diff --git a/.appveyor.yml b/.appveyor.yml new file mode 100644 index 0000000..c24b17e --- /dev/null +++ b/.appveyor.yml @@ -0,0 +1,35 @@ +# Build matrix / environment variables are explained on: +# https://www.appveyor.com/docs/appveyor-yml/ +# This file can be validated on: https://ci.appveyor.com/tools/validate-yaml + +version: "{build}" + +environment: + matrix: + # AppVeyor currently has no custom job name feature. + # http://help.appveyor.com/discussions/questions/1623-can-i-provide-a-friendly-name-for-jobs + - JOB: Visual Studio 2017 + APPVEYOR_BUILD_WORKER_IMAGE: Visual Studio 2017 + CMAKE_GENERATOR: Visual Studio 15 2017 + +platform: + - x86 + - x64 + +configuration: + - RelWithDebInfo + - Debug + +build_script: + - git submodule update --init --recursive + - mkdir build + - cd build + - if "%platform%"=="x64" set CMAKE_GENERATOR=%CMAKE_GENERATOR% Win64 + - cmake --version + - cmake .. -G "%CMAKE_GENERATOR%" + -DCMAKE_CONFIGURATION_TYPES="%CONFIGURATION%" + - cmake --build . --config "%CONFIGURATION%" + - cd .. + +test_script: + - cd build && ctest --verbose --build-config "%CONFIGURATION%" && cd .. diff --git a/.clang-format b/.clang-format new file mode 100644 index 0000000..f493f75 --- /dev/null +++ b/.clang-format @@ -0,0 +1,18 @@ +# Run manually to reformat a file: +# clang-format -i --style=file +# find . -iname '*.cc' -o -iname '*.h' -o -iname '*.h.in' | xargs clang-format -i --style=file +BasedOnStyle: Google +DerivePointerAlignment: false + +# Public headers are in a different location in the internal Google repository. +# Order them so that when imported to the authoritative repository they will be +# in correct alphabetical order. +IncludeCategories: + - Regex: '^(<|"(benchmarks|db|helpers)/)' + Priority: 1 + - Regex: '^"(leveldb)/' + Priority: 2 + - Regex: '^(<|"(issues|port|table|third_party|util)/)' + Priority: 3 + - Regex: '.*' + Priority: 4 diff --git a/.gitignore b/.gitignore index 0630251..c4b2425 100644 --- a/.gitignore +++ b/.gitignore @@ -1,9 +1,8 @@ -build_config.mk -*.a -*.o -*.dylib* -*.so -*.so.* -*_test -db_bench -leveldbutil +# Editors. +*.sw* +.vscode +.DS_Store + +# Build directory. +build/ +out/ diff --git a/.travis.yml b/.travis.yml new file mode 100644 index 0000000..436e037 --- /dev/null +++ b/.travis.yml @@ -0,0 +1,79 @@ +# Build matrix / environment variables are explained on: +# http://about.travis-ci.org/docs/user/build-configuration/ +# This file can be validated on: http://lint.travis-ci.org/ + +language: cpp +dist: xenial +osx_image: xcode10.2 + +compiler: +- gcc +- clang +os: +- linux +- osx + +env: +- BUILD_TYPE=Debug +- BUILD_TYPE=RelWithDebInfo + +addons: + apt: + sources: + - llvm-toolchain-xenial-8 + - ubuntu-toolchain-r-test + packages: + - clang-8 + - cmake + - gcc-8 + - g++-8 + - libgoogle-perftools-dev + - libkyotocabinet-dev + - libsnappy-dev + - libsqlite3-dev + - ninja-build + homebrew: + packages: + - cmake + - crc32c + - gcc@8 + - gperftools + - kyotocabinet + - llvm@8 + - ninja + - snappy + - sqlite3 + update: true + +install: +# The following Homebrew packages aren't linked by default, and need to be +# prepended to the path explicitly. +- if [ "$TRAVIS_OS_NAME" == "osx" ]; then + export PATH="$(brew --prefix llvm)/bin:$PATH"; + fi +# /usr/bin/gcc points to an older compiler on both Linux and macOS. +- if [ "$CXX" = "g++" ]; then export CXX="g++-8" CC="gcc-8"; fi +# /usr/bin/clang points to an older compiler on both Linux and macOS. +# +# Homebrew's llvm package doesn't ship a versioned clang++ binary, so the values +# below don't work on macOS. Fortunately, the path change above makes the +# default values (clang and clang++) resolve to the correct compiler on macOS. +- if [ "$TRAVIS_OS_NAME" == "linux" ]; then + if [ "$CXX" = "clang++" ]; then export CXX="clang++-8" CC="clang-8"; fi; + fi +- echo ${CC} +- echo ${CXX} +- ${CXX} --version +- cmake --version + +before_script: +- mkdir -p build && cd build +- cmake .. -G Ninja -DCMAKE_BUILD_TYPE=$BUILD_TYPE +- cmake --build . +- cd .. + +script: +- cd build && ctest --verbose && cd .. +- "if [ -f build/db_bench ] ; then build/db_bench ; fi" +- "if [ -f build/db_bench_sqlite3 ] ; then build/db_bench_sqlite3 ; fi" +- "if [ -f build/db_bench_tree_db ] ; then build/db_bench_tree_db ; fi" diff --git a/CMakeLists.txt b/CMakeLists.txt new file mode 100644 index 0000000..1ea33ab --- /dev/null +++ b/CMakeLists.txt @@ -0,0 +1,454 @@ +# Copyright 2017 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. + +cmake_minimum_required(VERSION 3.9) +# Keep the version below in sync with the one in db.h +project(leveldb VERSION 1.22.0 LANGUAGES C CXX) + +# This project can use C11, but will gracefully decay down to C89. +set(CMAKE_C_STANDARD 11) +set(CMAKE_C_STANDARD_REQUIRED OFF) +set(CMAKE_C_EXTENSIONS OFF) + +# This project requires C++11. +set(CMAKE_CXX_STANDARD 11) +set(CMAKE_CXX_STANDARD_REQUIRED ON) +set(CMAKE_CXX_EXTENSIONS OFF) + +if (WIN32) + set(LEVELDB_PLATFORM_NAME LEVELDB_PLATFORM_WINDOWS) + # TODO(cmumford): Make UNICODE configurable for Windows. + add_definitions(-D_UNICODE -DUNICODE) +else (WIN32) + set(LEVELDB_PLATFORM_NAME LEVELDB_PLATFORM_POSIX) +endif (WIN32) + +option(LEVELDB_BUILD_TESTS "Build LevelDB's unit tests" ON) +option(LEVELDB_BUILD_BENCHMARKS "Build LevelDB's benchmarks" ON) +option(LEVELDB_INSTALL "Install LevelDB's header and library" ON) + +include(TestBigEndian) +test_big_endian(LEVELDB_IS_BIG_ENDIAN) + +include(CheckIncludeFile) +check_include_file("unistd.h" HAVE_UNISTD_H) + +include(CheckLibraryExists) +check_library_exists(crc32c crc32c_value "" HAVE_CRC32C) +check_library_exists(snappy snappy_compress "" HAVE_SNAPPY) +check_library_exists(tcmalloc malloc "" HAVE_TCMALLOC) + +include(CheckCXXSymbolExists) +# Using check_cxx_symbol_exists() instead of check_c_symbol_exists() because +# we're including the header from C++, and feature detection should use the same +# compiler language that the project will use later. Principles aside, some +# versions of do not expose fdatasync() in in standard C mode +# (-std=c11), but do expose the function in standard C++ mode (-std=c++11). +check_cxx_symbol_exists(fdatasync "unistd.h" HAVE_FDATASYNC) +check_cxx_symbol_exists(F_FULLFSYNC "fcntl.h" HAVE_FULLFSYNC) +check_cxx_symbol_exists(O_CLOEXEC "fcntl.h" HAVE_O_CLOEXEC) + +include(CheckCXXSourceCompiles) + +set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wstrict-prototypes") + +# Test whether -Wthread-safety is available. See +# https://clang.llvm.org/docs/ThreadSafetyAnalysis.html +# -Werror is necessary because unknown attributes only generate warnings. +set(OLD_CMAKE_REQUIRED_FLAGS ${CMAKE_REQUIRED_FLAGS}) +list(APPEND CMAKE_REQUIRED_FLAGS -Werror -Wthread-safety) +check_cxx_source_compiles(" +struct __attribute__((lockable)) Lock { + void Acquire() __attribute__((exclusive_lock_function())); + void Release() __attribute__((unlock_function())); +}; +struct ThreadSafeType { + Lock lock_; + int data_ __attribute__((guarded_by(lock_))); +}; +int main() { return 0; } +" HAVE_CLANG_THREAD_SAFETY) +set(CMAKE_REQUIRED_FLAGS ${OLD_CMAKE_REQUIRED_FLAGS}) + +# Test whether C++17 __has_include is available. +check_cxx_source_compiles(" +#if defined(__has_include) && __has_include() +#include +#endif +int main() { std::string str; return 0; } +" HAVE_CXX17_HAS_INCLUDE) + +set(LEVELDB_PUBLIC_INCLUDE_DIR "include/leveldb") +set(LEVELDB_PORT_CONFIG_DIR "include/port") + +configure_file( + "${PROJECT_SOURCE_DIR}/port/port_config.h.in" + "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" +) + +include_directories( + "${PROJECT_BINARY_DIR}/include" + "${PROJECT_SOURCE_DIR}" +) + +if(BUILD_SHARED_LIBS) + # Only export LEVELDB_EXPORT symbols from the shared library. + add_compile_options(-fvisibility=hidden) +endif(BUILD_SHARED_LIBS) + +add_library(leveldb "") +target_sources(leveldb + PRIVATE + "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" + "${PROJECT_SOURCE_DIR}/db/builder.cc" + "${PROJECT_SOURCE_DIR}/db/builder.h" + "${PROJECT_SOURCE_DIR}/db/c.cc" + "${PROJECT_SOURCE_DIR}/db/db_impl.cc" + "${PROJECT_SOURCE_DIR}/db/db_impl.h" + "${PROJECT_SOURCE_DIR}/db/db_iter.cc" + "${PROJECT_SOURCE_DIR}/db/db_iter.h" + "${PROJECT_SOURCE_DIR}/db/dbformat.cc" + "${PROJECT_SOURCE_DIR}/db/dbformat.h" + "${PROJECT_SOURCE_DIR}/db/dumpfile.cc" + "${PROJECT_SOURCE_DIR}/db/filename.cc" + "${PROJECT_SOURCE_DIR}/db/filename.h" + "${PROJECT_SOURCE_DIR}/db/log_format.h" + "${PROJECT_SOURCE_DIR}/db/log_reader.cc" + "${PROJECT_SOURCE_DIR}/db/log_reader.h" + "${PROJECT_SOURCE_DIR}/db/log_writer.cc" + "${PROJECT_SOURCE_DIR}/db/log_writer.h" + "${PROJECT_SOURCE_DIR}/db/memtable.cc" + "${PROJECT_SOURCE_DIR}/db/memtable.h" + "${PROJECT_SOURCE_DIR}/db/repair.cc" + "${PROJECT_SOURCE_DIR}/db/skiplist.h" + "${PROJECT_SOURCE_DIR}/db/snapshot.h" + "${PROJECT_SOURCE_DIR}/db/table_cache.cc" + "${PROJECT_SOURCE_DIR}/db/table_cache.h" + "${PROJECT_SOURCE_DIR}/db/version_edit.cc" + "${PROJECT_SOURCE_DIR}/db/version_edit.h" + "${PROJECT_SOURCE_DIR}/db/version_set.cc" + "${PROJECT_SOURCE_DIR}/db/version_set.h" + "${PROJECT_SOURCE_DIR}/db/write_batch_internal.h" + "${PROJECT_SOURCE_DIR}/db/write_batch.cc" + "${PROJECT_SOURCE_DIR}/port/port_stdcxx.h" + "${PROJECT_SOURCE_DIR}/port/port.h" + "${PROJECT_SOURCE_DIR}/port/thread_annotations.h" + "${PROJECT_SOURCE_DIR}/table/block_builder.cc" + "${PROJECT_SOURCE_DIR}/table/block_builder.h" + "${PROJECT_SOURCE_DIR}/table/block.cc" + "${PROJECT_SOURCE_DIR}/table/block.h" + "${PROJECT_SOURCE_DIR}/table/filter_block.cc" + "${PROJECT_SOURCE_DIR}/table/filter_block.h" + "${PROJECT_SOURCE_DIR}/table/format.cc" + "${PROJECT_SOURCE_DIR}/table/format.h" + "${PROJECT_SOURCE_DIR}/table/iterator_wrapper.h" + "${PROJECT_SOURCE_DIR}/table/iterator.cc" + "${PROJECT_SOURCE_DIR}/table/merger.cc" + "${PROJECT_SOURCE_DIR}/table/merger.h" + "${PROJECT_SOURCE_DIR}/table/table_builder.cc" + "${PROJECT_SOURCE_DIR}/table/table.cc" + "${PROJECT_SOURCE_DIR}/table/two_level_iterator.cc" + "${PROJECT_SOURCE_DIR}/table/two_level_iterator.h" + "${PROJECT_SOURCE_DIR}/util/arena.cc" + "${PROJECT_SOURCE_DIR}/util/arena.h" + "${PROJECT_SOURCE_DIR}/util/bloom.cc" + "${PROJECT_SOURCE_DIR}/util/cache.cc" + "${PROJECT_SOURCE_DIR}/util/coding.cc" + "${PROJECT_SOURCE_DIR}/util/coding.h" + "${PROJECT_SOURCE_DIR}/util/comparator.cc" + "${PROJECT_SOURCE_DIR}/util/crc32c.cc" + "${PROJECT_SOURCE_DIR}/util/crc32c.h" + "${PROJECT_SOURCE_DIR}/util/env.cc" + "${PROJECT_SOURCE_DIR}/util/filter_policy.cc" + "${PROJECT_SOURCE_DIR}/util/hash.cc" + "${PROJECT_SOURCE_DIR}/util/hash.h" + "${PROJECT_SOURCE_DIR}/util/logging.cc" + "${PROJECT_SOURCE_DIR}/util/logging.h" + "${PROJECT_SOURCE_DIR}/util/mutexlock.h" + "${PROJECT_SOURCE_DIR}/util/no_destructor.h" + "${PROJECT_SOURCE_DIR}/util/options.cc" + "${PROJECT_SOURCE_DIR}/util/random.h" + "${PROJECT_SOURCE_DIR}/util/status.cc" + + # Only CMake 3.3+ supports PUBLIC sources in targets exported by "install". + $<$:PUBLIC> + "${LEVELDB_PUBLIC_INCLUDE_DIR}/c.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/cache.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/comparator.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/db.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/dumpfile.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/env.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/export.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/filter_policy.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/iterator.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/options.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/slice.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/status.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/table_builder.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/table.h" + "${LEVELDB_PUBLIC_INCLUDE_DIR}/write_batch.h" +) + +if (WIN32) + target_sources(leveldb + PRIVATE + "${PROJECT_SOURCE_DIR}/util/env_windows.cc" + "${PROJECT_SOURCE_DIR}/util/windows_logger.h" + ) +else (WIN32) + target_sources(leveldb + PRIVATE + "${PROJECT_SOURCE_DIR}/util/env_posix.cc" + "${PROJECT_SOURCE_DIR}/util/posix_logger.h" + ) +endif (WIN32) + +# MemEnv is not part of the interface and could be pulled to a separate library. +target_sources(leveldb + PRIVATE + "${PROJECT_SOURCE_DIR}/helpers/memenv/memenv.cc" + "${PROJECT_SOURCE_DIR}/helpers/memenv/memenv.h" +) + +target_include_directories(leveldb + PUBLIC + $ + $ +) + +set_target_properties(leveldb + PROPERTIES VERSION ${PROJECT_VERSION} SOVERSION ${PROJECT_VERSION_MAJOR}) + +target_compile_definitions(leveldb + PRIVATE + # Used by include/export.h when building shared libraries. + LEVELDB_COMPILE_LIBRARY + # Used by port/port.h. + ${LEVELDB_PLATFORM_NAME}=1 +) +if (NOT HAVE_CXX17_HAS_INCLUDE) + target_compile_definitions(leveldb + PRIVATE + LEVELDB_HAS_PORT_CONFIG_H=1 + ) +endif(NOT HAVE_CXX17_HAS_INCLUDE) + +if(BUILD_SHARED_LIBS) + target_compile_definitions(leveldb + PUBLIC + # Used by include/export.h. + LEVELDB_SHARED_LIBRARY + ) +endif(BUILD_SHARED_LIBS) + +if(HAVE_CLANG_THREAD_SAFETY) + target_compile_options(leveldb + PUBLIC + -Werror -Wthread-safety) +endif(HAVE_CLANG_THREAD_SAFETY) + +if(HAVE_CRC32C) + target_link_libraries(leveldb crc32c) +endif(HAVE_CRC32C) +if(HAVE_SNAPPY) + target_link_libraries(leveldb snappy) +endif(HAVE_SNAPPY) +if(HAVE_TCMALLOC) + target_link_libraries(leveldb tcmalloc) +endif(HAVE_TCMALLOC) + +# Needed by port_stdcxx.h +find_package(Threads REQUIRED) +target_link_libraries(leveldb Threads::Threads) + +add_executable(leveldbutil + "${PROJECT_SOURCE_DIR}/db/leveldbutil.cc" +) +target_link_libraries(leveldbutil leveldb) + +if(LEVELDB_BUILD_TESTS) + enable_testing() + + function(leveldb_test test_file) + get_filename_component(test_target_name "${test_file}" NAME_WE) + + add_executable("${test_target_name}" "") + target_sources("${test_target_name}" + PRIVATE + "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" + "${PROJECT_SOURCE_DIR}/util/testharness.cc" + "${PROJECT_SOURCE_DIR}/util/testharness.h" + "${PROJECT_SOURCE_DIR}/util/testutil.cc" + "${PROJECT_SOURCE_DIR}/util/testutil.h" + + "${test_file}" + ) + target_link_libraries("${test_target_name}" leveldb) + target_compile_definitions("${test_target_name}" + PRIVATE + ${LEVELDB_PLATFORM_NAME}=1 + ) + if (NOT HAVE_CXX17_HAS_INCLUDE) + target_compile_definitions("${test_target_name}" + PRIVATE + LEVELDB_HAS_PORT_CONFIG_H=1 + ) + endif(NOT HAVE_CXX17_HAS_INCLUDE) + + add_test(NAME "${test_target_name}" COMMAND "${test_target_name}") + endfunction(leveldb_test) + + leveldb_test("${PROJECT_SOURCE_DIR}/db/c_test.c") + leveldb_test("${PROJECT_SOURCE_DIR}/db/fault_injection_test.cc") + + leveldb_test("${PROJECT_SOURCE_DIR}/issues/issue178_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/issues/issue200_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/issues/issue320_test.cc") + + leveldb_test("${PROJECT_SOURCE_DIR}/util/env_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/status_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/no_destructor_test.cc") + + if(NOT BUILD_SHARED_LIBS) + leveldb_test("${PROJECT_SOURCE_DIR}/db/autocompact_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/corruption_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/db_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/dbformat_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/filename_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/log_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/recovery_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/skiplist_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/version_edit_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/version_set_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/db/write_batch_test.cc") + + leveldb_test("${PROJECT_SOURCE_DIR}/helpers/memenv/memenv_test.cc") + + leveldb_test("${PROJECT_SOURCE_DIR}/table/filter_block_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/table/table_test.cc") + + leveldb_test("${PROJECT_SOURCE_DIR}/util/arena_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/bloom_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/cache_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/coding_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/crc32c_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/hash_test.cc") + leveldb_test("${PROJECT_SOURCE_DIR}/util/logging_test.cc") + + # TODO(costan): This test also uses + # "${PROJECT_SOURCE_DIR}/util/env_{posix|windows}_test_helper.h" + if (WIN32) + leveldb_test("${PROJECT_SOURCE_DIR}/util/env_windows_test.cc") + else (WIN32) + leveldb_test("${PROJECT_SOURCE_DIR}/util/env_posix_test.cc") + endif (WIN32) + endif(NOT BUILD_SHARED_LIBS) +endif(LEVELDB_BUILD_TESTS) + +if(LEVELDB_BUILD_BENCHMARKS) + function(leveldb_benchmark bench_file) + get_filename_component(bench_target_name "${bench_file}" NAME_WE) + + add_executable("${bench_target_name}" "") + target_sources("${bench_target_name}" + PRIVATE + "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" + "${PROJECT_SOURCE_DIR}/util/histogram.cc" + "${PROJECT_SOURCE_DIR}/util/histogram.h" + "${PROJECT_SOURCE_DIR}/util/testharness.cc" + "${PROJECT_SOURCE_DIR}/util/testharness.h" + "${PROJECT_SOURCE_DIR}/util/testutil.cc" + "${PROJECT_SOURCE_DIR}/util/testutil.h" + + "${bench_file}" + ) + target_link_libraries("${bench_target_name}" leveldb) + target_compile_definitions("${bench_target_name}" + PRIVATE + ${LEVELDB_PLATFORM_NAME}=1 + ) + if (NOT HAVE_CXX17_HAS_INCLUDE) + target_compile_definitions("${bench_target_name}" + PRIVATE + LEVELDB_HAS_PORT_CONFIG_H=1 + ) + endif(NOT HAVE_CXX17_HAS_INCLUDE) + endfunction(leveldb_benchmark) + + if(NOT BUILD_SHARED_LIBS) + leveldb_benchmark("${PROJECT_SOURCE_DIR}/benchmarks/db_bench.cc") + endif(NOT BUILD_SHARED_LIBS) + + check_library_exists(sqlite3 sqlite3_open "" HAVE_SQLITE3) + if(HAVE_SQLITE3) + leveldb_benchmark("${PROJECT_SOURCE_DIR}/benchmarks/db_bench_sqlite3.cc") + target_link_libraries(db_bench_sqlite3 sqlite3) + endif(HAVE_SQLITE3) + + # check_library_exists is insufficient here because the library names have + # different manglings when compiled with clang or gcc, at least when installed + # with Homebrew on Mac. + set(OLD_CMAKE_REQURED_LIBRARIES ${CMAKE_REQUIRED_LIBRARIES}) + list(APPEND CMAKE_REQUIRED_LIBRARIES kyotocabinet) + check_cxx_source_compiles(" +#include + +int main() { + kyotocabinet::TreeDB* db = new kyotocabinet::TreeDB(); + delete db; + return 0; +} + " HAVE_KYOTOCABINET) + set(CMAKE_REQUIRED_LIBRARIES ${OLD_CMAKE_REQURED_LIBRARIES}) + if(HAVE_KYOTOCABINET) + leveldb_benchmark("${PROJECT_SOURCE_DIR}/benchmarks/db_bench_tree_db.cc") + target_link_libraries(db_bench_tree_db kyotocabinet) + endif(HAVE_KYOTOCABINET) +endif(LEVELDB_BUILD_BENCHMARKS) + +if(LEVELDB_INSTALL) + include(GNUInstallDirs) + install(TARGETS leveldb + EXPORT leveldbTargets + RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR} + LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR} + ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR} + ) + install( + FILES + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/c.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/cache.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/comparator.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/db.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/dumpfile.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/env.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/export.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/filter_policy.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/iterator.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/options.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/slice.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/status.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/table_builder.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/table.h" + "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/write_batch.h" + DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/leveldb + ) + + include(CMakePackageConfigHelpers) + write_basic_package_version_file( + "${PROJECT_BINARY_DIR}/leveldbConfigVersion.cmake" + COMPATIBILITY SameMajorVersion + ) + install( + EXPORT leveldbTargets + NAMESPACE leveldb:: + DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/leveldb" + ) + install( + FILES + "${PROJECT_SOURCE_DIR}/cmake/leveldbConfig.cmake" + "${PROJECT_BINARY_DIR}/leveldbConfigVersion.cmake" + DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/leveldb" + ) +endif(LEVELDB_INSTALL) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index cd600ff..a74572a 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -31,6 +31,6 @@ the CLA. ## Writing Code ## -If your contribution contains code, please make sure that it follows -[the style guide](http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml). +If your contribution contains code, please make sure that it follows +[the style guide](http://google.github.io/styleguide/cppguide.html). Otherwise we will have to ask you to make changes, and that's no fun for anyone. diff --git a/Makefile b/Makefile deleted file mode 100644 index 24f214a..0000000 --- a/Makefile +++ /dev/null @@ -1,231 +0,0 @@ -# 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. - -#----------------------------------------------- -# Uncomment exactly one of the lines labelled (A), (B), and (C) below -# to switch between compilation modes. - -# (A) Production use (optimized mode) -OPT ?= -O2 -DNDEBUG -# (B) Debug mode, w/ full line-level debugging symbols -# OPT ?= -g2 -# (C) Profiling mode: opt, but w/debugging symbols -# OPT ?= -O2 -g2 -DNDEBUG -#----------------------------------------------- - -# detect what platform we're building on -$(shell CC="$(CC)" CXX="$(CXX)" TARGET_OS="$(TARGET_OS)" \ - ./build_detect_platform build_config.mk ./) -# this file is generated by the previous line to set build flags and sources -include build_config.mk - -CFLAGS += -I. -I./include $(PLATFORM_CCFLAGS) $(OPT) -CXXFLAGS += -I. -I./include $(PLATFORM_CXXFLAGS) $(OPT) - -LDFLAGS += $(PLATFORM_LDFLAGS) -LIBS += $(PLATFORM_LIBS) - -LIBOBJECTS = $(SOURCES:.cc=.o) -MEMENVOBJECTS = $(MEMENV_SOURCES:.cc=.o) - -TESTUTIL = ./util/testutil.o -TESTHARNESS = ./util/testharness.o $(TESTUTIL) - -# Note: iOS should probably be using libtool, not ar. -ifeq ($(PLATFORM), IOS) -AR=xcrun ar -endif - -TESTS = \ - arena_test \ - autocompact_test \ - bloom_test \ - c_test \ - cache_test \ - coding_test \ - corruption_test \ - crc32c_test \ - db_test \ - dbformat_test \ - env_test \ - fault_injection_test \ - filename_test \ - filter_block_test \ - hash_test \ - issue178_test \ - issue200_test \ - log_test \ - memenv_test \ - skiplist_test \ - table_test \ - version_edit_test \ - version_set_test \ - write_batch_test - -PROGRAMS = db_bench leveldbutil $(TESTS) -BENCHMARKS = db_bench_sqlite3 db_bench_tree_db - -LIBRARY = libleveldb.a -MEMENVLIBRARY = libmemenv.a - -default: all - -# Should we build shared libraries? -ifneq ($(PLATFORM_SHARED_EXT),) - -ifneq ($(PLATFORM_SHARED_VERSIONED),true) -SHARED1 = libleveldb.$(PLATFORM_SHARED_EXT) -SHARED2 = $(SHARED1) -SHARED3 = $(SHARED1) -SHARED = $(SHARED1) -else -# Update db.h if you change these. -SHARED_MAJOR = 1 -SHARED_MINOR = 18 -SHARED1 = libleveldb.$(PLATFORM_SHARED_EXT) -SHARED2 = $(SHARED1).$(SHARED_MAJOR) -SHARED3 = $(SHARED1).$(SHARED_MAJOR).$(SHARED_MINOR) -SHARED = $(SHARED1) $(SHARED2) $(SHARED3) -$(SHARED1): $(SHARED3) - ln -fs $(SHARED3) $(SHARED1) -$(SHARED2): $(SHARED3) - ln -fs $(SHARED3) $(SHARED2) -endif - -$(SHARED3): - $(CXX) $(LDFLAGS) $(PLATFORM_SHARED_LDFLAGS)$(SHARED2) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(SOURCES) -o $(SHARED3) $(LIBS) - -endif # PLATFORM_SHARED_EXT - -all: $(SHARED) $(LIBRARY) - -check: all $(PROGRAMS) $(TESTS) - for t in $(TESTS); do echo "***** Running $$t"; ./$$t || exit 1; done - -clean: - -rm -f $(PROGRAMS) $(BENCHMARKS) $(LIBRARY) $(SHARED) $(MEMENVLIBRARY) */*.o */*/*.o ios-x86/*/*.o ios-arm/*/*.o build_config.mk - -rm -rf ios-x86/* ios-arm/* - -$(LIBRARY): $(LIBOBJECTS) - rm -f $@ - $(AR) -rs $@ $(LIBOBJECTS) - -db_bench: db/db_bench.o $(LIBOBJECTS) $(TESTUTIL) - $(CXX) $(LDFLAGS) db/db_bench.o $(LIBOBJECTS) $(TESTUTIL) -o $@ $(LIBS) - -db_bench_sqlite3: doc/bench/db_bench_sqlite3.o $(LIBOBJECTS) $(TESTUTIL) - $(CXX) $(LDFLAGS) doc/bench/db_bench_sqlite3.o $(LIBOBJECTS) $(TESTUTIL) -o $@ -lsqlite3 $(LIBS) - -db_bench_tree_db: doc/bench/db_bench_tree_db.o $(LIBOBJECTS) $(TESTUTIL) - $(CXX) $(LDFLAGS) doc/bench/db_bench_tree_db.o $(LIBOBJECTS) $(TESTUTIL) -o $@ -lkyotocabinet $(LIBS) - -leveldbutil: db/leveldb_main.o $(LIBOBJECTS) - $(CXX) $(LDFLAGS) db/leveldb_main.o $(LIBOBJECTS) -o $@ $(LIBS) - -arena_test: util/arena_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/arena_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -autocompact_test: db/autocompact_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/autocompact_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -bloom_test: util/bloom_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/bloom_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -c_test: db/c_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/c_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -cache_test: util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -corruption_test: db/corruption_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/corruption_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -crc32c_test: util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -db_test: db/db_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/db_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -dbformat_test: db/dbformat_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/dbformat_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -env_test: util/env_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/env_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -fault_injection_test: db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -filename_test: db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -filter_block_test: table/filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) table/filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -hash_test: util/hash_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) util/hash_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -issue178_test: issues/issue178_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) issues/issue178_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -issue200_test: issues/issue200_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) issues/issue200_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -log_test: db/log_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/log_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -table_test: table/table_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) table/table_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -skiplist_test: db/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -version_edit_test: db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -version_set_test: db/version_set_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/version_set_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -write_batch_test: db/write_batch_test.o $(LIBOBJECTS) $(TESTHARNESS) - $(CXX) $(LDFLAGS) db/write_batch_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS) - -$(MEMENVLIBRARY) : $(MEMENVOBJECTS) - rm -f $@ - $(AR) -rs $@ $(MEMENVOBJECTS) - -memenv_test : helpers/memenv/memenv_test.o $(MEMENVLIBRARY) $(LIBRARY) $(TESTHARNESS) - $(CXX) $(LDFLAGS) helpers/memenv/memenv_test.o $(MEMENVLIBRARY) $(LIBRARY) $(TESTHARNESS) -o $@ $(LIBS) - -ifeq ($(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 $(PLATFORMSROOT)/iPhoneOS.platform/version CFBundleShortVersionString) -IOSARCH=-arch armv6 -arch armv7 -arch armv7s -arch arm64 - -.cc.o: - mkdir -p ios-x86/$(dir $@) - xcrun -sdk iphonesimulator $(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 $(IOSARCH) -c $< -o ios-arm/$@ - xcrun lipo ios-x86/$@ ios-arm/$@ -create -output $@ - -.c.o: - mkdir -p ios-x86/$(dir $@) - xcrun -sdk iphonesimulator $(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 $(IOSARCH) -c $< -o ios-arm/$@ - xcrun lipo ios-x86/$@ ios-arm/$@ -create -output $@ - -else -.cc.o: - $(CXX) $(CXXFLAGS) -c $< -o $@ - -.c.o: - $(CC) $(CFLAGS) -c $< -o $@ -endif diff --git a/README.md b/README.md index 480affb..0b660ae 100644 --- a/README.md +++ b/README.md @@ -1,8 +1,12 @@ **LevelDB is a fast key-value storage library written at Google that provides an ordered mapping from string keys to string values.** +[![Build Status](https://travis-ci.org/google/leveldb.svg?branch=master)](https://travis-ci.org/google/leveldb) +[![Build status](https://ci.appveyor.com/api/projects/status/g2j5j4rfkda6eyw5/branch/master?svg=true)](https://ci.appveyor.com/project/pwnall/leveldb) + Authors: Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com) # Features + * Keys and values are arbitrary byte arrays. * Data is stored sorted by key. * Callers can provide a custom comparison function to override the sort order. @@ -10,16 +14,98 @@ Authors: Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com) * Multiple changes can be made in one atomic batch. * Users can create a transient snapshot to get a consistent view of data. * Forward and backward iteration is supported over the data. - * Data is automatically compressed using the [Snappy compression library](http://code.google.com/p/snappy). + * Data is automatically compressed using the [Snappy compression library](http://google.github.io/snappy/). * External activity (file system operations etc.) is relayed through a virtual interface so users can customize the operating system interactions. - * [Detailed documentation](http://htmlpreview.github.io/?https://github.com/google/leveldb/blob/master/doc/index.html) about how to use the library is included with the source code. +# Documentation + + [LevelDB library documentation](https://github.com/google/leveldb/blob/master/doc/index.md) is online and bundled with the source code. # Limitations + * This is not a SQL database. It does not have a relational data model, it does not support SQL queries, and it has no support for indexes. * Only a single process (possibly multi-threaded) can access a particular database at a time. * There is no client-server support builtin to the library. An application that needs such support will have to wrap their own server around the library. +# Building + +This project supports [CMake](https://cmake.org/) out of the box. + +### Build for POSIX + +Quick start: + +```bash +mkdir -p build && cd build +cmake -DCMAKE_BUILD_TYPE=Release .. && cmake --build . +``` + +### Building for Windows + +First generate the Visual Studio 2017 project/solution files: + +```cmd +mkdir build +cd build +cmake -G "Visual Studio 15" .. +``` +The default default will build for x86. For 64-bit run: + +```cmd +cmake -G "Visual Studio 15 Win64" .. +``` + +To compile the Windows solution from the command-line: + +```cmd +devenv /build Debug leveldb.sln +``` + +or open leveldb.sln in Visual Studio and build from within. + +Please see the CMake documentation and `CMakeLists.txt` for more advanced usage. + +# Contributing to the leveldb Project + +The leveldb project welcomes contributions. leveldb's primary goal is to be +a reliable and fast key/value store. Changes that are in line with the +features/limitations outlined above, and meet the requirements below, +will be considered. + +Contribution requirements: + +1. **Tested platforms only**. We _generally_ will only accept changes for + platforms that are compiled and tested. This means POSIX (for Linux and + macOS) or Windows. Very small changes will sometimes be accepted, but + consider that more of an exception than the rule. + +2. **Stable API**. We strive very hard to maintain a stable API. Changes that + require changes for projects using leveldb _might_ be rejected without + sufficient benefit to the project. + +3. **Tests**: All changes must be accompanied by a new (or changed) test, or + a sufficient explanation as to why a new (or changed) test is not required. + +4. **Consistent Style**: This project conforms to the + [Google C++ Style Guide](https://google.github.io/styleguide/cppguide.html). + To ensure your changes are properly formatted please run: + + ``` + clang-format -i --style=file + ``` + +## Submitting a Pull Request + +Before any pull request will be accepted the author must first sign a +Contributor License Agreement (CLA) at https://cla.developers.google.com/. + +In order to keep the commit timeline linear +[squash](https://git-scm.com/book/en/v2/Git-Tools-Rewriting-History#Squashing-Commits) +your changes down to a single commit and [rebase](https://git-scm.com/docs/git-rebase) +on google/leveldb/master. This keeps the commit timeline linear and more easily sync'ed +with the internal repository at Google. More information at GitHub's +[About Git rebase](https://help.github.com/articles/about-git-rebase/) page. + # Performance Here is a performance report (with explanations) from the run of the @@ -78,29 +164,30 @@ by the one or two disk seeks needed to fetch the data from disk. Write performance will be mostly unaffected by whether or not the working set fits in memory. - readrandom : 16.677 micros/op; (approximately 60,000 reads per second) - readseq : 0.476 micros/op; 232.3 MB/s - readreverse : 0.724 micros/op; 152.9 MB/s + readrandom : 16.677 micros/op; (approximately 60,000 reads per second) + readseq : 0.476 micros/op; 232.3 MB/s + readreverse : 0.724 micros/op; 152.9 MB/s LevelDB compacts its underlying storage data in the background to improve read performance. The results listed above were done immediately after a lot of random writes. The results after compactions (which are usually triggered automatically) are better. - readrandom : 11.602 micros/op; (approximately 85,000 reads per second) - readseq : 0.423 micros/op; 261.8 MB/s - readreverse : 0.663 micros/op; 166.9 MB/s + readrandom : 11.602 micros/op; (approximately 85,000 reads per second) + readseq : 0.423 micros/op; 261.8 MB/s + readreverse : 0.663 micros/op; 166.9 MB/s Some of the high cost of reads comes from repeated decompression of blocks read from disk. If we supply enough cache to the leveldb so it can hold the uncompressed blocks in memory, the read performance improves again: - readrandom : 9.775 micros/op; (approximately 100,000 reads per second before compaction) - readrandom : 5.215 micros/op; (approximately 190,000 reads per second after compaction) + readrandom : 9.775 micros/op; (approximately 100,000 reads per second before compaction) + readrandom : 5.215 micros/op; (approximately 190,000 reads per second after compaction) ## Repository contents -See doc/index.html for more explanation. See doc/impl.html for a brief overview of the implementation. +See [doc/index.md](doc/index.md) for more explanation. See +[doc/impl.md](doc/impl.md) for a brief overview of the implementation. The public interface is in include/*.h. Callers should not include or rely on the details of any other header files in this package. Those @@ -113,7 +200,7 @@ Guide to header files: * **include/options.h**: Control over the behavior of an entire database, and also control over the behavior of individual reads and writes. -* **include/comparator.h**: Abstraction for user-specified comparison function. +* **include/comparator.h**: Abstraction for user-specified comparison function. If you want just bytewise comparison of keys, you can use the default comparator, but clients can write their own comparator implementations if they want custom ordering (e.g. to handle different character encodings, etc.) @@ -130,7 +217,7 @@ length into some other byte array. * **include/status.h**: Status is returned from many of the public interfaces and is used to report success and various kinds of errors. -* **include/env.h**: +* **include/env.h**: Abstraction of the OS environment. A posix implementation of this interface is in util/env_posix.cc diff --git a/db/db_bench.cc b/benchmarks/db_bench.cc similarity index 83% rename from db/db_bench.cc rename to benchmarks/db_bench.cc index e5975bd..3696023 100644 --- a/db/db_bench.cc +++ b/benchmarks/db_bench.cc @@ -2,14 +2,14 @@ // 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_impl.h" -#include "db/version_set.h" +#include + #include "leveldb/cache.h" #include "leveldb/db.h" #include "leveldb/env.h" +#include "leveldb/filter_policy.h" #include "leveldb/write_batch.h" #include "port/port.h" #include "util/crc32c.h" @@ -35,7 +35,6 @@ // seekrandom -- N random seeks // open -- cost of opening a DB // crc32c -- repeated crc32c of 4K of data -// acquireload -- load N*1000 times // Meta operations: // compact -- Compact the entire DB // stats -- Print DB stats @@ -57,9 +56,7 @@ static const char* FLAGS_benchmarks = "fill100K," "crc32c," "snappycomp," - "snappyuncomp," - "acquireload," - ; + "snappyuncomp,"; // Number of key/values to place in database static int FLAGS_num = 1000000; @@ -84,6 +81,14 @@ static bool FLAGS_histogram = false; // (initialized to default value by "main") static int FLAGS_write_buffer_size = 0; +// Number of bytes written to each file. +// (initialized to default value by "main") +static int FLAGS_max_file_size = 0; + +// Approximate size of user data packed per block (before compression. +// (initialized to default value by "main") +static int FLAGS_block_size = 0; + // Number of bytes to use as a cache of uncompressed data. // Negative means use default settings. static int FLAGS_cache_size = -1; @@ -100,12 +105,16 @@ static int FLAGS_bloom_bits = -1; // benchmark will fail. static bool FLAGS_use_existing_db = false; +// If true, reuse existing log/MANIFEST files when re-opening a database. +static bool FLAGS_reuse_logs = false; + // Use the db with the following name. -static const char* FLAGS_db = NULL; +static const char* FLAGS_db = nullptr; namespace leveldb { namespace { +leveldb::Env* g_env = nullptr; // Helper for quickly generating random data. class RandomGenerator { @@ -139,17 +148,19 @@ class RandomGenerator { } }; +#if defined(__linux) static Slice TrimSpace(Slice s) { size_t start = 0; while (start < s.size() && isspace(s[start])) { start++; } size_t limit = s.size(); - while (limit > start && isspace(s[limit-1])) { + while (limit > start && isspace(s[limit - 1])) { limit--; } return Slice(s.data() + start, limit - start); } +#endif static void AppendWithSpace(std::string* str, Slice msg) { if (msg.empty()) return; @@ -176,14 +187,12 @@ class Stats { void Start() { next_report_ = 100; - last_op_finish_ = start_; hist_.Clear(); done_ = 0; bytes_ = 0; seconds_ = 0; - start_ = Env::Default()->NowMicros(); - finish_ = start_; message_.clear(); + start_ = finish_ = last_op_finish_ = g_env->NowMicros(); } void Merge(const Stats& other) { @@ -199,17 +208,15 @@ class Stats { } void Stop() { - finish_ = Env::Default()->NowMicros(); + finish_ = g_env->NowMicros(); seconds_ = (finish_ - start_) * 1e-6; } - void AddMessage(Slice msg) { - AppendWithSpace(&message_, msg); - } + void AddMessage(Slice msg) { AppendWithSpace(&message_, msg); } void FinishedSingleOp() { if (FLAGS_histogram) { - double now = Env::Default()->NowMicros(); + double now = g_env->NowMicros(); double micros = now - last_op_finish_; hist_.Add(micros); if (micros > 20000) { @@ -221,21 +228,26 @@ class Stats { done_++; if (done_ >= next_report_) { - 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; + 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 %d ops%30s\r", done_, ""); fflush(stderr); } } - void AddBytes(int64_t n) { - bytes_ += n; - } + 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 @@ -254,11 +266,8 @@ class Stats { } AppendWithSpace(&extra, message_); - fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", - name.ToString().c_str(), - seconds_ * 1e6 / done_, - (extra.empty() ? "" : " "), - extra.c_str()); + fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", name.ToString().c_str(), + seconds_ * 1e6 / done_, (extra.empty() ? "" : " "), extra.c_str()); if (FLAGS_histogram) { fprintf(stdout, "Microseconds per op:\n%s\n", hist_.ToString().c_str()); } @@ -269,8 +278,8 @@ class Stats { // State shared by all concurrent executions of the same benchmark. struct SharedState { port::Mutex mu; - port::CondVar cv; - int total; + port::CondVar cv GUARDED_BY(mu); + int total GUARDED_BY(mu); // Each thread goes through the following states: // (1) initializing @@ -278,24 +287,22 @@ struct SharedState { // (3) running // (4) done - int num_initialized; - int num_done; - bool start; + int num_initialized GUARDED_BY(mu); + int num_done GUARDED_BY(mu); + bool start GUARDED_BY(mu); - SharedState() : cv(&mu) { } + SharedState(int total) + : cv(&mu), total(total), num_initialized(0), num_done(0), start(false) {} }; // Per-thread state for concurrent executions of the same benchmark. struct ThreadState { - int tid; // 0..n-1 when running in n threads - Random rand; // Has different seeds for different threads + int tid; // 0..n-1 when running in n threads + Random rand; // Has different seeds for different threads Stats stats; SharedState* shared; - ThreadState(int index) - : tid(index), - rand(1000 + index) { - } + ThreadState(int index) : tid(index), rand(1000 + index), shared(nullptr) {} }; } // namespace @@ -321,20 +328,20 @@ class Benchmark { static_cast(FLAGS_value_size * FLAGS_compression_ratio + 0.5)); fprintf(stdout, "Entries: %d\n", num_); fprintf(stdout, "RawSize: %.1f MB (estimated)\n", - ((static_cast(kKeySize + FLAGS_value_size) * num_) - / 1048576.0)); + ((static_cast(kKeySize + FLAGS_value_size) * num_) / + 1048576.0)); fprintf(stdout, "FileSize: %.1f MB (estimated)\n", - (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) - / 1048576.0)); + (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) / + 1048576.0)); PrintWarnings(); fprintf(stdout, "------------------------------------------------\n"); } void PrintWarnings() { #if defined(__GNUC__) && !defined(__OPTIMIZE__) - fprintf(stdout, - "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n" - ); + fprintf( + stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); #endif #ifndef NDEBUG fprintf(stdout, @@ -352,22 +359,22 @@ class Benchmark { } void PrintEnvironment() { - fprintf(stderr, "LevelDB: version %d.%d\n", - kMajorVersion, kMinorVersion); + fprintf(stderr, "LevelDB: version %d.%d\n", kMajorVersion, + kMinorVersion); #if defined(__linux) - time_t now = time(NULL); + time_t now = time(nullptr); fprintf(stderr, "Date: %s", ctime(&now)); // ctime() adds newline FILE* cpuinfo = fopen("/proc/cpuinfo", "r"); - if (cpuinfo != NULL) { + if (cpuinfo != nullptr) { char line[1000]; int num_cpus = 0; std::string cpu_type; std::string cache_size; - while (fgets(line, sizeof(line), cpuinfo) != NULL) { + while (fgets(line, sizeof(line), cpuinfo) != nullptr) { const char* sep = strchr(line, ':'); - if (sep == NULL) { + if (sep == nullptr) { continue; } Slice key = TrimSpace(Slice(line, sep - 1 - line)); @@ -388,21 +395,21 @@ class Benchmark { public: Benchmark() - : cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : NULL), - filter_policy_(FLAGS_bloom_bits >= 0 - ? NewBloomFilterPolicy(FLAGS_bloom_bits) - : NULL), - db_(NULL), - num_(FLAGS_num), - value_size_(FLAGS_value_size), - entries_per_batch_(1), - reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), - heap_counter_(0) { + : cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : nullptr), + filter_policy_(FLAGS_bloom_bits >= 0 + ? NewBloomFilterPolicy(FLAGS_bloom_bits) + : nullptr), + db_(nullptr), + num_(FLAGS_num), + value_size_(FLAGS_value_size), + entries_per_batch_(1), + reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), + heap_counter_(0) { std::vector files; - Env::Default()->GetChildren(FLAGS_db, &files); + g_env->GetChildren(FLAGS_db, &files); for (size_t i = 0; i < files.size(); i++) { if (Slice(files[i]).starts_with("heap-")) { - Env::Default()->DeleteFile(std::string(FLAGS_db) + "/" + files[i]); + g_env->DeleteFile(std::string(FLAGS_db) + "/" + files[i]); } } if (!FLAGS_use_existing_db) { @@ -421,12 +428,12 @@ class Benchmark { Open(); const char* benchmarks = FLAGS_benchmarks; - while (benchmarks != NULL) { + while (benchmarks != nullptr) { const char* sep = strchr(benchmarks, ','); Slice name; - if (sep == NULL) { + if (sep == nullptr) { name = benchmarks; - benchmarks = NULL; + benchmarks = nullptr; } else { name = Slice(benchmarks, sep - benchmarks); benchmarks = sep + 1; @@ -439,7 +446,7 @@ class Benchmark { entries_per_batch_ = 1; write_options_ = WriteOptions(); - void (Benchmark::*method)(ThreadState*) = NULL; + void (Benchmark::*method)(ThreadState*) = nullptr; bool fresh_db = false; int num_threads = FLAGS_threads; @@ -496,8 +503,6 @@ class Benchmark { method = &Benchmark::Compact; } else if (name == Slice("crc32c")) { method = &Benchmark::Crc32c; - } else if (name == Slice("acquireload")) { - method = &Benchmark::AcquireLoad; } else if (name == Slice("snappycomp")) { method = &Benchmark::SnappyCompress; } else if (name == Slice("snappyuncomp")) { @@ -509,7 +514,7 @@ class Benchmark { } else if (name == Slice("sstables")) { PrintStats("leveldb.sstables"); } else { - if (name != Slice()) { // No error message for empty name + if (!name.empty()) { // No error message for empty name fprintf(stderr, "unknown benchmark '%s'\n", name.ToString().c_str()); } } @@ -518,16 +523,16 @@ class Benchmark { if (FLAGS_use_existing_db) { fprintf(stdout, "%-12s : skipped (--use_existing_db is true)\n", name.ToString().c_str()); - method = NULL; + method = nullptr; } else { delete db_; - db_ = NULL; + db_ = nullptr; DestroyDB(FLAGS_db, Options()); Open(); } } - if (method != NULL) { + if (method != nullptr) { RunBenchmark(num_threads, name, method); } } @@ -571,11 +576,7 @@ class Benchmark { 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; + SharedState shared(n); ThreadArg* arg = new ThreadArg[n]; for (int i = 0; i < n; i++) { @@ -584,7 +585,7 @@ class Benchmark { arg[i].shared = &shared; arg[i].thread = new ThreadState(i); arg[i].thread->shared = &shared; - Env::Default()->StartThread(ThreadBody, &arg[i]); + g_env->StartThread(ThreadBody, &arg[i]); } shared.mu.Lock(); @@ -629,22 +630,6 @@ class Benchmark { thread->stats.AddMessage(label); } - void AcquireLoad(ThreadState* thread) { - int dummy; - port::AtomicPointer ap(&dummy); - int count = 0; - void *ptr = NULL; - thread->stats.AddMessage("(each op is 1000 loads)"); - while (count < 100000) { - for (int i = 0; i < 1000; i++) { - ptr = ap.Acquire_Load(); - } - count++; - thread->stats.FinishedSingleOp(); - } - if (ptr == NULL) exit(1); // Disable unused variable warning. - } - void SnappyCompress(ThreadState* thread) { RandomGenerator gen; Slice input = gen.Generate(Options().block_size); @@ -678,8 +663,8 @@ class Benchmark { int64_t bytes = 0; char* uncompressed = new char[input.size()]; while (ok && bytes < 1024 * 1048576) { // Compress 1G - ok = port::Snappy_Uncompress(compressed.data(), compressed.size(), - uncompressed); + ok = port::Snappy_Uncompress(compressed.data(), compressed.size(), + uncompressed); bytes += input.size(); thread->stats.FinishedSingleOp(); } @@ -693,13 +678,17 @@ class Benchmark { } void Open() { - assert(db_ == NULL); + assert(db_ == nullptr); Options options; + options.env = g_env; options.create_if_missing = !FLAGS_use_existing_db; options.block_cache = cache_; options.write_buffer_size = FLAGS_write_buffer_size; + options.max_file_size = FLAGS_max_file_size; + options.block_size = FLAGS_block_size; options.max_open_files = FLAGS_open_files; options.filter_policy = filter_policy_; + options.reuse_logs = FLAGS_reuse_logs; Status s = DB::Open(options, FLAGS_db, &db_); if (!s.ok()) { fprintf(stderr, "open error: %s\n", s.ToString().c_str()); @@ -715,13 +704,9 @@ class Benchmark { } } - void WriteSeq(ThreadState* thread) { - DoWrite(thread, true); - } + void WriteSeq(ThreadState* thread) { DoWrite(thread, true); } - void WriteRandom(ThreadState* thread) { - DoWrite(thread, false); - } + void WriteRandom(ThreadState* thread) { DoWrite(thread, false); } void DoWrite(ThreadState* thread, bool seq) { if (num_ != FLAGS_num) { @@ -737,7 +722,7 @@ class Benchmark { for (int i = 0; i < num_; i += entries_per_batch_) { batch.Clear(); for (int j = 0; j < entries_per_batch_; j++) { - const int k = seq ? i+j : (thread->rand.Next() % FLAGS_num); + const int k = seq ? i + j : (thread->rand.Next() % FLAGS_num); char key[100]; snprintf(key, sizeof(key), "%016d", k); batch.Put(key, gen.Generate(value_size_)); @@ -847,7 +832,7 @@ class Benchmark { for (int i = 0; i < num_; i += entries_per_batch_) { batch.Clear(); for (int j = 0; j < entries_per_batch_; j++) { - const int k = seq ? i+j : (thread->rand.Next() % FLAGS_num); + const int k = seq ? i + j : (thread->rand.Next() % FLAGS_num); char key[100]; snprintf(key, sizeof(key), "%016d", k); batch.Delete(key); @@ -861,13 +846,9 @@ class Benchmark { } } - void DeleteSeq(ThreadState* thread) { - DoDelete(thread, true); - } + void DeleteSeq(ThreadState* thread) { DoDelete(thread, true); } - void DeleteRandom(ThreadState* thread) { - DoDelete(thread, false); - } + void DeleteRandom(ThreadState* thread) { DoDelete(thread, false); } void ReadWhileWriting(ThreadState* thread) { if (thread->tid > 0) { @@ -899,9 +880,7 @@ class Benchmark { } } - void Compact(ThreadState* thread) { - db_->CompactRange(NULL, NULL); - } + void Compact(ThreadState* thread) { db_->CompactRange(nullptr, nullptr); } void PrintStats(const char* key) { std::string stats; @@ -919,7 +898,7 @@ class Benchmark { char fname[100]; snprintf(fname, sizeof(fname), "%s/heap-%04d", FLAGS_db, ++heap_counter_); WritableFile* file; - Status s = Env::Default()->NewWritableFile(fname, &file); + Status s = g_env->NewWritableFile(fname, &file); if (!s.ok()) { fprintf(stderr, "%s\n", s.ToString().c_str()); return; @@ -928,7 +907,7 @@ class Benchmark { delete file; if (!ok) { fprintf(stderr, "heap profiling not supported\n"); - Env::Default()->DeleteFile(fname); + g_env->DeleteFile(fname); } } }; @@ -937,6 +916,8 @@ class Benchmark { int main(int argc, char** argv) { FLAGS_write_buffer_size = leveldb::Options().write_buffer_size; + FLAGS_max_file_size = leveldb::Options().max_file_size; + FLAGS_block_size = leveldb::Options().block_size; FLAGS_open_files = leveldb::Options().max_open_files; std::string default_db_path; @@ -954,6 +935,9 @@ int main(int argc, char** argv) { } else if (sscanf(argv[i], "--use_existing_db=%d%c", &n, &junk) == 1 && (n == 0 || n == 1)) { FLAGS_use_existing_db = n; + } else if (sscanf(argv[i], "--reuse_logs=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_reuse_logs = n; } else if (sscanf(argv[i], "--num=%d%c", &n, &junk) == 1) { FLAGS_num = n; } else if (sscanf(argv[i], "--reads=%d%c", &n, &junk) == 1) { @@ -964,6 +948,10 @@ int main(int argc, char** argv) { FLAGS_value_size = n; } else if (sscanf(argv[i], "--write_buffer_size=%d%c", &n, &junk) == 1) { FLAGS_write_buffer_size = n; + } else if (sscanf(argv[i], "--max_file_size=%d%c", &n, &junk) == 1) { + FLAGS_max_file_size = n; + } else if (sscanf(argv[i], "--block_size=%d%c", &n, &junk) == 1) { + FLAGS_block_size = n; } else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) { FLAGS_cache_size = n; } else if (sscanf(argv[i], "--bloom_bits=%d%c", &n, &junk) == 1) { @@ -978,11 +966,13 @@ int main(int argc, char** argv) { } } + leveldb::g_env = leveldb::Env::Default(); + // Choose a location for the test database if none given with --db= - if (FLAGS_db == NULL) { - leveldb::Env::Default()->GetTestDirectory(&default_db_path); - default_db_path += "/dbbench"; - FLAGS_db = default_db_path.c_str(); + if (FLAGS_db == nullptr) { + leveldb::g_env->GetTestDirectory(&default_db_path); + default_db_path += "/dbbench"; + FLAGS_db = default_db_path.c_str(); } leveldb::Benchmark benchmark; diff --git a/doc/bench/db_bench_sqlite3.cc b/benchmarks/db_bench_sqlite3.cc similarity index 83% rename from doc/bench/db_bench_sqlite3.cc rename to benchmarks/db_bench_sqlite3.cc index e63aaa8..f183f4f 100644 --- a/doc/bench/db_bench_sqlite3.cc +++ b/benchmarks/db_bench_sqlite3.cc @@ -2,9 +2,10 @@ // 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 "util/histogram.h" #include "util/random.h" #include "util/testutil.h" @@ -38,8 +39,7 @@ static const char* FLAGS_benchmarks = "fillrand100K," "fillseq100K," "readseq," - "readrand100K," - ; + "readrand100K,"; // Number of key/values to place in database static int FLAGS_num = 1000000; @@ -76,10 +76,9 @@ static bool FLAGS_transaction = true; static bool FLAGS_WAL_enabled = true; // Use the db with the following name. -static const char* FLAGS_db = NULL; +static const char* FLAGS_db = nullptr; -inline -static void ExecErrorCheck(int status, char *err_msg) { +inline static void ExecErrorCheck(int status, char* err_msg) { if (status != SQLITE_OK) { fprintf(stderr, "SQL error: %s\n", err_msg); sqlite3_free(err_msg); @@ -87,27 +86,25 @@ static void ExecErrorCheck(int status, char *err_msg) { } } -inline -static void StepErrorCheck(int status) { +inline static void StepErrorCheck(int status) { if (status != SQLITE_DONE) { fprintf(stderr, "SQL step error: status = %d\n", status); exit(1); } } -inline -static void ErrorCheck(int status) { +inline static void ErrorCheck(int status) { if (status != SQLITE_OK) { fprintf(stderr, "sqlite3 error: status = %d\n", status); exit(1); } } -inline -static void WalCheckpoint(sqlite3* db_) { +inline static void WalCheckpoint(sqlite3* db_) { // Flush all writes to disk if (FLAGS_WAL_enabled) { - sqlite3_wal_checkpoint_v2(db_, NULL, SQLITE_CHECKPOINT_FULL, NULL, NULL); + sqlite3_wal_checkpoint_v2(db_, nullptr, SQLITE_CHECKPOINT_FULL, nullptr, + nullptr); } } @@ -152,7 +149,7 @@ static Slice TrimSpace(Slice s) { start++; } int limit = s.size(); - while (limit > start && isspace(s[limit-1])) { + while (limit > start && isspace(s[limit - 1])) { limit--; } return Slice(s.data() + start, limit - start); @@ -176,7 +173,7 @@ class Benchmark { // State kept for progress messages int done_; - int next_report_; // When to report next + int next_report_; // When to report next void PrintHeader() { const int kKeySize = 16; @@ -185,17 +182,17 @@ class Benchmark { fprintf(stdout, "Values: %d bytes each\n", FLAGS_value_size); fprintf(stdout, "Entries: %d\n", num_); fprintf(stdout, "RawSize: %.1f MB (estimated)\n", - ((static_cast(kKeySize + FLAGS_value_size) * num_) - / 1048576.0)); + ((static_cast(kKeySize + FLAGS_value_size) * num_) / + 1048576.0)); PrintWarnings(); fprintf(stdout, "------------------------------------------------\n"); } void PrintWarnings() { #if defined(__GNUC__) && !defined(__OPTIMIZE__) - fprintf(stdout, - "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n" - ); + fprintf( + stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); #endif #ifndef NDEBUG fprintf(stdout, @@ -207,18 +204,18 @@ class Benchmark { fprintf(stderr, "SQLite: version %s\n", SQLITE_VERSION); #if defined(__linux) - time_t now = time(NULL); + time_t now = time(nullptr); fprintf(stderr, "Date: %s", ctime(&now)); // ctime() adds newline FILE* cpuinfo = fopen("/proc/cpuinfo", "r"); - if (cpuinfo != NULL) { + if (cpuinfo != nullptr) { char line[1000]; int num_cpus = 0; std::string cpu_type; std::string cache_size; - while (fgets(line, sizeof(line), cpuinfo) != NULL) { + while (fgets(line, sizeof(line), cpuinfo) != nullptr) { const char* sep = strchr(line, ':'); - if (sep == NULL) { + if (sep == nullptr) { continue; } Slice key = TrimSpace(Slice(line, sep - 1 - line)); @@ -261,13 +258,20 @@ class Benchmark { done_++; if (done_ >= next_report_) { - 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; + 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 %d ops%30s\r", done_, ""); fflush(stderr); } @@ -285,16 +289,14 @@ class Benchmark { snprintf(rate, sizeof(rate), "%6.1f MB/s", (bytes_ / 1048576.0) / (finish - start_)); if (!message_.empty()) { - message_ = std::string(rate) + " " + message_; + message_ = std::string(rate) + " " + message_; } else { message_ = rate; } } - fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", - name.ToString().c_str(), - (finish - start_) * 1e6 / done_, - (message_.empty() ? "" : " "), + fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", name.ToString().c_str(), + (finish - start_) * 1e6 / done_, (message_.empty() ? "" : " "), message_.c_str()); if (FLAGS_histogram) { fprintf(stdout, "Microseconds per op:\n%s\n", hist_.ToString().c_str()); @@ -303,22 +305,16 @@ class Benchmark { } public: - enum Order { - SEQUENTIAL, - RANDOM - }; - enum DBState { - FRESH, - EXISTING - }; + enum Order { SEQUENTIAL, RANDOM }; + enum DBState { FRESH, EXISTING }; Benchmark() - : db_(NULL), - db_num_(0), - num_(FLAGS_num), - reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), - bytes_(0), - rand_(301) { + : db_(nullptr), + db_num_(0), + num_(FLAGS_num), + reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), + bytes_(0), + rand_(301) { std::vector files; std::string test_dir; Env::Default()->GetTestDirectory(&test_dir); @@ -345,12 +341,12 @@ class Benchmark { Open(); const char* benchmarks = FLAGS_benchmarks; - while (benchmarks != NULL) { + while (benchmarks != nullptr) { const char* sep = strchr(benchmarks, ','); Slice name; - if (sep == NULL) { + if (sep == nullptr) { name = benchmarks; - benchmarks = NULL; + benchmarks = nullptr; } else { name = Slice(benchmarks, sep - benchmarks); benchmarks = sep + 1; @@ -415,20 +411,18 @@ class Benchmark { } void Open() { - assert(db_ == NULL); + assert(db_ == nullptr); int status; char file_name[100]; - char* err_msg = NULL; + char* err_msg = nullptr; db_num_++; // Open database std::string tmp_dir; Env::Default()->GetTestDirectory(&tmp_dir); - snprintf(file_name, sizeof(file_name), - "%s/dbbench_sqlite3-%d.db", - tmp_dir.c_str(), - db_num_); + snprintf(file_name, sizeof(file_name), "%s/dbbench_sqlite3-%d.db", + tmp_dir.c_str(), db_num_); status = sqlite3_open(file_name, &db_); if (status) { fprintf(stderr, "open error: %s\n", sqlite3_errmsg(db_)); @@ -439,7 +433,7 @@ class Benchmark { char cache_size[100]; snprintf(cache_size, sizeof(cache_size), "PRAGMA cache_size = %d", FLAGS_num_pages); - status = sqlite3_exec(db_, cache_size, NULL, NULL, &err_msg); + status = sqlite3_exec(db_, cache_size, nullptr, nullptr, &err_msg); ExecErrorCheck(status, err_msg); // FLAGS_page_size is defaulted to 1024 @@ -447,7 +441,7 @@ class Benchmark { char page_size[100]; snprintf(page_size, sizeof(page_size), "PRAGMA page_size = %d", FLAGS_page_size); - status = sqlite3_exec(db_, page_size, NULL, NULL, &err_msg); + status = sqlite3_exec(db_, page_size, nullptr, nullptr, &err_msg); ExecErrorCheck(status, err_msg); } @@ -457,26 +451,28 @@ class Benchmark { // LevelDB's default cache size is a combined 4 MB std::string WAL_checkpoint = "PRAGMA wal_autocheckpoint = 4096"; - status = sqlite3_exec(db_, WAL_stmt.c_str(), NULL, NULL, &err_msg); + status = sqlite3_exec(db_, WAL_stmt.c_str(), nullptr, nullptr, &err_msg); ExecErrorCheck(status, err_msg); - status = sqlite3_exec(db_, WAL_checkpoint.c_str(), NULL, NULL, &err_msg); + status = + sqlite3_exec(db_, WAL_checkpoint.c_str(), nullptr, nullptr, &err_msg); ExecErrorCheck(status, err_msg); } // Change locking mode to exclusive and create tables/index for database std::string locking_stmt = "PRAGMA locking_mode = EXCLUSIVE"; std::string create_stmt = - "CREATE TABLE test (key blob, value blob, PRIMARY KEY(key))"; - std::string stmt_array[] = { locking_stmt, create_stmt }; + "CREATE TABLE test (key blob, value blob, PRIMARY KEY(key))"; + std::string stmt_array[] = {locking_stmt, create_stmt}; int stmt_array_length = sizeof(stmt_array) / sizeof(std::string); for (int i = 0; i < stmt_array_length; i++) { - status = sqlite3_exec(db_, stmt_array[i].c_str(), NULL, NULL, &err_msg); + status = + sqlite3_exec(db_, stmt_array[i].c_str(), nullptr, nullptr, &err_msg); ExecErrorCheck(status, err_msg); } } - void Write(bool write_sync, Order order, DBState state, - int num_entries, int value_size, int entries_per_batch) { + void Write(bool write_sync, Order order, DBState state, int num_entries, + int value_size, int entries_per_batch) { // Create new database if state == FRESH if (state == FRESH) { if (FLAGS_use_existing_db) { @@ -484,7 +480,7 @@ class Benchmark { return; } sqlite3_close(db_); - db_ = NULL; + db_ = nullptr; Open(); Start(); } @@ -495,7 +491,7 @@ class Benchmark { message_ = msg; } - char* err_msg = NULL; + char* err_msg = nullptr; int status; sqlite3_stmt *replace_stmt, *begin_trans_stmt, *end_trans_stmt; @@ -504,20 +500,20 @@ class Benchmark { std::string end_trans_str = "END TRANSACTION;"; // Check for synchronous flag in options - std::string sync_stmt = (write_sync) ? "PRAGMA synchronous = FULL" : - "PRAGMA synchronous = OFF"; - status = sqlite3_exec(db_, sync_stmt.c_str(), NULL, NULL, &err_msg); + std::string sync_stmt = + (write_sync) ? "PRAGMA synchronous = FULL" : "PRAGMA synchronous = OFF"; + status = sqlite3_exec(db_, sync_stmt.c_str(), nullptr, nullptr, &err_msg); ExecErrorCheck(status, err_msg); // Preparing sqlite3 statements - status = sqlite3_prepare_v2(db_, replace_str.c_str(), -1, - &replace_stmt, NULL); + status = sqlite3_prepare_v2(db_, replace_str.c_str(), -1, &replace_stmt, + nullptr); ErrorCheck(status); status = sqlite3_prepare_v2(db_, begin_trans_str.c_str(), -1, - &begin_trans_stmt, NULL); + &begin_trans_stmt, nullptr); ErrorCheck(status); - status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1, - &end_trans_stmt, NULL); + status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1, &end_trans_stmt, + nullptr); ErrorCheck(status); bool transaction = (entries_per_batch > 1); @@ -535,16 +531,16 @@ class Benchmark { const char* value = gen_.Generate(value_size).data(); // Create values for key-value pair - const int k = (order == SEQUENTIAL) ? i + j : - (rand_.Next() % num_entries); + const int k = + (order == SEQUENTIAL) ? i + j : (rand_.Next() % num_entries); char key[100]; snprintf(key, sizeof(key), "%016d", k); // Bind KV values into replace_stmt status = sqlite3_bind_blob(replace_stmt, 1, key, 16, SQLITE_STATIC); ErrorCheck(status); - status = sqlite3_bind_blob(replace_stmt, 2, value, - value_size, SQLITE_STATIC); + status = sqlite3_bind_blob(replace_stmt, 2, value, value_size, + SQLITE_STATIC); ErrorCheck(status); // Execute replace_stmt @@ -588,12 +584,12 @@ class Benchmark { // Preparing sqlite3 statements status = sqlite3_prepare_v2(db_, begin_trans_str.c_str(), -1, - &begin_trans_stmt, NULL); + &begin_trans_stmt, nullptr); ErrorCheck(status); - status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1, - &end_trans_stmt, NULL); + status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1, &end_trans_stmt, + nullptr); ErrorCheck(status); - status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &read_stmt, NULL); + status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &read_stmt, nullptr); ErrorCheck(status); bool transaction = (entries_per_batch > 1); @@ -618,7 +614,8 @@ class Benchmark { ErrorCheck(status); // Execute read statement - while ((status = sqlite3_step(read_stmt)) == SQLITE_ROW) {} + while ((status = sqlite3_step(read_stmt)) == SQLITE_ROW) { + } StepErrorCheck(status); // Reset SQLite statement for another use @@ -648,10 +645,10 @@ class Benchmark { void ReadSequential() { int status; - sqlite3_stmt *pStmt; + sqlite3_stmt* pStmt; std::string read_str = "SELECT * FROM test ORDER BY key"; - status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &pStmt, NULL); + status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &pStmt, nullptr); ErrorCheck(status); for (int i = 0; i < reads_ && SQLITE_ROW == sqlite3_step(pStmt); i++) { bytes_ += sqlite3_column_bytes(pStmt, 1) + sqlite3_column_bytes(pStmt, 2); @@ -661,7 +658,6 @@ class Benchmark { status = sqlite3_finalize(pStmt); ErrorCheck(status); } - }; } // namespace leveldb @@ -706,10 +702,10 @@ int main(int argc, char** argv) { } // Choose a location for the test database if none given with --db= - if (FLAGS_db == NULL) { - leveldb::Env::Default()->GetTestDirectory(&default_db_path); - default_db_path += "/dbbench"; - FLAGS_db = default_db_path.c_str(); + if (FLAGS_db == nullptr) { + leveldb::Env::Default()->GetTestDirectory(&default_db_path); + default_db_path += "/dbbench"; + FLAGS_db = default_db_path.c_str(); } leveldb::Benchmark benchmark; diff --git a/doc/bench/db_bench_tree_db.cc b/benchmarks/db_bench_tree_db.cc similarity index 85% rename from doc/bench/db_bench_tree_db.cc rename to benchmarks/db_bench_tree_db.cc index 4ca381f..b2f6646 100644 --- a/doc/bench/db_bench_tree_db.cc +++ b/benchmarks/db_bench_tree_db.cc @@ -2,9 +2,10 @@ // 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 "util/histogram.h" #include "util/random.h" #include "util/testutil.h" @@ -34,8 +35,7 @@ static const char* FLAGS_benchmarks = "fillrand100K," "fillseq100K," "readseq100K," - "readrand100K," - ; + "readrand100K,"; // Number of key/values to place in database static int FLAGS_num = 1000000; @@ -69,11 +69,9 @@ static bool FLAGS_use_existing_db = false; static bool FLAGS_compression = true; // Use the db with the following name. -static const char* FLAGS_db = NULL; +static const char* FLAGS_db = nullptr; -inline -static void DBSynchronize(kyotocabinet::TreeDB* db_) -{ +inline static void DBSynchronize(kyotocabinet::TreeDB* db_) { // Synchronize will flush writes to disk if (!db_->synchronize()) { fprintf(stderr, "synchronize error: %s\n", db_->error().name()); @@ -121,7 +119,7 @@ static Slice TrimSpace(Slice s) { start++; } int limit = s.size(); - while (limit > start && isspace(s[limit-1])) { + while (limit > start && isspace(s[limit - 1])) { limit--; } return Slice(s.data() + start, limit - start); @@ -146,7 +144,7 @@ class Benchmark { // State kept for progress messages int done_; - int next_report_; // When to report next + int next_report_; // When to report next void PrintHeader() { const int kKeySize = 16; @@ -157,20 +155,20 @@ class Benchmark { static_cast(FLAGS_value_size * FLAGS_compression_ratio + 0.5)); fprintf(stdout, "Entries: %d\n", num_); fprintf(stdout, "RawSize: %.1f MB (estimated)\n", - ((static_cast(kKeySize + FLAGS_value_size) * num_) - / 1048576.0)); + ((static_cast(kKeySize + FLAGS_value_size) * num_) / + 1048576.0)); fprintf(stdout, "FileSize: %.1f MB (estimated)\n", - (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) - / 1048576.0)); + (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) / + 1048576.0)); PrintWarnings(); fprintf(stdout, "------------------------------------------------\n"); } void PrintWarnings() { #if defined(__GNUC__) && !defined(__OPTIMIZE__) - fprintf(stdout, - "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n" - ); + fprintf( + stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); #endif #ifndef NDEBUG fprintf(stdout, @@ -183,18 +181,18 @@ class Benchmark { kyotocabinet::VERSION, kyotocabinet::LIBVER, kyotocabinet::LIBREV); #if defined(__linux) - time_t now = time(NULL); + time_t now = time(nullptr); fprintf(stderr, "Date: %s", ctime(&now)); // ctime() adds newline FILE* cpuinfo = fopen("/proc/cpuinfo", "r"); - if (cpuinfo != NULL) { + if (cpuinfo != nullptr) { char line[1000]; int num_cpus = 0; std::string cpu_type; std::string cache_size; - while (fgets(line, sizeof(line), cpuinfo) != NULL) { + while (fgets(line, sizeof(line), cpuinfo) != nullptr) { const char* sep = strchr(line, ':'); - if (sep == NULL) { + if (sep == nullptr) { continue; } Slice key = TrimSpace(Slice(line, sep - 1 - line)); @@ -237,13 +235,20 @@ class Benchmark { done_++; if (done_ >= next_report_) { - 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; + 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 %d ops%30s\r", done_, ""); fflush(stderr); } @@ -261,16 +266,14 @@ class Benchmark { snprintf(rate, sizeof(rate), "%6.1f MB/s", (bytes_ / 1048576.0) / (finish - start_)); if (!message_.empty()) { - message_ = std::string(rate) + " " + message_; + message_ = std::string(rate) + " " + message_; } else { message_ = rate; } } - fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", - name.ToString().c_str(), - (finish - start_) * 1e6 / done_, - (message_.empty() ? "" : " "), + fprintf(stdout, "%-12s : %11.3f micros/op;%s%s\n", name.ToString().c_str(), + (finish - start_) * 1e6 / done_, (message_.empty() ? "" : " "), message_.c_str()); if (FLAGS_histogram) { fprintf(stdout, "Microseconds per op:\n%s\n", hist_.ToString().c_str()); @@ -279,21 +282,15 @@ class Benchmark { } public: - enum Order { - SEQUENTIAL, - RANDOM - }; - enum DBState { - FRESH, - EXISTING - }; + enum Order { SEQUENTIAL, RANDOM }; + enum DBState { FRESH, EXISTING }; Benchmark() - : db_(NULL), - num_(FLAGS_num), - reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), - bytes_(0), - rand_(301) { + : db_(nullptr), + num_(FLAGS_num), + reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), + bytes_(0), + rand_(301) { std::vector files; std::string test_dir; Env::Default()->GetTestDirectory(&test_dir); @@ -321,12 +318,12 @@ class Benchmark { Open(false); const char* benchmarks = FLAGS_benchmarks; - while (benchmarks != NULL) { + while (benchmarks != nullptr) { const char* sep = strchr(benchmarks, ','); Slice name; - if (sep == NULL) { + if (sep == nullptr) { name = benchmarks; - benchmarks = NULL; + benchmarks = nullptr; } else { name = Slice(benchmarks, sep - benchmarks); benchmarks = sep + 1; @@ -386,8 +383,8 @@ class Benchmark { } private: - void Open(bool sync) { - assert(db_ == NULL); + void Open(bool sync) { + assert(db_ == nullptr); // Initialize db_ db_ = new kyotocabinet::TreeDB(); @@ -395,16 +392,14 @@ class Benchmark { db_num_++; std::string test_dir; Env::Default()->GetTestDirectory(&test_dir); - snprintf(file_name, sizeof(file_name), - "%s/dbbench_polyDB-%d.kct", - test_dir.c_str(), - db_num_); + snprintf(file_name, sizeof(file_name), "%s/dbbench_polyDB-%d.kct", + test_dir.c_str(), db_num_); // Create tuning options and open the database - int open_options = kyotocabinet::PolyDB::OWRITER | - kyotocabinet::PolyDB::OCREATE; - int tune_options = kyotocabinet::TreeDB::TSMALL | - kyotocabinet::TreeDB::TLINEAR; + int open_options = + kyotocabinet::PolyDB::OWRITER | kyotocabinet::PolyDB::OCREATE; + int tune_options = + kyotocabinet::TreeDB::TSMALL | kyotocabinet::TreeDB::TLINEAR; if (FLAGS_compression) { tune_options |= kyotocabinet::TreeDB::TCOMPRESS; db_->tune_compressor(&comp_); @@ -412,7 +407,7 @@ class Benchmark { db_->tune_options(tune_options); db_->tune_page_cache(FLAGS_cache_size); db_->tune_page(FLAGS_page_size); - db_->tune_map(256LL<<20); + db_->tune_map(256LL << 20); if (sync) { open_options |= kyotocabinet::PolyDB::OAUTOSYNC; } @@ -421,8 +416,8 @@ class Benchmark { } } - void Write(bool sync, Order order, DBState state, - int num_entries, int value_size, int entries_per_batch) { + void Write(bool sync, Order order, DBState state, int num_entries, + int value_size, int entries_per_batch) { // Create new database if state == FRESH if (state == FRESH) { if (FLAGS_use_existing_db) { @@ -430,7 +425,7 @@ class Benchmark { return; } delete db_; - db_ = NULL; + db_ = nullptr; Open(sync); Start(); // Do not count time taken to destroy/open } @@ -442,8 +437,7 @@ class Benchmark { } // Write to database - for (int i = 0; i < num_entries; i++) - { + for (int i = 0; i < num_entries; i++) { const int k = (order == SEQUENTIAL) ? i : (rand_.Next() % num_entries); char key[100]; snprintf(key, sizeof(key), "%016d", k); @@ -516,10 +510,10 @@ int main(int argc, char** argv) { } // Choose a location for the test database if none given with --db= - if (FLAGS_db == NULL) { - leveldb::Env::Default()->GetTestDirectory(&default_db_path); - default_db_path += "/dbbench"; - FLAGS_db = default_db_path.c_str(); + if (FLAGS_db == nullptr) { + leveldb::Env::Default()->GetTestDirectory(&default_db_path); + default_db_path += "/dbbench"; + FLAGS_db = default_db_path.c_str(); } leveldb::Benchmark benchmark; diff --git a/build_detect_platform b/build_detect_platform deleted file mode 100755 index bb76c4f..0000000 --- a/build_detect_platform +++ /dev/null @@ -1,228 +0,0 @@ -#!/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 -# PLATFORM_LIBS Libraries flags -# PLATFORM_SHARED_EXT Extension for shared libraries -# PLATFORM_SHARED_LDFLAGS Flags for building shared library -# This flag is embedded just before the name -# of the shared library without intervening spaces -# 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_ATOMIC_PRESENT if is present -# -DLEVELDB_PLATFORM_POSIX for Posix-based platforms -# -DSNAPPY if the Snappy library is present -# - -OUTPUT=$1 -PREFIX=$2 -if test -z "$OUTPUT" || test -z "$PREFIX"; then - echo "usage: $0 " >&2 - exit 1 -fi - -# Delete existing output, if it exists -rm -f $OUTPUT -touch $OUTPUT - -if test -z "$CC"; then - CC=cc -fi - -if test -z "$CXX"; then - CXX=g++ -fi - -if test -z "$TMPDIR"; then - TMPDIR=/tmp -fi - -# Detect OS -if test -z "$TARGET_OS"; then - TARGET_OS=`uname -s` -fi - -COMMON_FLAGS= -CROSS_COMPILE= -PLATFORM_CCFLAGS= -PLATFORM_CXXFLAGS= -PLATFORM_LDFLAGS= -PLATFORM_LIBS= -PLATFORM_SHARED_EXT="so" -PLATFORM_SHARED_LDFLAGS="-shared -Wl,-soname -Wl," -PLATFORM_SHARED_CFLAGS="-fPIC" -PLATFORM_SHARED_VERSIONED=true - -MEMCMP_FLAG= -if [ "$CXX" = "g++" ]; then - # Use libc's memcmp instead of GCC's memcmp. This results in ~40% - # performance improvement on readrandom under gcc 4.4.3 on Linux/x86. - MEMCMP_FLAG="-fno-builtin-memcmp" -fi - -case "$TARGET_OS" in - CYGWIN_*) - PLATFORM=OS_LINUX - COMMON_FLAGS="$MEMCMP_FLAG -lpthread -DOS_LINUX -DCYGWIN" - PLATFORM_LDFLAGS="-lpthread" - PORT_FILE=port/port_posix.cc - ;; - Darwin) - PLATFORM=OS_MACOSX - COMMON_FLAGS="$MEMCMP_FLAG -DOS_MACOSX" - PLATFORM_SHARED_EXT=dylib - [ -z "$INSTALL_PATH" ] && INSTALL_PATH=`pwd` - PLATFORM_SHARED_LDFLAGS="-dynamiclib -install_name $INSTALL_PATH/" - PORT_FILE=port/port_posix.cc - ;; - Linux) - PLATFORM=OS_LINUX - COMMON_FLAGS="$MEMCMP_FLAG -pthread -DOS_LINUX" - PLATFORM_LDFLAGS="-pthread" - PORT_FILE=port/port_posix.cc - ;; - SunOS) - PLATFORM=OS_SOLARIS - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_SOLARIS" - PLATFORM_LIBS="-lpthread -lrt" - PORT_FILE=port/port_posix.cc - ;; - FreeBSD) - PLATFORM=OS_FREEBSD - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_FREEBSD" - PLATFORM_LIBS="-lpthread" - PORT_FILE=port/port_posix.cc - ;; - NetBSD) - PLATFORM=OS_NETBSD - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_NETBSD" - PLATFORM_LIBS="-lpthread -lgcc_s" - PORT_FILE=port/port_posix.cc - ;; - OpenBSD) - PLATFORM=OS_OPENBSD - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_OPENBSD" - PLATFORM_LDFLAGS="-pthread" - PORT_FILE=port/port_posix.cc - ;; - DragonFly) - PLATFORM=OS_DRAGONFLYBSD - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_DRAGONFLYBSD" - PLATFORM_LIBS="-lpthread" - PORT_FILE=port/port_posix.cc - ;; - OS_ANDROID_CROSSCOMPILE) - PLATFORM=OS_ANDROID - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_ANDROID -DLEVELDB_PLATFORM_POSIX" - PLATFORM_LDFLAGS="" # All pthread features are in the Android C library - PORT_FILE=port/port_posix.cc - CROSS_COMPILE=true - ;; - HP-UX) - PLATFORM=OS_HPUX - COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_HPUX" - PLATFORM_LDFLAGS="-pthread" - PORT_FILE=port/port_posix.cc - # man ld: +h internal_name - PLATFORM_SHARED_LDFLAGS="-shared -Wl,+h -Wl," - ;; - IOS) - PLATFORM=IOS - COMMON_FLAGS="$MEMCMP_FLAG -DOS_MACOSX" - [ -z "$INSTALL_PATH" ] && INSTALL_PATH=`pwd` - PORT_FILE=port/port_posix.cc - PLATFORM_SHARED_EXT= - PLATFORM_SHARED_LDFLAGS= - PLATFORM_SHARED_CFLAGS= - PLATFORM_SHARED_VERSIONED= - ;; - *) - echo "Unknown platform!" >&2 - exit 1 -esac - -# We want to make a list of all cc files within util, db, table, and helpers -# except for the test and benchmark files. By default, find will output a list -# of all files matching either rule, so we need to append -print to make the -# prune take effect. -DIRS="$PREFIX/db $PREFIX/util $PREFIX/table" - -set -f # temporarily disable globbing so that our patterns aren't expanded -PRUNE_TEST="-name *test*.cc -prune" -PRUNE_BENCH="-name *_bench.cc -prune" -PRUNE_TOOL="-name leveldb_main.cc -prune" -PORTABLE_FILES=`find $DIRS $PRUNE_TEST -o $PRUNE_BENCH -o $PRUNE_TOOL -o -name '*.cc' -print | sort | sed "s,^$PREFIX/,," | tr "\n" " "` - -set +f # re-enable globbing - -# The sources consist of the portable files, plus the platform-specific port -# file. -echo "SOURCES=$PORTABLE_FILES $PORT_FILE" >> $OUTPUT -echo "MEMENV_SOURCES=helpers/memenv/memenv.cc" >> $OUTPUT - -if [ "$CROSS_COMPILE" = "true" ]; then - # Cross-compiling; do not try any compilation tests. - true -else - CXXOUTPUT="${TMPDIR}/leveldb_build_detect_platform-cxx.$$" - - # If -std=c++0x works, use as fallback for when memory barriers - # are not available. - $CXX $CXXFLAGS -std=c++0x -x c++ - -o $CXXOUTPUT 2>/dev/null < - int main() {} -EOF - if [ "$?" = 0 ]; then - COMMON_FLAGS="$COMMON_FLAGS -DLEVELDB_PLATFORM_POSIX -DLEVELDB_ATOMIC_PRESENT" - PLATFORM_CXXFLAGS="-std=c++0x" - else - COMMON_FLAGS="$COMMON_FLAGS -DLEVELDB_PLATFORM_POSIX" - fi - - # Test whether Snappy library is installed - # http://code.google.com/p/snappy/ - $CXX $CXXFLAGS -x c++ - -o $CXXOUTPUT 2>/dev/null < - int main() {} -EOF - if [ "$?" = 0 ]; then - COMMON_FLAGS="$COMMON_FLAGS -DSNAPPY" - PLATFORM_LIBS="$PLATFORM_LIBS -lsnappy" - fi - - # Test whether tcmalloc is available - $CXX $CXXFLAGS -x c++ - -o $CXXOUTPUT -ltcmalloc 2>/dev/null </dev/null -fi - -PLATFORM_CCFLAGS="$PLATFORM_CCFLAGS $COMMON_FLAGS" -PLATFORM_CXXFLAGS="$PLATFORM_CXXFLAGS $COMMON_FLAGS" - -echo "CC=$CC" >> $OUTPUT -echo "CXX=$CXX" >> $OUTPUT -echo "PLATFORM=$PLATFORM" >> $OUTPUT -echo "PLATFORM_LDFLAGS=$PLATFORM_LDFLAGS" >> $OUTPUT -echo "PLATFORM_LIBS=$PLATFORM_LIBS" >> $OUTPUT -echo "PLATFORM_CCFLAGS=$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" >> $OUTPUT diff --git a/cmake/leveldbConfig.cmake b/cmake/leveldbConfig.cmake new file mode 100644 index 0000000..eea6e5c --- /dev/null +++ b/cmake/leveldbConfig.cmake @@ -0,0 +1 @@ +include("${CMAKE_CURRENT_LIST_DIR}/leveldbTargets.cmake") diff --git a/db/autocompact_test.cc b/db/autocompact_test.cc index d20a236..e6c97a0 100644 --- a/db/autocompact_test.cc +++ b/db/autocompact_test.cc @@ -2,9 +2,9 @@ // 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 "leveldb/db.h" #include "db/db_impl.h" #include "leveldb/cache.h" +#include "leveldb/db.h" #include "util/testharness.h" #include "util/testutil.h" @@ -12,11 +12,6 @@ namespace leveldb { class AutoCompactTest { public: - std::string dbname_; - Cache* tiny_cache_; - Options options_; - DB* db_; - AutoCompactTest() { dbname_ = test::TmpDir() + "/autocompact_test"; tiny_cache_ = NewLRUCache(100); @@ -47,6 +42,12 @@ class AutoCompactTest { } void DoReads(int n); + + private: + std::string dbname_; + Cache* tiny_cache_; + Options options_; + DB* db_; }; static const int kValueSize = 200 * 1024; @@ -81,17 +82,16 @@ void AutoCompactTest::DoReads(int n) { ASSERT_LT(read, 100) << "Taking too long to compact"; Iterator* iter = db_->NewIterator(ReadOptions()); for (iter->SeekToFirst(); - iter->Valid() && iter->key().ToString() < limit_key; - iter->Next()) { + iter->Valid() && iter->key().ToString() < limit_key; iter->Next()) { // Drop data } delete iter; // Wait a little bit to allow any triggered compactions to complete. Env::Default()->SleepForMicroseconds(1000000); uint64_t size = Size(Key(0), Key(n)); - fprintf(stderr, "iter %3d => %7.3f MB [other %7.3f MB]\n", - read+1, size/1048576.0, Size(Key(n), Key(kCount))/1048576.0); - if (size <= initial_size/10) { + fprintf(stderr, "iter %3d => %7.3f MB [other %7.3f MB]\n", read + 1, + size / 1048576.0, Size(Key(n), Key(kCount)) / 1048576.0); + if (size <= initial_size / 10) { break; } } @@ -100,19 +100,13 @@ void AutoCompactTest::DoReads(int n) { // is pretty much unchanged. const int64_t final_other_size = Size(Key(n), Key(kCount)); ASSERT_LE(final_other_size, initial_other_size + 1048576); - ASSERT_GE(final_other_size, initial_other_size/5 - 1048576); + ASSERT_GE(final_other_size, initial_other_size / 5 - 1048576); } -TEST(AutoCompactTest, ReadAll) { - DoReads(kCount); -} +TEST(AutoCompactTest, ReadAll) { DoReads(kCount); } -TEST(AutoCompactTest, ReadHalf) { - DoReads(kCount/2); -} +TEST(AutoCompactTest, ReadHalf) { DoReads(kCount / 2); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/builder.cc b/db/builder.cc index f419882..9520ee4 100644 --- a/db/builder.cc +++ b/db/builder.cc @@ -4,8 +4,8 @@ #include "db/builder.h" -#include "db/filename.h" #include "db/dbformat.h" +#include "db/filename.h" #include "db/table_cache.h" #include "db/version_edit.h" #include "leveldb/db.h" @@ -14,12 +14,8 @@ namespace leveldb { -Status BuildTable(const std::string& dbname, - Env* env, - const Options& options, - TableCache* table_cache, - Iterator* iter, - FileMetaData* meta) { +Status BuildTable(const std::string& dbname, Env* env, const Options& options, + TableCache* table_cache, Iterator* iter, FileMetaData* meta) { Status s; meta->file_size = 0; iter->SeekToFirst(); @@ -41,14 +37,10 @@ Status BuildTable(const std::string& dbname, } // Finish and check for builder errors + s = builder->Finish(); if (s.ok()) { - s = builder->Finish(); - if (s.ok()) { - meta->file_size = builder->FileSize(); - assert(meta->file_size > 0); - } - } else { - builder->Abandon(); + meta->file_size = builder->FileSize(); + assert(meta->file_size > 0); } delete builder; @@ -60,12 +52,11 @@ Status BuildTable(const std::string& dbname, s = file->Close(); } delete file; - file = NULL; + file = nullptr; if (s.ok()) { // Verify that the table is usable - Iterator* it = table_cache->NewIterator(ReadOptions(), - meta->number, + Iterator* it = table_cache->NewIterator(ReadOptions(), meta->number, meta->file_size); s = it->status(); delete it; diff --git a/db/builder.h b/db/builder.h index 62431fc..7bd0b80 100644 --- a/db/builder.h +++ b/db/builder.h @@ -22,12 +22,8 @@ class VersionEdit; // *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 Options& options, - TableCache* table_cache, - Iterator* iter, - FileMetaData* meta); +Status BuildTable(const std::string& dbname, Env* env, const Options& options, + TableCache* table_cache, Iterator* iter, FileMetaData* meta); } // namespace leveldb diff --git a/db/c.cc b/db/c.cc index 08ff0ad..1f6fd64 100644 --- a/db/c.cc +++ b/db/c.cc @@ -5,7 +5,7 @@ #include "leveldb/c.h" #include -#include + #include "leveldb/cache.h" #include "leveldb/comparator.h" #include "leveldb/db.h" @@ -43,69 +43,72 @@ using leveldb::WriteOptions; extern "C" { -struct leveldb_t { DB* rep; }; -struct leveldb_iterator_t { Iterator* rep; }; -struct leveldb_writebatch_t { WriteBatch rep; }; -struct leveldb_snapshot_t { const Snapshot* rep; }; -struct leveldb_readoptions_t { ReadOptions rep; }; -struct leveldb_writeoptions_t { WriteOptions rep; }; -struct leveldb_options_t { Options rep; }; -struct leveldb_cache_t { Cache* rep; }; -struct leveldb_seqfile_t { SequentialFile* rep; }; -struct leveldb_randomfile_t { RandomAccessFile* rep; }; -struct leveldb_writablefile_t { WritableFile* rep; }; -struct leveldb_logger_t { Logger* rep; }; -struct leveldb_filelock_t { FileLock* rep; }; +struct leveldb_t { + DB* rep; +}; +struct leveldb_iterator_t { + Iterator* rep; +}; +struct leveldb_writebatch_t { + WriteBatch rep; +}; +struct leveldb_snapshot_t { + const Snapshot* rep; +}; +struct leveldb_readoptions_t { + ReadOptions rep; +}; +struct leveldb_writeoptions_t { + WriteOptions rep; +}; +struct leveldb_options_t { + Options rep; +}; +struct leveldb_cache_t { + Cache* rep; +}; +struct leveldb_seqfile_t { + SequentialFile* rep; +}; +struct leveldb_randomfile_t { + RandomAccessFile* rep; +}; +struct leveldb_writablefile_t { + WritableFile* rep; +}; +struct leveldb_logger_t { + Logger* rep; +}; +struct leveldb_filelock_t { + FileLock* rep; +}; struct leveldb_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*); + ~leveldb_comparator_t() override { (*destructor_)(state_); } - virtual ~leveldb_comparator_t() { - (*destructor_)(state_); - } - - virtual int Compare(const Slice& a, const Slice& b) const { + 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 { - return (*name_)(state_); - } + 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 { } - virtual void FindShortSuccessor(std::string* key) const { } + void FindShortestSeparator(std::string*, const Slice&) const override {} + void FindShortSuccessor(std::string* key) const override {} + + void* state_; + void (*destructor_)(void*); + int (*compare_)(void*, const char* a, size_t alen, const char* b, + size_t blen); + const char* (*name_)(void*); }; struct leveldb_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); + ~leveldb_filterpolicy_t() override { (*destructor_)(state_); } - virtual ~leveldb_filterpolicy_t() { - (*destructor_)(state_); - } + const char* Name() const override { return (*name_)(state_); } - virtual const char* Name() const { - return (*name_)(state_); - } - - virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const { + 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++) { @@ -118,10 +121,19 @@ struct leveldb_filterpolicy_t : public FilterPolicy { free(filter); } - virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const { - return (*key_match_)(state_, key.data(), key.size(), - filter.data(), filter.size()); + bool KeyMayMatch(const Slice& key, const Slice& filter) const override { + return (*key_match_)(state_, key.data(), key.size(), filter.data(), + filter.size()); } + + 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); }; struct leveldb_env_t { @@ -130,10 +142,10 @@ struct leveldb_env_t { }; static bool SaveError(char** errptr, const Status& s) { - assert(errptr != NULL); + assert(errptr != nullptr); if (s.ok()) { return false; - } else if (*errptr == NULL) { + } else if (*errptr == nullptr) { *errptr = strdup(s.ToString().c_str()); } else { // TODO(sanjay): Merge with existing error? @@ -149,13 +161,11 @@ static char* CopyString(const std::string& str) { return result; } -leveldb_t* leveldb_open( - const leveldb_options_t* options, - const char* name, - char** errptr) { +leveldb_t* leveldb_open(const leveldb_options_t* options, const char* name, + char** errptr) { DB* db; if (SaveError(errptr, DB::Open(options->rep, std::string(name), &db))) { - return NULL; + return nullptr; } leveldb_t* result = new leveldb_t; result->rep = db; @@ -167,40 +177,27 @@ void leveldb_close(leveldb_t* db) { delete db; } -void leveldb_put( - leveldb_t* db, - const leveldb_writeoptions_t* options, - const char* key, size_t keylen, - const char* val, size_t vallen, - char** errptr) { +void leveldb_put(leveldb_t* db, const leveldb_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 leveldb_delete( - leveldb_t* db, - const leveldb_writeoptions_t* options, - const char* key, size_t keylen, - char** errptr) { +void leveldb_delete(leveldb_t* db, const leveldb_writeoptions_t* options, + const char* key, size_t keylen, char** errptr) { SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen))); } - -void leveldb_write( - leveldb_t* db, - const leveldb_writeoptions_t* options, - leveldb_writebatch_t* batch, - char** errptr) { +void leveldb_write(leveldb_t* db, const leveldb_writeoptions_t* options, + leveldb_writebatch_t* batch, char** errptr) { SaveError(errptr, db->rep->Write(options->rep, &batch->rep)); } -char* leveldb_get( - leveldb_t* db, - const leveldb_readoptions_t* options, - const char* key, size_t keylen, - size_t* vallen, - char** errptr) { - char* result = NULL; +char* leveldb_get(leveldb_t* db, const leveldb_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()) { @@ -216,45 +213,40 @@ char* leveldb_get( } leveldb_iterator_t* leveldb_create_iterator( - leveldb_t* db, - const leveldb_readoptions_t* options) { + leveldb_t* db, const leveldb_readoptions_t* options) { leveldb_iterator_t* result = new leveldb_iterator_t; result->rep = db->rep->NewIterator(options->rep); return result; } -const leveldb_snapshot_t* leveldb_create_snapshot( - leveldb_t* db) { +const leveldb_snapshot_t* leveldb_create_snapshot(leveldb_t* db) { leveldb_snapshot_t* result = new leveldb_snapshot_t; result->rep = db->rep->GetSnapshot(); return result; } -void leveldb_release_snapshot( - leveldb_t* db, - const leveldb_snapshot_t* snapshot) { +void leveldb_release_snapshot(leveldb_t* db, + const leveldb_snapshot_t* snapshot) { db->rep->ReleaseSnapshot(snapshot->rep); delete snapshot; } -char* leveldb_property_value( - leveldb_t* db, - const char* propname) { +char* leveldb_property_value(leveldb_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 NULL; + return nullptr; } } -void leveldb_approximate_sizes( - leveldb_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) { +void leveldb_approximate_sizes(leveldb_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]); @@ -264,28 +256,23 @@ void leveldb_approximate_sizes( delete[] ranges; } -void leveldb_compact_range( - leveldb_t* db, - const char* start_key, size_t start_key_len, - const char* limit_key, size_t limit_key_len) { +void leveldb_compact_range(leveldb_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( - // Pass NULL Slice if corresponding "const char*" is NULL - (start_key ? (a = Slice(start_key, start_key_len), &a) : NULL), - (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : NULL)); + // Pass null Slice if corresponding "const char*" is null + (start_key ? (a = Slice(start_key, start_key_len), &a) : nullptr), + (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : nullptr)); } -void leveldb_destroy_db( - const leveldb_options_t* options, - const char* name, - char** errptr) { +void leveldb_destroy_db(const leveldb_options_t* options, const char* name, + char** errptr) { SaveError(errptr, DestroyDB(name, options->rep)); } -void leveldb_repair_db( - const leveldb_options_t* options, - const char* name, - char** errptr) { +void leveldb_repair_db(const leveldb_options_t* options, const char* name, + char** errptr) { SaveError(errptr, RepairDB(name, options->rep)); } @@ -310,13 +297,9 @@ void leveldb_iter_seek(leveldb_iterator_t* iter, const char* k, size_t klen) { iter->rep->Seek(Slice(k, klen)); } -void leveldb_iter_next(leveldb_iterator_t* iter) { - iter->rep->Next(); -} +void leveldb_iter_next(leveldb_iterator_t* iter) { iter->rep->Next(); } -void leveldb_iter_prev(leveldb_iterator_t* iter) { - iter->rep->Prev(); -} +void leveldb_iter_prev(leveldb_iterator_t* iter) { iter->rep->Prev(); } const char* leveldb_iter_key(const leveldb_iterator_t* iter, size_t* klen) { Slice s = iter->rep->key(); @@ -338,41 +321,34 @@ leveldb_writebatch_t* leveldb_writebatch_create() { return new leveldb_writebatch_t; } -void leveldb_writebatch_destroy(leveldb_writebatch_t* b) { - delete b; -} +void leveldb_writebatch_destroy(leveldb_writebatch_t* b) { delete b; } -void leveldb_writebatch_clear(leveldb_writebatch_t* b) { - b->rep.Clear(); -} +void leveldb_writebatch_clear(leveldb_writebatch_t* b) { b->rep.Clear(); } -void leveldb_writebatch_put( - leveldb_writebatch_t* b, - const char* key, size_t klen, - const char* val, size_t vlen) { +void leveldb_writebatch_put(leveldb_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 leveldb_writebatch_delete( - leveldb_writebatch_t* b, - const char* key, size_t klen) { +void leveldb_writebatch_delete(leveldb_writebatch_t* b, const char* key, + size_t klen) { b->rep.Delete(Slice(key, klen)); } -void leveldb_writebatch_iterate( - leveldb_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)) { +void leveldb_writebatch_iterate(const leveldb_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) { + 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) { + void Delete(const Slice& key) override { (*deleted_)(state_, key.data(), key.size()); } }; @@ -383,47 +359,46 @@ void leveldb_writebatch_iterate( b->rep.Iterate(&handler); } -leveldb_options_t* leveldb_options_create() { - return new leveldb_options_t; +void leveldb_writebatch_append(leveldb_writebatch_t* destination, + const leveldb_writebatch_t* source) { + destination->rep.Append(source->rep); } -void leveldb_options_destroy(leveldb_options_t* options) { - delete options; -} +leveldb_options_t* leveldb_options_create() { return new leveldb_options_t; } -void leveldb_options_set_comparator( - leveldb_options_t* opt, - leveldb_comparator_t* cmp) { +void leveldb_options_destroy(leveldb_options_t* options) { delete options; } + +void leveldb_options_set_comparator(leveldb_options_t* opt, + leveldb_comparator_t* cmp) { opt->rep.comparator = cmp; } -void leveldb_options_set_filter_policy( - leveldb_options_t* opt, - leveldb_filterpolicy_t* policy) { +void leveldb_options_set_filter_policy(leveldb_options_t* opt, + leveldb_filterpolicy_t* policy) { opt->rep.filter_policy = policy; } -void leveldb_options_set_create_if_missing( - leveldb_options_t* opt, unsigned char v) { +void leveldb_options_set_create_if_missing(leveldb_options_t* opt, + unsigned char v) { opt->rep.create_if_missing = v; } -void leveldb_options_set_error_if_exists( - leveldb_options_t* opt, unsigned char v) { +void leveldb_options_set_error_if_exists(leveldb_options_t* opt, + unsigned char v) { opt->rep.error_if_exists = v; } -void leveldb_options_set_paranoid_checks( - leveldb_options_t* opt, unsigned char v) { +void leveldb_options_set_paranoid_checks(leveldb_options_t* opt, + unsigned char v) { opt->rep.paranoid_checks = v; } void leveldb_options_set_env(leveldb_options_t* opt, leveldb_env_t* env) { - opt->rep.env = (env ? env->rep : NULL); + opt->rep.env = (env ? env->rep : nullptr); } void leveldb_options_set_info_log(leveldb_options_t* opt, leveldb_logger_t* l) { - opt->rep.info_log = (l ? l->rep : NULL); + opt->rep.info_log = (l ? l->rep : nullptr); } void leveldb_options_set_write_buffer_size(leveldb_options_t* opt, size_t s) { @@ -446,17 +421,18 @@ void leveldb_options_set_block_restart_interval(leveldb_options_t* opt, int n) { opt->rep.block_restart_interval = n; } +void leveldb_options_set_max_file_size(leveldb_options_t* opt, size_t s) { + opt->rep.max_file_size = s; +} + void leveldb_options_set_compression(leveldb_options_t* opt, int t) { opt->rep.compression = static_cast(t); } leveldb_comparator_t* leveldb_comparator_create( - void* state, - void (*destructor)(void*), - int (*compare)( - void*, - const char* a, size_t alen, - const char* b, size_t blen), + void* state, void (*destructor)(void*), + int (*compare)(void*, const char* a, size_t alen, const char* b, + size_t blen), const char* (*name)(void*)) { leveldb_comparator_t* result = new leveldb_comparator_t; result->state_ = state; @@ -466,22 +442,15 @@ leveldb_comparator_t* leveldb_comparator_create( return result; } -void leveldb_comparator_destroy(leveldb_comparator_t* cmp) { - delete cmp; -} +void leveldb_comparator_destroy(leveldb_comparator_t* cmp) { delete cmp; } leveldb_filterpolicy_t* leveldb_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* 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), const char* (*name)(void*)) { leveldb_filterpolicy_t* result = new leveldb_filterpolicy_t; result->state_ = state; @@ -501,7 +470,8 @@ leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(int bits_per_key) { // they delegate to a NewBloomFilterPolicy() instead of user // supplied C functions. struct Wrapper : public leveldb_filterpolicy_t { - const FilterPolicy* rep_; + static void DoNothing(void*) {} + ~Wrapper() { delete rep_; } const char* Name() const { return rep_->Name(); } void CreateFilter(const Slice* keys, int n, std::string* dst) const { @@ -510,11 +480,12 @@ leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(int bits_per_key) { bool KeyMayMatch(const Slice& key, const Slice& filter) const { return rep_->KeyMayMatch(key, filter); } - static void DoNothing(void*) { } + + const FilterPolicy* rep_; }; Wrapper* wrapper = new Wrapper; wrapper->rep_ = NewBloomFilterPolicy(bits_per_key); - wrapper->state_ = NULL; + wrapper->state_ = nullptr; wrapper->destructor_ = &Wrapper::DoNothing; return wrapper; } @@ -523,37 +494,31 @@ leveldb_readoptions_t* leveldb_readoptions_create() { return new leveldb_readoptions_t; } -void leveldb_readoptions_destroy(leveldb_readoptions_t* opt) { - delete opt; -} +void leveldb_readoptions_destroy(leveldb_readoptions_t* opt) { delete opt; } -void leveldb_readoptions_set_verify_checksums( - leveldb_readoptions_t* opt, - unsigned char v) { +void leveldb_readoptions_set_verify_checksums(leveldb_readoptions_t* opt, + unsigned char v) { opt->rep.verify_checksums = v; } -void leveldb_readoptions_set_fill_cache( - leveldb_readoptions_t* opt, unsigned char v) { +void leveldb_readoptions_set_fill_cache(leveldb_readoptions_t* opt, + unsigned char v) { opt->rep.fill_cache = v; } -void leveldb_readoptions_set_snapshot( - leveldb_readoptions_t* opt, - const leveldb_snapshot_t* snap) { - opt->rep.snapshot = (snap ? snap->rep : NULL); +void leveldb_readoptions_set_snapshot(leveldb_readoptions_t* opt, + const leveldb_snapshot_t* snap) { + opt->rep.snapshot = (snap ? snap->rep : nullptr); } leveldb_writeoptions_t* leveldb_writeoptions_create() { return new leveldb_writeoptions_t; } -void leveldb_writeoptions_destroy(leveldb_writeoptions_t* opt) { - delete opt; -} +void leveldb_writeoptions_destroy(leveldb_writeoptions_t* opt) { delete opt; } -void leveldb_writeoptions_set_sync( - leveldb_writeoptions_t* opt, unsigned char v) { +void leveldb_writeoptions_set_sync(leveldb_writeoptions_t* opt, + unsigned char v) { opt->rep.sync = v; } @@ -580,16 +545,22 @@ void leveldb_env_destroy(leveldb_env_t* env) { delete env; } -void leveldb_free(void* ptr) { - free(ptr); +char* leveldb_env_get_test_directory(leveldb_env_t* env) { + std::string result; + if (!env->rep->GetTestDirectory(&result).ok()) { + return nullptr; + } + + char* buffer = static_cast(malloc(result.size() + 1)); + memcpy(buffer, result.data(), result.size()); + buffer[result.size()] = '\0'; + return buffer; } -int leveldb_major_version() { - return kMajorVersion; -} +void leveldb_free(void* ptr) { free(ptr); } -int leveldb_minor_version() { - return kMinorVersion; -} +int leveldb_major_version() { return kMajorVersion; } + +int leveldb_minor_version() { return kMinorVersion; } } // end extern "C" diff --git a/db/c_test.c b/db/c_test.c index 7cd5ee0..ae14b99 100644 --- a/db/c_test.c +++ b/db/c_test.c @@ -8,24 +8,14 @@ #include #include #include -#include -#include const char* phase = ""; -static char dbname[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)); \ @@ -162,21 +152,19 @@ int main(int argc, char** argv) { leveldb_options_t* options; leveldb_readoptions_t* roptions; leveldb_writeoptions_t* woptions; + char* dbname; char* err = NULL; int run = -1; CheckCondition(leveldb_major_version() >= 1); CheckCondition(leveldb_minor_version() >= 1); - snprintf(dbname, sizeof(dbname), - "%s/leveldb_c_test-%d", - GetTempDir(), - ((int) geteuid())); - StartPhase("create_objects"); cmp = leveldb_comparator_create(NULL, CmpDestroy, CmpCompare, CmpName); env = leveldb_create_default_env(); cache = leveldb_cache_create_lru(100000); + dbname = leveldb_env_get_test_directory(env); + CheckCondition(dbname != NULL); options = leveldb_options_create(); leveldb_options_set_comparator(options, cmp); @@ -189,6 +177,7 @@ int main(int argc, char** argv) { leveldb_options_set_max_open_files(options, 10); leveldb_options_set_block_size(options, 1024); leveldb_options_set_block_restart_interval(options, 8); + leveldb_options_set_max_file_size(options, 3 << 20); leveldb_options_set_compression(options, leveldb_no_compression); roptions = leveldb_readoptions_create(); @@ -239,12 +228,18 @@ int main(int argc, char** argv) { leveldb_writebatch_clear(wb); leveldb_writebatch_put(wb, "bar", 3, "b", 1); leveldb_writebatch_put(wb, "box", 3, "c", 1); - leveldb_writebatch_delete(wb, "bar", 3); + + leveldb_writebatch_t* wb2 = leveldb_writebatch_create(); + leveldb_writebatch_delete(wb2, "bar", 3); + leveldb_writebatch_append(wb, wb2); + leveldb_writebatch_destroy(wb2); + leveldb_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; leveldb_writebatch_iterate(wb, &pos, CheckPut, CheckDel); CheckCondition(pos == 3); @@ -381,6 +376,7 @@ int main(int argc, char** argv) { leveldb_options_destroy(options); leveldb_readoptions_destroy(roptions); leveldb_writeoptions_destroy(woptions); + leveldb_free(dbname); leveldb_cache_destroy(cache); leveldb_comparator_destroy(cmp); leveldb_env_destroy(env); diff --git a/db/corruption_test.cc b/db/corruption_test.cc index 96afc68..42f5237 100644 --- a/db/corruption_test.cc +++ b/db/corruption_test.cc @@ -2,20 +2,16 @@ // 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 "leveldb/db.h" - -#include -#include -#include #include -#include "leveldb/cache.h" -#include "leveldb/env.h" -#include "leveldb/table.h" -#include "leveldb/write_batch.h" + #include "db/db_impl.h" #include "db/filename.h" #include "db/log_format.h" #include "db/version_set.h" +#include "leveldb/cache.h" +#include "leveldb/db.h" +#include "leveldb/table.h" +#include "leveldb/write_batch.h" #include "util/logging.h" #include "util/testharness.h" #include "util/testutil.h" @@ -26,44 +22,35 @@ static const int kValueSize = 1000; class CorruptionTest { public: - test::ErrorEnv env_; - std::string dbname_; - Cache* tiny_cache_; - Options options_; - DB* db_; - - CorruptionTest() { - tiny_cache_ = NewLRUCache(100); + CorruptionTest() + : db_(nullptr), + dbname_("/memenv/corruption_test"), + tiny_cache_(NewLRUCache(100)) { options_.env = &env_; options_.block_cache = tiny_cache_; - dbname_ = test::TmpDir() + "/db_test"; DestroyDB(dbname_, options_); - db_ = NULL; options_.create_if_missing = true; Reopen(); options_.create_if_missing = false; } ~CorruptionTest() { - delete db_; - DestroyDB(dbname_, Options()); - delete tiny_cache_; + delete db_; + delete tiny_cache_; } Status TryReopen() { delete db_; - db_ = NULL; + db_ = nullptr; return DB::Open(options_, dbname_, &db_); } - void Reopen() { - ASSERT_OK(TryReopen()); - } + void Reopen() { ASSERT_OK(TryReopen()); } void RepairDB() { delete db_; - db_ = NULL; + db_ = nullptr; ASSERT_OK(::leveldb::RepairDB(dbname_, options_)); } @@ -71,7 +58,7 @@ class CorruptionTest { std::string key_space, value_space; WriteBatch batch; for (int i = 0; i < n; i++) { - //if ((i % 100) == 0) fprintf(stderr, "@ %d of %d\n", i, n); + // 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)); @@ -100,8 +87,7 @@ class CorruptionTest { // Ignore boundary keys. continue; } - if (!ConsumeDecimalNumber(&in, &key) || - !in.empty() || + if (!ConsumeDecimalNumber(&in, &key) || !in.empty() || key < next_expected) { bad_keys++; continue; @@ -126,14 +112,13 @@ class CorruptionTest { void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) { // Pick file to corrupt std::vector filenames; - ASSERT_OK(env_.GetChildren(dbname_, &filenames)); + ASSERT_OK(env_.target()->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 && + if (ParseFileName(filenames[i], &number, &type) && type == filetype && int(number) > picked_number) { // Pick latest file fname = dbname_ + "/" + filenames[i]; picked_number = number; @@ -141,35 +126,32 @@ class CorruptionTest { } ASSERT_TRUE(!fname.empty()) << filetype; - struct stat sbuf; - if (stat(fname.c_str(), &sbuf) != 0) { - const char* msg = strerror(errno); - ASSERT_TRUE(false) << fname << ": " << msg; - } + uint64_t file_size; + ASSERT_OK(env_.target()->GetFileSize(fname, &file_size)); if (offset < 0) { // Relative to end of file; make it absolute - if (-offset > sbuf.st_size) { + if (-offset > file_size) { offset = 0; } else { - offset = sbuf.st_size + offset; + offset = file_size + offset; } } - if (offset > sbuf.st_size) { - offset = sbuf.st_size; + if (offset > file_size) { + offset = file_size; } - if (offset + bytes_to_corrupt > sbuf.st_size) { - bytes_to_corrupt = sbuf.st_size - offset; + if (offset + bytes_to_corrupt > file_size) { + bytes_to_corrupt = file_size - offset; } // Do it std::string contents; - Status s = ReadFileToString(Env::Default(), fname, &contents); + Status s = ReadFileToString(env_.target(), 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); + s = WriteStringToFile(env_.target(), contents, fname); ASSERT_TRUE(s.ok()) << s.ToString(); } @@ -197,12 +179,20 @@ class CorruptionTest { Random r(k); return test::RandomString(&r, kValueSize, storage); } + + test::ErrorEnv env_; + Options options_; + DB* db_; + + private: + std::string dbname_; + Cache* tiny_cache_; }; TEST(CorruptionTest, Recovery) { Build(100); Check(100, 100); - Corrupt(kLogFile, 19, 1); // WriteBatch tag for first record + Corrupt(kLogFile, 19, 1); // WriteBatch tag for first record Corrupt(kLogFile, log::kBlockSize + 1000, 1); // Somewhere in second block Reopen(); @@ -237,8 +227,8 @@ TEST(CorruptionTest, TableFile) { Build(100); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_CompactMemTable(); - dbi->TEST_CompactRange(0, NULL, NULL); - dbi->TEST_CompactRange(1, NULL, NULL); + dbi->TEST_CompactRange(0, nullptr, nullptr); + dbi->TEST_CompactRange(1, nullptr, nullptr); Corrupt(kTableFile, 100, 1); Check(90, 99); @@ -251,8 +241,8 @@ TEST(CorruptionTest, TableFileRepair) { Build(100); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_CompactMemTable(); - dbi->TEST_CompactRange(0, NULL, NULL); - dbi->TEST_CompactRange(1, NULL, NULL); + dbi->TEST_CompactRange(0, nullptr, nullptr); + dbi->TEST_CompactRange(1, nullptr, nullptr); Corrupt(kTableFile, 100, 1); RepairDB(); @@ -302,7 +292,7 @@ TEST(CorruptionTest, CorruptedDescriptor) { ASSERT_OK(db_->Put(WriteOptions(), "foo", "hello")); DBImpl* dbi = reinterpret_cast(db_); dbi->TEST_CompactMemTable(); - dbi->TEST_CompactRange(0, NULL, NULL); + dbi->TEST_CompactRange(0, nullptr, nullptr); Corrupt(kDescriptorFile, 0, 1000); Status s = TryReopen(); @@ -343,7 +333,7 @@ TEST(CorruptionTest, CompactionInputErrorParanoid) { Corrupt(kTableFile, 100, 1); env_.SleepForMicroseconds(100000); } - dbi->CompactRange(NULL, NULL); + dbi->CompactRange(nullptr, nullptr); // Write must fail because of corrupted table std::string tmp1, tmp2; @@ -369,6 +359,4 @@ TEST(CorruptionTest, UnrelatedKeys) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/db_impl.cc b/db/db_impl.cc index cc1cea4..6e8fe61 100644 --- a/db/db_impl.cc +++ b/db/db_impl.cc @@ -4,12 +4,15 @@ #include "db/db_impl.h" -#include -#include -#include #include #include + +#include +#include +#include +#include #include + #include "db/builder.h" #include "db/db_iter.h" #include "db/dbformat.h" @@ -39,16 +42,33 @@ const int kNumNonTableCacheFiles = 10; // Information kept for every waiting writer struct DBImpl::Writer { + explicit Writer(port::Mutex* mu) + : batch(nullptr), sync(false), done(false), cv(mu) {} + Status status; WriteBatch* batch; bool sync; bool done; port::CondVar cv; - - explicit Writer(port::Mutex* mu) : cv(mu) { } }; struct DBImpl::CompactionState { + // Files produced by compaction + struct Output { + uint64_t number; + uint64_t file_size; + InternalKey smallest, largest; + }; + + Output* current_output() { return &outputs[outputs.size() - 1]; } + + explicit CompactionState(Compaction* c) + : compaction(c), + smallest_snapshot(0), + outfile(nullptr), + builder(nullptr), + total_bytes(0) {} + Compaction* const compaction; // Sequence numbers < smallest_snapshot are not significant since we @@ -57,12 +77,6 @@ struct DBImpl::CompactionState { // we can drop all entries for the same key with sequence numbers < S. SequenceNumber smallest_snapshot; - // Files produced by compaction - struct Output { - uint64_t number; - uint64_t file_size; - InternalKey smallest, largest; - }; std::vector outputs; // State kept for output being generated @@ -70,19 +84,10 @@ struct DBImpl::CompactionState { TableBuilder* builder; uint64_t total_bytes; - - Output* current_output() { return &outputs[outputs.size()-1]; } - - explicit CompactionState(Compaction* c) - : compaction(c), - outfile(NULL), - builder(NULL), - total_bytes(0) { - } }; // Fix user-supplied options to be reasonable -template +template static void ClipToRange(T* ptr, V minvalue, V maxvalue) { if (static_cast(*ptr) > maxvalue) *ptr = maxvalue; if (static_cast(*ptr) < minvalue) *ptr = minvalue; @@ -93,26 +98,32 @@ Options SanitizeOptions(const std::string& dbname, const Options& src) { Options result = src; result.comparator = icmp; - result.filter_policy = (src.filter_policy != NULL) ? ipolicy : NULL; - ClipToRange(&result.max_open_files, 64 + kNumNonTableCacheFiles, 50000); - ClipToRange(&result.write_buffer_size, 64<<10, 1<<30); - ClipToRange(&result.block_size, 1<<10, 4<<20); - if (result.info_log == NULL) { + result.filter_policy = (src.filter_policy != nullptr) ? ipolicy : nullptr; + ClipToRange(&result.max_open_files, 64 + kNumNonTableCacheFiles, 50000); + ClipToRange(&result.write_buffer_size, 64 << 10, 1 << 30); + ClipToRange(&result.max_file_size, 1 << 20, 1 << 30); + ClipToRange(&result.block_size, 1 << 10, 4 << 20); + if (result.info_log == nullptr) { // Open a log file in the same directory as the db src.env->CreateDir(dbname); // In case it does not exist src.env->RenameFile(InfoLogFileName(dbname), OldInfoLogFileName(dbname)); Status s = src.env->NewLogger(InfoLogFileName(dbname), &result.info_log); if (!s.ok()) { // No place suitable for logging - result.info_log = NULL; + result.info_log = nullptr; } } - if (result.block_cache == NULL) { + if (result.block_cache == nullptr) { result.block_cache = NewLRUCache(8 << 20); } return result; } +static int TableCacheSize(const Options& sanitized_options) { + // Reserve ten files or so for other uses and give the rest to TableCache. + return sanitized_options.max_open_files - kNumNonTableCacheFiles; +} + DBImpl::DBImpl(const Options& raw_options, const std::string& dbname) : env_(raw_options.env), internal_comparator_(raw_options.comparator), @@ -122,45 +133,39 @@ DBImpl::DBImpl(const Options& raw_options, const std::string& dbname) owns_info_log_(options_.info_log != raw_options.info_log), owns_cache_(options_.block_cache != raw_options.block_cache), dbname_(dbname), - db_lock_(NULL), - shutting_down_(NULL), - bg_cv_(&mutex_), - mem_(new MemTable(internal_comparator_)), - imm_(NULL), - logfile_(NULL), + table_cache_(new TableCache(dbname_, options_, TableCacheSize(options_))), + db_lock_(nullptr), + shutting_down_(false), + background_work_finished_signal_(&mutex_), + mem_(nullptr), + imm_(nullptr), + has_imm_(false), + logfile_(nullptr), logfile_number_(0), - log_(NULL), + log_(nullptr), seed_(0), tmp_batch_(new WriteBatch), - bg_compaction_scheduled_(false), - manual_compaction_(NULL) { - mem_->Ref(); - has_imm_.Release_Store(NULL); - - // Reserve ten files or so for other uses and give the rest to TableCache. - const int table_cache_size = options_.max_open_files - kNumNonTableCacheFiles; - table_cache_ = new TableCache(dbname_, &options_, table_cache_size); - - versions_ = new VersionSet(dbname_, &options_, table_cache_, - &internal_comparator_); -} + background_compaction_scheduled_(false), + manual_compaction_(nullptr), + versions_(new VersionSet(dbname_, &options_, table_cache_, + &internal_comparator_)) {} DBImpl::~DBImpl() { - // Wait for background work to finish + // Wait for background work to finish. mutex_.Lock(); - shutting_down_.Release_Store(this); // Any non-NULL value is ok - while (bg_compaction_scheduled_) { - bg_cv_.Wait(); + shutting_down_.store(true, std::memory_order_release); + while (background_compaction_scheduled_) { + background_work_finished_signal_.Wait(); } mutex_.Unlock(); - if (db_lock_ != NULL) { + if (db_lock_ != nullptr) { env_->UnlockFile(db_lock_); } delete versions_; - if (mem_ != NULL) mem_->Unref(); - if (imm_ != NULL) imm_->Unref(); + if (mem_ != nullptr) mem_->Unref(); + if (imm_ != nullptr) imm_->Unref(); delete tmp_batch_; delete log_; delete logfile_; @@ -237,7 +242,6 @@ void DBImpl::DeleteObsoleteFiles() { // Unlock while deleting obsolete files mutex_.Unlock(); - uint64_t number; FileType type; for (size_t i = 0; i < filenames.size(); i++) { @@ -272,8 +276,7 @@ void DBImpl::DeleteObsoleteFiles() { if (type == kTableFile) { table_cache_->Evict(number); } - Log(options_.info_log, "Delete type=%d #%lld\n", - int(type), + Log(options_.info_log, "Delete type=%d #%lld\n", static_cast(type), static_cast(number)); env_->DeleteFile(dbname_ + "/" + filenames[i]); } @@ -282,14 +285,14 @@ void DBImpl::DeleteObsoleteFiles() { mutex_.Lock(); } -Status DBImpl::Recover(VersionEdit* edit) { +Status DBImpl::Recover(VersionEdit* edit, bool* save_manifest) { mutex_.AssertHeld(); // Ignore error from CreateDir since the creation of the DB is // committed only when the descriptor is created, and this directory // may already exist from a previous failed creation attempt. env_->CreateDir(dbname_); - assert(db_lock_ == NULL); + assert(db_lock_ == nullptr); Status s = env_->LockFile(LockFileName(dbname_), &db_lock_); if (!s.ok()) { return s; @@ -307,82 +310,85 @@ Status DBImpl::Recover(VersionEdit* edit) { } } else { if (options_.error_if_exists) { - return Status::InvalidArgument( - dbname_, "exists (error_if_exists is true)"); + return Status::InvalidArgument(dbname_, + "exists (error_if_exists is true)"); } } - s = versions_->Recover(); - if (s.ok()) { - SequenceNumber max_sequence(0); + s = versions_->Recover(save_manifest); + if (!s.ok()) { + return s; + } + SequenceNumber max_sequence(0); - // 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 PrevLogNumber() is no longer used, but we pay - // attention to it in case we are recovering a database - // produced by an older version of leveldb. - const uint64_t min_log = versions_->LogNumber(); - const uint64_t prev_log = versions_->PrevLogNumber(); - std::vector filenames; - s = env_->GetChildren(dbname_, &filenames); + // 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 PrevLogNumber() is no longer used, but we pay + // attention to it in case we are recovering a database + // produced by an older version of leveldb. + const uint64_t min_log = versions_->LogNumber(); + const uint64_t prev_log = versions_->PrevLogNumber(); + std::vector filenames; + s = env_->GetChildren(dbname_, &filenames); + if (!s.ok()) { + return s; + } + std::set expected; + versions_->AddLiveFiles(&expected); + uint64_t number; + FileType type; + std::vector logs; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type)) { + expected.erase(number); + if (type == kLogFile && ((number >= min_log) || (number == prev_log))) + logs.push_back(number); + } + } + if (!expected.empty()) { + char buf[50]; + snprintf(buf, sizeof(buf), "%d missing files; e.g.", + static_cast(expected.size())); + return Status::Corruption(buf, TableFileName(dbname_, *(expected.begin()))); + } + + // Recover in the order in which the logs were generated + std::sort(logs.begin(), logs.end()); + for (size_t i = 0; i < logs.size(); i++) { + s = RecoverLogFile(logs[i], (i == logs.size() - 1), save_manifest, edit, + &max_sequence); if (!s.ok()) { return s; } - std::set expected; - versions_->AddLiveFiles(&expected); - uint64_t number; - FileType type; - std::vector logs; - for (size_t i = 0; i < filenames.size(); i++) { - if (ParseFileName(filenames[i], &number, &type)) { - expected.erase(number); - if (type == kLogFile && ((number >= min_log) || (number == prev_log))) - logs.push_back(number); - } - } - if (!expected.empty()) { - char buf[50]; - snprintf(buf, sizeof(buf), "%d missing files; e.g.", - static_cast(expected.size())); - return Status::Corruption(buf, TableFileName(dbname_, *(expected.begin()))); - } - // Recover in the order in which the logs were generated - std::sort(logs.begin(), logs.end()); - for (size_t i = 0; i < logs.size(); i++) { - s = RecoverLogFile(logs[i], edit, &max_sequence); - - // 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_->MarkFileNumberUsed(logs[i]); - } - - if (s.ok()) { - if (versions_->LastSequence() < max_sequence) { - versions_->SetLastSequence(max_sequence); - } - } + // 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_->MarkFileNumberUsed(logs[i]); } - return s; + if (versions_->LastSequence() < max_sequence) { + versions_->SetLastSequence(max_sequence); + } + + return Status::OK(); } -Status DBImpl::RecoverLogFile(uint64_t log_number, - VersionEdit* edit, +Status DBImpl::RecoverLogFile(uint64_t log_number, bool last_log, + bool* save_manifest, VersionEdit* edit, SequenceNumber* max_sequence) { struct LogReporter : public log::Reader::Reporter { Env* env; Logger* info_log; const char* fname; - Status* status; // NULL if options_.paranoid_checks==false - virtual void Corruption(size_t bytes, const Status& s) { + Status* status; // null if options_.paranoid_checks==false + void Corruption(size_t bytes, const Status& s) override { Log(info_log, "%s%s: dropping %d bytes; %s", - (this->status == NULL ? "(ignoring error) " : ""), - fname, static_cast(bytes), s.ToString().c_str()); - if (this->status != NULL && this->status->ok()) *this->status = s; + (this->status == nullptr ? "(ignoring error) " : ""), fname, + static_cast(bytes), s.ToString().c_str()); + if (this->status != nullptr && this->status->ok()) *this->status = s; } }; @@ -402,31 +408,30 @@ Status DBImpl::RecoverLogFile(uint64_t log_number, reporter.env = env_; reporter.info_log = options_.info_log; reporter.fname = fname.c_str(); - reporter.status = (options_.paranoid_checks ? &status : NULL); + reporter.status = (options_.paranoid_checks ? &status : nullptr); // We intentionally 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(file, &reporter, true/*checksum*/, - 0/*initial_offset*/); + log::Reader reader(file, &reporter, true /*checksum*/, 0 /*initial_offset*/); Log(options_.info_log, "Recovering log #%llu", - (unsigned long long) log_number); + (unsigned long long)log_number); // Read all the records and add to a memtable std::string scratch; Slice record; WriteBatch batch; - MemTable* mem = NULL; - while (reader.ReadRecord(&record, &scratch) && - status.ok()) { + int compactions = 0; + MemTable* mem = nullptr; + while (reader.ReadRecord(&record, &scratch) && status.ok()) { if (record.size() < 12) { - reporter.Corruption( - record.size(), Status::Corruption("log record too small")); + reporter.Corruption(record.size(), + Status::Corruption("log record too small")); continue; } WriteBatchInternal::SetContents(&batch, record); - if (mem == NULL) { + if (mem == nullptr) { mem = new MemTable(internal_comparator_); mem->Ref(); } @@ -435,33 +440,59 @@ Status DBImpl::RecoverLogFile(uint64_t log_number, if (!status.ok()) { break; } - const SequenceNumber last_seq = - WriteBatchInternal::Sequence(&batch) + - WriteBatchInternal::Count(&batch) - 1; + const SequenceNumber last_seq = WriteBatchInternal::Sequence(&batch) + + WriteBatchInternal::Count(&batch) - 1; if (last_seq > *max_sequence) { *max_sequence = last_seq; } if (mem->ApproximateMemoryUsage() > options_.write_buffer_size) { - status = WriteLevel0Table(mem, edit, NULL); + compactions++; + *save_manifest = true; + status = WriteLevel0Table(mem, edit, nullptr); + mem->Unref(); + mem = nullptr; if (!status.ok()) { // Reflect errors immediately so that conditions like full // file-systems cause the DB::Open() to fail. break; } - mem->Unref(); - mem = NULL; } } - if (status.ok() && mem != NULL) { - status = WriteLevel0Table(mem, edit, NULL); - // Reflect errors immediately so that conditions like full - // file-systems cause the DB::Open() to fail. + delete file; + + // See if we should keep reusing the last log file. + if (status.ok() && options_.reuse_logs && last_log && compactions == 0) { + assert(logfile_ == nullptr); + assert(log_ == nullptr); + assert(mem_ == nullptr); + uint64_t lfile_size; + if (env_->GetFileSize(fname, &lfile_size).ok() && + env_->NewAppendableFile(fname, &logfile_).ok()) { + Log(options_.info_log, "Reusing old log %s \n", fname.c_str()); + log_ = new log::Writer(logfile_, lfile_size); + logfile_number_ = log_number; + if (mem != nullptr) { + mem_ = mem; + mem = nullptr; + } else { + // mem can be nullptr if lognum exists but was empty. + mem_ = new MemTable(internal_comparator_); + mem_->Ref(); + } + } + } + + if (mem != nullptr) { + // mem did not get reused; compact it. + if (status.ok()) { + *save_manifest = true; + status = WriteLevel0Table(mem, edit, nullptr); + } + mem->Unref(); } - if (mem != NULL) mem->Unref(); - delete file; return status; } @@ -474,7 +505,7 @@ Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit, pending_outputs_.insert(meta.number); Iterator* iter = mem->NewIterator(); Log(options_.info_log, "Level-0 table #%llu: started", - (unsigned long long) meta.number); + (unsigned long long)meta.number); Status s; { @@ -484,24 +515,22 @@ Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit, } Log(options_.info_log, "Level-0 table #%llu: %lld bytes %s", - (unsigned long long) meta.number, - (unsigned long long) meta.file_size, + (unsigned long long)meta.number, (unsigned long long)meta.file_size, s.ToString().c_str()); delete iter; pending_outputs_.erase(meta.number); - // 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.file_size > 0) { const Slice min_user_key = meta.smallest.user_key(); const Slice max_user_key = meta.largest.user_key(); - if (base != NULL) { + if (base != nullptr) { level = base->PickLevelForMemTableOutput(min_user_key, max_user_key); } - edit->AddFile(level, meta.number, meta.file_size, - meta.smallest, meta.largest); + edit->AddFile(level, meta.number, meta.file_size, meta.smallest, + meta.largest); } CompactionStats stats; @@ -513,7 +542,7 @@ Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit, void DBImpl::CompactMemTable() { mutex_.AssertHeld(); - assert(imm_ != NULL); + assert(imm_ != nullptr); // Save the contents of the memtable as a new Table VersionEdit edit; @@ -522,7 +551,7 @@ void DBImpl::CompactMemTable() { Status s = WriteLevel0Table(imm_, &edit, base); base->Unref(); - if (s.ok() && shutting_down_.Acquire_Load()) { + if (s.ok() && shutting_down_.load(std::memory_order_acquire)) { s = Status::IOError("Deleting DB during memtable compaction"); } @@ -536,8 +565,8 @@ void DBImpl::CompactMemTable() { if (s.ok()) { // Commit to the new state imm_->Unref(); - imm_ = NULL; - has_imm_.Release_Store(NULL); + imm_ = nullptr; + has_imm_.store(false, std::memory_order_release); DeleteObsoleteFiles(); } else { RecordBackgroundError(s); @@ -555,13 +584,14 @@ void DBImpl::CompactRange(const Slice* begin, const Slice* end) { } } } - TEST_CompactMemTable(); // TODO(sanjay): Skip if memtable does not overlap + TEST_CompactMemTable(); // TODO(sanjay): Skip if memtable does not overlap for (int level = 0; level < max_level_with_files; level++) { TEST_CompactRange(level, begin, end); } } -void DBImpl::TEST_CompactRange(int level, const Slice* begin,const Slice* end) { +void DBImpl::TEST_CompactRange(int level, const Slice* begin, + const Slice* end) { assert(level >= 0); assert(level + 1 < config::kNumLevels); @@ -570,44 +600,45 @@ void DBImpl::TEST_CompactRange(int level, const Slice* begin,const Slice* end) { ManualCompaction manual; manual.level = level; manual.done = false; - if (begin == NULL) { - manual.begin = NULL; + if (begin == nullptr) { + manual.begin = nullptr; } else { begin_storage = InternalKey(*begin, kMaxSequenceNumber, kValueTypeForSeek); manual.begin = &begin_storage; } - if (end == NULL) { - manual.end = NULL; + if (end == nullptr) { + manual.end = nullptr; } else { end_storage = InternalKey(*end, 0, static_cast(0)); manual.end = &end_storage; } MutexLock l(&mutex_); - while (!manual.done && !shutting_down_.Acquire_Load() && bg_error_.ok()) { - if (manual_compaction_ == NULL) { // Idle + while (!manual.done && !shutting_down_.load(std::memory_order_acquire) && + bg_error_.ok()) { + if (manual_compaction_ == nullptr) { // Idle manual_compaction_ = &manual; MaybeScheduleCompaction(); } else { // Running either my compaction or another compaction. - bg_cv_.Wait(); + background_work_finished_signal_.Wait(); } } if (manual_compaction_ == &manual) { // Cancel my manual compaction since we aborted early for some reason. - manual_compaction_ = NULL; + manual_compaction_ = nullptr; } } Status DBImpl::TEST_CompactMemTable() { - // NULL batch means just wait for earlier writes to be done - Status s = Write(WriteOptions(), NULL); + // nullptr batch means just wait for earlier writes to be done + Status s = Write(WriteOptions(), nullptr); if (s.ok()) { // Wait until the compaction completes MutexLock l(&mutex_); - while (imm_ != NULL && bg_error_.ok()) { - bg_cv_.Wait(); + while (imm_ != nullptr && bg_error_.ok()) { + background_work_finished_signal_.Wait(); } - if (imm_ != NULL) { + if (imm_ != nullptr) { s = bg_error_; } } @@ -618,24 +649,23 @@ void DBImpl::RecordBackgroundError(const Status& s) { mutex_.AssertHeld(); if (bg_error_.ok()) { bg_error_ = s; - bg_cv_.SignalAll(); + background_work_finished_signal_.SignalAll(); } } void DBImpl::MaybeScheduleCompaction() { mutex_.AssertHeld(); - if (bg_compaction_scheduled_) { + if (background_compaction_scheduled_) { // Already scheduled - } else if (shutting_down_.Acquire_Load()) { + } else if (shutting_down_.load(std::memory_order_acquire)) { // DB is being deleted; no more background compactions } else if (!bg_error_.ok()) { // Already got an error; no more changes - } else if (imm_ == NULL && - manual_compaction_ == NULL && + } else if (imm_ == nullptr && manual_compaction_ == nullptr && !versions_->NeedsCompaction()) { // No work to be done } else { - bg_compaction_scheduled_ = true; + background_compaction_scheduled_ = true; env_->Schedule(&DBImpl::BGWork, this); } } @@ -646,8 +676,8 @@ void DBImpl::BGWork(void* db) { void DBImpl::BackgroundCall() { MutexLock l(&mutex_); - assert(bg_compaction_scheduled_); - if (shutting_down_.Acquire_Load()) { + assert(background_compaction_scheduled_); + if (shutting_down_.load(std::memory_order_acquire)) { // No more background work when shutting down. } else if (!bg_error_.ok()) { // No more background work after a background error. @@ -655,36 +685,35 @@ void DBImpl::BackgroundCall() { BackgroundCompaction(); } - bg_compaction_scheduled_ = false; + background_compaction_scheduled_ = false; // Previous compaction may have produced too many files in a level, // so reschedule another compaction if needed. MaybeScheduleCompaction(); - bg_cv_.SignalAll(); + background_work_finished_signal_.SignalAll(); } void DBImpl::BackgroundCompaction() { mutex_.AssertHeld(); - if (imm_ != NULL) { + if (imm_ != nullptr) { CompactMemTable(); return; } Compaction* c; - bool is_manual = (manual_compaction_ != NULL); + bool is_manual = (manual_compaction_ != nullptr); InternalKey manual_end; if (is_manual) { ManualCompaction* m = manual_compaction_; c = versions_->CompactRange(m->level, m->begin, m->end); - m->done = (c == NULL); - if (c != NULL) { + m->done = (c == nullptr); + if (c != nullptr) { manual_end = c->input(0, c->num_input_files(0) - 1)->largest; } Log(options_.info_log, "Manual compaction at level-%d from %s .. %s; will stop at %s\n", - m->level, - (m->begin ? m->begin->DebugString().c_str() : "(begin)"), + m->level, (m->begin ? m->begin->DebugString().c_str() : "(begin)"), (m->end ? m->end->DebugString().c_str() : "(end)"), (m->done ? "(end)" : manual_end.DebugString().c_str())); } else { @@ -692,26 +721,24 @@ void DBImpl::BackgroundCompaction() { } Status status; - if (c == NULL) { + if (c == nullptr) { // Nothing to do } else if (!is_manual && c->IsTrivialMove()) { // Move file to next level assert(c->num_input_files(0) == 1); FileMetaData* f = c->input(0, 0); c->edit()->DeleteFile(c->level(), f->number); - c->edit()->AddFile(c->level() + 1, f->number, f->file_size, - f->smallest, f->largest); + c->edit()->AddFile(c->level() + 1, f->number, f->file_size, f->smallest, + f->largest); status = versions_->LogAndApply(c->edit(), &mutex_); if (!status.ok()) { RecordBackgroundError(status); } VersionSet::LevelSummaryStorage tmp; Log(options_.info_log, "Moved #%lld to level-%d %lld bytes %s: %s\n", - static_cast(f->number), - c->level() + 1, + static_cast(f->number), c->level() + 1, static_cast(f->file_size), - status.ToString().c_str(), - versions_->LevelSummary(&tmp)); + status.ToString().c_str(), versions_->LevelSummary(&tmp)); } else { CompactionState* compact = new CompactionState(c); status = DoCompactionWork(compact); @@ -726,11 +753,10 @@ void DBImpl::BackgroundCompaction() { if (status.ok()) { // Done - } else if (shutting_down_.Acquire_Load()) { + } else if (shutting_down_.load(std::memory_order_acquire)) { // Ignore compaction errors found during shutting down } else { - Log(options_.info_log, - "Compaction error: %s", status.ToString().c_str()); + Log(options_.info_log, "Compaction error: %s", status.ToString().c_str()); } if (is_manual) { @@ -744,18 +770,18 @@ void DBImpl::BackgroundCompaction() { m->tmp_storage = manual_end; m->begin = &m->tmp_storage; } - manual_compaction_ = NULL; + manual_compaction_ = nullptr; } } void DBImpl::CleanupCompaction(CompactionState* compact) { mutex_.AssertHeld(); - if (compact->builder != NULL) { + if (compact->builder != nullptr) { // May happen if we get a shutdown call in the middle of compaction compact->builder->Abandon(); delete compact->builder; } else { - assert(compact->outfile == NULL); + assert(compact->outfile == nullptr); } delete compact->outfile; for (size_t i = 0; i < compact->outputs.size(); i++) { @@ -766,8 +792,8 @@ void DBImpl::CleanupCompaction(CompactionState* compact) { } Status DBImpl::OpenCompactionOutputFile(CompactionState* compact) { - assert(compact != NULL); - assert(compact->builder == NULL); + assert(compact != nullptr); + assert(compact->builder == nullptr); uint64_t file_number; { mutex_.Lock(); @@ -792,9 +818,9 @@ Status DBImpl::OpenCompactionOutputFile(CompactionState* compact) { Status DBImpl::FinishCompactionOutputFile(CompactionState* compact, Iterator* input) { - assert(compact != NULL); - assert(compact->outfile != NULL); - assert(compact->builder != NULL); + assert(compact != nullptr); + assert(compact->outfile != nullptr); + assert(compact->builder != nullptr); const uint64_t output_number = compact->current_output()->number; assert(output_number != 0); @@ -811,7 +837,7 @@ Status DBImpl::FinishCompactionOutputFile(CompactionState* compact, compact->current_output()->file_size = current_bytes; compact->total_bytes += current_bytes; delete compact->builder; - compact->builder = NULL; + compact->builder = nullptr; // Finish and check for file errors if (s.ok()) { @@ -821,34 +847,29 @@ Status DBImpl::FinishCompactionOutputFile(CompactionState* compact, s = compact->outfile->Close(); } delete compact->outfile; - compact->outfile = NULL; + compact->outfile = nullptr; if (s.ok() && current_entries > 0) { // Verify that the table is usable - Iterator* iter = table_cache_->NewIterator(ReadOptions(), - output_number, - current_bytes); + Iterator* iter = + table_cache_->NewIterator(ReadOptions(), output_number, current_bytes); s = iter->status(); delete iter; if (s.ok()) { - Log(options_.info_log, - "Generated table #%llu: %lld keys, %lld bytes", - (unsigned long long) output_number, - (unsigned long long) current_entries, - (unsigned long long) current_bytes); + Log(options_.info_log, "Generated table #%llu@%d: %lld keys, %lld bytes", + (unsigned long long)output_number, compact->compaction->level(), + (unsigned long long)current_entries, + (unsigned long long)current_bytes); } } return s; } - Status DBImpl::InstallCompactionResults(CompactionState* compact) { mutex_.AssertHeld(); - Log(options_.info_log, "Compacted %d@%d + %d@%d files => %lld bytes", - compact->compaction->num_input_files(0), - compact->compaction->level(), - compact->compaction->num_input_files(1), - compact->compaction->level() + 1, + Log(options_.info_log, "Compacted %d@%d + %d@%d files => %lld bytes", + compact->compaction->num_input_files(0), compact->compaction->level(), + compact->compaction->num_input_files(1), compact->compaction->level() + 1, static_cast(compact->total_bytes)); // Add compaction outputs @@ -856,9 +877,8 @@ Status DBImpl::InstallCompactionResults(CompactionState* compact) { const int level = compact->compaction->level(); for (size_t i = 0; i < compact->outputs.size(); i++) { const CompactionState::Output& out = compact->outputs[i]; - compact->compaction->edit()->AddFile( - level + 1, - out.number, out.file_size, out.smallest, out.largest); + compact->compaction->edit()->AddFile(level + 1, out.number, out.file_size, + out.smallest, out.largest); } return versions_->LogAndApply(compact->compaction->edit(), &mutex_); } @@ -867,39 +887,40 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { const uint64_t start_micros = env_->NowMicros(); int64_t imm_micros = 0; // Micros spent doing imm_ compactions - Log(options_.info_log, "Compacting %d@%d + %d@%d files", - compact->compaction->num_input_files(0), - compact->compaction->level(), + Log(options_.info_log, "Compacting %d@%d + %d@%d files", + compact->compaction->num_input_files(0), compact->compaction->level(), compact->compaction->num_input_files(1), compact->compaction->level() + 1); assert(versions_->NumLevelFiles(compact->compaction->level()) > 0); - assert(compact->builder == NULL); - assert(compact->outfile == NULL); + assert(compact->builder == nullptr); + assert(compact->outfile == nullptr); if (snapshots_.empty()) { compact->smallest_snapshot = versions_->LastSequence(); } else { - compact->smallest_snapshot = snapshots_.oldest()->number_; + compact->smallest_snapshot = snapshots_.oldest()->sequence_number(); } + Iterator* input = versions_->MakeInputIterator(compact->compaction); + // Release mutex while we're actually doing the compaction work mutex_.Unlock(); - Iterator* input = versions_->MakeInputIterator(compact->compaction); input->SeekToFirst(); Status status; ParsedInternalKey ikey; std::string current_user_key; bool has_current_user_key = false; SequenceNumber last_sequence_for_key = kMaxSequenceNumber; - for (; input->Valid() && !shutting_down_.Acquire_Load(); ) { + while (input->Valid() && !shutting_down_.load(std::memory_order_acquire)) { // Prioritize immutable compaction work - if (has_imm_.NoBarrier_Load() != NULL) { + if (has_imm_.load(std::memory_order_relaxed)) { const uint64_t imm_start = env_->NowMicros(); mutex_.Lock(); - if (imm_ != NULL) { + if (imm_ != nullptr) { CompactMemTable(); - bg_cv_.SignalAll(); // Wakeup MakeRoomForWrite() if necessary + // Wake up MakeRoomForWrite() if necessary. + background_work_finished_signal_.SignalAll(); } mutex_.Unlock(); imm_micros += (env_->NowMicros() - imm_start); @@ -907,7 +928,7 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { Slice key = input->key(); if (compact->compaction->ShouldStopBefore(key) && - compact->builder != NULL) { + compact->builder != nullptr) { status = FinishCompactionOutputFile(compact, input); if (!status.ok()) { break; @@ -923,8 +944,8 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { last_sequence_for_key = kMaxSequenceNumber; } else { if (!has_current_user_key || - user_comparator()->Compare(ikey.user_key, - Slice(current_user_key)) != 0) { + user_comparator()->Compare(ikey.user_key, Slice(current_user_key)) != + 0) { // First occurrence of this user key current_user_key.assign(ikey.user_key.data(), ikey.user_key.size()); has_current_user_key = true; @@ -933,7 +954,7 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { if (last_sequence_for_key <= compact->smallest_snapshot) { // Hidden by an newer entry for same user key - drop = true; // (A) + drop = true; // (A) } else if (ikey.type == kTypeDeletion && ikey.sequence <= compact->smallest_snapshot && compact->compaction->IsBaseLevelForKey(ikey.user_key)) { @@ -961,7 +982,7 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { if (!drop) { // Open output file if necessary - if (compact->builder == NULL) { + if (compact->builder == nullptr) { status = OpenCompactionOutputFile(compact); if (!status.ok()) { break; @@ -986,17 +1007,17 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { input->Next(); } - if (status.ok() && shutting_down_.Acquire_Load()) { + if (status.ok() && shutting_down_.load(std::memory_order_acquire)) { status = Status::IOError("Deleting DB during compaction"); } - if (status.ok() && compact->builder != NULL) { + if (status.ok() && compact->builder != nullptr) { status = FinishCompactionOutputFile(compact, input); } if (status.ok()) { status = input->status(); } delete input; - input = NULL; + input = nullptr; CompactionStats stats; stats.micros = env_->NowMicros() - start_micros - imm_micros; @@ -1019,34 +1040,37 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) { RecordBackgroundError(status); } VersionSet::LevelSummaryStorage tmp; - Log(options_.info_log, - "compacted to: %s", versions_->LevelSummary(&tmp)); + Log(options_.info_log, "compacted to: %s", versions_->LevelSummary(&tmp)); return status; } namespace { + struct IterState { - port::Mutex* mu; - Version* version; - MemTable* mem; - MemTable* imm; + port::Mutex* const mu; + Version* const version GUARDED_BY(mu); + MemTable* const mem GUARDED_BY(mu); + MemTable* const imm GUARDED_BY(mu); + + IterState(port::Mutex* mutex, MemTable* mem, MemTable* imm, Version* version) + : mu(mutex), version(version), mem(mem), imm(imm) {} }; static void CleanupIteratorState(void* arg1, void* arg2) { IterState* state = reinterpret_cast(arg1); state->mu->Lock(); state->mem->Unref(); - if (state->imm != NULL) state->imm->Unref(); + if (state->imm != nullptr) state->imm->Unref(); state->version->Unref(); state->mu->Unlock(); delete state; } -} // namespace + +} // anonymous namespace Iterator* DBImpl::NewInternalIterator(const ReadOptions& options, SequenceNumber* latest_snapshot, uint32_t* seed) { - IterState* cleanup = new IterState; mutex_.Lock(); *latest_snapshot = versions_->LastSequence(); @@ -1054,7 +1078,7 @@ Iterator* DBImpl::NewInternalIterator(const ReadOptions& options, std::vector list; list.push_back(mem_->NewIterator()); mem_->Ref(); - if (imm_ != NULL) { + if (imm_ != nullptr) { list.push_back(imm_->NewIterator()); imm_->Ref(); } @@ -1063,11 +1087,8 @@ Iterator* DBImpl::NewInternalIterator(const ReadOptions& options, NewMergingIterator(&internal_comparator_, &list[0], list.size()); versions_->current()->Ref(); - cleanup->mu = &mutex_; - cleanup->mem = mem_; - cleanup->imm = imm_; - cleanup->version = versions_->current(); - internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, NULL); + IterState* cleanup = new IterState(&mutex_, mem_, imm_, versions_->current()); + internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, nullptr); *seed = ++seed_; mutex_.Unlock(); @@ -1085,14 +1106,14 @@ int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes() { return versions_->MaxNextLevelOverlappingBytes(); } -Status DBImpl::Get(const ReadOptions& options, - const Slice& key, +Status DBImpl::Get(const ReadOptions& options, const Slice& key, std::string* value) { Status s; MutexLock l(&mutex_); SequenceNumber snapshot; - if (options.snapshot != NULL) { - snapshot = reinterpret_cast(options.snapshot)->number_; + if (options.snapshot != nullptr) { + snapshot = + static_cast(options.snapshot)->sequence_number(); } else { snapshot = versions_->LastSequence(); } @@ -1101,7 +1122,7 @@ Status DBImpl::Get(const ReadOptions& options, MemTable* imm = imm_; Version* current = versions_->current(); mem->Ref(); - if (imm != NULL) imm->Ref(); + if (imm != nullptr) imm->Ref(); current->Ref(); bool have_stat_update = false; @@ -1114,7 +1135,7 @@ Status DBImpl::Get(const ReadOptions& options, LookupKey lkey(key, snapshot); if (mem->Get(lkey, value, &s)) { // Done - } else if (imm != NULL && imm->Get(lkey, value, &s)) { + } else if (imm != nullptr && imm->Get(lkey, value, &s)) { // Done } else { s = current->Get(options, lkey, value, &stats); @@ -1127,7 +1148,7 @@ Status DBImpl::Get(const ReadOptions& options, MaybeScheduleCompaction(); } mem->Unref(); - if (imm != NULL) imm->Unref(); + if (imm != nullptr) imm->Unref(); current->Unref(); return s; } @@ -1136,12 +1157,12 @@ Iterator* DBImpl::NewIterator(const ReadOptions& options) { SequenceNumber latest_snapshot; uint32_t seed; Iterator* iter = NewInternalIterator(options, &latest_snapshot, &seed); - return NewDBIterator( - this, user_comparator(), iter, - (options.snapshot != NULL - ? reinterpret_cast(options.snapshot)->number_ - : latest_snapshot), - seed); + return NewDBIterator(this, user_comparator(), iter, + (options.snapshot != nullptr + ? static_cast(options.snapshot) + ->sequence_number() + : latest_snapshot), + seed); } void DBImpl::RecordReadSample(Slice key) { @@ -1156,9 +1177,9 @@ const Snapshot* DBImpl::GetSnapshot() { return snapshots_.New(versions_->LastSequence()); } -void DBImpl::ReleaseSnapshot(const Snapshot* s) { +void DBImpl::ReleaseSnapshot(const Snapshot* snapshot) { MutexLock l(&mutex_); - snapshots_.Delete(reinterpret_cast(s)); + snapshots_.Delete(static_cast(snapshot)); } // Convenience methods @@ -1170,9 +1191,9 @@ Status DBImpl::Delete(const WriteOptions& options, const Slice& key) { return DB::Delete(options, key); } -Status DBImpl::Write(const WriteOptions& options, WriteBatch* my_batch) { +Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) { Writer w(&mutex_); - w.batch = my_batch; + w.batch = updates; w.sync = options.sync; w.done = false; @@ -1186,10 +1207,10 @@ Status DBImpl::Write(const WriteOptions& options, WriteBatch* my_batch) { } // May temporarily unlock and wait. - Status status = MakeRoomForWrite(my_batch == NULL); + Status status = MakeRoomForWrite(updates == nullptr); uint64_t last_sequence = versions_->LastSequence(); Writer* last_writer = &w; - if (status.ok() && my_batch != NULL) { // NULL batch is for compactions + if (status.ok() && updates != nullptr) { // nullptr batch is for compactions WriteBatch* updates = BuildBatchGroup(&last_writer); WriteBatchInternal::SetSequence(updates, last_sequence + 1); last_sequence += WriteBatchInternal::Count(updates); @@ -1244,12 +1265,13 @@ Status DBImpl::Write(const WriteOptions& options, WriteBatch* my_batch) { } // REQUIRES: Writer list must be non-empty -// REQUIRES: First writer must have a non-NULL batch +// REQUIRES: First writer must have a non-null batch WriteBatch* DBImpl::BuildBatchGroup(Writer** last_writer) { + mutex_.AssertHeld(); assert(!writers_.empty()); Writer* first = writers_.front(); WriteBatch* result = first->batch; - assert(result != NULL); + assert(result != nullptr); size_t size = WriteBatchInternal::ByteSize(first->batch); @@ -1257,8 +1279,8 @@ WriteBatch* DBImpl::BuildBatchGroup(Writer** last_writer) { // 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); + if (size <= (128 << 10)) { + max_size = size + (128 << 10); } *last_writer = first; @@ -1271,7 +1293,7 @@ WriteBatch* DBImpl::BuildBatchGroup(Writer** last_writer) { break; } - if (w->batch != NULL) { + if (w->batch != nullptr) { size += WriteBatchInternal::ByteSize(w->batch); if (size > max_size) { // Do not make batch too big @@ -1304,9 +1326,8 @@ Status DBImpl::MakeRoomForWrite(bool force) { // Yield previous error s = bg_error_; break; - } else if ( - allow_delay && - versions_->NumLevelFiles(0) >= config::kL0_SlowdownWritesTrigger) { + } else if (allow_delay && versions_->NumLevelFiles(0) >= + config::kL0_SlowdownWritesTrigger) { // We are getting close to hitting a hard limit on the number of // L0 files. Rather than delaying a single write by several // seconds when we hit the hard limit, start delaying each @@ -1321,20 +1342,20 @@ Status DBImpl::MakeRoomForWrite(bool force) { (mem_->ApproximateMemoryUsage() <= options_.write_buffer_size)) { // There is room in current memtable break; - } else if (imm_ != NULL) { + } else if (imm_ != nullptr) { // We have filled up the current memtable, but the previous // one is still being compacted, so we wait. Log(options_.info_log, "Current memtable full; waiting...\n"); - bg_cv_.Wait(); + background_work_finished_signal_.Wait(); } else if (versions_->NumLevelFiles(0) >= config::kL0_StopWritesTrigger) { // There are too many level-0 files. Log(options_.info_log, "Too many L0 files; waiting...\n"); - bg_cv_.Wait(); + background_work_finished_signal_.Wait(); } else { // Attempt to switch to a new memtable and trigger compaction of old assert(versions_->PrevLogNumber() == 0); uint64_t new_log_number = versions_->NewFileNumber(); - WritableFile* lfile = NULL; + WritableFile* lfile = nullptr; s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile); if (!s.ok()) { // Avoid chewing through file number space in a tight loop. @@ -1347,10 +1368,10 @@ Status DBImpl::MakeRoomForWrite(bool force) { logfile_number_ = new_log_number; log_ = new log::Writer(lfile); imm_ = mem_; - has_imm_.Release_Store(imm_); + has_imm_.store(true, std::memory_order_release); mem_ = new MemTable(internal_comparator_); mem_->Ref(); - force = false; // Do not force another compaction if have room + force = false; // Do not force another compaction if have room MaybeScheduleCompaction(); } } @@ -1384,21 +1405,16 @@ bool DBImpl::GetProperty(const Slice& property, std::string* value) { snprintf(buf, sizeof(buf), " Compactions\n" "Level Files Size(MB) Time(sec) Read(MB) Write(MB)\n" - "--------------------------------------------------\n" - ); + "--------------------------------------------------\n"); value->append(buf); for (int level = 0; level < config::kNumLevels; level++) { int files = versions_->NumLevelFiles(level); if (stats_[level].micros > 0 || files > 0) { - snprintf( - buf, sizeof(buf), - "%3d %8d %8.0f %9.0f %8.0f %9.0f\n", - level, - files, - versions_->NumLevelBytes(level) / 1048576.0, - stats_[level].micros / 1e6, - stats_[level].bytes_read / 1048576.0, - stats_[level].bytes_written / 1048576.0); + snprintf(buf, sizeof(buf), "%3d %8d %8.0f %9.0f %8.0f %9.0f\n", level, + files, versions_->NumLevelBytes(level) / 1048576.0, + stats_[level].micros / 1e6, + stats_[level].bytes_read / 1048576.0, + stats_[level].bytes_written / 1048576.0); value->append(buf); } } @@ -1406,21 +1422,29 @@ bool DBImpl::GetProperty(const Slice& property, std::string* value) { } else if (in == "sstables") { *value = versions_->current()->DebugString(); return true; + } else if (in == "approximate-memory-usage") { + size_t total_usage = options_.block_cache->TotalCharge(); + if (mem_) { + total_usage += mem_->ApproximateMemoryUsage(); + } + if (imm_) { + total_usage += imm_->ApproximateMemoryUsage(); + } + char buf[50]; + snprintf(buf, sizeof(buf), "%llu", + static_cast(total_usage)); + value->append(buf); + return true; } return false; } -void DBImpl::GetApproximateSizes( - const Range* range, int n, - uint64_t* sizes) { +void DBImpl::GetApproximateSizes(const Range* range, int n, uint64_t* sizes) { // TODO(opt): better implementation - Version* v; - { - MutexLock l(&mutex_); - versions_->current()->Ref(); - v = versions_->current(); - } + MutexLock l(&mutex_); + Version* v = versions_->current(); + v->Ref(); for (int i = 0; i < n; i++) { // Convert user_key into a corresponding internal key. @@ -1431,10 +1455,7 @@ void DBImpl::GetApproximateSizes( sizes[i] = (limit >= start ? limit - start : 0); } - { - MutexLock l(&mutex_); - v->Unref(); - } + v->Unref(); } // Default implementations of convenience methods that subclasses of DB @@ -1451,17 +1472,19 @@ Status DB::Delete(const WriteOptions& opt, const Slice& key) { return Write(opt, &batch); } -DB::~DB() { } +DB::~DB() = default; -Status DB::Open(const Options& options, const std::string& dbname, - DB** dbptr) { - *dbptr = NULL; +Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) { + *dbptr = nullptr; DBImpl* impl = new DBImpl(options, dbname); impl->mutex_.Lock(); VersionEdit edit; - Status s = impl->Recover(&edit); // Handles create_if_missing, error_if_exists - if (s.ok()) { + // Recover handles create_if_missing, error_if_exists + bool save_manifest = false; + Status s = impl->Recover(&edit, &save_manifest); + if (s.ok() && impl->mem_ == nullptr) { + // Create new log and a corresponding memtable. uint64_t new_log_number = impl->versions_->NewFileNumber(); WritableFile* lfile; s = options.env->NewWritableFile(LogFileName(dbname, new_log_number), @@ -1471,15 +1494,22 @@ Status DB::Open(const Options& options, const std::string& dbname, impl->logfile_ = lfile; impl->logfile_number_ = new_log_number; impl->log_ = new log::Writer(lfile); - s = impl->versions_->LogAndApply(&edit, &impl->mutex_); - } - if (s.ok()) { - impl->DeleteObsoleteFiles(); - impl->MaybeScheduleCompaction(); + impl->mem_ = new MemTable(impl->internal_comparator_); + impl->mem_->Ref(); } } + if (s.ok() && save_manifest) { + edit.SetPrevLogNumber(0); // No older logs needed after recovery. + edit.SetLogNumber(impl->logfile_number_); + s = impl->versions_->LogAndApply(&edit, &impl->mutex_); + } + if (s.ok()) { + impl->DeleteObsoleteFiles(); + impl->MaybeScheduleCompaction(); + } impl->mutex_.Unlock(); if (s.ok()) { + assert(impl->mem_ != nullptr); *dbptr = impl; } else { delete impl; @@ -1487,21 +1517,20 @@ Status DB::Open(const Options& options, const std::string& dbname, return s; } -Snapshot::~Snapshot() { -} +Snapshot::~Snapshot() = default; Status DestroyDB(const std::string& dbname, const Options& options) { Env* env = options.env; std::vector filenames; - // Ignore error in case directory does not exist - env->GetChildren(dbname, &filenames); - if (filenames.empty()) { + Status result = env->GetChildren(dbname, &filenames); + if (!result.ok()) { + // Ignore error in case directory does not exist return Status::OK(); } FileLock* lock; const std::string lockname = LockFileName(dbname); - Status result = env->LockFile(lockname, &lock); + result = env->LockFile(lockname, &lock); if (result.ok()) { uint64_t number; FileType type; diff --git a/db/db_impl.h b/db/db_impl.h index cfc9981..685735c 100644 --- a/db/db_impl.h +++ b/db/db_impl.h @@ -5,8 +5,11 @@ #ifndef STORAGE_LEVELDB_DB_DB_IMPL_H_ #define STORAGE_LEVELDB_DB_DB_IMPL_H_ +#include #include #include +#include + #include "db/dbformat.h" #include "db/log_writer.h" #include "db/snapshot.h" @@ -26,21 +29,25 @@ class VersionSet; class DBImpl : public DB { public: DBImpl(const Options& options, const std::string& dbname); - virtual ~DBImpl(); + + DBImpl(const DBImpl&) = delete; + DBImpl& operator=(const DBImpl&) = delete; + + ~DBImpl() override; // Implementations of the DB interface - virtual Status Put(const WriteOptions&, const Slice& key, const Slice& value); - virtual Status Delete(const WriteOptions&, const Slice& key); - virtual Status Write(const WriteOptions& options, WriteBatch* updates); - virtual Status Get(const ReadOptions& options, - const Slice& key, - std::string* value); - virtual Iterator* NewIterator(const ReadOptions&); - virtual const Snapshot* GetSnapshot(); - virtual void ReleaseSnapshot(const Snapshot* snapshot); - virtual bool GetProperty(const Slice& property, std::string* value); - virtual void GetApproximateSizes(const Range* range, int n, uint64_t* sizes); - virtual void CompactRange(const Slice* begin, const Slice* end); + Status Put(const WriteOptions&, const Slice& key, + const Slice& value) override; + Status Delete(const WriteOptions&, const Slice& key) override; + Status Write(const WriteOptions& options, WriteBatch* updates) override; + Status Get(const ReadOptions& options, const Slice& key, + std::string* value) override; + Iterator* NewIterator(const ReadOptions&) override; + const Snapshot* GetSnapshot() override; + void ReleaseSnapshot(const Snapshot* snapshot) override; + bool GetProperty(const Slice& property, std::string* value) override; + void GetApproximateSizes(const Range* range, int n, uint64_t* sizes) override; + void CompactRange(const Slice* begin, const Slice* end) override; // Extra methods (for testing) that are not in the public DB interface @@ -69,6 +76,31 @@ class DBImpl : public DB { struct CompactionState; struct Writer; + // Information for a manual compaction + struct ManualCompaction { + int level; + bool done; + const InternalKey* begin; // null means beginning of key range + const InternalKey* end; // null means end of key range + InternalKey tmp_storage; // Used to keep track of compaction progress + }; + + // Per level compaction stats. stats_[level] stores the stats for + // compactions that produced data for the specified "level". + struct CompactionStats { + CompactionStats() : micros(0), bytes_read(0), bytes_written(0) {} + + void Add(const CompactionStats& c) { + this->micros += c.micros; + this->bytes_read += c.bytes_read; + this->bytes_written += c.bytes_written; + } + + int64_t micros; + int64_t bytes_read; + int64_t bytes_written; + }; + Iterator* NewInternalIterator(const ReadOptions&, SequenceNumber* latest_snapshot, uint32_t* seed); @@ -78,21 +110,21 @@ class DBImpl : public DB { // 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(VersionEdit* edit) EXCLUSIVE_LOCKS_REQUIRED(mutex_); + Status Recover(VersionEdit* edit, bool* save_manifest) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); void MaybeIgnoreError(Status* s) const; // Delete any unneeded files and stale in-memory entries. - void DeleteObsoleteFiles(); + void DeleteObsoleteFiles() EXCLUSIVE_LOCKS_REQUIRED(mutex_); // Compact the in-memory write buffer to disk. Switches to a new // log-file/memtable and writes a new descriptor iff successful. // Errors are recorded in bg_error_. void CompactMemTable() EXCLUSIVE_LOCKS_REQUIRED(mutex_); - Status RecoverLogFile(uint64_t log_number, - VersionEdit* edit, - SequenceNumber* max_sequence) + Status RecoverLogFile(uint64_t log_number, bool last_log, bool* save_manifest, + VersionEdit* edit, SequenceNumber* max_sequence) EXCLUSIVE_LOCKS_REQUIRED(mutex_); Status WriteLevel0Table(MemTable* mem, VersionEdit* edit, Version* base) @@ -100,14 +132,15 @@ class DBImpl : public DB { Status MakeRoomForWrite(bool force /* compact even if there is room? */) EXCLUSIVE_LOCKS_REQUIRED(mutex_); - WriteBatch* BuildBatchGroup(Writer** last_writer); + WriteBatch* BuildBatchGroup(Writer** last_writer) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); void RecordBackgroundError(const Status& s); void MaybeScheduleCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_); static void BGWork(void* db); void BackgroundCall(); - void BackgroundCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_); + void BackgroundCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_); void CleanupCompaction(CompactionState* compact) EXCLUSIVE_LOCKS_REQUIRED(mutex_); Status DoCompactionWork(CompactionState* compact) @@ -118,93 +151,66 @@ class DBImpl : public DB { Status InstallCompactionResults(CompactionState* compact) EXCLUSIVE_LOCKS_REQUIRED(mutex_); + const Comparator* user_comparator() const { + return internal_comparator_.user_comparator(); + } + // Constant after construction Env* const env_; const InternalKeyComparator internal_comparator_; const InternalFilterPolicy internal_filter_policy_; const Options options_; // options_.comparator == &internal_comparator_ - bool owns_info_log_; - bool owns_cache_; + const bool owns_info_log_; + const bool owns_cache_; const std::string dbname_; // table_cache_ provides its own synchronization - TableCache* table_cache_; + TableCache* const table_cache_; - // Lock over the persistent DB state. Non-NULL iff successfully acquired. + // Lock over the persistent DB state. Non-null iff successfully acquired. FileLock* db_lock_; // State below is protected by mutex_ port::Mutex mutex_; - port::AtomicPointer shutting_down_; - port::CondVar bg_cv_; // Signalled when background work finishes + std::atomic shutting_down_; + port::CondVar background_work_finished_signal_ GUARDED_BY(mutex_); MemTable* mem_; - MemTable* imm_; // Memtable being compacted - port::AtomicPointer has_imm_; // So bg thread can detect non-NULL imm_ + MemTable* imm_ GUARDED_BY(mutex_); // Memtable being compacted + std::atomic has_imm_; // So bg thread can detect non-null imm_ WritableFile* logfile_; - uint64_t logfile_number_; + uint64_t logfile_number_ GUARDED_BY(mutex_); log::Writer* log_; - uint32_t seed_; // For sampling. + uint32_t seed_ GUARDED_BY(mutex_); // For sampling. // Queue of writers. - std::deque writers_; - WriteBatch* tmp_batch_; + std::deque writers_ GUARDED_BY(mutex_); + WriteBatch* tmp_batch_ GUARDED_BY(mutex_); - SnapshotList snapshots_; + SnapshotList snapshots_ GUARDED_BY(mutex_); // Set of table files to protect from deletion because they are // part of ongoing compactions. - std::set pending_outputs_; + std::set pending_outputs_ GUARDED_BY(mutex_); // Has a background compaction been scheduled or is running? - bool bg_compaction_scheduled_; + bool background_compaction_scheduled_ GUARDED_BY(mutex_); - // Information for a manual compaction - struct ManualCompaction { - int level; - bool done; - const InternalKey* begin; // NULL means beginning of key range - const InternalKey* end; // NULL means end of key range - InternalKey tmp_storage; // Used to keep track of compaction progress - }; - ManualCompaction* manual_compaction_; + ManualCompaction* manual_compaction_ GUARDED_BY(mutex_); - VersionSet* versions_; + VersionSet* const versions_ GUARDED_BY(mutex_); // Have we encountered a background error in paranoid mode? - Status bg_error_; + Status bg_error_ GUARDED_BY(mutex_); - // Per level compaction stats. stats_[level] stores the stats for - // compactions that produced data for the specified "level". - struct CompactionStats { - int64_t micros; - int64_t bytes_read; - int64_t bytes_written; - - CompactionStats() : micros(0), bytes_read(0), bytes_written(0) { } - - void Add(const CompactionStats& c) { - this->micros += c.micros; - this->bytes_read += c.bytes_read; - this->bytes_written += c.bytes_written; - } - }; - CompactionStats stats_[config::kNumLevels]; - - // No copying allowed - DBImpl(const DBImpl&); - void operator=(const DBImpl&); - - const Comparator* user_comparator() const { - return internal_comparator_.user_comparator(); - } + CompactionStats stats_[config::kNumLevels] GUARDED_BY(mutex_); }; // 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 InternalFilterPolicy* ipolicy, - const Options& src); +Options SanitizeOptions(const std::string& db, + const InternalKeyComparator* icmp, + const InternalFilterPolicy* ipolicy, + const Options& src); } // namespace leveldb diff --git a/db/db_iter.cc b/db/db_iter.cc index 3b2035e..98715a9 100644 --- a/db/db_iter.cc +++ b/db/db_iter.cc @@ -4,9 +4,9 @@ #include "db/db_iter.h" -#include "db/filename.h" #include "db/db_impl.h" #include "db/dbformat.h" +#include "db/filename.h" #include "leveldb/env.h" #include "leveldb/iterator.h" #include "port/port.h" @@ -36,17 +36,14 @@ namespace { // 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 { +class DBIter : public Iterator { public: // 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 - }; + enum Direction { kForward, kReverse }; DBIter(DBImpl* db, const Comparator* cmp, Iterator* iter, SequenceNumber s, uint32_t seed) @@ -57,21 +54,22 @@ class DBIter: public Iterator { direction_(kForward), valid_(false), rnd_(seed), - bytes_counter_(RandomPeriod()) { - } - virtual ~DBIter() { - delete iter_; - } - virtual bool Valid() const { return valid_; } - virtual Slice key() const { + bytes_until_read_sampling_(RandomCompactionPeriod()) {} + + DBIter(const DBIter&) = delete; + DBIter& operator=(const DBIter&) = delete; + + ~DBIter() override { delete iter_; } + bool Valid() const override { return valid_; } + Slice key() const override { assert(valid_); return (direction_ == kForward) ? ExtractUserKey(iter_->key()) : saved_key_; } - virtual Slice value() const { + Slice value() const override { assert(valid_); return (direction_ == kForward) ? iter_->value() : saved_value_; } - virtual Status status() const { + Status status() const override { if (status_.ok()) { return iter_->status(); } else { @@ -79,11 +77,11 @@ class DBIter: public Iterator { } } - virtual void Next(); - virtual void Prev(); - virtual void Seek(const Slice& target); - virtual void SeekToFirst(); - virtual void SeekToLast(); + void Next() override; + void Prev() override; + void Seek(const Slice& target) override; + void SeekToFirst() override; + void SeekToLast() override; private: void FindNextUserEntry(bool skipping, std::string* skip); @@ -103,38 +101,35 @@ class DBIter: public Iterator { } } - // Pick next gap with average value of config::kReadBytesPeriod. - ssize_t RandomPeriod() { - return rnd_.Uniform(2*config::kReadBytesPeriod); + // Picks the number of bytes that can be read until a compaction is scheduled. + size_t RandomCompactionPeriod() { + return rnd_.Uniform(2 * config::kReadBytesPeriod); } DBImpl* db_; const Comparator* const user_comparator_; Iterator* const iter_; SequenceNumber const sequence_; - Status status_; - std::string saved_key_; // == current key when direction_==kReverse - std::string saved_value_; // == current raw value when direction_==kReverse + std::string saved_key_; // == current key when direction_==kReverse + std::string saved_value_; // == current raw value when direction_==kReverse Direction direction_; bool valid_; - Random rnd_; - ssize_t bytes_counter_; - - // No copying allowed - DBIter(const DBIter&); - void operator=(const DBIter&); + size_t bytes_until_read_sampling_; }; inline bool DBIter::ParseKey(ParsedInternalKey* ikey) { Slice k = iter_->key(); - ssize_t n = k.size() + iter_->value().size(); - bytes_counter_ -= n; - while (bytes_counter_ < 0) { - bytes_counter_ += RandomPeriod(); + + size_t bytes_read = k.size() + iter_->value().size(); + while (bytes_until_read_sampling_ < bytes_read) { + bytes_until_read_sampling_ += RandomCompactionPeriod(); db_->RecordReadSample(k); } + assert(bytes_until_read_sampling_ >= bytes_read); + bytes_until_read_sampling_ -= bytes_read; + if (!ParseInternalKey(k, ikey)) { status_ = Status::Corruption("corrupted internal key in DBIter"); return false; @@ -165,6 +160,15 @@ void DBIter::Next() { } else { // Store in saved_key_ the current key so we skip it below. SaveKey(ExtractUserKey(iter_->key()), &saved_key_); + + // iter_ is pointing to current key. We can now safely move to the next to + // avoid checking current key. + iter_->Next(); + if (!iter_->Valid()) { + valid_ = false; + saved_key_.clear(); + return; + } } FindNextUserEntry(true, &saved_key_); @@ -218,8 +222,8 @@ void DBIter::Prev() { ClearSavedValue(); return; } - if (user_comparator_->Compare(ExtractUserKey(iter_->key()), - saved_key_) < 0) { + if (user_comparator_->Compare(ExtractUserKey(iter_->key()), saved_key_) < + 0) { break; } } @@ -275,8 +279,8 @@ void DBIter::Seek(const Slice& target) { direction_ = kForward; ClearSavedValue(); saved_key_.clear(); - AppendInternalKey( - &saved_key_, ParsedInternalKey(target, sequence_, kValueTypeForSeek)); + AppendInternalKey(&saved_key_, + ParsedInternalKey(target, sequence_, kValueTypeForSeek)); iter_->Seek(saved_key_); if (iter_->Valid()) { FindNextUserEntry(false, &saved_key_ /* temporary storage */); @@ -305,12 +309,9 @@ void DBIter::SeekToLast() { } // anonymous namespace -Iterator* NewDBIterator( - DBImpl* db, - const Comparator* user_key_comparator, - Iterator* internal_iter, - SequenceNumber sequence, - uint32_t seed) { +Iterator* NewDBIterator(DBImpl* db, const Comparator* user_key_comparator, + Iterator* internal_iter, SequenceNumber sequence, + uint32_t seed) { return new DBIter(db, user_key_comparator, internal_iter, sequence, seed); } diff --git a/db/db_iter.h b/db/db_iter.h index 04927e9..fd93e91 100644 --- a/db/db_iter.h +++ b/db/db_iter.h @@ -6,8 +6,9 @@ #define STORAGE_LEVELDB_DB_DB_ITER_H_ #include -#include "leveldb/db.h" + #include "db/dbformat.h" +#include "leveldb/db.h" namespace leveldb { @@ -16,12 +17,9 @@ class DBImpl; // 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( - DBImpl* db, - const Comparator* user_key_comparator, - Iterator* internal_iter, - SequenceNumber sequence, - uint32_t seed); +Iterator* NewDBIterator(DBImpl* db, const Comparator* user_key_comparator, + Iterator* internal_iter, SequenceNumber sequence, + uint32_t seed); } // namespace leveldb diff --git a/db/db_test.cc b/db/db_test.cc index 0fed913..9a8faf1 100644 --- a/db/db_test.cc +++ b/db/db_test.cc @@ -3,14 +3,20 @@ // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "leveldb/db.h" -#include "leveldb/filter_policy.h" + +#include +#include + #include "db/db_impl.h" #include "db/filename.h" #include "db/version_set.h" #include "db/write_batch_internal.h" #include "leveldb/cache.h" #include "leveldb/env.h" +#include "leveldb/filter_policy.h" #include "leveldb/table.h" +#include "port/port.h" +#include "port/thread_annotations.h" #include "util/hash.h" #include "util/logging.h" #include "util/mutexlock.h" @@ -25,83 +31,116 @@ static std::string RandomString(Random* rnd, int len) { return r; } +static std::string RandomKey(Random* rnd) { + int len = + (rnd->OneIn(3) ? 1 // Short sometimes to encourage collisions + : (rnd->OneIn(100) ? rnd->Skewed(10) : rnd->Uniform(10))); + return test::RandomKey(rnd, len); +} + namespace { class AtomicCounter { - private: - port::Mutex mu_; - int count_; public: - AtomicCounter() : count_(0) { } - void Increment() { - IncrementBy(1); - } - void IncrementBy(int count) { + AtomicCounter() : count_(0) {} + void Increment() { IncrementBy(1); } + void IncrementBy(int count) LOCKS_EXCLUDED(mu_) { MutexLock l(&mu_); count_ += count; } - int Read() { + int Read() LOCKS_EXCLUDED(mu_) { MutexLock l(&mu_); return count_; } - void Reset() { + void Reset() LOCKS_EXCLUDED(mu_) { MutexLock l(&mu_); count_ = 0; } + + private: + port::Mutex mu_; + int count_ GUARDED_BY(mu_); }; void DelayMilliseconds(int millis) { Env::Default()->SleepForMicroseconds(millis * 1000); } -} +} // namespace -// Special Env used to delay background operations +// Test Env to override default Env behavior for testing. +class TestEnv : public EnvWrapper { + public: + explicit TestEnv(Env* base) : EnvWrapper(base), ignore_dot_files_(false) {} + + void SetIgnoreDotFiles(bool ignored) { ignore_dot_files_ = ignored; } + + Status GetChildren(const std::string& dir, + std::vector* result) override { + Status s = target()->GetChildren(dir, result); + if (!s.ok() || !ignore_dot_files_) { + return s; + } + + std::vector::iterator it = result->begin(); + while (it != result->end()) { + if ((*it == ".") || (*it == "..")) { + it = result->erase(it); + } else { + ++it; + } + } + + return s; + } + + private: + bool ignore_dot_files_; +}; + +// Special Env used to delay background operations. class SpecialEnv : public EnvWrapper { public: - // sstable/log Sync() calls are blocked while this pointer is non-NULL. - port::AtomicPointer delay_data_sync_; + // sstable/log Sync() calls are blocked while this pointer is non-null. + std::atomic delay_data_sync_; // sstable/log Sync() calls return an error. - port::AtomicPointer data_sync_error_; + std::atomic data_sync_error_; - // Simulate no-space errors while this pointer is non-NULL. - port::AtomicPointer no_space_; + // Simulate no-space errors while this pointer is non-null. + std::atomic no_space_; - // Simulate non-writable file system while this pointer is non-NULL - port::AtomicPointer non_writable_; + // Simulate non-writable file system while this pointer is non-null. + std::atomic non_writable_; - // Force sync of manifest files to fail while this pointer is non-NULL - port::AtomicPointer manifest_sync_error_; + // Force sync of manifest files to fail while this pointer is non-null. + std::atomic manifest_sync_error_; - // Force write to manifest files to fail while this pointer is non-NULL - port::AtomicPointer manifest_write_error_; + // Force write to manifest files to fail while this pointer is non-null. + std::atomic manifest_write_error_; bool count_random_reads_; AtomicCounter random_read_counter_; - explicit SpecialEnv(Env* base) : EnvWrapper(base) { - delay_data_sync_.Release_Store(NULL); - data_sync_error_.Release_Store(NULL); - no_space_.Release_Store(NULL); - non_writable_.Release_Store(NULL); - count_random_reads_ = false; - manifest_sync_error_.Release_Store(NULL); - manifest_write_error_.Release_Store(NULL); - } + explicit SpecialEnv(Env* base) + : EnvWrapper(base), + delay_data_sync_(false), + data_sync_error_(false), + no_space_(false), + non_writable_(false), + manifest_sync_error_(false), + manifest_write_error_(false), + count_random_reads_(false) {} Status NewWritableFile(const std::string& f, WritableFile** r) { class DataFile : public WritableFile { private: - SpecialEnv* env_; - WritableFile* base_; + SpecialEnv* const env_; + WritableFile* const base_; public: - DataFile(SpecialEnv* env, WritableFile* base) - : env_(env), - base_(base) { - } + DataFile(SpecialEnv* env, WritableFile* base) : env_(env), base_(base) {} ~DataFile() { delete base_; } Status Append(const Slice& data) { - if (env_->no_space_.Acquire_Load() != NULL) { + if (env_->no_space_.load(std::memory_order_acquire)) { // Drop writes on the floor return Status::OK(); } else { @@ -111,10 +150,10 @@ class SpecialEnv : public EnvWrapper { Status Close() { return base_->Close(); } Status Flush() { return base_->Flush(); } Status Sync() { - if (env_->data_sync_error_.Acquire_Load() != NULL) { + if (env_->data_sync_error_.load(std::memory_order_acquire)) { return Status::IOError("simulated data sync error"); } - while (env_->delay_data_sync_.Acquire_Load() != NULL) { + while (env_->delay_data_sync_.load(std::memory_order_acquire)) { DelayMilliseconds(100); } return base_->Sync(); @@ -124,11 +163,12 @@ class SpecialEnv : public EnvWrapper { private: SpecialEnv* env_; WritableFile* base_; + public: - ManifestFile(SpecialEnv* env, WritableFile* b) : env_(env), base_(b) { } + ManifestFile(SpecialEnv* env, WritableFile* b) : env_(env), base_(b) {} ~ManifestFile() { delete base_; } Status Append(const Slice& data) { - if (env_->manifest_write_error_.Acquire_Load() != NULL) { + if (env_->manifest_write_error_.load(std::memory_order_acquire)) { return Status::IOError("simulated writer error"); } else { return base_->Append(data); @@ -137,7 +177,7 @@ class SpecialEnv : public EnvWrapper { Status Close() { return base_->Close(); } Status Flush() { return base_->Flush(); } Status Sync() { - if (env_->manifest_sync_error_.Acquire_Load() != NULL) { + if (env_->manifest_sync_error_.load(std::memory_order_acquire)) { return Status::IOError("simulated sync error"); } else { return base_->Sync(); @@ -145,16 +185,16 @@ class SpecialEnv : public EnvWrapper { } }; - if (non_writable_.Acquire_Load() != NULL) { + if (non_writable_.load(std::memory_order_acquire)) { return Status::IOError("simulated write error"); } Status s = target()->NewWritableFile(f, r); if (s.ok()) { - if (strstr(f.c_str(), ".ldb") != NULL || - strstr(f.c_str(), ".log") != NULL) { + if (strstr(f.c_str(), ".ldb") != nullptr || + strstr(f.c_str(), ".log") != nullptr) { *r = new DataFile(this, *r); - } else if (strstr(f.c_str(), "MANIFEST") != NULL) { + } else if (strstr(f.c_str(), "MANIFEST") != nullptr) { *r = new ManifestFile(this, *r); } } @@ -166,13 +206,13 @@ class SpecialEnv : public EnvWrapper { private: RandomAccessFile* target_; AtomicCounter* counter_; + public: CountingFile(RandomAccessFile* target, AtomicCounter* counter) - : target_(target), counter_(counter) { - } - virtual ~CountingFile() { delete target_; } - virtual Status Read(uint64_t offset, size_t n, Slice* result, - char* scratch) const { + : target_(target), counter_(counter) {} + ~CountingFile() override { delete target_; } + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { counter_->Increment(); return target_->Read(offset, n, result, scratch); } @@ -187,18 +227,6 @@ class SpecialEnv : public EnvWrapper { }; class DBTest { - private: - const FilterPolicy* filter_policy_; - - // Sequence of option configurations to try - enum OptionConfig { - kDefault, - kFilter, - kUncompressed, - kEnd - }; - int option_config_; - public: std::string dbname_; SpecialEnv* env_; @@ -206,12 +234,11 @@ class DBTest { Options last_options_; - DBTest() : option_config_(kDefault), - env_(new SpecialEnv(Env::Default())) { + DBTest() : env_(new SpecialEnv(Env::Default())), option_config_(kDefault) { filter_policy_ = NewBloomFilterPolicy(10); dbname_ = test::TmpDir() + "/db_test"; DestroyDB(dbname_, Options()); - db_ = NULL; + db_ = nullptr; Reopen(); } @@ -237,7 +264,11 @@ class DBTest { // Return the current option configuration. Options CurrentOptions() { Options options; + options.reuse_logs = false; switch (option_config_) { + case kReuse: + options.reuse_logs = true; + break; case kFilter: options.filter_policy = filter_policy_; break; @@ -250,31 +281,27 @@ class DBTest { return options; } - DBImpl* dbfull() { - return reinterpret_cast(db_); - } + DBImpl* dbfull() { return reinterpret_cast(db_); } - void Reopen(Options* options = NULL) { - ASSERT_OK(TryReopen(options)); - } + void Reopen(Options* options = nullptr) { ASSERT_OK(TryReopen(options)); } void Close() { delete db_; - db_ = NULL; + db_ = nullptr; } - void DestroyAndReopen(Options* options = NULL) { + void DestroyAndReopen(Options* options = nullptr) { delete db_; - db_ = NULL; + db_ = nullptr; DestroyDB(dbname_, Options()); ASSERT_OK(TryReopen(options)); } Status TryReopen(Options* options) { delete db_; - db_ = NULL; + db_ = nullptr; Options opts; - if (options != NULL) { + if (options != nullptr) { opts = *options; } else { opts = CurrentOptions(); @@ -289,11 +316,9 @@ class DBTest { return db_->Put(WriteOptions(), k, v); } - Status Delete(const std::string& k) { - return db_->Delete(WriteOptions(), k); - } + Status Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); } - std::string Get(const std::string& k, const Snapshot* snapshot = NULL) { + std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) { ReadOptions options; options.snapshot = snapshot; std::string result; @@ -377,10 +402,9 @@ class DBTest { int NumTableFilesAtLevel(int level) { std::string property; - ASSERT_TRUE( - db_->GetProperty("leveldb.num-files-at-level" + NumberToString(level), - &property)); - return atoi(property.c_str()); + ASSERT_TRUE(db_->GetProperty( + "leveldb.num-files-at-level" + NumberToString(level), &property)); + return std::stoi(property); } int TotalTableFiles() { @@ -426,11 +450,12 @@ class DBTest { } // 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) { + // covering the range [small_key,large_key]. + void MakeTables(int n, const std::string& small_key, + const std::string& large_key) { for (int i = 0; i < n; i++) { - Put(small, "begin"); - Put(large, "end"); + Put(small_key, "begin"); + Put(large_key, "end"); dbfull()->TEST_CompactMemTable(); } } @@ -443,9 +468,9 @@ class DBTest { void DumpFileCounts(const char* label) { fprintf(stderr, "---\n%s:\n", label); - fprintf(stderr, "maxoverlap: %lld\n", - static_cast( - dbfull()->TEST_MaxNextLevelOverlappingBytes())); + fprintf( + stderr, "maxoverlap: %lld\n", + static_cast(dbfull()->TEST_MaxNextLevelOverlappingBytes())); for (int level = 0; level < config::kNumLevels; level++) { int num = NumTableFilesAtLevel(level); if (num > 0) { @@ -501,15 +526,42 @@ class DBTest { } return files_renamed; } + + private: + // Sequence of option configurations to try + enum OptionConfig { kDefault, kReuse, kFilter, kUncompressed, kEnd }; + + const FilterPolicy* filter_policy_; + int option_config_; }; TEST(DBTest, Empty) { do { - ASSERT_TRUE(db_ != NULL); + ASSERT_TRUE(db_ != nullptr); ASSERT_EQ("NOT_FOUND", Get("foo")); } while (ChangeOptions()); } +TEST(DBTest, EmptyKey) { + do { + ASSERT_OK(Put("", "v1")); + ASSERT_EQ("v1", Get("")); + ASSERT_OK(Put("", "v2")); + ASSERT_EQ("v2", Get("")); + } while (ChangeOptions()); +} + +TEST(DBTest, EmptyValue) { + do { + ASSERT_OK(Put("key", "v1")); + ASSERT_EQ("v1", Get("key")); + ASSERT_OK(Put("key", "")); + ASSERT_EQ("", Get("key")); + ASSERT_OK(Put("key", "v2")); + ASSERT_EQ("v2", Get("key")); + } while (ChangeOptions()); +} + TEST(DBTest, ReadWrite) { do { ASSERT_OK(Put("foo", "v1")); @@ -542,11 +594,13 @@ TEST(DBTest, GetFromImmutableLayer) { ASSERT_OK(Put("foo", "v1")); ASSERT_EQ("v1", Get("foo")); - env_->delay_data_sync_.Release_Store(env_); // Block sync calls - Put("k1", std::string(100000, 'x')); // Fill memtable - Put("k2", std::string(100000, 'y')); // Trigger compaction + // Block sync calls. + env_->delay_data_sync_.store(true, std::memory_order_release); + Put("k1", std::string(100000, 'x')); // Fill memtable. + Put("k2", std::string(100000, 'y')); // Trigger compaction. ASSERT_EQ("v1", Get("foo")); - env_->delay_data_sync_.Release_Store(NULL); // Release sync calls + // Release sync calls. + env_->delay_data_sync_.store(false, std::memory_order_release); } while (ChangeOptions()); } @@ -558,6 +612,17 @@ TEST(DBTest, GetFromVersions) { } while (ChangeOptions()); } +TEST(DBTest, GetMemUsage) { + do { + ASSERT_OK(Put("foo", "v1")); + std::string val; + ASSERT_TRUE(db_->GetProperty("leveldb.approximate-memory-usage", &val)); + int mem_usage = std::stoi(val); + ASSERT_GT(mem_usage, 0); + ASSERT_LT(mem_usage, 5 * 1024 * 1024); + } while (ChangeOptions()); +} + TEST(DBTest, GetSnapshot) { do { // Try with both a short key and a long key @@ -576,6 +641,55 @@ TEST(DBTest, GetSnapshot) { } while (ChangeOptions()); } +TEST(DBTest, GetIdenticalSnapshots) { + do { + // 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(key, "v1")); + const Snapshot* s1 = db_->GetSnapshot(); + const Snapshot* s2 = db_->GetSnapshot(); + const Snapshot* s3 = db_->GetSnapshot(); + ASSERT_OK(Put(key, "v2")); + ASSERT_EQ("v2", Get(key)); + ASSERT_EQ("v1", Get(key, s1)); + ASSERT_EQ("v1", Get(key, s2)); + ASSERT_EQ("v1", Get(key, s3)); + db_->ReleaseSnapshot(s1); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get(key)); + ASSERT_EQ("v1", Get(key, s2)); + db_->ReleaseSnapshot(s2); + ASSERT_EQ("v1", Get(key, s3)); + db_->ReleaseSnapshot(s3); + } + } while (ChangeOptions()); +} + +TEST(DBTest, IterateOverEmptySnapshot) { + do { + const Snapshot* snapshot = db_->GetSnapshot(); + ReadOptions read_options; + read_options.snapshot = snapshot; + ASSERT_OK(Put("foo", "v1")); + ASSERT_OK(Put("foo", "v2")); + + Iterator* iterator1 = db_->NewIterator(read_options); + iterator1->SeekToFirst(); + ASSERT_TRUE(!iterator1->Valid()); + delete iterator1; + + dbfull()->TEST_CompactMemTable(); + + Iterator* iterator2 = db_->NewIterator(read_options); + iterator2->SeekToFirst(); + ASSERT_TRUE(!iterator2->Valid()); + delete iterator2; + + db_->ReleaseSnapshot(snapshot); + } while (ChangeOptions()); +} + TEST(DBTest, GetLevel0Ordering) { do { // Check that we process level-0 files in correct order. The code @@ -630,8 +744,7 @@ TEST(DBTest, GetEncountersEmptyLevel) { // Step 1: First place sstables in levels 0 and 2 int compaction_count = 0; - while (NumTableFilesAtLevel(0) == 0 || - NumTableFilesAtLevel(2) == 0) { + while (NumTableFilesAtLevel(0) == 0 || NumTableFilesAtLevel(2) == 0) { ASSERT_LE(compaction_count, 100) << "could not fill levels 0 and 2"; compaction_count++; Put("a", "begin"); @@ -640,7 +753,7 @@ TEST(DBTest, GetEncountersEmptyLevel) { } // Step 2: clear level 1 if necessary. - dbfull()->TEST_CompactRange(1, NULL, NULL); + dbfull()->TEST_CompactRange(1, nullptr, nullptr); ASSERT_EQ(NumTableFilesAtLevel(0), 1); ASSERT_EQ(NumTableFilesAtLevel(1), 0); ASSERT_EQ(NumTableFilesAtLevel(2), 1); @@ -768,10 +881,10 @@ TEST(DBTest, IterMulti) { ASSERT_EQ(IterStatus(iter), "b->vb"); // Make sure iter stays at snapshot - ASSERT_OK(Put("a", "va2")); + ASSERT_OK(Put("a", "va2")); ASSERT_OK(Put("a2", "va3")); - ASSERT_OK(Put("b", "vb2")); - ASSERT_OK(Put("c", "vc2")); + ASSERT_OK(Put("b", "vb2")); + ASSERT_OK(Put("c", "vc2")); ASSERT_OK(Delete("b")); iter->SeekToFirst(); ASSERT_EQ(IterStatus(iter), "a->va"); @@ -962,7 +1075,7 @@ TEST(DBTest, RecoverWithLargeLog) { TEST(DBTest, CompactionsGenerateMultipleFiles) { Options options = CurrentOptions(); - options.write_buffer_size = 100000000; // Large write buffer + options.write_buffer_size = 100000000; // Large write buffer Reopen(&options); Random rnd(301); @@ -977,7 +1090,7 @@ TEST(DBTest, CompactionsGenerateMultipleFiles) { // Reopening moves updates to level-0 Reopen(&options); - dbfull()->TEST_CompactRange(0, NULL, NULL); + dbfull()->TEST_CompactRange(0, nullptr, nullptr); ASSERT_EQ(NumTableFilesAtLevel(0), 0); ASSERT_GT(NumTableFilesAtLevel(1), 1); @@ -1001,7 +1114,7 @@ TEST(DBTest, RepeatedWritesToSameKey) { for (int i = 0; i < 5 * kMaxFiles; i++) { Put("key", value); ASSERT_LE(TotalTableFiles(), kMaxFiles); - fprintf(stderr, "after %d: %d files\n", int(i+1), TotalTableFiles()); + fprintf(stderr, "after %d: %d files\n", i + 1, TotalTableFiles()); } } @@ -1028,29 +1141,28 @@ TEST(DBTest, SparseMerge) { } Put("C", "vc"); dbfull()->TEST_CompactMemTable(); - dbfull()->TEST_CompactRange(0, NULL, NULL); + dbfull()->TEST_CompactRange(0, nullptr, nullptr); // Make sparse update - Put("A", "va2"); + Put("A", "va2"); Put("B100", "bvalue2"); - Put("C", "vc2"); + Put("C", "vc2"); dbfull()->TEST_CompactMemTable(); // 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(), 20*1048576); - dbfull()->TEST_CompactRange(0, NULL, NULL); - ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576); - dbfull()->TEST_CompactRange(1, NULL, NULL); - ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576); + ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20 * 1048576); + dbfull()->TEST_CompactRange(0, nullptr, nullptr); + ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20 * 1048576); + dbfull()->TEST_CompactRange(1, nullptr, nullptr); + ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20 * 1048576); } 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)(val), (unsigned long long)(low), (unsigned long long)(high)); } return result; @@ -1059,7 +1171,7 @@ static bool Between(uint64_t val, uint64_t low, uint64_t high) { TEST(DBTest, ApproximateSizes) { do { Options options = CurrentOptions(); - options.write_buffer_size = 100000000; // Large write buffer + options.write_buffer_size = 100000000; // Large write buffer options.compression = kNoCompression; DestroyAndReopen(); @@ -1080,18 +1192,27 @@ TEST(DBTest, ApproximateSizes) { // 0 because GetApproximateSizes() does not account for memtable space ASSERT_TRUE(Between(Size("", Key(50)), 0, 0)); + if (options.reuse_logs) { + // Recovery will reuse memtable, and GetApproximateSizes() does not + // account for memtable usage; + Reopen(&options); + ASSERT_TRUE(Between(Size("", Key(50)), 0, 0)); + continue; + } + // Check sizes across recovery by reopening a few times for (int run = 0; run < 3; run++) { Reopen(&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)), S1*i, S2*i)); - ASSERT_TRUE(Between(Size("", Key(i)+".suffix"), S1*(i+1), S2*(i+1))); - ASSERT_TRUE(Between(Size(Key(i), Key(i+10)), S1*10, S2*10)); + ASSERT_TRUE(Between(Size("", Key(i)), S1 * i, S2 * i)); + ASSERT_TRUE(Between(Size("", Key(i) + ".suffix"), S1 * (i + 1), + S2 * (i + 1))); + ASSERT_TRUE(Between(Size(Key(i), Key(i + 10)), S1 * 10, S2 * 10)); } - ASSERT_TRUE(Between(Size("", Key(50)), S1*50, S2*50)); - ASSERT_TRUE(Between(Size("", Key(50)+".suffix"), S1*50, S2*50)); + ASSERT_TRUE(Between(Size("", Key(50)), S1 * 50, S2 * 50)); + ASSERT_TRUE(Between(Size("", Key(50) + ".suffix"), S1 * 50, S2 * 50)); std::string cstart_str = Key(compact_start); std::string cend_str = Key(compact_start + 9); @@ -1123,6 +1244,11 @@ TEST(DBTest, ApproximateSizes_MixOfSmallAndLarge) { ASSERT_OK(Put(Key(6), RandomString(&rnd, 300000))); ASSERT_OK(Put(Key(7), RandomString(&rnd, 10000))); + if (options.reuse_logs) { + // Need to force a memtable compaction since recovery does not do so. + ASSERT_OK(dbfull()->TEST_CompactMemTable()); + } + // Check sizes across recovery by reopening a few times for (int run = 0; run < 3; run++) { Reopen(&options); @@ -1139,7 +1265,7 @@ TEST(DBTest, ApproximateSizes_MixOfSmallAndLarge) { ASSERT_TRUE(Between(Size(Key(3), Key(5)), 110000, 111000)); - dbfull()->TEST_CompactRange(0, NULL, NULL); + dbfull()->TEST_CompactRange(0, nullptr, nullptr); } } while (ChangeOptions()); } @@ -1153,7 +1279,7 @@ TEST(DBTest, IteratorPinsRef) { // Write to force compactions Put("foo", "newvalue1"); for (int i = 0; i < 100; i++) { - ASSERT_OK(Put(Key(i), Key(i) + std::string(100000, 'v'))); // 100K values + ASSERT_OK(Put(Key(i), Key(i) + std::string(100000, 'v'))); // 100K values } Put("foo", "newvalue2"); @@ -1205,7 +1331,7 @@ TEST(DBTest, HiddenValuesAreRemoved) { Put("pastfoo", "v"); const Snapshot* snapshot = db_->GetSnapshot(); Put("foo", "tiny"); - Put("pastfoo2", "v2"); // Advance sequence number one more + Put("pastfoo2", "v2"); // Advance sequence number one more ASSERT_OK(dbfull()->TEST_CompactMemTable()); ASSERT_GT(NumTableFilesAtLevel(0), 0); @@ -1215,11 +1341,11 @@ TEST(DBTest, HiddenValuesAreRemoved) { db_->ReleaseSnapshot(snapshot); ASSERT_EQ(AllEntriesFor("foo"), "[ tiny, " + big + " ]"); Slice x("x"); - dbfull()->TEST_CompactRange(0, NULL, &x); + dbfull()->TEST_CompactRange(0, nullptr, &x); ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]"); ASSERT_EQ(NumTableFilesAtLevel(0), 0); ASSERT_GE(NumTableFilesAtLevel(1), 1); - dbfull()->TEST_CompactRange(1, NULL, &x); + dbfull()->TEST_CompactRange(1, nullptr, &x); ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]"); ASSERT_TRUE(Between(Size("", "pastfoo"), 0, 1000)); @@ -1230,14 +1356,14 @@ TEST(DBTest, DeletionMarkers1) { Put("foo", "v1"); ASSERT_OK(dbfull()->TEST_CompactMemTable()); const int last = config::kMaxMemCompactLevel; - ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo => v1 is now in last level + ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo => v1 is now in last level // Place a table at level last-1 to prevent merging with preceding mutation Put("a", "begin"); Put("z", "end"); dbfull()->TEST_CompactMemTable(); ASSERT_EQ(NumTableFilesAtLevel(last), 1); - ASSERT_EQ(NumTableFilesAtLevel(last-1), 1); + ASSERT_EQ(NumTableFilesAtLevel(last - 1), 1); Delete("foo"); Put("foo", "v2"); @@ -1245,11 +1371,11 @@ TEST(DBTest, DeletionMarkers1) { ASSERT_OK(dbfull()->TEST_CompactMemTable()); // Moves to level last-2 ASSERT_EQ(AllEntriesFor("foo"), "[ v2, DEL, v1 ]"); Slice z("z"); - dbfull()->TEST_CompactRange(last-2, NULL, &z); + dbfull()->TEST_CompactRange(last - 2, nullptr, &z); // DEL eliminated, but v1 remains because we aren't compacting that level // (DEL can be eliminated because v2 hides v1). ASSERT_EQ(AllEntriesFor("foo"), "[ v2, v1 ]"); - dbfull()->TEST_CompactRange(last-1, NULL, NULL); + dbfull()->TEST_CompactRange(last - 1, nullptr, nullptr); // Merging last-1 w/ last, so we are the base level for "foo", so // DEL is removed. (as is v1). ASSERT_EQ(AllEntriesFor("foo"), "[ v2 ]"); @@ -1259,23 +1385,23 @@ TEST(DBTest, DeletionMarkers2) { Put("foo", "v1"); ASSERT_OK(dbfull()->TEST_CompactMemTable()); const int last = config::kMaxMemCompactLevel; - ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo => v1 is now in last level + ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo => v1 is now in last level // Place a table at level last-1 to prevent merging with preceding mutation Put("a", "begin"); Put("z", "end"); dbfull()->TEST_CompactMemTable(); ASSERT_EQ(NumTableFilesAtLevel(last), 1); - ASSERT_EQ(NumTableFilesAtLevel(last-1), 1); + ASSERT_EQ(NumTableFilesAtLevel(last - 1), 1); Delete("foo"); ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); ASSERT_OK(dbfull()->TEST_CompactMemTable()); // Moves to level last-2 ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); - dbfull()->TEST_CompactRange(last-2, NULL, NULL); + dbfull()->TEST_CompactRange(last - 2, nullptr, nullptr); // DEL kept: "last" file overlaps ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); - dbfull()->TEST_CompactRange(last-1, NULL, NULL); + dbfull()->TEST_CompactRange(last - 1, nullptr, nullptr); // Merging last-1 w/ last, so we are the base level for "foo", so // DEL is removed. (as is v1). ASSERT_EQ(AllEntriesFor("foo"), "[ ]"); @@ -1285,7 +1411,8 @@ TEST(DBTest, OverlapInLevel0) { do { ASSERT_EQ(config::kMaxMemCompactLevel, 2) << "Fix test to match config"; - // Fill levels 1 and 2 to disable the pushing of new memtables to levels > 0. + // Fill levels 1 and 2 to disable the pushing of new memtables to levels > + // 0. ASSERT_OK(Put("100", "v100")); ASSERT_OK(Put("999", "v999")); dbfull()->TEST_CompactMemTable(); @@ -1308,8 +1435,8 @@ TEST(DBTest, OverlapInLevel0) { ASSERT_EQ("2,1,1", FilesPerLevel()); // Compact away the placeholder files we created initially - dbfull()->TEST_CompactRange(1, NULL, NULL); - dbfull()->TEST_CompactRange(2, NULL, NULL); + dbfull()->TEST_CompactRange(1, nullptr, nullptr); + dbfull()->TEST_CompactRange(2, nullptr, nullptr); ASSERT_EQ("2", FilesPerLevel()); // Do a memtable compaction. Before bug-fix, the compaction would @@ -1341,21 +1468,21 @@ TEST(DBTest, L0_CompactionBug_Issue44_a) { TEST(DBTest, L0_CompactionBug_Issue44_b) { Reopen(); - Put("",""); + Put("", ""); Reopen(); Delete("e"); - Put("",""); + Put("", ""); Reopen(); Put("c", "cv"); Reopen(); - Put("",""); + Put("", ""); Reopen(); - Put("",""); + Put("", ""); DelayMilliseconds(1000); // Wait for compaction to finish Reopen(); - Put("d","dv"); + Put("d", "dv"); Reopen(); - Put("",""); + Put("", ""); Reopen(); Delete("d"); Delete("b"); @@ -1365,17 +1492,26 @@ TEST(DBTest, L0_CompactionBug_Issue44_b) { ASSERT_EQ("(->)(c->cv)", Contents()); } +TEST(DBTest, Fflush_Issue474) { + static const int kNum = 100000; + Random rnd(test::RandomSeed()); + for (int i = 0; i < kNum; i++) { + fflush(nullptr); + ASSERT_OK(Put(RandomKey(&rnd), RandomString(&rnd, 100))); + } +} + TEST(DBTest, ComparatorCheck) { class NewComparator : public Comparator { public: - virtual const char* Name() const { return "leveldb.NewComparator"; } - virtual int Compare(const Slice& a, const Slice& b) const { + const char* Name() const override { return "leveldb.NewComparator"; } + 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 { + void FindShortestSeparator(std::string* s, const Slice& l) const override { BytewiseComparator()->FindShortestSeparator(s, l); } - virtual void FindShortSuccessor(std::string* key) const { + void FindShortSuccessor(std::string* key) const override { BytewiseComparator()->FindShortSuccessor(key); } }; @@ -1391,21 +1527,22 @@ TEST(DBTest, ComparatorCheck) { TEST(DBTest, CustomComparator) { class NumberComparator : public Comparator { public: - virtual const char* Name() const { return "test.NumberComparator"; } - virtual int Compare(const Slice& a, const Slice& b) const { + const char* Name() const override { return "test.NumberComparator"; } + 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 { - ToNumber(*s); // Check format - ToNumber(l); // Check format + 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 { - ToNumber(*key); // Check format + 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. - ASSERT_TRUE(x.size() >= 2 && x[0] == '[' && x[x.size()-1] == ']') + ASSERT_TRUE(x.size() >= 2 && x[0] == '[' && x[x.size() - 1] == ']') << EscapeString(x); int val; char ignored; @@ -1418,7 +1555,7 @@ TEST(DBTest, CustomComparator) { Options new_options = CurrentOptions(); new_options.create_if_missing = true; new_options.comparator = &cmp; - new_options.filter_policy = NULL; // Cannot use bloom filters + new_options.filter_policy = nullptr; // Cannot use bloom filters new_options.write_buffer_size = 1000; // Compact more often DestroyAndReopen(&new_options); ASSERT_OK(Put("[10]", "ten")); @@ -1436,7 +1573,7 @@ TEST(DBTest, CustomComparator) { for (int run = 0; run < 2; run++) { for (int i = 0; i < 1000; i++) { char buf[100]; - snprintf(buf, sizeof(buf), "[%d]", i*10); + snprintf(buf, sizeof(buf), "[%d]", i * 10); ASSERT_OK(Put(buf, buf)); } Compact("[0]", "[1000000]"); @@ -1473,7 +1610,7 @@ TEST(DBTest, ManualCompaction) { // Compact all MakeTables(1, "a", "z"); ASSERT_EQ("0,1,2", FilesPerLevel()); - db_->CompactRange(NULL, NULL); + db_->CompactRange(nullptr, nullptr); ASSERT_EQ("0,0,1", FilesPerLevel()); } @@ -1482,42 +1619,94 @@ TEST(DBTest, DBOpen_Options) { DestroyDB(dbname, Options()); // Does not exist, and create_if_missing == false: error - DB* db = NULL; + DB* db = nullptr; Options opts; opts.create_if_missing = false; Status s = DB::Open(opts, dbname, &db); - ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != NULL); - ASSERT_TRUE(db == NULL); + ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != nullptr); + ASSERT_TRUE(db == nullptr); // Does not exist, and create_if_missing == true: OK opts.create_if_missing = true; s = DB::Open(opts, dbname, &db); ASSERT_OK(s); - ASSERT_TRUE(db != NULL); + ASSERT_TRUE(db != nullptr); delete db; - db = NULL; + db = nullptr; // Does exist, and error_if_exists == true: error opts.create_if_missing = false; opts.error_if_exists = true; s = DB::Open(opts, dbname, &db); - ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != NULL); - ASSERT_TRUE(db == NULL); + ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != nullptr); + ASSERT_TRUE(db == nullptr); // Does exist, and error_if_exists == false: OK opts.create_if_missing = true; opts.error_if_exists = false; s = DB::Open(opts, dbname, &db); ASSERT_OK(s); - ASSERT_TRUE(db != NULL); + ASSERT_TRUE(db != nullptr); delete db; - db = NULL; + db = nullptr; +} + +TEST(DBTest, DestroyEmptyDir) { + std::string dbname = test::TmpDir() + "/db_empty_dir"; + TestEnv env(Env::Default()); + env.DeleteDir(dbname); + ASSERT_TRUE(!env.FileExists(dbname)); + + Options opts; + opts.env = &env; + + ASSERT_OK(env.CreateDir(dbname)); + ASSERT_TRUE(env.FileExists(dbname)); + std::vector children; + ASSERT_OK(env.GetChildren(dbname, &children)); + // The stock Env's do not filter out '.' and '..' special files. + ASSERT_EQ(2, children.size()); + ASSERT_OK(DestroyDB(dbname, opts)); + ASSERT_TRUE(!env.FileExists(dbname)); + + // Should also be destroyed if Env is filtering out dot files. + env.SetIgnoreDotFiles(true); + ASSERT_OK(env.CreateDir(dbname)); + ASSERT_TRUE(env.FileExists(dbname)); + ASSERT_OK(env.GetChildren(dbname, &children)); + ASSERT_EQ(0, children.size()); + ASSERT_OK(DestroyDB(dbname, opts)); + ASSERT_TRUE(!env.FileExists(dbname)); +} + +TEST(DBTest, DestroyOpenDB) { + std::string dbname = test::TmpDir() + "/open_db_dir"; + env_->DeleteDir(dbname); + ASSERT_TRUE(!env_->FileExists(dbname)); + + Options opts; + opts.create_if_missing = true; + DB* db = nullptr; + ASSERT_OK(DB::Open(opts, dbname, &db)); + ASSERT_TRUE(db != nullptr); + + // Must fail to destroy an open db. + ASSERT_TRUE(env_->FileExists(dbname)); + ASSERT_TRUE(!DestroyDB(dbname, Options()).ok()); + ASSERT_TRUE(env_->FileExists(dbname)); + + delete db; + db = nullptr; + + // Should succeed destroying a closed db. + ASSERT_OK(DestroyDB(dbname, Options())); + ASSERT_TRUE(!env_->FileExists(dbname)); } TEST(DBTest, Locking) { - DB* db2 = NULL; + DB* db2 = nullptr; Status s = DB::Open(CurrentOptions(), dbname_, &db2); ASSERT_TRUE(!s.ok()) << "Locking did not prevent re-opening db"; } @@ -1532,13 +1721,14 @@ TEST(DBTest, NoSpace) { ASSERT_EQ("v1", Get("foo")); Compact("a", "z"); const int num_files = CountFiles(); - env_->no_space_.Release_Store(env_); // Force out-of-space errors + // Force out-of-space errors. + env_->no_space_.store(true, std::memory_order_release); for (int i = 0; i < 10; i++) { - for (int level = 0; level < config::kNumLevels-1; level++) { - dbfull()->TEST_CompactRange(level, NULL, NULL); + for (int level = 0; level < config::kNumLevels - 1; level++) { + dbfull()->TEST_CompactRange(level, nullptr, nullptr); } } - env_->no_space_.Release_Store(NULL); + env_->no_space_.store(false, std::memory_order_release); ASSERT_LT(CountFiles(), num_files + 3); } @@ -1548,7 +1738,8 @@ TEST(DBTest, NonWritableFileSystem) { options.env = env_; Reopen(&options); ASSERT_OK(Put("foo", "v1")); - env_->non_writable_.Release_Store(env_); // Force errors for new files + // Force errors for new files. + env_->non_writable_.store(true, std::memory_order_release); std::string big(100000, 'x'); int errors = 0; for (int i = 0; i < 20; i++) { @@ -1559,7 +1750,7 @@ TEST(DBTest, NonWritableFileSystem) { } } ASSERT_GT(errors, 0); - env_->non_writable_.Release_Store(NULL); + env_->non_writable_.store(false, std::memory_order_release); } TEST(DBTest, WriteSyncError) { @@ -1569,7 +1760,7 @@ TEST(DBTest, WriteSyncError) { Options options = CurrentOptions(); options.env = env_; Reopen(&options); - env_->data_sync_error_.Release_Store(env_); + env_->data_sync_error_.store(true, std::memory_order_release); // (b) Normal write should succeed WriteOptions w; @@ -1583,7 +1774,7 @@ TEST(DBTest, WriteSyncError) { ASSERT_EQ("NOT_FOUND", Get("k2")); // (d) make sync behave normally - env_->data_sync_error_.Release_Store(NULL); + env_->data_sync_error_.store(false, std::memory_order_release); // (e) Do a non-sync write; should fail w.sync = false; @@ -1603,9 +1794,8 @@ TEST(DBTest, ManifestWriteError) { // 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++) { - port::AtomicPointer* error_type = (iter == 0) - ? &env_->manifest_sync_error_ - : &env_->manifest_write_error_; + std::atomic* error_type = (iter == 0) ? &env_->manifest_sync_error_ + : &env_->manifest_write_error_; // Insert foo=>bar mapping Options options = CurrentOptions(); @@ -1620,15 +1810,15 @@ TEST(DBTest, ManifestWriteError) { dbfull()->TEST_CompactMemTable(); ASSERT_EQ("bar", Get("foo")); const int last = config::kMaxMemCompactLevel; - ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo=>bar is now in last level + ASSERT_EQ(NumTableFilesAtLevel(last), 1); // foo=>bar is now in last level // Merging compaction (will fail) - error_type->Release_Store(env_); - dbfull()->TEST_CompactRange(last, NULL, NULL); // Should 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->Release_Store(NULL); + error_type->store(false, std::memory_order_release); Reopen(&options); ASSERT_EQ("bar", Get("foo")); } @@ -1648,8 +1838,7 @@ TEST(DBTest, MissingSSTFile) { options.paranoid_checks = true; Status s = TryReopen(&options); ASSERT_TRUE(!s.ok()); - ASSERT_TRUE(s.ToString().find("issing") != std::string::npos) - << s.ToString(); + ASSERT_TRUE(s.ToString().find("issing") != std::string::npos) << s.ToString(); } TEST(DBTest, StillReadSST) { @@ -1699,7 +1888,7 @@ TEST(DBTest, BloomFilter) { dbfull()->TEST_CompactMemTable(); // Prevent auto compactions triggered by seeks - env_->delay_data_sync_.Release_Store(env_); + env_->delay_data_sync_.store(true, std::memory_order_release); // Lookup present keys. Should rarely read from small sstable. env_->random_read_counter_.Reset(); @@ -1709,7 +1898,7 @@ TEST(DBTest, BloomFilter) { 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); + ASSERT_LE(reads, N + 2 * N / 100); // Lookup present keys. Should rarely read from either sstable. env_->random_read_counter_.Reset(); @@ -1718,9 +1907,9 @@ TEST(DBTest, BloomFilter) { } reads = env_->random_read_counter_.Read(); fprintf(stderr, "%d missing => %d reads\n", N, reads); - ASSERT_LE(reads, 3*N/100); + ASSERT_LE(reads, 3 * N / 100); - env_->delay_data_sync_.Release_Store(NULL); + env_->delay_data_sync_.store(false, std::memory_order_release); Close(); delete options.block_cache; delete options.filter_policy; @@ -1735,9 +1924,9 @@ static const int kNumKeys = 1000; struct MTState { DBTest* test; - port::AtomicPointer stop; - port::AtomicPointer counter[kNumThreads]; - port::AtomicPointer thread_done[kNumThreads]; + std::atomic stop; + std::atomic counter[kNumThreads]; + std::atomic thread_done[kNumThreads]; }; struct MTThread { @@ -1749,13 +1938,13 @@ static void MTThreadBody(void* arg) { MTThread* t = reinterpret_cast(arg); int id = t->id; DB* db = t->state->test->db_; - uintptr_t counter = 0; + int counter = 0; fprintf(stderr, "... starting thread %d\n", id); Random rnd(1000 + id); std::string value; char valbuf[1500]; - while (t->state->stop.Acquire_Load() == NULL) { - t->state->counter[id].Release_Store(reinterpret_cast(counter)); + while (!t->state->stop.load(std::memory_order_acquire)) { + t->state->counter[id].store(counter, std::memory_order_release); int key = rnd.Uniform(kNumKeys); char keybuf[20]; @@ -1764,8 +1953,8 @@ static void MTThreadBody(void* arg) { if (rnd.OneIn(2)) { // Write values of the form . // We add some padding for force compactions. - snprintf(valbuf, sizeof(valbuf), "%d.%d.%-1000d", - key, id, static_cast(counter)); + snprintf(valbuf, sizeof(valbuf), "%d.%d.%-1000d", key, id, + static_cast(counter)); ASSERT_OK(db->Put(WriteOptions(), Slice(keybuf), Slice(valbuf))); } else { // Read a value and verify that it matches the pattern written above. @@ -1780,14 +1969,13 @@ static void MTThreadBody(void* arg) { ASSERT_EQ(k, key); ASSERT_GE(w, 0); ASSERT_LT(w, kNumThreads); - ASSERT_LE(static_cast(c), reinterpret_cast( - t->state->counter[w].Acquire_Load())); + ASSERT_LE(c, t->state->counter[w].load(std::memory_order_acquire)); } } counter++; } - t->state->thread_done[id].Release_Store(t); - fprintf(stderr, "... stopping thread %d after %d ops\n", id, int(counter)); + t->state->thread_done[id].store(true, std::memory_order_release); + fprintf(stderr, "... stopping thread %d after %d ops\n", id, counter); } } // namespace @@ -1797,10 +1985,10 @@ TEST(DBTest, MultiThreaded) { // Initialize state MTState mt; mt.test = this; - mt.stop.Release_Store(0); + mt.stop.store(false, std::memory_order_release); for (int id = 0; id < kNumThreads; id++) { - mt.counter[id].Release_Store(0); - mt.thread_done[id].Release_Store(0); + mt.counter[id].store(false, std::memory_order_release); + mt.thread_done[id].store(false, std::memory_order_release); } // Start threads @@ -1815,9 +2003,9 @@ TEST(DBTest, MultiThreaded) { DelayMilliseconds(kTestSeconds * 1000); // Stop the threads and wait for them to finish - mt.stop.Release_Store(&mt); + mt.stop.store(true, std::memory_order_release); for (int id = 0; id < kNumThreads; id++) { - while (mt.thread_done[id].Acquire_Load() == NULL) { + while (!mt.thread_done[id].load(std::memory_order_acquire)) { DelayMilliseconds(100); } } @@ -1828,28 +2016,28 @@ namespace { typedef std::map KVMap; } -class ModelDB: public DB { +class ModelDB : public DB { public: class ModelSnapshot : public Snapshot { public: KVMap map_; }; - explicit ModelDB(const Options& options): options_(options) { } - ~ModelDB() { } - virtual Status Put(const WriteOptions& o, const Slice& k, const Slice& v) { + explicit ModelDB(const Options& options) : options_(options) {} + ~ModelDB() override = default; + Status Put(const WriteOptions& o, const Slice& k, const Slice& v) override { return DB::Put(o, k, v); } - virtual Status Delete(const WriteOptions& o, const Slice& key) { + Status Delete(const WriteOptions& o, const Slice& key) override { return DB::Delete(o, key); } - virtual Status Get(const ReadOptions& options, - const Slice& key, std::string* value) { - assert(false); // Not implemented + Status Get(const ReadOptions& options, const Slice& key, + std::string* value) override { + assert(false); // Not implemented return Status::NotFound(key); } - virtual Iterator* NewIterator(const ReadOptions& options) { - if (options.snapshot == NULL) { + Iterator* NewIterator(const ReadOptions& options) override { + if (options.snapshot == nullptr) { KVMap* saved = new KVMap; *saved = map_; return new ModelIter(saved, true); @@ -1859,68 +2047,65 @@ class ModelDB: public DB { return new ModelIter(snapshot_state, false); } } - virtual const Snapshot* GetSnapshot() { + const Snapshot* GetSnapshot() override { ModelSnapshot* snapshot = new ModelSnapshot; snapshot->map_ = map_; return snapshot; } - virtual void ReleaseSnapshot(const Snapshot* snapshot) { + void ReleaseSnapshot(const Snapshot* snapshot) override { delete reinterpret_cast(snapshot); } - virtual Status Write(const WriteOptions& options, WriteBatch* batch) { + 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) { + void Put(const Slice& key, const Slice& value) override { (*map_)[key.ToString()] = value.ToString(); } - virtual void Delete(const Slice& key) { - map_->erase(key.ToString()); - } + void Delete(const Slice& key) override { map_->erase(key.ToString()); } }; Handler handler; handler.map_ = &map_; return batch->Iterate(&handler); } - virtual bool GetProperty(const Slice& property, std::string* value) { + bool GetProperty(const Slice& property, std::string* value) override { return false; } - virtual void GetApproximateSizes(const Range* r, int n, uint64_t* sizes) { + void GetApproximateSizes(const Range* r, int n, uint64_t* sizes) override { for (int i = 0; i < n; i++) { sizes[i] = 0; } } - virtual void CompactRange(const Slice* start, const Slice* end) { - } + void CompactRange(const Slice* start, const Slice* end) override {} private: - class ModelIter: public Iterator { + class ModelIter : public Iterator { public: ModelIter(const KVMap* map, bool owned) - : map_(map), owned_(owned), iter_(map_->end()) { - } - ~ModelIter() { + : map_(map), owned_(owned), iter_(map_->end()) {} + ~ModelIter() override { if (owned_) delete map_; } - virtual bool Valid() const { return iter_ != map_->end(); } - virtual void SeekToFirst() { iter_ = map_->begin(); } - virtual void SeekToLast() { + bool Valid() const override { return iter_ != map_->end(); } + void SeekToFirst() override { iter_ = map_->begin(); } + void SeekToLast() override { if (map_->empty()) { iter_ = map_->end(); } else { iter_ = map_->find(map_->rbegin()->first); } } - virtual void Seek(const Slice& k) { + void Seek(const Slice& k) override { iter_ = map_->lower_bound(k.ToString()); } - virtual void Next() { ++iter_; } - virtual void Prev() { --iter_; } - virtual Slice key() const { return iter_->first; } - virtual Slice value() const { return iter_->second; } - virtual Status status() const { return Status::OK(); } + void Next() override { ++iter_; } + void Prev() override { --iter_; } + Slice key() const override { return iter_->first; } + Slice value() const override { return iter_->second; } + Status status() const override { return Status::OK(); } + private: const KVMap* const map_; const bool owned_; // Do we own map_ @@ -1930,16 +2115,7 @@ class ModelDB: public DB { KVMap map_; }; -static std::string RandomKey(Random* rnd) { - int len = (rnd->OneIn(3) - ? 1 // Short sometimes to encourage collisions - : (rnd->OneIn(100) ? rnd->Skewed(10) : rnd->Uniform(10))); - return test::RandomKey(rnd, len); -} - -static bool CompareIterators(int step, - DB* model, - DB* db, +static bool CompareIterators(int step, DB* model, DB* db, const Snapshot* model_snap, const Snapshot* db_snap) { ReadOptions options; @@ -1950,12 +2126,10 @@ static bool CompareIterators(int step, bool ok = true; int count = 0; for (miter->SeekToFirst(), dbiter->SeekToFirst(); - ok && miter->Valid() && dbiter->Valid(); - miter->Next(), dbiter->Next()) { + 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, + fprintf(stderr, "step %d: Key mismatch: '%s' vs. '%s'\n", step, EscapeString(miter->key()).c_str(), EscapeString(dbiter->key()).c_str()); ok = false; @@ -1964,8 +2138,7 @@ static bool CompareIterators(int step, 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(), + step, EscapeString(miter->key()).c_str(), EscapeString(miter->value()).c_str(), EscapeString(miter->value()).c_str()); ok = false; @@ -1990,8 +2163,8 @@ TEST(DBTest, Randomized) { do { ModelDB model(CurrentOptions()); const int N = 10000; - const Snapshot* model_snap = NULL; - const Snapshot* db_snap = NULL; + const Snapshot* model_snap = nullptr; + const Snapshot* db_snap = nullptr; std::string k, v; for (int step = 0; step < N; step++) { if (step % 100 == 0) { @@ -1999,22 +2172,19 @@ TEST(DBTest, Randomized) { } // TODO(sanjay): Test Get() works int p = rnd.Uniform(100); - if (p < 45) { // Put + if (p < 45) { // Put k = RandomKey(&rnd); - v = RandomString(&rnd, - rnd.OneIn(20) - ? 100 + rnd.Uniform(100) - : rnd.Uniform(8)); + 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 + } else if (p < 90) { // Delete k = RandomKey(&rnd); ASSERT_OK(model.Delete(WriteOptions(), k)); ASSERT_OK(db_->Delete(WriteOptions(), k)); - - } else { // Multi-element batch + } else { // Multi-element batch WriteBatch b; const int num = rnd.Uniform(8); for (int i = 0; i < num; i++) { @@ -2036,23 +2206,23 @@ TEST(DBTest, Randomized) { } if ((step % 100) == 0) { - ASSERT_TRUE(CompareIterators(step, &model, db_, NULL, NULL)); + 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 != NULL) model.ReleaseSnapshot(model_snap); - if (db_snap != NULL) db_->ReleaseSnapshot(db_snap); + if (model_snap != nullptr) model.ReleaseSnapshot(model_snap); + if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap); Reopen(); - ASSERT_TRUE(CompareIterators(step, &model, db_, NULL, NULL)); + ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr)); model_snap = model.GetSnapshot(); db_snap = db_->GetSnapshot(); } } - if (model_snap != NULL) model.ReleaseSnapshot(model_snap); - if (db_snap != NULL) db_->ReleaseSnapshot(db_snap); + if (model_snap != nullptr) model.ReleaseSnapshot(model_snap); + if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap); } while (ChangeOptions()); } @@ -2066,15 +2236,15 @@ void BM_LogAndApply(int iters, int num_base_files) { std::string dbname = test::TmpDir() + "/leveldb_test_benchmark"; DestroyDB(dbname, Options()); - DB* db = NULL; + DB* db = nullptr; Options opts; opts.create_if_missing = true; Status s = DB::Open(opts, dbname, &db); ASSERT_OK(s); - ASSERT_TRUE(db != NULL); + ASSERT_TRUE(db != nullptr); delete db; - db = NULL; + db = nullptr; Env* env = Env::Default(); @@ -2083,13 +2253,14 @@ void BM_LogAndApply(int iters, int num_base_files) { InternalKeyComparator cmp(BytewiseComparator()); Options options; - VersionSet vset(dbname, &options, NULL, &cmp); - ASSERT_OK(vset.Recover()); + VersionSet vset(dbname, &options, nullptr, &cmp); + bool save_manifest; + ASSERT_OK(vset.Recover(&save_manifest)); VersionEdit vbase; uint64_t fnum = 1; for (int i = 0; i < num_base_files; i++) { - InternalKey start(MakeKey(2*fnum), 1, kTypeValue); - InternalKey limit(MakeKey(2*fnum+1), 1, kTypeDeletion); + InternalKey start(MakeKey(2 * fnum), 1, kTypeValue); + InternalKey limit(MakeKey(2 * fnum + 1), 1, kTypeDeletion); vbase.AddFile(2, fnum++, 1 /* file size */, start, limit); } ASSERT_OK(vset.LogAndApply(&vbase, &mu)); @@ -2099,8 +2270,8 @@ void BM_LogAndApply(int iters, int num_base_files) { for (int i = 0; i < iters; i++) { VersionEdit vedit; vedit.DeleteFile(2, fnum); - InternalKey start(MakeKey(2*fnum), 1, kTypeValue); - InternalKey limit(MakeKey(2*fnum+1), 1, kTypeDeletion); + InternalKey start(MakeKey(2 * fnum), 1, kTypeValue); + InternalKey limit(MakeKey(2 * fnum + 1), 1, kTypeDeletion); vedit.AddFile(2, fnum++, 1 /* file size */, start, limit); vset.LogAndApply(&vedit, &mu); } @@ -2109,8 +2280,8 @@ void BM_LogAndApply(int iters, int num_base_files) { char buf[16]; snprintf(buf, sizeof(buf), "%d", num_base_files); fprintf(stderr, - "BM_LogAndApply/%-6s %8d iters : %9u us (%7.0f us / iter)\n", - buf, iters, us, ((float)us) / iters); + "BM_LogAndApply/%-6s %8d iters : %9u us (%7.0f us / iter)\n", buf, + iters, us, ((float)us) / iters); } } // namespace leveldb diff --git a/db/dbformat.cc b/db/dbformat.cc index 20a7ca4..459eddf 100644 --- a/db/dbformat.cc +++ b/db/dbformat.cc @@ -2,8 +2,12 @@ // 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/dbformat.h" + +#include + +#include + #include "port/port.h" #include "util/coding.h" @@ -21,26 +25,20 @@ void AppendInternalKey(std::string* result, const ParsedInternalKey& key) { } std::string ParsedInternalKey::DebugString() const { - char buf[50]; - snprintf(buf, sizeof(buf), "' @ %llu : %d", - (unsigned long long) sequence, - int(type)); - std::string result = "'"; - result += EscapeString(user_key.ToString()); - result += buf; - return result; + std::ostringstream ss; + ss << '\'' << EscapeString(user_key.ToString()) << "' @ " << sequence << " : " + << static_cast(type); + return ss.str(); } std::string InternalKey::DebugString() const { - std::string result; ParsedInternalKey parsed; if (ParseInternalKey(rep_, &parsed)) { - result = parsed.DebugString(); - } else { - result = "(bad)"; - result.append(EscapeString(rep_)); + return parsed.DebugString(); } - return result; + std::ostringstream ss; + ss << "(bad)" << EscapeString(rep_); + return ss.str(); } const char* InternalKeyComparator::Name() const { @@ -65,9 +63,8 @@ int InternalKeyComparator::Compare(const Slice& akey, const Slice& bkey) const { return r; } -void InternalKeyComparator::FindShortestSeparator( - std::string* start, - const Slice& limit) const { +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); @@ -77,7 +74,8 @@ void InternalKeyComparator::FindShortestSeparator( 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)); + PutFixed64(&tmp, + PackSequenceAndType(kMaxSequenceNumber, kValueTypeForSeek)); assert(this->Compare(*start, tmp) < 0); assert(this->Compare(tmp, limit) < 0); start->swap(tmp); @@ -92,15 +90,14 @@ void InternalKeyComparator::FindShortSuccessor(std::string* key) const { 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)); + PutFixed64(&tmp, + PackSequenceAndType(kMaxSequenceNumber, kValueTypeForSeek)); assert(this->Compare(*key, tmp) < 0); key->swap(tmp); } } -const char* InternalFilterPolicy::Name() const { - return user_policy_->Name(); -} +const char* InternalFilterPolicy::Name() const { return user_policy_->Name(); } void InternalFilterPolicy::CreateFilter(const Slice* keys, int n, std::string* dst) const { diff --git a/db/dbformat.h b/db/dbformat.h index ea897b1..abcb489 100644 --- a/db/dbformat.h +++ b/db/dbformat.h @@ -6,6 +6,7 @@ #define STORAGE_LEVELDB_DB_DBFORMAT_H_ #include + #include "leveldb/comparator.h" #include "leveldb/db.h" #include "leveldb/filter_policy.h" @@ -48,10 +49,7 @@ 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. -enum ValueType { - kTypeDeletion = 0x0, - kTypeValue = 0x1 -}; +enum ValueType { kTypeDeletion = 0x0, kTypeValue = 0x1 }; // 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 @@ -64,17 +62,16 @@ typedef uint64_t SequenceNumber; // 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); +static const SequenceNumber kMaxSequenceNumber = ((0x1ull << 56) - 1); struct ParsedInternalKey { Slice user_key; SequenceNumber sequence; ValueType type; - ParsedInternalKey() { } // Intentionally left uninitialized (for speed) + ParsedInternalKey() {} // Intentionally left uninitialized (for speed) ParsedInternalKey(const Slice& u, const SequenceNumber& seq, ValueType t) - : user_key(u), sequence(seq), type(t) { } + : user_key(u), sequence(seq), type(t) {} std::string DebugString() const; }; @@ -84,15 +81,13 @@ inline size_t InternalKeyEncodingLength(const ParsedInternalKey& key) { } // Append the serialization of "key" to *result. -extern void AppendInternalKey(std::string* result, - const ParsedInternalKey& key); +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); +bool ParseInternalKey(const Slice& internal_key, ParsedInternalKey* result); // Returns the user key portion of an internal key. inline Slice ExtractUserKey(const Slice& internal_key) { @@ -100,27 +95,19 @@ inline Slice ExtractUserKey(const Slice& internal_key) { 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_; + public: - explicit InternalKeyComparator(const Comparator* c) : user_comparator_(c) { } - 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; + explicit InternalKeyComparator(const Comparator* c) : user_comparator_(c) {} + const char* Name() const override; + int Compare(const Slice& a, const Slice& b) const override; + void FindShortestSeparator(std::string* start, + const Slice& limit) const override; + void FindShortSuccessor(std::string* key) const override; const Comparator* user_comparator() const { return user_comparator_; } @@ -131,11 +118,12 @@ class InternalKeyComparator : public Comparator { class InternalFilterPolicy : public FilterPolicy { private: const FilterPolicy* const user_policy_; + public: - explicit InternalFilterPolicy(const FilterPolicy* p) : user_policy_(p) { } - virtual const char* Name() const; - virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const; - virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const; + explicit InternalFilterPolicy(const FilterPolicy* p) : user_policy_(p) {} + const char* Name() const override; + void CreateFilter(const Slice* keys, int n, std::string* dst) const override; + bool KeyMayMatch(const Slice& key, const Slice& filter) const override; }; // Modules in this directory should keep internal keys wrapped inside @@ -144,13 +132,18 @@ class InternalFilterPolicy : public FilterPolicy { class InternalKey { private: std::string rep_; + public: - InternalKey() { } // Leave rep_ as empty to indicate it is invalid + 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)); } - void DecodeFrom(const Slice& s) { rep_.assign(s.data(), s.size()); } + bool DecodeFrom(const Slice& s) { + rep_.assign(s.data(), s.size()); + return !rep_.empty(); + } + Slice Encode() const { assert(!rep_.empty()); return rep_; @@ -168,8 +161,8 @@ class InternalKey { std::string DebugString() const; }; -inline int InternalKeyComparator::Compare( - const InternalKey& a, const InternalKey& b) const { +inline int InternalKeyComparator::Compare(const InternalKey& a, + const InternalKey& b) const { return Compare(a.Encode(), b.Encode()); } @@ -192,6 +185,9 @@ class LookupKey { // the specified sequence number. LookupKey(const Slice& user_key, SequenceNumber sequence); + LookupKey(const LookupKey&) = delete; + LookupKey& operator=(const LookupKey&) = delete; + ~LookupKey(); // Return a key suitable for lookup in a MemTable. @@ -214,11 +210,7 @@ class LookupKey { 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&); + char space_[200]; // Avoid allocation for short keys }; inline LookupKey::~LookupKey() { diff --git a/db/dbformat_test.cc b/db/dbformat_test.cc index 5d82f5d..1209369 100644 --- a/db/dbformat_test.cc +++ b/db/dbformat_test.cc @@ -8,8 +8,7 @@ namespace leveldb { -static std::string IKey(const std::string& user_key, - uint64_t seq, +static std::string IKey(const std::string& user_key, uint64_t seq, ValueType vt) { std::string encoded; AppendInternalKey(&encoded, ParsedInternalKey(user_key, seq, vt)); @@ -28,9 +27,7 @@ static std::string ShortSuccessor(const std::string& s) { return result; } -static void TestKey(const std::string& key, - uint64_t seq, - ValueType vt) { +static void TestKey(const std::string& key, uint64_t seq, ValueType vt) { std::string encoded = IKey(key, seq, vt); Slice in(encoded); @@ -44,16 +41,22 @@ static void TestKey(const std::string& key, ASSERT_TRUE(!ParseInternalKey(Slice("bar"), &decoded)); } -class FormatTest { }; +class FormatTest {}; TEST(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 - }; + 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 (int k = 0; k < sizeof(keys) / sizeof(keys[0]); k++) { for (int s = 0; s < sizeof(seq) / sizeof(seq[0]); s++) { TestKey(keys[k], seq[s], kTypeValue); @@ -62,40 +65,44 @@ TEST(FormatTest, InternalKey_EncodeDecode) { } } +TEST(FormatTest, InternalKey_DecodeFromEmpty) { + InternalKey internal_key; + + ASSERT_TRUE(!internal_key.DecodeFrom("")); +} + TEST(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))); + 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))); + 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))); + 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))); + 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))); + ASSERT_EQ( + IKey("foobar", 100, kTypeValue), + Shorten(IKey("foobar", 100, kTypeValue), IKey("foo", 200, kTypeValue))); } TEST(FormatTest, InternalKeyShortestSuccessor) { @@ -105,8 +112,20 @@ TEST(FormatTest, InternalKeyShortestSuccessor) { ShortSuccessor(IKey("\xff\xff", 100, kTypeValue))); } +TEST(FormatTest, ParsedInternalKeyDebugString) { + ParsedInternalKey key("The \"key\" in 'single quotes'", 42, kTypeValue); + + ASSERT_EQ("'The \"key\" in 'single quotes'' @ 42 : 1", key.DebugString()); +} + +TEST(FormatTest, InternalKeyDebugString) { + InternalKey key("The \"key\" in 'single quotes'", 42, kTypeValue); + ASSERT_EQ("'The \"key\" in 'single quotes'' @ 42 : 1", key.DebugString()); + + InternalKey invalid_key; + ASSERT_EQ("(bad)", invalid_key.DebugString()); +} + } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/dumpfile.cc b/db/dumpfile.cc index 61c47c2..77d5900 100644 --- a/db/dumpfile.cc +++ b/db/dumpfile.cc @@ -2,7 +2,10 @@ // 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 "leveldb/dumpfile.h" + #include + #include "db/dbformat.h" #include "db/filename.h" #include "db/log_reader.h" @@ -35,8 +38,7 @@ bool GuessType(const std::string& fname, FileType* type) { // Notified when log reader encounters corruption. class CorruptionReporter : public log::Reader::Reporter { public: - WritableFile* dst_; - virtual void Corruption(size_t bytes, const Status& status) { + void Corruption(size_t bytes, const Status& status) override { std::string r = "corruption: "; AppendNumberTo(&r, bytes); r += " bytes; "; @@ -44,6 +46,8 @@ class CorruptionReporter : public log::Reader::Reporter { r.push_back('\n'); dst_->Append(r); } + + WritableFile* dst_; }; // Print contents of a log file. (*func)() is called on every record. @@ -70,8 +74,7 @@ Status PrintLogContents(Env* env, const std::string& fname, // Called on every item found in a WriteBatch. class WriteBatchItemPrinter : public WriteBatch::Handler { public: - WritableFile* dst_; - virtual void Put(const Slice& key, const Slice& value) { + void Put(const Slice& key, const Slice& value) override { std::string r = " put '"; AppendEscapedStringTo(&r, key); r += "' '"; @@ -79,14 +82,15 @@ class WriteBatchItemPrinter : public WriteBatch::Handler { r += "'\n"; dst_->Append(r); } - virtual void Delete(const Slice& key) { + void Delete(const Slice& key) override { std::string r = " del '"; AppendEscapedStringTo(&r, key); r += "'\n"; dst_->Append(r); } -}; + WritableFile* dst_; +}; // Called on every log record (each one of which is a WriteBatch) // found in a kLogFile. @@ -142,8 +146,8 @@ Status DumpDescriptor(Env* env, const std::string& fname, WritableFile* dst) { Status DumpTable(Env* env, const std::string& fname, WritableFile* dst) { uint64_t file_size; - RandomAccessFile* file = NULL; - Table* table = NULL; + RandomAccessFile* file = nullptr; + Table* table = nullptr; Status s = env->GetFileSize(fname, &file_size); if (s.ok()) { s = env->NewRandomAccessFile(fname, &file); @@ -213,9 +217,12 @@ Status DumpFile(Env* env, const std::string& fname, WritableFile* dst) { return Status::InvalidArgument(fname + ": unknown file type"); } switch (ftype) { - case kLogFile: return DumpLog(env, fname, dst); - case kDescriptorFile: return DumpDescriptor(env, fname, dst); - case kTableFile: return DumpTable(env, fname, dst); + case kLogFile: + return DumpLog(env, fname, dst); + case kDescriptorFile: + return DumpDescriptor(env, fname, dst); + case kTableFile: + return DumpTable(env, fname, dst); default: break; } diff --git a/db/fault_injection_test.cc b/db/fault_injection_test.cc index 8e3e603..5b31bb8 100644 --- a/db/fault_injection_test.cc +++ b/db/fault_injection_test.cc @@ -6,18 +6,20 @@ // the last "sync". It then checks for data loss errors by purposely dropping // file data (or entire files) not protected by a "sync". -#include "leveldb/db.h" - #include #include + #include "db/db_impl.h" #include "db/filename.h" #include "db/log_format.h" #include "db/version_set.h" #include "leveldb/cache.h" +#include "leveldb/db.h" #include "leveldb/env.h" #include "leveldb/table.h" #include "leveldb/write_batch.h" +#include "port/port.h" +#include "port/thread_annotations.h" #include "util/logging.h" #include "util/mutexlock.h" #include "util/testharness.h" @@ -34,7 +36,7 @@ class FaultInjectionTestEnv; namespace { // Assume a filename, and not a directory name like "/foo/bar/" -static std::string GetDirName(const std::string filename) { +static std::string GetDirName(const std::string& filename) { size_t found = filename.find_last_of("/\\"); if (found == std::string::npos) { return ""; @@ -54,8 +56,7 @@ Status Truncate(const std::string& filename, uint64_t length) { SequentialFile* orig_file; Status s = env->NewSequentialFile(filename, &orig_file); - if (!s.ok()) - return s; + if (!s.ok()) return s; char* scratch = new char[length]; leveldb::Slice result; @@ -83,15 +84,15 @@ Status Truncate(const std::string& filename, uint64_t length) { struct FileState { std::string filename_; - ssize_t pos_; - ssize_t pos_at_last_sync_; - ssize_t pos_at_last_flush_; + int64_t pos_; + int64_t pos_at_last_sync_; + int64_t pos_at_last_flush_; FileState(const std::string& filename) : filename_(filename), pos_(-1), pos_at_last_sync_(-1), - pos_at_last_flush_(-1) { } + pos_at_last_flush_(-1) {} FileState() : pos_(-1), pos_at_last_sync_(-1), pos_at_last_flush_(-1) {} @@ -106,14 +107,13 @@ struct FileState { // is written to or sync'ed. class TestWritableFile : public WritableFile { public: - TestWritableFile(const std::string& fname, - WritableFile* f, + TestWritableFile(const FileState& state, WritableFile* f, FaultInjectionTestEnv* env); - virtual ~TestWritableFile(); - virtual Status Append(const Slice& data); - virtual Status Close(); - virtual Status Flush(); - virtual Status Sync(); + ~TestWritableFile() override; + Status Append(const Slice& data) override; + Status Close() override; + Status Flush() override; + Status Sync() override; private: FileState state_; @@ -126,12 +126,15 @@ class TestWritableFile : public WritableFile { class FaultInjectionTestEnv : public EnvWrapper { public: - FaultInjectionTestEnv() : EnvWrapper(Env::Default()), filesystem_active_(true) {} - virtual ~FaultInjectionTestEnv() { } - virtual Status NewWritableFile(const std::string& fname, - WritableFile** result); - virtual Status DeleteFile(const std::string& f); - virtual Status RenameFile(const std::string& s, const std::string& t); + FaultInjectionTestEnv() + : EnvWrapper(Env::Default()), filesystem_active_(true) {} + ~FaultInjectionTestEnv() override = default; + Status NewWritableFile(const std::string& fname, + WritableFile** result) override; + Status NewAppendableFile(const std::string& fname, + WritableFile** result) override; + Status DeleteFile(const std::string& f) override; + Status RenameFile(const std::string& s, const std::string& t) override; void WritableFileClosed(const FileState& state); Status DropUnsyncedFileData(); @@ -144,25 +147,26 @@ class FaultInjectionTestEnv : public EnvWrapper { // 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() const { return filesystem_active_; } - void SetFilesystemActive(bool active) { filesystem_active_ = active; } + bool IsFilesystemActive() LOCKS_EXCLUDED(mutex_) { + MutexLock l(&mutex_); + return filesystem_active_; + } + void SetFilesystemActive(bool active) LOCKS_EXCLUDED(mutex_) { + MutexLock l(&mutex_); + filesystem_active_ = active; + } private: port::Mutex mutex_; - std::map db_file_state_; - std::set new_files_since_last_dir_sync_; - bool filesystem_active_; // Record flushes, syncs, writes + std::map db_file_state_ GUARDED_BY(mutex_); + std::set new_files_since_last_dir_sync_ GUARDED_BY(mutex_); + bool filesystem_active_ GUARDED_BY(mutex_); // Record flushes, syncs, writes }; -TestWritableFile::TestWritableFile(const std::string& fname, - WritableFile* f, +TestWritableFile::TestWritableFile(const FileState& state, WritableFile* f, FaultInjectionTestEnv* env) - : state_(fname), - target_(f), - writable_file_opened_(true), - env_(env) { - assert(f != NULL); - state_.pos_ = 0; + : state_(state), target_(f), writable_file_opened_(true), env_(env) { + assert(f != nullptr); } TestWritableFile::~TestWritableFile() { @@ -228,9 +232,12 @@ Status FaultInjectionTestEnv::NewWritableFile(const std::string& fname, WritableFile* actual_writable_file; Status s = target()->NewWritableFile(fname, &actual_writable_file); if (s.ok()) { - *result = new TestWritableFile(fname, actual_writable_file, this); - // WritableFile doesn't append to files, so if the same file is opened again - // then it will be truncated - so forget our saved state. + FileState state(fname); + state.pos_ = 0; + *result = new TestWritableFile(state, actual_writable_file, this); + // NewWritableFile doesn't append to files, so if the same file is + // opened again then it will be truncated - so forget our saved + // state. UntrackFile(fname); MutexLock l(&mutex_); new_files_since_last_dir_sync_.insert(fname); @@ -238,13 +245,34 @@ Status FaultInjectionTestEnv::NewWritableFile(const std::string& fname, return s; } +Status FaultInjectionTestEnv::NewAppendableFile(const std::string& fname, + WritableFile** result) { + WritableFile* actual_writable_file; + Status s = target()->NewAppendableFile(fname, &actual_writable_file); + if (s.ok()) { + FileState state(fname); + state.pos_ = 0; + { + MutexLock l(&mutex_); + if (db_file_state_.count(fname) == 0) { + new_files_since_last_dir_sync_.insert(fname); + } else { + state = db_file_state_[fname]; + } + } + *result = new TestWritableFile(state, actual_writable_file, this); + } + return s; +} + Status FaultInjectionTestEnv::DropUnsyncedFileData() { 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; + for (const auto& kvp : db_file_state_) { + if (!s.ok()) { + break; + } + const FileState& state = kvp.second; if (!state.IsFullySynced()) { s = state.DropUnsyncedData(); } @@ -301,9 +329,9 @@ Status FaultInjectionTestEnv::RenameFile(const std::string& s, } void FaultInjectionTestEnv::ResetState() { - MutexLock l(&mutex_); - db_file_state_.clear(); - new_files_since_last_dir_sync_.clear(); + // Since we are not destroying the database, the existing files + // should keep their recorded synced/flushed state. Therefore + // we do not reset db_file_state_ and new_files_since_last_dir_sync_. SetFilesystemActive(true); } @@ -313,12 +341,14 @@ Status FaultInjectionTestEnv::DeleteFilesCreatedAfterLastDirSync() { std::set new_files(new_files_since_last_dir_sync_.begin(), new_files_since_last_dir_sync_.end()); mutex_.Unlock(); - Status s; - std::set::const_iterator it; - for (it = new_files.begin(); s.ok() && it != new_files.end(); ++it) { - s = DeleteFile(*it); + Status status; + for (const auto& new_file : new_files) { + Status delete_status = DeleteFile(new_file); + if (!delete_status.ok() && status.ok()) { + status = std::move(delete_status); + } } - return s; + return status; } void FaultInjectionTestEnv::WritableFileClosed(const FileState& state) { @@ -327,7 +357,7 @@ void FaultInjectionTestEnv::WritableFileClosed(const FileState& state) { } Status FileState::DropUnsyncedData() const { - ssize_t sync_pos = pos_at_last_sync_ == -1 ? 0 : pos_at_last_sync_; + int64_t sync_pos = pos_at_last_sync_ == -1 ? 0 : pos_at_last_sync_; return Truncate(filename_, sync_pos); } @@ -342,53 +372,28 @@ class FaultInjectionTest { Options options_; DB* db_; - FaultInjectionTest() : env_(NULL), tiny_cache_(NULL), db_(NULL) { NewDB(); } - - ~FaultInjectionTest() { ASSERT_OK(TearDown()); } - - Status NewDB() { - assert(db_ == NULL); - assert(tiny_cache_ == NULL); - assert(env_ == NULL); - - env_ = new FaultInjectionTestEnv(); - - options_ = Options(); + FaultInjectionTest() + : env_(new FaultInjectionTestEnv), + tiny_cache_(NewLRUCache(100)), + db_(nullptr) { + dbname_ = test::TmpDir() + "/fault_test"; + DestroyDB(dbname_, Options()); // Destroy any db from earlier run + options_.reuse_logs = true; options_.env = env_; options_.paranoid_checks = true; - - tiny_cache_ = NewLRUCache(100); options_.block_cache = tiny_cache_; - dbname_ = test::TmpDir() + "/fault_test"; - options_.create_if_missing = true; - Status s = OpenDB(); - options_.create_if_missing = false; - return s; } - Status SetUp() { - Status s = TearDown(); - if (s.ok()) { - s = NewDB(); - } - return s; - } - - Status TearDown() { + ~FaultInjectionTest() { CloseDB(); - - Status s = DestroyDB(dbname_, Options()); - + DestroyDB(dbname_, Options()); delete tiny_cache_; - tiny_cache_ = NULL; - delete env_; - env_ = NULL; - - return s; } + void ReuseLogs(bool reuse) { options_.reuse_logs = reuse; } + void Build(int start_idx, int num_vals) { std::string key_space, value_space; WriteBatch batch; @@ -447,19 +452,18 @@ class FaultInjectionTest { Status OpenDB() { delete db_; - db_ = NULL; + db_ = nullptr; env_->ResetState(); return DB::Open(options_, dbname_, &db_); } void CloseDB() { delete db_; - db_ = NULL; + db_ = nullptr; } void DeleteAllData() { Iterator* iter = db_->NewIterator(ReadOptions()); - WriteOptions options; for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { ASSERT_OK(db_->Delete(WriteOptions(), iter->key())); } @@ -483,60 +487,65 @@ class FaultInjectionTest { void PartialCompactTestPreFault(int num_pre_sync, int num_post_sync) { DeleteAllData(); Build(0, num_pre_sync); - db_->CompactRange(NULL, NULL); + db_->CompactRange(nullptr, nullptr); Build(num_pre_sync, num_post_sync); } void PartialCompactTestReopenWithFault(ResetMethod reset_method, - int num_pre_sync, - int num_post_sync) { + int num_pre_sync, int num_post_sync) { env_->SetFilesystemActive(false); CloseDB(); ResetDBState(reset_method); ASSERT_OK(OpenDB()); ASSERT_OK(Verify(0, num_pre_sync, FaultInjectionTest::VAL_EXPECT_NO_ERROR)); - ASSERT_OK(Verify(num_pre_sync, num_post_sync, FaultInjectionTest::VAL_EXPECT_ERROR)); + ASSERT_OK(Verify(num_pre_sync, num_post_sync, + FaultInjectionTest::VAL_EXPECT_ERROR)); } - void NoWriteTestPreFault() { - } + void NoWriteTestPreFault() {} void NoWriteTestReopenWithFault(ResetMethod reset_method) { CloseDB(); ResetDBState(reset_method); ASSERT_OK(OpenDB()); } + + void DoTest() { + Random rnd(0); + ASSERT_OK(OpenDB()); + 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(RESET_DROP_UNSYNCED_DATA, num_pre_sync, + num_post_sync); + + NoWriteTestPreFault(); + NoWriteTestReopenWithFault(RESET_DROP_UNSYNCED_DATA); + + PartialCompactTestPreFault(num_pre_sync, num_post_sync); + // No new files created so we expect all values since no files will be + // dropped. + PartialCompactTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES, + num_pre_sync + num_post_sync, 0); + + NoWriteTestPreFault(); + NoWriteTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES); + } + } }; -TEST(FaultInjectionTest, FaultTest) { - Random rnd(0); - ASSERT_OK(SetUp()); - for (size_t idx = 0; idx < kNumIterations; idx++) { - int num_pre_sync = rnd.Uniform(kMaxNumValues); - int num_post_sync = rnd.Uniform(kMaxNumValues); +TEST(FaultInjectionTest, FaultTestNoLogReuse) { + ReuseLogs(false); + DoTest(); +} - PartialCompactTestPreFault(num_pre_sync, num_post_sync); - PartialCompactTestReopenWithFault(RESET_DROP_UNSYNCED_DATA, - num_pre_sync, - num_post_sync); - - NoWriteTestPreFault(); - NoWriteTestReopenWithFault(RESET_DROP_UNSYNCED_DATA); - - PartialCompactTestPreFault(num_pre_sync, num_post_sync); - // No new files created so we expect all values since no files will be - // dropped. - PartialCompactTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES, - num_pre_sync + num_post_sync, - 0); - - NoWriteTestPreFault(); - NoWriteTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES); - } +TEST(FaultInjectionTest, FaultTestWithLogReuse) { + ReuseLogs(true); + DoTest(); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/filename.cc b/db/filename.cc index da32946..85de45c 100644 --- a/db/filename.cc +++ b/db/filename.cc @@ -2,9 +2,11 @@ // 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 #include -#include "db/filename.h" + #include "db/dbformat.h" #include "leveldb/env.h" #include "util/logging.h" @@ -12,31 +14,30 @@ namespace leveldb { // A utility routine: write "data" to the named file and Sync() it. -extern Status WriteStringToFileSync(Env* env, const Slice& data, - const std::string& fname); +Status WriteStringToFileSync(Env* env, const Slice& data, + const std::string& fname); -static std::string MakeFileName(const std::string& name, uint64_t number, +static std::string MakeFileName(const std::string& dbname, uint64_t number, const char* suffix) { char buf[100]; snprintf(buf, sizeof(buf), "/%06llu.%s", - static_cast(number), - suffix); - return name + buf; + static_cast(number), suffix); + return dbname + buf; } -std::string LogFileName(const std::string& name, uint64_t number) { +std::string LogFileName(const std::string& dbname, uint64_t number) { assert(number > 0); - return MakeFileName(name, number, "log"); + return MakeFileName(dbname, number, "log"); } -std::string TableFileName(const std::string& name, uint64_t number) { +std::string TableFileName(const std::string& dbname, uint64_t number) { assert(number > 0); - return MakeFileName(name, number, "ldb"); + return MakeFileName(dbname, number, "ldb"); } -std::string SSTTableFileName(const std::string& name, uint64_t number) { +std::string SSTTableFileName(const std::string& dbname, uint64_t number) { assert(number > 0); - return MakeFileName(name, number, "sst"); + return MakeFileName(dbname, number, "sst"); } std::string DescriptorFileName(const std::string& dbname, uint64_t number) { @@ -51,9 +52,7 @@ std::string CurrentFileName(const std::string& dbname) { return dbname + "/CURRENT"; } -std::string LockFileName(const std::string& dbname) { - return dbname + "/LOCK"; -} +std::string LockFileName(const std::string& dbname) { return dbname + "/LOCK"; } std::string TempFileName(const std::string& dbname, uint64_t number) { assert(number > 0); @@ -69,7 +68,6 @@ std::string OldInfoLogFileName(const std::string& dbname) { return dbname + "/LOG.old"; } - // Owned filenames have the form: // dbname/CURRENT // dbname/LOCK @@ -77,10 +75,9 @@ std::string OldInfoLogFileName(const std::string& dbname) { // dbname/LOG.old // dbname/MANIFEST-[0-9]+ // dbname/[0-9]+.(log|sst|ldb) -bool ParseFileName(const std::string& fname, - uint64_t* number, +bool ParseFileName(const std::string& filename, uint64_t* number, FileType* type) { - Slice rest(fname); + Slice rest(filename); if (rest == "CURRENT") { *number = 0; *type = kCurrentFile; diff --git a/db/filename.h b/db/filename.h index 87a7526..524e813 100644 --- a/db/filename.h +++ b/db/filename.h @@ -8,7 +8,9 @@ #define STORAGE_LEVELDB_DB_FILENAME_H_ #include + #include + #include "leveldb/slice.h" #include "leveldb/status.h" #include "port/port.h" @@ -30,55 +32,52 @@ enum FileType { // 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); +std::string LogFileName(const std::string& dbname, uint64_t number); // 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::string& dbname, uint64_t number); +std::string TableFileName(const std::string& dbname, uint64_t number); // Return the legacy file name for an sstable with the specified number // in the db named by "dbname". The result will be prefixed with // "dbname". -extern std::string SSTTableFileName(const std::string& dbname, uint64_t number); +std::string SSTTableFileName(const std::string& dbname, uint64_t number); // 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); +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); +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); +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); +std::string TempFileName(const std::string& dbname, uint64_t number); // Return the name of the info log file for "dbname". -extern std::string InfoLogFileName(const std::string& dbname); +std::string InfoLogFileName(const std::string& dbname); // Return the name of the old info log file for "dbname". -extern std::string OldInfoLogFileName(const std::string& dbname); +std::string OldInfoLogFileName(const std::string& dbname); // If filename is a leveldb 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. -extern bool ParseFileName(const std::string& filename, - uint64_t* number, - FileType* type); +bool ParseFileName(const std::string& filename, uint64_t* number, + FileType* type); // 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); - +Status SetCurrentFile(Env* env, const std::string& dbname, + uint64_t descriptor_number); } // namespace leveldb diff --git a/db/filename_test.cc b/db/filename_test.cc index a32556d..952f320 100644 --- a/db/filename_test.cc +++ b/db/filename_test.cc @@ -11,7 +11,7 @@ namespace leveldb { -class FileNameTest { }; +class FileNameTest {}; TEST(FileNameTest, Parse) { Slice db; @@ -24,17 +24,17 @@ TEST(FileNameTest, Parse) { uint64_t number; FileType type; } cases[] = { - { "100.log", 100, kLogFile }, - { "0.log", 0, kLogFile }, - { "0.sst", 0, kTableFile }, - { "0.ldb", 0, kTableFile }, - { "CURRENT", 0, kCurrentFile }, - { "LOCK", 0, kDBLockFile }, - { "MANIFEST-2", 2, kDescriptorFile }, - { "MANIFEST-7", 7, kDescriptorFile }, - { "LOG", 0, kInfoLogFile }, - { "LOG.old", 0, kInfoLogFile }, - { "18446744073709551615.log", 18446744073709551615ull, kLogFile }, + {"100.log", 100, kLogFile}, + {"0.log", 0, kLogFile}, + {"0.sst", 0, kTableFile}, + {"0.ldb", 0, kTableFile}, + {"CURRENT", 0, kCurrentFile}, + {"LOCK", 0, kDBLockFile}, + {"MANIFEST-2", 2, kDescriptorFile}, + {"MANIFEST-7", 7, kDescriptorFile}, + {"LOG", 0, kInfoLogFile}, + {"LOG.old", 0, kInfoLogFile}, + {"18446744073709551615.log", 18446744073709551615ull, kLogFile}, }; for (int i = 0; i < sizeof(cases) / sizeof(cases[0]); i++) { std::string f = cases[i].fname; @@ -44,30 +44,28 @@ TEST(FileNameTest, Parse) { } // Errors - static const char* errors[] = { - "", - "foo", - "foo-dx-100.log", - ".log", - "", - "manifest", - "CURREN", - "CURRENTX", - "MANIFES", - "MANIFEST", - "MANIFEST-", - "XMANIFEST-3", - "MANIFEST-3x", - "LOC", - "LOCKx", - "LO", - "LOGx", - "18446744073709551616.log", - "184467440737095516150.log", - "100", - "100.", - "100.lop" - }; + static const char* errors[] = {"", + "foo", + "foo-dx-100.log", + ".log", + "", + "manifest", + "CURREN", + "CURRENTX", + "MANIFES", + "MANIFEST", + "MANIFEST-", + "XMANIFEST-3", + "MANIFEST-3x", + "LOC", + "LOCKx", + "LO", + "LOGx", + "18446744073709551616.log", + "184467440737095516150.log", + "100", + "100.", + "100.lop"}; for (int i = 0; i < sizeof(errors) / sizeof(errors[0]); i++) { std::string f = errors[i]; ASSERT_TRUE(!ParseFileName(f, &number, &type)) << f; @@ -114,10 +112,20 @@ TEST(FileNameTest, Construction) { ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); ASSERT_EQ(999, number); ASSERT_EQ(kTempFile, type); + + fname = InfoLogFileName("foo"); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(0, number); + ASSERT_EQ(kInfoLogFile, type); + + fname = OldInfoLogFileName("foo"); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(0, number); + ASSERT_EQ(kInfoLogFile, type); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/leveldb_main.cc b/db/leveldbutil.cc similarity index 74% rename from db/leveldb_main.cc rename to db/leveldbutil.cc index 9f4b7dd..55cdcc5 100644 --- a/db/leveldb_main.cc +++ b/db/leveldbutil.cc @@ -3,6 +3,7 @@ // found in the LICENSE file. See the AUTHORS file for names of contributors. #include + #include "leveldb/dumpfile.h" #include "leveldb/env.h" #include "leveldb/status.h" @@ -12,13 +13,13 @@ namespace { class StdoutPrinter : public WritableFile { public: - virtual Status Append(const Slice& data) { + Status Append(const Slice& data) override { fwrite(data.data(), 1, data.size(), stdout); return Status::OK(); } - virtual Status Close() { return Status::OK(); } - virtual Status Flush() { return Status::OK(); } - virtual Status Sync() { return Status::OK(); } + Status Close() override { return Status::OK(); } + Status Flush() override { return Status::OK(); } + Status Sync() override { return Status::OK(); } }; bool HandleDumpCommand(Env* env, char** files, int num) { @@ -38,11 +39,9 @@ bool HandleDumpCommand(Env* env, char** files, int num) { } // namespace leveldb static void Usage() { - fprintf( - stderr, - "Usage: leveldbutil command...\n" - " dump files... -- dump contents of specified files\n" - ); + fprintf(stderr, + "Usage: leveldbutil command...\n" + " dump files... -- dump contents of specified files\n"); } int main(int argc, char** argv) { @@ -54,7 +53,7 @@ int main(int argc, char** argv) { } else { std::string command = argv[1]; if (command == "dump") { - ok = leveldb::HandleDumpCommand(env, argv+2, argc-2); + ok = leveldb::HandleDumpCommand(env, argv + 2, argc - 2); } else { Usage(); ok = false; diff --git a/db/log_format.h b/db/log_format.h index a8c06ef..356e69f 100644 --- a/db/log_format.h +++ b/db/log_format.h @@ -3,7 +3,7 @@ // 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. +// See ../doc/log_format.md for more detail. #ifndef STORAGE_LEVELDB_DB_LOG_FORMAT_H_ #define STORAGE_LEVELDB_DB_LOG_FORMAT_H_ diff --git a/db/log_reader.cc b/db/log_reader.cc index e44b66c..b770fee 100644 --- a/db/log_reader.cc +++ b/db/log_reader.cc @@ -5,6 +5,7 @@ #include "db/log_reader.h" #include + #include "leveldb/env.h" #include "util/coding.h" #include "util/crc32c.h" @@ -12,8 +13,7 @@ namespace leveldb { namespace log { -Reader::Reporter::~Reporter() { -} +Reader::Reporter::~Reporter() = default; Reader::Reader(SequentialFile* file, Reporter* reporter, bool checksum, uint64_t initial_offset) @@ -25,20 +25,17 @@ Reader::Reader(SequentialFile* file, Reporter* reporter, bool checksum, eof_(false), last_record_offset_(0), end_of_buffer_offset_(0), - initial_offset_(initial_offset) { -} + initial_offset_(initial_offset), + resyncing_(initial_offset > 0) {} -Reader::~Reader() { - delete[] backing_store_; -} +Reader::~Reader() { delete[] backing_store_; } bool Reader::SkipToInitialBlock() { - size_t offset_in_block = initial_offset_ % kBlockSize; + const size_t offset_in_block = initial_offset_ % kBlockSize; uint64_t block_start_location = initial_offset_ - offset_in_block; // Don't search a block if we'd be in the trailer if (offset_in_block > kBlockSize - 6) { - offset_in_block = 0; block_start_location += kBlockSize; } @@ -72,8 +69,25 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) { Slice fragment; while (true) { - uint64_t physical_record_offset = end_of_buffer_offset_ - buffer_.size(); const unsigned int record_type = ReadPhysicalRecord(&fragment); + + // ReadPhysicalRecord may have only had an empty trailer remaining in its + // internal buffer. Calculate the offset of the next physical record now + // that it has returned, properly accounting for its header size. + uint64_t physical_record_offset = + end_of_buffer_offset_ - buffer_.size() - kHeaderSize - fragment.size(); + + if (resyncing_) { + if (record_type == kMiddleType) { + continue; + } else if (record_type == kLastType) { + resyncing_ = false; + continue; + } else { + resyncing_ = false; + } + } + switch (record_type) { case kFullType: if (in_fragmented_record) { @@ -81,9 +95,7 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) { // 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. - if (scratch->empty()) { - in_fragmented_record = false; - } else { + if (!scratch->empty()) { ReportCorruption(scratch->size(), "partial record without end(1)"); } } @@ -99,9 +111,7 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) { // 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. - if (scratch->empty()) { - in_fragmented_record = false; - } else { + if (!scratch->empty()) { ReportCorruption(scratch->size(), "partial record without end(2)"); } } @@ -163,16 +173,14 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) { return false; } -uint64_t Reader::LastRecordOffset() { - return last_record_offset_; -} +uint64_t Reader::LastRecordOffset() { return last_record_offset_; } void Reader::ReportCorruption(uint64_t bytes, const char* reason) { ReportDrop(bytes, Status::Corruption(reason)); } void Reader::ReportDrop(uint64_t bytes, const Status& reason) { - if (reporter_ != NULL && + if (reporter_ != nullptr && end_of_buffer_offset_ - buffer_.size() - bytes >= initial_offset_) { reporter_->Corruption(static_cast(bytes), reason); } diff --git a/db/log_reader.h b/db/log_reader.h index 6aff791..001da89 100644 --- a/db/log_reader.h +++ b/db/log_reader.h @@ -32,7 +32,7 @@ class Reader { // Create a reader that will return log records from "*file". // "*file" must remain live while this Reader is in use. // - // If "reporter" is non-NULL, it is notified whenever some data is + // If "reporter" is non-null, it is notified whenever some data is // dropped due to a detected corruption. "*reporter" must remain // live while this Reader is in use. // @@ -43,6 +43,9 @@ class Reader { Reader(SequentialFile* file, Reporter* reporter, bool checksum, uint64_t initial_offset); + Reader(const Reader&) = delete; + Reader& operator=(const Reader&) = delete; + ~Reader(); // Read the next record into *record. Returns true if read @@ -58,21 +61,6 @@ class Reader { uint64_t LastRecordOffset(); private: - SequentialFile* const file_; - Reporter* const reporter_; - bool const checksum_; - char* const backing_store_; - Slice buffer_; - bool eof_; // Last Read() indicated EOF by returning < kBlockSize - - // 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_; - // Extend record types with the following special values enum { kEof = kMaxRecordType + 1, @@ -97,9 +85,25 @@ class Reader { void ReportCorruption(uint64_t bytes, const char* reason); void ReportDrop(uint64_t bytes, const Status& reason); - // No copying allowed - Reader(const Reader&); - void operator=(const Reader&); + SequentialFile* const file_; + Reporter* const reporter_; + bool const checksum_; + char* const backing_store_; + Slice buffer_; + bool eof_; // Last Read() indicated EOF by returning < kBlockSize + + // 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_; + + // True if we are resynchronizing after a seek (initial_offset_ > 0). In + // particular, a run of kMiddleType and kLastType records can be silently + // skipped in this mode + bool resyncing_; }; } // namespace log diff --git a/db/log_test.cc b/db/log_test.cc index dcf0562..0e31648 100644 --- a/db/log_test.cc +++ b/db/log_test.cc @@ -37,95 +37,29 @@ static std::string RandomSkewedString(int i, Random* rnd) { } class LogTest { - private: - class StringDest : public WritableFile { - public: - std::string contents_; - - virtual Status Close() { return Status::OK(); } - virtual Status Flush() { return Status::OK(); } - virtual Status Sync() { return Status::OK(); } - virtual Status Append(const Slice& slice) { - contents_.append(slice.data(), slice.size()); - return Status::OK(); - } - }; - - class StringSource : public SequentialFile { - public: - Slice contents_; - bool force_error_; - bool returned_partial_; - StringSource() : force_error_(false), returned_partial_(false) { } - - virtual Status Read(size_t n, Slice* result, char* scratch) { - ASSERT_TRUE(!returned_partial_) << "must not Read() after eof/error"; - - if (force_error_) { - force_error_ = false; - returned_partial_ = true; - return Status::Corruption("read error"); - } - - if (contents_.size() < n) { - n = contents_.size(); - returned_partial_ = true; - } - *result = Slice(contents_.data(), n); - contents_.remove_prefix(n); - return Status::OK(); - } - - virtual Status Skip(uint64_t n) { - 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) { - dropped_bytes_ += bytes; - message_.append(status.ToString()); - } - }; - - StringDest dest_; - StringSource source_; - ReportCollector report_; - bool reading_; - Writer writer_; - Reader reader_; - - // Record metadata for testing initial offset functionality - static size_t initial_offset_record_sizes_[]; - static uint64_t initial_offset_last_record_offsets_[]; - public: - LogTest() : reading_(false), - writer_(&dest_), - reader_(&source_, &report_, true/*checksum*/, - 0/*initial_offset*/) { + LogTest() + : reading_(false), + writer_(new Writer(&dest_)), + reader_(new Reader(&source_, &report_, true /*checksum*/, + 0 /*initial_offset*/)) {} + + ~LogTest() { + delete writer_; + delete reader_; + } + + void ReopenForAppend() { + delete writer_; + writer_ = new Writer(&dest_, dest_.contents_.size()); } void Write(const std::string& msg) { ASSERT_TRUE(!reading_) << "Write() after starting to read"; - writer_.AddRecord(Slice(msg)); + writer_->AddRecord(Slice(msg)); } - size_t WrittenBytes() const { - return dest_.contents_.size(); - } + size_t WrittenBytes() const { return dest_.contents_.size(); } std::string Read() { if (!reading_) { @@ -134,7 +68,7 @@ class LogTest { } std::string scratch; Slice record; - if (reader_.ReadRecord(&record, &scratch)) { + if (reader_->ReadRecord(&record, &scratch)) { return record.ToString(); } else { return "EOF"; @@ -155,22 +89,16 @@ class LogTest { void FixChecksum(int header_offset, int len) { // Compute crc of type/len/data - uint32_t crc = crc32c::Value(&dest_.contents_[header_offset+6], 1 + len); + uint32_t crc = crc32c::Value(&dest_.contents_[header_offset + 6], 1 + len); crc = crc32c::Mask(crc); EncodeFixed32(&dest_.contents_[header_offset], crc); } - void ForceError() { - source_.force_error_ = true; - } + void ForceError() { source_.force_error_ = true; } - size_t DroppedBytes() const { - return report_.dropped_bytes_; - } + size_t DroppedBytes() const { return report_.dropped_bytes_; } - std::string ReportMessage() const { - return report_.message_; - } + std::string ReportMessage() const { return report_.message_; } // Returns OK iff recorded error message contains "msg" std::string MatchError(const std::string& msg) const { @@ -182,18 +110,23 @@ class LogTest { } void WriteInitialOffsetLog() { - for (int i = 0; i < 4; i++) { + for (int i = 0; i < num_initial_offset_records_; i++) { std::string record(initial_offset_record_sizes_[i], static_cast('a' + i)); Write(record); } } + void StartReadingAt(uint64_t initial_offset) { + delete reader_; + reader_ = new Reader(&source_, &report_, true /*checksum*/, initial_offset); + } + void CheckOffsetPastEndReturnsNoRecords(uint64_t offset_past_end) { WriteInitialOffsetLog(); reading_ = true; source_.contents_ = Slice(dest_.contents_); - Reader* offset_reader = new Reader(&source_, &report_, true/*checksum*/, + Reader* offset_reader = new Reader(&source_, &report_, true /*checksum*/, WrittenBytes() + offset_past_end); Slice record; std::string scratch; @@ -206,38 +139,126 @@ class LogTest { WriteInitialOffsetLog(); reading_ = true; source_.contents_ = Slice(dest_.contents_); - Reader* offset_reader = new Reader(&source_, &report_, true/*checksum*/, - initial_offset); - 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]); + Reader* offset_reader = + new Reader(&source_, &report_, true /*checksum*/, initial_offset); + + // Read all records from expected_record_offset through the last one. + ASSERT_LT(expected_record_offset, num_initial_offset_records_); + for (; expected_record_offset < num_initial_offset_records_; + ++expected_record_offset) { + 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]); + } delete offset_reader; } + private: + class StringDest : public WritableFile { + public: + Status Close() override { return Status::OK(); } + Status Flush() override { return Status::OK(); } + Status Sync() override { return Status::OK(); } + Status Append(const Slice& slice) override { + contents_.append(slice.data(), slice.size()); + return Status::OK(); + } + + std::string contents_; + }; + + class StringSource : public SequentialFile { + public: + StringSource() : force_error_(false), returned_partial_(false) {} + + Status Read(size_t n, Slice* result, char* scratch) override { + ASSERT_TRUE(!returned_partial_) << "must not Read() after eof/error"; + + if (force_error_) { + force_error_ = false; + returned_partial_ = true; + return Status::Corruption("read error"); + } + + if (contents_.size() < n) { + n = contents_.size(); + returned_partial_ = true; + } + *result = Slice(contents_.data(), n); + contents_.remove_prefix(n); + return Status::OK(); + } + + Status Skip(uint64_t n) override { + if (n > contents_.size()) { + contents_.clear(); + return Status::NotFound("in-memory file skipped past end"); + } + + contents_.remove_prefix(n); + + return Status::OK(); + } + + Slice contents_; + bool force_error_; + bool returned_partial_; + }; + + class ReportCollector : public Reader::Reporter { + public: + ReportCollector() : dropped_bytes_(0) {} + void Corruption(size_t bytes, const Status& status) override { + dropped_bytes_ += bytes; + message_.append(status.ToString()); + } + + size_t dropped_bytes_; + std::string message_; + }; + + // Record metadata for testing initial offset functionality + static size_t initial_offset_record_sizes_[]; + static uint64_t initial_offset_last_record_offsets_[]; + static int num_initial_offset_records_; + + StringDest dest_; + StringSource source_; + ReportCollector report_; + bool reading_; + Writer* writer_; + Reader* reader_; }; -size_t LogTest::initial_offset_record_sizes_[] = - {10000, // Two sizable records in first block - 10000, - 2 * log::kBlockSize - 1000, // Span three blocks - 1}; +size_t LogTest::initial_offset_record_sizes_[] = { + 10000, // Two sizable records in first block + 10000, + 2 * log::kBlockSize - 1000, // Span three blocks + 1, + 13716, // Consume all but two bytes of block 3. + log::kBlockSize - kHeaderSize, // Consume the entirety of block 4. +}; -uint64_t LogTest::initial_offset_last_record_offsets_[] = - {0, - kHeaderSize + 10000, - 2 * (kHeaderSize + 10000), - 2 * (kHeaderSize + 10000) + - (2 * log::kBlockSize - 1000) + 3 * kHeaderSize}; +uint64_t LogTest::initial_offset_last_record_offsets_[] = { + 0, + kHeaderSize + 10000, + 2 * (kHeaderSize + 10000), + 2 * (kHeaderSize + 10000) + (2 * log::kBlockSize - 1000) + 3 * kHeaderSize, + 2 * (kHeaderSize + 10000) + (2 * log::kBlockSize - 1000) + 3 * kHeaderSize + + kHeaderSize + 1, + 3 * log::kBlockSize, +}; +// LogTest::initial_offset_last_record_offsets_ must be defined before this. +int LogTest::num_initial_offset_records_ = + sizeof(LogTest::initial_offset_last_record_offsets_) / sizeof(uint64_t); -TEST(LogTest, Empty) { - ASSERT_EQ("EOF", Read()); -} +TEST(LogTest, Empty) { ASSERT_EQ("EOF", Read()); } TEST(LogTest, ReadWrite) { Write("foo"); @@ -274,7 +295,7 @@ TEST(LogTest, Fragmentation) { TEST(LogTest, MarginalTrailer) { // Make a trailer that is exactly the same length as an empty record. - const int n = kBlockSize - 2*kHeaderSize; + const int n = kBlockSize - 2 * kHeaderSize; Write(BigString("foo", n)); ASSERT_EQ(kBlockSize - kHeaderSize, WrittenBytes()); Write(""); @@ -287,7 +308,7 @@ TEST(LogTest, MarginalTrailer) { TEST(LogTest, MarginalTrailer2) { // Make a trailer that is exactly the same length as an empty record. - const int n = kBlockSize - 2*kHeaderSize; + const int n = kBlockSize - 2 * kHeaderSize; Write(BigString("foo", n)); ASSERT_EQ(kBlockSize - kHeaderSize, WrittenBytes()); Write("bar"); @@ -299,7 +320,7 @@ TEST(LogTest, MarginalTrailer2) { } TEST(LogTest, ShortTrailer) { - const int n = kBlockSize - 2*kHeaderSize + 4; + const int n = kBlockSize - 2 * kHeaderSize + 4; Write(BigString("foo", n)); ASSERT_EQ(kBlockSize - kHeaderSize + 4, WrittenBytes()); Write(""); @@ -311,13 +332,22 @@ TEST(LogTest, ShortTrailer) { } TEST(LogTest, AlignedEof) { - const int n = kBlockSize - 2*kHeaderSize + 4; + const int n = kBlockSize - 2 * kHeaderSize + 4; Write(BigString("foo", n)); ASSERT_EQ(kBlockSize - kHeaderSize + 4, WrittenBytes()); ASSERT_EQ(BigString("foo", n), Read()); ASSERT_EQ("EOF", Read()); } +TEST(LogTest, OpenForAppend) { + Write("hello"); + ReopenForAppend(); + Write("world"); + ASSERT_EQ("hello", Read()); + ASSERT_EQ("world", Read()); + ASSERT_EQ("EOF", Read()); +} + TEST(LogTest, RandomRead) { const int N = 500; Random write_rnd(301); @@ -353,7 +383,7 @@ TEST(LogTest, BadRecordType) { TEST(LogTest, TruncatedTrailingRecordIsIgnored) { Write("foo"); - ShrinkSize(4); // Drop all payload as well as a header byte + 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(0, DroppedBytes()); @@ -445,6 +475,22 @@ TEST(LogTest, PartialLastIsIgnored) { ASSERT_EQ(0, DroppedBytes()); } +TEST(LogTest, SkipIntoMultiRecord) { + // Consider a fragmented record: + // first(R1), middle(R1), last(R1), first(R2) + // If initial_offset points to a record after first(R1) but before first(R2) + // incomplete fragment errors are not actual errors, and must be suppressed + // until a new first or full record is encountered. + Write(BigString("foo", 3 * kBlockSize)); + Write("correct"); + StartReadingAt(kBlockSize); + + ASSERT_EQ("correct", Read()); + ASSERT_EQ("", ReportMessage()); + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("EOF", Read()); +} + TEST(LogTest, ErrorJoinsRecords) { // Consider two fragmented records: // first(R1) last(R1) first(R2) last(R2) @@ -457,44 +503,30 @@ TEST(LogTest, ErrorJoinsRecords) { Write("correct"); // Wipe the middle block - for (int offset = kBlockSize; offset < 2*kBlockSize; offset++) { + for (int offset = kBlockSize; offset < 2 * kBlockSize; offset++) { SetByte(offset, 'x'); } ASSERT_EQ("correct", Read()); ASSERT_EQ("EOF", Read()); const size_t dropped = DroppedBytes(); - ASSERT_LE(dropped, 2*kBlockSize + 100); - ASSERT_GE(dropped, 2*kBlockSize); + ASSERT_LE(dropped, 2 * kBlockSize + 100); + ASSERT_GE(dropped, 2 * kBlockSize); } -TEST(LogTest, ReadStart) { - CheckInitialOffsetRecord(0, 0); -} +TEST(LogTest, ReadStart) { CheckInitialOffsetRecord(0, 0); } -TEST(LogTest, ReadSecondOneOff) { - CheckInitialOffsetRecord(1, 1); -} +TEST(LogTest, ReadSecondOneOff) { CheckInitialOffsetRecord(1, 1); } -TEST(LogTest, ReadSecondTenThousand) { - CheckInitialOffsetRecord(10000, 1); -} +TEST(LogTest, ReadSecondTenThousand) { CheckInitialOffsetRecord(10000, 1); } -TEST(LogTest, ReadSecondStart) { - CheckInitialOffsetRecord(10007, 1); -} +TEST(LogTest, ReadSecondStart) { CheckInitialOffsetRecord(10007, 1); } -TEST(LogTest, ReadThirdOneOff) { - CheckInitialOffsetRecord(10008, 2); -} +TEST(LogTest, ReadThirdOneOff) { CheckInitialOffsetRecord(10008, 2); } -TEST(LogTest, ReadThirdStart) { - CheckInitialOffsetRecord(20014, 2); -} +TEST(LogTest, ReadThirdStart) { CheckInitialOffsetRecord(20014, 2); } -TEST(LogTest, ReadFourthOneOff) { - CheckInitialOffsetRecord(20015, 3); -} +TEST(LogTest, ReadFourthOneOff) { CheckInitialOffsetRecord(20015, 3); } TEST(LogTest, ReadFourthFirstBlockTrailer) { CheckInitialOffsetRecord(log::kBlockSize - 4, 3); @@ -514,17 +546,15 @@ TEST(LogTest, ReadFourthStart) { 3); } -TEST(LogTest, ReadEnd) { - CheckOffsetPastEndReturnsNoRecords(0); +TEST(LogTest, ReadInitialOffsetIntoBlockPadding) { + CheckInitialOffsetRecord(3 * log::kBlockSize - 3, 5); } -TEST(LogTest, ReadPastEnd) { - CheckOffsetPastEndReturnsNoRecords(5); -} +TEST(LogTest, ReadEnd) { CheckOffsetPastEndReturnsNoRecords(0); } + +TEST(LogTest, ReadPastEnd) { CheckOffsetPastEndReturnsNoRecords(5); } } // namespace log } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/log_writer.cc b/db/log_writer.cc index 2da99ac..bfb16fb 100644 --- a/db/log_writer.cc +++ b/db/log_writer.cc @@ -5,6 +5,7 @@ #include "db/log_writer.h" #include + #include "leveldb/env.h" #include "util/coding.h" #include "util/crc32c.h" @@ -12,18 +13,24 @@ namespace leveldb { namespace log { -Writer::Writer(WritableFile* dest) - : dest_(dest), - block_offset_(0) { +static void InitTypeCrc(uint32_t* type_crc) { for (int i = 0; i <= kMaxRecordType; i++) { char t = static_cast(i); - type_crc_[i] = crc32c::Value(&t, 1); + type_crc[i] = crc32c::Value(&t, 1); } } -Writer::~Writer() { +Writer::Writer(WritableFile* dest) : dest_(dest), block_offset_(0) { + InitTypeCrc(type_crc_); } +Writer::Writer(WritableFile* dest, uint64_t dest_length) + : dest_(dest), block_offset_(dest_length % kBlockSize) { + InitTypeCrc(type_crc_); +} + +Writer::~Writer() = default; + Status Writer::AddRecord(const Slice& slice) { const char* ptr = slice.data(); size_t left = slice.size(); @@ -40,7 +47,7 @@ Status Writer::AddRecord(const Slice& slice) { // Switch to a new block if (leftover > 0) { // Fill the trailer (literal below relies on kHeaderSize being 7) - assert(kHeaderSize == 7); + static_assert(kHeaderSize == 7, ""); dest_->Append(Slice("\x00\x00\x00\x00\x00\x00", leftover)); } block_offset_ = 0; @@ -72,30 +79,31 @@ Status Writer::AddRecord(const Slice& slice) { return s; } -Status Writer::EmitPhysicalRecord(RecordType t, const char* ptr, size_t n) { - assert(n <= 0xffff); // Must fit in two bytes - assert(block_offset_ + kHeaderSize + n <= kBlockSize); +Status Writer::EmitPhysicalRecord(RecordType t, const char* ptr, + size_t length) { + assert(length <= 0xffff); // Must fit in two bytes + assert(block_offset_ + kHeaderSize + length <= kBlockSize); // Format the header char buf[kHeaderSize]; - buf[4] = static_cast(n & 0xff); - buf[5] = static_cast(n >> 8); + buf[4] = static_cast(length & 0xff); + buf[5] = static_cast(length >> 8); buf[6] = static_cast(t); // Compute the crc of the record type and the payload. - uint32_t crc = crc32c::Extend(type_crc_[t], ptr, n); - crc = crc32c::Mask(crc); // Adjust for storage + uint32_t crc = crc32c::Extend(type_crc_[t], ptr, length); + crc = crc32c::Mask(crc); // Adjust for storage EncodeFixed32(buf, crc); // Write the header and the payload Status s = dest_->Append(Slice(buf, kHeaderSize)); if (s.ok()) { - s = dest_->Append(Slice(ptr, n)); + s = dest_->Append(Slice(ptr, length)); if (s.ok()) { s = dest_->Flush(); } } - block_offset_ += kHeaderSize + n; + block_offset_ += kHeaderSize + length; return s; } diff --git a/db/log_writer.h b/db/log_writer.h index a3a954d..c0a2114 100644 --- a/db/log_writer.h +++ b/db/log_writer.h @@ -6,6 +6,7 @@ #define STORAGE_LEVELDB_DB_LOG_WRITER_H_ #include + #include "db/log_format.h" #include "leveldb/slice.h" #include "leveldb/status.h" @@ -22,24 +23,29 @@ class Writer { // "*dest" must be initially empty. // "*dest" must remain live while this Writer is in use. explicit Writer(WritableFile* dest); + + // Create a writer that will append data to "*dest". + // "*dest" must have initial length "dest_length". + // "*dest" must remain live while this Writer is in use. + Writer(WritableFile* dest, uint64_t dest_length); + + Writer(const Writer&) = delete; + Writer& operator=(const Writer&) = delete; + ~Writer(); Status AddRecord(const Slice& slice); private: + Status EmitPhysicalRecord(RecordType type, const char* ptr, size_t length); + WritableFile* dest_; - int block_offset_; // Current offset in block + int block_offset_; // Current offset in block // 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 diff --git a/db/memtable.cc b/db/memtable.cc index bfec0a7..00931d4 100644 --- a/db/memtable.cc +++ b/db/memtable.cc @@ -18,20 +18,15 @@ static Slice GetLengthPrefixedSlice(const char* data) { return Slice(p, len); } -MemTable::MemTable(const InternalKeyComparator& cmp) - : comparator_(cmp), - refs_(0), - table_(comparator_, &arena_) { -} +MemTable::MemTable(const InternalKeyComparator& comparator) + : comparator_(comparator), refs_(0), table_(comparator_, &arena_) {} -MemTable::~MemTable() { - assert(refs_ == 0); -} +MemTable::~MemTable() { assert(refs_ == 0); } size_t MemTable::ApproximateMemoryUsage() { return arena_.MemoryUsage(); } -int MemTable::KeyComparator::operator()(const char* aptr, const char* bptr) - const { +int MemTable::KeyComparator::operator()(const char* aptr, + const char* bptr) const { // Internal keys are encoded as length-prefixed strings. Slice a = GetLengthPrefixedSlice(aptr); Slice b = GetLengthPrefixedSlice(bptr); @@ -48,39 +43,37 @@ static const char* EncodeKey(std::string* scratch, const Slice& target) { return scratch->data(); } -class MemTableIterator: public Iterator { +class MemTableIterator : public Iterator { public: - explicit MemTableIterator(MemTable::Table* table) : iter_(table) { } + explicit MemTableIterator(MemTable::Table* table) : iter_(table) {} - virtual bool Valid() const { return iter_.Valid(); } - virtual void Seek(const Slice& k) { iter_.Seek(EncodeKey(&tmp_, k)); } - virtual void SeekToFirst() { iter_.SeekToFirst(); } - virtual void SeekToLast() { iter_.SeekToLast(); } - virtual void Next() { iter_.Next(); } - virtual void Prev() { iter_.Prev(); } - virtual Slice key() const { return GetLengthPrefixedSlice(iter_.key()); } - virtual Slice value() const { + MemTableIterator(const MemTableIterator&) = delete; + MemTableIterator& operator=(const MemTableIterator&) = delete; + + ~MemTableIterator() override = default; + + bool Valid() const override { return iter_.Valid(); } + void Seek(const Slice& k) override { iter_.Seek(EncodeKey(&tmp_, k)); } + void SeekToFirst() override { iter_.SeekToFirst(); } + void SeekToLast() override { iter_.SeekToLast(); } + void Next() override { iter_.Next(); } + void Prev() override { iter_.Prev(); } + Slice key() const override { return GetLengthPrefixedSlice(iter_.key()); } + Slice value() const override { Slice key_slice = GetLengthPrefixedSlice(iter_.key()); return GetLengthPrefixedSlice(key_slice.data() + key_slice.size()); } - virtual Status status() const { return Status::OK(); } + Status status() const override { return Status::OK(); } private: MemTable::Table::Iterator iter_; - std::string tmp_; // For passing to EncodeKey - - // No copying allowed - MemTableIterator(const MemTableIterator&); - void operator=(const MemTableIterator&); + std::string tmp_; // For passing to EncodeKey }; -Iterator* MemTable::NewIterator() { - return new MemTableIterator(&table_); -} +Iterator* MemTable::NewIterator() { return new MemTableIterator(&table_); } -void MemTable::Add(SequenceNumber s, ValueType type, - const Slice& key, +void MemTable::Add(SequenceNumber s, ValueType type, const Slice& key, const Slice& value) { // Format of an entry is concatenation of: // key_size : varint32 of internal_key.size() @@ -90,9 +83,9 @@ void MemTable::Add(SequenceNumber s, ValueType type, size_t key_size = key.size(); size_t val_size = value.size(); size_t internal_key_size = key_size + 8; - const size_t encoded_len = - VarintLength(internal_key_size) + internal_key_size + - VarintLength(val_size) + val_size; + const size_t encoded_len = VarintLength(internal_key_size) + + internal_key_size + VarintLength(val_size) + + val_size; char* buf = arena_.Allocate(encoded_len); char* p = EncodeVarint32(buf, internal_key_size); memcpy(p, key.data(), key_size); @@ -101,7 +94,7 @@ void MemTable::Add(SequenceNumber s, ValueType type, p += 8; p = EncodeVarint32(p, val_size); memcpy(p, value.data(), val_size); - assert((p + val_size) - buf == encoded_len); + assert(p + val_size == buf + encoded_len); table_.Insert(buf); } @@ -121,10 +114,9 @@ bool MemTable::Get(const LookupKey& key, std::string* value, Status* s) { // all entries with overly large sequence numbers. const char* entry = iter.key(); uint32_t key_length; - const char* key_ptr = GetVarint32Ptr(entry, entry+5, &key_length); + const char* key_ptr = GetVarint32Ptr(entry, entry + 5, &key_length); if (comparator_.comparator.user_comparator()->Compare( - Slice(key_ptr, key_length - 8), - key.user_key()) == 0) { + Slice(key_ptr, key_length - 8), key.user_key()) == 0) { // Correct user key const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8); switch (static_cast(tag & 0xff)) { diff --git a/db/memtable.h b/db/memtable.h index 92e90bb..9d986b1 100644 --- a/db/memtable.h +++ b/db/memtable.h @@ -6,15 +6,15 @@ #define STORAGE_LEVELDB_DB_MEMTABLE_H_ #include -#include "leveldb/db.h" + #include "db/dbformat.h" #include "db/skiplist.h" +#include "leveldb/db.h" #include "util/arena.h" namespace leveldb { class InternalKeyComparator; -class Mutex; class MemTableIterator; class MemTable { @@ -23,6 +23,9 @@ class MemTable { // is zero and the caller must call Ref() at least once. explicit MemTable(const InternalKeyComparator& comparator); + MemTable(const MemTable&) = delete; + MemTable& operator=(const MemTable&) = delete; + // Increase reference count. void Ref() { ++refs_; } @@ -36,10 +39,7 @@ class MemTable { } // 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. + // data structure. It is safe to call when MemTable is being modified. size_t ApproximateMemoryUsage(); // Return an iterator that yields the contents of the memtable. @@ -53,8 +53,7 @@ class MemTable { // 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. - void Add(SequenceNumber seq, ValueType type, - const Slice& key, + void Add(SequenceNumber seq, ValueType type, const Slice& key, const Slice& value); // If memtable contains a value for key, store it in *value and return true. @@ -64,26 +63,23 @@ class MemTable { bool Get(const LookupKey& key, std::string* value, Status* s); private: - ~MemTable(); // Private since only Unref() should be used to delete it - - struct KeyComparator { - const InternalKeyComparator comparator; - explicit KeyComparator(const InternalKeyComparator& c) : comparator(c) { } - int operator()(const char* a, const char* b) const; - }; friend class MemTableIterator; friend class MemTableBackwardIterator; + struct KeyComparator { + const InternalKeyComparator comparator; + explicit KeyComparator(const InternalKeyComparator& c) : comparator(c) {} + int operator()(const char* a, const char* b) const; + }; + typedef SkipList Table; + ~MemTable(); // Private since only Unref() should be used to delete it + KeyComparator comparator_; int refs_; Arena arena_; Table table_; - - // No copying allowed - MemTable(const MemTable&); - void operator=(const MemTable&); }; } // namespace leveldb diff --git a/db/recovery_test.cc b/db/recovery_test.cc new file mode 100644 index 0000000..547a959 --- /dev/null +++ b/db/recovery_test.cc @@ -0,0 +1,330 @@ +// 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. + +#include "db/db_impl.h" +#include "db/filename.h" +#include "db/version_set.h" +#include "db/write_batch_internal.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/write_batch.h" +#include "util/logging.h" +#include "util/testharness.h" +#include "util/testutil.h" + +namespace leveldb { + +class RecoveryTest { + public: + RecoveryTest() : env_(Env::Default()), db_(nullptr) { + dbname_ = test::TmpDir() + "/recovery_test"; + DestroyDB(dbname_, Options()); + Open(); + } + + ~RecoveryTest() { + Close(); + DestroyDB(dbname_, Options()); + } + + DBImpl* dbfull() const { return reinterpret_cast(db_); } + Env* env() const { return env_; } + + bool CanAppend() { + WritableFile* tmp; + Status s = env_->NewAppendableFile(CurrentFileName(dbname_), &tmp); + delete tmp; + if (s.IsNotSupportedError()) { + return false; + } else { + return true; + } + } + + void Close() { + delete db_; + db_ = nullptr; + } + + Status OpenWithStatus(Options* options = nullptr) { + Close(); + Options opts; + if (options != nullptr) { + opts = *options; + } else { + opts.reuse_logs = true; // TODO(sanjay): test both ways + opts.create_if_missing = true; + } + if (opts.env == nullptr) { + opts.env = env_; + } + return DB::Open(opts, dbname_, &db_); + } + + void Open(Options* options = nullptr) { + ASSERT_OK(OpenWithStatus(options)); + ASSERT_EQ(1, NumLogs()); + } + + Status Put(const std::string& k, const std::string& v) { + return db_->Put(WriteOptions(), k, v); + } + + std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) { + std::string result; + Status s = db_->Get(ReadOptions(), k, &result); + if (s.IsNotFound()) { + result = "NOT_FOUND"; + } else if (!s.ok()) { + result = s.ToString(); + } + return result; + } + + std::string ManifestFileName() { + std::string current; + ASSERT_OK(ReadFileToString(env_, CurrentFileName(dbname_), ¤t)); + size_t len = current.size(); + if (len > 0 && current[len - 1] == '\n') { + current.resize(len - 1); + } + return dbname_ + "/" + current; + } + + std::string LogName(uint64_t number) { return LogFileName(dbname_, number); } + + size_t DeleteLogFiles() { + // Linux allows unlinking open files, but Windows does not. + // Closing the db allows for file deletion. + Close(); + std::vector logs = GetFiles(kLogFile); + for (size_t i = 0; i < logs.size(); i++) { + ASSERT_OK(env_->DeleteFile(LogName(logs[i]))) << LogName(logs[i]); + } + return logs.size(); + } + + void DeleteManifestFile() { ASSERT_OK(env_->DeleteFile(ManifestFileName())); } + + uint64_t FirstLogFile() { return GetFiles(kLogFile)[0]; } + + std::vector GetFiles(FileType t) { + std::vector filenames; + ASSERT_OK(env_->GetChildren(dbname_, &filenames)); + std::vector result; + for (size_t i = 0; i < filenames.size(); i++) { + uint64_t number; + FileType type; + if (ParseFileName(filenames[i], &number, &type) && type == t) { + result.push_back(number); + } + } + return result; + } + + int NumLogs() { return GetFiles(kLogFile).size(); } + + int NumTables() { return GetFiles(kTableFile).size(); } + + uint64_t FileSize(const std::string& fname) { + uint64_t result; + ASSERT_OK(env_->GetFileSize(fname, &result)) << fname; + return result; + } + + void CompactMemTable() { dbfull()->TEST_CompactMemTable(); } + + // Directly construct a log file that sets key to val. + void MakeLogFile(uint64_t lognum, SequenceNumber seq, Slice key, Slice val) { + std::string fname = LogFileName(dbname_, lognum); + WritableFile* file; + ASSERT_OK(env_->NewWritableFile(fname, &file)); + log::Writer writer(file); + WriteBatch batch; + batch.Put(key, val); + WriteBatchInternal::SetSequence(&batch, seq); + ASSERT_OK(writer.AddRecord(WriteBatchInternal::Contents(&batch))); + ASSERT_OK(file->Flush()); + delete file; + } + + private: + std::string dbname_; + Env* env_; + DB* db_; +}; + +TEST(RecoveryTest, ManifestReused) { + if (!CanAppend()) { + fprintf(stderr, "skipping test because env does not support appending\n"); + return; + } + ASSERT_OK(Put("foo", "bar")); + Close(); + std::string old_manifest = ManifestFileName(); + Open(); + ASSERT_EQ(old_manifest, ManifestFileName()); + ASSERT_EQ("bar", Get("foo")); + Open(); + ASSERT_EQ(old_manifest, ManifestFileName()); + ASSERT_EQ("bar", Get("foo")); +} + +TEST(RecoveryTest, LargeManifestCompacted) { + if (!CanAppend()) { + fprintf(stderr, "skipping test because env does not support appending\n"); + return; + } + ASSERT_OK(Put("foo", "bar")); + Close(); + std::string old_manifest = ManifestFileName(); + + // Pad with zeroes to make manifest file very big. + { + uint64_t len = FileSize(old_manifest); + WritableFile* file; + ASSERT_OK(env()->NewAppendableFile(old_manifest, &file)); + std::string zeroes(3 * 1048576 - static_cast(len), 0); + ASSERT_OK(file->Append(zeroes)); + ASSERT_OK(file->Flush()); + delete file; + } + + Open(); + std::string new_manifest = ManifestFileName(); + ASSERT_NE(old_manifest, new_manifest); + ASSERT_GT(10000, FileSize(new_manifest)); + ASSERT_EQ("bar", Get("foo")); + + Open(); + ASSERT_EQ(new_manifest, ManifestFileName()); + ASSERT_EQ("bar", Get("foo")); +} + +TEST(RecoveryTest, NoLogFiles) { + ASSERT_OK(Put("foo", "bar")); + ASSERT_EQ(1, DeleteLogFiles()); + Open(); + ASSERT_EQ("NOT_FOUND", Get("foo")); + Open(); + ASSERT_EQ("NOT_FOUND", Get("foo")); +} + +TEST(RecoveryTest, LogFileReuse) { + if (!CanAppend()) { + fprintf(stderr, "skipping test because env does not support appending\n"); + return; + } + for (int i = 0; i < 2; i++) { + ASSERT_OK(Put("foo", "bar")); + if (i == 0) { + // Compact to ensure current log is empty + CompactMemTable(); + } + Close(); + ASSERT_EQ(1, NumLogs()); + uint64_t number = FirstLogFile(); + if (i == 0) { + ASSERT_EQ(0, FileSize(LogName(number))); + } else { + ASSERT_LT(0, FileSize(LogName(number))); + } + Open(); + ASSERT_EQ(1, NumLogs()); + ASSERT_EQ(number, FirstLogFile()) << "did not reuse log file"; + ASSERT_EQ("bar", Get("foo")); + Open(); + ASSERT_EQ(1, NumLogs()); + ASSERT_EQ(number, FirstLogFile()) << "did not reuse log file"; + ASSERT_EQ("bar", Get("foo")); + } +} + +TEST(RecoveryTest, MultipleMemTables) { + // Make a large log. + const int kNum = 1000; + for (int i = 0; i < kNum; i++) { + char buf[100]; + snprintf(buf, sizeof(buf), "%050d", i); + ASSERT_OK(Put(buf, buf)); + } + ASSERT_EQ(0, NumTables()); + Close(); + ASSERT_EQ(0, NumTables()); + ASSERT_EQ(1, NumLogs()); + uint64_t old_log_file = FirstLogFile(); + + // Force creation of multiple memtables by reducing the write buffer size. + Options opt; + opt.reuse_logs = true; + opt.write_buffer_size = (kNum * 100) / 2; + Open(&opt); + ASSERT_LE(2, NumTables()); + ASSERT_EQ(1, NumLogs()); + ASSERT_NE(old_log_file, FirstLogFile()) << "must not reuse log"; + for (int i = 0; i < kNum; i++) { + char buf[100]; + snprintf(buf, sizeof(buf), "%050d", i); + ASSERT_EQ(buf, Get(buf)); + } +} + +TEST(RecoveryTest, MultipleLogFiles) { + ASSERT_OK(Put("foo", "bar")); + Close(); + ASSERT_EQ(1, NumLogs()); + + // Make a bunch of uncompacted log files. + uint64_t old_log = FirstLogFile(); + MakeLogFile(old_log + 1, 1000, "hello", "world"); + MakeLogFile(old_log + 2, 1001, "hi", "there"); + MakeLogFile(old_log + 3, 1002, "foo", "bar2"); + + // Recover and check that all log files were processed. + Open(); + ASSERT_LE(1, NumTables()); + ASSERT_EQ(1, NumLogs()); + uint64_t new_log = FirstLogFile(); + ASSERT_LE(old_log + 3, new_log); + ASSERT_EQ("bar2", Get("foo")); + ASSERT_EQ("world", Get("hello")); + ASSERT_EQ("there", Get("hi")); + + // Test that previous recovery produced recoverable state. + Open(); + ASSERT_LE(1, NumTables()); + ASSERT_EQ(1, NumLogs()); + if (CanAppend()) { + ASSERT_EQ(new_log, FirstLogFile()); + } + ASSERT_EQ("bar2", Get("foo")); + ASSERT_EQ("world", Get("hello")); + ASSERT_EQ("there", Get("hi")); + + // Check that introducing an older log file does not cause it to be re-read. + Close(); + MakeLogFile(old_log + 1, 2000, "hello", "stale write"); + Open(); + ASSERT_LE(1, NumTables()); + ASSERT_EQ(1, NumLogs()); + if (CanAppend()) { + ASSERT_EQ(new_log, FirstLogFile()); + } + ASSERT_EQ("bar2", Get("foo")); + ASSERT_EQ("world", Get("hello")); + ASSERT_EQ("there", Get("hi")); +} + +TEST(RecoveryTest, ManifestMissing) { + ASSERT_OK(Put("foo", "bar")); + Close(); + DeleteManifestFile(); + + Status status = OpenWithStatus(); + ASSERT_TRUE(status.IsCorruption()); +} + +} // namespace leveldb + +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/repair.cc b/db/repair.cc index 4cd4bb0..d9d12ba 100644 --- a/db/repair.cc +++ b/db/repair.cc @@ -54,7 +54,7 @@ class Repairer { owns_cache_(options_.block_cache != options.block_cache), next_file_number_(1) { // TableCache can be small since we expect each table to be opened once. - table_cache_ = new TableCache(dbname_, &options_, 10); + table_cache_ = new TableCache(dbname_, options_, 10); } ~Repairer() { @@ -84,9 +84,7 @@ class Repairer { "recovered %d files; %llu bytes. " "Some data may have been lost. " "****", - dbname_.c_str(), - static_cast(tables_.size()), - bytes); + dbname_.c_str(), static_cast(tables_.size()), bytes); } return status; } @@ -97,22 +95,6 @@ class Repairer { SequenceNumber max_sequence; }; - std::string const dbname_; - Env* const env_; - InternalKeyComparator const icmp_; - InternalFilterPolicy const ipolicy_; - Options const options_; - bool owns_info_log_; - bool owns_cache_; - TableCache* table_cache_; - VersionEdit edit_; - - std::vector manifests_; - std::vector table_numbers_; - std::vector logs_; - std::vector tables_; - uint64_t next_file_number_; - Status FindFiles() { std::vector filenames; Status status = env_->GetChildren(dbname_, &filenames); @@ -152,8 +134,7 @@ class Repairer { Status status = ConvertLogToTable(logs_[i]); if (!status.ok()) { Log(options_.info_log, "Log #%llu: ignoring conversion error: %s", - (unsigned long long) logs_[i], - status.ToString().c_str()); + (unsigned long long)logs_[i], status.ToString().c_str()); } ArchiveFile(logname); } @@ -164,11 +145,10 @@ class Repairer { Env* env; Logger* info_log; uint64_t lognum; - virtual void Corruption(size_t bytes, const Status& s) { + void Corruption(size_t bytes, const Status& s) override { // We print error messages for corruption, but continue repairing. Log(info_log, "Log #%llu: dropping %d bytes; %s", - (unsigned long long) lognum, - static_cast(bytes), + (unsigned long long)lognum, static_cast(bytes), s.ToString().c_str()); } }; @@ -190,8 +170,8 @@ class Repairer { // corruptions cause entire commits to be skipped instead of // propagating bad information (like overly large sequence // numbers). - log::Reader reader(lfile, &reporter, false/*do not checksum*/, - 0/*initial_offset*/); + log::Reader reader(lfile, &reporter, false /*do not checksum*/, + 0 /*initial_offset*/); // Read all the records and add to a memtable std::string scratch; @@ -202,8 +182,8 @@ class Repairer { int counter = 0; while (reader.ReadRecord(&record, &scratch)) { if (record.size() < 12) { - reporter.Corruption( - record.size(), Status::Corruption("log record too small")); + reporter.Corruption(record.size(), + Status::Corruption("log record too small")); continue; } WriteBatchInternal::SetContents(&batch, record); @@ -212,8 +192,7 @@ class Repairer { counter += WriteBatchInternal::Count(&batch); } else { Log(options_.info_log, "Log #%llu: ignoring %s", - (unsigned long long) log, - status.ToString().c_str()); + (unsigned long long)log, status.ToString().c_str()); status = Status::OK(); // Keep going with rest of file } } @@ -227,16 +206,14 @@ class Repairer { status = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta); delete iter; mem->Unref(); - mem = NULL; + mem = nullptr; if (status.ok()) { if (meta.file_size > 0) { table_numbers_.push_back(meta.number); } } Log(options_.info_log, "Log #%llu: %d ops saved to Table #%llu %s", - (unsigned long long) log, - counter, - (unsigned long long) meta.number, + (unsigned long long)log, counter, (unsigned long long)meta.number, status.ToString().c_str()); return status; } @@ -272,8 +249,7 @@ class Repairer { ArchiveFile(TableFileName(dbname_, number)); ArchiveFile(SSTTableFileName(dbname_, number)); Log(options_.info_log, "Table #%llu: dropped: %s", - (unsigned long long) t.meta.number, - status.ToString().c_str()); + (unsigned long long)t.meta.number, status.ToString().c_str()); return; } @@ -287,8 +263,7 @@ class Repairer { Slice key = iter->key(); if (!ParseInternalKey(key, &parsed)) { Log(options_.info_log, "Table #%llu: unparsable key %s", - (unsigned long long) t.meta.number, - EscapeString(key).c_str()); + (unsigned long long)t.meta.number, EscapeString(key).c_str()); continue; } @@ -307,9 +282,7 @@ class Repairer { } delete iter; Log(options_.info_log, "Table #%llu: %d entries %s", - (unsigned long long) t.meta.number, - counter, - status.ToString().c_str()); + (unsigned long long)t.meta.number, counter, status.ToString().c_str()); if (status.ok()) { tables_.push_back(t); @@ -350,20 +323,20 @@ class Repairer { } } delete builder; - builder = NULL; + builder = nullptr; if (s.ok()) { s = file->Close(); } delete file; - file = NULL; + file = nullptr; if (counter > 0 && s.ok()) { std::string orig = TableFileName(dbname_, t.meta.number); s = env_->RenameFile(copy, orig); if (s.ok()) { Log(options_.info_log, "Table #%llu: %d entries repaired", - (unsigned long long) t.meta.number, counter); + (unsigned long long)t.meta.number, counter); tables_.push_back(t); } } @@ -395,11 +368,11 @@ class Repairer { 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.number, t.meta.file_size, - t.meta.smallest, t.meta.largest); + edit_.AddFile(0, t.meta.number, t.meta.file_size, t.meta.smallest, + t.meta.largest); } - //fprintf(stderr, "NewDescriptor:\n%s\n", edit_.DebugString().c_str()); + // fprintf(stderr, "NewDescriptor:\n%s\n", edit_.DebugString().c_str()); { log::Writer log(file); std::string record; @@ -410,7 +383,7 @@ class Repairer { status = file->Close(); } delete file; - file = NULL; + file = nullptr; if (!status.ok()) { env_->DeleteFile(tmp); @@ -438,18 +411,34 @@ class Repairer { // dir/lost/foo const char* slash = strrchr(fname.c_str(), '/'); std::string new_dir; - if (slash != NULL) { + 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 == NULL) ? fname.c_str() : slash + 1); + new_file.append((slash == nullptr) ? fname.c_str() : slash + 1); Status s = env_->RenameFile(fname, new_file); - Log(options_.info_log, "Archiving %s: %s\n", - fname.c_str(), s.ToString().c_str()); + Log(options_.info_log, "Archiving %s: %s\n", fname.c_str(), + s.ToString().c_str()); } + + const std::string dbname_; + Env* const env_; + InternalKeyComparator const icmp_; + InternalFilterPolicy const ipolicy_; + const Options options_; + bool owns_info_log_; + bool owns_cache_; + TableCache* table_cache_; + VersionEdit edit_; + + std::vector manifests_; + std::vector table_numbers_; + std::vector logs_; + std::vector tables_; + uint64_t next_file_number_; }; } // namespace diff --git a/db/skiplist.h b/db/skiplist.h index 8bd7776..a59b45b 100644 --- a/db/skiplist.h +++ b/db/skiplist.h @@ -27,9 +27,10 @@ // // ... prev vs. next pointer ordering ... -#include -#include -#include "port/port.h" +#include +#include +#include + #include "util/arena.h" #include "util/random.h" @@ -37,7 +38,7 @@ namespace leveldb { class Arena; -template +template class SkipList { private: struct Node; @@ -48,6 +49,9 @@ class SkipList { // must remain allocated for the lifetime of the skiplist object. explicit SkipList(Comparator cmp, Arena* arena); + SkipList(const SkipList&) = delete; + SkipList& operator=(const SkipList&) = delete; + // Insert key into the list. // REQUIRES: nothing that compares equal to key is currently in the list. void Insert(const Key& key); @@ -97,24 +101,10 @@ class SkipList { private: enum { kMaxHeight = 12 }; - // Immutable after construction - Comparator const compare_; - Arena* const arena_; // Arena used for allocations of nodes - - Node* const head_; - - // Modified only by Insert(). Read racily by readers, but stale - // values are ok. - port::AtomicPointer max_height_; // Height of the entire list - inline int GetMaxHeight() const { - return static_cast( - reinterpret_cast(max_height_.NoBarrier_Load())); + return max_height_.load(std::memory_order_relaxed); } - // Read/written only by Insert(). - Random rnd_; - Node* NewNode(const Key& key, int height); int RandomHeight(); bool Equal(const Key& a, const Key& b) const { return (compare_(a, b) == 0); } @@ -123,9 +113,9 @@ class SkipList { bool KeyIsAfterNode(const Key& key, Node* n) const; // Return the earliest node that comes at or after key. - // Return NULL if there is no such node. + // Return nullptr if there is no such node. // - // If prev is non-NULL, fills prev[level] with pointer to previous + // If prev is non-null, fills prev[level] with pointer to previous // node at "level" for every level in [0..max_height_-1]. Node* FindGreaterOrEqual(const Key& key, Node** prev) const; @@ -137,15 +127,24 @@ class SkipList { // Return head_ if list is empty. Node* FindLast() const; - // No copying allowed - SkipList(const SkipList&); - void operator=(const SkipList&); + // Immutable after construction + Comparator const compare_; + Arena* const arena_; // Arena 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 + + // Read/written only by Insert(). + Random rnd_; }; // Implementation details follow -template -struct SkipList::Node { - explicit Node(const Key& k) : key(k) { } +template +struct SkipList::Node { + explicit Node(const Key& k) : key(k) {} Key const key; @@ -155,92 +154,92 @@ struct SkipList::Node { assert(n >= 0); // Use an 'acquire load' so that we observe a fully initialized // version of the returned Node. - return reinterpret_cast(next_[n].Acquire_Load()); + 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].Release_Store(x); + 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 reinterpret_cast(next_[n].NoBarrier_Load()); + return next_[n].load(std::memory_order_relaxed); } void NoBarrier_SetNext(int n, Node* x) { assert(n >= 0); - next_[n].NoBarrier_Store(x); + next_[n].store(x, std::memory_order_relaxed); } private: // Array of length equal to the node height. next_[0] is lowest level link. - port::AtomicPointer next_[1]; + std::atomic next_[1]; }; -template -typename SkipList::Node* -SkipList::NewNode(const Key& key, int height) { - char* mem = arena_->AllocateAligned( - sizeof(Node) + sizeof(port::AtomicPointer) * (height - 1)); - return new (mem) Node(key); +template +typename SkipList::Node* SkipList::NewNode( + const Key& key, int height) { + char* const node_memory = arena_->AllocateAligned( + sizeof(Node) + sizeof(std::atomic) * (height - 1)); + return new (node_memory) Node(key); } -template -inline SkipList::Iterator::Iterator(const SkipList* list) { +template +inline SkipList::Iterator::Iterator(const SkipList* list) { list_ = list; - node_ = NULL; + node_ = nullptr; } -template -inline bool SkipList::Iterator::Valid() const { - return node_ != NULL; +template +inline bool SkipList::Iterator::Valid() const { + return node_ != nullptr; } -template -inline const Key& SkipList::Iterator::key() const { +template +inline const Key& SkipList::Iterator::key() const { assert(Valid()); return node_->key; } -template -inline void SkipList::Iterator::Next() { +template +inline void SkipList::Iterator::Next() { assert(Valid()); node_ = node_->Next(0); } -template -inline void SkipList::Iterator::Prev() { +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_ = NULL; + node_ = nullptr; } } -template -inline void SkipList::Iterator::Seek(const Key& target) { - node_ = list_->FindGreaterOrEqual(target, NULL); +template +inline void SkipList::Iterator::Seek(const Key& target) { + node_ = list_->FindGreaterOrEqual(target, nullptr); } -template -inline void SkipList::Iterator::SeekToFirst() { +template +inline void SkipList::Iterator::SeekToFirst() { node_ = list_->head_->Next(0); } -template -inline void SkipList::Iterator::SeekToLast() { +template +inline void SkipList::Iterator::SeekToLast() { node_ = list_->FindLast(); if (node_ == list_->head_) { - node_ = NULL; + node_ = nullptr; } } -template -int SkipList::RandomHeight() { +template +int SkipList::RandomHeight() { // Increase height with probability 1 in kBranching static const unsigned int kBranching = 4; int height = 1; @@ -252,15 +251,16 @@ int SkipList::RandomHeight() { return height; } -template -bool SkipList::KeyIsAfterNode(const Key& key, Node* n) const { - // NULL n is considered infinite - return (n != NULL) && (compare_(n->key, key) < 0); +template +bool SkipList::KeyIsAfterNode(const Key& key, Node* n) const { + // null n is considered infinite + return (n != nullptr) && (compare_(n->key, key) < 0); } -template -typename SkipList::Node* SkipList::FindGreaterOrEqual(const Key& key, Node** prev) - const { +template +typename SkipList::Node* +SkipList::FindGreaterOrEqual(const Key& key, + Node** prev) const { Node* x = head_; int level = GetMaxHeight() - 1; while (true) { @@ -269,7 +269,7 @@ typename SkipList::Node* SkipList::FindGreaterOr // Keep searching in this list x = next; } else { - if (prev != NULL) prev[level] = x; + if (prev != nullptr) prev[level] = x; if (level == 0) { return next; } else { @@ -280,15 +280,15 @@ typename SkipList::Node* SkipList::FindGreaterOr } } -template -typename SkipList::Node* -SkipList::FindLessThan(const Key& key) const { +template +typename SkipList::Node* +SkipList::FindLessThan(const Key& key) const { Node* x = head_; int level = GetMaxHeight() - 1; while (true) { assert(x == head_ || compare_(x->key, key) < 0); Node* next = x->Next(level); - if (next == NULL || compare_(next->key, key) >= 0) { + if (next == nullptr || compare_(next->key, key) >= 0) { if (level == 0) { return x; } else { @@ -301,14 +301,14 @@ SkipList::FindLessThan(const Key& key) const { } } -template -typename SkipList::Node* SkipList::FindLast() +template +typename SkipList::Node* SkipList::FindLast() const { Node* x = head_; int level = GetMaxHeight() - 1; while (true) { Node* next = x->Next(level); - if (next == NULL) { + if (next == nullptr) { if (level == 0) { return x; } else { @@ -321,43 +321,41 @@ typename SkipList::Node* SkipList::FindLast() } } -template -SkipList::SkipList(Comparator cmp, Arena* arena) +template +SkipList::SkipList(Comparator cmp, Arena* arena) : compare_(cmp), arena_(arena), head_(NewNode(0 /* any key will do */, kMaxHeight)), - max_height_(reinterpret_cast(1)), + max_height_(1), rnd_(0xdeadbeef) { for (int i = 0; i < kMaxHeight; i++) { - head_->SetNext(i, NULL); + head_->SetNext(i, nullptr); } } -template -void SkipList::Insert(const Key& key) { +template +void SkipList::Insert(const Key& key) { // TODO(opt): We can use a barrier-free variant of FindGreaterOrEqual() // here since Insert() is externally synchronized. Node* prev[kMaxHeight]; Node* x = FindGreaterOrEqual(key, prev); // Our data structure does not allow duplicate insertion - assert(x == NULL || !Equal(key, x->key)); + assert(x == nullptr || !Equal(key, x->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_ (NULL), or a new value set in + // 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 NULL sorts after all + // 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_.NoBarrier_Store(reinterpret_cast(height)); + max_height_.store(height, std::memory_order_relaxed); } x = NewNode(key, height); @@ -369,10 +367,10 @@ void SkipList::Insert(const Key& key) { } } -template -bool SkipList::Contains(const Key& key) const { - Node* x = FindGreaterOrEqual(key, NULL); - if (x != NULL && Equal(key, x->key)) { +template +bool SkipList::Contains(const Key& key) const { + Node* x = FindGreaterOrEqual(key, nullptr); + if (x != nullptr && Equal(key, x->key)) { return true; } else { return false; diff --git a/db/skiplist_test.cc b/db/skiplist_test.cc index c78f4b4..9fa2d96 100644 --- a/db/skiplist_test.cc +++ b/db/skiplist_test.cc @@ -3,8 +3,13 @@ // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "db/skiplist.h" + +#include #include + #include "leveldb/env.h" +#include "port/port.h" +#include "port/thread_annotations.h" #include "util/arena.h" #include "util/hash.h" #include "util/random.h" @@ -26,7 +31,7 @@ struct Comparator { } }; -class SkipTest { }; +class SkipTest {}; TEST(SkipTest, Empty) { Arena arena; @@ -112,8 +117,7 @@ TEST(SkipTest, InsertAndLookup) { // Compare against model iterator for (std::set::reverse_iterator model_iter = keys.rbegin(); - model_iter != keys.rend(); - ++model_iter) { + model_iter != keys.rend(); ++model_iter) { ASSERT_TRUE(iter.Valid()); ASSERT_EQ(*model_iter, iter.key()); iter.Prev(); @@ -126,7 +130,7 @@ TEST(SkipTest, InsertAndLookup) { // 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 +// constructed. 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. @@ -155,12 +159,12 @@ class ConcurrentTest { 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 }; + 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)); + static_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)); @@ -186,13 +190,11 @@ class ConcurrentTest { // Per-key generation struct State { - port::AtomicPointer generation[K]; - void Set(int k, intptr_t v) { - generation[k].Release_Store(reinterpret_cast(v)); - } - intptr_t Get(int k) { - return reinterpret_cast(generation[k].Acquire_Load()); + 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 (int k = 0; k < K; k++) { @@ -211,7 +213,7 @@ class ConcurrentTest { SkipList list_; public: - ConcurrentTest() : list_(Comparator(), &arena_) { } + ConcurrentTest() : list_(Comparator(), &arena_) {} // REQUIRES: External synchronization void WriteStep(Random* rnd) { @@ -250,11 +252,9 @@ class ConcurrentTest { // Note that generation 0 is never inserted, so it is ok if // <*,0,*> is missing. ASSERT_TRUE((gen(pos) == 0) || - (gen(pos) > initial_state.Get(key(pos))) - ) << "key: " << key(pos) - << "; gen: " << gen(pos) - << "; initgen: " - << initial_state.Get(key(pos)); + (gen(pos) > static_cast(initial_state.Get(key(pos))))) + << "key: " << key(pos) << "; gen: " << gen(pos) + << "; initgen: " << initial_state.Get(key(pos)); // Advance to next key in the valid key space if (key(pos) < key(current)) { @@ -298,21 +298,14 @@ class TestState { public: ConcurrentTest t_; int seed_; - port::AtomicPointer quit_flag_; + std::atomic quit_flag_; - enum ReaderState { - STARTING, - RUNNING, - DONE - }; + enum ReaderState { STARTING, RUNNING, DONE }; explicit TestState(int s) - : seed_(s), - quit_flag_(NULL), - state_(STARTING), - state_cv_(&mu_) {} + : seed_(s), quit_flag_(false), state_(STARTING), state_cv_(&mu_) {} - void Wait(ReaderState s) { + void Wait(ReaderState s) LOCKS_EXCLUDED(mu_) { mu_.Lock(); while (state_ != s) { state_cv_.Wait(); @@ -320,7 +313,7 @@ class TestState { mu_.Unlock(); } - void Change(ReaderState s) { + void Change(ReaderState s) LOCKS_EXCLUDED(mu_) { mu_.Lock(); state_ = s; state_cv_.Signal(); @@ -329,8 +322,8 @@ class TestState { private: port::Mutex mu_; - ReaderState state_; - port::CondVar state_cv_; + ReaderState state_ GUARDED_BY(mu_); + port::CondVar state_cv_ GUARDED_BY(mu_); }; static void ConcurrentReader(void* arg) { @@ -338,7 +331,7 @@ static void ConcurrentReader(void* arg) { Random rnd(state->seed_); int64_t reads = 0; state->Change(TestState::RUNNING); - while (!state->quit_flag_.Acquire_Load()) { + while (!state->quit_flag_.load(std::memory_order_acquire)) { state->t_.ReadStep(&rnd); ++reads; } @@ -360,7 +353,7 @@ static void RunConcurrent(int run) { for (int i = 0; i < kSize; i++) { state.t_.WriteStep(&rnd); } - state.quit_flag_.Release_Store(&state); // Any non-NULL arg will do + state.quit_flag_.store(true, std::memory_order_release); state.Wait(TestState::DONE); } } @@ -373,6 +366,4 @@ TEST(SkipTest, Concurrent5) { RunConcurrent(5); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/snapshot.h b/db/snapshot.h index 6ed413c..9f1d664 100644 --- a/db/snapshot.h +++ b/db/snapshot.h @@ -16,50 +16,78 @@ class SnapshotList; // Each SnapshotImpl corresponds to a particular sequence number. class SnapshotImpl : public Snapshot { public: - SequenceNumber number_; // const after creation + SnapshotImpl(SequenceNumber sequence_number) + : sequence_number_(sequence_number) {} + + SequenceNumber sequence_number() const { return sequence_number_; } private: friend class SnapshotList; - // SnapshotImpl is kept in a doubly-linked circular list + // SnapshotImpl is kept in a doubly-linked circular list. The SnapshotList + // implementation operates on the next/previous fields direcly. SnapshotImpl* prev_; SnapshotImpl* next_; - SnapshotList* list_; // just for sanity checks + const SequenceNumber sequence_number_; + +#if !defined(NDEBUG) + SnapshotList* list_ = nullptr; +#endif // !defined(NDEBUG) }; class SnapshotList { public: - SnapshotList() { - list_.prev_ = &list_; - list_.next_ = &list_; + SnapshotList() : head_(0) { + head_.prev_ = &head_; + head_.next_ = &head_; } - 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(SequenceNumber seq) { - SnapshotImpl* s = new SnapshotImpl; - s->number_ = seq; - s->list_ = this; - s->next_ = &list_; - s->prev_ = list_.prev_; - s->prev_->next_ = s; - s->next_->prev_ = s; - return s; + bool empty() const { return head_.next_ == &head_; } + SnapshotImpl* oldest() const { + assert(!empty()); + return head_.next_; + } + SnapshotImpl* newest() const { + assert(!empty()); + return head_.prev_; } - void Delete(const SnapshotImpl* s) { - assert(s->list_ == this); - s->prev_->next_ = s->next_; - s->next_->prev_ = s->prev_; - delete s; + // Creates a SnapshotImpl and appends it to the end of the list. + SnapshotImpl* New(SequenceNumber sequence_number) { + assert(empty() || newest()->sequence_number_ <= sequence_number); + + SnapshotImpl* snapshot = new SnapshotImpl(sequence_number); + +#if !defined(NDEBUG) + snapshot->list_ = this; +#endif // !defined(NDEBUG) + snapshot->next_ = &head_; + snapshot->prev_ = head_.prev_; + snapshot->prev_->next_ = snapshot; + snapshot->next_->prev_ = snapshot; + return snapshot; + } + + // Removes a SnapshotImpl from this list. + // + // The snapshot must have been created by calling New() on this list. + // + // The snapshot pointer should not be const, because its memory is + // deallocated. However, that would force us to change DB::ReleaseSnapshot(), + // which is in the API, and currently takes a const Snapshot. + void Delete(const SnapshotImpl* snapshot) { +#if !defined(NDEBUG) + assert(snapshot->list_ == this); +#endif // !defined(NDEBUG) + snapshot->prev_->next_ = snapshot->next_; + snapshot->next_->prev_ = snapshot->prev_; + delete snapshot; } private: // Dummy head of doubly-linked list of snapshots - SnapshotImpl list_; + SnapshotImpl head_; }; } // namespace leveldb diff --git a/db/table_cache.cc b/db/table_cache.cc index e3d82cd..73f05fd 100644 --- a/db/table_cache.cc +++ b/db/table_cache.cc @@ -29,18 +29,14 @@ static void UnrefEntry(void* arg1, void* arg2) { cache->Release(h); } -TableCache::TableCache(const std::string& dbname, - const Options* options, +TableCache::TableCache(const std::string& dbname, const Options& options, int entries) - : env_(options->env), + : env_(options.env), dbname_(dbname), options_(options), - cache_(NewLRUCache(entries)) { -} + cache_(NewLRUCache(entries)) {} -TableCache::~TableCache() { - delete cache_; -} +TableCache::~TableCache() { delete cache_; } Status TableCache::FindTable(uint64_t file_number, uint64_t file_size, Cache::Handle** handle) { @@ -49,10 +45,10 @@ Status TableCache::FindTable(uint64_t file_number, uint64_t file_size, EncodeFixed64(buf, file_number); Slice key(buf, sizeof(buf)); *handle = cache_->Lookup(key); - if (*handle == NULL) { + if (*handle == nullptr) { std::string fname = TableFileName(dbname_, file_number); - RandomAccessFile* file = NULL; - Table* table = NULL; + RandomAccessFile* file = nullptr; + Table* table = nullptr; s = env_->NewRandomAccessFile(fname, &file); if (!s.ok()) { std::string old_fname = SSTTableFileName(dbname_, file_number); @@ -61,11 +57,11 @@ Status TableCache::FindTable(uint64_t file_number, uint64_t file_size, } } if (s.ok()) { - s = Table::Open(*options_, file, file_size, &table); + s = Table::Open(options_, file, file_size, &table); } if (!s.ok()) { - assert(table == NULL); + assert(table == nullptr); delete file; // We do not cache error results so that if the error is transient, // or somebody repairs the file, we recover automatically. @@ -80,14 +76,13 @@ Status TableCache::FindTable(uint64_t file_number, uint64_t file_size, } Iterator* TableCache::NewIterator(const ReadOptions& options, - uint64_t file_number, - uint64_t file_size, + uint64_t file_number, uint64_t file_size, Table** tableptr) { - if (tableptr != NULL) { - *tableptr = NULL; + if (tableptr != nullptr) { + *tableptr = nullptr; } - Cache::Handle* handle = NULL; + Cache::Handle* handle = nullptr; Status s = FindTable(file_number, file_size, &handle); if (!s.ok()) { return NewErrorIterator(s); @@ -96,23 +91,21 @@ Iterator* TableCache::NewIterator(const ReadOptions& options, Table* table = reinterpret_cast(cache_->Value(handle))->table; Iterator* result = table->NewIterator(options); result->RegisterCleanup(&UnrefEntry, cache_, handle); - if (tableptr != NULL) { + if (tableptr != nullptr) { *tableptr = table; } return result; } -Status TableCache::Get(const ReadOptions& options, - uint64_t file_number, - uint64_t file_size, - const Slice& k, - void* arg, - void (*saver)(void*, const Slice&, const Slice&)) { - Cache::Handle* handle = NULL; +Status TableCache::Get(const ReadOptions& options, uint64_t file_number, + uint64_t file_size, const Slice& k, void* arg, + void (*handle_result)(void*, const Slice&, + const Slice&)) { + Cache::Handle* handle = nullptr; Status s = FindTable(file_number, file_size, &handle); if (s.ok()) { Table* t = reinterpret_cast(cache_->Value(handle))->table; - s = t->InternalGet(options, k, arg, saver); + s = t->InternalGet(options, k, arg, handle_result); cache_->Release(handle); } return s; diff --git a/db/table_cache.h b/db/table_cache.h index 8cf4aaf..93069c8 100644 --- a/db/table_cache.h +++ b/db/table_cache.h @@ -7,8 +7,10 @@ #ifndef STORAGE_LEVELDB_DB_TABLE_CACHE_H_ #define STORAGE_LEVELDB_DB_TABLE_CACHE_H_ -#include #include + +#include + #include "db/dbformat.h" #include "leveldb/cache.h" #include "leveldb/table.h" @@ -20,40 +22,35 @@ class Env; class TableCache { public: - TableCache(const std::string& dbname, const Options* options, int entries); + TableCache(const std::string& dbname, const Options& options, int entries); ~TableCache(); // Return an iterator for the specified file number (the corresponding // file length must be exactly "file_size" bytes). If "tableptr" is - // non-NULL, also sets "*tableptr" to point to the Table object - // underlying the returned iterator, or NULL 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 + // non-null, also sets "*tableptr" to point to the Table object + // underlying the returned iterator, or to 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. - Iterator* NewIterator(const ReadOptions& options, - uint64_t file_number, - uint64_t file_size, - Table** tableptr = NULL); + Iterator* NewIterator(const ReadOptions& options, uint64_t file_number, + uint64_t file_size, Table** tableptr = nullptr); // If a seek to internal key "k" in specified file finds an entry, // call (*handle_result)(arg, found_key, found_value). - Status Get(const ReadOptions& options, - uint64_t file_number, - uint64_t file_size, - const Slice& k, - void* arg, + Status Get(const ReadOptions& options, uint64_t file_number, + uint64_t file_size, const Slice& k, void* arg, void (*handle_result)(void*, const Slice&, const Slice&)); // Evict any entry for the specified file number void Evict(uint64_t file_number); private: + Status FindTable(uint64_t file_number, uint64_t file_size, Cache::Handle**); + Env* const env_; const std::string dbname_; - const Options* options_; + const Options& options_; Cache* cache_; - - Status FindTable(uint64_t file_number, uint64_t file_size, Cache::Handle**); }; } // namespace leveldb diff --git a/db/version_edit.cc b/db/version_edit.cc index f10a2d5..cd770ef 100644 --- a/db/version_edit.cc +++ b/db/version_edit.cc @@ -12,15 +12,15 @@ namespace leveldb { // 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, + kComparator = 1, + kLogNumber = 2, + kNextFileNumber = 3, + kLastSequence = 4, + kCompactPointer = 5, + kDeletedFile = 6, + kNewFile = 7, // 8 was used for large value refs - kPrevLogNumber = 9 + kPrevLogNumber = 9 }; void VersionEdit::Clear() { @@ -66,12 +66,10 @@ void VersionEdit::EncodeTo(std::string* dst) const { PutLengthPrefixedSlice(dst, compact_pointers_[i].second.Encode()); } - for (DeletedFileSet::const_iterator iter = deleted_files_.begin(); - iter != deleted_files_.end(); - ++iter) { + for (const auto& deleted_file_kvp : deleted_files_) { PutVarint32(dst, kDeletedFile); - PutVarint32(dst, iter->first); // level - PutVarint64(dst, iter->second); // file number + PutVarint32(dst, deleted_file_kvp.first); // level + PutVarint64(dst, deleted_file_kvp.second); // file number } for (size_t i = 0; i < new_files_.size(); i++) { @@ -88,8 +86,7 @@ void VersionEdit::EncodeTo(std::string* dst) const { static bool GetInternalKey(Slice* input, InternalKey* dst) { Slice str; if (GetLengthPrefixedSlice(input, &str)) { - dst->DecodeFrom(str); - return true; + return dst->DecodeFrom(str); } else { return false; } @@ -97,8 +94,7 @@ static bool GetInternalKey(Slice* input, InternalKey* dst) { static bool GetLevel(Slice* input, int* level) { uint32_t v; - if (GetVarint32(input, &v) && - v < config::kNumLevels) { + if (GetVarint32(input, &v) && v < config::kNumLevels) { *level = v; return true; } else { @@ -109,7 +105,7 @@ static bool GetLevel(Slice* input, int* level) { Status VersionEdit::DecodeFrom(const Slice& src) { Clear(); Slice input = src; - const char* msg = NULL; + const char* msg = nullptr; uint32_t tag; // Temporary storage for parsing @@ -119,7 +115,7 @@ Status VersionEdit::DecodeFrom(const Slice& src) { Slice str; InternalKey key; - while (msg == NULL && GetVarint32(&input, &tag)) { + while (msg == nullptr && GetVarint32(&input, &tag)) { switch (tag) { case kComparator: if (GetLengthPrefixedSlice(&input, &str)) { @@ -163,8 +159,7 @@ Status VersionEdit::DecodeFrom(const Slice& src) { break; case kCompactPointer: - if (GetLevel(&input, &level) && - GetInternalKey(&input, &key)) { + if (GetLevel(&input, &level) && GetInternalKey(&input, &key)) { compact_pointers_.push_back(std::make_pair(level, key)); } else { msg = "compaction pointer"; @@ -172,8 +167,7 @@ Status VersionEdit::DecodeFrom(const Slice& src) { break; case kDeletedFile: - if (GetLevel(&input, &level) && - GetVarint64(&input, &number)) { + if (GetLevel(&input, &level) && GetVarint64(&input, &number)) { deleted_files_.insert(std::make_pair(level, number)); } else { msg = "deleted file"; @@ -181,8 +175,7 @@ Status VersionEdit::DecodeFrom(const Slice& src) { break; case kNewFile: - if (GetLevel(&input, &level) && - GetVarint64(&input, &f.number) && + if (GetLevel(&input, &level) && GetVarint64(&input, &f.number) && GetVarint64(&input, &f.file_size) && GetInternalKey(&input, &f.smallest) && GetInternalKey(&input, &f.largest)) { @@ -198,12 +191,12 @@ Status VersionEdit::DecodeFrom(const Slice& src) { } } - if (msg == NULL && !input.empty()) { + if (msg == nullptr && !input.empty()) { msg = "invalid tag"; } Status result; - if (msg != NULL) { + if (msg != nullptr) { result = Status::Corruption("VersionEdit", msg); } return result; @@ -238,13 +231,11 @@ std::string VersionEdit::DebugString() const { r.append(" "); r.append(compact_pointers_[i].second.DebugString()); } - for (DeletedFileSet::const_iterator iter = deleted_files_.begin(); - iter != deleted_files_.end(); - ++iter) { + for (const auto& deleted_files_kvp : deleted_files_) { r.append("\n DeleteFile: "); - AppendNumberTo(&r, iter->first); + AppendNumberTo(&r, deleted_files_kvp.first); r.append(" "); - AppendNumberTo(&r, iter->second); + AppendNumberTo(&r, deleted_files_kvp.second); } for (size_t i = 0; i < new_files_.size(); i++) { const FileMetaData& f = new_files_[i].second; diff --git a/db/version_edit.h b/db/version_edit.h index eaef77b..86b2f22 100644 --- a/db/version_edit.h +++ b/db/version_edit.h @@ -8,6 +8,7 @@ #include #include #include + #include "db/dbformat.h" namespace leveldb { @@ -15,20 +16,20 @@ namespace leveldb { class VersionSet; struct FileMetaData { - int refs; - int allowed_seeks; // Seeks allowed until compaction - uint64_t number; - uint64_t file_size; // File size in bytes - InternalKey smallest; // Smallest internal key served by table - InternalKey largest; // Largest internal key served by table + FileMetaData() : refs(0), allowed_seeks(1 << 30), file_size(0) {} - FileMetaData() : refs(0), allowed_seeks(1 << 30), file_size(0) { } + int refs; + int allowed_seeks; // Seeks allowed until compaction + uint64_t number; + uint64_t file_size; // File size in bytes + InternalKey smallest; // Smallest internal key served by table + InternalKey largest; // Largest internal key served by table }; class VersionEdit { public: VersionEdit() { Clear(); } - ~VersionEdit() { } + ~VersionEdit() = default; void Clear(); @@ -59,10 +60,8 @@ class VersionEdit { // 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, - uint64_t file_size, - const InternalKey& smallest, - const InternalKey& largest) { + void AddFile(int level, uint64_t file, uint64_t file_size, + const InternalKey& smallest, const InternalKey& largest) { FileMetaData f; f.number = file; f.file_size = file_size; @@ -84,7 +83,7 @@ class VersionEdit { private: friend class VersionSet; - typedef std::set< std::pair > DeletedFileSet; + typedef std::set > DeletedFileSet; std::string comparator_; uint64_t log_number_; @@ -97,9 +96,9 @@ class VersionEdit { bool has_next_file_number_; bool has_last_sequence_; - std::vector< std::pair > compact_pointers_; + std::vector > compact_pointers_; DeletedFileSet deleted_files_; - std::vector< std::pair > new_files_; + std::vector > new_files_; }; } // namespace leveldb diff --git a/db/version_edit_test.cc b/db/version_edit_test.cc index 280310b..0b7cda8 100644 --- a/db/version_edit_test.cc +++ b/db/version_edit_test.cc @@ -17,7 +17,7 @@ static void TestEncodeDecode(const VersionEdit& edit) { ASSERT_EQ(encoded, encoded2); } -class VersionEditTest { }; +class VersionEditTest {}; TEST(VersionEditTest, EncodeDecode) { static const uint64_t kBig = 1ull << 50; @@ -41,6 +41,4 @@ TEST(VersionEditTest, EncodeDecode) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/version_set.cc b/db/version_set.cc index aa83df5..b62a2d0 100644 --- a/db/version_set.cc +++ b/db/version_set.cc @@ -4,8 +4,10 @@ #include "db/version_set.h" -#include #include + +#include + #include "db/filename.h" #include "db/log_reader.h" #include "db/log_writer.h" @@ -20,21 +22,29 @@ namespace leveldb { -static const int kTargetFileSize = 2 * 1048576; +static size_t TargetFileSize(const Options* options) { + return options->max_file_size; +} // Maximum bytes of overlaps in grandparent (i.e., level+2) before we // stop building a single file in a level->level+1 compaction. -static const int64_t kMaxGrandParentOverlapBytes = 10 * kTargetFileSize; +static int64_t MaxGrandParentOverlapBytes(const Options* options) { + return 10 * TargetFileSize(options); +} // 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 this many bytes. -static const int64_t kExpandedCompactionByteSizeLimit = 25 * kTargetFileSize; +static int64_t ExpandedCompactionByteSizeLimit(const Options* options) { + return 25 * TargetFileSize(options); +} -static double MaxBytesForLevel(int level) { +static double MaxBytesForLevel(const Options* options, int level) { // Note: the result for level zero is not really used since we set // the level-0 compaction threshold based on number of files. - double result = 10 * 1048576.0; // Result for both level-0 and level-1 + + // Result for both level-0 and level-1 + double result = 10. * 1048576.0; while (level > 1) { result *= 10; level--; @@ -42,8 +52,9 @@ static double MaxBytesForLevel(int level) { return result; } -static uint64_t MaxFileSizeForLevel(int level) { - return kTargetFileSize; // We could vary per level to reduce number of files? +static uint64_t MaxFileSizeForLevel(const Options* options, int level) { + // We could vary per level to reduce number of files? + return TargetFileSize(options); } static int64_t TotalFileSize(const std::vector& files) { @@ -75,8 +86,7 @@ Version::~Version() { } int FindFile(const InternalKeyComparator& icmp, - const std::vector& files, - const Slice& key) { + const std::vector& files, const Slice& key) { uint32_t left = 0; uint32_t right = files.size(); while (left < right) { @@ -95,26 +105,25 @@ int FindFile(const InternalKeyComparator& icmp, return right; } -static bool AfterFile(const Comparator* ucmp, - const Slice* user_key, const FileMetaData* f) { - // NULL user_key occurs before all keys and is therefore never after *f - return (user_key != NULL && +static bool AfterFile(const Comparator* ucmp, const Slice* user_key, + const FileMetaData* f) { + // null user_key occurs before all keys and is therefore never after *f + return (user_key != nullptr && ucmp->Compare(*user_key, f->largest.user_key()) > 0); } -static bool BeforeFile(const Comparator* ucmp, - const Slice* user_key, const FileMetaData* f) { - // NULL user_key occurs after all keys and is therefore never before *f - return (user_key != NULL && +static bool BeforeFile(const Comparator* ucmp, const Slice* user_key, + const FileMetaData* f) { + // null user_key occurs after all keys and is therefore never before *f + return (user_key != nullptr && ucmp->Compare(*user_key, f->smallest.user_key()) < 0); } -bool SomeFileOverlapsRange( - const InternalKeyComparator& icmp, - bool disjoint_sorted_files, - const std::vector& files, - const Slice* smallest_user_key, - const Slice* largest_user_key) { +bool SomeFileOverlapsRange(const InternalKeyComparator& icmp, + bool disjoint_sorted_files, + const std::vector& files, + 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 @@ -132,10 +141,11 @@ bool SomeFileOverlapsRange( // Binary search over file list uint32_t index = 0; - if (smallest_user_key != NULL) { + if (smallest_user_key != nullptr) { // Find the earliest possible internal key for smallest_user_key - InternalKey small(*smallest_user_key, kMaxSequenceNumber,kValueTypeForSeek); - index = FindFile(icmp, files, small.Encode()); + InternalKey small_key(*smallest_user_key, kMaxSequenceNumber, + kValueTypeForSeek); + index = FindFile(icmp, files, small_key.Encode()); } if (index >= files.size()) { @@ -155,25 +165,21 @@ class Version::LevelFileNumIterator : public Iterator { public: LevelFileNumIterator(const InternalKeyComparator& icmp, const std::vector* flist) - : icmp_(icmp), - flist_(flist), - index_(flist->size()) { // Marks as invalid + : icmp_(icmp), flist_(flist), index_(flist->size()) { // Marks as invalid } - virtual bool Valid() const { - return index_ < flist_->size(); - } - virtual void Seek(const Slice& target) { + bool Valid() const override { return index_ < flist_->size(); } + void Seek(const Slice& target) override { index_ = FindFile(icmp_, *flist_, target); } - virtual void SeekToFirst() { index_ = 0; } - virtual void SeekToLast() { + void SeekToFirst() override { index_ = 0; } + void SeekToLast() override { index_ = flist_->empty() ? 0 : flist_->size() - 1; } - virtual void Next() { + void Next() override { assert(Valid()); index_++; } - virtual void Prev() { + void Prev() override { assert(Valid()); if (index_ == 0) { index_ = flist_->size(); // Marks as invalid @@ -181,17 +187,18 @@ class Version::LevelFileNumIterator : public Iterator { index_--; } } - Slice key() const { + Slice key() const override { assert(Valid()); return (*flist_)[index_]->largest.Encode(); } - Slice value() const { + Slice value() const override { assert(Valid()); EncodeFixed64(value_buf_, (*flist_)[index_]->number); - EncodeFixed64(value_buf_+8, (*flist_)[index_]->file_size); + EncodeFixed64(value_buf_ + 8, (*flist_)[index_]->file_size); return Slice(value_buf_, sizeof(value_buf_)); } - virtual Status status() const { return Status::OK(); } + Status status() const override { return Status::OK(); } + private: const InternalKeyComparator icmp_; const std::vector* const flist_; @@ -201,16 +208,14 @@ class Version::LevelFileNumIterator : public Iterator { mutable char value_buf_[16]; }; -static Iterator* GetFileIterator(void* arg, - const ReadOptions& options, +static Iterator* GetFileIterator(void* arg, const ReadOptions& options, const Slice& file_value) { TableCache* cache = reinterpret_cast(arg); if (file_value.size() != 16) { return NewErrorIterator( Status::Corruption("FileReader invoked with unexpected value")); } else { - return cache->NewIterator(options, - DecodeFixed64(file_value.data()), + return cache->NewIterator(options, DecodeFixed64(file_value.data()), DecodeFixed64(file_value.data() + 8)); } } @@ -218,17 +223,16 @@ static Iterator* GetFileIterator(void* arg, Iterator* Version::NewConcatenatingIterator(const ReadOptions& options, int level) const { return NewTwoLevelIterator( - new LevelFileNumIterator(vset_->icmp_, &files_[level]), - &GetFileIterator, vset_->table_cache_, options); + new LevelFileNumIterator(vset_->icmp_, &files_[level]), &GetFileIterator, + vset_->table_cache_, options); } void Version::AddIterators(const ReadOptions& options, std::vector* iters) { // Merge all level zero files together since they may overlap for (size_t i = 0; i < files_[0].size(); i++) { - iters->push_back( - vset_->table_cache_->NewIterator( - options, files_[0][i]->number, files_[0][i]->file_size)); + iters->push_back(vset_->table_cache_->NewIterator( + options, files_[0][i]->number, files_[0][i]->file_size)); } // For levels > 0, we can use a concatenating iterator that sequentially @@ -255,7 +259,7 @@ struct Saver { Slice user_key; std::string* value; }; -} +} // namespace static void SaveValue(void* arg, const Slice& ikey, const Slice& v) { Saver* s = reinterpret_cast(arg); ParsedInternalKey parsed_key; @@ -275,8 +279,7 @@ static bool NewestFirst(FileMetaData* a, FileMetaData* b) { return a->number > b->number; } -void Version::ForEachOverlapping(Slice user_key, Slice internal_key, - void* arg, +void Version::ForEachOverlapping(Slice user_key, Slice internal_key, void* arg, bool (*func)(void*, int, FileMetaData*)) { // TODO(sanjay): Change Version::Get() to use this function. const Comparator* ucmp = vset_->icmp_.user_comparator(); @@ -320,23 +323,21 @@ void Version::ForEachOverlapping(Slice user_key, Slice internal_key, } } -Status Version::Get(const ReadOptions& options, - const LookupKey& k, - std::string* value, - GetStats* stats) { +Status Version::Get(const ReadOptions& options, const LookupKey& k, + std::string* value, GetStats* stats) { Slice ikey = k.internal_key(); Slice user_key = k.user_key(); const Comparator* ucmp = vset_->icmp_.user_comparator(); Status s; - stats->seek_file = NULL; + stats->seek_file = nullptr; stats->seek_file_level = -1; - FileMetaData* last_file_read = NULL; + FileMetaData* last_file_read = nullptr; int last_file_read_level = -1; // We can search level-by-level since entries never hop across // levels. Therefore we are guaranteed that if we find data - // in an smaller level, later levels are irrelevant. + // in a smaller level, later levels are irrelevant. std::vector tmp; FileMetaData* tmp2; for (int level = 0; level < config::kNumLevels; level++) { @@ -365,13 +366,13 @@ Status Version::Get(const ReadOptions& options, // Binary search to find earliest index whose largest key >= ikey. uint32_t index = FindFile(vset_->icmp_, files_[level], ikey); if (index >= num_files) { - files = NULL; + files = nullptr; num_files = 0; } else { tmp2 = files[index]; if (ucmp->Compare(user_key, tmp2->smallest.user_key()) < 0) { // All of "tmp2" is past any data for user_key - files = NULL; + files = nullptr; num_files = 0; } else { files = &tmp2; @@ -381,7 +382,7 @@ Status Version::Get(const ReadOptions& options, } for (uint32_t i = 0; i < num_files; ++i) { - if (last_file_read != NULL && stats->seek_file == NULL) { + if (last_file_read != nullptr && stats->seek_file == nullptr) { // We have had more than one seek for this read. Charge the 1st file. stats->seek_file = last_file_read; stats->seek_file_level = last_file_read_level; @@ -396,14 +397,14 @@ Status Version::Get(const ReadOptions& options, saver.ucmp = ucmp; saver.user_key = user_key; saver.value = value; - s = vset_->table_cache_->Get(options, f->number, f->file_size, - ikey, &saver, SaveValue); + s = vset_->table_cache_->Get(options, f->number, f->file_size, ikey, + &saver, SaveValue); if (!s.ok()) { return s; } switch (saver.state) { case kNotFound: - break; // Keep searching in other files + break; // Keep searching in other files case kFound: return s; case kDeleted: @@ -421,9 +422,9 @@ Status Version::Get(const ReadOptions& options, bool Version::UpdateStats(const GetStats& stats) { FileMetaData* f = stats.seek_file; - if (f != NULL) { + if (f != nullptr) { f->allowed_seeks--; - if (f->allowed_seeks <= 0 && file_to_compact_ == NULL) { + if (f->allowed_seeks <= 0 && file_to_compact_ == nullptr) { file_to_compact_ = f; file_to_compact_level_ = stats.seek_file_level; return true; @@ -470,9 +471,7 @@ bool Version::RecordReadSample(Slice internal_key) { return false; } -void Version::Ref() { - ++refs_; -} +void Version::Ref() { ++refs_; } void Version::Unref() { assert(this != &vset_->dummy_versions_); @@ -483,16 +482,14 @@ void Version::Unref() { } } -bool Version::OverlapInLevel(int level, - const Slice* smallest_user_key, +bool Version::OverlapInLevel(int level, const Slice* smallest_user_key, const Slice* largest_user_key) { return SomeFileOverlapsRange(vset_->icmp_, (level > 0), files_[level], smallest_user_key, largest_user_key); } -int Version::PickLevelForMemTableOutput( - const Slice& smallest_user_key, - const Slice& largest_user_key) { +int Version::PickLevelForMemTableOutput(const Slice& smallest_user_key, + const Slice& largest_user_key) { int level = 0; if (!OverlapInLevel(0, &smallest_user_key, &largest_user_key)) { // Push to next level if there is no overlap in next level, @@ -508,7 +505,7 @@ int Version::PickLevelForMemTableOutput( // Check that file does not overlap too many grandparent bytes. GetOverlappingInputs(level + 2, &start, &limit, &overlaps); const int64_t sum = TotalFileSize(overlaps); - if (sum > kMaxGrandParentOverlapBytes) { + if (sum > MaxGrandParentOverlapBytes(vset_->options_)) { break; } } @@ -519,40 +516,39 @@ int Version::PickLevelForMemTableOutput( } // Store in "*inputs" all files in "level" that overlap [begin,end] -void Version::GetOverlappingInputs( - int level, - const InternalKey* begin, - const InternalKey* end, - std::vector* inputs) { +void Version::GetOverlappingInputs(int level, const InternalKey* begin, + const InternalKey* end, + std::vector* inputs) { assert(level >= 0); assert(level < config::kNumLevels); inputs->clear(); Slice user_begin, user_end; - if (begin != NULL) { + if (begin != nullptr) { user_begin = begin->user_key(); } - if (end != NULL) { + if (end != nullptr) { user_end = end->user_key(); } const Comparator* user_cmp = vset_->icmp_.user_comparator(); - for (size_t i = 0; i < files_[level].size(); ) { + for (size_t i = 0; i < files_[level].size();) { FileMetaData* f = files_[level][i++]; const Slice file_start = f->smallest.user_key(); const Slice file_limit = f->largest.user_key(); - if (begin != NULL && user_cmp->Compare(file_limit, user_begin) < 0) { + if (begin != nullptr && user_cmp->Compare(file_limit, user_begin) < 0) { // "f" is completely before specified range; skip it - } else if (end != NULL && user_cmp->Compare(file_start, user_end) > 0) { + } else if (end != nullptr && user_cmp->Compare(file_start, user_end) > 0) { // "f" is completely after specified range; skip it } else { inputs->push_back(f); if (level == 0) { // Level-0 files may overlap each other. So check if the newly // added file has expanded the range. If so, restart search. - if (begin != NULL && user_cmp->Compare(file_start, user_begin) < 0) { + if (begin != nullptr && user_cmp->Compare(file_start, user_begin) < 0) { user_begin = file_start; inputs->clear(); i = 0; - } else if (end != NULL && user_cmp->Compare(file_limit, user_end) > 0) { + } else if (end != nullptr && + user_cmp->Compare(file_limit, user_end) > 0) { user_end = file_limit; inputs->clear(); i = 0; @@ -620,9 +616,7 @@ class VersionSet::Builder { public: // Initialize a builder with the files from *base and other info from *vset - Builder(VersionSet* vset, Version* base) - : vset_(vset), - base_(base) { + Builder(VersionSet* vset, Version* base) : vset_(vset), base_(base) { base_->Ref(); BySmallestKey cmp; cmp.internal_comparator = &vset_->icmp_; @@ -636,8 +630,8 @@ class VersionSet::Builder { const FileSet* added = levels_[level].added_files; std::vector to_unref; to_unref.reserve(added->size()); - for (FileSet::const_iterator it = added->begin(); - it != added->end(); ++it) { + for (FileSet::const_iterator it = added->begin(); it != added->end(); + ++it) { to_unref.push_back(*it); } delete added; @@ -662,12 +656,9 @@ class VersionSet::Builder { } // Delete files - const VersionEdit::DeletedFileSet& del = edit->deleted_files_; - for (VersionEdit::DeletedFileSet::const_iterator iter = del.begin(); - iter != del.end(); - ++iter) { - const int level = iter->first; - const uint64_t number = iter->second; + for (const auto& deleted_file_set_kvp : edit->deleted_files_) { + const int level = deleted_file_set_kvp.first; + const uint64_t number = deleted_file_set_kvp.second; levels_[level].deleted_files.insert(number); } @@ -690,7 +681,7 @@ class VersionSet::Builder { // same as the compaction of 40KB of data. We are a little // conservative and allow approximately one seek for every 16KB // of data before triggering a compaction. - f->allowed_seeks = (f->file_size / 16384); + f->allowed_seeks = static_cast((f->file_size / 16384U)); if (f->allowed_seeks < 100) f->allowed_seeks = 100; levels_[level].deleted_files.erase(f->number); @@ -708,20 +699,17 @@ class VersionSet::Builder { const std::vector& base_files = base_->files_[level]; std::vector::const_iterator base_iter = base_files.begin(); std::vector::const_iterator base_end = base_files.end(); - const FileSet* added = levels_[level].added_files; - v->files_[level].reserve(base_files.size() + added->size()); - for (FileSet::const_iterator added_iter = added->begin(); - added_iter != added->end(); - ++added_iter) { + const FileSet* added_files = levels_[level].added_files; + v->files_[level].reserve(base_files.size() + added_files->size()); + for (const auto& added_file : *added_files) { // Add all smaller files listed in base_ - for (std::vector::const_iterator bpos - = std::upper_bound(base_iter, base_end, *added_iter, cmp); - base_iter != bpos; - ++base_iter) { + for (std::vector::const_iterator bpos = + std::upper_bound(base_iter, base_end, added_file, cmp); + base_iter != bpos; ++base_iter) { MaybeAddFile(v, level, *base_iter); } - MaybeAddFile(v, level, *added_iter); + MaybeAddFile(v, level, added_file); } // Add remaining base files @@ -733,7 +721,7 @@ class VersionSet::Builder { // Make sure there is no overlap in levels > 0 if (level > 0) { for (uint32_t i = 1; i < v->files_[level].size(); i++) { - const InternalKey& prev_end = v->files_[level][i-1]->largest; + const InternalKey& prev_end = v->files_[level][i - 1]->largest; const InternalKey& this_begin = v->files_[level][i]->smallest; if (vset_->icmp_.Compare(prev_end, this_begin) >= 0) { fprintf(stderr, "overlapping ranges in same level %s vs. %s\n", @@ -754,7 +742,7 @@ class VersionSet::Builder { std::vector* files = &v->files_[level]; if (level > 0 && !files->empty()) { // Must not overlap - assert(vset_->icmp_.Compare((*files)[files->size()-1]->largest, + assert(vset_->icmp_.Compare((*files)[files->size() - 1]->largest, f->smallest) < 0); } f->refs++; @@ -763,8 +751,7 @@ class VersionSet::Builder { } }; -VersionSet::VersionSet(const std::string& dbname, - const Options* options, +VersionSet::VersionSet(const std::string& dbname, const Options* options, TableCache* table_cache, const InternalKeyComparator* cmp) : env_(options->env), @@ -777,10 +764,10 @@ VersionSet::VersionSet(const std::string& dbname, last_sequence_(0), log_number_(0), prev_log_number_(0), - descriptor_file_(NULL), - descriptor_log_(NULL), + descriptor_file_(nullptr), + descriptor_log_(nullptr), dummy_versions_(this), - current_(NULL) { + current_(nullptr) { AppendVersion(new Version(this)); } @@ -795,7 +782,7 @@ void VersionSet::AppendVersion(Version* v) { // Make "v" current assert(v->refs_ == 0); assert(v != current_); - if (current_ != NULL) { + if (current_ != nullptr) { current_->Unref(); } current_ = v; @@ -835,10 +822,10 @@ Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) { // a temporary file that contains a snapshot of the current version. std::string new_manifest_file; Status s; - if (descriptor_log_ == NULL) { + if (descriptor_log_ == nullptr) { // No reason to unlock *mu here since we only hit this path in the // first call to LogAndApply (when opening the database). - assert(descriptor_file_ == NULL); + assert(descriptor_file_ == nullptr); new_manifest_file = DescriptorFileName(dbname_, manifest_file_number_); edit->SetNextFile(next_file_number_); s = env_->NewWritableFile(new_manifest_file, &descriptor_file_); @@ -884,8 +871,8 @@ Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) { if (!new_manifest_file.empty()) { delete descriptor_log_; delete descriptor_file_; - descriptor_log_ = NULL; - descriptor_file_ = NULL; + descriptor_log_ = nullptr; + descriptor_file_ = nullptr; env_->DeleteFile(new_manifest_file); } } @@ -893,10 +880,10 @@ Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) { return s; } -Status VersionSet::Recover() { +Status VersionSet::Recover(bool* save_manifest) { struct LogReporter : public log::Reader::Reporter { Status* status; - virtual void Corruption(size_t bytes, const Status& s) { + void Corruption(size_t bytes, const Status& s) override { if (this->status->ok()) *this->status = s; } }; @@ -907,7 +894,7 @@ Status VersionSet::Recover() { if (!s.ok()) { return s; } - if (current.empty() || current[current.size()-1] != '\n') { + if (current.empty() || current[current.size() - 1] != '\n') { return Status::Corruption("CURRENT file does not end with newline"); } current.resize(current.size() - 1); @@ -916,6 +903,10 @@ Status VersionSet::Recover() { SequentialFile* file; s = env_->NewSequentialFile(dscname, &file); if (!s.ok()) { + if (s.IsNotFound()) { + return Status::Corruption("CURRENT points to a non-existent file", + s.ToString()); + } return s; } @@ -932,7 +923,8 @@ Status VersionSet::Recover() { { LogReporter reporter; reporter.status = &s; - log::Reader reader(file, &reporter, true/*checksum*/, 0/*initial_offset*/); + log::Reader reader(file, &reporter, true /*checksum*/, + 0 /*initial_offset*/); Slice record; std::string scratch; while (reader.ReadRecord(&record, &scratch) && s.ok()) { @@ -973,7 +965,7 @@ Status VersionSet::Recover() { } } delete file; - file = NULL; + file = nullptr; if (s.ok()) { if (!have_next_file) { @@ -1003,11 +995,49 @@ Status VersionSet::Recover() { last_sequence_ = last_sequence; log_number_ = log_number; prev_log_number_ = prev_log_number; + + // See if we can reuse the existing MANIFEST file. + if (ReuseManifest(dscname, current)) { + // No need to save new manifest + } else { + *save_manifest = true; + } } return s; } +bool VersionSet::ReuseManifest(const std::string& dscname, + const std::string& dscbase) { + if (!options_->reuse_logs) { + return false; + } + FileType manifest_type; + uint64_t manifest_number; + uint64_t manifest_size; + if (!ParseFileName(dscbase, &manifest_number, &manifest_type) || + manifest_type != kDescriptorFile || + !env_->GetFileSize(dscname, &manifest_size).ok() || + // Make new compacted MANIFEST if old one is too big + manifest_size >= TargetFileSize(options_)) { + return false; + } + + assert(descriptor_file_ == nullptr); + assert(descriptor_log_ == nullptr); + Status r = env_->NewAppendableFile(dscname, &descriptor_file_); + if (!r.ok()) { + Log(options_->info_log, "Reuse MANIFEST: %s\n", r.ToString().c_str()); + assert(descriptor_file_ == nullptr); + return false; + } + + Log(options_->info_log, "Reusing MANIFEST %s\n", dscname.c_str()); + descriptor_log_ = new log::Writer(descriptor_file_, manifest_size); + manifest_file_number_ = manifest_number; + return true; +} + void VersionSet::MarkFileNumberUsed(uint64_t number) { if (next_file_number_ <= number) { next_file_number_ = number + 1; @@ -1019,7 +1049,7 @@ void VersionSet::Finalize(Version* v) { int best_level = -1; double best_score = -1; - for (int level = 0; level < config::kNumLevels-1; level++) { + for (int level = 0; level < config::kNumLevels - 1; level++) { double score; if (level == 0) { // We treat level-0 specially by bounding the number of files @@ -1034,11 +1064,12 @@ void VersionSet::Finalize(Version* v) { // setting, or very high compression ratios, or lots of // overwrites/deletions). score = v->files_[level].size() / - static_cast(config::kL0_CompactionTrigger); + static_cast(config::kL0_CompactionTrigger); } else { // Compute the ratio of current size to size limit. const uint64_t level_bytes = TotalFileSize(v->files_[level]); - score = static_cast(level_bytes) / MaxBytesForLevel(level); + score = + static_cast(level_bytes) / MaxBytesForLevel(options_, level); } if (score > best_score) { @@ -1089,16 +1120,12 @@ int VersionSet::NumLevelFiles(int level) const { const char* VersionSet::LevelSummary(LevelSummaryStorage* scratch) const { // Update code if kNumLevels changes - assert(config::kNumLevels == 7); + static_assert(config::kNumLevels == 7, ""); snprintf(scratch->buffer, sizeof(scratch->buffer), - "files[ %d %d %d %d %d %d %d ]", - int(current_->files_[0].size()), - int(current_->files_[1].size()), - int(current_->files_[2].size()), - int(current_->files_[3].size()), - int(current_->files_[4].size()), - int(current_->files_[5].size()), - int(current_->files_[6].size())); + "files[ %d %d %d %d %d %d %d ]", int(current_->files_[0].size()), + int(current_->files_[1].size()), int(current_->files_[2].size()), + int(current_->files_[3].size()), int(current_->files_[4].size()), + int(current_->files_[5].size()), int(current_->files_[6].size())); return scratch->buffer; } @@ -1124,7 +1151,7 @@ uint64_t VersionSet::ApproximateOffsetOf(Version* v, const InternalKey& ikey) { Table* tableptr; Iterator* iter = table_cache_->NewIterator( ReadOptions(), files[i]->number, files[i]->file_size, &tableptr); - if (tableptr != NULL) { + if (tableptr != nullptr) { result += tableptr->ApproximateOffsetOf(ikey.Encode()); } delete iter; @@ -1135,8 +1162,7 @@ uint64_t VersionSet::ApproximateOffsetOf(Version* v, const InternalKey& ikey) { } void VersionSet::AddLiveFiles(std::set* live) { - for (Version* v = dummy_versions_.next_; - v != &dummy_versions_; + for (Version* v = dummy_versions_.next_; v != &dummy_versions_; v = v->next_) { for (int level = 0; level < config::kNumLevels; level++) { const std::vector& files = v->files_[level]; @@ -1159,7 +1185,7 @@ int64_t VersionSet::MaxNextLevelOverlappingBytes() { for (int level = 1; level < config::kNumLevels - 1; level++) { for (size_t i = 0; i < current_->files_[level].size(); i++) { const FileMetaData* f = current_->files_[level][i]; - current_->GetOverlappingInputs(level+1, &f->smallest, &f->largest, + current_->GetOverlappingInputs(level + 1, &f->smallest, &f->largest, &overlaps); const int64_t sum = TotalFileSize(overlaps); if (sum > result) { @@ -1174,8 +1200,7 @@ int64_t VersionSet::MaxNextLevelOverlappingBytes() { // *smallest, *largest. // REQUIRES: inputs is not empty void VersionSet::GetRange(const std::vector& inputs, - InternalKey* smallest, - InternalKey* largest) { + InternalKey* smallest, InternalKey* largest) { assert(!inputs.empty()); smallest->Clear(); largest->Clear(); @@ -1200,8 +1225,7 @@ void VersionSet::GetRange(const std::vector& inputs, // REQUIRES: inputs is not empty void VersionSet::GetRange2(const std::vector& inputs1, const std::vector& inputs2, - InternalKey* smallest, - InternalKey* largest) { + InternalKey* smallest, InternalKey* largest) { std::vector all = inputs1; all.insert(all.end(), inputs2.begin(), inputs2.end()); GetRange(all, smallest, largest); @@ -1223,8 +1247,8 @@ Iterator* VersionSet::MakeInputIterator(Compaction* c) { if (c->level() + which == 0) { const std::vector& files = c->inputs_[which]; for (size_t i = 0; i < files.size(); i++) { - list[num++] = table_cache_->NewIterator( - options, files[i]->number, files[i]->file_size); + list[num++] = table_cache_->NewIterator(options, files[i]->number, + files[i]->file_size); } } else { // Create concatenating iterator for the files from this level @@ -1247,12 +1271,12 @@ Compaction* VersionSet::PickCompaction() { // We prefer compactions triggered by too much data in a level over // the compactions triggered by seeks. const bool size_compaction = (current_->compaction_score_ >= 1); - const bool seek_compaction = (current_->file_to_compact_ != NULL); + const bool seek_compaction = (current_->file_to_compact_ != nullptr); if (size_compaction) { level = current_->compaction_level_; assert(level >= 0); - assert(level+1 < config::kNumLevels); - c = new Compaction(level); + assert(level + 1 < config::kNumLevels); + c = new Compaction(options_, level); // Pick the first file that comes after compact_pointer_[level] for (size_t i = 0; i < current_->files_[level].size(); i++) { @@ -1269,10 +1293,10 @@ Compaction* VersionSet::PickCompaction() { } } else if (seek_compaction) { level = current_->file_to_compact_level_; - c = new Compaction(level); + c = new Compaction(options_, level); c->inputs_[0].push_back(current_->file_to_compact_); } else { - return NULL; + return nullptr; } c->input_version_ = current_; @@ -1294,12 +1318,94 @@ Compaction* VersionSet::PickCompaction() { return c; } +// Finds the largest key in a vector of files. Returns true if files it not +// empty. +bool FindLargestKey(const InternalKeyComparator& icmp, + const std::vector& files, + InternalKey* largest_key) { + if (files.empty()) { + return false; + } + *largest_key = files[0]->largest; + for (size_t i = 1; i < files.size(); ++i) { + FileMetaData* f = files[i]; + if (icmp.Compare(f->largest, *largest_key) > 0) { + *largest_key = f->largest; + } + } + return true; +} + +// Finds minimum file b2=(l2, u2) in level file for which l2 > u1 and +// user_key(l2) = user_key(u1) +FileMetaData* FindSmallestBoundaryFile( + const InternalKeyComparator& icmp, + const std::vector& level_files, + const InternalKey& largest_key) { + const Comparator* user_cmp = icmp.user_comparator(); + FileMetaData* smallest_boundary_file = nullptr; + for (size_t i = 0; i < level_files.size(); ++i) { + FileMetaData* f = level_files[i]; + if (icmp.Compare(f->smallest, largest_key) > 0 && + user_cmp->Compare(f->smallest.user_key(), largest_key.user_key()) == + 0) { + if (smallest_boundary_file == nullptr || + icmp.Compare(f->smallest, smallest_boundary_file->smallest) < 0) { + smallest_boundary_file = f; + } + } + } + return smallest_boundary_file; +} + +// Extracts the largest file b1 from |compaction_files| and then searches for a +// b2 in |level_files| for which user_key(u1) = user_key(l2). If it finds such a +// file b2 (known as a boundary file) it adds it to |compaction_files| and then +// searches again using this new upper bound. +// +// If there are two blocks, b1=(l1, u1) and b2=(l2, u2) and +// user_key(u1) = user_key(l2), and if we compact b1 but not b2 then a +// subsequent get operation will yield an incorrect result because it will +// return the record from b2 in level i rather than from b1 because it searches +// level by level for records matching the supplied user key. +// +// parameters: +// in level_files: List of files to search for boundary files. +// in/out compaction_files: List of files to extend by adding boundary files. +void AddBoundaryInputs(const InternalKeyComparator& icmp, + const std::vector& level_files, + std::vector* compaction_files) { + InternalKey largest_key; + + // Quick return if compaction_files is empty. + if (!FindLargestKey(icmp, *compaction_files, &largest_key)) { + return; + } + + bool continue_searching = true; + while (continue_searching) { + FileMetaData* smallest_boundary_file = + FindSmallestBoundaryFile(icmp, level_files, largest_key); + + // If a boundary file was found advance largest_key, otherwise we're done. + if (smallest_boundary_file != NULL) { + compaction_files->push_back(smallest_boundary_file); + largest_key = smallest_boundary_file->largest; + } else { + continue_searching = false; + } + } +} + void VersionSet::SetupOtherInputs(Compaction* c) { const int level = c->level(); InternalKey smallest, largest; + + AddBoundaryInputs(icmp_, current_->files_[level], &c->inputs_[0]); GetRange(c->inputs_[0], &smallest, &largest); - current_->GetOverlappingInputs(level+1, &smallest, &largest, &c->inputs_[1]); + current_->GetOverlappingInputs(level + 1, &smallest, &largest, + &c->inputs_[1]); // Get entire range covered by compaction InternalKey all_start, all_limit; @@ -1310,26 +1416,24 @@ void VersionSet::SetupOtherInputs(Compaction* c) { if (!c->inputs_[1].empty()) { std::vector expanded0; current_->GetOverlappingInputs(level, &all_start, &all_limit, &expanded0); + AddBoundaryInputs(icmp_, current_->files_[level], &expanded0); const int64_t inputs0_size = TotalFileSize(c->inputs_[0]); const int64_t inputs1_size = TotalFileSize(c->inputs_[1]); const int64_t expanded0_size = TotalFileSize(expanded0); if (expanded0.size() > c->inputs_[0].size() && - inputs1_size + expanded0_size < kExpandedCompactionByteSizeLimit) { + inputs1_size + expanded0_size < + ExpandedCompactionByteSizeLimit(options_)) { InternalKey new_start, new_limit; GetRange(expanded0, &new_start, &new_limit); std::vector expanded1; - current_->GetOverlappingInputs(level+1, &new_start, &new_limit, + current_->GetOverlappingInputs(level + 1, &new_start, &new_limit, &expanded1); if (expanded1.size() == c->inputs_[1].size()) { Log(options_->info_log, "Expanding@%d %d+%d (%ld+%ld bytes) to %d+%d (%ld+%ld bytes)\n", - level, - int(c->inputs_[0].size()), - int(c->inputs_[1].size()), - long(inputs0_size), long(inputs1_size), - int(expanded0.size()), - int(expanded1.size()), - long(expanded0_size), long(inputs1_size)); + level, int(c->inputs_[0].size()), int(c->inputs_[1].size()), + long(inputs0_size), long(inputs1_size), int(expanded0.size()), + int(expanded1.size()), long(expanded0_size), long(inputs1_size)); smallest = new_start; largest = new_limit; c->inputs_[0] = expanded0; @@ -1346,13 +1450,6 @@ void VersionSet::SetupOtherInputs(Compaction* c) { &c->grandparents_); } - if (false) { - Log(options_->info_log, "Compacting %d '%s' .. '%s'", - level, - smallest.DebugString().c_str(), - largest.DebugString().c_str()); - } - // Update the place where we will do the next compaction for this level. // We update this immediately instead of waiting for the VersionEdit // to be applied so that if the compaction fails, we will try a different @@ -1361,14 +1458,12 @@ void VersionSet::SetupOtherInputs(Compaction* c) { c->edit_.SetCompactPointer(level, largest); } -Compaction* VersionSet::CompactRange( - int level, - const InternalKey* begin, - const InternalKey* end) { +Compaction* VersionSet::CompactRange(int level, const InternalKey* begin, + const InternalKey* end) { std::vector inputs; current_->GetOverlappingInputs(level, begin, end, &inputs); if (inputs.empty()) { - return NULL; + return nullptr; } // Avoid compacting too much in one shot in case the range is large. @@ -1376,7 +1471,7 @@ Compaction* VersionSet::CompactRange( // and we must not pick one file and drop another older file if the // two files overlap. if (level > 0) { - const uint64_t limit = MaxFileSizeForLevel(level); + const uint64_t limit = MaxFileSizeForLevel(options_, level); uint64_t total = 0; for (size_t i = 0; i < inputs.size(); i++) { uint64_t s = inputs[i]->file_size; @@ -1388,7 +1483,7 @@ Compaction* VersionSet::CompactRange( } } - Compaction* c = new Compaction(level); + Compaction* c = new Compaction(options_, level); c->input_version_ = current_; c->input_version_->Ref(); c->inputs_[0] = inputs; @@ -1396,10 +1491,10 @@ Compaction* VersionSet::CompactRange( return c; } -Compaction::Compaction(int level) +Compaction::Compaction(const Options* options, int level) : level_(level), - max_output_file_size_(MaxFileSizeForLevel(level)), - input_version_(NULL), + max_output_file_size_(MaxFileSizeForLevel(options, level)), + input_version_(nullptr), grandparent_index_(0), seen_key_(false), overlapped_bytes_(0) { @@ -1409,18 +1504,19 @@ Compaction::Compaction(int level) } Compaction::~Compaction() { - if (input_version_ != NULL) { + if (input_version_ != nullptr) { input_version_->Unref(); } } bool Compaction::IsTrivialMove() const { + const VersionSet* vset = input_version_->vset_; // 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. - return (num_input_files(0) == 1 && - num_input_files(1) == 0 && - TotalFileSize(grandparents_) <= kMaxGrandParentOverlapBytes); + return (num_input_files(0) == 1 && num_input_files(1) == 0 && + TotalFileSize(grandparents_) <= + MaxGrandParentOverlapBytes(vset->options_)); } void Compaction::AddInputDeletions(VersionEdit* edit) { @@ -1436,7 +1532,7 @@ bool Compaction::IsBaseLevelForKey(const Slice& user_key) { const Comparator* user_cmp = input_version_->vset_->icmp_.user_comparator(); for (int lvl = level_ + 2; lvl < config::kNumLevels; lvl++) { const std::vector& files = input_version_->files_[lvl]; - for (; level_ptrs_[lvl] < files.size(); ) { + while (level_ptrs_[lvl] < files.size()) { FileMetaData* f = files[level_ptrs_[lvl]]; if (user_cmp->Compare(user_key, f->largest.user_key()) <= 0) { // We've advanced far enough @@ -1453,11 +1549,13 @@ bool Compaction::IsBaseLevelForKey(const Slice& user_key) { } bool Compaction::ShouldStopBefore(const Slice& internal_key) { + const VersionSet* vset = input_version_->vset_; // Scan to find earliest grandparent file that contains key. - const InternalKeyComparator* icmp = &input_version_->vset_->icmp_; + const InternalKeyComparator* icmp = &vset->icmp_; while (grandparent_index_ < grandparents_.size() && - icmp->Compare(internal_key, - grandparents_[grandparent_index_]->largest.Encode()) > 0) { + icmp->Compare(internal_key, + grandparents_[grandparent_index_]->largest.Encode()) > + 0) { if (seen_key_) { overlapped_bytes_ += grandparents_[grandparent_index_]->file_size; } @@ -1465,7 +1563,7 @@ bool Compaction::ShouldStopBefore(const Slice& internal_key) { } seen_key_ = true; - if (overlapped_bytes_ > kMaxGrandParentOverlapBytes) { + if (overlapped_bytes_ > MaxGrandParentOverlapBytes(vset->options_)) { // Too much overlap for current output; start new output overlapped_bytes_ = 0; return true; @@ -1475,9 +1573,9 @@ bool Compaction::ShouldStopBefore(const Slice& internal_key) { } void Compaction::ReleaseInputs() { - if (input_version_ != NULL) { + if (input_version_ != nullptr) { input_version_->Unref(); - input_version_ = NULL; + input_version_ = nullptr; } } diff --git a/db/version_set.h b/db/version_set.h index 8dc14b8..69f3d70 100644 --- a/db/version_set.h +++ b/db/version_set.h @@ -18,6 +18,7 @@ #include #include #include + #include "db/dbformat.h" #include "db/version_edit.h" #include "port/port.h" @@ -25,7 +26,9 @@ namespace leveldb { -namespace log { class Writer; } +namespace log { +class Writer; +} class Compaction; class Iterator; @@ -39,30 +42,23 @@ class WritableFile; // Return the smallest index i such that files[i]->largest >= key. // Return files.size() if there is no such file. // REQUIRES: "files" contains a sorted list of non-overlapping files. -extern int FindFile(const InternalKeyComparator& icmp, - const std::vector& files, - const Slice& key); +int FindFile(const InternalKeyComparator& icmp, + const std::vector& files, const Slice& key); // Returns true iff some file in "files" overlaps the user key range // [*smallest,*largest]. -// smallest==NULL represents a key smaller than all keys in the DB. -// largest==NULL represents a key largest than all keys in the DB. +// 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, files[] contains disjoint ranges // in sorted order. -extern bool SomeFileOverlapsRange( - const InternalKeyComparator& icmp, - bool disjoint_sorted_files, - const std::vector& files, - const Slice* smallest_user_key, - const Slice* largest_user_key); +bool SomeFileOverlapsRange(const InternalKeyComparator& icmp, + bool disjoint_sorted_files, + const std::vector& files, + const Slice* smallest_user_key, + const Slice* largest_user_key); 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&, std::vector* iters); - // Lookup the value for key. If found, store it in *val and // return OK. Else return a non-OK status. Fills *stats. // REQUIRES: lock is not held @@ -70,6 +66,12 @@ class Version { FileMetaData* seek_file; int seek_file_level; }; + + // 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&, std::vector* iters); + Status Get(const ReadOptions&, const LookupKey& key, std::string* val, GetStats* stats); @@ -91,16 +93,15 @@ class Version { void GetOverlappingInputs( int level, - const InternalKey* begin, // NULL means before all keys - const InternalKey* end, // NULL means after all keys + const InternalKey* begin, // nullptr means before all keys + const InternalKey* end, // nullptr means after all keys std::vector* inputs); // 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, + // smallest_user_key==nullptr represents a key smaller than all the DB's keys. + // largest_user_key==nullptr represents a key largest than all the DB's keys. + bool OverlapInLevel(int level, const Slice* smallest_user_key, const Slice* largest_user_key); // Return the level at which we should place a new memtable compaction @@ -118,6 +119,22 @@ class Version { friend class VersionSet; class LevelFileNumIterator; + + explicit Version(VersionSet* vset) + : vset_(vset), + next_(this), + prev_(this), + refs_(0), + file_to_compact_(nullptr), + file_to_compact_level_(-1), + compaction_score_(-1), + compaction_level_(-1) {} + + Version(const Version&) = delete; + Version& operator=(const Version&) = delete; + + ~Version(); + Iterator* NewConcatenatingIterator(const ReadOptions&, int level) const; // Call func(arg, level, f) for every file that overlaps user_key in @@ -125,14 +142,13 @@ class Version { // false, makes no more calls. // // REQUIRES: user portion of internal_key == user_key. - void ForEachOverlapping(Slice user_key, Slice internal_key, - void* arg, + void ForEachOverlapping(Slice user_key, Slice internal_key, void* arg, bool (*func)(void*, int, FileMetaData*)); - 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 + 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 // List of files per level std::vector files_[config::kNumLevels]; @@ -146,28 +162,15 @@ class Version { // are initialized by Finalize(). double compaction_score_; int compaction_level_; - - explicit Version(VersionSet* vset) - : vset_(vset), next_(this), prev_(this), refs_(0), - file_to_compact_(NULL), - file_to_compact_level_(-1), - compaction_score_(-1), - compaction_level_(-1) { - } - - ~Version(); - - // No copying allowed - Version(const Version&); - void operator=(const Version&); }; class VersionSet { public: - VersionSet(const std::string& dbname, - const Options* options, - TableCache* table_cache, - const InternalKeyComparator*); + VersionSet(const std::string& dbname, const Options* options, + TableCache* table_cache, const InternalKeyComparator*); + VersionSet(const VersionSet&) = delete; + VersionSet& operator=(const VersionSet&) = delete; + ~VersionSet(); // Apply *edit to the current version to form a new descriptor that @@ -179,7 +182,7 @@ class VersionSet { EXCLUSIVE_LOCKS_REQUIRED(mu); // Recover the last saved descriptor from persistent storage. - Status Recover(); + Status Recover(bool* save_manifest); // Return the current version. Version* current() const { return current_; } @@ -225,19 +228,17 @@ class VersionSet { uint64_t PrevLogNumber() const { return prev_log_number_; } // Pick level and inputs for a new compaction. - // Returns NULL if there is no compaction to be done. + // 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. Compaction* PickCompaction(); // Return a compaction object for compacting the range [begin,end] in - // the specified level. Returns NULL if there is nothing in that + // the specified level. Returns nullptr if there is nothing in that // level that overlaps the specified range. Caller should delete // the result. - Compaction* CompactRange( - int level, - const InternalKey* begin, - const InternalKey* end); + Compaction* CompactRange(int level, const InternalKey* begin, + const InternalKey* end); // Return the maximum overlapping data (in bytes) at next level for any // file at a level >= 1. @@ -250,7 +251,7 @@ class VersionSet { // Returns true iff some level needs a compaction. bool NeedsCompaction() const { Version* v = current_; - return (v->compaction_score_ >= 1) || (v->file_to_compact_ != NULL); + return (v->compaction_score_ >= 1) || (v->file_to_compact_ != nullptr); } // Add all files listed in any live version to *live. @@ -274,16 +275,16 @@ class VersionSet { friend class Compaction; friend class Version; + bool ReuseManifest(const std::string& dscname, const std::string& dscbase); + void Finalize(Version* v); - void GetRange(const std::vector& inputs, - InternalKey* smallest, + void GetRange(const std::vector& inputs, InternalKey* smallest, InternalKey* largest); void GetRange2(const std::vector& inputs1, const std::vector& inputs2, - InternalKey* smallest, - InternalKey* largest); + InternalKey* smallest, InternalKey* largest); void SetupOtherInputs(Compaction* c); @@ -312,10 +313,6 @@ class VersionSet { // Per-level key at which the next compaction at that level should start. // Either an empty string, or a valid InternalKey. std::string compact_pointer_[config::kNumLevels]; - - // No copying allowed - VersionSet(const VersionSet&); - void operator=(const VersionSet&); }; // A Compaction encapsulates information about a compaction. @@ -364,7 +361,7 @@ class Compaction { friend class Version; friend class VersionSet; - explicit Compaction(int level); + Compaction(const Options* options, int level); int level_; uint64_t max_output_file_size_; @@ -372,9 +369,9 @@ class Compaction { VersionEdit edit_; // Each compaction reads inputs from "level_" and "level_+1" - std::vector inputs_[2]; // The two sets of inputs + std::vector inputs_[2]; // The two sets of inputs - // State used to check for number of of overlapping grandparent files + // State used to check for number of overlapping grandparent files // (parent == level_ + 1, grandparent == level_ + 2) std::vector grandparents_; size_t grandparent_index_; // Index in grandparent_starts_ diff --git a/db/version_set_test.cc b/db/version_set_test.cc index 501e34d..c1056a1 100644 --- a/db/version_set_test.cc +++ b/db/version_set_test.cc @@ -11,10 +11,7 @@ namespace leveldb { class FindFileTest { public: - std::vector files_; - bool disjoint_sorted_files_; - - FindFileTest() : disjoint_sorted_files_(true) { } + FindFileTest() : disjoint_sorted_files_(true) {} ~FindFileTest() { for (int i = 0; i < files_.size(); i++) { @@ -40,20 +37,25 @@ class FindFileTest { bool Overlaps(const char* smallest, const char* largest) { InternalKeyComparator cmp(BytewiseComparator()); - Slice s(smallest != NULL ? smallest : ""); - Slice l(largest != NULL ? largest : ""); + Slice s(smallest != nullptr ? smallest : ""); + Slice l(largest != nullptr ? largest : ""); return SomeFileOverlapsRange(cmp, disjoint_sorted_files_, files_, - (smallest != NULL ? &s : NULL), - (largest != NULL ? &l : NULL)); + (smallest != nullptr ? &s : nullptr), + (largest != nullptr ? &l : nullptr)); } + + bool disjoint_sorted_files_; + + private: + std::vector files_; }; TEST(FindFileTest, Empty) { ASSERT_EQ(0, Find("foo")); - ASSERT_TRUE(! Overlaps("a", "z")); - ASSERT_TRUE(! Overlaps(NULL, "z")); - ASSERT_TRUE(! Overlaps("a", NULL)); - ASSERT_TRUE(! Overlaps(NULL, NULL)); + ASSERT_TRUE(!Overlaps("a", "z")); + ASSERT_TRUE(!Overlaps(nullptr, "z")); + ASSERT_TRUE(!Overlaps("a", nullptr)); + ASSERT_TRUE(!Overlaps(nullptr, nullptr)); } TEST(FindFileTest, Single) { @@ -65,8 +67,8 @@ TEST(FindFileTest, Single) { ASSERT_EQ(1, Find("q1")); ASSERT_EQ(1, Find("z")); - ASSERT_TRUE(! Overlaps("a", "b")); - ASSERT_TRUE(! Overlaps("z1", "z2")); + 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")); @@ -78,15 +80,14 @@ TEST(FindFileTest, Single) { ASSERT_TRUE(Overlaps("q", "q")); ASSERT_TRUE(Overlaps("q", "q1")); - ASSERT_TRUE(! Overlaps(NULL, "j")); - ASSERT_TRUE(! Overlaps("r", NULL)); - ASSERT_TRUE(Overlaps(NULL, "p")); - ASSERT_TRUE(Overlaps(NULL, "p1")); - ASSERT_TRUE(Overlaps("q", NULL)); - ASSERT_TRUE(Overlaps(NULL, NULL)); + 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(FindFileTest, Multiple) { Add("150", "200"); Add("200", "250"); @@ -110,10 +111,10 @@ TEST(FindFileTest, Multiple) { 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", "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")); @@ -130,25 +131,25 @@ TEST(FindFileTest, MultipleNullBoundaries) { Add("200", "250"); Add("300", "350"); Add("400", "450"); - ASSERT_TRUE(! Overlaps(NULL, "149")); - ASSERT_TRUE(! Overlaps("451", NULL)); - ASSERT_TRUE(Overlaps(NULL, NULL)); - ASSERT_TRUE(Overlaps(NULL, "150")); - ASSERT_TRUE(Overlaps(NULL, "199")); - ASSERT_TRUE(Overlaps(NULL, "200")); - ASSERT_TRUE(Overlaps(NULL, "201")); - ASSERT_TRUE(Overlaps(NULL, "400")); - ASSERT_TRUE(Overlaps(NULL, "800")); - ASSERT_TRUE(Overlaps("100", NULL)); - ASSERT_TRUE(Overlaps("200", NULL)); - ASSERT_TRUE(Overlaps("449", NULL)); - ASSERT_TRUE(Overlaps("450", NULL)); + 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(FindFileTest, OverlapSequenceChecks) { Add("200", "200", 5000, 3000); - ASSERT_TRUE(! Overlaps("199", "199")); - ASSERT_TRUE(! Overlaps("201", "300")); + 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")); @@ -158,8 +159,8 @@ TEST(FindFileTest, OverlappingFiles) { Add("150", "600"); Add("400", "500"); disjoint_sorted_files_ = false; - ASSERT_TRUE(! Overlaps("100", "149")); - ASSERT_TRUE(! Overlaps("601", "700")); + 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")); @@ -172,8 +173,160 @@ TEST(FindFileTest, OverlappingFiles) { ASSERT_TRUE(Overlaps("600", "700")); } +void AddBoundaryInputs(const InternalKeyComparator& icmp, + const std::vector& level_files, + std::vector* compaction_files); + +class AddBoundaryInputsTest { + public: + std::vector level_files_; + std::vector compaction_files_; + std::vector all_files_; + InternalKeyComparator icmp_; + + AddBoundaryInputsTest() : icmp_(BytewiseComparator()) {} + + ~AddBoundaryInputsTest() { + for (size_t i = 0; i < all_files_.size(); ++i) { + delete all_files_[i]; + } + all_files_.clear(); + } + + FileMetaData* CreateFileMetaData(uint64_t number, InternalKey smallest, + InternalKey largest) { + FileMetaData* f = new FileMetaData(); + f->number = number; + f->smallest = smallest; + f->largest = largest; + all_files_.push_back(f); + return f; + } +}; + +TEST(AddBoundaryInputsTest, TestEmptyFileSets) { + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_TRUE(compaction_files_.empty()); + ASSERT_TRUE(level_files_.empty()); +} + +TEST(AddBoundaryInputsTest, TestEmptyLevelFiles) { + FileMetaData* f1 = + CreateFileMetaData(1, InternalKey("100", 2, kTypeValue), + InternalKey(InternalKey("100", 1, kTypeValue))); + compaction_files_.push_back(f1); + + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_EQ(1, compaction_files_.size()); + ASSERT_EQ(f1, compaction_files_[0]); + ASSERT_TRUE(level_files_.empty()); +} + +TEST(AddBoundaryInputsTest, TestEmptyCompactionFiles) { + FileMetaData* f1 = + CreateFileMetaData(1, InternalKey("100", 2, kTypeValue), + InternalKey(InternalKey("100", 1, kTypeValue))); + level_files_.push_back(f1); + + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_TRUE(compaction_files_.empty()); + ASSERT_EQ(1, level_files_.size()); + ASSERT_EQ(f1, level_files_[0]); +} + +TEST(AddBoundaryInputsTest, TestNoBoundaryFiles) { + FileMetaData* f1 = + CreateFileMetaData(1, InternalKey("100", 2, kTypeValue), + InternalKey(InternalKey("100", 1, kTypeValue))); + FileMetaData* f2 = + CreateFileMetaData(1, InternalKey("200", 2, kTypeValue), + InternalKey(InternalKey("200", 1, kTypeValue))); + FileMetaData* f3 = + CreateFileMetaData(1, InternalKey("300", 2, kTypeValue), + InternalKey(InternalKey("300", 1, kTypeValue))); + + level_files_.push_back(f3); + level_files_.push_back(f2); + level_files_.push_back(f1); + compaction_files_.push_back(f2); + compaction_files_.push_back(f3); + + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_EQ(2, compaction_files_.size()); +} + +TEST(AddBoundaryInputsTest, TestOneBoundaryFiles) { + FileMetaData* f1 = + CreateFileMetaData(1, InternalKey("100", 3, kTypeValue), + InternalKey(InternalKey("100", 2, kTypeValue))); + FileMetaData* f2 = + CreateFileMetaData(1, InternalKey("100", 1, kTypeValue), + InternalKey(InternalKey("200", 3, kTypeValue))); + FileMetaData* f3 = + CreateFileMetaData(1, InternalKey("300", 2, kTypeValue), + InternalKey(InternalKey("300", 1, kTypeValue))); + + level_files_.push_back(f3); + level_files_.push_back(f2); + level_files_.push_back(f1); + compaction_files_.push_back(f1); + + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_EQ(2, compaction_files_.size()); + ASSERT_EQ(f1, compaction_files_[0]); + ASSERT_EQ(f2, compaction_files_[1]); +} + +TEST(AddBoundaryInputsTest, TestTwoBoundaryFiles) { + FileMetaData* f1 = + CreateFileMetaData(1, InternalKey("100", 6, kTypeValue), + InternalKey(InternalKey("100", 5, kTypeValue))); + FileMetaData* f2 = + CreateFileMetaData(1, InternalKey("100", 2, kTypeValue), + InternalKey(InternalKey("300", 1, kTypeValue))); + FileMetaData* f3 = + CreateFileMetaData(1, InternalKey("100", 4, kTypeValue), + InternalKey(InternalKey("100", 3, kTypeValue))); + + level_files_.push_back(f2); + level_files_.push_back(f3); + level_files_.push_back(f1); + compaction_files_.push_back(f1); + + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_EQ(3, compaction_files_.size()); + ASSERT_EQ(f1, compaction_files_[0]); + ASSERT_EQ(f3, compaction_files_[1]); + ASSERT_EQ(f2, compaction_files_[2]); +} + +TEST(AddBoundaryInputsTest, TestDisjoinFilePointers) { + FileMetaData* f1 = + CreateFileMetaData(1, InternalKey("100", 6, kTypeValue), + InternalKey(InternalKey("100", 5, kTypeValue))); + FileMetaData* f2 = + CreateFileMetaData(1, InternalKey("100", 6, kTypeValue), + InternalKey(InternalKey("100", 5, kTypeValue))); + FileMetaData* f3 = + CreateFileMetaData(1, InternalKey("100", 2, kTypeValue), + InternalKey(InternalKey("300", 1, kTypeValue))); + FileMetaData* f4 = + CreateFileMetaData(1, InternalKey("100", 4, kTypeValue), + InternalKey(InternalKey("100", 3, kTypeValue))); + + level_files_.push_back(f2); + level_files_.push_back(f3); + level_files_.push_back(f4); + + compaction_files_.push_back(f1); + + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_EQ(3, compaction_files_.size()); + ASSERT_EQ(f1, compaction_files_[0]); + ASSERT_EQ(f4, compaction_files_[1]); + ASSERT_EQ(f3, compaction_files_[2]); +} + } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/db/write_batch.cc b/db/write_batch.cc index 33f4a42..b54313c 100644 --- a/db/write_batch.cc +++ b/db/write_batch.cc @@ -15,10 +15,10 @@ #include "leveldb/write_batch.h" -#include "leveldb/db.h" #include "db/dbformat.h" #include "db/memtable.h" #include "db/write_batch_internal.h" +#include "leveldb/db.h" #include "util/coding.h" namespace leveldb { @@ -26,19 +26,19 @@ namespace leveldb { // WriteBatch header has an 8-byte sequence number followed by a 4-byte count. static const size_t kHeader = 12; -WriteBatch::WriteBatch() { - Clear(); -} +WriteBatch::WriteBatch() { Clear(); } -WriteBatch::~WriteBatch() { } +WriteBatch::~WriteBatch() = default; -WriteBatch::Handler::~Handler() { } +WriteBatch::Handler::~Handler() = default; void WriteBatch::Clear() { rep_.clear(); rep_.resize(kHeader); } +size_t WriteBatch::ApproximateSize() const { return rep_.size(); } + Status WriteBatch::Iterate(Handler* handler) const { Slice input(rep_); if (input.size() < kHeader) { @@ -108,25 +108,28 @@ void WriteBatch::Delete(const Slice& key) { PutLengthPrefixedSlice(&rep_, key); } +void WriteBatch::Append(const WriteBatch& source) { + WriteBatchInternal::Append(this, &source); +} + namespace { class MemTableInserter : public WriteBatch::Handler { public: SequenceNumber sequence_; MemTable* mem_; - virtual void Put(const Slice& key, const Slice& value) { + void Put(const Slice& key, const Slice& value) override { mem_->Add(sequence_, kTypeValue, key, value); sequence_++; } - virtual void Delete(const Slice& key) { + void Delete(const Slice& key) override { mem_->Add(sequence_, kTypeDeletion, key, Slice()); sequence_++; } }; } // namespace -Status WriteBatchInternal::InsertInto(const WriteBatch* b, - MemTable* memtable) { +Status WriteBatchInternal::InsertInto(const WriteBatch* b, MemTable* memtable) { MemTableInserter inserter; inserter.sequence_ = WriteBatchInternal::Sequence(b); inserter.mem_ = memtable; diff --git a/db/write_batch_internal.h b/db/write_batch_internal.h index 9448ef7..fce86e3 100644 --- a/db/write_batch_internal.h +++ b/db/write_batch_internal.h @@ -29,13 +29,9 @@ class WriteBatchInternal { // this batch. static void SetSequence(WriteBatch* batch, SequenceNumber seq); - static Slice Contents(const WriteBatch* batch) { - return Slice(batch->rep_); - } + static Slice Contents(const WriteBatch* batch) { return Slice(batch->rep_); } - static size_t ByteSize(const WriteBatch* batch) { - return batch->rep_.size(); - } + static size_t ByteSize(const WriteBatch* batch) { return batch->rep_.size(); } static void SetContents(WriteBatch* batch, const Slice& contents); @@ -46,5 +42,4 @@ class WriteBatchInternal { } // namespace leveldb - #endif // STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_ diff --git a/db/write_batch_test.cc b/db/write_batch_test.cc index 9064e3d..c32317f 100644 --- a/db/write_batch_test.cc +++ b/db/write_batch_test.cc @@ -52,7 +52,7 @@ static std::string PrintContents(WriteBatch* b) { return state; } -class WriteBatchTest { }; +class WriteBatchTest {}; TEST(WriteBatchTest, Empty) { WriteBatch batch; @@ -68,10 +68,11 @@ TEST(WriteBatchTest, Multiple) { WriteBatchInternal::SetSequence(&batch, 100); ASSERT_EQ(100, 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( + "Put(baz, boo)@102" + "Delete(box)@101" + "Put(foo, bar)@100", + PrintContents(&batch)); } TEST(WriteBatchTest, Corruption) { @@ -81,40 +82,56 @@ TEST(WriteBatchTest, Corruption) { WriteBatchInternal::SetSequence(&batch, 200); Slice contents = WriteBatchInternal::Contents(&batch); WriteBatchInternal::SetContents(&batch, - Slice(contents.data(),contents.size()-1)); - ASSERT_EQ("Put(foo, bar)@200" - "ParseError()", - PrintContents(&batch)); + Slice(contents.data(), contents.size() - 1)); + ASSERT_EQ( + "Put(foo, bar)@200" + "ParseError()", + PrintContents(&batch)); } TEST(WriteBatchTest, Append) { WriteBatch b1, b2; WriteBatchInternal::SetSequence(&b1, 200); WriteBatchInternal::SetSequence(&b2, 300); - WriteBatchInternal::Append(&b1, &b2); - ASSERT_EQ("", - PrintContents(&b1)); + b1.Append(b2); + ASSERT_EQ("", PrintContents(&b1)); b2.Put("a", "va"); - WriteBatchInternal::Append(&b1, &b2); - ASSERT_EQ("Put(a, va)@200", - PrintContents(&b1)); + b1.Append(b2); + ASSERT_EQ("Put(a, va)@200", PrintContents(&b1)); b2.Clear(); b2.Put("b", "vb"); - WriteBatchInternal::Append(&b1, &b2); - ASSERT_EQ("Put(a, va)@200" - "Put(b, vb)@201", - PrintContents(&b1)); + b1.Append(b2); + ASSERT_EQ( + "Put(a, va)@200" + "Put(b, vb)@201", + PrintContents(&b1)); 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)); + b1.Append(b2); + ASSERT_EQ( + "Put(a, va)@200" + "Put(b, vb)@202" + "Put(b, vb)@201" + "Delete(foo)@203", + PrintContents(&b1)); +} + +TEST(WriteBatchTest, ApproximateSize) { + WriteBatch batch; + size_t empty_size = batch.ApproximateSize(); + + batch.Put(Slice("foo"), Slice("bar")); + size_t one_key_size = batch.ApproximateSize(); + ASSERT_LT(empty_size, one_key_size); + + batch.Put(Slice("baz"), Slice("boo")); + size_t two_keys_size = batch.ApproximateSize(); + ASSERT_LT(one_key_size, two_keys_size); + + batch.Delete(Slice("box")); + size_t post_delete_size = batch.ApproximateSize(); + ASSERT_LT(two_keys_size, post_delete_size); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/doc/benchmark.html b/doc/benchmark.html index c463977..f3fd771 100644 --- a/doc/benchmark.html +++ b/doc/benchmark.html @@ -90,9 +90,9 @@ div.bsql {

Benchmark Source Code

We wrote benchmark tools for SQLite and Kyoto TreeDB based on LevelDB's db_bench. The code for each of the benchmarks resides here:

Custom Build Specifications

diff --git a/doc/doc.css b/doc/doc.css deleted file mode 100644 index 700c564..0000000 --- a/doc/doc.css +++ /dev/null @@ -1,89 +0,0 @@ -body { - 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; -} diff --git a/doc/impl.html b/doc/impl.html deleted file mode 100644 index 6a468be..0000000 --- a/doc/impl.html +++ /dev/null @@ -1,213 +0,0 @@ - - - - -Leveldb file layout and compactions - - - - -

Files

- -The implementation of leveldb is similar in spirit to the -representation of a single - -Bigtable tablet (section 5.3). -However the organization of the files that make up the representation -is somewhat different and is explained below. - -

-Each database is represented by a set of files stored in a directory. -There are several different types of files as documented below: -

-

Log files

-

-A log file (*.log) stores a sequence of recent updates. Each update -is appended to the current log file. When the log file reaches a -pre-determined size (approximately 4MB by default), it is converted -to a sorted table (see below) and a new log file is created for future -updates. -

-A copy of the current log file is kept in an in-memory structure (the -memtable). This copy is consulted on every read so that read -operations reflect all logged updates. -

-

Sorted tables

-

-A sorted table (*.sst) stores a sequence of entries sorted by key. -Each entry is either a value for the key, or a deletion marker for the -key. (Deletion markers are kept around to hide obsolete values -present in older sorted tables). -

-The set of sorted tables are organized into a sequence of levels. The -sorted table generated from a log file is placed in a special young -level (also called level-0). When the number of young files exceeds a -certain threshold (currently four), all of the young files are merged -together with all of the overlapping level-1 files to produce a -sequence of new level-1 files (we create a new level-1 file for every -2MB of data.) -

-Files in the young level may contain overlapping keys. However files -in other levels have distinct non-overlapping key ranges. Consider -level number L where L >= 1. When the combined size of files in -level-L exceeds (10^L) MB (i.e., 10MB for level-1, 100MB for level-2, -...), one file in level-L, and all of the overlapping files in -level-(L+1) are merged to form a set of new files for level-(L+1). -These merges have the effect of gradually migrating new updates from -the young level to the largest level using only bulk reads and writes -(i.e., minimizing expensive seeks). - -

Manifest

-

-A MANIFEST file lists the set of sorted tables that make up each -level, the corresponding key ranges, and other important metadata. -A new MANIFEST file (with a new number embedded in the file name) -is created whenever the database is reopened. The MANIFEST file is -formatted as a log, and changes made to the serving state (as files -are added or removed) are appended to this log. -

-

Current

-

-CURRENT is a simple text file that contains the name of the latest -MANIFEST file. -

-

Info logs

-

-Informational messages are printed to files named LOG and LOG.old. -

-

Others

-

-Other files used for miscellaneous purposes may also be present -(LOCK, *.dbtmp). - -

Level 0

-When the log file grows above a certain size (1MB by default): -
    -
  • Create a brand new memtable and log file and direct future updates here -
  • In the background: -
      -
    • Write the contents of the previous memtable to an sstable -
    • Discard the memtable -
    • Delete the old log file and the old memtable -
    • Add the new sstable to the young (level-0) level. -
    -
- -

Compactions

- -

-When the size of level L exceeds its limit, we compact it in a -background thread. The compaction picks a file from level L and all -overlapping files from the next level L+1. Note that if a level-L -file overlaps only part of a level-(L+1) file, the entire file at -level-(L+1) is used as an input to the compaction and will be -discarded after the compaction. Aside: because level-0 is special -(files in it may overlap each other), we treat compactions from -level-0 to level-1 specially: a level-0 compaction may pick more than -one level-0 file in case some of these files overlap each other. - -

-A compaction merges the contents of the picked files to produce a -sequence of level-(L+1) files. We switch to producing a new -level-(L+1) file after the current output file has reached the target -file size (2MB). We also switch to a new output file when the key -range of the current output file has grown enough to overlap more than -ten level-(L+2) files. This last rule ensures that a later compaction -of a level-(L+1) file will not pick up too much data from level-(L+2). - -

-The old files are discarded and the new files are added to the serving -state. - -

-Compactions for a particular level rotate through the key space. In -more detail, for each level L, we remember the ending key of the last -compaction at level L. The next compaction for level L will pick the -first file that starts after this key (wrapping around to the -beginning of the key space if there is no such file). - -

-Compactions drop overwritten values. They also drop deletion markers -if there are no higher numbered levels that contain a file whose range -overlaps the current key. - -

Timing

- -Level-0 compactions will read up to four 1MB files from level-0, and -at worst all the level-1 files (10MB). I.e., we will read 14MB and -write 14MB. - -

-Other than the special level-0 compactions, we will pick one 2MB file -from level L. In the worst case, this will overlap ~ 12 files from -level L+1 (10 because level-(L+1) is ten times the size of level-L, -and another two at the boundaries since the file ranges at level-L -will usually not be aligned with the file ranges at level-L+1). The -compaction will therefore read 26MB and write 26MB. Assuming a disk -IO rate of 100MB/s (ballpark range for modern drives), the worst -compaction cost will be approximately 0.5 second. - -

-If we throttle the background writing to something small, say 10% of -the full 100MB/s speed, a compaction may take up to 5 seconds. If the -user is writing at 10MB/s, we might build up lots of level-0 files -(~50 to hold the 5*10MB). This may significantly increase the cost of -reads due to the overhead of merging more files together on every -read. - -

-Solution 1: To reduce this problem, we might want to increase the log -switching threshold when the number of level-0 files is large. Though -the downside is that the larger this threshold, the more memory we will -need to hold the corresponding memtable. - -

-Solution 2: We might want to decrease write rate artificially when the -number of level-0 files goes up. - -

-Solution 3: We work on reducing the cost of very wide merges. -Perhaps most of the level-0 files will have their blocks sitting -uncompressed in the cache and we will only need to worry about the -O(N) complexity in the merging iterator. - -

Number of files

- -Instead of always making 2MB files, we could make larger files for -larger levels to reduce the total file count, though at the expense of -more bursty compactions. Alternatively, we could shard the set of -files into multiple directories. - -

-An experiment on an ext3 filesystem on Feb 04, 2011 shows -the following timings to do 100K file opens in directories with -varying number of files: - - - - - -
Files in directoryMicroseconds to open a file
10009
1000010
10000016
-So maybe even the sharding is not necessary on modern filesystems? - -

Recovery

- -
    -
  • Read CURRENT to find name of the latest committed MANIFEST -
  • Read the named MANIFEST file -
  • Clean up stale files -
  • We could open all sstables here, but it is probably better to be lazy... -
  • Convert log chunk to a new level-0 sstable -
  • Start directing new writes to a new log file with recovered sequence# -
- -

Garbage collection of files

- -DeleteObsoleteFiles() is called at the end of every -compaction and at the end of recovery. It finds the names of all -files in the database. It deletes all log files that are not the -current log file. It deletes all table files that are not referenced -from some level and are not the output of an active compaction. - - - diff --git a/doc/impl.md b/doc/impl.md new file mode 100644 index 0000000..cacabb9 --- /dev/null +++ b/doc/impl.md @@ -0,0 +1,172 @@ +## Files + +The implementation of leveldb is similar in spirit to the representation of a +single [Bigtable tablet (section 5.3)](http://research.google.com/archive/bigtable.html). +However the organization of the files that make up the representation is +somewhat different and is explained below. + +Each database is represented by a set of files stored in a directory. There are +several different types of files as documented below: + +### Log files + +A log file (*.log) stores a sequence of recent updates. Each update is appended +to the current log file. When the log file reaches a pre-determined size +(approximately 4MB by default), it is converted to a sorted table (see below) +and a new log file is created for future updates. + +A copy of the current log file is kept in an in-memory structure (the +`memtable`). This copy is consulted on every read so that read operations +reflect all logged updates. + +## Sorted tables + +A sorted table (*.ldb) stores a sequence of entries sorted by key. Each entry is +either a value for the key, or a deletion marker for the key. (Deletion markers +are kept around to hide obsolete values present in older sorted tables). + +The set of sorted tables are organized into a sequence of levels. The sorted +table generated from a log file is placed in a special **young** level (also +called level-0). When the number of young files exceeds a certain threshold +(currently four), all of the young files are merged together with all of the +overlapping level-1 files to produce a sequence of new level-1 files (we create +a new level-1 file for every 2MB of data.) + +Files in the young level may contain overlapping keys. However files in other +levels have distinct non-overlapping key ranges. Consider level number L where +L >= 1. When the combined size of files in level-L exceeds (10^L) MB (i.e., 10MB +for level-1, 100MB for level-2, ...), one file in level-L, and all of the +overlapping files in level-(L+1) are merged to form a set of new files for +level-(L+1). These merges have the effect of gradually migrating new updates +from the young level to the largest level using only bulk reads and writes +(i.e., minimizing expensive seeks). + +### Manifest + +A MANIFEST file lists the set of sorted tables that make up each level, the +corresponding key ranges, and other important metadata. A new MANIFEST file +(with a new number embedded in the file name) is created whenever the database +is reopened. The MANIFEST file is formatted as a log, and changes made to the +serving state (as files are added or removed) are appended to this log. + +### Current + +CURRENT is a simple text file that contains the name of the latest MANIFEST +file. + +### Info logs + +Informational messages are printed to files named LOG and LOG.old. + +### Others + +Other files used for miscellaneous purposes may also be present (LOCK, *.dbtmp). + +## Level 0 + +When the log file grows above a certain size (4MB by default): +Create a brand new memtable and log file and direct future updates here. + +In the background: + +1. Write the contents of the previous memtable to an sstable. +2. Discard the memtable. +3. Delete the old log file and the old memtable. +4. Add the new sstable to the young (level-0) level. + +## Compactions + +When the size of level L exceeds its limit, we compact it in a background +thread. The compaction picks a file from level L and all overlapping files from +the next level L+1. Note that if a level-L file overlaps only part of a +level-(L+1) file, the entire file at level-(L+1) is used as an input to the +compaction and will be discarded after the compaction. Aside: because level-0 +is special (files in it may overlap each other), we treat compactions from +level-0 to level-1 specially: a level-0 compaction may pick more than one +level-0 file in case some of these files overlap each other. + +A compaction merges the contents of the picked files to produce a sequence of +level-(L+1) files. We switch to producing a new level-(L+1) file after the +current output file has reached the target file size (2MB). We also switch to a +new output file when the key range of the current output file has grown enough +to overlap more than ten level-(L+2) files. This last rule ensures that a later +compaction of a level-(L+1) file will not pick up too much data from +level-(L+2). + +The old files are discarded and the new files are added to the serving state. + +Compactions for a particular level rotate through the key space. In more detail, +for each level L, we remember the ending key of the last compaction at level L. +The next compaction for level L will pick the first file that starts after this +key (wrapping around to the beginning of the key space if there is no such +file). + +Compactions drop overwritten values. They also drop deletion markers if there +are no higher numbered levels that contain a file whose range overlaps the +current key. + +### Timing + +Level-0 compactions will read up to four 1MB files from level-0, and at worst +all the level-1 files (10MB). I.e., we will read 14MB and write 14MB. + +Other than the special level-0 compactions, we will pick one 2MB file from level +L. In the worst case, this will overlap ~ 12 files from level L+1 (10 because +level-(L+1) is ten times the size of level-L, and another two at the boundaries +since the file ranges at level-L will usually not be aligned with the file +ranges at level-L+1). The compaction will therefore read 26MB and write 26MB. +Assuming a disk IO rate of 100MB/s (ballpark range for modern drives), the worst +compaction cost will be approximately 0.5 second. + +If we throttle the background writing to something small, say 10% of the full +100MB/s speed, a compaction may take up to 5 seconds. If the user is writing at +10MB/s, we might build up lots of level-0 files (~50 to hold the 5*10MB). This +may significantly increase the cost of reads due to the overhead of merging more +files together on every read. + +Solution 1: To reduce this problem, we might want to increase the log switching +threshold when the number of level-0 files is large. Though the downside is that +the larger this threshold, the more memory we will need to hold the +corresponding memtable. + +Solution 2: We might want to decrease write rate artificially when the number of +level-0 files goes up. + +Solution 3: We work on reducing the cost of very wide merges. Perhaps most of +the level-0 files will have their blocks sitting uncompressed in the cache and +we will only need to worry about the O(N) complexity in the merging iterator. + +### Number of files + +Instead of always making 2MB files, we could make larger files for larger levels +to reduce the total file count, though at the expense of more bursty +compactions. Alternatively, we could shard the set of files into multiple +directories. + +An experiment on an ext3 filesystem on Feb 04, 2011 shows the following timings +to do 100K file opens in directories with varying number of files: + + +| Files in directory | Microseconds to open a file | +|-------------------:|----------------------------:| +| 1000 | 9 | +| 10000 | 10 | +| 100000 | 16 | + +So maybe even the sharding is not necessary on modern filesystems? + +## Recovery + +* Read CURRENT to find name of the latest committed MANIFEST +* Read the named MANIFEST file +* Clean up stale files +* We could open all sstables here, but it is probably better to be lazy... +* Convert log chunk to a new level-0 sstable +* Start directing new writes to a new log file with recovered sequence# + +## Garbage collection of files + +`DeleteObsoleteFiles()` is called at the end of every compaction and at the end +of recovery. It finds the names of all files in the database. It deletes all log +files that are not the current log file. It deletes all table files that are not +referenced from some level and are not the output of an active compaction. diff --git a/doc/index.html b/doc/index.html deleted file mode 100644 index 3ed0ed9..0000000 --- a/doc/index.html +++ /dev/null @@ -1,549 +0,0 @@ - - - - -Leveldb - - - -

Leveldb

-
Jeff Dean, Sanjay Ghemawat
-

-The leveldb 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 leveldb 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 "leveldb/db.h"
-
-  leveldb::DB* db;
-  leveldb::Options options;
-  options.create_if_missing = true;
-  leveldb::Status status = leveldb::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 leveldb::DB::Open call: -
-  options.error_if_exists = true;
-
-

Status

-

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

-

-   leveldb::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;
-  leveldb::Status s = db->Get(leveldb::ReadOptions(), key1, &value);
-  if (s.ok()) s = db->Put(leveldb::WriteOptions(), key2, value);
-  if (s.ok()) s = db->Delete(leveldb::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 "leveldb/write_batch.h"
-  ...
-  std::string value;
-  leveldb::Status s = db->Get(leveldb::ReadOptions(), key1, &value);
-  if (s.ok()) {
-    leveldb::WriteBatch batch;
-    batch.Delete(key1);
-    batch.Put(key2, value);
-    s = db->Write(leveldb::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.) -
-  leveldb::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. - -

-

Concurrency

-

-A database may only be opened by one process at a time. -The leveldb implementation acquires a lock from the -operating system to prevent misuse. Within a single process, the -same leveldb::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. -

-

Iteration

-

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

-

-  leveldb::Iterator* it = db->NewIterator(leveldb::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: -

-

-  leveldb::ReadOptions options;
-  options.snapshot = db->GetSnapshot();
-  ... apply some updates to db ...
-  leveldb::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 leveldb::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, leveldb methods do not return null-terminated -C-style strings since leveldb keys and values are allowed to -contain '\0' bytes. -

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

-

-   leveldb::Slice s1 = "hello";
-
-   std::string str("world");
-   leveldb::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: -

-

-   leveldb::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 leveldb::Comparator that expresses these rules: -

-

-  class TwoPartComparator : public leveldb::Comparator {
-   public:
-    // Three-way comparison function:
-    //   if a < b: negative result
-    //   if a > b: positive result
-    //   else: zero result
-    int Compare(const leveldb::Slice& a, const leveldb::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 leveldb::Slice&) const { }
-    void FindShortSuccessor(std::string*) const { }
-  };
-
-Now create a database using this custom comparator: -

-

-  TwoPartComparator cmp;
-  leveldb::DB* db;
-  leveldb::Options options;
-  options.create_if_missing = true;
-  options.comparator = &cmp;
-  leveldb::Status status = leveldb::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 leveldb::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. -

-

Performance

-

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

-

Block size

-

-leveldb 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. -

-

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: -

-

-  leveldb::Options options;
-  options.compression = leveldb::kNoCompression;
-  ... leveldb::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.cache is non-NULL, it is used to cache frequently used -uncompressed block contents. -

-

-  #include "leveldb/cache.h"
-
-  leveldb::Options options;
-  options.cache = leveldb::NewLRUCache(100 * 1048576);  // 100MB cache
-  leveldb::DB* db;
-  leveldb::DB::Open(options, name, &db);
-  ... use the db ...
-  delete db
-  delete options.cache;
-
-Note that the cache holds uncompressed data, and therefore it should -be sized according to application level data sizes, without any -reduction from compression. (Caching of compressed blocks is left to -the operating system buffer cache, or any custom Env -implementation provided by the client.) -

-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: -

-

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

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 leveldb. 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 leveldb 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. -

-   leveldb::Options options;
-   options.filter_policy = NewBloomFilterPolicy(10);
-   leveldb::DB* db;
-   leveldb::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 NewBloomFilterPolicy). 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. -NewBloomFilterPolicy 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 leveldb::FilterPolicy {
-   private:
-    FilterPolicy* builtin_policy_;
-   public:
-    CustomFilterPolicy() : builtin_policy_(NewBloomFilterPolicy(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 leveldb/filter_policy.h for detail. -

-

Checksums

-

-leveldb 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 leveldb::RepairDB function - may be used to recover as much of the data as possible -

    -

-

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. -

-

-   leveldb::Range ranges[2];
-   ranges[0] = leveldb::Range("a", "c");
-   ranges[1] = leveldb::Range("x", "z");
-   uint64_t sizes[2];
-   leveldb::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 -leveldb implementation are routed through a leveldb::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 leveldb on other activities in the system. -

-

-  class SlowEnv : public leveldb::Env {
-    .. implementation of the Env interface ...
-  };
-
-  SlowEnv env;
-  leveldb::Options options;
-  options.env = &env;
-  Status s = leveldb::DB::Open(options, ...);
-
-

Porting

-

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

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

Other Information

- -

-Details about the leveldb implementation may be found in -the following documents: -

- - - diff --git a/doc/index.md b/doc/index.md new file mode 100644 index 0000000..ea4609d --- /dev/null +++ b/doc/index.md @@ -0,0 +1,523 @@ +leveldb +======= + +_Jeff Dean, Sanjay Ghemawat_ + +The leveldb 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 leveldb 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: + +```c++ +#include +#include "leveldb/db.h" + +leveldb::DB* db; +leveldb::Options options; +options.create_if_missing = true; +leveldb::Status status = leveldb::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 `leveldb::DB::Open` call: + +```c++ +options.error_if_exists = true; +``` + +## Status + +You may have noticed the `leveldb::Status` type above. Values of this type are +returned by most functions in leveldb that may encounter an error. You can check +if such a result is ok, and also print an associated error message: + +```c++ +leveldb::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: + +```c++ +... 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. + +```c++ +std::string value; +leveldb::Status s = db->Get(leveldb::ReadOptions(), key1, &value); +if (s.ok()) s = db->Put(leveldb::WriteOptions(), key2, value); +if (s.ok()) s = db->Delete(leveldb::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: + +```c++ +#include "leveldb/write_batch.h" +... +std::string value; +leveldb::Status s = db->Get(leveldb::ReadOptions(), key1, &value); +if (s.ok()) { + leveldb::WriteBatch batch; + batch.Delete(key1); + batch.Put(key2, value); + s = db->Write(leveldb::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.) + +```c++ +leveldb::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. + +## Concurrency + +A database may only be opened by one process at a time. The leveldb +implementation acquires a lock from the operating system to prevent misuse. +Within a single process, the same `leveldb::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. + +## Iteration + +The following example demonstrates how to print all key,value pairs in a +database. + +```c++ +leveldb::Iterator* it = db->NewIterator(leveldb::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): + +```c++ +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.) + +```c++ +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: + +```c++ +leveldb::ReadOptions options; +options.snapshot = db->GetSnapshot(); +... apply some updates to db ... +leveldb::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 `leveldb::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, leveldb methods do not return +null-terminated C-style strings since leveldb keys and values are allowed to +contain `'\0'` bytes. + +C++ strings and null-terminated C-style strings can be easily converted to a +Slice: + +```c++ +leveldb::Slice s1 = "hello"; + +std::string str("world"); +leveldb::Slice s2 = str; +``` + +A Slice can be easily converted back to a C++ string: + +```c++ +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: + +```c++ +leveldb::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 `leveldb::Comparator` that expresses these rules: + +```c++ +class TwoPartComparator : public leveldb::Comparator { + public: + // Three-way comparison function: + // if a < b: negative result + // if a > b: positive result + // else: zero result + int Compare(const leveldb::Slice& a, const leveldb::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 leveldb::Slice&) const {} + void FindShortSuccessor(std::string*) const {} +}; +``` + +Now create a database using this custom comparator: + +```c++ +TwoPartComparator cmp; +leveldb::DB* db; +leveldb::Options options; +options.create_if_missing = true; +options.comparator = &cmp; +leveldb::Status status = leveldb::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 `leveldb::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. + +## Performance + +Performance can be tuned by changing the default values of the types defined in +`include/leveldb/options.h`. + +### Block size + +leveldb 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. + +### 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: + +```c++ +leveldb::Options options; +options.compression = leveldb::kNoCompression; +... leveldb::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. + +```c++ +#include "leveldb/cache.h" + +leveldb::Options options; +options.block_cache = leveldb::NewLRUCache(100 * 1048576); // 100MB cache +leveldb::DB* db; +leveldb::DB::Open(options, name, &db); +... use the db ... +delete db +delete options.block_cache; +``` + +Note that the cache holds uncompressed data, and therefore it should be sized +according to application level data sizes, without any reduction from +compression. (Caching of compressed blocks is left to the operating system +buffer cache, or any custom Env implementation provided by the client.) + +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: + +```c++ +leveldb::ReadOptions options; +options.fill_cache = false; +leveldb::Iterator* it = db->NewIterator(options); +for (it->SeekToFirst(); it->Valid(); it->Next()) { + ... +} +``` + +### 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 leveldb. +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 leveldb 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. + +```c++ +leveldb::Options options; +options.filter_policy = NewBloomFilterPolicy(10); +leveldb::DB* db; +leveldb::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 `NewBloomFilterPolicy`). 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. +`NewBloomFilterPolicy` must not be used with such a comparator. Instead, the +application should provide a custom filter policy that also ignores trailing +spaces. For example: + +```c++ +class CustomFilterPolicy : public leveldb::FilterPolicy { + private: + FilterPolicy* builtin_policy_; + + public: + CustomFilterPolicy() : builtin_policy_(NewBloomFilterPolicy(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 trimmed(n); + for (int i = 0; i < n; i++) { + trimmed[i] = RemoveTrailingSpaces(keys[i]); + } + return builtin_policy_->CreateFilter(&trimmed[i], n, dst); + } +}; +``` + +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 +`leveldb/filter_policy.h` for detail. + +## Checksums + +leveldb 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 `leveldb::RepairDB` function may be used to recover as much +of the data as possible + +## 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. + +```c++ +leveldb::Range ranges[2]; +ranges[0] = leveldb::Range("a", "c"); +ranges[1] = leveldb::Range("x", "z"); +uint64_t sizes[2]; +leveldb::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 leveldb +implementation are routed through a `leveldb::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 leveldb on other activities in the system. + +```c++ +class SlowEnv : public leveldb::Env { + ... implementation of the Env interface ... +}; + +SlowEnv env; +leveldb::Options options; +options.env = &env; +Status s = leveldb::DB::Open(options, ...); +``` + +## Porting + +leveldb may be ported to a new platform by providing platform specific +implementations of the types/methods/functions exported by +`leveldb/port/port.h`. See `leveldb/port/port_example.h` for more details. + +In addition, the new platform may need a new default `leveldb::Env` +implementation. See `leveldb/util/env_posix.h` for an example. + +## Other Information + +Details about the leveldb implementation may be found in the following +documents: + +1. [Implementation notes](impl.md) +2. [Format of an immutable Table file](table_format.md) +3. [Format of a log file](log_format.md) diff --git a/doc/log_format.md b/doc/log_format.md new file mode 100644 index 0000000..f32cb5d --- /dev/null +++ b/doc/log_format.md @@ -0,0 +1,75 @@ +leveldb Log format +================== +The 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[] ; little-endian + length: uint16 // little-endian + 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 MIDDLE 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. diff --git a/doc/log_format.txt b/doc/log_format.txt deleted file mode 100644 index 4cca5ef..0000000 --- a/doc/log_format.txt +++ /dev/null @@ -1,75 +0,0 @@ -The 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[] ; little-endian - length: uint16 // little-endian - 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 MIDDLE 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. diff --git a/doc/table_format.md b/doc/table_format.md new file mode 100644 index 0000000..5fe7e72 --- /dev/null +++ b/doc/table_format.md @@ -0,0 +1,107 @@ +leveldb File format +=================== + + + [data block 1] + [data block 2] + ... + [data block N] + [meta block 1] + ... + [meta block K] + [metaindex block] + [index block] + [Footer] (fixed size; starts at file_size - sizeof(Footer)) + + +The file contains internal pointers. Each such pointer is called +a BlockHandle and contains the following information: + + offset: varint64 + size: varint64 + +See [varints](https://developers.google.com/protocol-buffers/docs/encoding#varints) +for an explanation of varint64 format. + +1. The sequence of key/value pairs in the file are stored in sorted +order and partitioned into a sequence of data blocks. These blocks +come one after another at the beginning of the file. Each data block +is formatted according to the code in `block_builder.cc`, and then +optionally compressed. + +2. After the data blocks we store a bunch of meta blocks. The +supported meta block types are described below. More meta block types +may be added in the future. Each meta block is again formatted using +`block_builder.cc` and then optionally compressed. + +3. A "metaindex" block. It contains one entry for every other meta +block where the key is the name of the meta block and the value is a +BlockHandle pointing to that meta block. + +4. An "index" block. This block contains one entry per data block, +where the key is a string >= last key in that data block and before +the first key in the successive data block. The value is the +BlockHandle for the data block. + +5. At the very end of the file is a fixed length footer that contains +the BlockHandle of the metaindex and index blocks as well as a magic number. + + metaindex_handle: char[p]; // Block handle for metaindex + index_handle: char[q]; // Block handle for index + padding: char[40-p-q];// zeroed bytes to make fixed length + // (40==2*BlockHandle::kMaxEncodedLength) + magic: fixed64; // == 0xdb4775248b80fb57 (little-endian) + +## "filter" Meta Block + +If a `FilterPolicy` was specified when the database was opened, a +filter block is stored in each table. The "metaindex" block contains +an entry that maps from `filter.` to the BlockHandle for the filter +block where `` is the string returned by the filter policy's +`Name()` method. + +The filter block stores a sequence of filters, where filter i contains +the output of `FilterPolicy::CreateFilter()` on all keys that are stored +in a block whose file offset falls within the range + + [ i*base ... (i+1)*base-1 ] + +Currently, "base" is 2KB. So for example, if blocks X and Y start in +the range `[ 0KB .. 2KB-1 ]`, all of the keys in X and Y will be +converted to a filter by calling `FilterPolicy::CreateFilter()`, and the +resulting filter will be stored as the first filter in the filter +block. + +The filter block is formatted as follows: + + [filter 0] + [filter 1] + [filter 2] + ... + [filter N-1] + + [offset of filter 0] : 4 bytes + [offset of filter 1] : 4 bytes + [offset of filter 2] : 4 bytes + ... + [offset of filter N-1] : 4 bytes + + [offset of beginning of offset array] : 4 bytes + lg(base) : 1 byte + +The offset array at the end of the filter block allows efficient +mapping from a data block offset to the corresponding filter. + +## "stats" Meta Block + +This meta block contains a bunch of stats. The key is the name +of the statistic. The value contains the statistic. + +TODO(postrelease): record following stats. + + data size + index size + key size (uncompressed) + value size (uncompressed) + number of entries + number of data blocks diff --git a/doc/table_format.txt b/doc/table_format.txt deleted file mode 100644 index ca8f9b4..0000000 --- a/doc/table_format.txt +++ /dev/null @@ -1,104 +0,0 @@ -File format -=========== - - - [data block 1] - [data block 2] - ... - [data block N] - [meta block 1] - ... - [meta block K] - [metaindex block] - [index block] - [Footer] (fixed size; starts at file_size - sizeof(Footer)) - - -The file contains internal pointers. Each such pointer is called -a BlockHandle and contains the following information: - offset: varint64 - size: varint64 -See https://developers.google.com/protocol-buffers/docs/encoding#varints -for an explanation of varint64 format. - -(1) The sequence of key/value pairs in the file are stored in sorted -order and partitioned into a sequence of data blocks. These blocks -come one after another at the beginning of the file. Each data block -is formatted according to the code in block_builder.cc, and then -optionally compressed. - -(2) After the data blocks we store a bunch of meta blocks. The -supported meta block types are described below. More meta block types -may be added in the future. Each meta block is again formatted using -block_builder.cc and then optionally compressed. - -(3) A "metaindex" block. It contains one entry for every other meta -block where the key is the name of the meta block and the value is a -BlockHandle pointing to that meta block. - -(4) An "index" block. This block contains one entry per data block, -where the key is a string >= last key in that data block and before -the first key in the successive data block. The value is the -BlockHandle for the data block. - -(6) At the very end of the file is a fixed length footer that contains -the BlockHandle of the metaindex and index blocks as well as a magic number. - metaindex_handle: char[p]; // Block handle for metaindex - index_handle: char[q]; // Block handle for index - padding: char[40-p-q]; // zeroed bytes to make fixed length - // (40==2*BlockHandle::kMaxEncodedLength) - magic: fixed64; // == 0xdb4775248b80fb57 (little-endian) - -"filter" Meta Block -------------------- - -If a "FilterPolicy" was specified when the database was opened, a -filter block is stored in each table. The "metaindex" block contains -an entry that maps from "filter." to the BlockHandle for the filter -block where "" is the string returned by the filter policy's -"Name()" method. - -The filter block stores a sequence of filters, where filter i contains -the output of FilterPolicy::CreateFilter() on all keys that are stored -in a block whose file offset falls within the range - - [ i*base ... (i+1)*base-1 ] - -Currently, "base" is 2KB. So for example, if blocks X and Y start in -the range [ 0KB .. 2KB-1 ], all of the keys in X and Y will be -converted to a filter by calling FilterPolicy::CreateFilter(), and the -resulting filter will be stored as the first filter in the filter -block. - -The filter block is formatted as follows: - - [filter 0] - [filter 1] - [filter 2] - ... - [filter N-1] - - [offset of filter 0] : 4 bytes - [offset of filter 1] : 4 bytes - [offset of filter 2] : 4 bytes - ... - [offset of filter N-1] : 4 bytes - - [offset of beginning of offset array] : 4 bytes - lg(base) : 1 byte - -The offset array at the end of the filter block allows efficient -mapping from a data block offset to the corresponding filter. - -"stats" Meta Block ------------------- - -This meta block contains a bunch of stats. The key is the name -of the statistic. The value contains the statistic. -TODO(postrelease): record following stats. - data size - index size - key size (uncompressed) - value size (uncompressed) - number of entries - number of data blocks diff --git a/helpers/memenv/memenv.cc b/helpers/memenv/memenv.cc index 43ef2e0..31d2bc0 100644 --- a/helpers/memenv/memenv.cc +++ b/helpers/memenv/memenv.cc @@ -4,14 +4,18 @@ #include "helpers/memenv/memenv.h" +#include + +#include +#include +#include +#include + #include "leveldb/env.h" #include "leveldb/status.h" #include "port/port.h" +#include "port/thread_annotations.h" #include "util/mutexlock.h" -#include -#include -#include -#include namespace leveldb { @@ -23,6 +27,10 @@ class FileState { // and the caller must call Ref() at least once. FileState() : refs_(0), size_(0) {} + // No copying allowed. + FileState(const FileState&) = delete; + FileState& operator=(const FileState&) = delete; + // Increase the reference count. void Ref() { MutexLock lock(&refs_mutex_); @@ -47,9 +55,22 @@ class FileState { } } - uint64_t Size() const { return size_; } + uint64_t Size() const { + MutexLock lock(&blocks_mutex_); + return size_; + } + + void Truncate() { + MutexLock lock(&blocks_mutex_); + for (char*& block : blocks_) { + delete[] block; + } + blocks_.clear(); + size_ = 0; + } Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const { + MutexLock lock(&blocks_mutex_); if (offset > size_) { return Status::IOError("Offset greater than file size."); } @@ -62,16 +83,9 @@ class FileState { return Status::OK(); } - assert(offset / kBlockSize <= SIZE_MAX); + assert(offset / kBlockSize <= std::numeric_limits::max()); size_t block = static_cast(offset / kBlockSize); size_t block_offset = offset % kBlockSize; - - if (n <= kBlockSize - block_offset) { - // The requested bytes are all in the first block. - *result = Slice(blocks_[block] + block_offset, n); - return Status::OK(); - } - size_t bytes_to_copy = n; char* dst = scratch; @@ -96,6 +110,7 @@ class FileState { const char* src = data.data(); size_t src_len = data.size(); + MutexLock lock(&blocks_mutex_); while (src_len > 0) { size_t avail; size_t offset = size_ % kBlockSize; @@ -122,28 +137,17 @@ class FileState { } private: - // Private since only Unref() should be used to delete it. - ~FileState() { - for (std::vector::iterator i = blocks_.begin(); i != blocks_.end(); - ++i) { - delete [] *i; - } - } + enum { kBlockSize = 8 * 1024 }; - // No copying allowed. - FileState(const FileState&); - void operator=(const FileState&); + // Private since only Unref() should be used to delete it. + ~FileState() { Truncate(); } port::Mutex refs_mutex_; - int refs_; // Protected by refs_mutex_; + int refs_ GUARDED_BY(refs_mutex_); - // The following fields are not protected by any mutex. They are only mutable - // while the file is being written, and concurrent access is not allowed - // to writable files. - std::vector blocks_; - uint64_t size_; - - enum { kBlockSize = 8 * 1024 }; + mutable port::Mutex blocks_mutex_; + std::vector blocks_ GUARDED_BY(blocks_mutex_); + uint64_t size_ GUARDED_BY(blocks_mutex_); }; class SequentialFileImpl : public SequentialFile { @@ -152,11 +156,9 @@ class SequentialFileImpl : public SequentialFile { file_->Ref(); } - ~SequentialFileImpl() { - file_->Unref(); - } + ~SequentialFileImpl() override { file_->Unref(); } - virtual Status Read(size_t n, Slice* result, char* scratch) { + Status Read(size_t n, Slice* result, char* scratch) override { Status s = file_->Read(pos_, n, result, scratch); if (s.ok()) { pos_ += result->size(); @@ -164,7 +166,7 @@ class SequentialFileImpl : public SequentialFile { return s; } - virtual Status Skip(uint64_t n) { + Status Skip(uint64_t n) override { if (pos_ > file_->Size()) { return Status::IOError("pos_ > file_->Size()"); } @@ -183,16 +185,12 @@ class SequentialFileImpl : public SequentialFile { class RandomAccessFileImpl : public RandomAccessFile { public: - explicit RandomAccessFileImpl(FileState* file) : file_(file) { - file_->Ref(); - } + explicit RandomAccessFileImpl(FileState* file) : file_(file) { file_->Ref(); } - ~RandomAccessFileImpl() { - file_->Unref(); - } + ~RandomAccessFileImpl() override { file_->Unref(); } - virtual Status Read(uint64_t offset, size_t n, Slice* result, - char* scratch) const { + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { return file_->Read(offset, n, result, scratch); } @@ -202,21 +200,15 @@ class RandomAccessFileImpl : public RandomAccessFile { class WritableFileImpl : public WritableFile { public: - WritableFileImpl(FileState* file) : file_(file) { - file_->Ref(); - } + WritableFileImpl(FileState* file) : file_(file) { file_->Ref(); } - ~WritableFileImpl() { - file_->Unref(); - } + ~WritableFileImpl() override { file_->Unref(); } - virtual Status Append(const Slice& data) { - return file_->Append(data); - } + Status Append(const Slice& data) override { return file_->Append(data); } - virtual Status Close() { return Status::OK(); } - virtual Status Flush() { return Status::OK(); } - virtual Status Sync() { return Status::OK(); } + Status Close() override { return Status::OK(); } + Status Flush() override { return Status::OK(); } + Status Sync() override { return Status::OK(); } private: FileState* file_; @@ -224,25 +216,25 @@ class WritableFileImpl : public WritableFile { class NoOpLogger : public Logger { public: - virtual void Logv(const char* format, va_list ap) { } + void Logv(const char* format, va_list ap) override {} }; class InMemoryEnv : public EnvWrapper { public: - explicit InMemoryEnv(Env* base_env) : EnvWrapper(base_env) { } + explicit InMemoryEnv(Env* base_env) : EnvWrapper(base_env) {} - virtual ~InMemoryEnv() { - for (FileSystem::iterator i = file_map_.begin(); i != file_map_.end(); ++i){ - i->second->Unref(); + ~InMemoryEnv() override { + for (const auto& kvp : file_map_) { + kvp.second->Unref(); } } // Partial implementation of the Env interface. - virtual Status NewSequentialFile(const std::string& fname, - SequentialFile** result) { + Status NewSequentialFile(const std::string& fname, + SequentialFile** result) override { MutexLock lock(&mutex_); if (file_map_.find(fname) == file_map_.end()) { - *result = NULL; + *result = nullptr; return Status::IOError(fname, "File not found"); } @@ -250,11 +242,11 @@ class InMemoryEnv : public EnvWrapper { return Status::OK(); } - virtual Status NewRandomAccessFile(const std::string& fname, - RandomAccessFile** result) { + Status NewRandomAccessFile(const std::string& fname, + RandomAccessFile** result) override { MutexLock lock(&mutex_); if (file_map_.find(fname) == file_map_.end()) { - *result = NULL; + *result = nullptr; return Status::IOError(fname, "File not found"); } @@ -262,33 +254,51 @@ class InMemoryEnv : public EnvWrapper { return Status::OK(); } - virtual Status NewWritableFile(const std::string& fname, - WritableFile** result) { + Status NewWritableFile(const std::string& fname, + WritableFile** result) override { MutexLock lock(&mutex_); - if (file_map_.find(fname) != file_map_.end()) { - DeleteFileInternal(fname); - } + FileSystem::iterator it = file_map_.find(fname); - FileState* file = new FileState(); - file->Ref(); - file_map_[fname] = file; + FileState* file; + if (it == file_map_.end()) { + // File is not currently open. + file = new FileState(); + file->Ref(); + file_map_[fname] = file; + } else { + file = it->second; + file->Truncate(); + } *result = new WritableFileImpl(file); return Status::OK(); } - virtual bool FileExists(const std::string& fname) { + Status NewAppendableFile(const std::string& fname, + WritableFile** result) override { + MutexLock lock(&mutex_); + FileState** sptr = &file_map_[fname]; + FileState* file = *sptr; + if (file == nullptr) { + file = new FileState(); + file->Ref(); + } + *result = new WritableFileImpl(file); + return Status::OK(); + } + + bool FileExists(const std::string& fname) override { MutexLock lock(&mutex_); return file_map_.find(fname) != file_map_.end(); } - virtual Status GetChildren(const std::string& dir, - std::vector* result) { + Status GetChildren(const std::string& dir, + std::vector* result) override { MutexLock lock(&mutex_); result->clear(); - for (FileSystem::iterator i = file_map_.begin(); i != file_map_.end(); ++i){ - const std::string& filename = i->first; + for (const auto& kvp : file_map_) { + const std::string& filename = kvp.first; if (filename.size() >= dir.size() + 1 && filename[dir.size()] == '/' && Slice(filename).starts_with(Slice(dir))) { @@ -299,7 +309,8 @@ class InMemoryEnv : public EnvWrapper { return Status::OK(); } - void DeleteFileInternal(const std::string& fname) { + void DeleteFileInternal(const std::string& fname) + EXCLUSIVE_LOCKS_REQUIRED(mutex_) { if (file_map_.find(fname) == file_map_.end()) { return; } @@ -308,7 +319,7 @@ class InMemoryEnv : public EnvWrapper { file_map_.erase(fname); } - virtual Status DeleteFile(const std::string& fname) { + Status DeleteFile(const std::string& fname) override { MutexLock lock(&mutex_); if (file_map_.find(fname) == file_map_.end()) { return Status::IOError(fname, "File not found"); @@ -318,15 +329,11 @@ class InMemoryEnv : public EnvWrapper { return Status::OK(); } - virtual Status CreateDir(const std::string& dirname) { - return Status::OK(); - } + Status CreateDir(const std::string& dirname) override { return Status::OK(); } - virtual Status DeleteDir(const std::string& dirname) { - return Status::OK(); - } + Status DeleteDir(const std::string& dirname) override { return Status::OK(); } - virtual Status GetFileSize(const std::string& fname, uint64_t* file_size) { + Status GetFileSize(const std::string& fname, uint64_t* file_size) override { MutexLock lock(&mutex_); if (file_map_.find(fname) == file_map_.end()) { return Status::IOError(fname, "File not found"); @@ -336,8 +343,8 @@ class InMemoryEnv : public EnvWrapper { return Status::OK(); } - virtual Status RenameFile(const std::string& src, - const std::string& target) { + Status RenameFile(const std::string& src, + const std::string& target) override { MutexLock lock(&mutex_); if (file_map_.find(src) == file_map_.end()) { return Status::IOError(src, "File not found"); @@ -349,22 +356,22 @@ class InMemoryEnv : public EnvWrapper { return Status::OK(); } - virtual Status LockFile(const std::string& fname, FileLock** lock) { + Status LockFile(const std::string& fname, FileLock** lock) override { *lock = new FileLock; return Status::OK(); } - virtual Status UnlockFile(FileLock* lock) { + Status UnlockFile(FileLock* lock) override { delete lock; return Status::OK(); } - virtual Status GetTestDirectory(std::string* path) { + Status GetTestDirectory(std::string* path) override { *path = "/test"; return Status::OK(); } - virtual Status NewLogger(const std::string& fname, Logger** result) { + Status NewLogger(const std::string& fname, Logger** result) override { *result = new NoOpLogger; return Status::OK(); } @@ -372,14 +379,13 @@ class InMemoryEnv : public EnvWrapper { private: // Map from filenames to FileState objects, representing a simple file system. typedef std::map FileSystem; + port::Mutex mutex_; - FileSystem file_map_; // Protected by mutex_. + FileSystem file_map_ GUARDED_BY(mutex_); }; } // namespace -Env* NewMemEnv(Env* base_env) { - return new InMemoryEnv(base_env); -} +Env* NewMemEnv(Env* base_env) { return new InMemoryEnv(base_env); } } // namespace leveldb diff --git a/helpers/memenv/memenv.h b/helpers/memenv/memenv.h index 03b88de..3d929e4 100644 --- a/helpers/memenv/memenv.h +++ b/helpers/memenv/memenv.h @@ -5,6 +5,8 @@ #ifndef STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_ #define STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_ +#include "leveldb/export.h" + namespace leveldb { class Env; @@ -13,7 +15,7 @@ class Env; // 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); +LEVELDB_EXPORT Env* NewMemEnv(Env* base_env); } // namespace leveldb diff --git a/helpers/memenv/memenv_test.cc b/helpers/memenv/memenv_test.cc index a44310f..94ad06b 100644 --- a/helpers/memenv/memenv_test.cc +++ b/helpers/memenv/memenv_test.cc @@ -4,25 +4,22 @@ #include "helpers/memenv/memenv.h" +#include +#include + #include "db/db_impl.h" #include "leveldb/db.h" #include "leveldb/env.h" #include "util/testharness.h" -#include -#include namespace leveldb { class MemEnvTest { public: - Env* env_; + MemEnvTest() : env_(NewMemEnv(Env::Default())) {} + ~MemEnvTest() { delete env_; } - MemEnvTest() - : env_(NewMemEnv(Env::Default())) { - } - ~MemEnvTest() { - delete env_; - } + Env* env_; }; TEST(MemEnvTest, Basics) { @@ -40,6 +37,8 @@ TEST(MemEnvTest, Basics) { // Create a file. ASSERT_OK(env_->NewWritableFile("/dir/f", &writable_file)); + ASSERT_OK(env_->GetFileSize("/dir/f", &file_size)); + ASSERT_EQ(0, file_size); delete writable_file; // Check that the file exists. @@ -55,9 +54,16 @@ TEST(MemEnvTest, Basics) { ASSERT_OK(writable_file->Append("abc")); delete writable_file; - // Check for expected size. + // Check that append works. + ASSERT_OK(env_->NewAppendableFile("/dir/f", &writable_file)); ASSERT_OK(env_->GetFileSize("/dir/f", &file_size)); ASSERT_EQ(3, file_size); + ASSERT_OK(writable_file->Append("hello")); + delete writable_file; + + // Check for expected size. + ASSERT_OK(env_->GetFileSize("/dir/f", &file_size)); + ASSERT_EQ(8, file_size); // Check that renaming works. ASSERT_TRUE(!env_->RenameFile("/dir/non_existent", "/dir/g").ok()); @@ -65,7 +71,7 @@ TEST(MemEnvTest, Basics) { ASSERT_TRUE(!env_->FileExists("/dir/f")); ASSERT_TRUE(env_->FileExists("/dir/g")); ASSERT_OK(env_->GetFileSize("/dir/g", &file_size)); - ASSERT_EQ(3, file_size); + ASSERT_EQ(8, file_size); // Check that opening non-existent file fails. SequentialFile* seq_file; @@ -100,25 +106,25 @@ TEST(MemEnvTest, ReadWrite) { // Read sequentially. ASSERT_OK(env_->NewSequentialFile("/dir/f", &seq_file)); - ASSERT_OK(seq_file->Read(5, &result, scratch)); // Read "hello". + ASSERT_OK(seq_file->Read(5, &result, scratch)); // Read "hello". ASSERT_EQ(0, result.compare("hello")); ASSERT_OK(seq_file->Skip(1)); - ASSERT_OK(seq_file->Read(1000, &result, scratch)); // Read "world". + ASSERT_OK(seq_file->Read(1000, &result, scratch)); // Read "world". ASSERT_EQ(0, result.compare("world")); - ASSERT_OK(seq_file->Read(1000, &result, scratch)); // Try reading past EOF. + ASSERT_OK(seq_file->Read(1000, &result, scratch)); // Try reading past EOF. ASSERT_EQ(0, result.size()); - ASSERT_OK(seq_file->Skip(100)); // Try to skip past end of file. + ASSERT_OK(seq_file->Skip(100)); // Try to skip past end of file. ASSERT_OK(seq_file->Read(1000, &result, scratch)); ASSERT_EQ(0, result.size()); delete seq_file; // Random reads. ASSERT_OK(env_->NewRandomAccessFile("/dir/f", &rand_file)); - ASSERT_OK(rand_file->Read(6, 5, &result, scratch)); // Read "world". + ASSERT_OK(rand_file->Read(6, 5, &result, scratch)); // Read "world". ASSERT_EQ(0, result.compare("world")); - ASSERT_OK(rand_file->Read(0, 5, &result, scratch)); // Read "hello". + ASSERT_OK(rand_file->Read(0, 5, &result, scratch)); // Read "hello". ASSERT_EQ(0, result.compare("hello")); - ASSERT_OK(rand_file->Read(10, 100, &result, scratch)); // Read "d". + ASSERT_OK(rand_file->Read(10, 100, &result, scratch)); // Read "d". ASSERT_EQ(0, result.compare("d")); // Too high offset. @@ -167,7 +173,7 @@ TEST(MemEnvTest, LargeWrite) { SequentialFile* seq_file; Slice result; ASSERT_OK(env_->NewSequentialFile("/dir/f", &seq_file)); - ASSERT_OK(seq_file->Read(3, &result, scratch)); // Read "foo". + ASSERT_OK(seq_file->Read(3, &result, scratch)); // Read "foo". ASSERT_EQ(0, result.compare("foo")); size_t read = 0; @@ -179,7 +185,30 @@ TEST(MemEnvTest, LargeWrite) { } ASSERT_TRUE(write_data == read_data); delete seq_file; - delete [] scratch; + delete[] scratch; +} + +TEST(MemEnvTest, OverwriteOpenFile) { + const char kWrite1Data[] = "Write #1 data"; + const size_t kFileDataLen = sizeof(kWrite1Data) - 1; + const std::string kTestFileName = test::TmpDir() + "/leveldb-TestFile.dat"; + + ASSERT_OK(WriteStringToFile(env_, kWrite1Data, kTestFileName)); + + RandomAccessFile* rand_file; + ASSERT_OK(env_->NewRandomAccessFile(kTestFileName, &rand_file)); + + const char kWrite2Data[] = "Write #2 data"; + ASSERT_OK(WriteStringToFile(env_, kWrite2Data, kTestFileName)); + + // Verify that overwriting an open file will result in the new file data + // being read from files opened before the write. + Slice result; + char scratch[kFileDataLen]; + ASSERT_OK(rand_file->Read(0, kFileDataLen, &result, scratch)); + ASSERT_EQ(0, result.compare(kWrite2Data)); + + delete rand_file; } TEST(MemEnvTest, DBTest) { @@ -227,6 +256,4 @@ TEST(MemEnvTest, DBTest) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/include/leveldb/c.h b/include/leveldb/c.h index 1048fe3..04e383c 100644 --- a/include/leveldb/c.h +++ b/include/leveldb/c.h @@ -48,225 +48,205 @@ extern "C" { #include #include +#include "leveldb/export.h" + /* Exported types */ -typedef struct leveldb_t leveldb_t; -typedef struct leveldb_cache_t leveldb_cache_t; -typedef struct leveldb_comparator_t leveldb_comparator_t; -typedef struct leveldb_env_t leveldb_env_t; -typedef struct leveldb_filelock_t leveldb_filelock_t; -typedef struct leveldb_filterpolicy_t leveldb_filterpolicy_t; -typedef struct leveldb_iterator_t leveldb_iterator_t; -typedef struct leveldb_logger_t leveldb_logger_t; -typedef struct leveldb_options_t leveldb_options_t; -typedef struct leveldb_randomfile_t leveldb_randomfile_t; -typedef struct leveldb_readoptions_t leveldb_readoptions_t; -typedef struct leveldb_seqfile_t leveldb_seqfile_t; -typedef struct leveldb_snapshot_t leveldb_snapshot_t; -typedef struct leveldb_writablefile_t leveldb_writablefile_t; -typedef struct leveldb_writebatch_t leveldb_writebatch_t; -typedef struct leveldb_writeoptions_t leveldb_writeoptions_t; +typedef struct leveldb_t leveldb_t; +typedef struct leveldb_cache_t leveldb_cache_t; +typedef struct leveldb_comparator_t leveldb_comparator_t; +typedef struct leveldb_env_t leveldb_env_t; +typedef struct leveldb_filelock_t leveldb_filelock_t; +typedef struct leveldb_filterpolicy_t leveldb_filterpolicy_t; +typedef struct leveldb_iterator_t leveldb_iterator_t; +typedef struct leveldb_logger_t leveldb_logger_t; +typedef struct leveldb_options_t leveldb_options_t; +typedef struct leveldb_randomfile_t leveldb_randomfile_t; +typedef struct leveldb_readoptions_t leveldb_readoptions_t; +typedef struct leveldb_seqfile_t leveldb_seqfile_t; +typedef struct leveldb_snapshot_t leveldb_snapshot_t; +typedef struct leveldb_writablefile_t leveldb_writablefile_t; +typedef struct leveldb_writebatch_t leveldb_writebatch_t; +typedef struct leveldb_writeoptions_t leveldb_writeoptions_t; /* DB operations */ -extern leveldb_t* leveldb_open( - const leveldb_options_t* options, - const char* name, - char** errptr); +LEVELDB_EXPORT leveldb_t* leveldb_open(const leveldb_options_t* options, + const char* name, char** errptr); -extern void leveldb_close(leveldb_t* db); +LEVELDB_EXPORT void leveldb_close(leveldb_t* db); -extern void leveldb_put( - leveldb_t* db, - const leveldb_writeoptions_t* options, - const char* key, size_t keylen, - const char* val, size_t vallen, - char** errptr); +LEVELDB_EXPORT void leveldb_put(leveldb_t* db, + const leveldb_writeoptions_t* options, + const char* key, size_t keylen, const char* val, + size_t vallen, char** errptr); -extern void leveldb_delete( - leveldb_t* db, - const leveldb_writeoptions_t* options, - const char* key, size_t keylen, - char** errptr); +LEVELDB_EXPORT void leveldb_delete(leveldb_t* db, + const leveldb_writeoptions_t* options, + const char* key, size_t keylen, + char** errptr); -extern void leveldb_write( - leveldb_t* db, - const leveldb_writeoptions_t* options, - leveldb_writebatch_t* batch, - char** errptr); +LEVELDB_EXPORT void leveldb_write(leveldb_t* db, + const leveldb_writeoptions_t* options, + leveldb_writebatch_t* batch, char** errptr); /* Returns NULL if not found. A malloc()ed array otherwise. Stores the length of the array in *vallen. */ -extern char* leveldb_get( - leveldb_t* db, - const leveldb_readoptions_t* options, - const char* key, size_t keylen, - size_t* vallen, - char** errptr); +LEVELDB_EXPORT char* leveldb_get(leveldb_t* db, + const leveldb_readoptions_t* options, + const char* key, size_t keylen, size_t* vallen, + char** errptr); -extern leveldb_iterator_t* leveldb_create_iterator( - leveldb_t* db, - const leveldb_readoptions_t* options); +LEVELDB_EXPORT leveldb_iterator_t* leveldb_create_iterator( + leveldb_t* db, const leveldb_readoptions_t* options); -extern const leveldb_snapshot_t* leveldb_create_snapshot( - leveldb_t* db); +LEVELDB_EXPORT const leveldb_snapshot_t* leveldb_create_snapshot(leveldb_t* db); -extern void leveldb_release_snapshot( - leveldb_t* db, - const leveldb_snapshot_t* snapshot); +LEVELDB_EXPORT void leveldb_release_snapshot( + leveldb_t* db, const leveldb_snapshot_t* snapshot); /* Returns NULL if property name is unknown. Else returns a pointer to a malloc()-ed null-terminated value. */ -extern char* leveldb_property_value( - leveldb_t* db, - const char* propname); +LEVELDB_EXPORT char* leveldb_property_value(leveldb_t* db, + const char* propname); -extern void leveldb_approximate_sizes( - leveldb_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); +LEVELDB_EXPORT void leveldb_approximate_sizes( + leveldb_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 void leveldb_compact_range( - leveldb_t* db, - const char* start_key, size_t start_key_len, - const char* limit_key, size_t limit_key_len); +LEVELDB_EXPORT void leveldb_compact_range(leveldb_t* db, const char* start_key, + size_t start_key_len, + const char* limit_key, + size_t limit_key_len); /* Management operations */ -extern void leveldb_destroy_db( - const leveldb_options_t* options, - const char* name, - char** errptr); +LEVELDB_EXPORT void leveldb_destroy_db(const leveldb_options_t* options, + const char* name, char** errptr); -extern void leveldb_repair_db( - const leveldb_options_t* options, - const char* name, - char** errptr); +LEVELDB_EXPORT void leveldb_repair_db(const leveldb_options_t* options, + const char* name, char** errptr); /* Iterator */ -extern void leveldb_iter_destroy(leveldb_iterator_t*); -extern unsigned char leveldb_iter_valid(const leveldb_iterator_t*); -extern void leveldb_iter_seek_to_first(leveldb_iterator_t*); -extern void leveldb_iter_seek_to_last(leveldb_iterator_t*); -extern void leveldb_iter_seek(leveldb_iterator_t*, const char* k, size_t klen); -extern void leveldb_iter_next(leveldb_iterator_t*); -extern void leveldb_iter_prev(leveldb_iterator_t*); -extern const char* leveldb_iter_key(const leveldb_iterator_t*, size_t* klen); -extern const char* leveldb_iter_value(const leveldb_iterator_t*, size_t* vlen); -extern void leveldb_iter_get_error(const leveldb_iterator_t*, char** errptr); +LEVELDB_EXPORT void leveldb_iter_destroy(leveldb_iterator_t*); +LEVELDB_EXPORT unsigned char leveldb_iter_valid(const leveldb_iterator_t*); +LEVELDB_EXPORT void leveldb_iter_seek_to_first(leveldb_iterator_t*); +LEVELDB_EXPORT void leveldb_iter_seek_to_last(leveldb_iterator_t*); +LEVELDB_EXPORT void leveldb_iter_seek(leveldb_iterator_t*, const char* k, + size_t klen); +LEVELDB_EXPORT void leveldb_iter_next(leveldb_iterator_t*); +LEVELDB_EXPORT void leveldb_iter_prev(leveldb_iterator_t*); +LEVELDB_EXPORT const char* leveldb_iter_key(const leveldb_iterator_t*, + size_t* klen); +LEVELDB_EXPORT const char* leveldb_iter_value(const leveldb_iterator_t*, + size_t* vlen); +LEVELDB_EXPORT void leveldb_iter_get_error(const leveldb_iterator_t*, + char** errptr); /* Write batch */ -extern leveldb_writebatch_t* leveldb_writebatch_create(); -extern void leveldb_writebatch_destroy(leveldb_writebatch_t*); -extern void leveldb_writebatch_clear(leveldb_writebatch_t*); -extern void leveldb_writebatch_put( - leveldb_writebatch_t*, - const char* key, size_t klen, - const char* val, size_t vlen); -extern void leveldb_writebatch_delete( - leveldb_writebatch_t*, - const char* key, size_t klen); -extern void leveldb_writebatch_iterate( - leveldb_writebatch_t*, - void* state, +LEVELDB_EXPORT leveldb_writebatch_t* leveldb_writebatch_create(void); +LEVELDB_EXPORT void leveldb_writebatch_destroy(leveldb_writebatch_t*); +LEVELDB_EXPORT void leveldb_writebatch_clear(leveldb_writebatch_t*); +LEVELDB_EXPORT void leveldb_writebatch_put(leveldb_writebatch_t*, + const char* key, size_t klen, + const char* val, size_t vlen); +LEVELDