurSeedMultiplier * hash_cnt); } if (use_module_hash) { return value % table_size_; } else { return value & (table_size_ - 1); } } // Cuckoo Table is designed for applications that require fast point lookups // but not fast range scans. // // Some assumptions: // - Key length and Value length are fixed. // - Does not support Snapshot. // - Does not support Merge operations. class CuckooTableFactory : public TableFactory { public: explicit CuckooTableFactory(const CuckooTableOptions& table_options) : table_options_(table_options) {} ~CuckooTableFactory() {} const char* Name() const override { return "CuckooTable"; } Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table) const override; TableBuilder* NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const override; // Sanitizes the specified DB Options. Status SanitizeOptions(const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const override { return Status::OK(); } std::string GetPrintableTableOptions() const override; void* GetOptions() override { return &table_options_; } private: CuckooTableOptions table_options_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/cuckoo_table_reader.cc000066400000000000000000000322361267530634000205360ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "table/cuckoo_table_reader.h" #include #include #include #include #include #include "rocksdb/iterator.h" #include "rocksdb/table.h" #include "table/internal_iterator.h" #include "table/meta_blocks.h" #include "table/cuckoo_table_factory.h" #include "table/get_context.h" #include "util/arena.h" #include "util/coding.h" namespace rocksdb { namespace { const uint64_t CACHE_LINE_MASK = ~((uint64_t)CACHE_LINE_SIZE - 1); const uint32_t kInvalidIndex = std::numeric_limits::max(); } extern const uint64_t kCuckooTableMagicNumber; CuckooTableReader::CuckooTableReader( const ImmutableCFOptions& ioptions, std::unique_ptr&& file, uint64_t file_size, const Comparator* comparator, uint64_t (*get_slice_hash)(const Slice&, uint32_t, uint64_t)) : file_(std::move(file)), ucomp_(comparator), get_slice_hash_(get_slice_hash) { if (!ioptions.allow_mmap_reads) { status_ = Status::InvalidArgument("File is not mmaped"); } TableProperties* props = nullptr; status_ = ReadTableProperties(file_.get(), file_size, kCuckooTableMagicNumber, ioptions.env, ioptions.info_log, &props); if (!status_.ok()) { return; } table_props_.reset(props); auto& user_props = props->user_collected_properties; auto hash_funs = user_props.find(CuckooTablePropertyNames::kNumHashFunc); if (hash_funs == user_props.end()) { status_ = Status::Corruption("Number of hash functions not found"); return; } num_hash_func_ = *reinterpret_cast(hash_funs->second.data()); auto unused_key = user_props.find(CuckooTablePropertyNames::kEmptyKey); if (unused_key == user_props.end()) { status_ = Status::Corruption("Empty bucket value not found"); return; } unused_key_ = unused_key->second; key_length_ = static_cast(props->fixed_key_len); auto user_key_len = user_props.find(CuckooTablePropertyNames::kUserKeyLength); if (user_key_len == user_props.end()) { status_ = Status::Corruption("User key length not found"); return; } user_key_length_ = *reinterpret_cast( user_key_len->second.data()); auto value_length = user_props.find(CuckooTablePropertyNames::kValueLength); if (value_length == user_props.end()) { status_ = Status::Corruption("Value length not found"); return; } value_length_ = *reinterpret_cast( value_length->second.data()); bucket_length_ = key_length_ + value_length_; auto hash_table_size = user_props.find( CuckooTablePropertyNames::kHashTableSize); if (hash_table_size == user_props.end()) { status_ = Status::Corruption("Hash table size not found"); return; } table_size_ = *reinterpret_cast( hash_table_size->second.data()); auto is_last_level = user_props.find(CuckooTablePropertyNames::kIsLastLevel); if (is_last_level == user_props.end()) { status_ = Status::Corruption("Is last level not found"); return; } is_last_level_ = *reinterpret_cast(is_last_level->second.data()); auto identity_as_first_hash = user_props.find( CuckooTablePropertyNames::kIdentityAsFirstHash); if (identity_as_first_hash == user_props.end()) { status_ = Status::Corruption("identity as first hash not found"); return; } identity_as_first_hash_ = *reinterpret_cast( identity_as_first_hash->second.data()); auto use_module_hash = user_props.find( CuckooTablePropertyNames::kUseModuleHash); if (use_module_hash == user_props.end()) { status_ = Status::Corruption("hash type is not found"); return; } use_module_hash_ = *reinterpret_cast( use_module_hash->second.data()); auto cuckoo_block_size = user_props.find( CuckooTablePropertyNames::kCuckooBlockSize); if (cuckoo_block_size == user_props.end()) { status_ = Status::Corruption("Cuckoo block size not found"); return; } cuckoo_block_size_ = *reinterpret_cast( cuckoo_block_size->second.data()); cuckoo_block_bytes_minus_one_ = cuckoo_block_size_ * bucket_length_ - 1; status_ = file_->Read(0, file_size, &file_data_, nullptr); } Status CuckooTableReader::Get(const ReadOptions& readOptions, const Slice& key, GetContext* get_context, bool skip_filters) { assert(key.size() == key_length_ + (is_last_level_ ? 8 : 0)); Slice user_key = ExtractUserKey(key); for (uint32_t hash_cnt = 0; hash_cnt < num_hash_func_; ++hash_cnt) { uint64_t offset = bucket_length_ * CuckooHash( user_key, hash_cnt, use_module_hash_, table_size_, identity_as_first_hash_, get_slice_hash_); const char* bucket = &file_data_.data()[offset]; for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_; ++block_idx, bucket += bucket_length_) { if (ucomp_->Equal(Slice(unused_key_.data(), user_key.size()), Slice(bucket, user_key.size()))) { return Status::OK(); } // Here, we compare only the user key part as we support only one entry // per user key and we don't support snapshot. if (ucomp_->Equal(user_key, Slice(bucket, user_key.size()))) { Slice value(bucket + key_length_, value_length_); if (is_last_level_) { // Sequence number is not stored at the last level, so we will use // kMaxSequenceNumber since it is unknown. This could cause some // transactions to fail to lock a key due to known sequence number. // However, it is expected for anyone to use a CuckooTable in a // TransactionDB. get_context->SaveValue(value, kMaxSequenceNumber); } else { Slice full_key(bucket, key_length_); ParsedInternalKey found_ikey; ParseInternalKey(full_key, &found_ikey); get_context->SaveValue(found_ikey, value); } // We don't support merge operations. So, we return here. return Status::OK(); } } } return Status::OK(); } void CuckooTableReader::Prepare(const Slice& key) { // Prefetch the first Cuckoo Block. Slice user_key = ExtractUserKey(key); uint64_t addr = reinterpret_cast(file_data_.data()) + bucket_length_ * CuckooHash(user_key, 0, use_module_hash_, table_size_, identity_as_first_hash_, nullptr); uint64_t end_addr = addr + cuckoo_block_bytes_minus_one_; for (addr &= CACHE_LINE_MASK; addr < end_addr; addr += CACHE_LINE_SIZE) { PREFETCH(reinterpret_cast(addr), 0, 3); } } class CuckooTableIterator : public InternalIterator { public: explicit CuckooTableIterator(CuckooTableReader* reader); ~CuckooTableIterator() {} bool Valid() const override; void SeekToFirst() override; void SeekToLast() override; void Seek(const Slice& target) override; void Next() override; void Prev() override; Slice key() const override; Slice value() const override; Status status() const override { return status_; } void InitIfNeeded(); private: struct BucketComparator { BucketComparator(const Slice& file_data, const Comparator* ucomp, uint32_t bucket_len, uint32_t user_key_len, const Slice& target = Slice()) : file_data_(file_data), ucomp_(ucomp), bucket_len_(bucket_len), user_key_len_(user_key_len), target_(target) {} bool operator()(const uint32_t first, const uint32_t second) const { const char* first_bucket = (first == kInvalidIndex) ? target_.data() : &file_data_.data()[first * bucket_len_]; const char* second_bucket = (second == kInvalidIndex) ? target_.data() : &file_data_.data()[second * bucket_len_]; return ucomp_->Compare(Slice(first_bucket, user_key_len_), Slice(second_bucket, user_key_len_)) < 0; } private: const Slice file_data_; const Comparator* ucomp_; const uint32_t bucket_len_; const uint32_t user_key_len_; const Slice target_; }; const BucketComparator bucket_comparator_; void PrepareKVAtCurrIdx(); CuckooTableReader* reader_; bool initialized_; Status status_; // Contains a map of keys to bucket_id sorted in key order. std::vector sorted_bucket_ids_; // We assume that the number of items can be stored in uint32 (4 Billion). uint32_t curr_key_idx_; Slice curr_value_; IterKey curr_key_; // No copying allowed CuckooTableIterator(const CuckooTableIterator&) = delete; void operator=(const Iterator&) = delete; }; CuckooTableIterator::CuckooTableIterator(CuckooTableReader* reader) : bucket_comparator_(reader->file_data_, reader->ucomp_, reader->bucket_length_, reader->user_key_length_), reader_(reader), initialized_(false), curr_key_idx_(kInvalidIndex) { sorted_bucket_ids_.clear(); curr_value_.clear(); curr_key_.Clear(); } void CuckooTableIterator::InitIfNeeded() { if (initialized_) { return; } sorted_bucket_ids_.reserve(reader_->GetTableProperties()->num_entries); uint64_t num_buckets = reader_->table_size_ + reader_->cuckoo_block_size_ - 1; assert(num_buckets < kInvalidIndex); const char* bucket = reader_->file_data_.data(); for (uint32_t bucket_id = 0; bucket_id < num_buckets; ++bucket_id) { if (Slice(bucket, reader_->key_length_) != Slice(reader_->unused_key_)) { sorted_bucket_ids_.push_back(bucket_id); } bucket += reader_->bucket_length_; } assert(sorted_bucket_ids_.size() == reader_->GetTableProperties()->num_entries); std::sort(sorted_bucket_ids_.begin(), sorted_bucket_ids_.end(), bucket_comparator_); curr_key_idx_ = kInvalidIndex; initialized_ = true; } void CuckooTableIterator::SeekToFirst() { InitIfNeeded(); curr_key_idx_ = 0; PrepareKVAtCurrIdx(); } void CuckooTableIterator::SeekToLast() { InitIfNeeded(); curr_key_idx_ = static_cast(sorted_bucket_ids_.size()) - 1; PrepareKVAtCurrIdx(); } void CuckooTableIterator::Seek(const Slice& target) { InitIfNeeded(); const BucketComparator seek_comparator( reader_->file_data_, reader_->ucomp_, reader_->bucket_length_, reader_->user_key_length_, ExtractUserKey(target)); auto seek_it = std::lower_bound(sorted_bucket_ids_.begin(), sorted_bucket_ids_.end(), kInvalidIndex, seek_comparator); curr_key_idx_ = static_cast(std::distance(sorted_bucket_ids_.begin(), seek_it)); PrepareKVAtCurrIdx(); } bool CuckooTableIterator::Valid() const { return curr_key_idx_ < sorted_bucket_ids_.size(); } void CuckooTableIterator::PrepareKVAtCurrIdx() { if (!Valid()) { curr_value_.clear(); curr_key_.Clear(); return; } uint32_t id = sorted_bucket_ids_[curr_key_idx_]; const char* offset = reader_->file_data_.data() + id * reader_->bucket_length_; if (reader_->is_last_level_) { // Always return internal key. curr_key_.SetInternalKey(Slice(offset, reader_->user_key_length_), 0, kTypeValue); } else { curr_key_.SetKey(Slice(offset, reader_->key_length_)); } curr_value_ = Slice(offset + reader_->key_length_, reader_->value_length_); } void CuckooTableIterator::Next() { if (!Valid()) { curr_value_.clear(); curr_key_.Clear(); return; } ++curr_key_idx_; PrepareKVAtCurrIdx(); } void CuckooTableIterator::Prev() { if (curr_key_idx_ == 0) { curr_key_idx_ = static_cast(sorted_bucket_ids_.size()); } if (!Valid()) { curr_value_.clear(); curr_key_.Clear(); return; } --curr_key_idx_; PrepareKVAtCurrIdx(); } Slice CuckooTableIterator::key() const { assert(Valid()); return curr_key_.GetKey(); } Slice CuckooTableIterator::value() const { assert(Valid()); return curr_value_; } extern InternalIterator* NewErrorInternalIterator(const Status& status, Arena* arena); InternalIterator* CuckooTableReader::NewIterator( const ReadOptions& read_options, Arena* arena, bool skip_filters) { if (!status().ok()) { return NewErrorInternalIterator( Status::Corruption("CuckooTableReader status is not okay."), arena); } if (read_options.total_order_seek) { return NewErrorInternalIterator( Status::InvalidArgument("total_order_seek is not supported."), arena); } CuckooTableIterator* iter; if (arena == nullptr) { iter = new CuckooTableIterator(this); } else { auto iter_mem = arena->AllocateAligned(sizeof(CuckooTableIterator)); iter = new (iter_mem) CuckooTableIterator(this); } return iter; } size_t CuckooTableReader::ApproximateMemoryUsage() const { return 0; } } // namespace rocksdb #endif rocksdb-4.5.1/table/cuckoo_table_reader.h000066400000000000000000000054461267530634000204030ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include #include "db/dbformat.h" #include "rocksdb/env.h" #include "rocksdb/options.h" #include "table/table_reader.h" #include "util/file_reader_writer.h" namespace rocksdb { class Arena; class TableReader; class InternalIterator; class CuckooTableReader: public TableReader { public: CuckooTableReader(const ImmutableCFOptions& ioptions, std::unique_ptr&& file, uint64_t file_size, const Comparator* user_comparator, uint64_t (*get_slice_hash)(const Slice&, uint32_t, uint64_t)); ~CuckooTableReader() {} std::shared_ptr GetTableProperties() const override { return table_props_; } Status status() const { return status_; } Status Get(const ReadOptions& read_options, const Slice& key, GetContext* get_context, bool skip_filters = false) override; InternalIterator* NewIterator(const ReadOptions&, Arena* arena = nullptr, bool skip_filters = false) override; void Prepare(const Slice& target) override; // Report an approximation of how much memory has been used. size_t ApproximateMemoryUsage() const override; // Following methods are not implemented for Cuckoo Table Reader uint64_t ApproximateOffsetOf(const Slice& key) override { return 0; } void SetupForCompaction() override {} // End of methods not implemented. private: friend class CuckooTableIterator; void LoadAllKeys(std::vector>* key_to_bucket_id); std::unique_ptr file_; Slice file_data_; bool is_last_level_; bool identity_as_first_hash_; bool use_module_hash_; std::shared_ptr table_props_; Status status_; uint32_t num_hash_func_; std::string unused_key_; uint32_t key_length_; uint32_t user_key_length_; uint32_t value_length_; uint32_t bucket_length_; uint32_t cuckoo_block_size_; uint32_t cuckoo_block_bytes_minus_one_; uint64_t table_size_; const Comparator* ucomp_; uint64_t (*get_slice_hash_)(const Slice& s, uint32_t index, uint64_t max_num_buckets); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/cuckoo_table_reader_test.cc000066400000000000000000000462031267530634000215740ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #ifndef GFLAGS #include int main() { fprintf(stderr, "Please install gflags to run this test... Skipping...\n"); return 0; } #else #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include #include #include #include #include "table/meta_blocks.h" #include "table/cuckoo_table_builder.h" #include "table/cuckoo_table_reader.h" #include "table/cuckoo_table_factory.h" #include "table/get_context.h" #include "util/arena.h" #include "util/random.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" using GFLAGS::ParseCommandLineFlags; using GFLAGS::SetUsageMessage; DEFINE_string(file_dir, "", "Directory where the files will be created" " for benchmark. Added for using tmpfs."); DEFINE_bool(enable_perf, false, "Run Benchmark Tests too."); DEFINE_bool(write, false, "Should write new values to file in performance tests?"); DEFINE_bool(identity_as_first_hash, true, "use identity as first hash"); namespace rocksdb { namespace { const uint32_t kNumHashFunc = 10; // Methods, variables related to Hash functions. std::unordered_map> hash_map; void AddHashLookups(const std::string& s, uint64_t bucket_id, uint32_t num_hash_fun) { std::vector v; for (uint32_t i = 0; i < num_hash_fun; i++) { v.push_back(bucket_id + i); } hash_map[s] = v; } uint64_t GetSliceHash(const Slice& s, uint32_t index, uint64_t max_num_buckets) { return hash_map[s.ToString()][index]; } } // namespace class CuckooReaderTest : public testing::Test { public: using testing::Test::SetUp; CuckooReaderTest() { options.allow_mmap_reads = true; env = options.env; env_options = EnvOptions(options); } void SetUp(int num) { num_items = num; hash_map.clear(); keys.clear(); keys.resize(num_items); user_keys.clear(); user_keys.resize(num_items); values.clear(); values.resize(num_items); } std::string NumToStr(int64_t i) { return std::string(reinterpret_cast(&i), sizeof(i)); } void CreateCuckooFileAndCheckReader( const Comparator* ucomp = BytewiseComparator()) { std::unique_ptr writable_file; ASSERT_OK(env->NewWritableFile(fname, &writable_file, env_options)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), env_options)); CuckooTableBuilder builder(file_writer.get(), 0.9, kNumHashFunc, 100, ucomp, 2, false, false, GetSliceHash); ASSERT_OK(builder.status()); for (uint32_t key_idx = 0; key_idx < num_items; ++key_idx) { builder.Add(Slice(keys[key_idx]), Slice(values[key_idx])); ASSERT_OK(builder.status()); ASSERT_EQ(builder.NumEntries(), key_idx + 1); } ASSERT_OK(builder.Finish()); ASSERT_EQ(num_items, builder.NumEntries()); file_size = builder.FileSize(); ASSERT_OK(file_writer->Close()); // Check reader now. std::unique_ptr read_file; ASSERT_OK(env->NewRandomAccessFile(fname, &read_file, env_options)); unique_ptr file_reader( new RandomAccessFileReader(std::move(read_file))); const ImmutableCFOptions ioptions(options); CuckooTableReader reader(ioptions, std::move(file_reader), file_size, ucomp, GetSliceHash); ASSERT_OK(reader.status()); // Assume no merge/deletion for (uint32_t i = 0; i < num_items; ++i) { std::string value; GetContext get_context(ucomp, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(user_keys[i]), &value, nullptr, nullptr, nullptr); ASSERT_OK(reader.Get(ReadOptions(), Slice(keys[i]), &get_context)); ASSERT_EQ(values[i], value); } } void UpdateKeys(bool with_zero_seqno) { for (uint32_t i = 0; i < num_items; i++) { ParsedInternalKey ikey(user_keys[i], with_zero_seqno ? 0 : i + 1000, kTypeValue); keys[i].clear(); AppendInternalKey(&keys[i], ikey); } } void CheckIterator(const Comparator* ucomp = BytewiseComparator()) { std::unique_ptr read_file; ASSERT_OK(env->NewRandomAccessFile(fname, &read_file, env_options)); unique_ptr file_reader( new RandomAccessFileReader(std::move(read_file))); const ImmutableCFOptions ioptions(options); CuckooTableReader reader(ioptions, std::move(file_reader), file_size, ucomp, GetSliceHash); ASSERT_OK(reader.status()); InternalIterator* it = reader.NewIterator(ReadOptions(), nullptr); ASSERT_OK(it->status()); ASSERT_TRUE(!it->Valid()); it->SeekToFirst(); int cnt = 0; while (it->Valid()) { ASSERT_OK(it->status()); ASSERT_TRUE(Slice(keys[cnt]) == it->key()); ASSERT_TRUE(Slice(values[cnt]) == it->value()); ++cnt; it->Next(); } ASSERT_EQ(static_cast(cnt), num_items); it->SeekToLast(); cnt = static_cast(num_items) - 1; ASSERT_TRUE(it->Valid()); while (it->Valid()) { ASSERT_OK(it->status()); ASSERT_TRUE(Slice(keys[cnt]) == it->key()); ASSERT_TRUE(Slice(values[cnt]) == it->value()); --cnt; it->Prev(); } ASSERT_EQ(cnt, -1); cnt = static_cast(num_items) / 2; it->Seek(keys[cnt]); while (it->Valid()) { ASSERT_OK(it->status()); ASSERT_TRUE(Slice(keys[cnt]) == it->key()); ASSERT_TRUE(Slice(values[cnt]) == it->value()); ++cnt; it->Next(); } ASSERT_EQ(static_cast(cnt), num_items); delete it; Arena arena; it = reader.NewIterator(ReadOptions(), &arena); ASSERT_OK(it->status()); ASSERT_TRUE(!it->Valid()); it->Seek(keys[num_items/2]); ASSERT_TRUE(it->Valid()); ASSERT_OK(it->status()); ASSERT_TRUE(keys[num_items/2] == it->key()); ASSERT_TRUE(values[num_items/2] == it->value()); ASSERT_OK(it->status()); it->~InternalIterator(); } std::vector keys; std::vector user_keys; std::vector values; uint64_t num_items; std::string fname; uint64_t file_size; Options options; Env* env; EnvOptions env_options; }; TEST_F(CuckooReaderTest, WhenKeyExists) { SetUp(kNumHashFunc); fname = test::TmpDir() + "/CuckooReader_WhenKeyExists"; for (uint64_t i = 0; i < num_items; i++) { user_keys[i] = "key" + NumToStr(i); ParsedInternalKey ikey(user_keys[i], i + 1000, kTypeValue); AppendInternalKey(&keys[i], ikey); values[i] = "value" + NumToStr(i); // Give disjoint hash values. AddHashLookups(user_keys[i], i, kNumHashFunc); } CreateCuckooFileAndCheckReader(); // Last level file. UpdateKeys(true); CreateCuckooFileAndCheckReader(); // Test with collision. Make all hash values collide. hash_map.clear(); for (uint32_t i = 0; i < num_items; i++) { AddHashLookups(user_keys[i], 0, kNumHashFunc); } UpdateKeys(false); CreateCuckooFileAndCheckReader(); // Last level file. UpdateKeys(true); CreateCuckooFileAndCheckReader(); } TEST_F(CuckooReaderTest, WhenKeyExistsWithUint64Comparator) { SetUp(kNumHashFunc); fname = test::TmpDir() + "/CuckooReaderUint64_WhenKeyExists"; for (uint64_t i = 0; i < num_items; i++) { user_keys[i].resize(8); memcpy(&user_keys[i][0], static_cast(&i), 8); ParsedInternalKey ikey(user_keys[i], i + 1000, kTypeValue); AppendInternalKey(&keys[i], ikey); values[i] = "value" + NumToStr(i); // Give disjoint hash values. AddHashLookups(user_keys[i], i, kNumHashFunc); } CreateCuckooFileAndCheckReader(test::Uint64Comparator()); // Last level file. UpdateKeys(true); CreateCuckooFileAndCheckReader(test::Uint64Comparator()); // Test with collision. Make all hash values collide. hash_map.clear(); for (uint32_t i = 0; i < num_items; i++) { AddHashLookups(user_keys[i], 0, kNumHashFunc); } UpdateKeys(false); CreateCuckooFileAndCheckReader(test::Uint64Comparator()); // Last level file. UpdateKeys(true); CreateCuckooFileAndCheckReader(test::Uint64Comparator()); } TEST_F(CuckooReaderTest, CheckIterator) { SetUp(2*kNumHashFunc); fname = test::TmpDir() + "/CuckooReader_CheckIterator"; for (uint64_t i = 0; i < num_items; i++) { user_keys[i] = "key" + NumToStr(i); ParsedInternalKey ikey(user_keys[i], 1000, kTypeValue); AppendInternalKey(&keys[i], ikey); values[i] = "value" + NumToStr(i); // Give disjoint hash values, in reverse order. AddHashLookups(user_keys[i], num_items-i-1, kNumHashFunc); } CreateCuckooFileAndCheckReader(); CheckIterator(); // Last level file. UpdateKeys(true); CreateCuckooFileAndCheckReader(); CheckIterator(); } TEST_F(CuckooReaderTest, CheckIteratorUint64) { SetUp(2*kNumHashFunc); fname = test::TmpDir() + "/CuckooReader_CheckIterator"; for (uint64_t i = 0; i < num_items; i++) { user_keys[i].resize(8); memcpy(&user_keys[i][0], static_cast(&i), 8); ParsedInternalKey ikey(user_keys[i], 1000, kTypeValue); AppendInternalKey(&keys[i], ikey); values[i] = "value" + NumToStr(i); // Give disjoint hash values, in reverse order. AddHashLookups(user_keys[i], num_items-i-1, kNumHashFunc); } CreateCuckooFileAndCheckReader(test::Uint64Comparator()); CheckIterator(test::Uint64Comparator()); // Last level file. UpdateKeys(true); CreateCuckooFileAndCheckReader(test::Uint64Comparator()); CheckIterator(test::Uint64Comparator()); } TEST_F(CuckooReaderTest, WhenKeyNotFound) { // Add keys with colliding hash values. SetUp(kNumHashFunc); fname = test::TmpDir() + "/CuckooReader_WhenKeyNotFound"; for (uint64_t i = 0; i < num_items; i++) { user_keys[i] = "key" + NumToStr(i); ParsedInternalKey ikey(user_keys[i], i + 1000, kTypeValue); AppendInternalKey(&keys[i], ikey); values[i] = "value" + NumToStr(i); // Make all hash values collide. AddHashLookups(user_keys[i], 0, kNumHashFunc); } auto* ucmp = BytewiseComparator(); CreateCuckooFileAndCheckReader(); std::unique_ptr read_file; ASSERT_OK(env->NewRandomAccessFile(fname, &read_file, env_options)); unique_ptr file_reader( new RandomAccessFileReader(std::move(read_file))); const ImmutableCFOptions ioptions(options); CuckooTableReader reader(ioptions, std::move(file_reader), file_size, ucmp, GetSliceHash); ASSERT_OK(reader.status()); // Search for a key with colliding hash values. std::string not_found_user_key = "key" + NumToStr(num_items); std::string not_found_key; AddHashLookups(not_found_user_key, 0, kNumHashFunc); ParsedInternalKey ikey(not_found_user_key, 1000, kTypeValue); AppendInternalKey(¬_found_key, ikey); std::string value; GetContext get_context(ucmp, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(not_found_key), &value, nullptr, nullptr, nullptr); ASSERT_OK(reader.Get(ReadOptions(), Slice(not_found_key), &get_context)); ASSERT_TRUE(value.empty()); ASSERT_OK(reader.status()); // Search for a key with an independent hash value. std::string not_found_user_key2 = "key" + NumToStr(num_items + 1); AddHashLookups(not_found_user_key2, kNumHashFunc, kNumHashFunc); ParsedInternalKey ikey2(not_found_user_key2, 1000, kTypeValue); std::string not_found_key2; AppendInternalKey(¬_found_key2, ikey2); GetContext get_context2(ucmp, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(not_found_key2), &value, nullptr, nullptr, nullptr); ASSERT_OK(reader.Get(ReadOptions(), Slice(not_found_key2), &get_context2)); ASSERT_TRUE(value.empty()); ASSERT_OK(reader.status()); // Test read when key is unused key. std::string unused_key = reader.GetTableProperties()->user_collected_properties.at( CuckooTablePropertyNames::kEmptyKey); // Add hash values that map to empty buckets. AddHashLookups(ExtractUserKey(unused_key).ToString(), kNumHashFunc, kNumHashFunc); GetContext get_context3(ucmp, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(unused_key), &value, nullptr, nullptr, nullptr); ASSERT_OK(reader.Get(ReadOptions(), Slice(unused_key), &get_context3)); ASSERT_TRUE(value.empty()); ASSERT_OK(reader.status()); } // Performance tests namespace { void GetKeys(uint64_t num, std::vector* keys) { keys->clear(); IterKey k; k.SetInternalKey("", 0, kTypeValue); std::string internal_key_suffix = k.GetKey().ToString(); ASSERT_EQ(static_cast(8), internal_key_suffix.size()); for (uint64_t key_idx = 0; key_idx < num; ++key_idx) { uint64_t value = 2 * key_idx; std::string new_key(reinterpret_cast(&value), sizeof(value)); new_key += internal_key_suffix; keys->push_back(new_key); } } std::string GetFileName(uint64_t num) { if (FLAGS_file_dir.empty()) { FLAGS_file_dir = test::TmpDir(); } return FLAGS_file_dir + "/cuckoo_read_benchmark" + ToString(num/1000000) + "Mkeys"; } // Create last level file as we are interested in measuring performance of // last level file only. void WriteFile(const std::vector& keys, const uint64_t num, double hash_ratio) { Options options; options.allow_mmap_reads = true; Env* env = options.env; EnvOptions env_options = EnvOptions(options); std::string fname = GetFileName(num); std::unique_ptr writable_file; ASSERT_OK(env->NewWritableFile(fname, &writable_file, env_options)); unique_ptr file_writer( new WritableFileWriter(std::move(writable_file), env_options)); CuckooTableBuilder builder(file_writer.get(), hash_ratio, 64, 1000, test::Uint64Comparator(), 5, false, FLAGS_identity_as_first_hash, nullptr); ASSERT_OK(builder.status()); for (uint64_t key_idx = 0; key_idx < num; ++key_idx) { // Value is just a part of key. builder.Add(Slice(keys[key_idx]), Slice(&keys[key_idx][0], 4)); ASSERT_EQ(builder.NumEntries(), key_idx + 1); ASSERT_OK(builder.status()); } ASSERT_OK(builder.Finish()); ASSERT_EQ(num, builder.NumEntries()); ASSERT_OK(file_writer->Close()); uint64_t file_size; env->GetFileSize(fname, &file_size); std::unique_ptr read_file; ASSERT_OK(env->NewRandomAccessFile(fname, &read_file, env_options)); unique_ptr file_reader( new RandomAccessFileReader(std::move(read_file))); const ImmutableCFOptions ioptions(options); CuckooTableReader reader(ioptions, std::move(file_reader), file_size, test::Uint64Comparator(), nullptr); ASSERT_OK(reader.status()); ReadOptions r_options; std::string value; // Assume only the fast path is triggered GetContext get_context(nullptr, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(), &value, nullptr, nullptr, nullptr); for (uint64_t i = 0; i < num; ++i) { value.clear(); ASSERT_OK(reader.Get(r_options, Slice(keys[i]), &get_context)); ASSERT_TRUE(Slice(keys[i]) == Slice(&keys[i][0], 4)); } } void ReadKeys(uint64_t num, uint32_t batch_size) { Options options; options.allow_mmap_reads = true; Env* env = options.env; EnvOptions env_options = EnvOptions(options); std::string fname = GetFileName(num); uint64_t file_size; env->GetFileSize(fname, &file_size); std::unique_ptr read_file; ASSERT_OK(env->NewRandomAccessFile(fname, &read_file, env_options)); unique_ptr file_reader( new RandomAccessFileReader(std::move(read_file))); const ImmutableCFOptions ioptions(options); CuckooTableReader reader(ioptions, std::move(file_reader), file_size, test::Uint64Comparator(), nullptr); ASSERT_OK(reader.status()); const UserCollectedProperties user_props = reader.GetTableProperties()->user_collected_properties; const uint32_t num_hash_fun = *reinterpret_cast( user_props.at(CuckooTablePropertyNames::kNumHashFunc).data()); const uint64_t table_size = *reinterpret_cast( user_props.at(CuckooTablePropertyNames::kHashTableSize).data()); fprintf(stderr, "With %" PRIu64 " items, utilization is %.2f%%, number of" " hash functions: %u.\n", num, num * 100.0 / (table_size), num_hash_fun); ReadOptions r_options; std::vector keys; keys.reserve(num); for (uint64_t i = 0; i < num; ++i) { keys.push_back(2 * i); } std::random_shuffle(keys.begin(), keys.end()); std::string value; // Assume only the fast path is triggered GetContext get_context(nullptr, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(), &value, nullptr, nullptr, nullptr); uint64_t start_time = env->NowMicros(); if (batch_size > 0) { for (uint64_t i = 0; i < num; i += batch_size) { for (uint64_t j = i; j < i+batch_size && j < num; ++j) { reader.Prepare(Slice(reinterpret_cast(&keys[j]), 16)); } for (uint64_t j = i; j < i+batch_size && j < num; ++j) { reader.Get(r_options, Slice(reinterpret_cast(&keys[j]), 16), &get_context); } } } else { for (uint64_t i = 0; i < num; i++) { reader.Get(r_options, Slice(reinterpret_cast(&keys[i]), 16), &get_context); } } float time_per_op = (env->NowMicros() - start_time) * 1.0f / num; fprintf(stderr, "Time taken per op is %.3fus (%.1f Mqps) with batch size of %u\n", time_per_op, 1.0 / time_per_op, batch_size); } } // namespace. TEST_F(CuckooReaderTest, TestReadPerformance) { if (!FLAGS_enable_perf) { return; } double hash_ratio = 0.95; // These numbers are chosen to have a hash utilizaiton % close to // 0.9, 0.75, 0.6 and 0.5 respectively. // They all create 128 M buckets. std::vector nums = {120*1024*1024, 100*1024*1024, 80*1024*1024, 70*1024*1024}; #ifndef NDEBUG fprintf(stdout, "WARNING: Not compiled with DNDEBUG. Performance tests may be slow.\n"); #endif for (uint64_t num : nums) { if (FLAGS_write || Env::Default()->FileExists(GetFileName(num)).IsNotFound()) { std::vector all_keys; GetKeys(num, &all_keys); WriteFile(all_keys, num, hash_ratio); } ReadKeys(num, 0); ReadKeys(num, 10); ReadKeys(num, 25); ReadKeys(num, 50); ReadKeys(num, 100); fprintf(stderr, "\n"); } } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); ParseCommandLineFlags(&argc, &argv, true); return RUN_ALL_TESTS(); } #endif // GFLAGS. #else #include int main(int argc, char** argv) { fprintf(stderr, "SKIPPED as Cuckoo table is not supported in ROCKSDB_LITE\n"); return 0; } #endif // ROCKSDB_LITE rocksdb-4.5.1/table/filter_block.h000066400000000000000000000061041267530634000170560ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2012 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // // A filter block is stored near the end of a Table file. It contains // filters (e.g., bloom filters) for all data blocks in the table combined // into a single filter block. // // It is a base class for BlockBasedFilter and FullFilter. // These two are both used in BlockBasedTable. The first one contain filter // For a part of keys in sst file, the second contain filter for all keys // in sst file. #pragma once #include #include #include #include #include #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "util/hash.h" #include "format.h" namespace rocksdb { const uint64_t kNotValid = ULLONG_MAX; class FilterPolicy; // A FilterBlockBuilder is used to construct all of the filters for a // particular Table. It generates a single string which is stored as // a special block in the Table. // // The sequence of calls to FilterBlockBuilder must match the regexp: // (StartBlock Add*)* Finish // // BlockBased/Full FilterBlock would be called in the same way. class FilterBlockBuilder { public: explicit FilterBlockBuilder() {} virtual ~FilterBlockBuilder() {} virtual bool IsBlockBased() = 0; // If is blockbased filter virtual void StartBlock(uint64_t block_offset) = 0; // Start new block filter virtual void Add(const Slice& key) = 0; // Add a key to current filter virtual Slice Finish() = 0; // Generate Filter private: // No copying allowed FilterBlockBuilder(const FilterBlockBuilder&); void operator=(const FilterBlockBuilder&); }; // A FilterBlockReader is used to parse filter from SST table. // KeyMayMatch and PrefixMayMatch would trigger filter checking // // BlockBased/Full FilterBlock would be called in the same way. class FilterBlockReader { public: explicit FilterBlockReader() {} virtual ~FilterBlockReader() {} virtual bool IsBlockBased() = 0; // If is blockbased filter virtual bool KeyMayMatch(const Slice& key, uint64_t block_offset = kNotValid) = 0; virtual bool PrefixMayMatch(const Slice& prefix, uint64_t block_offset = kNotValid) = 0; virtual size_t ApproximateMemoryUsage() const = 0; // convert this object to a human readable form virtual std::string ToString() const { std::string error_msg("Unsupported filter \n"); return error_msg; } private: // No copying allowed FilterBlockReader(const FilterBlockReader&); void operator=(const FilterBlockReader&); }; } // namespace rocksdb rocksdb-4.5.1/table/flush_block_policy.cc000066400000000000000000000051241267530634000204300ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "rocksdb/options.h" #include "rocksdb/flush_block_policy.h" #include "rocksdb/slice.h" #include "table/block_builder.h" #include namespace rocksdb { // Flush block by size class FlushBlockBySizePolicy : public FlushBlockPolicy { public: // @params block_size: Approximate size of user data packed per // block. // @params block_size_deviation: This is used to close a block before it // reaches the configured FlushBlockBySizePolicy(const uint64_t block_size, const uint64_t block_size_deviation, const BlockBuilder& data_block_builder) : block_size_(block_size), block_size_deviation_(block_size_deviation), data_block_builder_(data_block_builder) { } virtual bool Update(const Slice& key, const Slice& value) override { // it makes no sense to flush when the data block is empty if (data_block_builder_.empty()) { return false; } auto curr_size = data_block_builder_.CurrentSizeEstimate(); // Do flush if one of the below two conditions is true: // 1) if the current estimated size already exceeds the block size, // 2) block_size_deviation is set and the estimated size after appending // the kv will exceed the block size and the current size is under the // the deviation. return curr_size >= block_size_ || BlockAlmostFull(key, value); } private: bool BlockAlmostFull(const Slice& key, const Slice& value) const { const auto curr_size = data_block_builder_.CurrentSizeEstimate(); const auto estimated_size_after = data_block_builder_.EstimateSizeAfterKV(key, value); return estimated_size_after > block_size_ && block_size_deviation_ > 0 && curr_size * 100 > block_size_ * (100 - block_size_deviation_); } const uint64_t block_size_; const uint64_t block_size_deviation_; const BlockBuilder& data_block_builder_; }; FlushBlockPolicy* FlushBlockBySizePolicyFactory::NewFlushBlockPolicy( const BlockBasedTableOptions& table_options, const BlockBuilder& data_block_builder) const { return new FlushBlockBySizePolicy( table_options.block_size, table_options.block_size_deviation, data_block_builder); } } // namespace rocksdb rocksdb-4.5.1/table/format.cc000066400000000000000000000367671267530634000160670ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/format.h" #include #include #include "rocksdb/env.h" #include "table/block.h" #include "util/coding.h" #include "util/compression.h" #include "util/crc32c.h" #include "util/file_reader_writer.h" #include "util/perf_context_imp.h" #include "util/string_util.h" #include "util/xxhash.h" namespace rocksdb { extern const uint64_t kLegacyBlockBasedTableMagicNumber; extern const uint64_t kBlockBasedTableMagicNumber; #ifndef ROCKSDB_LITE extern const uint64_t kLegacyPlainTableMagicNumber; extern const uint64_t kPlainTableMagicNumber; #else // ROCKSDB_LITE doesn't have plain table const uint64_t kLegacyPlainTableMagicNumber = 0; const uint64_t kPlainTableMagicNumber = 0; #endif const uint32_t DefaultStackBufferSize = 5000; void BlockHandle::EncodeTo(std::string* dst) const { // Sanity check that all fields have been set assert(offset_ != ~static_cast(0)); assert(size_ != ~static_cast(0)); PutVarint64(dst, offset_); PutVarint64(dst, size_); } Status BlockHandle::DecodeFrom(Slice* input) { if (GetVarint64(input, &offset_) && GetVarint64(input, &size_)) { return Status::OK(); } else { return Status::Corruption("bad block handle"); } } // Return a string that contains the copy of handle. std::string BlockHandle::ToString(bool hex) const { std::string handle_str; EncodeTo(&handle_str); if (hex) { std::string result; char buf[10]; for (size_t i = 0; i < handle_str.size(); i++) { snprintf(buf, sizeof(buf), "%02X", static_cast(handle_str[i])); result += buf; } return result; } else { return handle_str; } } const BlockHandle BlockHandle::kNullBlockHandle(0, 0); namespace { inline bool IsLegacyFooterFormat(uint64_t magic_number) { return magic_number == kLegacyBlockBasedTableMagicNumber || magic_number == kLegacyPlainTableMagicNumber; } inline uint64_t UpconvertLegacyFooterFormat(uint64_t magic_number) { if (magic_number == kLegacyBlockBasedTableMagicNumber) { return kBlockBasedTableMagicNumber; } if (magic_number == kLegacyPlainTableMagicNumber) { return kPlainTableMagicNumber; } assert(false); return 0; } } // namespace // legacy footer format: // metaindex handle (varint64 offset, varint64 size) // index handle (varint64 offset, varint64 size) // to make the total size 2 * BlockHandle::kMaxEncodedLength // table_magic_number (8 bytes) // new footer format: // checksum (char, 1 byte) // metaindex handle (varint64 offset, varint64 size) // index handle (varint64 offset, varint64 size) // to make the total size 2 * BlockHandle::kMaxEncodedLength + 1 // footer version (4 bytes) // table_magic_number (8 bytes) void Footer::EncodeTo(std::string* dst) const { assert(HasInitializedTableMagicNumber()); if (IsLegacyFooterFormat(table_magic_number())) { // has to be default checksum with legacy footer assert(checksum_ == kCRC32c); const size_t original_size = dst->size(); metaindex_handle_.EncodeTo(dst); index_handle_.EncodeTo(dst); dst->resize(original_size + 2 * BlockHandle::kMaxEncodedLength); // Padding PutFixed32(dst, static_cast(table_magic_number() & 0xffffffffu)); PutFixed32(dst, static_cast(table_magic_number() >> 32)); assert(dst->size() == original_size + kVersion0EncodedLength); } else { const size_t original_size = dst->size(); dst->push_back(static_cast(checksum_)); metaindex_handle_.EncodeTo(dst); index_handle_.EncodeTo(dst); dst->resize(original_size + kNewVersionsEncodedLength - 12); // Padding PutFixed32(dst, version()); PutFixed32(dst, static_cast(table_magic_number() & 0xffffffffu)); PutFixed32(dst, static_cast(table_magic_number() >> 32)); assert(dst->size() == original_size + kNewVersionsEncodedLength); } } Footer::Footer(uint64_t _table_magic_number, uint32_t _version) : version_(_version), checksum_(kCRC32c), table_magic_number_(_table_magic_number) { // This should be guaranteed by constructor callers assert(!IsLegacyFooterFormat(_table_magic_number) || version_ == 0); } Status Footer::DecodeFrom(Slice* input) { assert(!HasInitializedTableMagicNumber()); assert(input != nullptr); assert(input->size() >= kMinEncodedLength); const char *magic_ptr = input->data() + input->size() - kMagicNumberLengthByte; const uint32_t magic_lo = DecodeFixed32(magic_ptr); const uint32_t magic_hi = DecodeFixed32(magic_ptr + 4); uint64_t magic = ((static_cast(magic_hi) << 32) | (static_cast(magic_lo))); // We check for legacy formats here and silently upconvert them bool legacy = IsLegacyFooterFormat(magic); if (legacy) { magic = UpconvertLegacyFooterFormat(magic); } set_table_magic_number(magic); if (legacy) { // The size is already asserted to be at least kMinEncodedLength // at the beginning of the function input->remove_prefix(input->size() - kVersion0EncodedLength); version_ = 0 /* legacy */; checksum_ = kCRC32c; } else { version_ = DecodeFixed32(magic_ptr - 4); // Footer version 1 and higher will always occupy exactly this many bytes. // It consists of the checksum type, two block handles, padding, // a version number, and a magic number if (input->size() < kNewVersionsEncodedLength) { return Status::Corruption("input is too short to be an sstable"); } else { input->remove_prefix(input->size() - kNewVersionsEncodedLength); } uint32_t chksum; if (!GetVarint32(input, &chksum)) { return Status::Corruption("bad checksum type"); } checksum_ = static_cast(chksum); } Status result = metaindex_handle_.DecodeFrom(input); if (result.ok()) { result = index_handle_.DecodeFrom(input); } if (result.ok()) { // We skip over any leftover data (just padding for now) in "input" const char* end = magic_ptr + kMagicNumberLengthByte; *input = Slice(end, input->data() + input->size() - end); } return result; } std::string Footer::ToString() const { std::string result, handle_; result.reserve(1024); bool legacy = IsLegacyFooterFormat(table_magic_number_); if (legacy) { result.append("metaindex handle: " + metaindex_handle_.ToString() + "\n "); result.append("index handle: " + index_handle_.ToString() + "\n "); result.append("table_magic_number: " + rocksdb::ToString(table_magic_number_) + "\n "); } else { result.append("checksum: " + rocksdb::ToString(checksum_) + "\n "); result.append("metaindex handle: " + metaindex_handle_.ToString() + "\n "); result.append("index handle: " + index_handle_.ToString() + "\n "); result.append("footer version: " + rocksdb::ToString(version_) + "\n "); result.append("table_magic_number: " + rocksdb::ToString(table_magic_number_) + "\n "); } return result; } Status ReadFooterFromFile(RandomAccessFileReader* file, uint64_t file_size, Footer* footer, uint64_t enforce_table_magic_number) { if (file_size < Footer::kMinEncodedLength) { return Status::Corruption("file is too short to be an sstable"); } char footer_space[Footer::kMaxEncodedLength]; Slice footer_input; size_t read_offset = (file_size > Footer::kMaxEncodedLength) ? static_cast(file_size - Footer::kMaxEncodedLength) : 0; Status s = file->Read(read_offset, Footer::kMaxEncodedLength, &footer_input, footer_space); if (!s.ok()) return s; // Check that we actually read the whole footer from the file. It may be // that size isn't correct. if (footer_input.size() < Footer::kMinEncodedLength) { return Status::Corruption("file is too short to be an sstable"); } s = footer->DecodeFrom(&footer_input); if (!s.ok()) { return s; } if (enforce_table_magic_number != 0 && enforce_table_magic_number != footer->table_magic_number()) { return Status::Corruption("Bad table magic number"); } return Status::OK(); } // Without anonymous namespace here, we fail the warning -Wmissing-prototypes namespace { // Read a block and check its CRC // contents is the result of reading. // According to the implementation of file->Read, contents may not point to buf Status ReadBlock(RandomAccessFileReader* file, const Footer& footer, const ReadOptions& options, const BlockHandle& handle, Slice* contents, /* result of reading */ char* buf) { size_t n = static_cast(handle.size()); Status s; { PERF_TIMER_GUARD(block_read_time); s = file->Read(handle.offset(), n + kBlockTrailerSize, contents, buf); } PERF_COUNTER_ADD(block_read_count, 1); PERF_COUNTER_ADD(block_read_byte, n + kBlockTrailerSize); if (!s.ok()) { return s; } if (contents->size() != n + kBlockTrailerSize) { return Status::Corruption("truncated block read"); } // Check the crc of the type and the block contents const char* data = contents->data(); // Pointer to where Read put the data if (options.verify_checksums) { PERF_TIMER_GUARD(block_checksum_time); uint32_t value = DecodeFixed32(data + n + 1); uint32_t actual = 0; switch (footer.checksum()) { case kCRC32c: value = crc32c::Unmask(value); actual = crc32c::Value(data, n + 1); break; case kxxHash: actual = XXH32(data, static_cast(n) + 1, 0); break; default: s = Status::Corruption("unknown checksum type"); } if (s.ok() && actual != value) { s = Status::Corruption("block checksum mismatch"); } if (!s.ok()) { return s; } } return s; } } // namespace Status ReadBlockContents(RandomAccessFileReader* file, const Footer& footer, const ReadOptions& options, const BlockHandle& handle, BlockContents* contents, Env* env, bool decompression_requested) { Status status; Slice slice; size_t n = static_cast(handle.size()); std::unique_ptr heap_buf; char stack_buf[DefaultStackBufferSize]; char* used_buf = nullptr; rocksdb::CompressionType compression_type; if (decompression_requested && n + kBlockTrailerSize < DefaultStackBufferSize) { // If we've got a small enough hunk of data, read it in to the // trivially allocated stack buffer instead of needing a full malloc() used_buf = &stack_buf[0]; } else { heap_buf = std::unique_ptr(new char[n + kBlockTrailerSize]); used_buf = heap_buf.get(); } status = ReadBlock(file, footer, options, handle, &slice, used_buf); if (!status.ok()) { return status; } PERF_TIMER_GUARD(block_decompress_time); compression_type = static_cast(slice.data()[n]); if (decompression_requested && compression_type != kNoCompression) { return UncompressBlockContents(slice.data(), n, contents, footer.version()); } if (slice.data() != used_buf) { *contents = BlockContents(Slice(slice.data(), n), false, compression_type); return status; } if (used_buf == &stack_buf[0]) { heap_buf = std::unique_ptr(new char[n]); memcpy(heap_buf.get(), stack_buf, n); } *contents = BlockContents(std::move(heap_buf), n, true, compression_type); return status; } // // The 'data' points to the raw block contents that was read in from file. // This method allocates a new heap buffer and the raw block // contents are uncompresed into this buffer. This // buffer is returned via 'result' and it is upto the caller to // free this buffer. // format_version is the block format as defined in include/rocksdb/table.h Status UncompressBlockContents(const char* data, size_t n, BlockContents* contents, uint32_t format_version) { std::unique_ptr ubuf; int decompress_size = 0; assert(data[n] != kNoCompression); switch (data[n]) { case kSnappyCompression: { size_t ulength = 0; static char snappy_corrupt_msg[] = "Snappy not supported or corrupted Snappy compressed block contents"; if (!Snappy_GetUncompressedLength(data, n, &ulength)) { return Status::Corruption(snappy_corrupt_msg); } ubuf = std::unique_ptr(new char[ulength]); if (!Snappy_Uncompress(data, n, ubuf.get())) { return Status::Corruption(snappy_corrupt_msg); } *contents = BlockContents(std::move(ubuf), ulength, true, kNoCompression); break; } case kZlibCompression: ubuf = std::unique_ptr(Zlib_Uncompress( data, n, &decompress_size, GetCompressFormatForVersion(kZlibCompression, format_version))); if (!ubuf) { static char zlib_corrupt_msg[] = "Zlib not supported or corrupted Zlib compressed block contents"; return Status::Corruption(zlib_corrupt_msg); } *contents = BlockContents(std::move(ubuf), decompress_size, true, kNoCompression); break; case kBZip2Compression: ubuf = std::unique_ptr(BZip2_Uncompress( data, n, &decompress_size, GetCompressFormatForVersion(kBZip2Compression, format_version))); if (!ubuf) { static char bzip2_corrupt_msg[] = "Bzip2 not supported or corrupted Bzip2 compressed block contents"; return Status::Corruption(bzip2_corrupt_msg); } *contents = BlockContents(std::move(ubuf), decompress_size, true, kNoCompression); break; case kLZ4Compression: ubuf = std::unique_ptr(LZ4_Uncompress( data, n, &decompress_size, GetCompressFormatForVersion(kLZ4Compression, format_version))); if (!ubuf) { static char lz4_corrupt_msg[] = "LZ4 not supported or corrupted LZ4 compressed block contents"; return Status::Corruption(lz4_corrupt_msg); } *contents = BlockContents(std::move(ubuf), decompress_size, true, kNoCompression); break; case kLZ4HCCompression: ubuf = std::unique_ptr(LZ4_Uncompress( data, n, &decompress_size, GetCompressFormatForVersion(kLZ4HCCompression, format_version))); if (!ubuf) { static char lz4hc_corrupt_msg[] = "LZ4HC not supported or corrupted LZ4HC compressed block contents"; return Status::Corruption(lz4hc_corrupt_msg); } *contents = BlockContents(std::move(ubuf), decompress_size, true, kNoCompression); break; case kZSTDNotFinalCompression: ubuf = std::unique_ptr(ZSTD_Uncompress(data, n, &decompress_size)); if (!ubuf) { static char zstd_corrupt_msg[] = "ZSTD not supported or corrupted ZSTD compressed block contents"; return Status::Corruption(zstd_corrupt_msg); } *contents = BlockContents(std::move(ubuf), decompress_size, true, kNoCompression); break; default: return Status::Corruption("bad block type"); } return Status::OK(); } } // namespace rocksdb rocksdb-4.5.1/table/format.h000066400000000000000000000206641267530634000157160ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include "rocksdb/slice.h" #include "rocksdb/status.h" #include "rocksdb/options.h" #include "rocksdb/table.h" namespace rocksdb { class Block; class RandomAccessFile; struct ReadOptions; // the length of the magic number in bytes. const int kMagicNumberLengthByte = 8; // BlockHandle is a pointer to the extent of a file that stores a data // block or a meta block. class BlockHandle { public: BlockHandle(); BlockHandle(uint64_t offset, uint64_t size); // The offset of the block in the file. uint64_t offset() const { return offset_; } void set_offset(uint64_t _offset) { offset_ = _offset; } // The size of the stored block uint64_t size() const { return size_; } void set_size(uint64_t _size) { size_ = _size; } void EncodeTo(std::string* dst) const; Status DecodeFrom(Slice* input); // Return a string that contains the copy of handle. std::string ToString(bool hex = true) const; // if the block handle's offset and size are both "0", we will view it // as a null block handle that points to no where. bool IsNull() const { return offset_ == 0 && size_ == 0; } static const BlockHandle& NullBlockHandle() { return kNullBlockHandle; } // Maximum encoding length of a BlockHandle enum { kMaxEncodedLength = 10 + 10 }; private: uint64_t offset_ = 0; uint64_t size_ = 0; static const BlockHandle kNullBlockHandle; }; inline uint32_t GetCompressFormatForVersion(CompressionType compression_type, uint32_t version) { // snappy is not versioned assert(compression_type != kSnappyCompression && compression_type != kNoCompression); // As of version 2, we encode compressed block with // compress_format_version == 2. Before that, the version is 1. // DO NOT CHANGE THIS FUNCTION, it affects disk format return version >= 2 ? 2 : 1; } inline bool BlockBasedTableSupportedVersion(uint32_t version) { return version <= 2; } // Footer encapsulates the fixed information stored at the tail // end of every table file. class Footer { public: // Constructs a footer without specifying its table magic number. // In such case, the table magic number of such footer should be // initialized via @ReadFooterFromFile(). // Use this when you plan to load Footer with DecodeFrom(). Never use this // when you plan to EncodeTo. Footer() : Footer(kInvalidTableMagicNumber, 0) {} // Use this constructor when you plan to write out the footer using // EncodeTo(). Never use this constructor with DecodeFrom(). Footer(uint64_t table_magic_number, uint32_t version); // The version of the footer in this file uint32_t version() const { return version_; } // The checksum type used in this file ChecksumType checksum() const { return checksum_; } void set_checksum(const ChecksumType c) { checksum_ = c; } // The block handle for the metaindex block of the table const BlockHandle& metaindex_handle() const { return metaindex_handle_; } void set_metaindex_handle(const BlockHandle& h) { metaindex_handle_ = h; } // The block handle for the index block of the table const BlockHandle& index_handle() const { return index_handle_; } void set_index_handle(const BlockHandle& h) { index_handle_ = h; } uint64_t table_magic_number() const { return table_magic_number_; } void EncodeTo(std::string* dst) const; // Set the current footer based on the input slice. // // REQUIRES: table_magic_number_ is not set (i.e., // HasInitializedTableMagicNumber() is true). The function will initialize the // magic number Status DecodeFrom(Slice* input); // Encoded length of a Footer. Note that the serialization of a Footer will // always occupy at least kMinEncodedLength bytes. If fields are changed // the version number should be incremented and kMaxEncodedLength should be // increased accordingly. enum { // Footer version 0 (legacy) will always occupy exactly this many bytes. // It consists of two block handles, padding, and a magic number. kVersion0EncodedLength = 2 * BlockHandle::kMaxEncodedLength + 8, // Footer of versions 1 and higher will always occupy exactly this many // bytes. It consists of the checksum type, two block handles, padding, // a version number (bigger than 1), and a magic number kNewVersionsEncodedLength = 1 + 2 * BlockHandle::kMaxEncodedLength + 4 + 8, kMinEncodedLength = kVersion0EncodedLength, kMaxEncodedLength = kNewVersionsEncodedLength, }; static const uint64_t kInvalidTableMagicNumber = 0; // convert this object to a human readable form std::string ToString() const; private: // REQUIRES: magic number wasn't initialized. void set_table_magic_number(uint64_t magic_number) { assert(!HasInitializedTableMagicNumber()); table_magic_number_ = magic_number; } // return true if @table_magic_number_ is set to a value different // from @kInvalidTableMagicNumber. bool HasInitializedTableMagicNumber() const { return (table_magic_number_ != kInvalidTableMagicNumber); } uint32_t version_; ChecksumType checksum_; BlockHandle metaindex_handle_; BlockHandle index_handle_; uint64_t table_magic_number_ = 0; }; // Read the footer from file // If enforce_table_magic_number != 0, ReadFooterFromFile() will return // corruption if table_magic number is not equal to enforce_table_magic_number Status ReadFooterFromFile(RandomAccessFileReader* file, uint64_t file_size, Footer* footer, uint64_t enforce_table_magic_number = 0); // 1-byte type + 32-bit crc static const size_t kBlockTrailerSize = 5; struct BlockContents { Slice data; // Actual contents of data bool cachable; // True iff data can be cached CompressionType compression_type; std::unique_ptr allocation; BlockContents() : cachable(false), compression_type(kNoCompression) {} BlockContents(const Slice& _data, bool _cachable, CompressionType _compression_type) : data(_data), cachable(_cachable), compression_type(_compression_type) {} BlockContents(std::unique_ptr&& _data, size_t _size, bool _cachable, CompressionType _compression_type) : data(_data.get(), _size), cachable(_cachable), compression_type(_compression_type), allocation(std::move(_data)) {} BlockContents(BlockContents&& other) { *this = std::move(other); } BlockContents& operator=(BlockContents&& other) { data = std::move(other.data); cachable = other.cachable; compression_type = other.compression_type; allocation = std::move(other.allocation); return *this; } }; // Read the block identified by "handle" from "file". On failure // return non-OK. On success fill *result and return OK. extern Status ReadBlockContents(RandomAccessFileReader* file, const Footer& footer, const ReadOptions& options, const BlockHandle& handle, BlockContents* contents, Env* env, bool do_uncompress); // The 'data' points to the raw block contents read in from file. // This method allocates a new heap buffer and the raw block // contents are uncompresed into this buffer. This buffer is // returned via 'result' and it is upto the caller to // free this buffer. // For description of compress_format_version and possible values, see // util/compression.h extern Status UncompressBlockContents(const char* data, size_t n, BlockContents* contents, uint32_t compress_format_version); // Implementation details follow. Clients should ignore, inline BlockHandle::BlockHandle() : BlockHandle(~static_cast(0), ~static_cast(0)) { } inline BlockHandle::BlockHandle(uint64_t _offset, uint64_t _size) : offset_(_offset), size_(_size) {} } // namespace rocksdb rocksdb-4.5.1/table/full_filter_block.cc000066400000000000000000000062561267530634000202460ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/full_filter_block.h" #include "rocksdb/filter_policy.h" #include "port/port.h" #include "util/coding.h" #include "util/perf_context_imp.h" namespace rocksdb { FullFilterBlockBuilder::FullFilterBlockBuilder( const SliceTransform* prefix_extractor, bool whole_key_filtering, FilterBitsBuilder* filter_bits_builder) : prefix_extractor_(prefix_extractor), whole_key_filtering_(whole_key_filtering), num_added_(0) { assert(filter_bits_builder != nullptr); filter_bits_builder_.reset(filter_bits_builder); } void FullFilterBlockBuilder::Add(const Slice& key) { if (whole_key_filtering_) { AddKey(key); } if (prefix_extractor_ && prefix_extractor_->InDomain(key)) { AddPrefix(key); } } // Add key to filter if needed inline void FullFilterBlockBuilder::AddKey(const Slice& key) { filter_bits_builder_->AddKey(key); num_added_++; } // Add prefix to filter if needed inline void FullFilterBlockBuilder::AddPrefix(const Slice& key) { Slice prefix = prefix_extractor_->Transform(key); filter_bits_builder_->AddKey(prefix); num_added_++; } Slice FullFilterBlockBuilder::Finish() { if (num_added_ != 0) { num_added_ = 0; return filter_bits_builder_->Finish(&filter_data_); } return Slice(); } FullFilterBlockReader::FullFilterBlockReader( const SliceTransform* prefix_extractor, bool whole_key_filtering, const Slice& contents, FilterBitsReader* filter_bits_reader) : prefix_extractor_(prefix_extractor), whole_key_filtering_(whole_key_filtering), contents_(contents) { assert(filter_bits_reader != nullptr); filter_bits_reader_.reset(filter_bits_reader); } FullFilterBlockReader::FullFilterBlockReader( const SliceTransform* prefix_extractor, bool whole_key_filtering, BlockContents&& contents, FilterBitsReader* filter_bits_reader) : FullFilterBlockReader(prefix_extractor, whole_key_filtering, contents.data, filter_bits_reader) { block_contents_ = std::move(contents); } bool FullFilterBlockReader::KeyMayMatch(const Slice& key, uint64_t block_offset) { assert(block_offset == kNotValid); if (!whole_key_filtering_) { return true; } return MayMatch(key); } bool FullFilterBlockReader::PrefixMayMatch(const Slice& prefix, uint64_t block_offset) { assert(block_offset == kNotValid); if (!prefix_extractor_) { return true; } return MayMatch(prefix); } bool FullFilterBlockReader::MayMatch(const Slice& entry) { if (contents_.size() != 0) { if (filter_bits_reader_->MayMatch(entry)) { PERF_COUNTER_ADD(bloom_sst_hit_count, 1); return true; } else { PERF_COUNTER_ADD(bloom_sst_miss_count, 1); return false; } } return true; // remain the same with block_based filter } size_t FullFilterBlockReader::ApproximateMemoryUsage() const { return contents_.size(); } } // namespace rocksdb rocksdb-4.5.1/table/full_filter_block.h000066400000000000000000000100701267530634000200750ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include #include #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "rocksdb/slice_transform.h" #include "db/dbformat.h" #include "util/hash.h" #include "table/filter_block.h" namespace rocksdb { class FilterPolicy; class FilterBitsBuilder; class FilterBitsReader; // A FullFilterBlockBuilder is used to construct a full filter for a // particular Table. It generates a single string which is stored as // a special block in the Table. // The format of full filter block is: // +----------------------------------------------------------------+ // | full filter for all keys in sst file | // +----------------------------------------------------------------+ // The full filter can be very large. At the end of it, we put // num_probes: how many hash functions are used in bloom filter // class FullFilterBlockBuilder : public FilterBlockBuilder { public: explicit FullFilterBlockBuilder(const SliceTransform* prefix_extractor, bool whole_key_filtering, FilterBitsBuilder* filter_bits_builder); // bits_builder is created in filter_policy, it should be passed in here // directly. and be deleted here ~FullFilterBlockBuilder() {} virtual bool IsBlockBased() override { return false; } virtual void StartBlock(uint64_t block_offset) override {} virtual void Add(const Slice& key) override; virtual Slice Finish() override; private: // important: all of these might point to invalid addresses // at the time of destruction of this filter block. destructor // should NOT dereference them. const SliceTransform* prefix_extractor_; bool whole_key_filtering_; uint32_t num_added_; std::unique_ptr filter_bits_builder_; std::unique_ptr filter_data_; void AddKey(const Slice& key); void AddPrefix(const Slice& key); // No copying allowed FullFilterBlockBuilder(const FullFilterBlockBuilder&); void operator=(const FullFilterBlockBuilder&); }; // A FilterBlockReader is used to parse filter from SST table. // KeyMayMatch and PrefixMayMatch would trigger filter checking class FullFilterBlockReader : public FilterBlockReader { public: // REQUIRES: "contents" and filter_bits_reader must stay live // while *this is live. explicit FullFilterBlockReader(const SliceTransform* prefix_extractor, bool whole_key_filtering, const Slice& contents, FilterBitsReader* filter_bits_reader); explicit FullFilterBlockReader(const SliceTransform* prefix_extractor, bool whole_key_filtering, BlockContents&& contents, FilterBitsReader* filter_bits_reader); // bits_reader is created in filter_policy, it should be passed in here // directly. and be deleted here ~FullFilterBlockReader() {} virtual bool IsBlockBased() override { return false; } virtual bool KeyMayMatch(const Slice& key, uint64_t block_offset = kNotValid) override; virtual bool PrefixMayMatch(const Slice& prefix, uint64_t block_offset = kNotValid) override; virtual size_t ApproximateMemoryUsage() const override; private: const SliceTransform* prefix_extractor_; bool whole_key_filtering_; std::unique_ptr filter_bits_reader_; Slice contents_; BlockContents block_contents_; std::unique_ptr filter_data_; bool MayMatch(const Slice& entry); // No copying allowed FullFilterBlockReader(const FullFilterBlockReader&); void operator=(const FullFilterBlockReader&); }; } // namespace rocksdb rocksdb-4.5.1/table/full_filter_block_test.cc000066400000000000000000000130741267530634000213010ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/full_filter_block.h" #include "rocksdb/filter_policy.h" #include "util/coding.h" #include "util/hash.h" #include "util/logging.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class TestFilterBitsBuilder : public FilterBitsBuilder { public: explicit TestFilterBitsBuilder() {} // Add Key to filter virtual void AddKey(const Slice& key) override { hash_entries_.push_back(Hash(key.data(), key.size(), 1)); } // Generate the filter using the keys that are added virtual Slice Finish(std::unique_ptr* buf) override { uint32_t len = static_cast(hash_entries_.size()) * 4; char* data = new char[len]; for (size_t i = 0; i < hash_entries_.size(); i++) { EncodeFixed32(data + i * 4, hash_entries_[i]); } const char* const_data = data; buf->reset(const_data); return Slice(data, len); } private: std::vector hash_entries_; }; class TestFilterBitsReader : public FilterBitsReader { public: explicit TestFilterBitsReader(const Slice& contents) : data_(contents.data()), len_(static_cast(contents.size())) {} virtual bool MayMatch(const Slice& entry) override { uint32_t h = Hash(entry.data(), entry.size(), 1); for (size_t i = 0; i + 4 <= len_; i += 4) { if (h == DecodeFixed32(data_ + i)) { return true; } } return false; } private: const char* data_; uint32_t len_; }; class TestHashFilter : public FilterPolicy { public: virtual const char* Name() const override { return "TestHashFilter"; } virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const override { for (int i = 0; i < n; i++) { uint32_t h = Hash(keys[i].data(), keys[i].size(), 1); PutFixed32(dst, h); } } virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const override { uint32_t h = Hash(key.data(), key.size(), 1); for (unsigned int i = 0; i + 4 <= filter.size(); i += 4) { if (h == DecodeFixed32(filter.data() + i)) { return true; } } return false; } virtual FilterBitsBuilder* GetFilterBitsBuilder() const override { return new TestFilterBitsBuilder(); } virtual FilterBitsReader* GetFilterBitsReader(const Slice& contents) const override { return new TestFilterBitsReader(contents); } }; class PluginFullFilterBlockTest : public testing::Test { public: BlockBasedTableOptions table_options_; PluginFullFilterBlockTest() { table_options_.filter_policy.reset(new TestHashFilter()); } }; TEST_F(PluginFullFilterBlockTest, PluginEmptyBuilder) { FullFilterBlockBuilder builder( nullptr, true, table_options_.filter_policy->GetFilterBitsBuilder()); Slice block = builder.Finish(); ASSERT_EQ("", EscapeString(block)); FullFilterBlockReader reader( nullptr, true, block, table_options_.filter_policy->GetFilterBitsReader(block)); // Remain same symantic with blockbased filter ASSERT_TRUE(reader.KeyMayMatch("foo")); } TEST_F(PluginFullFilterBlockTest, PluginSingleChunk) { FullFilterBlockBuilder builder( nullptr, true, table_options_.filter_policy->GetFilterBitsBuilder()); builder.Add("foo"); builder.Add("bar"); builder.Add("box"); builder.Add("box"); builder.Add("hello"); Slice block = builder.Finish(); FullFilterBlockReader reader( nullptr, true, block, table_options_.filter_policy->GetFilterBitsReader(block)); ASSERT_TRUE(reader.KeyMayMatch("foo")); ASSERT_TRUE(reader.KeyMayMatch("bar")); ASSERT_TRUE(reader.KeyMayMatch("box")); ASSERT_TRUE(reader.KeyMayMatch("hello")); ASSERT_TRUE(reader.KeyMayMatch("foo")); ASSERT_TRUE(!reader.KeyMayMatch("missing")); ASSERT_TRUE(!reader.KeyMayMatch("other")); } class FullFilterBlockTest : public testing::Test { public: BlockBasedTableOptions table_options_; FullFilterBlockTest() { table_options_.filter_policy.reset(NewBloomFilterPolicy(10, false)); } ~FullFilterBlockTest() {} }; TEST_F(FullFilterBlockTest, EmptyBuilder) { FullFilterBlockBuilder builder( nullptr, true, table_options_.filter_policy->GetFilterBitsBuilder()); Slice block = builder.Finish(); ASSERT_EQ("", EscapeString(block)); FullFilterBlockReader reader( nullptr, true, block, table_options_.filter_policy->GetFilterBitsReader(block)); // Remain same symantic with blockbased filter ASSERT_TRUE(reader.KeyMayMatch("foo")); } TEST_F(FullFilterBlockTest, SingleChunk) { FullFilterBlockBuilder builder( nullptr, true, table_options_.filter_policy->GetFilterBitsBuilder()); builder.Add("foo"); builder.Add("bar"); builder.Add("box"); builder.Add("box"); builder.Add("hello"); Slice block = builder.Finish(); FullFilterBlockReader reader( nullptr, true, block, table_options_.filter_policy->GetFilterBitsReader(block)); ASSERT_TRUE(reader.KeyMayMatch("foo")); ASSERT_TRUE(reader.KeyMayMatch("bar")); ASSERT_TRUE(reader.KeyMayMatch("box")); ASSERT_TRUE(reader.KeyMayMatch("hello")); ASSERT_TRUE(reader.KeyMayMatch("foo")); ASSERT_TRUE(!reader.KeyMayMatch("missing")); ASSERT_TRUE(!reader.KeyMayMatch("other")); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/table/get_context.cc000066400000000000000000000137561267530634000171130ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/get_context.h" #include "rocksdb/env.h" #include "rocksdb/merge_operator.h" #include "rocksdb/statistics.h" #include "util/perf_context_imp.h" #include "util/statistics.h" namespace rocksdb { namespace { void appendToReplayLog(std::string* replay_log, ValueType type, Slice value) { #ifndef ROCKSDB_LITE if (replay_log) { if (replay_log->empty()) { // Optimization: in the common case of only one operation in the // log, we allocate the exact amount of space needed. replay_log->reserve(1 + VarintLength(value.size()) + value.size()); } replay_log->push_back(type); PutLengthPrefixedSlice(replay_log, value); } #endif // ROCKSDB_LITE } } // namespace GetContext::GetContext(const Comparator* ucmp, const MergeOperator* merge_operator, Logger* logger, Statistics* statistics, GetState init_state, const Slice& user_key, std::string* ret_value, bool* value_found, MergeContext* merge_context, Env* env, SequenceNumber* seq) : ucmp_(ucmp), merge_operator_(merge_operator), logger_(logger), statistics_(statistics), state_(init_state), user_key_(user_key), value_(ret_value), value_found_(value_found), merge_context_(merge_context), env_(env), seq_(seq), replay_log_(nullptr) { if (seq_) { *seq_ = kMaxSequenceNumber; } } // Called from TableCache::Get and Table::Get when file/block in which // key may exist are not there in TableCache/BlockCache respectively. In this // case we can't guarantee that key does not exist and are not permitted to do // IO to be certain.Set the status=kFound and value_found=false to let the // caller know that key may exist but is not there in memory void GetContext::MarkKeyMayExist() { state_ = kFound; if (value_found_ != nullptr) { *value_found_ = false; } } void GetContext::SaveValue(const Slice& value, SequenceNumber seq) { assert(state_ == kNotFound); appendToReplayLog(replay_log_, kTypeValue, value); state_ = kFound; if (value_ != nullptr) { value_->assign(value.data(), value.size()); } } bool GetContext::SaveValue(const ParsedInternalKey& parsed_key, const Slice& value) { assert((state_ != kMerge && parsed_key.type != kTypeMerge) || merge_context_ != nullptr); if (ucmp_->Equal(parsed_key.user_key, user_key_)) { appendToReplayLog(replay_log_, parsed_key.type, value); if (seq_ != nullptr) { // Set the sequence number if it is uninitialized if (*seq_ == kMaxSequenceNumber) { *seq_ = parsed_key.sequence; } } // Key matches. Process it switch (parsed_key.type) { case kTypeValue: assert(state_ == kNotFound || state_ == kMerge); if (kNotFound == state_) { state_ = kFound; if (value_ != nullptr) { value_->assign(value.data(), value.size()); } } else if (kMerge == state_) { assert(merge_operator_ != nullptr); state_ = kFound; if (value_ != nullptr) { bool merge_success = false; { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); merge_success = merge_operator_->FullMerge( user_key_, &value, merge_context_->GetOperands(), value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanosSafe()); } if (!merge_success) { RecordTick(statistics_, NUMBER_MERGE_FAILURES); state_ = kCorrupt; } } } return false; case kTypeDeletion: case kTypeSingleDeletion: // TODO(noetzli): Verify correctness once merge of single-deletes // is supported assert(state_ == kNotFound || state_ == kMerge); if (kNotFound == state_) { state_ = kDeleted; } else if (kMerge == state_) { state_ = kFound; if (value_ != nullptr) { bool merge_success = false; { StopWatchNano timer(env_, statistics_ != nullptr); PERF_TIMER_GUARD(merge_operator_time_nanos); merge_success = merge_operator_->FullMerge( user_key_, nullptr, merge_context_->GetOperands(), value_, logger_); RecordTick(statistics_, MERGE_OPERATION_TOTAL_TIME, timer.ElapsedNanosSafe()); } if (!merge_success) { RecordTick(statistics_, NUMBER_MERGE_FAILURES); state_ = kCorrupt; } } } return false; case kTypeMerge: assert(state_ == kNotFound || state_ == kMerge); state_ = kMerge; merge_context_->PushOperand(value); return true; default: assert(false); break; } } // state_ could be Corrupt, merge or notfound return false; } void replayGetContextLog(const Slice& replay_log, const Slice& user_key, GetContext* get_context) { #ifndef ROCKSDB_LITE Slice s = replay_log; while (s.size()) { auto type = static_cast(*s.data()); s.remove_prefix(1); Slice value; bool ret = GetLengthPrefixedSlice(&s, &value); assert(ret); (void)ret; // Since SequenceNumber is not stored and unknown, we will use // kMaxSequenceNumber. get_context->SaveValue( ParsedInternalKey(user_key, kMaxSequenceNumber, type), value); } #else // ROCKSDB_LITE assert(false); #endif // ROCKSDB_LITE } } // namespace rocksdb rocksdb-4.5.1/table/get_context.h000066400000000000000000000050201267530634000167360ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include "db/merge_context.h" #include "rocksdb/env.h" #include "rocksdb/types.h" namespace rocksdb { class MergeContext; class GetContext { public: enum GetState { kNotFound, kFound, kDeleted, kCorrupt, kMerge // saver contains the current merge result (the operands) }; GetContext(const Comparator* ucmp, const MergeOperator* merge_operator, Logger* logger, Statistics* statistics, GetState init_state, const Slice& user_key, std::string* ret_value, bool* value_found, MergeContext* merge_context, Env* env, SequenceNumber* seq = nullptr); void MarkKeyMayExist(); // Records this key, value, and any meta-data (such as sequence number and // state) into this GetContext. // // Returns True if more keys need to be read (due to merges) or // False if the complete value has been found. bool SaveValue(const ParsedInternalKey& parsed_key, const Slice& value); // Simplified version of the previous function. Should only be used when we // know that the operation is a Put. void SaveValue(const Slice& value, SequenceNumber seq); GetState State() const { return state_; } // If a non-null string is passed, all the SaveValue calls will be // logged into the string. The operations can then be replayed on // another GetContext with replayGetContextLog. void SetReplayLog(std::string* replay_log) { replay_log_ = replay_log; } // Do we need to fetch the SequenceNumber for this key? bool NeedToReadSequence() const { return (seq_ != nullptr); } private: const Comparator* ucmp_; const MergeOperator* merge_operator_; // the merge operations encountered; Logger* logger_; Statistics* statistics_; GetState state_; Slice user_key_; std::string* value_; bool* value_found_; // Is value set correctly? Used by KeyMayExist MergeContext* merge_context_; Env* env_; // If a key is found, seq_ will be set to the SequenceNumber of most recent // write to the key or kMaxSequenceNumber if unknown SequenceNumber* seq_; std::string* replay_log_; }; void replayGetContextLog(const Slice& replay_log, const Slice& user_key, GetContext* get_context); } // namespace rocksdb rocksdb-4.5.1/table/internal_iterator.h000066400000000000000000000067571267530634000201620ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #pragma once #include "rocksdb/iterator.h" #include "rocksdb/status.h" namespace rocksdb { class InternalIterator : public Cleanable { public: InternalIterator() {} virtual ~InternalIterator() {} // An iterator is either positioned at a key/value pair, or // not valid. This method returns true iff the iterator is valid. virtual bool Valid() const = 0; // Position at the first key in the source. The iterator is Valid() // after this call iff the source is not empty. virtual void SeekToFirst() = 0; // Position at the last key in the source. The iterator is // Valid() after this call iff the source is not empty. virtual void SeekToLast() = 0; // Position at the first key in the source that at or past target // The iterator is Valid() after this call iff the source contains // an entry that comes at or past target. virtual void Seek(const Slice& target) = 0; // Moves to the next entry in the source. After this call, Valid() is // true iff the iterator was not positioned at the last entry in the source. // REQUIRES: Valid() virtual void Next() = 0; // Moves to the previous entry in the source. After this call, Valid() is // true iff the iterator was not positioned at the first entry in source. // REQUIRES: Valid() virtual void Prev() = 0; // Return the key for the current entry. The underlying storage for // the returned slice is valid only until the next modification of // the iterator. // REQUIRES: Valid() virtual Slice key() const = 0; // Return the value for the current entry. The underlying storage for // the returned slice is valid only until the next modification of // the iterator. // REQUIRES: !AtEnd() && !AtStart() virtual Slice value() const = 0; // If an error has occurred, return it. Else return an ok status. // If non-blocking IO is requested and this operation cannot be // satisfied without doing some IO, then this returns Status::Incomplete(). virtual Status status() const = 0; // Make sure that all current and future data blocks used by this iterator // will be pinned in memory and will not be released except when // ReleasePinnedData() is called or the iterator is deleted. virtual Status PinData() { return Status::NotSupported(""); } // Release all blocks that were pinned because of PinData() and no future // blocks will be pinned. virtual Status ReleasePinnedData() { return Status::NotSupported(""); } // If true, this means that the Slice returned by key() is valid as long // as the iterator is not deleted and ReleasePinnedData() is not called. // // IsKeyPinned() is guaranteed to always return true if // - PinData() is called // - DB tables were created with BlockBasedTableOptions::use_delta_encoding // set to false. virtual bool IsKeyPinned() const { return false; } private: // No copying allowed InternalIterator(const InternalIterator&) = delete; InternalIterator& operator=(const InternalIterator&) = delete; }; // Return an empty iterator (yields nothing). extern InternalIterator* NewEmptyInternalIterator(); // Return an empty iterator with the specified status. extern InternalIterator* NewErrorInternalIterator(const Status& status); } // namespace rocksdb rocksdb-4.5.1/table/iter_heap.h000066400000000000000000000023551267530634000163630ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // #pragma once #include "rocksdb/comparator.h" #include "table/iterator_wrapper.h" namespace rocksdb { // When used with std::priority_queue, this comparison functor puts the // iterator with the max/largest key on top. class MaxIteratorComparator { public: MaxIteratorComparator(const Comparator* comparator) : comparator_(comparator) {} bool operator()(IteratorWrapper* a, IteratorWrapper* b) const { return comparator_->Compare(a->key(), b->key()) < 0; } private: const Comparator* comparator_; }; // When used with std::priority_queue, this comparison functor puts the // iterator with the min/smallest key on top. class MinIteratorComparator { public: MinIteratorComparator(const Comparator* comparator) : comparator_(comparator) {} bool operator()(IteratorWrapper* a, IteratorWrapper* b) const { return comparator_->Compare(a->key(), b->key()) > 0; } private: const Comparator* comparator_; }; } // namespace rocksdb rocksdb-4.5.1/table/iterator.cc000066400000000000000000000071021267530634000164050ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "rocksdb/iterator.h" #include "table/internal_iterator.h" #include "table/iterator_wrapper.h" #include "util/arena.h" namespace rocksdb { Cleanable::Cleanable() { cleanup_.function = nullptr; cleanup_.next = nullptr; } Cleanable::~Cleanable() { if (cleanup_.function != nullptr) { (*cleanup_.function)(cleanup_.arg1, cleanup_.arg2); for (Cleanup* c = cleanup_.next; c != nullptr; ) { (*c->function)(c->arg1, c->arg2); Cleanup* next = c->next; delete c; c = next; } } } void Cleanable::RegisterCleanup(CleanupFunction func, void* arg1, void* arg2) { assert(func != nullptr); Cleanup* c; if (cleanup_.function == nullptr) { c = &cleanup_; } else { c = new Cleanup; c->next = cleanup_.next; cleanup_.next = c; } c->function = func; c->arg1 = arg1; c->arg2 = arg2; } namespace { class EmptyIterator : public Iterator { public: explicit EmptyIterator(const Status& s) : status_(s) { } virtual bool Valid() const override { return false; } virtual void Seek(const Slice& target) override {} virtual void SeekToFirst() override {} virtual void SeekToLast() override {} virtual void Next() override { assert(false); } virtual void Prev() override { assert(false); } Slice key() const override { assert(false); return Slice(); } Slice value() const override { assert(false); return Slice(); } virtual Status status() const override { return status_; } private: Status status_; }; class EmptyInternalIterator : public InternalIterator { public: explicit EmptyInternalIterator(const Status& s) : status_(s) {} virtual bool Valid() const override { return false; } virtual void Seek(const Slice& target) override {} virtual void SeekToFirst() override {} virtual void SeekToLast() override {} virtual void Next() override { assert(false); } virtual void Prev() override { assert(false); } Slice key() const override { assert(false); return Slice(); } Slice value() const override { assert(false); return Slice(); } virtual Status status() const override { return status_; } private: Status status_; }; } // namespace Iterator* NewEmptyIterator() { return new EmptyIterator(Status::OK()); } Iterator* NewErrorIterator(const Status& status) { return new EmptyIterator(status); } InternalIterator* NewEmptyInternalIterator() { return new EmptyInternalIterator(Status::OK()); } InternalIterator* NewEmptyInternalIterator(Arena* arena) { if (arena == nullptr) { return NewEmptyInternalIterator(); } else { auto mem = arena->AllocateAligned(sizeof(EmptyIterator)); return new (mem) EmptyInternalIterator(Status::OK()); } } InternalIterator* NewErrorInternalIterator(const Status& status) { return new EmptyInternalIterator(status); } InternalIterator* NewErrorInternalIterator(const Status& status, Arena* arena) { if (arena == nullptr) { return NewErrorInternalIterator(status); } else { auto mem = arena->AllocateAligned(sizeof(EmptyIterator)); return new (mem) EmptyInternalIterator(status); } } } // namespace rocksdb rocksdb-4.5.1/table/iterator_wrapper.h000066400000000000000000000107451267530634000200160ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include "table/internal_iterator.h" namespace rocksdb { // A internal wrapper class with an interface similar to Iterator that // caches the valid() and key() results for an underlying iterator. // This can help avoid virtual function calls and also gives better // cache locality. class IteratorWrapper { public: IteratorWrapper() : iter_(nullptr), iters_pinned_(false), valid_(false) {} explicit IteratorWrapper(InternalIterator* _iter) : iter_(nullptr), iters_pinned_(false) { Set(_iter); } ~IteratorWrapper() {} InternalIterator* iter() const { return iter_; } // Takes the ownership of "_iter" and will delete it when destroyed. // Next call to Set() will destroy "_iter" except if PinData() was called. void Set(InternalIterator* _iter) { if (iters_pinned_ && iter_) { // keep old iterator until ReleasePinnedData() is called pinned_iters_.insert(iter_); } else { delete iter_; } iter_ = _iter; if (iter_ == nullptr) { valid_ = false; } else { Update(); if (iters_pinned_) { // Pin new iterator Status s = iter_->PinData(); assert(s.ok()); } } } Status PinData() { Status s; if (iters_pinned_) { return s; } if (iter_) { s = iter_->PinData(); } if (s.ok()) { iters_pinned_ = true; } return s; } Status ReleasePinnedData() { Status s; if (!iters_pinned_) { return s; } if (iter_) { s = iter_->ReleasePinnedData(); } if (s.ok()) { iters_pinned_ = false; // No need to call ReleasePinnedData() for pinned_iters_ // since we will delete them DeletePinnedIterators(false); } return s; } bool IsKeyPinned() const { assert(iter_); return iters_pinned_ && iter_->IsKeyPinned(); } void DeleteIter(bool is_arena_mode) { if (iter_ && pinned_iters_.find(iter_) == pinned_iters_.end()) { DestroyIterator(iter_, is_arena_mode); } DeletePinnedIterators(is_arena_mode); } // Iterator interface methods bool Valid() const { return valid_; } Slice key() const { assert(Valid()); return key_; } Slice value() const { assert(Valid()); return iter_->value(); } // Methods below require iter() != nullptr Status status() const { assert(iter_); return iter_->status(); } void Next() { assert(iter_); iter_->Next(); Update(); } void Prev() { assert(iter_); iter_->Prev(); Update(); } void Seek(const Slice& k) { assert(iter_); iter_->Seek(k); Update(); } void SeekToFirst() { assert(iter_); iter_->SeekToFirst(); Update(); } void SeekToLast() { assert(iter_); iter_->SeekToLast(); Update(); } private: void Update() { valid_ = iter_->Valid(); if (valid_) { key_ = iter_->key(); } } void DeletePinnedIterators(bool is_arena_mode) { for (auto it : pinned_iters_) { DestroyIterator(it, is_arena_mode); } pinned_iters_.clear(); } inline void DestroyIterator(InternalIterator* it, bool is_arena_mode) { if (!is_arena_mode) { delete it; } else { it->~InternalIterator(); } } InternalIterator* iter_; // If set to true, current and future iterators wont be deleted. bool iters_pinned_; // List of past iterators that are pinned and wont be deleted as long as // iters_pinned_ is true. When we are pinning iterators this set will contain // iterators of previous data blocks to keep them from being deleted. std::set pinned_iters_; bool valid_; Slice key_; }; class Arena; // Return an empty iterator (yields nothing) allocated from arena. extern InternalIterator* NewEmptyInternalIterator(Arena* arena); // Return an empty iterator with the specified status, allocated arena. extern InternalIterator* NewErrorInternalIterator(const Status& status, Arena* arena); } // namespace rocksdb rocksdb-4.5.1/table/merger.cc000066400000000000000000000256131267530634000160440ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/merger.h" #include #include "rocksdb/comparator.h" #include "rocksdb/iterator.h" #include "rocksdb/options.h" #include "table/internal_iterator.h" #include "table/iter_heap.h" #include "table/iterator_wrapper.h" #include "util/arena.h" #include "util/heap.h" #include "util/stop_watch.h" #include "util/sync_point.h" #include "util/perf_context_imp.h" #include "util/autovector.h" namespace rocksdb { // Without anonymous namespace here, we fail the warning -Wmissing-prototypes namespace { typedef BinaryHeap MergerMaxIterHeap; typedef BinaryHeap MergerMinIterHeap; } // namespace const size_t kNumIterReserve = 4; class MergingIterator : public InternalIterator { public: MergingIterator(const Comparator* comparator, InternalIterator** children, int n, bool is_arena_mode) : data_pinned_(false), is_arena_mode_(is_arena_mode), comparator_(comparator), current_(nullptr), direction_(kForward), minHeap_(comparator_) { children_.resize(n); for (int i = 0; i < n; i++) { children_[i].Set(children[i]); } for (auto& child : children_) { if (child.Valid()) { minHeap_.push(&child); } } current_ = CurrentForward(); } virtual void AddIterator(InternalIterator* iter) { assert(direction_ == kForward); children_.emplace_back(iter); if (data_pinned_) { Status s = iter->PinData(); assert(s.ok()); } auto new_wrapper = children_.back(); if (new_wrapper.Valid()) { minHeap_.push(&new_wrapper); current_ = CurrentForward(); } } virtual ~MergingIterator() { for (auto& child : children_) { child.DeleteIter(is_arena_mode_); } } virtual bool Valid() const override { return (current_ != nullptr); } virtual void SeekToFirst() override { ClearHeaps(); for (auto& child : children_) { child.SeekToFirst(); if (child.Valid()) { minHeap_.push(&child); } } direction_ = kForward; current_ = CurrentForward(); } virtual void SeekToLast() override { ClearHeaps(); InitMaxHeap(); for (auto& child : children_) { child.SeekToLast(); if (child.Valid()) { maxHeap_->push(&child); } } direction_ = kReverse; current_ = CurrentReverse(); } virtual void Seek(const Slice& target) override { ClearHeaps(); for (auto& child : children_) { { PERF_TIMER_GUARD(seek_child_seek_time); child.Seek(target); } PERF_COUNTER_ADD(seek_child_seek_count, 1); if (child.Valid()) { PERF_TIMER_GUARD(seek_min_heap_time); minHeap_.push(&child); } } direction_ = kForward; { PERF_TIMER_GUARD(seek_min_heap_time); current_ = CurrentForward(); } } virtual void Next() override { assert(Valid()); // Ensure that all children are positioned after key(). // If we are moving in the forward direction, it is already // true for all of the non-current children since current_ is // the smallest child and key() == current_->key(). if (direction_ != kForward) { // Otherwise, advance the non-current children. We advance current_ // just after the if-block. ClearHeaps(); for (auto& child : children_) { if (&child != current_) { child.Seek(key()); if (child.Valid() && comparator_->Equal(key(), child.key())) { child.Next(); } } if (child.Valid()) { minHeap_.push(&child); } } direction_ = kForward; // The loop advanced all non-current children to be > key() so current_ // should still be strictly the smallest key. assert(current_ == CurrentForward()); } // For the heap modifications below to be correct, current_ must be the // current top of the heap. assert(current_ == CurrentForward()); // as the current points to the current record. move the iterator forward. current_->Next(); if (current_->Valid()) { // current is still valid after the Next() call above. Call // replace_top() to restore the heap property. When the same child // iterator yields a sequence of keys, this is cheap. minHeap_.replace_top(current_); } else { // current stopped being valid, remove it from the heap. minHeap_.pop(); } current_ = CurrentForward(); } virtual void Prev() override { assert(Valid()); // Ensure that all children are positioned before key(). // If we are moving in the reverse direction, it is already // true for all of the non-current children since current_ is // the largest child and key() == current_->key(). if (direction_ != kReverse) { // Otherwise, retreat the non-current children. We retreat current_ // just after the if-block. ClearHeaps(); InitMaxHeap(); for (auto& child : children_) { if (&child != current_) { child.Seek(key()); if (child.Valid()) { // Child is at first entry >= key(). Step back one to be < key() TEST_SYNC_POINT_CALLBACK("MergeIterator::Prev:BeforePrev", &child); child.Prev(); } else { // Child has no entries >= key(). Position at last entry. TEST_SYNC_POINT("MergeIterator::Prev:BeforeSeekToLast"); child.SeekToLast(); } } if (child.Valid()) { maxHeap_->push(&child); } } direction_ = kReverse; // Note that we don't do assert(current_ == CurrentReverse()) here // because it is possible to have some keys larger than the seek-key // inserted between Seek() and SeekToLast(), which makes current_ not // equal to CurrentReverse(). current_ = CurrentReverse(); } // For the heap modifications below to be correct, current_ must be the // current top of the heap. assert(current_ == CurrentReverse()); current_->Prev(); if (current_->Valid()) { // current is still valid after the Prev() call above. Call // replace_top() to restore the heap property. When the same child // iterator yields a sequence of keys, this is cheap. maxHeap_->replace_top(current_); } else { // current stopped being valid, remove it from the heap. maxHeap_->pop(); } current_ = CurrentReverse(); } virtual Slice key() const override { assert(Valid()); return current_->key(); } virtual Slice value() const override { assert(Valid()); return current_->value(); } virtual Status status() const override { Status s; for (auto& child : children_) { s = child.status(); if (!s.ok()) { break; } } return s; } virtual Status PinData() override { Status s; if (data_pinned_) { return s; } for (size_t i = 0; i < children_.size(); i++) { s = children_[i].PinData(); if (!s.ok()) { // We failed to pin an iterator, clean up for (size_t j = 0; j < i; j++) { children_[j].ReleasePinnedData(); } break; } } data_pinned_ = s.ok(); return s; } virtual Status ReleasePinnedData() override { Status s; if (!data_pinned_) { return s; } for (auto& child : children_) { Status release_status = child.ReleasePinnedData(); if (s.ok() && !release_status.ok()) { s = release_status; } } data_pinned_ = false; return s; } virtual bool IsKeyPinned() const override { assert(Valid()); return current_->IsKeyPinned(); } private: bool data_pinned_; // Clears heaps for both directions, used when changing direction or seeking void ClearHeaps(); // Ensures that maxHeap_ is initialized when starting to go in the reverse // direction void InitMaxHeap(); bool is_arena_mode_; const Comparator* comparator_; autovector children_; // Cached pointer to child iterator with the current key, or nullptr if no // child iterators are valid. This is the top of minHeap_ or maxHeap_ // depending on the direction. IteratorWrapper* current_; // Which direction is the iterator moving? enum Direction { kForward, kReverse }; Direction direction_; MergerMinIterHeap minHeap_; // Max heap is used for reverse iteration, which is way less common than // forward. Lazily initialize it to save memory. std::unique_ptr maxHeap_; IteratorWrapper* CurrentForward() const { assert(direction_ == kForward); return !minHeap_.empty() ? minHeap_.top() : nullptr; } IteratorWrapper* CurrentReverse() const { assert(direction_ == kReverse); assert(maxHeap_); return !maxHeap_->empty() ? maxHeap_->top() : nullptr; } }; void MergingIterator::ClearHeaps() { minHeap_.clear(); if (maxHeap_) { maxHeap_->clear(); } } void MergingIterator::InitMaxHeap() { if (!maxHeap_) { maxHeap_.reset(new MergerMaxIterHeap(comparator_)); } } InternalIterator* NewMergingIterator(const Comparator* cmp, InternalIterator** list, int n, Arena* arena) { assert(n >= 0); if (n == 0) { return NewEmptyInternalIterator(arena); } else if (n == 1) { return list[0]; } else { if (arena == nullptr) { return new MergingIterator(cmp, list, n, false); } else { auto mem = arena->AllocateAligned(sizeof(MergingIterator)); return new (mem) MergingIterator(cmp, list, n, true); } } } MergeIteratorBuilder::MergeIteratorBuilder(const Comparator* comparator, Arena* a) : first_iter(nullptr), use_merging_iter(false), arena(a) { auto mem = arena->AllocateAligned(sizeof(MergingIterator)); merge_iter = new (mem) MergingIterator(comparator, nullptr, 0, true); } void MergeIteratorBuilder::AddIterator(InternalIterator* iter) { if (!use_merging_iter && first_iter != nullptr) { merge_iter->AddIterator(first_iter); use_merging_iter = true; } if (use_merging_iter) { merge_iter->AddIterator(iter); } else { first_iter = iter; } } InternalIterator* MergeIteratorBuilder::Finish() { if (!use_merging_iter) { return first_iter; } else { auto ret = merge_iter; merge_iter = nullptr; return ret; } } } // namespace rocksdb rocksdb-4.5.1/table/merger.h000066400000000000000000000037621267530634000157070ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include "rocksdb/types.h" namespace rocksdb { class Comparator; class InternalIterator; class Env; class Arena; // Return an iterator that provided the union of the data in // children[0,n-1]. Takes ownership of the child iterators and // will delete them when the result iterator is deleted. // // The result does no duplicate suppression. I.e., if a particular // key is present in K child iterators, it will be yielded K times. // // REQUIRES: n >= 0 extern InternalIterator* NewMergingIterator(const Comparator* comparator, InternalIterator** children, int n, Arena* arena = nullptr); class MergingIterator; // A builder class to build a merging iterator by adding iterators one by one. class MergeIteratorBuilder { public: // comparator: the comparator used in merging comparator // arena: where the merging iterator needs to be allocated from. explicit MergeIteratorBuilder(const Comparator* comparator, Arena* arena); ~MergeIteratorBuilder() {} // Add iter to the merging iterator. void AddIterator(InternalIterator* iter); // Get arena used to build the merging iterator. It is called one a child // iterator needs to be allocated. Arena* GetArena() { return arena; } // Return the result merging iterator. InternalIterator* Finish(); private: MergingIterator* merge_iter; InternalIterator* first_iter; bool use_merging_iter; Arena* arena; }; } // namespace rocksdb rocksdb-4.5.1/table/merger_test.cc000066400000000000000000000104221267530634000170730ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include #include #include "table/merger.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { class MergerTest : public testing::Test { public: MergerTest() : rnd_(3), merging_iterator_(nullptr), single_iterator_(nullptr) {} ~MergerTest() = default; std::vector GenerateStrings(size_t len, int string_len) { std::vector ret; for (size_t i = 0; i < len; ++i) { ret.push_back(test::RandomHumanReadableString(&rnd_, string_len)); } return ret; } void AssertEquivalence() { auto a = merging_iterator_.get(); auto b = single_iterator_.get(); if (!a->Valid()) { ASSERT_TRUE(!b->Valid()); } else { ASSERT_TRUE(b->Valid()); ASSERT_EQ(b->key().ToString(), a->key().ToString()); ASSERT_EQ(b->value().ToString(), a->value().ToString()); } } void SeekToRandom() { Seek(test::RandomHumanReadableString(&rnd_, 5)); } void Seek(std::string target) { merging_iterator_->Seek(target); single_iterator_->Seek(target); } void SeekToFirst() { merging_iterator_->SeekToFirst(); single_iterator_->SeekToFirst(); } void SeekToLast() { merging_iterator_->SeekToLast(); single_iterator_->SeekToLast(); } void Next(int times) { for (int i = 0; i < times && merging_iterator_->Valid(); ++i) { AssertEquivalence(); merging_iterator_->Next(); single_iterator_->Next(); } AssertEquivalence(); } void Prev(int times) { for (int i = 0; i < times && merging_iterator_->Valid(); ++i) { AssertEquivalence(); merging_iterator_->Prev(); single_iterator_->Prev(); } AssertEquivalence(); } void NextAndPrev(int times) { for (int i = 0; i < times && merging_iterator_->Valid(); ++i) { AssertEquivalence(); if (rnd_.OneIn(2)) { merging_iterator_->Prev(); single_iterator_->Prev(); } else { merging_iterator_->Next(); single_iterator_->Next(); } } AssertEquivalence(); } void Generate(size_t num_iterators, size_t strings_per_iterator, int letters_per_string) { std::vector small_iterators; for (size_t i = 0; i < num_iterators; ++i) { auto strings = GenerateStrings(strings_per_iterator, letters_per_string); small_iterators.push_back(new test::VectorIterator(strings)); all_keys_.insert(all_keys_.end(), strings.begin(), strings.end()); } merging_iterator_.reset( NewMergingIterator(BytewiseComparator(), &small_iterators[0], static_cast(small_iterators.size()))); single_iterator_.reset(new test::VectorIterator(all_keys_)); } Random rnd_; std::unique_ptr merging_iterator_; std::unique_ptr single_iterator_; std::vector all_keys_; }; TEST_F(MergerTest, SeekToRandomNextTest) { Generate(1000, 50, 50); for (int i = 0; i < 10; ++i) { SeekToRandom(); AssertEquivalence(); Next(50000); } } TEST_F(MergerTest, SeekToRandomNextSmallStringsTest) { Generate(1000, 50, 2); for (int i = 0; i < 10; ++i) { SeekToRandom(); AssertEquivalence(); Next(50000); } } TEST_F(MergerTest, SeekToRandomPrevTest) { Generate(1000, 50, 50); for (int i = 0; i < 10; ++i) { SeekToRandom(); AssertEquivalence(); Prev(50000); } } TEST_F(MergerTest, SeekToRandomRandomTest) { Generate(200, 50, 50); for (int i = 0; i < 3; ++i) { SeekToRandom(); AssertEquivalence(); NextAndPrev(5000); } } TEST_F(MergerTest, SeekToFirstTest) { Generate(1000, 50, 50); for (int i = 0; i < 10; ++i) { SeekToFirst(); AssertEquivalence(); Next(50000); } } TEST_F(MergerTest, SeekToLastTest) { Generate(1000, 50, 50); for (int i = 0; i < 10; ++i) { SeekToLast(); AssertEquivalence(); Prev(50000); } } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/table/meta_blocks.cc000066400000000000000000000262461267530634000170510ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/meta_blocks.h" #include #include #include "db/table_properties_collector.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "table/block.h" #include "table/format.h" #include "table/internal_iterator.h" #include "table/table_properties_internal.h" #include "util/coding.h" namespace rocksdb { MetaIndexBuilder::MetaIndexBuilder() : meta_index_block_(new BlockBuilder(1 /* restart interval */)) {} void MetaIndexBuilder::Add(const std::string& key, const BlockHandle& handle) { std::string handle_encoding; handle.EncodeTo(&handle_encoding); meta_block_handles_.insert({key, handle_encoding}); } Slice MetaIndexBuilder::Finish() { for (const auto& metablock : meta_block_handles_) { meta_index_block_->Add(metablock.first, metablock.second); } return meta_index_block_->Finish(); } PropertyBlockBuilder::PropertyBlockBuilder() : properties_block_(new BlockBuilder(1 /* restart interval */)) {} void PropertyBlockBuilder::Add(const std::string& name, const std::string& val) { props_.insert({name, val}); } void PropertyBlockBuilder::Add(const std::string& name, uint64_t val) { assert(props_.find(name) == props_.end()); std::string dst; PutVarint64(&dst, val); Add(name, dst); } void PropertyBlockBuilder::Add( const UserCollectedProperties& user_collected_properties) { for (const auto& prop : user_collected_properties) { Add(prop.first, prop.second); } } void PropertyBlockBuilder::AddTableProperty(const TableProperties& props) { Add(TablePropertiesNames::kRawKeySize, props.raw_key_size); Add(TablePropertiesNames::kRawValueSize, props.raw_value_size); Add(TablePropertiesNames::kDataSize, props.data_size); Add(TablePropertiesNames::kIndexSize, props.index_size); Add(TablePropertiesNames::kNumEntries, props.num_entries); Add(TablePropertiesNames::kNumDataBlocks, props.num_data_blocks); Add(TablePropertiesNames::kFilterSize, props.filter_size); Add(TablePropertiesNames::kFormatVersion, props.format_version); Add(TablePropertiesNames::kFixedKeyLen, props.fixed_key_len); if (!props.filter_policy_name.empty()) { Add(TablePropertiesNames::kFilterPolicy, props.filter_policy_name); } } Slice PropertyBlockBuilder::Finish() { for (const auto& prop : props_) { properties_block_->Add(prop.first, prop.second); } return properties_block_->Finish(); } void LogPropertiesCollectionError( Logger* info_log, const std::string& method, const std::string& name) { assert(method == "Add" || method == "Finish"); std::string msg = "Encountered error when calling TablePropertiesCollector::" + method + "() with collector name: " + name; Log(InfoLogLevel::ERROR_LEVEL, info_log, "%s", msg.c_str()); } bool NotifyCollectTableCollectorsOnAdd( const Slice& key, const Slice& value, uint64_t file_size, const std::vector>& collectors, Logger* info_log) { bool all_succeeded = true; for (auto& collector : collectors) { Status s = collector->InternalAdd(key, value, file_size); all_succeeded = all_succeeded && s.ok(); if (!s.ok()) { LogPropertiesCollectionError(info_log, "Add" /* method */, collector->Name()); } } return all_succeeded; } bool NotifyCollectTableCollectorsOnFinish( const std::vector>& collectors, Logger* info_log, PropertyBlockBuilder* builder) { bool all_succeeded = true; for (auto& collector : collectors) { UserCollectedProperties user_collected_properties; Status s = collector->Finish(&user_collected_properties); all_succeeded = all_succeeded && s.ok(); if (!s.ok()) { LogPropertiesCollectionError(info_log, "Finish" /* method */, collector->Name()); } else { builder->Add(user_collected_properties); } } return all_succeeded; } Status ReadProperties(const Slice& handle_value, RandomAccessFileReader* file, const Footer& footer, Env* env, Logger* logger, TableProperties** table_properties) { assert(table_properties); Slice v = handle_value; BlockHandle handle; if (!handle.DecodeFrom(&v).ok()) { return Status::InvalidArgument("Failed to decode properties block handle"); } BlockContents block_contents; ReadOptions read_options; read_options.verify_checksums = false; Status s; s = ReadBlockContents(file, footer, read_options, handle, &block_contents, env, false); if (!s.ok()) { return s; } Block properties_block(std::move(block_contents)); std::unique_ptr iter( properties_block.NewIterator(BytewiseComparator())); auto new_table_properties = new TableProperties(); // All pre-defined properties of type uint64_t std::unordered_map predefined_uint64_properties = { {TablePropertiesNames::kDataSize, &new_table_properties->data_size}, {TablePropertiesNames::kIndexSize, &new_table_properties->index_size}, {TablePropertiesNames::kFilterSize, &new_table_properties->filter_size}, {TablePropertiesNames::kRawKeySize, &new_table_properties->raw_key_size}, {TablePropertiesNames::kRawValueSize, &new_table_properties->raw_value_size}, {TablePropertiesNames::kNumDataBlocks, &new_table_properties->num_data_blocks}, {TablePropertiesNames::kNumEntries, &new_table_properties->num_entries}, {TablePropertiesNames::kFormatVersion, &new_table_properties->format_version}, {TablePropertiesNames::kFixedKeyLen, &new_table_properties->fixed_key_len}, }; std::string last_key; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { s = iter->status(); if (!s.ok()) { break; } auto key = iter->key().ToString(); // properties block is strictly sorted with no duplicate key. assert(last_key.empty() || BytewiseComparator()->Compare(key, last_key) > 0); last_key = key; auto raw_val = iter->value(); auto pos = predefined_uint64_properties.find(key); if (pos != predefined_uint64_properties.end()) { // handle predefined rocksdb properties uint64_t val; if (!GetVarint64(&raw_val, &val)) { // skip malformed value auto error_msg = "Detect malformed value in properties meta-block:" "\tkey: " + key + "\tval: " + raw_val.ToString(); Log(InfoLogLevel::ERROR_LEVEL, logger, "%s", error_msg.c_str()); continue; } *(pos->second) = val; } else if (key == TablePropertiesNames::kFilterPolicy) { new_table_properties->filter_policy_name = raw_val.ToString(); } else { // handle user-collected properties new_table_properties->user_collected_properties.insert( {key, raw_val.ToString()}); } } if (s.ok()) { *table_properties = new_table_properties; } else { delete new_table_properties; } return s; } Status ReadTableProperties(RandomAccessFileReader* file, uint64_t file_size, uint64_t table_magic_number, Env* env, Logger* info_log, TableProperties** properties) { // -- Read metaindex block Footer footer; auto s = ReadFooterFromFile(file, file_size, &footer, table_magic_number); if (!s.ok()) { return s; } auto metaindex_handle = footer.metaindex_handle(); BlockContents metaindex_contents; ReadOptions read_options; read_options.verify_checksums = false; s = ReadBlockContents(file, footer, read_options, metaindex_handle, &metaindex_contents, env, false); if (!s.ok()) { return s; } Block metaindex_block(std::move(metaindex_contents)); std::unique_ptr meta_iter( metaindex_block.NewIterator(BytewiseComparator())); // -- Read property block bool found_properties_block = true; s = SeekToPropertiesBlock(meta_iter.get(), &found_properties_block); if (!s.ok()) { return s; } TableProperties table_properties; if (found_properties_block == true) { s = ReadProperties(meta_iter->value(), file, footer, env, info_log, properties); } else { s = Status::NotFound(); } return s; } Status FindMetaBlock(InternalIterator* meta_index_iter, const std::string& meta_block_name, BlockHandle* block_handle) { meta_index_iter->Seek(meta_block_name); if (meta_index_iter->status().ok() && meta_index_iter->Valid() && meta_index_iter->key() == meta_block_name) { Slice v = meta_index_iter->value(); return block_handle->DecodeFrom(&v); } else { return Status::Corruption("Cannot find the meta block", meta_block_name); } } Status FindMetaBlock(RandomAccessFileReader* file, uint64_t file_size, uint64_t table_magic_number, Env* env, const std::string& meta_block_name, BlockHandle* block_handle) { Footer footer; auto s = ReadFooterFromFile(file, file_size, &footer, table_magic_number); if (!s.ok()) { return s; } auto metaindex_handle = footer.metaindex_handle(); BlockContents metaindex_contents; ReadOptions read_options; read_options.verify_checksums = false; s = ReadBlockContents(file, footer, read_options, metaindex_handle, &metaindex_contents, env, false); if (!s.ok()) { return s; } Block metaindex_block(std::move(metaindex_contents)); std::unique_ptr meta_iter; meta_iter.reset(metaindex_block.NewIterator(BytewiseComparator())); return FindMetaBlock(meta_iter.get(), meta_block_name, block_handle); } Status ReadMetaBlock(RandomAccessFileReader* file, uint64_t file_size, uint64_t table_magic_number, Env* env, const std::string& meta_block_name, BlockContents* contents) { Status status; Footer footer; status = ReadFooterFromFile(file, file_size, &footer, table_magic_number); if (!status.ok()) { return status; } // Reading metaindex block auto metaindex_handle = footer.metaindex_handle(); BlockContents metaindex_contents; ReadOptions read_options; read_options.verify_checksums = false; status = ReadBlockContents(file, footer, read_options, metaindex_handle, &metaindex_contents, env, false); if (!status.ok()) { return status; } // Finding metablock Block metaindex_block(std::move(metaindex_contents)); std::unique_ptr meta_iter; meta_iter.reset(metaindex_block.NewIterator(BytewiseComparator())); BlockHandle block_handle; status = FindMetaBlock(meta_iter.get(), meta_block_name, &block_handle); if (!status.ok()) { return status; } // Reading metablock return ReadBlockContents(file, footer, read_options, block_handle, contents, env, false); } } // namespace rocksdb rocksdb-4.5.1/table/meta_blocks.h000066400000000000000000000114721267530634000167060ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include #include "db/builder.h" #include "db/table_properties_collector.h" #include "util/kv_map.h" #include "rocksdb/comparator.h" #include "rocksdb/options.h" #include "rocksdb/slice.h" #include "table/block_builder.h" #include "table/format.h" namespace rocksdb { class BlockBuilder; class BlockHandle; class Env; class Footer; class Logger; class RandomAccessFile; struct TableProperties; class InternalIterator; class MetaIndexBuilder { public: MetaIndexBuilder(const MetaIndexBuilder&) = delete; MetaIndexBuilder& operator=(const MetaIndexBuilder&) = delete; MetaIndexBuilder(); void Add(const std::string& key, const BlockHandle& handle); // Write all the added key/value pairs to the block and return the contents // of the block. Slice Finish(); private: // store the sorted key/handle of the metablocks. stl_wrappers::KVMap meta_block_handles_; std::unique_ptr meta_index_block_; }; class PropertyBlockBuilder { public: PropertyBlockBuilder(const PropertyBlockBuilder&) = delete; PropertyBlockBuilder& operator=(const PropertyBlockBuilder&) = delete; PropertyBlockBuilder(); void AddTableProperty(const TableProperties& props); void Add(const std::string& key, uint64_t value); void Add(const std::string& key, const std::string& value); void Add(const UserCollectedProperties& user_collected_properties); // Write all the added entries to the block and return the block contents Slice Finish(); private: std::unique_ptr properties_block_; stl_wrappers::KVMap props_; }; // Were we encounter any error occurs during user-defined statistics collection, // we'll write the warning message to info log. void LogPropertiesCollectionError( Logger* info_log, const std::string& method, const std::string& name); // Utility functions help table builder to trigger batch events for user // defined property collectors. // Return value indicates if there is any error occurred; if error occurred, // the warning message will be logged. // NotifyCollectTableCollectorsOnAdd() triggers the `Add` event for all // property collectors. bool NotifyCollectTableCollectorsOnAdd( const Slice& key, const Slice& value, uint64_t file_size, const std::vector>& collectors, Logger* info_log); // NotifyCollectTableCollectorsOnAdd() triggers the `Finish` event for all // property collectors. The collected properties will be added to `builder`. bool NotifyCollectTableCollectorsOnFinish( const std::vector>& collectors, Logger* info_log, PropertyBlockBuilder* builder); // Read the properties from the table. // @returns a status to indicate if the operation succeeded. On success, // *table_properties will point to a heap-allocated TableProperties // object, otherwise value of `table_properties` will not be modified. Status ReadProperties(const Slice& handle_value, RandomAccessFileReader* file, const Footer& footer, Env* env, Logger* logger, TableProperties** table_properties); // Directly read the properties from the properties block of a plain table. // @returns a status to indicate if the operation succeeded. On success, // *table_properties will point to a heap-allocated TableProperties // object, otherwise value of `table_properties` will not be modified. Status ReadTableProperties(RandomAccessFileReader* file, uint64_t file_size, uint64_t table_magic_number, Env* env, Logger* info_log, TableProperties** properties); // Find the meta block from the meta index block. Status FindMetaBlock(InternalIterator* meta_index_iter, const std::string& meta_block_name, BlockHandle* block_handle); // Find the meta block Status FindMetaBlock(RandomAccessFileReader* file, uint64_t file_size, uint64_t table_magic_number, Env* env, const std::string& meta_block_name, BlockHandle* block_handle); // Read the specified meta block with name meta_block_name // from `file` and initialize `contents` with contents of this block. // Return Status::OK in case of success. Status ReadMetaBlock(RandomAccessFileReader* file, uint64_t file_size, uint64_t table_magic_number, Env* env, const std::string& meta_block_name, BlockContents* contents); } // namespace rocksdb rocksdb-4.5.1/table/mock_table.cc000066400000000000000000000105311267530634000166540ustar00rootroot00000000000000// Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/mock_table.h" #include "db/dbformat.h" #include "port/port.h" #include "rocksdb/table_properties.h" #include "table/get_context.h" #include "util/coding.h" #include "util/file_reader_writer.h" namespace rocksdb { namespace mock { namespace { const InternalKeyComparator icmp_(BytewiseComparator()); } // namespace stl_wrappers::KVMap MakeMockFile( std::initializer_list> l) { return stl_wrappers::KVMap(l, stl_wrappers::LessOfComparator(&icmp_)); } InternalIterator* MockTableReader::NewIterator(const ReadOptions&, Arena* arena, bool skip_filters) { return new MockTableIterator(table_); } Status MockTableReader::Get(const ReadOptions&, const Slice& key, GetContext* get_context, bool skip_filters) { std::unique_ptr iter(new MockTableIterator(table_)); for (iter->Seek(key); iter->Valid(); iter->Next()) { ParsedInternalKey parsed_key; if (!ParseInternalKey(iter->key(), &parsed_key)) { return Status::Corruption(Slice()); } if (!get_context->SaveValue(parsed_key, iter->value())) { break; } } return Status::OK(); } std::shared_ptr MockTableReader::GetTableProperties() const { return std::shared_ptr(new TableProperties()); } MockTableFactory::MockTableFactory() : next_id_(1) {} Status MockTableFactory::NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader) const { uint32_t id = GetIDFromFile(file.get()); MutexLock lock_guard(&file_system_.mutex); auto it = file_system_.files.find(id); if (it == file_system_.files.end()) { return Status::IOError("Mock file not found"); } table_reader->reset(new MockTableReader(it->second)); return Status::OK(); } TableBuilder* MockTableFactory::NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const { uint32_t id = GetAndWriteNextID(file); return new MockTableBuilder(id, &file_system_); } Status MockTableFactory::CreateMockTable(Env* env, const std::string& fname, stl_wrappers::KVMap file_contents) { std::unique_ptr file; auto s = env->NewWritableFile(fname, &file, EnvOptions()); if (!s.ok()) { return s; } WritableFileWriter file_writer(std::move(file), EnvOptions()); uint32_t id = GetAndWriteNextID(&file_writer); file_system_.files.insert({id, std::move(file_contents)}); return Status::OK(); } uint32_t MockTableFactory::GetAndWriteNextID(WritableFileWriter* file) const { uint32_t next_id = next_id_.fetch_add(1); char buf[4]; EncodeFixed32(buf, next_id); file->Append(Slice(buf, 4)); return next_id; } uint32_t MockTableFactory::GetIDFromFile(RandomAccessFileReader* file) const { char buf[4]; Slice result; file->Read(0, 4, &result, buf); assert(result.size() == 4); return DecodeFixed32(buf); } void MockTableFactory::AssertSingleFile( const stl_wrappers::KVMap& file_contents) { ASSERT_EQ(file_system_.files.size(), 1U); ASSERT_TRUE(file_contents == file_system_.files.begin()->second); } void MockTableFactory::AssertLatestFile( const stl_wrappers::KVMap& file_contents) { ASSERT_GE(file_system_.files.size(), 1U); auto latest = file_system_.files.end(); --latest; if (file_contents != latest->second) { std::cout << "Wrong content! Content of latest file:" << std::endl; for (const auto& kv : latest->second) { ParsedInternalKey ikey; std::string key, value; std::tie(key, value) = kv; ParseInternalKey(Slice(key), &ikey); std::cout << ikey.DebugString(false) << " -> " << value << std::endl; } ASSERT_TRUE(false); } } } // namespace mock } // namespace rocksdb rocksdb-4.5.1/table/mock_table.h000066400000000000000000000131401267530634000165150ustar00rootroot00000000000000// Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. // Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include #include #include #include #include #include #include #include "util/kv_map.h" #include "port/port.h" #include "rocksdb/comparator.h" #include "rocksdb/table.h" #include "table/internal_iterator.h" #include "table/table_builder.h" #include "table/table_reader.h" #include "util/mutexlock.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { namespace mock { stl_wrappers::KVMap MakeMockFile( std::initializer_list> l = {}); struct MockTableFileSystem { port::Mutex mutex; std::map files; }; class MockTableReader : public TableReader { public: explicit MockTableReader(const stl_wrappers::KVMap& table) : table_(table) {} InternalIterator* NewIterator(const ReadOptions&, Arena* arena, bool skip_filters = false) override; Status Get(const ReadOptions&, const Slice& key, GetContext* get_context, bool skip_filters = false) override; uint64_t ApproximateOffsetOf(const Slice& key) override { return 0; } virtual size_t ApproximateMemoryUsage() const override { return 0; } void SetupForCompaction() override {} std::shared_ptr GetTableProperties() const override; ~MockTableReader() {} private: const stl_wrappers::KVMap& table_; }; class MockTableIterator : public InternalIterator { public: explicit MockTableIterator(const stl_wrappers::KVMap& table) : table_(table) { itr_ = table_.end(); } bool Valid() const override { return itr_ != table_.end(); } void SeekToFirst() override { itr_ = table_.begin(); } void SeekToLast() override { itr_ = table_.end(); --itr_; } void Seek(const Slice& target) override { std::string str_target(target.data(), target.size()); itr_ = table_.lower_bound(str_target); } void Next() override { ++itr_; } void Prev() override { if (itr_ == table_.begin()) { itr_ = table_.end(); } else { --itr_; } } Slice key() const override { return Slice(itr_->first); } Slice value() const override { return Slice(itr_->second); } Status status() const override { return Status::OK(); } private: const stl_wrappers::KVMap& table_; stl_wrappers::KVMap::const_iterator itr_; }; class MockTableBuilder : public TableBuilder { public: MockTableBuilder(uint32_t id, MockTableFileSystem* file_system) : id_(id), file_system_(file_system) { table_ = MakeMockFile({}); } // REQUIRES: Either Finish() or Abandon() has been called. ~MockTableBuilder() {} // Add key,value to the table being constructed. // REQUIRES: key is after any previously added key according to comparator. // REQUIRES: Finish(), Abandon() have not been called void Add(const Slice& key, const Slice& value) override { table_.insert({key.ToString(), value.ToString()}); } // Return non-ok iff some error has been detected. Status status() const override { return Status::OK(); } Status Finish() override { MutexLock lock_guard(&file_system_->mutex); file_system_->files.insert({id_, table_}); return Status::OK(); } void Abandon() override {} uint64_t NumEntries() const override { return table_.size(); } uint64_t FileSize() const override { return table_.size(); } TableProperties GetTableProperties() const override { return TableProperties(); } private: uint32_t id_; MockTableFileSystem* file_system_; stl_wrappers::KVMap table_; }; class MockTableFactory : public TableFactory { public: MockTableFactory(); const char* Name() const override { return "MockTable"; } Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader) const override; TableBuilder* NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_familly_id, WritableFileWriter* file) const override; // This function will directly create mock table instead of going through // MockTableBuilder. file_contents has to have a format of . Those key-value pairs will then be inserted into the mock table. Status CreateMockTable(Env* env, const std::string& fname, stl_wrappers::KVMap file_contents); virtual Status SanitizeOptions( const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const override { return Status::OK(); } virtual std::string GetPrintableTableOptions() const override { return std::string(); } // This function will assert that only a single file exists and that the // contents are equal to file_contents void AssertSingleFile(const stl_wrappers::KVMap& file_contents); void AssertLatestFile(const stl_wrappers::KVMap& file_contents); private: uint32_t GetAndWriteNextID(WritableFileWriter* file) const; uint32_t GetIDFromFile(RandomAccessFileReader* file) const; mutable MockTableFileSystem file_system_; mutable std::atomic next_id_; }; } // namespace mock } // namespace rocksdb rocksdb-4.5.1/table/plain_table_builder.cc000066400000000000000000000217451267530634000205450ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "table/plain_table_builder.h" #include #include #include #include #include "rocksdb/comparator.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/options.h" #include "rocksdb/table.h" #include "table/plain_table_factory.h" #include "db/dbformat.h" #include "table/block_builder.h" #include "table/bloom_block.h" #include "table/plain_table_index.h" #include "table/format.h" #include "table/meta_blocks.h" #include "util/coding.h" #include "util/crc32c.h" #include "util/file_reader_writer.h" #include "util/stop_watch.h" namespace rocksdb { namespace { // a utility that helps writing block content to the file // @offset will advance if @block_contents was successfully written. // @block_handle the block handle this particular block. Status WriteBlock(const Slice& block_contents, WritableFileWriter* file, uint64_t* offset, BlockHandle* block_handle) { block_handle->set_offset(*offset); block_handle->set_size(block_contents.size()); Status s = file->Append(block_contents); if (s.ok()) { *offset += block_contents.size(); } return s; } } // namespace // kPlainTableMagicNumber was picked by running // echo rocksdb.table.plain | sha1sum // and taking the leading 64 bits. extern const uint64_t kPlainTableMagicNumber = 0x8242229663bf9564ull; extern const uint64_t kLegacyPlainTableMagicNumber = 0x4f3418eb7a8f13b8ull; PlainTableBuilder::PlainTableBuilder( const ImmutableCFOptions& ioptions, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* file, uint32_t user_key_len, EncodingType encoding_type, size_t index_sparseness, uint32_t bloom_bits_per_key, uint32_t num_probes, size_t huge_page_tlb_size, double hash_table_ratio, bool store_index_in_file) : ioptions_(ioptions), bloom_block_(num_probes), file_(file), bloom_bits_per_key_(bloom_bits_per_key), huge_page_tlb_size_(huge_page_tlb_size), encoder_(encoding_type, user_key_len, ioptions.prefix_extractor, index_sparseness), store_index_in_file_(store_index_in_file), prefix_extractor_(ioptions.prefix_extractor) { // Build index block and save it in the file if hash_table_ratio > 0 if (store_index_in_file_) { assert(hash_table_ratio > 0 || IsTotalOrderMode()); index_builder_.reset( new PlainTableIndexBuilder(&arena_, ioptions, index_sparseness, hash_table_ratio, huge_page_tlb_size_)); assert(bloom_bits_per_key_ > 0); properties_.user_collected_properties [PlainTablePropertyNames::kBloomVersion] = "1"; // For future use } properties_.fixed_key_len = user_key_len; // for plain table, we put all the data in a big chuck. properties_.num_data_blocks = 1; // Fill it later if store_index_in_file_ == true properties_.index_size = 0; properties_.filter_size = 0; // To support roll-back to previous version, now still use version 0 for // plain encoding. properties_.format_version = (encoding_type == kPlain) ? 0 : 1; if (ioptions_.prefix_extractor) { properties_.user_collected_properties [PlainTablePropertyNames::kPrefixExtractorName] = ioptions_.prefix_extractor->Name(); } std::string val; PutFixed32(&val, static_cast(encoder_.GetEncodingType())); properties_.user_collected_properties [PlainTablePropertyNames::kEncodingType] = val; for (auto& collector_factories : *int_tbl_prop_collector_factories) { table_properties_collectors_.emplace_back( collector_factories->CreateIntTblPropCollector(column_family_id)); } } PlainTableBuilder::~PlainTableBuilder() { } void PlainTableBuilder::Add(const Slice& key, const Slice& value) { // temp buffer for metadata bytes between key and value. char meta_bytes_buf[6]; size_t meta_bytes_buf_size = 0; ParsedInternalKey internal_key; ParseInternalKey(key, &internal_key); // Store key hash if (store_index_in_file_) { if (ioptions_.prefix_extractor == nullptr) { keys_or_prefixes_hashes_.push_back(GetSliceHash(internal_key.user_key)); } else { Slice prefix = ioptions_.prefix_extractor->Transform(internal_key.user_key); keys_or_prefixes_hashes_.push_back(GetSliceHash(prefix)); } } // Write value assert(offset_ <= std::numeric_limits::max()); auto prev_offset = static_cast(offset_); // Write out the key encoder_.AppendKey(key, file_, &offset_, meta_bytes_buf, &meta_bytes_buf_size); if (SaveIndexInFile()) { index_builder_->AddKeyPrefix(GetPrefix(internal_key), prev_offset); } // Write value length uint32_t value_size = static_cast(value.size()); char* end_ptr = EncodeVarint32(meta_bytes_buf + meta_bytes_buf_size, value_size); assert(end_ptr <= meta_bytes_buf + sizeof(meta_bytes_buf)); meta_bytes_buf_size = end_ptr - meta_bytes_buf; file_->Append(Slice(meta_bytes_buf, meta_bytes_buf_size)); // Write value file_->Append(value); offset_ += value_size + meta_bytes_buf_size; properties_.num_entries++; properties_.raw_key_size += key.size(); properties_.raw_value_size += value.size(); // notify property collectors NotifyCollectTableCollectorsOnAdd( key, value, offset_, table_properties_collectors_, ioptions_.info_log); } Status PlainTableBuilder::status() const { return status_; } Status PlainTableBuilder::Finish() { assert(!closed_); closed_ = true; properties_.data_size = offset_; // Write the following blocks // 1. [meta block: bloom] - optional // 2. [meta block: index] - optional // 3. [meta block: properties] // 4. [metaindex block] // 5. [footer] MetaIndexBuilder meta_index_builer; if (store_index_in_file_ && (properties_.num_entries > 0)) { assert(properties_.num_entries <= std::numeric_limits::max()); bloom_block_.SetTotalBits( &arena_, static_cast(properties_.num_entries) * bloom_bits_per_key_, ioptions_.bloom_locality, huge_page_tlb_size_, ioptions_.info_log); PutVarint32(&properties_.user_collected_properties [PlainTablePropertyNames::kNumBloomBlocks], bloom_block_.GetNumBlocks()); bloom_block_.AddKeysHashes(keys_or_prefixes_hashes_); BlockHandle bloom_block_handle; auto finish_result = bloom_block_.Finish(); properties_.filter_size = finish_result.size(); auto s = WriteBlock(finish_result, file_, &offset_, &bloom_block_handle); if (!s.ok()) { return s; } BlockHandle index_block_handle; finish_result = index_builder_->Finish(); properties_.index_size = finish_result.size(); s = WriteBlock(finish_result, file_, &offset_, &index_block_handle); if (!s.ok()) { return s; } meta_index_builer.Add(BloomBlockBuilder::kBloomBlock, bloom_block_handle); meta_index_builer.Add(PlainTableIndexBuilder::kPlainTableIndexBlock, index_block_handle); } // Calculate bloom block size and index block size PropertyBlockBuilder property_block_builder; // -- Add basic properties property_block_builder.AddTableProperty(properties_); property_block_builder.Add(properties_.user_collected_properties); // -- Add user collected properties NotifyCollectTableCollectorsOnFinish(table_properties_collectors_, ioptions_.info_log, &property_block_builder); // -- Write property block BlockHandle property_block_handle; auto s = WriteBlock( property_block_builder.Finish(), file_, &offset_, &property_block_handle ); if (!s.ok()) { return s; } meta_index_builer.Add(kPropertiesBlock, property_block_handle); // -- write metaindex block BlockHandle metaindex_block_handle; s = WriteBlock( meta_index_builer.Finish(), file_, &offset_, &metaindex_block_handle ); if (!s.ok()) { return s; } // Write Footer // no need to write out new footer if we're using default checksum Footer footer(kLegacyPlainTableMagicNumber, 0); footer.set_metaindex_handle(metaindex_block_handle); footer.set_index_handle(BlockHandle::NullBlockHandle()); std::string footer_encoding; footer.EncodeTo(&footer_encoding); s = file_->Append(footer_encoding); if (s.ok()) { offset_ += footer_encoding.size(); } return s; } void PlainTableBuilder::Abandon() { closed_ = true; } uint64_t PlainTableBuilder::NumEntries() const { return properties_.num_entries; } uint64_t PlainTableBuilder::FileSize() const { return offset_; } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_builder.h000066400000000000000000000110421267530634000203740ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "rocksdb/options.h" #include "rocksdb/status.h" #include "table/table_builder.h" #include "table/plain_table_key_coding.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "table/bloom_block.h" #include "table/plain_table_index.h" namespace rocksdb { class BlockBuilder; class BlockHandle; class WritableFile; class TableBuilder; class PlainTableBuilder: public TableBuilder { public: // Create a builder that will store the contents of the table it is // building in *file. Does not close the file. It is up to the // caller to close the file after calling Finish(). The output file // will be part of level specified by 'level'. A value of -1 means // that the caller does not know which level the output file will reside. PlainTableBuilder( const ImmutableCFOptions& ioptions, const std::vector>* int_tbl_prop_collector_factories, uint32_t column_family_id, WritableFileWriter* file, uint32_t user_key_size, EncodingType encoding_type, size_t index_sparseness, uint32_t bloom_bits_per_key, uint32_t num_probes = 6, size_t huge_page_tlb_size = 0, double hash_table_ratio = 0, bool store_index_in_file = false); // REQUIRES: Either Finish() or Abandon() has been called. ~PlainTableBuilder(); // Add key,value to the table being constructed. // REQUIRES: key is after any previously added key according to comparator. // REQUIRES: Finish(), Abandon() have not been called void Add(const Slice& key, const Slice& value) override; // Return non-ok iff some error has been detected. Status status() const override; // Finish building the table. Stops using the file passed to the // constructor after this function returns. // REQUIRES: Finish(), Abandon() have not been called Status Finish() override; // Indicate that the contents of this builder should be abandoned. Stops // using the file passed to the constructor after this function returns. // If the caller is not going to call Finish(), it must call Abandon() // before destroying this builder. // REQUIRES: Finish(), Abandon() have not been called void Abandon() override; // Number of calls to Add() so far. uint64_t NumEntries() const override; // Size of the file generated so far. If invoked after a successful // Finish() call, returns the size of the final generated file. uint64_t FileSize() const override; TableProperties GetTableProperties() const override { return properties_; } bool SaveIndexInFile() const { return store_index_in_file_; } private: Arena arena_; const ImmutableCFOptions& ioptions_; std::vector> table_properties_collectors_; BloomBlockBuilder bloom_block_; std::unique_ptr index_builder_; WritableFileWriter* file_; uint64_t offset_ = 0; uint32_t bloom_bits_per_key_; size_t huge_page_tlb_size_; Status status_; TableProperties properties_; PlainTableKeyEncoder encoder_; bool store_index_in_file_; std::vector keys_or_prefixes_hashes_; bool closed_ = false; // Either Finish() or Abandon() has been called. const SliceTransform* prefix_extractor_; Slice GetPrefix(const Slice& target) const { assert(target.size() >= 8); // target is internal key return GetPrefixFromUserKey(GetUserKey(target)); } Slice GetPrefix(const ParsedInternalKey& target) const { return GetPrefixFromUserKey(target.user_key); } Slice GetUserKey(const Slice& key) const { return Slice(key.data(), key.size() - 8); } Slice GetPrefixFromUserKey(const Slice& user_key) const { if (!IsTotalOrderMode()) { return prefix_extractor_->Transform(user_key); } else { // Use empty slice as prefix if prefix_extractor is not set. // In that case, // it falls back to pure binary search and // total iterator seek is supported. return Slice(); } } bool IsTotalOrderMode() const { return (prefix_extractor_ == nullptr); } // No copying allowed PlainTableBuilder(const PlainTableBuilder&) = delete; void operator=(const PlainTableBuilder&) = delete; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_factory.cc000066400000000000000000000071721267530634000205640ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "table/plain_table_factory.h" #include #include #include "db/dbformat.h" #include "table/plain_table_builder.h" #include "table/plain_table_reader.h" #include "port/port.h" namespace rocksdb { Status PlainTableFactory::NewTableReader( const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table) const { return PlainTableReader::Open( table_reader_options.ioptions, table_reader_options.env_options, table_reader_options.internal_comparator, std::move(file), file_size, table, table_options_.bloom_bits_per_key, table_options_.hash_table_ratio, table_options_.index_sparseness, table_options_.huge_page_tlb_size, table_options_.full_scan_mode); } TableBuilder* PlainTableFactory::NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const { // Ignore the skip_filters flag. PlainTable format is optimized for small // in-memory dbs. The skip_filters optimization is not useful for plain // tables // return new PlainTableBuilder( table_builder_options.ioptions, table_builder_options.int_tbl_prop_collector_factories, column_family_id, file, table_options_.user_key_len, table_options_.encoding_type, table_options_.index_sparseness, table_options_.bloom_bits_per_key, 6, table_options_.huge_page_tlb_size, table_options_.hash_table_ratio, table_options_.store_index_in_file); } std::string PlainTableFactory::GetPrintableTableOptions() const { std::string ret; ret.reserve(20000); const int kBufferSize = 200; char buffer[kBufferSize]; snprintf(buffer, kBufferSize, " user_key_len: %u\n", table_options_.user_key_len); ret.append(buffer); snprintf(buffer, kBufferSize, " bloom_bits_per_key: %d\n", table_options_.bloom_bits_per_key); ret.append(buffer); snprintf(buffer, kBufferSize, " hash_table_ratio: %lf\n", table_options_.hash_table_ratio); ret.append(buffer); snprintf(buffer, kBufferSize, " index_sparseness: %" ROCKSDB_PRIszt "\n", table_options_.index_sparseness); ret.append(buffer); snprintf(buffer, kBufferSize, " huge_page_tlb_size: %" ROCKSDB_PRIszt "\n", table_options_.huge_page_tlb_size); ret.append(buffer); snprintf(buffer, kBufferSize, " encoding_type: %d\n", table_options_.encoding_type); ret.append(buffer); snprintf(buffer, kBufferSize, " full_scan_mode: %d\n", table_options_.full_scan_mode); ret.append(buffer); snprintf(buffer, kBufferSize, " store_index_in_file: %d\n", table_options_.store_index_in_file); ret.append(buffer); return ret; } const PlainTableOptions& PlainTableFactory::table_options() const { return table_options_; } extern TableFactory* NewPlainTableFactory(const PlainTableOptions& options) { return new PlainTableFactory(options); } const std::string PlainTablePropertyNames::kPrefixExtractorName = "rocksdb.prefix.extractor.name"; const std::string PlainTablePropertyNames::kEncodingType = "rocksdb.plain.table.encoding.type"; const std::string PlainTablePropertyNames::kBloomVersion = "rocksdb.plain.table.bloom.version"; const std::string PlainTablePropertyNames::kNumBloomBlocks = "rocksdb.plain.table.bloom.numblocks"; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_factory.h000066400000000000000000000156061267530634000204270ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include "rocksdb/options.h" #include "rocksdb/table.h" namespace rocksdb { struct EnvOptions; using std::unique_ptr; class Status; class RandomAccessFile; class WritableFile; class Table; class TableBuilder; // IndexedTable requires fixed length key, configured as a constructor // parameter of the factory class. Output file format: // +-------------+-----------------+ // | version | user_key_length | // +------------++------------+-----------------+ <= key1 offset // | encoded key1 | value_size | | // +------------+-------------+-------------+ | // | value1 | // | | // +--------------------------+-------------+---+ <= key2 offset // | encoded key2 | value_size | | // +------------+-------------+-------------+ | // | value2 | // | | // | ...... | // +-----------------+--------------------------+ // // When the key encoding type is kPlain. Key part is encoded as: // +------------+--------------------+ // | [key_size] | internal key | // +------------+--------------------+ // for the case of user_key_len = kPlainTableVariableLength case, // and simply: // +----------------------+ // | internal key | // +----------------------+ // for user_key_len != kPlainTableVariableLength case. // // If key encoding type is kPrefix. Keys are encoding in this format. // There are three ways to encode a key: // (1) Full Key // +---------------+---------------+-------------------+ // | Full Key Flag | Full Key Size | Full Internal Key | // +---------------+---------------+-------------------+ // which simply encodes a full key // // (2) A key shared the same prefix as the previous key, which is encoded as // format of (1). // +-------------+-------------+-------------+-------------+------------+ // | Prefix Flag | Prefix Size | Suffix Flag | Suffix Size | Key Suffix | // +-------------+-------------+-------------+-------------+------------+ // where key is the suffix part of the key, including the internal bytes. // the actual key will be constructed by concatenating prefix part of the // previous key, with the suffix part of the key here, with sizes given here. // // (3) A key shared the same prefix as the previous key, which is encoded as // the format of (2). // +-----------------+-----------------+------------------------+ // | Key Suffix Flag | Key Suffix Size | Suffix of Internal Key | // +-----------------+-----------------+------------------------+ // The key will be constructed by concatenating previous key's prefix (which is // also a prefix which the last key encoded in the format of (1)) and the // key given here. // // For example, we for following keys (prefix and suffix are separated by // spaces): // 0000 0001 // 0000 00021 // 0000 0002 // 00011 00 // 0002 0001 // Will be encoded like this: // FK 8 00000001 // PF 4 SF 5 00021 // SF 4 0002 // FK 7 0001100 // FK 8 00020001 // (where FK means full key flag, PF means prefix flag and SF means suffix flag) // // All those "key flag + key size" shown above are in this format: // The 8 bits of the first byte: // +----+----+----+----+----+----+----+----+ // | Type | Size | // +----+----+----+----+----+----+----+----+ // Type indicates: full key, prefix, or suffix. // The last 6 bits are for size. If the size bits are not all 1, it means the // size of the key. Otherwise, varint32 is read after this byte. This varint // value + 0x3F (the value of all 1) will be the key size. // // For example, full key with length 16 will be encoded as (binary): // 00 010000 // (00 means full key) // and a prefix with 100 bytes will be encoded as: // 01 111111 00100101 // (63) (37) // (01 means key suffix) // // All the internal keys above (including kPlain and kPrefix) are encoded in // this format: // There are two types: // (1) normal internal key format // +----------- ...... -------------+----+---+---+---+---+---+---+---+ // | user key |type| sequence ID | // +----------- ..... --------------+----+---+---+---+---+---+---+---+ // (2) Special case for keys whose sequence ID is 0 and is value type // +----------- ...... -------------+----+ // | user key |0x80| // +----------- ..... --------------+----+ // To save 7 bytes for the special case where sequence ID = 0. // // class PlainTableFactory : public TableFactory { public: ~PlainTableFactory() {} // user_key_len is the length of the user key. If it is set to be // kPlainTableVariableLength, then it means variable length. Otherwise, all // the keys need to have the fix length of this value. bloom_bits_per_key is // number of bits used for bloom filer per key. hash_table_ratio is // the desired utilization of the hash table used for prefix hashing. // hash_table_ratio = number of prefixes / #buckets in the hash table // hash_table_ratio = 0 means skip hash table but only replying on binary // search. // index_sparseness determines index interval for keys // inside the same prefix. It will be the maximum number of linear search // required after hash and binary search. // index_sparseness = 0 means index for every key. // huge_page_tlb_size determines whether to allocate hash indexes from huge // page TLB and the page size if allocating from there. See comments of // Arena::AllocateAligned() for details. explicit PlainTableFactory( const PlainTableOptions& _table_options = PlainTableOptions()) : table_options_(_table_options) {} const char* Name() const override { return "PlainTable"; } Status NewTableReader(const TableReaderOptions& table_reader_options, unique_ptr&& file, uint64_t file_size, unique_ptr* table) const override; TableBuilder* NewTableBuilder( const TableBuilderOptions& table_builder_options, uint32_t column_family_id, WritableFileWriter* file) const override; std::string GetPrintableTableOptions() const override; const PlainTableOptions& table_options() const; static const char kValueTypeSeqId0 = char(0xFF); // Sanitizes the specified DB Options. Status SanitizeOptions(const DBOptions& db_opts, const ColumnFamilyOptions& cf_opts) const override { return Status::OK(); } void* GetOptions() override { return &table_options_; } private: PlainTableOptions table_options_; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_index.cc000066400000000000000000000164451267530634000202270ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #include #include "table/plain_table_index.h" #include "util/coding.h" #include "util/hash.h" namespace rocksdb { namespace { inline uint32_t GetBucketIdFromHash(uint32_t hash, uint32_t num_buckets) { assert(num_buckets > 0); return hash % num_buckets; } } Status PlainTableIndex::InitFromRawData(Slice data) { if (!GetVarint32(&data, &index_size_)) { return Status::Corruption("Couldn't read the index size!"); } assert(index_size_ > 0); if (!GetVarint32(&data, &num_prefixes_)) { return Status::Corruption("Couldn't read the index size!"); } sub_index_size_ = static_cast(data.size()) - index_size_ * kOffsetLen; char* index_data_begin = const_cast(data.data()); index_ = reinterpret_cast(index_data_begin); sub_index_ = reinterpret_cast(index_ + index_size_); return Status::OK(); } PlainTableIndex::IndexSearchResult PlainTableIndex::GetOffset( uint32_t prefix_hash, uint32_t* bucket_value) const { int bucket = GetBucketIdFromHash(prefix_hash, index_size_); *bucket_value = index_[bucket]; if ((*bucket_value & kSubIndexMask) == kSubIndexMask) { *bucket_value ^= kSubIndexMask; return kSubindex; } if (*bucket_value >= kMaxFileSize) { return kNoPrefixForBucket; } else { // point directly to the file return kDirectToFile; } } void PlainTableIndexBuilder::IndexRecordList::AddRecord(uint32_t hash, uint32_t offset) { if (num_records_in_current_group_ == kNumRecordsPerGroup) { current_group_ = AllocateNewGroup(); num_records_in_current_group_ = 0; } auto& new_record = current_group_[num_records_in_current_group_++]; new_record.hash = hash; new_record.offset = offset; new_record.next = nullptr; } void PlainTableIndexBuilder::AddKeyPrefix(Slice key_prefix_slice, uint32_t key_offset) { if (is_first_record_ || prev_key_prefix_ != key_prefix_slice.ToString()) { ++num_prefixes_; if (!is_first_record_) { keys_per_prefix_hist_.Add(num_keys_per_prefix_); } num_keys_per_prefix_ = 0; prev_key_prefix_ = key_prefix_slice.ToString(); prev_key_prefix_hash_ = GetSliceHash(key_prefix_slice); due_index_ = true; } if (due_index_) { // Add an index key for every kIndexIntervalForSamePrefixKeys keys record_list_.AddRecord(prev_key_prefix_hash_, key_offset); due_index_ = false; } num_keys_per_prefix_++; if (index_sparseness_ == 0 || num_keys_per_prefix_ % index_sparseness_ == 0) { due_index_ = true; } is_first_record_ = false; } Slice PlainTableIndexBuilder::Finish() { AllocateIndex(); std::vector hash_to_offsets(index_size_, nullptr); std::vector entries_per_bucket(index_size_, 0); BucketizeIndexes(&hash_to_offsets, &entries_per_bucket); keys_per_prefix_hist_.Add(num_keys_per_prefix_); Log(InfoLogLevel::INFO_LEVEL, ioptions_.info_log, "Number of Keys per prefix Histogram: %s", keys_per_prefix_hist_.ToString().c_str()); // From the temp data structure, populate indexes. return FillIndexes(hash_to_offsets, entries_per_bucket); } void PlainTableIndexBuilder::AllocateIndex() { if (prefix_extractor_ == nullptr || hash_table_ratio_ <= 0) { // Fall back to pure binary search if the user fails to specify a prefix // extractor. index_size_ = 1; } else { double hash_table_size_multipier = 1.0 / hash_table_ratio_; index_size_ = static_cast(num_prefixes_ * hash_table_size_multipier) + 1; assert(index_size_ > 0); } } void PlainTableIndexBuilder::BucketizeIndexes( std::vector* hash_to_offsets, std::vector* entries_per_bucket) { bool first = true; uint32_t prev_hash = 0; size_t num_records = record_list_.GetNumRecords(); for (size_t i = 0; i < num_records; i++) { IndexRecord* index_record = record_list_.At(i); uint32_t cur_hash = index_record->hash; if (first || prev_hash != cur_hash) { prev_hash = cur_hash; first = false; } uint32_t bucket = GetBucketIdFromHash(cur_hash, index_size_); IndexRecord* prev_bucket_head = (*hash_to_offsets)[bucket]; index_record->next = prev_bucket_head; (*hash_to_offsets)[bucket] = index_record; (*entries_per_bucket)[bucket]++; } sub_index_size_ = 0; for (auto entry_count : *entries_per_bucket) { if (entry_count <= 1) { continue; } // Only buckets with more than 1 entry will have subindex. sub_index_size_ += VarintLength(entry_count); // total bytes needed to store these entries' in-file offsets. sub_index_size_ += entry_count * PlainTableIndex::kOffsetLen; } } Slice PlainTableIndexBuilder::FillIndexes( const std::vector& hash_to_offsets, const std::vector& entries_per_bucket) { Log(InfoLogLevel::DEBUG_LEVEL, ioptions_.info_log, "Reserving %" PRIu32 " bytes for plain table's sub_index", sub_index_size_); auto total_allocate_size = GetTotalSize(); char* allocated = arena_->AllocateAligned( total_allocate_size, huge_page_tlb_size_, ioptions_.info_log); auto temp_ptr = EncodeVarint32(allocated, index_size_); uint32_t* index = reinterpret_cast(EncodeVarint32(temp_ptr, num_prefixes_)); char* sub_index = reinterpret_cast(index + index_size_); uint32_t sub_index_offset = 0; for (uint32_t i = 0; i < index_size_; i++) { uint32_t num_keys_for_bucket = entries_per_bucket[i]; switch (num_keys_for_bucket) { case 0: // No key for bucket index[i] = PlainTableIndex::kMaxFileSize; break; case 1: // point directly to the file offset index[i] = hash_to_offsets[i]->offset; break; default: // point to second level indexes. index[i] = sub_index_offset | PlainTableIndex::kSubIndexMask; char* prev_ptr = &sub_index[sub_index_offset]; char* cur_ptr = EncodeVarint32(prev_ptr, num_keys_for_bucket); sub_index_offset += static_cast(cur_ptr - prev_ptr); char* sub_index_pos = &sub_index[sub_index_offset]; IndexRecord* record = hash_to_offsets[i]; int j; for (j = num_keys_for_bucket - 1; j >= 0 && record; j--, record = record->next) { EncodeFixed32(sub_index_pos + j * sizeof(uint32_t), record->offset); } assert(j == -1 && record == nullptr); sub_index_offset += PlainTableIndex::kOffsetLen * num_keys_for_bucket; assert(sub_index_offset <= sub_index_size_); break; } } assert(sub_index_offset == sub_index_size_); Log(InfoLogLevel::DEBUG_LEVEL, ioptions_.info_log, "hash table size: %d, suffix_map length %" ROCKSDB_PRIszt, index_size_, sub_index_size_); return Slice(allocated, GetTotalSize()); } const std::string PlainTableIndexBuilder::kPlainTableIndexBlock = "PlainTableIndexBlock"; }; // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_index.h000066400000000000000000000160741267530634000200670ustar00rootroot00000000000000// Copyright (c) 2014, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include #include "db/dbformat.h" #include "rocksdb/options.h" #include "util/murmurhash.h" #include "util/hash.h" #include "util/arena.h" #include "util/histogram.h" namespace rocksdb { // PlainTableIndex contains buckets size of index_size_, each is a // 32-bit integer. The lower 31 bits contain an offset value (explained below) // and the first bit of the integer indicates type of the offset. // // +--------------+------------------------------------------------------+ // | Flag (1 bit) | Offset to binary search buffer or file (31 bits) + // +--------------+------------------------------------------------------+ // // Explanation for the "flag bit": // // 0 indicates that the bucket contains only one prefix (no conflict when // hashing this prefix), whose first row starts from this offset of the // file. // 1 indicates that the bucket contains more than one prefixes, or there // are too many rows for one prefix so we need a binary search for it. In // this case, the offset indicates the offset of sub_index_ holding the // binary search indexes of keys for those rows. Those binary search indexes // are organized in this way: // // The first 4 bytes, indicate how many indexes (N) are stored after it. After // it, there are N 32-bit integers, each points of an offset of the file, // which // points to starting of a row. Those offsets need to be guaranteed to be in // ascending order so the keys they are pointing to are also in ascending // order // to make sure we can use them to do binary searches. Below is visual // presentation of a bucket. // // // number_of_records: varint32 // record 1 file offset: fixedint32 // record 2 file offset: fixedint32 // .... // record N file offset: fixedint32 // class PlainTableIndex { public: enum IndexSearchResult { kNoPrefixForBucket = 0, kDirectToFile = 1, kSubindex = 2 }; explicit PlainTableIndex(Slice data) { InitFromRawData(data); } PlainTableIndex() : index_size_(0), sub_index_size_(0), num_prefixes_(0), index_(nullptr), sub_index_(nullptr) {} IndexSearchResult GetOffset(uint32_t prefix_hash, uint32_t* bucket_value) const; Status InitFromRawData(Slice data); const char* GetSubIndexBasePtrAndUpperBound(uint32_t offset, uint32_t* upper_bound) const { const char* index_ptr = &sub_index_[offset]; return GetVarint32Ptr(index_ptr, index_ptr + 4, upper_bound); } uint32_t GetIndexSize() const { return index_size_; } uint32_t GetSubIndexSize() const { return sub_index_size_; } uint32_t GetNumPrefixes() const { return num_prefixes_; } static const uint64_t kMaxFileSize = (1u << 31) - 1; static const uint32_t kSubIndexMask = 0x80000000; static const size_t kOffsetLen = sizeof(uint32_t); private: uint32_t index_size_; uint32_t sub_index_size_; uint32_t num_prefixes_; uint32_t* index_; char* sub_index_; }; // PlainTableIndexBuilder is used to create plain table index. // After calling Finish(), it returns Slice, which is usually // used either to initialize PlainTableIndex or // to save index to sst file. // For more details about the index, please refer to: // https://github.com/facebook/rocksdb/wiki/PlainTable-Format // #wiki-in-memory-index-format class PlainTableIndexBuilder { public: PlainTableIndexBuilder(Arena* arena, const ImmutableCFOptions& ioptions, size_t index_sparseness, double hash_table_ratio, size_t huge_page_tlb_size) : arena_(arena), ioptions_(ioptions), record_list_(kRecordsPerGroup), is_first_record_(true), due_index_(false), num_prefixes_(0), num_keys_per_prefix_(0), prev_key_prefix_hash_(0), index_sparseness_(index_sparseness), prefix_extractor_(ioptions.prefix_extractor), hash_table_ratio_(hash_table_ratio), huge_page_tlb_size_(huge_page_tlb_size) {} void AddKeyPrefix(Slice key_prefix_slice, uint32_t key_offset); Slice Finish(); uint32_t GetTotalSize() const { return VarintLength(index_size_) + VarintLength(num_prefixes_) + PlainTableIndex::kOffsetLen * index_size_ + sub_index_size_; } static const std::string kPlainTableIndexBlock; private: struct IndexRecord { uint32_t hash; // hash of the prefix uint32_t offset; // offset of a row IndexRecord* next; }; // Helper class to track all the index records class IndexRecordList { public: explicit IndexRecordList(size_t num_records_per_group) : kNumRecordsPerGroup(num_records_per_group), current_group_(nullptr), num_records_in_current_group_(num_records_per_group) {} ~IndexRecordList() { for (size_t i = 0; i < groups_.size(); i++) { delete[] groups_[i]; } } void AddRecord(uint32_t hash, uint32_t offset); size_t GetNumRecords() const { return (groups_.size() - 1) * kNumRecordsPerGroup + num_records_in_current_group_; } IndexRecord* At(size_t index) { return &(groups_[index / kNumRecordsPerGroup] [index % kNumRecordsPerGroup]); } private: IndexRecord* AllocateNewGroup() { IndexRecord* result = new IndexRecord[kNumRecordsPerGroup]; groups_.push_back(result); return result; } // Each group in `groups_` contains fix-sized records (determined by // kNumRecordsPerGroup). Which can help us minimize the cost if resizing // occurs. const size_t kNumRecordsPerGroup; IndexRecord* current_group_; // List of arrays allocated std::vector groups_; size_t num_records_in_current_group_; }; void AllocateIndex(); // Internal helper function to bucket index record list to hash buckets. void BucketizeIndexes(std::vector* hash_to_offsets, std::vector* entries_per_bucket); // Internal helper class to fill the indexes and bloom filters to internal // data structures. Slice FillIndexes(const std::vector& hash_to_offsets, const std::vector& entries_per_bucket); Arena* arena_; const ImmutableCFOptions ioptions_; HistogramImpl keys_per_prefix_hist_; IndexRecordList record_list_; bool is_first_record_; bool due_index_; uint32_t num_prefixes_; uint32_t num_keys_per_prefix_; uint32_t prev_key_prefix_hash_; size_t index_sparseness_; uint32_t index_size_; uint32_t sub_index_size_; const SliceTransform* prefix_extractor_; double hash_table_ratio_; size_t huge_page_tlb_size_; std::string prev_key_prefix_; static const size_t kRecordsPerGroup = 256; }; }; // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_key_coding.cc000066400000000000000000000424531267530634000212310ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef ROCKSDB_LITE #include "table/plain_table_key_coding.h" #include #include #include "db/dbformat.h" #include "table/plain_table_reader.h" #include "table/plain_table_factory.h" #include "util/file_reader_writer.h" namespace rocksdb { enum PlainTableEntryType : unsigned char { kFullKey = 0, kPrefixFromPreviousKey = 1, kKeySuffix = 2, }; namespace { // Control byte: // First two bits indicate type of entry // Other bytes are inlined sizes. If all bits are 1 (0x03F), overflow bytes // are used. key_size-0x3F will be encoded as a variint32 after this bytes. const unsigned char kSizeInlineLimit = 0x3F; // Return 0 for error size_t EncodeSize(PlainTableEntryType type, uint32_t key_size, char* out_buffer) { out_buffer[0] = type << 6; if (key_size < static_cast(kSizeInlineLimit)) { // size inlined out_buffer[0] |= static_cast(key_size); return 1; } else { out_buffer[0] |= kSizeInlineLimit; char* ptr = EncodeVarint32(out_buffer + 1, key_size - kSizeInlineLimit); return ptr - out_buffer; } } } // namespace // Fill bytes_read with number of bytes read. inline Status PlainTableKeyDecoder::DecodeSize(uint32_t start_offset, PlainTableEntryType* entry_type, uint32_t* key_size, uint32_t* bytes_read) { Slice next_byte_slice; bool success = file_reader_.Read(start_offset, 1, &next_byte_slice); if (!success) { return file_reader_.status(); } *entry_type = static_cast( (static_cast(next_byte_slice[0]) & ~kSizeInlineLimit) >> 6); char inline_key_size = next_byte_slice[0] & kSizeInlineLimit; if (inline_key_size < kSizeInlineLimit) { *key_size = inline_key_size; *bytes_read = 1; return Status::OK(); } else { uint32_t extra_size; uint32_t tmp_bytes_read; success = file_reader_.ReadVarint32(start_offset + 1, &extra_size, &tmp_bytes_read); if (!success) { return file_reader_.status(); } assert(tmp_bytes_read > 0); *key_size = kSizeInlineLimit + extra_size; *bytes_read = tmp_bytes_read + 1; return Status::OK(); } } Status PlainTableKeyEncoder::AppendKey(const Slice& key, WritableFileWriter* file, uint64_t* offset, char* meta_bytes_buf, size_t* meta_bytes_buf_size) { ParsedInternalKey parsed_key; if (!ParseInternalKey(key, &parsed_key)) { return Status::Corruption(Slice()); } Slice key_to_write = key; // Portion of internal key to write out. uint32_t user_key_size = static_cast(key.size() - 8); if (encoding_type_ == kPlain) { if (fixed_user_key_len_ == kPlainTableVariableLength) { // Write key length char key_size_buf[5]; // tmp buffer for key size as varint32 char* ptr = EncodeVarint32(key_size_buf, user_key_size); assert(ptr <= key_size_buf + sizeof(key_size_buf)); auto len = ptr - key_size_buf; Status s = file->Append(Slice(key_size_buf, len)); if (!s.ok()) { return s; } *offset += len; } } else { assert(encoding_type_ == kPrefix); char size_bytes[12]; size_t size_bytes_pos = 0; Slice prefix = prefix_extractor_->Transform(Slice(key.data(), user_key_size)); if (key_count_for_prefix_ == 0 || prefix != pre_prefix_.GetKey() || key_count_for_prefix_ % index_sparseness_ == 0) { key_count_for_prefix_ = 1; pre_prefix_.SetKey(prefix); size_bytes_pos += EncodeSize(kFullKey, user_key_size, size_bytes); Status s = file->Append(Slice(size_bytes, size_bytes_pos)); if (!s.ok()) { return s; } *offset += size_bytes_pos; } else { key_count_for_prefix_++; if (key_count_for_prefix_ == 2) { // For second key within a prefix, need to encode prefix length size_bytes_pos += EncodeSize(kPrefixFromPreviousKey, static_cast(pre_prefix_.GetKey().size()), size_bytes + size_bytes_pos); } uint32_t prefix_len = static_cast(pre_prefix_.GetKey().size()); size_bytes_pos += EncodeSize(kKeySuffix, user_key_size - prefix_len, size_bytes + size_bytes_pos); Status s = file->Append(Slice(size_bytes, size_bytes_pos)); if (!s.ok()) { return s; } *offset += size_bytes_pos; key_to_write = Slice(key.data() + prefix_len, key.size() - prefix_len); } } // Encode full key // For value size as varint32 (up to 5 bytes). // If the row is of value type with seqId 0, flush the special flag together // in this buffer to safe one file append call, which takes 1 byte. if (parsed_key.sequence == 0 && parsed_key.type == kTypeValue) { Status s = file->Append(Slice(key_to_write.data(), key_to_write.size() - 8)); if (!s.ok()) { return s; } *offset += key_to_write.size() - 8; meta_bytes_buf[*meta_bytes_buf_size] = PlainTableFactory::kValueTypeSeqId0; *meta_bytes_buf_size += 1; } else { file->Append(key_to_write); *offset += key_to_write.size(); } return Status::OK(); } Slice PlainTableFileReader::GetFromBuffer(Buffer* buffer, uint32_t file_offset, uint32_t len) { assert(file_offset + len <= file_info_->data_end_offset); return Slice(buffer->buf.get() + (file_offset - buffer->buf_start_offset), len); } bool PlainTableFileReader::ReadNonMmap(uint32_t file_offset, uint32_t len, Slice* out) { const uint32_t kPrefetchSize = 256u; // Try to read from buffers. for (uint32_t i = 0; i < num_buf_; i++) { Buffer* buffer = buffers_[num_buf_ - 1 - i].get(); if (file_offset >= buffer->buf_start_offset && file_offset + len <= buffer->buf_start_offset + buffer->buf_len) { *out = GetFromBuffer(buffer, file_offset, len); return true; } } Buffer* new_buffer; // Data needed is not in any of the buffer. Allocate a new buffer. if (num_buf_ < buffers_.size()) { // Add a new buffer new_buffer = new Buffer(); buffers_[num_buf_++].reset(new_buffer); } else { // Now simply replace the last buffer. Can improve the placement policy // if needed. new_buffer = buffers_[num_buf_ - 1].get(); } assert(file_offset + len <= file_info_->data_end_offset); uint32_t size_to_read = std::min(file_info_->data_end_offset - file_offset, std::max(kPrefetchSize, len)); if (size_to_read > new_buffer->buf_capacity) { new_buffer->buf.reset(new char[size_to_read]); new_buffer->buf_capacity = size_to_read; new_buffer->buf_len = 0; } Slice read_result; Status s = file_info_->file->Read(file_offset, size_to_read, &read_result, new_buffer->buf.get()); if (!s.ok()) { status_ = s; return false; } new_buffer->buf_start_offset = file_offset; new_buffer->buf_len = size_to_read; *out = GetFromBuffer(new_buffer, file_offset, len); return true; } inline bool PlainTableFileReader::ReadVarint32(uint32_t offset, uint32_t* out, uint32_t* bytes_read) { if (file_info_->is_mmap_mode) { const char* start = file_info_->file_data.data() + offset; const char* limit = file_info_->file_data.data() + file_info_->data_end_offset; const char* key_ptr = GetVarint32Ptr(start, limit, out); assert(key_ptr != nullptr); *bytes_read = static_cast(key_ptr - start); return true; } else { return ReadVarint32NonMmap(offset, out, bytes_read); } } bool PlainTableFileReader::ReadVarint32NonMmap(uint32_t offset, uint32_t* out, uint32_t* bytes_read) { const char* start; const char* limit; const uint32_t kMaxVarInt32Size = 6u; uint32_t bytes_to_read = std::min(file_info_->data_end_offset - offset, kMaxVarInt32Size); Slice bytes; if (!Read(offset, bytes_to_read, &bytes)) { return false; } start = bytes.data(); limit = bytes.data() + bytes.size(); const char* key_ptr = GetVarint32Ptr(start, limit, out); *bytes_read = (key_ptr != nullptr) ? static_cast(key_ptr - start) : 0; return true; } Status PlainTableKeyDecoder::ReadInternalKey( uint32_t file_offset, uint32_t user_key_size, ParsedInternalKey* parsed_key, uint32_t* bytes_read, bool* internal_key_valid, Slice* internal_key) { Slice tmp_slice; bool success = file_reader_.Read(file_offset, user_key_size + 1, &tmp_slice); if (!success) { return file_reader_.status(); } if (tmp_slice[user_key_size] == PlainTableFactory::kValueTypeSeqId0) { // Special encoding for the row with seqID=0 parsed_key->user_key = Slice(tmp_slice.data(), user_key_size); parsed_key->sequence = 0; parsed_key->type = kTypeValue; *bytes_read += user_key_size + 1; *internal_key_valid = false; } else { success = file_reader_.Read(file_offset, user_key_size + 8, internal_key); if (!success) { return file_reader_.status(); } *internal_key_valid = true; if (!ParseInternalKey(*internal_key, parsed_key)) { return Status::Corruption( Slice("Incorrect value type found when reading the next key")); } *bytes_read += user_key_size + 8; } return Status::OK(); } Status PlainTableKeyDecoder::NextPlainEncodingKey(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, uint32_t* bytes_read, bool* seekable) { uint32_t user_key_size = 0; Status s; if (fixed_user_key_len_ != kPlainTableVariableLength) { user_key_size = fixed_user_key_len_; } else { uint32_t tmp_size = 0; uint32_t tmp_read; bool success = file_reader_.ReadVarint32(start_offset, &tmp_size, &tmp_read); if (!success) { return file_reader_.status(); } assert(tmp_read > 0); user_key_size = tmp_size; *bytes_read = tmp_read; } // dummy initial value to avoid compiler complain bool decoded_internal_key_valid = true; Slice decoded_internal_key; s = ReadInternalKey(start_offset + *bytes_read, user_key_size, parsed_key, bytes_read, &decoded_internal_key_valid, &decoded_internal_key); if (!s.ok()) { return s; } if (!file_reader_.file_info()->is_mmap_mode) { cur_key_.SetInternalKey(*parsed_key); parsed_key->user_key = Slice(cur_key_.GetKey().data(), user_key_size); if (internal_key != nullptr) { *internal_key = cur_key_.GetKey(); } } else if (internal_key != nullptr) { if (decoded_internal_key_valid) { *internal_key = decoded_internal_key; } else { // Need to copy out the internal key cur_key_.SetInternalKey(*parsed_key); *internal_key = cur_key_.GetKey(); } } return Status::OK(); } Status PlainTableKeyDecoder::NextPrefixEncodingKey( uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, uint32_t* bytes_read, bool* seekable) { PlainTableEntryType entry_type; bool expect_suffix = false; Status s; do { uint32_t size = 0; // dummy initial value to avoid compiler complain bool decoded_internal_key_valid = true; uint32_t my_bytes_read = 0; s = DecodeSize(start_offset + *bytes_read, &entry_type, &size, &my_bytes_read); if (!s.ok()) { return s; } if (my_bytes_read == 0) { return Status::Corruption("Unexpected EOF when reading size of the key"); } *bytes_read += my_bytes_read; switch (entry_type) { case kFullKey: { expect_suffix = false; Slice decoded_internal_key; s = ReadInternalKey(start_offset + *bytes_read, size, parsed_key, bytes_read, &decoded_internal_key_valid, &decoded_internal_key); if (!s.ok()) { return s; } if (!file_reader_.file_info()->is_mmap_mode || (internal_key != nullptr && !decoded_internal_key_valid)) { // In non-mmap mode, always need to make a copy of keys returned to // users, because after reading value for the key, the key might // be invalid. cur_key_.SetInternalKey(*parsed_key); saved_user_key_ = cur_key_.GetKey(); if (!file_reader_.file_info()->is_mmap_mode) { parsed_key->user_key = Slice(cur_key_.GetKey().data(), size); } if (internal_key != nullptr) { *internal_key = cur_key_.GetKey(); } } else { if (internal_key != nullptr) { *internal_key = decoded_internal_key; } saved_user_key_ = parsed_key->user_key; } break; } case kPrefixFromPreviousKey: { if (seekable != nullptr) { *seekable = false; } prefix_len_ = size; assert(prefix_extractor_ == nullptr || prefix_extractor_->Transform(saved_user_key_).size() == prefix_len_); // Need read another size flag for suffix expect_suffix = true; break; } case kKeySuffix: { expect_suffix = false; if (seekable != nullptr) { *seekable = false; } Slice tmp_slice; s = ReadInternalKey(start_offset + *bytes_read, size, parsed_key, bytes_read, &decoded_internal_key_valid, &tmp_slice); if (!s.ok()) { return s; } if (!file_reader_.file_info()->is_mmap_mode) { // In non-mmap mode, we need to make a copy of keys returned to // users, because after reading value for the key, the key might // be invalid. // saved_user_key_ points to cur_key_. We are making a copy of // the prefix part to another string, and construct the current // key from the prefix part and the suffix part back to cur_key_. std::string tmp = Slice(saved_user_key_.data(), prefix_len_).ToString(); cur_key_.Reserve(prefix_len_ + size); cur_key_.SetInternalKey(tmp, *parsed_key); parsed_key->user_key = Slice(cur_key_.GetKey().data(), prefix_len_ + size); saved_user_key_ = cur_key_.GetKey(); } else { cur_key_.Reserve(prefix_len_ + size); cur_key_.SetInternalKey(Slice(saved_user_key_.data(), prefix_len_), *parsed_key); } parsed_key->user_key = ExtractUserKey(cur_key_.GetKey()); if (internal_key != nullptr) { *internal_key = cur_key_.GetKey(); } break; } default: return Status::Corruption("Un-identified size flag."); } } while (expect_suffix); // Another round if suffix is expected. return Status::OK(); } Status PlainTableKeyDecoder::NextKey(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, Slice* value, uint32_t* bytes_read, bool* seekable) { assert(value != nullptr); Status s = NextKeyNoValue(start_offset, parsed_key, internal_key, bytes_read, seekable); if (s.ok()) { assert(bytes_read != nullptr); uint32_t value_size; uint32_t value_size_bytes; bool success = file_reader_.ReadVarint32(start_offset + *bytes_read, &value_size, &value_size_bytes); if (!success) { return file_reader_.status(); } if (value_size_bytes == 0) { return Status::Corruption( "Unexpected EOF when reading the next value's size."); } *bytes_read += value_size_bytes; success = file_reader_.Read(start_offset + *bytes_read, value_size, value); if (!success) { return file_reader_.status(); } *bytes_read += value_size; } return s; } Status PlainTableKeyDecoder::NextKeyNoValue(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, uint32_t* bytes_read, bool* seekable) { *bytes_read = 0; if (seekable != nullptr) { *seekable = true; } Status s; if (encoding_type_ == kPlain) { return NextPlainEncodingKey(start_offset, parsed_key, internal_key, bytes_read, seekable); } else { assert(encoding_type_ == kPrefix); return NextPrefixEncodingKey(start_offset, parsed_key, internal_key, bytes_read, seekable); } } } // namespace rocksdb #endif // ROCKSDB_LIT rocksdb-4.5.1/table/plain_table_key_coding.h000066400000000000000000000166351267530634000210760ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #ifndef ROCKSDB_LITE #include #include "rocksdb/slice.h" #include "db/dbformat.h" #include "table/plain_table_reader.h" namespace rocksdb { class WritableFile; struct ParsedInternalKey; struct PlainTableReaderFileInfo; enum PlainTableEntryType : unsigned char; // Helper class to write out a key to an output file // Actual data format of the key is documented in plain_table_factory.h class PlainTableKeyEncoder { public: explicit PlainTableKeyEncoder(EncodingType encoding_type, uint32_t user_key_len, const SliceTransform* prefix_extractor, size_t index_sparseness) : encoding_type_((prefix_extractor != nullptr) ? encoding_type : kPlain), fixed_user_key_len_(user_key_len), prefix_extractor_(prefix_extractor), index_sparseness_((index_sparseness > 1) ? index_sparseness : 1), key_count_for_prefix_(0) {} // key: the key to write out, in the format of internal key. // file: the output file to write out // offset: offset in the file. Needs to be updated after appending bytes // for the key // meta_bytes_buf: buffer for extra meta bytes // meta_bytes_buf_size: offset to append extra meta bytes. Will be updated // if meta_bytes_buf is updated. Status AppendKey(const Slice& key, WritableFileWriter* file, uint64_t* offset, char* meta_bytes_buf, size_t* meta_bytes_buf_size); // Return actual encoding type to be picked EncodingType GetEncodingType() { return encoding_type_; } private: EncodingType encoding_type_; uint32_t fixed_user_key_len_; const SliceTransform* prefix_extractor_; const size_t index_sparseness_; size_t key_count_for_prefix_; IterKey pre_prefix_; }; class PlainTableFileReader { public: explicit PlainTableFileReader(const PlainTableReaderFileInfo* _file_info) : file_info_(_file_info), num_buf_(0) {} // In mmaped mode, the results point to mmaped area of the file, which // means it is always valid before closing the file. // In non-mmap mode, the results point to an internal buffer. If the caller // makes another read call, the results may not be valid. So callers should // make a copy when needed. // In order to save read calls to files, we keep two internal buffers: // the first read and the most recent read. This is efficient because it // columns these two common use cases: // (1) hash index only identify one location, we read the key to verify // the location, and read key and value if it is the right location. // (2) after hash index checking, we identify two locations (because of // hash bucket conflicts), we binary search the two location to see // which one is what we need and start to read from the location. // These two most common use cases will be covered by the two buffers // so that we don't need to re-read the same location. // Currently we keep a fixed size buffer. If a read doesn't exactly fit // the buffer, we replace the second buffer with the location user reads. // // If return false, status code is stored in status_. bool Read(uint32_t file_offset, uint32_t len, Slice* out) { if (file_info_->is_mmap_mode) { assert(file_offset + len <= file_info_->data_end_offset); *out = Slice(file_info_->file_data.data() + file_offset, len); return true; } else { return ReadNonMmap(file_offset, len, out); } } // If return false, status code is stored in status_. bool ReadNonMmap(uint32_t file_offset, uint32_t len, Slice* output); // *bytes_read = 0 means eof. false means failure and status is saved // in status_. Not directly returning Status to save copying status // object to map previous performance of mmap mode. inline bool ReadVarint32(uint32_t offset, uint32_t* output, uint32_t* bytes_read); bool ReadVarint32NonMmap(uint32_t offset, uint32_t* output, uint32_t* bytes_read); Status status() const { return status_; } const PlainTableReaderFileInfo* file_info() { return file_info_; } private: const PlainTableReaderFileInfo* file_info_; struct Buffer { Buffer() : buf_start_offset(0), buf_len(0), buf_capacity(0) {} std::unique_ptr buf; uint32_t buf_start_offset; uint32_t buf_len; uint32_t buf_capacity; }; // Keep buffers for two recent reads. std::array, 2> buffers_; uint32_t num_buf_; Status status_; Slice GetFromBuffer(Buffer* buf, uint32_t file_offset, uint32_t len); }; // A helper class to decode keys from input buffer // Actual data format of the key is documented in plain_table_factory.h class PlainTableKeyDecoder { public: explicit PlainTableKeyDecoder(const PlainTableReaderFileInfo* file_info, EncodingType encoding_type, uint32_t user_key_len, const SliceTransform* prefix_extractor) : file_reader_(file_info), encoding_type_(encoding_type), prefix_len_(0), fixed_user_key_len_(user_key_len), prefix_extractor_(prefix_extractor), in_prefix_(false) {} // Find the next key. // start: char array where the key starts. // limit: boundary of the char array // parsed_key: the output of the result key // internal_key: if not null, fill with the output of the result key in // un-parsed format // bytes_read: how many bytes read from start. Output // seekable: whether key can be read from this place. Used when building // indexes. Output. Status NextKey(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, Slice* value, uint32_t* bytes_read, bool* seekable = nullptr); Status NextKeyNoValue(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, uint32_t* bytes_read, bool* seekable = nullptr); PlainTableFileReader file_reader_; EncodingType encoding_type_; uint32_t prefix_len_; uint32_t fixed_user_key_len_; Slice saved_user_key_; IterKey cur_key_; const SliceTransform* prefix_extractor_; bool in_prefix_; private: Status NextPlainEncodingKey(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, uint32_t* bytes_read, bool* seekable = nullptr); Status NextPrefixEncodingKey(uint32_t start_offset, ParsedInternalKey* parsed_key, Slice* internal_key, uint32_t* bytes_read, bool* seekable = nullptr); Status ReadInternalKey(uint32_t file_offset, uint32_t user_key_size, ParsedInternalKey* parsed_key, uint32_t* bytes_read, bool* internal_key_valid, Slice* internal_key); inline Status DecodeSize(uint32_t start_offset, PlainTableEntryType* entry_type, uint32_t* key_size, uint32_t* bytes_read); }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_reader.cc000066400000000000000000000557431267530634000203660ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #ifndef ROCKSDB_LITE #include "table/plain_table_reader.h" #include #include #include "db/dbformat.h" #include "rocksdb/cache.h" #include "rocksdb/comparator.h" #include "rocksdb/env.h" #include "rocksdb/filter_policy.h" #include "rocksdb/options.h" #include "rocksdb/statistics.h" #include "table/block.h" #include "table/bloom_block.h" #include "table/filter_block.h" #include "table/format.h" #include "table/internal_iterator.h" #include "table/meta_blocks.h" #include "table/two_level_iterator.h" #include "table/plain_table_factory.h" #include "table/plain_table_key_coding.h" #include "table/get_context.h" #include "util/arena.h" #include "util/coding.h" #include "util/dynamic_bloom.h" #include "util/hash.h" #include "util/histogram.h" #include "util/murmurhash.h" #include "util/perf_context_imp.h" #include "util/stop_watch.h" #include "util/string_util.h" namespace rocksdb { namespace { // Safely getting a uint32_t element from a char array, where, starting from // `base`, every 4 bytes are considered as an fixed 32 bit integer. inline uint32_t GetFixed32Element(const char* base, size_t offset) { return DecodeFixed32(base + offset * sizeof(uint32_t)); } } // namespace // Iterator to iterate IndexedTable class PlainTableIterator : public InternalIterator { public: explicit PlainTableIterator(PlainTableReader* table, bool use_prefix_seek); ~PlainTableIterator(); bool Valid() const override; void SeekToFirst() override; void SeekToLast() override; void Seek(const Slice& target) override; void Next() override; void Prev() override; Slice key() const override; Slice value() const override; Status status() const override; private: PlainTableReader* table_; PlainTableKeyDecoder decoder_; bool use_prefix_seek_; uint32_t offset_; uint32_t next_offset_; Slice key_; Slice value_; Status status_; // No copying allowed PlainTableIterator(const PlainTableIterator&) = delete; void operator=(const Iterator&) = delete; }; extern const uint64_t kPlainTableMagicNumber; PlainTableReader::PlainTableReader(const ImmutableCFOptions& ioptions, unique_ptr&& file, const EnvOptions& storage_options, const InternalKeyComparator& icomparator, EncodingType encoding_type, uint64_t file_size, const TableProperties* table_properties) : internal_comparator_(icomparator), encoding_type_(encoding_type), full_scan_mode_(false), user_key_len_(static_cast(table_properties->fixed_key_len)), prefix_extractor_(ioptions.prefix_extractor), enable_bloom_(false), bloom_(6, nullptr), file_info_(std::move(file), storage_options, static_cast(table_properties->data_size)), ioptions_(ioptions), file_size_(file_size), table_properties_(nullptr) {} PlainTableReader::~PlainTableReader() { } Status PlainTableReader::Open(const ImmutableCFOptions& ioptions, const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, unique_ptr&& file, uint64_t file_size, unique_ptr* table_reader, const int bloom_bits_per_key, double hash_table_ratio, size_t index_sparseness, size_t huge_page_tlb_size, bool full_scan_mode) { if (file_size > PlainTableIndex::kMaxFileSize) { return Status::NotSupported("File is too large for PlainTableReader!"); } TableProperties* props = nullptr; auto s = ReadTableProperties(file.get(), file_size, kPlainTableMagicNumber, ioptions.env, ioptions.info_log, &props); if (!s.ok()) { return s; } assert(hash_table_ratio >= 0.0); auto& user_props = props->user_collected_properties; auto prefix_extractor_in_file = user_props.find(PlainTablePropertyNames::kPrefixExtractorName); if (!full_scan_mode && prefix_extractor_in_file != user_props.end()) { if (!ioptions.prefix_extractor) { return Status::InvalidArgument( "Prefix extractor is missing when opening a PlainTable built " "using a prefix extractor"); } else if (prefix_extractor_in_file->second.compare( ioptions.prefix_extractor->Name()) != 0) { return Status::InvalidArgument( "Prefix extractor given doesn't match the one used to build " "PlainTable"); } } EncodingType encoding_type = kPlain; auto encoding_type_prop = user_props.find(PlainTablePropertyNames::kEncodingType); if (encoding_type_prop != user_props.end()) { encoding_type = static_cast( DecodeFixed32(encoding_type_prop->second.c_str())); } std::unique_ptr new_reader(new PlainTableReader( ioptions, std::move(file), env_options, internal_comparator, encoding_type, file_size, props)); s = new_reader->MmapDataIfNeeded(); if (!s.ok()) { return s; } if (!full_scan_mode) { s = new_reader->PopulateIndex(props, bloom_bits_per_key, hash_table_ratio, index_sparseness, huge_page_tlb_size); if (!s.ok()) { return s; } } else { // Flag to indicate it is a full scan mode so that none of the indexes // can be used. new_reader->full_scan_mode_ = true; } *table_reader = std::move(new_reader); return s; } void PlainTableReader::SetupForCompaction() { } InternalIterator* PlainTableReader::NewIterator(const ReadOptions& options, Arena* arena, bool skip_filters) { if (options.total_order_seek && !IsTotalOrderMode()) { return NewErrorInternalIterator( Status::InvalidArgument("total_order_seek not supported"), arena); } if (arena == nullptr) { return new PlainTableIterator(this, prefix_extractor_ != nullptr); } else { auto mem = arena->AllocateAligned(sizeof(PlainTableIterator)); return new (mem) PlainTableIterator(this, prefix_extractor_ != nullptr); } } Status PlainTableReader::PopulateIndexRecordList( PlainTableIndexBuilder* index_builder, vector* prefix_hashes) { Slice prev_key_prefix_slice; std::string prev_key_prefix_buf; uint32_t pos = data_start_offset_; bool is_first_record = true; Slice key_prefix_slice; PlainTableKeyDecoder decoder(&file_info_, encoding_type_, user_key_len_, ioptions_.prefix_extractor); while (pos < file_info_.data_end_offset) { uint32_t key_offset = pos; ParsedInternalKey key; Slice value_slice; bool seekable = false; Status s = Next(&decoder, &pos, &key, nullptr, &value_slice, &seekable); if (!s.ok()) { return s; } key_prefix_slice = GetPrefix(key); if (enable_bloom_) { bloom_.AddHash(GetSliceHash(key.user_key)); } else { if (is_first_record || prev_key_prefix_slice != key_prefix_slice) { if (!is_first_record) { prefix_hashes->push_back(GetSliceHash(prev_key_prefix_slice)); } if (file_info_.is_mmap_mode) { prev_key_prefix_slice = key_prefix_slice; } else { prev_key_prefix_buf = key_prefix_slice.ToString(); prev_key_prefix_slice = prev_key_prefix_buf; } } } index_builder->AddKeyPrefix(GetPrefix(key), key_offset); if (!seekable && is_first_record) { return Status::Corruption("Key for a prefix is not seekable"); } is_first_record = false; } prefix_hashes->push_back(GetSliceHash(key_prefix_slice)); auto s = index_.InitFromRawData(index_builder->Finish()); return s; } void PlainTableReader::AllocateAndFillBloom(int bloom_bits_per_key, int num_prefixes, size_t huge_page_tlb_size, vector* prefix_hashes) { if (!IsTotalOrderMode()) { uint32_t bloom_total_bits = num_prefixes * bloom_bits_per_key; if (bloom_total_bits > 0) { enable_bloom_ = true; bloom_.SetTotalBits(&arena_, bloom_total_bits, ioptions_.bloom_locality, huge_page_tlb_size, ioptions_.info_log); FillBloom(prefix_hashes); } } } void PlainTableReader::FillBloom(vector* prefix_hashes) { assert(bloom_.IsInitialized()); for (auto prefix_hash : *prefix_hashes) { bloom_.AddHash(prefix_hash); } } Status PlainTableReader::MmapDataIfNeeded() { if (file_info_.is_mmap_mode) { // Get mmapped memory. return file_info_.file->Read(0, file_size_, &file_info_.file_data, nullptr); } return Status::OK(); } Status PlainTableReader::PopulateIndex(TableProperties* props, int bloom_bits_per_key, double hash_table_ratio, size_t index_sparseness, size_t huge_page_tlb_size) { assert(props != nullptr); table_properties_.reset(props); BlockContents bloom_block_contents; auto s = ReadMetaBlock(file_info_.file.get(), file_size_, kPlainTableMagicNumber, ioptions_.env, BloomBlockBuilder::kBloomBlock, &bloom_block_contents); bool index_in_file = s.ok(); BlockContents index_block_contents; s = ReadMetaBlock( file_info_.file.get(), file_size_, kPlainTableMagicNumber, ioptions_.env, PlainTableIndexBuilder::kPlainTableIndexBlock, &index_block_contents); index_in_file &= s.ok(); Slice* bloom_block; if (index_in_file) { // If bloom_block_contents.allocation is not empty (which will be the case // for non-mmap mode), it holds the alloated memory for the bloom block. // It needs to be kept alive to keep `bloom_block` valid. bloom_block_alloc_ = std::move(bloom_block_contents.allocation); bloom_block = &bloom_block_contents.data; } else { bloom_block = nullptr; } // index_in_file == true only if there are kBloomBlock and // kPlainTableIndexBlock in file Slice* index_block; if (index_in_file) { // If index_block_contents.allocation is not empty (which will be the case // for non-mmap mode), it holds the alloated memory for the index block. // It needs to be kept alive to keep `index_block` valid. index_block_alloc_ = std::move(index_block_contents.allocation); index_block = &index_block_contents.data; } else { index_block = nullptr; } if ((ioptions_.prefix_extractor == nullptr) && (hash_table_ratio != 0)) { // ioptions.prefix_extractor is requried for a hash-based look-up. return Status::NotSupported( "PlainTable requires a prefix extractor enable prefix hash mode."); } // First, read the whole file, for every kIndexIntervalForSamePrefixKeys rows // for a prefix (starting from the first one), generate a record of (hash, // offset) and append it to IndexRecordList, which is a data structure created // to store them. if (!index_in_file) { // Allocate bloom filter here for total order mode. if (IsTotalOrderMode()) { uint32_t num_bloom_bits = static_cast(table_properties_->num_entries) * bloom_bits_per_key; if (num_bloom_bits > 0) { enable_bloom_ = true; bloom_.SetTotalBits(&arena_, num_bloom_bits, ioptions_.bloom_locality, huge_page_tlb_size, ioptions_.info_log); } } } else { enable_bloom_ = true; auto num_blocks_property = props->user_collected_properties.find( PlainTablePropertyNames::kNumBloomBlocks); uint32_t num_blocks = 0; if (num_blocks_property != props->user_collected_properties.end()) { Slice temp_slice(num_blocks_property->second); if (!GetVarint32(&temp_slice, &num_blocks)) { num_blocks = 0; } } // cast away const qualifier, because bloom_ won't be changed bloom_.SetRawData( const_cast( reinterpret_cast(bloom_block->data())), static_cast(bloom_block->size()) * 8, num_blocks); } PlainTableIndexBuilder index_builder(&arena_, ioptions_, index_sparseness, hash_table_ratio, huge_page_tlb_size); std::vector prefix_hashes; if (!index_in_file) { s = PopulateIndexRecordList(&index_builder, &prefix_hashes); if (!s.ok()) { return s; } } else { s = index_.InitFromRawData(*index_block); if (!s.ok()) { return s; } } if (!index_in_file) { // Calculated bloom filter size and allocate memory for // bloom filter based on the number of prefixes, then fill it. AllocateAndFillBloom(bloom_bits_per_key, index_.GetNumPrefixes(), huge_page_tlb_size, &prefix_hashes); } // Fill two table properties. if (!index_in_file) { props->user_collected_properties["plain_table_hash_table_size"] = ToString(index_.GetIndexSize() * PlainTableIndex::kOffsetLen); props->user_collected_properties["plain_table_sub_index_size"] = ToString(index_.GetSubIndexSize()); } else { props->user_collected_properties["plain_table_hash_table_size"] = ToString(0); props->user_collected_properties["plain_table_sub_index_size"] = ToString(0); } return Status::OK(); } Status PlainTableReader::GetOffset(PlainTableKeyDecoder* decoder, const Slice& target, const Slice& prefix, uint32_t prefix_hash, bool& prefix_matched, uint32_t* offset) const { prefix_matched = false; uint32_t prefix_index_offset; auto res = index_.GetOffset(prefix_hash, &prefix_index_offset); if (res == PlainTableIndex::kNoPrefixForBucket) { *offset = file_info_.data_end_offset; return Status::OK(); } else if (res == PlainTableIndex::kDirectToFile) { *offset = prefix_index_offset; return Status::OK(); } // point to sub-index, need to do a binary search uint32_t upper_bound; const char* base_ptr = index_.GetSubIndexBasePtrAndUpperBound(prefix_index_offset, &upper_bound); uint32_t low = 0; uint32_t high = upper_bound; ParsedInternalKey mid_key; ParsedInternalKey parsed_target; if (!ParseInternalKey(target, &parsed_target)) { return Status::Corruption(Slice()); } // The key is between [low, high). Do a binary search between it. while (high - low > 1) { uint32_t mid = (high + low) / 2; uint32_t file_offset = GetFixed32Element(base_ptr, mid); uint32_t tmp; Status s = decoder->NextKeyNoValue(file_offset, &mid_key, nullptr, &tmp); if (!s.ok()) { return s; } int cmp_result = internal_comparator_.Compare(mid_key, parsed_target); if (cmp_result < 0) { low = mid; } else { if (cmp_result == 0) { // Happen to have found the exact key or target is smaller than the // first key after base_offset. prefix_matched = true; *offset = file_offset; return Status::OK(); } else { high = mid; } } } // Both of the key at the position low or low+1 could share the same // prefix as target. We need to rule out one of them to avoid to go // to the wrong prefix. ParsedInternalKey low_key; uint32_t tmp; uint32_t low_key_offset = GetFixed32Element(base_ptr, low); Status s = decoder->NextKeyNoValue(low_key_offset, &low_key, nullptr, &tmp); if (!s.ok()) { return s; } if (GetPrefix(low_key) == prefix) { prefix_matched = true; *offset = low_key_offset; } else if (low + 1 < upper_bound) { // There is possible a next prefix, return it prefix_matched = false; *offset = GetFixed32Element(base_ptr, low + 1); } else { // target is larger than a key of the last prefix in this bucket // but with a different prefix. Key does not exist. *offset = file_info_.data_end_offset; } return Status::OK(); } bool PlainTableReader::MatchBloom(uint32_t hash) const { if (!enable_bloom_) { return true; } if (bloom_.MayContainHash(hash)) { PERF_COUNTER_ADD(bloom_sst_hit_count, 1); return true; } else { PERF_COUNTER_ADD(bloom_sst_miss_count, 1); return false; } } Status PlainTableReader::Next(PlainTableKeyDecoder* decoder, uint32_t* offset, ParsedInternalKey* parsed_key, Slice* internal_key, Slice* value, bool* seekable) const { if (*offset == file_info_.data_end_offset) { *offset = file_info_.data_end_offset; return Status::OK(); } if (*offset > file_info_.data_end_offset) { return Status::Corruption("Offset is out of file size"); } uint32_t bytes_read; Status s = decoder->NextKey(*offset, parsed_key, internal_key, value, &bytes_read, seekable); if (!s.ok()) { return s; } *offset = *offset + bytes_read; return Status::OK(); } void PlainTableReader::Prepare(const Slice& target) { if (enable_bloom_) { uint32_t prefix_hash = GetSliceHash(GetPrefix(target)); bloom_.Prefetch(prefix_hash); } } Status PlainTableReader::Get(const ReadOptions& ro, const Slice& target, GetContext* get_context, bool skip_filters) { // Check bloom filter first. Slice prefix_slice; uint32_t prefix_hash; if (IsTotalOrderMode()) { if (full_scan_mode_) { status_ = Status::InvalidArgument("Get() is not allowed in full scan mode."); } // Match whole user key for bloom filter check. if (!MatchBloom(GetSliceHash(GetUserKey(target)))) { return Status::OK(); } // in total order mode, there is only one bucket 0, and we always use empty // prefix. prefix_slice = Slice(); prefix_hash = 0; } else { prefix_slice = GetPrefix(target); prefix_hash = GetSliceHash(prefix_slice); if (!MatchBloom(prefix_hash)) { return Status::OK(); } } uint32_t offset; bool prefix_match; PlainTableKeyDecoder decoder(&file_info_, encoding_type_, user_key_len_, ioptions_.prefix_extractor); Status s = GetOffset(&decoder, target, prefix_slice, prefix_hash, prefix_match, &offset); if (!s.ok()) { return s; } ParsedInternalKey found_key; ParsedInternalKey parsed_target; if (!ParseInternalKey(target, &parsed_target)) { return Status::Corruption(Slice()); } Slice found_value; while (offset < file_info_.data_end_offset) { s = Next(&decoder, &offset, &found_key, nullptr, &found_value); if (!s.ok()) { return s; } if (!prefix_match) { // Need to verify prefix for the first key found if it is not yet // checked. if (GetPrefix(found_key) != prefix_slice) { return Status::OK(); } prefix_match = true; } // TODO(ljin): since we know the key comparison result here, // can we enable the fast path? if (internal_comparator_.Compare(found_key, parsed_target) >= 0) { if (!get_context->SaveValue(found_key, found_value)) { break; } } } return Status::OK(); } uint64_t PlainTableReader::ApproximateOffsetOf(const Slice& key) { return 0; } PlainTableIterator::PlainTableIterator(PlainTableReader* table, bool use_prefix_seek) : table_(table), decoder_(&table_->file_info_, table_->encoding_type_, table_->user_key_len_, table_->prefix_extractor_), use_prefix_seek_(use_prefix_seek) { next_offset_ = offset_ = table_->file_info_.data_end_offset; } PlainTableIterator::~PlainTableIterator() { } bool PlainTableIterator::Valid() const { return offset_ < table_->file_info_.data_end_offset && offset_ >= table_->data_start_offset_; } void PlainTableIterator::SeekToFirst() { next_offset_ = table_->data_start_offset_; if (next_offset_ >= table_->file_info_.data_end_offset) { next_offset_ = offset_ = table_->file_info_.data_end_offset; } else { Next(); } } void PlainTableIterator::SeekToLast() { assert(false); status_ = Status::NotSupported("SeekToLast() is not supported in PlainTable"); } void PlainTableIterator::Seek(const Slice& target) { // If the user doesn't set prefix seek option and we are not able to do a // total Seek(). assert failure. if (!use_prefix_seek_) { if (table_->full_scan_mode_) { status_ = Status::InvalidArgument("Seek() is not allowed in full scan mode."); offset_ = next_offset_ = table_->file_info_.data_end_offset; return; } else if (table_->GetIndexSize() > 1) { assert(false); status_ = Status::NotSupported( "PlainTable cannot issue non-prefix seek unless in total order " "mode."); offset_ = next_offset_ = table_->file_info_.data_end_offset; return; } } Slice prefix_slice = table_->GetPrefix(target); uint32_t prefix_hash = 0; // Bloom filter is ignored in total-order mode. if (!table_->IsTotalOrderMode()) { prefix_hash = GetSliceHash(prefix_slice); if (!table_->MatchBloom(prefix_hash)) { offset_ = next_offset_ = table_->file_info_.data_end_offset; return; } } bool prefix_match; status_ = table_->GetOffset(&decoder_, target, prefix_slice, prefix_hash, prefix_match, &next_offset_); if (!status_.ok()) { offset_ = next_offset_ = table_->file_info_.data_end_offset; return; } if (next_offset_ < table_->file_info_.data_end_offset) { for (Next(); status_.ok() && Valid(); Next()) { if (!prefix_match) { // Need to verify the first key's prefix if (table_->GetPrefix(key()) != prefix_slice) { offset_ = next_offset_ = table_->file_info_.data_end_offset; break; } prefix_match = true; } if (table_->internal_comparator_.Compare(key(), target) >= 0) { break; } } } else { offset_ = table_->file_info_.data_end_offset; } } void PlainTableIterator::Next() { offset_ = next_offset_; if (offset_ < table_->file_info_.data_end_offset) { Slice tmp_slice; ParsedInternalKey parsed_key; status_ = table_->Next(&decoder_, &next_offset_, &parsed_key, &key_, &value_); if (!status_.ok()) { offset_ = next_offset_ = table_->file_info_.data_end_offset; } } } void PlainTableIterator::Prev() { assert(false); } Slice PlainTableIterator::key() const { assert(Valid()); return key_; } Slice PlainTableIterator::value() const { assert(Valid()); return value_; } Status PlainTableIterator::status() const { return status_; } } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/plain_table_reader.h000066400000000000000000000200151267530634000202100ustar00rootroot00000000000000// Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #ifndef ROCKSDB_LITE #include #include #include #include #include #include "db/dbformat.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "rocksdb/table_properties.h" #include "table/table_reader.h" #include "table/plain_table_factory.h" #include "table/plain_table_index.h" #include "util/arena.h" #include "util/dynamic_bloom.h" #include "util/file_reader_writer.h" namespace rocksdb { class Block; struct BlockContents; class BlockHandle; class Footer; struct Options; class RandomAccessFile; struct ReadOptions; class TableCache; class TableReader; class InternalKeyComparator; class PlainTableKeyDecoder; class GetContext; class InternalIterator; using std::unique_ptr; using std::unordered_map; using std::vector; extern const uint32_t kPlainTableVariableLength; struct PlainTableReaderFileInfo { bool is_mmap_mode; Slice file_data; uint32_t data_end_offset; unique_ptr file; PlainTableReaderFileInfo(unique_ptr&& _file, const EnvOptions& storage_options, uint32_t _data_size_offset) : is_mmap_mode(storage_options.use_mmap_reads), data_end_offset(_data_size_offset), file(std::move(_file)) {} }; // Based on following output file format shown in plain_table_factory.h // When opening the output file, IndexedTableReader creates a hash table // from key prefixes to offset of the output file. IndexedTable will decide // whether it points to the data offset of the first key with the key prefix // or the offset of it. If there are too many keys share this prefix, it will // create a binary search-able index from the suffix to offset on disk. // // The implementation of IndexedTableReader requires output file is mmaped class PlainTableReader: public TableReader { public: static Status Open(const ImmutableCFOptions& ioptions, const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, unique_ptr&& file, uint64_t file_size, unique_ptr* table, const int bloom_bits_per_key, double hash_table_ratio, size_t index_sparseness, size_t huge_page_tlb_size, bool full_scan_mode); InternalIterator* NewIterator(const ReadOptions&, Arena* arena = nullptr, bool skip_filters = false) override; void Prepare(const Slice& target) override; Status Get(const ReadOptions&, const Slice& key, GetContext* get_context, bool skip_filters = false) override; uint64_t ApproximateOffsetOf(const Slice& key) override; uint32_t GetIndexSize() const { return index_.GetIndexSize(); } void SetupForCompaction() override; std::shared_ptr GetTableProperties() const override { return table_properties_; } virtual size_t ApproximateMemoryUsage() const override { return arena_.MemoryAllocatedBytes(); } PlainTableReader(const ImmutableCFOptions& ioptions, unique_ptr&& file, const EnvOptions& env_options, const InternalKeyComparator& internal_comparator, EncodingType encoding_type, uint64_t file_size, const TableProperties* table_properties); virtual ~PlainTableReader(); protected: // Check bloom filter to see whether it might contain this prefix. // The hash of the prefix is given, since it can be reused for index lookup // too. virtual bool MatchBloom(uint32_t hash) const; // PopulateIndex() builds index of keys. It must be called before any query // to the table. // // props: the table properties object that need to be stored. Ownership of // the object will be passed. // Status PopulateIndex(TableProperties* props, int bloom_bits_per_key, double hash_table_ratio, size_t index_sparseness, size_t huge_page_tlb_size); Status MmapDataIfNeeded(); private: const InternalKeyComparator internal_comparator_; EncodingType encoding_type_; // represents plain table's current status. Status status_; PlainTableIndex index_; bool full_scan_mode_; // data_start_offset_ and data_end_offset_ defines the range of the // sst file that stores data. const uint32_t data_start_offset_ = 0; const uint32_t user_key_len_; const SliceTransform* prefix_extractor_; static const size_t kNumInternalBytes = 8; // Bloom filter is used to rule out non-existent key bool enable_bloom_; DynamicBloom bloom_; PlainTableReaderFileInfo file_info_; Arena arena_; std::unique_ptr index_block_alloc_; std::unique_ptr bloom_block_alloc_; const ImmutableCFOptions& ioptions_; uint64_t file_size_; std::shared_ptr table_properties_; bool IsFixedLength() const { return user_key_len_ != kPlainTableVariableLength; } size_t GetFixedInternalKeyLength() const { return user_key_len_ + kNumInternalBytes; } Slice GetPrefix(const Slice& target) const { assert(target.size() >= 8); // target is internal key return GetPrefixFromUserKey(GetUserKey(target)); } Slice GetPrefix(const ParsedInternalKey& target) const { return GetPrefixFromUserKey(target.user_key); } Slice GetUserKey(const Slice& key) const { return Slice(key.data(), key.size() - 8); } Slice GetPrefixFromUserKey(const Slice& user_key) const { if (!IsTotalOrderMode()) { return prefix_extractor_->Transform(user_key); } else { // Use empty slice as prefix if prefix_extractor is not set. // In that case, // it falls back to pure binary search and // total iterator seek is supported. return Slice(); } } friend class TableCache; friend class PlainTableIterator; // Internal helper function to generate an IndexRecordList object from all // the rows, which contains index records as a list. // If bloom_ is not null, all the keys' full-key hash will be added to the // bloom filter. Status PopulateIndexRecordList(PlainTableIndexBuilder* index_builder, vector* prefix_hashes); // Internal helper function to allocate memory for bloom filter and fill it void AllocateAndFillBloom(int bloom_bits_per_key, int num_prefixes, size_t huge_page_tlb_size, vector* prefix_hashes); void FillBloom(vector* prefix_hashes); // Read the key and value at `offset` to parameters for keys, the and // `seekable`. // On success, `offset` will be updated as the offset for the next key. // `parsed_key` will be key in parsed format. // if `internal_key` is not empty, it will be filled with key with slice // format. // if `seekable` is not null, it will return whether we can directly read // data using this offset. Status Next(PlainTableKeyDecoder* decoder, uint32_t* offset, ParsedInternalKey* parsed_key, Slice* internal_key, Slice* value, bool* seekable = nullptr) const; // Get file offset for key target. // return value prefix_matched is set to true if the offset is confirmed // for a key with the same prefix as target. Status GetOffset(PlainTableKeyDecoder* decoder, const Slice& target, const Slice& prefix, uint32_t prefix_hash, bool& prefix_matched, uint32_t* offset) const; bool IsTotalOrderMode() const { return (prefix_extractor_ == nullptr); } // No copying allowed explicit PlainTableReader(const TableReader&) = delete; void operator=(const TableReader&) = delete; }; } // namespace rocksdb #endif // ROCKSDB_LITE rocksdb-4.5.1/table/scoped_arena_iterator.h000066400000000000000000000017171267530634000207600ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include "table/internal_iterator.h" namespace rocksdb { class ScopedArenaIterator { public: explicit ScopedArenaIterator(InternalIterator* iter = nullptr) : iter_(iter) {} InternalIterator* operator->() { return iter_; } void set(InternalIterator* iter) { iter_ = iter; } InternalIterator* get() { return iter_; } ~ScopedArenaIterator() { iter_->~InternalIterator(); } private: InternalIterator* iter_; }; } // namespace rocksdb rocksdb-4.5.1/table/sst_file_writer.cc000066400000000000000000000137001267530634000177610ustar00rootroot00000000000000// Copyright (c) 2015, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "rocksdb/sst_file_writer.h" #include #include "db/dbformat.h" #include "rocksdb/table.h" #include "table/block_based_table_builder.h" #include "util/file_reader_writer.h" #include "util/string_util.h" namespace rocksdb { const std::string ExternalSstFilePropertyNames::kVersion = "rocksdb.external_sst_file.version"; // PropertiesCollector used to add properties specific to tables // generated by SstFileWriter class SstFileWriter::SstFileWriterPropertiesCollector : public IntTblPropCollector { public: explicit SstFileWriterPropertiesCollector(int32_t version) : version_(version) {} virtual Status InternalAdd(const Slice& key, const Slice& value, uint64_t file_size) override { // Intentionally left blank. Have no interest in collecting stats for // individual key/value pairs. return Status::OK(); } virtual Status Finish(UserCollectedProperties* properties) override { std::string version_val; PutFixed32(&version_val, static_cast(version_)); properties->insert({ExternalSstFilePropertyNames::kVersion, version_val}); return Status::OK(); } virtual const char* Name() const override { return "SstFileWriterPropertiesCollector"; } virtual UserCollectedProperties GetReadableProperties() const override { return {{ExternalSstFilePropertyNames::kVersion, ToString(version_)}}; } private: int32_t version_; }; class SstFileWriter::SstFileWriterPropertiesCollectorFactory : public IntTblPropCollectorFactory { public: explicit SstFileWriterPropertiesCollectorFactory(int32_t version) : version_(version) {} virtual IntTblPropCollector* CreateIntTblPropCollector( uint32_t column_family_id) override { return new SstFileWriterPropertiesCollector(version_); } virtual const char* Name() const override { return "SstFileWriterPropertiesCollector"; } private: int32_t version_; }; struct SstFileWriter::Rep { Rep(const EnvOptions& _env_options, const ImmutableCFOptions& _ioptions, const Comparator* _user_comparator) : env_options(_env_options), ioptions(_ioptions), internal_comparator(_user_comparator) {} std::unique_ptr file_writer; std::unique_ptr builder; EnvOptions env_options; ImmutableCFOptions ioptions; InternalKeyComparator internal_comparator; ExternalSstFileInfo file_info; }; SstFileWriter::SstFileWriter(const EnvOptions& env_options, const ImmutableCFOptions& ioptions, const Comparator* user_comparator) : rep_(new Rep(env_options, ioptions, user_comparator)) {} SstFileWriter::~SstFileWriter() { delete rep_; } Status SstFileWriter::Open(const std::string& file_path) { Rep* r = rep_; Status s; std::unique_ptr sst_file; s = r->ioptions.env->NewWritableFile(file_path, &sst_file, r->env_options); if (!s.ok()) { return s; } CompressionType compression_type = r->ioptions.compression; if (!r->ioptions.compression_per_level.empty()) { // Use the compression of the last level if we have per level compression compression_type = *(r->ioptions.compression_per_level.rbegin()); } std::vector> int_tbl_prop_collector_factories; int_tbl_prop_collector_factories.emplace_back( new SstFileWriterPropertiesCollectorFactory(1 /* version */)); TableBuilderOptions table_builder_options( r->ioptions, r->internal_comparator, &int_tbl_prop_collector_factories, compression_type, r->ioptions.compression_opts, false); r->file_writer.reset( new WritableFileWriter(std::move(sst_file), r->env_options)); r->builder.reset(r->ioptions.table_factory->NewTableBuilder( table_builder_options, TablePropertiesCollectorFactory::Context::kUnknownColumnFamily, r->file_writer.get())); r->file_info.file_path = file_path; r->file_info.file_size = 0; r->file_info.num_entries = 0; r->file_info.sequence_number = 0; r->file_info.version = 1; return s; } Status SstFileWriter::Add(const Slice& user_key, const Slice& value) { Rep* r = rep_; if (!r->builder) { return Status::InvalidArgument("File is not opened"); } if (r->file_info.num_entries == 0) { r->file_info.smallest_key = user_key.ToString(); } else { if (r->internal_comparator.user_comparator()->Compare( user_key, r->file_info.largest_key) <= 0) { // Make sure that keys are added in order return Status::InvalidArgument("Keys must be added in order"); } } // update file info r->file_info.num_entries++; r->file_info.largest_key = user_key.ToString(); r->file_info.file_size = r->builder->FileSize(); InternalKey ikey(user_key, 0 /* Sequence Number */, ValueType::kTypeValue /* Put */); r->builder->Add(ikey.Encode(), value); return Status::OK(); } Status SstFileWriter::Finish(ExternalSstFileInfo* file_info) { Rep* r = rep_; if (!r->builder) { return Status::InvalidArgument("File is not opened"); } if (r->file_info.num_entries == 0) { return Status::InvalidArgument("Cannot create sst file with no entries"); } Status s = r->builder->Finish(); if (s.ok()) { if (!r->ioptions.disable_data_sync) { s = r->file_writer->Sync(r->ioptions.use_fsync); } if (s.ok()) { s = r->file_writer->Close(); } } else { r->builder->Abandon(); } if (!s.ok()) { r->ioptions.env->DeleteFile(r->file_info.file_path); } if (s.ok() && file_info != nullptr) { r->file_info.file_size = r->builder->FileSize(); *file_info = r->file_info; } r->builder.reset(); return s; } } // namespace rocksdb rocksdb-4.5.1/table/table_builder.h000066400000000000000000000102671267530634000172210ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include #include #include #include #include "db/table_properties_collector.h" #include "rocksdb/options.h" #include "rocksdb/table_properties.h" #include "util/file_reader_writer.h" #include "util/mutable_cf_options.h" namespace rocksdb { class Slice; class Status; struct TableReaderOptions { // @param skip_filters Disables loading/accessing the filter block TableReaderOptions(const ImmutableCFOptions& _ioptions, const EnvOptions& _env_options, const InternalKeyComparator& _internal_comparator, bool _skip_filters = false) : ioptions(_ioptions), env_options(_env_options), internal_comparator(_internal_comparator), skip_filters(_skip_filters) {} const ImmutableCFOptions& ioptions; const EnvOptions& env_options; const InternalKeyComparator& internal_comparator; // This is only used for BlockBasedTable (reader) bool skip_filters; }; struct TableBuilderOptions { TableBuilderOptions( const ImmutableCFOptions& _ioptions, const InternalKeyComparator& _internal_comparator, const std::vector>* _int_tbl_prop_collector_factories, CompressionType _compression_type, const CompressionOptions& _compression_opts, bool _skip_filters) : ioptions(_ioptions), internal_comparator(_internal_comparator), int_tbl_prop_collector_factories(_int_tbl_prop_collector_factories), compression_type(_compression_type), compression_opts(_compression_opts), skip_filters(_skip_filters) {} const ImmutableCFOptions& ioptions; const InternalKeyComparator& internal_comparator; const std::vector>* int_tbl_prop_collector_factories; CompressionType compression_type; const CompressionOptions& compression_opts; // This is only used for BlockBasedTableBuilder bool skip_filters = false; }; // TableBuilder provides the interface used to build a Table // (an immutable and sorted map from keys to values). // // Multiple threads can invoke const methods on a TableBuilder without // external synchronization, but if any of the threads may call a // non-const method, all threads accessing the same TableBuilder must use // external synchronization. class TableBuilder { public: // REQUIRES: Either Finish() or Abandon() has been called. virtual ~TableBuilder() {} // Add key,value to the table being constructed. // REQUIRES: key is after any previously added key according to comparator. // REQUIRES: Finish(), Abandon() have not been called virtual void Add(const Slice& key, const Slice& value) = 0; // Return non-ok iff some error has been detected. virtual Status status() const = 0; // Finish building the table. // REQUIRES: Finish(), Abandon() have not been called virtual Status Finish() = 0; // Indicate that the contents of this builder should be abandoned. // If the caller is not going to call Finish(), it must call Abandon() // before destroying this builder. // REQUIRES: Finish(), Abandon() have not been called virtual void Abandon() = 0; // Number of calls to Add() so far. virtual uint64_t NumEntries() const = 0; // Size of the file generated so far. If invoked after a successful // Finish() call, returns the size of the final generated file. virtual uint64_t FileSize() const = 0; // If the user defined table properties collector suggest the file to // be further compacted. virtual bool NeedCompact() const { return false; } // Returns table properties virtual TableProperties GetTableProperties() const = 0; }; } // namespace rocksdb rocksdb-4.5.1/table/table_properties.cc000066400000000000000000000107631267530634000201260ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #include "table/table_properties_internal.h" #include "rocksdb/table_properties.h" #include "rocksdb/iterator.h" #include "rocksdb/env.h" #include "port/port.h" #include "table/internal_iterator.h" #include "util/string_util.h" namespace rocksdb { const uint32_t TablePropertiesCollectorFactory::Context::kUnknownColumnFamily = port::kMaxInt32; namespace { void AppendProperty( std::string& props, const std::string& key, const std::string& value, const std::string& prop_delim, const std::string& kv_delim) { props.append(key); props.append(kv_delim); props.append(value); props.append(prop_delim); } template void AppendProperty( std::string& props, const std::string& key, const TValue& value, const std::string& prop_delim, const std::string& kv_delim) { AppendProperty( props, key, ToString(value), prop_delim, kv_delim ); } } std::string TableProperties::ToString( const std::string& prop_delim, const std::string& kv_delim) const { std::string result; result.reserve(1024); // Basic Info AppendProperty(result, "# data blocks", num_data_blocks, prop_delim, kv_delim); AppendProperty(result, "# entries", num_entries, prop_delim, kv_delim); AppendProperty(result, "raw key size", raw_key_size, prop_delim, kv_delim); AppendProperty(result, "raw average key size", num_entries != 0 ? 1.0 * raw_key_size / num_entries : 0.0, prop_delim, kv_delim); AppendProperty(result, "raw value size", raw_value_size, prop_delim, kv_delim); AppendProperty(result, "raw average value size", num_entries != 0 ? 1.0 * raw_value_size / num_entries : 0.0, prop_delim, kv_delim); AppendProperty(result, "data block size", data_size, prop_delim, kv_delim); AppendProperty(result, "index block size", index_size, prop_delim, kv_delim); AppendProperty(result, "filter block size", filter_size, prop_delim, kv_delim); AppendProperty(result, "(estimated) table size", data_size + index_size + filter_size, prop_delim, kv_delim); AppendProperty( result, "filter policy name", filter_policy_name.empty() ? std::string("N/A") : filter_policy_name, prop_delim, kv_delim); return result; } void TableProperties::Add(const TableProperties& tp) { data_size += tp.data_size; index_size += tp.index_size; filter_size += tp.filter_size; raw_key_size += tp.raw_key_size; raw_value_size += tp.raw_value_size; num_data_blocks += tp.num_data_blocks; num_entries += tp.num_entries; } const std::string TablePropertiesNames::kDataSize = "rocksdb.data.size"; const std::string TablePropertiesNames::kIndexSize = "rocksdb.index.size"; const std::string TablePropertiesNames::kFilterSize = "rocksdb.filter.size"; const std::string TablePropertiesNames::kRawKeySize = "rocksdb.raw.key.size"; const std::string TablePropertiesNames::kRawValueSize = "rocksdb.raw.value.size"; const std::string TablePropertiesNames::kNumDataBlocks = "rocksdb.num.data.blocks"; const std::string TablePropertiesNames::kNumEntries = "rocksdb.num.entries"; const std::string TablePropertiesNames::kFilterPolicy = "rocksdb.filter.policy"; const std::string TablePropertiesNames::kFormatVersion = "rocksdb.format.version"; const std::string TablePropertiesNames::kFixedKeyLen = "rocksdb.fixed.key.length"; extern const std::string kPropertiesBlock = "rocksdb.properties"; // Old property block name for backward compatibility extern const std::string kPropertiesBlockOldName = "rocksdb.stats"; // Seek to the properties block. // Return true if it successfully seeks to the properties block. Status SeekToPropertiesBlock(InternalIterator* meta_iter, bool* is_found) { *is_found = true; meta_iter->Seek(kPropertiesBlock); if (meta_iter->status().ok() && (!meta_iter->Valid() || meta_iter->key() != kPropertiesBlock)) { meta_iter->Seek(kPropertiesBlockOldName); if (meta_iter->status().ok() && (!meta_iter->Valid() || meta_iter->key() != kPropertiesBlockOldName)) { *is_found = false; } } return meta_iter->status(); } } // namespace rocksdb rocksdb-4.5.1/table/table_properties_internal.h000066400000000000000000000011711267530634000216550ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #pragma once #include "rocksdb/status.h" #include "rocksdb/iterator.h" namespace rocksdb { class InternalIterator; // Seek to the properties block. // If it successfully seeks to the properties block, "is_found" will be // set to true. Status SeekToPropertiesBlock(InternalIterator* meta_iter, bool* is_found); } // namespace rocksdb rocksdb-4.5.1/table/table_reader.h000066400000000000000000000100071267530634000170250ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include namespace rocksdb { class Iterator; struct ParsedInternalKey; class Slice; class Arena; struct ReadOptions; struct TableProperties; class GetContext; class InternalIterator; // A Table is a sorted map from strings to strings. Tables are // immutable and persistent. A Table may be safely accessed from // multiple threads without external synchronization. class TableReader { public: virtual ~TableReader() {} // Returns a new iterator over the table contents. // The result of NewIterator() is initially invalid (caller must // call one of the Seek methods on the iterator before using it). // arena: If not null, the arena needs to be used to allocate the Iterator. // When destroying the iterator, the caller will not call "delete" // but Iterator::~Iterator() directly. The destructor needs to destroy // all the states but those allocated in arena. // skip_filters: disables checking the bloom filters even if they exist. This // option is effective only for block-based table format. virtual InternalIterator* NewIterator(const ReadOptions&, Arena* arena = nullptr, bool skip_filters = false) = 0; // Given a key, return an approximate byte offset in the file where // the data for that key begins (or would begin if the key were // present in the file). The returned value is in terms of file // bytes, and so includes effects like compression of the underlying data. // E.g., the approximate offset of the last key in the table will // be close to the file length. virtual uint64_t ApproximateOffsetOf(const Slice& key) = 0; // Set up the table for Compaction. Might change some parameters with // posix_fadvise virtual void SetupForCompaction() = 0; virtual std::shared_ptr GetTableProperties() const = 0; // Prepare work that can be done before the real Get() virtual void Prepare(const Slice& target) {} // Report an approximation of how much memory has been used. virtual size_t ApproximateMemoryUsage() const = 0; // Calls get_context->SaveValue() repeatedly, starting with // the entry found after a call to Seek(key), until it returns false. // May not make such a call if filter policy says that key is not present. // // get_context->MarkKeyMayExist needs to be called when it is configured to be // memory only and the key is not found in the block cache. // // readOptions is the options for the read // key is the key to search for // skip_filters: disables checking the bloom filters even if they exist. This // option is effective only for block-based table format. virtual Status Get(const ReadOptions& readOptions, const Slice& key, GetContext* get_context, bool skip_filters = false) = 0; // Prefetch data corresponding to a give range of keys // Typically this functionality is required for table implementations that // persists the data on a non volatile storage medium like disk/SSD virtual Status Prefetch(const Slice* begin = nullptr, const Slice* end = nullptr) { (void) begin; (void) end; // Default implementation is NOOP. // The child class should implement functionality when applicable return Status::OK(); } // convert db file to a human readable form virtual Status DumpTable(WritableFile* out_file) { return Status::NotSupported("DumpTable() not supported"); } }; } // namespace rocksdb rocksdb-4.5.1/table/table_reader_bench.cc000066400000000000000000000277021267530634000203340ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. #ifndef GFLAGS #include int main() { fprintf(stderr, "Please install gflags to run rocksdb tools\n"); return 1; } #else #include #include "rocksdb/db.h" #include "rocksdb/slice_transform.h" #include "rocksdb/table.h" #include "db/db_impl.h" #include "db/dbformat.h" #include "table/block_based_table_factory.h" #include "table/internal_iterator.h" #include "table/plain_table_factory.h" #include "table/table_builder.h" #include "table/get_context.h" #include "util/file_reader_writer.h" #include "util/histogram.h" #include "util/testharness.h" #include "util/testutil.h" using GFLAGS::ParseCommandLineFlags; using GFLAGS::SetUsageMessage; namespace rocksdb { namespace { // Make a key that i determines the first 4 characters and j determines the // last 4 characters. static std::string MakeKey(int i, int j, bool through_db) { char buf[100]; snprintf(buf, sizeof(buf), "%04d__key___%04d", i, j); if (through_db) { return std::string(buf); } // If we directly query table, which operates on internal keys // instead of user keys, we need to add 8 bytes of internal // information (row type etc) to user key to make an internal // key. InternalKey key(std::string(buf), 0, ValueType::kTypeValue); return key.Encode().ToString(); } uint64_t Now(Env* env, bool measured_by_nanosecond) { return measured_by_nanosecond ? env->NowNanos() : env->NowMicros(); } } // namespace // A very simple benchmark that. // Create a table with roughly numKey1 * numKey2 keys, // where there are numKey1 prefixes of the key, each has numKey2 number of // distinguished key, differing in the suffix part. // If if_query_empty_keys = false, query the existing keys numKey1 * numKey2 // times randomly. // If if_query_empty_keys = true, query numKey1 * numKey2 random empty keys. // Print out the total time. // If through_db=true, a full DB will be created and queries will be against // it. Otherwise, operations will be directly through table level. // // If for_terator=true, instead of just query one key each time, it queries // a range sharing the same prefix. namespace { void TableReaderBenchmark(Options& opts, EnvOptions& env_options, ReadOptions& read_options, int num_keys1, int num_keys2, int num_iter, int prefix_len, bool if_query_empty_keys, bool for_iterator, bool through_db, bool measured_by_nanosecond) { rocksdb::InternalKeyComparator ikc(opts.comparator); std::string file_name = test::TmpDir() + "/rocksdb_table_reader_benchmark"; std::string dbname = test::TmpDir() + "/rocksdb_table_reader_bench_db"; WriteOptions wo; Env* env = Env::Default(); TableBuilder* tb = nullptr; DB* db = nullptr; Status s; const ImmutableCFOptions ioptions(opts); unique_ptr file_writer; if (!through_db) { unique_ptr file; env->NewWritableFile(file_name, &file, env_options); std::vector > int_tbl_prop_collector_factories; file_writer.reset(new WritableFileWriter(std::move(file), env_options)); tb = opts.table_factory->NewTableBuilder( TableBuilderOptions(ioptions, ikc, &int_tbl_prop_collector_factories, CompressionType::kNoCompression, CompressionOptions(), false), 0, file_writer.get()); } else { s = DB::Open(opts, dbname, &db); ASSERT_OK(s); ASSERT_TRUE(db != nullptr); } // Populate slightly more than 1M keys for (int i = 0; i < num_keys1; i++) { for (int j = 0; j < num_keys2; j++) { std::string key = MakeKey(i * 2, j, through_db); if (!through_db) { tb->Add(key, key); } else { db->Put(wo, key, key); } } } if (!through_db) { tb->Finish(); file_writer->Close(); } else { db->Flush(FlushOptions()); } unique_ptr table_reader; if (!through_db) { unique_ptr raf; s = env->NewRandomAccessFile(file_name, &raf, env_options); if (!s.ok()) { fprintf(stderr, "Create File Error: %s\n", s.ToString().c_str()); exit(1); } uint64_t file_size; env->GetFileSize(file_name, &file_size); unique_ptr file_reader( new RandomAccessFileReader(std::move(raf))); s = opts.table_factory->NewTableReader( TableReaderOptions(ioptions, env_options, ikc), std::move(file_reader), file_size, &table_reader); if (!s.ok()) { fprintf(stderr, "Open Table Error: %s\n", s.ToString().c_str()); exit(1); } } Random rnd(301); std::string result; HistogramImpl hist; for (int it = 0; it < num_iter; it++) { for (int i = 0; i < num_keys1; i++) { for (int j = 0; j < num_keys2; j++) { int r1 = rnd.Uniform(num_keys1) * 2; int r2 = rnd.Uniform(num_keys2); if (if_query_empty_keys) { r1++; r2 = num_keys2 * 2 - r2; } if (!for_iterator) { // Query one existing key; std::string key = MakeKey(r1, r2, through_db); uint64_t start_time = Now(env, measured_by_nanosecond); if (!through_db) { std::string value; MergeContext merge_context; GetContext get_context(ioptions.comparator, ioptions.merge_operator, ioptions.info_log, ioptions.statistics, GetContext::kNotFound, Slice(key), &value, nullptr, &merge_context, env); s = table_reader->Get(read_options, key, &get_context); } else { s = db->Get(read_options, key, &result); } hist.Add(Now(env, measured_by_nanosecond) - start_time); } else { int r2_len; if (if_query_empty_keys) { r2_len = 0; } else { r2_len = rnd.Uniform(num_keys2) + 1; if (r2_len + r2 > num_keys2) { r2_len = num_keys2 - r2; } } std::string start_key = MakeKey(r1, r2, through_db); std::string end_key = MakeKey(r1, r2 + r2_len, through_db); uint64_t total_time = 0; uint64_t start_time = Now(env, measured_by_nanosecond); Iterator* iter = nullptr; InternalIterator* iiter = nullptr; if (!through_db) { iiter = table_reader->NewIterator(read_options); } else { iter = db->NewIterator(read_options); } int count = 0; for (through_db ? iter->Seek(start_key) : iiter->Seek(start_key); through_db ? iter->Valid() : iiter->Valid(); through_db ? iter->Next() : iiter->Next()) { if (if_query_empty_keys) { break; } // verify key; total_time += Now(env, measured_by_nanosecond) - start_time; assert(Slice(MakeKey(r1, r2 + count, through_db)) == (through_db ? iter->key() : iiter->key())); start_time = Now(env, measured_by_nanosecond); if (++count >= r2_len) { break; } } if (count != r2_len) { fprintf( stderr, "Iterator cannot iterate expected number of entries. " "Expected %d but got %d\n", r2_len, count); assert(false); } delete iter; total_time += Now(env, measured_by_nanosecond) - start_time; hist.Add(total_time); } } } } fprintf( stderr, "===================================================" "====================================================\n" "InMemoryTableSimpleBenchmark: %20s num_key1: %5d " "num_key2: %5d %10s\n" "===================================================" "====================================================" "\nHistogram (unit: %s): \n%s", opts.table_factory->Name(), num_keys1, num_keys2, for_iterator ? "iterator" : (if_query_empty_keys ? "empty" : "non_empty"), measured_by_nanosecond ? "nanosecond" : "microsecond", hist.ToString().c_str()); if (!through_db) { env->DeleteFile(file_name); } else { delete db; db = nullptr; DestroyDB(dbname, opts); } } } // namespace } // namespace rocksdb DEFINE_bool(query_empty, false, "query non-existing keys instead of existing " "ones."); DEFINE_int32(num_keys1, 4096, "number of distinguish prefix of keys"); DEFINE_int32(num_keys2, 512, "number of distinguish keys for each prefix"); DEFINE_int32(iter, 3, "query non-existing keys instead of existing ones"); DEFINE_int32(prefix_len, 16, "Prefix length used for iterators and indexes"); DEFINE_bool(iterator, false, "For test iterator"); DEFINE_bool(through_db, false, "If enable, a DB instance will be created and " "the query will be against DB. Otherwise, will be directly against " "a table reader."); DEFINE_bool(mmap_read, true, "Whether use mmap read"); DEFINE_string(table_factory, "block_based", "Table factory to use: `block_based` (default), `plain_table` or " "`cuckoo_hash`."); DEFINE_string(time_unit, "microsecond", "The time unit used for measuring performance. User can specify " "`microsecond` (default) or `nanosecond`"); int main(int argc, char** argv) { SetUsageMessage(std::string("\nUSAGE:\n") + std::string(argv[0]) + " [OPTIONS]..."); ParseCommandLineFlags(&argc, &argv, true); std::shared_ptr tf; rocksdb::Options options; if (FLAGS_prefix_len < 16) { options.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform( FLAGS_prefix_len)); } rocksdb::ReadOptions ro; rocksdb::EnvOptions env_options; options.create_if_missing = true; options.compression = rocksdb::CompressionType::kNoCompression; if (FLAGS_table_factory == "cuckoo_hash") { #ifndef ROCKSDB_LITE options.allow_mmap_reads = FLAGS_mmap_read; env_options.use_mmap_reads = FLAGS_mmap_read; rocksdb::CuckooTableOptions table_options; table_options.hash_table_ratio = 0.75; tf.reset(rocksdb::NewCuckooTableFactory(table_options)); #else fprintf(stderr, "Plain table is not supported in lite mode\n"); exit(1); #endif // ROCKSDB_LITE } else if (FLAGS_table_factory == "plain_table") { #ifndef ROCKSDB_LITE options.allow_mmap_reads = FLAGS_mmap_read; env_options.use_mmap_reads = FLAGS_mmap_read; rocksdb::PlainTableOptions plain_table_options; plain_table_options.user_key_len = 16; plain_table_options.bloom_bits_per_key = (FLAGS_prefix_len == 16) ? 0 : 8; plain_table_options.hash_table_ratio = 0.75; tf.reset(new rocksdb::PlainTableFactory(plain_table_options)); options.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform( FLAGS_prefix_len)); #else fprintf(stderr, "Cuckoo table is not supported in lite mode\n"); exit(1); #endif // ROCKSDB_LITE } else if (FLAGS_table_factory == "block_based") { tf.reset(new rocksdb::BlockBasedTableFactory()); } else { fprintf(stderr, "Invalid table type %s\n", FLAGS_table_factory.c_str()); } if (tf) { // if user provides invalid options, just fall back to microsecond. bool measured_by_nanosecond = FLAGS_time_unit == "nanosecond"; options.table_factory = tf; rocksdb::TableReaderBenchmark(options, env_options, ro, FLAGS_num_keys1, FLAGS_num_keys2, FLAGS_iter, FLAGS_prefix_len, FLAGS_query_empty, FLAGS_iterator, FLAGS_through_db, measured_by_nanosecond); } else { return 1; } return 0; } #endif // GFLAGS rocksdb-4.5.1/table/table_test.cc000066400000000000000000002443631267530634000167160ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include #include #include #include #include #include #include #include #include "db/dbformat.h" #include "db/memtable.h" #include "db/write_batch_internal.h" #include "db/writebuffer.h" #include "memtable/stl_wrappers.h" #include "rocksdb/cache.h" #include "rocksdb/db.h" #include "rocksdb/env.h" #include "rocksdb/iterator.h" #include "rocksdb/memtablerep.h" #include "rocksdb/perf_context.h" #include "rocksdb/slice_transform.h" #include "rocksdb/statistics.h" #include "table/block.h" #include "table/block_based_table_builder.h" #include "table/block_based_table_factory.h" #include "table/block_based_table_reader.h" #include "table/block_builder.h" #include "table/format.h" #include "table/get_context.h" #include "table/internal_iterator.h" #include "table/meta_blocks.h" #include "table/plain_table_factory.h" #include "table/scoped_arena_iterator.h" #include "util/compression.h" #include "util/random.h" #include "util/statistics.h" #include "util/string_util.h" #include "util/testharness.h" #include "util/testutil.h" namespace rocksdb { extern const uint64_t kLegacyBlockBasedTableMagicNumber; extern const uint64_t kLegacyPlainTableMagicNumber; extern const uint64_t kBlockBasedTableMagicNumber; extern const uint64_t kPlainTableMagicNumber; namespace { // Return reverse of "key". // Used to test non-lexicographic comparators. std::string Reverse(const Slice& key) { auto rev = key.ToString(); std::reverse(rev.begin(), rev.end()); return rev; } class ReverseKeyComparator : public Comparator { public: virtual const char* Name() const override { return "rocksdb.ReverseBytewiseComparator"; } virtual int Compare(const Slice& a, const Slice& b) const override { return BytewiseComparator()->Compare(Reverse(a), Reverse(b)); } virtual void FindShortestSeparator(std::string* start, const Slice& limit) const override { std::string s = Reverse(*start); std::string l = Reverse(limit); BytewiseComparator()->FindShortestSeparator(&s, l); *start = Reverse(s); } virtual void FindShortSuccessor(std::string* key) const override { std::string s = Reverse(*key); BytewiseComparator()->FindShortSuccessor(&s); *key = Reverse(s); } }; ReverseKeyComparator reverse_key_comparator; void Increment(const Comparator* cmp, std::string* key) { if (cmp == BytewiseComparator()) { key->push_back('\0'); } else { assert(cmp == &reverse_key_comparator); std::string rev = Reverse(*key); rev.push_back('\0'); *key = Reverse(rev); } } } // namespace // Helper class for tests to unify the interface between // BlockBuilder/TableBuilder and Block/Table. class Constructor { public: explicit Constructor(const Comparator* cmp) : data_(stl_wrappers::LessOfComparator(cmp)) {} virtual ~Constructor() { } void Add(const std::string& key, const Slice& value) { data_[key] = value.ToString(); } // Finish constructing the data structure with all the keys that have // been added so far. Returns the keys in sorted order in "*keys" // and stores the key/value pairs in "*kvmap" void Finish(const Options& options, const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, std::vector* keys, stl_wrappers::KVMap* kvmap) { last_internal_key_ = &internal_comparator; *kvmap = data_; keys->clear(); for (const auto& kv : data_) { keys->push_back(kv.first); } data_.clear(); Status s = FinishImpl(options, ioptions, table_options, internal_comparator, *kvmap); ASSERT_TRUE(s.ok()) << s.ToString(); } // Construct the data structure from the data in "data" virtual Status FinishImpl(const Options& options, const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const stl_wrappers::KVMap& data) = 0; virtual InternalIterator* NewIterator() const = 0; virtual const stl_wrappers::KVMap& data() { return data_; } virtual bool IsArenaMode() const { return false; } virtual DB* db() const { return nullptr; } // Overridden in DBConstructor virtual bool AnywayDeleteIterator() const { return false; } protected: const InternalKeyComparator* last_internal_key_; private: stl_wrappers::KVMap data_; }; class BlockConstructor: public Constructor { public: explicit BlockConstructor(const Comparator* cmp) : Constructor(cmp), comparator_(cmp), block_(nullptr) { } ~BlockConstructor() { delete block_; } virtual Status FinishImpl(const Options& options, const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const stl_wrappers::KVMap& kv_map) override { delete block_; block_ = nullptr; BlockBuilder builder(table_options.block_restart_interval); for (const auto kv : kv_map) { builder.Add(kv.first, kv.second); } // Open the block data_ = builder.Finish().ToString(); BlockContents contents; contents.data = data_; contents.cachable = false; block_ = new Block(std::move(contents)); return Status::OK(); } virtual InternalIterator* NewIterator() const override { return block_->NewIterator(comparator_); } private: const Comparator* comparator_; std::string data_; Block* block_; BlockConstructor(); }; // A helper class that converts internal format keys into user keys class KeyConvertingIterator : public InternalIterator { public: explicit KeyConvertingIterator(InternalIterator* iter, bool arena_mode = false) : iter_(iter), arena_mode_(arena_mode) {} virtual ~KeyConvertingIterator() { if (arena_mode_) { iter_->~InternalIterator(); } else { delete iter_; } } virtual bool Valid() const override { return iter_->Valid(); } virtual void Seek(const Slice& target) override { ParsedInternalKey ikey(target, kMaxSequenceNumber, kTypeValue); std::string encoded; AppendInternalKey(&encoded, ikey); iter_->Seek(encoded); } virtual void SeekToFirst() override { iter_->SeekToFirst(); } virtual void SeekToLast() override { iter_->SeekToLast(); } virtual void Next() override { iter_->Next(); } virtual void Prev() override { iter_->Prev(); } virtual Slice key() const override { assert(Valid()); ParsedInternalKey parsed_key; if (!ParseInternalKey(iter_->key(), &parsed_key)) { status_ = Status::Corruption("malformed internal key"); return Slice("corrupted key"); } return parsed_key.user_key; } virtual Slice value() const override { return iter_->value(); } virtual Status status() const override { return status_.ok() ? iter_->status() : status_; } private: mutable Status status_; InternalIterator* iter_; bool arena_mode_; // No copying allowed KeyConvertingIterator(const KeyConvertingIterator&); void operator=(const KeyConvertingIterator&); }; class TableConstructor: public Constructor { public: explicit TableConstructor(const Comparator* cmp, bool convert_to_internal_key = false) : Constructor(cmp), convert_to_internal_key_(convert_to_internal_key) {} ~TableConstructor() { Reset(); } virtual Status FinishImpl(const Options& options, const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const stl_wrappers::KVMap& kv_map) override { Reset(); soptions.use_mmap_reads = ioptions.allow_mmap_reads; file_writer_.reset(test::GetWritableFileWriter(new test::StringSink())); unique_ptr builder; std::vector> int_tbl_prop_collector_factories; builder.reset(ioptions.table_factory->NewTableBuilder( TableBuilderOptions(ioptions, internal_comparator, &int_tbl_prop_collector_factories, options.compression, CompressionOptions(), false), TablePropertiesCollectorFactory::Context::kUnknownColumnFamily, file_writer_.get())); for (const auto kv : kv_map) { if (convert_to_internal_key_) { ParsedInternalKey ikey(kv.first, kMaxSequenceNumber, kTypeValue); std::string encoded; AppendInternalKey(&encoded, ikey); builder->Add(encoded, kv.second); } else { builder->Add(kv.first, kv.second); } EXPECT_TRUE(builder->status().ok()); } Status s = builder->Finish(); file_writer_->Flush(); EXPECT_TRUE(s.ok()) << s.ToString(); EXPECT_EQ(GetSink()->contents().size(), builder->FileSize()); // Open the table uniq_id_ = cur_uniq_id_++; file_reader_.reset(test::GetRandomAccessFileReader(new test::StringSource( GetSink()->contents(), uniq_id_, ioptions.allow_mmap_reads))); return ioptions.table_factory->NewTableReader( TableReaderOptions(ioptions, soptions, internal_comparator), std::move(file_reader_), GetSink()->contents().size(), &table_reader_); } virtual InternalIterator* NewIterator() const override { ReadOptions ro; InternalIterator* iter = table_reader_->NewIterator(ro); if (convert_to_internal_key_) { return new KeyConvertingIterator(iter); } else { return iter; } } uint64_t ApproximateOffsetOf(const Slice& key) const { return table_reader_->ApproximateOffsetOf(key); } virtual Status Reopen(const ImmutableCFOptions& ioptions) { file_reader_.reset(test::GetRandomAccessFileReader(new test::StringSource( GetSink()->contents(), uniq_id_, ioptions.allow_mmap_reads))); return ioptions.table_factory->NewTableReader( TableReaderOptions(ioptions, soptions, *last_internal_key_), std::move(file_reader_), GetSink()->contents().size(), &table_reader_); } virtual TableReader* GetTableReader() { return table_reader_.get(); } virtual bool AnywayDeleteIterator() const override { return convert_to_internal_key_; } private: void Reset() { uniq_id_ = 0; table_reader_.reset(); file_writer_.reset(); file_reader_.reset(); } test::StringSink* GetSink() { return static_cast(file_writer_->writable_file()); } uint64_t uniq_id_; unique_ptr file_writer_; unique_ptr file_reader_; unique_ptr table_reader_; bool convert_to_internal_key_; TableConstructor(); static uint64_t cur_uniq_id_; EnvOptions soptions; }; uint64_t TableConstructor::cur_uniq_id_ = 1; class MemTableConstructor: public Constructor { public: explicit MemTableConstructor(const Comparator* cmp, WriteBuffer* wb) : Constructor(cmp), internal_comparator_(cmp), write_buffer_(wb), table_factory_(new SkipListFactory) { options_.memtable_factory = table_factory_; ImmutableCFOptions ioptions(options_); memtable_ = new MemTable(internal_comparator_, ioptions, MutableCFOptions(options_, ioptions), wb, kMaxSequenceNumber); memtable_->Ref(); } ~MemTableConstructor() { delete memtable_->Unref(); } virtual Status FinishImpl(const Options&, const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const stl_wrappers::KVMap& kv_map) override { delete memtable_->Unref(); ImmutableCFOptions mem_ioptions(ioptions); memtable_ = new MemTable(internal_comparator_, mem_ioptions, MutableCFOptions(options_, mem_ioptions), write_buffer_, kMaxSequenceNumber); memtable_->Ref(); int seq = 1; for (const auto kv : kv_map) { memtable_->Add(seq, kTypeValue, kv.first, kv.second); seq++; } return Status::OK(); } virtual InternalIterator* NewIterator() const override { return new KeyConvertingIterator( memtable_->NewIterator(ReadOptions(), &arena_), true); } virtual bool AnywayDeleteIterator() const override { return true; } virtual bool IsArenaMode() const override { return true; } private: mutable Arena arena_; InternalKeyComparator internal_comparator_; Options options_; WriteBuffer* write_buffer_; MemTable* memtable_; std::shared_ptr table_factory_; }; class InternalIteratorFromIterator : public InternalIterator { public: explicit InternalIteratorFromIterator(Iterator* it) : it_(it) {} virtual bool Valid() const override { return it_->Valid(); } virtual void Seek(const Slice& target) override { it_->Seek(target); } virtual void SeekToFirst() override { it_->SeekToFirst(); } virtual void SeekToLast() override { it_->SeekToLast(); } virtual void Next() override { it_->Next(); } virtual void Prev() override { it_->Prev(); } Slice key() const override { return it_->key(); } Slice value() const override { return it_->value(); } virtual Status status() const override { return it_->status(); } private: unique_ptr it_; }; class DBConstructor: public Constructor { public: explicit DBConstructor(const Comparator* cmp) : Constructor(cmp), comparator_(cmp) { db_ = nullptr; NewDB(); } ~DBConstructor() { delete db_; } virtual Status FinishImpl(const Options& options, const ImmutableCFOptions& ioptions, const BlockBasedTableOptions& table_options, const InternalKeyComparator& internal_comparator, const stl_wrappers::KVMap& kv_map) override { delete db_; db_ = nullptr; NewDB(); for (const auto kv : kv_map) { WriteBatch batch; batch.Put(kv.first, kv.second); EXPECT_TRUE(db_->Write(WriteOptions(), &batch).ok()); } return Status::OK(); } virtual InternalIterator* NewIterator() const override { return new InternalIteratorFromIterator(db_->NewIterator(ReadOptions())); } virtual DB* db() const override { return db_; } private: void NewDB() { std::string name = test::TmpDir() + "/table_testdb"; Options options; options.comparator = comparator_; Status status = DestroyDB(name, options); ASSERT_TRUE(status.ok()) << status.ToString(); options.create_if_missing = true; options.error_if_exists = true; options.write_buffer_size = 10000; // Something small to force merging status = DB::Open(options, name, &db_); ASSERT_TRUE(status.ok()) << status.ToString(); } const Comparator* comparator_; DB* db_; }; enum TestType { BLOCK_BASED_TABLE_TEST, #ifndef ROCKSDB_LITE PLAIN_TABLE_SEMI_FIXED_PREFIX, PLAIN_TABLE_FULL_STR_PREFIX, PLAIN_TABLE_TOTAL_ORDER, #endif // !ROCKSDB_LITE BLOCK_TEST, MEMTABLE_TEST, DB_TEST }; struct TestArgs { TestType type; bool reverse_compare; int restart_interval; CompressionType compression; uint32_t format_version; bool use_mmap; }; static std::vector GenerateArgList() { std::vector test_args; std::vector test_types = { BLOCK_BASED_TABLE_TEST, #ifndef ROCKSDB_LITE PLAIN_TABLE_SEMI_FIXED_PREFIX, PLAIN_TABLE_FULL_STR_PREFIX, PLAIN_TABLE_TOTAL_ORDER, #endif // !ROCKSDB_LITE BLOCK_TEST, MEMTABLE_TEST, DB_TEST}; std::vector reverse_compare_types = {false, true}; std::vector restart_intervals = {16, 1, 1024}; // Only add compression if it is supported std::vector> compression_types; compression_types.emplace_back(kNoCompression, false); if (Snappy_Supported()) { compression_types.emplace_back(kSnappyCompression, false); } if (Zlib_Supported()) { compression_types.emplace_back(kZlibCompression, false); compression_types.emplace_back(kZlibCompression, true); } if (BZip2_Supported()) { compression_types.emplace_back(kBZip2Compression, false); compression_types.emplace_back(kBZip2Compression, true); } if (LZ4_Supported()) { compression_types.emplace_back(kLZ4Compression, false); compression_types.emplace_back(kLZ4Compression, true); compression_types.emplace_back(kLZ4HCCompression, false); compression_types.emplace_back(kLZ4HCCompression, true); } if (ZSTD_Supported()) { compression_types.emplace_back(kZSTDNotFinalCompression, false); compression_types.emplace_back(kZSTDNotFinalCompression, true); } for (auto test_type : test_types) { for (auto reverse_compare : reverse_compare_types) { #ifndef ROCKSDB_LITE if (test_type == PLAIN_TABLE_SEMI_FIXED_PREFIX || test_type == PLAIN_TABLE_FULL_STR_PREFIX || test_type == PLAIN_TABLE_TOTAL_ORDER) { // Plain table doesn't use restart index or compression. TestArgs one_arg; one_arg.type = test_type; one_arg.reverse_compare = reverse_compare; one_arg.restart_interval = restart_intervals[0]; one_arg.compression = compression_types[0].first; one_arg.use_mmap = true; test_args.push_back(one_arg); one_arg.use_mmap = false; test_args.push_back(one_arg); continue; } #endif // !ROCKSDB_LITE for (auto restart_interval : restart_intervals) { for (auto compression_type : compression_types) { TestArgs one_arg; one_arg.type = test_type; one_arg.reverse_compare = reverse_compare; one_arg.restart_interval = restart_interval; one_arg.compression = compression_type.first; one_arg.format_version = compression_type.second ? 2 : 1; one_arg.use_mmap = false; test_args.push_back(one_arg); } } } } return test_args; } // In order to make all tests run for plain table format, including // those operating on empty keys, create a new prefix transformer which // return fixed prefix if the slice is not shorter than the prefix length, // and the full slice if it is shorter. class FixedOrLessPrefixTransform : public SliceTransform { private: const size_t prefix_len_; public: explicit FixedOrLessPrefixTransform(size_t prefix_len) : prefix_len_(prefix_len) { } virtual const char* Name() const override { return "rocksdb.FixedPrefix"; } virtual Slice Transform(const Slice& src) const override { assert(InDomain(src)); if (src.size() < prefix_len_) { return src; } return Slice(src.data(), prefix_len_); } virtual bool InDomain(const Slice& src) const override { return true; } virtual bool InRange(const Slice& dst) const override { return (dst.size() <= prefix_len_); } }; class HarnessTest : public testing::Test { public: HarnessTest() : ioptions_(options_), constructor_(nullptr), write_buffer_(options_.db_write_buffer_size) {} void Init(const TestArgs& args) { delete constructor_; constructor_ = nullptr; options_ = Options(); options_.compression = args.compression; // Use shorter block size for tests to exercise block boundary // conditions more. if (args.reverse_compare) { options_.comparator = &reverse_key_comparator; } internal_comparator_.reset( new test::PlainInternalKeyComparator(options_.comparator)); support_prev_ = true; only_support_prefix_seek_ = false; options_.allow_mmap_reads = args.use_mmap; switch (args.type) { case BLOCK_BASED_TABLE_TEST: table_options_.flush_block_policy_factory.reset( new FlushBlockBySizePolicyFactory()); table_options_.block_size = 256; table_options_.block_restart_interval = args.restart_interval; table_options_.index_block_restart_interval = args.restart_interval; table_options_.format_version = args.format_version; options_.table_factory.reset( new BlockBasedTableFactory(table_options_)); constructor_ = new TableConstructor(options_.comparator); break; // Plain table is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE case PLAIN_TABLE_SEMI_FIXED_PREFIX: support_prev_ = false; only_support_prefix_seek_ = true; options_.prefix_extractor.reset(new FixedOrLessPrefixTransform(2)); options_.table_factory.reset(NewPlainTableFactory()); constructor_ = new TableConstructor(options_.comparator, true); internal_comparator_.reset( new InternalKeyComparator(options_.comparator)); break; case PLAIN_TABLE_FULL_STR_PREFIX: support_prev_ = false; only_support_prefix_seek_ = true; options_.prefix_extractor.reset(NewNoopTransform()); options_.table_factory.reset(NewPlainTableFactory()); constructor_ = new TableConstructor(options_.comparator, true); internal_comparator_.reset( new InternalKeyComparator(options_.comparator)); break; case PLAIN_TABLE_TOTAL_ORDER: support_prev_ = false; only_support_prefix_seek_ = false; options_.prefix_extractor = nullptr; { PlainTableOptions plain_table_options; plain_table_options.user_key_len = kPlainTableVariableLength; plain_table_options.bloom_bits_per_key = 0; plain_table_options.hash_table_ratio = 0; options_.table_factory.reset( NewPlainTableFactory(plain_table_options)); } constructor_ = new TableConstructor(options_.comparator, true); internal_comparator_.reset( new InternalKeyComparator(options_.comparator)); break; #endif // !ROCKSDB_LITE case BLOCK_TEST: table_options_.block_size = 256; options_.table_factory.reset( new BlockBasedTableFactory(table_options_)); constructor_ = new BlockConstructor(options_.comparator); break; case MEMTABLE_TEST: table_options_.block_size = 256; options_.table_factory.reset( new BlockBasedTableFactory(table_options_)); constructor_ = new MemTableConstructor(options_.comparator, &write_buffer_); break; case DB_TEST: table_options_.block_size = 256; options_.table_factory.reset( new BlockBasedTableFactory(table_options_)); constructor_ = new DBConstructor(options_.comparator); break; } ioptions_ = ImmutableCFOptions(options_); } ~HarnessTest() { delete constructor_; } void Add(const std::string& key, const std::string& value) { constructor_->Add(key, value); } void Test(Random* rnd) { std::vector keys; stl_wrappers::KVMap data; constructor_->Finish(options_, ioptions_, table_options_, *internal_comparator_, &keys, &data); TestForwardScan(keys, data); if (support_prev_) { TestBackwardScan(keys, data); } TestRandomAccess(rnd, keys, data); } void TestForwardScan(const std::vector& keys, const stl_wrappers::KVMap& data) { InternalIterator* iter = constructor_->NewIterator(); ASSERT_TRUE(!iter->Valid()); iter->SeekToFirst(); for (stl_wrappers::KVMap::const_iterator model_iter = data.begin(); model_iter != data.end(); ++model_iter) { ASSERT_EQ(ToString(data, model_iter), ToString(iter)); iter->Next(); } ASSERT_TRUE(!iter->Valid()); if (constructor_->IsArenaMode() && !constructor_->AnywayDeleteIterator()) { iter->~InternalIterator(); } else { delete iter; } } void TestBackwardScan(const std::vector& keys, const stl_wrappers::KVMap& data) { InternalIterator* iter = constructor_->NewIterator(); ASSERT_TRUE(!iter->Valid()); iter->SeekToLast(); for (stl_wrappers::KVMap::const_reverse_iterator model_iter = data.rbegin(); model_iter != data.rend(); ++model_iter) { ASSERT_EQ(ToString(data, model_iter), ToString(iter)); iter->Prev(); } ASSERT_TRUE(!iter->Valid()); if (constructor_->IsArenaMode() && !constructor_->AnywayDeleteIterator()) { iter->~InternalIterator(); } else { delete iter; } } void TestRandomAccess(Random* rnd, const std::vector& keys, const stl_wrappers::KVMap& data) { static const bool kVerbose = false; InternalIterator* iter = constructor_->NewIterator(); ASSERT_TRUE(!iter->Valid()); stl_wrappers::KVMap::const_iterator model_iter = data.begin(); if (kVerbose) fprintf(stderr, "---\n"); for (int i = 0; i < 200; i++) { const int toss = rnd->Uniform(support_prev_ ? 5 : 3); switch (toss) { case 0: { if (iter->Valid()) { if (kVerbose) fprintf(stderr, "Next\n"); iter->Next(); ++model_iter; ASSERT_EQ(ToString(data, model_iter), ToString(iter)); } break; } case 1: { if (kVerbose) fprintf(stderr, "SeekToFirst\n"); iter->SeekToFirst(); model_iter = data.begin(); ASSERT_EQ(ToString(data, model_iter), ToString(iter)); break; } case 2: { std::string key = PickRandomKey(rnd, keys); model_iter = data.lower_bound(key); if (kVerbose) fprintf(stderr, "Seek '%s'\n", EscapeString(key).c_str()); iter->Seek(Slice(key)); ASSERT_EQ(ToString(data, model_iter), ToString(iter)); break; } case 3: { if (iter->Valid()) { if (kVerbose) fprintf(stderr, "Prev\n"); iter->Prev(); if (model_iter == data.begin()) { model_iter = data.end(); // Wrap around to invalid value } else { --model_iter; } ASSERT_EQ(ToString(data, model_iter), ToString(iter)); } break; } case 4: { if (kVerbose) fprintf(stderr, "SeekToLast\n"); iter->SeekToLast(); if (keys.empty()) { model_iter = data.end(); } else { std::string last = data.rbegin()->first; model_iter = data.lower_bound(last); } ASSERT_EQ(ToString(data, model_iter), ToString(iter)); break; } } } if (constructor_->IsArenaMode() && !constructor_->AnywayDeleteIterator()) { iter->~InternalIterator(); } else { delete iter; } } std::string ToString(const stl_wrappers::KVMap& data, const stl_wrappers::KVMap::const_iterator& it) { if (it == data.end()) { return "END"; } else { return "'" + it->first + "->" + it->second + "'"; } } std::string ToString(const stl_wrappers::KVMap& data, const stl_wrappers::KVMap::const_reverse_iterator& it) { if (it == data.rend()) { return "END"; } else { return "'" + it->first + "->" + it->second + "'"; } } std::string ToString(const InternalIterator* it) { if (!it->Valid()) { return "END"; } else { return "'" + it->key().ToString() + "->" + it->value().ToString() + "'"; } } std::string PickRandomKey(Random* rnd, const std::vector& keys) { if (keys.empty()) { return "foo"; } else { const int index = rnd->Uniform(static_cast(keys.size())); std::string result = keys[index]; switch (rnd->Uniform(support_prev_ ? 3 : 1)) { case 0: // Return an existing key break; case 1: { // Attempt to return something smaller than an existing key if (result.size() > 0 && result[result.size() - 1] > '\0' && (!only_support_prefix_seek_ || options_.prefix_extractor->Transform(result).size() < result.size())) { result[result.size() - 1]--; } break; } case 2: { // Return something larger than an existing key Increment(options_.comparator, &result); break; } } return result; } } // Returns nullptr if not running against a DB DB* db() const { return constructor_->db(); } private: Options options_ = Options(); ImmutableCFOptions ioptions_; BlockBasedTableOptions table_options_ = BlockBasedTableOptions(); Constructor* constructor_; WriteBuffer write_buffer_; bool support_prev_; bool only_support_prefix_seek_; shared_ptr internal_comparator_; }; static bool Between(uint64_t val, uint64_t low, uint64_t high) { bool result = (val >= low) && (val <= high); if (!result) { fprintf(stderr, "Value %llu is not in range [%llu, %llu]\n", (unsigned long long)(val), (unsigned long long)(low), (unsigned long long)(high)); } return result; } // Tests against all kinds of tables class TableTest : public testing::Test { public: const InternalKeyComparator& GetPlainInternalComparator( const Comparator* comp) { if (!plain_internal_comparator) { plain_internal_comparator.reset( new test::PlainInternalKeyComparator(comp)); } return *plain_internal_comparator; } private: std::unique_ptr plain_internal_comparator; }; class GeneralTableTest : public TableTest {}; class BlockBasedTableTest : public TableTest {}; class PlainTableTest : public TableTest {}; class TablePropertyTest : public testing::Test {}; // This test serves as the living tutorial for the prefix scan of user collected // properties. TEST_F(TablePropertyTest, PrefixScanTest) { UserCollectedProperties props{{"num.111.1", "1"}, {"num.111.2", "2"}, {"num.111.3", "3"}, {"num.333.1", "1"}, {"num.333.2", "2"}, {"num.333.3", "3"}, {"num.555.1", "1"}, {"num.555.2", "2"}, {"num.555.3", "3"}, }; // prefixes that exist for (const std::string& prefix : {"num.111", "num.333", "num.555"}) { int num = 0; for (auto pos = props.lower_bound(prefix); pos != props.end() && pos->first.compare(0, prefix.size(), prefix) == 0; ++pos) { ++num; auto key = prefix + "." + ToString(num); ASSERT_EQ(key, pos->first); ASSERT_EQ(ToString(num), pos->second); } ASSERT_EQ(3, num); } // prefixes that don't exist for (const std::string& prefix : {"num.000", "num.222", "num.444", "num.666"}) { auto pos = props.lower_bound(prefix); ASSERT_TRUE(pos == props.end() || pos->first.compare(0, prefix.size(), prefix) != 0); } } // This test include all the basic checks except those for index size and block // size, which will be conducted in separated unit tests. TEST_F(BlockBasedTableTest, BasicBlockBasedTableProperties) { TableConstructor c(BytewiseComparator()); c.Add("a1", "val1"); c.Add("b2", "val2"); c.Add("c3", "val3"); c.Add("d4", "val4"); c.Add("e5", "val5"); c.Add("f6", "val6"); c.Add("g7", "val7"); c.Add("h8", "val8"); c.Add("j9", "val9"); std::vector keys; stl_wrappers::KVMap kvmap; Options options; options.compression = kNoCompression; BlockBasedTableOptions table_options; table_options.block_restart_interval = 1; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); auto& props = *c.GetTableReader()->GetTableProperties(); ASSERT_EQ(kvmap.size(), props.num_entries); auto raw_key_size = kvmap.size() * 2ul; auto raw_value_size = kvmap.size() * 4ul; ASSERT_EQ(raw_key_size, props.raw_key_size); ASSERT_EQ(raw_value_size, props.raw_value_size); ASSERT_EQ(1ul, props.num_data_blocks); ASSERT_EQ("", props.filter_policy_name); // no filter policy is used // Verify data size. BlockBuilder block_builder(1); for (const auto& item : kvmap) { block_builder.Add(item.first, item.second); } Slice content = block_builder.Finish(); ASSERT_EQ(content.size() + kBlockTrailerSize, props.data_size); } TEST_F(BlockBasedTableTest, FilterPolicyNameProperties) { TableConstructor c(BytewiseComparator(), true); c.Add("a1", "val1"); std::vector keys; stl_wrappers::KVMap kvmap; BlockBasedTableOptions table_options; table_options.filter_policy.reset(NewBloomFilterPolicy(10)); Options options; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); auto& props = *c.GetTableReader()->GetTableProperties(); ASSERT_EQ("rocksdb.BuiltinBloomFilter", props.filter_policy_name); } // // BlockBasedTableTest::PrefetchTest // void AssertKeysInCache(BlockBasedTable* table_reader, const std::vector& keys_in_cache, const std::vector& keys_not_in_cache) { for (auto key : keys_in_cache) { ASSERT_TRUE(table_reader->TEST_KeyInCache(ReadOptions(), key)); } for (auto key : keys_not_in_cache) { ASSERT_TRUE(!table_reader->TEST_KeyInCache(ReadOptions(), key)); } } void PrefetchRange(TableConstructor* c, Options* opt, BlockBasedTableOptions* table_options, const std::vector& keys, const char* key_begin, const char* key_end, const std::vector& keys_in_cache, const std::vector& keys_not_in_cache, const Status expected_status = Status::OK()) { // reset the cache and reopen the table table_options->block_cache = NewLRUCache(16 * 1024 * 1024); opt->table_factory.reset(NewBlockBasedTableFactory(*table_options)); const ImmutableCFOptions ioptions2(*opt); ASSERT_OK(c->Reopen(ioptions2)); // prefetch auto* table_reader = dynamic_cast(c->GetTableReader()); // empty string replacement is a trick so we don't crash the test Slice begin(key_begin ? key_begin : ""); Slice end(key_end ? key_end : ""); Status s = table_reader->Prefetch(key_begin ? &begin : nullptr, key_end ? &end : nullptr); ASSERT_TRUE(s.code() == expected_status.code()); // assert our expectation in cache warmup AssertKeysInCache(table_reader, keys_in_cache, keys_not_in_cache); } TEST_F(BlockBasedTableTest, PrefetchTest) { // The purpose of this test is to test the prefetching operation built into // BlockBasedTable. Options opt; unique_ptr ikc; ikc.reset(new test::PlainInternalKeyComparator(opt.comparator)); opt.compression = kNoCompression; BlockBasedTableOptions table_options; table_options.block_size = 1024; // big enough so we don't ever lose cached values. table_options.block_cache = NewLRUCache(16 * 1024 * 1024); opt.table_factory.reset(NewBlockBasedTableFactory(table_options)); TableConstructor c(BytewiseComparator()); c.Add("k01", "hello"); c.Add("k02", "hello2"); c.Add("k03", std::string(10000, 'x')); c.Add("k04", std::string(200000, 'x')); c.Add("k05", std::string(300000, 'x')); c.Add("k06", "hello3"); c.Add("k07", std::string(100000, 'x')); std::vector keys; stl_wrappers::KVMap kvmap; const ImmutableCFOptions ioptions(opt); c.Finish(opt, ioptions, table_options, *ikc, &keys, &kvmap); // We get the following data spread : // // Data block Index // ======================== // [ k01 k02 k03 ] k03 // [ k04 ] k04 // [ k05 ] k05 // [ k06 k07 ] k07 // Simple PrefetchRange(&c, &opt, &table_options, keys, /*key_range=*/ "k01", "k05", /*keys_in_cache=*/ {"k01", "k02", "k03", "k04", "k05"}, /*keys_not_in_cache=*/ {"k06", "k07"}); PrefetchRange(&c, &opt, &table_options, keys, "k01", "k01", {"k01", "k02", "k03"}, {"k04", "k05", "k06", "k07"}); // odd PrefetchRange(&c, &opt, &table_options, keys, "a", "z", {"k01", "k02", "k03", "k04", "k05", "k06", "k07"}, {}); PrefetchRange(&c, &opt, &table_options, keys, "k00", "k00", {"k01", "k02", "k03"}, {"k04", "k05", "k06", "k07"}); // Edge cases PrefetchRange(&c, &opt, &table_options, keys, "k00", "k06", {"k01", "k02", "k03", "k04", "k05", "k06", "k07"}, {}); PrefetchRange(&c, &opt, &table_options, keys, "k00", "zzz", {"k01", "k02", "k03", "k04", "k05", "k06", "k07"}, {}); // null keys PrefetchRange(&c, &opt, &table_options, keys, nullptr, nullptr, {"k01", "k02", "k03", "k04", "k05", "k06", "k07"}, {}); PrefetchRange(&c, &opt, &table_options, keys, "k04", nullptr, {"k04", "k05", "k06", "k07"}, {"k01", "k02", "k03"}); PrefetchRange(&c, &opt, &table_options, keys, nullptr, "k05", {"k01", "k02", "k03", "k04", "k05"}, {"k06", "k07"}); // invalid PrefetchRange(&c, &opt, &table_options, keys, "k06", "k00", {}, {}, Status::InvalidArgument(Slice("k06 "), Slice("k07"))); } TEST_F(BlockBasedTableTest, TotalOrderSeekOnHashIndex) { BlockBasedTableOptions table_options; for (int i = 0; i < 4; ++i) { Options options; // Make each key/value an individual block table_options.block_size = 64; switch (i) { case 0: // Binary search index table_options.index_type = BlockBasedTableOptions::kBinarySearch; options.table_factory.reset(new BlockBasedTableFactory(table_options)); break; case 1: // Hash search index table_options.index_type = BlockBasedTableOptions::kHashSearch; options.table_factory.reset(new BlockBasedTableFactory(table_options)); options.prefix_extractor.reset(NewFixedPrefixTransform(4)); break; case 2: // Hash search index with hash_index_allow_collision table_options.index_type = BlockBasedTableOptions::kHashSearch; table_options.hash_index_allow_collision = true; options.table_factory.reset(new BlockBasedTableFactory(table_options)); options.prefix_extractor.reset(NewFixedPrefixTransform(4)); break; case 3: default: // Hash search index with filter policy table_options.index_type = BlockBasedTableOptions::kHashSearch; table_options.filter_policy.reset(NewBloomFilterPolicy(10)); options.table_factory.reset(new BlockBasedTableFactory(table_options)); options.prefix_extractor.reset(NewFixedPrefixTransform(4)); break; } TableConstructor c(BytewiseComparator(), true); c.Add("aaaa1", std::string('a', 56)); c.Add("bbaa1", std::string('a', 56)); c.Add("cccc1", std::string('a', 56)); c.Add("bbbb1", std::string('a', 56)); c.Add("baaa1", std::string('a', 56)); c.Add("abbb1", std::string('a', 56)); c.Add("cccc2", std::string('a', 56)); std::vector keys; stl_wrappers::KVMap kvmap; const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); auto props = c.GetTableReader()->GetTableProperties(); ASSERT_EQ(7u, props->num_data_blocks); auto* reader = c.GetTableReader(); ReadOptions ro; ro.total_order_seek = true; std::unique_ptr iter(reader->NewIterator(ro)); iter->Seek(InternalKey("b", 0, kTypeValue).Encode()); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("baaa1", ExtractUserKey(iter->key()).ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("bbaa1", ExtractUserKey(iter->key()).ToString()); iter->Seek(InternalKey("bb", 0, kTypeValue).Encode()); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("bbaa1", ExtractUserKey(iter->key()).ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("bbbb1", ExtractUserKey(iter->key()).ToString()); iter->Seek(InternalKey("bbb", 0, kTypeValue).Encode()); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("bbbb1", ExtractUserKey(iter->key()).ToString()); iter->Next(); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("cccc1", ExtractUserKey(iter->key()).ToString()); } } TEST_F(BlockBasedTableTest, NoopTransformSeek) { BlockBasedTableOptions table_options; table_options.filter_policy.reset(NewBloomFilterPolicy(10)); Options options; options.comparator = BytewiseComparator(); options.table_factory.reset(new BlockBasedTableFactory(table_options)); options.prefix_extractor.reset(NewNoopTransform()); TableConstructor c(options.comparator); // To tickle the PrefixMayMatch bug it is important that the // user-key is a single byte so that the index key exactly matches // the user-key. InternalKey key("a", 1, kTypeValue); c.Add(key.Encode().ToString(), "b"); std::vector keys; stl_wrappers::KVMap kvmap; const ImmutableCFOptions ioptions(options); const InternalKeyComparator internal_comparator(options.comparator); c.Finish(options, ioptions, table_options, internal_comparator, &keys, &kvmap); auto* reader = c.GetTableReader(); for (int i = 0; i < 2; ++i) { ReadOptions ro; ro.total_order_seek = (i == 0); std::unique_ptr iter(reader->NewIterator(ro)); iter->Seek(key.Encode()); ASSERT_OK(iter->status()); ASSERT_TRUE(iter->Valid()); ASSERT_EQ("a", ExtractUserKey(iter->key()).ToString()); } } static std::string RandomString(Random* rnd, int len) { std::string r; test::RandomString(rnd, len, &r); return r; } void AddInternalKey(TableConstructor* c, const std::string& prefix, int suffix_len = 800) { static Random rnd(1023); InternalKey k(prefix + RandomString(&rnd, 800), 0, kTypeValue); c->Add(k.Encode().ToString(), "v"); } TEST_F(TableTest, HashIndexTest) { TableConstructor c(BytewiseComparator()); // keys with prefix length 3, make sure the key/value is big enough to fill // one block AddInternalKey(&c, "0015"); AddInternalKey(&c, "0035"); AddInternalKey(&c, "0054"); AddInternalKey(&c, "0055"); AddInternalKey(&c, "0056"); AddInternalKey(&c, "0057"); AddInternalKey(&c, "0058"); AddInternalKey(&c, "0075"); AddInternalKey(&c, "0076"); AddInternalKey(&c, "0095"); std::vector keys; stl_wrappers::KVMap kvmap; Options options; options.prefix_extractor.reset(NewFixedPrefixTransform(3)); BlockBasedTableOptions table_options; table_options.index_type = BlockBasedTableOptions::kHashSearch; table_options.hash_index_allow_collision = true; table_options.block_size = 1700; table_options.block_cache = NewLRUCache(1024); options.table_factory.reset(NewBlockBasedTableFactory(table_options)); std::unique_ptr comparator( new InternalKeyComparator(BytewiseComparator())); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, *comparator, &keys, &kvmap); auto reader = c.GetTableReader(); auto props = reader->GetTableProperties(); ASSERT_EQ(5u, props->num_data_blocks); std::unique_ptr hash_iter( reader->NewIterator(ReadOptions())); // -- Find keys do not exist, but have common prefix. std::vector prefixes = {"001", "003", "005", "007", "009"}; std::vector lower_bound = {keys[0], keys[1], keys[2], keys[7], keys[9], }; // find the lower bound of the prefix for (size_t i = 0; i < prefixes.size(); ++i) { hash_iter->Seek(InternalKey(prefixes[i], 0, kTypeValue).Encode()); ASSERT_OK(hash_iter->status()); ASSERT_TRUE(hash_iter->Valid()); // seek the first element in the block ASSERT_EQ(lower_bound[i], hash_iter->key().ToString()); ASSERT_EQ("v", hash_iter->value().ToString()); } // find the upper bound of prefixes std::vector upper_bound = {keys[1], keys[2], keys[7], keys[9], }; // find existing keys for (const auto& item : kvmap) { auto ukey = ExtractUserKey(item.first).ToString(); hash_iter->Seek(ukey); // ASSERT_OK(regular_iter->status()); ASSERT_OK(hash_iter->status()); // ASSERT_TRUE(regular_iter->Valid()); ASSERT_TRUE(hash_iter->Valid()); ASSERT_EQ(item.first, hash_iter->key().ToString()); ASSERT_EQ(item.second, hash_iter->value().ToString()); } for (size_t i = 0; i < prefixes.size(); ++i) { // the key is greater than any existing keys. auto key = prefixes[i] + "9"; hash_iter->Seek(InternalKey(key, 0, kTypeValue).Encode()); ASSERT_OK(hash_iter->status()); if (i == prefixes.size() - 1) { // last key ASSERT_TRUE(!hash_iter->Valid()); } else { ASSERT_TRUE(hash_iter->Valid()); // seek the first element in the block ASSERT_EQ(upper_bound[i], hash_iter->key().ToString()); ASSERT_EQ("v", hash_iter->value().ToString()); } } // find keys with prefix that don't match any of the existing prefixes. std::vector non_exist_prefixes = {"002", "004", "006", "008"}; for (const auto& prefix : non_exist_prefixes) { hash_iter->Seek(InternalKey(prefix, 0, kTypeValue).Encode()); // regular_iter->Seek(prefix); ASSERT_OK(hash_iter->status()); // Seek to non-existing prefixes should yield either invalid, or a // key with prefix greater than the target. if (hash_iter->Valid()) { Slice ukey = ExtractUserKey(hash_iter->key()); Slice ukey_prefix = options.prefix_extractor->Transform(ukey); ASSERT_TRUE(BytewiseComparator()->Compare(prefix, ukey_prefix) < 0); } } } // It's very hard to figure out the index block size of a block accurately. // To make sure we get the index size, we just make sure as key number // grows, the filter block size also grows. TEST_F(BlockBasedTableTest, IndexSizeStat) { uint64_t last_index_size = 0; // we need to use random keys since the pure human readable texts // may be well compressed, resulting insignifcant change of index // block size. Random rnd(test::RandomSeed()); std::vector keys; for (int i = 0; i < 100; ++i) { keys.push_back(RandomString(&rnd, 10000)); } // Each time we load one more key to the table. the table index block // size is expected to be larger than last time's. for (size_t i = 1; i < keys.size(); ++i) { TableConstructor c(BytewiseComparator()); for (size_t j = 0; j < i; ++j) { c.Add(keys[j], "val"); } std::vector ks; stl_wrappers::KVMap kvmap; Options options; options.compression = kNoCompression; BlockBasedTableOptions table_options; table_options.block_restart_interval = 1; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &ks, &kvmap); auto index_size = c.GetTableReader()->GetTableProperties()->index_size; ASSERT_GT(index_size, last_index_size); last_index_size = index_size; } } TEST_F(BlockBasedTableTest, NumBlockStat) { Random rnd(test::RandomSeed()); TableConstructor c(BytewiseComparator()); Options options; options.compression = kNoCompression; BlockBasedTableOptions table_options; table_options.block_restart_interval = 1; table_options.block_size = 1000; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); for (int i = 0; i < 10; ++i) { // the key/val are slightly smaller than block size, so that each block // holds roughly one key/value pair. c.Add(RandomString(&rnd, 900), "val"); } std::vector ks; stl_wrappers::KVMap kvmap; const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &ks, &kvmap); ASSERT_EQ(kvmap.size(), c.GetTableReader()->GetTableProperties()->num_data_blocks); } // A simple tool that takes the snapshot of block cache statistics. class BlockCachePropertiesSnapshot { public: explicit BlockCachePropertiesSnapshot(Statistics* statistics) { block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_MISS); block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_HIT); index_block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_INDEX_MISS); index_block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_INDEX_HIT); data_block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_DATA_MISS); data_block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_DATA_HIT); filter_block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_FILTER_MISS); filter_block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_FILTER_HIT); block_cache_bytes_read = statistics->getTickerCount(BLOCK_CACHE_BYTES_READ); block_cache_bytes_write = statistics->getTickerCount(BLOCK_CACHE_BYTES_WRITE); } void AssertIndexBlockStat(int64_t expected_index_block_cache_miss, int64_t expected_index_block_cache_hit) { ASSERT_EQ(expected_index_block_cache_miss, index_block_cache_miss); ASSERT_EQ(expected_index_block_cache_hit, index_block_cache_hit); } void AssertFilterBlockStat(int64_t expected_filter_block_cache_miss, int64_t expected_filter_block_cache_hit) { ASSERT_EQ(expected_filter_block_cache_miss, filter_block_cache_miss); ASSERT_EQ(expected_filter_block_cache_hit, filter_block_cache_hit); } // Check if the fetched props matches the expected ones. // TODO(kailiu) Use this only when you disabled filter policy! void AssertEqual(int64_t expected_index_block_cache_miss, int64_t expected_index_block_cache_hit, int64_t expected_data_block_cache_miss, int64_t expected_data_block_cache_hit) const { ASSERT_EQ(expected_index_block_cache_miss, index_block_cache_miss); ASSERT_EQ(expected_index_block_cache_hit, index_block_cache_hit); ASSERT_EQ(expected_data_block_cache_miss, data_block_cache_miss); ASSERT_EQ(expected_data_block_cache_hit, data_block_cache_hit); ASSERT_EQ(expected_index_block_cache_miss + expected_data_block_cache_miss, block_cache_miss); ASSERT_EQ(expected_index_block_cache_hit + expected_data_block_cache_hit, block_cache_hit); } int64_t GetCacheBytesRead() { return block_cache_bytes_read; } int64_t GetCacheBytesWrite() { return block_cache_bytes_write; } private: int64_t block_cache_miss = 0; int64_t block_cache_hit = 0; int64_t index_block_cache_miss = 0; int64_t index_block_cache_hit = 0; int64_t data_block_cache_miss = 0; int64_t data_block_cache_hit = 0; int64_t filter_block_cache_miss = 0; int64_t filter_block_cache_hit = 0; int64_t block_cache_bytes_read = 0; int64_t block_cache_bytes_write = 0; }; // Make sure, by default, index/filter blocks were pre-loaded (meaning we won't // use block cache to store them). TEST_F(BlockBasedTableTest, BlockCacheDisabledTest) { Options options; options.create_if_missing = true; options.statistics = CreateDBStatistics(); BlockBasedTableOptions table_options; table_options.block_cache = NewLRUCache(1024); table_options.filter_policy.reset(NewBloomFilterPolicy(10)); options.table_factory.reset(new BlockBasedTableFactory(table_options)); std::vector keys; stl_wrappers::KVMap kvmap; TableConstructor c(BytewiseComparator(), true); c.Add("key", "value"); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); // preloading filter/index blocks is enabled. auto reader = dynamic_cast(c.GetTableReader()); ASSERT_TRUE(reader->TEST_filter_block_preloaded()); ASSERT_TRUE(reader->TEST_index_reader_preloaded()); { // nothing happens in the beginning BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertIndexBlockStat(0, 0); props.AssertFilterBlockStat(0, 0); } { GetContext get_context(options.comparator, nullptr, nullptr, nullptr, GetContext::kNotFound, Slice(), nullptr, nullptr, nullptr, nullptr); // a hack that just to trigger BlockBasedTable::GetFilter. reader->Get(ReadOptions(), "non-exist-key", &get_context); BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertIndexBlockStat(0, 0); props.AssertFilterBlockStat(0, 0); } } // Due to the difficulities of the intersaction between statistics, this test // only tests the case when "index block is put to block cache" TEST_F(BlockBasedTableTest, FilterBlockInBlockCache) { // -- Table construction Options options; options.create_if_missing = true; options.statistics = CreateDBStatistics(); // Enable the cache for index/filter blocks BlockBasedTableOptions table_options; table_options.block_cache = NewLRUCache(1024); table_options.cache_index_and_filter_blocks = true; options.table_factory.reset(new BlockBasedTableFactory(table_options)); std::vector keys; stl_wrappers::KVMap kvmap; TableConstructor c(BytewiseComparator()); c.Add("key", "value"); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); // preloading filter/index blocks is prohibited. auto* reader = dynamic_cast(c.GetTableReader()); ASSERT_TRUE(!reader->TEST_filter_block_preloaded()); ASSERT_TRUE(!reader->TEST_index_reader_preloaded()); // -- PART 1: Open with regular block cache. // Since block_cache is disabled, no cache activities will be involved. unique_ptr iter; int64_t last_cache_bytes_read = 0; // At first, no block will be accessed. { BlockCachePropertiesSnapshot props(options.statistics.get()); // index will be added to block cache. props.AssertEqual(1, // index block miss 0, 0, 0); ASSERT_EQ(props.GetCacheBytesRead(), 0); ASSERT_EQ(props.GetCacheBytesWrite(), table_options.block_cache->GetUsage()); last_cache_bytes_read = props.GetCacheBytesRead(); } // Only index block will be accessed { iter.reset(c.NewIterator()); BlockCachePropertiesSnapshot props(options.statistics.get()); // NOTE: to help better highlight the "detla" of each ticker, I use // + to indicate the increment of changed // value; other numbers remain the same. props.AssertEqual(1, 0 + 1, // index block hit 0, 0); // Cache hit, bytes read from cache should increase ASSERT_GT(props.GetCacheBytesRead(), last_cache_bytes_read); ASSERT_EQ(props.GetCacheBytesWrite(), table_options.block_cache->GetUsage()); last_cache_bytes_read = props.GetCacheBytesRead(); } // Only data block will be accessed { iter->SeekToFirst(); BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertEqual(1, 1, 0 + 1, // data block miss 0); // Cache miss, Bytes read from cache should not change ASSERT_EQ(props.GetCacheBytesRead(), last_cache_bytes_read); ASSERT_EQ(props.GetCacheBytesWrite(), table_options.block_cache->GetUsage()); last_cache_bytes_read = props.GetCacheBytesRead(); } // Data block will be in cache { iter.reset(c.NewIterator()); iter->SeekToFirst(); BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertEqual(1, 1 + 1, /* index block hit */ 1, 0 + 1 /* data block hit */); // Cache hit, bytes read from cache should increase ASSERT_GT(props.GetCacheBytesRead(), last_cache_bytes_read); ASSERT_EQ(props.GetCacheBytesWrite(), table_options.block_cache->GetUsage()); last_cache_bytes_read = props.GetCacheBytesRead(); } // release the iterator so that the block cache can reset correctly. iter.reset(); // -- PART 2: Open with very small block cache // In this test, no block will ever get hit since the block cache is // too small to fit even one entry. table_options.block_cache = NewLRUCache(1); options.statistics = CreateDBStatistics(); options.table_factory.reset(new BlockBasedTableFactory(table_options)); const ImmutableCFOptions ioptions2(options); c.Reopen(ioptions2); { BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertEqual(1, // index block miss 0, 0, 0); // Cache miss, Bytes read from cache should not change ASSERT_EQ(props.GetCacheBytesRead(), 0); } { // Both index and data block get accessed. // It first cache index block then data block. But since the cache size // is only 1, index block will be purged after data block is inserted. iter.reset(c.NewIterator()); BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertEqual(1 + 1, // index block miss 0, 0, // data block miss 0); // Cache hit, bytes read from cache should increase ASSERT_EQ(props.GetCacheBytesRead(), 0); } { // SeekToFirst() accesses data block. With similar reason, we expect data // block's cache miss. iter->SeekToFirst(); BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertEqual(2, 0, 0 + 1, // data block miss 0); // Cache miss, Bytes read from cache should not change ASSERT_EQ(props.GetCacheBytesRead(), 0); } iter.reset(); // -- PART 3: Open table with bloom filter enabled but not in SST file table_options.block_cache = NewLRUCache(4096); table_options.cache_index_and_filter_blocks = false; options.table_factory.reset(NewBlockBasedTableFactory(table_options)); TableConstructor c3(BytewiseComparator()); std::string user_key = "k01"; InternalKey internal_key(user_key, 0, kTypeValue); c3.Add(internal_key.Encode().ToString(), "hello"); ImmutableCFOptions ioptions3(options); // Generate table without filter policy c3.Finish(options, ioptions3, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); // Open table with filter policy table_options.filter_policy.reset(NewBloomFilterPolicy(1)); options.table_factory.reset(new BlockBasedTableFactory(table_options)); options.statistics = CreateDBStatistics(); ImmutableCFOptions ioptions4(options); ASSERT_OK(c3.Reopen(ioptions4)); reader = dynamic_cast(c3.GetTableReader()); ASSERT_TRUE(!reader->TEST_filter_block_preloaded()); std::string value; GetContext get_context(options.comparator, nullptr, nullptr, nullptr, GetContext::kNotFound, user_key, &value, nullptr, nullptr, nullptr); ASSERT_OK(reader->Get(ReadOptions(), user_key, &get_context)); ASSERT_EQ(value, "hello"); BlockCachePropertiesSnapshot props(options.statistics.get()); props.AssertFilterBlockStat(0, 0); } void ValidateBlockSizeDeviation(int value, int expected) { BlockBasedTableOptions table_options; table_options.block_size_deviation = value; BlockBasedTableFactory* factory = new BlockBasedTableFactory(table_options); const BlockBasedTableOptions* normalized_table_options = (const BlockBasedTableOptions*)factory->GetOptions(); ASSERT_EQ(normalized_table_options->block_size_deviation, expected); delete factory; } void ValidateBlockRestartInterval(int value, int expected) { BlockBasedTableOptions table_options; table_options.block_restart_interval = value; BlockBasedTableFactory* factory = new BlockBasedTableFactory(table_options); const BlockBasedTableOptions* normalized_table_options = (const BlockBasedTableOptions*)factory->GetOptions(); ASSERT_EQ(normalized_table_options->block_restart_interval, expected); delete factory; } TEST_F(BlockBasedTableTest, InvalidOptions) { // invalid values for block_size_deviation (<0 or >100) are silently set to 0 ValidateBlockSizeDeviation(-10, 0); ValidateBlockSizeDeviation(-1, 0); ValidateBlockSizeDeviation(0, 0); ValidateBlockSizeDeviation(1, 1); ValidateBlockSizeDeviation(99, 99); ValidateBlockSizeDeviation(100, 100); ValidateBlockSizeDeviation(101, 0); ValidateBlockSizeDeviation(1000, 0); // invalid values for block_restart_interval (<1) are silently set to 1 ValidateBlockRestartInterval(-10, 1); ValidateBlockRestartInterval(-1, 1); ValidateBlockRestartInterval(0, 1); ValidateBlockRestartInterval(1, 1); ValidateBlockRestartInterval(2, 2); ValidateBlockRestartInterval(1000, 1000); } TEST_F(BlockBasedTableTest, BlockReadCountTest) { // bloom_filter_type = 0 -- block-based filter // bloom_filter_type = 0 -- full filter for (int bloom_filter_type = 0; bloom_filter_type < 2; ++bloom_filter_type) { for (int index_and_filter_in_cache = 0; index_and_filter_in_cache < 2; ++index_and_filter_in_cache) { Options options; options.create_if_missing = true; BlockBasedTableOptions table_options; table_options.block_cache = NewLRUCache(1, 0); table_options.cache_index_and_filter_blocks = index_and_filter_in_cache; table_options.filter_policy.reset( NewBloomFilterPolicy(10, bloom_filter_type == 0)); options.table_factory.reset(new BlockBasedTableFactory(table_options)); std::vector keys; stl_wrappers::KVMap kvmap; TableConstructor c(BytewiseComparator()); std::string user_key = "k04"; InternalKey internal_key(user_key, 0, kTypeValue); std::string encoded_key = internal_key.Encode().ToString(); c.Add(encoded_key, "hello"); ImmutableCFOptions ioptions(options); // Generate table with filter policy c.Finish(options, ioptions, table_options, GetPlainInternalComparator(options.comparator), &keys, &kvmap); auto reader = c.GetTableReader(); std::string value; GetContext get_context(options.comparator, nullptr, nullptr, nullptr, GetContext::kNotFound, user_key, &value, nullptr, nullptr, nullptr); perf_context.Reset(); ASSERT_OK(reader->Get(ReadOptions(), encoded_key, &get_context)); if (index_and_filter_in_cache) { // data, index and filter block ASSERT_EQ(perf_context.block_read_count, 3); } else { // just the data block ASSERT_EQ(perf_context.block_read_count, 1); } ASSERT_EQ(get_context.State(), GetContext::kFound); ASSERT_EQ(value, "hello"); // Get non-existing key user_key = "does-not-exist"; internal_key = InternalKey(user_key, 0, kTypeValue); encoded_key = internal_key.Encode().ToString(); get_context = GetContext(options.comparator, nullptr, nullptr, nullptr, GetContext::kNotFound, user_key, &value, nullptr, nullptr, nullptr); perf_context.Reset(); ASSERT_OK(reader->Get(ReadOptions(), encoded_key, &get_context)); ASSERT_EQ(get_context.State(), GetContext::kNotFound); if (index_and_filter_in_cache) { if (bloom_filter_type == 0) { // with block-based, we read index and then the filter ASSERT_EQ(perf_context.block_read_count, 2); } else { // with full-filter, we read filter first and then we stop ASSERT_EQ(perf_context.block_read_count, 1); } } else { // filter is already in memory and it figures out that the key doesn't // exist ASSERT_EQ(perf_context.block_read_count, 0); } } } } TEST_F(BlockBasedTableTest, BlockCacheLeak) { // Check that when we reopen a table we don't lose access to blocks already // in the cache. This test checks whether the Table actually makes use of the // unique ID from the file. Options opt; unique_ptr ikc; ikc.reset(new test::PlainInternalKeyComparator(opt.comparator)); opt.compression = kNoCompression; BlockBasedTableOptions table_options; table_options.block_size = 1024; // big enough so we don't ever lose cached values. table_options.block_cache = NewLRUCache(16 * 1024 * 1024); opt.table_factory.reset(NewBlockBasedTableFactory(table_options)); TableConstructor c(BytewiseComparator()); c.Add("k01", "hello"); c.Add("k02", "hello2"); c.Add("k03", std::string(10000, 'x')); c.Add("k04", std::string(200000, 'x')); c.Add("k05", std::string(300000, 'x')); c.Add("k06", "hello3"); c.Add("k07", std::string(100000, 'x')); std::vector keys; stl_wrappers::KVMap kvmap; const ImmutableCFOptions ioptions(opt); c.Finish(opt, ioptions, table_options, *ikc, &keys, &kvmap); unique_ptr iter(c.NewIterator()); iter->SeekToFirst(); while (iter->Valid()) { iter->key(); iter->value(); iter->Next(); } ASSERT_OK(iter->status()); const ImmutableCFOptions ioptions1(opt); ASSERT_OK(c.Reopen(ioptions1)); auto table_reader = dynamic_cast(c.GetTableReader()); for (const std::string& key : keys) { ASSERT_TRUE(table_reader->TEST_KeyInCache(ReadOptions(), key)); } // rerun with different block cache table_options.block_cache = NewLRUCache(16 * 1024 * 1024); opt.table_factory.reset(NewBlockBasedTableFactory(table_options)); const ImmutableCFOptions ioptions2(opt); ASSERT_OK(c.Reopen(ioptions2)); table_reader = dynamic_cast(c.GetTableReader()); for (const std::string& key : keys) { ASSERT_TRUE(!table_reader->TEST_KeyInCache(ReadOptions(), key)); } } // Plain table is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE TEST_F(PlainTableTest, BasicPlainTableProperties) { PlainTableOptions plain_table_options; plain_table_options.user_key_len = 8; plain_table_options.bloom_bits_per_key = 8; plain_table_options.hash_table_ratio = 0; PlainTableFactory factory(plain_table_options); test::StringSink sink; unique_ptr file_writer( test::GetWritableFileWriter(new test::StringSink())); Options options; const ImmutableCFOptions ioptions(options); InternalKeyComparator ikc(options.comparator); std::vector> int_tbl_prop_collector_factories; std::unique_ptr builder(factory.NewTableBuilder( TableBuilderOptions(ioptions, ikc, &int_tbl_prop_collector_factories, kNoCompression, CompressionOptions(), false), TablePropertiesCollectorFactory::Context::kUnknownColumnFamily, file_writer.get())); for (char c = 'a'; c <= 'z'; ++c) { std::string key(8, c); key.append("\1 "); // PlainTable expects internal key structure std::string value(28, c + 42); builder->Add(key, value); } ASSERT_OK(builder->Finish()); file_writer->Flush(); test::StringSink* ss = static_cast(file_writer->writable_file()); unique_ptr file_reader( test::GetRandomAccessFileReader( new test::StringSource(ss->contents(), 72242, true))); TableProperties* props = nullptr; auto s = ReadTableProperties(file_reader.get(), ss->contents().size(), kPlainTableMagicNumber, Env::Default(), nullptr, &props); std::unique_ptr props_guard(props); ASSERT_OK(s); ASSERT_EQ(0ul, props->index_size); ASSERT_EQ(0ul, props->filter_size); ASSERT_EQ(16ul * 26, props->raw_key_size); ASSERT_EQ(28ul * 26, props->raw_value_size); ASSERT_EQ(26ul, props->num_entries); ASSERT_EQ(1ul, props->num_data_blocks); } #endif // !ROCKSDB_LITE TEST_F(GeneralTableTest, ApproximateOffsetOfPlain) { TableConstructor c(BytewiseComparator()); c.Add("k01", "hello"); c.Add("k02", "hello2"); c.Add("k03", std::string(10000, 'x')); c.Add("k04", std::string(200000, 'x')); c.Add("k05", std::string(300000, 'x')); c.Add("k06", "hello3"); c.Add("k07", std::string(100000, 'x')); std::vector keys; stl_wrappers::KVMap kvmap; Options options; test::PlainInternalKeyComparator internal_comparator(options.comparator); options.compression = kNoCompression; BlockBasedTableOptions table_options; table_options.block_size = 1024; const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, internal_comparator, &keys, &kvmap); ASSERT_TRUE(Between(c.ApproximateOffsetOf("abc"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k01"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k01a"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k02"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k03"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k04"), 10000, 11000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k04a"), 210000, 211000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k05"), 210000, 211000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k06"), 510000, 511000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k07"), 510000, 511000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("xyz"), 610000, 612000)); } static void DoCompressionTest(CompressionType comp) { Random rnd(301); TableConstructor c(BytewiseComparator()); std::string tmp; c.Add("k01", "hello"); c.Add("k02", test::CompressibleString(&rnd, 0.25, 10000, &tmp)); c.Add("k03", "hello3"); c.Add("k04", test::CompressibleString(&rnd, 0.25, 10000, &tmp)); std::vector keys; stl_wrappers::KVMap kvmap; Options options; test::PlainInternalKeyComparator ikc(options.comparator); options.compression = comp; BlockBasedTableOptions table_options; table_options.block_size = 1024; const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, ikc, &keys, &kvmap); ASSERT_TRUE(Between(c.ApproximateOffsetOf("abc"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k01"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k02"), 0, 0)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k03"), 2000, 3000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("k04"), 2000, 3000)); ASSERT_TRUE(Between(c.ApproximateOffsetOf("xyz"), 4000, 6100)); } TEST_F(GeneralTableTest, ApproximateOffsetOfCompressed) { std::vector compression_state; if (!Snappy_Supported()) { fprintf(stderr, "skipping snappy compression tests\n"); } else { compression_state.push_back(kSnappyCompression); } if (!Zlib_Supported()) { fprintf(stderr, "skipping zlib compression tests\n"); } else { compression_state.push_back(kZlibCompression); } // TODO(kailiu) DoCompressionTest() doesn't work with BZip2. /* if (!BZip2_Supported()) { fprintf(stderr, "skipping bzip2 compression tests\n"); } else { compression_state.push_back(kBZip2Compression); } */ if (!LZ4_Supported()) { fprintf(stderr, "skipping lz4 and lz4hc compression tests\n"); } else { compression_state.push_back(kLZ4Compression); compression_state.push_back(kLZ4HCCompression); } for (auto state : compression_state) { DoCompressionTest(state); } } TEST_F(HarnessTest, Randomized) { std::vector args = GenerateArgList(); for (unsigned int i = 0; i < args.size(); i++) { Init(args[i]); Random rnd(test::RandomSeed() + 5); for (int num_entries = 0; num_entries < 2000; num_entries += (num_entries < 50 ? 1 : 200)) { if ((num_entries % 10) == 0) { fprintf(stderr, "case %d of %d: num_entries = %d\n", (i + 1), static_cast(args.size()), num_entries); } for (int e = 0; e < num_entries; e++) { std::string v; Add(test::RandomKey(&rnd, rnd.Skewed(4)), test::RandomString(&rnd, rnd.Skewed(5), &v).ToString()); } Test(&rnd); } } } #ifndef ROCKSDB_LITE TEST_F(HarnessTest, RandomizedLongDB) { Random rnd(test::RandomSeed()); TestArgs args = {DB_TEST, false, 16, kNoCompression, 0, false}; Init(args); int num_entries = 100000; for (int e = 0; e < num_entries; e++) { std::string v; Add(test::RandomKey(&rnd, rnd.Skewed(4)), test::RandomString(&rnd, rnd.Skewed(5), &v).ToString()); } Test(&rnd); // We must have created enough data to force merging int files = 0; for (int level = 0; level < db()->NumberLevels(); level++) { std::string value; char name[100]; snprintf(name, sizeof(name), "rocksdb.num-files-at-level%d", level); ASSERT_TRUE(db()->GetProperty(name, &value)); files += atoi(value.c_str()); } ASSERT_GT(files, 0); } #endif // ROCKSDB_LITE class MemTableTest : public testing::Test {}; TEST_F(MemTableTest, Simple) { InternalKeyComparator cmp(BytewiseComparator()); auto table_factory = std::make_shared(); Options options; options.memtable_factory = table_factory; ImmutableCFOptions ioptions(options); WriteBuffer wb(options.db_write_buffer_size); MemTable* memtable = new MemTable(cmp, ioptions, MutableCFOptions(options, ioptions), &wb, kMaxSequenceNumber); memtable->Ref(); WriteBatch batch; WriteBatchInternal::SetSequence(&batch, 100); batch.Put(std::string("k1"), std::string("v1")); batch.Put(std::string("k2"), std::string("v2")); batch.Put(std::string("k3"), std::string("v3")); batch.Put(std::string("largekey"), std::string("vlarge")); ColumnFamilyMemTablesDefault cf_mems_default(memtable); ASSERT_TRUE( WriteBatchInternal::InsertInto(&batch, &cf_mems_default, nullptr).ok()); Arena arena; ScopedArenaIterator iter(memtable->NewIterator(ReadOptions(), &arena)); iter->SeekToFirst(); while (iter->Valid()) { fprintf(stderr, "key: '%s' -> '%s'\n", iter->key().ToString().c_str(), iter->value().ToString().c_str()); iter->Next(); } delete memtable->Unref(); } // Test the empty key TEST_F(HarnessTest, SimpleEmptyKey) { auto args = GenerateArgList(); for (const auto& arg : args) { Init(arg); Random rnd(test::RandomSeed() + 1); Add("", "v"); Test(&rnd); } } TEST_F(HarnessTest, SimpleSingle) { auto args = GenerateArgList(); for (const auto& arg : args) { Init(arg); Random rnd(test::RandomSeed() + 2); Add("abc", "v"); Test(&rnd); } } TEST_F(HarnessTest, SimpleMulti) { auto args = GenerateArgList(); for (const auto& arg : args) { Init(arg); Random rnd(test::RandomSeed() + 3); Add("abc", "v"); Add("abcd", "v"); Add("ac", "v2"); Test(&rnd); } } TEST_F(HarnessTest, SimpleSpecialKey) { auto args = GenerateArgList(); for (const auto& arg : args) { Init(arg); Random rnd(test::RandomSeed() + 4); Add("\xff\xff", "v3"); Test(&rnd); } } TEST_F(HarnessTest, FooterTests) { { // upconvert legacy block based std::string encoded; Footer footer(kLegacyBlockBasedTableMagicNumber, 0); BlockHandle meta_index(10, 5), index(20, 15); footer.set_metaindex_handle(meta_index); footer.set_index_handle(index); footer.EncodeTo(&encoded); Footer decoded_footer; Slice encoded_slice(encoded); decoded_footer.DecodeFrom(&encoded_slice); ASSERT_EQ(decoded_footer.table_magic_number(), kBlockBasedTableMagicNumber); ASSERT_EQ(decoded_footer.checksum(), kCRC32c); ASSERT_EQ(decoded_footer.metaindex_handle().offset(), meta_index.offset()); ASSERT_EQ(decoded_footer.metaindex_handle().size(), meta_index.size()); ASSERT_EQ(decoded_footer.index_handle().offset(), index.offset()); ASSERT_EQ(decoded_footer.index_handle().size(), index.size()); ASSERT_EQ(decoded_footer.version(), 0U); } { // xxhash block based std::string encoded; Footer footer(kBlockBasedTableMagicNumber, 1); BlockHandle meta_index(10, 5), index(20, 15); footer.set_metaindex_handle(meta_index); footer.set_index_handle(index); footer.set_checksum(kxxHash); footer.EncodeTo(&encoded); Footer decoded_footer; Slice encoded_slice(encoded); decoded_footer.DecodeFrom(&encoded_slice); ASSERT_EQ(decoded_footer.table_magic_number(), kBlockBasedTableMagicNumber); ASSERT_EQ(decoded_footer.checksum(), kxxHash); ASSERT_EQ(decoded_footer.metaindex_handle().offset(), meta_index.offset()); ASSERT_EQ(decoded_footer.metaindex_handle().size(), meta_index.size()); ASSERT_EQ(decoded_footer.index_handle().offset(), index.offset()); ASSERT_EQ(decoded_footer.index_handle().size(), index.size()); ASSERT_EQ(decoded_footer.version(), 1U); } // Plain table is not supported in ROCKSDB_LITE #ifndef ROCKSDB_LITE { // upconvert legacy plain table std::string encoded; Footer footer(kLegacyPlainTableMagicNumber, 0); BlockHandle meta_index(10, 5), index(20, 15); footer.set_metaindex_handle(meta_index); footer.set_index_handle(index); footer.EncodeTo(&encoded); Footer decoded_footer; Slice encoded_slice(encoded); decoded_footer.DecodeFrom(&encoded_slice); ASSERT_EQ(decoded_footer.table_magic_number(), kPlainTableMagicNumber); ASSERT_EQ(decoded_footer.checksum(), kCRC32c); ASSERT_EQ(decoded_footer.metaindex_handle().offset(), meta_index.offset()); ASSERT_EQ(decoded_footer.metaindex_handle().size(), meta_index.size()); ASSERT_EQ(decoded_footer.index_handle().offset(), index.offset()); ASSERT_EQ(decoded_footer.index_handle().size(), index.size()); ASSERT_EQ(decoded_footer.version(), 0U); } { // xxhash block based std::string encoded; Footer footer(kPlainTableMagicNumber, 1); BlockHandle meta_index(10, 5), index(20, 15); footer.set_metaindex_handle(meta_index); footer.set_index_handle(index); footer.set_checksum(kxxHash); footer.EncodeTo(&encoded); Footer decoded_footer; Slice encoded_slice(encoded); decoded_footer.DecodeFrom(&encoded_slice); ASSERT_EQ(decoded_footer.table_magic_number(), kPlainTableMagicNumber); ASSERT_EQ(decoded_footer.checksum(), kxxHash); ASSERT_EQ(decoded_footer.metaindex_handle().offset(), meta_index.offset()); ASSERT_EQ(decoded_footer.metaindex_handle().size(), meta_index.size()); ASSERT_EQ(decoded_footer.index_handle().offset(), index.offset()); ASSERT_EQ(decoded_footer.index_handle().size(), index.size()); ASSERT_EQ(decoded_footer.version(), 1U); } #endif // !ROCKSDB_LITE { // version == 2 std::string encoded; Footer footer(kBlockBasedTableMagicNumber, 2); BlockHandle meta_index(10, 5), index(20, 15); footer.set_metaindex_handle(meta_index); footer.set_index_handle(index); footer.EncodeTo(&encoded); Footer decoded_footer; Slice encoded_slice(encoded); decoded_footer.DecodeFrom(&encoded_slice); ASSERT_EQ(decoded_footer.table_magic_number(), kBlockBasedTableMagicNumber); ASSERT_EQ(decoded_footer.checksum(), kCRC32c); ASSERT_EQ(decoded_footer.metaindex_handle().offset(), meta_index.offset()); ASSERT_EQ(decoded_footer.metaindex_handle().size(), meta_index.size()); ASSERT_EQ(decoded_footer.index_handle().offset(), index.offset()); ASSERT_EQ(decoded_footer.index_handle().size(), index.size()); ASSERT_EQ(decoded_footer.version(), 2U); } } class IndexBlockRestartIntervalTest : public BlockBasedTableTest, public ::testing::WithParamInterface { public: static std::vector GetRestartValues() { return {-1, 0, 1, 8, 16, 32}; } }; INSTANTIATE_TEST_CASE_P( IndexBlockRestartIntervalTest, IndexBlockRestartIntervalTest, ::testing::ValuesIn(IndexBlockRestartIntervalTest::GetRestartValues())); TEST_P(IndexBlockRestartIntervalTest, IndexBlockRestartInterval) { const int kKeysInTable = 10000; const int kKeySize = 100; const int kValSize = 500; int index_block_restart_interval = GetParam(); Options options; BlockBasedTableOptions table_options; table_options.block_size = 64; // small block size to get big index block table_options.index_block_restart_interval = index_block_restart_interval; options.table_factory.reset(new BlockBasedTableFactory(table_options)); TableConstructor c(BytewiseComparator()); static Random rnd(301); for (int i = 0; i < kKeysInTable; i++) { InternalKey k(RandomString(&rnd, kKeySize), 0, kTypeValue); c.Add(k.Encode().ToString(), RandomString(&rnd, kValSize)); } std::vector keys; stl_wrappers::KVMap kvmap; std::unique_ptr comparator( new InternalKeyComparator(BytewiseComparator())); const ImmutableCFOptions ioptions(options); c.Finish(options, ioptions, table_options, *comparator, &keys, &kvmap); auto reader = c.GetTableReader(); std::unique_ptr db_iter(reader->NewIterator(ReadOptions())); // Test point lookup for (auto& kv : kvmap) { db_iter->Seek(kv.first); ASSERT_TRUE(db_iter->Valid()); ASSERT_OK(db_iter->status()); ASSERT_EQ(db_iter->key(), kv.first); ASSERT_EQ(db_iter->value(), kv.second); } // Test iterating auto kv_iter = kvmap.begin(); for (db_iter->SeekToFirst(); db_iter->Valid(); db_iter->Next()) { ASSERT_EQ(db_iter->key(), kv_iter->first); ASSERT_EQ(db_iter->value(), kv_iter->second); kv_iter++; } ASSERT_EQ(kv_iter, kvmap.end()); } } // namespace rocksdb int main(int argc, char** argv) { ::testing::InitGoogleTest(&argc, argv); return RUN_ALL_TESTS(); } rocksdb-4.5.1/table/two_level_iterator.cc000066400000000000000000000152061267530634000204710ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "table/two_level_iterator.h" #include "rocksdb/options.h" #include "rocksdb/table.h" #include "table/block.h" #include "table/format.h" #include "util/arena.h" namespace rocksdb { namespace { class TwoLevelIterator : public InternalIterator { public: explicit TwoLevelIterator(TwoLevelIteratorState* state, InternalIterator* first_level_iter, bool need_free_iter_and_state); virtual ~TwoLevelIterator() { first_level_iter_.DeleteIter(!need_free_iter_and_state_); second_level_iter_.DeleteIter(false); if (need_free_iter_and_state_) { delete state_; } else { state_->~TwoLevelIteratorState(); } } virtual void Seek(const Slice& target) override; virtual void SeekToFirst() override; virtual void SeekToLast() override; virtual void Next() override; virtual void Prev() override; virtual bool Valid() const override { return second_level_iter_.Valid(); } virtual Slice key() const override { assert(Valid()); return second_level_iter_.key(); } virtual Slice value() const override { assert(Valid()); return second_level_iter_.value(); } virtual Status status() const override { // It'd be nice if status() returned a const Status& instead of a Status if (!first_level_iter_.status().ok()) { return first_level_iter_.status(); } else if (second_level_iter_.iter() != nullptr && !second_level_iter_.status().ok()) { return second_level_iter_.status(); } else { return status_; } } virtual Status PinData() override { return second_level_iter_.PinData(); } virtual Status ReleasePinnedData() override { return second_level_iter_.ReleasePinnedData(); } virtual bool IsKeyPinned() const override { return second_level_iter_.iter() ? second_level_iter_.IsKeyPinned() : false; } private: void SaveError(const Status& s) { if (status_.ok() && !s.ok()) status_ = s; } void SkipEmptyDataBlocksForward(); void SkipEmptyDataBlocksBackward(); void SetSecondLevelIterator(InternalIterator* iter); void InitDataBlock(); TwoLevelIteratorState* state_; IteratorWrapper first_level_iter_; IteratorWrapper second_level_iter_; // May be nullptr bool need_free_iter_and_state_; Status status_; // If second_level_iter is non-nullptr, then "data_block_handle_" holds the // "index_value" passed to block_function_ to create the second_level_iter. std::string data_block_handle_; }; TwoLevelIterator::TwoLevelIterator(TwoLevelIteratorState* state, InternalIterator* first_level_iter, bool need_free_iter_and_state) : state_(state), first_level_iter_(first_level_iter), need_free_iter_and_state_(need_free_iter_and_state) {} void TwoLevelIterator::Seek(const Slice& target) { if (state_->check_prefix_may_match && !state_->PrefixMayMatch(target)) { SetSecondLevelIterator(nullptr); return; } first_level_iter_.Seek(target); InitDataBlock(); if (second_level_iter_.iter() != nullptr) { second_level_iter_.Seek(target); } SkipEmptyDataBlocksForward(); } void TwoLevelIterator::SeekToFirst() { first_level_iter_.SeekToFirst(); InitDataBlock(); if (second_level_iter_.iter() != nullptr) { second_level_iter_.SeekToFirst(); } SkipEmptyDataBlocksForward(); } void TwoLevelIterator::SeekToLast() { first_level_iter_.SeekToLast(); InitDataBlock(); if (second_level_iter_.iter() != nullptr) { second_level_iter_.SeekToLast(); } SkipEmptyDataBlocksBackward(); } void TwoLevelIterator::Next() { assert(Valid()); second_level_iter_.Next(); SkipEmptyDataBlocksForward(); } void TwoLevelIterator::Prev() { assert(Valid()); second_level_iter_.Prev(); SkipEmptyDataBlocksBackward(); } void TwoLevelIterator::SkipEmptyDataBlocksForward() { while (second_level_iter_.iter() == nullptr || (!second_level_iter_.Valid() && !second_level_iter_.status().IsIncomplete())) { // Move to next block if (!first_level_iter_.Valid()) { SetSecondLevelIterator(nullptr); return; } first_level_iter_.Next(); InitDataBlock(); if (second_level_iter_.iter() != nullptr) { second_level_iter_.SeekToFirst(); } } } void TwoLevelIterator::SkipEmptyDataBlocksBackward() { while (second_level_iter_.iter() == nullptr || (!second_level_iter_.Valid() && !second_level_iter_.status().IsIncomplete())) { // Move to next block if (!first_level_iter_.Valid()) { SetSecondLevelIterator(nullptr); return; } first_level_iter_.Prev(); InitDataBlock(); if (second_level_iter_.iter() != nullptr) { second_level_iter_.SeekToLast(); } } } void TwoLevelIterator::SetSecondLevelIterator(InternalIterator* iter) { if (second_level_iter_.iter() != nullptr) { SaveError(second_level_iter_.status()); } second_level_iter_.Set(iter); } void TwoLevelIterator::InitDataBlock() { if (!first_level_iter_.Valid()) { SetSecondLevelIterator(nullptr); } else { Slice handle = first_level_iter_.value(); if (second_level_iter_.iter() != nullptr && !second_level_iter_.status().IsIncomplete() && handle.compare(data_block_handle_) == 0) { // second_level_iter is already constructed with this iterator, so // no need to change anything } else { InternalIterator* iter = state_->NewSecondaryIterator(handle); data_block_handle_.assign(handle.data(), handle.size()); SetSecondLevelIterator(iter); } } } } // namespace InternalIterator* NewTwoLevelIterator(TwoLevelIteratorState* state, InternalIterator* first_level_iter, Arena* arena, bool need_free_iter_and_state) { if (arena == nullptr) { return new TwoLevelIterator(state, first_level_iter, need_free_iter_and_state); } else { auto mem = arena->AllocateAligned(sizeof(TwoLevelIterator)); return new (mem) TwoLevelIterator(state, first_level_iter, need_free_iter_and_state); } } } // namespace rocksdb rocksdb-4.5.1/table/two_level_iterator.h000066400000000000000000000041371267530634000203340ustar00rootroot00000000000000// Copyright (c) 2013, Facebook, Inc. All rights reserved. // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. An additional grant // of patent rights can be found in the PATENTS file in the same directory. // // Copyright (c) 2011 The LevelDB Authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. See the AUTHORS file for names of contributors. #pragma once #include "rocksdb/iterator.h" #include "rocksdb/env.h" #include "table/iterator_wrapper.h" namespace rocksdb { struct ReadOptions; class InternalKeyComparator; class Arena; struct TwoLevelIteratorState { explicit TwoLevelIteratorState(bool _check_prefix_may_match) : check_prefix_may_match(_check_prefix_may_match) {} virtual ~TwoLevelIteratorState() {} virtual InternalIterator* NewSecondaryIterator(const Slice& handle) = 0; virtual bool PrefixMayMatch(const Slice& internal_key) = 0; // If call PrefixMayMatch() bool check_prefix_may_match; }; // Return a new two level iterator. A two-level iterator contains an // index iterator whose values point to a sequence of blocks where // each block is itself a sequence of key,value pairs. The returned // two-level iterator yields the concatenation of all key/value pairs // in the sequence of blocks. Takes ownership of "index_iter" and // will delete it when no longer needed. // // Uses a supplied function to convert an index_iter value into // an iterator over the contents of the corresponding block. // arena: If not null, the arena is used to allocate the Iterator. // When destroying the iterator, the destructor will destroy // all the states but those allocated in arena. // need_free_iter_and_state: free `state` and `first_level_iter` if // true. Otherwise, just call destructor. extern InternalIterator* NewTwoLevelIterator( TwoLevelIteratorState* state, InternalIterator* first_level_iter, Arena* arena = nullptr, bool need_free_iter_and_state = true); } // namespace rocksdb rocksdb-4.5.1/third-party/000077500000000000000000000000001267530634000154255ustar00rootroot00000000000000rocksdb-4.5.1/third-party/fbson/000077500000000000000000000000001267530634000165345ustar00rootroot00000000000000rocksdb-4.5.1/third-party/fbson/COMMIT.md000066400000000000000000000005071267530634000200500ustar00rootroot00000000000000fbson commit: https://github.com/facebook/mysql-5.6/commit/55ef9ff25c934659a70b4094e9b406c48e9dd43d # TODO. * Had to convert zero sized array to [1] sized arrays due to the fact that MS Compiler complains about it not being standard. At some point need to contribute this change back to MySql where this code was taken from. rocksdb-4.5.1/third-party/fbson/FbsonDocument.h000066400000000000000000000563471267530634000214720ustar00rootroot00000000000000/* * Copyright (c) 2014, Facebook, Inc. * All rights reserved. * * This source code is licensed under the BSD-style license found in the * LICENSE file in the root directory of this source tree. An additional grant * of patent rights can be found in the PATENTS file in the same directory. * */ /* * This header defines FbsonDocument, FbsonKeyValue, and various value classes * which are derived from FbsonValue, and a forward iterator for container * values - essentially everything that is related to FBSON binary data * structures. * * Implementation notes: * * None of the classes in this header file can be instantiated directly (i.e. * you cannot create a FbsonKeyValue or FbsonValue object - all constructors * are declared non-public). We use the classes as wrappers on the packed FBSON * bytes (serialized), and cast the classes (types) to the underlying packed * byte array. * * For the same reason, we cannot define any FBSON value class to be virtual, * since we never call constructors, and will not instantiate vtbl and vptrs. * * Therefore, the classes are defined as packed structures (i.e. no data * alignment and padding), and the private member variables of the classes are * defined precisely in the same order as the FBSON spec. This ensures we * access the packed FBSON bytes correctly. * * The packed structures are highly optimized for in-place operations with low * overhead. The reads (and in-place writes) are performed directly on packed * bytes. There is no memory allocation at all at runtime. * * For updates/writes of values that will expand the original FBSON size, the * write will fail, and the caller needs to handle buffer increase. * * ** Iterator ** * Both ObjectVal class and ArrayVal class have iterator type that you can use * to declare an iterator on a container object to go through the key-value * pairs or value list. The iterator has both non-const and const types. * * Note: iterators are forward direction only. * * ** Query ** * Querying into containers is through the member functions find (for key/value * pairs) and get (for array elements), and is in streaming style. We don't * need to read/scan the whole FBSON packed bytes in order to return results. * Once the key/index is found, we will stop search. You can use text to query * both objects and array (for array, text will be converted to integer index), * and use index to retrieve from array. Array index is 0-based. * * ** External dictionary ** * During query processing, you can also pass a call-back function, so the * search will first try to check if the key string exists in the dictionary. * If so, search will be based on the id instead of the key string. * * @author Tian Xia */ #ifndef FBSON_FBSONDOCUMENT_H #define FBSON_FBSONDOCUMENT_H #include #include #include namespace fbson { #pragma pack(push, 1) #define FBSON_VER 1 // forward declaration class FbsonValue; class ObjectVal; /* * FbsonDocument is the main object that accesses and queries FBSON packed * bytes. NOTE: FbsonDocument only allows object container as the top level * FBSON value. However, you can use the static method "createValue" to get any * FbsonValue object from the packed bytes. * * FbsonDocument object also dereferences to an object container value * (ObjectVal) once FBSON is loaded. * * ** Load ** * FbsonDocument is usable after loading packed bytes (memory location) into * the object. We only need the header and first few bytes of the payload after * header to verify the FBSON. * * Note: creating an FbsonDocument (through createDocument) does not allocate * any memory. The document object is an efficient wrapper on the packed bytes * which is accessed directly. * * ** Query ** * Query is through dereferencing into ObjectVal. */ class FbsonDocument { public: // create an FbsonDocument object from FBSON packed bytes static FbsonDocument* createDocument(const char* pb, uint32_t size); // create an FbsonValue from FBSON packed bytes static FbsonValue* createValue(const char* pb, uint32_t size); uint8_t version() { return header_.ver_; } FbsonValue* getValue() { return ((FbsonValue*)payload_); } ObjectVal* operator->() { return ((ObjectVal*)payload_); } const ObjectVal* operator->() const { return ((const ObjectVal*)payload_); } private: /* * FbsonHeader class defines FBSON header (internal to FbsonDocument). * * Currently it only contains version information (1-byte). We may expand the * header to include checksum of the FBSON binary for more security. */ struct FbsonHeader { uint8_t ver_; } header_; char payload_[1]; FbsonDocument(); FbsonDocument(const FbsonDocument&) = delete; FbsonDocument& operator=(const FbsonDocument&) = delete; }; /* * FbsonFwdIteratorT implements FBSON's iterator template. * * Note: it is an FORWARD iterator only due to the design of FBSON format. */ template class FbsonFwdIteratorT { typedef Iter_Type iterator; typedef typename std::iterator_traits::pointer pointer; typedef typename std::iterator_traits::reference reference; public: explicit FbsonFwdIteratorT(const iterator& i) : current_(i) {} // allow non-const to const iterator conversion (same container type) template FbsonFwdIteratorT(const FbsonFwdIteratorT& rhs) : current_(rhs.base()) {} bool operator==(const FbsonFwdIteratorT& rhs) const { return (current_ == rhs.current_); } bool operator!=(const FbsonFwdIteratorT& rhs) const { return !operator==(rhs); } bool operator<(const FbsonFwdIteratorT& rhs) const { return (current_ < rhs.current_); } bool operator>(const FbsonFwdIteratorT& rhs) const { return !operator<(rhs); } FbsonFwdIteratorT& operator++() { current_ = (iterator)(((char*)current_) + current_->numPackedBytes()); return *this; } FbsonFwdIteratorT operator++(int) { auto tmp = *this; current_ = (iterator)(((char*)current_) + current_->numPackedBytes()); return tmp; } explicit operator pointer() { return current_; } reference operator*() const { return *current_; } pointer operator->() const { return current_; } iterator base() const { return current_; } private: iterator current_; }; typedef int (*hDictInsert)(const char* key, unsigned len); typedef int (*hDictFind)(const char* key, unsigned len); /* * FbsonType defines 10 primitive types and 2 container types, as described * below. * * primitive_value ::= * 0x00 //null value (0 byte) * | 0x01 //boolean true (0 byte) * | 0x02 //boolean false (0 byte) * | 0x03 int8 //char/int8 (1 byte) * | 0x04 int16 //int16 (2 bytes) * | 0x05 int32 //int32 (4 bytes) * | 0x06 int64 //int64 (8 bytes) * | 0x07 double //floating point (8 bytes) * | 0x08 string //variable length string * | 0x09 binary //variable length binary * * container ::= * 0x0A int32 key_value_list //object, int32 is the total bytes of the object * | 0x0B int32 value_list //array, int32 is the total bytes of the array */ enum class FbsonType : char { T_Null = 0x00, T_True = 0x01, T_False = 0x02, T_Int8 = 0x03, T_Int16 = 0x04, T_Int32 = 0x05, T_Int64 = 0x06, T_Double = 0x07, T_String = 0x08, T_Binary = 0x09, T_Object = 0x0A, T_Array = 0x0B, NUM_TYPES, }; typedef std::underlying_type::type FbsonTypeUnder; /* * FbsonKeyValue class defines FBSON key type, as described below. * * key ::= * 0x00 int8 //1-byte dictionary id * | int8 (byte*) //int8 (>0) is the size of the key string * * value ::= primitive_value | container * * FbsonKeyValue can be either an id mapping to the key string in an external * dictionary, or it is the original key string. Whether to read an id or a * string is decided by the first byte (size_). * * Note: a key object must be followed by a value object. Therefore, a key * object implicitly refers to a key-value pair, and you can get the value * object right after the key object. The function numPackedBytes hence * indicates the total size of the key-value pair, so that we will be able go * to next pair from the key. * * ** Dictionary size ** * By default, the dictionary size is 255 (1-byte). Users can define * "USE_LARGE_DICT" to increase the dictionary size to 655535 (2-byte). */ class FbsonKeyValue { public: #ifdef USE_LARGE_DICT static const int sMaxKeyId = 65535; typedef uint16_t keyid_type; #else static const int sMaxKeyId = 255; typedef uint8_t keyid_type; #endif // #ifdef USE_LARGE_DICT static const uint8_t sMaxKeyLen = 64; // size of the key. 0 indicates it is stored as id uint8_t klen() const { return size_; } // get the key string. Note the string may not be null terminated. const char* getKeyStr() const { return key_.str_; } keyid_type getKeyId() const { return key_.id_; } unsigned int keyPackedBytes() const { return size_ ? (sizeof(size_) + size_) : (sizeof(size_) + sizeof(keyid_type)); } FbsonValue* value() const { return (FbsonValue*)(((char*)this) + keyPackedBytes()); } // size of the total packed bytes (key+value) unsigned int numPackedBytes() const; private: uint8_t size_; union key_ { keyid_type id_; char str_[1]; } key_; FbsonKeyValue(); }; /* * FbsonValue is the base class of all FBSON types. It contains only one member * variable - type info, which can be retrieved by member functions is[Type]() * or type(). */ class FbsonValue { public: static const uint32_t sMaxValueLen = 1 << 24; // 16M bool isNull() const { return (type_ == FbsonType::T_Null); } bool isTrue() const { return (type_ == FbsonType::T_True); } bool isFalse() const { return (type_ == FbsonType::T_False); } bool isInt8() const { return (type_ == FbsonType::T_Int8); } bool isInt16() const { return (type_ == FbsonType::T_Int16); } bool isInt32() const { return (type_ == FbsonType::T_Int32); } bool isInt64() const { return (type_ == FbsonType::T_Int64); } bool isDouble() const { return (type_ == FbsonType::T_Double); } bool isString() const { return (type_ == FbsonType::T_String); } bool isBinary() const { return (type_ == FbsonType::T_Binary); } bool isObject() const { return (type_ == FbsonType::T_Object); } bool isArray() const { return (type_ == FbsonType::T_Array); } FbsonType type() const { return type_; } // size of the total packed bytes unsigned int numPackedBytes() const; // size of the value in bytes unsigned int size() const; // get the raw byte array of the value const char* getValuePtr() const; // find the FBSON value by a key path string (null terminated) FbsonValue* findPath(const char* key_path, const char* delim = ".", hDictFind handler = nullptr) { return findPath(key_path, (unsigned int)strlen(key_path), delim, handler); } // find the FBSON value by a key path string (with length) FbsonValue* findPath(const char* key_path, unsigned int len, const char* delim, hDictFind handler); protected: FbsonType type_; // type info FbsonValue(); }; /* * NumerValT is the template class (derived from FbsonValue) of all number * types (integers and double). */ template class NumberValT : public FbsonValue { public: T val() const { return num_; } unsigned int numPackedBytes() const { return sizeof(FbsonValue) + sizeof(T); } // catch all unknow specialization of the template class bool setVal(T value) { return false; } private: T num_; NumberValT(); }; typedef NumberValT Int8Val; // override setVal for Int8Val template <> inline bool Int8Val::setVal(int8_t value) { if (!isInt8()) { return false; } num_ = value; return true; } typedef NumberValT Int16Val; // override setVal for Int16Val template <> inline bool Int16Val::setVal(int16_t value) { if (!isInt16()) { return false; } num_ = value; return true; } typedef NumberValT Int32Val; // override setVal for Int32Val template <> inline bool Int32Val::setVal(int32_t value) { if (!isInt32()) { return false; } num_ = value; return true; } typedef NumberValT Int64Val; // override setVal for Int64Val template <> inline bool Int64Val::setVal(int64_t value) { if (!isInt64()) { return false; } num_ = value; return true; } typedef NumberValT DoubleVal; // override setVal for DoubleVal template <> inline bool DoubleVal::setVal(double value) { if (!isDouble()) { return false; } num_ = value; return true; } /* * BlobVal is the base class (derived from FbsonValue) for string and binary * types. The size_ indicates the total bytes of the payload_. */ class BlobVal : public FbsonValue { public: // size of the blob payload only unsigned int getBlobLen() const { return size_; } // return the blob as byte array const char* getBlob() const { return payload_; } // size of the total packed bytes unsigned int numPackedBytes() const { return sizeof(FbsonValue) + sizeof(size_) + size_; } protected: uint32_t size_; char payload_[1]; // set new blob bytes bool internalSetVal(const char* blob, uint32_t blobSize) { // if we cannot fit the new blob, fail the operation if (blobSize > size_) { return false; } memcpy(payload_, blob, blobSize); // Set the reset of the bytes to 0. Note we cannot change the size_ of the // current payload, as all values are packed. memset(payload_ + blobSize, 0, size_ - blobSize); return true; } BlobVal(); private: // Disable as this class can only be allocated dynamically BlobVal(const BlobVal&) = delete; BlobVal& operator=(const BlobVal&) = delete; }; /* * Binary type */ class BinaryVal : public BlobVal { public: bool setVal(const char* blob, uint32_t blobSize) { if (!isBinary()) { return false; } return internalSetVal(blob, blobSize); } private: BinaryVal(); }; /* * String type * Note: FBSON string may not be a c-string (NULL-terminated) */ class StringVal : public BlobVal { public: bool setVal(const char* str, uint32_t blobSize) { if (!isString()) { return false; } return internalSetVal(str, blobSize); } private: StringVal(); }; /* * ContainerVal is the base class (derived from FbsonValue) for object and * array types. The size_ indicates the total bytes of the payload_. */ class ContainerVal : public FbsonValue { public: // size of the container payload only unsigned int getContainerSize() const { return size_; } // return the container payload as byte array const char* getPayload() const { return payload_; } // size of the total packed bytes unsigned int numPackedBytes() const { return sizeof(FbsonValue) + sizeof(size_) + size_; } protected: uint32_t size_; char payload_[1]; ContainerVal(); ContainerVal(const ContainerVal&) = delete; ContainerVal& operator=(const ContainerVal&) = delete; }; /* * Object type */ class ObjectVal : public ContainerVal { public: // find the FBSON value by a key string (null terminated) FbsonValue* find(const char* key, hDictFind handler = nullptr) const { if (!key) return nullptr; return find(key, (unsigned int)strlen(key), handler); } // find the FBSON value by a key string (with length) FbsonValue* find(const char* key, unsigned int klen, hDictFind handler = nullptr) const { if (!key || !klen) return nullptr; int key_id = -1; if (handler && (key_id = handler(key, klen)) >= 0) { return find(key_id); } return internalFind(key, klen); } // find the FBSON value by a key dictionary ID FbsonValue* find(int key_id) const { if (key_id < 0 || key_id > FbsonKeyValue::sMaxKeyId) return nullptr; const char* pch = payload_; const char* fence = payload_ + size_; while (pch < fence) { FbsonKeyValue* pkey = (FbsonKeyValue*)(pch); if (!pkey->klen() && key_id == pkey->getKeyId()) { return pkey->value(); } pch += pkey->numPackedBytes(); } assert(pch == fence); return nullptr; } typedef FbsonKeyValue value_type; typedef value_type* pointer; typedef const value_type* const_pointer; typedef FbsonFwdIteratorT iterator; typedef FbsonFwdIteratorT const_iterator; iterator begin() { return iterator((pointer)payload_); } const_iterator begin() const { return const_iterator((pointer)payload_); } iterator end() { return iterator((pointer)(payload_ + size_)); } const_iterator end() const { return const_iterator((pointer)(payload_ + size_)); } private: FbsonValue* internalFind(const char* key, unsigned int klen) const { const char* pch = payload_; const char* fence = payload_ + size_; while (pch < fence) { FbsonKeyValue* pkey = (FbsonKeyValue*)(pch); if (klen == pkey->klen() && strncmp(key, pkey->getKeyStr(), klen) == 0) { return pkey->value(); } pch += pkey->numPackedBytes(); } assert(pch == fence); return nullptr; } private: ObjectVal(); }; /* * Array type */ class ArrayVal : public ContainerVal { public: // get the FBSON value at index FbsonValue* get(int idx) const { if (idx < 0) return nullptr; const char* pch = payload_; const char* fence = payload_ + size_; while (pch < fence && idx-- > 0) pch += ((FbsonValue*)pch)->numPackedBytes(); if (idx == -1) return (FbsonValue*)pch; else { assert(pch == fence); return nullptr; } } // Get number of elements in array unsigned int numElem() const { const char* pch = payload_; const char* fence = payload_ + size_; unsigned int num = 0; while (pch < fence) { ++num; pch += ((FbsonValue*)pch)->numPackedBytes(); } assert(pch == fence); return num; } typedef FbsonValue value_type; typedef value_type* pointer; typedef const value_type* const_pointer; typedef FbsonFwdIteratorT iterator; typedef FbsonFwdIteratorT const_iterator; iterator begin() { return iterator((pointer)payload_); } const_iterator begin() const { return const_iterator((pointer)payload_); } iterator end() { return iterator((pointer)(payload_ + size_)); } const_iterator end() const { return const_iterator((pointer)(payload_ + size_)); } private: ArrayVal(); }; inline FbsonDocument* FbsonDocument::createDocument(const char* pb, uint32_t size) { if (!pb || size < sizeof(FbsonHeader) + sizeof(FbsonValue)) { return nullptr; } FbsonDocument* doc = (FbsonDocument*)pb; if (doc->header_.ver_ != FBSON_VER) { return nullptr; } FbsonValue* val = (FbsonValue*)doc->payload_; if (!val->isObject() || size != sizeof(FbsonHeader) + val->numPackedBytes()) { return nullptr; } return doc; } inline FbsonValue* FbsonDocument::createValue(const char* pb, uint32_t size) { if (!pb || size < sizeof(FbsonHeader) + sizeof(FbsonValue)) { return nullptr; } FbsonDocument* doc = (FbsonDocument*)pb; if (doc->header_.ver_ != FBSON_VER) { return nullptr; } FbsonValue* val = (FbsonValue*)doc->payload_; if (size != sizeof(FbsonHeader) + val->numPackedBytes()) { return nullptr; } return val; } inline unsigned int FbsonKeyValue::numPackedBytes() const { unsigned int ks = keyPackedBytes(); FbsonValue* val = (FbsonValue*)(((char*)this) + ks); return ks + val->numPackedBytes(); } // Poor man's "virtual" function FbsonValue::numPackedBytes inline unsigned int FbsonValue::numPackedBytes() const { switch (type_) { case FbsonType::T_Null: case FbsonType::T_True: case FbsonType::T_False: { return sizeof(type_); } case FbsonType::T_Int8: { return sizeof(type_) + sizeof(int8_t); } case FbsonType::T_Int16: { return sizeof(type_) + sizeof(int16_t); } case FbsonType::T_Int32: { return sizeof(type_) + sizeof(int32_t); } case FbsonType::T_Int64: { return sizeof(type_) + sizeof(int64_t); } case FbsonType::T_Double: { return sizeof(type_) + sizeof(double); } case FbsonType::T_String: case FbsonType::T_Binary: { return ((BlobVal*)(this))->numPackedBytes(); } case FbsonType::T_Object: case FbsonType::T_Array: { return ((ContainerVal*)(this))->numPackedBytes(); } default: return 0; } } inline unsigned int FbsonValue::size() const { switch (type_) { case FbsonType::T_Int8: { return sizeof(int8_t); } case FbsonType::T_Int16: { return sizeof(int16_t); } case FbsonType::T_Int32: { return sizeof(int32_t); } case FbsonType::T_Int64: { return sizeof(int64_t); } case FbsonType::T_Double: { return sizeof(double); } case FbsonType::T_String: case FbsonType::T_Binary: { return ((BlobVal*)(this))->getBlobLen(); } case FbsonType::T_Object: case FbsonType::T_Array: { return ((ContainerVal*)(this))->getContainerSize(); } case FbsonType::T_Null: case FbsonType::T_True: case FbsonType::T_False: default: return 0; } } inline const char* FbsonValue::getValuePtr() const { switch (type_) { case FbsonType::T_Int8: case FbsonType::T_Int16: case FbsonType::T_Int32: case FbsonType::T_Int64: case FbsonType::T_Double: return ((char*)this) + sizeof(FbsonType); case FbsonType::T_String: case FbsonType::T_Binary: return ((BlobVal*)(this))->getBlob(); case FbsonType::T_Object: case FbsonType::T_Array: return ((ContainerVal*)(this))->getPayload(); case FbsonType::T_Null: case FbsonType::T_True: case FbsonType::T_False: default: return nullptr; } } inline FbsonValue* FbsonValue::findPath(const char* key_path, unsigned int kp_len, const char* delim = ".", hDictFind handler = nullptr) { if (!key_path || !kp_len) return nullptr; if (!delim) delim = "."; // default delimiter FbsonValue* pval = this; const char* fence = key_path + kp_len; char idx_buf[21]; // buffer to parse array index (integer value) while (pval && key_path < fence) { const char* key = key_path; unsigned int klen = 0; // find the current key for (; key_path != fence && *key_path != *delim; ++key_path, ++klen) ; if (!klen) return nullptr; switch (pval->type_) { case FbsonType::T_Object: { pval = ((ObjectVal*)pval)->find(key, klen, handler); break; } case FbsonType::T_Array: { // parse string into an integer (array index) if (klen >= sizeof(idx_buf)) return nullptr; memcpy(idx_buf, key, klen); idx_buf[klen] = 0; char* end = nullptr; int index = (int)strtol(idx_buf, &end, 10); if (end && !*end) pval = ((fbson::ArrayVal*)pval)->get(index); else // incorrect index string return nullptr; break; } default: return nullptr; } // skip the delimiter if (key_path < fence) { ++key_path; if (key_path == fence) // we have a trailing delimiter at the end return nullptr; } } return pval; } #pragma pack(pop) } // namespace fbson #endif // FBSON_FBSONDOCUMENT_H rocksdb-4.5.1/third-party/fbson/FbsonJsonParser.h000066400000000000000000000432541267530634000217730ustar00rootroot00000000000000/* * Copyright (c) 2014, Facebook, Inc. * All rights reserved. * * This source code is licensed under the BSD-style license found in the * LICENSE file in the root directory of this source tree. An additional grant * of patent rights can be found in the PATENTS file in the same directory. * */ /* * This file defines FbsonJsonParserT (template) and FbsonJsonParser. * * FbsonJsonParserT is a template class which implements a JSON parser. * FbsonJsonParserT parses JSON text, and serialize it to FBSON binary format * by using FbsonWriterT object. By default, FbsonJsonParserT creates a new * FbsonWriterT object with an output stream object. However, you can also * pass in your FbsonWriterT or any stream object that implements some basic * interface of std::ostream (see FbsonStream.h). * * FbsonJsonParser specializes FbsonJsonParserT with FbsonOutStream type (see * FbsonStream.h). So unless you want to provide own a different output stream * type, use FbsonJsonParser object. * * ** Parsing JSON ** * FbsonJsonParserT parses JSON string, and directly serializes into FBSON * packed bytes. There are three ways to parse a JSON string: (1) using * c-string, (2) using string with len, (3) using std::istream object. You can * use custome streambuf to redirect output. FbsonOutBuffer is a streambuf used * internally if the input is raw character buffer. * * You can reuse an FbsonJsonParserT object to parse/serialize multiple JSON * strings, and the previous FBSON will be overwritten. * * If parsing fails (returned false), the error code will be set to one of * FbsonErrType, and can be retrieved by calling getErrorCode(). * * ** External dictionary ** * During parsing a JSON string, you can pass a call-back function to map a key * string to an id, and store the dictionary id in FBSON to save space. The * purpose of using an external dictionary is more towards a collection of * documents (which has common keys) rather than a single document, so that * space saving will be siginificant. * * ** Endianness ** * Note: FBSON serialization doesn't assume endianness of the server. However * you will need to ensure that the endianness at the reader side is the same * as that at the writer side (if they are on different machines). Otherwise, * proper conversion is needed when a number value is returned to the * caller/writer. * * @author Tian Xia */ #ifndef FBSON_FBSONPARSER_H #define FBSON_FBSONPARSER_H #include #include #include "FbsonDocument.h" #include "FbsonWriter.h" namespace fbson { const char* const kJsonDelim = " ,]}\t\r\n"; const char* const kWhiteSpace = " \t\n\r"; /* * Error codes */ enum class FbsonErrType { E_NONE = 0, E_INVALID_VER, E_EMPTY_STR, E_OUTPUT_FAIL, E_INVALID_DOCU, E_INVALID_VALUE, E_INVALID_KEY, E_INVALID_STR, E_INVALID_OBJ, E_INVALID_ARR, E_INVALID_HEX, E_INVALID_OCTAL, E_INVALID_DECIMAL, E_INVALID_EXPONENT, E_HEX_OVERFLOW, E_OCTAL_OVERFLOW, E_DECIMAL_OVERFLOW, E_DOUBLE_OVERFLOW, E_EXPONENT_OVERFLOW, }; /* * Template FbsonJsonParserT */ template class FbsonJsonParserT { public: FbsonJsonParserT() : err_(FbsonErrType::E_NONE) {} explicit FbsonJsonParserT(OS_TYPE& os) : writer_(os), err_(FbsonErrType::E_NONE) {} // parse a UTF-8 JSON string bool parse(const std::string& str, hDictInsert handler = nullptr) { return parse(str.c_str(), (unsigned int)str.size(), handler); } // parse a UTF-8 JSON c-style string (NULL terminated) bool parse(const char* c_str, hDictInsert handler = nullptr) { return parse(c_str, (unsigned int)strlen(c_str), handler); } // parse a UTF-8 JSON string with length bool parse(const char* pch, unsigned int len, hDictInsert handler = nullptr) { if (!pch || len == 0) { err_ = FbsonErrType::E_EMPTY_STR; return false; } FbsonInBuffer sb(pch, len); std::istream in(&sb); return parse(in, handler); } // parse UTF-8 JSON text from an input stream bool parse(std::istream& in, hDictInsert handler = nullptr) { bool res = false; // reset output stream writer_.reset(); trim(in); if (in.peek() == '{') { in.ignore(); res = parseObject(in, handler); } else if (in.peek() == '[') { in.ignore(); res = parseArray(in, handler); } else { err_ = FbsonErrType::E_INVALID_DOCU; } trim(in); if (res && !in.eof()) { err_ = FbsonErrType::E_INVALID_DOCU; return false; } return res; } FbsonWriterT& getWriter() { return writer_; } FbsonErrType getErrorCode() { return err_; } // clear error code void clearErr() { err_ = FbsonErrType::E_NONE; } private: // parse a JSON object (comma-separated list of key-value pairs) bool parseObject(std::istream& in, hDictInsert handler) { if (!writer_.writeStartObject()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } trim(in); if (in.peek() == '}') { in.ignore(); // empty object if (!writer_.writeEndObject()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } while (in.good()) { if (in.get() != '"') { err_ = FbsonErrType::E_INVALID_KEY; return false; } if (!parseKVPair(in, handler)) { return false; } trim(in); char ch = in.get(); if (ch == '}') { // end of the object if (!writer_.writeEndObject()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } else if (ch != ',') { err_ = FbsonErrType::E_INVALID_OBJ; return false; } trim(in); } err_ = FbsonErrType::E_INVALID_OBJ; return false; } // parse a JSON array (comma-separated list of values) bool parseArray(std::istream& in, hDictInsert handler) { if (!writer_.writeStartArray()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } trim(in); if (in.peek() == ']') { in.ignore(); // empty array if (!writer_.writeEndArray()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } while (in.good()) { if (!parseValue(in, handler)) { return false; } trim(in); char ch = in.get(); if (ch == ']') { // end of the array if (!writer_.writeEndArray()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } else if (ch != ',') { err_ = FbsonErrType::E_INVALID_ARR; return false; } trim(in); } err_ = FbsonErrType::E_INVALID_ARR; return false; } // parse a key-value pair, separated by ":" bool parseKVPair(std::istream& in, hDictInsert handler) { if (parseKey(in, handler) && parseValue(in, handler)) { return true; } return false; } // parse a key (must be string) bool parseKey(std::istream& in, hDictInsert handler) { char key[FbsonKeyValue::sMaxKeyLen]; int i = 0; while (in.good() && in.peek() != '"' && i < FbsonKeyValue::sMaxKeyLen) { key[i++] = in.get(); } if (!in.good() || in.peek() != '"' || i == 0) { err_ = FbsonErrType::E_INVALID_KEY; return false; } in.ignore(); // discard '"' int key_id = -1; if (handler) { key_id = handler(key, i); } if (key_id < 0) { writer_.writeKey(key, i); } else { writer_.writeKey(key_id); } trim(in); if (in.get() != ':') { err_ = FbsonErrType::E_INVALID_OBJ; return false; } return true; } // parse a value bool parseValue(std::istream& in, hDictInsert handler) { bool res = false; trim(in); switch (in.peek()) { case 'N': case 'n': { in.ignore(); res = parseNull(in); break; } case 'T': case 't': { in.ignore(); res = parseTrue(in); break; } case 'F': case 'f': { in.ignore(); res = parseFalse(in); break; } case '"': { in.ignore(); res = parseString(in); break; } case '{': { in.ignore(); res = parseObject(in, handler); break; } case '[': { in.ignore(); res = parseArray(in, handler); break; } default: { res = parseNumber(in); break; } } return res; } // parse NULL value bool parseNull(std::istream& in) { if (tolower(in.get()) == 'u' && tolower(in.get()) == 'l' && tolower(in.get()) == 'l') { writer_.writeNull(); return true; } err_ = FbsonErrType::E_INVALID_VALUE; return false; } // parse TRUE value bool parseTrue(std::istream& in) { if (tolower(in.get()) == 'r' && tolower(in.get()) == 'u' && tolower(in.get()) == 'e') { writer_.writeBool(true); return true; } err_ = FbsonErrType::E_INVALID_VALUE; return false; } // parse FALSE value bool parseFalse(std::istream& in) { if (tolower(in.get()) == 'a' && tolower(in.get()) == 'l' && tolower(in.get()) == 's' && tolower(in.get()) == 'e') { writer_.writeBool(false); return true; } err_ = FbsonErrType::E_INVALID_VALUE; return false; } // parse a string bool parseString(std::istream& in) { if (!writer_.writeStartString()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } bool escaped = false; char buffer[4096]; // write 4KB at a time int nread = 0; while (in.good()) { char ch = in.get(); if (ch != '"' || escaped) { buffer[nread++] = ch; if (nread == 4096) { // flush buffer if (!writer_.writeString(buffer, nread)) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } nread = 0; } // set/reset escape if (ch == '\\' || escaped) { escaped = !escaped; } } else { // write all remaining bytes in the buffer if (nread > 0) { if (!writer_.writeString(buffer, nread)) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } } // end writing string if (!writer_.writeEndString()) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } } err_ = FbsonErrType::E_INVALID_STR; return false; } // parse a number // Number format can be hex, octal, or decimal (including float). // Only decimal can have (+/-) sign prefix. bool parseNumber(std::istream& in) { bool ret = false; switch (in.peek()) { case '0': { in.ignore(); if (in.peek() == 'x' || in.peek() == 'X') { in.ignore(); ret = parseHex(in); } else if (in.peek() == '.') { in.ignore(); ret = parseDouble(in, 0, 0, 1); } else { ret = parseOctal(in); } break; } case '-': { in.ignore(); ret = parseDecimal(in, -1); break; } case '+': in.ignore(); // fall through default: ret = parseDecimal(in, 1); break; } return ret; } // parse a number in hex format bool parseHex(std::istream& in) { uint64_t val = 0; int num_digits = 0; char ch = tolower(in.peek()); while (in.good() && !strchr(kJsonDelim, ch) && (++num_digits) <= 16) { if (ch >= '0' && ch <= '9') { val = (val << 4) + (ch - '0'); } else if (ch >= 'a' && ch <= 'f') { val = (val << 4) + (ch - 'a' + 10); } else { // unrecognized hex digit err_ = FbsonErrType::E_INVALID_HEX; return false; } in.ignore(); ch = tolower(in.peek()); } int size = 0; if (num_digits <= 2) { size = writer_.writeInt8((int8_t)val); } else if (num_digits <= 4) { size = writer_.writeInt16((int16_t)val); } else if (num_digits <= 8) { size = writer_.writeInt32((int32_t)val); } else if (num_digits <= 16) { size = writer_.writeInt64(val); } else { err_ = FbsonErrType::E_HEX_OVERFLOW; return false; } if (size == 0) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } // parse a number in octal format bool parseOctal(std::istream& in) { int64_t val = 0; char ch = in.peek(); while (in.good() && !strchr(kJsonDelim, ch)) { if (ch >= '0' && ch <= '7') { val = val * 8 + (ch - '0'); } else { err_ = FbsonErrType::E_INVALID_OCTAL; return false; } // check if the number overflows if (val < 0) { err_ = FbsonErrType::E_OCTAL_OVERFLOW; return false; } in.ignore(); ch = in.peek(); } int size = 0; if (val <= std::numeric_limits::max()) { size = writer_.writeInt8((int8_t)val); } else if (val <= std::numeric_limits::max()) { size = writer_.writeInt16((int16_t)val); } else if (val <= std::numeric_limits::max()) { size = writer_.writeInt32((int32_t)val); } else { // val <= INT64_MAX size = writer_.writeInt64(val); } if (size == 0) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } // parse a number in decimal (including float) bool parseDecimal(std::istream& in, int sign) { int64_t val = 0; int precision = 0; char ch = 0; while (in.good() && (ch = in.peek()) == '0') in.ignore(); while (in.good() && !strchr(kJsonDelim, ch)) { if (ch >= '0' && ch <= '9') { val = val * 10 + (ch - '0'); ++precision; } else if (ch == '.') { // note we don't pop out '.' return parseDouble(in, static_cast(val), precision, sign); } else { err_ = FbsonErrType::E_INVALID_DECIMAL; return false; } in.ignore(); // if the number overflows int64_t, first parse it as double iff we see a // decimal point later. Otherwise, will treat it as overflow if (val < 0 && val > std::numeric_limits::min()) { return parseDouble(in, static_cast(val), precision, sign); } ch = in.peek(); } if (sign < 0) { val = -val; } int size = 0; if (val >= std::numeric_limits::min() && val <= std::numeric_limits::max()) { size = writer_.writeInt8((int8_t)val); } else if (val >= std::numeric_limits::min() && val <= std::numeric_limits::max()) { size = writer_.writeInt16((int16_t)val); } else if (val >= std::numeric_limits::min() && val <= std::numeric_limits::max()) { size = writer_.writeInt32((int32_t)val); } else { // val <= INT64_MAX size = writer_.writeInt64(val); } if (size == 0) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } // parse IEEE745 double precision: // Significand precision length - 15 // Maximum exponent value - 308 // // "If a decimal string with at most 15 significant digits is converted to // IEEE 754 double precision representation and then converted back to a // string with the same number of significant digits, then the final string // should match the original" bool parseDouble(std::istream& in, double val, int precision, int sign) { int integ = precision; int frac = 0; bool is_frac = false; char ch = in.peek(); if (ch == '.') { is_frac = true; in.ignore(); ch = in.peek(); } int exp = 0; while (in.good() && !strchr(kJsonDelim, ch)) { if (ch >= '0' && ch <= '9') { if (precision < 15) { val = val * 10 + (ch - '0'); if (is_frac) { ++frac; } else { ++integ; } ++precision; } else if (!is_frac) { ++exp; } } else if (ch == 'e' || ch == 'E') { in.ignore(); int exp2; if (!parseExponent(in, exp2)) { return false; } exp += exp2; // check if exponent overflows if (exp > 308 || exp < -308) { err_ = FbsonErrType::E_EXPONENT_OVERFLOW; return false; } is_frac = true; break; } in.ignore(); ch = in.peek(); } if (!is_frac) { err_ = FbsonErrType::E_DECIMAL_OVERFLOW; return false; } val *= std::pow(10, exp - frac); if (std::isnan(val) || std::isinf(val)) { err_ = FbsonErrType::E_DOUBLE_OVERFLOW; return false; } if (sign < 0) { val = -val; } if (writer_.writeDouble(val) == 0) { err_ = FbsonErrType::E_OUTPUT_FAIL; return false; } return true; } // parse the exponent part of a double number bool parseExponent(std::istream& in, int& exp) { bool neg = false; char ch = in.peek(); if (ch == '+') { in.ignore(); ch = in.peek(); } else if (ch == '-') { neg = true; in.ignore(); ch = in.peek(); } exp = 0; while (in.good() && !strchr(kJsonDelim, ch)) { if (ch >= '0' && ch <= '9') { exp = exp * 10 + (ch - '0'); } else { err_ = FbsonErrType::E_INVALID_EXPONENT; return false; } if (exp > 308) { err_ = FbsonErrType::E_EXPONENT_OVERFLOW; return false; } in.ignore(); ch = in.peek(); } if (neg) { exp = -exp; } return true; } void trim(std::istream& in) { while (in.good() && strchr(kWhiteSpace, in.peek())) { in.ignore(); } } private: FbsonWriterT writer_; FbsonErrType err_; }; typedef FbsonJsonParserT FbsonJsonParser; } // namespace fbson #endif // FBSON_FBSONPARSER_H rocksdb-4.5.1/third-party/fbson/FbsonStream.h000066400000000000000000000106521267530634000211340ustar00rootroot00000000000000/* * Copyright (c) 2014, Facebook, Inc. * All rights reserved. * * This source code is licensed under the BSD-style license found in the * LICENSE file in the root directory of this source tree. An additional grant * of patent rights can be found in the PATENTS file in the same directory. * */ /* * This header file defines FbsonInBuffer and FbsonOutStream classes. * * ** Input Buffer ** * FbsonInBuffer is a customer input buffer to wrap raw character buffer. Its * object instances are used to create std::istream objects interally. * * ** Output Stream ** * FbsonOutStream is a custom output stream classes, to contain the FBSON * serialized binary. The class is conveniently used to specialize templates of * FbsonParser and FbsonWriter. * * @author Tian Xia */ #ifndef FBSON_FBSONSTREAM_H #define FBSON_FBSONSTREAM_H #ifndef __STDC_FORMAT_MACROS #define __STDC_FORMAT_MACROS #endif #if defined OS_WIN && !defined snprintf #define snprintf _snprintf #endif #include #include namespace fbson { // lengths includes sign #define MAX_INT_DIGITS 11 #define MAX_INT64_DIGITS 20 #define MAX_DOUBLE_DIGITS 23 // 1(sign)+16(significant)+1(decimal)+5(exponent) /* * FBSON's implementation of input buffer */ class FbsonInBuffer : public std::streambuf { public: FbsonInBuffer(const char* str, uint32_t len) { // this is read buffer and the str will not be changed // so we use const_cast (ugly!) to remove constness char* pch(const_cast(str)); setg(pch, pch, pch + len); } }; /* * FBSON's implementation of output stream. * * This is a wrapper of a char buffer. By default, the buffer capacity is 1024 * bytes. We will double the buffer if realloc is needed for writes. */ class FbsonOutStream : public std::ostream { public: explicit FbsonOutStream(uint32_t capacity = 1024) : std::ostream(nullptr), head_(nullptr), size_(0), capacity_(capacity), alloc_(true) { if (capacity_ == 0) { capacity_ = 1024; } head_ = (char*)malloc(capacity_); } FbsonOutStream(char* buffer, uint32_t capacity) : std::ostream(nullptr), head_(buffer), size_(0), capacity_(capacity), alloc_(false) { assert(buffer && capacity_ > 0); } ~FbsonOutStream() { if (alloc_) { free(head_); } } void put(char c) { write(&c, 1); } void write(const char* c_str) { write(c_str, (uint32_t)strlen(c_str)); } void write(const char* bytes, uint32_t len) { if (len == 0) return; if (size_ + len > capacity_) { realloc(len); } memcpy(head_ + size_, bytes, len); size_ += len; } // write the integer to string void write(int i) { // snprintf automatically adds a NULL, so we need one more char if (size_ + MAX_INT_DIGITS + 1 > capacity_) { realloc(MAX_INT_DIGITS + 1); } int len = snprintf(head_ + size_, MAX_INT_DIGITS + 1, "%d", i); assert(len > 0); size_ += len; } // write the 64bit integer to string void write(int64_t l) { // snprintf automatically adds a NULL, so we need one more char if (size_ + MAX_INT64_DIGITS + 1 > capacity_) { realloc(MAX_INT64_DIGITS + 1); } int len = snprintf(head_ + size_, MAX_INT64_DIGITS + 1, "%" PRIi64, l); assert(len > 0); size_ += len; } // write the double to string void write(double d) { // snprintf automatically adds a NULL, so we need one more char if (size_ + MAX_DOUBLE_DIGITS + 1 > capacity_) { realloc(MAX_DOUBLE_DIGITS + 1); } int len = snprintf(head_ + size_, MAX_DOUBLE_DIGITS + 1, "%.15g", d); assert(len > 0); size_ += len; } pos_type tellp() const { return size_; } void seekp(pos_type pos) { size_ = (uint32_t)pos; } const char* getBuffer() const { return head_; } pos_type getSize() const { return tellp(); } private: void realloc(uint32_t len) { assert(capacity_ > 0); capacity_ *= 2; while (capacity_ < size_ + len) { capacity_ *= 2; } if (alloc_) { char* new_buf = (char*)::realloc(head_, capacity_); assert(new_buf); head_ = new_buf; } else { char* new_buf = (char*)::malloc(capacity_); assert(new_buf); memcpy(new_buf, head_, size_); head_ = new_buf; alloc_ = true; } } private: char* head_; uint32_t size_; uint32_t capacity_; bool alloc_; }; } // namespace fbson #endif // FBSON_FBSONSTREAM_H rocksdb-4.5.1/third-party/fbson/FbsonUtil.h000066400000000000000000000065241267530634000206210ustar00rootroot00000000000000/* * Copyright (c) 2014, Facebook, Inc. * All rights reserved. * * This source code is licensed under the BSD-style license found in the * LICENSE file in the root directory of this source tree. An additional grant * of patent rights can be found in the PATENTS file in the same directory. * */ /* * This header file defines miscellaneous utility classes. * * @author Tian Xia */ #ifndef FBSON_FBSONUTIL_H #define FBSON_FBSONUTIL_H #include #include "FbsonDocument.h" namespace fbson { #define OUT_BUF_SIZE 1024 /* * FbsonToJson converts an FbsonValue object to a JSON string. */ class FbsonToJson { public: FbsonToJson() : os_(buffer_, OUT_BUF_SIZE) {} // get json string const char* json(const FbsonValue* pval) { os_.clear(); os_.seekp(0); if (pval) { intern_json(pval); } os_.put(0); return os_.getBuffer(); } private: // recursively convert FbsonValue void intern_json(const FbsonValue* val) { switch (val->type()) { case FbsonType::T_Null: { os_.write("null", 4); break; } case FbsonType::T_True: { os_.write("true", 4); break; } case FbsonType::T_False: { os_.write("false", 5); break; } case FbsonType::T_Int8: { os_.write(((Int8Val*)val)->val()); break; } case FbsonType::T_Int16: { os_.write(((Int16Val*)val)->val()); break; } case FbsonType::T_Int32: { os_.write(((Int32Val*)val)->val()); break; } case FbsonType::T_Int64: { os_.write(((Int64Val*)val)->val()); break; } case FbsonType::T_Double: { os_.write(((DoubleVal*)val)->val()); break; } case FbsonType::T_String: { os_.put('"'); os_.write(((StringVal*)val)->getBlob(), ((StringVal*)val)->getBlobLen()); os_.put('"'); break; } case FbsonType::T_Binary: { os_.write("\"", 9); os_.write(((BinaryVal*)val)->getBlob(), ((BinaryVal*)val)->getBlobLen()); os_.write("\"", 9); break; } case FbsonType::T_Object: { object_to_json((ObjectVal*)val); break; } case FbsonType::T_Array: { array_to_json((ArrayVal*)val); break; } default: break; } } // convert object void object_to_json(const ObjectVal* val) { os_.put('{'); auto iter = val->begin(); auto iter_fence = val->end(); while (iter < iter_fence) { // write key if (iter->klen()) { os_.put('"'); os_.write(iter->getKeyStr(), iter->klen()); os_.put('"'); } else { os_.write(iter->getKeyId()); } os_.put(':'); // convert value intern_json(iter->value()); ++iter; if (iter != iter_fence) { os_.put(','); } } assert(iter == iter_fence); os_.put('}'); } // convert array to json void array_to_json(const ArrayVal* val) { os_.put('['); auto iter = val->begin(); auto iter_fence = val->end(); while (iter != iter_fence) { // convert value intern_json((const FbsonValue*)iter); ++iter; if (iter != iter_fence) { os_.put(','); } } assert(iter == iter_fence); os_.put(']'); } private: FbsonOutStream os_; char buffer_[OUT_BUF_SIZE]; }; } // namespace fbson #endif // FBSON_FBSONUTIL_H rocksdb-4.5.1/third-party/fbson/FbsonWriter.h000066400000000000000000000244401267530634000211550ustar00rootroot00000000000000/* * Copyright (c) 2014, Facebook, Inc. * All rights reserved. * * This source code is licensed under the BSD-style license found in the * LICENSE file in the root directory of this source tree. An additional grant * of patent rights can be found in the PATENTS file in the same directory. * */ /* * This file defines FbsonWriterT (template) and FbsonWriter. * * FbsonWriterT is a template class which implements an FBSON serializer. * Users call various write functions of FbsonWriterT object to write values * directly to FBSON packed bytes. All write functions of value or key return * the number of bytes written to FBSON, or 0 if there is an error. To write an * object, an array, or a string, you must call writeStart[..] before writing * values or key, and call writeEnd[..] after finishing at the end. * * By default, an FbsonWriterT object creates an output stream buffer. * Alternatively, you can also pass any output stream object to a writer, as * long as the stream object implements some basic functions of std::ostream * (such as FbsonOutStream, see FbsonStream.h). * * FbsonWriter specializes FbsonWriterT with FbsonOutStream type (see * FbsonStream.h). So unless you want to provide own a different output stream * type, use FbsonParser object. * * @author Tian Xia */ #ifndef FBSON_FBSONWRITER_H #define FBSON_FBSONWRITER_H #include #include "FbsonDocument.h" #include "FbsonStream.h" namespace fbson { template class FbsonWriterT { public: FbsonWriterT() : alloc_(true), hasHdr_(false), kvState_(WS_Value), str_pos_(0) { os_ = new OS_TYPE(); } explicit FbsonWriterT(OS_TYPE& os) : os_(&os), alloc_(false), hasHdr_(false), kvState_(WS_Value), str_pos_(0) {} ~FbsonWriterT() { if (alloc_) { delete os_; } } void reset() { os_->clear(); os_->seekp(0); hasHdr_ = false; kvState_ = WS_Value; for (; !stack_.empty(); stack_.pop()) ; } // write a key string (or key id if an external dict is provided) uint32_t writeKey(const char* key, uint8_t len, hDictInsert handler = nullptr) { if (len && !stack_.empty() && verifyKeyState()) { int key_id = -1; if (handler) { key_id = handler(key, len); } uint32_t size = sizeof(uint8_t); if (key_id < 0) { os_->put(len); os_->write(key, len); size += len; } else if (key_id <= FbsonKeyValue::sMaxKeyId) { FbsonKeyValue::keyid_type idx = key_id; os_->put(0); os_->write((char*)&idx, sizeof(FbsonKeyValue::keyid_type)); size += sizeof(FbsonKeyValue::keyid_type); } else { // key id overflow assert(0); return 0; } kvState_ = WS_Key; return size; } return 0; } // write a key id uint32_t writeKey(FbsonKeyValue::keyid_type idx) { if (!stack_.empty() && verifyKeyState()) { os_->put(0); os_->write((char*)&idx, sizeof(FbsonKeyValue::keyid_type)); kvState_ = WS_Key; return sizeof(uint8_t) + sizeof(FbsonKeyValue::keyid_type); } return 0; } uint32_t writeNull() { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Null); kvState_ = WS_Value; return sizeof(FbsonValue); } return 0; } uint32_t writeBool(bool b) { if (!stack_.empty() && verifyValueState()) { if (b) { os_->put((FbsonTypeUnder)FbsonType::T_True); } else { os_->put((FbsonTypeUnder)FbsonType::T_False); } kvState_ = WS_Value; return sizeof(FbsonValue); } return 0; } uint32_t writeInt8(int8_t v) { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Int8); os_->put(v); kvState_ = WS_Value; return sizeof(Int8Val); } return 0; } uint32_t writeInt16(int16_t v) { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Int16); os_->write((char*)&v, sizeof(int16_t)); kvState_ = WS_Value; return sizeof(Int16Val); } return 0; } uint32_t writeInt32(int32_t v) { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Int32); os_->write((char*)&v, sizeof(int32_t)); kvState_ = WS_Value; return sizeof(Int32Val); } return 0; } uint32_t writeInt64(int64_t v) { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Int64); os_->write((char*)&v, sizeof(int64_t)); kvState_ = WS_Value; return sizeof(Int64Val); } return 0; } uint32_t writeDouble(double v) { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Double); os_->write((char*)&v, sizeof(double)); kvState_ = WS_Value; return sizeof(DoubleVal); } return 0; } // must call writeStartString before writing a string val bool writeStartString() { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_String); str_pos_ = os_->tellp(); // fill the size bytes with 0 for now uint32_t size = 0; os_->write((char*)&size, sizeof(uint32_t)); kvState_ = WS_String; return true; } return false; } // finish writing a string val bool writeEndString() { if (kvState_ == WS_String) { std::streampos cur_pos = os_->tellp(); int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); assert(size >= 0); os_->seekp(str_pos_); os_->write((char*)&size, sizeof(uint32_t)); os_->seekp(cur_pos); kvState_ = WS_Value; return true; } return false; } uint32_t writeString(const char* str, uint32_t len) { if (kvState_ == WS_String) { os_->write(str, len); return len; } return 0; } uint32_t writeString(char ch) { if (kvState_ == WS_String) { os_->put(ch); return 1; } return 0; } // must call writeStartBinary before writing a binary val bool writeStartBinary() { if (!stack_.empty() && verifyValueState()) { os_->put((FbsonTypeUnder)FbsonType::T_Binary); str_pos_ = os_->tellp(); // fill the size bytes with 0 for now uint32_t size = 0; os_->write((char*)&size, sizeof(uint32_t)); kvState_ = WS_Binary; return true; } return false; } // finish writing a binary val bool writeEndBinary() { if (kvState_ == WS_Binary) { std::streampos cur_pos = os_->tellp(); int32_t size = (int32_t)(cur_pos - str_pos_ - sizeof(uint32_t)); assert(size >= 0); os_->seekp(str_pos_); os_->write((char*)&size, sizeof(uint32_t)); os_->seekp(cur_pos); kvState_ = WS_Value; return true; } return false; } uint32_t writeBinary(const char* bin, uint32_t len) { if (kvState_ == WS_Binary) { os_->write(bin, len); return len; } return 0; } // must call writeStartObject before writing an object val bool writeStartObject() { if (stack_.empty() || verifyValueState()) { if (stack_.empty()) { // if this is a new FBSON, write the header if (!hasHdr_) { writeHeader(); } else return false; } os_->put((FbsonTypeUnder)FbsonType::T_Object); // save the size position stack_.push(WriteInfo({WS_Object, os_->tellp()})); // fill the size bytes with 0 for now uint32_t size = 0; os_->write((char*)&size, sizeof(uint32_t)); kvState_ = WS_Value; return true; } return false; } // finish writing an object val bool writeEndObject() { if (!stack_.empty() && stack_.top().state == WS_Object && kvState_ == WS_Value) { WriteInfo& ci = stack_.top(); std::streampos cur_pos = os_->tellp(); int32_t size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); assert(size >= 0); os_->seekp(ci.sz_pos); os_->write((char*)&size, sizeof(uint32_t)); os_->seekp(cur_pos); stack_.pop(); return true; } return false; } // must call writeStartArray before writing an array val bool writeStartArray() { if (stack_.empty() || verifyValueState()) { if (stack_.empty()) { // if this is a new FBSON, write the header if (!hasHdr_) { writeHeader(); } else return false; } os_->put((FbsonTypeUnder)FbsonType::T_Array); // save the size position stack_.push(WriteInfo({WS_Array, os_->tellp()})); // fill the size bytes with 0 for now uint32_t size = 0; os_->write((char*)&size, sizeof(uint32_t)); kvState_ = WS_Value; return true; } return false; } // finish writing an array val bool writeEndArray() { if (!stack_.empty() && stack_.top().state == WS_Array && kvState_ == WS_Value) { WriteInfo& ci = stack_.top(); std::streampos cur_pos = os_->tellp(); int32_t size = (int32_t)(cur_pos - ci.sz_pos - sizeof(uint32_t)); assert(size >= 0); os_->seekp(ci.sz_pos); os_->write((char*)&size, sizeof(uint32_t)); os_->seekp(cur_pos); stack_.pop(); return true; } return false; } OS_TYPE* getOutput() { return os_; } private: // verify we are in the right state before writing a value bool verifyValueState() { assert(!stack_.empty()); return (stack_.top().state == WS_Object && kvState_ == WS_Key) || (stack_.top().state == WS_Array && kvState_ == WS_Value); } // verify we are in the right state before writing a key bool verifyKeyState() { assert(!stack_.empty()); return stack_.top().state == WS_Object && kvState_ == WS_Value; } void writeHeader() { os_->put(FBSON_VER); hasHdr_ = true; } private: enum WriteState { WS_NONE, WS_Array, WS_Object, WS_Key, WS_Value, WS_String, WS_Binary, }; struct WriteInfo { WriteState state; std::streampos sz_pos; }; private: OS_TYPE* os_; bool alloc_; bool hasHdr_; WriteState kvState_; // key or value state std::streampos str_pos_; std::stack stack_; }; typedef FbsonWriterT FbsonWriter; } // namespace fbson #endif // FBSON_FBSONWRITER_H rocksdb-4.5.1/third-party/flashcache/000077500000000000000000000000001267530634000175065ustar00rootroot00000000000000rocksdb-4.5.1/third-party/flashcache/flashcache_ioctl.h000066400000000000000000000041041267530634000231310ustar00rootroot00000000000000/**************************************************************************** * flashcache_ioctl.h * FlashCache: Device mapper target for block-level disk caching * * Copyright 2010 Facebook, Inc. * Author: Mohan Srinivasan (mohan@facebook.com) * * Based on DM-Cache: * Copyright (C) International Business Machines Corp., 2006 * Author: Ming Zhao (mingzhao@ufl.edu) * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; under version 2 of the License. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program. If not, see . ****************************************************************************/ #ifdef OS_LINUX #ifndef FLASHCACHE_IOCTL_H #define FLASHCACHE_IOCTL_H #include #define FLASHCACHE_IOCTL 0xfe enum { FLASHCACHEADDNCPID_CMD=200, FLASHCACHEDELNCPID_CMD, FLASHCACHEDELNCALL_CMD, FLASHCACHEADDWHITELIST_CMD, FLASHCACHEDELWHITELIST_CMD, FLASHCACHEDELWHITELISTALL_CMD, }; #define FLASHCACHEADDNCPID _IOW(FLASHCACHE_IOCTL, FLASHCACHEADDNCPID_CMD, pid_t) #define FLASHCACHEDELNCPID _IOW(FLASHCACHE_IOCTL, FLASHCACHEDELNCPID_CMD, pid_t) #define FLASHCACHEDELNCALL _IOW(FLASHCACHE_IOCTL, FLASHCACHEDELNCALL_CMD, pid_t) #define FLASHCACHEADDBLACKLIST FLASHCACHEADDNCPID #define FLASHCACHEDELBLACKLIST FLASHCACHEDELNCPID #define FLASHCACHEDELALLBLACKLIST FLASHCACHEDELNCALL #define FLASHCACHEADDWHITELIST _IOW(FLASHCACHE_IOCTL, FLASHCACHEADDWHITELIST_CMD, pid_t) #define FLASHCACHEDELWHITELIST _IOW(FLASHCACHE_IOCTL, FLASHCACHEDELWHITELIST_CMD, pid_t) #define FLASHCACHEDELALLWHITELIST _IOW(FLASHCACHE_IOCTL, FLASHCACHEDELWHITELISTALL_CMD, pid_t) #endif /* FLASHCACHE_IOCTL_H */ #endif /* OS_LINUX */ rocksdb-4.5.1/third-party/gtest-1.7.0/000077500000000000000000000000001267530634000172145ustar00rootroot00000000000000rocksdb-4.5.1/third-party/gtest-1.7.0/fused-src/000077500000000000000000000000001267530634000211075ustar00rootroot00000000000000rocksdb-4.5.1/third-party/gtest-1.7.0/fused-src/gtest/000077500000000000000000000000001267530634000222355ustar00rootroot00000000000000rocksdb-4.5.1/third-party/gtest-1.7.0/fused-src/gtest/CMakeLists.txt000066400000000000000000000000401267530634000247670ustar00rootroot00000000000000add_library(gtest gtest-all.cc) rocksdb-4.5.1/third-party/gtest-1.7.0/fused-src/gtest/gtest-all.cc000066400000000000000000013402271267530634000244510ustar00rootroot00000000000000// Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: mheule@google.com (Markus Heule) // // Google C++ Testing Framework (Google Test) // // Sometimes it's desirable to build Google Test by compiling a single file. // This file serves this purpose. // This line ensures that gtest.h can be compiled on its own, even // when it's fused. #include "gtest/gtest.h" // The following lines pull in the real gtest *.cc files. // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // // The Google C++ Testing Framework (Google Test) // Copyright 2007, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // // Utilities for testing Google Test itself and code that uses Google Test // (e.g. frameworks built on top of Google Test). #ifndef GTEST_INCLUDE_GTEST_GTEST_SPI_H_ #define GTEST_INCLUDE_GTEST_GTEST_SPI_H_ namespace testing { // This helper class can be used to mock out Google Test failure reporting // so that we can test Google Test or code that builds on Google Test. // // An object of this class appends a TestPartResult object to the // TestPartResultArray object given in the constructor whenever a Google Test // failure is reported. It can either intercept only failures that are // generated in the same thread that created this object or it can intercept // all generated failures. The scope of this mock object can be controlled with // the second argument to the two arguments constructor. class GTEST_API_ ScopedFakeTestPartResultReporter : public TestPartResultReporterInterface { public: // The two possible mocking modes of this object. enum InterceptMode { INTERCEPT_ONLY_CURRENT_THREAD, // Intercepts only thread local failures. INTERCEPT_ALL_THREADS // Intercepts all failures. }; // The c'tor sets this object as the test part result reporter used // by Google Test. The 'result' parameter specifies where to report the // results. This reporter will only catch failures generated in the current // thread. DEPRECATED explicit ScopedFakeTestPartResultReporter(TestPartResultArray* result); // Same as above, but you can choose the interception scope of this object. ScopedFakeTestPartResultReporter(InterceptMode intercept_mode, TestPartResultArray* result); // The d'tor restores the previous test part result reporter. virtual ~ScopedFakeTestPartResultReporter(); // Appends the TestPartResult object to the TestPartResultArray // received in the constructor. // // This method is from the TestPartResultReporterInterface // interface. virtual void ReportTestPartResult(const TestPartResult& result); private: void Init(); const InterceptMode intercept_mode_; TestPartResultReporterInterface* old_reporter_; TestPartResultArray* const result_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedFakeTestPartResultReporter); }; namespace internal { // A helper class for implementing EXPECT_FATAL_FAILURE() and // EXPECT_NONFATAL_FAILURE(). Its destructor verifies that the given // TestPartResultArray contains exactly one failure that has the given // type and contains the given substring. If that's not the case, a // non-fatal failure will be generated. class GTEST_API_ SingleFailureChecker { public: // The constructor remembers the arguments. SingleFailureChecker(const TestPartResultArray* results, TestPartResult::Type type, const string& substr); ~SingleFailureChecker(); private: const TestPartResultArray* const results_; const TestPartResult::Type type_; const string substr_; GTEST_DISALLOW_COPY_AND_ASSIGN_(SingleFailureChecker); }; } // namespace internal } // namespace testing // A set of macros for testing Google Test assertions or code that's expected // to generate Google Test fatal failures. It verifies that the given // statement will cause exactly one fatal Google Test failure with 'substr' // being part of the failure message. // // There are two different versions of this macro. EXPECT_FATAL_FAILURE only // affects and considers failures generated in the current thread and // EXPECT_FATAL_FAILURE_ON_ALL_THREADS does the same but for all threads. // // The verification of the assertion is done correctly even when the statement // throws an exception or aborts the current function. // // Known restrictions: // - 'statement' cannot reference local non-static variables or // non-static members of the current object. // - 'statement' cannot return a value. // - You cannot stream a failure message to this macro. // // Note that even though the implementations of the following two // macros are much alike, we cannot refactor them to use a common // helper macro, due to some peculiarity in how the preprocessor // works. The AcceptsMacroThatExpandsToUnprotectedComma test in // gtest_unittest.cc will fail to compile if we do that. #define EXPECT_FATAL_FAILURE(statement, substr) \ do { \ class GTestExpectFatalFailureHelper {\ public:\ static void Execute() { statement; }\ };\ ::testing::TestPartResultArray gtest_failures;\ ::testing::internal::SingleFailureChecker gtest_checker(\ >est_failures, ::testing::TestPartResult::kFatalFailure, (substr));\ {\ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ ::testing::ScopedFakeTestPartResultReporter:: \ INTERCEPT_ONLY_CURRENT_THREAD, >est_failures);\ GTestExpectFatalFailureHelper::Execute();\ }\ } while (::testing::internal::AlwaysFalse()) #define EXPECT_FATAL_FAILURE_ON_ALL_THREADS(statement, substr) \ do { \ class GTestExpectFatalFailureHelper {\ public:\ static void Execute() { statement; }\ };\ ::testing::TestPartResultArray gtest_failures;\ ::testing::internal::SingleFailureChecker gtest_checker(\ >est_failures, ::testing::TestPartResult::kFatalFailure, (substr));\ {\ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ ::testing::ScopedFakeTestPartResultReporter:: \ INTERCEPT_ALL_THREADS, >est_failures);\ GTestExpectFatalFailureHelper::Execute();\ }\ } while (::testing::internal::AlwaysFalse()) // A macro for testing Google Test assertions or code that's expected to // generate Google Test non-fatal failures. It asserts that the given // statement will cause exactly one non-fatal Google Test failure with 'substr' // being part of the failure message. // // There are two different versions of this macro. EXPECT_NONFATAL_FAILURE only // affects and considers failures generated in the current thread and // EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS does the same but for all threads. // // 'statement' is allowed to reference local variables and members of // the current object. // // The verification of the assertion is done correctly even when the statement // throws an exception or aborts the current function. // // Known restrictions: // - You cannot stream a failure message to this macro. // // Note that even though the implementations of the following two // macros are much alike, we cannot refactor them to use a common // helper macro, due to some peculiarity in how the preprocessor // works. If we do that, the code won't compile when the user gives // EXPECT_NONFATAL_FAILURE() a statement that contains a macro that // expands to code containing an unprotected comma. The // AcceptsMacroThatExpandsToUnprotectedComma test in gtest_unittest.cc // catches that. // // For the same reason, we have to write // if (::testing::internal::AlwaysTrue()) { statement; } // instead of // GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) // to avoid an MSVC warning on unreachable code. #define EXPECT_NONFATAL_FAILURE(statement, substr) \ do {\ ::testing::TestPartResultArray gtest_failures;\ ::testing::internal::SingleFailureChecker gtest_checker(\ >est_failures, ::testing::TestPartResult::kNonFatalFailure, \ (substr));\ {\ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ ::testing::ScopedFakeTestPartResultReporter:: \ INTERCEPT_ONLY_CURRENT_THREAD, >est_failures);\ if (::testing::internal::AlwaysTrue()) { statement; }\ }\ } while (::testing::internal::AlwaysFalse()) #define EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(statement, substr) \ do {\ ::testing::TestPartResultArray gtest_failures;\ ::testing::internal::SingleFailureChecker gtest_checker(\ >est_failures, ::testing::TestPartResult::kNonFatalFailure, \ (substr));\ {\ ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ ::testing::ScopedFakeTestPartResultReporter::INTERCEPT_ALL_THREADS, \ >est_failures);\ if (::testing::internal::AlwaysTrue()) { statement; }\ }\ } while (::testing::internal::AlwaysFalse()) #endif // GTEST_INCLUDE_GTEST_GTEST_SPI_H_ #include #include #include #include #include #include #include #include #include #include #include #include #include #include // NOLINT #include #include #if GTEST_OS_LINUX // TODO(kenton@google.com): Use autoconf to detect availability of // gettimeofday(). # define GTEST_HAS_GETTIMEOFDAY_ 1 # include // NOLINT # include // NOLINT # include // NOLINT // Declares vsnprintf(). This header is not available on Windows. # include // NOLINT # include // NOLINT # include // NOLINT # include // NOLINT # include #elif GTEST_OS_SYMBIAN # define GTEST_HAS_GETTIMEOFDAY_ 1 # include // NOLINT #elif GTEST_OS_ZOS # define GTEST_HAS_GETTIMEOFDAY_ 1 # include // NOLINT // On z/OS we additionally need strings.h for strcasecmp. # include // NOLINT #elif GTEST_OS_WINDOWS_MOBILE // We are on Windows CE. # include // NOLINT # undef min #elif GTEST_OS_WINDOWS // We are on Windows proper. # include // NOLINT # include // NOLINT # include // NOLINT # include // NOLINT # if GTEST_OS_WINDOWS_MINGW // MinGW has gettimeofday() but not _ftime64(). // TODO(kenton@google.com): Use autoconf to detect availability of // gettimeofday(). // TODO(kenton@google.com): There are other ways to get the time on // Windows, like GetTickCount() or GetSystemTimeAsFileTime(). MinGW // supports these. consider using them instead. # define GTEST_HAS_GETTIMEOFDAY_ 1 # include // NOLINT # endif // GTEST_OS_WINDOWS_MINGW // cpplint thinks that the header is already included, so we want to // silence it. # include // NOLINT # undef min #else // Assume other platforms have gettimeofday(). // TODO(kenton@google.com): Use autoconf to detect availability of // gettimeofday(). # define GTEST_HAS_GETTIMEOFDAY_ 1 // cpplint thinks that the header is already included, so we want to // silence it. # include // NOLINT # include // NOLINT #endif // GTEST_OS_LINUX #if GTEST_HAS_EXCEPTIONS # include #endif #if GTEST_CAN_STREAM_RESULTS_ # include // NOLINT # include // NOLINT # include // NOLINT # include // NOLINT #endif // Indicates that this translation unit is part of Google Test's // implementation. It must come before gtest-internal-inl.h is // included, or there will be a compiler error. This trick is to // prevent a user from accidentally including gtest-internal-inl.h in // his code. #define GTEST_IMPLEMENTATION_ 1 // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // Utility functions and classes used by the Google C++ testing framework. // // Author: wan@google.com (Zhanyong Wan) // // This file contains purely Google Test's internal implementation. Please // DO NOT #INCLUDE IT IN A USER PROGRAM. #ifndef GTEST_SRC_GTEST_INTERNAL_INL_H_ #define GTEST_SRC_GTEST_INTERNAL_INL_H_ // GTEST_IMPLEMENTATION_ is defined to 1 iff the current translation unit is // part of Google Test's implementation; otherwise it's undefined. #if !GTEST_IMPLEMENTATION_ // If this file is included from the user's code, just say no. # error "gtest-internal-inl.h is part of Google Test's internal implementation." # error "It must not be included except by Google Test itself." #endif // GTEST_IMPLEMENTATION_ #ifndef _WIN32_WCE # include #endif // !_WIN32_WCE #include #include // For strtoll/_strtoul64/malloc/free. #include // For memmove. #include #include #include #if GTEST_CAN_STREAM_RESULTS_ # include // NOLINT # include // NOLINT #endif #if GTEST_OS_WINDOWS # include // NOLINT #endif // GTEST_OS_WINDOWS namespace testing { // Declares the flags. // // We don't want the users to modify this flag in the code, but want // Google Test's own unit tests to be able to access it. Therefore we // declare it here as opposed to in gtest.h. GTEST_DECLARE_bool_(death_test_use_fork); namespace internal { // The value of GetTestTypeId() as seen from within the Google Test // library. This is solely for testing GetTestTypeId(). GTEST_API_ extern const TypeId kTestTypeIdInGoogleTest; // Names of the flags (needed for parsing Google Test flags). const char kAlsoRunDisabledTestsFlag[] = "also_run_disabled_tests"; const char kBreakOnFailureFlag[] = "break_on_failure"; const char kCatchExceptionsFlag[] = "catch_exceptions"; const char kColorFlag[] = "color"; const char kFilterFlag[] = "filter"; const char kListTestsFlag[] = "list_tests"; const char kOutputFlag[] = "output"; const char kPrintTimeFlag[] = "print_time"; const char kRandomSeedFlag[] = "random_seed"; const char kRepeatFlag[] = "repeat"; const char kShuffleFlag[] = "shuffle"; const char kStackTraceDepthFlag[] = "stack_trace_depth"; const char kStreamResultToFlag[] = "stream_result_to"; const char kThrowOnFailureFlag[] = "throw_on_failure"; // A valid random seed must be in [1, kMaxRandomSeed]. const int kMaxRandomSeed = 99999; // g_help_flag is true iff the --help flag or an equivalent form is // specified on the command line. GTEST_API_ extern bool g_help_flag; // Returns the current time in milliseconds. GTEST_API_ TimeInMillis GetTimeInMillis(); // Returns true iff Google Test should use colors in the output. GTEST_API_ bool ShouldUseColor(bool stdout_is_tty); // Formats the given time in milliseconds as seconds. GTEST_API_ std::string FormatTimeInMillisAsSeconds(TimeInMillis ms); // Converts the given time in milliseconds to a date string in the ISO 8601 // format, without the timezone information. N.B.: due to the use the // non-reentrant localtime() function, this function is not thread safe. Do // not use it in any code that can be called from multiple threads. GTEST_API_ std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms); // Parses a string for an Int32 flag, in the form of "--flag=value". // // On success, stores the value of the flag in *value, and returns // true. On failure, returns false without changing *value. GTEST_API_ bool ParseInt32Flag( const char* str, const char* flag, Int32* value); // Returns a random seed in range [1, kMaxRandomSeed] based on the // given --gtest_random_seed flag value. inline int GetRandomSeedFromFlag(Int32 random_seed_flag) { const unsigned int raw_seed = (random_seed_flag == 0) ? static_cast(GetTimeInMillis()) : static_cast(random_seed_flag); // Normalizes the actual seed to range [1, kMaxRandomSeed] such that // it's easy to type. const int normalized_seed = static_cast((raw_seed - 1U) % static_cast(kMaxRandomSeed)) + 1; return normalized_seed; } // Returns the first valid random seed after 'seed'. The behavior is // undefined if 'seed' is invalid. The seed after kMaxRandomSeed is // considered to be 1. inline int GetNextRandomSeed(int seed) { GTEST_CHECK_(1 <= seed && seed <= kMaxRandomSeed) << "Invalid random seed " << seed << " - must be in [1, " << kMaxRandomSeed << "]."; const int next_seed = seed + 1; return (next_seed > kMaxRandomSeed) ? 1 : next_seed; } // This class saves the values of all Google Test flags in its c'tor, and // restores them in its d'tor. class GTestFlagSaver { public: // The c'tor. GTestFlagSaver() { also_run_disabled_tests_ = GTEST_FLAG(also_run_disabled_tests); break_on_failure_ = GTEST_FLAG(break_on_failure); catch_exceptions_ = GTEST_FLAG(catch_exceptions); color_ = GTEST_FLAG(color); death_test_style_ = GTEST_FLAG(death_test_style); death_test_use_fork_ = GTEST_FLAG(death_test_use_fork); filter_ = GTEST_FLAG(filter); internal_run_death_test_ = GTEST_FLAG(internal_run_death_test); list_tests_ = GTEST_FLAG(list_tests); output_ = GTEST_FLAG(output); print_time_ = GTEST_FLAG(print_time); random_seed_ = GTEST_FLAG(random_seed); repeat_ = GTEST_FLAG(repeat); shuffle_ = GTEST_FLAG(shuffle); stack_trace_depth_ = GTEST_FLAG(stack_trace_depth); stream_result_to_ = GTEST_FLAG(stream_result_to); throw_on_failure_ = GTEST_FLAG(throw_on_failure); } // The d'tor is not virtual. DO NOT INHERIT FROM THIS CLASS. ~GTestFlagSaver() { GTEST_FLAG(also_run_disabled_tests) = also_run_disabled_tests_; GTEST_FLAG(break_on_failure) = break_on_failure_; GTEST_FLAG(catch_exceptions) = catch_exceptions_; GTEST_FLAG(color) = color_; GTEST_FLAG(death_test_style) = death_test_style_; GTEST_FLAG(death_test_use_fork) = death_test_use_fork_; GTEST_FLAG(filter) = filter_; GTEST_FLAG(internal_run_death_test) = internal_run_death_test_; GTEST_FLAG(list_tests) = list_tests_; GTEST_FLAG(output) = output_; GTEST_FLAG(print_time) = print_time_; GTEST_FLAG(random_seed) = random_seed_; GTEST_FLAG(repeat) = repeat_; GTEST_FLAG(shuffle) = shuffle_; GTEST_FLAG(stack_trace_depth) = stack_trace_depth_; GTEST_FLAG(stream_result_to) = stream_result_to_; GTEST_FLAG(throw_on_failure) = throw_on_failure_; } private: // Fields for saving the original values of flags. bool also_run_disabled_tests_; bool break_on_failure_; bool catch_exceptions_; std::string color_; std::string death_test_style_; bool death_test_use_fork_; std::string filter_; std::string internal_run_death_test_; bool list_tests_; std::string output_; bool print_time_; internal::Int32 random_seed_; internal::Int32 repeat_; bool shuffle_; internal::Int32 stack_trace_depth_; std::string stream_result_to_; bool throw_on_failure_; } GTEST_ATTRIBUTE_UNUSED_; // Converts a Unicode code point to a narrow string in UTF-8 encoding. // code_point parameter is of type UInt32 because wchar_t may not be // wide enough to contain a code point. // If the code_point is not a valid Unicode code point // (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted // to "(Invalid Unicode 0xXXXXXXXX)". GTEST_API_ std::string CodePointToUtf8(UInt32 code_point); // Converts a wide string to a narrow string in UTF-8 encoding. // The wide string is assumed to have the following encoding: // UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS) // UTF-32 if sizeof(wchar_t) == 4 (on Linux) // Parameter str points to a null-terminated wide string. // Parameter num_chars may additionally limit the number // of wchar_t characters processed. -1 is used when the entire string // should be processed. // If the string contains code points that are not valid Unicode code points // (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output // as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding // and contains invalid UTF-16 surrogate pairs, values in those pairs // will be encoded as individual Unicode characters from Basic Normal Plane. GTEST_API_ std::string WideStringToUtf8(const wchar_t* str, int num_chars); // Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file // if the variable is present. If a file already exists at this location, this // function will write over it. If the variable is present, but the file cannot // be created, prints an error and exits. void WriteToShardStatusFileIfNeeded(); // Checks whether sharding is enabled by examining the relevant // environment variable values. If the variables are present, // but inconsistent (e.g., shard_index >= total_shards), prints // an error and exits. If in_subprocess_for_death_test, sharding is // disabled because it must only be applied to the original test // process. Otherwise, we could filter out death tests we intended to execute. GTEST_API_ bool ShouldShard(const char* total_shards_str, const char* shard_index_str, bool in_subprocess_for_death_test); // Parses the environment variable var as an Int32. If it is unset, // returns default_val. If it is not an Int32, prints an error and // and aborts. GTEST_API_ Int32 Int32FromEnvOrDie(const char* env_var, Int32 default_val); // Given the total number of shards, the shard index, and the test id, // returns true iff the test should be run on this shard. The test id is // some arbitrary but unique non-negative integer assigned to each test // method. Assumes that 0 <= shard_index < total_shards. GTEST_API_ bool ShouldRunTestOnShard( int total_shards, int shard_index, int test_id); // STL container utilities. // Returns the number of elements in the given container that satisfy // the given predicate. template inline int CountIf(const Container& c, Predicate predicate) { // Implemented as an explicit loop since std::count_if() in libCstd on // Solaris has a non-standard signature. int count = 0; for (typename Container::const_iterator it = c.begin(); it != c.end(); ++it) { if (predicate(*it)) ++count; } return count; } // Applies a function/functor to each element in the container. template void ForEach(const Container& c, Functor functor) { std::for_each(c.begin(), c.end(), functor); } // Returns the i-th element of the vector, or default_value if i is not // in range [0, v.size()). template inline E GetElementOr(const std::vector& v, int i, E default_value) { return (i < 0 || i >= static_cast(v.size())) ? default_value : v[i]; } // Performs an in-place shuffle of a range of the vector's elements. // 'begin' and 'end' are element indices as an STL-style range; // i.e. [begin, end) are shuffled, where 'end' == size() means to // shuffle to the end of the vector. template void ShuffleRange(internal::Random* random, int begin, int end, std::vector* v) { const int size = static_cast(v->size()); GTEST_CHECK_(0 <= begin && begin <= size) << "Invalid shuffle range start " << begin << ": must be in range [0, " << size << "]."; GTEST_CHECK_(begin <= end && end <= size) << "Invalid shuffle range finish " << end << ": must be in range [" << begin << ", " << size << "]."; // Fisher-Yates shuffle, from // http://en.wikipedia.org/wiki/Fisher-Yates_shuffle for (int range_width = end - begin; range_width >= 2; range_width--) { const int last_in_range = begin + range_width - 1; const int selected = begin + random->Generate(range_width); std::swap((*v)[selected], (*v)[last_in_range]); } } // Performs an in-place shuffle of the vector's elements. template inline void Shuffle(internal::Random* random, std::vector* v) { ShuffleRange(random, 0, static_cast(v->size()), v); } // A function for deleting an object. Handy for being used as a // functor. template static void Delete(T* x) { delete x; } // A predicate that checks the key of a TestProperty against a known key. // // TestPropertyKeyIs is copyable. class TestPropertyKeyIs { public: // Constructor. // // TestPropertyKeyIs has NO default constructor. explicit TestPropertyKeyIs(const std::string& key) : key_(key) {} // Returns true iff the test name of test property matches on key_. bool operator()(const TestProperty& test_property) const { return test_property.key() == key_; } private: std::string key_; }; // Class UnitTestOptions. // // This class contains functions for processing options the user // specifies when running the tests. It has only static members. // // In most cases, the user can specify an option using either an // environment variable or a command line flag. E.g. you can set the // test filter using either GTEST_FILTER or --gtest_filter. If both // the variable and the flag are present, the latter overrides the // former. class GTEST_API_ UnitTestOptions { public: // Functions for processing the gtest_output flag. // Returns the output format, or "" for normal printed output. static std::string GetOutputFormat(); // Returns the absolute path of the requested output file, or the // default (test_detail.xml in the original working directory) if // none was explicitly specified. static std::string GetAbsolutePathToOutputFile(); // Functions for processing the gtest_filter flag. // Returns true iff the wildcard pattern matches the string. The // first ':' or '\0' character in pattern marks the end of it. // // This recursive algorithm isn't very efficient, but is clear and // works well enough for matching test names, which are short. static bool PatternMatchesString(const char *pattern, const char *str); // Returns true iff the user-specified filter matches the test case // name and the test name. static bool FilterMatchesTest(const std::string &test_case_name, const std::string &test_name); #if GTEST_OS_WINDOWS // Function for supporting the gtest_catch_exception flag. // Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the // given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise. // This function is useful as an __except condition. static int GTestShouldProcessSEH(DWORD exception_code); #endif // GTEST_OS_WINDOWS // Returns true if "name" matches the ':' separated list of glob-style // filters in "filter". static bool MatchesFilter(const std::string& name, const char* filter); }; // Returns the current application's name, removing directory path if that // is present. Used by UnitTestOptions::GetOutputFile. GTEST_API_ FilePath GetCurrentExecutableName(); // The role interface for getting the OS stack trace as a string. class OsStackTraceGetterInterface { public: OsStackTraceGetterInterface() {} virtual ~OsStackTraceGetterInterface() {} // Returns the current OS stack trace as an std::string. Parameters: // // max_depth - the maximum number of stack frames to be included // in the trace. // skip_count - the number of top frames to be skipped; doesn't count // against max_depth. virtual string CurrentStackTrace(int max_depth, int skip_count) = 0; // UponLeavingGTest() should be called immediately before Google Test calls // user code. It saves some information about the current stack that // CurrentStackTrace() will use to find and hide Google Test stack frames. virtual void UponLeavingGTest() = 0; private: GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetterInterface); }; // A working implementation of the OsStackTraceGetterInterface interface. class OsStackTraceGetter : public OsStackTraceGetterInterface { public: OsStackTraceGetter() : caller_frame_(NULL) {} virtual string CurrentStackTrace(int max_depth, int skip_count) GTEST_LOCK_EXCLUDED_(mutex_); virtual void UponLeavingGTest() GTEST_LOCK_EXCLUDED_(mutex_); // This string is inserted in place of stack frames that are part of // Google Test's implementation. static const char* const kElidedFramesMarker; private: Mutex mutex_; // protects all internal state // We save the stack frame below the frame that calls user code. // We do this because the address of the frame immediately below // the user code changes between the call to UponLeavingGTest() // and any calls to CurrentStackTrace() from within the user code. void* caller_frame_; GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetter); }; // Information about a Google Test trace point. struct TraceInfo { const char* file; int line; std::string message; }; // This is the default global test part result reporter used in UnitTestImpl. // This class should only be used by UnitTestImpl. class DefaultGlobalTestPartResultReporter : public TestPartResultReporterInterface { public: explicit DefaultGlobalTestPartResultReporter(UnitTestImpl* unit_test); // Implements the TestPartResultReporterInterface. Reports the test part // result in the current test. virtual void ReportTestPartResult(const TestPartResult& result); private: UnitTestImpl* const unit_test_; GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultGlobalTestPartResultReporter); }; // This is the default per thread test part result reporter used in // UnitTestImpl. This class should only be used by UnitTestImpl. class DefaultPerThreadTestPartResultReporter : public TestPartResultReporterInterface { public: explicit DefaultPerThreadTestPartResultReporter(UnitTestImpl* unit_test); // Implements the TestPartResultReporterInterface. The implementation just // delegates to the current global test part result reporter of *unit_test_. virtual void ReportTestPartResult(const TestPartResult& result); private: UnitTestImpl* const unit_test_; GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultPerThreadTestPartResultReporter); }; // The private implementation of the UnitTest class. We don't protect // the methods under a mutex, as this class is not accessible by a // user and the UnitTest class that delegates work to this class does // proper locking. class GTEST_API_ UnitTestImpl { public: explicit UnitTestImpl(UnitTest* parent); virtual ~UnitTestImpl(); // There are two different ways to register your own TestPartResultReporter. // You can register your own repoter to listen either only for test results // from the current thread or for results from all threads. // By default, each per-thread test result repoter just passes a new // TestPartResult to the global test result reporter, which registers the // test part result for the currently running test. // Returns the global test part result reporter. TestPartResultReporterInterface* GetGlobalTestPartResultReporter(); // Sets the global test part result reporter. void SetGlobalTestPartResultReporter( TestPartResultReporterInterface* reporter); // Returns the test part result reporter for the current thread. TestPartResultReporterInterface* GetTestPartResultReporterForCurrentThread(); // Sets the test part result reporter for the current thread. void SetTestPartResultReporterForCurrentThread( TestPartResultReporterInterface* reporter); // Gets the number of successful test cases. int successful_test_case_count() const; // Gets the number of failed test cases. int failed_test_case_count() const; // Gets the number of all test cases. int total_test_case_count() const; // Gets the number of all test cases that contain at least one test // that should run. int test_case_to_run_count() const; // Gets the number of successful tests. int successful_test_count() const; // Gets the number of failed tests. int failed_test_count() const; // Gets the number of disabled tests that will be reported in the XML report. int reportable_disabled_test_count() const; // Gets the number of disabled tests. int disabled_test_count() const; // Gets the number of tests to be printed in the XML report. int reportable_test_count() const; // Gets the number of all tests. int total_test_count() const; // Gets the number of tests that should run. int test_to_run_count() const; // Gets the time of the test program start, in ms from the start of the // UNIX epoch. TimeInMillis start_timestamp() const { return start_timestamp_; } // Gets the elapsed time, in milliseconds. TimeInMillis elapsed_time() const { return elapsed_time_; } // Returns true iff the unit test passed (i.e. all test cases passed). bool Passed() const { return !Failed(); } // Returns true iff the unit test failed (i.e. some test case failed // or something outside of all tests failed). bool Failed() const { return failed_test_case_count() > 0 || ad_hoc_test_result()->Failed(); } // Gets the i-th test case among all the test cases. i can range from 0 to // total_test_case_count() - 1. If i is not in that range, returns NULL. const TestCase* GetTestCase(int i) const { const int index = GetElementOr(test_case_indices_, i, -1); return index < 0 ? NULL : test_cases_[i]; } // Gets the i-th test case among all the test cases. i can range from 0 to // total_test_case_count() - 1. If i is not in that range, returns NULL. TestCase* GetMutableTestCase(int i) { const int index = GetElementOr(test_case_indices_, i, -1); return index < 0 ? NULL : test_cases_[index]; } // Provides access to the event listener list. TestEventListeners* listeners() { return &listeners_; } // Returns the TestResult for the test that's currently running, or // the TestResult for the ad hoc test if no test is running. TestResult* current_test_result(); // Returns the TestResult for the ad hoc test. const TestResult* ad_hoc_test_result() const { return &ad_hoc_test_result_; } // Sets the OS stack trace getter. // // Does nothing if the input and the current OS stack trace getter // are the same; otherwise, deletes the old getter and makes the // input the current getter. void set_os_stack_trace_getter(OsStackTraceGetterInterface* getter); // Returns the current OS stack trace getter if it is not NULL; // otherwise, creates an OsStackTraceGetter, makes it the current // getter, and returns it. OsStackTraceGetterInterface* os_stack_trace_getter(); // Returns the current OS stack trace as an std::string. // // The maximum number of stack frames to be included is specified by // the gtest_stack_trace_depth flag. The skip_count parameter // specifies the number of top frames to be skipped, which doesn't // count against the number of frames to be included. // // For example, if Foo() calls Bar(), which in turn calls // CurrentOsStackTraceExceptTop(1), Foo() will be included in the // trace but Bar() and CurrentOsStackTraceExceptTop() won't. std::string CurrentOsStackTraceExceptTop(int skip_count) GTEST_NO_INLINE_; // Finds and returns a TestCase with the given name. If one doesn't // exist, creates one and returns it. // // Arguments: // // test_case_name: name of the test case // type_param: the name of the test's type parameter, or NULL if // this is not a typed or a type-parameterized test. // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case TestCase* GetTestCase(const char* test_case_name, const char* type_param, Test::SetUpTestCaseFunc set_up_tc, Test::TearDownTestCaseFunc tear_down_tc); // Adds a TestInfo to the unit test. // // Arguments: // // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case // test_info: the TestInfo object void AddTestInfo(Test::SetUpTestCaseFunc set_up_tc, Test::TearDownTestCaseFunc tear_down_tc, TestInfo* test_info) { // In order to support thread-safe death tests, we need to // remember the original working directory when the test program // was first invoked. We cannot do this in RUN_ALL_TESTS(), as // the user may have changed the current directory before calling // RUN_ALL_TESTS(). Therefore we capture the current directory in // AddTestInfo(), which is called to register a TEST or TEST_F // before main() is reached. if (original_working_dir_.IsEmpty()) { original_working_dir_.Set(FilePath::GetCurrentDir()); GTEST_CHECK_(!original_working_dir_.IsEmpty()) << "Failed to get the current working directory."; } GetTestCase(test_info->test_case_name(), test_info->type_param(), set_up_tc, tear_down_tc)->AddTestInfo(test_info); } #if GTEST_HAS_PARAM_TEST // Returns ParameterizedTestCaseRegistry object used to keep track of // value-parameterized tests and instantiate and register them. internal::ParameterizedTestCaseRegistry& parameterized_test_registry() { return parameterized_test_registry_; } #endif // GTEST_HAS_PARAM_TEST // Sets the TestCase object for the test that's currently running. void set_current_test_case(TestCase* a_current_test_case) { current_test_case_ = a_current_test_case; } // Sets the TestInfo object for the test that's currently running. If // current_test_info is NULL, the assertion results will be stored in // ad_hoc_test_result_. void set_current_test_info(TestInfo* a_current_test_info) { current_test_info_ = a_current_test_info; } // Registers all parameterized tests defined using TEST_P and // INSTANTIATE_TEST_CASE_P, creating regular tests for each test/parameter // combination. This method can be called more then once; it has guards // protecting from registering the tests more then once. If // value-parameterized tests are disabled, RegisterParameterizedTests is // present but does nothing. void RegisterParameterizedTests(); // Runs all tests in this UnitTest object, prints the result, and // returns true if all tests are successful. If any exception is // thrown during a test, this test is considered to be failed, but // the rest of the tests will still be run. bool RunAllTests(); // Clears the results of all tests, except the ad hoc tests. void ClearNonAdHocTestResult() { ForEach(test_cases_, TestCase::ClearTestCaseResult); } // Clears the results of ad-hoc test assertions. void ClearAdHocTestResult() { ad_hoc_test_result_.Clear(); } // Adds a TestProperty to the current TestResult object when invoked in a // context of a test or a test case, or to the global property set. If the // result already contains a property with the same key, the value will be // updated. void RecordProperty(const TestProperty& test_property); enum ReactionToSharding { HONOR_SHARDING_PROTOCOL, IGNORE_SHARDING_PROTOCOL }; // Matches the full name of each test against the user-specified // filter to decide whether the test should run, then records the // result in each TestCase and TestInfo object. // If shard_tests == HONOR_SHARDING_PROTOCOL, further filters tests // based on sharding variables in the environment. // Returns the number of tests that should run. int FilterTests(ReactionToSharding shard_tests); // Prints the names of the tests matching the user-specified filter flag. void ListTestsMatchingFilter(); const TestCase* current_test_case() const { return current_test_case_; } TestInfo* current_test_info() { return current_test_info_; } const TestInfo* current_test_info() const { return current_test_info_; } // Returns the vector of environments that need to be set-up/torn-down // before/after the tests are run. std::vector& environments() { return environments_; } // Getters for the per-thread Google Test trace stack. std::vector& gtest_trace_stack() { return *(gtest_trace_stack_.pointer()); } const std::vector& gtest_trace_stack() const { return gtest_trace_stack_.get(); } #if GTEST_HAS_DEATH_TEST void InitDeathTestSubprocessControlInfo() { internal_run_death_test_flag_.reset(ParseInternalRunDeathTestFlag()); } // Returns a pointer to the parsed --gtest_internal_run_death_test // flag, or NULL if that flag was not specified. // This information is useful only in a death test child process. // Must not be called before a call to InitGoogleTest. const InternalRunDeathTestFlag* internal_run_death_test_flag() const { return internal_run_death_test_flag_.get(); } // Returns a pointer to the current death test factory. internal::DeathTestFactory* death_test_factory() { return death_test_factory_.get(); } void SuppressTestEventsIfInSubprocess(); friend class ReplaceDeathTestFactory; #endif // GTEST_HAS_DEATH_TEST // Initializes the event listener performing XML output as specified by // UnitTestOptions. Must not be called before InitGoogleTest. void ConfigureXmlOutput(); #if GTEST_CAN_STREAM_RESULTS_ // Initializes the event listener for streaming test results to a socket. // Must not be called before InitGoogleTest. void ConfigureStreamingOutput(); #endif // Performs initialization dependent upon flag values obtained in // ParseGoogleTestFlagsOnly. Is called from InitGoogleTest after the call to // ParseGoogleTestFlagsOnly. In case a user neglects to call InitGoogleTest // this function is also called from RunAllTests. Since this function can be // called more than once, it has to be idempotent. void PostFlagParsingInit(); // Gets the random seed used at the start of the current test iteration. int random_seed() const { return random_seed_; } // Gets the random number generator. internal::Random* random() { return &random_; } // Shuffles all test cases, and the tests within each test case, // making sure that death tests are still run first. void ShuffleTests(); // Restores the test cases and tests to their order before the first shuffle. void UnshuffleTests(); // Returns the value of GTEST_FLAG(catch_exceptions) at the moment // UnitTest::Run() starts. bool catch_exceptions() const { return catch_exceptions_; } private: friend class ::testing::UnitTest; // Used by UnitTest::Run() to capture the state of // GTEST_FLAG(catch_exceptions) at the moment it starts. void set_catch_exceptions(bool value) { catch_exceptions_ = value; } // The UnitTest object that owns this implementation object. UnitTest* const parent_; // The working directory when the first TEST() or TEST_F() was // executed. internal::FilePath original_working_dir_; // The default test part result reporters. DefaultGlobalTestPartResultReporter default_global_test_part_result_reporter_; DefaultPerThreadTestPartResultReporter default_per_thread_test_part_result_reporter_; // Points to (but doesn't own) the global test part result reporter. TestPartResultReporterInterface* global_test_part_result_repoter_; // Protects read and write access to global_test_part_result_reporter_. internal::Mutex global_test_part_result_reporter_mutex_; // Points to (but doesn't own) the per-thread test part result reporter. internal::ThreadLocal per_thread_test_part_result_reporter_; // The vector of environments that need to be set-up/torn-down // before/after the tests are run. std::vector environments_; // The vector of TestCases in their original order. It owns the // elements in the vector. std::vector test_cases_; // Provides a level of indirection for the test case list to allow // easy shuffling and restoring the test case order. The i-th // element of this vector is the index of the i-th test case in the // shuffled order. std::vector test_case_indices_; #if GTEST_HAS_PARAM_TEST // ParameterizedTestRegistry object used to register value-parameterized // tests. internal::ParameterizedTestCaseRegistry parameterized_test_registry_; // Indicates whether RegisterParameterizedTests() has been called already. bool parameterized_tests_registered_; #endif // GTEST_HAS_PARAM_TEST // Index of the last death test case registered. Initially -1. int last_death_test_case_; // This points to the TestCase for the currently running test. It // changes as Google Test goes through one test case after another. // When no test is running, this is set to NULL and Google Test // stores assertion results in ad_hoc_test_result_. Initially NULL. TestCase* current_test_case_; // This points to the TestInfo for the currently running test. It // changes as Google Test goes through one test after another. When // no test is running, this is set to NULL and Google Test stores // assertion results in ad_hoc_test_result_. Initially NULL. TestInfo* current_test_info_; // Normally, a user only writes assertions inside a TEST or TEST_F, // or inside a function called by a TEST or TEST_F. Since Google // Test keeps track of which test is current running, it can // associate such an assertion with the test it belongs to. // // If an assertion is encountered when no TEST or TEST_F is running, // Google Test attributes the assertion result to an imaginary "ad hoc" // test, and records the result in ad_hoc_test_result_. TestResult ad_hoc_test_result_; // The list of event listeners that can be used to track events inside // Google Test. TestEventListeners listeners_; // The OS stack trace getter. Will be deleted when the UnitTest // object is destructed. By default, an OsStackTraceGetter is used, // but the user can set this field to use a custom getter if that is // desired. OsStackTraceGetterInterface* os_stack_trace_getter_; // True iff PostFlagParsingInit() has been called. bool post_flag_parse_init_performed_; // The random number seed used at the beginning of the test run. int random_seed_; // Our random number generator. internal::Random random_; // The time of the test program start, in ms from the start of the // UNIX epoch. TimeInMillis start_timestamp_; // How long the test took to run, in milliseconds. TimeInMillis elapsed_time_; #if GTEST_HAS_DEATH_TEST // The decomposed components of the gtest_internal_run_death_test flag, // parsed when RUN_ALL_TESTS is called. internal::scoped_ptr internal_run_death_test_flag_; internal::scoped_ptr death_test_factory_; #endif // GTEST_HAS_DEATH_TEST // A per-thread stack of traces created by the SCOPED_TRACE() macro. internal::ThreadLocal > gtest_trace_stack_; // The value of GTEST_FLAG(catch_exceptions) at the moment RunAllTests() // starts. bool catch_exceptions_; GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTestImpl); }; // class UnitTestImpl // Convenience function for accessing the global UnitTest // implementation object. inline UnitTestImpl* GetUnitTestImpl() { return UnitTest::GetInstance()->impl(); } #if GTEST_USES_SIMPLE_RE // Internal helper functions for implementing the simple regular // expression matcher. GTEST_API_ bool IsInSet(char ch, const char* str); GTEST_API_ bool IsAsciiDigit(char ch); GTEST_API_ bool IsAsciiPunct(char ch); GTEST_API_ bool IsRepeat(char ch); GTEST_API_ bool IsAsciiWhiteSpace(char ch); GTEST_API_ bool IsAsciiWordChar(char ch); GTEST_API_ bool IsValidEscape(char ch); GTEST_API_ bool AtomMatchesChar(bool escaped, char pattern, char ch); GTEST_API_ bool ValidateRegex(const char* regex); GTEST_API_ bool MatchRegexAtHead(const char* regex, const char* str); GTEST_API_ bool MatchRepetitionAndRegexAtHead( bool escaped, char ch, char repeat, const char* regex, const char* str); GTEST_API_ bool MatchRegexAnywhere(const char* regex, const char* str); #endif // GTEST_USES_SIMPLE_RE // Parses the command line for Google Test flags, without initializing // other parts of Google Test. GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, char** argv); GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv); #if GTEST_HAS_DEATH_TEST // Returns the message describing the last system error, regardless of the // platform. GTEST_API_ std::string GetLastErrnoDescription(); // Attempts to parse a string into a positive integer pointed to by the // number parameter. Returns true if that is possible. // GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we can use // it here. template bool ParseNaturalNumber(const ::std::string& str, Integer* number) { // Fail fast if the given string does not begin with a digit; // this bypasses strtoXXX's "optional leading whitespace and plus // or minus sign" semantics, which are undesirable here. if (str.empty() || !IsDigit(str[0])) { return false; } errno = 0; char* end; // BiggestConvertible is the largest integer type that system-provided // string-to-number conversion routines can return. # if GTEST_OS_WINDOWS && !defined(__GNUC__) // MSVC and C++ Builder define __int64 instead of the standard long long. typedef unsigned __int64 BiggestConvertible; const BiggestConvertible parsed = _strtoui64(str.c_str(), &end, 10); # else typedef unsigned long long BiggestConvertible; // NOLINT const BiggestConvertible parsed = strtoull(str.c_str(), &end, 10); # endif // GTEST_OS_WINDOWS && !defined(__GNUC__) const bool parse_success = *end == '\0' && errno == 0; // TODO(vladl@google.com): Convert this to compile time assertion when it is // available. GTEST_CHECK_(sizeof(Integer) <= sizeof(parsed)); const Integer result = static_cast(parsed); if (parse_success && static_cast(result) == parsed) { *number = result; return true; } return false; } #endif // GTEST_HAS_DEATH_TEST // TestResult contains some private methods that should be hidden from // Google Test user but are required for testing. This class allow our tests // to access them. // // This class is supplied only for the purpose of testing Google Test's own // constructs. Do not use it in user tests, either directly or indirectly. class TestResultAccessor { public: static void RecordProperty(TestResult* test_result, const std::string& xml_element, const TestProperty& property) { test_result->RecordProperty(xml_element, property); } static void ClearTestPartResults(TestResult* test_result) { test_result->ClearTestPartResults(); } static const std::vector& test_part_results( const TestResult& test_result) { return test_result.test_part_results(); } }; #if GTEST_CAN_STREAM_RESULTS_ // Streams test results to the given port on the given host machine. class StreamingListener : public EmptyTestEventListener { public: // Abstract base class for writing strings to a socket. class AbstractSocketWriter { public: virtual ~AbstractSocketWriter() {} // Sends a string to the socket. virtual void Send(const string& message) = 0; // Closes the socket. virtual void CloseConnection() {} // Sends a string and a newline to the socket. void SendLn(const string& message) { Send(message + "\n"); } }; // Concrete class for actually writing strings to a socket. class SocketWriter : public AbstractSocketWriter { public: SocketWriter(const string& host, const string& port) : sockfd_(-1), host_name_(host), port_num_(port) { MakeConnection(); } virtual ~SocketWriter() { if (sockfd_ != -1) CloseConnection(); } // Sends a string to the socket. virtual void Send(const string& message) { GTEST_CHECK_(sockfd_ != -1) << "Send() can be called only when there is a connection."; const int len = static_cast(message.length()); if (write(sockfd_, message.c_str(), len) != len) { GTEST_LOG_(WARNING) << "stream_result_to: failed to stream to " << host_name_ << ":" << port_num_; } } private: // Creates a client socket and connects to the server. void MakeConnection(); // Closes the socket. void CloseConnection() { GTEST_CHECK_(sockfd_ != -1) << "CloseConnection() can be called only when there is a connection."; close(sockfd_); sockfd_ = -1; } int sockfd_; // socket file descriptor const string host_name_; const string port_num_; GTEST_DISALLOW_COPY_AND_ASSIGN_(SocketWriter); }; // class SocketWriter // Escapes '=', '&', '%', and '\n' characters in str as "%xx". static string UrlEncode(const char* str); StreamingListener(const string& host, const string& port) : socket_writer_(new SocketWriter(host, port)) { Start(); } explicit StreamingListener(AbstractSocketWriter* socket_writer) : socket_writer_(socket_writer) { Start(); } void OnTestProgramStart(const UnitTest& /* unit_test */) { SendLn("event=TestProgramStart"); } void OnTestProgramEnd(const UnitTest& unit_test) { // Note that Google Test current only report elapsed time for each // test iteration, not for the entire test program. SendLn("event=TestProgramEnd&passed=" + FormatBool(unit_test.Passed())); // Notify the streaming server to stop. socket_writer_->CloseConnection(); } void OnTestIterationStart(const UnitTest& /* unit_test */, int iteration) { SendLn("event=TestIterationStart&iteration=" + StreamableToString(iteration)); } void OnTestIterationEnd(const UnitTest& unit_test, int /* iteration */) { SendLn("event=TestIterationEnd&passed=" + FormatBool(unit_test.Passed()) + "&elapsed_time=" + StreamableToString(unit_test.elapsed_time()) + "ms"); } void OnTestCaseStart(const TestCase& test_case) { SendLn(std::string("event=TestCaseStart&name=") + test_case.name()); } void OnTestCaseEnd(const TestCase& test_case) { SendLn("event=TestCaseEnd&passed=" + FormatBool(test_case.Passed()) + "&elapsed_time=" + StreamableToString(test_case.elapsed_time()) + "ms"); } void OnTestStart(const TestInfo& test_info) { SendLn(std::string("event=TestStart&name=") + test_info.name()); } void OnTestEnd(const TestInfo& test_info) { SendLn("event=TestEnd&passed=" + FormatBool((test_info.result())->Passed()) + "&elapsed_time=" + StreamableToString((test_info.result())->elapsed_time()) + "ms"); } void OnTestPartResult(const TestPartResult& test_part_result) { const char* file_name = test_part_result.file_name(); if (file_name == NULL) file_name = ""; SendLn("event=TestPartResult&file=" + UrlEncode(file_name) + "&line=" + StreamableToString(test_part_result.line_number()) + "&message=" + UrlEncode(test_part_result.message())); } private: // Sends the given message and a newline to the socket. void SendLn(const string& message) { socket_writer_->SendLn(message); } // Called at the start of streaming to notify the receiver what // protocol we are using. void Start() { SendLn("gtest_streaming_protocol_version=1.0"); } string FormatBool(bool value) { return value ? "1" : "0"; } const scoped_ptr socket_writer_; GTEST_DISALLOW_COPY_AND_ASSIGN_(StreamingListener); }; // class StreamingListener #endif // GTEST_CAN_STREAM_RESULTS_ } // namespace internal } // namespace testing #endif // GTEST_SRC_GTEST_INTERNAL_INL_H_ #undef GTEST_IMPLEMENTATION_ #if GTEST_OS_WINDOWS # define vsnprintf _vsnprintf #endif // GTEST_OS_WINDOWS namespace testing { using internal::CountIf; using internal::ForEach; using internal::GetElementOr; using internal::Shuffle; // Constants. // A test whose test case name or test name matches this filter is // disabled and not run. static const char kDisableTestFilter[] = "DISABLED_*:*/DISABLED_*"; // A test case whose name matches this filter is considered a death // test case and will be run before test cases whose name doesn't // match this filter. static const char kDeathTestCaseFilter[] = "*DeathTest:*DeathTest/*"; // A test filter that matches everything. static const char kUniversalFilter[] = "*"; // The default output file for XML output. static const char kDefaultOutputFile[] = "test_detail.xml"; // The environment variable name for the test shard index. static const char kTestShardIndex[] = "GTEST_SHARD_INDEX"; // The environment variable name for the total number of test shards. static const char kTestTotalShards[] = "GTEST_TOTAL_SHARDS"; // The environment variable name for the test shard status file. static const char kTestShardStatusFile[] = "GTEST_SHARD_STATUS_FILE"; namespace internal { // The text used in failure messages to indicate the start of the // stack trace. const char kStackTraceMarker[] = "\nStack trace:\n"; // g_help_flag is true iff the --help flag or an equivalent form is // specified on the command line. bool g_help_flag = false; } // namespace internal static const char* GetDefaultFilter() { return kUniversalFilter; } GTEST_DEFINE_bool_( also_run_disabled_tests, internal::BoolFromGTestEnv("also_run_disabled_tests", false), "Run disabled tests too, in addition to the tests normally being run."); GTEST_DEFINE_bool_( break_on_failure, internal::BoolFromGTestEnv("break_on_failure", false), "True iff a failed assertion should be a debugger break-point."); GTEST_DEFINE_bool_( catch_exceptions, internal::BoolFromGTestEnv("catch_exceptions", true), "True iff " GTEST_NAME_ " should catch exceptions and treat them as test failures."); GTEST_DEFINE_string_( color, internal::StringFromGTestEnv("color", "auto"), "Whether to use colors in the output. Valid values: yes, no, " "and auto. 'auto' means to use colors if the output is " "being sent to a terminal and the TERM environment variable " "is set to a terminal type that supports colors."); GTEST_DEFINE_string_( filter, internal::StringFromGTestEnv("filter", GetDefaultFilter()), "A colon-separated list of glob (not regex) patterns " "for filtering the tests to run, optionally followed by a " "'-' and a : separated list of negative patterns (tests to " "exclude). A test is run if it matches one of the positive " "patterns and does not match any of the negative patterns."); GTEST_DEFINE_bool_(list_tests, false, "List all tests without running them."); GTEST_DEFINE_string_( output, internal::StringFromGTestEnv("output", ""), "A format (currently must be \"xml\"), optionally followed " "by a colon and an output file name or directory. A directory " "is indicated by a trailing pathname separator. " "Examples: \"xml:filename.xml\", \"xml::directoryname/\". " "If a directory is specified, output files will be created " "within that directory, with file-names based on the test " "executable's name and, if necessary, made unique by adding " "digits."); GTEST_DEFINE_bool_( print_time, internal::BoolFromGTestEnv("print_time", true), "True iff " GTEST_NAME_ " should display elapsed time in text output."); GTEST_DEFINE_int32_( random_seed, internal::Int32FromGTestEnv("random_seed", 0), "Random number seed to use when shuffling test orders. Must be in range " "[1, 99999], or 0 to use a seed based on the current time."); GTEST_DEFINE_int32_( repeat, internal::Int32FromGTestEnv("repeat", 1), "How many times to repeat each test. Specify a negative number " "for repeating forever. Useful for shaking out flaky tests."); GTEST_DEFINE_bool_( show_internal_stack_frames, false, "True iff " GTEST_NAME_ " should include internal stack frames when " "printing test failure stack traces."); GTEST_DEFINE_bool_( shuffle, internal::BoolFromGTestEnv("shuffle", false), "True iff " GTEST_NAME_ " should randomize tests' order on every run."); GTEST_DEFINE_int32_( stack_trace_depth, internal::Int32FromGTestEnv("stack_trace_depth", kMaxStackTraceDepth), "The maximum number of stack frames to print when an " "assertion fails. The valid range is 0 through 100, inclusive."); GTEST_DEFINE_string_( stream_result_to, internal::StringFromGTestEnv("stream_result_to", ""), "This flag specifies the host name and the port number on which to stream " "test results. Example: \"localhost:555\". The flag is effective only on " "Linux."); GTEST_DEFINE_bool_( throw_on_failure, internal::BoolFromGTestEnv("throw_on_failure", false), "When this flag is specified, a failed assertion will throw an exception " "if exceptions are enabled or exit the program with a non-zero code " "otherwise."); namespace internal { // Generates a random number from [0, range), using a Linear // Congruential Generator (LCG). Crashes if 'range' is 0 or greater // than kMaxRange. UInt32 Random::Generate(UInt32 range) { // These constants are the same as are used in glibc's rand(3). state_ = (1103515245U*state_ + 12345U) % kMaxRange; GTEST_CHECK_(range > 0) << "Cannot generate a number in the range [0, 0)."; GTEST_CHECK_(range <= kMaxRange) << "Generation of a number in [0, " << range << ") was requested, " << "but this can only generate numbers in [0, " << kMaxRange << ")."; // Converting via modulus introduces a bit of downward bias, but // it's simple, and a linear congruential generator isn't too good // to begin with. return state_ % range; } // GTestIsInitialized() returns true iff the user has initialized // Google Test. Useful for catching the user mistake of not initializing // Google Test before calling RUN_ALL_TESTS(). // // A user must call testing::InitGoogleTest() to initialize Google // Test. g_init_gtest_count is set to the number of times // InitGoogleTest() has been called. We don't protect this variable // under a mutex as it is only accessed in the main thread. GTEST_API_ int g_init_gtest_count = 0; static bool GTestIsInitialized() { return g_init_gtest_count != 0; } // Iterates over a vector of TestCases, keeping a running sum of the // results of calling a given int-returning method on each. // Returns the sum. static int SumOverTestCaseList(const std::vector& case_list, int (TestCase::*method)() const) { int sum = 0; for (size_t i = 0; i < case_list.size(); i++) { sum += (case_list[i]->*method)(); } return sum; } // Returns true iff the test case passed. static bool TestCasePassed(const TestCase* test_case) { return test_case->should_run() && test_case->Passed(); } // Returns true iff the test case failed. static bool TestCaseFailed(const TestCase* test_case) { return test_case->should_run() && test_case->Failed(); } // Returns true iff test_case contains at least one test that should // run. static bool ShouldRunTestCase(const TestCase* test_case) { return test_case->should_run(); } // AssertHelper constructor. AssertHelper::AssertHelper(TestPartResult::Type type, const char* file, int line, const char* message) : data_(new AssertHelperData(type, file, line, message)) { } AssertHelper::~AssertHelper() { delete data_; } // Message assignment, for assertion streaming support. void AssertHelper::operator=(const Message& message) const { UnitTest::GetInstance()-> AddTestPartResult(data_->type, data_->file, data_->line, AppendUserMessage(data_->message, message), UnitTest::GetInstance()->impl() ->CurrentOsStackTraceExceptTop(1) // Skips the stack frame for this function itself. ); // NOLINT } // Mutex for linked pointers. GTEST_API_ GTEST_DEFINE_STATIC_MUTEX_(g_linked_ptr_mutex); // Application pathname gotten in InitGoogleTest. std::string g_executable_path; // Returns the current application's name, removing directory path if that // is present. FilePath GetCurrentExecutableName() { FilePath result; #if GTEST_OS_WINDOWS result.Set(FilePath(g_executable_path).RemoveExtension("exe")); #else result.Set(FilePath(g_executable_path)); #endif // GTEST_OS_WINDOWS return result.RemoveDirectoryName(); } // Functions for processing the gtest_output flag. // Returns the output format, or "" for normal printed output. std::string UnitTestOptions::GetOutputFormat() { const char* const gtest_output_flag = GTEST_FLAG(output).c_str(); if (gtest_output_flag == NULL) return std::string(""); const char* const colon = strchr(gtest_output_flag, ':'); return (colon == NULL) ? std::string(gtest_output_flag) : std::string(gtest_output_flag, colon - gtest_output_flag); } // Returns the name of the requested output file, or the default if none // was explicitly specified. std::string UnitTestOptions::GetAbsolutePathToOutputFile() { const char* const gtest_output_flag = GTEST_FLAG(output).c_str(); if (gtest_output_flag == NULL) return ""; const char* const colon = strchr(gtest_output_flag, ':'); if (colon == NULL) return internal::FilePath::ConcatPaths( internal::FilePath( UnitTest::GetInstance()->original_working_dir()), internal::FilePath(kDefaultOutputFile)).string(); internal::FilePath output_name(colon + 1); if (!output_name.IsAbsolutePath()) // TODO(wan@google.com): on Windows \some\path is not an absolute // path (as its meaning depends on the current drive), yet the // following logic for turning it into an absolute path is wrong. // Fix it. output_name = internal::FilePath::ConcatPaths( internal::FilePath(UnitTest::GetInstance()->original_working_dir()), internal::FilePath(colon + 1)); if (!output_name.IsDirectory()) return output_name.string(); internal::FilePath result(internal::FilePath::GenerateUniqueFileName( output_name, internal::GetCurrentExecutableName(), GetOutputFormat().c_str())); return result.string(); } // Returns true iff the wildcard pattern matches the string. The // first ':' or '\0' character in pattern marks the end of it. // // This recursive algorithm isn't very efficient, but is clear and // works well enough for matching test names, which are short. bool UnitTestOptions::PatternMatchesString(const char *pattern, const char *str) { switch (*pattern) { case '\0': case ':': // Either ':' or '\0' marks the end of the pattern. return *str == '\0'; case '?': // Matches any single character. return *str != '\0' && PatternMatchesString(pattern + 1, str + 1); case '*': // Matches any string (possibly empty) of characters. return (*str != '\0' && PatternMatchesString(pattern, str + 1)) || PatternMatchesString(pattern + 1, str); default: // Non-special character. Matches itself. return *pattern == *str && PatternMatchesString(pattern + 1, str + 1); } } bool UnitTestOptions::MatchesFilter( const std::string& name, const char* filter) { const char *cur_pattern = filter; for (;;) { if (PatternMatchesString(cur_pattern, name.c_str())) { return true; } // Finds the next pattern in the filter. cur_pattern = strchr(cur_pattern, ':'); // Returns if no more pattern can be found. if (cur_pattern == NULL) { return false; } // Skips the pattern separater (the ':' character). cur_pattern++; } } // Returns true iff the user-specified filter matches the test case // name and the test name. bool UnitTestOptions::FilterMatchesTest(const std::string &test_case_name, const std::string &test_name) { const std::string& full_name = test_case_name + "." + test_name.c_str(); // Split --gtest_filter at '-', if there is one, to separate into // positive filter and negative filter portions const char* const p = GTEST_FLAG(filter).c_str(); const char* const dash = strchr(p, '-'); std::string positive; std::string negative; if (dash == NULL) { positive = GTEST_FLAG(filter).c_str(); // Whole string is a positive filter negative = ""; } else { positive = std::string(p, dash); // Everything up to the dash negative = std::string(dash + 1); // Everything after the dash if (positive.empty()) { // Treat '-test1' as the same as '*-test1' positive = kUniversalFilter; } } // A filter is a colon-separated list of patterns. It matches a // test if any pattern in it matches the test. return (MatchesFilter(full_name, positive.c_str()) && !MatchesFilter(full_name, negative.c_str())); } #if GTEST_HAS_SEH // Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the // given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise. // This function is useful as an __except condition. int UnitTestOptions::GTestShouldProcessSEH(DWORD exception_code) { // Google Test should handle a SEH exception if: // 1. the user wants it to, AND // 2. this is not a breakpoint exception, AND // 3. this is not a C++ exception (VC++ implements them via SEH, // apparently). // // SEH exception code for C++ exceptions. // (see http://support.microsoft.com/kb/185294 for more information). const DWORD kCxxExceptionCode = 0xe06d7363; bool should_handle = true; if (!GTEST_FLAG(catch_exceptions)) should_handle = false; else if (exception_code == EXCEPTION_BREAKPOINT) should_handle = false; else if (exception_code == kCxxExceptionCode) should_handle = false; return should_handle ? EXCEPTION_EXECUTE_HANDLER : EXCEPTION_CONTINUE_SEARCH; } #endif // GTEST_HAS_SEH } // namespace internal // The c'tor sets this object as the test part result reporter used by // Google Test. The 'result' parameter specifies where to report the // results. Intercepts only failures from the current thread. ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter( TestPartResultArray* result) : intercept_mode_(INTERCEPT_ONLY_CURRENT_THREAD), result_(result) { Init(); } // The c'tor sets this object as the test part result reporter used by // Google Test. The 'result' parameter specifies where to report the // results. ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter( InterceptMode intercept_mode, TestPartResultArray* result) : intercept_mode_(intercept_mode), result_(result) { Init(); } void ScopedFakeTestPartResultReporter::Init() { internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); if (intercept_mode_ == INTERCEPT_ALL_THREADS) { old_reporter_ = impl->GetGlobalTestPartResultReporter(); impl->SetGlobalTestPartResultReporter(this); } else { old_reporter_ = impl->GetTestPartResultReporterForCurrentThread(); impl->SetTestPartResultReporterForCurrentThread(this); } } // The d'tor restores the test part result reporter used by Google Test // before. ScopedFakeTestPartResultReporter::~ScopedFakeTestPartResultReporter() { internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); if (intercept_mode_ == INTERCEPT_ALL_THREADS) { impl->SetGlobalTestPartResultReporter(old_reporter_); } else { impl->SetTestPartResultReporterForCurrentThread(old_reporter_); } } // Increments the test part result count and remembers the result. // This method is from the TestPartResultReporterInterface interface. void ScopedFakeTestPartResultReporter::ReportTestPartResult( const TestPartResult& result) { result_->Append(result); } namespace internal { // Returns the type ID of ::testing::Test. We should always call this // instead of GetTypeId< ::testing::Test>() to get the type ID of // testing::Test. This is to work around a suspected linker bug when // using Google Test as a framework on Mac OS X. The bug causes // GetTypeId< ::testing::Test>() to return different values depending // on whether the call is from the Google Test framework itself or // from user test code. GetTestTypeId() is guaranteed to always // return the same value, as it always calls GetTypeId<>() from the // gtest.cc, which is within the Google Test framework. TypeId GetTestTypeId() { return GetTypeId(); } // The value of GetTestTypeId() as seen from within the Google Test // library. This is solely for testing GetTestTypeId(). extern const TypeId kTestTypeIdInGoogleTest = GetTestTypeId(); // This predicate-formatter checks that 'results' contains a test part // failure of the given type and that the failure message contains the // given substring. AssertionResult HasOneFailure(const char* /* results_expr */, const char* /* type_expr */, const char* /* substr_expr */, const TestPartResultArray& results, TestPartResult::Type type, const string& substr) { const std::string expected(type == TestPartResult::kFatalFailure ? "1 fatal failure" : "1 non-fatal failure"); Message msg; if (results.size() != 1) { msg << "Expected: " << expected << "\n" << " Actual: " << results.size() << " failures"; for (int i = 0; i < results.size(); i++) { msg << "\n" << results.GetTestPartResult(i); } return AssertionFailure() << msg; } const TestPartResult& r = results.GetTestPartResult(0); if (r.type() != type) { return AssertionFailure() << "Expected: " << expected << "\n" << " Actual:\n" << r; } if (strstr(r.message(), substr.c_str()) == NULL) { return AssertionFailure() << "Expected: " << expected << " containing \"" << substr << "\"\n" << " Actual:\n" << r; } return AssertionSuccess(); } // The constructor of SingleFailureChecker remembers where to look up // test part results, what type of failure we expect, and what // substring the failure message should contain. SingleFailureChecker:: SingleFailureChecker( const TestPartResultArray* results, TestPartResult::Type type, const string& substr) : results_(results), type_(type), substr_(substr) {} // The destructor of SingleFailureChecker verifies that the given // TestPartResultArray contains exactly one failure that has the given // type and contains the given substring. If that's not the case, a // non-fatal failure will be generated. SingleFailureChecker::~SingleFailureChecker() { EXPECT_PRED_FORMAT3(HasOneFailure, *results_, type_, substr_); } DefaultGlobalTestPartResultReporter::DefaultGlobalTestPartResultReporter( UnitTestImpl* unit_test) : unit_test_(unit_test) {} void DefaultGlobalTestPartResultReporter::ReportTestPartResult( const TestPartResult& result) { unit_test_->current_test_result()->AddTestPartResult(result); unit_test_->listeners()->repeater()->OnTestPartResult(result); } DefaultPerThreadTestPartResultReporter::DefaultPerThreadTestPartResultReporter( UnitTestImpl* unit_test) : unit_test_(unit_test) {} void DefaultPerThreadTestPartResultReporter::ReportTestPartResult( const TestPartResult& result) { unit_test_->GetGlobalTestPartResultReporter()->ReportTestPartResult(result); } // Returns the global test part result reporter. TestPartResultReporterInterface* UnitTestImpl::GetGlobalTestPartResultReporter() { internal::MutexLock lock(&global_test_part_result_reporter_mutex_); return global_test_part_result_repoter_; } // Sets the global test part result reporter. void UnitTestImpl::SetGlobalTestPartResultReporter( TestPartResultReporterInterface* reporter) { internal::MutexLock lock(&global_test_part_result_reporter_mutex_); global_test_part_result_repoter_ = reporter; } // Returns the test part result reporter for the current thread. TestPartResultReporterInterface* UnitTestImpl::GetTestPartResultReporterForCurrentThread() { return per_thread_test_part_result_reporter_.get(); } // Sets the test part result reporter for the current thread. void UnitTestImpl::SetTestPartResultReporterForCurrentThread( TestPartResultReporterInterface* reporter) { per_thread_test_part_result_reporter_.set(reporter); } // Gets the number of successful test cases. int UnitTestImpl::successful_test_case_count() const { return CountIf(test_cases_, TestCasePassed); } // Gets the number of failed test cases. int UnitTestImpl::failed_test_case_count() const { return CountIf(test_cases_, TestCaseFailed); } // Gets the number of all test cases. int UnitTestImpl::total_test_case_count() const { return static_cast(test_cases_.size()); } // Gets the number of all test cases that contain at least one test // that should run. int UnitTestImpl::test_case_to_run_count() const { return CountIf(test_cases_, ShouldRunTestCase); } // Gets the number of successful tests. int UnitTestImpl::successful_test_count() const { return SumOverTestCaseList(test_cases_, &TestCase::successful_test_count); } // Gets the number of failed tests. int UnitTestImpl::failed_test_count() const { return SumOverTestCaseList(test_cases_, &TestCase::failed_test_count); } // Gets the number of disabled tests that will be reported in the XML report. int UnitTestImpl::reportable_disabled_test_count() const { return SumOverTestCaseList(test_cases_, &TestCase::reportable_disabled_test_count); } // Gets the number of disabled tests. int UnitTestImpl::disabled_test_count() const { return SumOverTestCaseList(test_cases_, &TestCase::disabled_test_count); } // Gets the number of tests to be printed in the XML report. int UnitTestImpl::reportable_test_count() const { return SumOverTestCaseList(test_cases_, &TestCase::reportable_test_count); } // Gets the number of all tests. int UnitTestImpl::total_test_count() const { return SumOverTestCaseList(test_cases_, &TestCase::total_test_count); } // Gets the number of tests that should run. int UnitTestImpl::test_to_run_count() const { return SumOverTestCaseList(test_cases_, &TestCase::test_to_run_count); } // Returns the current OS stack trace as an std::string. // // The maximum number of stack frames to be included is specified by // the gtest_stack_trace_depth flag. The skip_count parameter // specifies the number of top frames to be skipped, which doesn't // count against the number of frames to be included. // // For example, if Foo() calls Bar(), which in turn calls // CurrentOsStackTraceExceptTop(1), Foo() will be included in the // trace but Bar() and CurrentOsStackTraceExceptTop() won't. std::string UnitTestImpl::CurrentOsStackTraceExceptTop(int skip_count) { (void)skip_count; return ""; } // Returns the current time in milliseconds. TimeInMillis GetTimeInMillis() { #if GTEST_OS_WINDOWS_MOBILE || defined(__BORLANDC__) // Difference between 1970-01-01 and 1601-01-01 in milliseconds. // http://analogous.blogspot.com/2005/04/epoch.html const TimeInMillis kJavaEpochToWinFileTimeDelta = static_cast(116444736UL) * 100000UL; const DWORD kTenthMicrosInMilliSecond = 10000; SYSTEMTIME now_systime; FILETIME now_filetime; ULARGE_INTEGER now_int64; // TODO(kenton@google.com): Shouldn't this just use // GetSystemTimeAsFileTime()? GetSystemTime(&now_systime); if (SystemTimeToFileTime(&now_systime, &now_filetime)) { now_int64.LowPart = now_filetime.dwLowDateTime; now_int64.HighPart = now_filetime.dwHighDateTime; now_int64.QuadPart = (now_int64.QuadPart / kTenthMicrosInMilliSecond) - kJavaEpochToWinFileTimeDelta; return now_int64.QuadPart; } return 0; #elif GTEST_OS_WINDOWS && !GTEST_HAS_GETTIMEOFDAY_ __timeb64 now; // MSVC 8 deprecates _ftime64(), so we want to suppress warning 4996 // (deprecated function) there. // TODO(kenton@google.com): Use GetTickCount()? Or use // SystemTimeToFileTime() GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996) _ftime64(&now); GTEST_DISABLE_MSC_WARNINGS_POP_() return static_cast(now.time) * 1000 + now.millitm; #elif GTEST_HAS_GETTIMEOFDAY_ struct timeval now; gettimeofday(&now, NULL); return static_cast(now.tv_sec) * 1000 + now.tv_usec / 1000; #else # error "Don't know how to get the current time on your system." #endif } // Utilities // class String. #if GTEST_OS_WINDOWS_MOBILE // Creates a UTF-16 wide string from the given ANSI string, allocating // memory using new. The caller is responsible for deleting the return // value using delete[]. Returns the wide string, or NULL if the // input is NULL. LPCWSTR String::AnsiToUtf16(const char* ansi) { if (!ansi) return NULL; const int length = strlen(ansi); const int unicode_length = MultiByteToWideChar(CP_ACP, 0, ansi, length, NULL, 0); WCHAR* unicode = new WCHAR[unicode_length + 1]; MultiByteToWideChar(CP_ACP, 0, ansi, length, unicode, unicode_length); unicode[unicode_length] = 0; return unicode; } // Creates an ANSI string from the given wide string, allocating // memory using new. The caller is responsible for deleting the return // value using delete[]. Returns the ANSI string, or NULL if the // input is NULL. const char* String::Utf16ToAnsi(LPCWSTR utf16_str) { if (!utf16_str) return NULL; const int ansi_length = WideCharToMultiByte(CP_ACP, 0, utf16_str, -1, NULL, 0, NULL, NULL); char* ansi = new char[ansi_length + 1]; WideCharToMultiByte(CP_ACP, 0, utf16_str, -1, ansi, ansi_length, NULL, NULL); ansi[ansi_length] = 0; return ansi; } #endif // GTEST_OS_WINDOWS_MOBILE // Compares two C strings. Returns true iff they have the same content. // // Unlike strcmp(), this function can handle NULL argument(s). A NULL // C string is considered different to any non-NULL C string, // including the empty string. bool String::CStringEquals(const char * lhs, const char * rhs) { if ( lhs == NULL ) return rhs == NULL; if ( rhs == NULL ) return false; return strcmp(lhs, rhs) == 0; } #if GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING // Converts an array of wide chars to a narrow string using the UTF-8 // encoding, and streams the result to the given Message object. static void StreamWideCharsToMessage(const wchar_t* wstr, size_t length, Message* msg) { for (size_t i = 0; i != length; ) { // NOLINT if (wstr[i] != L'\0') { *msg << WideStringToUtf8(wstr + i, static_cast(length - i)); while (i != length && wstr[i] != L'\0') i++; } else { *msg << '\0'; i++; } } } #endif // GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING } // namespace internal // Constructs an empty Message. // We allocate the stringstream separately because otherwise each use of // ASSERT/EXPECT in a procedure adds over 200 bytes to the procedure's // stack frame leading to huge stack frames in some cases; gcc does not reuse // the stack space. Message::Message() : ss_(new ::std::stringstream) { // By default, we want there to be enough precision when printing // a double to a Message. *ss_ << std::setprecision(std::numeric_limits::digits10 + 2); } // These two overloads allow streaming a wide C string to a Message // using the UTF-8 encoding. Message& Message::operator <<(const wchar_t* wide_c_str) { return *this << internal::String::ShowWideCString(wide_c_str); } Message& Message::operator <<(wchar_t* wide_c_str) { return *this << internal::String::ShowWideCString(wide_c_str); } #if GTEST_HAS_STD_WSTRING // Converts the given wide string to a narrow string using the UTF-8 // encoding, and streams the result to this Message object. Message& Message::operator <<(const ::std::wstring& wstr) { internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this); return *this; } #endif // GTEST_HAS_STD_WSTRING #if GTEST_HAS_GLOBAL_WSTRING // Converts the given wide string to a narrow string using the UTF-8 // encoding, and streams the result to this Message object. Message& Message::operator <<(const ::wstring& wstr) { internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this); return *this; } #endif // GTEST_HAS_GLOBAL_WSTRING // Gets the text streamed to this object so far as an std::string. // Each '\0' character in the buffer is replaced with "\\0". std::string Message::GetString() const { return internal::StringStreamToString(ss_.get()); } // AssertionResult constructors. // Used in EXPECT_TRUE/FALSE(assertion_result). AssertionResult::AssertionResult(const AssertionResult& other) : success_(other.success_), message_(other.message_.get() != NULL ? new ::std::string(*other.message_) : static_cast< ::std::string*>(NULL)) { } // Swaps two AssertionResults. void AssertionResult::swap(AssertionResult& other) { using std::swap; swap(success_, other.success_); swap(message_, other.message_); } // Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE. AssertionResult AssertionResult::operator!() const { AssertionResult negation(!success_); if (message_.get() != NULL) negation << *message_; return negation; } // Makes a successful assertion result. AssertionResult AssertionSuccess() { return AssertionResult(true); } // Makes a failed assertion result. AssertionResult AssertionFailure() { return AssertionResult(false); } // Makes a failed assertion result with the given failure message. // Deprecated; use AssertionFailure() << message. AssertionResult AssertionFailure(const Message& message) { return AssertionFailure() << message; } namespace internal { namespace edit_distance { std::vector CalculateOptimalEdits(const std::vector& left, const std::vector& right) { std::vector > costs( left.size() + 1, std::vector(right.size() + 1)); std::vector > best_move( left.size() + 1, std::vector(right.size() + 1)); // Populate for empty right. for (size_t l_i = 0; l_i < costs.size(); ++l_i) { costs[l_i][0] = static_cast(l_i); best_move[l_i][0] = kRemove; } // Populate for empty left. for (size_t r_i = 1; r_i < costs[0].size(); ++r_i) { costs[0][r_i] = static_cast(r_i); best_move[0][r_i] = kAdd; } for (size_t l_i = 0; l_i < left.size(); ++l_i) { for (size_t r_i = 0; r_i < right.size(); ++r_i) { if (left[l_i] == right[r_i]) { // Found a match. Consume it. costs[l_i + 1][r_i + 1] = costs[l_i][r_i]; best_move[l_i + 1][r_i + 1] = kMatch; continue; } const double add = costs[l_i + 1][r_i]; const double remove = costs[l_i][r_i + 1]; const double replace = costs[l_i][r_i]; if (add < remove && add < replace) { costs[l_i + 1][r_i + 1] = add + 1; best_move[l_i + 1][r_i + 1] = kAdd; } else if (remove < add && remove < replace) { costs[l_i + 1][r_i + 1] = remove + 1; best_move[l_i + 1][r_i + 1] = kRemove; } else { // We make replace a little more expensive than add/remove to lower // their priority. costs[l_i + 1][r_i + 1] = replace + 1.00001; best_move[l_i + 1][r_i + 1] = kReplace; } } } // Reconstruct the best path. We do it in reverse order. std::vector best_path; for (size_t l_i = left.size(), r_i = right.size(); l_i > 0 || r_i > 0;) { EditType move = best_move[l_i][r_i]; best_path.push_back(move); l_i -= move != kAdd; r_i -= move != kRemove; } std::reverse(best_path.begin(), best_path.end()); return best_path; } namespace { // Helper class to convert string into ids with deduplication. class InternalStrings { public: size_t GetId(const std::string& str) { IdMap::iterator it = ids_.find(str); if (it != ids_.end()) return it->second; size_t id = ids_.size(); return ids_[str] = id; } private: typedef std::map IdMap; IdMap ids_; }; } // namespace std::vector CalculateOptimalEdits( const std::vector& left, const std::vector& right) { std::vector left_ids, right_ids; { InternalStrings intern_table; for (size_t i = 0; i < left.size(); ++i) { left_ids.push_back(intern_table.GetId(left[i])); } for (size_t i = 0; i < right.size(); ++i) { right_ids.push_back(intern_table.GetId(right[i])); } } return CalculateOptimalEdits(left_ids, right_ids); } namespace { // Helper class that holds the state for one hunk and prints it out to the // stream. // It reorders adds/removes when possible to group all removes before all // adds. It also adds the hunk header before printint into the stream. class Hunk { public: Hunk(size_t left_start, size_t right_start) : left_start_(left_start), right_start_(right_start), adds_(), removes_(), common_() {} void PushLine(char edit, const char* line) { switch (edit) { case ' ': ++common_; FlushEdits(); hunk_.push_back(std::make_pair(' ', line)); break; case '-': ++removes_; hunk_removes_.push_back(std::make_pair('-', line)); break; case '+': ++adds_; hunk_adds_.push_back(std::make_pair('+', line)); break; } } void PrintTo(std::ostream* os) { PrintHeader(os); FlushEdits(); for (std::list >::const_iterator it = hunk_.begin(); it != hunk_.end(); ++it) { *os << it->first << it->second << "\n"; } } bool has_edits() const { return adds_ || removes_; } private: void FlushEdits() { hunk_.splice(hunk_.end(), hunk_removes_); hunk_.splice(hunk_.end(), hunk_adds_); } // Print a unified diff header for one hunk. // The format is // "@@ -, +, @@" // where the left/right parts are omitted if unnecessary. void PrintHeader(std::ostream* ss) const { *ss << "@@ "; if (removes_) { *ss << "-" << left_start_ << "," << (removes_ + common_); } if (removes_ && adds_) { *ss << " "; } if (adds_) { *ss << "+" << right_start_ << "," << (adds_ + common_); } *ss << " @@\n"; } size_t left_start_, right_start_; size_t adds_, removes_, common_; std::list > hunk_, hunk_adds_, hunk_removes_; }; } // namespace // Create a list of diff hunks in Unified diff format. // Each hunk has a header generated by PrintHeader above plus a body with // lines prefixed with ' ' for no change, '-' for deletion and '+' for // addition. // 'context' represents the desired unchanged prefix/suffix around the diff. // If two hunks are close enough that their contexts overlap, then they are // joined into one hunk. std::string CreateUnifiedDiff(const std::vector& left, const std::vector& right, size_t context) { const std::vector edits = CalculateOptimalEdits(left, right); size_t l_i = 0, r_i = 0, edit_i = 0; std::stringstream ss; while (edit_i < edits.size()) { // Find first edit. while (edit_i < edits.size() && edits[edit_i] == kMatch) { ++l_i; ++r_i; ++edit_i; } // Find the first line to include in the hunk. const size_t prefix_context = std::min(l_i, context); Hunk hunk(l_i - prefix_context + 1, r_i - prefix_context + 1); for (size_t i = prefix_context; i > 0; --i) { hunk.PushLine(' ', left[l_i - i].c_str()); } // Iterate the edits until we found enough suffix for the hunk or the input // is over. size_t n_suffix = 0; for (; edit_i < edits.size(); ++edit_i) { if (n_suffix >= context) { // Continue only if the next hunk is very close. std::vector::const_iterator it = edits.begin() + edit_i; while (it != edits.end() && *it == kMatch) ++it; if (it == edits.end() || (it - edits.begin()) - edit_i >= context) { // There is no next edit or it is too far away. break; } } EditType edit = edits[edit_i]; // Reset count when a non match is found. n_suffix = edit == kMatch ? n_suffix + 1 : 0; if (edit == kMatch || edit == kRemove || edit == kReplace) { hunk.PushLine(edit == kMatch ? ' ' : '-', left[l_i].c_str()); } if (edit == kAdd || edit == kReplace) { hunk.PushLine('+', right[r_i].c_str()); } // Advance indices, depending on edit type. l_i += edit != kAdd; r_i += edit != kRemove; } if (!hunk.has_edits()) { // We are done. We don't want this hunk. break; } hunk.PrintTo(&ss); } return ss.str(); } } // namespace edit_distance namespace { // The string representation of the values received in EqFailure() are already // escaped. Split them on escaped '\n' boundaries. Leave all other escaped // characters the same. std::vector SplitEscapedString(const std::string& str) { std::vector lines; size_t start = 0, end = str.size(); if (end > 2 && str[0] == '"' && str[end - 1] == '"') { ++start; --end; } bool escaped = false; for (size_t i = start; i + 1 < end; ++i) { if (escaped) { escaped = false; if (str[i] == 'n') { lines.push_back(str.substr(start, i - start - 1)); start = i + 1; } } else { escaped = str[i] == '\\'; } } lines.push_back(str.substr(start, end - start)); return lines; } } // namespace // Constructs and returns the message for an equality assertion // (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure. // // The first four parameters are the expressions used in the assertion // and their values, as strings. For example, for ASSERT_EQ(foo, bar) // where foo is 5 and bar is 6, we have: // // expected_expression: "foo" // actual_expression: "bar" // expected_value: "5" // actual_value: "6" // // The ignoring_case parameter is true iff the assertion is a // *_STRCASEEQ*. When it's true, the string " (ignoring case)" will // be inserted into the message. AssertionResult EqFailure(const char* expected_expression, const char* actual_expression, const std::string& expected_value, const std::string& actual_value, bool ignoring_case) { Message msg; msg << "Value of: " << actual_expression; if (actual_value != actual_expression) { msg << "\n Actual: " << actual_value; } msg << "\nExpected: " << expected_expression; if (ignoring_case) { msg << " (ignoring case)"; } if (expected_value != expected_expression) { msg << "\nWhich is: " << expected_value; } if (!expected_value.empty() && !actual_value.empty()) { const std::vector expected_lines = SplitEscapedString(expected_value); const std::vector actual_lines = SplitEscapedString(actual_value); if (expected_lines.size() > 1 || actual_lines.size() > 1) { msg << "\nWith diff:\n" << edit_distance::CreateUnifiedDiff(expected_lines, actual_lines); } } return AssertionFailure() << msg; } // Constructs a failure message for Boolean assertions such as EXPECT_TRUE. std::string GetBoolAssertionFailureMessage( const AssertionResult& assertion_result, const char* expression_text, const char* actual_predicate_value, const char* expected_predicate_value) { const char* actual_message = assertion_result.message(); Message msg; msg << "Value of: " << expression_text << "\n Actual: " << actual_predicate_value; if (actual_message[0] != '\0') msg << " (" << actual_message << ")"; msg << "\nExpected: " << expected_predicate_value; return msg.GetString(); } // Helper function for implementing ASSERT_NEAR. AssertionResult DoubleNearPredFormat(const char* expr1, const char* expr2, const char* abs_error_expr, double val1, double val2, double abs_error) { const double diff = fabs(val1 - val2); if (diff <= abs_error) return AssertionSuccess(); // TODO(wan): do not print the value of an expression if it's // already a literal. return AssertionFailure() << "The difference between " << expr1 << " and " << expr2 << " is " << diff << ", which exceeds " << abs_error_expr << ", where\n" << expr1 << " evaluates to " << val1 << ",\n" << expr2 << " evaluates to " << val2 << ", and\n" << abs_error_expr << " evaluates to " << abs_error << "."; } // Helper template for implementing FloatLE() and DoubleLE(). template AssertionResult FloatingPointLE(const char* expr1, const char* expr2, RawType val1, RawType val2) { // Returns success if val1 is less than val2, if (val1 < val2) { return AssertionSuccess(); } // or if val1 is almost equal to val2. const FloatingPoint lhs(val1), rhs(val2); if (lhs.AlmostEquals(rhs)) { return AssertionSuccess(); } // Note that the above two checks will both fail if either val1 or // val2 is NaN, as the IEEE floating-point standard requires that // any predicate involving a NaN must return false. ::std::stringstream val1_ss; val1_ss << std::setprecision(std::numeric_limits::digits10 + 2) << val1; ::std::stringstream val2_ss; val2_ss << std::setprecision(std::numeric_limits::digits10 + 2) << val2; return AssertionFailure() << "Expected: (" << expr1 << ") <= (" << expr2 << ")\n" << " Actual: " << StringStreamToString(&val1_ss) << " vs " << StringStreamToString(&val2_ss); } } // namespace internal // Asserts that val1 is less than, or almost equal to, val2. Fails // otherwise. In particular, it fails if either val1 or val2 is NaN. AssertionResult FloatLE(const char* expr1, const char* expr2, float val1, float val2) { return internal::FloatingPointLE(expr1, expr2, val1, val2); } // Asserts that val1 is less than, or almost equal to, val2. Fails // otherwise. In particular, it fails if either val1 or val2 is NaN. AssertionResult DoubleLE(const char* expr1, const char* expr2, double val1, double val2) { return internal::FloatingPointLE(expr1, expr2, val1, val2); } namespace internal { // The helper function for {ASSERT|EXPECT}_EQ with int or enum // arguments. AssertionResult CmpHelperEQ(const char* expected_expression, const char* actual_expression, BiggestInt expected, BiggestInt actual) { if (expected == actual) { return AssertionSuccess(); } return EqFailure(expected_expression, actual_expression, FormatForComparisonFailureMessage(expected, actual), FormatForComparisonFailureMessage(actual, expected), false); } // A macro for implementing the helper functions needed to implement // ASSERT_?? and EXPECT_?? with integer or enum arguments. It is here // just to avoid copy-and-paste of similar code. #define GTEST_IMPL_CMP_HELPER_(op_name, op)\ AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \ BiggestInt val1, BiggestInt val2) {\ if (val1 op val2) {\ return AssertionSuccess();\ } else {\ return AssertionFailure() \ << "Expected: (" << expr1 << ") " #op " (" << expr2\ << "), actual: " << FormatForComparisonFailureMessage(val1, val2)\ << " vs " << FormatForComparisonFailureMessage(val2, val1);\ }\ } // Implements the helper function for {ASSERT|EXPECT}_NE with int or // enum arguments. GTEST_IMPL_CMP_HELPER_(NE, !=) // Implements the helper function for {ASSERT|EXPECT}_LE with int or // enum arguments. GTEST_IMPL_CMP_HELPER_(LE, <=) // Implements the helper function for {ASSERT|EXPECT}_LT with int or // enum arguments. GTEST_IMPL_CMP_HELPER_(LT, < ) // Implements the helper function for {ASSERT|EXPECT}_GE with int or // enum arguments. GTEST_IMPL_CMP_HELPER_(GE, >=) // Implements the helper function for {ASSERT|EXPECT}_GT with int or // enum arguments. GTEST_IMPL_CMP_HELPER_(GT, > ) #undef GTEST_IMPL_CMP_HELPER_ // The helper function for {ASSERT|EXPECT}_STREQ. AssertionResult CmpHelperSTREQ(const char* expected_expression, const char* actual_expression, const char* expected, const char* actual) { if (String::CStringEquals(expected, actual)) { return AssertionSuccess(); } return EqFailure(expected_expression, actual_expression, PrintToString(expected), PrintToString(actual), false); } // The helper function for {ASSERT|EXPECT}_STRCASEEQ. AssertionResult CmpHelperSTRCASEEQ(const char* expected_expression, const char* actual_expression, const char* expected, const char* actual) { if (String::CaseInsensitiveCStringEquals(expected, actual)) { return AssertionSuccess(); } return EqFailure(expected_expression, actual_expression, PrintToString(expected), PrintToString(actual), true); } // The helper function for {ASSERT|EXPECT}_STRNE. AssertionResult CmpHelperSTRNE(const char* s1_expression, const char* s2_expression, const char* s1, const char* s2) { if (!String::CStringEquals(s1, s2)) { return AssertionSuccess(); } else { return AssertionFailure() << "Expected: (" << s1_expression << ") != (" << s2_expression << "), actual: \"" << s1 << "\" vs \"" << s2 << "\""; } } // The helper function for {ASSERT|EXPECT}_STRCASENE. AssertionResult CmpHelperSTRCASENE(const char* s1_expression, const char* s2_expression, const char* s1, const char* s2) { if (!String::CaseInsensitiveCStringEquals(s1, s2)) { return AssertionSuccess(); } else { return AssertionFailure() << "Expected: (" << s1_expression << ") != (" << s2_expression << ") (ignoring case), actual: \"" << s1 << "\" vs \"" << s2 << "\""; } } } // namespace internal namespace { // Helper functions for implementing IsSubString() and IsNotSubstring(). // This group of overloaded functions return true iff needle is a // substring of haystack. NULL is considered a substring of itself // only. bool IsSubstringPred(const char* needle, const char* haystack) { if (needle == NULL || haystack == NULL) return needle == haystack; return strstr(haystack, needle) != NULL; } bool IsSubstringPred(const wchar_t* needle, const wchar_t* haystack) { if (needle == NULL || haystack == NULL) return needle == haystack; return wcsstr(haystack, needle) != NULL; } // StringType here can be either ::std::string or ::std::wstring. template bool IsSubstringPred(const StringType& needle, const StringType& haystack) { return haystack.find(needle) != StringType::npos; } // This function implements either IsSubstring() or IsNotSubstring(), // depending on the value of the expected_to_be_substring parameter. // StringType here can be const char*, const wchar_t*, ::std::string, // or ::std::wstring. template AssertionResult IsSubstringImpl( bool expected_to_be_substring, const char* needle_expr, const char* haystack_expr, const StringType& needle, const StringType& haystack) { if (IsSubstringPred(needle, haystack) == expected_to_be_substring) return AssertionSuccess(); const bool is_wide_string = sizeof(needle[0]) > 1; const char* const begin_string_quote = is_wide_string ? "L\"" : "\""; return AssertionFailure() << "Value of: " << needle_expr << "\n" << " Actual: " << begin_string_quote << needle << "\"\n" << "Expected: " << (expected_to_be_substring ? "" : "not ") << "a substring of " << haystack_expr << "\n" << "Which is: " << begin_string_quote << haystack << "\""; } } // namespace // IsSubstring() and IsNotSubstring() check whether needle is a // substring of haystack (NULL is considered a substring of itself // only), and return an appropriate error message when they fail. AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const char* needle, const char* haystack) { return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); } AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const wchar_t* needle, const wchar_t* haystack) { return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); } AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const char* needle, const char* haystack) { return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); } AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const wchar_t* needle, const wchar_t* haystack) { return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); } AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const ::std::string& needle, const ::std::string& haystack) { return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); } AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const ::std::string& needle, const ::std::string& haystack) { return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); } #if GTEST_HAS_STD_WSTRING AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const ::std::wstring& needle, const ::std::wstring& haystack) { return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); } AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const ::std::wstring& needle, const ::std::wstring& haystack) { return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); } #endif // GTEST_HAS_STD_WSTRING namespace internal { #if GTEST_OS_WINDOWS namespace { // Helper function for IsHRESULT{SuccessFailure} predicates AssertionResult HRESULTFailureHelper(const char* expr, const char* expected, long hr) { // NOLINT # if GTEST_OS_WINDOWS_MOBILE // Windows CE doesn't support FormatMessage. const char error_text[] = ""; # else // Looks up the human-readable system message for the HRESULT code // and since we're not passing any params to FormatMessage, we don't // want inserts expanded. const DWORD kFlags = FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS; const DWORD kBufSize = 4096; // Gets the system's human readable message string for this HRESULT. char error_text[kBufSize] = { '\0' }; DWORD message_length = ::FormatMessageA(kFlags, 0, // no source, we're asking system hr, // the error 0, // no line width restrictions error_text, // output buffer kBufSize, // buf size NULL); // no arguments for inserts // Trims tailing white space (FormatMessage leaves a trailing CR-LF) for (; message_length && IsSpace(error_text[message_length - 1]); --message_length) { error_text[message_length - 1] = '\0'; } # endif // GTEST_OS_WINDOWS_MOBILE const std::string error_hex("0x" + String::FormatHexInt(hr)); return ::testing::AssertionFailure() << "Expected: " << expr << " " << expected << ".\n" << " Actual: " << error_hex << " " << error_text << "\n"; } } // namespace AssertionResult IsHRESULTSuccess(const char* expr, long hr) { // NOLINT if (SUCCEEDED(hr)) { return AssertionSuccess(); } return HRESULTFailureHelper(expr, "succeeds", hr); } AssertionResult IsHRESULTFailure(const char* expr, long hr) { // NOLINT if (FAILED(hr)) { return AssertionSuccess(); } return HRESULTFailureHelper(expr, "fails", hr); } #endif // GTEST_OS_WINDOWS // Utility functions for encoding Unicode text (wide strings) in // UTF-8. // A Unicode code-point can have upto 21 bits, and is encoded in UTF-8 // like this: // // Code-point length Encoding // 0 - 7 bits 0xxxxxxx // 8 - 11 bits 110xxxxx 10xxxxxx // 12 - 16 bits 1110xxxx 10xxxxxx 10xxxxxx // 17 - 21 bits 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx // The maximum code-point a one-byte UTF-8 sequence can represent. const UInt32 kMaxCodePoint1 = (static_cast(1) << 7) - 1; // The maximum code-point a two-byte UTF-8 sequence can represent. const UInt32 kMaxCodePoint2 = (static_cast(1) << (5 + 6)) - 1; // The maximum code-point a three-byte UTF-8 sequence can represent. const UInt32 kMaxCodePoint3 = (static_cast(1) << (4 + 2*6)) - 1; // The maximum code-point a four-byte UTF-8 sequence can represent. const UInt32 kMaxCodePoint4 = (static_cast(1) << (3 + 3*6)) - 1; // Chops off the n lowest bits from a bit pattern. Returns the n // lowest bits. As a side effect, the original bit pattern will be // shifted to the right by n bits. inline UInt32 ChopLowBits(UInt32* bits, int n) { const UInt32 low_bits = *bits & ((static_cast(1) << n) - 1); *bits >>= n; return low_bits; } // Converts a Unicode code point to a narrow string in UTF-8 encoding. // code_point parameter is of type UInt32 because wchar_t may not be // wide enough to contain a code point. // If the code_point is not a valid Unicode code point // (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted // to "(Invalid Unicode 0xXXXXXXXX)". std::string CodePointToUtf8(UInt32 code_point) { if (code_point > kMaxCodePoint4) { return "(Invalid Unicode 0x" + String::FormatHexInt(code_point) + ")"; } char str[5]; // Big enough for the largest valid code point. if (code_point <= kMaxCodePoint1) { str[1] = '\0'; str[0] = static_cast(code_point); // 0xxxxxxx } else if (code_point <= kMaxCodePoint2) { str[2] = '\0'; str[1] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx str[0] = static_cast(0xC0 | code_point); // 110xxxxx } else if (code_point <= kMaxCodePoint3) { str[3] = '\0'; str[2] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx str[1] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx str[0] = static_cast(0xE0 | code_point); // 1110xxxx } else { // code_point <= kMaxCodePoint4 str[4] = '\0'; str[3] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx str[2] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx str[1] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx str[0] = static_cast(0xF0 | code_point); // 11110xxx } return str; } // The following two functions only make sense if the the system // uses UTF-16 for wide string encoding. All supported systems // with 16 bit wchar_t (Windows, Cygwin, Symbian OS) do use UTF-16. // Determines if the arguments constitute UTF-16 surrogate pair // and thus should be combined into a single Unicode code point // using CreateCodePointFromUtf16SurrogatePair. inline bool IsUtf16SurrogatePair(wchar_t first, wchar_t second) { return sizeof(wchar_t) == 2 && (first & 0xFC00) == 0xD800 && (second & 0xFC00) == 0xDC00; } // Creates a Unicode code point from UTF16 surrogate pair. inline UInt32 CreateCodePointFromUtf16SurrogatePair(wchar_t first, wchar_t second) { const UInt32 mask = (1 << 10) - 1; return (sizeof(wchar_t) == 2) ? (((first & mask) << 10) | (second & mask)) + 0x10000 : // This function should not be called when the condition is // false, but we provide a sensible default in case it is. static_cast(first); } // Converts a wide string to a narrow string in UTF-8 encoding. // The wide string is assumed to have the following encoding: // UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS) // UTF-32 if sizeof(wchar_t) == 4 (on Linux) // Parameter str points to a null-terminated wide string. // Parameter num_chars may additionally limit the number // of wchar_t characters processed. -1 is used when the entire string // should be processed. // If the string contains code points that are not valid Unicode code points // (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output // as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding // and contains invalid UTF-16 surrogate pairs, values in those pairs // will be encoded as individual Unicode characters from Basic Normal Plane. std::string WideStringToUtf8(const wchar_t* str, int num_chars) { if (num_chars == -1) num_chars = static_cast(wcslen(str)); ::std::stringstream stream; for (int i = 0; i < num_chars; ++i) { UInt32 unicode_code_point; if (str[i] == L'\0') { break; } else if (i + 1 < num_chars && IsUtf16SurrogatePair(str[i], str[i + 1])) { unicode_code_point = CreateCodePointFromUtf16SurrogatePair(str[i], str[i + 1]); i++; } else { unicode_code_point = static_cast(str[i]); } stream << CodePointToUtf8(unicode_code_point); } return StringStreamToString(&stream); } // Converts a wide C string to an std::string using the UTF-8 encoding. // NULL will be converted to "(null)". std::string String::ShowWideCString(const wchar_t * wide_c_str) { if (wide_c_str == NULL) return "(null)"; return internal::WideStringToUtf8(wide_c_str, -1); } // Compares two wide C strings. Returns true iff they have the same // content. // // Unlike wcscmp(), this function can handle NULL argument(s). A NULL // C string is considered different to any non-NULL C string, // including the empty string. bool String::WideCStringEquals(const wchar_t * lhs, const wchar_t * rhs) { if (lhs == NULL) return rhs == NULL; if (rhs == NULL) return false; return wcscmp(lhs, rhs) == 0; } // Helper function for *_STREQ on wide strings. AssertionResult CmpHelperSTREQ(const char* expected_expression, const char* actual_expression, const wchar_t* expected, const wchar_t* actual) { if (String::WideCStringEquals(expected, actual)) { return AssertionSuccess(); } return EqFailure(expected_expression, actual_expression, PrintToString(expected), PrintToString(actual), false); } // Helper function for *_STRNE on wide strings. AssertionResult CmpHelperSTRNE(const char* s1_expression, const char* s2_expression, const wchar_t* s1, const wchar_t* s2) { if (!String::WideCStringEquals(s1, s2)) { return AssertionSuccess(); } return AssertionFailure() << "Expected: (" << s1_expression << ") != (" << s2_expression << "), actual: " << PrintToString(s1) << " vs " << PrintToString(s2); } // Compares two C strings, ignoring case. Returns true iff they have // the same content. // // Unlike strcasecmp(), this function can handle NULL argument(s). A // NULL C string is considered different to any non-NULL C string, // including the empty string. bool String::CaseInsensitiveCStringEquals(const char * lhs, const char * rhs) { if (lhs == NULL) return rhs == NULL; if (rhs == NULL) return false; return posix::StrCaseCmp(lhs, rhs) == 0; } // Compares two wide C strings, ignoring case. Returns true iff they // have the same content. // // Unlike wcscasecmp(), this function can handle NULL argument(s). // A NULL C string is considered different to any non-NULL wide C string, // including the empty string. // NB: The implementations on different platforms slightly differ. // On windows, this method uses _wcsicmp which compares according to LC_CTYPE // environment variable. On GNU platform this method uses wcscasecmp // which compares according to LC_CTYPE category of the current locale. // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the // current locale. bool String::CaseInsensitiveWideCStringEquals(const wchar_t* lhs, const wchar_t* rhs) { if (lhs == NULL) return rhs == NULL; if (rhs == NULL) return false; #if GTEST_OS_WINDOWS return _wcsicmp(lhs, rhs) == 0; #elif GTEST_OS_LINUX && !GTEST_OS_LINUX_ANDROID return wcscasecmp(lhs, rhs) == 0; #else // Android, Mac OS X and Cygwin don't define wcscasecmp. // Other unknown OSes may not define it either. wint_t left, right; do { left = towlower(*lhs++); right = towlower(*rhs++); } while (left && left == right); return left == right; #endif // OS selector } // Returns true iff str ends with the given suffix, ignoring case. // Any string is considered to end with an empty suffix. bool String::EndsWithCaseInsensitive( const std::string& str, const std::string& suffix) { const size_t str_len = str.length(); const size_t suffix_len = suffix.length(); return (str_len >= suffix_len) && CaseInsensitiveCStringEquals(str.c_str() + str_len - suffix_len, suffix.c_str()); } // Formats an int value as "%02d". std::string String::FormatIntWidth2(int value) { std::stringstream ss; ss << std::setfill('0') << std::setw(2) << value; return ss.str(); } // Formats an int value as "%X". std::string String::FormatHexInt(int value) { std::stringstream ss; ss << std::hex << std::uppercase << value; return ss.str(); } // Formats a byte as "%02X". std::string String::FormatByte(unsigned char value) { std::stringstream ss; ss << std::setfill('0') << std::setw(2) << std::hex << std::uppercase << static_cast(value); return ss.str(); } // Converts the buffer in a stringstream to an std::string, converting NUL // bytes to "\\0" along the way. std::string StringStreamToString(::std::stringstream* ss) { const ::std::string& str = ss->str(); const char* const start = str.c_str(); const char* const end = start + str.length(); std::string result; result.reserve(2 * (end - start)); for (const char* ch = start; ch != end; ++ch) { if (*ch == '\0') { result += "\\0"; // Replaces NUL with "\\0"; } else { result += *ch; } } return result; } // Appends the user-supplied message to the Google-Test-generated message. std::string AppendUserMessage(const std::string& gtest_msg, const Message& user_msg) { // Appends the user message if it's non-empty. const std::string user_msg_string = user_msg.GetString(); if (user_msg_string.empty()) { return gtest_msg; } return gtest_msg + "\n" + user_msg_string; } } // namespace internal // class TestResult // Creates an empty TestResult. TestResult::TestResult() : death_test_count_(0), elapsed_time_(0) { } // D'tor. TestResult::~TestResult() { } // Returns the i-th test part result among all the results. i can // range from 0 to total_part_count() - 1. If i is not in that range, // aborts the program. const TestPartResult& TestResult::GetTestPartResult(int i) const { if (i < 0 || i >= total_part_count()) internal::posix::Abort(); return test_part_results_.at(i); } // Returns the i-th test property. i can range from 0 to // test_property_count() - 1. If i is not in that range, aborts the // program. const TestProperty& TestResult::GetTestProperty(int i) const { if (i < 0 || i >= test_property_count()) internal::posix::Abort(); return test_properties_.at(i); } // Clears the test part results. void TestResult::ClearTestPartResults() { test_part_results_.clear(); } // Adds a test part result to the list. void TestResult::AddTestPartResult(const TestPartResult& test_part_result) { test_part_results_.push_back(test_part_result); } // Adds a test property to the list. If a property with the same key as the // supplied property is already represented, the value of this test_property // replaces the old value for that key. void TestResult::RecordProperty(const std::string& xml_element, const TestProperty& test_property) { if (!ValidateTestProperty(xml_element, test_property)) { return; } internal::MutexLock lock(&test_properites_mutex_); const std::vector::iterator property_with_matching_key = std::find_if(test_properties_.begin(), test_properties_.end(), internal::TestPropertyKeyIs(test_property.key())); if (property_with_matching_key == test_properties_.end()) { test_properties_.push_back(test_property); return; } property_with_matching_key->SetValue(test_property.value()); } // The list of reserved attributes used in the element of XML // output. static const char* const kReservedTestSuitesAttributes[] = { "disabled", "errors", "failures", "name", "random_seed", "tests", "time", "timestamp" }; // The list of reserved attributes used in the element of XML // output. static const char* const kReservedTestSuiteAttributes[] = { "disabled", "errors", "failures", "name", "tests", "time" }; // The list of reserved attributes used in the element of XML output. static const char* const kReservedTestCaseAttributes[] = { "classname", "name", "status", "time", "type_param", "value_param" }; template std::vector ArrayAsVector(const char* const (&array)[kSize]) { return std::vector(array, array + kSize); } static std::vector GetReservedAttributesForElement( const std::string& xml_element) { if (xml_element == "testsuites") { return ArrayAsVector(kReservedTestSuitesAttributes); } else if (xml_element == "testsuite") { return ArrayAsVector(kReservedTestSuiteAttributes); } else if (xml_element == "testcase") { return ArrayAsVector(kReservedTestCaseAttributes); } else { GTEST_CHECK_(false) << "Unrecognized xml_element provided: " << xml_element; } // This code is unreachable but some compilers may not realizes that. return std::vector(); } static std::string FormatWordList(const std::vector& words) { Message word_list; for (size_t i = 0; i < words.size(); ++i) { if (i > 0 && words.size() > 2) { word_list << ", "; } if (i == words.size() - 1) { word_list << "and "; } word_list << "'" << words[i] << "'"; } return word_list.GetString(); } bool ValidateTestPropertyName(const std::string& property_name, const std::vector& reserved_names) { if (std::find(reserved_names.begin(), reserved_names.end(), property_name) != reserved_names.end()) { ADD_FAILURE() << "Reserved key used in RecordProperty(): " << property_name << " (" << FormatWordList(reserved_names) << " are reserved by " << GTEST_NAME_ << ")"; return false; } return true; } // Adds a failure if the key is a reserved attribute of the element named // xml_element. Returns true if the property is valid. bool TestResult::ValidateTestProperty(const std::string& xml_element, const TestProperty& test_property) { return ValidateTestPropertyName(test_property.key(), GetReservedAttributesForElement(xml_element)); } // Clears the object. void TestResult::Clear() { test_part_results_.clear(); test_properties_.clear(); death_test_count_ = 0; elapsed_time_ = 0; } // Returns true iff the test failed. bool TestResult::Failed() const { for (int i = 0; i < total_part_count(); ++i) { if (GetTestPartResult(i).failed()) return true; } return false; } // Returns true iff the test part fatally failed. static bool TestPartFatallyFailed(const TestPartResult& result) { return result.fatally_failed(); } // Returns true iff the test fatally failed. bool TestResult::HasFatalFailure() const { return CountIf(test_part_results_, TestPartFatallyFailed) > 0; } // Returns true iff the test part non-fatally failed. static bool TestPartNonfatallyFailed(const TestPartResult& result) { return result.nonfatally_failed(); } // Returns true iff the test has a non-fatal failure. bool TestResult::HasNonfatalFailure() const { return CountIf(test_part_results_, TestPartNonfatallyFailed) > 0; } // Gets the number of all test parts. This is the sum of the number // of successful test parts and the number of failed test parts. int TestResult::total_part_count() const { return static_cast(test_part_results_.size()); } // Returns the number of the test properties. int TestResult::test_property_count() const { return static_cast(test_properties_.size()); } // class Test // Creates a Test object. // The c'tor saves the values of all Google Test flags. Test::Test() : gtest_flag_saver_(new internal::GTestFlagSaver) { } // The d'tor restores the values of all Google Test flags. Test::~Test() { delete gtest_flag_saver_; } // Sets up the test fixture. // // A sub-class may override this. void Test::SetUp() { } // Tears down the test fixture. // // A sub-class may override this. void Test::TearDown() { } // Allows user supplied key value pairs to be recorded for later output. void Test::RecordProperty(const std::string& key, const std::string& value) { UnitTest::GetInstance()->RecordProperty(key, value); } // Allows user supplied key value pairs to be recorded for later output. void Test::RecordProperty(const std::string& key, int value) { Message value_message; value_message << value; RecordProperty(key, value_message.GetString().c_str()); } namespace internal { void ReportFailureInUnknownLocation(TestPartResult::Type result_type, const std::string& message) { // This function is a friend of UnitTest and as such has access to // AddTestPartResult. UnitTest::GetInstance()->AddTestPartResult( result_type, NULL, // No info about the source file where the exception occurred. -1, // We have no info on which line caused the exception. message, ""); // No stack trace, either. } } // namespace internal // Google Test requires all tests in the same test case to use the same test // fixture class. This function checks if the current test has the // same fixture class as the first test in the current test case. If // yes, it returns true; otherwise it generates a Google Test failure and // returns false. bool Test::HasSameFixtureClass() { internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); const TestCase* const test_case = impl->current_test_case(); // Info about the first test in the current test case. const TestInfo* const first_test_info = test_case->test_info_list()[0]; const internal::TypeId first_fixture_id = first_test_info->fixture_class_id_; const char* const first_test_name = first_test_info->name(); // Info about the current test. const TestInfo* const this_test_info = impl->current_test_info(); const internal::TypeId this_fixture_id = this_test_info->fixture_class_id_; const char* const this_test_name = this_test_info->name(); if (this_fixture_id != first_fixture_id) { // Is the first test defined using TEST? const bool first_is_TEST = first_fixture_id == internal::GetTestTypeId(); // Is this test defined using TEST? const bool this_is_TEST = this_fixture_id == internal::GetTestTypeId(); if (first_is_TEST || this_is_TEST) { // Both TEST and TEST_F appear in same test case, which is incorrect. // Tell the user how to fix this. // Gets the name of the TEST and the name of the TEST_F. Note // that first_is_TEST and this_is_TEST cannot both be true, as // the fixture IDs are different for the two tests. const char* const TEST_name = first_is_TEST ? first_test_name : this_test_name; const char* const TEST_F_name = first_is_TEST ? this_test_name : first_test_name; ADD_FAILURE() << "All tests in the same test case must use the same test fixture\n" << "class, so mixing TEST_F and TEST in the same test case is\n" << "illegal. In test case " << this_test_info->test_case_name() << ",\n" << "test " << TEST_F_name << " is defined using TEST_F but\n" << "test " << TEST_name << " is defined using TEST. You probably\n" << "want to change the TEST to TEST_F or move it to another test\n" << "case."; } else { // Two fixture classes with the same name appear in two different // namespaces, which is not allowed. Tell the user how to fix this. ADD_FAILURE() << "All tests in the same test case must use the same test fixture\n" << "class. However, in test case " << this_test_info->test_case_name() << ",\n" << "you defined test " << first_test_name << " and test " << this_test_name << "\n" << "using two different test fixture classes. This can happen if\n" << "the two classes are from different namespaces or translation\n" << "units and have the same name. You should probably rename one\n" << "of the classes to put the tests into different test cases."; } return false; } return true; } #if GTEST_HAS_SEH // Adds an "exception thrown" fatal failure to the current test. This // function returns its result via an output parameter pointer because VC++ // prohibits creation of objects with destructors on stack in functions // using __try (see error C2712). static std::string* FormatSehExceptionMessage(DWORD exception_code, const char* location) { Message message; message << "SEH exception with code 0x" << std::setbase(16) << exception_code << std::setbase(10) << " thrown in " << location << "."; return new std::string(message.GetString()); } #endif // GTEST_HAS_SEH namespace internal { #if GTEST_HAS_EXCEPTIONS // Adds an "exception thrown" fatal failure to the current test. static std::string FormatCxxExceptionMessage(const char* description, const char* location) { Message message; if (description != NULL) { message << "C++ exception with description \"" << description << "\""; } else { message << "Unknown C++ exception"; } message << " thrown in " << location << "."; return message.GetString(); } static std::string PrintTestPartResultToString( const TestPartResult& test_part_result); GoogleTestFailureException::GoogleTestFailureException( const TestPartResult& failure) : ::std::runtime_error(PrintTestPartResultToString(failure).c_str()) {} #endif // GTEST_HAS_EXCEPTIONS // We put these helper functions in the internal namespace as IBM's xlC // compiler rejects the code if they were declared static. // Runs the given method and handles SEH exceptions it throws, when // SEH is supported; returns the 0-value for type Result in case of an // SEH exception. (Microsoft compilers cannot handle SEH and C++ // exceptions in the same function. Therefore, we provide a separate // wrapper function for handling SEH exceptions.) template Result HandleSehExceptionsInMethodIfSupported( T* object, Result (T::*method)(), const char* location) { #if GTEST_HAS_SEH __try { return (object->*method)(); } __except (internal::UnitTestOptions::GTestShouldProcessSEH( // NOLINT GetExceptionCode())) { // We create the exception message on the heap because VC++ prohibits // creation of objects with destructors on stack in functions using __try // (see error C2712). std::string* exception_message = FormatSehExceptionMessage( GetExceptionCode(), location); internal::ReportFailureInUnknownLocation(TestPartResult::kFatalFailure, *exception_message); delete exception_message; return static_cast(0); } #else (void)location; return (object->*method)(); #endif // GTEST_HAS_SEH } // Runs the given method and catches and reports C++ and/or SEH-style // exceptions, if they are supported; returns the 0-value for type // Result in case of an SEH exception. template Result HandleExceptionsInMethodIfSupported( T* object, Result (T::*method)(), const char* location) { // NOTE: The user code can affect the way in which Google Test handles // exceptions by setting GTEST_FLAG(catch_exceptions), but only before // RUN_ALL_TESTS() starts. It is technically possible to check the flag // after the exception is caught and either report or re-throw the // exception based on the flag's value: // // try { // // Perform the test method. // } catch (...) { // if (GTEST_FLAG(catch_exceptions)) // // Report the exception as failure. // else // throw; // Re-throws the original exception. // } // // However, the purpose of this flag is to allow the program to drop into // the debugger when the exception is thrown. On most platforms, once the // control enters the catch block, the exception origin information is // lost and the debugger will stop the program at the point of the // re-throw in this function -- instead of at the point of the original // throw statement in the code under test. For this reason, we perform // the check early, sacrificing the ability to affect Google Test's // exception handling in the method where the exception is thrown. if (internal::GetUnitTestImpl()->catch_exceptions()) { #if GTEST_HAS_EXCEPTIONS try { return HandleSehExceptionsInMethodIfSupported(object, method, location); } catch (const internal::GoogleTestFailureException&) { // NOLINT // This exception type can only be thrown by a failed Google // Test assertion with the intention of letting another testing // framework catch it. Therefore we just re-throw it. throw; } catch (const std::exception& e) { // NOLINT internal::ReportFailureInUnknownLocation( TestPartResult::kFatalFailure, FormatCxxExceptionMessage(e.what(), location)); } catch (...) { // NOLINT internal::ReportFailureInUnknownLocation( TestPartResult::kFatalFailure, FormatCxxExceptionMessage(NULL, location)); } return static_cast(0); #else return HandleSehExceptionsInMethodIfSupported(object, method, location); #endif // GTEST_HAS_EXCEPTIONS } else { return (object->*method)(); } } } // namespace internal // Runs the test and updates the test result. void Test::Run() { if (!HasSameFixtureClass()) return; internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); impl->os_stack_trace_getter()->UponLeavingGTest(); internal::HandleExceptionsInMethodIfSupported(this, &Test::SetUp, "SetUp()"); // We will run the test only if SetUp() was successful. if (!HasFatalFailure()) { impl->os_stack_trace_getter()->UponLeavingGTest(); internal::HandleExceptionsInMethodIfSupported( this, &Test::TestBody, "the test body"); } // However, we want to clean up as much as possible. Hence we will // always call TearDown(), even if SetUp() or the test body has // failed. impl->os_stack_trace_getter()->UponLeavingGTest(); internal::HandleExceptionsInMethodIfSupported( this, &Test::TearDown, "TearDown()"); } // Returns true iff the current test has a fatal failure. bool Test::HasFatalFailure() { return internal::GetUnitTestImpl()->current_test_result()->HasFatalFailure(); } // Returns true iff the current test has a non-fatal failure. bool Test::HasNonfatalFailure() { return internal::GetUnitTestImpl()->current_test_result()-> HasNonfatalFailure(); } // class TestInfo // Constructs a TestInfo object. It assumes ownership of the test factory // object. TestInfo::TestInfo(const std::string& a_test_case_name, const std::string& a_name, const char* a_type_param, const char* a_value_param, internal::TypeId fixture_class_id, internal::TestFactoryBase* factory) : test_case_name_(a_test_case_name), name_(a_name), type_param_(a_type_param ? new std::string(a_type_param) : NULL), value_param_(a_value_param ? new std::string(a_value_param) : NULL), fixture_class_id_(fixture_class_id), should_run_(false), is_disabled_(false), matches_filter_(false), factory_(factory), result_() {} // Destructs a TestInfo object. TestInfo::~TestInfo() { delete factory_; } namespace internal { // Creates a new TestInfo object and registers it with Google Test; // returns the created object. // // Arguments: // // test_case_name: name of the test case // name: name of the test // type_param: the name of the test's type parameter, or NULL if // this is not a typed or a type-parameterized test. // value_param: text representation of the test's value parameter, // or NULL if this is not a value-parameterized test. // fixture_class_id: ID of the test fixture class // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case // factory: pointer to the factory that creates a test object. // The newly created TestInfo instance will assume // ownership of the factory object. TestInfo* MakeAndRegisterTestInfo( const char* test_case_name, const char* name, const char* type_param, const char* value_param, TypeId fixture_class_id, SetUpTestCaseFunc set_up_tc, TearDownTestCaseFunc tear_down_tc, TestFactoryBase* factory) { TestInfo* const test_info = new TestInfo(test_case_name, name, type_param, value_param, fixture_class_id, factory); GetUnitTestImpl()->AddTestInfo(set_up_tc, tear_down_tc, test_info); return test_info; } #if GTEST_HAS_PARAM_TEST void ReportInvalidTestCaseType(const char* test_case_name, const char* file, int line) { Message errors; errors << "Attempted redefinition of test case " << test_case_name << ".\n" << "All tests in the same test case must use the same test fixture\n" << "class. However, in test case " << test_case_name << ", you tried\n" << "to define a test using a fixture class different from the one\n" << "used earlier. This can happen if the two fixture classes are\n" << "from different namespaces and have the same name. You should\n" << "probably rename one of the classes to put the tests into different\n" << "test cases."; fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(), errors.GetString().c_str()); } #endif // GTEST_HAS_PARAM_TEST } // namespace internal namespace { // A predicate that checks the test name of a TestInfo against a known // value. // // This is used for implementation of the TestCase class only. We put // it in the anonymous namespace to prevent polluting the outer // namespace. // // TestNameIs is copyable. class TestNameIs { public: // Constructor. // // TestNameIs has NO default constructor. explicit TestNameIs(const char* name) : name_(name) {} // Returns true iff the test name of test_info matches name_. bool operator()(const TestInfo * test_info) const { return test_info && test_info->name() == name_; } private: std::string name_; }; } // namespace namespace internal { // This method expands all parameterized tests registered with macros TEST_P // and INSTANTIATE_TEST_CASE_P into regular tests and registers those. // This will be done just once during the program runtime. void UnitTestImpl::RegisterParameterizedTests() { #if GTEST_HAS_PARAM_TEST if (!parameterized_tests_registered_) { parameterized_test_registry_.RegisterTests(); parameterized_tests_registered_ = true; } #endif } } // namespace internal // Creates the test object, runs it, records its result, and then // deletes it. void TestInfo::Run() { if (!should_run_) return; // Tells UnitTest where to store test result. internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); impl->set_current_test_info(this); TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater(); // Notifies the unit test event listeners that a test is about to start. repeater->OnTestStart(*this); const TimeInMillis start = internal::GetTimeInMillis(); impl->os_stack_trace_getter()->UponLeavingGTest(); // Creates the test object. Test* const test = internal::HandleExceptionsInMethodIfSupported( factory_, &internal::TestFactoryBase::CreateTest, "the test fixture's constructor"); // Runs the test only if the test object was created and its // constructor didn't generate a fatal failure. if ((test != NULL) && !Test::HasFatalFailure()) { // This doesn't throw as all user code that can throw are wrapped into // exception handling code. test->Run(); } // Deletes the test object. impl->os_stack_trace_getter()->UponLeavingGTest(); internal::HandleExceptionsInMethodIfSupported( test, &Test::DeleteSelf_, "the test fixture's destructor"); result_.set_elapsed_time(internal::GetTimeInMillis() - start); // Notifies the unit test event listener that a test has just finished. repeater->OnTestEnd(*this); // Tells UnitTest to stop associating assertion results to this // test. impl->set_current_test_info(NULL); } // class TestCase // Gets the number of successful tests in this test case. int TestCase::successful_test_count() const { return CountIf(test_info_list_, TestPassed); } // Gets the number of failed tests in this test case. int TestCase::failed_test_count() const { return CountIf(test_info_list_, TestFailed); } // Gets the number of disabled tests that will be reported in the XML report. int TestCase::reportable_disabled_test_count() const { return CountIf(test_info_list_, TestReportableDisabled); } // Gets the number of disabled tests in this test case. int TestCase::disabled_test_count() const { return CountIf(test_info_list_, TestDisabled); } // Gets the number of tests to be printed in the XML report. int TestCase::reportable_test_count() const { return CountIf(test_info_list_, TestReportable); } // Get the number of tests in this test case that should run. int TestCase::test_to_run_count() const { return CountIf(test_info_list_, ShouldRunTest); } // Gets the number of all tests. int TestCase::total_test_count() const { return static_cast(test_info_list_.size()); } // Creates a TestCase with the given name. // // Arguments: // // name: name of the test case // a_type_param: the name of the test case's type parameter, or NULL if // this is not a typed or a type-parameterized test case. // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case TestCase::TestCase(const char* a_name, const char* a_type_param, Test::SetUpTestCaseFunc set_up_tc, Test::TearDownTestCaseFunc tear_down_tc) : name_(a_name), type_param_(a_type_param ? new std::string(a_type_param) : NULL), set_up_tc_(set_up_tc), tear_down_tc_(tear_down_tc), should_run_(false), elapsed_time_(0) { } // Destructor of TestCase. TestCase::~TestCase() { // Deletes every Test in the collection. ForEach(test_info_list_, internal::Delete); } // Returns the i-th test among all the tests. i can range from 0 to // total_test_count() - 1. If i is not in that range, returns NULL. const TestInfo* TestCase::GetTestInfo(int i) const { const int index = GetElementOr(test_indices_, i, -1); return index < 0 ? NULL : test_info_list_[index]; } // Returns the i-th test among all the tests. i can range from 0 to // total_test_count() - 1. If i is not in that range, returns NULL. TestInfo* TestCase::GetMutableTestInfo(int i) { const int index = GetElementOr(test_indices_, i, -1); return index < 0 ? NULL : test_info_list_[index]; } // Adds a test to this test case. Will delete the test upon // destruction of the TestCase object. void TestCase::AddTestInfo(TestInfo * test_info) { test_info_list_.push_back(test_info); test_indices_.push_back(static_cast(test_indices_.size())); } // Runs every test in this TestCase. void TestCase::Run() { if (!should_run_) return; internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); impl->set_current_test_case(this); TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater(); repeater->OnTestCaseStart(*this); impl->os_stack_trace_getter()->UponLeavingGTest(); internal::HandleExceptionsInMethodIfSupported( this, &TestCase::RunSetUpTestCase, "SetUpTestCase()"); const internal::TimeInMillis start = internal::GetTimeInMillis(); for (int i = 0; i < total_test_count(); i++) { GetMutableTestInfo(i)->Run(); } elapsed_time_ = internal::GetTimeInMillis() - start; impl->os_stack_trace_getter()->UponLeavingGTest(); internal::HandleExceptionsInMethodIfSupported( this, &TestCase::RunTearDownTestCase, "TearDownTestCase()"); repeater->OnTestCaseEnd(*this); impl->set_current_test_case(NULL); } // Clears the results of all tests in this test case. void TestCase::ClearResult() { ad_hoc_test_result_.Clear(); ForEach(test_info_list_, TestInfo::ClearTestResult); } // Shuffles the tests in this test case. void TestCase::ShuffleTests(internal::Random* random) { Shuffle(random, &test_indices_); } // Restores the test order to before the first shuffle. void TestCase::UnshuffleTests() { for (size_t i = 0; i < test_indices_.size(); i++) { test_indices_[i] = static_cast(i); } } // Formats a countable noun. Depending on its quantity, either the // singular form or the plural form is used. e.g. // // FormatCountableNoun(1, "formula", "formuli") returns "1 formula". // FormatCountableNoun(5, "book", "books") returns "5 books". static std::string FormatCountableNoun(int count, const char * singular_form, const char * plural_form) { return internal::StreamableToString(count) + " " + (count == 1 ? singular_form : plural_form); } // Formats the count of tests. static std::string FormatTestCount(int test_count) { return FormatCountableNoun(test_count, "test", "tests"); } // Formats the count of test cases. static std::string FormatTestCaseCount(int test_case_count) { return FormatCountableNoun(test_case_count, "test case", "test cases"); } // Converts a TestPartResult::Type enum to human-friendly string // representation. Both kNonFatalFailure and kFatalFailure are translated // to "Failure", as the user usually doesn't care about the difference // between the two when viewing the test result. static const char * TestPartResultTypeToString(TestPartResult::Type type) { switch (type) { case TestPartResult::kSuccess: return "Success"; case TestPartResult::kNonFatalFailure: case TestPartResult::kFatalFailure: #ifdef _MSC_VER return "error: "; #else return "Failure\n"; #endif default: return "Unknown result type"; } } namespace internal { // Prints a TestPartResult to an std::string. static std::string PrintTestPartResultToString( const TestPartResult& test_part_result) { return (Message() << internal::FormatFileLocation(test_part_result.file_name(), test_part_result.line_number()) << " " << TestPartResultTypeToString(test_part_result.type()) << test_part_result.message()).GetString(); } // Prints a TestPartResult. static void PrintTestPartResult(const TestPartResult& test_part_result) { const std::string& result = PrintTestPartResultToString(test_part_result); printf("%s\n", result.c_str()); fflush(stdout); // If the test program runs in Visual Studio or a debugger, the // following statements add the test part result message to the Output // window such that the user can double-click on it to jump to the // corresponding source code location; otherwise they do nothing. #if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE // We don't call OutputDebugString*() on Windows Mobile, as printing // to stdout is done by OutputDebugString() there already - we don't // want the same message printed twice. ::OutputDebugStringA(result.c_str()); ::OutputDebugStringA("\n"); #endif } // class PrettyUnitTestResultPrinter enum GTestColor { COLOR_DEFAULT, COLOR_RED, COLOR_GREEN, COLOR_YELLOW }; #if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \ !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT // Returns the character attribute for the given color. WORD GetColorAttribute(GTestColor color) { switch (color) { case COLOR_RED: return FOREGROUND_RED; case COLOR_GREEN: return FOREGROUND_GREEN; case COLOR_YELLOW: return FOREGROUND_RED | FOREGROUND_GREEN; default: return 0; } } #else // Returns the ANSI color code for the given color. COLOR_DEFAULT is // an invalid input. const char* GetAnsiColorCode(GTestColor color) { switch (color) { case COLOR_RED: return "1"; case COLOR_GREEN: return "2"; case COLOR_YELLOW: return "3"; default: return NULL; }; } #endif // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE // Returns true iff Google Test should use colors in the output. bool ShouldUseColor(bool stdout_is_tty) { const char* const gtest_color = GTEST_FLAG(color).c_str(); if (String::CaseInsensitiveCStringEquals(gtest_color, "auto")) { #if GTEST_OS_WINDOWS // On Windows the TERM variable is usually not set, but the // console there does support colors. return stdout_is_tty; #else // On non-Windows platforms, we rely on the TERM variable. const char* const term = posix::GetEnv("TERM"); const bool term_supports_color = String::CStringEquals(term, "xterm") || String::CStringEquals(term, "xterm-color") || String::CStringEquals(term, "xterm-256color") || String::CStringEquals(term, "screen") || String::CStringEquals(term, "screen-256color") || String::CStringEquals(term, "linux") || String::CStringEquals(term, "cygwin"); return stdout_is_tty && term_supports_color; #endif // GTEST_OS_WINDOWS } return String::CaseInsensitiveCStringEquals(gtest_color, "yes") || String::CaseInsensitiveCStringEquals(gtest_color, "true") || String::CaseInsensitiveCStringEquals(gtest_color, "t") || String::CStringEquals(gtest_color, "1"); // We take "yes", "true", "t", and "1" as meaning "yes". If the // value is neither one of these nor "auto", we treat it as "no" to // be conservative. } // Helpers for printing colored strings to stdout. Note that on Windows, we // cannot simply emit special characters and have the terminal change colors. // This routine must actually emit the characters rather than return a string // that would be colored when printed, as can be done on Linux. void ColoredPrintf(GTestColor color, const char* fmt, ...) { va_list args; va_start(args, fmt); #if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS || \ GTEST_OS_IOS || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT const bool use_color = AlwaysFalse(); #else static const bool in_color_mode = ShouldUseColor(posix::IsATTY(posix::FileNo(stdout)) != 0); const bool use_color = in_color_mode && (color != COLOR_DEFAULT); #endif // GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS // The '!= 0' comparison is necessary to satisfy MSVC 7.1. if (!use_color) { vprintf(fmt, args); va_end(args); return; } #if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \ !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT const HANDLE stdout_handle = GetStdHandle(STD_OUTPUT_HANDLE); // Gets the current text color. CONSOLE_SCREEN_BUFFER_INFO buffer_info; GetConsoleScreenBufferInfo(stdout_handle, &buffer_info); const WORD old_color_attrs = buffer_info.wAttributes; // We need to flush the stream buffers into the console before each // SetConsoleTextAttribute call lest it affect the text that is already // printed but has not yet reached the console. fflush(stdout); SetConsoleTextAttribute(stdout_handle, GetColorAttribute(color) | FOREGROUND_INTENSITY); vprintf(fmt, args); fflush(stdout); // Restores the text color. SetConsoleTextAttribute(stdout_handle, old_color_attrs); #else printf("\033[0;3%sm", GetAnsiColorCode(color)); vprintf(fmt, args); printf("\033[m"); // Resets the terminal to default. #endif // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE va_end(args); } // Text printed in Google Test's text output and --gunit_list_tests // output to label the type parameter and value parameter for a test. static const char kTypeParamLabel[] = "TypeParam"; static const char kValueParamLabel[] = "GetParam()"; void PrintFullTestCommentIfPresent(const TestInfo& test_info) { const char* const type_param = test_info.type_param(); const char* const value_param = test_info.value_param(); if (type_param != NULL || value_param != NULL) { printf(", where "); if (type_param != NULL) { printf("%s = %s", kTypeParamLabel, type_param); if (value_param != NULL) printf(" and "); } if (value_param != NULL) { printf("%s = %s", kValueParamLabel, value_param); } } } // This class implements the TestEventListener interface. // // Class PrettyUnitTestResultPrinter is copyable. class PrettyUnitTestResultPrinter : public TestEventListener { public: PrettyUnitTestResultPrinter() {} static void PrintTestName(const char * test_case, const char * test) { printf("%s.%s", test_case, test); } // The following methods override what's in the TestEventListener class. virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {} virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration); virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test); virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {} virtual void OnTestCaseStart(const TestCase& test_case); virtual void OnTestStart(const TestInfo& test_info); virtual void OnTestPartResult(const TestPartResult& result); virtual void OnTestEnd(const TestInfo& test_info); virtual void OnTestCaseEnd(const TestCase& test_case); virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test); virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {} virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration); virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {} private: static void PrintFailedTests(const UnitTest& unit_test); }; // Fired before each iteration of tests starts. void PrettyUnitTestResultPrinter::OnTestIterationStart( const UnitTest& unit_test, int iteration) { if (GTEST_FLAG(repeat) != 1) printf("\nRepeating all tests (iteration %d) . . .\n\n", iteration + 1); const char* const filter = GTEST_FLAG(filter).c_str(); // Prints the filter if it's not *. This reminds the user that some // tests may be skipped. if (!String::CStringEquals(filter, kUniversalFilter)) { ColoredPrintf(COLOR_YELLOW, "Note: %s filter = %s\n", GTEST_NAME_, filter); } if (internal::ShouldShard(kTestTotalShards, kTestShardIndex, false)) { const Int32 shard_index = Int32FromEnvOrDie(kTestShardIndex, -1); ColoredPrintf(COLOR_YELLOW, "Note: This is test shard %d of %s.\n", static_cast(shard_index) + 1, internal::posix::GetEnv(kTestTotalShards)); } if (GTEST_FLAG(shuffle)) { ColoredPrintf(COLOR_YELLOW, "Note: Randomizing tests' orders with a seed of %d .\n", unit_test.random_seed()); } ColoredPrintf(COLOR_GREEN, "[==========] "); printf("Running %s from %s.\n", FormatTestCount(unit_test.test_to_run_count()).c_str(), FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str()); fflush(stdout); } void PrettyUnitTestResultPrinter::OnEnvironmentsSetUpStart( const UnitTest& /*unit_test*/) { ColoredPrintf(COLOR_GREEN, "[----------] "); printf("Global test environment set-up.\n"); fflush(stdout); } void PrettyUnitTestResultPrinter::OnTestCaseStart(const TestCase& test_case) { const std::string counts = FormatCountableNoun(test_case.test_to_run_count(), "test", "tests"); ColoredPrintf(COLOR_GREEN, "[----------] "); printf("%s from %s", counts.c_str(), test_case.name()); if (test_case.type_param() == NULL) { printf("\n"); } else { printf(", where %s = %s\n", kTypeParamLabel, test_case.type_param()); } fflush(stdout); } void PrettyUnitTestResultPrinter::OnTestStart(const TestInfo& test_info) { ColoredPrintf(COLOR_GREEN, "[ RUN ] "); PrintTestName(test_info.test_case_name(), test_info.name()); printf("\n"); fflush(stdout); } // Called after an assertion failure. void PrettyUnitTestResultPrinter::OnTestPartResult( const TestPartResult& result) { // If the test part succeeded, we don't need to do anything. if (result.type() == TestPartResult::kSuccess) return; // Print failure message from the assertion (e.g. expected this and got that). PrintTestPartResult(result); fflush(stdout); } void PrettyUnitTestResultPrinter::OnTestEnd(const TestInfo& test_info) { if (test_info.result()->Passed()) { ColoredPrintf(COLOR_GREEN, "[ OK ] "); } else { ColoredPrintf(COLOR_RED, "[ FAILED ] "); } PrintTestName(test_info.test_case_name(), test_info.name()); if (test_info.result()->Failed()) PrintFullTestCommentIfPresent(test_info); if (GTEST_FLAG(print_time)) { printf(" (%s ms)\n", internal::StreamableToString( test_info.result()->elapsed_time()).c_str()); } else { printf("\n"); } fflush(stdout); } void PrettyUnitTestResultPrinter::OnTestCaseEnd(const TestCase& test_case) { if (!GTEST_FLAG(print_time)) return; const std::string counts = FormatCountableNoun(test_case.test_to_run_count(), "test", "tests"); ColoredPrintf(COLOR_GREEN, "[----------] "); printf("%s from %s (%s ms total)\n\n", counts.c_str(), test_case.name(), internal::StreamableToString(test_case.elapsed_time()).c_str()); fflush(stdout); } void PrettyUnitTestResultPrinter::OnEnvironmentsTearDownStart( const UnitTest& /*unit_test*/) { ColoredPrintf(COLOR_GREEN, "[----------] "); printf("Global test environment tear-down\n"); fflush(stdout); } // Internal helper for printing the list of failed tests. void PrettyUnitTestResultPrinter::PrintFailedTests(const UnitTest& unit_test) { const int failed_test_count = unit_test.failed_test_count(); if (failed_test_count == 0) { return; } for (int i = 0; i < unit_test.total_test_case_count(); ++i) { const TestCase& test_case = *unit_test.GetTestCase(i); if (!test_case.should_run() || (test_case.failed_test_count() == 0)) { continue; } for (int j = 0; j < test_case.total_test_count(); ++j) { const TestInfo& test_info = *test_case.GetTestInfo(j); if (!test_info.should_run() || test_info.result()->Passed()) { continue; } ColoredPrintf(COLOR_RED, "[ FAILED ] "); printf("%s.%s", test_case.name(), test_info.name()); PrintFullTestCommentIfPresent(test_info); printf("\n"); } } } void PrettyUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test, int /*iteration*/) { ColoredPrintf(COLOR_GREEN, "[==========] "); printf("%s from %s ran.", FormatTestCount(unit_test.test_to_run_count()).c_str(), FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str()); if (GTEST_FLAG(print_time)) { printf(" (%s ms total)", internal::StreamableToString(unit_test.elapsed_time()).c_str()); } printf("\n"); ColoredPrintf(COLOR_GREEN, "[ PASSED ] "); printf("%s.\n", FormatTestCount(unit_test.successful_test_count()).c_str()); int num_failures = unit_test.failed_test_count(); if (!unit_test.Passed()) { const int failed_test_count = unit_test.failed_test_count(); ColoredPrintf(COLOR_RED, "[ FAILED ] "); printf("%s, listed below:\n", FormatTestCount(failed_test_count).c_str()); PrintFailedTests(unit_test); printf("\n%2d FAILED %s\n", num_failures, num_failures == 1 ? "TEST" : "TESTS"); } int num_disabled = unit_test.reportable_disabled_test_count(); if (num_disabled && !GTEST_FLAG(also_run_disabled_tests)) { if (!num_failures) { printf("\n"); // Add a spacer if no FAILURE banner is displayed. } ColoredPrintf(COLOR_YELLOW, " YOU HAVE %d DISABLED %s\n\n", num_disabled, num_disabled == 1 ? "TEST" : "TESTS"); } // Ensure that Google Test output is printed before, e.g., heapchecker output. fflush(stdout); } // End PrettyUnitTestResultPrinter // class TestEventRepeater // // This class forwards events to other event listeners. class TestEventRepeater : public TestEventListener { public: TestEventRepeater() : forwarding_enabled_(true) {} virtual ~TestEventRepeater(); void Append(TestEventListener *listener); TestEventListener* Release(TestEventListener* listener); // Controls whether events will be forwarded to listeners_. Set to false // in death test child processes. bool forwarding_enabled() const { return forwarding_enabled_; } void set_forwarding_enabled(bool enable) { forwarding_enabled_ = enable; } virtual void OnTestProgramStart(const UnitTest& unit_test); virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration); virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test); virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test); virtual void OnTestCaseStart(const TestCase& test_case); virtual void OnTestStart(const TestInfo& test_info); virtual void OnTestPartResult(const TestPartResult& result); virtual void OnTestEnd(const TestInfo& test_info); virtual void OnTestCaseEnd(const TestCase& test_case); virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test); virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test); virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration); virtual void OnTestProgramEnd(const UnitTest& unit_test); private: // Controls whether events will be forwarded to listeners_. Set to false // in death test child processes. bool forwarding_enabled_; // The list of listeners that receive events. std::vector listeners_; GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventRepeater); }; TestEventRepeater::~TestEventRepeater() { ForEach(listeners_, Delete); } void TestEventRepeater::Append(TestEventListener *listener) { listeners_.push_back(listener); } // TODO(vladl@google.com): Factor the search functionality into Vector::Find. TestEventListener* TestEventRepeater::Release(TestEventListener *listener) { for (size_t i = 0; i < listeners_.size(); ++i) { if (listeners_[i] == listener) { listeners_.erase(listeners_.begin() + i); return listener; } } return NULL; } // Since most methods are very similar, use macros to reduce boilerplate. // This defines a member that forwards the call to all listeners. #define GTEST_REPEATER_METHOD_(Name, Type) \ void TestEventRepeater::Name(const Type& parameter) { \ if (forwarding_enabled_) { \ for (size_t i = 0; i < listeners_.size(); i++) { \ listeners_[i]->Name(parameter); \ } \ } \ } // This defines a member that forwards the call to all listeners in reverse // order. #define GTEST_REVERSE_REPEATER_METHOD_(Name, Type) \ void TestEventRepeater::Name(const Type& parameter) { \ if (forwarding_enabled_) { \ for (int i = static_cast(listeners_.size()) - 1; i >= 0; i--) { \ listeners_[i]->Name(parameter); \ } \ } \ } GTEST_REPEATER_METHOD_(OnTestProgramStart, UnitTest) GTEST_REPEATER_METHOD_(OnEnvironmentsSetUpStart, UnitTest) GTEST_REPEATER_METHOD_(OnTestCaseStart, TestCase) GTEST_REPEATER_METHOD_(OnTestStart, TestInfo) GTEST_REPEATER_METHOD_(OnTestPartResult, TestPartResult) GTEST_REPEATER_METHOD_(OnEnvironmentsTearDownStart, UnitTest) GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsSetUpEnd, UnitTest) GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsTearDownEnd, UnitTest) GTEST_REVERSE_REPEATER_METHOD_(OnTestEnd, TestInfo) GTEST_REVERSE_REPEATER_METHOD_(OnTestCaseEnd, TestCase) GTEST_REVERSE_REPEATER_METHOD_(OnTestProgramEnd, UnitTest) #undef GTEST_REPEATER_METHOD_ #undef GTEST_REVERSE_REPEATER_METHOD_ void TestEventRepeater::OnTestIterationStart(const UnitTest& unit_test, int iteration) { if (forwarding_enabled_) { for (size_t i = 0; i < listeners_.size(); i++) { listeners_[i]->OnTestIterationStart(unit_test, iteration); } } } void TestEventRepeater::OnTestIterationEnd(const UnitTest& unit_test, int iteration) { if (forwarding_enabled_) { for (int i = static_cast(listeners_.size()) - 1; i >= 0; i--) { listeners_[i]->OnTestIterationEnd(unit_test, iteration); } } } // End TestEventRepeater // This class generates an XML output file. class XmlUnitTestResultPrinter : public EmptyTestEventListener { public: explicit XmlUnitTestResultPrinter(const char* output_file); virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration); private: // Is c a whitespace character that is normalized to a space character // when it appears in an XML attribute value? static bool IsNormalizableWhitespace(char c) { return c == 0x9 || c == 0xA || c == 0xD; } // May c appear in a well-formed XML document? static bool IsValidXmlCharacter(char c) { return IsNormalizableWhitespace(c) || c >= 0x20; } // Returns an XML-escaped copy of the input string str. If // is_attribute is true, the text is meant to appear as an attribute // value, and normalizable whitespace is preserved by replacing it // with character references. static std::string EscapeXml(const std::string& str, bool is_attribute); // Returns the given string with all characters invalid in XML removed. static std::string RemoveInvalidXmlCharacters(const std::string& str); // Convenience wrapper around EscapeXml when str is an attribute value. static std::string EscapeXmlAttribute(const std::string& str) { return EscapeXml(str, true); } // Convenience wrapper around EscapeXml when str is not an attribute value. static std::string EscapeXmlText(const char* str) { return EscapeXml(str, false); } // Verifies that the given attribute belongs to the given element and // streams the attribute as XML. static void OutputXmlAttribute(std::ostream* stream, const std::string& element_name, const std::string& name, const std::string& value); // Streams an XML CDATA section, escaping invalid CDATA sequences as needed. static void OutputXmlCDataSection(::std::ostream* stream, const char* data); // Streams an XML representation of a TestInfo object. static void OutputXmlTestInfo(::std::ostream* stream, const char* test_case_name, const TestInfo& test_info); // Prints an XML representation of a TestCase object static void PrintXmlTestCase(::std::ostream* stream, const TestCase& test_case); // Prints an XML summary of unit_test to output stream out. static void PrintXmlUnitTest(::std::ostream* stream, const UnitTest& unit_test); // Produces a string representing the test properties in a result as space // delimited XML attributes based on the property key="value" pairs. // When the std::string is not empty, it includes a space at the beginning, // to delimit this attribute from prior attributes. static std::string TestPropertiesAsXmlAttributes(const TestResult& result); // The output file. const std::string output_file_; GTEST_DISALLOW_COPY_AND_ASSIGN_(XmlUnitTestResultPrinter); }; // Creates a new XmlUnitTestResultPrinter. XmlUnitTestResultPrinter::XmlUnitTestResultPrinter(const char* output_file) : output_file_(output_file) { if (output_file_.c_str() == NULL || output_file_.empty()) { fprintf(stderr, "XML output file may not be null\n"); fflush(stderr); exit(EXIT_FAILURE); } } // Called after the unit test ends. void XmlUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test, int /*iteration*/) { FILE* xmlout = NULL; FilePath output_file(output_file_); FilePath output_dir(output_file.RemoveFileName()); if (output_dir.CreateDirectoriesRecursively()) { xmlout = posix::FOpen(output_file_.c_str(), "w"); } if (xmlout == NULL) { // TODO(wan): report the reason of the failure. // // We don't do it for now as: // // 1. There is no urgent need for it. // 2. It's a bit involved to make the errno variable thread-safe on // all three operating systems (Linux, Windows, and Mac OS). // 3. To interpret the meaning of errno in a thread-safe way, // we need the strerror_r() function, which is not available on // Windows. fprintf(stderr, "Unable to open file \"%s\"\n", output_file_.c_str()); fflush(stderr); exit(EXIT_FAILURE); } std::stringstream stream; PrintXmlUnitTest(&stream, unit_test); fprintf(xmlout, "%s", StringStreamToString(&stream).c_str()); fclose(xmlout); } // Returns an XML-escaped copy of the input string str. If is_attribute // is true, the text is meant to appear as an attribute value, and // normalizable whitespace is preserved by replacing it with character // references. // // Invalid XML characters in str, if any, are stripped from the output. // It is expected that most, if not all, of the text processed by this // module will consist of ordinary English text. // If this module is ever modified to produce version 1.1 XML output, // most invalid characters can be retained using character references. // TODO(wan): It might be nice to have a minimally invasive, human-readable // escaping scheme for invalid characters, rather than dropping them. std::string XmlUnitTestResultPrinter::EscapeXml( const std::string& str, bool is_attribute) { Message m; for (size_t i = 0; i < str.size(); ++i) { const char ch = str[i]; switch (ch) { case '<': m << "<"; break; case '>': m << ">"; break; case '&': m << "&"; break; case '\'': if (is_attribute) m << "'"; else m << '\''; break; case '"': if (is_attribute) m << """; else m << '"'; break; default: if (IsValidXmlCharacter(ch)) { if (is_attribute && IsNormalizableWhitespace(ch)) m << "&#x" << String::FormatByte(static_cast(ch)) << ";"; else m << ch; } break; } } return m.GetString(); } // Returns the given string with all characters invalid in XML removed. // Currently invalid characters are dropped from the string. An // alternative is to replace them with certain characters such as . or ?. std::string XmlUnitTestResultPrinter::RemoveInvalidXmlCharacters( const std::string& str) { std::string output; output.reserve(str.size()); for (std::string::const_iterator it = str.begin(); it != str.end(); ++it) if (IsValidXmlCharacter(*it)) output.push_back(*it); return output; } // The following routines generate an XML representation of a UnitTest // object. // // This is how Google Test concepts map to the DTD: // // <-- corresponds to a UnitTest object // <-- corresponds to a TestCase object // <-- corresponds to a TestInfo object // ... // ... // ... // <-- individual assertion failures // // // // Formats the given time in milliseconds as seconds. std::string FormatTimeInMillisAsSeconds(TimeInMillis ms) { ::std::stringstream ss; ss << ms/1000.0; return ss.str(); } static bool PortableLocaltime(time_t seconds, struct tm* out) { #if defined(_MSC_VER) return localtime_s(out, &seconds) == 0; #elif defined(__MINGW32__) || defined(__MINGW64__) // MINGW provides neither localtime_r nor localtime_s, but uses // Windows' localtime(), which has a thread-local tm buffer. struct tm* tm_ptr = localtime(&seconds); // NOLINT if (tm_ptr == NULL) return false; *out = *tm_ptr; return true; #else return localtime_r(&seconds, out) != NULL; #endif } // Converts the given epoch time in milliseconds to a date string in the ISO // 8601 format, without the timezone information. std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms) { struct tm time_struct; if (!PortableLocaltime(static_cast(ms / 1000), &time_struct)) return ""; // YYYY-MM-DDThh:mm:ss return StreamableToString(time_struct.tm_year + 1900) + "-" + String::FormatIntWidth2(time_struct.tm_mon + 1) + "-" + String::FormatIntWidth2(time_struct.tm_mday) + "T" + String::FormatIntWidth2(time_struct.tm_hour) + ":" + String::FormatIntWidth2(time_struct.tm_min) + ":" + String::FormatIntWidth2(time_struct.tm_sec); } // Streams an XML CDATA section, escaping invalid CDATA sequences as needed. void XmlUnitTestResultPrinter::OutputXmlCDataSection(::std::ostream* stream, const char* data) { const char* segment = data; *stream << ""); if (next_segment != NULL) { stream->write( segment, static_cast(next_segment - segment)); *stream << "]]>]]>"); } else { *stream << segment; break; } } *stream << "]]>"; } void XmlUnitTestResultPrinter::OutputXmlAttribute( std::ostream* stream, const std::string& element_name, const std::string& name, const std::string& value) { const std::vector& allowed_names = GetReservedAttributesForElement(element_name); GTEST_CHECK_(std::find(allowed_names.begin(), allowed_names.end(), name) != allowed_names.end()) << "Attribute " << name << " is not allowed for element <" << element_name << ">."; *stream << " " << name << "=\"" << EscapeXmlAttribute(value) << "\""; } // Prints an XML representation of a TestInfo object. // TODO(wan): There is also value in printing properties with the plain printer. void XmlUnitTestResultPrinter::OutputXmlTestInfo(::std::ostream* stream, const char* test_case_name, const TestInfo& test_info) { const TestResult& result = *test_info.result(); const std::string kTestcase = "testcase"; *stream << " \n"; } const string location = internal::FormatCompilerIndependentFileLocation( part.file_name(), part.line_number()); const string summary = location + "\n" + part.summary(); *stream << " "; const string detail = location + "\n" + part.message(); OutputXmlCDataSection(stream, RemoveInvalidXmlCharacters(detail).c_str()); *stream << "\n"; } } if (failures == 0) *stream << " />\n"; else *stream << " \n"; } // Prints an XML representation of a TestCase object void XmlUnitTestResultPrinter::PrintXmlTestCase(std::ostream* stream, const TestCase& test_case) { const std::string kTestsuite = "testsuite"; *stream << " <" << kTestsuite; OutputXmlAttribute(stream, kTestsuite, "name", test_case.name()); OutputXmlAttribute(stream, kTestsuite, "tests", StreamableToString(test_case.reportable_test_count())); OutputXmlAttribute(stream, kTestsuite, "failures", StreamableToString(test_case.failed_test_count())); OutputXmlAttribute( stream, kTestsuite, "disabled", StreamableToString(test_case.reportable_disabled_test_count())); OutputXmlAttribute(stream, kTestsuite, "errors", "0"); OutputXmlAttribute(stream, kTestsuite, "time", FormatTimeInMillisAsSeconds(test_case.elapsed_time())); *stream << TestPropertiesAsXmlAttributes(test_case.ad_hoc_test_result()) << ">\n"; for (int i = 0; i < test_case.total_test_count(); ++i) { if (test_case.GetTestInfo(i)->is_reportable()) OutputXmlTestInfo(stream, test_case.name(), *test_case.GetTestInfo(i)); } *stream << " \n"; } // Prints an XML summary of unit_test to output stream out. void XmlUnitTestResultPrinter::PrintXmlUnitTest(std::ostream* stream, const UnitTest& unit_test) { const std::string kTestsuites = "testsuites"; *stream << "\n"; *stream << "<" << kTestsuites; OutputXmlAttribute(stream, kTestsuites, "tests", StreamableToString(unit_test.reportable_test_count())); OutputXmlAttribute(stream, kTestsuites, "failures", StreamableToString(unit_test.failed_test_count())); OutputXmlAttribute( stream, kTestsuites, "disabled", StreamableToString(unit_test.reportable_disabled_test_count())); OutputXmlAttribute(stream, kTestsuites, "errors", "0"); OutputXmlAttribute( stream, kTestsuites, "timestamp", FormatEpochTimeInMillisAsIso8601(unit_test.start_timestamp())); OutputXmlAttribute(stream, kTestsuites, "time", FormatTimeInMillisAsSeconds(unit_test.elapsed_time())); if (GTEST_FLAG(shuffle)) { OutputXmlAttribute(stream, kTestsuites, "random_seed", StreamableToString(unit_test.random_seed())); } *stream << TestPropertiesAsXmlAttributes(unit_test.ad_hoc_test_result()); OutputXmlAttribute(stream, kTestsuites, "name", "AllTests"); *stream << ">\n"; for (int i = 0; i < unit_test.total_test_case_count(); ++i) { if (unit_test.GetTestCase(i)->reportable_test_count() > 0) PrintXmlTestCase(stream, *unit_test.GetTestCase(i)); } *stream << "\n"; } // Produces a string representing the test properties in a result as space // delimited XML attributes based on the property key="value" pairs. std::string XmlUnitTestResultPrinter::TestPropertiesAsXmlAttributes( const TestResult& result) { Message attributes; for (int i = 0; i < result.test_property_count(); ++i) { const TestProperty& property = result.GetTestProperty(i); attributes << " " << property.key() << "=" << "\"" << EscapeXmlAttribute(property.value()) << "\""; } return attributes.GetString(); } // End XmlUnitTestResultPrinter #if GTEST_CAN_STREAM_RESULTS_ // Checks if str contains '=', '&', '%' or '\n' characters. If yes, // replaces them by "%xx" where xx is their hexadecimal value. For // example, replaces "=" with "%3D". This algorithm is O(strlen(str)) // in both time and space -- important as the input str may contain an // arbitrarily long test failure message and stack trace. string StreamingListener::UrlEncode(const char* str) { string result; result.reserve(strlen(str) + 1); for (char ch = *str; ch != '\0'; ch = *++str) { switch (ch) { case '%': case '=': case '&': case '\n': result.append("%" + String::FormatByte(static_cast(ch))); break; default: result.push_back(ch); break; } } return result; } void StreamingListener::SocketWriter::MakeConnection() { GTEST_CHECK_(sockfd_ == -1) << "MakeConnection() can't be called when there is already a connection."; addrinfo hints; memset(&hints, 0, sizeof(hints)); hints.ai_family = AF_UNSPEC; // To allow both IPv4 and IPv6 addresses. hints.ai_socktype = SOCK_STREAM; addrinfo* servinfo = NULL; // Use the getaddrinfo() to get a linked list of IP addresses for // the given host name. const int error_num = getaddrinfo( host_name_.c_str(), port_num_.c_str(), &hints, &servinfo); if (error_num != 0) { GTEST_LOG_(WARNING) << "stream_result_to: getaddrinfo() failed: " << gai_strerror(error_num); } // Loop through all the results and connect to the first we can. for (addrinfo* cur_addr = servinfo; sockfd_ == -1 && cur_addr != NULL; cur_addr = cur_addr->ai_next) { sockfd_ = socket( cur_addr->ai_family, cur_addr->ai_socktype, cur_addr->ai_protocol); if (sockfd_ != -1) { // Connect the client socket to the server socket. if (connect(sockfd_, cur_addr->ai_addr, cur_addr->ai_addrlen) == -1) { close(sockfd_); sockfd_ = -1; } } } freeaddrinfo(servinfo); // all done with this structure if (sockfd_ == -1) { GTEST_LOG_(WARNING) << "stream_result_to: failed to connect to " << host_name_ << ":" << port_num_; } } // End of class Streaming Listener #endif // GTEST_CAN_STREAM_RESULTS__ // Class ScopedTrace // Pushes the given source file location and message onto a per-thread // trace stack maintained by Google Test. ScopedTrace::ScopedTrace(const char* file, int line, const Message& message) GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) { TraceInfo trace; trace.file = file; trace.line = line; trace.message = message.GetString(); UnitTest::GetInstance()->PushGTestTrace(trace); } // Pops the info pushed by the c'tor. ScopedTrace::~ScopedTrace() GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) { UnitTest::GetInstance()->PopGTestTrace(); } // class OsStackTraceGetter // Returns the current OS stack trace as an std::string. Parameters: // // max_depth - the maximum number of stack frames to be included // in the trace. // skip_count - the number of top frames to be skipped; doesn't count // against max_depth. // string OsStackTraceGetter::CurrentStackTrace(int /* max_depth */, int /* skip_count */) GTEST_LOCK_EXCLUDED_(mutex_) { return ""; } void OsStackTraceGetter::UponLeavingGTest() GTEST_LOCK_EXCLUDED_(mutex_) { } const char* const OsStackTraceGetter::kElidedFramesMarker = "... " GTEST_NAME_ " internal frames ..."; // A helper class that creates the premature-exit file in its // constructor and deletes the file in its destructor. class ScopedPrematureExitFile { public: explicit ScopedPrematureExitFile(const char* premature_exit_filepath) : premature_exit_filepath_(premature_exit_filepath) { // If a path to the premature-exit file is specified... if (premature_exit_filepath != NULL && *premature_exit_filepath != '\0') { // create the file with a single "0" character in it. I/O // errors are ignored as there's nothing better we can do and we // don't want to fail the test because of this. FILE* pfile = posix::FOpen(premature_exit_filepath, "w"); fwrite("0", 1, 1, pfile); fclose(pfile); } } ~ScopedPrematureExitFile() { if (premature_exit_filepath_ != NULL && *premature_exit_filepath_ != '\0') { remove(premature_exit_filepath_); } } private: const char* const premature_exit_filepath_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedPrematureExitFile); }; } // namespace internal // class TestEventListeners TestEventListeners::TestEventListeners() : repeater_(new internal::TestEventRepeater()), default_result_printer_(NULL), default_xml_generator_(NULL) { } TestEventListeners::~TestEventListeners() { delete repeater_; } // Returns the standard listener responsible for the default console // output. Can be removed from the listeners list to shut down default // console output. Note that removing this object from the listener list // with Release transfers its ownership to the user. void TestEventListeners::Append(TestEventListener* listener) { repeater_->Append(listener); } // Removes the given event listener from the list and returns it. It then // becomes the caller's responsibility to delete the listener. Returns // NULL if the listener is not found in the list. TestEventListener* TestEventListeners::Release(TestEventListener* listener) { if (listener == default_result_printer_) default_result_printer_ = NULL; else if (listener == default_xml_generator_) default_xml_generator_ = NULL; return repeater_->Release(listener); } // Returns repeater that broadcasts the TestEventListener events to all // subscribers. TestEventListener* TestEventListeners::repeater() { return repeater_; } // Sets the default_result_printer attribute to the provided listener. // The listener is also added to the listener list and previous // default_result_printer is removed from it and deleted. The listener can // also be NULL in which case it will not be added to the list. Does // nothing if the previous and the current listener objects are the same. void TestEventListeners::SetDefaultResultPrinter(TestEventListener* listener) { if (default_result_printer_ != listener) { // It is an error to pass this method a listener that is already in the // list. delete Release(default_result_printer_); default_result_printer_ = listener; if (listener != NULL) Append(listener); } } // Sets the default_xml_generator attribute to the provided listener. The // listener is also added to the listener list and previous // default_xml_generator is removed from it and deleted. The listener can // also be NULL in which case it will not be added to the list. Does // nothing if the previous and the current listener objects are the same. void TestEventListeners::SetDefaultXmlGenerator(TestEventListener* listener) { if (default_xml_generator_ != listener) { // It is an error to pass this method a listener that is already in the // list. delete Release(default_xml_generator_); default_xml_generator_ = listener; if (listener != NULL) Append(listener); } } // Controls whether events will be forwarded by the repeater to the // listeners in the list. bool TestEventListeners::EventForwardingEnabled() const { return repeater_->forwarding_enabled(); } void TestEventListeners::SuppressEventForwarding() { repeater_->set_forwarding_enabled(false); } // class UnitTest // Gets the singleton UnitTest object. The first time this method is // called, a UnitTest object is constructed and returned. Consecutive // calls will return the same object. // // We don't protect this under mutex_ as a user is not supposed to // call this before main() starts, from which point on the return // value will never change. UnitTest* UnitTest::GetInstance() { // When compiled with MSVC 7.1 in optimized mode, destroying the // UnitTest object upon exiting the program messes up the exit code, // causing successful tests to appear failed. We have to use a // different implementation in this case to bypass the compiler bug. // This implementation makes the compiler happy, at the cost of // leaking the UnitTest object. // CodeGear C++Builder insists on a public destructor for the // default implementation. Use this implementation to keep good OO // design with private destructor. #if (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__) static UnitTest* const instance = new UnitTest; return instance; #else static UnitTest instance; return &instance; #endif // (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__) } // Gets the number of successful test cases. int UnitTest::successful_test_case_count() const { return impl()->successful_test_case_count(); } // Gets the number of failed test cases. int UnitTest::failed_test_case_count() const { return impl()->failed_test_case_count(); } // Gets the number of all test cases. int UnitTest::total_test_case_count() const { return impl()->total_test_case_count(); } // Gets the number of all test cases that contain at least one test // that should run. int UnitTest::test_case_to_run_count() const { return impl()->test_case_to_run_count(); } // Gets the number of successful tests. int UnitTest::successful_test_count() const { return impl()->successful_test_count(); } // Gets the number of failed tests. int UnitTest::failed_test_count() const { return impl()->failed_test_count(); } // Gets the number of disabled tests that will be reported in the XML report. int UnitTest::reportable_disabled_test_count() const { return impl()->reportable_disabled_test_count(); } // Gets the number of disabled tests. int UnitTest::disabled_test_count() const { return impl()->disabled_test_count(); } // Gets the number of tests to be printed in the XML report. int UnitTest::reportable_test_count() const { return impl()->reportable_test_count(); } // Gets the number of all tests. int UnitTest::total_test_count() const { return impl()->total_test_count(); } // Gets the number of tests that should run. int UnitTest::test_to_run_count() const { return impl()->test_to_run_count(); } // Gets the time of the test program start, in ms from the start of the // UNIX epoch. internal::TimeInMillis UnitTest::start_timestamp() const { return impl()->start_timestamp(); } // Gets the elapsed time, in milliseconds. internal::TimeInMillis UnitTest::elapsed_time() const { return impl()->elapsed_time(); } // Returns true iff the unit test passed (i.e. all test cases passed). bool UnitTest::Passed() const { return impl()->Passed(); } // Returns true iff the unit test failed (i.e. some test case failed // or something outside of all tests failed). bool UnitTest::Failed() const { return impl()->Failed(); } // Gets the i-th test case among all the test cases. i can range from 0 to // total_test_case_count() - 1. If i is not in that range, returns NULL. const TestCase* UnitTest::GetTestCase(int i) const { return impl()->GetTestCase(i); } // Returns the TestResult containing information on test failures and // properties logged outside of individual test cases. const TestResult& UnitTest::ad_hoc_test_result() const { return *impl()->ad_hoc_test_result(); } // Gets the i-th test case among all the test cases. i can range from 0 to // total_test_case_count() - 1. If i is not in that range, returns NULL. TestCase* UnitTest::GetMutableTestCase(int i) { return impl()->GetMutableTestCase(i); } // Returns the list of event listeners that can be used to track events // inside Google Test. TestEventListeners& UnitTest::listeners() { return *impl()->listeners(); } // Registers and returns a global test environment. When a test // program is run, all global test environments will be set-up in the // order they were registered. After all tests in the program have // finished, all global test environments will be torn-down in the // *reverse* order they were registered. // // The UnitTest object takes ownership of the given environment. // // We don't protect this under mutex_, as we only support calling it // from the main thread. Environment* UnitTest::AddEnvironment(Environment* env) { if (env == NULL) { return NULL; } impl_->environments().push_back(env); return env; } // Adds a TestPartResult to the current TestResult object. All Google Test // assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) eventually call // this to report their results. The user code should use the // assertion macros instead of calling this directly. void UnitTest::AddTestPartResult( TestPartResult::Type result_type, const char* file_name, int line_number, const std::string& message, const std::string& os_stack_trace) GTEST_LOCK_EXCLUDED_(mutex_) { Message msg; msg << message; internal::MutexLock lock(&mutex_); if (impl_->gtest_trace_stack().size() > 0) { msg << "\n" << GTEST_NAME_ << " trace:"; for (int i = static_cast(impl_->gtest_trace_stack().size()); i > 0; --i) { const internal::TraceInfo& trace = impl_->gtest_trace_stack()[i - 1]; msg << "\n" << internal::FormatFileLocation(trace.file, trace.line) << " " << trace.message; } } if (os_stack_trace.c_str() != NULL && !os_stack_trace.empty()) { msg << internal::kStackTraceMarker << os_stack_trace; } const TestPartResult result = TestPartResult(result_type, file_name, line_number, msg.GetString().c_str()); impl_->GetTestPartResultReporterForCurrentThread()-> ReportTestPartResult(result); if (result_type != TestPartResult::kSuccess) { // gtest_break_on_failure takes precedence over // gtest_throw_on_failure. This allows a user to set the latter // in the code (perhaps in order to use Google Test assertions // with another testing framework) and specify the former on the // command line for debugging. if (GTEST_FLAG(break_on_failure)) { #if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT // Using DebugBreak on Windows allows gtest to still break into a debugger // when a failure happens and both the --gtest_break_on_failure and // the --gtest_catch_exceptions flags are specified. DebugBreak(); #else // Dereference NULL through a volatile pointer to prevent the compiler // from removing. We use this rather than abort() or __builtin_trap() for // portability: Symbian doesn't implement abort() well, and some debuggers // don't correctly trap abort(). *static_cast(NULL) = 1; #endif // GTEST_OS_WINDOWS } else if (GTEST_FLAG(throw_on_failure)) { #if GTEST_HAS_EXCEPTIONS throw internal::GoogleTestFailureException(result); #else // We cannot call abort() as it generates a pop-up in debug mode // that cannot be suppressed in VC 7.1 or below. exit(1); #endif } } } // Adds a TestProperty to the current TestResult object when invoked from // inside a test, to current TestCase's ad_hoc_test_result_ when invoked // from SetUpTestCase or TearDownTestCase, or to the global property set // when invoked elsewhere. If the result already contains a property with // the same key, the value will be updated. void UnitTest::RecordProperty(const std::string& key, const std::string& value) { impl_->RecordProperty(TestProperty(key, value)); } // Runs all tests in this UnitTest object and prints the result. // Returns 0 if successful, or 1 otherwise. // // We don't protect this under mutex_, as we only support calling it // from the main thread. int UnitTest::Run() { const bool in_death_test_child_process = internal::GTEST_FLAG(internal_run_death_test).length() > 0; // Google Test implements this protocol for catching that a test // program exits before returning control to Google Test: // // 1. Upon start, Google Test creates a file whose absolute path // is specified by the environment variable // TEST_PREMATURE_EXIT_FILE. // 2. When Google Test has finished its work, it deletes the file. // // This allows a test runner to set TEST_PREMATURE_EXIT_FILE before // running a Google-Test-based test program and check the existence // of the file at the end of the test execution to see if it has // exited prematurely. // If we are in the child process of a death test, don't // create/delete the premature exit file, as doing so is unnecessary // and will confuse the parent process. Otherwise, create/delete // the file upon entering/leaving this function. If the program // somehow exits before this function has a chance to return, the // premature-exit file will be left undeleted, causing a test runner // that understands the premature-exit-file protocol to report the // test as having failed. const internal::ScopedPrematureExitFile premature_exit_file( in_death_test_child_process ? NULL : internal::posix::GetEnv("TEST_PREMATURE_EXIT_FILE")); // Captures the value of GTEST_FLAG(catch_exceptions). This value will be // used for the duration of the program. impl()->set_catch_exceptions(GTEST_FLAG(catch_exceptions)); #if GTEST_HAS_SEH // Either the user wants Google Test to catch exceptions thrown by the // tests or this is executing in the context of death test child // process. In either case the user does not want to see pop-up dialogs // about crashes - they are expected. if (impl()->catch_exceptions() || in_death_test_child_process) { # if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT // SetErrorMode doesn't exist on CE. SetErrorMode(SEM_FAILCRITICALERRORS | SEM_NOALIGNMENTFAULTEXCEPT | SEM_NOGPFAULTERRORBOX | SEM_NOOPENFILEERRORBOX); # endif // !GTEST_OS_WINDOWS_MOBILE # if (defined(_MSC_VER) || GTEST_OS_WINDOWS_MINGW) && !GTEST_OS_WINDOWS_MOBILE // Death test children can be terminated with _abort(). On Windows, // _abort() can show a dialog with a warning message. This forces the // abort message to go to stderr instead. _set_error_mode(_OUT_TO_STDERR); # endif # if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE // In the debug version, Visual Studio pops up a separate dialog // offering a choice to debug the aborted program. We need to suppress // this dialog or it will pop up for every EXPECT/ASSERT_DEATH statement // executed. Google Test will notify the user of any unexpected // failure via stderr. // // VC++ doesn't define _set_abort_behavior() prior to the version 8.0. // Users of prior VC versions shall suffer the agony and pain of // clicking through the countless debug dialogs. // TODO(vladl@google.com): find a way to suppress the abort dialog() in the // debug mode when compiled with VC 7.1 or lower. if (!GTEST_FLAG(break_on_failure)) _set_abort_behavior( 0x0, // Clear the following flags: _WRITE_ABORT_MSG | _CALL_REPORTFAULT); // pop-up window, core dump. # endif } #endif // GTEST_HAS_SEH return internal::HandleExceptionsInMethodIfSupported( impl(), &internal::UnitTestImpl::RunAllTests, "auxiliary test code (environments or event listeners)") ? 0 : 1; } // Returns the working directory when the first TEST() or TEST_F() was // executed. const char* UnitTest::original_working_dir() const { return impl_->original_working_dir_.c_str(); } // Returns the TestCase object for the test that's currently running, // or NULL if no test is running. const TestCase* UnitTest::current_test_case() const GTEST_LOCK_EXCLUDED_(mutex_) { internal::MutexLock lock(&mutex_); return impl_->current_test_case(); } // Returns the TestInfo object for the test that's currently running, // or NULL if no test is running. const TestInfo* UnitTest::current_test_info() const GTEST_LOCK_EXCLUDED_(mutex_) { internal::MutexLock lock(&mutex_); return impl_->current_test_info(); } // Returns the random seed used at the start of the current test run. int UnitTest::random_seed() const { return impl_->random_seed(); } #if GTEST_HAS_PARAM_TEST // Returns ParameterizedTestCaseRegistry object used to keep track of // value-parameterized tests and instantiate and register them. internal::ParameterizedTestCaseRegistry& UnitTest::parameterized_test_registry() GTEST_LOCK_EXCLUDED_(mutex_) { return impl_->parameterized_test_registry(); } #endif // GTEST_HAS_PARAM_TEST // Creates an empty UnitTest. UnitTest::UnitTest() { impl_ = new internal::UnitTestImpl(this); } // Destructor of UnitTest. UnitTest::~UnitTest() { delete impl_; } // Pushes a trace defined by SCOPED_TRACE() on to the per-thread // Google Test trace stack. void UnitTest::PushGTestTrace(const internal::TraceInfo& trace) GTEST_LOCK_EXCLUDED_(mutex_) { internal::MutexLock lock(&mutex_); impl_->gtest_trace_stack().push_back(trace); } // Pops a trace from the per-thread Google Test trace stack. void UnitTest::PopGTestTrace() GTEST_LOCK_EXCLUDED_(mutex_) { internal::MutexLock lock(&mutex_); impl_->gtest_trace_stack().pop_back(); } namespace internal { UnitTestImpl::UnitTestImpl(UnitTest* parent) : parent_(parent), GTEST_DISABLE_MSC_WARNINGS_PUSH_(4355 /* using this in initializer */) default_global_test_part_result_reporter_(this), default_per_thread_test_part_result_reporter_(this), GTEST_DISABLE_MSC_WARNINGS_POP_() global_test_part_result_repoter_( &default_global_test_part_result_reporter_), per_thread_test_part_result_reporter_( &default_per_thread_test_part_result_reporter_), #if GTEST_HAS_PARAM_TEST parameterized_test_registry_(), parameterized_tests_registered_(false), #endif // GTEST_HAS_PARAM_TEST last_death_test_case_(-1), current_test_case_(NULL), current_test_info_(NULL), ad_hoc_test_result_(), os_stack_trace_getter_(NULL), post_flag_parse_init_performed_(false), random_seed_(0), // Will be overridden by the flag before first use. random_(0), // Will be reseeded before first use. start_timestamp_(0), elapsed_time_(0), #if GTEST_HAS_DEATH_TEST death_test_factory_(new DefaultDeathTestFactory), #endif // Will be overridden by the flag before first use. catch_exceptions_(false) { listeners()->SetDefaultResultPrinter(new PrettyUnitTestResultPrinter); } UnitTestImpl::~UnitTestImpl() { // Deletes every TestCase. ForEach(test_cases_, internal::Delete); // Deletes every Environment. ForEach(environments_, internal::Delete); delete os_stack_trace_getter_; } // Adds a TestProperty to the current TestResult object when invoked in a // context of a test, to current test case's ad_hoc_test_result when invoke // from SetUpTestCase/TearDownTestCase, or to the global property set // otherwise. If the result already contains a property with the same key, // the value will be updated. void UnitTestImpl::RecordProperty(const TestProperty& test_property) { std::string xml_element; TestResult* test_result; // TestResult appropriate for property recording. if (current_test_info_ != NULL) { xml_element = "testcase"; test_result = &(current_test_info_->result_); } else if (current_test_case_ != NULL) { xml_element = "testsuite"; test_result = &(current_test_case_->ad_hoc_test_result_); } else { xml_element = "testsuites"; test_result = &ad_hoc_test_result_; } test_result->RecordProperty(xml_element, test_property); } #if GTEST_HAS_DEATH_TEST // Disables event forwarding if the control is currently in a death test // subprocess. Must not be called before InitGoogleTest. void UnitTestImpl::SuppressTestEventsIfInSubprocess() { if (internal_run_death_test_flag_.get() != NULL) listeners()->SuppressEventForwarding(); } #endif // GTEST_HAS_DEATH_TEST // Initializes event listeners performing XML output as specified by // UnitTestOptions. Must not be called before InitGoogleTest. void UnitTestImpl::ConfigureXmlOutput() { const std::string& output_format = UnitTestOptions::GetOutputFormat(); if (output_format == "xml") { listeners()->SetDefaultXmlGenerator(new XmlUnitTestResultPrinter( UnitTestOptions::GetAbsolutePathToOutputFile().c_str())); } else if (output_format != "") { printf("WARNING: unrecognized output format \"%s\" ignored.\n", output_format.c_str()); fflush(stdout); } } #if GTEST_CAN_STREAM_RESULTS_ // Initializes event listeners for streaming test results in string form. // Must not be called before InitGoogleTest. void UnitTestImpl::ConfigureStreamingOutput() { const std::string& target = GTEST_FLAG(stream_result_to); if (!target.empty()) { const size_t pos = target.find(':'); if (pos != std::string::npos) { listeners()->Append(new StreamingListener(target.substr(0, pos), target.substr(pos+1))); } else { printf("WARNING: unrecognized streaming target \"%s\" ignored.\n", target.c_str()); fflush(stdout); } } } #endif // GTEST_CAN_STREAM_RESULTS_ // Performs initialization dependent upon flag values obtained in // ParseGoogleTestFlagsOnly. Is called from InitGoogleTest after the call to // ParseGoogleTestFlagsOnly. In case a user neglects to call InitGoogleTest // this function is also called from RunAllTests. Since this function can be // called more than once, it has to be idempotent. void UnitTestImpl::PostFlagParsingInit() { // Ensures that this function does not execute more than once. if (!post_flag_parse_init_performed_) { post_flag_parse_init_performed_ = true; #if GTEST_HAS_DEATH_TEST InitDeathTestSubprocessControlInfo(); SuppressTestEventsIfInSubprocess(); #endif // GTEST_HAS_DEATH_TEST // Registers parameterized tests. This makes parameterized tests // available to the UnitTest reflection API without running // RUN_ALL_TESTS. RegisterParameterizedTests(); // Configures listeners for XML output. This makes it possible for users // to shut down the default XML output before invoking RUN_ALL_TESTS. ConfigureXmlOutput(); #if GTEST_CAN_STREAM_RESULTS_ // Configures listeners for streaming test results to the specified server. ConfigureStreamingOutput(); #endif // GTEST_CAN_STREAM_RESULTS_ } } // A predicate that checks the name of a TestCase against a known // value. // // This is used for implementation of the UnitTest class only. We put // it in the anonymous namespace to prevent polluting the outer // namespace. // // TestCaseNameIs is copyable. class TestCaseNameIs { public: // Constructor. explicit TestCaseNameIs(const std::string& name) : name_(name) {} // Returns true iff the name of test_case matches name_. bool operator()(const TestCase* test_case) const { return test_case != NULL && strcmp(test_case->name(), name_.c_str()) == 0; } private: std::string name_; }; // Finds and returns a TestCase with the given name. If one doesn't // exist, creates one and returns it. It's the CALLER'S // RESPONSIBILITY to ensure that this function is only called WHEN THE // TESTS ARE NOT SHUFFLED. // // Arguments: // // test_case_name: name of the test case // type_param: the name of the test case's type parameter, or NULL if // this is not a typed or a type-parameterized test case. // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case TestCase* UnitTestImpl::GetTestCase(const char* test_case_name, const char* type_param, Test::SetUpTestCaseFunc set_up_tc, Test::TearDownTestCaseFunc tear_down_tc) { // Can we find a TestCase with the given name? const std::vector::const_iterator test_case = std::find_if(test_cases_.begin(), test_cases_.end(), TestCaseNameIs(test_case_name)); if (test_case != test_cases_.end()) return *test_case; // No. Let's create one. TestCase* const new_test_case = new TestCase(test_case_name, type_param, set_up_tc, tear_down_tc); // Is this a death test case? if (internal::UnitTestOptions::MatchesFilter(test_case_name, kDeathTestCaseFilter)) { // Yes. Inserts the test case after the last death test case // defined so far. This only works when the test cases haven't // been shuffled. Otherwise we may end up running a death test // after a non-death test. ++last_death_test_case_; test_cases_.insert(test_cases_.begin() + last_death_test_case_, new_test_case); } else { // No. Appends to the end of the list. test_cases_.push_back(new_test_case); } test_case_indices_.push_back(static_cast(test_case_indices_.size())); return new_test_case; } // Helpers for setting up / tearing down the given environment. They // are for use in the ForEach() function. static void SetUpEnvironment(Environment* env) { env->SetUp(); } static void TearDownEnvironment(Environment* env) { env->TearDown(); } // Runs all tests in this UnitTest object, prints the result, and // returns true if all tests are successful. If any exception is // thrown during a test, the test is considered to be failed, but the // rest of the tests will still be run. // // When parameterized tests are enabled, it expands and registers // parameterized tests first in RegisterParameterizedTests(). // All other functions called from RunAllTests() may safely assume that // parameterized tests are ready to be counted and run. bool UnitTestImpl::RunAllTests() { // Makes sure InitGoogleTest() was called. if (!GTestIsInitialized()) { printf("%s", "\nThis test program did NOT call ::testing::InitGoogleTest " "before calling RUN_ALL_TESTS(). Please fix it.\n"); return false; } // Do not run any test if the --help flag was specified. if (g_help_flag) return true; // Repeats the call to the post-flag parsing initialization in case the // user didn't call InitGoogleTest. PostFlagParsingInit(); // Even if sharding is not on, test runners may want to use the // GTEST_SHARD_STATUS_FILE to query whether the test supports the sharding // protocol. internal::WriteToShardStatusFileIfNeeded(); // True iff we are in a subprocess for running a thread-safe-style // death test. bool in_subprocess_for_death_test = false; #if GTEST_HAS_DEATH_TEST in_subprocess_for_death_test = (internal_run_death_test_flag_.get() != NULL); #endif // GTEST_HAS_DEATH_TEST const bool should_shard = ShouldShard(kTestTotalShards, kTestShardIndex, in_subprocess_for_death_test); // Compares the full test names with the filter to decide which // tests to run. const bool has_tests_to_run = FilterTests(should_shard ? HONOR_SHARDING_PROTOCOL : IGNORE_SHARDING_PROTOCOL) > 0; // Lists the tests and exits if the --gtest_list_tests flag was specified. if (GTEST_FLAG(list_tests)) { // This must be called *after* FilterTests() has been called. ListTestsMatchingFilter(); return true; } random_seed_ = GTEST_FLAG(shuffle) ? GetRandomSeedFromFlag(GTEST_FLAG(random_seed)) : 0; // True iff at least one test has failed. bool failed = false; TestEventListener* repeater = listeners()->repeater(); start_timestamp_ = GetTimeInMillis(); repeater->OnTestProgramStart(*parent_); // How many times to repeat the tests? We don't want to repeat them // when we are inside the subprocess of a death test. const int repeat = in_subprocess_for_death_test ? 1 : GTEST_FLAG(repeat); // Repeats forever if the repeat count is negative. const bool forever = repeat < 0; for (int i = 0; forever || i != repeat; i++) { // We want to preserve failures generated by ad-hoc test // assertions executed before RUN_ALL_TESTS(). ClearNonAdHocTestResult(); const TimeInMillis start = GetTimeInMillis(); // Shuffles test cases and tests if requested. if (has_tests_to_run && GTEST_FLAG(shuffle)) { random()->Reseed(random_seed_); // This should be done before calling OnTestIterationStart(), // such that a test event listener can see the actual test order // in the event. ShuffleTests(); } // Tells the unit test event listeners that the tests are about to start. repeater->OnTestIterationStart(*parent_, i); // Runs each test case if there is at least one test to run. if (has_tests_to_run) { // Sets up all environments beforehand. repeater->OnEnvironmentsSetUpStart(*parent_); ForEach(environments_, SetUpEnvironment); repeater->OnEnvironmentsSetUpEnd(*parent_); // Runs the tests only if there was no fatal failure during global // set-up. if (!Test::HasFatalFailure()) { for (int test_index = 0; test_index < total_test_case_count(); test_index++) { GetMutableTestCase(test_index)->Run(); } } // Tears down all environments in reverse order afterwards. repeater->OnEnvironmentsTearDownStart(*parent_); std::for_each(environments_.rbegin(), environments_.rend(), TearDownEnvironment); repeater->OnEnvironmentsTearDownEnd(*parent_); } elapsed_time_ = GetTimeInMillis() - start; // Tells the unit test event listener that the tests have just finished. repeater->OnTestIterationEnd(*parent_, i); // Gets the result and clears it. if (!Passed()) { failed = true; } // Restores the original test order after the iteration. This // allows the user to quickly repro a failure that happens in the // N-th iteration without repeating the first (N - 1) iterations. // This is not enclosed in "if (GTEST_FLAG(shuffle)) { ... }", in // case the user somehow changes the value of the flag somewhere // (it's always safe to unshuffle the tests). UnshuffleTests(); if (GTEST_FLAG(shuffle)) { // Picks a new random seed for each iteration. random_seed_ = GetNextRandomSeed(random_seed_); } } repeater->OnTestProgramEnd(*parent_); return !failed; } // Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file // if the variable is present. If a file already exists at this location, this // function will write over it. If the variable is present, but the file cannot // be created, prints an error and exits. void WriteToShardStatusFileIfNeeded() { const char* const test_shard_file = posix::GetEnv(kTestShardStatusFile); if (test_shard_file != NULL) { FILE* const file = posix::FOpen(test_shard_file, "w"); if (file == NULL) { ColoredPrintf(COLOR_RED, "Could not write to the test shard status file \"%s\" " "specified by the %s environment variable.\n", test_shard_file, kTestShardStatusFile); fflush(stdout); exit(EXIT_FAILURE); } fclose(file); } } // Checks whether sharding is enabled by examining the relevant // environment variable values. If the variables are present, // but inconsistent (i.e., shard_index >= total_shards), prints // an error and exits. If in_subprocess_for_death_test, sharding is // disabled because it must only be applied to the original test // process. Otherwise, we could filter out death tests we intended to execute. bool ShouldShard(const char* total_shards_env, const char* shard_index_env, bool in_subprocess_for_death_test) { if (in_subprocess_for_death_test) { return false; } const Int32 total_shards = Int32FromEnvOrDie(total_shards_env, -1); const Int32 shard_index = Int32FromEnvOrDie(shard_index_env, -1); if (total_shards == -1 && shard_index == -1) { return false; } else if (total_shards == -1 && shard_index != -1) { const Message msg = Message() << "Invalid environment variables: you have " << kTestShardIndex << " = " << shard_index << ", but have left " << kTestTotalShards << " unset.\n"; ColoredPrintf(COLOR_RED, msg.GetString().c_str()); fflush(stdout); exit(EXIT_FAILURE); } else if (total_shards != -1 && shard_index == -1) { const Message msg = Message() << "Invalid environment variables: you have " << kTestTotalShards << " = " << total_shards << ", but have left " << kTestShardIndex << " unset.\n"; ColoredPrintf(COLOR_RED, msg.GetString().c_str()); fflush(stdout); exit(EXIT_FAILURE); } else if (shard_index < 0 || shard_index >= total_shards) { const Message msg = Message() << "Invalid environment variables: we require 0 <= " << kTestShardIndex << " < " << kTestTotalShards << ", but you have " << kTestShardIndex << "=" << shard_index << ", " << kTestTotalShards << "=" << total_shards << ".\n"; ColoredPrintf(COLOR_RED, msg.GetString().c_str()); fflush(stdout); exit(EXIT_FAILURE); } return total_shards > 1; } // Parses the environment variable var as an Int32. If it is unset, // returns default_val. If it is not an Int32, prints an error // and aborts. Int32 Int32FromEnvOrDie(const char* var, Int32 default_val) { const char* str_val = posix::GetEnv(var); if (str_val == NULL) { return default_val; } Int32 result; if (!ParseInt32(Message() << "The value of environment variable " << var, str_val, &result)) { exit(EXIT_FAILURE); } return result; } // Given the total number of shards, the shard index, and the test id, // returns true iff the test should be run on this shard. The test id is // some arbitrary but unique non-negative integer assigned to each test // method. Assumes that 0 <= shard_index < total_shards. bool ShouldRunTestOnShard(int total_shards, int shard_index, int test_id) { return (test_id % total_shards) == shard_index; } // Compares the name of each test with the user-specified filter to // decide whether the test should be run, then records the result in // each TestCase and TestInfo object. // If shard_tests == true, further filters tests based on sharding // variables in the environment - see // http://code.google.com/p/googletest/wiki/GoogleTestAdvancedGuide. // Returns the number of tests that should run. int UnitTestImpl::FilterTests(ReactionToSharding shard_tests) { const Int32 total_shards = shard_tests == HONOR_SHARDING_PROTOCOL ? Int32FromEnvOrDie(kTestTotalShards, -1) : -1; const Int32 shard_index = shard_tests == HONOR_SHARDING_PROTOCOL ? Int32FromEnvOrDie(kTestShardIndex, -1) : -1; // num_runnable_tests are the number of tests that will // run across all shards (i.e., match filter and are not disabled). // num_selected_tests are the number of tests to be run on // this shard. int num_runnable_tests = 0; int num_selected_tests = 0; for (size_t i = 0; i < test_cases_.size(); i++) { TestCase* const test_case = test_cases_[i]; const std::string &test_case_name = test_case->name(); test_case->set_should_run(false); for (size_t j = 0; j < test_case->test_info_list().size(); j++) { TestInfo* const test_info = test_case->test_info_list()[j]; const std::string test_name(test_info->name()); // A test is disabled if test case name or test name matches // kDisableTestFilter. const bool is_disabled = internal::UnitTestOptions::MatchesFilter(test_case_name, kDisableTestFilter) || internal::UnitTestOptions::MatchesFilter(test_name, kDisableTestFilter); test_info->is_disabled_ = is_disabled; const bool matches_filter = internal::UnitTestOptions::FilterMatchesTest(test_case_name, test_name); test_info->matches_filter_ = matches_filter; const bool is_runnable = (GTEST_FLAG(also_run_disabled_tests) || !is_disabled) && matches_filter; const bool is_selected = is_runnable && (shard_tests == IGNORE_SHARDING_PROTOCOL || ShouldRunTestOnShard(total_shards, shard_index, num_runnable_tests)); num_runnable_tests += is_runnable; num_selected_tests += is_selected; test_info->should_run_ = is_selected; test_case->set_should_run(test_case->should_run() || is_selected); } } return num_selected_tests; } // Prints the given C-string on a single line by replacing all '\n' // characters with string "\\n". If the output takes more than // max_length characters, only prints the first max_length characters // and "...". static void PrintOnOneLine(const char* str, int max_length) { if (str != NULL) { for (int i = 0; *str != '\0'; ++str) { if (i >= max_length) { printf("..."); break; } if (*str == '\n') { printf("\\n"); i += 2; } else { printf("%c", *str); ++i; } } } } // Prints the names of the tests matching the user-specified filter flag. void UnitTestImpl::ListTestsMatchingFilter() { // Print at most this many characters for each type/value parameter. const int kMaxParamLength = 250; for (size_t i = 0; i < test_cases_.size(); i++) { const TestCase* const test_case = test_cases_[i]; bool printed_test_case_name = false; for (size_t j = 0; j < test_case->test_info_list().size(); j++) { const TestInfo* const test_info = test_case->test_info_list()[j]; if (test_info->matches_filter_) { if (!printed_test_case_name) { printed_test_case_name = true; printf("%s.", test_case->name()); if (test_case->type_param() != NULL) { printf(" # %s = ", kTypeParamLabel); // We print the type parameter on a single line to make // the output easy to parse by a program. PrintOnOneLine(test_case->type_param(), kMaxParamLength); } printf("\n"); } printf(" %s", test_info->name()); if (test_info->value_param() != NULL) { printf(" # %s = ", kValueParamLabel); // We print the value parameter on a single line to make the // output easy to parse by a program. PrintOnOneLine(test_info->value_param(), kMaxParamLength); } printf("\n"); } } } fflush(stdout); } // Sets the OS stack trace getter. // // Does nothing if the input and the current OS stack trace getter are // the same; otherwise, deletes the old getter and makes the input the // current getter. void UnitTestImpl::set_os_stack_trace_getter( OsStackTraceGetterInterface* getter) { if (os_stack_trace_getter_ != getter) { delete os_stack_trace_getter_; os_stack_trace_getter_ = getter; } } // Returns the current OS stack trace getter if it is not NULL; // otherwise, creates an OsStackTraceGetter, makes it the current // getter, and returns it. OsStackTraceGetterInterface* UnitTestImpl::os_stack_trace_getter() { if (os_stack_trace_getter_ == NULL) { os_stack_trace_getter_ = new OsStackTraceGetter; } return os_stack_trace_getter_; } // Returns the TestResult for the test that's currently running, or // the TestResult for the ad hoc test if no test is running. TestResult* UnitTestImpl::current_test_result() { return current_test_info_ ? &(current_test_info_->result_) : &ad_hoc_test_result_; } // Shuffles all test cases, and the tests within each test case, // making sure that death tests are still run first. void UnitTestImpl::ShuffleTests() { // Shuffles the death test cases. ShuffleRange(random(), 0, last_death_test_case_ + 1, &test_case_indices_); // Shuffles the non-death test cases. ShuffleRange(random(), last_death_test_case_ + 1, static_cast(test_cases_.size()), &test_case_indices_); // Shuffles the tests inside each test case. for (size_t i = 0; i < test_cases_.size(); i++) { test_cases_[i]->ShuffleTests(random()); } } // Restores the test cases and tests to their order before the first shuffle. void UnitTestImpl::UnshuffleTests() { for (size_t i = 0; i < test_cases_.size(); i++) { // Unshuffles the tests in each test case. test_cases_[i]->UnshuffleTests(); // Resets the index of each test case. test_case_indices_[i] = static_cast(i); } } // Returns the current OS stack trace as an std::string. // // The maximum number of stack frames to be included is specified by // the gtest_stack_trace_depth flag. The skip_count parameter // specifies the number of top frames to be skipped, which doesn't // count against the number of frames to be included. // // For example, if Foo() calls Bar(), which in turn calls // GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in // the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't. std::string GetCurrentOsStackTraceExceptTop(UnitTest* /*unit_test*/, int skip_count) { // We pass skip_count + 1 to skip this wrapper function in addition // to what the user really wants to skip. return GetUnitTestImpl()->CurrentOsStackTraceExceptTop(skip_count + 1); } // Used by the GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_ macro to // suppress unreachable code warnings. namespace { class ClassUniqueToAlwaysTrue {}; } bool IsTrue(bool condition) { return condition; } bool AlwaysTrue() { #if GTEST_HAS_EXCEPTIONS // This condition is always false so AlwaysTrue() never actually throws, // but it makes the compiler think that it may throw. if (IsTrue(false)) throw ClassUniqueToAlwaysTrue(); #endif // GTEST_HAS_EXCEPTIONS return true; } // If *pstr starts with the given prefix, modifies *pstr to be right // past the prefix and returns true; otherwise leaves *pstr unchanged // and returns false. None of pstr, *pstr, and prefix can be NULL. bool SkipPrefix(const char* prefix, const char** pstr) { const size_t prefix_len = strlen(prefix); if (strncmp(*pstr, prefix, prefix_len) == 0) { *pstr += prefix_len; return true; } return false; } // Parses a string as a command line flag. The string should have // the format "--flag=value". When def_optional is true, the "=value" // part can be omitted. // // Returns the value of the flag, or NULL if the parsing failed. const char* ParseFlagValue(const char* str, const char* flag, bool def_optional) { // str and flag must not be NULL. if (str == NULL || flag == NULL) return NULL; // The flag must start with "--" followed by GTEST_FLAG_PREFIX_. const std::string flag_str = std::string("--") + GTEST_FLAG_PREFIX_ + flag; const size_t flag_len = flag_str.length(); if (strncmp(str, flag_str.c_str(), flag_len) != 0) return NULL; // Skips the flag name. const char* flag_end = str + flag_len; // When def_optional is true, it's OK to not have a "=value" part. if (def_optional && (flag_end[0] == '\0')) { return flag_end; } // If def_optional is true and there are more characters after the // flag name, or if def_optional is false, there must be a '=' after // the flag name. if (flag_end[0] != '=') return NULL; // Returns the string after "=". return flag_end + 1; } // Parses a string for a bool flag, in the form of either // "--flag=value" or "--flag". // // In the former case, the value is taken as true as long as it does // not start with '0', 'f', or 'F'. // // In the latter case, the value is taken as true. // // On success, stores the value of the flag in *value, and returns // true. On failure, returns false without changing *value. bool ParseBoolFlag(const char* str, const char* flag, bool* value) { // Gets the value of the flag as a string. const char* const value_str = ParseFlagValue(str, flag, true); // Aborts if the parsing failed. if (value_str == NULL) return false; // Converts the string value to a bool. *value = !(*value_str == '0' || *value_str == 'f' || *value_str == 'F'); return true; } // Parses a string for an Int32 flag, in the form of // "--flag=value". // // On success, stores the value of the flag in *value, and returns // true. On failure, returns false without changing *value. bool ParseInt32Flag(const char* str, const char* flag, Int32* value) { // Gets the value of the flag as a string. const char* const value_str = ParseFlagValue(str, flag, false); // Aborts if the parsing failed. if (value_str == NULL) return false; // Sets *value to the value of the flag. return ParseInt32(Message() << "The value of flag --" << flag, value_str, value); } // Parses a string for a string flag, in the form of // "--flag=value". // // On success, stores the value of the flag in *value, and returns // true. On failure, returns false without changing *value. bool ParseStringFlag(const char* str, const char* flag, std::string* value) { // Gets the value of the flag as a string. const char* const value_str = ParseFlagValue(str, flag, false); // Aborts if the parsing failed. if (value_str == NULL) return false; // Sets *value to the value of the flag. *value = value_str; return true; } // Determines whether a string has a prefix that Google Test uses for its // flags, i.e., starts with GTEST_FLAG_PREFIX_ or GTEST_FLAG_PREFIX_DASH_. // If Google Test detects that a command line flag has its prefix but is not // recognized, it will print its help message. Flags starting with // GTEST_INTERNAL_PREFIX_ followed by "internal_" are considered Google Test // internal flags and do not trigger the help message. static bool HasGoogleTestFlagPrefix(const char* str) { return (SkipPrefix("--", &str) || SkipPrefix("-", &str) || SkipPrefix("/", &str)) && !SkipPrefix(GTEST_FLAG_PREFIX_ "internal_", &str) && (SkipPrefix(GTEST_FLAG_PREFIX_, &str) || SkipPrefix(GTEST_FLAG_PREFIX_DASH_, &str)); } // Prints a string containing code-encoded text. The following escape // sequences can be used in the string to control the text color: // // @@ prints a single '@' character. // @R changes the color to red. // @G changes the color to green. // @Y changes the color to yellow. // @D changes to the default terminal text color. // // TODO(wan@google.com): Write tests for this once we add stdout // capturing to Google Test. static void PrintColorEncoded(const char* str) { GTestColor color = COLOR_DEFAULT; // The current color. // Conceptually, we split the string into segments divided by escape // sequences. Then we print one segment at a time. At the end of // each iteration, the str pointer advances to the beginning of the // next segment. for (;;) { const char* p = strchr(str, '@'); if (p == NULL) { ColoredPrintf(color, "%s", str); return; } ColoredPrintf(color, "%s", std::string(str, p).c_str()); const char ch = p[1]; str = p + 2; if (ch == '@') { ColoredPrintf(color, "@"); } else if (ch == 'D') { color = COLOR_DEFAULT; } else if (ch == 'R') { color = COLOR_RED; } else if (ch == 'G') { color = COLOR_GREEN; } else if (ch == 'Y') { color = COLOR_YELLOW; } else { --str; } } } static const char kColorEncodedHelpMessage[] = "This program contains tests written using " GTEST_NAME_ ". You can use the\n" "following command line flags to control its behavior:\n" "\n" "Test Selection:\n" " @G--" GTEST_FLAG_PREFIX_ "list_tests@D\n" " List the names of all tests instead of running them. The name of\n" " TEST(Foo, Bar) is \"Foo.Bar\".\n" " @G--" GTEST_FLAG_PREFIX_ "filter=@YPOSTIVE_PATTERNS" "[@G-@YNEGATIVE_PATTERNS]@D\n" " Run only the tests whose name matches one of the positive patterns but\n" " none of the negative patterns. '?' matches any single character; '*'\n" " matches any substring; ':' separates two patterns.\n" " @G--" GTEST_FLAG_PREFIX_ "also_run_disabled_tests@D\n" " Run all disabled tests too.\n" "\n" "Test Execution:\n" " @G--" GTEST_FLAG_PREFIX_ "repeat=@Y[COUNT]@D\n" " Run the tests repeatedly; use a negative count to repeat forever.\n" " @G--" GTEST_FLAG_PREFIX_ "shuffle@D\n" " Randomize tests' orders on every iteration.\n" " @G--" GTEST_FLAG_PREFIX_ "random_seed=@Y[NUMBER]@D\n" " Random number seed to use for shuffling test orders (between 1 and\n" " 99999, or 0 to use a seed based on the current time).\n" "\n" "Test Output:\n" " @G--" GTEST_FLAG_PREFIX_ "color=@Y(@Gyes@Y|@Gno@Y|@Gauto@Y)@D\n" " Enable/disable colored output. The default is @Gauto@D.\n" " -@G-" GTEST_FLAG_PREFIX_ "print_time=0@D\n" " Don't print the elapsed time of each test.\n" " @G--" GTEST_FLAG_PREFIX_ "output=xml@Y[@G:@YDIRECTORY_PATH@G" GTEST_PATH_SEP_ "@Y|@G:@YFILE_PATH]@D\n" " Generate an XML report in the given directory or with the given file\n" " name. @YFILE_PATH@D defaults to @Gtest_details.xml@D.\n" #if GTEST_CAN_STREAM_RESULTS_ " @G--" GTEST_FLAG_PREFIX_ "stream_result_to=@YHOST@G:@YPORT@D\n" " Stream test results to the given server.\n" #endif // GTEST_CAN_STREAM_RESULTS_ "\n" "Assertion Behavior:\n" #if GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS " @G--" GTEST_FLAG_PREFIX_ "death_test_style=@Y(@Gfast@Y|@Gthreadsafe@Y)@D\n" " Set the default death test style.\n" #endif // GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS " @G--" GTEST_FLAG_PREFIX_ "break_on_failure@D\n" " Turn assertion failures into debugger break-points.\n" " @G--" GTEST_FLAG_PREFIX_ "throw_on_failure@D\n" " Turn assertion failures into C++ exceptions.\n" " @G--" GTEST_FLAG_PREFIX_ "catch_exceptions=0@D\n" " Do not report exceptions as test failures. Instead, allow them\n" " to crash the program or throw a pop-up (on Windows).\n" "\n" "Except for @G--" GTEST_FLAG_PREFIX_ "list_tests@D, you can alternatively set " "the corresponding\n" "environment variable of a flag (all letters in upper-case). For example, to\n" "disable colored text output, you can either specify @G--" GTEST_FLAG_PREFIX_ "color=no@D or set\n" "the @G" GTEST_FLAG_PREFIX_UPPER_ "COLOR@D environment variable to @Gno@D.\n" "\n" "For more information, please read the " GTEST_NAME_ " documentation at\n" "@G" GTEST_PROJECT_URL_ "@D. If you find a bug in " GTEST_NAME_ "\n" "(not one in your own code or tests), please report it to\n" "@G<" GTEST_DEV_EMAIL_ ">@D.\n"; // Parses the command line for Google Test flags, without initializing // other parts of Google Test. The type parameter CharType can be // instantiated to either char or wchar_t. template void ParseGoogleTestFlagsOnlyImpl(int* argc, CharType** argv) { for (int i = 1; i < *argc; i++) { const std::string arg_string = StreamableToString(argv[i]); const char* const arg = arg_string.c_str(); using internal::ParseBoolFlag; using internal::ParseInt32Flag; using internal::ParseStringFlag; // Do we see a Google Test flag? if (ParseBoolFlag(arg, kAlsoRunDisabledTestsFlag, >EST_FLAG(also_run_disabled_tests)) || ParseBoolFlag(arg, kBreakOnFailureFlag, >EST_FLAG(break_on_failure)) || ParseBoolFlag(arg, kCatchExceptionsFlag, >EST_FLAG(catch_exceptions)) || ParseStringFlag(arg, kColorFlag, >EST_FLAG(color)) || ParseStringFlag(arg, kDeathTestStyleFlag, >EST_FLAG(death_test_style)) || ParseBoolFlag(arg, kDeathTestUseFork, >EST_FLAG(death_test_use_fork)) || ParseStringFlag(arg, kFilterFlag, >EST_FLAG(filter)) || ParseStringFlag(arg, kInternalRunDeathTestFlag, >EST_FLAG(internal_run_death_test)) || ParseBoolFlag(arg, kListTestsFlag, >EST_FLAG(list_tests)) || ParseStringFlag(arg, kOutputFlag, >EST_FLAG(output)) || ParseBoolFlag(arg, kPrintTimeFlag, >EST_FLAG(print_time)) || ParseInt32Flag(arg, kRandomSeedFlag, >EST_FLAG(random_seed)) || ParseInt32Flag(arg, kRepeatFlag, >EST_FLAG(repeat)) || ParseBoolFlag(arg, kShuffleFlag, >EST_FLAG(shuffle)) || ParseInt32Flag(arg, kStackTraceDepthFlag, >EST_FLAG(stack_trace_depth)) || ParseStringFlag(arg, kStreamResultToFlag, >EST_FLAG(stream_result_to)) || ParseBoolFlag(arg, kThrowOnFailureFlag, >EST_FLAG(throw_on_failure)) ) { // Yes. Shift the remainder of the argv list left by one. Note // that argv has (*argc + 1) elements, the last one always being // NULL. The following loop moves the trailing NULL element as // well. for (int j = i; j != *argc; j++) { argv[j] = argv[j + 1]; } // Decrements the argument count. (*argc)--; // We also need to decrement the iterator as we just removed // an element. i--; } else if (arg_string == "--help" || arg_string == "-h" || arg_string == "-?" || arg_string == "/?" || HasGoogleTestFlagPrefix(arg)) { // Both help flag and unrecognized Google Test flags (excluding // internal ones) trigger help display. g_help_flag = true; } } if (g_help_flag) { // We print the help here instead of in RUN_ALL_TESTS(), as the // latter may not be called at all if the user is using Google // Test with another testing framework. PrintColorEncoded(kColorEncodedHelpMessage); } } // Parses the command line for Google Test flags, without initializing // other parts of Google Test. void ParseGoogleTestFlagsOnly(int* argc, char** argv) { ParseGoogleTestFlagsOnlyImpl(argc, argv); } void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv) { ParseGoogleTestFlagsOnlyImpl(argc, argv); } // The internal implementation of InitGoogleTest(). // // The type parameter CharType can be instantiated to either char or // wchar_t. template void InitGoogleTestImpl(int* argc, CharType** argv) { g_init_gtest_count++; // We don't want to run the initialization code twice. if (g_init_gtest_count != 1) return; if (*argc <= 0) return; internal::g_executable_path = internal::StreamableToString(argv[0]); #if GTEST_HAS_DEATH_TEST g_argvs.clear(); for (int i = 0; i != *argc; i++) { g_argvs.push_back(StreamableToString(argv[i])); } #endif // GTEST_HAS_DEATH_TEST ParseGoogleTestFlagsOnly(argc, argv); GetUnitTestImpl()->PostFlagParsingInit(); } } // namespace internal // Initializes Google Test. This must be called before calling // RUN_ALL_TESTS(). In particular, it parses a command line for the // flags that Google Test recognizes. Whenever a Google Test flag is // seen, it is removed from argv, and *argc is decremented. // // No value is returned. Instead, the Google Test flag variables are // updated. // // Calling the function for the second time has no user-visible effect. void InitGoogleTest(int* argc, char** argv) { internal::InitGoogleTestImpl(argc, argv); } // This overloaded version can be used in Windows programs compiled in // UNICODE mode. void InitGoogleTest(int* argc, wchar_t** argv) { internal::InitGoogleTestImpl(argc, argv); } } // namespace testing // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan), vladl@google.com (Vlad Losev) // // This file implements death tests. #if GTEST_HAS_DEATH_TEST # if GTEST_OS_MAC # include # endif // GTEST_OS_MAC # include # include # include # if GTEST_OS_LINUX # include # endif // GTEST_OS_LINUX # include # if GTEST_OS_WINDOWS # include # else # include # include # endif // GTEST_OS_WINDOWS # if GTEST_OS_QNX # include # endif // GTEST_OS_QNX #endif // GTEST_HAS_DEATH_TEST // Indicates that this translation unit is part of Google Test's // implementation. It must come before gtest-internal-inl.h is // included, or there will be a compiler error. This trick exists to // prevent the accidental inclusion of gtest-internal-inl.h in the // user's code. #define GTEST_IMPLEMENTATION_ 1 #undef GTEST_IMPLEMENTATION_ namespace testing { // Constants. // The default death test style. static const char kDefaultDeathTestStyle[] = "fast"; GTEST_DEFINE_string_( death_test_style, internal::StringFromGTestEnv("death_test_style", kDefaultDeathTestStyle), "Indicates how to run a death test in a forked child process: " "\"threadsafe\" (child process re-executes the test binary " "from the beginning, running only the specific death test) or " "\"fast\" (child process runs the death test immediately " "after forking)."); GTEST_DEFINE_bool_( death_test_use_fork, internal::BoolFromGTestEnv("death_test_use_fork", false), "Instructs to use fork()/_exit() instead of clone() in death tests. " "Ignored and always uses fork() on POSIX systems where clone() is not " "implemented. Useful when running under valgrind or similar tools if " "those do not support clone(). Valgrind 3.3.1 will just fail if " "it sees an unsupported combination of clone() flags. " "It is not recommended to use this flag w/o valgrind though it will " "work in 99% of the cases. Once valgrind is fixed, this flag will " "most likely be removed."); namespace internal { GTEST_DEFINE_string_( internal_run_death_test, "", "Indicates the file, line number, temporal index of " "the single death test to run, and a file descriptor to " "which a success code may be sent, all separated by " "the '|' characters. This flag is specified if and only if the current " "process is a sub-process launched for running a thread-safe " "death test. FOR INTERNAL USE ONLY."); } // namespace internal #if GTEST_HAS_DEATH_TEST namespace internal { // Valid only for fast death tests. Indicates the code is running in the // child process of a fast style death test. static bool g_in_fast_death_test_child = false; // Returns a Boolean value indicating whether the caller is currently // executing in the context of the death test child process. Tools such as // Valgrind heap checkers may need this to modify their behavior in death // tests. IMPORTANT: This is an internal utility. Using it may break the // implementation of death tests. User code MUST NOT use it. bool InDeathTestChild() { # if GTEST_OS_WINDOWS // On Windows, death tests are thread-safe regardless of the value of the // death_test_style flag. return !GTEST_FLAG(internal_run_death_test).empty(); # else if (GTEST_FLAG(death_test_style) == "threadsafe") return !GTEST_FLAG(internal_run_death_test).empty(); else return g_in_fast_death_test_child; #endif } } // namespace internal // ExitedWithCode constructor. ExitedWithCode::ExitedWithCode(int exit_code) : exit_code_(exit_code) { } // ExitedWithCode function-call operator. bool ExitedWithCode::operator()(int exit_status) const { # if GTEST_OS_WINDOWS return exit_status == exit_code_; # else return WIFEXITED(exit_status) && WEXITSTATUS(exit_status) == exit_code_; # endif // GTEST_OS_WINDOWS } # if !GTEST_OS_WINDOWS // KilledBySignal constructor. KilledBySignal::KilledBySignal(int signum) : signum_(signum) { } // KilledBySignal function-call operator. bool KilledBySignal::operator()(int exit_status) const { return WIFSIGNALED(exit_status) && WTERMSIG(exit_status) == signum_; } # endif // !GTEST_OS_WINDOWS namespace internal { // Utilities needed for death tests. // Generates a textual description of a given exit code, in the format // specified by wait(2). static std::string ExitSummary(int exit_code) { Message m; # if GTEST_OS_WINDOWS m << "Exited with exit status " << exit_code; # else if (WIFEXITED(exit_code)) { m << "Exited with exit status " << WEXITSTATUS(exit_code); } else if (WIFSIGNALED(exit_code)) { m << "Terminated by signal " << WTERMSIG(exit_code); } # ifdef WCOREDUMP if (WCOREDUMP(exit_code)) { m << " (core dumped)"; } # endif # endif // GTEST_OS_WINDOWS return m.GetString(); } // Returns true if exit_status describes a process that was terminated // by a signal, or exited normally with a nonzero exit code. bool ExitedUnsuccessfully(int exit_status) { return !ExitedWithCode(0)(exit_status); } # if !GTEST_OS_WINDOWS // Generates a textual failure message when a death test finds more than // one thread running, or cannot determine the number of threads, prior // to executing the given statement. It is the responsibility of the // caller not to pass a thread_count of 1. static std::string DeathTestThreadWarning(size_t thread_count) { Message msg; msg << "Death tests use fork(), which is unsafe particularly" << " in a threaded context. For this test, " << GTEST_NAME_ << " "; if (thread_count == 0) msg << "couldn't detect the number of threads."; else msg << "detected " << thread_count << " threads."; return msg.GetString(); } # endif // !GTEST_OS_WINDOWS // Flag characters for reporting a death test that did not die. static const char kDeathTestLived = 'L'; static const char kDeathTestReturned = 'R'; static const char kDeathTestThrew = 'T'; static const char kDeathTestInternalError = 'I'; // An enumeration describing all of the possible ways that a death test can // conclude. DIED means that the process died while executing the test // code; LIVED means that process lived beyond the end of the test code; // RETURNED means that the test statement attempted to execute a return // statement, which is not allowed; THREW means that the test statement // returned control by throwing an exception. IN_PROGRESS means the test // has not yet concluded. // TODO(vladl@google.com): Unify names and possibly values for // AbortReason, DeathTestOutcome, and flag characters above. enum DeathTestOutcome { IN_PROGRESS, DIED, LIVED, RETURNED, THREW }; // Routine for aborting the program which is safe to call from an // exec-style death test child process, in which case the error // message is propagated back to the parent process. Otherwise, the // message is simply printed to stderr. In either case, the program // then exits with status 1. void DeathTestAbort(const std::string& message) { // On a POSIX system, this function may be called from a threadsafe-style // death test child process, which operates on a very small stack. Use // the heap for any additional non-minuscule memory requirements. const InternalRunDeathTestFlag* const flag = GetUnitTestImpl()->internal_run_death_test_flag(); if (flag != NULL) { FILE* parent = posix::FDOpen(flag->write_fd(), "w"); fputc(kDeathTestInternalError, parent); fprintf(parent, "%s", message.c_str()); fflush(parent); _exit(1); } else { fprintf(stderr, "%s", message.c_str()); fflush(stderr); posix::Abort(); } } // A replacement for CHECK that calls DeathTestAbort if the assertion // fails. # define GTEST_DEATH_TEST_CHECK_(expression) \ do { \ if (!::testing::internal::IsTrue(expression)) { \ DeathTestAbort( \ ::std::string("CHECK failed: File ") + __FILE__ + ", line " \ + ::testing::internal::StreamableToString(__LINE__) + ": " \ + #expression); \ } \ } while (::testing::internal::AlwaysFalse()) // This macro is similar to GTEST_DEATH_TEST_CHECK_, but it is meant for // evaluating any system call that fulfills two conditions: it must return // -1 on failure, and set errno to EINTR when it is interrupted and // should be tried again. The macro expands to a loop that repeatedly // evaluates the expression as long as it evaluates to -1 and sets // errno to EINTR. If the expression evaluates to -1 but errno is // something other than EINTR, DeathTestAbort is called. # define GTEST_DEATH_TEST_CHECK_SYSCALL_(expression) \ do { \ int gtest_retval; \ do { \ gtest_retval = (expression); \ } while (gtest_retval == -1 && errno == EINTR); \ if (gtest_retval == -1) { \ DeathTestAbort( \ ::std::string("CHECK failed: File ") + __FILE__ + ", line " \ + ::testing::internal::StreamableToString(__LINE__) + ": " \ + #expression + " != -1"); \ } \ } while (::testing::internal::AlwaysFalse()) // Returns the message describing the last system error in errno. std::string GetLastErrnoDescription() { return errno == 0 ? "" : posix::StrError(errno); } // This is called from a death test parent process to read a failure // message from the death test child process and log it with the FATAL // severity. On Windows, the message is read from a pipe handle. On other // platforms, it is read from a file descriptor. static void FailFromInternalError(int fd) { Message error; char buffer[256]; int num_read; do { while ((num_read = posix::Read(fd, buffer, 255)) > 0) { buffer[num_read] = '\0'; error << buffer; } } while (num_read == -1 && errno == EINTR); if (num_read == 0) { GTEST_LOG_(FATAL) << error.GetString(); } else { const int last_error = errno; GTEST_LOG_(FATAL) << "Error while reading death test internal: " << GetLastErrnoDescription() << " [" << last_error << "]"; } } // Death test constructor. Increments the running death test count // for the current test. DeathTest::DeathTest() { TestInfo* const info = GetUnitTestImpl()->current_test_info(); if (info == NULL) { DeathTestAbort("Cannot run a death test outside of a TEST or " "TEST_F construct"); } } // Creates and returns a death test by dispatching to the current // death test factory. bool DeathTest::Create(const char* statement, const RE* regex, const char* file, int line, DeathTest** test) { return GetUnitTestImpl()->death_test_factory()->Create( statement, regex, file, line, test); } const char* DeathTest::LastMessage() { return last_death_test_message_.c_str(); } void DeathTest::set_last_death_test_message(const std::string& message) { last_death_test_message_ = message; } std::string DeathTest::last_death_test_message_; // Provides cross platform implementation for some death functionality. class DeathTestImpl : public DeathTest { protected: DeathTestImpl(const char* a_statement, const RE* a_regex) : statement_(a_statement), regex_(a_regex), spawned_(false), status_(-1), outcome_(IN_PROGRESS), read_fd_(-1), write_fd_(-1) {} // read_fd_ is expected to be closed and cleared by a derived class. ~DeathTestImpl() { GTEST_DEATH_TEST_CHECK_(read_fd_ == -1); } void Abort(AbortReason reason); virtual bool Passed(bool status_ok); const char* statement() const { return statement_; } const RE* regex() const { return regex_; } bool spawned() const { return spawned_; } void set_spawned(bool is_spawned) { spawned_ = is_spawned; } int status() const { return status_; } void set_status(int a_status) { status_ = a_status; } DeathTestOutcome outcome() const { return outcome_; } void set_outcome(DeathTestOutcome an_outcome) { outcome_ = an_outcome; } int read_fd() const { return read_fd_; } void set_read_fd(int fd) { read_fd_ = fd; } int write_fd() const { return write_fd_; } void set_write_fd(int fd) { write_fd_ = fd; } // Called in the parent process only. Reads the result code of the death // test child process via a pipe, interprets it to set the outcome_ // member, and closes read_fd_. Outputs diagnostics and terminates in // case of unexpected codes. void ReadAndInterpretStatusByte(); private: // The textual content of the code this object is testing. This class // doesn't own this string and should not attempt to delete it. const char* const statement_; // The regular expression which test output must match. DeathTestImpl // doesn't own this object and should not attempt to delete it. const RE* const regex_; // True if the death test child process has been successfully spawned. bool spawned_; // The exit status of the child process. int status_; // How the death test concluded. DeathTestOutcome outcome_; // Descriptor to the read end of the pipe to the child process. It is // always -1 in the child process. The child keeps its write end of the // pipe in write_fd_. int read_fd_; // Descriptor to the child's write end of the pipe to the parent process. // It is always -1 in the parent process. The parent keeps its end of the // pipe in read_fd_. int write_fd_; }; // Called in the parent process only. Reads the result code of the death // test child process via a pipe, interprets it to set the outcome_ // member, and closes read_fd_. Outputs diagnostics and terminates in // case of unexpected codes. void DeathTestImpl::ReadAndInterpretStatusByte() { char flag; int bytes_read; // The read() here blocks until data is available (signifying the // failure of the death test) or until the pipe is closed (signifying // its success), so it's okay to call this in the parent before // the child process has exited. do { bytes_read = posix::Read(read_fd(), &flag, 1); } while (bytes_read == -1 && errno == EINTR); if (bytes_read == 0) { set_outcome(DIED); } else if (bytes_read == 1) { switch (flag) { case kDeathTestReturned: set_outcome(RETURNED); break; case kDeathTestThrew: set_outcome(THREW); break; case kDeathTestLived: set_outcome(LIVED); break; case kDeathTestInternalError: FailFromInternalError(read_fd()); // Does not return. break; default: GTEST_LOG_(FATAL) << "Death test child process reported " << "unexpected status byte (" << static_cast(flag) << ")"; } } else { GTEST_LOG_(FATAL) << "Read from death test child process failed: " << GetLastErrnoDescription(); } GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Close(read_fd())); set_read_fd(-1); } // Signals that the death test code which should have exited, didn't. // Should be called only in a death test child process. // Writes a status byte to the child's status file descriptor, then // calls _exit(1). void DeathTestImpl::Abort(AbortReason reason) { // The parent process considers the death test to be a failure if // it finds any data in our pipe. So, here we write a single flag byte // to the pipe, then exit. const char status_ch = reason == TEST_DID_NOT_DIE ? kDeathTestLived : reason == TEST_THREW_EXCEPTION ? kDeathTestThrew : kDeathTestReturned; GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Write(write_fd(), &status_ch, 1)); // We are leaking the descriptor here because on some platforms (i.e., // when built as Windows DLL), destructors of global objects will still // run after calling _exit(). On such systems, write_fd_ will be // indirectly closed from the destructor of UnitTestImpl, causing double // close if it is also closed here. On debug configurations, double close // may assert. As there are no in-process buffers to flush here, we are // relying on the OS to close the descriptor after the process terminates // when the destructors are not run. _exit(1); // Exits w/o any normal exit hooks (we were supposed to crash) } // Returns an indented copy of stderr output for a death test. // This makes distinguishing death test output lines from regular log lines // much easier. static ::std::string FormatDeathTestOutput(const ::std::string& output) { ::std::string ret; for (size_t at = 0; ; ) { const size_t line_end = output.find('\n', at); ret += "[ DEATH ] "; if (line_end == ::std::string::npos) { ret += output.substr(at); break; } ret += output.substr(at, line_end + 1 - at); at = line_end + 1; } return ret; } // Assesses the success or failure of a death test, using both private // members which have previously been set, and one argument: // // Private data members: // outcome: An enumeration describing how the death test // concluded: DIED, LIVED, THREW, or RETURNED. The death test // fails in the latter three cases. // status: The exit status of the child process. On *nix, it is in the // in the format specified by wait(2). On Windows, this is the // value supplied to the ExitProcess() API or a numeric code // of the exception that terminated the program. // regex: A regular expression object to be applied to // the test's captured standard error output; the death test // fails if it does not match. // // Argument: // status_ok: true if exit_status is acceptable in the context of // this particular death test, which fails if it is false // // Returns true iff all of the above conditions are met. Otherwise, the // first failing condition, in the order given above, is the one that is // reported. Also sets the last death test message string. bool DeathTestImpl::Passed(bool status_ok) { if (!spawned()) return false; const std::string error_message = GetCapturedStderr(); bool success = false; Message buffer; buffer << "Death test: " << statement() << "\n"; switch (outcome()) { case LIVED: buffer << " Result: failed to die.\n" << " Error msg:\n" << FormatDeathTestOutput(error_message); break; case THREW: buffer << " Result: threw an exception.\n" << " Error msg:\n" << FormatDeathTestOutput(error_message); break; case RETURNED: buffer << " Result: illegal return in test statement.\n" << " Error msg:\n" << FormatDeathTestOutput(error_message); break; case DIED: if (status_ok) { const bool matched = RE::PartialMatch(error_message.c_str(), *regex()); if (matched) { success = true; } else { buffer << " Result: died but not with expected error.\n" << " Expected: " << regex()->pattern() << "\n" << "Actual msg:\n" << FormatDeathTestOutput(error_message); } } else { buffer << " Result: died but not with expected exit code:\n" << " " << ExitSummary(status()) << "\n" << "Actual msg:\n" << FormatDeathTestOutput(error_message); } break; case IN_PROGRESS: default: GTEST_LOG_(FATAL) << "DeathTest::Passed somehow called before conclusion of test"; } DeathTest::set_last_death_test_message(buffer.GetString()); return success; } # if GTEST_OS_WINDOWS // WindowsDeathTest implements death tests on Windows. Due to the // specifics of starting new processes on Windows, death tests there are // always threadsafe, and Google Test considers the // --gtest_death_test_style=fast setting to be equivalent to // --gtest_death_test_style=threadsafe there. // // A few implementation notes: Like the Linux version, the Windows // implementation uses pipes for child-to-parent communication. But due to // the specifics of pipes on Windows, some extra steps are required: // // 1. The parent creates a communication pipe and stores handles to both // ends of it. // 2. The parent starts the child and provides it with the information // necessary to acquire the handle to the write end of the pipe. // 3. The child acquires the write end of the pipe and signals the parent // using a Windows event. // 4. Now the parent can release the write end of the pipe on its side. If // this is done before step 3, the object's reference count goes down to // 0 and it is destroyed, preventing the child from acquiring it. The // parent now has to release it, or read operations on the read end of // the pipe will not return when the child terminates. // 5. The parent reads child's output through the pipe (outcome code and // any possible error messages) from the pipe, and its stderr and then // determines whether to fail the test. // // Note: to distinguish Win32 API calls from the local method and function // calls, the former are explicitly resolved in the global namespace. // class WindowsDeathTest : public DeathTestImpl { public: WindowsDeathTest(const char* a_statement, const RE* a_regex, const char* file, int line) : DeathTestImpl(a_statement, a_regex), file_(file), line_(line) {} // All of these virtual functions are inherited from DeathTest. virtual int Wait(); virtual TestRole AssumeRole(); private: // The name of the file in which the death test is located. const char* const file_; // The line number on which the death test is located. const int line_; // Handle to the write end of the pipe to the child process. AutoHandle write_handle_; // Child process handle. AutoHandle child_handle_; // Event the child process uses to signal the parent that it has // acquired the handle to the write end of the pipe. After seeing this // event the parent can release its own handles to make sure its // ReadFile() calls return when the child terminates. AutoHandle event_handle_; }; // Waits for the child in a death test to exit, returning its exit // status, or 0 if no child process exists. As a side effect, sets the // outcome data member. int WindowsDeathTest::Wait() { if (!spawned()) return 0; // Wait until the child either signals that it has acquired the write end // of the pipe or it dies. const HANDLE wait_handles[2] = { child_handle_.Get(), event_handle_.Get() }; switch (::WaitForMultipleObjects(2, wait_handles, FALSE, // Waits for any of the handles. INFINITE)) { case WAIT_OBJECT_0: case WAIT_OBJECT_0 + 1: break; default: GTEST_DEATH_TEST_CHECK_(false); // Should not get here. } // The child has acquired the write end of the pipe or exited. // We release the handle on our side and continue. write_handle_.Reset(); event_handle_.Reset(); ReadAndInterpretStatusByte(); // Waits for the child process to exit if it haven't already. This // returns immediately if the child has already exited, regardless of // whether previous calls to WaitForMultipleObjects synchronized on this // handle or not. GTEST_DEATH_TEST_CHECK_( WAIT_OBJECT_0 == ::WaitForSingleObject(child_handle_.Get(), INFINITE)); DWORD status_code; GTEST_DEATH_TEST_CHECK_( ::GetExitCodeProcess(child_handle_.Get(), &status_code) != FALSE); child_handle_.Reset(); set_status(static_cast(status_code)); return status(); } // The AssumeRole process for a Windows death test. It creates a child // process with the same executable as the current process to run the // death test. The child process is given the --gtest_filter and // --gtest_internal_run_death_test flags such that it knows to run the // current death test only. DeathTest::TestRole WindowsDeathTest::AssumeRole() { const UnitTestImpl* const impl = GetUnitTestImpl(); const InternalRunDeathTestFlag* const flag = impl->internal_run_death_test_flag(); const TestInfo* const info = impl->current_test_info(); const int death_test_index = info->result()->death_test_count(); if (flag != NULL) { // ParseInternalRunDeathTestFlag() has performed all the necessary // processing. set_write_fd(flag->write_fd()); return EXECUTE_TEST; } // WindowsDeathTest uses an anonymous pipe to communicate results of // a death test. SECURITY_ATTRIBUTES handles_are_inheritable = { sizeof(SECURITY_ATTRIBUTES), NULL, TRUE }; HANDLE read_handle, write_handle; GTEST_DEATH_TEST_CHECK_( ::CreatePipe(&read_handle, &write_handle, &handles_are_inheritable, 0) // Default buffer size. != FALSE); set_read_fd(::_open_osfhandle(reinterpret_cast(read_handle), O_RDONLY)); write_handle_.Reset(write_handle); event_handle_.Reset(::CreateEvent( &handles_are_inheritable, TRUE, // The event will automatically reset to non-signaled state. FALSE, // The initial state is non-signalled. NULL)); // The even is unnamed. GTEST_DEATH_TEST_CHECK_(event_handle_.Get() != NULL); const std::string filter_flag = std::string("--") + GTEST_FLAG_PREFIX_ + kFilterFlag + "=" + info->test_case_name() + "." + info->name(); const std::string internal_flag = std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + "=" + file_ + "|" + StreamableToString(line_) + "|" + StreamableToString(death_test_index) + "|" + StreamableToString(static_cast(::GetCurrentProcessId())) + // size_t has the same width as pointers on both 32-bit and 64-bit // Windows platforms. // See http://msdn.microsoft.com/en-us/library/tcxf1dw6.aspx. "|" + StreamableToString(reinterpret_cast(write_handle)) + "|" + StreamableToString(reinterpret_cast(event_handle_.Get())); char executable_path[_MAX_PATH + 1]; // NOLINT GTEST_DEATH_TEST_CHECK_( _MAX_PATH + 1 != ::GetModuleFileNameA(NULL, executable_path, _MAX_PATH)); std::string command_line = std::string(::GetCommandLineA()) + " " + filter_flag + " \"" + internal_flag + "\""; DeathTest::set_last_death_test_message(""); CaptureStderr(); // Flush the log buffers since the log streams are shared with the child. FlushInfoLog(); // The child process will share the standard handles with the parent. STARTUPINFOA startup_info; memset(&startup_info, 0, sizeof(STARTUPINFO)); startup_info.dwFlags = STARTF_USESTDHANDLES; startup_info.hStdInput = ::GetStdHandle(STD_INPUT_HANDLE); startup_info.hStdOutput = ::GetStdHandle(STD_OUTPUT_HANDLE); startup_info.hStdError = ::GetStdHandle(STD_ERROR_HANDLE); PROCESS_INFORMATION process_info; GTEST_DEATH_TEST_CHECK_(::CreateProcessA( executable_path, const_cast(command_line.c_str()), NULL, // Retuned process handle is not inheritable. NULL, // Retuned thread handle is not inheritable. TRUE, // Child inherits all inheritable handles (for write_handle_). 0x0, // Default creation flags. NULL, // Inherit the parent's environment. UnitTest::GetInstance()->original_working_dir(), &startup_info, &process_info) != FALSE); child_handle_.Reset(process_info.hProcess); ::CloseHandle(process_info.hThread); set_spawned(true); return OVERSEE_TEST; } # else // We are not on Windows. // ForkingDeathTest provides implementations for most of the abstract // methods of the DeathTest interface. Only the AssumeRole method is // left undefined. class ForkingDeathTest : public DeathTestImpl { public: ForkingDeathTest(const char* statement, const RE* regex); // All of these virtual functions are inherited from DeathTest. virtual int Wait(); protected: void set_child_pid(pid_t child_pid) { child_pid_ = child_pid; } private: // PID of child process during death test; 0 in the child process itself. pid_t child_pid_; }; // Constructs a ForkingDeathTest. ForkingDeathTest::ForkingDeathTest(const char* a_statement, const RE* a_regex) : DeathTestImpl(a_statement, a_regex), child_pid_(-1) {} // Waits for the child in a death test to exit, returning its exit // status, or 0 if no child process exists. As a side effect, sets the // outcome data member. int ForkingDeathTest::Wait() { if (!spawned()) return 0; ReadAndInterpretStatusByte(); int status_value; GTEST_DEATH_TEST_CHECK_SYSCALL_(waitpid(child_pid_, &status_value, 0)); set_status(status_value); return status_value; } // A concrete death test class that forks, then immediately runs the test // in the child process. class NoExecDeathTest : public ForkingDeathTest { public: NoExecDeathTest(const char* a_statement, const RE* a_regex) : ForkingDeathTest(a_statement, a_regex) { } virtual TestRole AssumeRole(); }; // The AssumeRole process for a fork-and-run death test. It implements a // straightforward fork, with a simple pipe to transmit the status byte. DeathTest::TestRole NoExecDeathTest::AssumeRole() { const size_t thread_count = GetThreadCount(); if (thread_count != 1) { GTEST_LOG_(WARNING) << DeathTestThreadWarning(thread_count); } int pipe_fd[2]; GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1); DeathTest::set_last_death_test_message(""); CaptureStderr(); // When we fork the process below, the log file buffers are copied, but the // file descriptors are shared. We flush all log files here so that closing // the file descriptors in the child process doesn't throw off the // synchronization between descriptors and buffers in the parent process. // This is as close to the fork as possible to avoid a race condition in case // there are multiple threads running before the death test, and another // thread writes to the log file. FlushInfoLog(); const pid_t child_pid = fork(); GTEST_DEATH_TEST_CHECK_(child_pid != -1); set_child_pid(child_pid); if (child_pid == 0) { GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[0])); set_write_fd(pipe_fd[1]); // Redirects all logging to stderr in the child process to prevent // concurrent writes to the log files. We capture stderr in the parent // process and append the child process' output to a log. LogToStderr(); // Event forwarding to the listeners of event listener API mush be shut // down in death test subprocesses. GetUnitTestImpl()->listeners()->SuppressEventForwarding(); g_in_fast_death_test_child = true; return EXECUTE_TEST; } else { GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1])); set_read_fd(pipe_fd[0]); set_spawned(true); return OVERSEE_TEST; } } // A concrete death test class that forks and re-executes the main // program from the beginning, with command-line flags set that cause // only this specific death test to be run. class ExecDeathTest : public ForkingDeathTest { public: ExecDeathTest(const char* a_statement, const RE* a_regex, const char* file, int line) : ForkingDeathTest(a_statement, a_regex), file_(file), line_(line) { } virtual TestRole AssumeRole(); private: static ::std::vector GetArgvsForDeathTestChildProcess() { ::std::vector args = GetInjectableArgvs(); return args; } // The name of the file in which the death test is located. const char* const file_; // The line number on which the death test is located. const int line_; }; // Utility class for accumulating command-line arguments. class Arguments { public: Arguments() { args_.push_back(NULL); } ~Arguments() { for (std::vector::iterator i = args_.begin(); i != args_.end(); ++i) { free(*i); } } void AddArgument(const char* argument) { args_.insert(args_.end() - 1, posix::StrDup(argument)); } template void AddArguments(const ::std::vector& arguments) { for (typename ::std::vector::const_iterator i = arguments.begin(); i != arguments.end(); ++i) { args_.insert(args_.end() - 1, posix::StrDup(i->c_str())); } } char* const* Argv() { return &args_[0]; } private: std::vector args_; }; // A struct that encompasses the arguments to the child process of a // threadsafe-style death test process. struct ExecDeathTestArgs { char* const* argv; // Command-line arguments for the child's call to exec int close_fd; // File descriptor to close; the read end of a pipe }; # if GTEST_OS_MAC inline char** GetEnviron() { // When Google Test is built as a framework on MacOS X, the environ variable // is unavailable. Apple's documentation (man environ) recommends using // _NSGetEnviron() instead. return *_NSGetEnviron(); } # else // Some POSIX platforms expect you to declare environ. extern "C" makes // it reside in the global namespace. extern "C" char** environ; inline char** GetEnviron() { return environ; } # endif // GTEST_OS_MAC # if !GTEST_OS_QNX // The main function for a threadsafe-style death test child process. // This function is called in a clone()-ed process and thus must avoid // any potentially unsafe operations like malloc or libc functions. static int ExecDeathTestChildMain(void* child_arg) { ExecDeathTestArgs* const args = static_cast(child_arg); GTEST_DEATH_TEST_CHECK_SYSCALL_(close(args->close_fd)); // We need to execute the test program in the same environment where // it was originally invoked. Therefore we change to the original // working directory first. const char* const original_dir = UnitTest::GetInstance()->original_working_dir(); // We can safely call chdir() as it's a direct system call. if (chdir(original_dir) != 0) { DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " + GetLastErrnoDescription()); return EXIT_FAILURE; } // We can safely call execve() as it's a direct system call. We // cannot use execvp() as it's a libc function and thus potentially // unsafe. Since execve() doesn't search the PATH, the user must // invoke the test program via a valid path that contains at least // one path separator. execve(args->argv[0], args->argv, GetEnviron()); DeathTestAbort(std::string("execve(") + args->argv[0] + ", ...) in " + original_dir + " failed: " + GetLastErrnoDescription()); return EXIT_FAILURE; } # endif // !GTEST_OS_QNX // Two utility routines that together determine the direction the stack // grows. // This could be accomplished more elegantly by a single recursive // function, but we want to guard against the unlikely possibility of // a smart compiler optimizing the recursion away. // // GTEST_NO_INLINE_ is required to prevent GCC 4.6 from inlining // StackLowerThanAddress into StackGrowsDown, which then doesn't give // correct answer. void StackLowerThanAddress(const void* ptr, bool* result) GTEST_NO_INLINE_; void StackLowerThanAddress(const void* ptr, bool* result) { int dummy; *result = (&dummy < ptr); } // Make sure AddressSanitizer does not tamper with the stack here. GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ bool StackGrowsDown() { int dummy; bool result; StackLowerThanAddress(&dummy, &result); return result; } // Spawns a child process with the same executable as the current process in // a thread-safe manner and instructs it to run the death test. The // implementation uses fork(2) + exec. On systems where clone(2) is // available, it is used instead, being slightly more thread-safe. On QNX, // fork supports only single-threaded environments, so this function uses // spawn(2) there instead. The function dies with an error message if // anything goes wrong. static pid_t ExecDeathTestSpawnChild(char* const* argv, int close_fd) { ExecDeathTestArgs args = { argv, close_fd }; pid_t child_pid = -1; # if GTEST_OS_QNX // Obtains the current directory and sets it to be closed in the child // process. const int cwd_fd = open(".", O_RDONLY); GTEST_DEATH_TEST_CHECK_(cwd_fd != -1); GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(cwd_fd, F_SETFD, FD_CLOEXEC)); // We need to execute the test program in the same environment where // it was originally invoked. Therefore we change to the original // working directory first. const char* const original_dir = UnitTest::GetInstance()->original_working_dir(); // We can safely call chdir() as it's a direct system call. if (chdir(original_dir) != 0) { DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " + GetLastErrnoDescription()); return EXIT_FAILURE; } int fd_flags; // Set close_fd to be closed after spawn. GTEST_DEATH_TEST_CHECK_SYSCALL_(fd_flags = fcntl(close_fd, F_GETFD)); GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(close_fd, F_SETFD, fd_flags | FD_CLOEXEC)); struct inheritance inherit = {0}; // spawn is a system call. child_pid = spawn(args.argv[0], 0, NULL, &inherit, args.argv, GetEnviron()); // Restores the current working directory. GTEST_DEATH_TEST_CHECK_(fchdir(cwd_fd) != -1); GTEST_DEATH_TEST_CHECK_SYSCALL_(close(cwd_fd)); # else // GTEST_OS_QNX # if GTEST_OS_LINUX // When a SIGPROF signal is received while fork() or clone() are executing, // the process may hang. To avoid this, we ignore SIGPROF here and re-enable // it after the call to fork()/clone() is complete. struct sigaction saved_sigprof_action; struct sigaction ignore_sigprof_action; memset(&ignore_sigprof_action, 0, sizeof(ignore_sigprof_action)); sigemptyset(&ignore_sigprof_action.sa_mask); ignore_sigprof_action.sa_handler = SIG_IGN; GTEST_DEATH_TEST_CHECK_SYSCALL_(sigaction( SIGPROF, &ignore_sigprof_action, &saved_sigprof_action)); # endif // GTEST_OS_LINUX # if GTEST_HAS_CLONE const bool use_fork = GTEST_FLAG(death_test_use_fork); if (!use_fork) { static const bool stack_grows_down = StackGrowsDown(); const size_t stack_size = getpagesize(); // MMAP_ANONYMOUS is not defined on Mac, so we use MAP_ANON instead. void* const stack = mmap(NULL, stack_size, PROT_READ | PROT_WRITE, MAP_ANON | MAP_PRIVATE, -1, 0); GTEST_DEATH_TEST_CHECK_(stack != MAP_FAILED); // Maximum stack alignment in bytes: For a downward-growing stack, this // amount is subtracted from size of the stack space to get an address // that is within the stack space and is aligned on all systems we care // about. As far as I know there is no ABI with stack alignment greater // than 64. We assume stack and stack_size already have alignment of // kMaxStackAlignment. const size_t kMaxStackAlignment = 64; void* const stack_top = static_cast(stack) + (stack_grows_down ? stack_size - kMaxStackAlignment : 0); GTEST_DEATH_TEST_CHECK_(stack_size > kMaxStackAlignment && reinterpret_cast(stack_top) % kMaxStackAlignment == 0); child_pid = clone(&ExecDeathTestChildMain, stack_top, SIGCHLD, &args); GTEST_DEATH_TEST_CHECK_(munmap(stack, stack_size) != -1); } # else const bool use_fork = true; # endif // GTEST_HAS_CLONE if (use_fork && (child_pid = fork()) == 0) { ExecDeathTestChildMain(&args); _exit(0); } # endif // GTEST_OS_QNX # if GTEST_OS_LINUX GTEST_DEATH_TEST_CHECK_SYSCALL_( sigaction(SIGPROF, &saved_sigprof_action, NULL)); # endif // GTEST_OS_LINUX GTEST_DEATH_TEST_CHECK_(child_pid != -1); return child_pid; } // The AssumeRole process for a fork-and-exec death test. It re-executes the // main program from the beginning, setting the --gtest_filter // and --gtest_internal_run_death_test flags to cause only the current // death test to be re-run. DeathTest::TestRole ExecDeathTest::AssumeRole() { const UnitTestImpl* const impl = GetUnitTestImpl(); const InternalRunDeathTestFlag* const flag = impl->internal_run_death_test_flag(); const TestInfo* const info = impl->current_test_info(); const int death_test_index = info->result()->death_test_count(); if (flag != NULL) { set_write_fd(flag->write_fd()); return EXECUTE_TEST; } int pipe_fd[2]; GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1); // Clear the close-on-exec flag on the write end of the pipe, lest // it be closed when the child process does an exec: GTEST_DEATH_TEST_CHECK_(fcntl(pipe_fd[1], F_SETFD, 0) != -1); const std::string filter_flag = std::string("--") + GTEST_FLAG_PREFIX_ + kFilterFlag + "=" + info->test_case_name() + "." + info->name(); const std::string internal_flag = std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + "=" + file_ + "|" + StreamableToString(line_) + "|" + StreamableToString(death_test_index) + "|" + StreamableToString(pipe_fd[1]); Arguments args; args.AddArguments(GetArgvsForDeathTestChildProcess()); args.AddArgument(filter_flag.c_str()); args.AddArgument(internal_flag.c_str()); DeathTest::set_last_death_test_message(""); CaptureStderr(); // See the comment in NoExecDeathTest::AssumeRole for why the next line // is necessary. FlushInfoLog(); const pid_t child_pid = ExecDeathTestSpawnChild(args.Argv(), pipe_fd[0]); GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1])); set_child_pid(child_pid); set_read_fd(pipe_fd[0]); set_spawned(true); return OVERSEE_TEST; } # endif // !GTEST_OS_WINDOWS // Creates a concrete DeathTest-derived class that depends on the // --gtest_death_test_style flag, and sets the pointer pointed to // by the "test" argument to its address. If the test should be // skipped, sets that pointer to NULL. Returns true, unless the // flag is set to an invalid value. bool DefaultDeathTestFactory::Create(const char* statement, const RE* regex, const char* file, int line, DeathTest** test) { UnitTestImpl* const impl = GetUnitTestImpl(); const InternalRunDeathTestFlag* const flag = impl->internal_run_death_test_flag(); const int death_test_index = impl->current_test_info() ->increment_death_test_count(); if (flag != NULL) { if (death_test_index > flag->index()) { DeathTest::set_last_death_test_message( "Death test count (" + StreamableToString(death_test_index) + ") somehow exceeded expected maximum (" + StreamableToString(flag->index()) + ")"); return false; } if (!(flag->file() == file && flag->line() == line && flag->index() == death_test_index)) { *test = NULL; return true; } } # if GTEST_OS_WINDOWS if (GTEST_FLAG(death_test_style) == "threadsafe" || GTEST_FLAG(death_test_style) == "fast") { *test = new WindowsDeathTest(statement, regex, file, line); } # else if (GTEST_FLAG(death_test_style) == "threadsafe") { *test = new ExecDeathTest(statement, regex, file, line); } else if (GTEST_FLAG(death_test_style) == "fast") { *test = new NoExecDeathTest(statement, regex); } # endif // GTEST_OS_WINDOWS else { // NOLINT - this is more readable than unbalanced brackets inside #if. DeathTest::set_last_death_test_message( "Unknown death test style \"" + GTEST_FLAG(death_test_style) + "\" encountered"); return false; } return true; } // Splits a given string on a given delimiter, populating a given // vector with the fields. GTEST_HAS_DEATH_TEST implies that we have // ::std::string, so we can use it here. static void SplitString(const ::std::string& str, char delimiter, ::std::vector< ::std::string>* dest) { ::std::vector< ::std::string> parsed; ::std::string::size_type pos = 0; while (::testing::internal::AlwaysTrue()) { const ::std::string::size_type colon = str.find(delimiter, pos); if (colon == ::std::string::npos) { parsed.push_back(str.substr(pos)); break; } else { parsed.push_back(str.substr(pos, colon - pos)); pos = colon + 1; } } dest->swap(parsed); } # if GTEST_OS_WINDOWS // Recreates the pipe and event handles from the provided parameters, // signals the event, and returns a file descriptor wrapped around the pipe // handle. This function is called in the child process only. int GetStatusFileDescriptor(unsigned int parent_process_id, size_t write_handle_as_size_t, size_t event_handle_as_size_t) { AutoHandle parent_process_handle(::OpenProcess(PROCESS_DUP_HANDLE, FALSE, // Non-inheritable. parent_process_id)); if (parent_process_handle.Get() == INVALID_HANDLE_VALUE) { DeathTestAbort("Unable to open parent process " + StreamableToString(parent_process_id)); } // TODO(vladl@google.com): Replace the following check with a // compile-time assertion when available. GTEST_CHECK_(sizeof(HANDLE) <= sizeof(size_t)); const HANDLE write_handle = reinterpret_cast(write_handle_as_size_t); HANDLE dup_write_handle; // The newly initialized handle is accessible only in in the parent // process. To obtain one accessible within the child, we need to use // DuplicateHandle. if (!::DuplicateHandle(parent_process_handle.Get(), write_handle, ::GetCurrentProcess(), &dup_write_handle, 0x0, // Requested privileges ignored since // DUPLICATE_SAME_ACCESS is used. FALSE, // Request non-inheritable handler. DUPLICATE_SAME_ACCESS)) { DeathTestAbort("Unable to duplicate the pipe handle " + StreamableToString(write_handle_as_size_t) + " from the parent process " + StreamableToString(parent_process_id)); } const HANDLE event_handle = reinterpret_cast(event_handle_as_size_t); HANDLE dup_event_handle; if (!::DuplicateHandle(parent_process_handle.Get(), event_handle, ::GetCurrentProcess(), &dup_event_handle, 0x0, FALSE, DUPLICATE_SAME_ACCESS)) { DeathTestAbort("Unable to duplicate the event handle " + StreamableToString(event_handle_as_size_t) + " from the parent process " + StreamableToString(parent_process_id)); } const int write_fd = ::_open_osfhandle(reinterpret_cast(dup_write_handle), O_APPEND); if (write_fd == -1) { DeathTestAbort("Unable to convert pipe handle " + StreamableToString(write_handle_as_size_t) + " to a file descriptor"); } // Signals the parent that the write end of the pipe has been acquired // so the parent can release its own write end. ::SetEvent(dup_event_handle); return write_fd; } # endif // GTEST_OS_WINDOWS // Returns a newly created InternalRunDeathTestFlag object with fields // initialized from the GTEST_FLAG(internal_run_death_test) flag if // the flag is specified; otherwise returns NULL. InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag() { if (GTEST_FLAG(internal_run_death_test) == "") return NULL; // GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we // can use it here. int line = -1; int index = -1; ::std::vector< ::std::string> fields; SplitString(GTEST_FLAG(internal_run_death_test).c_str(), '|', &fields); int write_fd = -1; # if GTEST_OS_WINDOWS unsigned int parent_process_id = 0; size_t write_handle_as_size_t = 0; size_t event_handle_as_size_t = 0; if (fields.size() != 6 || !ParseNaturalNumber(fields[1], &line) || !ParseNaturalNumber(fields[2], &index) || !ParseNaturalNumber(fields[3], &parent_process_id) || !ParseNaturalNumber(fields[4], &write_handle_as_size_t) || !ParseNaturalNumber(fields[5], &event_handle_as_size_t)) { DeathTestAbort("Bad --gtest_internal_run_death_test flag: " + GTEST_FLAG(internal_run_death_test)); } write_fd = GetStatusFileDescriptor(parent_process_id, write_handle_as_size_t, event_handle_as_size_t); # else if (fields.size() != 4 || !ParseNaturalNumber(fields[1], &line) || !ParseNaturalNumber(fields[2], &index) || !ParseNaturalNumber(fields[3], &write_fd)) { DeathTestAbort("Bad --gtest_internal_run_death_test flag: " + GTEST_FLAG(internal_run_death_test)); } # endif // GTEST_OS_WINDOWS return new InternalRunDeathTestFlag(fields[0], line, index, write_fd); } } // namespace internal #endif // GTEST_HAS_DEATH_TEST } // namespace testing // Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: keith.ray@gmail.com (Keith Ray) #include #if GTEST_OS_WINDOWS_MOBILE # include #elif GTEST_OS_WINDOWS # include # include #elif GTEST_OS_SYMBIAN // Symbian OpenC has PATH_MAX in sys/syslimits.h # include #else # include # include // Some Linux distributions define PATH_MAX here. #endif // GTEST_OS_WINDOWS_MOBILE #if GTEST_OS_WINDOWS # define GTEST_PATH_MAX_ _MAX_PATH #elif defined(PATH_MAX) # define GTEST_PATH_MAX_ PATH_MAX #elif defined(_XOPEN_PATH_MAX) # define GTEST_PATH_MAX_ _XOPEN_PATH_MAX #else # define GTEST_PATH_MAX_ _POSIX_PATH_MAX #endif // GTEST_OS_WINDOWS namespace testing { namespace internal { #if GTEST_OS_WINDOWS // On Windows, '\\' is the standard path separator, but many tools and the // Windows API also accept '/' as an alternate path separator. Unless otherwise // noted, a file path can contain either kind of path separators, or a mixture // of them. const char kPathSeparator = '\\'; const char kAlternatePathSeparator = '/'; const char kAlternatePathSeparatorString[] = "/"; # if GTEST_OS_WINDOWS_MOBILE // Windows CE doesn't have a current directory. You should not use // the current directory in tests on Windows CE, but this at least // provides a reasonable fallback. const char kCurrentDirectoryString[] = "\\"; // Windows CE doesn't define INVALID_FILE_ATTRIBUTES const DWORD kInvalidFileAttributes = 0xffffffff; # else const char kCurrentDirectoryString[] = ".\\"; # endif // GTEST_OS_WINDOWS_MOBILE #else const char kPathSeparator = '/'; const char kCurrentDirectoryString[] = "./"; #endif // GTEST_OS_WINDOWS // Returns whether the given character is a valid path separator. static bool IsPathSeparator(char c) { #if GTEST_HAS_ALT_PATH_SEP_ return (c == kPathSeparator) || (c == kAlternatePathSeparator); #else return c == kPathSeparator; #endif } // Returns the current working directory, or "" if unsuccessful. FilePath FilePath::GetCurrentDir() { #if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT // Windows CE doesn't have a current directory, so we just return // something reasonable. return FilePath(kCurrentDirectoryString); #elif GTEST_OS_WINDOWS char cwd[GTEST_PATH_MAX_ + 1] = { '\0' }; return FilePath(_getcwd(cwd, sizeof(cwd)) == NULL ? "" : cwd); #else char cwd[GTEST_PATH_MAX_ + 1] = { '\0' }; char* result = getcwd(cwd, sizeof(cwd)); # if GTEST_OS_NACL // getcwd will likely fail in NaCl due to the sandbox, so return something // reasonable. The user may have provided a shim implementation for getcwd, // however, so fallback only when failure is detected. return FilePath(result == NULL ? kCurrentDirectoryString : cwd); # endif // GTEST_OS_NACL return FilePath(result == NULL ? "" : cwd); #endif // GTEST_OS_WINDOWS_MOBILE } // Returns a copy of the FilePath with the case-insensitive extension removed. // Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns // FilePath("dir/file"). If a case-insensitive extension is not // found, returns a copy of the original FilePath. FilePath FilePath::RemoveExtension(const char* extension) const { const std::string dot_extension = std::string(".") + extension; if (String::EndsWithCaseInsensitive(pathname_, dot_extension)) { return FilePath(pathname_.substr( 0, pathname_.length() - dot_extension.length())); } return *this; } // Returns a pointer to the last occurence of a valid path separator in // the FilePath. On Windows, for example, both '/' and '\' are valid path // separators. Returns NULL if no path separator was found. const char* FilePath::FindLastPathSeparator() const { const char* const last_sep = strrchr(c_str(), kPathSeparator); #if GTEST_HAS_ALT_PATH_SEP_ const char* const last_alt_sep = strrchr(c_str(), kAlternatePathSeparator); // Comparing two pointers of which only one is NULL is undefined. if (last_alt_sep != NULL && (last_sep == NULL || last_alt_sep > last_sep)) { return last_alt_sep; } #endif return last_sep; } // Returns a copy of the FilePath with the directory part removed. // Example: FilePath("path/to/file").RemoveDirectoryName() returns // FilePath("file"). If there is no directory part ("just_a_file"), it returns // the FilePath unmodified. If there is no file part ("just_a_dir/") it // returns an empty FilePath (""). // On Windows platform, '\' is the path separator, otherwise it is '/'. FilePath FilePath::RemoveDirectoryName() const { const char* const last_sep = FindLastPathSeparator(); return last_sep ? FilePath(last_sep + 1) : *this; } // RemoveFileName returns the directory path with the filename removed. // Example: FilePath("path/to/file").RemoveFileName() returns "path/to/". // If the FilePath is "a_file" or "/a_file", RemoveFileName returns // FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does // not have a file, like "just/a/dir/", it returns the FilePath unmodified. // On Windows platform, '\' is the path separator, otherwise it is '/'. FilePath FilePath::RemoveFileName() const { const char* const last_sep = FindLastPathSeparator(); std::string dir; if (last_sep) { dir = std::string(c_str(), last_sep + 1 - c_str()); } else { dir = kCurrentDirectoryString; } return FilePath(dir); } // Helper functions for naming files in a directory for xml output. // Given directory = "dir", base_name = "test", number = 0, // extension = "xml", returns "dir/test.xml". If number is greater // than zero (e.g., 12), returns "dir/test_12.xml". // On Windows platform, uses \ as the separator rather than /. FilePath FilePath::MakeFileName(const FilePath& directory, const FilePath& base_name, int number, const char* extension) { std::string file; if (number == 0) { file = base_name.string() + "." + extension; } else { file = base_name.string() + "_" + StreamableToString(number) + "." + extension; } return ConcatPaths(directory, FilePath(file)); } // Given directory = "dir", relative_path = "test.xml", returns "dir/test.xml". // On Windows, uses \ as the separator rather than /. FilePath FilePath::ConcatPaths(const FilePath& directory, const FilePath& relative_path) { if (directory.IsEmpty()) return relative_path; const FilePath dir(directory.RemoveTrailingPathSeparator()); return FilePath(dir.string() + kPathSeparator + relative_path.string()); } // Returns true if pathname describes something findable in the file-system, // either a file, directory, or whatever. bool FilePath::FileOrDirectoryExists() const { #if GTEST_OS_WINDOWS_MOBILE LPCWSTR unicode = String::AnsiToUtf16(pathname_.c_str()); const DWORD attributes = GetFileAttributes(unicode); delete [] unicode; return attributes != kInvalidFileAttributes; #else posix::StatStruct file_stat; return posix::Stat(pathname_.c_str(), &file_stat) == 0; #endif // GTEST_OS_WINDOWS_MOBILE } // Returns true if pathname describes a directory in the file-system // that exists. bool FilePath::DirectoryExists() const { bool result = false; #if GTEST_OS_WINDOWS // Don't strip off trailing separator if path is a root directory on // Windows (like "C:\\"). const FilePath& path(IsRootDirectory() ? *this : RemoveTrailingPathSeparator()); #else const FilePath& path(*this); #endif #if GTEST_OS_WINDOWS_MOBILE LPCWSTR unicode = String::AnsiToUtf16(path.c_str()); const DWORD attributes = GetFileAttributes(unicode); delete [] unicode; if ((attributes != kInvalidFileAttributes) && (attributes & FILE_ATTRIBUTE_DIRECTORY)) { result = true; } #else posix::StatStruct file_stat; result = posix::Stat(path.c_str(), &file_stat) == 0 && posix::IsDir(file_stat); #endif // GTEST_OS_WINDOWS_MOBILE return result; } // Returns true if pathname describes a root directory. (Windows has one // root directory per disk drive.) bool FilePath::IsRootDirectory() const { #if GTEST_OS_WINDOWS // TODO(wan@google.com): on Windows a network share like // \\server\share can be a root directory, although it cannot be the // current directory. Handle this properly. return pathname_.length() == 3 && IsAbsolutePath(); #else return pathname_.length() == 1 && IsPathSeparator(pathname_.c_str()[0]); #endif } // Returns true if pathname describes an absolute path. bool FilePath::IsAbsolutePath() const { const char* const name = pathname_.c_str(); #if GTEST_OS_WINDOWS return pathname_.length() >= 3 && ((name[0] >= 'a' && name[0] <= 'z') || (name[0] >= 'A' && name[0] <= 'Z')) && name[1] == ':' && IsPathSeparator(name[2]); #else return IsPathSeparator(name[0]); #endif } // Returns a pathname for a file that does not currently exist. The pathname // will be directory/base_name.extension or // directory/base_name_.extension if directory/base_name.extension // already exists. The number will be incremented until a pathname is found // that does not already exist. // Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'. // There could be a race condition if two or more processes are calling this // function at the same time -- they could both pick the same filename. FilePath FilePath::GenerateUniqueFileName(const FilePath& directory, const FilePath& base_name, const char* extension) { FilePath full_pathname; int number = 0; do { full_pathname.Set(MakeFileName(directory, base_name, number++, extension)); } while (full_pathname.FileOrDirectoryExists()); return full_pathname; } // Returns true if FilePath ends with a path separator, which indicates that // it is intended to represent a directory. Returns false otherwise. // This does NOT check that a directory (or file) actually exists. bool FilePath::IsDirectory() const { return !pathname_.empty() && IsPathSeparator(pathname_.c_str()[pathname_.length() - 1]); } // Create directories so that path exists. Returns true if successful or if // the directories already exist; returns false if unable to create directories // for any reason. bool FilePath::CreateDirectoriesRecursively() const { if (!this->IsDirectory()) { return false; } if (pathname_.length() == 0 || this->DirectoryExists()) { return true; } const FilePath parent(this->RemoveTrailingPathSeparator().RemoveFileName()); return parent.CreateDirectoriesRecursively() && this->CreateFolder(); } // Create the directory so that path exists. Returns true if successful or // if the directory already exists; returns false if unable to create the // directory for any reason, including if the parent directory does not // exist. Not named "CreateDirectory" because that's a macro on Windows. bool FilePath::CreateFolder() const { #if GTEST_OS_WINDOWS_MOBILE FilePath removed_sep(this->RemoveTrailingPathSeparator()); LPCWSTR unicode = String::AnsiToUtf16(removed_sep.c_str()); int result = CreateDirectory(unicode, NULL) ? 0 : -1; delete [] unicode; #elif GTEST_OS_WINDOWS int result = _mkdir(pathname_.c_str()); #else int result = mkdir(pathname_.c_str(), 0777); #endif // GTEST_OS_WINDOWS_MOBILE if (result == -1) { return this->DirectoryExists(); // An error is OK if the directory exists. } return true; // No error. } // If input name has a trailing separator character, remove it and return the // name, otherwise return the name string unmodified. // On Windows platform, uses \ as the separator, other platforms use /. FilePath FilePath::RemoveTrailingPathSeparator() const { return IsDirectory() ? FilePath(pathname_.substr(0, pathname_.length() - 1)) : *this; } // Removes any redundant separators that might be in the pathname. // For example, "bar///foo" becomes "bar/foo". Does not eliminate other // redundancies that might be in a pathname involving "." or "..". // TODO(wan@google.com): handle Windows network shares (e.g. \\server\share). void FilePath::Normalize() { if (pathname_.c_str() == NULL) { pathname_ = ""; return; } const char* src = pathname_.c_str(); char* const dest = new char[pathname_.length() + 1]; char* dest_ptr = dest; memset(dest_ptr, 0, pathname_.length() + 1); while (*src != '\0') { *dest_ptr = *src; if (!IsPathSeparator(*src)) { src++; } else { #if GTEST_HAS_ALT_PATH_SEP_ if (*dest_ptr == kAlternatePathSeparator) { *dest_ptr = kPathSeparator; } #endif while (IsPathSeparator(*src)) src++; } dest_ptr++; } *dest_ptr = '\0'; pathname_ = dest; delete[] dest; } } // namespace internal } // namespace testing // Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) #include #include #include #include #if GTEST_OS_WINDOWS # include # include # include # include // Used in ThreadLocal. #else # include #endif // GTEST_OS_WINDOWS #if GTEST_OS_MAC # include # include # include #endif // GTEST_OS_MAC #if GTEST_OS_QNX # include # include # include #endif // GTEST_OS_QNX // Indicates that this translation unit is part of Google Test's // implementation. It must come before gtest-internal-inl.h is // included, or there will be a compiler error. This trick exists to // prevent the accidental inclusion of gtest-internal-inl.h in the // user's code. #define GTEST_IMPLEMENTATION_ 1 #undef GTEST_IMPLEMENTATION_ namespace testing { namespace internal { #if defined(_MSC_VER) || defined(__BORLANDC__) // MSVC and C++Builder do not provide a definition of STDERR_FILENO. const int kStdOutFileno = 1; const int kStdErrFileno = 2; #else const int kStdOutFileno = STDOUT_FILENO; const int kStdErrFileno = STDERR_FILENO; #endif // _MSC_VER #if GTEST_OS_MAC // Returns the number of threads running in the process, or 0 to indicate that // we cannot detect it. size_t GetThreadCount() { const task_t task = mach_task_self(); mach_msg_type_number_t thread_count; thread_act_array_t thread_list; const kern_return_t status = task_threads(task, &thread_list, &thread_count); if (status == KERN_SUCCESS) { // task_threads allocates resources in thread_list and we need to free them // to avoid leaks. vm_deallocate(task, reinterpret_cast(thread_list), sizeof(thread_t) * thread_count); return static_cast(thread_count); } else { return 0; } } #elif GTEST_OS_QNX // Returns the number of threads running in the process, or 0 to indicate that // we cannot detect it. size_t GetThreadCount() { const int fd = open("/proc/self/as", O_RDONLY); if (fd < 0) { return 0; } procfs_info process_info; const int status = devctl(fd, DCMD_PROC_INFO, &process_info, sizeof(process_info), NULL); close(fd); if (status == EOK) { return static_cast(process_info.num_threads); } else { return 0; } } #else size_t GetThreadCount() { // There's no portable way to detect the number of threads, so we just // return 0 to indicate that we cannot detect it. return 0; } #endif // GTEST_OS_MAC #if GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS void SleepMilliseconds(int n) { ::Sleep(n); } AutoHandle::AutoHandle() : handle_(INVALID_HANDLE_VALUE) {} AutoHandle::AutoHandle(Handle handle) : handle_(handle) {} AutoHandle::~AutoHandle() { Reset(); } AutoHandle::Handle AutoHandle::Get() const { return handle_; } void AutoHandle::Reset() { Reset(INVALID_HANDLE_VALUE); } void AutoHandle::Reset(HANDLE handle) { // Resetting with the same handle we already own is invalid. if (handle_ != handle) { if (IsCloseable()) { ::CloseHandle(handle_); } handle_ = handle; } else { GTEST_CHECK_(!IsCloseable()) << "Resetting a valid handle to itself is likely a programmer error " "and thus not allowed."; } } bool AutoHandle::IsCloseable() const { // Different Windows APIs may use either of these values to represent an // invalid handle. return handle_ != NULL && handle_ != INVALID_HANDLE_VALUE; } Notification::Notification() : event_(::CreateEvent(NULL, // Default security attributes. TRUE, // Do not reset automatically. FALSE, // Initially unset. NULL)) { // Anonymous event. GTEST_CHECK_(event_.Get() != NULL); } void Notification::Notify() { GTEST_CHECK_(::SetEvent(event_.Get()) != FALSE); } void Notification::WaitForNotification() { GTEST_CHECK_( ::WaitForSingleObject(event_.Get(), INFINITE) == WAIT_OBJECT_0); } Mutex::Mutex() : type_(kDynamic), owner_thread_id_(0), critical_section_init_phase_(0), critical_section_(new CRITICAL_SECTION) { ::InitializeCriticalSection(critical_section_); } Mutex::~Mutex() { // Static mutexes are leaked intentionally. It is not thread-safe to try // to clean them up. // TODO(yukawa): Switch to Slim Reader/Writer (SRW) Locks, which requires // nothing to clean it up but is available only on Vista and later. // http://msdn.microsoft.com/en-us/library/windows/desktop/aa904937.aspx if (type_ == kDynamic) { ::DeleteCriticalSection(critical_section_); delete critical_section_; critical_section_ = NULL; } } void Mutex::Lock() { ThreadSafeLazyInit(); ::EnterCriticalSection(critical_section_); owner_thread_id_ = ::GetCurrentThreadId(); } void Mutex::Unlock() { ThreadSafeLazyInit(); // We don't protect writing to owner_thread_id_ here, as it's the // caller's responsibility to ensure that the current thread holds the // mutex when this is called. owner_thread_id_ = 0; ::LeaveCriticalSection(critical_section_); } // Does nothing if the current thread holds the mutex. Otherwise, crashes // with high probability. void Mutex::AssertHeld() { ThreadSafeLazyInit(); GTEST_CHECK_(owner_thread_id_ == ::GetCurrentThreadId()) << "The current thread is not holding the mutex @" << this; } // Initializes owner_thread_id_ and critical_section_ in static mutexes. void Mutex::ThreadSafeLazyInit() { // Dynamic mutexes are initialized in the constructor. if (type_ == kStatic) { switch ( ::InterlockedCompareExchange(&critical_section_init_phase_, 1L, 0L)) { case 0: // If critical_section_init_phase_ was 0 before the exchange, we // are the first to test it and need to perform the initialization. owner_thread_id_ = 0; critical_section_ = new CRITICAL_SECTION; ::InitializeCriticalSection(critical_section_); // Updates the critical_section_init_phase_ to 2 to signal // initialization complete. GTEST_CHECK_(::InterlockedCompareExchange( &critical_section_init_phase_, 2L, 1L) == 1L); break; case 1: // Somebody else is already initializing the mutex; spin until they // are done. while (::InterlockedCompareExchange(&critical_section_init_phase_, 2L, 2L) != 2L) { // Possibly yields the rest of the thread's time slice to other // threads. ::Sleep(0); } break; case 2: break; // The mutex is already initialized and ready for use. default: GTEST_CHECK_(false) << "Unexpected value of critical_section_init_phase_ " << "while initializing a static mutex."; } } } namespace { class ThreadWithParamSupport : public ThreadWithParamBase { public: static HANDLE CreateThread(Runnable* runnable, Notification* thread_can_start) { ThreadMainParam* param = new ThreadMainParam(runnable, thread_can_start); DWORD thread_id; // TODO(yukawa): Consider to use _beginthreadex instead. HANDLE thread_handle = ::CreateThread( NULL, // Default security. 0, // Default stack size. &ThreadWithParamSupport::ThreadMain, param, // Parameter to ThreadMainStatic 0x0, // Default creation flags. &thread_id); // Need a valid pointer for the call to work under Win98. GTEST_CHECK_(thread_handle != NULL) << "CreateThread failed with error " << ::GetLastError() << "."; if (thread_handle == NULL) { delete param; } return thread_handle; } private: struct ThreadMainParam { ThreadMainParam(Runnable* runnable, Notification* thread_can_start) : runnable_(runnable), thread_can_start_(thread_can_start) { } scoped_ptr runnable_; // Does not own. Notification* thread_can_start_; }; static DWORD WINAPI ThreadMain(void* ptr) { // Transfers ownership. scoped_ptr param(static_cast(ptr)); if (param->thread_can_start_ != NULL) param->thread_can_start_->WaitForNotification(); param->runnable_->Run(); return 0; } // Prohibit instantiation. ThreadWithParamSupport(); GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParamSupport); }; } // namespace ThreadWithParamBase::ThreadWithParamBase(Runnable *runnable, Notification* thread_can_start) : thread_(ThreadWithParamSupport::CreateThread(runnable, thread_can_start)) { } ThreadWithParamBase::~ThreadWithParamBase() { Join(); } void ThreadWithParamBase::Join() { GTEST_CHECK_(::WaitForSingleObject(thread_.Get(), INFINITE) == WAIT_OBJECT_0) << "Failed to join the thread with error " << ::GetLastError() << "."; } // Maps a thread to a set of ThreadIdToThreadLocals that have values // instantiated on that thread and notifies them when the thread exits. A // ThreadLocal instance is expected to persist until all threads it has // values on have terminated. class ThreadLocalRegistryImpl { public: // Registers thread_local_instance as having value on the current thread. // Returns a value that can be used to identify the thread from other threads. static ThreadLocalValueHolderBase* GetValueOnCurrentThread( const ThreadLocalBase* thread_local_instance) { DWORD current_thread = ::GetCurrentThreadId(); MutexLock lock(&mutex_); ThreadIdToThreadLocals* const thread_to_thread_locals = GetThreadLocalsMapLocked(); ThreadIdToThreadLocals::iterator thread_local_pos = thread_to_thread_locals->find(current_thread); if (thread_local_pos == thread_to_thread_locals->end()) { thread_local_pos = thread_to_thread_locals->insert( std::make_pair(current_thread, ThreadLocalValues())).first; StartWatcherThreadFor(current_thread); } ThreadLocalValues& thread_local_values = thread_local_pos->second; ThreadLocalValues::iterator value_pos = thread_local_values.find(thread_local_instance); if (value_pos == thread_local_values.end()) { value_pos = thread_local_values .insert(std::make_pair( thread_local_instance, linked_ptr( thread_local_instance->NewValueForCurrentThread()))) .first; } return value_pos->second.get(); } static void OnThreadLocalDestroyed( const ThreadLocalBase* thread_local_instance) { std::vector > value_holders; // Clean up the ThreadLocalValues data structure while holding the lock, but // defer the destruction of the ThreadLocalValueHolderBases. { MutexLock lock(&mutex_); ThreadIdToThreadLocals* const thread_to_thread_locals = GetThreadLocalsMapLocked(); for (ThreadIdToThreadLocals::iterator it = thread_to_thread_locals->begin(); it != thread_to_thread_locals->end(); ++it) { ThreadLocalValues& thread_local_values = it->second; ThreadLocalValues::iterator value_pos = thread_local_values.find(thread_local_instance); if (value_pos != thread_local_values.end()) { value_holders.push_back(value_pos->second); thread_local_values.erase(value_pos); // This 'if' can only be successful at most once, so theoretically we // could break out of the loop here, but we don't bother doing so. } } } // Outside the lock, let the destructor for 'value_holders' deallocate the // ThreadLocalValueHolderBases. } static void OnThreadExit(DWORD thread_id) { GTEST_CHECK_(thread_id != 0) << ::GetLastError(); std::vector > value_holders; // Clean up the ThreadIdToThreadLocals data structure while holding the // lock, but defer the destruction of the ThreadLocalValueHolderBases. { MutexLock lock(&mutex_); ThreadIdToThreadLocals* const thread_to_thread_locals = GetThreadLocalsMapLocked(); ThreadIdToThreadLocals::iterator thread_local_pos = thread_to_thread_locals->find(thread_id); if (thread_local_pos != thread_to_thread_locals->end()) { ThreadLocalValues& thread_local_values = thread_local_pos->second; for (ThreadLocalValues::iterator value_pos = thread_local_values.begin(); value_pos != thread_local_values.end(); ++value_pos) { value_holders.push_back(value_pos->second); } thread_to_thread_locals->erase(thread_local_pos); } } // Outside the lock, let the destructor for 'value_holders' deallocate the // ThreadLocalValueHolderBases. } private: // In a particular thread, maps a ThreadLocal object to its value. typedef std::map > ThreadLocalValues; // Stores all ThreadIdToThreadLocals having values in a thread, indexed by // thread's ID. typedef std::map ThreadIdToThreadLocals; // Holds the thread id and thread handle that we pass from // StartWatcherThreadFor to WatcherThreadFunc. typedef std::pair ThreadIdAndHandle; static void StartWatcherThreadFor(DWORD thread_id) { // The returned handle will be kept in thread_map and closed by // watcher_thread in WatcherThreadFunc. HANDLE thread = ::OpenThread(SYNCHRONIZE | THREAD_QUERY_INFORMATION, FALSE, thread_id); GTEST_CHECK_(thread != NULL); // We need to to pass a valid thread ID pointer into CreateThread for it // to work correctly under Win98. DWORD watcher_thread_id; HANDLE watcher_thread = ::CreateThread( NULL, // Default security. 0, // Default stack size &ThreadLocalRegistryImpl::WatcherThreadFunc, reinterpret_cast(new ThreadIdAndHandle(thread_id, thread)), CREATE_SUSPENDED, &watcher_thread_id); GTEST_CHECK_(watcher_thread != NULL); // Give the watcher thread the same priority as ours to avoid being // blocked by it. ::SetThreadPriority(watcher_thread, ::GetThreadPriority(::GetCurrentThread())); ::ResumeThread(watcher_thread); ::CloseHandle(watcher_thread); } // Monitors exit from a given thread and notifies those // ThreadIdToThreadLocals about thread termination. static DWORD WINAPI WatcherThreadFunc(LPVOID param) { const ThreadIdAndHandle* tah = reinterpret_cast(param); GTEST_CHECK_( ::WaitForSingleObject(tah->second, INFINITE) == WAIT_OBJECT_0); OnThreadExit(tah->first); ::CloseHandle(tah->second); delete tah; return 0; } // Returns map of thread local instances. static ThreadIdToThreadLocals* GetThreadLocalsMapLocked() { mutex_.AssertHeld(); static ThreadIdToThreadLocals* map = new ThreadIdToThreadLocals; return map; } // Protects access to GetThreadLocalsMapLocked() and its return value. static Mutex mutex_; // Protects access to GetThreadMapLocked() and its return value. static Mutex thread_map_mutex_; }; Mutex ThreadLocalRegistryImpl::mutex_(Mutex::kStaticMutex); Mutex ThreadLocalRegistryImpl::thread_map_mutex_(Mutex::kStaticMutex); ThreadLocalValueHolderBase* ThreadLocalRegistry::GetValueOnCurrentThread( const ThreadLocalBase* thread_local_instance) { return ThreadLocalRegistryImpl::GetValueOnCurrentThread( thread_local_instance); } void ThreadLocalRegistry::OnThreadLocalDestroyed( const ThreadLocalBase* thread_local_instance) { ThreadLocalRegistryImpl::OnThreadLocalDestroyed(thread_local_instance); } #endif // GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS #if GTEST_USES_POSIX_RE // Implements RE. Currently only needed for death tests. RE::~RE() { if (is_valid_) { // regfree'ing an invalid regex might crash because the content // of the regex is undefined. Since the regex's are essentially // the same, one cannot be valid (or invalid) without the other // being so too. regfree(&partial_regex_); regfree(&full_regex_); } free(const_cast(pattern_)); } // Returns true iff regular expression re matches the entire str. bool RE::FullMatch(const char* str, const RE& re) { if (!re.is_valid_) return false; regmatch_t match; return regexec(&re.full_regex_, str, 1, &match, 0) == 0; } // Returns true iff regular expression re matches a substring of str // (including str itself). bool RE::PartialMatch(const char* str, const RE& re) { if (!re.is_valid_) return false; regmatch_t match; return regexec(&re.partial_regex_, str, 1, &match, 0) == 0; } // Initializes an RE from its string representation. void RE::Init(const char* regex) { pattern_ = posix::StrDup(regex); // Reserves enough bytes to hold the regular expression used for a // full match. const size_t full_regex_len = strlen(regex) + 10; char* const full_pattern = new char[full_regex_len]; snprintf(full_pattern, full_regex_len, "^(%s)$", regex); is_valid_ = regcomp(&full_regex_, full_pattern, REG_EXTENDED) == 0; // We want to call regcomp(&partial_regex_, ...) even if the // previous expression returns false. Otherwise partial_regex_ may // not be properly initialized can may cause trouble when it's // freed. // // Some implementation of POSIX regex (e.g. on at least some // versions of Cygwin) doesn't accept the empty string as a valid // regex. We change it to an equivalent form "()" to be safe. if (is_valid_) { const char* const partial_regex = (*regex == '\0') ? "()" : regex; is_valid_ = regcomp(&partial_regex_, partial_regex, REG_EXTENDED) == 0; } EXPECT_TRUE(is_valid_) << "Regular expression \"" << regex << "\" is not a valid POSIX Extended regular expression."; delete[] full_pattern; } #elif GTEST_USES_SIMPLE_RE // Returns true iff ch appears anywhere in str (excluding the // terminating '\0' character). bool IsInSet(char ch, const char* str) { return ch != '\0' && strchr(str, ch) != NULL; } // Returns true iff ch belongs to the given classification. Unlike // similar functions in , these aren't affected by the // current locale. bool IsAsciiDigit(char ch) { return '0' <= ch && ch <= '9'; } bool IsAsciiPunct(char ch) { return IsInSet(ch, "^-!\"#$%&'()*+,./:;<=>?@[\\]_`{|}~"); } bool IsRepeat(char ch) { return IsInSet(ch, "?*+"); } bool IsAsciiWhiteSpace(char ch) { return IsInSet(ch, " \f\n\r\t\v"); } bool IsAsciiWordChar(char ch) { return ('a' <= ch && ch <= 'z') || ('A' <= ch && ch <= 'Z') || ('0' <= ch && ch <= '9') || ch == '_'; } // Returns true iff "\\c" is a supported escape sequence. bool IsValidEscape(char c) { return (IsAsciiPunct(c) || IsInSet(c, "dDfnrsStvwW")); } // Returns true iff the given atom (specified by escaped and pattern) // matches ch. The result is undefined if the atom is invalid. bool AtomMatchesChar(bool escaped, char pattern_char, char ch) { if (escaped) { // "\\p" where p is pattern_char. switch (pattern_char) { case 'd': return IsAsciiDigit(ch); case 'D': return !IsAsciiDigit(ch); case 'f': return ch == '\f'; case 'n': return ch == '\n'; case 'r': return ch == '\r'; case 's': return IsAsciiWhiteSpace(ch); case 'S': return !IsAsciiWhiteSpace(ch); case 't': return ch == '\t'; case 'v': return ch == '\v'; case 'w': return IsAsciiWordChar(ch); case 'W': return !IsAsciiWordChar(ch); } return IsAsciiPunct(pattern_char) && pattern_char == ch; } return (pattern_char == '.' && ch != '\n') || pattern_char == ch; } // Helper function used by ValidateRegex() to format error messages. std::string FormatRegexSyntaxError(const char* regex, int index) { return (Message() << "Syntax error at index " << index << " in simple regular expression \"" << regex << "\": ").GetString(); } // Generates non-fatal failures and returns false if regex is invalid; // otherwise returns true. bool ValidateRegex(const char* regex) { if (regex == NULL) { // TODO(wan@google.com): fix the source file location in the // assertion failures to match where the regex is used in user // code. ADD_FAILURE() << "NULL is not a valid simple regular expression."; return false; } bool is_valid = true; // True iff ?, *, or + can follow the previous atom. bool prev_repeatable = false; for (int i = 0; regex[i]; i++) { if (regex[i] == '\\') { // An escape sequence i++; if (regex[i] == '\0') { ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1) << "'\\' cannot appear at the end."; return false; } if (!IsValidEscape(regex[i])) { ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1) << "invalid escape sequence \"\\" << regex[i] << "\"."; is_valid = false; } prev_repeatable = true; } else { // Not an escape sequence. const char ch = regex[i]; if (ch == '^' && i > 0) { ADD_FAILURE() << FormatRegexSyntaxError(regex, i) << "'^' can only appear at the beginning."; is_valid = false; } else if (ch == '$' && regex[i + 1] != '\0') { ADD_FAILURE() << FormatRegexSyntaxError(regex, i) << "'$' can only appear at the end."; is_valid = false; } else if (IsInSet(ch, "()[]{}|")) { ADD_FAILURE() << FormatRegexSyntaxError(regex, i) << "'" << ch << "' is unsupported."; is_valid = false; } else if (IsRepeat(ch) && !prev_repeatable) { ADD_FAILURE() << FormatRegexSyntaxError(regex, i) << "'" << ch << "' can only follow a repeatable token."; is_valid = false; } prev_repeatable = !IsInSet(ch, "^$?*+"); } } return is_valid; } // Matches a repeated regex atom followed by a valid simple regular // expression. The regex atom is defined as c if escaped is false, // or \c otherwise. repeat is the repetition meta character (?, *, // or +). The behavior is undefined if str contains too many // characters to be indexable by size_t, in which case the test will // probably time out anyway. We are fine with this limitation as // std::string has it too. bool MatchRepetitionAndRegexAtHead( bool escaped, char c, char repeat, const char* regex, const char* str) { const size_t min_count = (repeat == '+') ? 1 : 0; const size_t max_count = (repeat == '?') ? 1 : static_cast(-1) - 1; // We cannot call numeric_limits::max() as it conflicts with the // max() macro on Windows. for (size_t i = 0; i <= max_count; ++i) { // We know that the atom matches each of the first i characters in str. if (i >= min_count && MatchRegexAtHead(regex, str + i)) { // We have enough matches at the head, and the tail matches too. // Since we only care about *whether* the pattern matches str // (as opposed to *how* it matches), there is no need to find a // greedy match. return true; } if (str[i] == '\0' || !AtomMatchesChar(escaped, c, str[i])) return false; } return false; } // Returns true iff regex matches a prefix of str. regex must be a // valid simple regular expression and not start with "^", or the // result is undefined. bool MatchRegexAtHead(const char* regex, const char* str) { if (*regex == '\0') // An empty regex matches a prefix of anything. return true; // "$" only matches the end of a string. Note that regex being // valid guarantees that there's nothing after "$" in it. if (*regex == '$') return *str == '\0'; // Is the first thing in regex an escape sequence? const bool escaped = *regex == '\\'; if (escaped) ++regex; if (IsRepeat(regex[1])) { // MatchRepetitionAndRegexAtHead() calls MatchRegexAtHead(), so // here's an indirect recursion. It terminates as the regex gets // shorter in each recursion. return MatchRepetitionAndRegexAtHead( escaped, regex[0], regex[1], regex + 2, str); } else { // regex isn't empty, isn't "$", and doesn't start with a // repetition. We match the first atom of regex with the first // character of str and recurse. return (*str != '\0') && AtomMatchesChar(escaped, *regex, *str) && MatchRegexAtHead(regex + 1, str + 1); } } // Returns true iff regex matches any substring of str. regex must be // a valid simple regular expression, or the result is undefined. // // The algorithm is recursive, but the recursion depth doesn't exceed // the regex length, so we won't need to worry about running out of // stack space normally. In rare cases the time complexity can be // exponential with respect to the regex length + the string length, // but usually it's must faster (often close to linear). bool MatchRegexAnywhere(const char* regex, const char* str) { if (regex == NULL || str == NULL) return false; if (*regex == '^') return MatchRegexAtHead(regex + 1, str); // A successful match can be anywhere in str. do { if (MatchRegexAtHead(regex, str)) return true; } while (*str++ != '\0'); return false; } // Implements the RE class. RE::~RE() { free(const_cast(pattern_)); free(const_cast(full_pattern_)); } // Returns true iff regular expression re matches the entire str. bool RE::FullMatch(const char* str, const RE& re) { return re.is_valid_ && MatchRegexAnywhere(re.full_pattern_, str); } // Returns true iff regular expression re matches a substring of str // (including str itself). bool RE::PartialMatch(const char* str, const RE& re) { return re.is_valid_ && MatchRegexAnywhere(re.pattern_, str); } // Initializes an RE from its string representation. void RE::Init(const char* regex) { pattern_ = full_pattern_ = NULL; if (regex != NULL) { pattern_ = posix::StrDup(regex); } is_valid_ = ValidateRegex(regex); if (!is_valid_) { // No need to calculate the full pattern when the regex is invalid. return; } const size_t len = strlen(regex); // Reserves enough bytes to hold the regular expression used for a // full match: we need space to prepend a '^', append a '$', and // terminate the string with '\0'. char* buffer = static_cast(malloc(len + 3)); full_pattern_ = buffer; if (*regex != '^') *buffer++ = '^'; // Makes sure full_pattern_ starts with '^'. // We don't use snprintf or strncpy, as they trigger a warning when // compiled with VC++ 8.0. memcpy(buffer, regex, len); buffer += len; if (len == 0 || regex[len - 1] != '$') *buffer++ = '$'; // Makes sure full_pattern_ ends with '$'. *buffer = '\0'; } #endif // GTEST_USES_POSIX_RE const char kUnknownFile[] = "unknown file"; // Formats a source file path and a line number as they would appear // in an error message from the compiler used to compile this code. GTEST_API_ ::std::string FormatFileLocation(const char* file, int line) { const std::string file_name(file == NULL ? kUnknownFile : file); if (line < 0) { return file_name + ":"; } #ifdef _MSC_VER return file_name + "(" + StreamableToString(line) + "):"; #else return file_name + ":" + StreamableToString(line) + ":"; #endif // _MSC_VER } // Formats a file location for compiler-independent XML output. // Although this function is not platform dependent, we put it next to // FormatFileLocation in order to contrast the two functions. // Note that FormatCompilerIndependentFileLocation() does NOT append colon // to the file location it produces, unlike FormatFileLocation(). GTEST_API_ ::std::string FormatCompilerIndependentFileLocation( const char* file, int line) { const std::string file_name(file == NULL ? kUnknownFile : file); if (line < 0) return file_name; else return file_name + ":" + StreamableToString(line); } GTestLog::GTestLog(GTestLogSeverity severity, const char* file, int line) : severity_(severity) { const char* const marker = severity == GTEST_INFO ? "[ INFO ]" : severity == GTEST_WARNING ? "[WARNING]" : severity == GTEST_ERROR ? "[ ERROR ]" : "[ FATAL ]"; GetStream() << ::std::endl << marker << " " << FormatFileLocation(file, line).c_str() << ": "; } // Flushes the buffers and, if severity is GTEST_FATAL, aborts the program. GTestLog::~GTestLog() { GetStream() << ::std::endl; if (severity_ == GTEST_FATAL) { fflush(stderr); posix::Abort(); } } // Disable Microsoft deprecation warnings for POSIX functions called from // this class (creat, dup, dup2, and close) GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996) #if GTEST_HAS_STREAM_REDIRECTION // Object that captures an output stream (stdout/stderr). class CapturedStream { public: // The ctor redirects the stream to a temporary file. explicit CapturedStream(int fd) : fd_(fd), uncaptured_fd_(dup(fd)) { # if GTEST_OS_WINDOWS char temp_dir_path[MAX_PATH + 1] = { '\0' }; // NOLINT char temp_file_path[MAX_PATH + 1] = { '\0' }; // NOLINT ::GetTempPathA(sizeof(temp_dir_path), temp_dir_path); const UINT success = ::GetTempFileNameA(temp_dir_path, "gtest_redir", 0, // Generate unique file name. temp_file_path); GTEST_CHECK_(success != 0) << "Unable to create a temporary file in " << temp_dir_path; const int captured_fd = creat(temp_file_path, _S_IREAD | _S_IWRITE); GTEST_CHECK_(captured_fd != -1) << "Unable to open temporary file " << temp_file_path; filename_ = temp_file_path; # else // There's no guarantee that a test has write access to the current // directory, so we create the temporary file in the /tmp directory // instead. We use /tmp on most systems, and /sdcard on Android. // That's because Android doesn't have /tmp. # if GTEST_OS_LINUX_ANDROID // Note: Android applications are expected to call the framework's // Context.getExternalStorageDirectory() method through JNI to get // the location of the world-writable SD Card directory. However, // this requires a Context handle, which cannot be retrieved // globally from native code. Doing so also precludes running the // code as part of a regular standalone executable, which doesn't // run in a Dalvik process (e.g. when running it through 'adb shell'). // // The location /sdcard is directly accessible from native code // and is the only location (unofficially) supported by the Android // team. It's generally a symlink to the real SD Card mount point // which can be /mnt/sdcard, /mnt/sdcard0, /system/media/sdcard, or // other OEM-customized locations. Never rely on these, and always // use /sdcard. char name_template[] = "/sdcard/gtest_captured_stream.XXXXXX"; # else char name_template[] = "/tmp/captured_stream.XXXXXX"; # endif // GTEST_OS_LINUX_ANDROID const int captured_fd = mkstemp(name_template); filename_ = name_template; # endif // GTEST_OS_WINDOWS fflush(NULL); dup2(captured_fd, fd_); close(captured_fd); } ~CapturedStream() { remove(filename_.c_str()); } std::string GetCapturedString() { if (uncaptured_fd_ != -1) { // Restores the original stream. fflush(NULL); dup2(uncaptured_fd_, fd_); close(uncaptured_fd_); uncaptured_fd_ = -1; } FILE* const file = posix::FOpen(filename_.c_str(), "r"); const std::string content = ReadEntireFile(file); posix::FClose(file); return content; } private: // Reads the entire content of a file as an std::string. static std::string ReadEntireFile(FILE* file); // Returns the size (in bytes) of a file. static size_t GetFileSize(FILE* file); const int fd_; // A stream to capture. int uncaptured_fd_; // Name of the temporary file holding the stderr output. ::std::string filename_; GTEST_DISALLOW_COPY_AND_ASSIGN_(CapturedStream); }; // Returns the size (in bytes) of a file. size_t CapturedStream::GetFileSize(FILE* file) { fseek(file, 0, SEEK_END); return static_cast(ftell(file)); } // Reads the entire content of a file as a string. std::string CapturedStream::ReadEntireFile(FILE* file) { const size_t file_size = GetFileSize(file); char* const buffer = new char[file_size]; size_t bytes_last_read = 0; // # of bytes read in the last fread() size_t bytes_read = 0; // # of bytes read so far fseek(file, 0, SEEK_SET); // Keeps reading the file until we cannot read further or the // pre-determined file size is reached. do { bytes_last_read = fread(buffer+bytes_read, 1, file_size-bytes_read, file); bytes_read += bytes_last_read; } while (bytes_last_read > 0 && bytes_read < file_size); const std::string content(buffer, bytes_read); delete[] buffer; return content; } GTEST_DISABLE_MSC_WARNINGS_POP_() static CapturedStream* g_captured_stderr = NULL; static CapturedStream* g_captured_stdout = NULL; // Starts capturing an output stream (stdout/stderr). void CaptureStream(int fd, const char* stream_name, CapturedStream** stream) { if (*stream != NULL) { GTEST_LOG_(FATAL) << "Only one " << stream_name << " capturer can exist at a time."; } *stream = new CapturedStream(fd); } // Stops capturing the output stream and returns the captured string. std::string GetCapturedStream(CapturedStream** captured_stream) { const std::string content = (*captured_stream)->GetCapturedString(); delete *captured_stream; *captured_stream = NULL; return content; } // Starts capturing stdout. void CaptureStdout() { CaptureStream(kStdOutFileno, "stdout", &g_captured_stdout); } // Starts capturing stderr. void CaptureStderr() { CaptureStream(kStdErrFileno, "stderr", &g_captured_stderr); } // Stops capturing stdout and returns the captured string. std::string GetCapturedStdout() { return GetCapturedStream(&g_captured_stdout); } // Stops capturing stderr and returns the captured string. std::string GetCapturedStderr() { return GetCapturedStream(&g_captured_stderr); } #endif // GTEST_HAS_STREAM_REDIRECTION #if GTEST_HAS_DEATH_TEST // A copy of all command line arguments. Set by InitGoogleTest(). ::std::vector g_argvs; static const ::std::vector* g_injected_test_argvs = NULL; // Owned. void SetInjectableArgvs(const ::std::vector* argvs) { if (g_injected_test_argvs != argvs) delete g_injected_test_argvs; g_injected_test_argvs = argvs; } const ::std::vector& GetInjectableArgvs() { if (g_injected_test_argvs != NULL) { return *g_injected_test_argvs; } return g_argvs; } #endif // GTEST_HAS_DEATH_TEST #if GTEST_OS_WINDOWS_MOBILE namespace posix { void Abort() { DebugBreak(); TerminateProcess(GetCurrentProcess(), 1); } } // namespace posix #endif // GTEST_OS_WINDOWS_MOBILE // Returns the name of the environment variable corresponding to the // given flag. For example, FlagToEnvVar("foo") will return // "GTEST_FOO" in the open-source version. static std::string FlagToEnvVar(const char* flag) { const std::string full_flag = (Message() << GTEST_FLAG_PREFIX_ << flag).GetString(); Message env_var; for (size_t i = 0; i != full_flag.length(); i++) { env_var << ToUpper(full_flag.c_str()[i]); } return env_var.GetString(); } // Parses 'str' for a 32-bit signed integer. If successful, writes // the result to *value and returns true; otherwise leaves *value // unchanged and returns false. bool ParseInt32(const Message& src_text, const char* str, Int32* value) { // Parses the environment variable as a decimal integer. char* end = NULL; const long long_value = strtol(str, &end, 10); // NOLINT // Has strtol() consumed all characters in the string? if (*end != '\0') { // No - an invalid character was encountered. Message msg; msg << "WARNING: " << src_text << " is expected to be a 32-bit integer, but actually" << " has value \"" << str << "\".\n"; printf("%s", msg.GetString().c_str()); fflush(stdout); return false; } // Is the parsed value in the range of an Int32? const Int32 result = static_cast(long_value); if (long_value == LONG_MAX || long_value == LONG_MIN || // The parsed value overflows as a long. (strtol() returns // LONG_MAX or LONG_MIN when the input overflows.) result != long_value // The parsed value overflows as an Int32. ) { Message msg; msg << "WARNING: " << src_text << " is expected to be a 32-bit integer, but actually" << " has value " << str << ", which overflows.\n"; printf("%s", msg.GetString().c_str()); fflush(stdout); return false; } *value = result; return true; } // Reads and returns the Boolean environment variable corresponding to // the given flag; if it's not set, returns default_value. // // The value is considered true iff it's not "0". bool BoolFromGTestEnv(const char* flag, bool default_value) { const std::string env_var = FlagToEnvVar(flag); const char* const string_value = posix::GetEnv(env_var.c_str()); return string_value == NULL ? default_value : strcmp(string_value, "0") != 0; } // Reads and returns a 32-bit integer stored in the environment // variable corresponding to the given flag; if it isn't set or // doesn't represent a valid 32-bit integer, returns default_value. Int32 Int32FromGTestEnv(const char* flag, Int32 default_value) { const std::string env_var = FlagToEnvVar(flag); const char* const string_value = posix::GetEnv(env_var.c_str()); if (string_value == NULL) { // The environment variable is not set. return default_value; } Int32 result = default_value; if (!ParseInt32(Message() << "Environment variable " << env_var, string_value, &result)) { printf("The default value %s is used.\n", (Message() << default_value).GetString().c_str()); fflush(stdout); return default_value; } return result; } // Reads and returns the string environment variable corresponding to // the given flag; if it's not set, returns default_value. const char* StringFromGTestEnv(const char* flag, const char* default_value) { const std::string env_var = FlagToEnvVar(flag); const char* const value = posix::GetEnv(env_var.c_str()); return value == NULL ? default_value : value; } } // namespace internal } // namespace testing // Copyright 2007, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // Google Test - The Google C++ Testing Framework // // This file implements a universal value printer that can print a // value of any type T: // // void ::testing::internal::UniversalPrinter::Print(value, ostream_ptr); // // It uses the << operator when possible, and prints the bytes in the // object otherwise. A user can override its behavior for a class // type Foo by defining either operator<<(::std::ostream&, const Foo&) // or void PrintTo(const Foo&, ::std::ostream*) in the namespace that // defines Foo. #include #include #include #include // NOLINT #include namespace testing { namespace { using ::std::ostream; // Prints a segment of bytes in the given object. GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ void PrintByteSegmentInObjectTo(const unsigned char* obj_bytes, size_t start, size_t count, ostream* os) { char text[5] = ""; for (size_t i = 0; i != count; i++) { const size_t j = start + i; if (i != 0) { // Organizes the bytes into groups of 2 for easy parsing by // human. if ((j % 2) == 0) *os << ' '; else *os << '-'; } GTEST_SNPRINTF_(text, sizeof(text), "%02X", obj_bytes[j]); *os << text; } } // Prints the bytes in the given value to the given ostream. void PrintBytesInObjectToImpl(const unsigned char* obj_bytes, size_t count, ostream* os) { // Tells the user how big the object is. *os << count << "-byte object <"; const size_t kThreshold = 132; const size_t kChunkSize = 64; // If the object size is bigger than kThreshold, we'll have to omit // some details by printing only the first and the last kChunkSize // bytes. // TODO(wan): let the user control the threshold using a flag. if (count < kThreshold) { PrintByteSegmentInObjectTo(obj_bytes, 0, count, os); } else { PrintByteSegmentInObjectTo(obj_bytes, 0, kChunkSize, os); *os << " ... "; // Rounds up to 2-byte boundary. const size_t resume_pos = (count - kChunkSize + 1)/2*2; PrintByteSegmentInObjectTo(obj_bytes, resume_pos, count - resume_pos, os); } *os << ">"; } } // namespace namespace internal2 { // Delegates to PrintBytesInObjectToImpl() to print the bytes in the // given object. The delegation simplifies the implementation, which // uses the << operator and thus is easier done outside of the // ::testing::internal namespace, which contains a << operator that // sometimes conflicts with the one in STL. void PrintBytesInObjectTo(const unsigned char* obj_bytes, size_t count, ostream* os) { PrintBytesInObjectToImpl(obj_bytes, count, os); } } // namespace internal2 namespace internal { // Depending on the value of a char (or wchar_t), we print it in one // of three formats: // - as is if it's a printable ASCII (e.g. 'a', '2', ' '), // - as a hexidecimal escape sequence (e.g. '\x7F'), or // - as a special escape sequence (e.g. '\r', '\n'). enum CharFormat { kAsIs, kHexEscape, kSpecialEscape }; // Returns true if c is a printable ASCII character. We test the // value of c directly instead of calling isprint(), which is buggy on // Windows Mobile. inline bool IsPrintableAscii(wchar_t c) { return 0x20 <= c && c <= 0x7E; } // Prints a wide or narrow char c as a character literal without the // quotes, escaping it when necessary; returns how c was formatted. // The template argument UnsignedChar is the unsigned version of Char, // which is the type of c. template static CharFormat PrintAsCharLiteralTo(Char c, ostream* os) { switch (static_cast(c)) { case L'\0': *os << "\\0"; break; case L'\'': *os << "\\'"; break; case L'\\': *os << "\\\\"; break; case L'\a': *os << "\\a"; break; case L'\b': *os << "\\b"; break; case L'\f': *os << "\\f"; break; case L'\n': *os << "\\n"; break; case L'\r': *os << "\\r"; break; case L'\t': *os << "\\t"; break; case L'\v': *os << "\\v"; break; default: if (IsPrintableAscii(c)) { *os << static_cast(c); return kAsIs; } else { *os << "\\x" + String::FormatHexInt(static_cast(c)); return kHexEscape; } } return kSpecialEscape; } // Prints a wchar_t c as if it's part of a string literal, escaping it when // necessary; returns how c was formatted. static CharFormat PrintAsStringLiteralTo(wchar_t c, ostream* os) { switch (c) { case L'\'': *os << "'"; return kAsIs; case L'"': *os << "\\\""; return kSpecialEscape; default: return PrintAsCharLiteralTo(c, os); } } // Prints a char c as if it's part of a string literal, escaping it when // necessary; returns how c was formatted. static CharFormat PrintAsStringLiteralTo(char c, ostream* os) { return PrintAsStringLiteralTo( static_cast(static_cast(c)), os); } // Prints a wide or narrow character c and its code. '\0' is printed // as "'\\0'", other unprintable characters are also properly escaped // using the standard C++ escape sequence. The template argument // UnsignedChar is the unsigned version of Char, which is the type of c. template void PrintCharAndCodeTo(Char c, ostream* os) { // First, print c as a literal in the most readable form we can find. *os << ((sizeof(c) > 1) ? "L'" : "'"); const CharFormat format = PrintAsCharLiteralTo(c, os); *os << "'"; // To aid user debugging, we also print c's code in decimal, unless // it's 0 (in which case c was printed as '\\0', making the code // obvious). if (c == 0) return; *os << " (" << static_cast(c); // For more convenience, we print c's code again in hexidecimal, // unless c was already printed in the form '\x##' or the code is in // [1, 9]. if (format == kHexEscape || (1 <= c && c <= 9)) { // Do nothing. } else { *os << ", 0x" << String::FormatHexInt(static_cast(c)); } *os << ")"; } void PrintTo(unsigned char c, ::std::ostream* os) { PrintCharAndCodeTo(c, os); } void PrintTo(signed char c, ::std::ostream* os) { PrintCharAndCodeTo(c, os); } // Prints a wchar_t as a symbol if it is printable or as its internal // code otherwise and also as its code. L'\0' is printed as "L'\\0'". void PrintTo(wchar_t wc, ostream* os) { PrintCharAndCodeTo(wc, os); } // Prints the given array of characters to the ostream. CharType must be either // char or wchar_t. // The array starts at begin, the length is len, it may include '\0' characters // and may not be NUL-terminated. template GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ static void PrintCharsAsStringTo( const CharType* begin, size_t len, ostream* os) { const char* const kQuoteBegin = sizeof(CharType) == 1 ? "\"" : "L\""; *os << kQuoteBegin; bool is_previous_hex = false; for (size_t index = 0; index < len; ++index) { const CharType cur = begin[index]; if (is_previous_hex && IsXDigit(cur)) { // Previous character is of '\x..' form and this character can be // interpreted as another hexadecimal digit in its number. Break string to // disambiguate. *os << "\" " << kQuoteBegin; } is_previous_hex = PrintAsStringLiteralTo(cur, os) == kHexEscape; } *os << "\""; } // Prints a (const) char/wchar_t array of 'len' elements, starting at address // 'begin'. CharType must be either char or wchar_t. template GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ static void UniversalPrintCharArray( const CharType* begin, size_t len, ostream* os) { // The code // const char kFoo[] = "foo"; // generates an array of 4, not 3, elements, with the last one being '\0'. // // Therefore when printing a char array, we don't print the last element if // it's '\0', such that the output matches the string literal as it's // written in the source code. if (len > 0 && begin[len - 1] == '\0') { PrintCharsAsStringTo(begin, len - 1, os); return; } // If, however, the last element in the array is not '\0', e.g. // const char kFoo[] = { 'f', 'o', 'o' }; // we must print the entire array. We also print a message to indicate // that the array is not NUL-terminated. PrintCharsAsStringTo(begin, len, os); *os << " (no terminating NUL)"; } // Prints a (const) char array of 'len' elements, starting at address 'begin'. void UniversalPrintArray(const char* begin, size_t len, ostream* os) { UniversalPrintCharArray(begin, len, os); } // Prints a (const) wchar_t array of 'len' elements, starting at address // 'begin'. void UniversalPrintArray(const wchar_t* begin, size_t len, ostream* os) { UniversalPrintCharArray(begin, len, os); } // Prints the given C string to the ostream. void PrintTo(const char* s, ostream* os) { if (s == NULL) { *os << "NULL"; } else { *os << ImplicitCast_(s) << " pointing to "; PrintCharsAsStringTo(s, strlen(s), os); } } // MSVC compiler can be configured to define whar_t as a typedef // of unsigned short. Defining an overload for const wchar_t* in that case // would cause pointers to unsigned shorts be printed as wide strings, // possibly accessing more memory than intended and causing invalid // memory accesses. MSVC defines _NATIVE_WCHAR_T_DEFINED symbol when // wchar_t is implemented as a native type. #if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED) // Prints the given wide C string to the ostream. void PrintTo(const wchar_t* s, ostream* os) { if (s == NULL) { *os << "NULL"; } else { *os << ImplicitCast_(s) << " pointing to "; PrintCharsAsStringTo(s, std::wcslen(s), os); } } #endif // wchar_t is native // Prints a ::string object. #if GTEST_HAS_GLOBAL_STRING void PrintStringTo(const ::string& s, ostream* os) { PrintCharsAsStringTo(s.data(), s.size(), os); } #endif // GTEST_HAS_GLOBAL_STRING void PrintStringTo(const ::std::string& s, ostream* os) { PrintCharsAsStringTo(s.data(), s.size(), os); } // Prints a ::wstring object. #if GTEST_HAS_GLOBAL_WSTRING void PrintWideStringTo(const ::wstring& s, ostream* os) { PrintCharsAsStringTo(s.data(), s.size(), os); } #endif // GTEST_HAS_GLOBAL_WSTRING #if GTEST_HAS_STD_WSTRING void PrintWideStringTo(const ::std::wstring& s, ostream* os) { PrintCharsAsStringTo(s.data(), s.size(), os); } #endif // GTEST_HAS_STD_WSTRING } // namespace internal } // namespace testing // Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: mheule@google.com (Markus Heule) // // The Google C++ Testing Framework (Google Test) // Indicates that this translation unit is part of Google Test's // implementation. It must come before gtest-internal-inl.h is // included, or there will be a compiler error. This trick exists to // prevent the accidental inclusion of gtest-internal-inl.h in the // user's code. #define GTEST_IMPLEMENTATION_ 1 #undef GTEST_IMPLEMENTATION_ namespace testing { using internal::GetUnitTestImpl; // Gets the summary of the failure message by omitting the stack trace // in it. std::string TestPartResult::ExtractSummary(const char* message) { const char* const stack_trace = strstr(message, internal::kStackTraceMarker); return stack_trace == NULL ? message : std::string(message, stack_trace); } // Prints a TestPartResult object. std::ostream& operator<<(std::ostream& os, const TestPartResult& result) { return os << result.file_name() << ":" << result.line_number() << ": " << (result.type() == TestPartResult::kSuccess ? "Success" : result.type() == TestPartResult::kFatalFailure ? "Fatal failure" : "Non-fatal failure") << ":\n" << result.message() << std::endl; } // Appends a TestPartResult to the array. void TestPartResultArray::Append(const TestPartResult& result) { array_.push_back(result); } // Returns the TestPartResult at the given index (0-based). const TestPartResult& TestPartResultArray::GetTestPartResult(int index) const { if (index < 0 || index >= size()) { printf("\nInvalid index (%d) into TestPartResultArray.\n", index); internal::posix::Abort(); } return array_[index]; } // Returns the number of TestPartResult objects in the array. int TestPartResultArray::size() const { return static_cast(array_.size()); } namespace internal { HasNewFatalFailureHelper::HasNewFatalFailureHelper() : has_new_fatal_failure_(false), original_reporter_(GetUnitTestImpl()-> GetTestPartResultReporterForCurrentThread()) { GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(this); } HasNewFatalFailureHelper::~HasNewFatalFailureHelper() { GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread( original_reporter_); } void HasNewFatalFailureHelper::ReportTestPartResult( const TestPartResult& result) { if (result.fatally_failed()) has_new_fatal_failure_ = true; original_reporter_->ReportTestPartResult(result); } } // namespace internal } // namespace testing // Copyright 2008 Google Inc. // All Rights Reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) namespace testing { namespace internal { #if GTEST_HAS_TYPED_TEST_P // Skips to the first non-space char in str. Returns an empty string if str // contains only whitespace characters. static const char* SkipSpaces(const char* str) { while (IsSpace(*str)) str++; return str; } static std::vector SplitIntoTestNames(const char* src) { std::vector name_vec; src = SkipSpaces(src); for (; src != NULL; src = SkipComma(src)) { name_vec.push_back(StripTrailingSpaces(GetPrefixUntilComma(src))); } return name_vec; } // Verifies that registered_tests match the test names in // defined_test_names_; returns registered_tests if successful, or // aborts the program otherwise. const char* TypedTestCasePState::VerifyRegisteredTestNames( const char* file, int line, const char* registered_tests) { typedef ::std::set::const_iterator DefinedTestIter; registered_ = true; std::vector name_vec = SplitIntoTestNames(registered_tests); Message errors; std::set tests; for (std::vector::const_iterator name_it = name_vec.begin(); name_it != name_vec.end(); ++name_it) { const std::string& name = *name_it; if (tests.count(name) != 0) { errors << "Test " << name << " is listed more than once.\n"; continue; } bool found = false; for (DefinedTestIter it = defined_test_names_.begin(); it != defined_test_names_.end(); ++it) { if (name == *it) { found = true; break; } } if (found) { tests.insert(name); } else { errors << "No test named " << name << " can be found in this test case.\n"; } } for (DefinedTestIter it = defined_test_names_.begin(); it != defined_test_names_.end(); ++it) { if (tests.count(*it) == 0) { errors << "You forgot to list test " << *it << ".\n"; } } const std::string& errors_str = errors.GetString(); if (errors_str != "") { fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(), errors_str.c_str()); fflush(stderr); posix::Abort(); } return registered_tests; } #endif // GTEST_HAS_TYPED_TEST_P } // namespace internal } // namespace testing rocksdb-4.5.1/third-party/gtest-1.7.0/fused-src/gtest/gtest.h000066400000000000000000032035301267530634000235420ustar00rootroot00000000000000// Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // // The Google C++ Testing Framework (Google Test) // // This header file defines the public API for Google Test. It should be // included by any test program that uses Google Test. // // IMPORTANT NOTE: Due to limitation of the C++ language, we have to // leave some internal implementation details in this header file. // They are clearly marked by comments like this: // // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. // // Such code is NOT meant to be used by a user directly, and is subject // to CHANGE WITHOUT NOTICE. Therefore DO NOT DEPEND ON IT in a user // program! // // Acknowledgment: Google Test borrowed the idea of automatic test // registration from Barthelemy Dagenais' (barthelemy@prologique.com) // easyUnit framework. #ifndef GTEST_INCLUDE_GTEST_GTEST_H_ #define GTEST_INCLUDE_GTEST_GTEST_H_ #include #include #include // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee) // // The Google C++ Testing Framework (Google Test) // // This header file declares functions and macros used internally by // Google Test. They are subject to change without notice. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_ // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: wan@google.com (Zhanyong Wan) // // Low-level types and utilities for porting Google Test to various // platforms. All macros ending with _ and symbols defined in an // internal namespace are subject to change without notice. Code // outside Google Test MUST NOT USE THEM DIRECTLY. Macros that don't // end with _ are part of Google Test's public API and can be used by // code outside Google Test. // // This file is fundamental to Google Test. All other Google Test source // files are expected to #include this. Therefore, it cannot #include // any other Google Test header. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_ // Environment-describing macros // ----------------------------- // // Google Test can be used in many different environments. Macros in // this section tell Google Test what kind of environment it is being // used in, such that Google Test can provide environment-specific // features and implementations. // // Google Test tries to automatically detect the properties of its // environment, so users usually don't need to worry about these // macros. However, the automatic detection is not perfect. // Sometimes it's necessary for a user to define some of the following // macros in the build script to override Google Test's decisions. // // If the user doesn't define a macro in the list, Google Test will // provide a default definition. After this header is #included, all // macros in this list will be defined to either 1 or 0. // // Notes to maintainers: // - Each macro here is a user-tweakable knob; do not grow the list // lightly. // - Use #if to key off these macros. Don't use #ifdef or "#if // defined(...)", which will not work as these macros are ALWAYS // defined. // // GTEST_HAS_CLONE - Define it to 1/0 to indicate that clone(2) // is/isn't available. // GTEST_HAS_EXCEPTIONS - Define it to 1/0 to indicate that exceptions // are enabled. // GTEST_HAS_GLOBAL_STRING - Define it to 1/0 to indicate that ::string // is/isn't available (some systems define // ::string, which is different to std::string). // GTEST_HAS_GLOBAL_WSTRING - Define it to 1/0 to indicate that ::string // is/isn't available (some systems define // ::wstring, which is different to std::wstring). // GTEST_HAS_POSIX_RE - Define it to 1/0 to indicate that POSIX regular // expressions are/aren't available. // GTEST_HAS_PTHREAD - Define it to 1/0 to indicate that // is/isn't available. // GTEST_HAS_RTTI - Define it to 1/0 to indicate that RTTI is/isn't // enabled. // GTEST_HAS_STD_WSTRING - Define it to 1/0 to indicate that // std::wstring does/doesn't work (Google Test can // be used where std::wstring is unavailable). // GTEST_HAS_TR1_TUPLE - Define it to 1/0 to indicate tr1::tuple // is/isn't available. // GTEST_HAS_SEH - Define it to 1/0 to indicate whether the // compiler supports Microsoft's "Structured // Exception Handling". // GTEST_HAS_STREAM_REDIRECTION // - Define it to 1/0 to indicate whether the // platform supports I/O stream redirection using // dup() and dup2(). // GTEST_USE_OWN_TR1_TUPLE - Define it to 1/0 to indicate whether Google // Test's own tr1 tuple implementation should be // used. Unused when the user sets // GTEST_HAS_TR1_TUPLE to 0. // GTEST_LANG_CXX11 - Define it to 1/0 to indicate that Google Test // is building in C++11/C++98 mode. // GTEST_LINKED_AS_SHARED_LIBRARY // - Define to 1 when compiling tests that use // Google Test as a shared library (known as // DLL on Windows). // GTEST_CREATE_SHARED_LIBRARY // - Define to 1 when compiling Google Test itself // as a shared library. // Platform-indicating macros // -------------------------- // // Macros indicating the platform on which Google Test is being used // (a macro is defined to 1 if compiled on the given platform; // otherwise UNDEFINED -- it's never defined to 0.). Google Test // defines these macros automatically. Code outside Google Test MUST // NOT define them. // // GTEST_OS_AIX - IBM AIX // GTEST_OS_CYGWIN - Cygwin // GTEST_OS_FREEBSD - FreeBSD // GTEST_OS_HPUX - HP-UX // GTEST_OS_LINUX - Linux // GTEST_OS_LINUX_ANDROID - Google Android // GTEST_OS_MAC - Mac OS X // GTEST_OS_IOS - iOS // GTEST_OS_NACL - Google Native Client (NaCl) // GTEST_OS_OPENBSD - OpenBSD // GTEST_OS_QNX - QNX // GTEST_OS_SOLARIS - Sun Solaris // GTEST_OS_SYMBIAN - Symbian // GTEST_OS_WINDOWS - Windows (Desktop, MinGW, or Mobile) // GTEST_OS_WINDOWS_DESKTOP - Windows Desktop // GTEST_OS_WINDOWS_MINGW - MinGW // GTEST_OS_WINDOWS_MOBILE - Windows Mobile // GTEST_OS_WINDOWS_PHONE - Windows Phone // GTEST_OS_WINDOWS_RT - Windows Store App/WinRT // GTEST_OS_ZOS - z/OS // // Among the platforms, Cygwin, Linux, Max OS X, and Windows have the // most stable support. Since core members of the Google Test project // don't have access to other platforms, support for them may be less // stable. If you notice any problems on your platform, please notify // googletestframework@googlegroups.com (patches for fixing them are // even more welcome!). // // It is possible that none of the GTEST_OS_* macros are defined. // Feature-indicating macros // ------------------------- // // Macros indicating which Google Test features are available (a macro // is defined to 1 if the corresponding feature is supported; // otherwise UNDEFINED -- it's never defined to 0.). Google Test // defines these macros automatically. Code outside Google Test MUST // NOT define them. // // These macros are public so that portable tests can be written. // Such tests typically surround code using a feature with an #if // which controls that code. For example: // // #if GTEST_HAS_DEATH_TEST // EXPECT_DEATH(DoSomethingDeadly()); // #endif // // GTEST_HAS_COMBINE - the Combine() function (for value-parameterized // tests) // GTEST_HAS_DEATH_TEST - death tests // GTEST_HAS_PARAM_TEST - value-parameterized tests // GTEST_HAS_TYPED_TEST - typed tests // GTEST_HAS_TYPED_TEST_P - type-parameterized tests // GTEST_IS_THREADSAFE - Google Test is thread-safe. // GTEST_USES_POSIX_RE - enhanced POSIX regex is used. Do not confuse with // GTEST_HAS_POSIX_RE (see above) which users can // define themselves. // GTEST_USES_SIMPLE_RE - our own simple regex is used; // the above two are mutually exclusive. // GTEST_CAN_COMPARE_NULL - accepts untyped NULL in EXPECT_EQ(). // Misc public macros // ------------------ // // GTEST_FLAG(flag_name) - references the variable corresponding to // the given Google Test flag. // Internal utilities // ------------------ // // The following macros and utilities are for Google Test's INTERNAL // use only. Code outside Google Test MUST NOT USE THEM DIRECTLY. // // Macros for basic C++ coding: // GTEST_AMBIGUOUS_ELSE_BLOCKER_ - for disabling a gcc warning. // GTEST_ATTRIBUTE_UNUSED_ - declares that a class' instances or a // variable don't have to be used. // GTEST_DISALLOW_ASSIGN_ - disables operator=. // GTEST_DISALLOW_COPY_AND_ASSIGN_ - disables copy ctor and operator=. // GTEST_MUST_USE_RESULT_ - declares that a function's result must be used. // GTEST_INTENTIONAL_CONST_COND_PUSH_ - start code section where MSVC C4127 is // suppressed (constant conditional). // GTEST_INTENTIONAL_CONST_COND_POP_ - finish code section where MSVC C4127 // is suppressed. // // C++11 feature wrappers: // // testing::internal::move - portability wrapper for std::move. // // Synchronization: // Mutex, MutexLock, ThreadLocal, GetThreadCount() // - synchronization primitives. // // Template meta programming: // is_pointer - as in TR1; needed on Symbian and IBM XL C/C++ only. // IteratorTraits - partial implementation of std::iterator_traits, which // is not available in libCstd when compiled with Sun C++. // // Smart pointers: // scoped_ptr - as in TR2. // // Regular expressions: // RE - a simple regular expression class using the POSIX // Extended Regular Expression syntax on UNIX-like // platforms, or a reduced regular exception syntax on // other platforms, including Windows. // // Logging: // GTEST_LOG_() - logs messages at the specified severity level. // LogToStderr() - directs all log messages to stderr. // FlushInfoLog() - flushes informational log messages. // // Stdout and stderr capturing: // CaptureStdout() - starts capturing stdout. // GetCapturedStdout() - stops capturing stdout and returns the captured // string. // CaptureStderr() - starts capturing stderr. // GetCapturedStderr() - stops capturing stderr and returns the captured // string. // // Integer types: // TypeWithSize - maps an integer to a int type. // Int32, UInt32, Int64, UInt64, TimeInMillis // - integers of known sizes. // BiggestInt - the biggest signed integer type. // // Command-line utilities: // GTEST_DECLARE_*() - declares a flag. // GTEST_DEFINE_*() - defines a flag. // GetInjectableArgvs() - returns the command line as a vector of strings. // // Environment variable utilities: // GetEnv() - gets the value of an environment variable. // BoolFromGTestEnv() - parses a bool environment variable. // Int32FromGTestEnv() - parses an Int32 environment variable. // StringFromGTestEnv() - parses a string environment variable. #include // for isspace, etc #include // for ptrdiff_t #include #include #include #ifndef _WIN32_WCE # include # include #endif // !_WIN32_WCE #if defined __APPLE__ # include # include #endif #include // NOLINT #include // NOLINT #include // NOLINT #include // NOLINT #include #define GTEST_DEV_EMAIL_ "googletestframework@@googlegroups.com" #define GTEST_FLAG_PREFIX_ "gtest_" #define GTEST_FLAG_PREFIX_DASH_ "gtest-" #define GTEST_FLAG_PREFIX_UPPER_ "GTEST_" #define GTEST_NAME_ "Google Test" #define GTEST_PROJECT_URL_ "http://code.google.com/p/googletest/" // Determines the version of gcc that is used to compile this. #ifdef __GNUC__ // 40302 means version 4.3.2. # define GTEST_GCC_VER_ \ (__GNUC__*10000 + __GNUC_MINOR__*100 + __GNUC_PATCHLEVEL__) #endif // __GNUC__ // Determines the platform on which Google Test is compiled. #ifdef __CYGWIN__ # define GTEST_OS_CYGWIN 1 #elif defined __SYMBIAN32__ # define GTEST_OS_SYMBIAN 1 #elif defined _WIN32 # define GTEST_OS_WINDOWS 1 # ifdef _WIN32_WCE # define GTEST_OS_WINDOWS_MOBILE 1 # elif defined(__MINGW__) || defined(__MINGW32__) # define GTEST_OS_WINDOWS_MINGW 1 # elif defined(WINAPI_FAMILY) # include # if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP) # define GTEST_OS_WINDOWS_DESKTOP 1 # elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_PHONE_APP) # define GTEST_OS_WINDOWS_PHONE 1 # elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_APP) # define GTEST_OS_WINDOWS_RT 1 # else // WINAPI_FAMILY defined but no known partition matched. // Default to desktop. # define GTEST_OS_WINDOWS_DESKTOP 1 # endif # else # define GTEST_OS_WINDOWS_DESKTOP 1 # endif // _WIN32_WCE #elif defined __APPLE__ # define GTEST_OS_MAC 1 # if TARGET_OS_IPHONE # define GTEST_OS_IOS 1 # endif #elif defined __FreeBSD__ # define GTEST_OS_FREEBSD 1 #elif defined __linux__ # define GTEST_OS_LINUX 1 # if defined __ANDROID__ # define GTEST_OS_LINUX_ANDROID 1 # endif #elif defined __MVS__ # define GTEST_OS_ZOS 1 #elif defined(__sun) && defined(__SVR4) # define GTEST_OS_SOLARIS 1 #elif defined(_AIX) # define GTEST_OS_AIX 1 #elif defined(__hpux) # define GTEST_OS_HPUX 1 #elif defined __native_client__ # define GTEST_OS_NACL 1 #elif defined __OpenBSD__ # define GTEST_OS_OPENBSD 1 #elif defined __QNX__ # define GTEST_OS_QNX 1 #endif // __CYGWIN__ // Macros for disabling Microsoft Visual C++ warnings. // // GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 4385) // /* code that triggers warnings C4800 and C4385 */ // GTEST_DISABLE_MSC_WARNINGS_POP_() #if _MSC_VER >= 1500 # define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings) \ __pragma(warning(push)) \ __pragma(warning(disable: warnings)) # define GTEST_DISABLE_MSC_WARNINGS_POP_() \ __pragma(warning(pop)) #else // Older versions of MSVC don't have __pragma. # define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings) # define GTEST_DISABLE_MSC_WARNINGS_POP_() #endif #ifndef GTEST_LANG_CXX11 // gcc and clang define __GXX_EXPERIMENTAL_CXX0X__ when // -std={c,gnu}++{0x,11} is passed. The C++11 standard specifies a // value for __cplusplus, and recent versions of clang, gcc, and // probably other compilers set that too in C++11 mode. # if __GXX_EXPERIMENTAL_CXX0X__ || __cplusplus >= 201103L // Compiling in at least C++11 mode. # define GTEST_LANG_CXX11 1 # else # define GTEST_LANG_CXX11 0 # endif #endif // Distinct from C++11 language support, some environments don't provide // proper C++11 library support. Notably, it's possible to build in // C++11 mode when targeting Mac OS X 10.6, which has an old libstdc++ // with no C++11 support. // // libstdc++ has sufficient C++11 support as of GCC 4.6.0, __GLIBCXX__ // 20110325, but maintenance releases in the 4.4 and 4.5 series followed // this date, so check for those versions by their date stamps. // https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html#abi.versioning #if GTEST_LANG_CXX11 && \ (!defined(__GLIBCXX__) || ( \ __GLIBCXX__ >= 20110325ul && /* GCC >= 4.6.0 */ \ /* Blacklist of patch releases of older branches: */ \ __GLIBCXX__ != 20110416ul && /* GCC 4.4.6 */ \ __GLIBCXX__ != 20120313ul && /* GCC 4.4.7 */ \ __GLIBCXX__ != 20110428ul && /* GCC 4.5.3 */ \ __GLIBCXX__ != 20120702ul)) /* GCC 4.5.4 */ # define GTEST_STDLIB_CXX11 1 #endif // Only use C++11 library features if the library provides them. #if GTEST_STDLIB_CXX11 # define GTEST_HAS_STD_BEGIN_AND_END_ 1 # define GTEST_HAS_STD_FORWARD_LIST_ 1 # define GTEST_HAS_STD_FUNCTION_ 1 # define GTEST_HAS_STD_INITIALIZER_LIST_ 1 # define GTEST_HAS_STD_MOVE_ 1 # define GTEST_HAS_STD_UNIQUE_PTR_ 1 #endif // C++11 specifies that provides std::tuple. // Some platforms still might not have it, however. #if GTEST_LANG_CXX11 # define GTEST_HAS_STD_TUPLE_ 1 # if defined(__clang__) // Inspired by http://clang.llvm.org/docs/LanguageExtensions.html#__has_include # if defined(__has_include) && !__has_include() # undef GTEST_HAS_STD_TUPLE_ # endif # elif defined(_MSC_VER) // Inspired by boost/config/stdlib/dinkumware.hpp # if defined(_CPPLIB_VER) && _CPPLIB_VER < 520 # undef GTEST_HAS_STD_TUPLE_ # endif # elif defined(__GLIBCXX__) // Inspired by boost/config/stdlib/libstdcpp3.hpp, // http://gcc.gnu.org/gcc-4.2/changes.html and // http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.200x # if __GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 2) # undef GTEST_HAS_STD_TUPLE_ # endif # endif #endif // Brings in definitions for functions used in the testing::internal::posix // namespace (read, write, close, chdir, isatty, stat). We do not currently // use them on Windows Mobile. #if GTEST_OS_WINDOWS # if !GTEST_OS_WINDOWS_MOBILE # include # include # endif // In order to avoid having to include , use forward declaration // assuming CRITICAL_SECTION is a typedef of _RTL_CRITICAL_SECTION. // This assumption is verified by // WindowsTypesTest.CRITICAL_SECTIONIs_RTL_CRITICAL_SECTION. struct _RTL_CRITICAL_SECTION; #else // This assumes that non-Windows OSes provide unistd.h. For OSes where this // is not the case, we need to include headers that provide the functions // mentioned above. # include # include #endif // GTEST_OS_WINDOWS #if GTEST_OS_LINUX_ANDROID // Used to define __ANDROID_API__ matching the target NDK API level. # include // NOLINT #endif // Defines this to true iff Google Test can use POSIX regular expressions. #ifndef GTEST_HAS_POSIX_RE # if GTEST_OS_LINUX_ANDROID // On Android, is only available starting with Gingerbread. # define GTEST_HAS_POSIX_RE (__ANDROID_API__ >= 9) # else # define GTEST_HAS_POSIX_RE (!GTEST_OS_WINDOWS) # endif #endif #if GTEST_HAS_POSIX_RE // On some platforms, needs someone to define size_t, and // won't compile otherwise. We can #include it here as we already // included , which is guaranteed to define size_t through // . # include // NOLINT # define GTEST_USES_POSIX_RE 1 #elif GTEST_OS_WINDOWS // is not available on Windows. Use our own simple regex // implementation instead. # define GTEST_USES_SIMPLE_RE 1 #else // may not be available on this platform. Use our own // simple regex implementation instead. # define GTEST_USES_SIMPLE_RE 1 #endif // GTEST_HAS_POSIX_RE #ifndef GTEST_HAS_EXCEPTIONS // The user didn't tell us whether exceptions are enabled, so we need // to figure it out. # if defined(_MSC_VER) || defined(__BORLANDC__) // MSVC's and C++Builder's implementations of the STL use the _HAS_EXCEPTIONS // macro to enable exceptions, so we'll do the same. // Assumes that exceptions are enabled by default. # ifndef _HAS_EXCEPTIONS # define _HAS_EXCEPTIONS 1 # endif // _HAS_EXCEPTIONS # define GTEST_HAS_EXCEPTIONS _HAS_EXCEPTIONS # elif defined(__clang__) // clang defines __EXCEPTIONS iff exceptions are enabled before clang 220714, // but iff cleanups are enabled after that. In Obj-C++ files, there can be // cleanups for ObjC exceptions which also need cleanups, even if C++ exceptions // are disabled. clang has __has_feature(cxx_exceptions) which checks for C++ // exceptions starting at clang r206352, but which checked for cleanups prior to // that. To reliably check for C++ exception availability with clang, check for // __EXCEPTIONS && __has_feature(cxx_exceptions). # define GTEST_HAS_EXCEPTIONS (__EXCEPTIONS && __has_feature(cxx_exceptions)) # elif defined(__GNUC__) && __EXCEPTIONS // gcc defines __EXCEPTIONS to 1 iff exceptions are enabled. # define GTEST_HAS_EXCEPTIONS 1 # elif defined(__SUNPRO_CC) // Sun Pro CC supports exceptions. However, there is no compile-time way of // detecting whether they are enabled or not. Therefore, we assume that // they are enabled unless the user tells us otherwise. # define GTEST_HAS_EXCEPTIONS 1 # elif defined(__IBMCPP__) && __EXCEPTIONS // xlC defines __EXCEPTIONS to 1 iff exceptions are enabled. # define GTEST_HAS_EXCEPTIONS 1 # elif defined(__HP_aCC) // Exception handling is in effect by default in HP aCC compiler. It has to // be turned of by +noeh compiler option if desired. # define GTEST_HAS_EXCEPTIONS 1 # else // For other compilers, we assume exceptions are disabled to be // conservative. # define GTEST_HAS_EXCEPTIONS 0 # endif // defined(_MSC_VER) || defined(__BORLANDC__) #endif // GTEST_HAS_EXCEPTIONS #if !defined(GTEST_HAS_STD_STRING) // Even though we don't use this macro any longer, we keep it in case // some clients still depend on it. # define GTEST_HAS_STD_STRING 1 #elif !GTEST_HAS_STD_STRING // The user told us that ::std::string isn't available. # error "Google Test cannot be used where ::std::string isn't available." #endif // !defined(GTEST_HAS_STD_STRING) #ifndef GTEST_HAS_GLOBAL_STRING // The user didn't tell us whether ::string is available, so we need // to figure it out. # define GTEST_HAS_GLOBAL_STRING 0 #endif // GTEST_HAS_GLOBAL_STRING #ifndef GTEST_HAS_STD_WSTRING // The user didn't tell us whether ::std::wstring is available, so we need // to figure it out. // TODO(wan@google.com): uses autoconf to detect whether ::std::wstring // is available. // Cygwin 1.7 and below doesn't support ::std::wstring. // Solaris' libc++ doesn't support it either. Android has // no support for it at least as recent as Froyo (2.2). # define GTEST_HAS_STD_WSTRING \ (!(GTEST_OS_LINUX_ANDROID || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS)) #endif // GTEST_HAS_STD_WSTRING #ifndef GTEST_HAS_GLOBAL_WSTRING // The user didn't tell us whether ::wstring is available, so we need // to figure it out. # define GTEST_HAS_GLOBAL_WSTRING \ (GTEST_HAS_STD_WSTRING && GTEST_HAS_GLOBAL_STRING) #endif // GTEST_HAS_GLOBAL_WSTRING // Determines whether RTTI is available. #ifndef GTEST_HAS_RTTI // The user didn't tell us whether RTTI is enabled, so we need to // figure it out. # ifdef _MSC_VER # ifdef _CPPRTTI // MSVC defines this macro iff RTTI is enabled. # define GTEST_HAS_RTTI 1 # else # define GTEST_HAS_RTTI 0 # endif // Starting with version 4.3.2, gcc defines __GXX_RTTI iff RTTI is enabled. # elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40302) # ifdef __GXX_RTTI // When building against STLport with the Android NDK and with // -frtti -fno-exceptions, the build fails at link time with undefined // references to __cxa_bad_typeid. Note sure if STL or toolchain bug, // so disable RTTI when detected. # if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR) && \ !defined(__EXCEPTIONS) # define GTEST_HAS_RTTI 0 # else # define GTEST_HAS_RTTI 1 # endif // GTEST_OS_LINUX_ANDROID && __STLPORT_MAJOR && !__EXCEPTIONS # else # define GTEST_HAS_RTTI 0 # endif // __GXX_RTTI // Clang defines __GXX_RTTI starting with version 3.0, but its manual recommends // using has_feature instead. has_feature(cxx_rtti) is supported since 2.7, the // first version with C++ support. # elif defined(__clang__) # define GTEST_HAS_RTTI __has_feature(cxx_rtti) // Starting with version 9.0 IBM Visual Age defines __RTTI_ALL__ to 1 if // both the typeid and dynamic_cast features are present. # elif defined(__IBMCPP__) && (__IBMCPP__ >= 900) # ifdef __RTTI_ALL__ # define GTEST_HAS_RTTI 1 # else # define GTEST_HAS_RTTI 0 # endif # else // For all other compilers, we assume RTTI is enabled. # define GTEST_HAS_RTTI 1 # endif // _MSC_VER #endif // GTEST_HAS_RTTI // It's this header's responsibility to #include when RTTI // is enabled. #if GTEST_HAS_RTTI # include #endif // Determines whether Google Test can use the pthreads library. #ifndef GTEST_HAS_PTHREAD // The user didn't tell us explicitly, so we make reasonable assumptions about // which platforms have pthreads support. // // To disable threading support in Google Test, add -DGTEST_HAS_PTHREAD=0 // to your compiler flags. # define GTEST_HAS_PTHREAD (GTEST_OS_LINUX || GTEST_OS_MAC || GTEST_OS_HPUX \ || GTEST_OS_QNX || GTEST_OS_FREEBSD || GTEST_OS_NACL) #endif // GTEST_HAS_PTHREAD #if GTEST_HAS_PTHREAD // gtest-port.h guarantees to #include when GTEST_HAS_PTHREAD is // true. # include // NOLINT // For timespec and nanosleep, used below. # include // NOLINT #endif // Determines whether Google Test can use tr1/tuple. You can define // this macro to 0 to prevent Google Test from using tuple (any // feature depending on tuple with be disabled in this mode). #ifndef GTEST_HAS_TR1_TUPLE # if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR) // STLport, provided with the Android NDK, has neither or . # define GTEST_HAS_TR1_TUPLE 0 # else // The user didn't tell us not to do it, so we assume it's OK. # define GTEST_HAS_TR1_TUPLE 1 # endif #endif // GTEST_HAS_TR1_TUPLE // Determines whether Google Test's own tr1 tuple implementation // should be used. #ifndef GTEST_USE_OWN_TR1_TUPLE // The user didn't tell us, so we need to figure it out. // We use our own TR1 tuple if we aren't sure the user has an // implementation of it already. At this time, libstdc++ 4.0.0+ and // MSVC 2010 are the only mainstream standard libraries that come // with a TR1 tuple implementation. NVIDIA's CUDA NVCC compiler // pretends to be GCC by defining __GNUC__ and friends, but cannot // compile GCC's tuple implementation. MSVC 2008 (9.0) provides TR1 // tuple in a 323 MB Feature Pack download, which we cannot assume the // user has. QNX's QCC compiler is a modified GCC but it doesn't // support TR1 tuple. libc++ only provides std::tuple, in C++11 mode, // and it can be used with some compilers that define __GNUC__. # if (defined(__GNUC__) && !defined(__CUDACC__) && (GTEST_GCC_VER_ >= 40000) \ && !GTEST_OS_QNX && !defined(_LIBCPP_VERSION)) || _MSC_VER >= 1600 # define GTEST_ENV_HAS_TR1_TUPLE_ 1 # endif // C++11 specifies that provides std::tuple. Use that if gtest is used // in C++11 mode and libstdc++ isn't very old (binaries targeting OS X 10.6 // can build with clang but need to use gcc4.2's libstdc++). # if GTEST_LANG_CXX11 && (!defined(__GLIBCXX__) || __GLIBCXX__ > 20110325) # define GTEST_ENV_HAS_STD_TUPLE_ 1 # endif # if GTEST_ENV_HAS_TR1_TUPLE_ || GTEST_ENV_HAS_STD_TUPLE_ # define GTEST_USE_OWN_TR1_TUPLE 0 # else # define GTEST_USE_OWN_TR1_TUPLE 1 # endif #endif // GTEST_USE_OWN_TR1_TUPLE // To avoid conditional compilation everywhere, we make it // gtest-port.h's responsibility to #include the header implementing // tuple. #if GTEST_HAS_STD_TUPLE_ # include // IWYU pragma: export # define GTEST_TUPLE_NAMESPACE_ ::std #endif // GTEST_HAS_STD_TUPLE_ // We include tr1::tuple even if std::tuple is available to define printers for // them. #if GTEST_HAS_TR1_TUPLE # ifndef GTEST_TUPLE_NAMESPACE_ # define GTEST_TUPLE_NAMESPACE_ ::std::tr1 # endif // GTEST_TUPLE_NAMESPACE_ # if GTEST_USE_OWN_TR1_TUPLE // This file was GENERATED by command: // pump.py gtest-tuple.h.pump // DO NOT EDIT BY HAND!!! // Copyright 2009 Google Inc. // All Rights Reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // Implements a subset of TR1 tuple needed by Google Test and Google Mock. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_ #include // For ::std::pair. // The compiler used in Symbian has a bug that prevents us from declaring the // tuple template as a friend (it complains that tuple is redefined). This // hack bypasses the bug by declaring the members that should otherwise be // private as public. // Sun Studio versions < 12 also have the above bug. #if defined(__SYMBIAN32__) || (defined(__SUNPRO_CC) && __SUNPRO_CC < 0x590) # define GTEST_DECLARE_TUPLE_AS_FRIEND_ public: #else # define GTEST_DECLARE_TUPLE_AS_FRIEND_ \ template friend class tuple; \ private: #endif // Visual Studio 2010, 2012, and 2013 define symbols in std::tr1 that conflict // with our own definitions. Therefore using our own tuple does not work on // those compilers. #if defined(_MSC_VER) && _MSC_VER >= 1600 /* 1600 is Visual Studio 2010 */ # error "gtest's tuple doesn't compile on Visual Studio 2010 or later. \ GTEST_USE_OWN_TR1_TUPLE must be set to 0 on those compilers." #endif // GTEST_n_TUPLE_(T) is the type of an n-tuple. #define GTEST_0_TUPLE_(T) tuple<> #define GTEST_1_TUPLE_(T) tuple #define GTEST_2_TUPLE_(T) tuple #define GTEST_3_TUPLE_(T) tuple #define GTEST_4_TUPLE_(T) tuple #define GTEST_5_TUPLE_(T) tuple #define GTEST_6_TUPLE_(T) tuple #define GTEST_7_TUPLE_(T) tuple #define GTEST_8_TUPLE_(T) tuple #define GTEST_9_TUPLE_(T) tuple #define GTEST_10_TUPLE_(T) tuple // GTEST_n_TYPENAMES_(T) declares a list of n typenames. #define GTEST_0_TYPENAMES_(T) #define GTEST_1_TYPENAMES_(T) typename T##0 #define GTEST_2_TYPENAMES_(T) typename T##0, typename T##1 #define GTEST_3_TYPENAMES_(T) typename T##0, typename T##1, typename T##2 #define GTEST_4_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3 #define GTEST_5_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3, typename T##4 #define GTEST_6_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3, typename T##4, typename T##5 #define GTEST_7_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3, typename T##4, typename T##5, typename T##6 #define GTEST_8_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3, typename T##4, typename T##5, typename T##6, typename T##7 #define GTEST_9_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3, typename T##4, typename T##5, typename T##6, \ typename T##7, typename T##8 #define GTEST_10_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \ typename T##3, typename T##4, typename T##5, typename T##6, \ typename T##7, typename T##8, typename T##9 // In theory, defining stuff in the ::std namespace is undefined // behavior. We can do this as we are playing the role of a standard // library vendor. namespace std { namespace tr1 { template class tuple; // Anything in namespace gtest_internal is Google Test's INTERNAL // IMPLEMENTATION DETAIL and MUST NOT BE USED DIRECTLY in user code. namespace gtest_internal { // ByRef::type is T if T is a reference; otherwise it's const T&. template struct ByRef { typedef const T& type; }; // NOLINT template struct ByRef { typedef T& type; }; // NOLINT // A handy wrapper for ByRef. #define GTEST_BY_REF_(T) typename ::std::tr1::gtest_internal::ByRef::type // AddRef::type is T if T is a reference; otherwise it's T&. This // is the same as tr1::add_reference::type. template struct AddRef { typedef T& type; }; // NOLINT template struct AddRef { typedef T& type; }; // NOLINT // A handy wrapper for AddRef. #define GTEST_ADD_REF_(T) typename ::std::tr1::gtest_internal::AddRef::type // A helper for implementing get(). template class Get; // A helper for implementing tuple_element. kIndexValid is true // iff k < the number of fields in tuple type T. template struct TupleElement; template struct TupleElement { typedef T0 type; }; template struct TupleElement { typedef T1 type; }; template struct TupleElement { typedef T2 type; }; template struct TupleElement { typedef T3 type; }; template struct TupleElement { typedef T4 type; }; template struct TupleElement { typedef T5 type; }; template struct TupleElement { typedef T6 type; }; template struct TupleElement { typedef T7 type; }; template struct TupleElement { typedef T8 type; }; template struct TupleElement { typedef T9 type; }; } // namespace gtest_internal template <> class tuple<> { public: tuple() {} tuple(const tuple& /* t */) {} tuple& operator=(const tuple& /* t */) { return *this; } }; template class GTEST_1_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_() {} explicit tuple(GTEST_BY_REF_(T0) f0) : f0_(f0) {} tuple(const tuple& t) : f0_(t.f0_) {} template tuple(const GTEST_1_TUPLE_(U)& t) : f0_(t.f0_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_1_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_1_TUPLE_(U)& t) { f0_ = t.f0_; return *this; } T0 f0_; }; template class GTEST_2_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1) : f0_(f0), f1_(f1) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_) {} template tuple(const GTEST_2_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_) {} template tuple(const ::std::pair& p) : f0_(p.first), f1_(p.second) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_2_TUPLE_(U)& t) { return CopyFrom(t); } template tuple& operator=(const ::std::pair& p) { f0_ = p.first; f1_ = p.second; return *this; } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_2_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; return *this; } T0 f0_; T1 f1_; }; template class GTEST_3_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2) : f0_(f0), f1_(f1), f2_(f2) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {} template tuple(const GTEST_3_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_3_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_3_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; return *this; } T0 f0_; T1 f1_; T2 f2_; }; template class GTEST_4_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3) : f0_(f0), f1_(f1), f2_(f2), f3_(f3) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_) {} template tuple(const GTEST_4_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_4_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_4_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; }; template class GTEST_5_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_(), f4_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_) {} template tuple(const GTEST_5_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_5_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_5_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; f4_ = t.f4_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; T4 f4_; }; template class GTEST_6_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4, GTEST_BY_REF_(T5) f5) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4), f5_(f5) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_) {} template tuple(const GTEST_6_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_6_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_6_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; f4_ = t.f4_; f5_ = t.f5_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; T4 f4_; T5 f5_; }; template class GTEST_7_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4, GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4), f5_(f5), f6_(f6) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {} template tuple(const GTEST_7_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_7_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_7_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; f4_ = t.f4_; f5_ = t.f5_; f6_ = t.f6_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; T4 f4_; T5 f5_; T6 f6_; }; template class GTEST_8_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4, GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4), f5_(f5), f6_(f6), f7_(f7) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {} template tuple(const GTEST_8_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_8_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_8_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; f4_ = t.f4_; f5_ = t.f5_; f6_ = t.f6_; f7_ = t.f7_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; T4 f4_; T5 f5_; T6 f6_; T7 f7_; }; template class GTEST_9_TUPLE_(T) { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4, GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7, GTEST_BY_REF_(T8) f8) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4), f5_(f5), f6_(f6), f7_(f7), f8_(f8) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {} template tuple(const GTEST_9_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_9_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_9_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; f4_ = t.f4_; f5_ = t.f5_; f6_ = t.f6_; f7_ = t.f7_; f8_ = t.f8_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; T4 f4_; T5 f5_; T6 f6_; T7 f7_; T8 f8_; }; template class tuple { public: template friend class gtest_internal::Get; tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_(), f9_() {} explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1, GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4, GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7, GTEST_BY_REF_(T8) f8, GTEST_BY_REF_(T9) f9) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4), f5_(f5), f6_(f6), f7_(f7), f8_(f8), f9_(f9) {} tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_), f9_(t.f9_) {} template tuple(const GTEST_10_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_), f9_(t.f9_) {} tuple& operator=(const tuple& t) { return CopyFrom(t); } template tuple& operator=(const GTEST_10_TUPLE_(U)& t) { return CopyFrom(t); } GTEST_DECLARE_TUPLE_AS_FRIEND_ template tuple& CopyFrom(const GTEST_10_TUPLE_(U)& t) { f0_ = t.f0_; f1_ = t.f1_; f2_ = t.f2_; f3_ = t.f3_; f4_ = t.f4_; f5_ = t.f5_; f6_ = t.f6_; f7_ = t.f7_; f8_ = t.f8_; f9_ = t.f9_; return *this; } T0 f0_; T1 f1_; T2 f2_; T3 f3_; T4 f4_; T5 f5_; T6 f6_; T7 f7_; T8 f8_; T9 f9_; }; // 6.1.3.2 Tuple creation functions. // Known limitations: we don't support passing an // std::tr1::reference_wrapper to make_tuple(). And we don't // implement tie(). inline tuple<> make_tuple() { return tuple<>(); } template inline GTEST_1_TUPLE_(T) make_tuple(const T0& f0) { return GTEST_1_TUPLE_(T)(f0); } template inline GTEST_2_TUPLE_(T) make_tuple(const T0& f0, const T1& f1) { return GTEST_2_TUPLE_(T)(f0, f1); } template inline GTEST_3_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2) { return GTEST_3_TUPLE_(T)(f0, f1, f2); } template inline GTEST_4_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3) { return GTEST_4_TUPLE_(T)(f0, f1, f2, f3); } template inline GTEST_5_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3, const T4& f4) { return GTEST_5_TUPLE_(T)(f0, f1, f2, f3, f4); } template inline GTEST_6_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3, const T4& f4, const T5& f5) { return GTEST_6_TUPLE_(T)(f0, f1, f2, f3, f4, f5); } template inline GTEST_7_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3, const T4& f4, const T5& f5, const T6& f6) { return GTEST_7_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6); } template inline GTEST_8_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3, const T4& f4, const T5& f5, const T6& f6, const T7& f7) { return GTEST_8_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7); } template inline GTEST_9_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3, const T4& f4, const T5& f5, const T6& f6, const T7& f7, const T8& f8) { return GTEST_9_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8); } template inline GTEST_10_TUPLE_(T) make_tuple(const T0& f0, const T1& f1, const T2& f2, const T3& f3, const T4& f4, const T5& f5, const T6& f6, const T7& f7, const T8& f8, const T9& f9) { return GTEST_10_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8, f9); } // 6.1.3.3 Tuple helper classes. template struct tuple_size; template struct tuple_size { static const int value = 0; }; template struct tuple_size { static const int value = 1; }; template struct tuple_size { static const int value = 2; }; template struct tuple_size { static const int value = 3; }; template struct tuple_size { static const int value = 4; }; template struct tuple_size { static const int value = 5; }; template struct tuple_size { static const int value = 6; }; template struct tuple_size { static const int value = 7; }; template struct tuple_size { static const int value = 8; }; template struct tuple_size { static const int value = 9; }; template struct tuple_size { static const int value = 10; }; template struct tuple_element { typedef typename gtest_internal::TupleElement< k < (tuple_size::value), k, Tuple>::type type; }; #define GTEST_TUPLE_ELEMENT_(k, Tuple) typename tuple_element::type // 6.1.3.4 Element access. namespace gtest_internal { template <> class Get<0> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple)) Field(Tuple& t) { return t.f0_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple)) ConstField(const Tuple& t) { return t.f0_; } }; template <> class Get<1> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple)) Field(Tuple& t) { return t.f1_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple)) ConstField(const Tuple& t) { return t.f1_; } }; template <> class Get<2> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple)) Field(Tuple& t) { return t.f2_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple)) ConstField(const Tuple& t) { return t.f2_; } }; template <> class Get<3> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple)) Field(Tuple& t) { return t.f3_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple)) ConstField(const Tuple& t) { return t.f3_; } }; template <> class Get<4> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple)) Field(Tuple& t) { return t.f4_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple)) ConstField(const Tuple& t) { return t.f4_; } }; template <> class Get<5> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple)) Field(Tuple& t) { return t.f5_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple)) ConstField(const Tuple& t) { return t.f5_; } }; template <> class Get<6> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple)) Field(Tuple& t) { return t.f6_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple)) ConstField(const Tuple& t) { return t.f6_; } }; template <> class Get<7> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple)) Field(Tuple& t) { return t.f7_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple)) ConstField(const Tuple& t) { return t.f7_; } }; template <> class Get<8> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple)) Field(Tuple& t) { return t.f8_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple)) ConstField(const Tuple& t) { return t.f8_; } }; template <> class Get<9> { public: template static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple)) Field(Tuple& t) { return t.f9_; } // NOLINT template static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple)) ConstField(const Tuple& t) { return t.f9_; } }; } // namespace gtest_internal template GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_10_TUPLE_(T))) get(GTEST_10_TUPLE_(T)& t) { return gtest_internal::Get::Field(t); } template GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_10_TUPLE_(T))) get(const GTEST_10_TUPLE_(T)& t) { return gtest_internal::Get::ConstField(t); } // 6.1.3.5 Relational operators // We only implement == and !=, as we don't have a need for the rest yet. namespace gtest_internal { // SameSizeTuplePrefixComparator::Eq(t1, t2) returns true if the // first k fields of t1 equals the first k fields of t2. // SameSizeTuplePrefixComparator(k1, k2) would be a compiler error if // k1 != k2. template struct SameSizeTuplePrefixComparator; template <> struct SameSizeTuplePrefixComparator<0, 0> { template static bool Eq(const Tuple1& /* t1 */, const Tuple2& /* t2 */) { return true; } }; template struct SameSizeTuplePrefixComparator { template static bool Eq(const Tuple1& t1, const Tuple2& t2) { return SameSizeTuplePrefixComparator::Eq(t1, t2) && ::std::tr1::get(t1) == ::std::tr1::get(t2); } }; } // namespace gtest_internal template inline bool operator==(const GTEST_10_TUPLE_(T)& t, const GTEST_10_TUPLE_(U)& u) { return gtest_internal::SameSizeTuplePrefixComparator< tuple_size::value, tuple_size::value>::Eq(t, u); } template inline bool operator!=(const GTEST_10_TUPLE_(T)& t, const GTEST_10_TUPLE_(U)& u) { return !(t == u); } // 6.1.4 Pairs. // Unimplemented. } // namespace tr1 } // namespace std #undef GTEST_0_TUPLE_ #undef GTEST_1_TUPLE_ #undef GTEST_2_TUPLE_ #undef GTEST_3_TUPLE_ #undef GTEST_4_TUPLE_ #undef GTEST_5_TUPLE_ #undef GTEST_6_TUPLE_ #undef GTEST_7_TUPLE_ #undef GTEST_8_TUPLE_ #undef GTEST_9_TUPLE_ #undef GTEST_10_TUPLE_ #undef GTEST_0_TYPENAMES_ #undef GTEST_1_TYPENAMES_ #undef GTEST_2_TYPENAMES_ #undef GTEST_3_TYPENAMES_ #undef GTEST_4_TYPENAMES_ #undef GTEST_5_TYPENAMES_ #undef GTEST_6_TYPENAMES_ #undef GTEST_7_TYPENAMES_ #undef GTEST_8_TYPENAMES_ #undef GTEST_9_TYPENAMES_ #undef GTEST_10_TYPENAMES_ #undef GTEST_DECLARE_TUPLE_AS_FRIEND_ #undef GTEST_BY_REF_ #undef GTEST_ADD_REF_ #undef GTEST_TUPLE_ELEMENT_ #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_ # elif GTEST_ENV_HAS_STD_TUPLE_ # include // C++11 puts its tuple into the ::std namespace rather than // ::std::tr1. gtest expects tuple to live in ::std::tr1, so put it there. // This causes undefined behavior, but supported compilers react in // the way we intend. namespace std { namespace tr1 { using ::std::get; using ::std::make_tuple; using ::std::tuple; using ::std::tuple_element; using ::std::tuple_size; } } # elif GTEST_OS_SYMBIAN // On Symbian, BOOST_HAS_TR1_TUPLE causes Boost's TR1 tuple library to // use STLport's tuple implementation, which unfortunately doesn't // work as the copy of STLport distributed with Symbian is incomplete. // By making sure BOOST_HAS_TR1_TUPLE is undefined, we force Boost to // use its own tuple implementation. # ifdef BOOST_HAS_TR1_TUPLE # undef BOOST_HAS_TR1_TUPLE # endif // BOOST_HAS_TR1_TUPLE // This prevents , which defines // BOOST_HAS_TR1_TUPLE, from being #included by Boost's . # define BOOST_TR1_DETAIL_CONFIG_HPP_INCLUDED # include // IWYU pragma: export // NOLINT # elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40000) // GCC 4.0+ implements tr1/tuple in the header. This does // not conform to the TR1 spec, which requires the header to be . # if !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302 // Until version 4.3.2, gcc has a bug that causes , // which is #included by , to not compile when RTTI is // disabled. _TR1_FUNCTIONAL is the header guard for // . Hence the following #define is a hack to prevent // from being included. # define _TR1_FUNCTIONAL 1 # include # undef _TR1_FUNCTIONAL // Allows the user to #include // if he chooses to. # else # include // NOLINT # endif // !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302 # else // If the compiler is not GCC 4.0+, we assume the user is using a // spec-conforming TR1 implementation. # include // IWYU pragma: export // NOLINT # endif // GTEST_USE_OWN_TR1_TUPLE #endif // GTEST_HAS_TR1_TUPLE // Determines whether clone(2) is supported. // Usually it will only be available on Linux, excluding // Linux on the Itanium architecture. // Also see http://linux.die.net/man/2/clone. #ifndef GTEST_HAS_CLONE // The user didn't tell us, so we need to figure it out. # if GTEST_OS_LINUX && !defined(__ia64__) # if GTEST_OS_LINUX_ANDROID // On Android, clone() is only available on ARM starting with Gingerbread. # if defined(__arm__) && __ANDROID_API__ >= 9 # define GTEST_HAS_CLONE 1 # else # define GTEST_HAS_CLONE 0 # endif # else # define GTEST_HAS_CLONE 1 # endif # else # define GTEST_HAS_CLONE 0 # endif // GTEST_OS_LINUX && !defined(__ia64__) #endif // GTEST_HAS_CLONE // Determines whether to support stream redirection. This is used to test // output correctness and to implement death tests. #ifndef GTEST_HAS_STREAM_REDIRECTION // By default, we assume that stream redirection is supported on all // platforms except known mobile ones. # if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || \ GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT # define GTEST_HAS_STREAM_REDIRECTION 0 # else # define GTEST_HAS_STREAM_REDIRECTION 1 # endif // !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_SYMBIAN #endif // GTEST_HAS_STREAM_REDIRECTION // Determines whether to support death tests. // Google Test does not support death tests for VC 7.1 and earlier as // abort() in a VC 7.1 application compiled as GUI in debug config // pops up a dialog window that cannot be suppressed programmatically. #if (GTEST_OS_LINUX || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS || \ (GTEST_OS_MAC && !GTEST_OS_IOS) || \ (GTEST_OS_WINDOWS_DESKTOP && _MSC_VER >= 1400) || \ GTEST_OS_WINDOWS_MINGW || GTEST_OS_AIX || GTEST_OS_HPUX || \ GTEST_OS_OPENBSD || GTEST_OS_QNX || GTEST_OS_FREEBSD) # define GTEST_HAS_DEATH_TEST 1 # include // NOLINT #endif // We don't support MSVC 7.1 with exceptions disabled now. Therefore // all the compilers we care about are adequate for supporting // value-parameterized tests. #define GTEST_HAS_PARAM_TEST 1 // Determines whether to support type-driven tests. // Typed tests need and variadic macros, which GCC, VC++ 8.0, // Sun Pro CC, IBM Visual Age, and HP aCC support. #if defined(__GNUC__) || (_MSC_VER >= 1400) || defined(__SUNPRO_CC) || \ defined(__IBMCPP__) || defined(__HP_aCC) # define GTEST_HAS_TYPED_TEST 1 # define GTEST_HAS_TYPED_TEST_P 1 #endif // Determines whether to support Combine(). This only makes sense when // value-parameterized tests are enabled. The implementation doesn't // work on Sun Studio since it doesn't understand templated conversion // operators. #if GTEST_HAS_PARAM_TEST && GTEST_HAS_TR1_TUPLE && !defined(__SUNPRO_CC) # define GTEST_HAS_COMBINE 1 #endif // Determines whether the system compiler uses UTF-16 for encoding wide strings. #define GTEST_WIDE_STRING_USES_UTF16_ \ (GTEST_OS_WINDOWS || GTEST_OS_CYGWIN || GTEST_OS_SYMBIAN || GTEST_OS_AIX) // Determines whether test results can be streamed to a socket. #if GTEST_OS_LINUX # define GTEST_CAN_STREAM_RESULTS_ 1 #endif // Defines some utility macros. // The GNU compiler emits a warning if nested "if" statements are followed by // an "else" statement and braces are not used to explicitly disambiguate the // "else" binding. This leads to problems with code like: // // if (gate) // ASSERT_*(condition) << "Some message"; // // The "switch (0) case 0:" idiom is used to suppress this. #ifdef __INTEL_COMPILER # define GTEST_AMBIGUOUS_ELSE_BLOCKER_ #else # define GTEST_AMBIGUOUS_ELSE_BLOCKER_ switch (0) case 0: default: // NOLINT #endif // Use this annotation at the end of a struct/class definition to // prevent the compiler from optimizing away instances that are never // used. This is useful when all interesting logic happens inside the // c'tor and / or d'tor. Example: // // struct Foo { // Foo() { ... } // } GTEST_ATTRIBUTE_UNUSED_; // // Also use it after a variable or parameter declaration to tell the // compiler the variable/parameter does not have to be used. #if defined(__GNUC__) && !defined(COMPILER_ICC) # define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused)) #elif defined(__clang__) # if __has_attribute(unused) # define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused)) # endif #endif #ifndef GTEST_ATTRIBUTE_UNUSED_ # define GTEST_ATTRIBUTE_UNUSED_ #endif // A macro to disallow operator= // This should be used in the private: declarations for a class. #define GTEST_DISALLOW_ASSIGN_(type)\ void operator=(type const &) // A macro to disallow copy constructor and operator= // This should be used in the private: declarations for a class. #define GTEST_DISALLOW_COPY_AND_ASSIGN_(type)\ type(type const &);\ GTEST_DISALLOW_ASSIGN_(type) // Tell the compiler to warn about unused return values for functions declared // with this macro. The macro should be used on function declarations // following the argument list: // // Sprocket* AllocateSprocket() GTEST_MUST_USE_RESULT_; #if defined(__GNUC__) && (GTEST_GCC_VER_ >= 30400) && !defined(COMPILER_ICC) # define GTEST_MUST_USE_RESULT_ __attribute__ ((warn_unused_result)) #else # define GTEST_MUST_USE_RESULT_ #endif // __GNUC__ && (GTEST_GCC_VER_ >= 30400) && !COMPILER_ICC // MS C++ compiler emits warning when a conditional expression is compile time // constant. In some contexts this warning is false positive and needs to be // suppressed. Use the following two macros in such cases: // // GTEST_INTENTIONAL_CONST_COND_PUSH_() // while (true) { // GTEST_INTENTIONAL_CONST_COND_POP_() // } # define GTEST_INTENTIONAL_CONST_COND_PUSH_() \ GTEST_DISABLE_MSC_WARNINGS_PUSH_(4127) # define GTEST_INTENTIONAL_CONST_COND_POP_() \ GTEST_DISABLE_MSC_WARNINGS_POP_() // Determine whether the compiler supports Microsoft's Structured Exception // Handling. This is supported by several Windows compilers but generally // does not exist on any other system. #ifndef GTEST_HAS_SEH // The user didn't tell us, so we need to figure it out. # if defined(_MSC_VER) || defined(__BORLANDC__) // These two compilers are known to support SEH. # define GTEST_HAS_SEH 1 # else // Assume no SEH. # define GTEST_HAS_SEH 0 # endif #define GTEST_IS_THREADSAFE \ (0 \ || (GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT) \ || GTEST_HAS_PTHREAD) #endif // GTEST_HAS_SEH #ifdef _MSC_VER # if GTEST_LINKED_AS_SHARED_LIBRARY # define GTEST_API_ __declspec(dllimport) # elif GTEST_CREATE_SHARED_LIBRARY # define GTEST_API_ __declspec(dllexport) # endif #endif // _MSC_VER #ifndef GTEST_API_ # define GTEST_API_ #endif #ifdef __GNUC__ // Ask the compiler to never inline a given function. # define GTEST_NO_INLINE_ __attribute__((noinline)) #else # define GTEST_NO_INLINE_ #endif // _LIBCPP_VERSION is defined by the libc++ library from the LLVM project. #if defined(__GLIBCXX__) || defined(_LIBCPP_VERSION) # define GTEST_HAS_CXXABI_H_ 1 #else # define GTEST_HAS_CXXABI_H_ 0 #endif // A function level attribute to disable checking for use of uninitialized // memory when built with MemorySanitizer. #if defined(__clang__) # if __has_feature(memory_sanitizer) # define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ \ __attribute__((no_sanitize_memory)) # else # define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ # endif // __has_feature(memory_sanitizer) #else # define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ #endif // __clang__ // A function level attribute to disable AddressSanitizer instrumentation. #if defined(__clang__) # if __has_feature(address_sanitizer) # define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ \ __attribute__((no_sanitize_address)) # else # define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ # endif // __has_feature(address_sanitizer) #else # define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ #endif // __clang__ // A function level attribute to disable ThreadSanitizer instrumentation. #if defined(__clang__) # if __has_feature(thread_sanitizer) # define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ \ __attribute__((no_sanitize_thread)) # else # define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ # endif // __has_feature(thread_sanitizer) #else # define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ #endif // __clang__ namespace testing { class Message; #if defined(GTEST_TUPLE_NAMESPACE_) // Import tuple and friends into the ::testing namespace. // It is part of our interface, having them in ::testing allows us to change // their types as needed. using GTEST_TUPLE_NAMESPACE_::get; using GTEST_TUPLE_NAMESPACE_::make_tuple; using GTEST_TUPLE_NAMESPACE_::tuple; using GTEST_TUPLE_NAMESPACE_::tuple_size; using GTEST_TUPLE_NAMESPACE_::tuple_element; #endif // defined(GTEST_TUPLE_NAMESPACE_) namespace internal { // A secret type that Google Test users don't know about. It has no // definition on purpose. Therefore it's impossible to create a // Secret object, which is what we want. class Secret; // The GTEST_COMPILE_ASSERT_ macro can be used to verify that a compile time // expression is true. For example, you could use it to verify the // size of a static array: // // GTEST_COMPILE_ASSERT_(GTEST_ARRAY_SIZE_(names) == NUM_NAMES, // names_incorrect_size); // // or to make sure a struct is smaller than a certain size: // // GTEST_COMPILE_ASSERT_(sizeof(foo) < 128, foo_too_large); // // The second argument to the macro is the name of the variable. If // the expression is false, most compilers will issue a warning/error // containing the name of the variable. #if GTEST_LANG_CXX11 # define GTEST_COMPILE_ASSERT_(expr, msg) static_assert(expr, #msg) #else // !GTEST_LANG_CXX11 template struct CompileAssert { }; # define GTEST_COMPILE_ASSERT_(expr, msg) \ typedef ::testing::internal::CompileAssert<(static_cast(expr))> \ msg[static_cast(expr) ? 1 : -1] GTEST_ATTRIBUTE_UNUSED_ #endif // !GTEST_LANG_CXX11 // Implementation details of GTEST_COMPILE_ASSERT_: // // (In C++11, we simply use static_assert instead of the following) // // - GTEST_COMPILE_ASSERT_ works by defining an array type that has -1 // elements (and thus is invalid) when the expression is false. // // - The simpler definition // // #define GTEST_COMPILE_ASSERT_(expr, msg) typedef char msg[(expr) ? 1 : -1] // // does not work, as gcc supports variable-length arrays whose sizes // are determined at run-time (this is gcc's extension and not part // of the C++ standard). As a result, gcc fails to reject the // following code with the simple definition: // // int foo; // GTEST_COMPILE_ASSERT_(foo, msg); // not supposed to compile as foo is // // not a compile-time constant. // // - By using the type CompileAssert<(bool(expr))>, we ensures that // expr is a compile-time constant. (Template arguments must be // determined at compile-time.) // // - The outter parentheses in CompileAssert<(bool(expr))> are necessary // to work around a bug in gcc 3.4.4 and 4.0.1. If we had written // // CompileAssert // // instead, these compilers will refuse to compile // // GTEST_COMPILE_ASSERT_(5 > 0, some_message); // // (They seem to think the ">" in "5 > 0" marks the end of the // template argument list.) // // - The array size is (bool(expr) ? 1 : -1), instead of simply // // ((expr) ? 1 : -1). // // This is to avoid running into a bug in MS VC 7.1, which // causes ((0.0) ? 1 : -1) to incorrectly evaluate to 1. // StaticAssertTypeEqHelper is used by StaticAssertTypeEq defined in gtest.h. // // This template is declared, but intentionally undefined. template struct StaticAssertTypeEqHelper; template struct StaticAssertTypeEqHelper { enum { value = true }; }; // Evaluates to the number of elements in 'array'. #define GTEST_ARRAY_SIZE_(array) (sizeof(array) / sizeof(array[0])) #if GTEST_HAS_GLOBAL_STRING typedef ::string string; #else typedef ::std::string string; #endif // GTEST_HAS_GLOBAL_STRING #if GTEST_HAS_GLOBAL_WSTRING typedef ::wstring wstring; #elif GTEST_HAS_STD_WSTRING typedef ::std::wstring wstring; #endif // GTEST_HAS_GLOBAL_WSTRING // A helper for suppressing warnings on constant condition. It just // returns 'condition'. GTEST_API_ bool IsTrue(bool condition); // Defines scoped_ptr. // This implementation of scoped_ptr is PARTIAL - it only contains // enough stuff to satisfy Google Test's need. template class scoped_ptr { public: typedef T element_type; explicit scoped_ptr(T* p = NULL) : ptr_(p) {} ~scoped_ptr() { reset(); } T& operator*() const { return *ptr_; } T* operator->() const { return ptr_; } T* get() const { return ptr_; } T* release() { T* const ptr = ptr_; ptr_ = NULL; return ptr; } void reset(T* p = NULL) { if (p != ptr_) { if (IsTrue(sizeof(T) > 0)) { // Makes sure T is a complete type. delete ptr_; } ptr_ = p; } } friend void swap(scoped_ptr& a, scoped_ptr& b) { using std::swap; swap(a.ptr_, b.ptr_); } private: T* ptr_; GTEST_DISALLOW_COPY_AND_ASSIGN_(scoped_ptr); }; // Defines RE. // A simple C++ wrapper for . It uses the POSIX Extended // Regular Expression syntax. class GTEST_API_ RE { public: // A copy constructor is required by the Standard to initialize object // references from r-values. RE(const RE& other) { Init(other.pattern()); } // Constructs an RE from a string. RE(const ::std::string& regex) { Init(regex.c_str()); } // NOLINT #if GTEST_HAS_GLOBAL_STRING RE(const ::string& regex) { Init(regex.c_str()); } // NOLINT #endif // GTEST_HAS_GLOBAL_STRING RE(const char* regex) { Init(regex); } // NOLINT ~RE(); // Returns the string representation of the regex. const char* pattern() const { return pattern_; } // FullMatch(str, re) returns true iff regular expression re matches // the entire str. // PartialMatch(str, re) returns true iff regular expression re // matches a substring of str (including str itself). // // TODO(wan@google.com): make FullMatch() and PartialMatch() work // when str contains NUL characters. static bool FullMatch(const ::std::string& str, const RE& re) { return FullMatch(str.c_str(), re); } static bool PartialMatch(const ::std::string& str, const RE& re) { return PartialMatch(str.c_str(), re); } #if GTEST_HAS_GLOBAL_STRING static bool FullMatch(const ::string& str, const RE& re) { return FullMatch(str.c_str(), re); } static bool PartialMatch(const ::string& str, const RE& re) { return PartialMatch(str.c_str(), re); } #endif // GTEST_HAS_GLOBAL_STRING static bool FullMatch(const char* str, const RE& re); static bool PartialMatch(const char* str, const RE& re); private: void Init(const char* regex); // We use a const char* instead of an std::string, as Google Test used to be // used where std::string is not available. TODO(wan@google.com): change to // std::string. const char* pattern_; bool is_valid_; #if GTEST_USES_POSIX_RE regex_t full_regex_; // For FullMatch(). regex_t partial_regex_; // For PartialMatch(). #else // GTEST_USES_SIMPLE_RE const char* full_pattern_; // For FullMatch(); #endif GTEST_DISALLOW_ASSIGN_(RE); }; // Formats a source file path and a line number as they would appear // in an error message from the compiler used to compile this code. GTEST_API_ ::std::string FormatFileLocation(const char* file, int line); // Formats a file location for compiler-independent XML output. // Although this function is not platform dependent, we put it next to // FormatFileLocation in order to contrast the two functions. GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(const char* file, int line); // Defines logging utilities: // GTEST_LOG_(severity) - logs messages at the specified severity level. The // message itself is streamed into the macro. // LogToStderr() - directs all log messages to stderr. // FlushInfoLog() - flushes informational log messages. enum GTestLogSeverity { GTEST_INFO, GTEST_WARNING, GTEST_ERROR, GTEST_FATAL }; // Formats log entry severity, provides a stream object for streaming the // log message, and terminates the message with a newline when going out of // scope. class GTEST_API_ GTestLog { public: GTestLog(GTestLogSeverity severity, const char* file, int line); // Flushes the buffers and, if severity is GTEST_FATAL, aborts the program. ~GTestLog(); ::std::ostream& GetStream() { return ::std::cerr; } private: const GTestLogSeverity severity_; GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestLog); }; #define GTEST_LOG_(severity) \ ::testing::internal::GTestLog(::testing::internal::GTEST_##severity, \ __FILE__, __LINE__).GetStream() inline void LogToStderr() {} inline void FlushInfoLog() { fflush(NULL); } // INTERNAL IMPLEMENTATION - DO NOT USE. // // GTEST_CHECK_ is an all-mode assert. It aborts the program if the condition // is not satisfied. // Synopsys: // GTEST_CHECK_(boolean_condition); // or // GTEST_CHECK_(boolean_condition) << "Additional message"; // // This checks the condition and if the condition is not satisfied // it prints message about the condition violation, including the // condition itself, plus additional message streamed into it, if any, // and then it aborts the program. It aborts the program irrespective of // whether it is built in the debug mode or not. #define GTEST_CHECK_(condition) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::IsTrue(condition)) \ ; \ else \ GTEST_LOG_(FATAL) << "Condition " #condition " failed. " // An all-mode assert to verify that the given POSIX-style function // call returns 0 (indicating success). Known limitation: this // doesn't expand to a balanced 'if' statement, so enclose the macro // in {} if you need to use it as the only statement in an 'if' // branch. #define GTEST_CHECK_POSIX_SUCCESS_(posix_call) \ if (const int gtest_error = (posix_call)) \ GTEST_LOG_(FATAL) << #posix_call << "failed with error " \ << gtest_error #if GTEST_HAS_STD_MOVE_ using std::move; #else // GTEST_HAS_STD_MOVE_ template const T& move(const T& t) { return t; } #endif // GTEST_HAS_STD_MOVE_ // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Use ImplicitCast_ as a safe version of static_cast for upcasting in // the type hierarchy (e.g. casting a Foo* to a SuperclassOfFoo* or a // const Foo*). When you use ImplicitCast_, the compiler checks that // the cast is safe. Such explicit ImplicitCast_s are necessary in // surprisingly many situations where C++ demands an exact type match // instead of an argument type convertable to a target type. // // The syntax for using ImplicitCast_ is the same as for static_cast: // // ImplicitCast_(expr) // // ImplicitCast_ would have been part of the C++ standard library, // but the proposal was submitted too late. It will probably make // its way into the language in the future. // // This relatively ugly name is intentional. It prevents clashes with // similar functions users may have (e.g., implicit_cast). The internal // namespace alone is not enough because the function can be found by ADL. template inline To ImplicitCast_(To x) { return ::testing::internal::move(x); } // When you upcast (that is, cast a pointer from type Foo to type // SuperclassOfFoo), it's fine to use ImplicitCast_<>, since upcasts // always succeed. When you downcast (that is, cast a pointer from // type Foo to type SubclassOfFoo), static_cast<> isn't safe, because // how do you know the pointer is really of type SubclassOfFoo? It // could be a bare Foo, or of type DifferentSubclassOfFoo. Thus, // when you downcast, you should use this macro. In debug mode, we // use dynamic_cast<> to double-check the downcast is legal (we die // if it's not). In normal mode, we do the efficient static_cast<> // instead. Thus, it's important to test in debug mode to make sure // the cast is legal! // This is the only place in the code we should use dynamic_cast<>. // In particular, you SHOULDN'T be using dynamic_cast<> in order to // do RTTI (eg code like this: // if (dynamic_cast(foo)) HandleASubclass1Object(foo); // if (dynamic_cast(foo)) HandleASubclass2Object(foo); // You should design the code some other way not to need this. // // This relatively ugly name is intentional. It prevents clashes with // similar functions users may have (e.g., down_cast). The internal // namespace alone is not enough because the function can be found by ADL. template // use like this: DownCast_(foo); inline To DownCast_(From* f) { // so we only accept pointers // Ensures that To is a sub-type of From *. This test is here only // for compile-time type checking, and has no overhead in an // optimized build at run-time, as it will be optimized away // completely. GTEST_INTENTIONAL_CONST_COND_PUSH_() if (false) { GTEST_INTENTIONAL_CONST_COND_POP_() const To to = NULL; ::testing::internal::ImplicitCast_(to); } #if GTEST_HAS_RTTI // RTTI: debug mode only! GTEST_CHECK_(f == NULL || dynamic_cast(f) != NULL); #endif return static_cast(f); } // Downcasts the pointer of type Base to Derived. // Derived must be a subclass of Base. The parameter MUST // point to a class of type Derived, not any subclass of it. // When RTTI is available, the function performs a runtime // check to enforce this. template Derived* CheckedDowncastToActualType(Base* base) { #if GTEST_HAS_RTTI GTEST_CHECK_(typeid(*base) == typeid(Derived)); return dynamic_cast(base); // NOLINT #else return static_cast(base); // Poor man's downcast. #endif } #if GTEST_HAS_STREAM_REDIRECTION // Defines the stderr capturer: // CaptureStdout - starts capturing stdout. // GetCapturedStdout - stops capturing stdout and returns the captured string. // CaptureStderr - starts capturing stderr. // GetCapturedStderr - stops capturing stderr and returns the captured string. // GTEST_API_ void CaptureStdout(); GTEST_API_ std::string GetCapturedStdout(); GTEST_API_ void CaptureStderr(); GTEST_API_ std::string GetCapturedStderr(); #endif // GTEST_HAS_STREAM_REDIRECTION #if GTEST_HAS_DEATH_TEST const ::std::vector& GetInjectableArgvs(); void SetInjectableArgvs(const ::std::vector* new_argvs); // A copy of all command line arguments. Set by InitGoogleTest(). extern ::std::vector g_argvs; #endif // GTEST_HAS_DEATH_TEST // Defines synchronization primitives. #if GTEST_IS_THREADSAFE # if GTEST_HAS_PTHREAD // Sleeps for (roughly) n milliseconds. This function is only for testing // Google Test's own constructs. Don't use it in user tests, either // directly or indirectly. inline void SleepMilliseconds(int n) { const timespec time = { 0, // 0 seconds. n * 1000L * 1000L, // And n ms. }; nanosleep(&time, NULL); } # endif // GTEST_HAS_PTHREAD # if 0 // OS detection # elif GTEST_HAS_PTHREAD // Allows a controller thread to pause execution of newly created // threads until notified. Instances of this class must be created // and destroyed in the controller thread. // // This class is only for testing Google Test's own constructs. Do not // use it in user tests, either directly or indirectly. class Notification { public: Notification() : notified_(false) { GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL)); } ~Notification() { pthread_mutex_destroy(&mutex_); } // Notifies all threads created with this notification to start. Must // be called from the controller thread. void Notify() { pthread_mutex_lock(&mutex_); notified_ = true; pthread_mutex_unlock(&mutex_); } // Blocks until the controller thread notifies. Must be called from a test // thread. void WaitForNotification() { for (;;) { pthread_mutex_lock(&mutex_); const bool notified = notified_; pthread_mutex_unlock(&mutex_); if (notified) break; SleepMilliseconds(10); } } private: pthread_mutex_t mutex_; bool notified_; GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification); }; # elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT GTEST_API_ void SleepMilliseconds(int n); // Provides leak-safe Windows kernel handle ownership. // Used in death tests and in threading support. class GTEST_API_ AutoHandle { public: // Assume that Win32 HANDLE type is equivalent to void*. Doing so allows us to // avoid including in this header file. Including is // undesirable because it defines a lot of symbols and macros that tend to // conflict with client code. This assumption is verified by // WindowsTypesTest.HANDLEIsVoidStar. typedef void* Handle; AutoHandle(); explicit AutoHandle(Handle handle); ~AutoHandle(); Handle Get() const; void Reset(); void Reset(Handle handle); private: // Returns true iff the handle is a valid handle object that can be closed. bool IsCloseable() const; Handle handle_; GTEST_DISALLOW_COPY_AND_ASSIGN_(AutoHandle); }; // Allows a controller thread to pause execution of newly created // threads until notified. Instances of this class must be created // and destroyed in the controller thread. // // This class is only for testing Google Test's own constructs. Do not // use it in user tests, either directly or indirectly. class GTEST_API_ Notification { public: Notification(); void Notify(); void WaitForNotification(); private: AutoHandle event_; GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification); }; # endif // OS detection // On MinGW, we can have both GTEST_OS_WINDOWS and GTEST_HAS_PTHREAD // defined, but we don't want to use MinGW's pthreads implementation, which // has conformance problems with some versions of the POSIX standard. # if GTEST_HAS_PTHREAD && !GTEST_OS_WINDOWS_MINGW // As a C-function, ThreadFuncWithCLinkage cannot be templated itself. // Consequently, it cannot select a correct instantiation of ThreadWithParam // in order to call its Run(). Introducing ThreadWithParamBase as a // non-templated base class for ThreadWithParam allows us to bypass this // problem. class ThreadWithParamBase { public: virtual ~ThreadWithParamBase() {} virtual void Run() = 0; }; // pthread_create() accepts a pointer to a function type with the C linkage. // According to the Standard (7.5/1), function types with different linkages // are different even if they are otherwise identical. Some compilers (for // example, SunStudio) treat them as different types. Since class methods // cannot be defined with C-linkage we need to define a free C-function to // pass into pthread_create(). extern "C" inline void* ThreadFuncWithCLinkage(void* thread) { static_cast(thread)->Run(); return NULL; } // Helper class for testing Google Test's multi-threading constructs. // To use it, write: // // void ThreadFunc(int param) { /* Do things with param */ } // Notification thread_can_start; // ... // // The thread_can_start parameter is optional; you can supply NULL. // ThreadWithParam thread(&ThreadFunc, 5, &thread_can_start); // thread_can_start.Notify(); // // These classes are only for testing Google Test's own constructs. Do // not use them in user tests, either directly or indirectly. template class ThreadWithParam : public ThreadWithParamBase { public: typedef void UserThreadFunc(T); ThreadWithParam(UserThreadFunc* func, T param, Notification* thread_can_start) : func_(func), param_(param), thread_can_start_(thread_can_start), finished_(false) { ThreadWithParamBase* const base = this; // The thread can be created only after all fields except thread_ // have been initialized. GTEST_CHECK_POSIX_SUCCESS_( pthread_create(&thread_, 0, &ThreadFuncWithCLinkage, base)); } ~ThreadWithParam() { Join(); } void Join() { if (!finished_) { GTEST_CHECK_POSIX_SUCCESS_(pthread_join(thread_, 0)); finished_ = true; } } virtual void Run() { if (thread_can_start_ != NULL) thread_can_start_->WaitForNotification(); func_(param_); } private: UserThreadFunc* const func_; // User-supplied thread function. const T param_; // User-supplied parameter to the thread function. // When non-NULL, used to block execution until the controller thread // notifies. Notification* const thread_can_start_; bool finished_; // true iff we know that the thread function has finished. pthread_t thread_; // The native thread object. GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam); }; # endif // GTEST_HAS_PTHREAD && !GTEST_OS_WINDOWS_MINGW # if 0 // OS detection # elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT // Mutex implements mutex on Windows platforms. It is used in conjunction // with class MutexLock: // // Mutex mutex; // ... // MutexLock lock(&mutex); // Acquires the mutex and releases it at the // // end of the current scope. // // A static Mutex *must* be defined or declared using one of the following // macros: // GTEST_DEFINE_STATIC_MUTEX_(g_some_mutex); // GTEST_DECLARE_STATIC_MUTEX_(g_some_mutex); // // (A non-static Mutex is defined/declared in the usual way). class GTEST_API_ Mutex { public: enum MutexType { kStatic = 0, kDynamic = 1 }; // We rely on kStaticMutex being 0 as it is to what the linker initializes // type_ in static mutexes. critical_section_ will be initialized lazily // in ThreadSafeLazyInit(). enum StaticConstructorSelector { kStaticMutex = 0 }; // This constructor intentionally does nothing. It relies on type_ being // statically initialized to 0 (effectively setting it to kStatic) and on // ThreadSafeLazyInit() to lazily initialize the rest of the members. explicit Mutex(StaticConstructorSelector /*dummy*/) {} Mutex(); ~Mutex(); void Lock(); void Unlock(); // Does nothing if the current thread holds the mutex. Otherwise, crashes // with high probability. void AssertHeld(); private: // Initializes owner_thread_id_ and critical_section_ in static mutexes. void ThreadSafeLazyInit(); // Per http://blogs.msdn.com/b/oldnewthing/archive/2004/02/23/78395.aspx, // we assume that 0 is an invalid value for thread IDs. unsigned int owner_thread_id_; // For static mutexes, we rely on these members being initialized to zeros // by the linker. MutexType type_; long critical_section_init_phase_; // NOLINT _RTL_CRITICAL_SECTION* critical_section_; GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex); }; # define GTEST_DECLARE_STATIC_MUTEX_(mutex) \ extern ::testing::internal::Mutex mutex # define GTEST_DEFINE_STATIC_MUTEX_(mutex) \ ::testing::internal::Mutex mutex(::testing::internal::Mutex::kStaticMutex) // We cannot name this class MutexLock because the ctor declaration would // conflict with a macro named MutexLock, which is defined on some // platforms. That macro is used as a defensive measure to prevent against // inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than // "MutexLock l(&mu)". Hence the typedef trick below. class GTestMutexLock { public: explicit GTestMutexLock(Mutex* mutex) : mutex_(mutex) { mutex_->Lock(); } ~GTestMutexLock() { mutex_->Unlock(); } private: Mutex* const mutex_; GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock); }; typedef GTestMutexLock MutexLock; // Base class for ValueHolder. Allows a caller to hold and delete a value // without knowing its type. class ThreadLocalValueHolderBase { public: virtual ~ThreadLocalValueHolderBase() {} }; // Provides a way for a thread to send notifications to a ThreadLocal // regardless of its parameter type. class ThreadLocalBase { public: // Creates a new ValueHolder object holding a default value passed to // this ThreadLocal's constructor and returns it. It is the caller's // responsibility not to call this when the ThreadLocal instance already // has a value on the current thread. virtual ThreadLocalValueHolderBase* NewValueForCurrentThread() const = 0; protected: ThreadLocalBase() {} virtual ~ThreadLocalBase() {} private: GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocalBase); }; // Maps a thread to a set of ThreadLocals that have values instantiated on that // thread and notifies them when the thread exits. A ThreadLocal instance is // expected to persist until all threads it has values on have terminated. class GTEST_API_ ThreadLocalRegistry { public: // Registers thread_local_instance as having value on the current thread. // Returns a value that can be used to identify the thread from other threads. static ThreadLocalValueHolderBase* GetValueOnCurrentThread( const ThreadLocalBase* thread_local_instance); // Invoked when a ThreadLocal instance is destroyed. static void OnThreadLocalDestroyed( const ThreadLocalBase* thread_local_instance); }; class GTEST_API_ ThreadWithParamBase { public: void Join(); protected: class Runnable { public: virtual ~Runnable() {} virtual void Run() = 0; }; ThreadWithParamBase(Runnable *runnable, Notification* thread_can_start); virtual ~ThreadWithParamBase(); private: AutoHandle thread_; }; // Helper class for testing Google Test's multi-threading constructs. template class ThreadWithParam : public ThreadWithParamBase { public: typedef void UserThreadFunc(T); ThreadWithParam(UserThreadFunc* func, T param, Notification* thread_can_start) : ThreadWithParamBase(new RunnableImpl(func, param), thread_can_start) { } virtual ~ThreadWithParam() {} private: class RunnableImpl : public Runnable { public: RunnableImpl(UserThreadFunc* func, T param) : func_(func), param_(param) { } virtual ~RunnableImpl() {} virtual void Run() { func_(param_); } private: UserThreadFunc* const func_; const T param_; GTEST_DISALLOW_COPY_AND_ASSIGN_(RunnableImpl); }; GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam); }; // Implements thread-local storage on Windows systems. // // // Thread 1 // ThreadLocal tl(100); // 100 is the default value for each thread. // // // Thread 2 // tl.set(150); // Changes the value for thread 2 only. // EXPECT_EQ(150, tl.get()); // // // Thread 1 // EXPECT_EQ(100, tl.get()); // In thread 1, tl has the original value. // tl.set(200); // EXPECT_EQ(200, tl.get()); // // The template type argument T must have a public copy constructor. // In addition, the default ThreadLocal constructor requires T to have // a public default constructor. // // The users of a TheadLocal instance have to make sure that all but one // threads (including the main one) using that instance have exited before // destroying it. Otherwise, the per-thread objects managed for them by the // ThreadLocal instance are not guaranteed to be destroyed on all platforms. // // Google Test only uses global ThreadLocal objects. That means they // will die after main() has returned. Therefore, no per-thread // object managed by Google Test will be leaked as long as all threads // using Google Test have exited when main() returns. template class ThreadLocal : public ThreadLocalBase { public: ThreadLocal() : default_() {} explicit ThreadLocal(const T& value) : default_(value) {} ~ThreadLocal() { ThreadLocalRegistry::OnThreadLocalDestroyed(this); } T* pointer() { return GetOrCreateValue(); } const T* pointer() const { return GetOrCreateValue(); } const T& get() const { return *pointer(); } void set(const T& value) { *pointer() = value; } private: // Holds a value of T. Can be deleted via its base class without the caller // knowing the type of T. class ValueHolder : public ThreadLocalValueHolderBase { public: explicit ValueHolder(const T& value) : value_(value) {} T* pointer() { return &value_; } private: T value_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder); }; T* GetOrCreateValue() const { return static_cast( ThreadLocalRegistry::GetValueOnCurrentThread(this))->pointer(); } virtual ThreadLocalValueHolderBase* NewValueForCurrentThread() const { return new ValueHolder(default_); } const T default_; // The default value for each thread. GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal); }; # elif GTEST_HAS_PTHREAD // MutexBase and Mutex implement mutex on pthreads-based platforms. class MutexBase { public: // Acquires this mutex. void Lock() { GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_lock(&mutex_)); owner_ = pthread_self(); has_owner_ = true; } // Releases this mutex. void Unlock() { // Since the lock is being released the owner_ field should no longer be // considered valid. We don't protect writing to has_owner_ here, as it's // the caller's responsibility to ensure that the current thread holds the // mutex when this is called. has_owner_ = false; GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_unlock(&mutex_)); } // Does nothing if the current thread holds the mutex. Otherwise, crashes // with high probability. void AssertHeld() const { GTEST_CHECK_(has_owner_ && pthread_equal(owner_, pthread_self())) << "The current thread is not holding the mutex @" << this; } // A static mutex may be used before main() is entered. It may even // be used before the dynamic initialization stage. Therefore we // must be able to initialize a static mutex object at link time. // This means MutexBase has to be a POD and its member variables // have to be public. public: pthread_mutex_t mutex_; // The underlying pthread mutex. // has_owner_ indicates whether the owner_ field below contains a valid thread // ID and is therefore safe to inspect (e.g., to use in pthread_equal()). All // accesses to the owner_ field should be protected by a check of this field. // An alternative might be to memset() owner_ to all zeros, but there's no // guarantee that a zero'd pthread_t is necessarily invalid or even different // from pthread_self(). bool has_owner_; pthread_t owner_; // The thread holding the mutex. }; // Forward-declares a static mutex. # define GTEST_DECLARE_STATIC_MUTEX_(mutex) \ extern ::testing::internal::MutexBase mutex // Defines and statically (i.e. at link time) initializes a static mutex. // The initialization list here does not explicitly initialize each field, // instead relying on default initialization for the unspecified fields. In // particular, the owner_ field (a pthread_t) is not explicitly initialized. // This allows initialization to work whether pthread_t is a scalar or struct. // The flag -Wmissing-field-initializers must not be specified for this to work. # define GTEST_DEFINE_STATIC_MUTEX_(mutex) \ ::testing::internal::MutexBase mutex = { PTHREAD_MUTEX_INITIALIZER, false } // The Mutex class can only be used for mutexes created at runtime. It // shares its API with MutexBase otherwise. class Mutex : public MutexBase { public: Mutex() { GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL)); has_owner_ = false; } ~Mutex() { GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_destroy(&mutex_)); } private: GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex); }; // We cannot name this class MutexLock because the ctor declaration would // conflict with a macro named MutexLock, which is defined on some // platforms. That macro is used as a defensive measure to prevent against // inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than // "MutexLock l(&mu)". Hence the typedef trick below. class GTestMutexLock { public: explicit GTestMutexLock(MutexBase* mutex) : mutex_(mutex) { mutex_->Lock(); } ~GTestMutexLock() { mutex_->Unlock(); } private: MutexBase* const mutex_; GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock); }; typedef GTestMutexLock MutexLock; // Helpers for ThreadLocal. // pthread_key_create() requires DeleteThreadLocalValue() to have // C-linkage. Therefore it cannot be templatized to access // ThreadLocal. Hence the need for class // ThreadLocalValueHolderBase. class ThreadLocalValueHolderBase { public: virtual ~ThreadLocalValueHolderBase() {} }; // Called by pthread to delete thread-local data stored by // pthread_setspecific(). extern "C" inline void DeleteThreadLocalValue(void* value_holder) { delete static_cast(value_holder); } // Implements thread-local storage on pthreads-based systems. template class ThreadLocal { public: ThreadLocal() : key_(CreateKey()), default_() {} explicit ThreadLocal(const T& value) : key_(CreateKey()), default_(value) {} ~ThreadLocal() { // Destroys the managed object for the current thread, if any. DeleteThreadLocalValue(pthread_getspecific(key_)); // Releases resources associated with the key. This will *not* // delete managed objects for other threads. GTEST_CHECK_POSIX_SUCCESS_(pthread_key_delete(key_)); } T* pointer() { return GetOrCreateValue(); } const T* pointer() const { return GetOrCreateValue(); } const T& get() const { return *pointer(); } void set(const T& value) { *pointer() = value; } private: // Holds a value of type T. class ValueHolder : public ThreadLocalValueHolderBase { public: explicit ValueHolder(const T& value) : value_(value) {} T* pointer() { return &value_; } private: T value_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder); }; static pthread_key_t CreateKey() { pthread_key_t key; // When a thread exits, DeleteThreadLocalValue() will be called on // the object managed for that thread. GTEST_CHECK_POSIX_SUCCESS_( pthread_key_create(&key, &DeleteThreadLocalValue)); return key; } T* GetOrCreateValue() const { ThreadLocalValueHolderBase* const holder = static_cast(pthread_getspecific(key_)); if (holder != NULL) { return CheckedDowncastToActualType(holder)->pointer(); } ValueHolder* const new_holder = new ValueHolder(default_); ThreadLocalValueHolderBase* const holder_base = new_holder; GTEST_CHECK_POSIX_SUCCESS_(pthread_setspecific(key_, holder_base)); return new_holder->pointer(); } // A key pthreads uses for looking up per-thread values. const pthread_key_t key_; const T default_; // The default value for each thread. GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal); }; # endif // OS detection #else // GTEST_IS_THREADSAFE // A dummy implementation of synchronization primitives (mutex, lock, // and thread-local variable). Necessary for compiling Google Test where // mutex is not supported - using Google Test in multiple threads is not // supported on such platforms. class Mutex { public: Mutex() {} void Lock() {} void Unlock() {} void AssertHeld() const {} }; # define GTEST_DECLARE_STATIC_MUTEX_(mutex) \ extern ::testing::internal::Mutex mutex # define GTEST_DEFINE_STATIC_MUTEX_(mutex) ::testing::internal::Mutex mutex // We cannot name this class MutexLock because the ctor declaration would // conflict with a macro named MutexLock, which is defined on some // platforms. That macro is used as a defensive measure to prevent against // inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than // "MutexLock l(&mu)". Hence the typedef trick below. class GTestMutexLock { public: explicit GTestMutexLock(Mutex*) {} // NOLINT }; typedef GTestMutexLock MutexLock; template class ThreadLocal { public: ThreadLocal() : value_() {} explicit ThreadLocal(const T& value) : value_(value) {} T* pointer() { return &value_; } const T* pointer() const { return &value_; } const T& get() const { return value_; } void set(const T& value) { value_ = value; } private: T value_; }; #endif // GTEST_IS_THREADSAFE // Returns the number of threads running in the process, or 0 to indicate that // we cannot detect it. GTEST_API_ size_t GetThreadCount(); // Passing non-POD classes through ellipsis (...) crashes the ARM // compiler and generates a warning in Sun Studio. The Nokia Symbian // and the IBM XL C/C++ compiler try to instantiate a copy constructor // for objects passed through ellipsis (...), failing for uncopyable // objects. We define this to ensure that only POD is passed through // ellipsis on these systems. #if defined(__SYMBIAN32__) || defined(__IBMCPP__) || defined(__SUNPRO_CC) // We lose support for NULL detection where the compiler doesn't like // passing non-POD classes through ellipsis (...). # define GTEST_ELLIPSIS_NEEDS_POD_ 1 #else # define GTEST_CAN_COMPARE_NULL 1 #endif // The Nokia Symbian and IBM XL C/C++ compilers cannot decide between // const T& and const T* in a function template. These compilers // _can_ decide between class template specializations for T and T*, // so a tr1::type_traits-like is_pointer works. #if defined(__SYMBIAN32__) || defined(__IBMCPP__) # define GTEST_NEEDS_IS_POINTER_ 1 #endif template struct bool_constant { typedef bool_constant type; static const bool value = bool_value; }; template const bool bool_constant::value; typedef bool_constant false_type; typedef bool_constant true_type; template struct is_pointer : public false_type {}; template struct is_pointer : public true_type {}; template struct IteratorTraits { typedef typename Iterator::value_type value_type; }; template struct IteratorTraits { typedef T value_type; }; template struct IteratorTraits { typedef T value_type; }; #if GTEST_OS_WINDOWS # define GTEST_PATH_SEP_ "\\" # define GTEST_HAS_ALT_PATH_SEP_ 1 // The biggest signed integer type the compiler supports. typedef __int64 BiggestInt; #else # define GTEST_PATH_SEP_ "/" # define GTEST_HAS_ALT_PATH_SEP_ 0 typedef long long BiggestInt; // NOLINT #endif // GTEST_OS_WINDOWS // Utilities for char. // isspace(int ch) and friends accept an unsigned char or EOF. char // may be signed, depending on the compiler (or compiler flags). // Therefore we need to cast a char to unsigned char before calling // isspace(), etc. inline bool IsAlpha(char ch) { return isalpha(static_cast(ch)) != 0; } inline bool IsAlNum(char ch) { return isalnum(static_cast(ch)) != 0; } inline bool IsDigit(char ch) { return isdigit(static_cast(ch)) != 0; } inline bool IsLower(char ch) { return islower(static_cast(ch)) != 0; } inline bool IsSpace(char ch) { return isspace(static_cast(ch)) != 0; } inline bool IsUpper(char ch) { return isupper(static_cast(ch)) != 0; } inline bool IsXDigit(char ch) { return isxdigit(static_cast(ch)) != 0; } inline bool IsXDigit(wchar_t ch) { const unsigned char low_byte = static_cast(ch); return ch == low_byte && isxdigit(low_byte) != 0; } inline char ToLower(char ch) { return static_cast(tolower(static_cast(ch))); } inline char ToUpper(char ch) { return static_cast(toupper(static_cast(ch))); } inline std::string StripTrailingSpaces(std::string str) { std::string::iterator it = str.end(); while (it != str.begin() && IsSpace(*--it)) it = str.erase(it); return str; } // The testing::internal::posix namespace holds wrappers for common // POSIX functions. These wrappers hide the differences between // Windows/MSVC and POSIX systems. Since some compilers define these // standard functions as macros, the wrapper cannot have the same name // as the wrapped function. namespace posix { // Functions with a different name on Windows. #if GTEST_OS_WINDOWS typedef struct _stat StatStruct; # ifdef __BORLANDC__ inline int IsATTY(int fd) { return isatty(fd); } inline int StrCaseCmp(const char* s1, const char* s2) { return stricmp(s1, s2); } inline char* StrDup(const char* src) { return strdup(src); } # else // !__BORLANDC__ # if GTEST_OS_WINDOWS_MOBILE inline int IsATTY(int /* fd */) { return 0; } # else inline int IsATTY(int fd) { return _isatty(fd); } # endif // GTEST_OS_WINDOWS_MOBILE inline int StrCaseCmp(const char* s1, const char* s2) { return _stricmp(s1, s2); } inline char* StrDup(const char* src) { return _strdup(src); } # endif // __BORLANDC__ # if GTEST_OS_WINDOWS_MOBILE inline int FileNo(FILE* file) { return reinterpret_cast(_fileno(file)); } // Stat(), RmDir(), and IsDir() are not needed on Windows CE at this // time and thus not defined there. # else inline int FileNo(FILE* file) { return _fileno(file); } inline int Stat(const char* path, StatStruct* buf) { return _stat(path, buf); } inline int RmDir(const char* dir) { return _rmdir(dir); } inline bool IsDir(const StatStruct& st) { return (_S_IFDIR & st.st_mode) != 0; } # endif // GTEST_OS_WINDOWS_MOBILE #else typedef struct stat StatStruct; inline int FileNo(FILE* file) { return fileno(file); } inline int IsATTY(int fd) { return isatty(fd); } inline int Stat(const char* path, StatStruct* buf) { return stat(path, buf); } inline int StrCaseCmp(const char* s1, const char* s2) { return strcasecmp(s1, s2); } inline char* StrDup(const char* src) { return strdup(src); } inline int RmDir(const char* dir) { return rmdir(dir); } inline bool IsDir(const StatStruct& st) { return S_ISDIR(st.st_mode); } #endif // GTEST_OS_WINDOWS // Functions deprecated by MSVC 8.0. GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996 /* deprecated function */) inline const char* StrNCpy(char* dest, const char* src, size_t n) { return strncpy(dest, src, n); } // ChDir(), FReopen(), FDOpen(), Read(), Write(), Close(), and // StrError() aren't needed on Windows CE at this time and thus not // defined there. #if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT inline int ChDir(const char* dir) { return chdir(dir); } #endif inline FILE* FOpen(const char* path, const char* mode) { return fopen(path, mode); } #if !GTEST_OS_WINDOWS_MOBILE inline FILE *FReopen(const char* path, const char* mode, FILE* stream) { return freopen(path, mode, stream); } inline FILE* FDOpen(int fd, const char* mode) { return fdopen(fd, mode); } #endif inline int FClose(FILE* fp) { return fclose(fp); } #if !GTEST_OS_WINDOWS_MOBILE inline int Read(int fd, void* buf, unsigned int count) { return static_cast(read(fd, buf, count)); } inline int Write(int fd, const void* buf, unsigned int count) { return static_cast(write(fd, buf, count)); } inline int Close(int fd) { return close(fd); } inline const char* StrError(int errnum) { return strerror(errnum); } #endif inline const char* GetEnv(const char* name) { #if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE | GTEST_OS_WINDOWS_RT // We are on Windows CE, which has no environment variables. static_cast(name); // To prevent 'unused argument' warning. return NULL; #elif defined(__BORLANDC__) || defined(__SunOS_5_8) || defined(__SunOS_5_9) // Environment variables which we programmatically clear will be set to the // empty string rather than unset (NULL). Handle that case. const char* const env = getenv(name); return (env != NULL && env[0] != '\0') ? env : NULL; #else return getenv(name); #endif } GTEST_DISABLE_MSC_WARNINGS_POP_() #if GTEST_OS_WINDOWS_MOBILE // Windows CE has no C library. The abort() function is used in // several places in Google Test. This implementation provides a reasonable // imitation of standard behaviour. void Abort(); #else inline void Abort() { abort(); } #endif // GTEST_OS_WINDOWS_MOBILE } // namespace posix // MSVC "deprecates" snprintf and issues warnings wherever it is used. In // order to avoid these warnings, we need to use _snprintf or _snprintf_s on // MSVC-based platforms. We map the GTEST_SNPRINTF_ macro to the appropriate // function in order to achieve that. We use macro definition here because // snprintf is a variadic function. #if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE // MSVC 2005 and above support variadic macros. # define GTEST_SNPRINTF_(buffer, size, format, ...) \ _snprintf_s(buffer, size, size, format, __VA_ARGS__) #elif defined(_MSC_VER) // Windows CE does not define _snprintf_s and MSVC prior to 2005 doesn't // complain about _snprintf. # define GTEST_SNPRINTF_ _snprintf #else # define GTEST_SNPRINTF_ snprintf #endif // The maximum number a BiggestInt can represent. This definition // works no matter BiggestInt is represented in one's complement or // two's complement. // // We cannot rely on numeric_limits in STL, as __int64 and long long // are not part of standard C++ and numeric_limits doesn't need to be // defined for them. const BiggestInt kMaxBiggestInt = ~(static_cast(1) << (8*sizeof(BiggestInt) - 1)); // This template class serves as a compile-time function from size to // type. It maps a size in bytes to a primitive type with that // size. e.g. // // TypeWithSize<4>::UInt // // is typedef-ed to be unsigned int (unsigned integer made up of 4 // bytes). // // Such functionality should belong to STL, but I cannot find it // there. // // Google Test uses this class in the implementation of floating-point // comparison. // // For now it only handles UInt (unsigned int) as that's all Google Test // needs. Other types can be easily added in the future if need // arises. template class TypeWithSize { public: // This prevents the user from using TypeWithSize with incorrect // values of N. typedef void UInt; }; // The specialization for size 4. template <> class TypeWithSize<4> { public: // unsigned int has size 4 in both gcc and MSVC. // // As base/basictypes.h doesn't compile on Windows, we cannot use // uint32, uint64, and etc here. typedef int Int; typedef unsigned int UInt; }; // The specialization for size 8. template <> class TypeWithSize<8> { public: #if GTEST_OS_WINDOWS typedef __int64 Int; typedef unsigned __int64 UInt; #else typedef long long Int; // NOLINT typedef unsigned long long UInt; // NOLINT #endif // GTEST_OS_WINDOWS }; // Integer types of known sizes. typedef TypeWithSize<4>::Int Int32; typedef TypeWithSize<4>::UInt UInt32; typedef TypeWithSize<8>::Int Int64; typedef TypeWithSize<8>::UInt UInt64; typedef TypeWithSize<8>::Int TimeInMillis; // Represents time in milliseconds. // Utilities for command line flags and environment variables. // Macro for referencing flags. #define GTEST_FLAG(name) FLAGS_gtest_##name // Macros for declaring flags. #define GTEST_DECLARE_bool_(name) GTEST_API_ extern bool GTEST_FLAG(name) #define GTEST_DECLARE_int32_(name) \ GTEST_API_ extern ::testing::internal::Int32 GTEST_FLAG(name) #define GTEST_DECLARE_string_(name) \ GTEST_API_ extern ::std::string GTEST_FLAG(name) // Macros for defining flags. #define GTEST_DEFINE_bool_(name, default_val, doc) \ GTEST_API_ bool GTEST_FLAG(name) = (default_val) #define GTEST_DEFINE_int32_(name, default_val, doc) \ GTEST_API_ ::testing::internal::Int32 GTEST_FLAG(name) = (default_val) #define GTEST_DEFINE_string_(name, default_val, doc) \ GTEST_API_ ::std::string GTEST_FLAG(name) = (default_val) // Thread annotations #define GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks) #define GTEST_LOCK_EXCLUDED_(locks) // Parses 'str' for a 32-bit signed integer. If successful, writes the result // to *value and returns true; otherwise leaves *value unchanged and returns // false. // TODO(chandlerc): Find a better way to refactor flag and environment parsing // out of both gtest-port.cc and gtest.cc to avoid exporting this utility // function. bool ParseInt32(const Message& src_text, const char* str, Int32* value); // Parses a bool/Int32/string from the environment variable // corresponding to the given Google Test flag. bool BoolFromGTestEnv(const char* flag, bool default_val); GTEST_API_ Int32 Int32FromGTestEnv(const char* flag, Int32 default_val); const char* StringFromGTestEnv(const char* flag, const char* default_val); } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_ #if GTEST_OS_LINUX # include # include # include # include #endif // GTEST_OS_LINUX #if GTEST_HAS_EXCEPTIONS # include #endif #include #include #include #include #include #include #include #include // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // // The Google C++ Testing Framework (Google Test) // // This header file defines the Message class. // // IMPORTANT NOTE: Due to limitation of the C++ language, we have to // leave some internal implementation details in this header file. // They are clearly marked by comments like this: // // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. // // Such code is NOT meant to be used by a user directly, and is subject // to CHANGE WITHOUT NOTICE. Therefore DO NOT DEPEND ON IT in a user // program! #ifndef GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_ #define GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_ #include // Ensures that there is at least one operator<< in the global namespace. // See Message& operator<<(...) below for why. void operator<<(const testing::internal::Secret&, int); namespace testing { // The Message class works like an ostream repeater. // // Typical usage: // // 1. You stream a bunch of values to a Message object. // It will remember the text in a stringstream. // 2. Then you stream the Message object to an ostream. // This causes the text in the Message to be streamed // to the ostream. // // For example; // // testing::Message foo; // foo << 1 << " != " << 2; // std::cout << foo; // // will print "1 != 2". // // Message is not intended to be inherited from. In particular, its // destructor is not virtual. // // Note that stringstream behaves differently in gcc and in MSVC. You // can stream a NULL char pointer to it in the former, but not in the // latter (it causes an access violation if you do). The Message // class hides this difference by treating a NULL char pointer as // "(null)". class GTEST_API_ Message { private: // The type of basic IO manipulators (endl, ends, and flush) for // narrow streams. typedef std::ostream& (*BasicNarrowIoManip)(std::ostream&); public: // Constructs an empty Message. Message(); // Copy constructor. Message(const Message& msg) : ss_(new ::std::stringstream) { // NOLINT *ss_ << msg.GetString(); } // Constructs a Message from a C-string. explicit Message(const char* str) : ss_(new ::std::stringstream) { *ss_ << str; } #if GTEST_OS_SYMBIAN // Streams a value (either a pointer or not) to this object. template inline Message& operator <<(const T& value) { StreamHelper(typename internal::is_pointer::type(), value); return *this; } #else // Streams a non-pointer value to this object. template inline Message& operator <<(const T& val) { // Some libraries overload << for STL containers. These // overloads are defined in the global namespace instead of ::std. // // C++'s symbol lookup rule (i.e. Koenig lookup) says that these // overloads are visible in either the std namespace or the global // namespace, but not other namespaces, including the testing // namespace which Google Test's Message class is in. // // To allow STL containers (and other types that has a << operator // defined in the global namespace) to be used in Google Test // assertions, testing::Message must access the custom << operator // from the global namespace. With this using declaration, // overloads of << defined in the global namespace and those // visible via Koenig lookup are both exposed in this function. using ::operator <<; *ss_ << val; return *this; } // Streams a pointer value to this object. // // This function is an overload of the previous one. When you // stream a pointer to a Message, this definition will be used as it // is more specialized. (The C++ Standard, section // [temp.func.order].) If you stream a non-pointer, then the // previous definition will be used. // // The reason for this overload is that streaming a NULL pointer to // ostream is undefined behavior. Depending on the compiler, you // may get "0", "(nil)", "(null)", or an access violation. To // ensure consistent result across compilers, we always treat NULL // as "(null)". template inline Message& operator <<(T* const& pointer) { // NOLINT if (pointer == NULL) { *ss_ << "(null)"; } else { *ss_ << pointer; } return *this; } #endif // GTEST_OS_SYMBIAN // Since the basic IO manipulators are overloaded for both narrow // and wide streams, we have to provide this specialized definition // of operator <<, even though its body is the same as the // templatized version above. Without this definition, streaming // endl or other basic IO manipulators to Message will confuse the // compiler. Message& operator <<(BasicNarrowIoManip val) { *ss_ << val; return *this; } // Instead of 1/0, we want to see true/false for bool values. Message& operator <<(bool b) { return *this << (b ? "true" : "false"); } // These two overloads allow streaming a wide C string to a Message // using the UTF-8 encoding. Message& operator <<(const wchar_t* wide_c_str); Message& operator <<(wchar_t* wide_c_str); #if GTEST_HAS_STD_WSTRING // Converts the given wide string to a narrow string using the UTF-8 // encoding, and streams the result to this Message object. Message& operator <<(const ::std::wstring& wstr); #endif // GTEST_HAS_STD_WSTRING #if GTEST_HAS_GLOBAL_WSTRING // Converts the given wide string to a narrow string using the UTF-8 // encoding, and streams the result to this Message object. Message& operator <<(const ::wstring& wstr); #endif // GTEST_HAS_GLOBAL_WSTRING // Gets the text streamed to this object so far as an std::string. // Each '\0' character in the buffer is replaced with "\\0". // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. std::string GetString() const; private: #if GTEST_OS_SYMBIAN // These are needed as the Nokia Symbian Compiler cannot decide between // const T& and const T* in a function template. The Nokia compiler _can_ // decide between class template specializations for T and T*, so a // tr1::type_traits-like is_pointer works, and we can overload on that. template inline void StreamHelper(internal::true_type /*is_pointer*/, T* pointer) { if (pointer == NULL) { *ss_ << "(null)"; } else { *ss_ << pointer; } } template inline void StreamHelper(internal::false_type /*is_pointer*/, const T& value) { // See the comments in Message& operator <<(const T&) above for why // we need this using statement. using ::operator <<; *ss_ << value; } #endif // GTEST_OS_SYMBIAN // We'll hold the text streamed to this object here. const internal::scoped_ptr< ::std::stringstream> ss_; // We declare (but don't implement) this to prevent the compiler // from implementing the assignment operator. void operator=(const Message&); }; // Streams a Message to an ostream. inline std::ostream& operator <<(std::ostream& os, const Message& sb) { return os << sb.GetString(); } namespace internal { // Converts a streamable value to an std::string. A NULL pointer is // converted to "(null)". When the input value is a ::string, // ::std::string, ::wstring, or ::std::wstring object, each NUL // character in it is replaced with "\\0". template std::string StreamableToString(const T& streamable) { return (Message() << streamable).GetString(); } } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_ // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee) // // The Google C++ Testing Framework (Google Test) // // This header file declares the String class and functions used internally by // Google Test. They are subject to change without notice. They should not used // by code external to Google Test. // // This header file is #included by . // It should not be #included by other files. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_ #ifdef __BORLANDC__ // string.h is not guaranteed to provide strcpy on C++ Builder. # include #endif #include #include namespace testing { namespace internal { // String - an abstract class holding static string utilities. class GTEST_API_ String { public: // Static utility methods // Clones a 0-terminated C string, allocating memory using new. The // caller is responsible for deleting the return value using // delete[]. Returns the cloned string, or NULL if the input is // NULL. // // This is different from strdup() in string.h, which allocates // memory using malloc(). static const char* CloneCString(const char* c_str); #if GTEST_OS_WINDOWS_MOBILE // Windows CE does not have the 'ANSI' versions of Win32 APIs. To be // able to pass strings to Win32 APIs on CE we need to convert them // to 'Unicode', UTF-16. // Creates a UTF-16 wide string from the given ANSI string, allocating // memory using new. The caller is responsible for deleting the return // value using delete[]. Returns the wide string, or NULL if the // input is NULL. // // The wide string is created using the ANSI codepage (CP_ACP) to // match the behaviour of the ANSI versions of Win32 calls and the // C runtime. static LPCWSTR AnsiToUtf16(const char* c_str); // Creates an ANSI string from the given wide string, allocating // memory using new. The caller is responsible for deleting the return // value using delete[]. Returns the ANSI string, or NULL if the // input is NULL. // // The returned string is created using the ANSI codepage (CP_ACP) to // match the behaviour of the ANSI versions of Win32 calls and the // C runtime. static const char* Utf16ToAnsi(LPCWSTR utf16_str); #endif // Compares two C strings. Returns true iff they have the same content. // // Unlike strcmp(), this function can handle NULL argument(s). A // NULL C string is considered different to any non-NULL C string, // including the empty string. static bool CStringEquals(const char* lhs, const char* rhs); // Converts a wide C string to a String using the UTF-8 encoding. // NULL will be converted to "(null)". If an error occurred during // the conversion, "(failed to convert from wide string)" is // returned. static std::string ShowWideCString(const wchar_t* wide_c_str); // Compares two wide C strings. Returns true iff they have the same // content. // // Unlike wcscmp(), this function can handle NULL argument(s). A // NULL C string is considered different to any non-NULL C string, // including the empty string. static bool WideCStringEquals(const wchar_t* lhs, const wchar_t* rhs); // Compares two C strings, ignoring case. Returns true iff they // have the same content. // // Unlike strcasecmp(), this function can handle NULL argument(s). // A NULL C string is considered different to any non-NULL C string, // including the empty string. static bool CaseInsensitiveCStringEquals(const char* lhs, const char* rhs); // Compares two wide C strings, ignoring case. Returns true iff they // have the same content. // // Unlike wcscasecmp(), this function can handle NULL argument(s). // A NULL C string is considered different to any non-NULL wide C string, // including the empty string. // NB: The implementations on different platforms slightly differ. // On windows, this method uses _wcsicmp which compares according to LC_CTYPE // environment variable. On GNU platform this method uses wcscasecmp // which compares according to LC_CTYPE category of the current locale. // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the // current locale. static bool CaseInsensitiveWideCStringEquals(const wchar_t* lhs, const wchar_t* rhs); // Returns true iff the given string ends with the given suffix, ignoring // case. Any string is considered to end with an empty suffix. static bool EndsWithCaseInsensitive( const std::string& str, const std::string& suffix); // Formats an int value as "%02d". static std::string FormatIntWidth2(int value); // "%02d" for width == 2 // Formats an int value as "%X". static std::string FormatHexInt(int value); // Formats a byte as "%02X". static std::string FormatByte(unsigned char value); private: String(); // Not meant to be instantiated. }; // class String // Gets the content of the stringstream's buffer as an std::string. Each '\0' // character in the buffer is replaced with "\\0". GTEST_API_ std::string StringStreamToString(::std::stringstream* stream); } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_ // Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: keith.ray@gmail.com (Keith Ray) // // Google Test filepath utilities // // This header file declares classes and functions used internally by // Google Test. They are subject to change without notice. // // This file is #included in . // Do not include this header file separately! #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_ namespace testing { namespace internal { // FilePath - a class for file and directory pathname manipulation which // handles platform-specific conventions (like the pathname separator). // Used for helper functions for naming files in a directory for xml output. // Except for Set methods, all methods are const or static, which provides an // "immutable value object" -- useful for peace of mind. // A FilePath with a value ending in a path separator ("like/this/") represents // a directory, otherwise it is assumed to represent a file. In either case, // it may or may not represent an actual file or directory in the file system. // Names are NOT checked for syntax correctness -- no checking for illegal // characters, malformed paths, etc. class GTEST_API_ FilePath { public: FilePath() : pathname_("") { } FilePath(const FilePath& rhs) : pathname_(rhs.pathname_) { } explicit FilePath(const std::string& pathname) : pathname_(pathname) { Normalize(); } FilePath& operator=(const FilePath& rhs) { Set(rhs); return *this; } void Set(const FilePath& rhs) { pathname_ = rhs.pathname_; } const std::string& string() const { return pathname_; } const char* c_str() const { return pathname_.c_str(); } // Returns the current working directory, or "" if unsuccessful. static FilePath GetCurrentDir(); // Given directory = "dir", base_name = "test", number = 0, // extension = "xml", returns "dir/test.xml". If number is greater // than zero (e.g., 12), returns "dir/test_12.xml". // On Windows platform, uses \ as the separator rather than /. static FilePath MakeFileName(const FilePath& directory, const FilePath& base_name, int number, const char* extension); // Given directory = "dir", relative_path = "test.xml", // returns "dir/test.xml". // On Windows, uses \ as the separator rather than /. static FilePath ConcatPaths(const FilePath& directory, const FilePath& relative_path); // Returns a pathname for a file that does not currently exist. The pathname // will be directory/base_name.extension or // directory/base_name_.extension if directory/base_name.extension // already exists. The number will be incremented until a pathname is found // that does not already exist. // Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'. // There could be a race condition if two or more processes are calling this // function at the same time -- they could both pick the same filename. static FilePath GenerateUniqueFileName(const FilePath& directory, const FilePath& base_name, const char* extension); // Returns true iff the path is "". bool IsEmpty() const { return pathname_.empty(); } // If input name has a trailing separator character, removes it and returns // the name, otherwise return the name string unmodified. // On Windows platform, uses \ as the separator, other platforms use /. FilePath RemoveTrailingPathSeparator() const; // Returns a copy of the FilePath with the directory part removed. // Example: FilePath("path/to/file").RemoveDirectoryName() returns // FilePath("file"). If there is no directory part ("just_a_file"), it returns // the FilePath unmodified. If there is no file part ("just_a_dir/") it // returns an empty FilePath (""). // On Windows platform, '\' is the path separator, otherwise it is '/'. FilePath RemoveDirectoryName() const; // RemoveFileName returns the directory path with the filename removed. // Example: FilePath("path/to/file").RemoveFileName() returns "path/to/". // If the FilePath is "a_file" or "/a_file", RemoveFileName returns // FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does // not have a file, like "just/a/dir/", it returns the FilePath unmodified. // On Windows platform, '\' is the path separator, otherwise it is '/'. FilePath RemoveFileName() const; // Returns a copy of the FilePath with the case-insensitive extension removed. // Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns // FilePath("dir/file"). If a case-insensitive extension is not // found, returns a copy of the original FilePath. FilePath RemoveExtension(const char* extension) const; // Creates directories so that path exists. Returns true if successful or if // the directories already exist; returns false if unable to create // directories for any reason. Will also return false if the FilePath does // not represent a directory (that is, it doesn't end with a path separator). bool CreateDirectoriesRecursively() const; // Create the directory so that path exists. Returns true if successful or // if the directory already exists; returns false if unable to create the // directory for any reason, including if the parent directory does not // exist. Not named "CreateDirectory" because that's a macro on Windows. bool CreateFolder() const; // Returns true if FilePath describes something in the file-system, // either a file, directory, or whatever, and that something exists. bool FileOrDirectoryExists() const; // Returns true if pathname describes a directory in the file-system // that exists. bool DirectoryExists() const; // Returns true if FilePath ends with a path separator, which indicates that // it is intended to represent a directory. Returns false otherwise. // This does NOT check that a directory (or file) actually exists. bool IsDirectory() const; // Returns true if pathname describes a root directory. (Windows has one // root directory per disk drive.) bool IsRootDirectory() const; // Returns true if pathname describes an absolute path. bool IsAbsolutePath() const; private: // Replaces multiple consecutive separators with a single separator. // For example, "bar///foo" becomes "bar/foo". Does not eliminate other // redundancies that might be in a pathname involving "." or "..". // // A pathname with multiple consecutive separators may occur either through // user error or as a result of some scripts or APIs that generate a pathname // with a trailing separator. On other platforms the same API or script // may NOT generate a pathname with a trailing "/". Then elsewhere that // pathname may have another "/" and pathname components added to it, // without checking for the separator already being there. // The script language and operating system may allow paths like "foo//bar" // but some of the functions in FilePath will not handle that correctly. In // particular, RemoveTrailingPathSeparator() only removes one separator, and // it is called in CreateDirectoriesRecursively() assuming that it will change // a pathname from directory syntax (trailing separator) to filename syntax. // // On Windows this method also replaces the alternate path separator '/' with // the primary path separator '\\', so that for example "bar\\/\\foo" becomes // "bar\\foo". void Normalize(); // Returns a pointer to the last occurence of a valid path separator in // the FilePath. On Windows, for example, both '/' and '\' are valid path // separators. Returns NULL if no path separator was found. const char* FindLastPathSeparator() const; std::string pathname_; }; // class FilePath } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_ // This file was GENERATED by command: // pump.py gtest-type-util.h.pump // DO NOT EDIT BY HAND!!! // Copyright 2008 Google Inc. // All Rights Reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // Type utilities needed for implementing typed and type-parameterized // tests. This file is generated by a SCRIPT. DO NOT EDIT BY HAND! // // Currently we support at most 50 types in a list, and at most 50 // type-parameterized tests in one type-parameterized test case. // Please contact googletestframework@googlegroups.com if you need // more. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ // #ifdef __GNUC__ is too general here. It is possible to use gcc without using // libstdc++ (which is where cxxabi.h comes from). # if GTEST_HAS_CXXABI_H_ # include # elif defined(__HP_aCC) # include # endif // GTEST_HASH_CXXABI_H_ namespace testing { namespace internal { // GetTypeName() returns a human-readable name of type T. // NB: This function is also used in Google Mock, so don't move it inside of // the typed-test-only section below. template std::string GetTypeName() { # if GTEST_HAS_RTTI const char* const name = typeid(T).name(); # if GTEST_HAS_CXXABI_H_ || defined(__HP_aCC) int status = 0; // gcc's implementation of typeid(T).name() mangles the type name, // so we have to demangle it. # if GTEST_HAS_CXXABI_H_ using abi::__cxa_demangle; # endif // GTEST_HAS_CXXABI_H_ char* const readable_name = __cxa_demangle(name, 0, 0, &status); const std::string name_str(status == 0 ? readable_name : name); free(readable_name); return name_str; # else return name; # endif // GTEST_HAS_CXXABI_H_ || __HP_aCC # else return ""; # endif // GTEST_HAS_RTTI } #if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P // AssertyTypeEq::type is defined iff T1 and T2 are the same // type. This can be used as a compile-time assertion to ensure that // two types are equal. template struct AssertTypeEq; template struct AssertTypeEq { typedef bool type; }; // A unique type used as the default value for the arguments of class // template Types. This allows us to simulate variadic templates // (e.g. Types, Type, and etc), which C++ doesn't // support directly. struct None {}; // The following family of struct and struct templates are used to // represent type lists. In particular, TypesN // represents a type list with N types (T1, T2, ..., and TN) in it. // Except for Types0, every struct in the family has two member types: // Head for the first type in the list, and Tail for the rest of the // list. // The empty type list. struct Types0 {}; // Type lists of length 1, 2, 3, and so on. template struct Types1 { typedef T1 Head; typedef Types0 Tail; }; template struct Types2 { typedef T1 Head; typedef Types1 Tail; }; template struct Types3 { typedef T1 Head; typedef Types2 Tail; }; template struct Types4 { typedef T1 Head; typedef Types3 Tail; }; template struct Types5 { typedef T1 Head; typedef Types4 Tail; }; template struct Types6 { typedef T1 Head; typedef Types5 Tail; }; template struct Types7 { typedef T1 Head; typedef Types6 Tail; }; template struct Types8 { typedef T1 Head; typedef Types7 Tail; }; template struct Types9 { typedef T1 Head; typedef Types8 Tail; }; template struct Types10 { typedef T1 Head; typedef Types9 Tail; }; template struct Types11 { typedef T1 Head; typedef Types10 Tail; }; template struct Types12 { typedef T1 Head; typedef Types11 Tail; }; template struct Types13 { typedef T1 Head; typedef Types12 Tail; }; template struct Types14 { typedef T1 Head; typedef Types13 Tail; }; template struct Types15 { typedef T1 Head; typedef Types14 Tail; }; template struct Types16 { typedef T1 Head; typedef Types15 Tail; }; template struct Types17 { typedef T1 Head; typedef Types16 Tail; }; template struct Types18 { typedef T1 Head; typedef Types17 Tail; }; template struct Types19 { typedef T1 Head; typedef Types18 Tail; }; template struct Types20 { typedef T1 Head; typedef Types19 Tail; }; template struct Types21 { typedef T1 Head; typedef Types20 Tail; }; template struct Types22 { typedef T1 Head; typedef Types21 Tail; }; template struct Types23 { typedef T1 Head; typedef Types22 Tail; }; template struct Types24 { typedef T1 Head; typedef Types23 Tail; }; template struct Types25 { typedef T1 Head; typedef Types24 Tail; }; template struct Types26 { typedef T1 Head; typedef Types25 Tail; }; template struct Types27 { typedef T1 Head; typedef Types26 Tail; }; template struct Types28 { typedef T1 Head; typedef Types27 Tail; }; template struct Types29 { typedef T1 Head; typedef Types28 Tail; }; template struct Types30 { typedef T1 Head; typedef Types29 Tail; }; template struct Types31 { typedef T1 Head; typedef Types30 Tail; }; template struct Types32 { typedef T1 Head; typedef Types31 Tail; }; template struct Types33 { typedef T1 Head; typedef Types32 Tail; }; template struct Types34 { typedef T1 Head; typedef Types33 Tail; }; template struct Types35 { typedef T1 Head; typedef Types34 Tail; }; template struct Types36 { typedef T1 Head; typedef Types35 Tail; }; template struct Types37 { typedef T1 Head; typedef Types36 Tail; }; template struct Types38 { typedef T1 Head; typedef Types37 Tail; }; template struct Types39 { typedef T1 Head; typedef Types38 Tail; }; template struct Types40 { typedef T1 Head; typedef Types39 Tail; }; template struct Types41 { typedef T1 Head; typedef Types40 Tail; }; template struct Types42 { typedef T1 Head; typedef Types41 Tail; }; template struct Types43 { typedef T1 Head; typedef Types42 Tail; }; template struct Types44 { typedef T1 Head; typedef Types43 Tail; }; template struct Types45 { typedef T1 Head; typedef Types44 Tail; }; template struct Types46 { typedef T1 Head; typedef Types45 Tail; }; template struct Types47 { typedef T1 Head; typedef Types46 Tail; }; template struct Types48 { typedef T1 Head; typedef Types47 Tail; }; template struct Types49 { typedef T1 Head; typedef Types48 Tail; }; template struct Types50 { typedef T1 Head; typedef Types49 Tail; }; } // namespace internal // We don't want to require the users to write TypesN<...> directly, // as that would require them to count the length. Types<...> is much // easier to write, but generates horrible messages when there is a // compiler error, as gcc insists on printing out each template // argument, even if it has the default value (this means Types // will appear as Types in the compiler // errors). // // Our solution is to combine the best part of the two approaches: a // user would write Types, and Google Test will translate // that to TypesN internally to make error messages // readable. The translation is done by the 'type' member of the // Types template. template struct Types { typedef internal::Types50 type; }; template <> struct Types { typedef internal::Types0 type; }; template struct Types { typedef internal::Types1 type; }; template struct Types { typedef internal::Types2 type; }; template struct Types { typedef internal::Types3 type; }; template struct Types { typedef internal::Types4 type; }; template struct Types { typedef internal::Types5 type; }; template struct Types { typedef internal::Types6 type; }; template struct Types { typedef internal::Types7 type; }; template struct Types { typedef internal::Types8 type; }; template struct Types { typedef internal::Types9 type; }; template struct Types { typedef internal::Types10 type; }; template struct Types { typedef internal::Types11 type; }; template struct Types { typedef internal::Types12 type; }; template struct Types { typedef internal::Types13 type; }; template struct Types { typedef internal::Types14 type; }; template struct Types { typedef internal::Types15 type; }; template struct Types { typedef internal::Types16 type; }; template struct Types { typedef internal::Types17 type; }; template struct Types { typedef internal::Types18 type; }; template struct Types { typedef internal::Types19 type; }; template struct Types { typedef internal::Types20 type; }; template struct Types { typedef internal::Types21 type; }; template struct Types { typedef internal::Types22 type; }; template struct Types { typedef internal::Types23 type; }; template struct Types { typedef internal::Types24 type; }; template struct Types { typedef internal::Types25 type; }; template struct Types { typedef internal::Types26 type; }; template struct Types { typedef internal::Types27 type; }; template struct Types { typedef internal::Types28 type; }; template struct Types { typedef internal::Types29 type; }; template struct Types { typedef internal::Types30 type; }; template struct Types { typedef internal::Types31 type; }; template struct Types { typedef internal::Types32 type; }; template struct Types { typedef internal::Types33 type; }; template struct Types { typedef internal::Types34 type; }; template struct Types { typedef internal::Types35 type; }; template struct Types { typedef internal::Types36 type; }; template struct Types { typedef internal::Types37 type; }; template struct Types { typedef internal::Types38 type; }; template struct Types { typedef internal::Types39 type; }; template struct Types { typedef internal::Types40 type; }; template struct Types { typedef internal::Types41 type; }; template struct Types { typedef internal::Types42 type; }; template struct Types { typedef internal::Types43 type; }; template struct Types { typedef internal::Types44 type; }; template struct Types { typedef internal::Types45 type; }; template struct Types { typedef internal::Types46 type; }; template struct Types { typedef internal::Types47 type; }; template struct Types { typedef internal::Types48 type; }; template struct Types { typedef internal::Types49 type; }; namespace internal { # define GTEST_TEMPLATE_ template class // The template "selector" struct TemplateSel is used to // represent Tmpl, which must be a class template with one type // parameter, as a type. TemplateSel::Bind::type is defined // as the type Tmpl. This allows us to actually instantiate the // template "selected" by TemplateSel. // // This trick is necessary for simulating typedef for class templates, // which C++ doesn't support directly. template struct TemplateSel { template struct Bind { typedef Tmpl type; }; }; # define GTEST_BIND_(TmplSel, T) \ TmplSel::template Bind::type // A unique struct template used as the default value for the // arguments of class template Templates. This allows us to simulate // variadic templates (e.g. Templates, Templates, // and etc), which C++ doesn't support directly. template struct NoneT {}; // The following family of struct and struct templates are used to // represent template lists. In particular, TemplatesN represents a list of N templates (T1, T2, ..., and TN). Except // for Templates0, every struct in the family has two member types: // Head for the selector of the first template in the list, and Tail // for the rest of the list. // The empty template list. struct Templates0 {}; // Template lists of length 1, 2, 3, and so on. template struct Templates1 { typedef TemplateSel Head; typedef Templates0 Tail; }; template struct Templates2 { typedef TemplateSel Head; typedef Templates1 Tail; }; template struct Templates3 { typedef TemplateSel Head; typedef Templates2 Tail; }; template struct Templates4 { typedef TemplateSel Head; typedef Templates3 Tail; }; template struct Templates5 { typedef TemplateSel Head; typedef Templates4 Tail; }; template struct Templates6 { typedef TemplateSel Head; typedef Templates5 Tail; }; template struct Templates7 { typedef TemplateSel Head; typedef Templates6 Tail; }; template struct Templates8 { typedef TemplateSel Head; typedef Templates7 Tail; }; template struct Templates9 { typedef TemplateSel Head; typedef Templates8 Tail; }; template struct Templates10 { typedef TemplateSel Head; typedef Templates9 Tail; }; template struct Templates11 { typedef TemplateSel Head; typedef Templates10 Tail; }; template struct Templates12 { typedef TemplateSel Head; typedef Templates11 Tail; }; template struct Templates13 { typedef TemplateSel Head; typedef Templates12 Tail; }; template struct Templates14 { typedef TemplateSel Head; typedef Templates13 Tail; }; template struct Templates15 { typedef TemplateSel Head; typedef Templates14 Tail; }; template struct Templates16 { typedef TemplateSel Head; typedef Templates15 Tail; }; template struct Templates17 { typedef TemplateSel Head; typedef Templates16 Tail; }; template struct Templates18 { typedef TemplateSel Head; typedef Templates17 Tail; }; template struct Templates19 { typedef TemplateSel Head; typedef Templates18 Tail; }; template struct Templates20 { typedef TemplateSel Head; typedef Templates19 Tail; }; template struct Templates21 { typedef TemplateSel Head; typedef Templates20 Tail; }; template struct Templates22 { typedef TemplateSel Head; typedef Templates21 Tail; }; template struct Templates23 { typedef TemplateSel Head; typedef Templates22 Tail; }; template struct Templates24 { typedef TemplateSel Head; typedef Templates23 Tail; }; template struct Templates25 { typedef TemplateSel Head; typedef Templates24 Tail; }; template struct Templates26 { typedef TemplateSel Head; typedef Templates25 Tail; }; template struct Templates27 { typedef TemplateSel Head; typedef Templates26 Tail; }; template struct Templates28 { typedef TemplateSel Head; typedef Templates27 Tail; }; template struct Templates29 { typedef TemplateSel Head; typedef Templates28 Tail; }; template struct Templates30 { typedef TemplateSel Head; typedef Templates29 Tail; }; template struct Templates31 { typedef TemplateSel Head; typedef Templates30 Tail; }; template struct Templates32 { typedef TemplateSel Head; typedef Templates31 Tail; }; template struct Templates33 { typedef TemplateSel Head; typedef Templates32 Tail; }; template struct Templates34 { typedef TemplateSel Head; typedef Templates33 Tail; }; template struct Templates35 { typedef TemplateSel Head; typedef Templates34 Tail; }; template struct Templates36 { typedef TemplateSel Head; typedef Templates35 Tail; }; template struct Templates37 { typedef TemplateSel Head; typedef Templates36 Tail; }; template struct Templates38 { typedef TemplateSel Head; typedef Templates37 Tail; }; template struct Templates39 { typedef TemplateSel Head; typedef Templates38 Tail; }; template struct Templates40 { typedef TemplateSel Head; typedef Templates39 Tail; }; template struct Templates41 { typedef TemplateSel Head; typedef Templates40 Tail; }; template struct Templates42 { typedef TemplateSel Head; typedef Templates41 Tail; }; template struct Templates43 { typedef TemplateSel Head; typedef Templates42 Tail; }; template struct Templates44 { typedef TemplateSel Head; typedef Templates43 Tail; }; template struct Templates45 { typedef TemplateSel Head; typedef Templates44 Tail; }; template struct Templates46 { typedef TemplateSel Head; typedef Templates45 Tail; }; template struct Templates47 { typedef TemplateSel Head; typedef Templates46 Tail; }; template struct Templates48 { typedef TemplateSel Head; typedef Templates47 Tail; }; template struct Templates49 { typedef TemplateSel Head; typedef Templates48 Tail; }; template struct Templates50 { typedef TemplateSel Head; typedef Templates49 Tail; }; // We don't want to require the users to write TemplatesN<...> directly, // as that would require them to count the length. Templates<...> is much // easier to write, but generates horrible messages when there is a // compiler error, as gcc insists on printing out each template // argument, even if it has the default value (this means Templates // will appear as Templates in the compiler // errors). // // Our solution is to combine the best part of the two approaches: a // user would write Templates, and Google Test will translate // that to TemplatesN internally to make error messages // readable. The translation is done by the 'type' member of the // Templates template. template struct Templates { typedef Templates50 type; }; template <> struct Templates { typedef Templates0 type; }; template struct Templates { typedef Templates1 type; }; template struct Templates { typedef Templates2 type; }; template struct Templates { typedef Templates3 type; }; template struct Templates { typedef Templates4 type; }; template struct Templates { typedef Templates5 type; }; template struct Templates { typedef Templates6 type; }; template struct Templates { typedef Templates7 type; }; template struct Templates { typedef Templates8 type; }; template struct Templates { typedef Templates9 type; }; template struct Templates { typedef Templates10 type; }; template struct Templates { typedef Templates11 type; }; template struct Templates { typedef Templates12 type; }; template struct Templates { typedef Templates13 type; }; template struct Templates { typedef Templates14 type; }; template struct Templates { typedef Templates15 type; }; template struct Templates { typedef Templates16 type; }; template struct Templates { typedef Templates17 type; }; template struct Templates { typedef Templates18 type; }; template struct Templates { typedef Templates19 type; }; template struct Templates { typedef Templates20 type; }; template struct Templates { typedef Templates21 type; }; template struct Templates { typedef Templates22 type; }; template struct Templates { typedef Templates23 type; }; template struct Templates { typedef Templates24 type; }; template struct Templates { typedef Templates25 type; }; template struct Templates { typedef Templates26 type; }; template struct Templates { typedef Templates27 type; }; template struct Templates { typedef Templates28 type; }; template struct Templates { typedef Templates29 type; }; template struct Templates { typedef Templates30 type; }; template struct Templates { typedef Templates31 type; }; template struct Templates { typedef Templates32 type; }; template struct Templates { typedef Templates33 type; }; template struct Templates { typedef Templates34 type; }; template struct Templates { typedef Templates35 type; }; template struct Templates { typedef Templates36 type; }; template struct Templates { typedef Templates37 type; }; template struct Templates { typedef Templates38 type; }; template struct Templates { typedef Templates39 type; }; template struct Templates { typedef Templates40 type; }; template struct Templates { typedef Templates41 type; }; template struct Templates { typedef Templates42 type; }; template struct Templates { typedef Templates43 type; }; template struct Templates { typedef Templates44 type; }; template struct Templates { typedef Templates45 type; }; template struct Templates { typedef Templates46 type; }; template struct Templates { typedef Templates47 type; }; template struct Templates { typedef Templates48 type; }; template struct Templates { typedef Templates49 type; }; // The TypeList template makes it possible to use either a single type // or a Types<...> list in TYPED_TEST_CASE() and // INSTANTIATE_TYPED_TEST_CASE_P(). template struct TypeList { typedef Types1 type; }; template struct TypeList > { typedef typename Types::type type; }; #endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ // Due to C++ preprocessor weirdness, we need double indirection to // concatenate two tokens when one of them is __LINE__. Writing // // foo ## __LINE__ // // will result in the token foo__LINE__, instead of foo followed by // the current line number. For more details, see // http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.6 #define GTEST_CONCAT_TOKEN_(foo, bar) GTEST_CONCAT_TOKEN_IMPL_(foo, bar) #define GTEST_CONCAT_TOKEN_IMPL_(foo, bar) foo ## bar class ProtocolMessage; namespace proto2 { class Message; } namespace testing { // Forward declarations. class AssertionResult; // Result of an assertion. class Message; // Represents a failure message. class Test; // Represents a test. class TestInfo; // Information about a test. class TestPartResult; // Result of a test part. class UnitTest; // A collection of test cases. template ::std::string PrintToString(const T& value); namespace internal { struct TraceInfo; // Information about a trace point. class ScopedTrace; // Implements scoped trace. class TestInfoImpl; // Opaque implementation of TestInfo class UnitTestImpl; // Opaque implementation of UnitTest // How many times InitGoogleTest() has been called. GTEST_API_ extern int g_init_gtest_count; // The text used in failure messages to indicate the start of the // stack trace. GTEST_API_ extern const char kStackTraceMarker[]; // Two overloaded helpers for checking at compile time whether an // expression is a null pointer literal (i.e. NULL or any 0-valued // compile-time integral constant). Their return values have // different sizes, so we can use sizeof() to test which version is // picked by the compiler. These helpers have no implementations, as // we only need their signatures. // // Given IsNullLiteralHelper(x), the compiler will pick the first // version if x can be implicitly converted to Secret*, and pick the // second version otherwise. Since Secret is a secret and incomplete // type, the only expression a user can write that has type Secret* is // a null pointer literal. Therefore, we know that x is a null // pointer literal if and only if the first version is picked by the // compiler. char IsNullLiteralHelper(Secret* p); char (&IsNullLiteralHelper(...))[2]; // NOLINT // A compile-time bool constant that is true if and only if x is a // null pointer literal (i.e. NULL or any 0-valued compile-time // integral constant). #ifdef GTEST_ELLIPSIS_NEEDS_POD_ // We lose support for NULL detection where the compiler doesn't like // passing non-POD classes through ellipsis (...). # define GTEST_IS_NULL_LITERAL_(x) false #else # define GTEST_IS_NULL_LITERAL_(x) \ (sizeof(::testing::internal::IsNullLiteralHelper(x)) == 1) #endif // GTEST_ELLIPSIS_NEEDS_POD_ // Appends the user-supplied message to the Google-Test-generated message. GTEST_API_ std::string AppendUserMessage( const std::string& gtest_msg, const Message& user_msg); #if GTEST_HAS_EXCEPTIONS // This exception is thrown by (and only by) a failed Google Test // assertion when GTEST_FLAG(throw_on_failure) is true (if exceptions // are enabled). We derive it from std::runtime_error, which is for // errors presumably detectable only at run time. Since // std::runtime_error inherits from std::exception, many testing // frameworks know how to extract and print the message inside it. class GTEST_API_ GoogleTestFailureException : public ::std::runtime_error { public: explicit GoogleTestFailureException(const TestPartResult& failure); }; #endif // GTEST_HAS_EXCEPTIONS // A helper class for creating scoped traces in user programs. class GTEST_API_ ScopedTrace { public: // The c'tor pushes the given source file location and message onto // a trace stack maintained by Google Test. ScopedTrace(const char* file, int line, const Message& message); // The d'tor pops the info pushed by the c'tor. // // Note that the d'tor is not virtual in order to be efficient. // Don't inherit from ScopedTrace! ~ScopedTrace(); private: GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedTrace); } GTEST_ATTRIBUTE_UNUSED_; // A ScopedTrace object does its job in its // c'tor and d'tor. Therefore it doesn't // need to be used otherwise. namespace edit_distance { // Returns the optimal edits to go from 'left' to 'right'. // All edits cost the same, with replace having lower priority than // add/remove. // Simple implementation of the Wagner–Fischer algorithm. // See http://en.wikipedia.org/wiki/Wagner-Fischer_algorithm enum EditType { kMatch, kAdd, kRemove, kReplace }; GTEST_API_ std::vector CalculateOptimalEdits( const std::vector& left, const std::vector& right); // Same as above, but the input is represented as strings. GTEST_API_ std::vector CalculateOptimalEdits( const std::vector& left, const std::vector& right); // Create a diff of the input strings in Unified diff format. GTEST_API_ std::string CreateUnifiedDiff(const std::vector& left, const std::vector& right, size_t context = 2); } // namespace edit_distance // Calculate the diff between 'left' and 'right' and return it in unified diff // format. // If not null, stores in 'total_line_count' the total number of lines found // in left + right. GTEST_API_ std::string DiffStrings(const std::string& left, const std::string& right, size_t* total_line_count); // Constructs and returns the message for an equality assertion // (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure. // // The first four parameters are the expressions used in the assertion // and their values, as strings. For example, for ASSERT_EQ(foo, bar) // where foo is 5 and bar is 6, we have: // // expected_expression: "foo" // actual_expression: "bar" // expected_value: "5" // actual_value: "6" // // The ignoring_case parameter is true iff the assertion is a // *_STRCASEEQ*. When it's true, the string " (ignoring case)" will // be inserted into the message. GTEST_API_ AssertionResult EqFailure(const char* expected_expression, const char* actual_expression, const std::string& expected_value, const std::string& actual_value, bool ignoring_case); // Constructs a failure message for Boolean assertions such as EXPECT_TRUE. GTEST_API_ std::string GetBoolAssertionFailureMessage( const AssertionResult& assertion_result, const char* expression_text, const char* actual_predicate_value, const char* expected_predicate_value); // This template class represents an IEEE floating-point number // (either single-precision or double-precision, depending on the // template parameters). // // The purpose of this class is to do more sophisticated number // comparison. (Due to round-off error, etc, it's very unlikely that // two floating-points will be equal exactly. Hence a naive // comparison by the == operation often doesn't work.) // // Format of IEEE floating-point: // // The most-significant bit being the leftmost, an IEEE // floating-point looks like // // sign_bit exponent_bits fraction_bits // // Here, sign_bit is a single bit that designates the sign of the // number. // // For float, there are 8 exponent bits and 23 fraction bits. // // For double, there are 11 exponent bits and 52 fraction bits. // // More details can be found at // http://en.wikipedia.org/wiki/IEEE_floating-point_standard. // // Template parameter: // // RawType: the raw floating-point type (either float or double) template class FloatingPoint { public: // Defines the unsigned integer type that has the same size as the // floating point number. typedef typename TypeWithSize::UInt Bits; // Constants. // # of bits in a number. static const size_t kBitCount = 8*sizeof(RawType); // # of fraction bits in a number. static const size_t kFractionBitCount = std::numeric_limits::digits - 1; // # of exponent bits in a number. static const size_t kExponentBitCount = kBitCount - 1 - kFractionBitCount; // The mask for the sign bit. static const Bits kSignBitMask = static_cast(1) << (kBitCount - 1); // The mask for the fraction bits. static const Bits kFractionBitMask = ~static_cast(0) >> (kExponentBitCount + 1); // The mask for the exponent bits. static const Bits kExponentBitMask = ~(kSignBitMask | kFractionBitMask); // How many ULP's (Units in the Last Place) we want to tolerate when // comparing two numbers. The larger the value, the more error we // allow. A 0 value means that two numbers must be exactly the same // to be considered equal. // // The maximum error of a single floating-point operation is 0.5 // units in the last place. On Intel CPU's, all floating-point // calculations are done with 80-bit precision, while double has 64 // bits. Therefore, 4 should be enough for ordinary use. // // See the following article for more details on ULP: // http://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/ static const size_t kMaxUlps = 4; // Constructs a FloatingPoint from a raw floating-point number. // // On an Intel CPU, passing a non-normalized NAN (Not a Number) // around may change its bits, although the new value is guaranteed // to be also a NAN. Therefore, don't expect this constructor to // preserve the bits in x when x is a NAN. explicit FloatingPoint(const RawType& x) { u_.value_ = x; } // Static methods // Reinterprets a bit pattern as a floating-point number. // // This function is needed to test the AlmostEquals() method. static RawType ReinterpretBits(const Bits bits) { FloatingPoint fp(0); fp.u_.bits_ = bits; return fp.u_.value_; } // Returns the floating-point number that represent positive infinity. static RawType Infinity() { return ReinterpretBits(kExponentBitMask); } // Returns the maximum representable finite floating-point number. static RawType Max(); // Non-static methods // Returns the bits that represents this number. const Bits &bits() const { return u_.bits_; } // Returns the exponent bits of this number. Bits exponent_bits() const { return kExponentBitMask & u_.bits_; } // Returns the fraction bits of this number. Bits fraction_bits() const { return kFractionBitMask & u_.bits_; } // Returns the sign bit of this number. Bits sign_bit() const { return kSignBitMask & u_.bits_; } // Returns true iff this is NAN (not a number). bool is_nan() const { // It's a NAN if the exponent bits are all ones and the fraction // bits are not entirely zeros. return (exponent_bits() == kExponentBitMask) && (fraction_bits() != 0); } // Returns true iff this number is at most kMaxUlps ULP's away from // rhs. In particular, this function: // // - returns false if either number is (or both are) NAN. // - treats really large numbers as almost equal to infinity. // - thinks +0.0 and -0.0 are 0 DLP's apart. bool AlmostEquals(const FloatingPoint& rhs) const { // The IEEE standard says that any comparison operation involving // a NAN must return false. if (is_nan() || rhs.is_nan()) return false; return DistanceBetweenSignAndMagnitudeNumbers(u_.bits_, rhs.u_.bits_) <= kMaxUlps; } private: // The data type used to store the actual floating-point number. union FloatingPointUnion { RawType value_; // The raw floating-point number. Bits bits_; // The bits that represent the number. }; // Converts an integer from the sign-and-magnitude representation to // the biased representation. More precisely, let N be 2 to the // power of (kBitCount - 1), an integer x is represented by the // unsigned number x + N. // // For instance, // // -N + 1 (the most negative number representable using // sign-and-magnitude) is represented by 1; // 0 is represented by N; and // N - 1 (the biggest number representable using // sign-and-magnitude) is represented by 2N - 1. // // Read http://en.wikipedia.org/wiki/Signed_number_representations // for more details on signed number representations. static Bits SignAndMagnitudeToBiased(const Bits &sam) { if (kSignBitMask & sam) { // sam represents a negative number. return ~sam + 1; } else { // sam represents a positive number. return kSignBitMask | sam; } } // Given two numbers in the sign-and-magnitude representation, // returns the distance between them as an unsigned number. static Bits DistanceBetweenSignAndMagnitudeNumbers(const Bits &sam1, const Bits &sam2) { const Bits biased1 = SignAndMagnitudeToBiased(sam1); const Bits biased2 = SignAndMagnitudeToBiased(sam2); return (biased1 >= biased2) ? (biased1 - biased2) : (biased2 - biased1); } FloatingPointUnion u_; }; // We cannot use std::numeric_limits::max() as it clashes with the max() // macro defined by . template <> inline float FloatingPoint::Max() { return FLT_MAX; } template <> inline double FloatingPoint::Max() { return DBL_MAX; } // Typedefs the instances of the FloatingPoint template class that we // care to use. typedef FloatingPoint Float; typedef FloatingPoint Double; // In order to catch the mistake of putting tests that use different // test fixture classes in the same test case, we need to assign // unique IDs to fixture classes and compare them. The TypeId type is // used to hold such IDs. The user should treat TypeId as an opaque // type: the only operation allowed on TypeId values is to compare // them for equality using the == operator. typedef const void* TypeId; template class TypeIdHelper { public: // dummy_ must not have a const type. Otherwise an overly eager // compiler (e.g. MSVC 7.1 & 8.0) may try to merge // TypeIdHelper::dummy_ for different Ts as an "optimization". static bool dummy_; }; template bool TypeIdHelper::dummy_ = false; // GetTypeId() returns the ID of type T. Different values will be // returned for different types. Calling the function twice with the // same type argument is guaranteed to return the same ID. template TypeId GetTypeId() { // The compiler is required to allocate a different // TypeIdHelper::dummy_ variable for each T used to instantiate // the template. Therefore, the address of dummy_ is guaranteed to // be unique. return &(TypeIdHelper::dummy_); } // Returns the type ID of ::testing::Test. Always call this instead // of GetTypeId< ::testing::Test>() to get the type ID of // ::testing::Test, as the latter may give the wrong result due to a // suspected linker bug when compiling Google Test as a Mac OS X // framework. GTEST_API_ TypeId GetTestTypeId(); // Defines the abstract factory interface that creates instances // of a Test object. class TestFactoryBase { public: virtual ~TestFactoryBase() {} // Creates a test instance to run. The instance is both created and destroyed // within TestInfoImpl::Run() virtual Test* CreateTest() = 0; protected: TestFactoryBase() {} private: GTEST_DISALLOW_COPY_AND_ASSIGN_(TestFactoryBase); }; // This class provides implementation of TeastFactoryBase interface. // It is used in TEST and TEST_F macros. template class TestFactoryImpl : public TestFactoryBase { public: virtual Test* CreateTest() { return new TestClass; } }; #if GTEST_OS_WINDOWS // Predicate-formatters for implementing the HRESULT checking macros // {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED} // We pass a long instead of HRESULT to avoid causing an // include dependency for the HRESULT type. GTEST_API_ AssertionResult IsHRESULTSuccess(const char* expr, long hr); // NOLINT GTEST_API_ AssertionResult IsHRESULTFailure(const char* expr, long hr); // NOLINT #endif // GTEST_OS_WINDOWS // Types of SetUpTestCase() and TearDownTestCase() functions. typedef void (*SetUpTestCaseFunc)(); typedef void (*TearDownTestCaseFunc)(); // Creates a new TestInfo object and registers it with Google Test; // returns the created object. // // Arguments: // // test_case_name: name of the test case // name: name of the test // type_param the name of the test's type parameter, or NULL if // this is not a typed or a type-parameterized test. // value_param text representation of the test's value parameter, // or NULL if this is not a type-parameterized test. // fixture_class_id: ID of the test fixture class // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case // factory: pointer to the factory that creates a test object. // The newly created TestInfo instance will assume // ownership of the factory object. GTEST_API_ TestInfo* MakeAndRegisterTestInfo( const char* test_case_name, const char* name, const char* type_param, const char* value_param, TypeId fixture_class_id, SetUpTestCaseFunc set_up_tc, TearDownTestCaseFunc tear_down_tc, TestFactoryBase* factory); // If *pstr starts with the given prefix, modifies *pstr to be right // past the prefix and returns true; otherwise leaves *pstr unchanged // and returns false. None of pstr, *pstr, and prefix can be NULL. GTEST_API_ bool SkipPrefix(const char* prefix, const char** pstr); #if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P // State of the definition of a type-parameterized test case. class GTEST_API_ TypedTestCasePState { public: TypedTestCasePState() : registered_(false) {} // Adds the given test name to defined_test_names_ and return true // if the test case hasn't been registered; otherwise aborts the // program. bool AddTestName(const char* file, int line, const char* case_name, const char* test_name) { if (registered_) { fprintf(stderr, "%s Test %s must be defined before " "REGISTER_TYPED_TEST_CASE_P(%s, ...).\n", FormatFileLocation(file, line).c_str(), test_name, case_name); fflush(stderr); posix::Abort(); } defined_test_names_.insert(test_name); return true; } // Verifies that registered_tests match the test names in // defined_test_names_; returns registered_tests if successful, or // aborts the program otherwise. const char* VerifyRegisteredTestNames( const char* file, int line, const char* registered_tests); private: bool registered_; ::std::set defined_test_names_; }; // Skips to the first non-space char after the first comma in 'str'; // returns NULL if no comma is found in 'str'. inline const char* SkipComma(const char* str) { const char* comma = strchr(str, ','); if (comma == NULL) { return NULL; } while (IsSpace(*(++comma))) {} return comma; } // Returns the prefix of 'str' before the first comma in it; returns // the entire string if it contains no comma. inline std::string GetPrefixUntilComma(const char* str) { const char* comma = strchr(str, ','); return comma == NULL ? str : std::string(str, comma); } // TypeParameterizedTest::Register() // registers a list of type-parameterized tests with Google Test. The // return value is insignificant - we just need to return something // such that we can call this function in a namespace scope. // // Implementation note: The GTEST_TEMPLATE_ macro declares a template // template parameter. It's defined in gtest-type-util.h. template class TypeParameterizedTest { public: // 'index' is the index of the test in the type list 'Types' // specified in INSTANTIATE_TYPED_TEST_CASE_P(Prefix, TestCase, // Types). Valid values for 'index' are [0, N - 1] where N is the // length of Types. static bool Register(const char* prefix, const char* case_name, const char* test_names, int index) { typedef typename Types::Head Type; typedef Fixture FixtureClass; typedef typename GTEST_BIND_(TestSel, Type) TestClass; // First, registers the first type-parameterized test in the type // list. MakeAndRegisterTestInfo( (std::string(prefix) + (prefix[0] == '\0' ? "" : "/") + case_name + "/" + StreamableToString(index)).c_str(), StripTrailingSpaces(GetPrefixUntilComma(test_names)).c_str(), GetTypeName().c_str(), NULL, // No value parameter. GetTypeId(), TestClass::SetUpTestCase, TestClass::TearDownTestCase, new TestFactoryImpl); // Next, recurses (at compile time) with the tail of the type list. return TypeParameterizedTest ::Register(prefix, case_name, test_names, index + 1); } }; // The base case for the compile time recursion. template class TypeParameterizedTest { public: static bool Register(const char* /*prefix*/, const char* /*case_name*/, const char* /*test_names*/, int /*index*/) { return true; } }; // TypeParameterizedTestCase::Register() // registers *all combinations* of 'Tests' and 'Types' with Google // Test. The return value is insignificant - we just need to return // something such that we can call this function in a namespace scope. template class TypeParameterizedTestCase { public: static bool Register(const char* prefix, const char* case_name, const char* test_names) { typedef typename Tests::Head Head; // First, register the first test in 'Test' for each type in 'Types'. TypeParameterizedTest::Register( prefix, case_name, test_names, 0); // Next, recurses (at compile time) with the tail of the test list. return TypeParameterizedTestCase ::Register(prefix, case_name, SkipComma(test_names)); } }; // The base case for the compile time recursion. template class TypeParameterizedTestCase { public: static bool Register(const char* /*prefix*/, const char* /*case_name*/, const char* /*test_names*/) { return true; } }; #endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P // Returns the current OS stack trace as an std::string. // // The maximum number of stack frames to be included is specified by // the gtest_stack_trace_depth flag. The skip_count parameter // specifies the number of top frames to be skipped, which doesn't // count against the number of frames to be included. // // For example, if Foo() calls Bar(), which in turn calls // GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in // the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't. GTEST_API_ std::string GetCurrentOsStackTraceExceptTop( UnitTest* unit_test, int skip_count); // Helpers for suppressing warnings on unreachable code or constant // condition. // Always returns true. GTEST_API_ bool AlwaysTrue(); // Always returns false. inline bool AlwaysFalse() { return !AlwaysTrue(); } // Helper for suppressing false warning from Clang on a const char* // variable declared in a conditional expression always being NULL in // the else branch. struct GTEST_API_ ConstCharPtr { ConstCharPtr(const char* str) : value(str) {} operator bool() const { return true; } const char* value; }; // A simple Linear Congruential Generator for generating random // numbers with a uniform distribution. Unlike rand() and srand(), it // doesn't use global state (and therefore can't interfere with user // code). Unlike rand_r(), it's portable. An LCG isn't very random, // but it's good enough for our purposes. class GTEST_API_ Random { public: static const UInt32 kMaxRange = 1u << 31; explicit Random(UInt32 seed) : state_(seed) {} void Reseed(UInt32 seed) { state_ = seed; } // Generates a random number from [0, range). Crashes if 'range' is // 0 or greater than kMaxRange. UInt32 Generate(UInt32 range); private: UInt32 state_; GTEST_DISALLOW_COPY_AND_ASSIGN_(Random); }; // Defining a variable of type CompileAssertTypesEqual will cause a // compiler error iff T1 and T2 are different types. template struct CompileAssertTypesEqual; template struct CompileAssertTypesEqual { }; // Removes the reference from a type if it is a reference type, // otherwise leaves it unchanged. This is the same as // tr1::remove_reference, which is not widely available yet. template struct RemoveReference { typedef T type; }; // NOLINT template struct RemoveReference { typedef T type; }; // NOLINT // A handy wrapper around RemoveReference that works when the argument // T depends on template parameters. #define GTEST_REMOVE_REFERENCE_(T) \ typename ::testing::internal::RemoveReference::type // Removes const from a type if it is a const type, otherwise leaves // it unchanged. This is the same as tr1::remove_const, which is not // widely available yet. template struct RemoveConst { typedef T type; }; // NOLINT template struct RemoveConst { typedef T type; }; // NOLINT // MSVC 8.0, Sun C++, and IBM XL C++ have a bug which causes the above // definition to fail to remove the const in 'const int[3]' and 'const // char[3][4]'. The following specialization works around the bug. template struct RemoveConst { typedef typename RemoveConst::type type[N]; }; #if defined(_MSC_VER) && _MSC_VER < 1400 // This is the only specialization that allows VC++ 7.1 to remove const in // 'const int[3] and 'const int[3][4]'. However, it causes trouble with GCC // and thus needs to be conditionally compiled. template struct RemoveConst { typedef typename RemoveConst::type type[N]; }; #endif // A handy wrapper around RemoveConst that works when the argument // T depends on template parameters. #define GTEST_REMOVE_CONST_(T) \ typename ::testing::internal::RemoveConst::type // Turns const U&, U&, const U, and U all into U. #define GTEST_REMOVE_REFERENCE_AND_CONST_(T) \ GTEST_REMOVE_CONST_(GTEST_REMOVE_REFERENCE_(T)) // Adds reference to a type if it is not a reference type, // otherwise leaves it unchanged. This is the same as // tr1::add_reference, which is not widely available yet. template struct AddReference { typedef T& type; }; // NOLINT template struct AddReference { typedef T& type; }; // NOLINT // A handy wrapper around AddReference that works when the argument T // depends on template parameters. #define GTEST_ADD_REFERENCE_(T) \ typename ::testing::internal::AddReference::type // Adds a reference to const on top of T as necessary. For example, // it transforms // // char ==> const char& // const char ==> const char& // char& ==> const char& // const char& ==> const char& // // The argument T must depend on some template parameters. #define GTEST_REFERENCE_TO_CONST_(T) \ GTEST_ADD_REFERENCE_(const GTEST_REMOVE_REFERENCE_(T)) // ImplicitlyConvertible::value is a compile-time bool // constant that's true iff type From can be implicitly converted to // type To. template class ImplicitlyConvertible { private: // We need the following helper functions only for their types. // They have no implementations. // MakeFrom() is an expression whose type is From. We cannot simply // use From(), as the type From may not have a public default // constructor. static typename AddReference::type MakeFrom(); // These two functions are overloaded. Given an expression // Helper(x), the compiler will pick the first version if x can be // implicitly converted to type To; otherwise it will pick the // second version. // // The first version returns a value of size 1, and the second // version returns a value of size 2. Therefore, by checking the // size of Helper(x), which can be done at compile time, we can tell // which version of Helper() is used, and hence whether x can be // implicitly converted to type To. static char Helper(To); static char (&Helper(...))[2]; // NOLINT // We have to put the 'public' section after the 'private' section, // or MSVC refuses to compile the code. public: #if defined(__BORLANDC__) // C++Builder cannot use member overload resolution during template // instantiation. The simplest workaround is to use its C++0x type traits // functions (C++Builder 2009 and above only). static const bool value = __is_convertible(From, To); #else // MSVC warns about implicitly converting from double to int for // possible loss of data, so we need to temporarily disable the // warning. GTEST_DISABLE_MSC_WARNINGS_PUSH_(4244) static const bool value = sizeof(Helper(ImplicitlyConvertible::MakeFrom())) == 1; GTEST_DISABLE_MSC_WARNINGS_POP_() #endif // __BORLANDC__ }; template const bool ImplicitlyConvertible::value; // IsAProtocolMessage::value is a compile-time bool constant that's // true iff T is type ProtocolMessage, proto2::Message, or a subclass // of those. template struct IsAProtocolMessage : public bool_constant< ImplicitlyConvertible::value || ImplicitlyConvertible::value> { }; // When the compiler sees expression IsContainerTest(0), if C is an // STL-style container class, the first overload of IsContainerTest // will be viable (since both C::iterator* and C::const_iterator* are // valid types and NULL can be implicitly converted to them). It will // be picked over the second overload as 'int' is a perfect match for // the type of argument 0. If C::iterator or C::const_iterator is not // a valid type, the first overload is not viable, and the second // overload will be picked. Therefore, we can determine whether C is // a container class by checking the type of IsContainerTest(0). // The value of the expression is insignificant. // // Note that we look for both C::iterator and C::const_iterator. The // reason is that C++ injects the name of a class as a member of the // class itself (e.g. you can refer to class iterator as either // 'iterator' or 'iterator::iterator'). If we look for C::iterator // only, for example, we would mistakenly think that a class named // iterator is an STL container. // // Also note that the simpler approach of overloading // IsContainerTest(typename C::const_iterator*) and // IsContainerTest(...) doesn't work with Visual Age C++ and Sun C++. typedef int IsContainer; template IsContainer IsContainerTest(int /* dummy */, typename C::iterator* /* it */ = NULL, typename C::const_iterator* /* const_it */ = NULL) { return 0; } typedef char IsNotContainer; template IsNotContainer IsContainerTest(long /* dummy */) { return '\0'; } // EnableIf::type is void when 'Cond' is true, and // undefined when 'Cond' is false. To use SFINAE to make a function // overload only apply when a particular expression is true, add // "typename EnableIf::type* = 0" as the last parameter. template struct EnableIf; template<> struct EnableIf { typedef void type; }; // NOLINT // Utilities for native arrays. // ArrayEq() compares two k-dimensional native arrays using the // elements' operator==, where k can be any integer >= 0. When k is // 0, ArrayEq() degenerates into comparing a single pair of values. template bool ArrayEq(const T* lhs, size_t size, const U* rhs); // This generic version is used when k is 0. template inline bool ArrayEq(const T& lhs, const U& rhs) { return lhs == rhs; } // This overload is used when k >= 1. template inline bool ArrayEq(const T(&lhs)[N], const U(&rhs)[N]) { return internal::ArrayEq(lhs, N, rhs); } // This helper reduces code bloat. If we instead put its logic inside // the previous ArrayEq() function, arrays with different sizes would // lead to different copies of the template code. template bool ArrayEq(const T* lhs, size_t size, const U* rhs) { for (size_t i = 0; i != size; i++) { if (!internal::ArrayEq(lhs[i], rhs[i])) return false; } return true; } // Finds the first element in the iterator range [begin, end) that // equals elem. Element may be a native array type itself. template Iter ArrayAwareFind(Iter begin, Iter end, const Element& elem) { for (Iter it = begin; it != end; ++it) { if (internal::ArrayEq(*it, elem)) return it; } return end; } // CopyArray() copies a k-dimensional native array using the elements' // operator=, where k can be any integer >= 0. When k is 0, // CopyArray() degenerates into copying a single value. template void CopyArray(const T* from, size_t size, U* to); // This generic version is used when k is 0. template inline void CopyArray(const T& from, U* to) { *to = from; } // This overload is used when k >= 1. template inline void CopyArray(const T(&from)[N], U(*to)[N]) { internal::CopyArray(from, N, *to); } // This helper reduces code bloat. If we instead put its logic inside // the previous CopyArray() function, arrays with different sizes // would lead to different copies of the template code. template void CopyArray(const T* from, size_t size, U* to) { for (size_t i = 0; i != size; i++) { internal::CopyArray(from[i], to + i); } } // The relation between an NativeArray object (see below) and the // native array it represents. // We use 2 different structs to allow non-copyable types to be used, as long // as RelationToSourceReference() is passed. struct RelationToSourceReference {}; struct RelationToSourceCopy {}; // Adapts a native array to a read-only STL-style container. Instead // of the complete STL container concept, this adaptor only implements // members useful for Google Mock's container matchers. New members // should be added as needed. To simplify the implementation, we only // support Element being a raw type (i.e. having no top-level const or // reference modifier). It's the client's responsibility to satisfy // this requirement. Element can be an array type itself (hence // multi-dimensional arrays are supported). template class NativeArray { public: // STL-style container typedefs. typedef Element value_type; typedef Element* iterator; typedef const Element* const_iterator; // Constructs from a native array. References the source. NativeArray(const Element* array, size_t count, RelationToSourceReference) { InitRef(array, count); } // Constructs from a native array. Copies the source. NativeArray(const Element* array, size_t count, RelationToSourceCopy) { InitCopy(array, count); } // Copy constructor. NativeArray(const NativeArray& rhs) { (this->*rhs.clone_)(rhs.array_, rhs.size_); } ~NativeArray() { if (clone_ != &NativeArray::InitRef) delete[] array_; } // STL-style container methods. size_t size() const { return size_; } const_iterator begin() const { return array_; } const_iterator end() const { return array_ + size_; } bool operator==(const NativeArray& rhs) const { return size() == rhs.size() && ArrayEq(begin(), size(), rhs.begin()); } private: enum { kCheckTypeIsNotConstOrAReference = StaticAssertTypeEqHelper< Element, GTEST_REMOVE_REFERENCE_AND_CONST_(Element)>::value, }; // Initializes this object with a copy of the input. void InitCopy(const Element* array, size_t a_size) { Element* const copy = new Element[a_size]; CopyArray(array, a_size, copy); array_ = copy; size_ = a_size; clone_ = &NativeArray::InitCopy; } // Initializes this object with a reference of the input. void InitRef(const Element* array, size_t a_size) { array_ = array; size_ = a_size; clone_ = &NativeArray::InitRef; } const Element* array_; size_t size_; void (NativeArray::*clone_)(const Element*, size_t); GTEST_DISALLOW_ASSIGN_(NativeArray); }; } // namespace internal } // namespace testing #define GTEST_MESSAGE_AT_(file, line, message, result_type) \ ::testing::internal::AssertHelper(result_type, file, line, message) \ = ::testing::Message() #define GTEST_MESSAGE_(message, result_type) \ GTEST_MESSAGE_AT_(__FILE__, __LINE__, message, result_type) #define GTEST_FATAL_FAILURE_(message) \ return GTEST_MESSAGE_(message, ::testing::TestPartResult::kFatalFailure) #define GTEST_NONFATAL_FAILURE_(message) \ GTEST_MESSAGE_(message, ::testing::TestPartResult::kNonFatalFailure) #define GTEST_SUCCESS_(message) \ GTEST_MESSAGE_(message, ::testing::TestPartResult::kSuccess) // Suppresses MSVC warnings 4072 (unreachable code) for the code following // statement if it returns or throws (or doesn't return or throw in some // situations). #define GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) \ if (::testing::internal::AlwaysTrue()) { statement; } #define GTEST_TEST_THROW_(statement, expected_exception, fail) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::ConstCharPtr gtest_msg = "") { \ bool gtest_caught_expected = false; \ try { \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ } \ catch (expected_exception const&) { \ gtest_caught_expected = true; \ } \ catch (...) { \ gtest_msg.value = \ "Expected: " #statement " throws an exception of type " \ #expected_exception ".\n Actual: it throws a different type."; \ goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \ } \ if (!gtest_caught_expected) { \ gtest_msg.value = \ "Expected: " #statement " throws an exception of type " \ #expected_exception ".\n Actual: it throws nothing."; \ goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \ } \ } else \ GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__): \ fail(gtest_msg.value) #define GTEST_TEST_NO_THROW_(statement, fail) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::AlwaysTrue()) { \ try { \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ } \ catch (...) { \ goto GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__); \ } \ } else \ GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__): \ fail("Expected: " #statement " doesn't throw an exception.\n" \ " Actual: it throws.") #define GTEST_TEST_ANY_THROW_(statement, fail) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::AlwaysTrue()) { \ bool gtest_caught_any = false; \ try { \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ } \ catch (...) { \ gtest_caught_any = true; \ } \ if (!gtest_caught_any) { \ goto GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__); \ } \ } else \ GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__): \ fail("Expected: " #statement " throws an exception.\n" \ " Actual: it doesn't.") // Implements Boolean test assertions such as EXPECT_TRUE. expression can be // either a boolean expression or an AssertionResult. text is a textual // represenation of expression as it was passed into the EXPECT_TRUE. #define GTEST_TEST_BOOLEAN_(expression, text, actual, expected, fail) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (const ::testing::AssertionResult gtest_ar_ = \ ::testing::AssertionResult(expression)) \ ; \ else \ fail(::testing::internal::GetBoolAssertionFailureMessage(\ gtest_ar_, text, #actual, #expected).c_str()) #define GTEST_TEST_NO_FATAL_FAILURE_(statement, fail) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::AlwaysTrue()) { \ ::testing::internal::HasNewFatalFailureHelper gtest_fatal_failure_checker; \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ if (gtest_fatal_failure_checker.has_new_fatal_failure()) { \ goto GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__); \ } \ } else \ GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__): \ fail("Expected: " #statement " doesn't generate new fatal " \ "failures in the current thread.\n" \ " Actual: it does.") // Expands to the name of the class that implements the given test. #define GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \ test_case_name##_##test_name##_Test // Helper macro for defining tests. #define GTEST_TEST_(test_case_name, test_name, parent_class, parent_id)\ class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) : public parent_class {\ public:\ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {}\ private:\ virtual void TestBody();\ static ::testing::TestInfo* const test_info_ GTEST_ATTRIBUTE_UNUSED_;\ GTEST_DISALLOW_COPY_AND_ASSIGN_(\ GTEST_TEST_CLASS_NAME_(test_case_name, test_name));\ };\ \ ::testing::TestInfo* const GTEST_TEST_CLASS_NAME_(test_case_name, test_name)\ ::test_info_ =\ ::testing::internal::MakeAndRegisterTestInfo(\ #test_case_name, #test_name, NULL, NULL, \ (parent_id), \ parent_class::SetUpTestCase, \ parent_class::TearDownTestCase, \ new ::testing::internal::TestFactoryImpl<\ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>);\ void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody() #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_ // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // // The Google C++ Testing Framework (Google Test) // // This header file defines the public API for death tests. It is // #included by gtest.h so a user doesn't need to include this // directly. #ifndef GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_ #define GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_ // Copyright 2005, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee) // // The Google C++ Testing Framework (Google Test) // // This header file defines internal utilities needed for implementing // death tests. They are subject to change without notice. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_ #include namespace testing { namespace internal { GTEST_DECLARE_string_(internal_run_death_test); // Names of the flags (needed for parsing Google Test flags). const char kDeathTestStyleFlag[] = "death_test_style"; const char kDeathTestUseFork[] = "death_test_use_fork"; const char kInternalRunDeathTestFlag[] = "internal_run_death_test"; #if GTEST_HAS_DEATH_TEST // DeathTest is a class that hides much of the complexity of the // GTEST_DEATH_TEST_ macro. It is abstract; its static Create method // returns a concrete class that depends on the prevailing death test // style, as defined by the --gtest_death_test_style and/or // --gtest_internal_run_death_test flags. // In describing the results of death tests, these terms are used with // the corresponding definitions: // // exit status: The integer exit information in the format specified // by wait(2) // exit code: The integer code passed to exit(3), _exit(2), or // returned from main() class GTEST_API_ DeathTest { public: // Create returns false if there was an error determining the // appropriate action to take for the current death test; for example, // if the gtest_death_test_style flag is set to an invalid value. // The LastMessage method will return a more detailed message in that // case. Otherwise, the DeathTest pointer pointed to by the "test" // argument is set. If the death test should be skipped, the pointer // is set to NULL; otherwise, it is set to the address of a new concrete // DeathTest object that controls the execution of the current test. static bool Create(const char* statement, const RE* regex, const char* file, int line, DeathTest** test); DeathTest(); virtual ~DeathTest() { } // A helper class that aborts a death test when it's deleted. class ReturnSentinel { public: explicit ReturnSentinel(DeathTest* test) : test_(test) { } ~ReturnSentinel() { test_->Abort(TEST_ENCOUNTERED_RETURN_STATEMENT); } private: DeathTest* const test_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ReturnSentinel); } GTEST_ATTRIBUTE_UNUSED_; // An enumeration of possible roles that may be taken when a death // test is encountered. EXECUTE means that the death test logic should // be executed immediately. OVERSEE means that the program should prepare // the appropriate environment for a child process to execute the death // test, then wait for it to complete. enum TestRole { OVERSEE_TEST, EXECUTE_TEST }; // An enumeration of the three reasons that a test might be aborted. enum AbortReason { TEST_ENCOUNTERED_RETURN_STATEMENT, TEST_THREW_EXCEPTION, TEST_DID_NOT_DIE }; // Assumes one of the above roles. virtual TestRole AssumeRole() = 0; // Waits for the death test to finish and returns its status. virtual int Wait() = 0; // Returns true if the death test passed; that is, the test process // exited during the test, its exit status matches a user-supplied // predicate, and its stderr output matches a user-supplied regular // expression. // The user-supplied predicate may be a macro expression rather // than a function pointer or functor, or else Wait and Passed could // be combined. virtual bool Passed(bool exit_status_ok) = 0; // Signals that the death test did not die as expected. virtual void Abort(AbortReason reason) = 0; // Returns a human-readable outcome message regarding the outcome of // the last death test. static const char* LastMessage(); static void set_last_death_test_message(const std::string& message); private: // A string containing a description of the outcome of the last death test. static std::string last_death_test_message_; GTEST_DISALLOW_COPY_AND_ASSIGN_(DeathTest); }; // Factory interface for death tests. May be mocked out for testing. class DeathTestFactory { public: virtual ~DeathTestFactory() { } virtual bool Create(const char* statement, const RE* regex, const char* file, int line, DeathTest** test) = 0; }; // A concrete DeathTestFactory implementation for normal use. class DefaultDeathTestFactory : public DeathTestFactory { public: virtual bool Create(const char* statement, const RE* regex, const char* file, int line, DeathTest** test); }; // Returns true if exit_status describes a process that was terminated // by a signal, or exited normally with a nonzero exit code. GTEST_API_ bool ExitedUnsuccessfully(int exit_status); // Traps C++ exceptions escaping statement and reports them as test // failures. Note that trapping SEH exceptions is not implemented here. # if GTEST_HAS_EXCEPTIONS # define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \ try { \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ } catch (const ::std::exception& gtest_exception) { \ fprintf(\ stderr, \ "\n%s: Caught std::exception-derived exception escaping the " \ "death test statement. Exception message: %s\n", \ ::testing::internal::FormatFileLocation(__FILE__, __LINE__).c_str(), \ gtest_exception.what()); \ fflush(stderr); \ death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \ } catch (...) { \ death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \ } # else # define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) # endif // This macro is for implementing ASSERT_DEATH*, EXPECT_DEATH*, // ASSERT_EXIT*, and EXPECT_EXIT*. # define GTEST_DEATH_TEST_(statement, predicate, regex, fail) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::AlwaysTrue()) { \ const ::testing::internal::RE& gtest_regex = (regex); \ ::testing::internal::DeathTest* gtest_dt; \ if (!::testing::internal::DeathTest::Create(#statement, >est_regex, \ __FILE__, __LINE__, >est_dt)) { \ goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \ } \ if (gtest_dt != NULL) { \ ::testing::internal::scoped_ptr< ::testing::internal::DeathTest> \ gtest_dt_ptr(gtest_dt); \ switch (gtest_dt->AssumeRole()) { \ case ::testing::internal::DeathTest::OVERSEE_TEST: \ if (!gtest_dt->Passed(predicate(gtest_dt->Wait()))) { \ goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \ } \ break; \ case ::testing::internal::DeathTest::EXECUTE_TEST: { \ ::testing::internal::DeathTest::ReturnSentinel \ gtest_sentinel(gtest_dt); \ GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, gtest_dt); \ gtest_dt->Abort(::testing::internal::DeathTest::TEST_DID_NOT_DIE); \ break; \ } \ default: \ break; \ } \ } \ } else \ GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__): \ fail(::testing::internal::DeathTest::LastMessage()) // The symbol "fail" here expands to something into which a message // can be streamed. // This macro is for implementing ASSERT/EXPECT_DEBUG_DEATH when compiled in // NDEBUG mode. In this case we need the statements to be executed, the regex is // ignored, and the macro must accept a streamed message even though the message // is never printed. # define GTEST_EXECUTE_STATEMENT_(statement, regex) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::AlwaysTrue()) { \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ } else \ ::testing::Message() // A class representing the parsed contents of the // --gtest_internal_run_death_test flag, as it existed when // RUN_ALL_TESTS was called. class InternalRunDeathTestFlag { public: InternalRunDeathTestFlag(const std::string& a_file, int a_line, int an_index, int a_write_fd) : file_(a_file), line_(a_line), index_(an_index), write_fd_(a_write_fd) {} ~InternalRunDeathTestFlag() { if (write_fd_ >= 0) posix::Close(write_fd_); } const std::string& file() const { return file_; } int line() const { return line_; } int index() const { return index_; } int write_fd() const { return write_fd_; } private: std::string file_; int line_; int index_; int write_fd_; GTEST_DISALLOW_COPY_AND_ASSIGN_(InternalRunDeathTestFlag); }; // Returns a newly created InternalRunDeathTestFlag object with fields // initialized from the GTEST_FLAG(internal_run_death_test) flag if // the flag is specified; otherwise returns NULL. InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag(); #else // GTEST_HAS_DEATH_TEST // This macro is used for implementing macros such as // EXPECT_DEATH_IF_SUPPORTED and ASSERT_DEATH_IF_SUPPORTED on systems where // death tests are not supported. Those macros must compile on such systems // iff EXPECT_DEATH and ASSERT_DEATH compile with the same parameters on // systems that support death tests. This allows one to write such a macro // on a system that does not support death tests and be sure that it will // compile on a death-test supporting system. // // Parameters: // statement - A statement that a macro such as EXPECT_DEATH would test // for program termination. This macro has to make sure this // statement is compiled but not executed, to ensure that // EXPECT_DEATH_IF_SUPPORTED compiles with a certain // parameter iff EXPECT_DEATH compiles with it. // regex - A regex that a macro such as EXPECT_DEATH would use to test // the output of statement. This parameter has to be // compiled but not evaluated by this macro, to ensure that // this macro only accepts expressions that a macro such as // EXPECT_DEATH would accept. // terminator - Must be an empty statement for EXPECT_DEATH_IF_SUPPORTED // and a return statement for ASSERT_DEATH_IF_SUPPORTED. // This ensures that ASSERT_DEATH_IF_SUPPORTED will not // compile inside functions where ASSERT_DEATH doesn't // compile. // // The branch that has an always false condition is used to ensure that // statement and regex are compiled (and thus syntactically correct) but // never executed. The unreachable code macro protects the terminator // statement from generating an 'unreachable code' warning in case // statement unconditionally returns or throws. The Message constructor at // the end allows the syntax of streaming additional messages into the // macro, for compilational compatibility with EXPECT_DEATH/ASSERT_DEATH. # define GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, terminator) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (::testing::internal::AlwaysTrue()) { \ GTEST_LOG_(WARNING) \ << "Death tests are not supported on this platform.\n" \ << "Statement '" #statement "' cannot be verified."; \ } else if (::testing::internal::AlwaysFalse()) { \ ::testing::internal::RE::PartialMatch(".*", (regex)); \ GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ terminator; \ } else \ ::testing::Message() #endif // GTEST_HAS_DEATH_TEST } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_ namespace testing { // This flag controls the style of death tests. Valid values are "threadsafe", // meaning that the death test child process will re-execute the test binary // from the start, running only a single death test, or "fast", // meaning that the child process will execute the test logic immediately // after forking. GTEST_DECLARE_string_(death_test_style); #if GTEST_HAS_DEATH_TEST namespace internal { // Returns a Boolean value indicating whether the caller is currently // executing in the context of the death test child process. Tools such as // Valgrind heap checkers may need this to modify their behavior in death // tests. IMPORTANT: This is an internal utility. Using it may break the // implementation of death tests. User code MUST NOT use it. GTEST_API_ bool InDeathTestChild(); } // namespace internal // The following macros are useful for writing death tests. // Here's what happens when an ASSERT_DEATH* or EXPECT_DEATH* is // executed: // // 1. It generates a warning if there is more than one active // thread. This is because it's safe to fork() or clone() only // when there is a single thread. // // 2. The parent process clone()s a sub-process and runs the death // test in it; the sub-process exits with code 0 at the end of the // death test, if it hasn't exited already. // // 3. The parent process waits for the sub-process to terminate. // // 4. The parent process checks the exit code and error message of // the sub-process. // // Examples: // // ASSERT_DEATH(server.SendMessage(56, "Hello"), "Invalid port number"); // for (int i = 0; i < 5; i++) { // EXPECT_DEATH(server.ProcessRequest(i), // "Invalid request .* in ProcessRequest()") // << "Failed to die on request " << i; // } // // ASSERT_EXIT(server.ExitNow(), ::testing::ExitedWithCode(0), "Exiting"); // // bool KilledBySIGHUP(int exit_code) { // return WIFSIGNALED(exit_code) && WTERMSIG(exit_code) == SIGHUP; // } // // ASSERT_EXIT(client.HangUpServer(), KilledBySIGHUP, "Hanging up!"); // // On the regular expressions used in death tests: // // On POSIX-compliant systems (*nix), we use the library, // which uses the POSIX extended regex syntax. // // On other platforms (e.g. Windows), we only support a simple regex // syntax implemented as part of Google Test. This limited // implementation should be enough most of the time when writing // death tests; though it lacks many features you can find in PCRE // or POSIX extended regex syntax. For example, we don't support // union ("x|y"), grouping ("(xy)"), brackets ("[xy]"), and // repetition count ("x{5,7}"), among others. // // Below is the syntax that we do support. We chose it to be a // subset of both PCRE and POSIX extended regex, so it's easy to // learn wherever you come from. In the following: 'A' denotes a // literal character, period (.), or a single \\ escape sequence; // 'x' and 'y' denote regular expressions; 'm' and 'n' are for // natural numbers. // // c matches any literal character c // \\d matches any decimal digit // \\D matches any character that's not a decimal digit // \\f matches \f // \\n matches \n // \\r matches \r // \\s matches any ASCII whitespace, including \n // \\S matches any character that's not a whitespace // \\t matches \t // \\v matches \v // \\w matches any letter, _, or decimal digit // \\W matches any character that \\w doesn't match // \\c matches any literal character c, which must be a punctuation // . matches any single character except \n // A? matches 0 or 1 occurrences of A // A* matches 0 or many occurrences of A // A+ matches 1 or many occurrences of A // ^ matches the beginning of a string (not that of each line) // $ matches the end of a string (not that of each line) // xy matches x followed by y // // If you accidentally use PCRE or POSIX extended regex features // not implemented by us, you will get a run-time failure. In that // case, please try to rewrite your regular expression within the // above syntax. // // This implementation is *not* meant to be as highly tuned or robust // as a compiled regex library, but should perform well enough for a // death test, which already incurs significant overhead by launching // a child process. // // Known caveats: // // A "threadsafe" style death test obtains the path to the test // program from argv[0] and re-executes it in the sub-process. For // simplicity, the current implementation doesn't search the PATH // when launching the sub-process. This means that the user must // invoke the test program via a path that contains at least one // path separator (e.g. path/to/foo_test and // /absolute/path/to/bar_test are fine, but foo_test is not). This // is rarely a problem as people usually don't put the test binary // directory in PATH. // // TODO(wan@google.com): make thread-safe death tests search the PATH. // Asserts that a given statement causes the program to exit, with an // integer exit status that satisfies predicate, and emitting error output // that matches regex. # define ASSERT_EXIT(statement, predicate, regex) \ GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_FATAL_FAILURE_) // Like ASSERT_EXIT, but continues on to successive tests in the // test case, if any: # define EXPECT_EXIT(statement, predicate, regex) \ GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_NONFATAL_FAILURE_) // Asserts that a given statement causes the program to exit, either by // explicitly exiting with a nonzero exit code or being killed by a // signal, and emitting error output that matches regex. # define ASSERT_DEATH(statement, regex) \ ASSERT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex) // Like ASSERT_DEATH, but continues on to successive tests in the // test case, if any: # define EXPECT_DEATH(statement, regex) \ EXPECT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex) // Two predicate classes that can be used in {ASSERT,EXPECT}_EXIT*: // Tests that an exit code describes a normal exit with a given exit code. class GTEST_API_ ExitedWithCode { public: explicit ExitedWithCode(int exit_code); bool operator()(int exit_status) const; private: // No implementation - assignment is unsupported. void operator=(const ExitedWithCode& other); const int exit_code_; }; # if !GTEST_OS_WINDOWS // Tests that an exit code describes an exit due to termination by a // given signal. class GTEST_API_ KilledBySignal { public: explicit KilledBySignal(int signum); bool operator()(int exit_status) const; private: const int signum_; }; # endif // !GTEST_OS_WINDOWS // EXPECT_DEBUG_DEATH asserts that the given statements die in debug mode. // The death testing framework causes this to have interesting semantics, // since the sideeffects of the call are only visible in opt mode, and not // in debug mode. // // In practice, this can be used to test functions that utilize the // LOG(DFATAL) macro using the following style: // // int DieInDebugOr12(int* sideeffect) { // if (sideeffect) { // *sideeffect = 12; // } // LOG(DFATAL) << "death"; // return 12; // } // // TEST(TestCase, TestDieOr12WorksInDgbAndOpt) { // int sideeffect = 0; // // Only asserts in dbg. // EXPECT_DEBUG_DEATH(DieInDebugOr12(&sideeffect), "death"); // // #ifdef NDEBUG // // opt-mode has sideeffect visible. // EXPECT_EQ(12, sideeffect); // #else // // dbg-mode no visible sideeffect. // EXPECT_EQ(0, sideeffect); // #endif // } // // This will assert that DieInDebugReturn12InOpt() crashes in debug // mode, usually due to a DCHECK or LOG(DFATAL), but returns the // appropriate fallback value (12 in this case) in opt mode. If you // need to test that a function has appropriate side-effects in opt // mode, include assertions against the side-effects. A general // pattern for this is: // // EXPECT_DEBUG_DEATH({ // // Side-effects here will have an effect after this statement in // // opt mode, but none in debug mode. // EXPECT_EQ(12, DieInDebugOr12(&sideeffect)); // }, "death"); // # ifdef NDEBUG # define EXPECT_DEBUG_DEATH(statement, regex) \ GTEST_EXECUTE_STATEMENT_(statement, regex) # define ASSERT_DEBUG_DEATH(statement, regex) \ GTEST_EXECUTE_STATEMENT_(statement, regex) # else # define EXPECT_DEBUG_DEATH(statement, regex) \ EXPECT_DEATH(statement, regex) # define ASSERT_DEBUG_DEATH(statement, regex) \ ASSERT_DEATH(statement, regex) # endif // NDEBUG for EXPECT_DEBUG_DEATH #endif // GTEST_HAS_DEATH_TEST // EXPECT_DEATH_IF_SUPPORTED(statement, regex) and // ASSERT_DEATH_IF_SUPPORTED(statement, regex) expand to real death tests if // death tests are supported; otherwise they just issue a warning. This is // useful when you are combining death test assertions with normal test // assertions in one test. #if GTEST_HAS_DEATH_TEST # define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \ EXPECT_DEATH(statement, regex) # define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \ ASSERT_DEATH(statement, regex) #else # define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \ GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, ) # define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \ GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, return) #endif } // namespace testing #endif // GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_ // This file was GENERATED by command: // pump.py gtest-param-test.h.pump // DO NOT EDIT BY HAND!!! // Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: vladl@google.com (Vlad Losev) // // Macros and functions for implementing parameterized tests // in Google C++ Testing Framework (Google Test) // // This file is generated by a SCRIPT. DO NOT EDIT BY HAND! // #ifndef GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_ #define GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_ // Value-parameterized tests allow you to test your code with different // parameters without writing multiple copies of the same test. // // Here is how you use value-parameterized tests: #if 0 // To write value-parameterized tests, first you should define a fixture // class. It is usually derived from testing::TestWithParam (see below for // another inheritance scheme that's sometimes useful in more complicated // class hierarchies), where the type of your parameter values. // TestWithParam is itself derived from testing::Test. T can be any // copyable type. If it's a raw pointer, you are responsible for managing the // lifespan of the pointed values. class FooTest : public ::testing::TestWithParam { // You can implement all the usual class fixture members here. }; // Then, use the TEST_P macro to define as many parameterized tests // for this fixture as you want. The _P suffix is for "parameterized" // or "pattern", whichever you prefer to think. TEST_P(FooTest, DoesBlah) { // Inside a test, access the test parameter with the GetParam() method // of the TestWithParam class: EXPECT_TRUE(foo.Blah(GetParam())); ... } TEST_P(FooTest, HasBlahBlah) { ... } // Finally, you can use INSTANTIATE_TEST_CASE_P to instantiate the test // case with any set of parameters you want. Google Test defines a number // of functions for generating test parameters. They return what we call // (surprise!) parameter generators. Here is a summary of them, which // are all in the testing namespace: // // // Range(begin, end [, step]) - Yields values {begin, begin+step, // begin+step+step, ...}. The values do not // include end. step defaults to 1. // Values(v1, v2, ..., vN) - Yields values {v1, v2, ..., vN}. // ValuesIn(container) - Yields values from a C-style array, an STL // ValuesIn(begin,end) container, or an iterator range [begin, end). // Bool() - Yields sequence {false, true}. // Combine(g1, g2, ..., gN) - Yields all combinations (the Cartesian product // for the math savvy) of the values generated // by the N generators. // // For more details, see comments at the definitions of these functions below // in this file. // // The following statement will instantiate tests from the FooTest test case // each with parameter values "meeny", "miny", and "moe". INSTANTIATE_TEST_CASE_P(InstantiationName, FooTest, Values("meeny", "miny", "moe")); // To distinguish different instances of the pattern, (yes, you // can instantiate it more then once) the first argument to the // INSTANTIATE_TEST_CASE_P macro is a prefix that will be added to the // actual test case name. Remember to pick unique prefixes for different // instantiations. The tests from the instantiation above will have // these names: // // * InstantiationName/FooTest.DoesBlah/0 for "meeny" // * InstantiationName/FooTest.DoesBlah/1 for "miny" // * InstantiationName/FooTest.DoesBlah/2 for "moe" // * InstantiationName/FooTest.HasBlahBlah/0 for "meeny" // * InstantiationName/FooTest.HasBlahBlah/1 for "miny" // * InstantiationName/FooTest.HasBlahBlah/2 for "moe" // // You can use these names in --gtest_filter. // // This statement will instantiate all tests from FooTest again, each // with parameter values "cat" and "dog": const char* pets[] = {"cat", "dog"}; INSTANTIATE_TEST_CASE_P(AnotherInstantiationName, FooTest, ValuesIn(pets)); // The tests from the instantiation above will have these names: // // * AnotherInstantiationName/FooTest.DoesBlah/0 for "cat" // * AnotherInstantiationName/FooTest.DoesBlah/1 for "dog" // * AnotherInstantiationName/FooTest.HasBlahBlah/0 for "cat" // * AnotherInstantiationName/FooTest.HasBlahBlah/1 for "dog" // // Please note that INSTANTIATE_TEST_CASE_P will instantiate all tests // in the given test case, whether their definitions come before or // AFTER the INSTANTIATE_TEST_CASE_P statement. // // Please also note that generator expressions (including parameters to the // generators) are evaluated in InitGoogleTest(), after main() has started. // This allows the user on one hand, to adjust generator parameters in order // to dynamically determine a set of tests to run and on the other hand, // give the user a chance to inspect the generated tests with Google Test // reflection API before RUN_ALL_TESTS() is executed. // // You can see samples/sample7_unittest.cc and samples/sample8_unittest.cc // for more examples. // // In the future, we plan to publish the API for defining new parameter // generators. But for now this interface remains part of the internal // implementation and is subject to change. // // // A parameterized test fixture must be derived from testing::Test and from // testing::WithParamInterface, where T is the type of the parameter // values. Inheriting from TestWithParam satisfies that requirement because // TestWithParam inherits from both Test and WithParamInterface. In more // complicated hierarchies, however, it is occasionally useful to inherit // separately from Test and WithParamInterface. For example: class BaseTest : public ::testing::Test { // You can inherit all the usual members for a non-parameterized test // fixture here. }; class DerivedTest : public BaseTest, public ::testing::WithParamInterface { // The usual test fixture members go here too. }; TEST_F(BaseTest, HasFoo) { // This is an ordinary non-parameterized test. } TEST_P(DerivedTest, DoesBlah) { // GetParam works just the same here as if you inherit from TestWithParam. EXPECT_TRUE(foo.Blah(GetParam())); } #endif // 0 #if !GTEST_OS_SYMBIAN # include #endif // scripts/fuse_gtest.py depends on gtest's own header being #included // *unconditionally*. Therefore these #includes cannot be moved // inside #if GTEST_HAS_PARAM_TEST. // Copyright 2008 Google Inc. // All Rights Reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: vladl@google.com (Vlad Losev) // Type and function utilities for implementing parameterized tests. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_ #include #include #include // scripts/fuse_gtest.py depends on gtest's own header being #included // *unconditionally*. Therefore these #includes cannot be moved // inside #if GTEST_HAS_PARAM_TEST. // Copyright 2003 Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Authors: Dan Egnor (egnor@google.com) // // A "smart" pointer type with reference tracking. Every pointer to a // particular object is kept on a circular linked list. When the last pointer // to an object is destroyed or reassigned, the object is deleted. // // Used properly, this deletes the object when the last reference goes away. // There are several caveats: // - Like all reference counting schemes, cycles lead to leaks. // - Each smart pointer is actually two pointers (8 bytes instead of 4). // - Every time a pointer is assigned, the entire list of pointers to that // object is traversed. This class is therefore NOT SUITABLE when there // will often be more than two or three pointers to a particular object. // - References are only tracked as long as linked_ptr<> objects are copied. // If a linked_ptr<> is converted to a raw pointer and back, BAD THINGS // will happen (double deletion). // // A good use of this class is storing object references in STL containers. // You can safely put linked_ptr<> in a vector<>. // Other uses may not be as good. // // Note: If you use an incomplete type with linked_ptr<>, the class // *containing* linked_ptr<> must have a constructor and destructor (even // if they do nothing!). // // Bill Gibbons suggested we use something like this. // // Thread Safety: // Unlike other linked_ptr implementations, in this implementation // a linked_ptr object is thread-safe in the sense that: // - it's safe to copy linked_ptr objects concurrently, // - it's safe to copy *from* a linked_ptr and read its underlying // raw pointer (e.g. via get()) concurrently, and // - it's safe to write to two linked_ptrs that point to the same // shared object concurrently. // TODO(wan@google.com): rename this to safe_linked_ptr to avoid // confusion with normal linked_ptr. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_ #include #include namespace testing { namespace internal { // Protects copying of all linked_ptr objects. GTEST_API_ GTEST_DECLARE_STATIC_MUTEX_(g_linked_ptr_mutex); // This is used internally by all instances of linked_ptr<>. It needs to be // a non-template class because different types of linked_ptr<> can refer to // the same object (linked_ptr(obj) vs linked_ptr(obj)). // So, it needs to be possible for different types of linked_ptr to participate // in the same circular linked list, so we need a single class type here. // // DO NOT USE THIS CLASS DIRECTLY YOURSELF. Use linked_ptr. class linked_ptr_internal { public: // Create a new circle that includes only this instance. void join_new() { next_ = this; } // Many linked_ptr operations may change p.link_ for some linked_ptr // variable p in the same circle as this object. Therefore we need // to prevent two such operations from occurring concurrently. // // Note that different types of linked_ptr objects can coexist in a // circle (e.g. linked_ptr, linked_ptr, and // linked_ptr). Therefore we must use a single mutex to // protect all linked_ptr objects. This can create serious // contention in production code, but is acceptable in a testing // framework. // Join an existing circle. void join(linked_ptr_internal const* ptr) GTEST_LOCK_EXCLUDED_(g_linked_ptr_mutex) { MutexLock lock(&g_linked_ptr_mutex); linked_ptr_internal const* p = ptr; while (p->next_ != ptr) { assert(p->next_ != this && "Trying to join() a linked ring we are already in. " "Is GMock thread safety enabled?"); p = p->next_; } p->next_ = this; next_ = ptr; } // Leave whatever circle we're part of. Returns true if we were the // last member of the circle. Once this is done, you can join() another. bool depart() GTEST_LOCK_EXCLUDED_(g_linked_ptr_mutex) { MutexLock lock(&g_linked_ptr_mutex); if (next_ == this) return true; linked_ptr_internal const* p = next_; while (p->next_ != this) { assert(p->next_ != next_ && "Trying to depart() a linked ring we are not in. " "Is GMock thread safety enabled?"); p = p->next_; } p->next_ = next_; return false; } private: mutable linked_ptr_internal const* next_; }; template class linked_ptr { public: typedef T element_type; // Take over ownership of a raw pointer. This should happen as soon as // possible after the object is created. explicit linked_ptr(T* ptr = NULL) { capture(ptr); } ~linked_ptr() { depart(); } // Copy an existing linked_ptr<>, adding ourselves to the list of references. template linked_ptr(linked_ptr const& ptr) { copy(&ptr); } linked_ptr(linked_ptr const& ptr) { // NOLINT assert(&ptr != this); copy(&ptr); } // Assignment releases the old value and acquires the new. template linked_ptr& operator=(linked_ptr const& ptr) { depart(); copy(&ptr); return *this; } linked_ptr& operator=(linked_ptr const& ptr) { if (&ptr != this) { depart(); copy(&ptr); } return *this; } // Smart pointer members. void reset(T* ptr = NULL) { depart(); capture(ptr); } T* get() const { return value_; } T* operator->() const { return value_; } T& operator*() const { return *value_; } bool operator==(T* p) const { return value_ == p; } bool operator!=(T* p) const { return value_ != p; } template bool operator==(linked_ptr const& ptr) const { return value_ == ptr.get(); } template bool operator!=(linked_ptr const& ptr) const { return value_ != ptr.get(); } private: template friend class linked_ptr; T* value_; linked_ptr_internal link_; void depart() { if (link_.depart()) delete value_; } void capture(T* ptr) { value_ = ptr; link_.join_new(); } template void copy(linked_ptr const* ptr) { value_ = ptr->get(); if (value_) link_.join(&ptr->link_); else link_.join_new(); } }; template inline bool operator==(T* ptr, const linked_ptr& x) { return ptr == x.get(); } template inline bool operator!=(T* ptr, const linked_ptr& x) { return ptr != x.get(); } // A function to convert T* into linked_ptr // Doing e.g. make_linked_ptr(new FooBarBaz(arg)) is a shorter notation // for linked_ptr >(new FooBarBaz(arg)) template linked_ptr make_linked_ptr(T* ptr) { return linked_ptr(ptr); } } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_ // Copyright 2007, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // Google Test - The Google C++ Testing Framework // // This file implements a universal value printer that can print a // value of any type T: // // void ::testing::internal::UniversalPrinter::Print(value, ostream_ptr); // // A user can teach this function how to print a class type T by // defining either operator<<() or PrintTo() in the namespace that // defines T. More specifically, the FIRST defined function in the // following list will be used (assuming T is defined in namespace // foo): // // 1. foo::PrintTo(const T&, ostream*) // 2. operator<<(ostream&, const T&) defined in either foo or the // global namespace. // // If none of the above is defined, it will print the debug string of // the value if it is a protocol buffer, or print the raw bytes in the // value otherwise. // // To aid debugging: when T is a reference type, the address of the // value is also printed; when T is a (const) char pointer, both the // pointer value and the NUL-terminated string it points to are // printed. // // We also provide some convenient wrappers: // // // Prints a value to a string. For a (const or not) char // // pointer, the NUL-terminated string (but not the pointer) is // // printed. // std::string ::testing::PrintToString(const T& value); // // // Prints a value tersely: for a reference type, the referenced // // value (but not the address) is printed; for a (const or not) char // // pointer, the NUL-terminated string (but not the pointer) is // // printed. // void ::testing::internal::UniversalTersePrint(const T& value, ostream*); // // // Prints value using the type inferred by the compiler. The difference // // from UniversalTersePrint() is that this function prints both the // // pointer and the NUL-terminated string for a (const or not) char pointer. // void ::testing::internal::UniversalPrint(const T& value, ostream*); // // // Prints the fields of a tuple tersely to a string vector, one // // element for each field. Tuple support must be enabled in // // gtest-port.h. // std::vector UniversalTersePrintTupleFieldsToStrings( // const Tuple& value); // // Known limitation: // // The print primitives print the elements of an STL-style container // using the compiler-inferred type of *iter where iter is a // const_iterator of the container. When const_iterator is an input // iterator but not a forward iterator, this inferred type may not // match value_type, and the print output may be incorrect. In // practice, this is rarely a problem as for most containers // const_iterator is a forward iterator. We'll fix this if there's an // actual need for it. Note that this fix cannot rely on value_type // being defined as many user-defined container types don't have // value_type. #ifndef GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_ #define GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_ #include // NOLINT #include #include #include #include #if GTEST_HAS_STD_TUPLE_ # include #endif namespace testing { // Definitions in the 'internal' and 'internal2' name spaces are // subject to change without notice. DO NOT USE THEM IN USER CODE! namespace internal2 { // Prints the given number of bytes in the given object to the given // ostream. GTEST_API_ void PrintBytesInObjectTo(const unsigned char* obj_bytes, size_t count, ::std::ostream* os); // For selecting which printer to use when a given type has neither << // nor PrintTo(). enum TypeKind { kProtobuf, // a protobuf type kConvertibleToInteger, // a type implicitly convertible to BiggestInt // (e.g. a named or unnamed enum type) kOtherType // anything else }; // TypeWithoutFormatter::PrintValue(value, os) is called // by the universal printer to print a value of type T when neither // operator<< nor PrintTo() is defined for T, where kTypeKind is the // "kind" of T as defined by enum TypeKind. template class TypeWithoutFormatter { public: // This default version is called when kTypeKind is kOtherType. static void PrintValue(const T& value, ::std::ostream* os) { PrintBytesInObjectTo(reinterpret_cast(&value), sizeof(value), os); } }; // We print a protobuf using its ShortDebugString() when the string // doesn't exceed this many characters; otherwise we print it using // DebugString() for better readability. const size_t kProtobufOneLinerMaxLength = 50; template class TypeWithoutFormatter { public: static void PrintValue(const T& value, ::std::ostream* os) { const ::testing::internal::string short_str = value.ShortDebugString(); const ::testing::internal::string pretty_str = short_str.length() <= kProtobufOneLinerMaxLength ? short_str : ("\n" + value.DebugString()); *os << ("<" + pretty_str + ">"); } }; template class TypeWithoutFormatter { public: // Since T has no << operator or PrintTo() but can be implicitly // converted to BiggestInt, we print it as a BiggestInt. // // Most likely T is an enum type (either named or unnamed), in which // case printing it as an integer is the desired behavior. In case // T is not an enum, printing it as an integer is the best we can do // given that it has no user-defined printer. static void PrintValue(const T& value, ::std::ostream* os) { const internal::BiggestInt kBigInt = value; *os << kBigInt; } }; // Prints the given value to the given ostream. If the value is a // protocol message, its debug string is printed; if it's an enum or // of a type implicitly convertible to BiggestInt, it's printed as an // integer; otherwise the bytes in the value are printed. This is // what UniversalPrinter::Print() does when it knows nothing about // type T and T has neither << operator nor PrintTo(). // // A user can override this behavior for a class type Foo by defining // a << operator in the namespace where Foo is defined. // // We put this operator in namespace 'internal2' instead of 'internal' // to simplify the implementation, as much code in 'internal' needs to // use << in STL, which would conflict with our own << were it defined // in 'internal'. // // Note that this operator<< takes a generic std::basic_ostream type instead of the more restricted std::ostream. If // we define it to take an std::ostream instead, we'll get an // "ambiguous overloads" compiler error when trying to print a type // Foo that supports streaming to std::basic_ostream, as the compiler cannot tell whether // operator<<(std::ostream&, const T&) or // operator<<(std::basic_stream, const Foo&) is more // specific. template ::std::basic_ostream& operator<<( ::std::basic_ostream& os, const T& x) { TypeWithoutFormatter::value ? kProtobuf : internal::ImplicitlyConvertible::value ? kConvertibleToInteger : kOtherType)>::PrintValue(x, &os); return os; } } // namespace internal2 } // namespace testing // This namespace MUST NOT BE NESTED IN ::testing, or the name look-up // magic needed for implementing UniversalPrinter won't work. namespace testing_internal { // Used to print a value that is not an STL-style container when the // user doesn't define PrintTo() for it. template void DefaultPrintNonContainerTo(const T& value, ::std::ostream* os) { // With the following statement, during unqualified name lookup, // testing::internal2::operator<< appears as if it was declared in // the nearest enclosing namespace that contains both // ::testing_internal and ::testing::internal2, i.e. the global // namespace. For more details, refer to the C++ Standard section // 7.3.4-1 [namespace.udir]. This allows us to fall back onto // testing::internal2::operator<< in case T doesn't come with a << // operator. // // We cannot write 'using ::testing::internal2::operator<<;', which // gcc 3.3 fails to compile due to a compiler bug. using namespace ::testing::internal2; // NOLINT // Assuming T is defined in namespace foo, in the next statement, // the compiler will consider all of: // // 1. foo::operator<< (thanks to Koenig look-up), // 2. ::operator<< (as the current namespace is enclosed in ::), // 3. testing::internal2::operator<< (thanks to the using statement above). // // The operator<< whose type matches T best will be picked. // // We deliberately allow #2 to be a candidate, as sometimes it's // impossible to define #1 (e.g. when foo is ::std, defining // anything in it is undefined behavior unless you are a compiler // vendor.). *os << value; } } // namespace testing_internal namespace testing { namespace internal { // UniversalPrinter::Print(value, ostream_ptr) prints the given // value to the given ostream. The caller must ensure that // 'ostream_ptr' is not NULL, or the behavior is undefined. // // We define UniversalPrinter as a class template (as opposed to a // function template), as we need to partially specialize it for // reference types, which cannot be done with function templates. template class UniversalPrinter; template void UniversalPrint(const T& value, ::std::ostream* os); // Used to print an STL-style container when the user doesn't define // a PrintTo() for it. template void DefaultPrintTo(IsContainer /* dummy */, false_type /* is not a pointer */, const C& container, ::std::ostream* os) { const size_t kMaxCount = 32; // The maximum number of elements to print. *os << '{'; size_t count = 0; for (typename C::const_iterator it = container.begin(); it != container.end(); ++it, ++count) { if (count > 0) { *os << ','; if (count == kMaxCount) { // Enough has been printed. *os << " ..."; break; } } *os << ' '; // We cannot call PrintTo(*it, os) here as PrintTo() doesn't // handle *it being a native array. internal::UniversalPrint(*it, os); } if (count > 0) { *os << ' '; } *os << '}'; } // Used to print a pointer that is neither a char pointer nor a member // pointer, when the user doesn't define PrintTo() for it. (A member // variable pointer or member function pointer doesn't really point to // a location in the address space. Their representation is // implementation-defined. Therefore they will be printed as raw // bytes.) template void DefaultPrintTo(IsNotContainer /* dummy */, true_type /* is a pointer */, T* p, ::std::ostream* os) { if (p == NULL) { *os << "NULL"; } else { // C++ doesn't allow casting from a function pointer to any object // pointer. // // IsTrue() silences warnings: "Condition is always true", // "unreachable code". if (IsTrue(ImplicitlyConvertible::value)) { // T is not a function type. We just call << to print p, // relying on ADL to pick up user-defined << for their pointer // types, if any. *os << p; } else { // T is a function type, so '*os << p' doesn't do what we want // (it just prints p as bool). We want to print p as a const // void*. However, we cannot cast it to const void* directly, // even using reinterpret_cast, as earlier versions of gcc // (e.g. 3.4.5) cannot compile the cast when p is a function // pointer. Casting to UInt64 first solves the problem. *os << reinterpret_cast( reinterpret_cast(p)); } } } // Used to print a non-container, non-pointer value when the user // doesn't define PrintTo() for it. template void DefaultPrintTo(IsNotContainer /* dummy */, false_type /* is not a pointer */, const T& value, ::std::ostream* os) { ::testing_internal::DefaultPrintNonContainerTo(value, os); } // Prints the given value using the << operator if it has one; // otherwise prints the bytes in it. This is what // UniversalPrinter::Print() does when PrintTo() is not specialized // or overloaded for type T. // // A user can override this behavior for a class type Foo by defining // an overload of PrintTo() in the namespace where Foo is defined. We // give the user this option as sometimes defining a << operator for // Foo is not desirable (e.g. the coding style may prevent doing it, // or there is already a << operator but it doesn't do what the user // wants). template void PrintTo(const T& value, ::std::ostream* os) { // DefaultPrintTo() is overloaded. The type of its first two // arguments determine which version will be picked. If T is an // STL-style container, the version for container will be called; if // T is a pointer, the pointer version will be called; otherwise the // generic version will be called. // // Note that we check for container types here, prior to we check // for protocol message types in our operator<<. The rationale is: // // For protocol messages, we want to give people a chance to // override Google Mock's format by defining a PrintTo() or // operator<<. For STL containers, other formats can be // incompatible with Google Mock's format for the container // elements; therefore we check for container types here to ensure // that our format is used. // // The second argument of DefaultPrintTo() is needed to bypass a bug // in Symbian's C++ compiler that prevents it from picking the right // overload between: // // PrintTo(const T& x, ...); // PrintTo(T* x, ...); DefaultPrintTo(IsContainerTest(0), is_pointer(), value, os); } // The following list of PrintTo() overloads tells // UniversalPrinter::Print() how to print standard types (built-in // types, strings, plain arrays, and pointers). // Overloads for various char types. GTEST_API_ void PrintTo(unsigned char c, ::std::ostream* os); GTEST_API_ void PrintTo(signed char c, ::std::ostream* os); inline void PrintTo(char c, ::std::ostream* os) { // When printing a plain char, we always treat it as unsigned. This // way, the output won't be affected by whether the compiler thinks // char is signed or not. PrintTo(static_cast(c), os); } // Overloads for other simple built-in types. inline void PrintTo(bool x, ::std::ostream* os) { *os << (x ? "true" : "false"); } // Overload for wchar_t type. // Prints a wchar_t as a symbol if it is printable or as its internal // code otherwise and also as its decimal code (except for L'\0'). // The L'\0' char is printed as "L'\\0'". The decimal code is printed // as signed integer when wchar_t is implemented by the compiler // as a signed type and is printed as an unsigned integer when wchar_t // is implemented as an unsigned type. GTEST_API_ void PrintTo(wchar_t wc, ::std::ostream* os); // Overloads for C strings. GTEST_API_ void PrintTo(const char* s, ::std::ostream* os); inline void PrintTo(char* s, ::std::ostream* os) { PrintTo(ImplicitCast_(s), os); } // signed/unsigned char is often used for representing binary data, so // we print pointers to it as void* to be safe. inline void PrintTo(const signed char* s, ::std::ostream* os) { PrintTo(ImplicitCast_(s), os); } inline void PrintTo(signed char* s, ::std::ostream* os) { PrintTo(ImplicitCast_(s), os); } inline void PrintTo(const unsigned char* s, ::std::ostream* os) { PrintTo(ImplicitCast_(s), os); } inline void PrintTo(unsigned char* s, ::std::ostream* os) { PrintTo(ImplicitCast_(s), os); } // MSVC can be configured to define wchar_t as a typedef of unsigned // short. It defines _NATIVE_WCHAR_T_DEFINED when wchar_t is a native // type. When wchar_t is a typedef, defining an overload for const // wchar_t* would cause unsigned short* be printed as a wide string, // possibly causing invalid memory accesses. #if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED) // Overloads for wide C strings GTEST_API_ void PrintTo(const wchar_t* s, ::std::ostream* os); inline void PrintTo(wchar_t* s, ::std::ostream* os) { PrintTo(ImplicitCast_(s), os); } #endif // Overload for C arrays. Multi-dimensional arrays are printed // properly. // Prints the given number of elements in an array, without printing // the curly braces. template void PrintRawArrayTo(const T a[], size_t count, ::std::ostream* os) { UniversalPrint(a[0], os); for (size_t i = 1; i != count; i++) { *os << ", "; UniversalPrint(a[i], os); } } // Overloads for ::string and ::std::string. #if GTEST_HAS_GLOBAL_STRING GTEST_API_ void PrintStringTo(const ::string&s, ::std::ostream* os); inline void PrintTo(const ::string& s, ::std::ostream* os) { PrintStringTo(s, os); } #endif // GTEST_HAS_GLOBAL_STRING GTEST_API_ void PrintStringTo(const ::std::string&s, ::std::ostream* os); inline void PrintTo(const ::std::string& s, ::std::ostream* os) { PrintStringTo(s, os); } // Overloads for ::wstring and ::std::wstring. #if GTEST_HAS_GLOBAL_WSTRING GTEST_API_ void PrintWideStringTo(const ::wstring&s, ::std::ostream* os); inline void PrintTo(const ::wstring& s, ::std::ostream* os) { PrintWideStringTo(s, os); } #endif // GTEST_HAS_GLOBAL_WSTRING #if GTEST_HAS_STD_WSTRING GTEST_API_ void PrintWideStringTo(const ::std::wstring&s, ::std::ostream* os); inline void PrintTo(const ::std::wstring& s, ::std::ostream* os) { PrintWideStringTo(s, os); } #endif // GTEST_HAS_STD_WSTRING #if GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_ // Helper function for printing a tuple. T must be instantiated with // a tuple type. template void PrintTupleTo(const T& t, ::std::ostream* os); #endif // GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_ #if GTEST_HAS_TR1_TUPLE // Overload for ::std::tr1::tuple. Needed for printing function arguments, // which are packed as tuples. // Overloaded PrintTo() for tuples of various arities. We support // tuples of up-to 10 fields. The following implementation works // regardless of whether tr1::tuple is implemented using the // non-standard variadic template feature or not. inline void PrintTo(const ::std::tr1::tuple<>& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo(const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } template void PrintTo( const ::std::tr1::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } #endif // GTEST_HAS_TR1_TUPLE #if GTEST_HAS_STD_TUPLE_ template void PrintTo(const ::std::tuple& t, ::std::ostream* os) { PrintTupleTo(t, os); } #endif // GTEST_HAS_STD_TUPLE_ // Overload for std::pair. template void PrintTo(const ::std::pair& value, ::std::ostream* os) { *os << '('; // We cannot use UniversalPrint(value.first, os) here, as T1 may be // a reference type. The same for printing value.second. UniversalPrinter::Print(value.first, os); *os << ", "; UniversalPrinter::Print(value.second, os); *os << ')'; } // Implements printing a non-reference type T by letting the compiler // pick the right overload of PrintTo() for T. template class UniversalPrinter { public: // MSVC warns about adding const to a function type, so we want to // disable the warning. GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180) // Note: we deliberately don't call this PrintTo(), as that name // conflicts with ::testing::internal::PrintTo in the body of the // function. static void Print(const T& value, ::std::ostream* os) { // By default, ::testing::internal::PrintTo() is used for printing // the value. // // Thanks to Koenig look-up, if T is a class and has its own // PrintTo() function defined in its namespace, that function will // be visible here. Since it is more specific than the generic ones // in ::testing::internal, it will be picked by the compiler in the // following statement - exactly what we want. PrintTo(value, os); } GTEST_DISABLE_MSC_WARNINGS_POP_() }; // UniversalPrintArray(begin, len, os) prints an array of 'len' // elements, starting at address 'begin'. template void UniversalPrintArray(const T* begin, size_t len, ::std::ostream* os) { if (len == 0) { *os << "{}"; } else { *os << "{ "; const size_t kThreshold = 18; const size_t kChunkSize = 8; // If the array has more than kThreshold elements, we'll have to // omit some details by printing only the first and the last // kChunkSize elements. // TODO(wan@google.com): let the user control the threshold using a flag. if (len <= kThreshold) { PrintRawArrayTo(begin, len, os); } else { PrintRawArrayTo(begin, kChunkSize, os); *os << ", ..., "; PrintRawArrayTo(begin + len - kChunkSize, kChunkSize, os); } *os << " }"; } } // This overload prints a (const) char array compactly. GTEST_API_ void UniversalPrintArray( const char* begin, size_t len, ::std::ostream* os); // This overload prints a (const) wchar_t array compactly. GTEST_API_ void UniversalPrintArray( const wchar_t* begin, size_t len, ::std::ostream* os); // Implements printing an array type T[N]. template class UniversalPrinter { public: // Prints the given array, omitting some elements when there are too // many. static void Print(const T (&a)[N], ::std::ostream* os) { UniversalPrintArray(a, N, os); } }; // Implements printing a reference type T&. template class UniversalPrinter { public: // MSVC warns about adding const to a function type, so we want to // disable the warning. GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180) static void Print(const T& value, ::std::ostream* os) { // Prints the address of the value. We use reinterpret_cast here // as static_cast doesn't compile when T is a function type. *os << "@" << reinterpret_cast(&value) << " "; // Then prints the value itself. UniversalPrint(value, os); } GTEST_DISABLE_MSC_WARNINGS_POP_() }; // Prints a value tersely: for a reference type, the referenced value // (but not the address) is printed; for a (const) char pointer, the // NUL-terminated string (but not the pointer) is printed. template class UniversalTersePrinter { public: static void Print(const T& value, ::std::ostream* os) { UniversalPrint(value, os); } }; template class UniversalTersePrinter { public: static void Print(const T& value, ::std::ostream* os) { UniversalPrint(value, os); } }; template class UniversalTersePrinter { public: static void Print(const T (&value)[N], ::std::ostream* os) { UniversalPrinter::Print(value, os); } }; template <> class UniversalTersePrinter { public: static void Print(const char* str, ::std::ostream* os) { if (str == NULL) { *os << "NULL"; } else { UniversalPrint(string(str), os); } } }; template <> class UniversalTersePrinter { public: static void Print(char* str, ::std::ostream* os) { UniversalTersePrinter::Print(str, os); } }; #if GTEST_HAS_STD_WSTRING template <> class UniversalTersePrinter { public: static void Print(const wchar_t* str, ::std::ostream* os) { if (str == NULL) { *os << "NULL"; } else { UniversalPrint(::std::wstring(str), os); } } }; #endif template <> class UniversalTersePrinter { public: static void Print(wchar_t* str, ::std::ostream* os) { UniversalTersePrinter::Print(str, os); } }; template void UniversalTersePrint(const T& value, ::std::ostream* os) { UniversalTersePrinter::Print(value, os); } // Prints a value using the type inferred by the compiler. The // difference between this and UniversalTersePrint() is that for a // (const) char pointer, this prints both the pointer and the // NUL-terminated string. template void UniversalPrint(const T& value, ::std::ostream* os) { // A workarond for the bug in VC++ 7.1 that prevents us from instantiating // UniversalPrinter with T directly. typedef T T1; UniversalPrinter::Print(value, os); } typedef ::std::vector Strings; // TuplePolicy must provide: // - tuple_size // size of tuple TupleT. // - get(const TupleT& t) // static function extracting element I of tuple TupleT. // - tuple_element::type // type of element I of tuple TupleT. template struct TuplePolicy; #if GTEST_HAS_TR1_TUPLE template struct TuplePolicy { typedef TupleT Tuple; static const size_t tuple_size = ::std::tr1::tuple_size::value; template struct tuple_element : ::std::tr1::tuple_element {}; template static typename AddReference< const typename ::std::tr1::tuple_element::type>::type get( const Tuple& tuple) { return ::std::tr1::get(tuple); } }; template const size_t TuplePolicy::tuple_size; #endif // GTEST_HAS_TR1_TUPLE #if GTEST_HAS_STD_TUPLE_ template struct TuplePolicy< ::std::tuple > { typedef ::std::tuple Tuple; static const size_t tuple_size = ::std::tuple_size::value; template struct tuple_element : ::std::tuple_element {}; template static const typename ::std::tuple_element::type& get( const Tuple& tuple) { return ::std::get(tuple); } }; template const size_t TuplePolicy< ::std::tuple >::tuple_size; #endif // GTEST_HAS_STD_TUPLE_ #if GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_ // This helper template allows PrintTo() for tuples and // UniversalTersePrintTupleFieldsToStrings() to be defined by // induction on the number of tuple fields. The idea is that // TuplePrefixPrinter::PrintPrefixTo(t, os) prints the first N // fields in tuple t, and can be defined in terms of // TuplePrefixPrinter. // // The inductive case. template struct TuplePrefixPrinter { // Prints the first N fields of a tuple. template static void PrintPrefixTo(const Tuple& t, ::std::ostream* os) { TuplePrefixPrinter::PrintPrefixTo(t, os); GTEST_INTENTIONAL_CONST_COND_PUSH_() if (N > 1) { GTEST_INTENTIONAL_CONST_COND_POP_() *os << ", "; } UniversalPrinter< typename TuplePolicy::template tuple_element::type> ::Print(TuplePolicy::template get(t), os); } // Tersely prints the first N fields of a tuple to a string vector, // one element for each field. template static void TersePrintPrefixToStrings(const Tuple& t, Strings* strings) { TuplePrefixPrinter::TersePrintPrefixToStrings(t, strings); ::std::stringstream ss; UniversalTersePrint(TuplePolicy::template get(t), &ss); strings->push_back(ss.str()); } }; // Base case. template <> struct TuplePrefixPrinter<0> { template static void PrintPrefixTo(const Tuple&, ::std::ostream*) {} template static void TersePrintPrefixToStrings(const Tuple&, Strings*) {} }; // Helper function for printing a tuple. // Tuple must be either std::tr1::tuple or std::tuple type. template void PrintTupleTo(const Tuple& t, ::std::ostream* os) { *os << "("; TuplePrefixPrinter::tuple_size>::PrintPrefixTo(t, os); *os << ")"; } // Prints the fields of a tuple tersely to a string vector, one // element for each field. See the comment before // UniversalTersePrint() for how we define "tersely". template Strings UniversalTersePrintTupleFieldsToStrings(const Tuple& value) { Strings result; TuplePrefixPrinter::tuple_size>:: TersePrintPrefixToStrings(value, &result); return result; } #endif // GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_ } // namespace internal template ::std::string PrintToString(const T& value) { ::std::stringstream ss; internal::UniversalTersePrinter::Print(value, &ss); return ss.str(); } } // namespace testing #endif // GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_ #if GTEST_HAS_PARAM_TEST namespace testing { namespace internal { // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Outputs a message explaining invalid registration of different // fixture class for the same test case. This may happen when // TEST_P macro is used to define two tests with the same name // but in different namespaces. GTEST_API_ void ReportInvalidTestCaseType(const char* test_case_name, const char* file, int line); template class ParamGeneratorInterface; template class ParamGenerator; // Interface for iterating over elements provided by an implementation // of ParamGeneratorInterface. template class ParamIteratorInterface { public: virtual ~ParamIteratorInterface() {} // A pointer to the base generator instance. // Used only for the purposes of iterator comparison // to make sure that two iterators belong to the same generator. virtual const ParamGeneratorInterface* BaseGenerator() const = 0; // Advances iterator to point to the next element // provided by the generator. The caller is responsible // for not calling Advance() on an iterator equal to // BaseGenerator()->End(). virtual void Advance() = 0; // Clones the iterator object. Used for implementing copy semantics // of ParamIterator. virtual ParamIteratorInterface* Clone() const = 0; // Dereferences the current iterator and provides (read-only) access // to the pointed value. It is the caller's responsibility not to call // Current() on an iterator equal to BaseGenerator()->End(). // Used for implementing ParamGenerator::operator*(). virtual const T* Current() const = 0; // Determines whether the given iterator and other point to the same // element in the sequence generated by the generator. // Used for implementing ParamGenerator::operator==(). virtual bool Equals(const ParamIteratorInterface& other) const = 0; }; // Class iterating over elements provided by an implementation of // ParamGeneratorInterface. It wraps ParamIteratorInterface // and implements the const forward iterator concept. template class ParamIterator { public: typedef T value_type; typedef const T& reference; typedef ptrdiff_t difference_type; // ParamIterator assumes ownership of the impl_ pointer. ParamIterator(const ParamIterator& other) : impl_(other.impl_->Clone()) {} ParamIterator& operator=(const ParamIterator& other) { if (this != &other) impl_.reset(other.impl_->Clone()); return *this; } const T& operator*() const { return *impl_->Current(); } const T* operator->() const { return impl_->Current(); } // Prefix version of operator++. ParamIterator& operator++() { impl_->Advance(); return *this; } // Postfix version of operator++. ParamIterator operator++(int /*unused*/) { ParamIteratorInterface* clone = impl_->Clone(); impl_->Advance(); return ParamIterator(clone); } bool operator==(const ParamIterator& other) const { return impl_.get() == other.impl_.get() || impl_->Equals(*other.impl_); } bool operator!=(const ParamIterator& other) const { return !(*this == other); } private: friend class ParamGenerator; explicit ParamIterator(ParamIteratorInterface* impl) : impl_(impl) {} scoped_ptr > impl_; }; // ParamGeneratorInterface is the binary interface to access generators // defined in other translation units. template class ParamGeneratorInterface { public: typedef T ParamType; virtual ~ParamGeneratorInterface() {} // Generator interface definition virtual ParamIteratorInterface* Begin() const = 0; virtual ParamIteratorInterface* End() const = 0; }; // Wraps ParamGeneratorInterface and provides general generator syntax // compatible with the STL Container concept. // This class implements copy initialization semantics and the contained // ParamGeneratorInterface instance is shared among all copies // of the original object. This is possible because that instance is immutable. template class ParamGenerator { public: typedef ParamIterator iterator; explicit ParamGenerator(ParamGeneratorInterface* impl) : impl_(impl) {} ParamGenerator(const ParamGenerator& other) : impl_(other.impl_) {} ParamGenerator& operator=(const ParamGenerator& other) { impl_ = other.impl_; return *this; } iterator begin() const { return iterator(impl_->Begin()); } iterator end() const { return iterator(impl_->End()); } private: linked_ptr > impl_; }; // Generates values from a range of two comparable values. Can be used to // generate sequences of user-defined types that implement operator+() and // operator<(). // This class is used in the Range() function. template class RangeGenerator : public ParamGeneratorInterface { public: RangeGenerator(T begin, T end, IncrementT step) : begin_(begin), end_(end), step_(step), end_index_(CalculateEndIndex(begin, end, step)) {} virtual ~RangeGenerator() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, begin_, 0, step_); } virtual ParamIteratorInterface* End() const { return new Iterator(this, end_, end_index_, step_); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, T value, int index, IncrementT step) : base_(base), value_(value), index_(index), step_(step) {} virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } virtual void Advance() { value_ = value_ + step_; index_++; } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const T* Current() const { return &value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const int other_index = CheckedDowncastToActualType(&other)->index_; return index_ == other_index; } private: Iterator(const Iterator& other) : ParamIteratorInterface(), base_(other.base_), value_(other.value_), index_(other.index_), step_(other.step_) {} // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; T value_; int index_; const IncrementT step_; }; // class RangeGenerator::Iterator static int CalculateEndIndex(const T& begin, const T& end, const IncrementT& step) { int end_index = 0; for (T i = begin; i < end; i = i + step) end_index++; return end_index; } // No implementation - assignment is unsupported. void operator=(const RangeGenerator& other); const T begin_; const T end_; const IncrementT step_; // The index for the end() iterator. All the elements in the generated // sequence are indexed (0-based) to aid iterator comparison. const int end_index_; }; // class RangeGenerator // Generates values from a pair of STL-style iterators. Used in the // ValuesIn() function. The elements are copied from the source range // since the source can be located on the stack, and the generator // is likely to persist beyond that stack frame. template class ValuesInIteratorRangeGenerator : public ParamGeneratorInterface { public: template ValuesInIteratorRangeGenerator(ForwardIterator begin, ForwardIterator end) : container_(begin, end) {} virtual ~ValuesInIteratorRangeGenerator() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, container_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, container_.end()); } private: typedef typename ::std::vector ContainerType; class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, typename ContainerType::const_iterator iterator) : base_(base), iterator_(iterator) {} virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } virtual void Advance() { ++iterator_; value_.reset(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } // We need to use cached value referenced by iterator_ because *iterator_ // can return a temporary object (and of type other then T), so just // having "return &*iterator_;" doesn't work. // value_ is updated here and not in Advance() because Advance() // can advance iterator_ beyond the end of the range, and we cannot // detect that fact. The client code, on the other hand, is // responsible for not calling Current() on an out-of-range iterator. virtual const T* Current() const { if (value_.get() == NULL) value_.reset(new T(*iterator_)); return value_.get(); } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; return iterator_ == CheckedDowncastToActualType(&other)->iterator_; } private: Iterator(const Iterator& other) // The explicit constructor call suppresses a false warning // emitted by gcc when supplied with the -Wextra option. : ParamIteratorInterface(), base_(other.base_), iterator_(other.iterator_) {} const ParamGeneratorInterface* const base_; typename ContainerType::const_iterator iterator_; // A cached value of *iterator_. We keep it here to allow access by // pointer in the wrapping iterator's operator->(). // value_ needs to be mutable to be accessed in Current(). // Use of scoped_ptr helps manage cached value's lifetime, // which is bound by the lifespan of the iterator itself. mutable scoped_ptr value_; }; // class ValuesInIteratorRangeGenerator::Iterator // No implementation - assignment is unsupported. void operator=(const ValuesInIteratorRangeGenerator& other); const ContainerType container_; }; // class ValuesInIteratorRangeGenerator // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Stores a parameter value and later creates tests parameterized with that // value. template class ParameterizedTestFactory : public TestFactoryBase { public: typedef typename TestClass::ParamType ParamType; explicit ParameterizedTestFactory(ParamType parameter) : parameter_(parameter) {} virtual Test* CreateTest() { TestClass::SetParam(¶meter_); return new TestClass(); } private: const ParamType parameter_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestFactory); }; // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // TestMetaFactoryBase is a base class for meta-factories that create // test factories for passing into MakeAndRegisterTestInfo function. template class TestMetaFactoryBase { public: virtual ~TestMetaFactoryBase() {} virtual TestFactoryBase* CreateTestFactory(ParamType parameter) = 0; }; // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // TestMetaFactory creates test factories for passing into // MakeAndRegisterTestInfo function. Since MakeAndRegisterTestInfo receives // ownership of test factory pointer, same factory object cannot be passed // into that method twice. But ParameterizedTestCaseInfo is going to call // it for each Test/Parameter value combination. Thus it needs meta factory // creator class. template class TestMetaFactory : public TestMetaFactoryBase { public: typedef typename TestCase::ParamType ParamType; TestMetaFactory() {} virtual TestFactoryBase* CreateTestFactory(ParamType parameter) { return new ParameterizedTestFactory(parameter); } private: GTEST_DISALLOW_COPY_AND_ASSIGN_(TestMetaFactory); }; // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // ParameterizedTestCaseInfoBase is a generic interface // to ParameterizedTestCaseInfo classes. ParameterizedTestCaseInfoBase // accumulates test information provided by TEST_P macro invocations // and generators provided by INSTANTIATE_TEST_CASE_P macro invocations // and uses that information to register all resulting test instances // in RegisterTests method. The ParameterizeTestCaseRegistry class holds // a collection of pointers to the ParameterizedTestCaseInfo objects // and calls RegisterTests() on each of them when asked. class ParameterizedTestCaseInfoBase { public: virtual ~ParameterizedTestCaseInfoBase() {} // Base part of test case name for display purposes. virtual const string& GetTestCaseName() const = 0; // Test case id to verify identity. virtual TypeId GetTestCaseTypeId() const = 0; // UnitTest class invokes this method to register tests in this // test case right before running them in RUN_ALL_TESTS macro. // This method should not be called more then once on any single // instance of a ParameterizedTestCaseInfoBase derived class. virtual void RegisterTests() = 0; protected: ParameterizedTestCaseInfoBase() {} private: GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfoBase); }; // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // ParameterizedTestCaseInfo accumulates tests obtained from TEST_P // macro invocations for a particular test case and generators // obtained from INSTANTIATE_TEST_CASE_P macro invocations for that // test case. It registers tests with all values generated by all // generators when asked. template class ParameterizedTestCaseInfo : public ParameterizedTestCaseInfoBase { public: // ParamType and GeneratorCreationFunc are private types but are required // for declarations of public methods AddTestPattern() and // AddTestCaseInstantiation(). typedef typename TestCase::ParamType ParamType; // A function that returns an instance of appropriate generator type. typedef ParamGenerator(GeneratorCreationFunc)(); explicit ParameterizedTestCaseInfo(const char* name) : test_case_name_(name) {} // Test case base name for display purposes. virtual const string& GetTestCaseName() const { return test_case_name_; } // Test case id to verify identity. virtual TypeId GetTestCaseTypeId() const { return GetTypeId(); } // TEST_P macro uses AddTestPattern() to record information // about a single test in a LocalTestInfo structure. // test_case_name is the base name of the test case (without invocation // prefix). test_base_name is the name of an individual test without // parameter index. For the test SequenceA/FooTest.DoBar/1 FooTest is // test case base name and DoBar is test base name. void AddTestPattern(const char* test_case_name, const char* test_base_name, TestMetaFactoryBase* meta_factory) { tests_.push_back(linked_ptr(new TestInfo(test_case_name, test_base_name, meta_factory))); } // INSTANTIATE_TEST_CASE_P macro uses AddGenerator() to record information // about a generator. int AddTestCaseInstantiation(const string& instantiation_name, GeneratorCreationFunc* func, const char* /* file */, int /* line */) { instantiations_.push_back(::std::make_pair(instantiation_name, func)); return 0; // Return value used only to run this method in namespace scope. } // UnitTest class invokes this method to register tests in this test case // test cases right before running tests in RUN_ALL_TESTS macro. // This method should not be called more then once on any single // instance of a ParameterizedTestCaseInfoBase derived class. // UnitTest has a guard to prevent from calling this method more then once. virtual void RegisterTests() { for (typename TestInfoContainer::iterator test_it = tests_.begin(); test_it != tests_.end(); ++test_it) { linked_ptr test_info = *test_it; for (typename InstantiationContainer::iterator gen_it = instantiations_.begin(); gen_it != instantiations_.end(); ++gen_it) { const string& instantiation_name = gen_it->first; ParamGenerator generator((*gen_it->second)()); string test_case_name; if ( !instantiation_name.empty() ) test_case_name = instantiation_name + "/"; test_case_name += test_info->test_case_base_name; int i = 0; for (typename ParamGenerator::iterator param_it = generator.begin(); param_it != generator.end(); ++param_it, ++i) { Message test_name_stream; test_name_stream << test_info->test_base_name << "/" << i; MakeAndRegisterTestInfo( test_case_name.c_str(), test_name_stream.GetString().c_str(), NULL, // No type parameter. PrintToString(*param_it).c_str(), GetTestCaseTypeId(), TestCase::SetUpTestCase, TestCase::TearDownTestCase, test_info->test_meta_factory->CreateTestFactory(*param_it)); } // for param_it } // for gen_it } // for test_it } // RegisterTests private: // LocalTestInfo structure keeps information about a single test registered // with TEST_P macro. struct TestInfo { TestInfo(const char* a_test_case_base_name, const char* a_test_base_name, TestMetaFactoryBase* a_test_meta_factory) : test_case_base_name(a_test_case_base_name), test_base_name(a_test_base_name), test_meta_factory(a_test_meta_factory) {} const string test_case_base_name; const string test_base_name; const scoped_ptr > test_meta_factory; }; typedef ::std::vector > TestInfoContainer; // Keeps pairs of // received from INSTANTIATE_TEST_CASE_P macros. typedef ::std::vector > InstantiationContainer; const string test_case_name_; TestInfoContainer tests_; InstantiationContainer instantiations_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfo); }; // class ParameterizedTestCaseInfo // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // ParameterizedTestCaseRegistry contains a map of ParameterizedTestCaseInfoBase // classes accessed by test case names. TEST_P and INSTANTIATE_TEST_CASE_P // macros use it to locate their corresponding ParameterizedTestCaseInfo // descriptors. class ParameterizedTestCaseRegistry { public: ParameterizedTestCaseRegistry() {} ~ParameterizedTestCaseRegistry() { for (TestCaseInfoContainer::iterator it = test_case_infos_.begin(); it != test_case_infos_.end(); ++it) { delete *it; } } // Looks up or creates and returns a structure containing information about // tests and instantiations of a particular test case. template ParameterizedTestCaseInfo* GetTestCasePatternHolder( const char* test_case_name, const char* file, int line) { ParameterizedTestCaseInfo* typed_test_info = NULL; for (TestCaseInfoContainer::iterator it = test_case_infos_.begin(); it != test_case_infos_.end(); ++it) { if ((*it)->GetTestCaseName() == test_case_name) { if ((*it)->GetTestCaseTypeId() != GetTypeId()) { // Complain about incorrect usage of Google Test facilities // and terminate the program since we cannot guaranty correct // test case setup and tear-down in this case. ReportInvalidTestCaseType(test_case_name, file, line); posix::Abort(); } else { // At this point we are sure that the object we found is of the same // type we are looking for, so we downcast it to that type // without further checks. typed_test_info = CheckedDowncastToActualType< ParameterizedTestCaseInfo >(*it); } break; } } if (typed_test_info == NULL) { typed_test_info = new ParameterizedTestCaseInfo(test_case_name); test_case_infos_.push_back(typed_test_info); } return typed_test_info; } void RegisterTests() { for (TestCaseInfoContainer::iterator it = test_case_infos_.begin(); it != test_case_infos_.end(); ++it) { (*it)->RegisterTests(); } } private: typedef ::std::vector TestCaseInfoContainer; TestCaseInfoContainer test_case_infos_; GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseRegistry); }; } // namespace internal } // namespace testing #endif // GTEST_HAS_PARAM_TEST #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_ // This file was GENERATED by command: // pump.py gtest-param-util-generated.h.pump // DO NOT EDIT BY HAND!!! // Copyright 2008 Google Inc. // All Rights Reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: vladl@google.com (Vlad Losev) // Type and function utilities for implementing parameterized tests. // This file is generated by a SCRIPT. DO NOT EDIT BY HAND! // // Currently Google Test supports at most 50 arguments in Values, // and at most 10 arguments in Combine. Please contact // googletestframework@googlegroups.com if you need more. // Please note that the number of arguments to Combine is limited // by the maximum arity of the implementation of tuple which is // currently set at 10. #ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_ #define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_ // scripts/fuse_gtest.py depends on gtest's own header being #included // *unconditionally*. Therefore these #includes cannot be moved // inside #if GTEST_HAS_PARAM_TEST. #if GTEST_HAS_PARAM_TEST namespace testing { // Forward declarations of ValuesIn(), which is implemented in // include/gtest/gtest-param-test.h. template internal::ParamGenerator< typename ::testing::internal::IteratorTraits::value_type> ValuesIn(ForwardIterator begin, ForwardIterator end); template internal::ParamGenerator ValuesIn(const T (&array)[N]); template internal::ParamGenerator ValuesIn( const Container& container); namespace internal { // Used in the Values() function to provide polymorphic capabilities. template class ValueArray1 { public: explicit ValueArray1(T1 v1) : v1_(v1) {} template operator ParamGenerator() const { return ValuesIn(&v1_, &v1_ + 1); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray1& other); const T1 v1_; }; template class ValueArray2 { public: ValueArray2(T1 v1, T2 v2) : v1_(v1), v2_(v2) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray2& other); const T1 v1_; const T2 v2_; }; template class ValueArray3 { public: ValueArray3(T1 v1, T2 v2, T3 v3) : v1_(v1), v2_(v2), v3_(v3) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray3& other); const T1 v1_; const T2 v2_; const T3 v3_; }; template class ValueArray4 { public: ValueArray4(T1 v1, T2 v2, T3 v3, T4 v4) : v1_(v1), v2_(v2), v3_(v3), v4_(v4) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray4& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; }; template class ValueArray5 { public: ValueArray5(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray5& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; }; template class ValueArray6 { public: ValueArray6(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray6& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; }; template class ValueArray7 { public: ValueArray7(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray7& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; }; template class ValueArray8 { public: ValueArray8(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray8& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; }; template class ValueArray9 { public: ValueArray9(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray9& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; }; template class ValueArray10 { public: ValueArray10(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray10& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; }; template class ValueArray11 { public: ValueArray11(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray11& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; }; template class ValueArray12 { public: ValueArray12(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray12& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; }; template class ValueArray13 { public: ValueArray13(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray13& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; }; template class ValueArray14 { public: ValueArray14(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray14& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; }; template class ValueArray15 { public: ValueArray15(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray15& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; }; template class ValueArray16 { public: ValueArray16(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray16& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; }; template class ValueArray17 { public: ValueArray17(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray17& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; }; template class ValueArray18 { public: ValueArray18(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray18& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; }; template class ValueArray19 { public: ValueArray19(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray19& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; }; template class ValueArray20 { public: ValueArray20(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray20& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; }; template class ValueArray21 { public: ValueArray21(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray21& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; }; template class ValueArray22 { public: ValueArray22(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray22& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; }; template class ValueArray23 { public: ValueArray23(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray23& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; }; template class ValueArray24 { public: ValueArray24(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray24& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; }; template class ValueArray25 { public: ValueArray25(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray25& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; }; template class ValueArray26 { public: ValueArray26(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray26& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; }; template class ValueArray27 { public: ValueArray27(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray27& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; }; template class ValueArray28 { public: ValueArray28(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray28& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; }; template class ValueArray29 { public: ValueArray29(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray29& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; }; template class ValueArray30 { public: ValueArray30(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray30& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; }; template class ValueArray31 { public: ValueArray31(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray31& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; }; template class ValueArray32 { public: ValueArray32(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray32& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; }; template class ValueArray33 { public: ValueArray33(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray33& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; }; template class ValueArray34 { public: ValueArray34(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray34& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; }; template class ValueArray35 { public: ValueArray35(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray35& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; }; template class ValueArray36 { public: ValueArray36(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray36& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; }; template class ValueArray37 { public: ValueArray37(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray37& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; }; template class ValueArray38 { public: ValueArray38(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray38& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; }; template class ValueArray39 { public: ValueArray39(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray39& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; }; template class ValueArray40 { public: ValueArray40(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray40& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; }; template class ValueArray41 { public: ValueArray41(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray41& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; }; template class ValueArray42 { public: ValueArray42(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray42& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; }; template class ValueArray43 { public: ValueArray43(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray43& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; }; template class ValueArray44 { public: ValueArray44(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray44& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; }; template class ValueArray45 { public: ValueArray45(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_), static_cast(v45_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray45& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; const T45 v45_; }; template class ValueArray46 { public: ValueArray46(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_), static_cast(v45_), static_cast(v46_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray46& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; const T45 v45_; const T46 v46_; }; template class ValueArray47 { public: ValueArray47(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46), v47_(v47) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_), static_cast(v45_), static_cast(v46_), static_cast(v47_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray47& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; const T45 v45_; const T46 v46_; const T47 v47_; }; template class ValueArray48 { public: ValueArray48(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46), v47_(v47), v48_(v48) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_), static_cast(v45_), static_cast(v46_), static_cast(v47_), static_cast(v48_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray48& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; const T45 v45_; const T46 v46_; const T47 v47_; const T48 v48_; }; template class ValueArray49 { public: ValueArray49(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_), static_cast(v45_), static_cast(v46_), static_cast(v47_), static_cast(v48_), static_cast(v49_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray49& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; const T45 v45_; const T46 v46_; const T47 v47_; const T48 v48_; const T49 v49_; }; template class ValueArray50 { public: ValueArray50(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49, T50 v50) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49), v50_(v50) {} template operator ParamGenerator() const { const T array[] = {static_cast(v1_), static_cast(v2_), static_cast(v3_), static_cast(v4_), static_cast(v5_), static_cast(v6_), static_cast(v7_), static_cast(v8_), static_cast(v9_), static_cast(v10_), static_cast(v11_), static_cast(v12_), static_cast(v13_), static_cast(v14_), static_cast(v15_), static_cast(v16_), static_cast(v17_), static_cast(v18_), static_cast(v19_), static_cast(v20_), static_cast(v21_), static_cast(v22_), static_cast(v23_), static_cast(v24_), static_cast(v25_), static_cast(v26_), static_cast(v27_), static_cast(v28_), static_cast(v29_), static_cast(v30_), static_cast(v31_), static_cast(v32_), static_cast(v33_), static_cast(v34_), static_cast(v35_), static_cast(v36_), static_cast(v37_), static_cast(v38_), static_cast(v39_), static_cast(v40_), static_cast(v41_), static_cast(v42_), static_cast(v43_), static_cast(v44_), static_cast(v45_), static_cast(v46_), static_cast(v47_), static_cast(v48_), static_cast(v49_), static_cast(v50_)}; return ValuesIn(array); } private: // No implementation - assignment is unsupported. void operator=(const ValueArray50& other); const T1 v1_; const T2 v2_; const T3 v3_; const T4 v4_; const T5 v5_; const T6 v6_; const T7 v7_; const T8 v8_; const T9 v9_; const T10 v10_; const T11 v11_; const T12 v12_; const T13 v13_; const T14 v14_; const T15 v15_; const T16 v16_; const T17 v17_; const T18 v18_; const T19 v19_; const T20 v20_; const T21 v21_; const T22 v22_; const T23 v23_; const T24 v24_; const T25 v25_; const T26 v26_; const T27 v27_; const T28 v28_; const T29 v29_; const T30 v30_; const T31 v31_; const T32 v32_; const T33 v33_; const T34 v34_; const T35 v35_; const T36 v36_; const T37 v37_; const T38 v38_; const T39 v39_; const T40 v40_; const T41 v41_; const T42 v42_; const T43 v43_; const T44 v44_; const T45 v45_; const T46 v46_; const T47 v47_; const T48 v48_; const T49 v49_; const T50 v50_; }; # if GTEST_HAS_COMBINE // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Generates values from the Cartesian product of values produced // by the argument generators. // template class CartesianProductGenerator2 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator2(const ParamGenerator& g1, const ParamGenerator& g2) : g1_(g1), g2_(g2) {} virtual ~CartesianProductGenerator2() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current2_; if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; ParamType current_value_; }; // class CartesianProductGenerator2::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator2& other); const ParamGenerator g1_; const ParamGenerator g2_; }; // class CartesianProductGenerator2 template class CartesianProductGenerator3 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator3(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3) : g1_(g1), g2_(g2), g3_(g3) {} virtual ~CartesianProductGenerator3() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current3_; if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; ParamType current_value_; }; // class CartesianProductGenerator3::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator3& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; }; // class CartesianProductGenerator3 template class CartesianProductGenerator4 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator4(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4) : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {} virtual ~CartesianProductGenerator4() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current4_; if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; ParamType current_value_; }; // class CartesianProductGenerator4::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator4& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; }; // class CartesianProductGenerator4 template class CartesianProductGenerator5 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator5(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4, const ParamGenerator& g5) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {} virtual ~CartesianProductGenerator5() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end(), g5_, g5_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4, const ParamGenerator& g5, const typename ParamGenerator::iterator& current5) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4), begin5_(g5.begin()), end5_(g5.end()), current5_(current5) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current5_; if (current5_ == end5_) { current5_ = begin5_; ++current4_; } if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_ && current5_ == typed_other->current5_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_), begin5_(other.begin5_), end5_(other.end5_), current5_(other.current5_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_, *current5_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_ || current5_ == end5_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; const typename ParamGenerator::iterator begin5_; const typename ParamGenerator::iterator end5_; typename ParamGenerator::iterator current5_; ParamType current_value_; }; // class CartesianProductGenerator5::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator5& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; const ParamGenerator g5_; }; // class CartesianProductGenerator5 template class CartesianProductGenerator6 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator6(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4, const ParamGenerator& g5, const ParamGenerator& g6) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {} virtual ~CartesianProductGenerator6() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4, const ParamGenerator& g5, const typename ParamGenerator::iterator& current5, const ParamGenerator& g6, const typename ParamGenerator::iterator& current6) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4), begin5_(g5.begin()), end5_(g5.end()), current5_(current5), begin6_(g6.begin()), end6_(g6.end()), current6_(current6) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current6_; if (current6_ == end6_) { current6_ = begin6_; ++current5_; } if (current5_ == end5_) { current5_ = begin5_; ++current4_; } if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_ && current5_ == typed_other->current5_ && current6_ == typed_other->current6_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_), begin5_(other.begin5_), end5_(other.end5_), current5_(other.current5_), begin6_(other.begin6_), end6_(other.end6_), current6_(other.current6_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_, *current5_, *current6_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_ || current5_ == end5_ || current6_ == end6_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; const typename ParamGenerator::iterator begin5_; const typename ParamGenerator::iterator end5_; typename ParamGenerator::iterator current5_; const typename ParamGenerator::iterator begin6_; const typename ParamGenerator::iterator end6_; typename ParamGenerator::iterator current6_; ParamType current_value_; }; // class CartesianProductGenerator6::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator6& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; const ParamGenerator g5_; const ParamGenerator g6_; }; // class CartesianProductGenerator6 template class CartesianProductGenerator7 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator7(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4, const ParamGenerator& g5, const ParamGenerator& g6, const ParamGenerator& g7) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {} virtual ~CartesianProductGenerator7() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_, g7_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4, const ParamGenerator& g5, const typename ParamGenerator::iterator& current5, const ParamGenerator& g6, const typename ParamGenerator::iterator& current6, const ParamGenerator& g7, const typename ParamGenerator::iterator& current7) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4), begin5_(g5.begin()), end5_(g5.end()), current5_(current5), begin6_(g6.begin()), end6_(g6.end()), current6_(current6), begin7_(g7.begin()), end7_(g7.end()), current7_(current7) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current7_; if (current7_ == end7_) { current7_ = begin7_; ++current6_; } if (current6_ == end6_) { current6_ = begin6_; ++current5_; } if (current5_ == end5_) { current5_ = begin5_; ++current4_; } if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_ && current5_ == typed_other->current5_ && current6_ == typed_other->current6_ && current7_ == typed_other->current7_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_), begin5_(other.begin5_), end5_(other.end5_), current5_(other.current5_), begin6_(other.begin6_), end6_(other.end6_), current6_(other.current6_), begin7_(other.begin7_), end7_(other.end7_), current7_(other.current7_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_, *current5_, *current6_, *current7_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_ || current5_ == end5_ || current6_ == end6_ || current7_ == end7_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; const typename ParamGenerator::iterator begin5_; const typename ParamGenerator::iterator end5_; typename ParamGenerator::iterator current5_; const typename ParamGenerator::iterator begin6_; const typename ParamGenerator::iterator end6_; typename ParamGenerator::iterator current6_; const typename ParamGenerator::iterator begin7_; const typename ParamGenerator::iterator end7_; typename ParamGenerator::iterator current7_; ParamType current_value_; }; // class CartesianProductGenerator7::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator7& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; const ParamGenerator g5_; const ParamGenerator g6_; const ParamGenerator g7_; }; // class CartesianProductGenerator7 template class CartesianProductGenerator8 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator8(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4, const ParamGenerator& g5, const ParamGenerator& g6, const ParamGenerator& g7, const ParamGenerator& g8) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8) {} virtual ~CartesianProductGenerator8() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_, g7_.begin(), g8_, g8_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_, g8_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4, const ParamGenerator& g5, const typename ParamGenerator::iterator& current5, const ParamGenerator& g6, const typename ParamGenerator::iterator& current6, const ParamGenerator& g7, const typename ParamGenerator::iterator& current7, const ParamGenerator& g8, const typename ParamGenerator::iterator& current8) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4), begin5_(g5.begin()), end5_(g5.end()), current5_(current5), begin6_(g6.begin()), end6_(g6.end()), current6_(current6), begin7_(g7.begin()), end7_(g7.end()), current7_(current7), begin8_(g8.begin()), end8_(g8.end()), current8_(current8) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current8_; if (current8_ == end8_) { current8_ = begin8_; ++current7_; } if (current7_ == end7_) { current7_ = begin7_; ++current6_; } if (current6_ == end6_) { current6_ = begin6_; ++current5_; } if (current5_ == end5_) { current5_ = begin5_; ++current4_; } if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_ && current5_ == typed_other->current5_ && current6_ == typed_other->current6_ && current7_ == typed_other->current7_ && current8_ == typed_other->current8_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_), begin5_(other.begin5_), end5_(other.end5_), current5_(other.current5_), begin6_(other.begin6_), end6_(other.end6_), current6_(other.current6_), begin7_(other.begin7_), end7_(other.end7_), current7_(other.current7_), begin8_(other.begin8_), end8_(other.end8_), current8_(other.current8_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_, *current5_, *current6_, *current7_, *current8_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_ || current5_ == end5_ || current6_ == end6_ || current7_ == end7_ || current8_ == end8_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; const typename ParamGenerator::iterator begin5_; const typename ParamGenerator::iterator end5_; typename ParamGenerator::iterator current5_; const typename ParamGenerator::iterator begin6_; const typename ParamGenerator::iterator end6_; typename ParamGenerator::iterator current6_; const typename ParamGenerator::iterator begin7_; const typename ParamGenerator::iterator end7_; typename ParamGenerator::iterator current7_; const typename ParamGenerator::iterator begin8_; const typename ParamGenerator::iterator end8_; typename ParamGenerator::iterator current8_; ParamType current_value_; }; // class CartesianProductGenerator8::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator8& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; const ParamGenerator g5_; const ParamGenerator g6_; const ParamGenerator g7_; const ParamGenerator g8_; }; // class CartesianProductGenerator8 template class CartesianProductGenerator9 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator9(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4, const ParamGenerator& g5, const ParamGenerator& g6, const ParamGenerator& g7, const ParamGenerator& g8, const ParamGenerator& g9) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8), g9_(g9) {} virtual ~CartesianProductGenerator9() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_, g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_, g8_.end(), g9_, g9_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4, const ParamGenerator& g5, const typename ParamGenerator::iterator& current5, const ParamGenerator& g6, const typename ParamGenerator::iterator& current6, const ParamGenerator& g7, const typename ParamGenerator::iterator& current7, const ParamGenerator& g8, const typename ParamGenerator::iterator& current8, const ParamGenerator& g9, const typename ParamGenerator::iterator& current9) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4), begin5_(g5.begin()), end5_(g5.end()), current5_(current5), begin6_(g6.begin()), end6_(g6.end()), current6_(current6), begin7_(g7.begin()), end7_(g7.end()), current7_(current7), begin8_(g8.begin()), end8_(g8.end()), current8_(current8), begin9_(g9.begin()), end9_(g9.end()), current9_(current9) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current9_; if (current9_ == end9_) { current9_ = begin9_; ++current8_; } if (current8_ == end8_) { current8_ = begin8_; ++current7_; } if (current7_ == end7_) { current7_ = begin7_; ++current6_; } if (current6_ == end6_) { current6_ = begin6_; ++current5_; } if (current5_ == end5_) { current5_ = begin5_; ++current4_; } if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_ && current5_ == typed_other->current5_ && current6_ == typed_other->current6_ && current7_ == typed_other->current7_ && current8_ == typed_other->current8_ && current9_ == typed_other->current9_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_), begin5_(other.begin5_), end5_(other.end5_), current5_(other.current5_), begin6_(other.begin6_), end6_(other.end6_), current6_(other.current6_), begin7_(other.begin7_), end7_(other.end7_), current7_(other.current7_), begin8_(other.begin8_), end8_(other.end8_), current8_(other.current8_), begin9_(other.begin9_), end9_(other.end9_), current9_(other.current9_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_, *current5_, *current6_, *current7_, *current8_, *current9_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_ || current5_ == end5_ || current6_ == end6_ || current7_ == end7_ || current8_ == end8_ || current9_ == end9_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; const typename ParamGenerator::iterator begin5_; const typename ParamGenerator::iterator end5_; typename ParamGenerator::iterator current5_; const typename ParamGenerator::iterator begin6_; const typename ParamGenerator::iterator end6_; typename ParamGenerator::iterator current6_; const typename ParamGenerator::iterator begin7_; const typename ParamGenerator::iterator end7_; typename ParamGenerator::iterator current7_; const typename ParamGenerator::iterator begin8_; const typename ParamGenerator::iterator end8_; typename ParamGenerator::iterator current8_; const typename ParamGenerator::iterator begin9_; const typename ParamGenerator::iterator end9_; typename ParamGenerator::iterator current9_; ParamType current_value_; }; // class CartesianProductGenerator9::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator9& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; const ParamGenerator g5_; const ParamGenerator g6_; const ParamGenerator g7_; const ParamGenerator g8_; const ParamGenerator g9_; }; // class CartesianProductGenerator9 template class CartesianProductGenerator10 : public ParamGeneratorInterface< ::testing::tuple > { public: typedef ::testing::tuple ParamType; CartesianProductGenerator10(const ParamGenerator& g1, const ParamGenerator& g2, const ParamGenerator& g3, const ParamGenerator& g4, const ParamGenerator& g5, const ParamGenerator& g6, const ParamGenerator& g7, const ParamGenerator& g8, const ParamGenerator& g9, const ParamGenerator& g10) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8), g9_(g9), g10_(g10) {} virtual ~CartesianProductGenerator10() {} virtual ParamIteratorInterface* Begin() const { return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_, g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_, g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin(), g10_, g10_.begin()); } virtual ParamIteratorInterface* End() const { return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(), g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_, g8_.end(), g9_, g9_.end(), g10_, g10_.end()); } private: class Iterator : public ParamIteratorInterface { public: Iterator(const ParamGeneratorInterface* base, const ParamGenerator& g1, const typename ParamGenerator::iterator& current1, const ParamGenerator& g2, const typename ParamGenerator::iterator& current2, const ParamGenerator& g3, const typename ParamGenerator::iterator& current3, const ParamGenerator& g4, const typename ParamGenerator::iterator& current4, const ParamGenerator& g5, const typename ParamGenerator::iterator& current5, const ParamGenerator& g6, const typename ParamGenerator::iterator& current6, const ParamGenerator& g7, const typename ParamGenerator::iterator& current7, const ParamGenerator& g8, const typename ParamGenerator::iterator& current8, const ParamGenerator& g9, const typename ParamGenerator::iterator& current9, const ParamGenerator& g10, const typename ParamGenerator::iterator& current10) : base_(base), begin1_(g1.begin()), end1_(g1.end()), current1_(current1), begin2_(g2.begin()), end2_(g2.end()), current2_(current2), begin3_(g3.begin()), end3_(g3.end()), current3_(current3), begin4_(g4.begin()), end4_(g4.end()), current4_(current4), begin5_(g5.begin()), end5_(g5.end()), current5_(current5), begin6_(g6.begin()), end6_(g6.end()), current6_(current6), begin7_(g7.begin()), end7_(g7.end()), current7_(current7), begin8_(g8.begin()), end8_(g8.end()), current8_(current8), begin9_(g9.begin()), end9_(g9.end()), current9_(current9), begin10_(g10.begin()), end10_(g10.end()), current10_(current10) { ComputeCurrentValue(); } virtual ~Iterator() {} virtual const ParamGeneratorInterface* BaseGenerator() const { return base_; } // Advance should not be called on beyond-of-range iterators // so no component iterators must be beyond end of range, either. virtual void Advance() { assert(!AtEnd()); ++current10_; if (current10_ == end10_) { current10_ = begin10_; ++current9_; } if (current9_ == end9_) { current9_ = begin9_; ++current8_; } if (current8_ == end8_) { current8_ = begin8_; ++current7_; } if (current7_ == end7_) { current7_ = begin7_; ++current6_; } if (current6_ == end6_) { current6_ = begin6_; ++current5_; } if (current5_ == end5_) { current5_ = begin5_; ++current4_; } if (current4_ == end4_) { current4_ = begin4_; ++current3_; } if (current3_ == end3_) { current3_ = begin3_; ++current2_; } if (current2_ == end2_) { current2_ = begin2_; ++current1_; } ComputeCurrentValue(); } virtual ParamIteratorInterface* Clone() const { return new Iterator(*this); } virtual const ParamType* Current() const { return ¤t_value_; } virtual bool Equals(const ParamIteratorInterface& other) const { // Having the same base generator guarantees that the other // iterator is of the same type and we can downcast. GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) << "The program attempted to compare iterators " << "from different generators." << std::endl; const Iterator* typed_other = CheckedDowncastToActualType(&other); // We must report iterators equal if they both point beyond their // respective ranges. That can happen in a variety of fashions, // so we have to consult AtEnd(). return (AtEnd() && typed_other->AtEnd()) || ( current1_ == typed_other->current1_ && current2_ == typed_other->current2_ && current3_ == typed_other->current3_ && current4_ == typed_other->current4_ && current5_ == typed_other->current5_ && current6_ == typed_other->current6_ && current7_ == typed_other->current7_ && current8_ == typed_other->current8_ && current9_ == typed_other->current9_ && current10_ == typed_other->current10_); } private: Iterator(const Iterator& other) : base_(other.base_), begin1_(other.begin1_), end1_(other.end1_), current1_(other.current1_), begin2_(other.begin2_), end2_(other.end2_), current2_(other.current2_), begin3_(other.begin3_), end3_(other.end3_), current3_(other.current3_), begin4_(other.begin4_), end4_(other.end4_), current4_(other.current4_), begin5_(other.begin5_), end5_(other.end5_), current5_(other.current5_), begin6_(other.begin6_), end6_(other.end6_), current6_(other.current6_), begin7_(other.begin7_), end7_(other.end7_), current7_(other.current7_), begin8_(other.begin8_), end8_(other.end8_), current8_(other.current8_), begin9_(other.begin9_), end9_(other.end9_), current9_(other.current9_), begin10_(other.begin10_), end10_(other.end10_), current10_(other.current10_) { ComputeCurrentValue(); } void ComputeCurrentValue() { if (!AtEnd()) current_value_ = ParamType(*current1_, *current2_, *current3_, *current4_, *current5_, *current6_, *current7_, *current8_, *current9_, *current10_); } bool AtEnd() const { // We must report iterator past the end of the range when either of the // component iterators has reached the end of its range. return current1_ == end1_ || current2_ == end2_ || current3_ == end3_ || current4_ == end4_ || current5_ == end5_ || current6_ == end6_ || current7_ == end7_ || current8_ == end8_ || current9_ == end9_ || current10_ == end10_; } // No implementation - assignment is unsupported. void operator=(const Iterator& other); const ParamGeneratorInterface* const base_; // begin[i]_ and end[i]_ define the i-th range that Iterator traverses. // current[i]_ is the actual traversing iterator. const typename ParamGenerator::iterator begin1_; const typename ParamGenerator::iterator end1_; typename ParamGenerator::iterator current1_; const typename ParamGenerator::iterator begin2_; const typename ParamGenerator::iterator end2_; typename ParamGenerator::iterator current2_; const typename ParamGenerator::iterator begin3_; const typename ParamGenerator::iterator end3_; typename ParamGenerator::iterator current3_; const typename ParamGenerator::iterator begin4_; const typename ParamGenerator::iterator end4_; typename ParamGenerator::iterator current4_; const typename ParamGenerator::iterator begin5_; const typename ParamGenerator::iterator end5_; typename ParamGenerator::iterator current5_; const typename ParamGenerator::iterator begin6_; const typename ParamGenerator::iterator end6_; typename ParamGenerator::iterator current6_; const typename ParamGenerator::iterator begin7_; const typename ParamGenerator::iterator end7_; typename ParamGenerator::iterator current7_; const typename ParamGenerator::iterator begin8_; const typename ParamGenerator::iterator end8_; typename ParamGenerator::iterator current8_; const typename ParamGenerator::iterator begin9_; const typename ParamGenerator::iterator end9_; typename ParamGenerator::iterator current9_; const typename ParamGenerator::iterator begin10_; const typename ParamGenerator::iterator end10_; typename ParamGenerator::iterator current10_; ParamType current_value_; }; // class CartesianProductGenerator10::Iterator // No implementation - assignment is unsupported. void operator=(const CartesianProductGenerator10& other); const ParamGenerator g1_; const ParamGenerator g2_; const ParamGenerator g3_; const ParamGenerator g4_; const ParamGenerator g5_; const ParamGenerator g6_; const ParamGenerator g7_; const ParamGenerator g8_; const ParamGenerator g9_; const ParamGenerator g10_; }; // class CartesianProductGenerator10 // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Helper classes providing Combine() with polymorphic features. They allow // casting CartesianProductGeneratorN to ParamGenerator if T is // convertible to U. // template class CartesianProductHolder2 { public: CartesianProductHolder2(const Generator1& g1, const Generator2& g2) : g1_(g1), g2_(g2) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator2( static_cast >(g1_), static_cast >(g2_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder2& other); const Generator1 g1_; const Generator2 g2_; }; // class CartesianProductHolder2 template class CartesianProductHolder3 { public: CartesianProductHolder3(const Generator1& g1, const Generator2& g2, const Generator3& g3) : g1_(g1), g2_(g2), g3_(g3) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator3( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder3& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; }; // class CartesianProductHolder3 template class CartesianProductHolder4 { public: CartesianProductHolder4(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4) : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator4( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder4& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; }; // class CartesianProductHolder4 template class CartesianProductHolder5 { public: CartesianProductHolder5(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator5( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_), static_cast >(g5_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder5& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; const Generator5 g5_; }; // class CartesianProductHolder5 template class CartesianProductHolder6 { public: CartesianProductHolder6(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator6( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_), static_cast >(g5_), static_cast >(g6_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder6& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; const Generator5 g5_; const Generator6 g6_; }; // class CartesianProductHolder6 template class CartesianProductHolder7 { public: CartesianProductHolder7(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator7( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_), static_cast >(g5_), static_cast >(g6_), static_cast >(g7_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder7& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; const Generator5 g5_; const Generator6 g6_; const Generator7 g7_; }; // class CartesianProductHolder7 template class CartesianProductHolder8 { public: CartesianProductHolder8(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7, const Generator8& g8) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator8( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_), static_cast >(g5_), static_cast >(g6_), static_cast >(g7_), static_cast >(g8_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder8& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; const Generator5 g5_; const Generator6 g6_; const Generator7 g7_; const Generator8 g8_; }; // class CartesianProductHolder8 template class CartesianProductHolder9 { public: CartesianProductHolder9(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7, const Generator8& g8, const Generator9& g9) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8), g9_(g9) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator9( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_), static_cast >(g5_), static_cast >(g6_), static_cast >(g7_), static_cast >(g8_), static_cast >(g9_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder9& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; const Generator5 g5_; const Generator6 g6_; const Generator7 g7_; const Generator8 g8_; const Generator9 g9_; }; // class CartesianProductHolder9 template class CartesianProductHolder10 { public: CartesianProductHolder10(const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7, const Generator8& g8, const Generator9& g9, const Generator10& g10) : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8), g9_(g9), g10_(g10) {} template operator ParamGenerator< ::testing::tuple >() const { return ParamGenerator< ::testing::tuple >( new CartesianProductGenerator10( static_cast >(g1_), static_cast >(g2_), static_cast >(g3_), static_cast >(g4_), static_cast >(g5_), static_cast >(g6_), static_cast >(g7_), static_cast >(g8_), static_cast >(g9_), static_cast >(g10_))); } private: // No implementation - assignment is unsupported. void operator=(const CartesianProductHolder10& other); const Generator1 g1_; const Generator2 g2_; const Generator3 g3_; const Generator4 g4_; const Generator5 g5_; const Generator6 g6_; const Generator7 g7_; const Generator8 g8_; const Generator9 g9_; const Generator10 g10_; }; // class CartesianProductHolder10 # endif // GTEST_HAS_COMBINE } // namespace internal } // namespace testing #endif // GTEST_HAS_PARAM_TEST #endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_ #if GTEST_HAS_PARAM_TEST namespace testing { // Functions producing parameter generators. // // Google Test uses these generators to produce parameters for value- // parameterized tests. When a parameterized test case is instantiated // with a particular generator, Google Test creates and runs tests // for each element in the sequence produced by the generator. // // In the following sample, tests from test case FooTest are instantiated // each three times with parameter values 3, 5, and 8: // // class FooTest : public TestWithParam { ... }; // // TEST_P(FooTest, TestThis) { // } // TEST_P(FooTest, TestThat) { // } // INSTANTIATE_TEST_CASE_P(TestSequence, FooTest, Values(3, 5, 8)); // // Range() returns generators providing sequences of values in a range. // // Synopsis: // Range(start, end) // - returns a generator producing a sequence of values {start, start+1, // start+2, ..., }. // Range(start, end, step) // - returns a generator producing a sequence of values {start, start+step, // start+step+step, ..., }. // Notes: // * The generated sequences never include end. For example, Range(1, 5) // returns a generator producing a sequence {1, 2, 3, 4}. Range(1, 9, 2) // returns a generator producing {1, 3, 5, 7}. // * start and end must have the same type. That type may be any integral or // floating-point type or a user defined type satisfying these conditions: // * It must be assignable (have operator=() defined). // * It must have operator+() (operator+(int-compatible type) for // two-operand version). // * It must have operator<() defined. // Elements in the resulting sequences will also have that type. // * Condition start < end must be satisfied in order for resulting sequences // to contain any elements. // template internal::ParamGenerator Range(T start, T end, IncrementT step) { return internal::ParamGenerator( new internal::RangeGenerator(start, end, step)); } template internal::ParamGenerator Range(T start, T end) { return Range(start, end, 1); } // ValuesIn() function allows generation of tests with parameters coming from // a container. // // Synopsis: // ValuesIn(const T (&array)[N]) // - returns a generator producing sequences with elements from // a C-style array. // ValuesIn(const Container& container) // - returns a generator producing sequences with elements from // an STL-style container. // ValuesIn(Iterator begin, Iterator end) // - returns a generator producing sequences with elements from // a range [begin, end) defined by a pair of STL-style iterators. These // iterators can also be plain C pointers. // // Please note that ValuesIn copies the values from the containers // passed in and keeps them to generate tests in RUN_ALL_TESTS(). // // Examples: // // This instantiates tests from test case StringTest // each with C-string values of "foo", "bar", and "baz": // // const char* strings[] = {"foo", "bar", "baz"}; // INSTANTIATE_TEST_CASE_P(StringSequence, SrtingTest, ValuesIn(strings)); // // This instantiates tests from test case StlStringTest // each with STL strings with values "a" and "b": // // ::std::vector< ::std::string> GetParameterStrings() { // ::std::vector< ::std::string> v; // v.push_back("a"); // v.push_back("b"); // return v; // } // // INSTANTIATE_TEST_CASE_P(CharSequence, // StlStringTest, // ValuesIn(GetParameterStrings())); // // // This will also instantiate tests from CharTest // each with parameter values 'a' and 'b': // // ::std::list GetParameterChars() { // ::std::list list; // list.push_back('a'); // list.push_back('b'); // return list; // } // ::std::list l = GetParameterChars(); // INSTANTIATE_TEST_CASE_P(CharSequence2, // CharTest, // ValuesIn(l.begin(), l.end())); // template internal::ParamGenerator< typename ::testing::internal::IteratorTraits::value_type> ValuesIn(ForwardIterator begin, ForwardIterator end) { typedef typename ::testing::internal::IteratorTraits ::value_type ParamType; return internal::ParamGenerator( new internal::ValuesInIteratorRangeGenerator(begin, end)); } template internal::ParamGenerator ValuesIn(const T (&array)[N]) { return ValuesIn(array, array + N); } template internal::ParamGenerator ValuesIn( const Container& container) { return ValuesIn(container.begin(), container.end()); } // Values() allows generating tests from explicitly specified list of // parameters. // // Synopsis: // Values(T v1, T v2, ..., T vN) // - returns a generator producing sequences with elements v1, v2, ..., vN. // // For example, this instantiates tests from test case BarTest each // with values "one", "two", and "three": // // INSTANTIATE_TEST_CASE_P(NumSequence, BarTest, Values("one", "two", "three")); // // This instantiates tests from test case BazTest each with values 1, 2, 3.5. // The exact type of values will depend on the type of parameter in BazTest. // // INSTANTIATE_TEST_CASE_P(FloatingNumbers, BazTest, Values(1, 2, 3.5)); // // Currently, Values() supports from 1 to 50 parameters. // template internal::ValueArray1 Values(T1 v1) { return internal::ValueArray1(v1); } template internal::ValueArray2 Values(T1 v1, T2 v2) { return internal::ValueArray2(v1, v2); } template internal::ValueArray3 Values(T1 v1, T2 v2, T3 v3) { return internal::ValueArray3(v1, v2, v3); } template internal::ValueArray4 Values(T1 v1, T2 v2, T3 v3, T4 v4) { return internal::ValueArray4(v1, v2, v3, v4); } template internal::ValueArray5 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5) { return internal::ValueArray5(v1, v2, v3, v4, v5); } template internal::ValueArray6 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6) { return internal::ValueArray6(v1, v2, v3, v4, v5, v6); } template internal::ValueArray7 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7) { return internal::ValueArray7(v1, v2, v3, v4, v5, v6, v7); } template internal::ValueArray8 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8) { return internal::ValueArray8(v1, v2, v3, v4, v5, v6, v7, v8); } template internal::ValueArray9 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9) { return internal::ValueArray9(v1, v2, v3, v4, v5, v6, v7, v8, v9); } template internal::ValueArray10 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10) { return internal::ValueArray10(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10); } template internal::ValueArray11 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11) { return internal::ValueArray11(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11); } template internal::ValueArray12 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12) { return internal::ValueArray12(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12); } template internal::ValueArray13 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13) { return internal::ValueArray13(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13); } template internal::ValueArray14 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14) { return internal::ValueArray14(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14); } template internal::ValueArray15 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15) { return internal::ValueArray15(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15); } template internal::ValueArray16 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16) { return internal::ValueArray16(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16); } template internal::ValueArray17 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17) { return internal::ValueArray17(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17); } template internal::ValueArray18 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18) { return internal::ValueArray18(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18); } template internal::ValueArray19 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19) { return internal::ValueArray19(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19); } template internal::ValueArray20 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20) { return internal::ValueArray20(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20); } template internal::ValueArray21 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21) { return internal::ValueArray21(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21); } template internal::ValueArray22 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22) { return internal::ValueArray22(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22); } template internal::ValueArray23 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23) { return internal::ValueArray23(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23); } template internal::ValueArray24 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) { return internal::ValueArray24(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24); } template internal::ValueArray25 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25) { return internal::ValueArray25(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25); } template internal::ValueArray26 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26) { return internal::ValueArray26(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26); } template internal::ValueArray27 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27) { return internal::ValueArray27(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27); } template internal::ValueArray28 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28) { return internal::ValueArray28(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28); } template internal::ValueArray29 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29) { return internal::ValueArray29(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29); } template internal::ValueArray30 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30) { return internal::ValueArray30(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30); } template internal::ValueArray31 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31) { return internal::ValueArray31(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31); } template internal::ValueArray32 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32) { return internal::ValueArray32(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32); } template internal::ValueArray33 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33) { return internal::ValueArray33(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33); } template internal::ValueArray34 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34) { return internal::ValueArray34(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34); } template internal::ValueArray35 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35) { return internal::ValueArray35(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35); } template internal::ValueArray36 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36) { return internal::ValueArray36(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36); } template internal::ValueArray37 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37) { return internal::ValueArray37(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37); } template internal::ValueArray38 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38) { return internal::ValueArray38(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38); } template internal::ValueArray39 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39) { return internal::ValueArray39(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39); } template internal::ValueArray40 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40) { return internal::ValueArray40(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40); } template internal::ValueArray41 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41) { return internal::ValueArray41(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41); } template internal::ValueArray42 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42) { return internal::ValueArray42(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42); } template internal::ValueArray43 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43) { return internal::ValueArray43(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43); } template internal::ValueArray44 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44) { return internal::ValueArray44(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44); } template internal::ValueArray45 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45) { return internal::ValueArray45(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45); } template internal::ValueArray46 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46) { return internal::ValueArray46(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46); } template internal::ValueArray47 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47) { return internal::ValueArray47(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47); } template internal::ValueArray48 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48) { return internal::ValueArray48(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48); } template internal::ValueArray49 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49) { return internal::ValueArray49(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48, v49); } template internal::ValueArray50 Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49, T50 v50) { return internal::ValueArray50(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48, v49, v50); } // Bool() allows generating tests with parameters in a set of (false, true). // // Synopsis: // Bool() // - returns a generator producing sequences with elements {false, true}. // // It is useful when testing code that depends on Boolean flags. Combinations // of multiple flags can be tested when several Bool()'s are combined using // Combine() function. // // In the following example all tests in the test case FlagDependentTest // will be instantiated twice with parameters false and true. // // class FlagDependentTest : public testing::TestWithParam { // virtual void SetUp() { // external_flag = GetParam(); // } // } // INSTANTIATE_TEST_CASE_P(BoolSequence, FlagDependentTest, Bool()); // inline internal::ParamGenerator Bool() { return Values(false, true); } # if GTEST_HAS_COMBINE // Combine() allows the user to combine two or more sequences to produce // values of a Cartesian product of those sequences' elements. // // Synopsis: // Combine(gen1, gen2, ..., genN) // - returns a generator producing sequences with elements coming from // the Cartesian product of elements from the sequences generated by // gen1, gen2, ..., genN. The sequence elements will have a type of // tuple where T1, T2, ..., TN are the types // of elements from sequences produces by gen1, gen2, ..., genN. // // Combine can have up to 10 arguments. This number is currently limited // by the maximum number of elements in the tuple implementation used by Google // Test. // // Example: // // This will instantiate tests in test case AnimalTest each one with // the parameter values tuple("cat", BLACK), tuple("cat", WHITE), // tuple("dog", BLACK), and tuple("dog", WHITE): // // enum Color { BLACK, GRAY, WHITE }; // class AnimalTest // : public testing::TestWithParam > {...}; // // TEST_P(AnimalTest, AnimalLooksNice) {...} // // INSTANTIATE_TEST_CASE_P(AnimalVariations, AnimalTest, // Combine(Values("cat", "dog"), // Values(BLACK, WHITE))); // // This will instantiate tests in FlagDependentTest with all variations of two // Boolean flags: // // class FlagDependentTest // : public testing::TestWithParam > { // virtual void SetUp() { // // Assigns external_flag_1 and external_flag_2 values from the tuple. // tie(external_flag_1, external_flag_2) = GetParam(); // } // }; // // TEST_P(FlagDependentTest, TestFeature1) { // // Test your code using external_flag_1 and external_flag_2 here. // } // INSTANTIATE_TEST_CASE_P(TwoBoolSequence, FlagDependentTest, // Combine(Bool(), Bool())); // template internal::CartesianProductHolder2 Combine( const Generator1& g1, const Generator2& g2) { return internal::CartesianProductHolder2( g1, g2); } template internal::CartesianProductHolder3 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3) { return internal::CartesianProductHolder3( g1, g2, g3); } template internal::CartesianProductHolder4 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4) { return internal::CartesianProductHolder4( g1, g2, g3, g4); } template internal::CartesianProductHolder5 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5) { return internal::CartesianProductHolder5( g1, g2, g3, g4, g5); } template internal::CartesianProductHolder6 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6) { return internal::CartesianProductHolder6( g1, g2, g3, g4, g5, g6); } template internal::CartesianProductHolder7 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7) { return internal::CartesianProductHolder7( g1, g2, g3, g4, g5, g6, g7); } template internal::CartesianProductHolder8 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7, const Generator8& g8) { return internal::CartesianProductHolder8( g1, g2, g3, g4, g5, g6, g7, g8); } template internal::CartesianProductHolder9 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7, const Generator8& g8, const Generator9& g9) { return internal::CartesianProductHolder9( g1, g2, g3, g4, g5, g6, g7, g8, g9); } template internal::CartesianProductHolder10 Combine( const Generator1& g1, const Generator2& g2, const Generator3& g3, const Generator4& g4, const Generator5& g5, const Generator6& g6, const Generator7& g7, const Generator8& g8, const Generator9& g9, const Generator10& g10) { return internal::CartesianProductHolder10( g1, g2, g3, g4, g5, g6, g7, g8, g9, g10); } # endif // GTEST_HAS_COMBINE # define TEST_P(test_case_name, test_name) \ class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \ : public test_case_name { \ public: \ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {} \ virtual void TestBody(); \ private: \ static int AddToRegistry() { \ ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \ GetTestCasePatternHolder(\ #test_case_name, __FILE__, __LINE__)->AddTestPattern(\ #test_case_name, \ #test_name, \ new ::testing::internal::TestMetaFactory< \ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>()); \ return 0; \ } \ static int gtest_registering_dummy_; \ GTEST_DISALLOW_COPY_AND_ASSIGN_(\ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)); \ }; \ int GTEST_TEST_CLASS_NAME_(test_case_name, \ test_name)::gtest_registering_dummy_ = \ GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::AddToRegistry(); \ void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody() # define INSTANTIATE_TEST_CASE_P(prefix, test_case_name, generator) \ ::testing::internal::ParamGenerator \ gtest_##prefix##test_case_name##_EvalGenerator_() { return generator; } \ int gtest_##prefix##test_case_name##_dummy_ = \ ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \ GetTestCasePatternHolder(\ #test_case_name, __FILE__, __LINE__)->AddTestCaseInstantiation(\ #prefix, \ >est_##prefix##test_case_name##_EvalGenerator_, \ __FILE__, __LINE__) } // namespace testing #endif // GTEST_HAS_PARAM_TEST #endif // GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_ // Copyright 2006, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) // // Google C++ Testing Framework definitions useful in production code. #ifndef GTEST_INCLUDE_GTEST_GTEST_PROD_H_ #define GTEST_INCLUDE_GTEST_GTEST_PROD_H_ // When you need to test the private or protected members of a class, // use the FRIEND_TEST macro to declare your tests as friends of the // class. For example: // // class MyClass { // private: // void MyMethod(); // FRIEND_TEST(MyClassTest, MyMethod); // }; // // class MyClassTest : public testing::Test { // // ... // }; // // TEST_F(MyClassTest, MyMethod) { // // Can call MyClass::MyMethod() here. // } #define FRIEND_TEST(test_case_name, test_name)\ friend class test_case_name##_##test_name##_Test #endif // GTEST_INCLUDE_GTEST_GTEST_PROD_H_ // Copyright 2008, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: mheule@google.com (Markus Heule) // #ifndef GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_ #define GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_ #include #include namespace testing { // A copyable object representing the result of a test part (i.e. an // assertion or an explicit FAIL(), ADD_FAILURE(), or SUCCESS()). // // Don't inherit from TestPartResult as its destructor is not virtual. class GTEST_API_ TestPartResult { public: // The possible outcomes of a test part (i.e. an assertion or an // explicit SUCCEED(), FAIL(), or ADD_FAILURE()). enum Type { kSuccess, // Succeeded. kNonFatalFailure, // Failed but the test can continue. kFatalFailure // Failed and the test should be terminated. }; // C'tor. TestPartResult does NOT have a default constructor. // Always use this constructor (with parameters) to create a // TestPartResult object. TestPartResult(Type a_type, const char* a_file_name, int a_line_number, const char* a_message) : type_(a_type), file_name_(a_file_name == NULL ? "" : a_file_name), line_number_(a_line_number), summary_(ExtractSummary(a_message)), message_(a_message) { } // Gets the outcome of the test part. Type type() const { return type_; } // Gets the name of the source file where the test part took place, or // NULL if it's unknown. const char* file_name() const { return file_name_.empty() ? NULL : file_name_.c_str(); } // Gets the line in the source file where the test part took place, // or -1 if it's unknown. int line_number() const { return line_number_; } // Gets the summary of the failure message. const char* summary() const { return summary_.c_str(); } // Gets the message associated with the test part. const char* message() const { return message_.c_str(); } // Returns true iff the test part passed. bool passed() const { return type_ == kSuccess; } // Returns true iff the test part failed. bool failed() const { return type_ != kSuccess; } // Returns true iff the test part non-fatally failed. bool nonfatally_failed() const { return type_ == kNonFatalFailure; } // Returns true iff the test part fatally failed. bool fatally_failed() const { return type_ == kFatalFailure; } private: Type type_; // Gets the summary of the failure message by omitting the stack // trace in it. static std::string ExtractSummary(const char* message); // The name of the source file where the test part took place, or // "" if the source file is unknown. std::string file_name_; // The line in the source file where the test part took place, or -1 // if the line number is unknown. int line_number_; std::string summary_; // The test failure summary. std::string message_; // The test failure message. }; // Prints a TestPartResult object. std::ostream& operator<<(std::ostream& os, const TestPartResult& result); // An array of TestPartResult objects. // // Don't inherit from TestPartResultArray as its destructor is not // virtual. class GTEST_API_ TestPartResultArray { public: TestPartResultArray() {} // Appends the given TestPartResult to the array. void Append(const TestPartResult& result); // Returns the TestPartResult at the given index (0-based). const TestPartResult& GetTestPartResult(int index) const; // Returns the number of TestPartResult objects in the array. int size() const; private: std::vector array_; GTEST_DISALLOW_COPY_AND_ASSIGN_(TestPartResultArray); }; // This interface knows how to report a test part result. class TestPartResultReporterInterface { public: virtual ~TestPartResultReporterInterface() {} virtual void ReportTestPartResult(const TestPartResult& result) = 0; }; namespace internal { // This helper class is used by {ASSERT|EXPECT}_NO_FATAL_FAILURE to check if a // statement generates new fatal failures. To do so it registers itself as the // current test part result reporter. Besides checking if fatal failures were // reported, it only delegates the reporting to the former result reporter. // The original result reporter is restored in the destructor. // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. class GTEST_API_ HasNewFatalFailureHelper : public TestPartResultReporterInterface { public: HasNewFatalFailureHelper(); virtual ~HasNewFatalFailureHelper(); virtual void ReportTestPartResult(const TestPartResult& result); bool has_new_fatal_failure() const { return has_new_fatal_failure_; } private: bool has_new_fatal_failure_; TestPartResultReporterInterface* original_reporter_; GTEST_DISALLOW_COPY_AND_ASSIGN_(HasNewFatalFailureHelper); }; } // namespace internal } // namespace testing #endif // GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_ // Copyright 2008 Google Inc. // All Rights Reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // // Author: wan@google.com (Zhanyong Wan) #ifndef GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_ #define GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_ // This header implements typed tests and type-parameterized tests. // Typed (aka type-driven) tests repeat the same test for types in a // list. You must know which types you want to test with when writing // typed tests. Here's how you do it: #if 0 // First, define a fixture class template. It should be parameterized // by a type. Remember to derive it from testing::Test. template class FooTest : public testing::Test { public: ... typedef std::list List; static T shared_; T value_; }; // Next, associate a list of types with the test case, which will be // repeated for each type in the list. The typedef is necessary for // the macro to parse correctly. typedef testing::Types MyTypes; TYPED_TEST_CASE(FooTest, MyTypes); // If the type list contains only one type, you can write that type // directly without Types<...>: // TYPED_TEST_CASE(FooTest, int); // Then, use TYPED_TEST() instead of TEST_F() to define as many typed // tests for this test case as you want. TYPED_TEST(FooTest, DoesBlah) { // Inside a test, refer to TypeParam to get the type parameter. // Since we are inside a derived class template, C++ requires use to // visit the members of FooTest via 'this'. TypeParam n = this->value_; // To visit static members of the fixture, add the TestFixture:: // prefix. n += TestFixture::shared_; // To refer to typedefs in the fixture, add the "typename // TestFixture::" prefix. typename TestFixture::List values; values.push_back(n); ... } TYPED_TEST(FooTest, HasPropertyA) { ... } #endif // 0 // Type-parameterized tests are abstract test patterns parameterized // by a type. Compared with typed tests, type-parameterized tests // allow you to define the test pattern without knowing what the type // parameters are. The defined pattern can be instantiated with // different types any number of times, in any number of translation // units. // // If you are designing an interface or concept, you can define a // suite of type-parameterized tests to verify properties that any // valid implementation of the interface/concept should have. Then, // each implementation can easily instantiate the test suite to verify // that it conforms to the requirements, without having to write // similar tests repeatedly. Here's an example: #if 0 // First, define a fixture class template. It should be parameterized // by a type. Remember to derive it from testing::Test. template class FooTest : public testing::Test { ... }; // Next, declare that you will define a type-parameterized test case // (the _P suffix is for "parameterized" or "pattern", whichever you // prefer): TYPED_TEST_CASE_P(FooTest); // Then, use TYPED_TEST_P() to define as many type-parameterized tests // for this type-parameterized test case as you want. TYPED_TEST_P(FooTest, DoesBlah) { // Inside a test, refer to TypeParam to get the type parameter. TypeParam n = 0; ... } TYPED_TEST_P(FooTest, HasPropertyA) { ... } // Now the tricky part: you need to register all test patterns before // you can instantiate them. The first argument of the macro is the // test case name; the rest are the names of the tests in this test // case. REGISTER_TYPED_TEST_CASE_P(FooTest, DoesBlah, HasPropertyA); // Finally, you are free to instantiate the pattern with the types you // want. If you put the above code in a header file, you can #include // it in multiple C++ source files and instantiate it multiple times. // // To distinguish different instances of the pattern, the first // argument to the INSTANTIATE_* macro is a prefix that will be added // to the actual test case name. Remember to pick unique prefixes for // different instances. typedef testing::Types MyTypes; INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, MyTypes); // If the type list contains only one type, you can write that type // directly without Types<...>: // INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, int); #endif // 0 // Implements typed tests. #if GTEST_HAS_TYPED_TEST // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Expands to the name of the typedef for the type parameters of the // given test case. # define GTEST_TYPE_PARAMS_(TestCaseName) gtest_type_params_##TestCaseName##_ // The 'Types' template argument below must have spaces around it // since some compilers may choke on '>>' when passing a template // instance (e.g. Types) # define TYPED_TEST_CASE(CaseName, Types) \ typedef ::testing::internal::TypeList< Types >::type \ GTEST_TYPE_PARAMS_(CaseName) # define TYPED_TEST(CaseName, TestName) \ template \ class GTEST_TEST_CLASS_NAME_(CaseName, TestName) \ : public CaseName { \ private: \ typedef CaseName TestFixture; \ typedef gtest_TypeParam_ TypeParam; \ virtual void TestBody(); \ }; \ bool gtest_##CaseName##_##TestName##_registered_ GTEST_ATTRIBUTE_UNUSED_ = \ ::testing::internal::TypeParameterizedTest< \ CaseName, \ ::testing::internal::TemplateSel< \ GTEST_TEST_CLASS_NAME_(CaseName, TestName)>, \ GTEST_TYPE_PARAMS_(CaseName)>::Register(\ "", #CaseName, #TestName, 0); \ template \ void GTEST_TEST_CLASS_NAME_(CaseName, TestName)::TestBody() #endif // GTEST_HAS_TYPED_TEST // Implements type-parameterized tests. #if GTEST_HAS_TYPED_TEST_P // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Expands to the namespace name that the type-parameterized tests for // the given type-parameterized test case are defined in. The exact // name of the namespace is subject to change without notice. # define GTEST_CASE_NAMESPACE_(TestCaseName) \ gtest_case_##TestCaseName##_ // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // // Expands to the name of the variable used to remember the names of // the defined tests in the given test case. # define GTEST_TYPED_TEST_CASE_P_STATE_(TestCaseName) \ gtest_typed_test_case_p_state_##TestCaseName##_ // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE DIRECTLY. // // Expands to the name of the variable used to remember the names of // the registered tests in the given test case. # define GTEST_REGISTERED_TEST_NAMES_(TestCaseName) \ gtest_registered_test_names_##TestCaseName##_ // The variables defined in the type-parameterized test macros are // static as typically these macros are used in a .h file that can be // #included in multiple translation units linked together. # define TYPED_TEST_CASE_P(CaseName) \ static ::testing::internal::TypedTestCasePState \ GTEST_TYPED_TEST_CASE_P_STATE_(CaseName) # define TYPED_TEST_P(CaseName, TestName) \ namespace GTEST_CASE_NAMESPACE_(CaseName) { \ template \ class TestName : public CaseName { \ private: \ typedef CaseName TestFixture; \ typedef gtest_TypeParam_ TypeParam; \ virtual void TestBody(); \ }; \ static bool gtest_##TestName##_defined_ GTEST_ATTRIBUTE_UNUSED_ = \ GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).AddTestName(\ __FILE__, __LINE__, #CaseName, #TestName); \ } \ template \ void GTEST_CASE_NAMESPACE_(CaseName)::TestName::TestBody() # define REGISTER_TYPED_TEST_CASE_P(CaseName, ...) \ namespace GTEST_CASE_NAMESPACE_(CaseName) { \ typedef ::testing::internal::Templates<__VA_ARGS__>::type gtest_AllTests_; \ } \ static const char* const GTEST_REGISTERED_TEST_NAMES_(CaseName) = \ GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).VerifyRegisteredTestNames(\ __FILE__, __LINE__, #__VA_ARGS__) // The 'Types' template argument below must have spaces around it // since some compilers may choke on '>>' when passing a template // instance (e.g. Types) # define INSTANTIATE_TYPED_TEST_CASE_P(Prefix, CaseName, Types) \ bool gtest_##Prefix##_##CaseName GTEST_ATTRIBUTE_UNUSED_ = \ ::testing::internal::TypeParameterizedTestCase::type>::Register(\ #Prefix, #CaseName, GTEST_REGISTERED_TEST_NAMES_(CaseName)) #endif // GTEST_HAS_TYPED_TEST_P #endif // GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_ // Depending on the platform, different string classes are available. // On Linux, in addition to ::std::string, Google also makes use of // class ::string, which has the same interface as ::std::string, but // has a different implementation. // // You can define GTEST_HAS_GLOBAL_STRING to 1 to indicate that // ::string is available AND is a distinct type to ::std::string, or // define it to 0 to indicate otherwise. // // If ::std::string and ::string are the same class on your platform // due to aliasing, you should define GTEST_HAS_GLOBAL_STRING to 0. // // If you do not define GTEST_HAS_GLOBAL_STRING, it is defined // heuristically. namespace testing { // Declares the flags. // This flag temporary enables the disabled tests. GTEST_DECLARE_bool_(also_run_disabled_tests); // This flag brings the debugger on an assertion failure. GTEST_DECLARE_bool_(break_on_failure); // This flag controls whether Google Test catches all test-thrown exceptions // and logs them as failures. GTEST_DECLARE_bool_(catch_exceptions); // This flag enables using colors in terminal output. Available values are // "yes" to enable colors, "no" (disable colors), or "auto" (the default) // to let Google Test decide. GTEST_DECLARE_string_(color); // This flag sets up the filter to select by name using a glob pattern // the tests to run. If the filter is not given all tests are executed. GTEST_DECLARE_string_(filter); // This flag causes the Google Test to list tests. None of the tests listed // are actually run if the flag is provided. GTEST_DECLARE_bool_(list_tests); // This flag controls whether Google Test emits a detailed XML report to a file // in addition to its normal textual output. GTEST_DECLARE_string_(output); // This flags control whether Google Test prints the elapsed time for each // test. GTEST_DECLARE_bool_(print_time); // This flag specifies the random number seed. GTEST_DECLARE_int32_(random_seed); // This flag sets how many times the tests are repeated. The default value // is 1. If the value is -1 the tests are repeating forever. GTEST_DECLARE_int32_(repeat); // This flag controls whether Google Test includes Google Test internal // stack frames in failure stack traces. GTEST_DECLARE_bool_(show_internal_stack_frames); // When this flag is specified, tests' order is randomized on every iteration. GTEST_DECLARE_bool_(shuffle); // This flag specifies the maximum number of stack frames to be // printed in a failure message. GTEST_DECLARE_int32_(stack_trace_depth); // When this flag is specified, a failed assertion will throw an // exception if exceptions are enabled, or exit the program with a // non-zero code otherwise. GTEST_DECLARE_bool_(throw_on_failure); // When this flag is set with a "host:port" string, on supported // platforms test results are streamed to the specified port on // the specified host machine. GTEST_DECLARE_string_(stream_result_to); // The upper limit for valid stack trace depths. const int kMaxStackTraceDepth = 100; namespace internal { class AssertHelper; class DefaultGlobalTestPartResultReporter; class ExecDeathTest; class NoExecDeathTest; class FinalSuccessChecker; class GTestFlagSaver; class StreamingListenerTest; class TestResultAccessor; class TestEventListenersAccessor; class TestEventRepeater; class UnitTestRecordPropertyTestHelper; class WindowsDeathTest; class UnitTestImpl* GetUnitTestImpl(); void ReportFailureInUnknownLocation(TestPartResult::Type result_type, const std::string& message); } // namespace internal // The friend relationship of some of these classes is cyclic. // If we don't forward declare them the compiler might confuse the classes // in friendship clauses with same named classes on the scope. class Test; class TestCase; class TestInfo; class UnitTest; // A class for indicating whether an assertion was successful. When // the assertion wasn't successful, the AssertionResult object // remembers a non-empty message that describes how it failed. // // To create an instance of this class, use one of the factory functions // (AssertionSuccess() and AssertionFailure()). // // This class is useful for two purposes: // 1. Defining predicate functions to be used with Boolean test assertions // EXPECT_TRUE/EXPECT_FALSE and their ASSERT_ counterparts // 2. Defining predicate-format functions to be // used with predicate assertions (ASSERT_PRED_FORMAT*, etc). // // For example, if you define IsEven predicate: // // testing::AssertionResult IsEven(int n) { // if ((n % 2) == 0) // return testing::AssertionSuccess(); // else // return testing::AssertionFailure() << n << " is odd"; // } // // Then the failed expectation EXPECT_TRUE(IsEven(Fib(5))) // will print the message // // Value of: IsEven(Fib(5)) // Actual: false (5 is odd) // Expected: true // // instead of a more opaque // // Value of: IsEven(Fib(5)) // Actual: false // Expected: true // // in case IsEven is a simple Boolean predicate. // // If you expect your predicate to be reused and want to support informative // messages in EXPECT_FALSE and ASSERT_FALSE (negative assertions show up // about half as often as positive ones in our tests), supply messages for // both success and failure cases: // // testing::AssertionResult IsEven(int n) { // if ((n % 2) == 0) // return testing::AssertionSuccess() << n << " is even"; // else // return testing::AssertionFailure() << n << " is odd"; // } // // Then a statement EXPECT_FALSE(IsEven(Fib(6))) will print // // Value of: IsEven(Fib(6)) // Actual: true (8 is even) // Expected: false // // NB: Predicates that support negative Boolean assertions have reduced // performance in positive ones so be careful not to use them in tests // that have lots (tens of thousands) of positive Boolean assertions. // // To use this class with EXPECT_PRED_FORMAT assertions such as: // // // Verifies that Foo() returns an even number. // EXPECT_PRED_FORMAT1(IsEven, Foo()); // // you need to define: // // testing::AssertionResult IsEven(const char* expr, int n) { // if ((n % 2) == 0) // return testing::AssertionSuccess(); // else // return testing::AssertionFailure() // << "Expected: " << expr << " is even\n Actual: it's " << n; // } // // If Foo() returns 5, you will see the following message: // // Expected: Foo() is even // Actual: it's 5 // class GTEST_API_ AssertionResult { public: // Copy constructor. // Used in EXPECT_TRUE/FALSE(assertion_result). AssertionResult(const AssertionResult& other); GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 /* forcing value to bool */) // Used in the EXPECT_TRUE/FALSE(bool_expression). // // T must be contextually convertible to bool. // // The second parameter prevents this overload from being considered if // the argument is implicitly convertible to AssertionResult. In that case // we want AssertionResult's copy constructor to be used. template explicit AssertionResult( const T& success, typename internal::EnableIf< !internal::ImplicitlyConvertible::value>::type* /*enabler*/ = NULL) : success_(success) {} GTEST_DISABLE_MSC_WARNINGS_POP_() // Assignment operator. AssertionResult& operator=(AssertionResult other) { swap(other); return *this; } // Returns true iff the assertion succeeded. operator bool() const { return success_; } // NOLINT // Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE. AssertionResult operator!() const; // Returns the text streamed into this AssertionResult. Test assertions // use it when they fail (i.e., the predicate's outcome doesn't match the // assertion's expectation). When nothing has been streamed into the // object, returns an empty string. const char* message() const { return message_.get() != NULL ? message_->c_str() : ""; } // TODO(vladl@google.com): Remove this after making sure no clients use it. // Deprecated; please use message() instead. const char* failure_message() const { return message(); } // Streams a custom failure message into this object. template AssertionResult& operator<<(const T& value) { AppendMessage(Message() << value); return *this; } // Allows streaming basic output manipulators such as endl or flush into // this object. AssertionResult& operator<<( ::std::ostream& (*basic_manipulator)(::std::ostream& stream)) { AppendMessage(Message() << basic_manipulator); return *this; } private: // Appends the contents of message to message_. void AppendMessage(const Message& a_message) { if (message_.get() == NULL) message_.reset(new ::std::string); message_->append(a_message.GetString().c_str()); } // Swap the contents of this AssertionResult with other. void swap(AssertionResult& other); // Stores result of the assertion predicate. bool success_; // Stores the message describing the condition in case the expectation // construct is not satisfied with the predicate's outcome. // Referenced via a pointer to avoid taking too much stack frame space // with test assertions. internal::scoped_ptr< ::std::string> message_; }; // Makes a successful assertion result. GTEST_API_ AssertionResult AssertionSuccess(); // Makes a failed assertion result. GTEST_API_ AssertionResult AssertionFailure(); // Makes a failed assertion result with the given failure message. // Deprecated; use AssertionFailure() << msg. GTEST_API_ AssertionResult AssertionFailure(const Message& msg); // The abstract class that all tests inherit from. // // In Google Test, a unit test program contains one or many TestCases, and // each TestCase contains one or many Tests. // // When you define a test using the TEST macro, you don't need to // explicitly derive from Test - the TEST macro automatically does // this for you. // // The only time you derive from Test is when defining a test fixture // to be used a TEST_F. For example: // // class FooTest : public testing::Test { // protected: // void SetUp() override { ... } // void TearDown() override { ... } // ... // }; // // TEST_F(FooTest, Bar) { ... } // TEST_F(FooTest, Baz) { ... } // // Test is not copyable. class GTEST_API_ Test { public: friend class TestInfo; // Defines types for pointers to functions that set up and tear down // a test case. typedef internal::SetUpTestCaseFunc SetUpTestCaseFunc; typedef internal::TearDownTestCaseFunc TearDownTestCaseFunc; // The d'tor is virtual as we intend to inherit from Test. virtual ~Test(); // Sets up the stuff shared by all tests in this test case. // // Google Test will call Foo::SetUpTestCase() before running the first // test in test case Foo. Hence a sub-class can define its own // SetUpTestCase() method to shadow the one defined in the super // class. static void SetUpTestCase() {} // Tears down the stuff shared by all tests in this test case. // // Google Test will call Foo::TearDownTestCase() after running the last // test in test case Foo. Hence a sub-class can define its own // TearDownTestCase() method to shadow the one defined in the super // class. static void TearDownTestCase() {} // Returns true iff the current test has a fatal failure. static bool HasFatalFailure(); // Returns true iff the current test has a non-fatal failure. static bool HasNonfatalFailure(); // Returns true iff the current test has a (either fatal or // non-fatal) failure. static bool HasFailure() { return HasFatalFailure() || HasNonfatalFailure(); } // Logs a property for the current test, test case, or for the entire // invocation of the test program when used outside of the context of a // test case. Only the last value for a given key is remembered. These // are public static so they can be called from utility functions that are // not members of the test fixture. Calls to RecordProperty made during // lifespan of the test (from the moment its constructor starts to the // moment its destructor finishes) will be output in XML as attributes of // the element. Properties recorded from fixture's // SetUpTestCase or TearDownTestCase are logged as attributes of the // corresponding element. Calls to RecordProperty made in the // global context (before or after invocation of RUN_ALL_TESTS and from // SetUp/TearDown method of Environment objects registered with Google // Test) will be output as attributes of the element. static void RecordProperty(const std::string& key, const std::string& value); static void RecordProperty(const std::string& key, int value); protected: // Creates a Test object. Test(); // Sets up the test fixture. virtual void SetUp(); // Tears down the test fixture. virtual void TearDown(); private: // Returns true iff the current test has the same fixture class as // the first test in the current test case. static bool HasSameFixtureClass(); // Runs the test after the test fixture has been set up. // // A sub-class must implement this to define the test logic. // // DO NOT OVERRIDE THIS FUNCTION DIRECTLY IN A USER PROGRAM. // Instead, use the TEST or TEST_F macro. virtual void TestBody() = 0; // Sets up, executes, and tears down the test. void Run(); // Deletes self. We deliberately pick an unusual name for this // internal method to avoid clashing with names used in user TESTs. void DeleteSelf_() { delete this; } // Uses a GTestFlagSaver to save and restore all Google Test flags. const internal::GTestFlagSaver* const gtest_flag_saver_; // Often a user misspells SetUp() as Setup() and spends a long time // wondering why it is never called by Google Test. The declaration of // the following method is solely for catching such an error at // compile time: // // - The return type is deliberately chosen to be not void, so it // will be a conflict if void Setup() is declared in the user's // test fixture. // // - This method is private, so it will be another compiler error // if the method is called from the user's test fixture. // // DO NOT OVERRIDE THIS FUNCTION. // // If you see an error about overriding the following function or // about it being private, you have mis-spelled SetUp() as Setup(). struct Setup_should_be_spelled_SetUp {}; virtual Setup_should_be_spelled_SetUp* Setup() { return NULL; } // We disallow copying Tests. GTEST_DISALLOW_COPY_AND_ASSIGN_(Test); }; typedef internal::TimeInMillis TimeInMillis; // A copyable object representing a user specified test property which can be // output as a key/value string pair. // // Don't inherit from TestProperty as its destructor is not virtual. class TestProperty { public: // C'tor. TestProperty does NOT have a default constructor. // Always use this constructor (with parameters) to create a // TestProperty object. TestProperty(const std::string& a_key, const std::string& a_value) : key_(a_key), value_(a_value) { } // Gets the user supplied key. const char* key() const { return key_.c_str(); } // Gets the user supplied value. const char* value() const { return value_.c_str(); } // Sets a new value, overriding the one supplied in the constructor. void SetValue(const std::string& new_value) { value_ = new_value; } private: // The key supplied by the user. std::string key_; // The value supplied by the user. std::string value_; }; // The result of a single Test. This includes a list of // TestPartResults, a list of TestProperties, a count of how many // death tests there are in the Test, and how much time it took to run // the Test. // // TestResult is not copyable. class GTEST_API_ TestResult { public: // Creates an empty TestResult. TestResult(); // D'tor. Do not inherit from TestResult. ~TestResult(); // Gets the number of all test parts. This is the sum of the number // of successful test parts and the number of failed test parts. int total_part_count() const; // Returns the number of the test properties. int test_property_count() const; // Returns true iff the test passed (i.e. no test part failed). bool Passed() const { return !Failed(); } // Returns true iff the test failed. bool Failed() const; // Returns true iff the test fatally failed. bool HasFatalFailure() const; // Returns true iff the test has a non-fatal failure. bool HasNonfatalFailure() const; // Returns the elapsed time, in milliseconds. TimeInMillis elapsed_time() const { return elapsed_time_; } // Returns the i-th test part result among all the results. i can range // from 0 to test_property_count() - 1. If i is not in that range, aborts // the program. const TestPartResult& GetTestPartResult(int i) const; // Returns the i-th test property. i can range from 0 to // test_property_count() - 1. If i is not in that range, aborts the // program. const TestProperty& GetTestProperty(int i) const; private: friend class TestInfo; friend class TestCase; friend class UnitTest; friend class internal::DefaultGlobalTestPartResultReporter; friend class internal::ExecDeathTest; friend class internal::TestResultAccessor; friend class internal::UnitTestImpl; friend class internal::WindowsDeathTest; // Gets the vector of TestPartResults. const std::vector& test_part_results() const { return test_part_results_; } // Gets the vector of TestProperties. const std::vector& test_properties() const { return test_properties_; } // Sets the elapsed time. void set_elapsed_time(TimeInMillis elapsed) { elapsed_time_ = elapsed; } // Adds a test property to the list. The property is validated and may add // a non-fatal failure if invalid (e.g., if it conflicts with reserved // key names). If a property is already recorded for the same key, the // value will be updated, rather than storing multiple values for the same // key. xml_element specifies the element for which the property is being // recorded and is used for validation. void RecordProperty(const std::string& xml_element, const TestProperty& test_property); // Adds a failure if the key is a reserved attribute of Google Test // testcase tags. Returns true if the property is valid. // TODO(russr): Validate attribute names are legal and human readable. static bool ValidateTestProperty(const std::string& xml_element, const TestProperty& test_property); // Adds a test part result to the list. void AddTestPartResult(const TestPartResult& test_part_result); // Returns the death test count. int death_test_count() const { return death_test_count_; } // Increments the death test count, returning the new count. int increment_death_test_count() { return ++death_test_count_; } // Clears the test part results. void ClearTestPartResults(); // Clears the object. void Clear(); // Protects mutable state of the property vector and of owned // properties, whose values may be updated. internal::Mutex test_properites_mutex_; // The vector of TestPartResults std::vector test_part_results_; // The vector of TestProperties std::vector test_properties_; // Running count of death tests. int death_test_count_; // The elapsed time, in milliseconds. TimeInMillis elapsed_time_; // We disallow copying TestResult. GTEST_DISALLOW_COPY_AND_ASSIGN_(TestResult); }; // class TestResult // A TestInfo object stores the following information about a test: // // Test case name // Test name // Whether the test should be run // A function pointer that creates the test object when invoked // Test result // // The constructor of TestInfo registers itself with the UnitTest // singleton such that the RUN_ALL_TESTS() macro knows which tests to // run. class GTEST_API_ TestInfo { public: // Destructs a TestInfo object. This function is not virtual, so // don't inherit from TestInfo. ~TestInfo(); // Returns the test case name. const char* test_case_name() const { return test_case_name_.c_str(); } // Returns the test name. const char* name() const { return name_.c_str(); } // Returns the name of the parameter type, or NULL if this is not a typed // or a type-parameterized test. const char* type_param() const { if (type_param_.get() != NULL) return type_param_->c_str(); return NULL; } // Returns the text representation of the value parameter, or NULL if this // is not a value-parameterized test. const char* value_param() const { if (value_param_.get() != NULL) return value_param_->c_str(); return NULL; } // Returns true if this test should run, that is if the test is not // disabled (or it is disabled but the also_run_disabled_tests flag has // been specified) and its full name matches the user-specified filter. // // Google Test allows the user to filter the tests by their full names. // The full name of a test Bar in test case Foo is defined as // "Foo.Bar". Only the tests that match the filter will run. // // A filter is a colon-separated list of glob (not regex) patterns, // optionally followed by a '-' and a colon-separated list of // negative patterns (tests to exclude). A test is run if it // matches one of the positive patterns and does not match any of // the negative patterns. // // For example, *A*:Foo.* is a filter that matches any string that // contains the character 'A' or starts with "Foo.". bool should_run() const { return should_run_; } // Returns true iff this test will appear in the XML report. bool is_reportable() const { // For now, the XML report includes all tests matching the filter. // In the future, we may trim tests that are excluded because of // sharding. return matches_filter_; } // Returns the result of the test. const TestResult* result() const { return &result_; } private: #if GTEST_HAS_DEATH_TEST friend class internal::DefaultDeathTestFactory; #endif // GTEST_HAS_DEATH_TEST friend class Test; friend class TestCase; friend class internal::UnitTestImpl; friend class internal::StreamingListenerTest; friend TestInfo* internal::MakeAndRegisterTestInfo( const char* test_case_name, const char* name, const char* type_param, const char* value_param, internal::TypeId fixture_class_id, Test::SetUpTestCaseFunc set_up_tc, Test::TearDownTestCaseFunc tear_down_tc, internal::TestFactoryBase* factory); // Constructs a TestInfo object. The newly constructed instance assumes // ownership of the factory object. TestInfo(const std::string& test_case_name, const std::string& name, const char* a_type_param, // NULL if not a type-parameterized test const char* a_value_param, // NULL if not a value-parameterized test internal::TypeId fixture_class_id, internal::TestFactoryBase* factory); // Increments the number of death tests encountered in this test so // far. int increment_death_test_count() { return result_.increment_death_test_count(); } // Creates the test object, runs it, records its result, and then // deletes it. void Run(); static void ClearTestResult(TestInfo* test_info) { test_info->result_.Clear(); } // These fields are immutable properties of the test. const std::string test_case_name_; // Test case name const std::string name_; // Test name // Name of the parameter type, or NULL if this is not a typed or a // type-parameterized test. const internal::scoped_ptr type_param_; // Text representation of the value parameter, or NULL if this is not a // value-parameterized test. const internal::scoped_ptr value_param_; const internal::TypeId fixture_class_id_; // ID of the test fixture class bool should_run_; // True iff this test should run bool is_disabled_; // True iff this test is disabled bool matches_filter_; // True if this test matches the // user-specified filter. internal::TestFactoryBase* const factory_; // The factory that creates // the test object // This field is mutable and needs to be reset before running the // test for the second time. TestResult result_; GTEST_DISALLOW_COPY_AND_ASSIGN_(TestInfo); }; // A test case, which consists of a vector of TestInfos. // // TestCase is not copyable. class GTEST_API_ TestCase { public: // Creates a TestCase with the given name. // // TestCase does NOT have a default constructor. Always use this // constructor to create a TestCase object. // // Arguments: // // name: name of the test case // a_type_param: the name of the test's type parameter, or NULL if // this is not a type-parameterized test. // set_up_tc: pointer to the function that sets up the test case // tear_down_tc: pointer to the function that tears down the test case TestCase(const char* name, const char* a_type_param, Test::SetUpTestCaseFunc set_up_tc, Test::TearDownTestCaseFunc tear_down_tc); // Destructor of TestCase. virtual ~TestCase(); // Gets the name of the TestCase. const char* name() const { return name_.c_str(); } // Returns the name of the parameter type, or NULL if this is not a // type-parameterized test case. const char* type_param() const { if (type_param_.get() != NULL) return type_param_->c_str(); return NULL; } // Returns true if any test in this test case should run. bool should_run() const { return should_run_; } // Gets the number of successful tests in this test case. int successful_test_count() const; // Gets the number of failed tests in this test case. int failed_test_count() const; // Gets the number of disabled tests that will be reported in the XML report. int reportable_disabled_test_count() const; // Gets the number of disabled tests in this test case. int disabled_test_count() const; // Gets the number of tests to be printed in the XML report. int reportable_test_count() const; // Get the number of tests in this test case that should run. int test_to_run_count() const; // Gets the number of all tests in this test case. int total_test_count() const; // Returns true iff the test case passed. bool Passed() const { return !Failed(); } // Returns true iff the test case failed. bool Failed() const { return failed_test_count() > 0; } // Returns the elapsed time, in milliseconds. TimeInMillis elapsed_time() const { return elapsed_time_; } // Returns the i-th test among all the tests. i can range from 0 to // total_test_count() - 1. If i is not in that range, returns NULL. const TestInfo* GetTestInfo(int i) const; // Returns the TestResult that holds test properties recorded during // execution of SetUpTestCase and TearDownTestCase. const TestResult& ad_hoc_test_result() const { return ad_hoc_test_result_; } private: friend class Test; friend class internal::UnitTestImpl; // Gets the (mutable) vector of TestInfos in this TestCase. std::vector& test_info_list() { return test_info_list_; } // Gets the (immutable) vector of TestInfos in this TestCase. const std::vector& test_info_list() const { return test_info_list_; } // Returns the i-th test among all the tests. i can range from 0 to // total_test_count() - 1. If i is not in that range, returns NULL. TestInfo* GetMutableTestInfo(int i); // Sets the should_run member. void set_should_run(bool should) { should_run_ = should; } // Adds a TestInfo to this test case. Will delete the TestInfo upon // destruction of the TestCase object. void AddTestInfo(TestInfo * test_info); // Clears the results of all tests in this test case. void ClearResult(); // Clears the results of all tests in the given test case. static void ClearTestCaseResult(TestCase* test_case) { test_case->ClearResult(); } // Runs every test in this TestCase. void Run(); // Runs SetUpTestCase() for this TestCase. This wrapper is needed // for catching exceptions thrown from SetUpTestCase(). void RunSetUpTestCase() { (*set_up_tc_)(); } // Runs TearDownTestCase() for this TestCase. This wrapper is // needed for catching exceptions thrown from TearDownTestCase(). void RunTearDownTestCase() { (*tear_down_tc_)(); } // Returns true iff test passed. static bool TestPassed(const TestInfo* test_info) { return test_info->should_run() && test_info->result()->Passed(); } // Returns true iff test failed. static bool TestFailed(const TestInfo* test_info) { return test_info->should_run() && test_info->result()->Failed(); } // Returns true iff the test is disabled and will be reported in the XML // report. static bool TestReportableDisabled(const TestInfo* test_info) { return test_info->is_reportable() && test_info->is_disabled_; } // Returns true iff test is disabled. static bool TestDisabled(const TestInfo* test_info) { return test_info->is_disabled_; } // Returns true iff this test will appear in the XML report. static bool TestReportable(const TestInfo* test_info) { return test_info->is_reportable(); } // Returns true if the given test should run. static bool ShouldRunTest(const TestInfo* test_info) { return test_info->should_run(); } // Shuffles the tests in this test case. void ShuffleTests(internal::Random* random); // Restores the test order to before the first shuffle. void UnshuffleTests(); // Name of the test case. std::string name_; // Name of the parameter type, or NULL if this is not a typed or a // type-parameterized test. const internal::scoped_ptr type_param_; // The vector of TestInfos in their original order. It owns the // elements in the vector. std::vector test_info_list_; // Provides a level of indirection for the test list to allow easy // shuffling and restoring the test order. The i-th element in this // vector is the index of the i-th test in the shuffled test list. std::vector test_indices_; // Pointer to the function that sets up the test case. Test::SetUpTestCaseFunc set_up_tc_; // Pointer to the function that tears down the test case. Test::TearDownTestCaseFunc tear_down_tc_; // True iff any test in this test case should run. bool should_run_; // Elapsed time, in milliseconds. TimeInMillis elapsed_time_; // Holds test properties recorded during execution of SetUpTestCase and // TearDownTestCase. TestResult ad_hoc_test_result_; // We disallow copying TestCases. GTEST_DISALLOW_COPY_AND_ASSIGN_(TestCase); }; // An Environment object is capable of setting up and tearing down an // environment. You should subclass this to define your own // environment(s). // // An Environment object does the set-up and tear-down in virtual // methods SetUp() and TearDown() instead of the constructor and the // destructor, as: // // 1. You cannot safely throw from a destructor. This is a problem // as in some cases Google Test is used where exceptions are enabled, and // we may want to implement ASSERT_* using exceptions where they are // available. // 2. You cannot use ASSERT_* directly in a constructor or // destructor. class Environment { public: // The d'tor is virtual as we need to subclass Environment. virtual ~Environment() {} // Override this to define how to set up the environment. virtual void SetUp() {} // Override this to define how to tear down the environment. virtual void TearDown() {} private: // If you see an error about overriding the following function or // about it being private, you have mis-spelled SetUp() as Setup(). struct Setup_should_be_spelled_SetUp {}; virtual Setup_should_be_spelled_SetUp* Setup() { return NULL; } }; // The interface for tracing execution of tests. The methods are organized in // the order the corresponding events are fired. class TestEventListener { public: virtual ~TestEventListener() {} // Fired before any test activity starts. virtual void OnTestProgramStart(const UnitTest& unit_test) = 0; // Fired before each iteration of tests starts. There may be more than // one iteration if GTEST_FLAG(repeat) is set. iteration is the iteration // index, starting from 0. virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration) = 0; // Fired before environment set-up for each iteration of tests starts. virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test) = 0; // Fired after environment set-up for each iteration of tests ends. virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test) = 0; // Fired before the test case starts. virtual void OnTestCaseStart(const TestCase& test_case) = 0; // Fired before the test starts. virtual void OnTestStart(const TestInfo& test_info) = 0; // Fired after a failed assertion or a SUCCEED() invocation. virtual void OnTestPartResult(const TestPartResult& test_part_result) = 0; // Fired after the test ends. virtual void OnTestEnd(const TestInfo& test_info) = 0; // Fired after the test case ends. virtual void OnTestCaseEnd(const TestCase& test_case) = 0; // Fired before environment tear-down for each iteration of tests starts. virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test) = 0; // Fired after environment tear-down for each iteration of tests ends. virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test) = 0; // Fired after each iteration of tests finishes. virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration) = 0; // Fired after all test activities have ended. virtual void OnTestProgramEnd(const UnitTest& unit_test) = 0; }; // The convenience class for users who need to override just one or two // methods and are not concerned that a possible change to a signature of // the methods they override will not be caught during the build. For // comments about each method please see the definition of TestEventListener // above. class EmptyTestEventListener : public TestEventListener { public: virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {} virtual void OnTestIterationStart(const UnitTest& /*unit_test*/, int /*iteration*/) {} virtual void OnEnvironmentsSetUpStart(const UnitTest& /*unit_test*/) {} virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {} virtual void OnTestCaseStart(const TestCase& /*test_case*/) {} virtual void OnTestStart(const TestInfo& /*test_info*/) {} virtual void OnTestPartResult(const TestPartResult& /*test_part_result*/) {} virtual void OnTestEnd(const TestInfo& /*test_info*/) {} virtual void OnTestCaseEnd(const TestCase& /*test_case*/) {} virtual void OnEnvironmentsTearDownStart(const UnitTest& /*unit_test*/) {} virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {} virtual void OnTestIterationEnd(const UnitTest& /*unit_test*/, int /*iteration*/) {} virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {} }; // TestEventListeners lets users add listeners to track events in Google Test. class GTEST_API_ TestEventListeners { public: TestEventListeners(); ~TestEventListeners(); // Appends an event listener to the end of the list. Google Test assumes // the ownership of the listener (i.e. it will delete the listener when // the test program finishes). void Append(TestEventListener* listener); // Removes the given event listener from the list and returns it. It then // becomes the caller's responsibility to delete the listener. Returns // NULL if the listener is not found in the list. TestEventListener* Release(TestEventListener* listener); // Returns the standard listener responsible for the default console // output. Can be removed from the listeners list to shut down default // console output. Note that removing this object from the listener list // with Release transfers its ownership to the caller and makes this // function return NULL the next time. TestEventListener* default_result_printer() const { return default_result_printer_; } // Returns the standard listener responsible for the default XML output // controlled by the --gtest_output=xml flag. Can be removed from the // listeners list by users who want to shut down the default XML output // controlled by this flag and substitute it with custom one. Note that // removing this object from the listener list with Release transfers its // ownership to the caller and makes this function return NULL the next // time. TestEventListener* default_xml_generator() const { return default_xml_generator_; } private: friend class TestCase; friend class TestInfo; friend class internal::DefaultGlobalTestPartResultReporter; friend class internal::NoExecDeathTest; friend class internal::TestEventListenersAccessor; friend class internal::UnitTestImpl; // Returns repeater that broadcasts the TestEventListener events to all // subscribers. TestEventListener* repeater(); // Sets the default_result_printer attribute to the provided listener. // The listener is also added to the listener list and previous // default_result_printer is removed from it and deleted. The listener can // also be NULL in which case it will not be added to the list. Does // nothing if the previous and the current listener objects are the same. void SetDefaultResultPrinter(TestEventListener* listener); // Sets the default_xml_generator attribute to the provided listener. The // listener is also added to the listener list and previous // default_xml_generator is removed from it and deleted. The listener can // also be NULL in which case it will not be added to the list. Does // nothing if the previous and the current listener objects are the same. void SetDefaultXmlGenerator(TestEventListener* listener); // Controls whether events will be forwarded by the repeater to the // listeners in the list. bool EventForwardingEnabled() const; void SuppressEventForwarding(); // The actual list of listeners. internal::TestEventRepeater* repeater_; // Listener responsible for the standard result output. TestEventListener* default_result_printer_; // Listener responsible for the creation of the XML output file. TestEventListener* default_xml_generator_; // We disallow copying TestEventListeners. GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventListeners); }; // A UnitTest consists of a vector of TestCases. // // This is a singleton class. The only instance of UnitTest is // created when UnitTest::GetInstance() is first called. This // instance is never deleted. // // UnitTest is not copyable. // // This class is thread-safe as long as the methods are called // according to their specification. class GTEST_API_ UnitTest { public: // Gets the singleton UnitTest object. The first time this method // is called, a UnitTest object is constructed and returned. // Consecutive calls will return the same object. static UnitTest* GetInstance(); // Runs all tests in this UnitTest object and prints the result. // Returns 0 if successful, or 1 otherwise. // // This method can only be called from the main thread. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. int Run() GTEST_MUST_USE_RESULT_; // Returns the working directory when the first TEST() or TEST_F() // was executed. The UnitTest object owns the string. const char* original_working_dir() const; // Returns the TestCase object for the test that's currently running, // or NULL if no test is running. const TestCase* current_test_case() const GTEST_LOCK_EXCLUDED_(mutex_); // Returns the TestInfo object for the test that's currently running, // or NULL if no test is running. const TestInfo* current_test_info() const GTEST_LOCK_EXCLUDED_(mutex_); // Returns the random seed used at the start of the current test run. int random_seed() const; #if GTEST_HAS_PARAM_TEST // Returns the ParameterizedTestCaseRegistry object used to keep track of // value-parameterized tests and instantiate and register them. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. internal::ParameterizedTestCaseRegistry& parameterized_test_registry() GTEST_LOCK_EXCLUDED_(mutex_); #endif // GTEST_HAS_PARAM_TEST // Gets the number of successful test cases. int successful_test_case_count() const; // Gets the number of failed test cases. int failed_test_case_count() const; // Gets the number of all test cases. int total_test_case_count() const; // Gets the number of all test cases that contain at least one test // that should run. int test_case_to_run_count() const; // Gets the number of successful tests. int successful_test_count() const; // Gets the number of failed tests. int failed_test_count() const; // Gets the number of disabled tests that will be reported in the XML report. int reportable_disabled_test_count() const; // Gets the number of disabled tests. int disabled_test_count() const; // Gets the number of tests to be printed in the XML report. int reportable_test_count() const; // Gets the number of all tests. int total_test_count() const; // Gets the number of tests that should run. int test_to_run_count() const; // Gets the time of the test program start, in ms from the start of the // UNIX epoch. TimeInMillis start_timestamp() const; // Gets the elapsed time, in milliseconds. TimeInMillis elapsed_time() const; // Returns true iff the unit test passed (i.e. all test cases passed). bool Passed() const; // Returns true iff the unit test failed (i.e. some test case failed // or something outside of all tests failed). bool Failed() const; // Gets the i-th test case among all the test cases. i can range from 0 to // total_test_case_count() - 1. If i is not in that range, returns NULL. const TestCase* GetTestCase(int i) const; // Returns the TestResult containing information on test failures and // properties logged outside of individual test cases. const TestResult& ad_hoc_test_result() const; // Returns the list of event listeners that can be used to track events // inside Google Test. TestEventListeners& listeners(); private: // Registers and returns a global test environment. When a test // program is run, all global test environments will be set-up in // the order they were registered. After all tests in the program // have finished, all global test environments will be torn-down in // the *reverse* order they were registered. // // The UnitTest object takes ownership of the given environment. // // This method can only be called from the main thread. Environment* AddEnvironment(Environment* env); // Adds a TestPartResult to the current TestResult object. All // Google Test assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) // eventually call this to report their results. The user code // should use the assertion macros instead of calling this directly. void AddTestPartResult(TestPartResult::Type result_type, const char* file_name, int line_number, const std::string& message, const std::string& os_stack_trace) GTEST_LOCK_EXCLUDED_(mutex_); // Adds a TestProperty to the current TestResult object when invoked from // inside a test, to current TestCase's ad_hoc_test_result_ when invoked // from SetUpTestCase or TearDownTestCase, or to the global property set // when invoked elsewhere. If the result already contains a property with // the same key, the value will be updated. void RecordProperty(const std::string& key, const std::string& value); // Gets the i-th test case among all the test cases. i can range from 0 to // total_test_case_count() - 1. If i is not in that range, returns NULL. TestCase* GetMutableTestCase(int i); // Accessors for the implementation object. internal::UnitTestImpl* impl() { return impl_; } const internal::UnitTestImpl* impl() const { return impl_; } // These classes and funcions are friends as they need to access private // members of UnitTest. friend class Test; friend class internal::AssertHelper; friend class internal::ScopedTrace; friend class internal::StreamingListenerTest; friend class internal::UnitTestRecordPropertyTestHelper; friend Environment* AddGlobalTestEnvironment(Environment* env); friend internal::UnitTestImpl* internal::GetUnitTestImpl(); friend void internal::ReportFailureInUnknownLocation( TestPartResult::Type result_type, const std::string& message); // Creates an empty UnitTest. UnitTest(); // D'tor virtual ~UnitTest(); // Pushes a trace defined by SCOPED_TRACE() on to the per-thread // Google Test trace stack. void PushGTestTrace(const internal::TraceInfo& trace) GTEST_LOCK_EXCLUDED_(mutex_); // Pops a trace from the per-thread Google Test trace stack. void PopGTestTrace() GTEST_LOCK_EXCLUDED_(mutex_); // Protects mutable state in *impl_. This is mutable as some const // methods need to lock it too. mutable internal::Mutex mutex_; // Opaque implementation object. This field is never changed once // the object is constructed. We don't mark it as const here, as // doing so will cause a warning in the constructor of UnitTest. // Mutable state in *impl_ is protected by mutex_. internal::UnitTestImpl* impl_; // We disallow copying UnitTest. GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTest); }; // A convenient wrapper for adding an environment for the test // program. // // You should call this before RUN_ALL_TESTS() is called, probably in // main(). If you use gtest_main, you need to call this before main() // starts for it to take effect. For example, you can define a global // variable like this: // // testing::Environment* const foo_env = // testing::AddGlobalTestEnvironment(new FooEnvironment); // // However, we strongly recommend you to write your own main() and // call AddGlobalTestEnvironment() there, as relying on initialization // of global variables makes the code harder to read and may cause // problems when you register multiple environments from different // translation units and the environments have dependencies among them // (remember that the compiler doesn't guarantee the order in which // global variables from different translation units are initialized). inline Environment* AddGlobalTestEnvironment(Environment* env) { return UnitTest::GetInstance()->AddEnvironment(env); } // Initializes Google Test. This must be called before calling // RUN_ALL_TESTS(). In particular, it parses a command line for the // flags that Google Test recognizes. Whenever a Google Test flag is // seen, it is removed from argv, and *argc is decremented. // // No value is returned. Instead, the Google Test flag variables are // updated. // // Calling the function for the second time has no user-visible effect. GTEST_API_ void InitGoogleTest(int* argc, char** argv); // This overloaded version can be used in Windows programs compiled in // UNICODE mode. GTEST_API_ void InitGoogleTest(int* argc, wchar_t** argv); namespace internal { // FormatForComparison::Format(value) formats a // value of type ToPrint that is an operand of a comparison assertion // (e.g. ASSERT_EQ). OtherOperand is the type of the other operand in // the comparison, and is used to help determine the best way to // format the value. In particular, when the value is a C string // (char pointer) and the other operand is an STL string object, we // want to format the C string as a string, since we know it is // compared by value with the string object. If the value is a char // pointer but the other operand is not an STL string object, we don't // know whether the pointer is supposed to point to a NUL-terminated // string, and thus want to print it as a pointer to be safe. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. // The default case. template class FormatForComparison { public: static ::std::string Format(const ToPrint& value) { return ::testing::PrintToString(value); } }; // Array. template class FormatForComparison { public: static ::std::string Format(const ToPrint* value) { return FormatForComparison::Format(value); } }; // By default, print C string as pointers to be safe, as we don't know // whether they actually point to a NUL-terminated string. #define GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(CharType) \ template \ class FormatForComparison { \ public: \ static ::std::string Format(CharType* value) { \ return ::testing::PrintToString(static_cast(value)); \ } \ } GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(char); GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const char); GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(wchar_t); GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const wchar_t); #undef GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_ // If a C string is compared with an STL string object, we know it's meant // to point to a NUL-terminated string, and thus can print it as a string. #define GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(CharType, OtherStringType) \ template <> \ class FormatForComparison { \ public: \ static ::std::string Format(CharType* value) { \ return ::testing::PrintToString(value); \ } \ } GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::std::string); GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::std::string); #if GTEST_HAS_GLOBAL_STRING GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::string); GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::string); #endif #if GTEST_HAS_GLOBAL_WSTRING GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::wstring); GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::wstring); #endif #if GTEST_HAS_STD_WSTRING GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::std::wstring); GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::std::wstring); #endif #undef GTEST_IMPL_FORMAT_C_STRING_AS_STRING_ // Formats a comparison assertion (e.g. ASSERT_EQ, EXPECT_LT, and etc) // operand to be used in a failure message. The type (but not value) // of the other operand may affect the format. This allows us to // print a char* as a raw pointer when it is compared against another // char* or void*, and print it as a C string when it is compared // against an std::string object, for example. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. template std::string FormatForComparisonFailureMessage( const T1& value, const T2& /* other_operand */) { return FormatForComparison::Format(value); } // Separate the error generating code from the code path to reduce the stack // frame size of CmpHelperEQ. This helps reduce the overhead of some sanitizers // when calling EXPECT_* in a tight loop. template AssertionResult CmpHelperEQFailure(const char* expected_expression, const char* actual_expression, const T1& expected, const T2& actual) { return EqFailure(expected_expression, actual_expression, FormatForComparisonFailureMessage(expected, actual), FormatForComparisonFailureMessage(actual, expected), false); } // The helper function for {ASSERT|EXPECT}_EQ. template AssertionResult CmpHelperEQ(const char* expected_expression, const char* actual_expression, const T1& expected, const T2& actual) { GTEST_DISABLE_MSC_WARNINGS_PUSH_(4389 /* signed/unsigned mismatch */) if (expected == actual) { return AssertionSuccess(); } GTEST_DISABLE_MSC_WARNINGS_POP_() return CmpHelperEQFailure(expected_expression, actual_expression, expected, actual); } // With this overloaded version, we allow anonymous enums to be used // in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous enums // can be implicitly cast to BiggestInt. GTEST_API_ AssertionResult CmpHelperEQ(const char* expected_expression, const char* actual_expression, BiggestInt expected, BiggestInt actual); // The helper class for {ASSERT|EXPECT}_EQ. The template argument // lhs_is_null_literal is true iff the first argument to ASSERT_EQ() // is a null pointer literal. The following default implementation is // for lhs_is_null_literal being false. template class EqHelper { public: // This templatized version is for the general case. template static AssertionResult Compare(const char* expected_expression, const char* actual_expression, const T1& expected, const T2& actual) { return CmpHelperEQ(expected_expression, actual_expression, expected, actual); } // With this overloaded version, we allow anonymous enums to be used // in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous // enums can be implicitly cast to BiggestInt. // // Even though its body looks the same as the above version, we // cannot merge the two, as it will make anonymous enums unhappy. static AssertionResult Compare(const char* expected_expression, const char* actual_expression, BiggestInt expected, BiggestInt actual) { return CmpHelperEQ(expected_expression, actual_expression, expected, actual); } }; // This specialization is used when the first argument to ASSERT_EQ() // is a null pointer literal, like NULL, false, or 0. template <> class EqHelper { public: // We define two overloaded versions of Compare(). The first // version will be picked when the second argument to ASSERT_EQ() is // NOT a pointer, e.g. ASSERT_EQ(0, AnIntFunction()) or // EXPECT_EQ(false, a_bool). template static AssertionResult Compare( const char* expected_expression, const char* actual_expression, const T1& expected, const T2& actual, // The following line prevents this overload from being considered if T2 // is not a pointer type. We need this because ASSERT_EQ(NULL, my_ptr) // expands to Compare("", "", NULL, my_ptr), which requires a conversion // to match the Secret* in the other overload, which would otherwise make // this template match better. typename EnableIf::value>::type* = 0) { return CmpHelperEQ(expected_expression, actual_expression, expected, actual); } // This version will be picked when the second argument to ASSERT_EQ() is a // pointer, e.g. ASSERT_EQ(NULL, a_pointer). template static AssertionResult Compare( const char* expected_expression, const char* actual_expression, // We used to have a second template parameter instead of Secret*. That // template parameter would deduce to 'long', making this a better match // than the first overload even without the first overload's EnableIf. // Unfortunately, gcc with -Wconversion-null warns when "passing NULL to // non-pointer argument" (even a deduced integral argument), so the old // implementation caused warnings in user code. Secret* /* expected (NULL) */, T* actual) { // We already know that 'expected' is a null pointer. return CmpHelperEQ(expected_expression, actual_expression, static_cast(NULL), actual); } }; // Separate the error generating code from the code path to reduce the stack // frame size of CmpHelperOP. This helps reduce the overhead of some sanitizers // when calling EXPECT_OP in a tight loop. template AssertionResult CmpHelperOpFailure(const char* expr1, const char* expr2, const T1& val1, const T2& val2, const char* op) { return AssertionFailure() << "Expected: (" << expr1 << ") " << op << " (" << expr2 << "), actual: " << FormatForComparisonFailureMessage(val1, val2) << " vs " << FormatForComparisonFailureMessage(val2, val1); } // A macro for implementing the helper functions needed to implement // ASSERT_?? and EXPECT_??. It is here just to avoid copy-and-paste // of similar code. // // For each templatized helper function, we also define an overloaded // version for BiggestInt in order to reduce code bloat and allow // anonymous enums to be used with {ASSERT|EXPECT}_?? when compiled // with gcc 4. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. #define GTEST_IMPL_CMP_HELPER_(op_name, op)\ template \ AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \ const T1& val1, const T2& val2) {\ if (val1 op val2) {\ return AssertionSuccess();\ } else {\ return CmpHelperOpFailure(expr1, expr2, val1, val2, #op);\ }\ }\ GTEST_API_ AssertionResult CmpHelper##op_name(\ const char* expr1, const char* expr2, BiggestInt val1, BiggestInt val2) // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. // Implements the helper function for {ASSERT|EXPECT}_NE GTEST_IMPL_CMP_HELPER_(NE, !=); // Implements the helper function for {ASSERT|EXPECT}_LE GTEST_IMPL_CMP_HELPER_(LE, <=); // Implements the helper function for {ASSERT|EXPECT}_LT GTEST_IMPL_CMP_HELPER_(LT, <); // Implements the helper function for {ASSERT|EXPECT}_GE GTEST_IMPL_CMP_HELPER_(GE, >=); // Implements the helper function for {ASSERT|EXPECT}_GT GTEST_IMPL_CMP_HELPER_(GT, >); #undef GTEST_IMPL_CMP_HELPER_ // The helper function for {ASSERT|EXPECT}_STREQ. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult CmpHelperSTREQ(const char* expected_expression, const char* actual_expression, const char* expected, const char* actual); // The helper function for {ASSERT|EXPECT}_STRCASEEQ. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult CmpHelperSTRCASEEQ(const char* expected_expression, const char* actual_expression, const char* expected, const char* actual); // The helper function for {ASSERT|EXPECT}_STRNE. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult CmpHelperSTRNE(const char* s1_expression, const char* s2_expression, const char* s1, const char* s2); // The helper function for {ASSERT|EXPECT}_STRCASENE. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult CmpHelperSTRCASENE(const char* s1_expression, const char* s2_expression, const char* s1, const char* s2); // Helper function for *_STREQ on wide strings. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult CmpHelperSTREQ(const char* expected_expression, const char* actual_expression, const wchar_t* expected, const wchar_t* actual); // Helper function for *_STRNE on wide strings. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult CmpHelperSTRNE(const char* s1_expression, const char* s2_expression, const wchar_t* s1, const wchar_t* s2); } // namespace internal // IsSubstring() and IsNotSubstring() are intended to be used as the // first argument to {EXPECT,ASSERT}_PRED_FORMAT2(), not by // themselves. They check whether needle is a substring of haystack // (NULL is considered a substring of itself only), and return an // appropriate error message when they fail. // // The {needle,haystack}_expr arguments are the stringified // expressions that generated the two real arguments. GTEST_API_ AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const char* needle, const char* haystack); GTEST_API_ AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const wchar_t* needle, const wchar_t* haystack); GTEST_API_ AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const char* needle, const char* haystack); GTEST_API_ AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const wchar_t* needle, const wchar_t* haystack); GTEST_API_ AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const ::std::string& needle, const ::std::string& haystack); GTEST_API_ AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const ::std::string& needle, const ::std::string& haystack); #if GTEST_HAS_STD_WSTRING GTEST_API_ AssertionResult IsSubstring( const char* needle_expr, const char* haystack_expr, const ::std::wstring& needle, const ::std::wstring& haystack); GTEST_API_ AssertionResult IsNotSubstring( const char* needle_expr, const char* haystack_expr, const ::std::wstring& needle, const ::std::wstring& haystack); #endif // GTEST_HAS_STD_WSTRING namespace internal { // Helper template function for comparing floating-points. // // Template parameter: // // RawType: the raw floating-point type (either float or double) // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. template AssertionResult CmpHelperFloatingPointEQ(const char* expected_expression, const char* actual_expression, RawType expected, RawType actual) { const FloatingPoint lhs(expected), rhs(actual); if (lhs.AlmostEquals(rhs)) { return AssertionSuccess(); } ::std::stringstream expected_ss; expected_ss << std::setprecision(std::numeric_limits::digits10 + 2) << expected; ::std::stringstream actual_ss; actual_ss << std::setprecision(std::numeric_limits::digits10 + 2) << actual; return EqFailure(expected_expression, actual_expression, StringStreamToString(&expected_ss), StringStreamToString(&actual_ss), false); } // Helper function for implementing ASSERT_NEAR. // // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. GTEST_API_ AssertionResult DoubleNearPredFormat(const char* expr1, const char* expr2, const char* abs_error_expr, double val1, double val2, double abs_error); // INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. // A class that enables one to stream messages to assertion macros class GTEST_API_ AssertHelper { public: // Constructor. AssertHelper(TestPartResult::Type type, const char* file, int line, const char* message); ~AssertHelper(); // Message assignment is a semantic trick to enable assertion // streaming; see the GTEST_MESSAGE_ macro below. void operator=(const Message& message) const; private: // We put our data in a struct so that the size of the AssertHelper class can // be as small as possible. This is important because gcc is incapable of // re-using stack space even for temporary variables, so every EXPECT_EQ // reserves stack space for another AssertHelper. struct AssertHelperData { AssertHelperData(TestPartResult::Type t, const char* srcfile, int line_num, const char* msg) : type(t), file(srcfile), line(line_num), message(msg) { } TestPartResult::Type const type; const char* const file; int const line; std::string const message; private: GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelperData); }; AssertHelperData* const data_; GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelper); }; } // namespace internal #if GTEST_HAS_PARAM_TEST // The pure interface class that all value-parameterized tests inherit from. // A value-parameterized class must inherit from both ::testing::Test and // ::testing::WithParamInterface. In most cases that just means inheriting // from ::testing::TestWithParam, but more complicated test hierarchies // may need to inherit from Test and WithParamInterface at different levels. // // This interface has support for accessing the test parameter value via // the GetParam() method. // // Use it with one of the parameter generator defining functions, like Range(), // Values(), ValuesIn(), Bool(), and Combine(). // // class FooTest : public ::testing::TestWithParam { // protected: // FooTest() { // // Can use GetParam() here. // } // virtual ~FooTest() { // // Can use GetParam() here. // } // virtual void SetUp() { // // Can use GetParam() here. // } // virtual void TearDown { // // Can use GetParam() here. // } // }; // TEST_P(FooTest, DoesBar) { // // Can use GetParam() method here. // Foo foo; // ASSERT_TRUE(foo.DoesBar(GetParam())); // } // INSTANTIATE_TEST_CASE_P(OneToTenRange, FooTest, ::testing::Range(1, 10)); template class WithParamInterface { public: typedef T ParamType; virtual ~WithParamInterface() {} // The current parameter value. Is also available in the test fixture's // constructor. This member function is non-static, even though it only // references static data, to reduce the opportunity for incorrect uses // like writing 'WithParamInterface::GetParam()' for a test that // uses a fixture whose parameter type is int. const ParamType& GetParam() const { GTEST_CHECK_(parameter_ != NULL) << "GetParam() can only be called inside a value-parameterized test " << "-- did you intend to write TEST_P instead of TEST_F?"; return *parameter_; } private: // Sets parameter value. The caller is responsible for making sure the value // remains alive and unchanged throughout the current test. static void SetParam(const ParamType* parameter) { parameter_ = parameter; } // Static value used for accessing parameter during a test lifetime. static const ParamType* parameter_; // TestClass must be a subclass of WithParamInterface and Test. template friend class internal::ParameterizedTestFactory; }; template const T* WithParamInterface::parameter_ = NULL; // Most value-parameterized classes can ignore the existence of // WithParamInterface, and can just inherit from ::testing::TestWithParam. template class TestWithParam : public Test, public WithParamInterface { }; #endif // GTEST_HAS_PARAM_TEST // Macros for indicating success/failure in test code. // ADD_FAILURE unconditionally adds a failure to the current test. // SUCCEED generates a success - it doesn't automatically make the // current test successful, as a test is only successful when it has // no failure. // // EXPECT_* verifies that a certain condition is satisfied. If not, // it behaves like ADD_FAILURE. In particular: // // EXPECT_TRUE verifies that a Boolean condition is true. // EXPECT_FALSE verifies that a Boolean condition is false. // // FAIL and ASSERT_* are similar to ADD_FAILURE and EXPECT_*, except // that they will also abort the current function on failure. People // usually want the fail-fast behavior of FAIL and ASSERT_*, but those // writing data-driven tests often find themselves using ADD_FAILURE // and EXPECT_* more. // Generates a nonfatal failure with a generic message. #define ADD_FAILURE() GTEST_NONFATAL_FAILURE_("Failed") // Generates a nonfatal failure at the given source file location with // a generic message. #define ADD_FAILURE_AT(file, line) \ GTEST_MESSAGE_AT_(file, line, "Failed", \ ::testing::TestPartResult::kNonFatalFailure) // Generates a fatal failure with a generic message. #define GTEST_FAIL() GTEST_FATAL_FAILURE_("Failed") // Define this macro to 1 to omit the definition of FAIL(), which is a // generic name and clashes with some other libraries. #if !GTEST_DONT_DEFINE_FAIL # define FAIL() GTEST_FAIL() #endif // Generates a success with a generic message. #define GTEST_SUCCEED() GTEST_SUCCESS_("Succeeded") // Define this macro to 1 to omit the definition of SUCCEED(), which // is a generic name and clashes with some other libraries. #if !GTEST_DONT_DEFINE_SUCCEED # define SUCCEED() GTEST_SUCCEED() #endif // Macros for testing exceptions. // // * {ASSERT|EXPECT}_THROW(statement, expected_exception): // Tests that the statement throws the expected exception. // * {ASSERT|EXPECT}_NO_THROW(statement): // Tests that the statement doesn't throw any exception. // * {ASSERT|EXPECT}_ANY_THROW(statement): // Tests that the statement throws an exception. #define EXPECT_THROW(statement, expected_exception) \ GTEST_TEST_THROW_(statement, expected_exception, GTEST_NONFATAL_FAILURE_) #define EXPECT_NO_THROW(statement) \ GTEST_TEST_NO_THROW_(statement, GTEST_NONFATAL_FAILURE_) #define EXPECT_ANY_THROW(statement) \ GTEST_TEST_ANY_THROW_(statement, GTEST_NONFATAL_FAILURE_) #define ASSERT_THROW(statement, expected_exception) \ GTEST_TEST_THROW_(statement, expected_exception, GTEST_FATAL_FAILURE_) #define ASSERT_NO_THROW(statement) \ GTEST_TEST_NO_THROW_(statement, GTEST_FATAL_FAILURE_) #define ASSERT_ANY_THROW(statement) \ GTEST_TEST_ANY_THROW_(statement, GTEST_FATAL_FAILURE_) // Boolean assertions. Condition can be either a Boolean expression or an // AssertionResult. For more information on how to use AssertionResult with // these macros see comments on that class. #define EXPECT_TRUE(condition) \ GTEST_TEST_BOOLEAN_(condition, #condition, false, true, \ GTEST_NONFATAL_FAILURE_) #define EXPECT_FALSE(condition) \ GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \ GTEST_NONFATAL_FAILURE_) #define ASSERT_TRUE(condition) \ GTEST_TEST_BOOLEAN_(condition, #condition, false, true, \ GTEST_FATAL_FAILURE_) #define ASSERT_FALSE(condition) \ GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \ GTEST_FATAL_FAILURE_) // Includes the auto-generated header that implements a family of // generic predicate assertion macros. // Copyright 2006, Google Inc. // All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. // This file is AUTOMATICALLY GENERATED on 10/31/2011 by command // 'gen_gtest_pred_impl.py 5'. DO NOT EDIT BY HAND! // // Implements a family of generic predicate assertion macros. #ifndef GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_ #define GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_ // Makes sure this header is not included before gtest.h. #ifndef GTEST_INCLUDE_GTEST_GTEST_H_ # error Do not include gtest_pred_impl.h directly. Include gtest.h instead. #endif // GTEST_INCLUDE_GTEST_GTEST_H_ // This header implements a family of generic predicate assertion // macros: // // ASSERT_PRED_FORMAT1(pred_format, v1) // ASSERT_PRED_FORMAT2(pred_format, v1, v2) // ... // // where pred_format is a function or functor that takes n (in the // case of ASSERT_PRED_FORMATn) values and their source expression // text, and returns a testing::AssertionResult. See the definition // of ASSERT_EQ in gtest.h for an example. // // If you don't care about formatting, you can use the more // restrictive version: // // ASSERT_PRED1(pred, v1) // ASSERT_PRED2(pred, v1, v2) // ... // // where pred is an n-ary function or functor that returns bool, // and the values v1, v2, ..., must support the << operator for // streaming to std::ostream. // // We also define the EXPECT_* variations. // // For now we only support predicates whose arity is at most 5. // Please email googletestframework@googlegroups.com if you need // support for higher arities. // GTEST_ASSERT_ is the basic statement to which all of the assertions // in this file reduce. Don't use this in your code. #define GTEST_ASSERT_(expression, on_failure) \ GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ if (const ::testing::AssertionResult gtest_ar = (expression)) \ ; \ else \ on_failure(gtest_ar.failure_message()) // Helper function for implementing {EXPECT|ASSERT}_PRED1. Don't use // this in your code. template AssertionResult AssertPred1Helper(const char* pred_text, const char* e1, Pred pred, const T1& v1) { if (pred(v1)) return AssertionSuccess(); return AssertionFailure() << pred_text << "(" << e1 << ") evaluates to false, where" << "\n" << e1 << " evaluates to " << v1; } // Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT1. // Don't use this in your code. #define GTEST_PRED_FORMAT1_(pred_format, v1, on_failure)\ GTEST_ASSERT_(pred_format(#v1, v1), \ on_failure) // Internal macro for implementing {EXPECT|ASSERT}_PRED1. Don't use // this in your code. #define GTEST_PRED1_(pred, v1, on_failure)\ GTEST_ASSERT_(::testing::AssertPred1Helper(#pred, \ #v1, \ pred, \ v1), on_failure) // Unary predicate assertion macros. #define EXPECT_PRED_FORMAT1(pred_format, v1) \ GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_NONFATAL_FAILURE_) #define EXPECT_PRED1(pred, v1) \ GTEST_PRED1_(pred, v1, GTEST_NONFATAL_FAILURE_) #define ASSERT_PRED_FORMAT1(pred_format, v1) \ GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_FATAL_FAILURE_) #define ASSERT_PRED1(pred, v1) \ GTEST_PRED1_(pred, v1, GTEST_FATAL_FAILURE_) // Helper function for implementing {EXPECT|ASSERT}_PRED2. Don't use // this in your code. template AssertionResult AssertPred2Helper(const char* pred_text, const char* e1, const char* e2, Pred pred, const T1& v1, const T2& v2) { if (pred(v1, v2)) return AssertionSuccess(); return AssertionFailure() << pred_text << "(" << e1 << ", " << e2 << ") evaluates to false, where" << "\n" << e1 << " evaluates to " << v1 << "\n" << e2 << " evaluates to " << v2; } // Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT2. // Don't use this in your code. #define GTEST_PRED_FORMAT2_(pred_format, v1, v2, on_failure)\ GTEST_ASSERT_(pred_format(#v1, #v2, v1, v2), \ on_failure) // Internal macro for implementing {EXPECT|ASSERT}_PRED2. Don't use // this in your code. #define GTEST_PRED2_(pred, v1, v2, on_failure)\ GTEST_ASSERT_(::testing::AssertPred2Helper(#pred, \ #v1, \ #v2, \ pred, \ v1, \ v2), on_failure) // Binary predicate assertion macros. #define EXPECT_PRED_FORMAT2(pred_format, v1, v2) \ GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_NONFATAL_FAILURE_) #define EXPECT_PRED2(pred, v1, v2) \ GTEST_PRED2_(pred, v1, v2, GTEST_NONFATAL_FAILURE_) #define ASSERT_PRED_FORMAT2(pred_format, v1, v2) \ GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_FATAL_FAILURE_) #define ASSERT_PRED2(pred, v1, v2) \ GTEST_PRED2_(pred, v1, v2, GTEST_FATAL_FAILURE_) // Helper function for implementing {EXPECT|ASSERT}_PRED3. Don't use // this in your code. template AssertionResult AssertPred3Helper(const char* pred_text, const char* e1, const char* e2, const char* e3, Pred pred, const T1& v1, const T2& v2, const T3& v3) { if (pred(v1, v2, v3)) return AssertionSuccess(); return AssertionFailure() << pred_text << "(" << e1 << ", " << e2 << ", " << e3 << ") evaluates to false, where" << "\n" << e1 << " evaluates to " << v1 << "\n" << e2 << " evaluates to " << v2 << "\n" << e3 << " evaluates to " << v3; } // Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT3. // Don't use this in your code. #define GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, on_failure)\ GTEST_ASSERT_(pred_format(#v1, #v2, #v3, v1, v2, v3), \ on_failure) // Internal macro for implementing {EXPECT|ASSERT}_PRED3. Don't use // this in your code. #define GTEST_PRED3_(pred, v1, v2, v3, on_failure)\ GTEST_ASSERT_(::testing::AssertPred3Helper(#pred, \ #v1, \ #v2, \ #v3, \ pred, \ v1, \ v2, \ v3), on_failure) // Ternary predicate assertion macros. #define EXPECT_PRED_FORMAT3(pred_format, v1, v2, v3) \ GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_NONFATAL_FAILURE_) #define EXPECT_PRED3(pred, v1, v2, v3) \ GTEST_PRED3_(pred, v1, v2, v3, GTEST_NONFATAL_FAILURE_) #define ASSERT_PRED_FORMAT3(pred_format, v1, v2, v3) \ GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_FATAL_FAILURE_) #define ASSERT_PRED3(pred, v1, v2, v3) \ GTEST_PRED3_(pred, v1, v2, v3, GTEST_FATAL_FAILURE_) // Helper function for implementing {EXPECT|ASSERT}_PRED4. Don't use // this in your code. template AssertionResult AssertPred4Helper(const char* pred_text, const char* e1, const char* e2, const char* e3, const char* e4, Pred pred, const T1& v1, const T2& v2, const T3& v3, const T4& v4) { if (pred(v1, v2, v3, v4)) return AssertionSuccess(); return AssertionFailure() << pred_text << "(" << e1 << ", " << e2 << ", " << e3 << ", " << e4 << ") evaluates to false, where" << "\n" << e1 << " evaluates to " << v1 << "\n" << e2 << " evaluates to " << v2 << "\n" << e3 << " evaluates to " << v3 << "\n" << e4 << " evaluates to " << v4; } // Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT4. // Don't use this in your code. #define GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, on_failure)\ GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, v1, v2, v3, v4), \ on_failure) // Internal macro for implementing {EXPECT|ASSERT}_PRED4. Don't use // this in your code. #define GTEST_PRED4_(pred, v1, v2, v3, v4, on_failure)\ GTEST_ASSERT_(::testing::AssertPred4Helper(#pred, \ #v1, \ #v2, \ #v3, \ #v4, \ pred, \ v1, \ v2, \ v3, \ v4), on_failure) // 4-ary predicate assertion macros. #define EXPECT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \ GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_) #define EXPECT_PRED4(pred, v1, v2, v3, v4) \ GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_) #define ASSERT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \ GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_FATAL_FAILURE_) #define ASSERT_PRED4(pred, v1, v2, v3, v4) \ GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_FATAL_FAILURE_) // Helper function for implementing {EXPECT|ASSERT}_PRED5. Don't use // this in your code. template AssertionResult AssertPred5Helper(const char* pred_text, const char* e1, const char* e2, const char* e3, const char* e4, const char* e5, Pred pred, const T1& v1, const T2& v2, const T3& v3, const T4& v4, const T5& v5) { if (pred(v1, v2, v3, v4, v5)) return AssertionSuccess(); return AssertionFailure() << pred_text << "(" << e1 << ", " << e2 << ", " << e3 << ", " << e4 << ", " << e5 << ") evaluates to false, where" << "\n" << e1 << " evaluates to " << v1 << "\n" << e2 << " evaluates to " << v2 << "\n" << e3 << " evaluates to " << v3 << "\n" << e4 << " evaluates to " << v4 << "\n" << e5 << " evaluates to " << v5; } // Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT5. // Don't use this in your code. #define GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, on_failure)\ GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, #v5, v1, v2, v3, v4, v5), \ on_failure) // Internal macro for implementing {EXPECT|ASSERT}_PRED5. Don't use // this in your code. #define GTEST_PRED5_(pred, v1, v2, v3, v4, v5, on_failure)\ GTEST_ASSERT_(::testing::AssertPred5Helper(#pred, \ #v1, \ #v2, \ #v3, \ #v4, \ #v5, \ pred, \ v1, \ v2, \ v3, \ v4, \ v5), on_failure) // 5-ary predicate assertion macros. #define EXPECT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \ GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_) #define EXPECT_PRED5(pred, v1, v2, v3, v4, v5) \ GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_) #define ASSERT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \ GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_) #define ASSERT_PRED5(pred, v1, v2, v3, v4, v5) \ GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_) #endif // GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_ // Macros for testing equalities and inequalities. // // * {ASSERT|EXPECT}_EQ(expected, actual): Tests that expected == actual // * {ASSERT|EXPECT}_NE(v1, v2): Tests that v1 != v2 // * {ASSERT|EXPECT}_LT(v1, v2): Tests that v1 < v2 // * {ASSERT|EXPECT}_LE(v1, v2): Tests that v1 <= v2 // * {ASSERT|EXPECT}_GT(v1, v2): Tests that v1 > v2 // * {ASSERT|EXPECT}_GE(v1, v2): Tests that v1 >= v2 // // When they are not, Google Test prints both the tested expressions and // their actual values. The values must be compatible built-in types, // or you will get a compiler error. By "compatible" we mean that the // values can be compared by the respective operator. // // Note: // // 1. It is possible to make a user-defined type work with // {ASSERT|EXPECT}_??(), but that requires overloading the // comparison operators and is thus discouraged by the Google C++ // Usage Guide. Therefore, you are advised to use the // {ASSERT|EXPECT}_TRUE() macro to assert that two objects are // equal. // // 2. The {ASSERT|EXPECT}_??() macros do pointer comparisons on // pointers (in particular, C strings). Therefore, if you use it // with two C strings, you are testing how their locations in memory // are related, not how their content is related. To compare two C // strings by content, use {ASSERT|EXPECT}_STR*(). // // 3. {ASSERT|EXPECT}_EQ(expected, actual) is preferred to // {ASSERT|EXPECT}_TRUE(expected == actual), as the former tells you // what the actual value is when it fails, and similarly for the // other comparisons. // // 4. Do not depend on the order in which {ASSERT|EXPECT}_??() // evaluate their arguments, which is undefined. // // 5. These macros evaluate their arguments exactly once. // // Examples: // // EXPECT_NE(5, Foo()); // EXPECT_EQ(NULL, a_pointer); // ASSERT_LT(i, array_size); // ASSERT_GT(records.size(), 0) << "There is no record left."; #define EXPECT_EQ(expected, actual) \ EXPECT_PRED_FORMAT2(::testing::internal:: \ EqHelper::Compare, \ expected, actual) #define EXPECT_NE(expected, actual) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperNE, expected, actual) #define EXPECT_LE(val1, val2) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2) #define EXPECT_LT(val1, val2) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2) #define EXPECT_GE(val1, val2) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2) #define EXPECT_GT(val1, val2) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2) #define GTEST_ASSERT_EQ(expected, actual) \ ASSERT_PRED_FORMAT2(::testing::internal:: \ EqHelper::Compare, \ expected, actual) #define GTEST_ASSERT_NE(val1, val2) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2) #define GTEST_ASSERT_LE(val1, val2) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2) #define GTEST_ASSERT_LT(val1, val2) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2) #define GTEST_ASSERT_GE(val1, val2) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2) #define GTEST_ASSERT_GT(val1, val2) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2) // Define macro GTEST_DONT_DEFINE_ASSERT_XY to 1 to omit the definition of // ASSERT_XY(), which clashes with some users' own code. #if !GTEST_DONT_DEFINE_ASSERT_EQ # define ASSERT_EQ(val1, val2) GTEST_ASSERT_EQ(val1, val2) #endif #if !GTEST_DONT_DEFINE_ASSERT_NE # define ASSERT_NE(val1, val2) GTEST_ASSERT_NE(val1, val2) #endif #if !GTEST_DONT_DEFINE_ASSERT_LE # define ASSERT_LE(val1, val2) GTEST_ASSERT_LE(val1, val2) #endif #if !GTEST_DONT_DEFINE_ASSERT_LT # define ASSERT_LT(val1, val2) GTEST_ASSERT_LT(val1, val2) #endif #if !GTEST_DONT_DEFINE_ASSERT_GE # define ASSERT_GE(val1, val2) GTEST_ASSERT_GE(val1, val2) #endif #if !GTEST_DONT_DEFINE_ASSERT_GT # define ASSERT_GT(val1, val2) GTEST_ASSERT_GT(val1, val2) #endif // C-string Comparisons. All tests treat NULL and any non-NULL string // as different. Two NULLs are equal. // // * {ASSERT|EXPECT}_STREQ(s1, s2): Tests that s1 == s2 // * {ASSERT|EXPECT}_STRNE(s1, s2): Tests that s1 != s2 // * {ASSERT|EXPECT}_STRCASEEQ(s1, s2): Tests that s1 == s2, ignoring case // * {ASSERT|EXPECT}_STRCASENE(s1, s2): Tests that s1 != s2, ignoring case // // For wide or narrow string objects, you can use the // {ASSERT|EXPECT}_??() macros. // // Don't depend on the order in which the arguments are evaluated, // which is undefined. // // These macros evaluate their arguments exactly once. #define EXPECT_STREQ(expected, actual) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, expected, actual) #define EXPECT_STRNE(s1, s2) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2) #define EXPECT_STRCASEEQ(expected, actual) \ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, expected, actual) #define EXPECT_STRCASENE(s1, s2)\ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2) #define ASSERT_STREQ(expected, actual) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, expected, actual) #define ASSERT_STRNE(s1, s2) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2) #define ASSERT_STRCASEEQ(expected, actual) \ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, expected, actual) #define ASSERT_STRCASENE(s1, s2)\ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2) // Macros for comparing floating-point numbers. // // * {ASSERT|EXPECT}_FLOAT_EQ(expected, actual): // Tests that two float values are almost equal. // * {ASSERT|EXPECT}_DOUBLE_EQ(expected, actual): // Tests that two double values are almost equal. // * {ASSERT|EXPECT}_NEAR(v1, v2, abs_error): // Tests that v1 and v2 are within the given distance to each other. // // Google Test uses ULP-based comparison to automatically pick a default // error bound that is appropriate for the operands. See the // FloatingPoint template class in gtest-internal.h if you are // interested in the implementation details. #define EXPECT_FLOAT_EQ(expected, actual)\ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ expected, actual) #define EXPECT_DOUBLE_EQ(expected, actual)\ EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ expected, actual) #define ASSERT_FLOAT_EQ(expected, actual)\ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ expected, actual) #define ASSERT_DOUBLE_EQ(expected, actual)\ ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ expected, actual) #define EXPECT_NEAR(val1, val2, abs_error)\ EXPECT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \ val1, val2, abs_error) #define ASSERT_NEAR(val1, val2, abs_error)\ ASSERT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \ val1, val2, abs_error) // These predicate format functions work on floating-point values, and // can be used in {ASSERT|EXPECT}_PRED_FORMAT2*(), e.g. // // EXPECT_PRED_FORMAT2(testing::DoubleLE, Foo(), 5.0); // Asserts that val1 is less than, or almost equal to, val2. Fails // otherwise. In particular, it fails if either val1 or val2 is NaN. GTEST_API_ AssertionResult FloatLE(const char* expr1, const char* expr2, float val1, float val2); GTEST_API_ AssertionResult DoubleLE(const char* expr1, const char* expr2, double val1, double val2); #if GTEST_OS_WINDOWS // Macros that test for HRESULT failure and success, these are only useful // on Windows, and rely on Windows SDK macros and APIs to compile. // // * {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}(expr) // // When expr unexpectedly fails or succeeds, Google Test prints the // expected result and the actual result with both a human-readable // string representation of the error, if available, as well as the // hex result code. # define EXPECT_HRESULT_SUCCEEDED(expr) \ EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr)) # define ASSERT_HRESULT_SUCCEEDED(expr) \ ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr)) # define EXPECT_HRESULT_FAILED(expr) \ EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr)) # define ASSERT_HRESULT_FAILED(expr) \ ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr)) #endif // GTEST_OS_WINDOWS // Macros that execute statement and check that it doesn't generate new fatal // failures in the current thread. // // * {ASSERT|EXPECT}_NO_FATAL_FAILURE(statement); // // Examples: // // EXPECT_NO_FATAL_FAILURE(Process()); // ASSERT_NO_FATAL_FAILURE(Process()) << "Process() failed"; // #define ASSERT_NO_FATAL_FAILURE(statement) \ GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_FATAL_FAILURE_) #define EXPECT_NO_FATAL_FAILURE(statement) \ GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_NONFATAL_FAILURE_) // Causes a trace (including the source file path, the current line // number, and the given message) to be included in every test failure // message generated by code in the current scope. The effect is // undone when the control leaves the current scope. // // The message argument can be anything streamable to std::ostream. // // In the implementation, we include the current line number as part // of the dummy variable name, thus allowing multiple SCOPED_TRACE()s // to appear in the same block - as long as they are on different // lines. #define SCOPED_TRACE(message) \ ::testing::internal::ScopedTrace GTEST_CONCAT_TOKEN_(gtest_trace_, __LINE__)(\ __FILE__, __LINE__, ::testing::Message() << (message)) // Compile-time assertion for type equality. // StaticAssertTypeEq() compiles iff type1 and type2 are // the same type. The value it returns is not interesting. // // Instead of making StaticAssertTypeEq a class template, we make it a // function template that invokes a helper class template. This // prevents a user from misusing StaticAssertTypeEq by // defining objects of that type. // // CAVEAT: // // When used inside a method of a class template, // StaticAssertTypeEq() is effective ONLY IF the method is // instantiated. For example, given: // // template class Foo { // public: // void Bar() { testing::StaticAssertTypeEq(); } // }; // // the code: // // void Test1() { Foo foo; } // // will NOT generate a compiler error, as Foo::Bar() is never // actually instantiated. Instead, you need: // // void Test2() { Foo foo; foo.Bar(); } // // to cause a compiler error. template bool StaticAssertTypeEq() { (void)internal::StaticAssertTypeEqHelper(); return true; } // Defines a test. // // The first parameter is the name of the test case, and the second // parameter is the name of the test within the test case. // // The convention is to end the test case name with "Test". For // example, a test case for the Foo class can be named FooTest. // // Test code should appear between braces after an invocation of // this macro. Example: // // TEST(FooTest, InitializesCorrectly) { // Foo foo; // EXPECT_TRUE(foo.StatusIsOK()); // } // Note that we call GetTestTypeId() instead of GetTypeId< // ::testing::Test>() here to get the type ID of testing::Test. This // is to work around a suspected linker bug when using Google Test as // a framework on Mac OS X. The bug causes GetTypeId< // ::testing::Test>() to return different values depending on whether // the call is from the Google Test framework itself or from user test // code. GetTestTypeId() is guaranteed to always return the same // value, as it always calls GetTypeId<>() from the Google Test // framework. #define GTEST_TEST(test_case_name, test_name)\ GTEST_TEST_(test_case_name, test_name, \ ::testing::Test, ::testing::internal::GetTestTypeId()) // Define this macro to 1 to omit the definition of TEST(), which // is a generic name and clashes with some other libraries. #if !GTEST_DONT_DEFINE_TEST # define TEST(test_case_name, test_name) GTEST_TEST(test_case_name, test_name) #endif // Defines a test that uses a test fixture. // // The first parameter is the name of the test fixture class, which // also doubles as the test case name. The second parameter is the // name of the test within the test case. // // A test fixture class must be declared earlier. The user should put // his test code between braces after using this macro. Example: // // class FooTest : public testing::Test { // protected: // virtual void SetUp() { b_.AddElement(3); } // // Foo a_; // Foo b_; // }; // // TEST_F(FooTest, InitializesCorrectly) { // EXPECT_TRUE(a_.StatusIsOK()); // } // // TEST_F(FooTest, ReturnsElementCountCorrectly) { // EXPECT_EQ(0, a_.size()); // EXPECT_EQ(1, b_.size()); // } #define TEST_F(test_fixture, test_name)\ GTEST_TEST_(test_fixture, test_name, test_fixture, \ ::testing::internal::GetTypeId()) } // namespace testing // Use this function in main() to run all tests. It returns 0 if all // tests are successful, or 1 otherwise. // // RUN_ALL_TESTS() should be invoked after the command line has been // parsed by InitGoogleTest(). // // This function was formerly a macro; thus, it is in the global // namespace and has an all-caps name. int RUN_ALL_TESTS() GTEST_MUST_USE_RESULT_; inline int RUN_ALL_TESTS() { return ::testing::UnitTest::GetInstance()->Run(); } #endif // GTEST_INCLUDE_GTEST_GTEST_H_ rocksdb-4.5.1/thirdparty.inc000066400000000000000000000157231267530634000160530ustar00rootroot00000000000000# Edit definitions below to specify paths to include files and libraries of all 3rd party libraries # # Edit these lines to set defaults for use of external libraries # set(USE_GFLAGS_DEFAULT 0) # GFLAGS is disabled by default, enable with -DGFLAGS=1 cmake command line agrument set(USE_SNAPPY_DEFAULT 0) # SNAPPY is disabled by default, enable with -DSNAPPY=1 cmake command line agrument set(USE_LZ4_DEFAULT 0) # LZ4 is disabled by default, enable with -DLZ4=1 cmake command line agrument set(USE_ZLIB_DEFAULT 0) # ZLIB is disabled by default, enable with -DZLIB=1 cmake command line agrument set(USE_JEMALLOC_DEFAULT 0) # JEMALLOC is disabled by default, enable with -DJEMALLOC=1 cmake command line agrument # # This example assumes all the libraries locate in directories under THIRDPARTY_HOME environment variable # Set environment variable THIRDPARTY_HOME to point to your third party libraries home (Unix style dir separators) # or change the paths below to reflect where the libraries actually reside # set (THIRDPARTY_LIBS "") # Initialization, don't touch # # Edit these 4 lines to define paths to GFLAGS # set(GFLAGS_HOME $ENV{THIRDPARTY_HOME}/Gflags.Library) set(GFLAGS_INCLUDE ${GFLAGS_HOME}/inc/include) set(GFLAGS_LIB_DEBUG ${GFLAGS_HOME}/bin/debug/amd64/gflags.lib) set(GFLAGS_LIB_RELEASE ${GFLAGS_HOME}/bin/retail/amd64/gflags.lib) # ================================================== GFLAGS ================================================== # # Don't touch these lines # if (DEFINED GFLAGS) set(USE_GFLAGS ${GFLAGS}) else () set(USE_GFLAGS ${USE_GFLAGS_DEFAULT}) endif () if (${USE_GFLAGS} EQUAL 1) message(STATUS "GFLAGS library is enabled") if(DEFINED ENV{GFLAGS_INCLUDE}) set(GFLAGS_INCLUDE $ENV{GFLAGS_INCLUDE}) endif() if(DEFINED ENV{GFLAGS_LIB_DEBUG}) set(GFLAGS_LIB_DEBUG $ENV{GFLAGS_LIB_DEBUG}) endif() if(DEFINED ENV{GFLAGS_LIB_RELEASE}) set(GFLAGS_LIB_RELEASE $ENV{GFLAGS_LIB_RELEASE}) endif() set(GFLAGS_CXX_FLAGS -DGFLAGS=gflags) set(GFLAGS_LIBS debug ${GFLAGS_LIB_DEBUG} optimized ${GFLAGS_LIB_RELEASE}) add_definitions(${GFLAGS_CXX_FLAGS}) include_directories(${GFLAGS_INCLUDE}) set (THIRDPARTY_LIBS ${THIRDPARTY_LIBS} ${GFLAGS_LIBS}) else () message(STATUS "GFLAGS library is disabled") endif () # ================================================== SNAPPY ================================================== # # Edit these 4 lines to define paths to Snappy # set(SNAPPY_HOME $ENV{THIRDPARTY_HOME}/Snappy.Library) set(SNAPPY_INCLUDE ${SNAPPY_HOME}/inc/inc) set(SNAPPY_LIB_DEBUG ${SNAPPY_HOME}/bin/debug/amd64/snappy.lib) set(SNAPPY_LIB_RELEASE ${SNAPPY_HOME}/bin/retail/amd64/snappy.lib) # # Don't touch these lines # if (DEFINED SNAPPY) set(USE_SNAPPY ${SNAPPY}) else () set(USE_SNAPPY ${USE_SNAPPY_DEFAULT}) endif () if (${USE_SNAPPY} EQUAL 1) message(STATUS "SNAPPY library is enabled") if(DEFINED ENV{SNAPPY_INCLUDE}) set(SNAPPY_INCLUDE $ENV{SNAPPY_INCLUDE}) endif() if(DEFINED ENV{SNAPPY_LIB_DEBUG}) set(SNAPPY_LIB_DEBUG $ENV{SNAPPY_LIB_DEBUG}) endif() if(DEFINED ENV{SNAPPY_LIB_RELEASE}) set(SNAPPY_LIB_RELEASE $ENV{SNAPPY_LIB_RELEASE}) endif() set(SNAPPY_CXX_FLAGS -DSNAPPY) set(SNAPPY_LIBS debug ${SNAPPY_LIB_DEBUG} optimized ${SNAPPY_LIB_RELEASE}) add_definitions(${SNAPPY_CXX_FLAGS}) include_directories(${SNAPPY_INCLUDE}) set (THIRDPARTY_LIBS ${THIRDPARTY_LIBS} ${SNAPPY_LIBS}) else () message(STATUS "SNAPPY library is disabled") endif () # ================================================== LZ4 ================================================== # # Edit these 4 lines to define paths to LZ4 # set(LZ4_HOME $ENV{THIRDPARTY_HOME}/LZ4.Library) set(LZ4_INCLUDE ${LZ4_HOME}/inc/include) set(LZ4_LIB_DEBUG ${LZ4_HOME}/bin/debug/amd64/lz4.lib) set(LZ4_LIB_RELEASE ${LZ4_HOME}/bin/retail/amd64/lz4.lib) # # Don't touch these lines # if (DEFINED LZ4) set(USE_LZ4 ${LZ4}) else () set(USE_LZ4 ${USE_LZ4_DEFAULT}) endif () if (${USE_LZ4} EQUAL 1) message(STATUS "LZ4 library is enabled") if(DEFINED ENV{LZ4_INCLUDE}) set(LZ4_INCLUDE $ENV{LZ4_INCLUDE}) endif() if(DEFINED ENV{LZ4_LIB_DEBUG}) set(LZ4_LIB_DEBUG $ENV{LZ4_LIB_DEBUG}) endif() if(DEFINED ENV{LZ4_LIB_RELEASE}) set(LZ4_LIB_RELEASE $ENV{LZ4_LIB_RELEASE}) endif() set(LZ4_CXX_FLAGS -DLZ4) set(LZ4_LIBS debug ${LZ4_LIB_DEBUG} optimized ${LZ4_LIB_RELEASE}) add_definitions(${LZ4_CXX_FLAGS}) include_directories(${LZ4_INCLUDE}) set (THIRDPARTY_LIBS ${THIRDPARTY_LIBS} ${LZ4_LIBS}) else () message(STATUS "LZ4 library is disabled") endif () # ================================================== ZLIB ================================================== # # Edit these 4 lines to define paths to ZLIB # set(ZLIB_HOME $ENV{THIRDPARTY_HOME}/ZLIB.Library) set(ZLIB_INCLUDE ${ZLIB_HOME}/inc/include) set(ZLIB_LIB_DEBUG ${ZLIB_HOME}/bin/debug/amd64/zlib.lib) set(ZLIB_LIB_RELEASE ${ZLIB_HOME}/bin/retail/amd64/zlib.lib) # # Don't touch these lines # if (DEFINED ZLIB) set(USE_ZLIB ${ZLIB}) else () set(USE_ZLIB ${USE_ZLIB_DEFAULT}) endif () if (${USE_ZLIB} EQUAL 1) message(STATUS "ZLIB library is enabled") if(DEFINED ENV{ZLIB_INCLUDE}) set(ZLIB_INCLUDE $ENV{ZLIB_INCLUDE}) endif() if(DEFINED ENV{ZLIB_LIB_DEBUG}) set(ZLIB_LIB_DEBUG $ENV{ZLIB_LIB_DEBUG}) endif() if(DEFINED ENV{ZLIB_LIB_RELEASE}) set(ZLIB_LIB_RELEASE $ENV{ZLIB_LIB_RELEASE}) endif() set(ZLIB_CXX_FLAGS -DZLIB) set(ZLIB_LIBS debug ${ZLIB_LIB_DEBUG} optimized ${ZLIB_LIB_RELEASE}) add_definitions(${ZLIB_CXX_FLAGS}) include_directories(${ZLIB_INCLUDE}) set (THIRDPARTY_LIBS ${THIRDPARTY_LIBS} ${ZLIB_LIBS}) else () message(STATUS "ZLIB library is disabled") endif () # # Edit these 4 lines to define paths to Jemalloc # set(JEMALLOC_HOME $ENV{THIRDPARTY_HOME}/Jemalloc.Library) set(JEMALLOC_INCLUDE ${JEMALLOC_HOME}/inc/include) set(JEMALLOC_LIB_DEBUG ${JEMALLOC_HOME}/bin/debug/amd64/jemalloc.lib) set(JEMALLOC_LIB_RELEASE ${JEMALLOC_HOME}/bin/retail/amd64/jemalloc.lib) # ================================================== JEMALLOC ================================================== # # Don't touch these lines # if (DEFINED JEMALLOC) set(USE_JEMALLOC ${JEMALLOC}) else () set(USE_JEMALLOC ${USE_JEMALLOC_DEFAULT}) endif () if (${USE_JEMALLOC} EQUAL 1) message(STATUS "JEMALLOC library is enabled") set(JEMALLOC_CXX_FLAGS -DJEMALLOC) if(DEFINED ENV{JEMALLOC_INCLUDE}) set(JEMALLOC_INCLUDE $ENV{JEMALLOC_INCLUDE}) endif() if(DEFINED ENV{JEMALLOC_LIB_DEBUG}) set(JEMALLOC_LIB_DEBUG $ENV{JEMALLOC_LIB_DEBUG}) endif() if(DEFINED ENV{JEMALLOC_LIB_RELEASE}) set(JEMALLOC_LIB_RELEASE $ENV{JEMALLOC_LIB_RELEASE}) endif() set(JEMALLOC_LIBS debug ${JEMALLOC_LIB_DEBUG} optimized ${JEMALLOC_LIB_RELEASE}) add_definitions(${JEMALLOC_CXX_FLAGS}) include_directories(${JEMALLOC_INCLUDE}) set (THIRDPARTY_LIBS ${THIRDPARTY_LIBS} ${JEMALLOC_LIBS}) set (ARTIFACT_SUFFIX "_je") else () set (ARTIFACT_SUFFIX "") message(STATUS "JEMALLOC library is disabled") endif () rocksdb-4.5.1/tools/000077500000000000000000000000001267530634000143165ustar00rootroot00000000000000rocksdb-4.5.1/tools/Dockerfile000066400000000000000000000001211267530634000163020ustar00rootroot00000000000000FROM buildpack-deps:wheezy ADD ./ldb /rocksdb/tools/ldb CMD /rocksdb/tools/ldb rocksdb-4.5.1/tools/auto_sanity_test.sh000077500000000000000000000052231267530634000202550ustar00rootroot00000000000000TMP_DIR="/tmp/rocksdb-sanity-test" if [ "$#" -lt 2 ]; then echo "usage: ./auto_sanity_test.sh [new_commit] [old_commit]" echo "Missing either [new_commit] or [old_commit], perform sanity check with the latest and 10th latest commits." recent_commits=`git log | grep -e "^commit [a-z0-9]\+$"| head -n10 | sed -e 's/commit //g'` commit_new=`echo "$recent_commits" | head -n1` commit_old=`echo "$recent_commits" | tail -n1` echo "the most recent commits are:" echo "$recent_commits" else commit_new=$1 commit_old=$2 fi if [ ! -d $TMP_DIR ]; then mkdir $TMP_DIR fi dir_new="${TMP_DIR}/${commit_new}" dir_old="${TMP_DIR}/${commit_old}" function makestuff() { echo "make clean" make clean > /dev/null echo "make db_sanity_test -j32" make db_sanity_test -j32 > /dev/null if [ $? -ne 0 ]; then echo "[ERROR] Failed to perform 'make db_sanity_test'" exit 1 fi } rm -r -f $dir_new rm -r -f $dir_old echo "Running db sanity check with commits $commit_new and $commit_old." echo "=============================================================" echo "Making build $commit_new" git checkout $commit_new if [ $? -ne 0 ]; then echo "[ERROR] Can't checkout $commit_new" exit 1 fi makestuff mv db_sanity_test new_db_sanity_test echo "Creating db based on the new commit --- $commit_new" ./new_db_sanity_test $dir_new create cp ./tools/db_sanity_test.cc $dir_new cp ./tools/auto_sanity_test.sh $dir_new echo "=============================================================" echo "Making build $commit_old" git checkout $commit_old if [ $? -ne 0 ]; then echo "[ERROR] Can't checkout $commit_old" exit 1 fi cp -f $dir_new/db_sanity_test.cc ./tools/. cp -f $dir_new/auto_sanity_test.sh ./tools/. makestuff mv db_sanity_test old_db_sanity_test echo "Creating db based on the old commit --- $commit_old" ./old_db_sanity_test $dir_old create echo "=============================================================" echo "[Backward Compatibility Check]" echo "Verifying old db $dir_old using the new commit --- $commit_new" ./new_db_sanity_test $dir_old verify if [ $? -ne 0 ]; then echo "[ERROR] Backward Compatibility Check fails:" echo " Verification of $dir_old using commit $commit_new failed." exit 2 fi echo "=============================================================" echo "[Forward Compatibility Check]" echo "Verifying new db $dir_new using the old commit --- $commit_old" ./old_db_sanity_test $dir_new verify if [ $? -ne 0 ]; then echo "[ERROR] Forward Compatibility Check fails:" echo " $dir_new using commit $commit_old failed." exit 2 fi rm old_db_sanity_test rm new_db_sanity_test rm -rf $dir_new rm -rf $dir_old echo "Auto sanity test passed!" rocksdb-4.5.1/tools/benchmark.sh000077500000000000000000000304321267530634000166110ustar00rootroot00000000000000#!/bin/bash # REQUIRE: db_bench binary exists in the current directory if [ $# -ne 1 ]; then echo -n "./benchmark.sh [bulkload/fillseq/overwrite/filluniquerandom/" echo "readrandom/readwhilewriting/readwhilemerging/updaterandom/" echo "mergerandom/randomtransaction]" exit 0 fi # size constants K=1024 M=$((1024 * K)) G=$((1024 * M)) if [ -z $DB_DIR ]; then echo "DB_DIR is not defined" exit 0 fi if [ -z $WAL_DIR ]; then echo "WAL_DIR is not defined" exit 0 fi output_dir=${OUTPUT_DIR:-/tmp/} if [ ! -d $output_dir ]; then mkdir -p $output_dir fi # all multithreaded tests run with sync=1 unless # $DB_BENCH_NO_SYNC is defined syncval="1" if [ ! -z $DB_BENCH_NO_SYNC ]; then echo "Turning sync off for all multithreaded tests" syncval="0"; fi num_threads=${NUM_THREADS:-16} mb_written_per_sec=${MB_WRITE_PER_SEC:-0} # Only for tests that do range scans num_nexts_per_seek=${NUM_NEXTS_PER_SEEK:-10} cache_size=${CACHE_SIZE:-$((1 * G))} duration=${DURATION:-0} num_keys=${NUM_KEYS:-$((1 * G))} key_size=20 value_size=${VALUE_SIZE:-400} block_size=${BLOCK_SIZE:-8192} const_params=" --db=$DB_DIR \ --wal_dir=$WAL_DIR \ --disable_data_sync=0 \ \ --num=$num_keys \ --num_levels=6 \ --key_size=$key_size \ --value_size=$value_size \ --block_size=$block_size \ --cache_size=$cache_size \ --cache_numshardbits=6 \ --compression_type=snappy \ --min_level_to_compress=3 \ --compression_ratio=0.5 \ --level_compaction_dynamic_level_bytes=true \ --bytes_per_sync=$((8 * M)) \ --cache_index_and_filter_blocks=0 \ --benchmark_write_rate_limit=$(( 1024 * 1024 * $mb_written_per_sec )) \ \ --hard_rate_limit=3 \ --rate_limit_delay_max_milliseconds=1000000 \ --write_buffer_size=$((128 * M)) \ --max_write_buffer_number=8 \ --target_file_size_base=$((128 * M)) \ --max_bytes_for_level_base=$((1 * G)) \ \ --verify_checksum=1 \ --delete_obsolete_files_period_micros=$((60 * M)) \ --max_grandparent_overlap_factor=8 \ --max_bytes_for_level_multiplier=8 \ \ --statistics=0 \ --stats_per_interval=1 \ --stats_interval_seconds=60 \ --histogram=1 \ \ --memtablerep=skip_list \ --bloom_bits=10 \ --open_files=-1" l0_config=" --level0_file_num_compaction_trigger=4 \ --level0_slowdown_writes_trigger=12 \ --level0_stop_writes_trigger=20" if [ $duration -gt 0 ]; then const_params="$const_params --duration=$duration" fi params_w="$const_params $l0_config --max_background_compactions=16 --max_background_flushes=7" params_bulkload="$const_params --max_background_compactions=16 --max_background_flushes=7 \ --level0_file_num_compaction_trigger=$((10 * M)) \ --level0_slowdown_writes_trigger=$((10 * M)) \ --level0_stop_writes_trigger=$((10 * M))" function summarize_result { test_out=$1 test_name=$2 bench_name=$3 # Note that this function assumes that the benchmark executes long enough so # that "Compaction Stats" is written to stdout at least once. If it won't # happen then empty output from grep when searching for "Sum" will cause # syntax errors. uptime=$( grep ^Uptime\(secs $test_out | tail -1 | awk '{ printf "%.0f", $2 }' ) stall_time=$( grep "^Cumulative stall" $test_out | tail -1 | awk '{ print $3 }' ) stall_pct=$( grep "^Cumulative stall" $test_out| tail -1 | awk '{ print $5 }' ) ops_sec=$( grep ^${bench_name} $test_out | awk '{ print $5 }' ) mb_sec=$( grep ^${bench_name} $test_out | awk '{ print $7 }' ) lo_wgb=$( grep "^ L0" $test_out | tail -1 | awk '{ print $8 }' ) sum_wgb=$( grep "^ Sum" $test_out | tail -1 | awk '{ print $8 }' ) sum_size=$( grep "^ Sum" $test_out | tail -1 | awk '{ printf "%.1f", $3 / 1024.0 }' ) wamp=$( echo "scale=1; $sum_wgb / $lo_wgb" | bc ) wmb_ps=$( echo "scale=1; ( $sum_wgb * 1024.0 ) / $uptime" | bc ) usecs_op=$( grep ^${bench_name} $test_out | awk '{ printf "%.1f", $3 }' ) p50=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.1f", $3 }' ) p75=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.1f", $5 }' ) p99=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.0f", $7 }' ) p999=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.0f", $9 }' ) p9999=$( grep "^Percentiles:" $test_out | tail -1 | awk '{ printf "%.0f", $11 }' ) echo -e "$ops_sec\t$mb_sec\t$sum_size\t$lo_wgb\t$sum_wgb\t$wamp\t$wmb_ps\t$usecs_op\t$p50\t$p75\t$p99\t$p999\t$p9999\t$uptime\t$stall_time\t$stall_pct\t$test_name" \ >> $output_dir/report.txt } function run_bulkload { # This runs with a vector memtable and the WAL disabled to load faster. It is still crash safe and the # client can discover where to restart a load after a crash. I think this is a good way to load. echo "Bulk loading $num_keys random keys" cmd="./db_bench --benchmarks=fillrandom \ --use_existing_db=0 \ --disable_auto_compactions=1 \ --sync=0 \ $params_bulkload \ --threads=1 \ --memtablerep=vector \ --disable_wal=1 \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/benchmark_bulkload_fillrandom.log" echo $cmd | tee $output_dir/benchmark_bulkload_fillrandom.log eval $cmd summarize_result $output_dir/benchmark_bulkload_fillrandom.log bulkload fillrandom echo "Compacting..." cmd="./db_bench --benchmarks=compact \ --use_existing_db=1 \ --disable_auto_compactions=1 \ --sync=0 \ $params_w \ --threads=1 \ 2>&1 | tee -a $output_dir/benchmark_bulkload_compact.log" echo $cmd | tee $output_dir/benchmark_bulkload_compact.log eval $cmd } function run_fillseq { # This runs with a vector memtable. WAL can be either disabled or enabled # depending on the input parameter (1 for disabled, 0 for enabled). The main # benefit behind disabling WAL is to make loading faster. It is still crash # safe and the client can discover where to restart a load after a crash. I # think this is a good way to load. # Make sure that we'll have unique names for all the files so that data won't # be overwritten. if [ $1 == 1 ]; then log_file_name=$output_dir/benchmark_fillseq.wal_disabled.v${value_size}.log test_name=fillseq.wal_disabled.v${value_size} else log_file_name=$output_dir/benchmark_fillseq.wal_enabled.v${value_size}.log test_name=fillseq.wal_enabled.v${value_size} fi echo "Loading $num_keys keys sequentially" cmd="./db_bench --benchmarks=fillseq \ --use_existing_db=0 \ --sync=0 \ $params_w \ --min_level_to_compress=0 \ --threads=1 \ --memtablerep=vector \ --disable_wal=$1 \ --seed=$( date +%s ) \ 2>&1 | tee -a $log_file_name" echo $cmd | tee $log_file_name eval $cmd # The constant "fillseq" which we pass to db_bench is the benchmark name. summarize_result $log_file_name $test_name fillseq } function run_change { operation=$1 echo "Do $num_keys random $operation" out_name="benchmark_${operation}.t${num_threads}.s${syncval}.log" cmd="./db_bench --benchmarks=$operation \ --use_existing_db=1 \ --sync=$syncval \ $params_w \ --threads=$num_threads \ --merge_operator=\"put\" \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} ${operation}.t${num_threads}.s${syncval} $operation } function run_filluniquerandom { echo "Loading $num_keys unique keys randomly" cmd="./db_bench --benchmarks=filluniquerandom \ --use_existing_db=0 \ --sync=0 \ $params_w \ --threads=1 \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/benchmark_filluniquerandom.log" echo $cmd | tee $output_dir/benchmark_filluniquerandom.log eval $cmd summarize_result $output_dir/benchmark_filluniquerandom.log filluniquerandom filluniquerandom } function run_readrandom { echo "Reading $num_keys random keys" out_name="benchmark_readrandom.t${num_threads}.log" cmd="./db_bench --benchmarks=readrandom \ --use_existing_db=1 \ $params_w \ --threads=$num_threads \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} readrandom.t${num_threads} readrandom } function run_readwhile { operation=$1 echo "Reading $num_keys random keys while $operation" out_name="benchmark_readwhile${operation}.t${num_threads}.log" cmd="./db_bench --benchmarks=readwhile${operation} \ --use_existing_db=1 \ --sync=$syncval \ $params_w \ --threads=$num_threads \ --merge_operator=\"put\" \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} readwhile${operation}.t${num_threads} readwhile${operation} } function run_rangewhile { operation=$1 full_name=$2 reverse_arg=$3 out_name="benchmark_${full_name}.t${num_threads}.log" echo "Range scan $num_keys random keys while ${operation} for reverse_iter=${reverse_arg}" cmd="./db_bench --benchmarks=seekrandomwhile${operation} \ --use_existing_db=1 \ --sync=$syncval \ $params_w \ --threads=$num_threads \ --merge_operator=\"put\" \ --seek_nexts=$num_nexts_per_seek \ --reverse_iterator=$reverse_arg \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} ${full_name}.t${num_threads} seekrandomwhile${operation} } function run_range { full_name=$1 reverse_arg=$2 out_name="benchmark_${full_name}.t${num_threads}.log" echo "Range scan $num_keys random keys for reverse_iter=${reverse_arg}" cmd="./db_bench --benchmarks=seekrandom \ --use_existing_db=1 \ $params_w \ --threads=$num_threads \ --seek_nexts=$num_nexts_per_seek \ --reverse_iterator=$reverse_arg \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} ${full_name}.t${num_threads} seekrandom } function run_randomtransaction { echo "..." cmd="./db_bench $params_r --benchmarks=randomtransaction \ --num=$num_keys \ --transaction_db \ --threads=5 \ --transaction_sets=5 \ 2>&1 | tee $output_dir/benchmark_randomtransaction.log" echo $cmd | tee $output_dir/benchmark_rangescanwhilewriting.log eval $cmd } function now() { echo `date +"%s"` } report="$output_dir/report.txt" schedule="$output_dir/schedule.txt" echo "===== Benchmark =====" # Run!!! IFS=',' read -a jobs <<< $1 for job in ${jobs[@]}; do if [ $job != debug ]; then echo "Start $job at `date`" | tee -a $schedule fi start=$(now) if [ $job = bulkload ]; then run_bulkload elif [ $job = fillseq_disable_wal ]; then run_fillseq 1 elif [ $job = fillseq_enable_wal ]; then run_fillseq 0 elif [ $job = overwrite ]; then run_change overwrite elif [ $job = updaterandom ]; then run_change updaterandom elif [ $job = mergerandom ]; then run_change mergerandom elif [ $job = filluniquerandom ]; then run_filluniquerandom elif [ $job = readrandom ]; then run_readrandom elif [ $job = fwdrange ]; then run_range $job false elif [ $job = revrange ]; then run_range $job true elif [ $job = readwhilewriting ]; then run_readwhile writing elif [ $job = readwhilemerging ]; then run_readwhile merging elif [ $job = fwdrangewhilewriting ]; then run_rangewhile writing $job false elif [ $job = revrangewhilewriting ]; then run_rangewhile writing $job true elif [ $job = fwdrangewhilemerging ]; then run_rangewhile merging $job false elif [ $job = revrangewhilemerging ]; then run_rangewhile merging $job true elif [ $job = randomtransaction ]; then run_randomtransaction elif [ $job = debug ]; then num_keys=1000; # debug echo "Setting num_keys to $num_keys" else echo "unknown job $job" exit fi end=$(now) if [ $job != debug ]; then echo "Complete $job in $((end-start)) seconds" | tee -a $schedule fi echo -e "ops/sec\tmb/sec\tSize-GB\tL0_MB\tSum_GB\tW-Amp\tW-MB/s\tusec/op\tp50\tp75\tp99\tp99.9\tp99.99\tUptime\tStall-time\tStall%\tTest" tail -1 $output_dir/report.txt done rocksdb-4.5.1/tools/benchmark_leveldb.sh000077500000000000000000000121371267530634000203100ustar00rootroot00000000000000#!/bin/bash # REQUIRE: db_bench binary exists in the current directory # # This should be used with the LevelDB fork listed here to use additional test options. # For more details on the changes see the blog post listed below. # https://github.com/mdcallag/leveldb-1 # http://smalldatum.blogspot.com/2015/04/comparing-leveldb-and-rocksdb-take-2.html if [ $# -ne 1 ]; then echo -n "./benchmark.sh [fillseq/overwrite/readrandom/readwhilewriting]" exit 0 fi # size constants K=1024 M=$((1024 * K)) G=$((1024 * M)) if [ -z $DB_DIR ]; then echo "DB_DIR is not defined" exit 0 fi output_dir=${OUTPUT_DIR:-/tmp/} if [ ! -d $output_dir ]; then mkdir -p $output_dir fi # all multithreaded tests run with sync=1 unless # $DB_BENCH_NO_SYNC is defined syncval="1" if [ ! -z $DB_BENCH_NO_SYNC ]; then echo "Turning sync off for all multithreaded tests" syncval="0"; fi num_threads=${NUM_THREADS:-16} # Only for *whilewriting, *whilemerging writes_per_second=${WRITES_PER_SECOND:-$((10 * K))} cache_size=${CACHE_SIZE:-$((1 * G))} num_keys=${NUM_KEYS:-$((1 * G))} key_size=20 value_size=${VALUE_SIZE:-400} block_size=${BLOCK_SIZE:-4096} const_params=" --db=$DB_DIR \ \ --num=$num_keys \ --value_size=$value_size \ --cache_size=$cache_size \ --compression_ratio=0.5 \ \ --write_buffer_size=$((2 * M)) \ \ --histogram=1 \ \ --bloom_bits=10 \ --open_files=$((20 * K))" params_w="$const_params " function summarize_result { test_out=$1 test_name=$2 bench_name=$3 nthr=$4 usecs_op=$( grep ^${bench_name} $test_out | awk '{ printf "%.1f", $3 }' ) mb_sec=$( grep ^${bench_name} $test_out | awk '{ printf "%.1f", $5 }' ) ops=$( grep "^Count:" $test_out | awk '{ print $2 }' ) ops_sec=$( echo "scale=0; (1000000.0 * $nthr) / $usecs_op" | bc ) avg=$( grep "^Count:" $test_out | awk '{ printf "%.1f", $4 }' ) p50=$( grep "^Min:" $test_out | awk '{ printf "%.1f", $4 }' ) echo -e "$ops_sec\t$mb_sec\t$usecs_op\t$avg\t$p50\t$test_name" \ >> $output_dir/report.txt } function run_fillseq { # This runs with a vector memtable and the WAL disabled to load faster. It is still crash safe and the # client can discover where to restart a load after a crash. I think this is a good way to load. echo "Loading $num_keys keys sequentially" cmd="./db_bench --benchmarks=fillseq \ --use_existing_db=0 \ --sync=0 \ $params_w \ --threads=1 \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/benchmark_fillseq.v${value_size}.log" echo $cmd | tee $output_dir/benchmark_fillseq.v${value_size}.log eval $cmd summarize_result $output_dir/benchmark_fillseq.v${value_size}.log fillseq.v${value_size} fillseq 1 } function run_change { operation=$1 echo "Do $num_keys random $operation" out_name="benchmark_${operation}.t${num_threads}.s${syncval}.log" cmd="./db_bench --benchmarks=$operation \ --use_existing_db=1 \ --sync=$syncval \ $params_w \ --threads=$num_threads \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} ${operation}.t${num_threads}.s${syncval} $operation $num_threads } function run_readrandom { echo "Reading $num_keys random keys" out_name="benchmark_readrandom.t${num_threads}.log" cmd="./db_bench --benchmarks=readrandom \ --use_existing_db=1 \ $params_w \ --threads=$num_threads \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} readrandom.t${num_threads} readrandom $num_threads } function run_readwhile { operation=$1 echo "Reading $num_keys random keys while $operation" out_name="benchmark_readwhile${operation}.t${num_threads}.log" cmd="./db_bench --benchmarks=readwhile${operation} \ --use_existing_db=1 \ --sync=$syncval \ $params_w \ --threads=$num_threads \ --writes_per_second=$writes_per_second \ --seed=$( date +%s ) \ 2>&1 | tee -a $output_dir/${out_name}" echo $cmd | tee $output_dir/${out_name} eval $cmd summarize_result $output_dir/${out_name} readwhile${operation}.t${num_threads} readwhile${operation} $num_threads } function now() { echo `date +"%s"` } report="$output_dir/report.txt" schedule="$output_dir/schedule.txt" echo "===== Benchmark =====" # Run!!! IFS=',' read -a jobs <<< $1 for job in ${jobs[@]}; do if [ $job != debug ]; then echo "Start $job at `date`" | tee -a $schedule fi start=$(now) if [ $job = fillseq ]; then run_fillseq elif [ $job = overwrite ]; then run_change overwrite elif [ $job = readrandom ]; then run_readrandom elif [ $job = readwhilewriting ]; then run_readwhile writing elif [ $job = debug ]; then num_keys=1000; # debug echo "Setting num_keys to $num_keys" else echo "unknown job $job" exit fi end=$(now) if [ $job != debug ]; then echo "Complete $job in $((end-start)) seconds" | tee -a $schedule fi echo -e "ops/sec\tmb/sec\tusec/op\tavg\tp50\tTest" tail -1 $output_dir/report.txt done rocksdb-4.5.1/tools/check_format_compatible.sh000077500000000000000000000065661267530634000215160ustar00rootroot00000000000000#!/bin/bash # # A shell script to load some pre generated data file to a DB using ldb tool # ./ldb needs to be avaible to be executed. # # Usage: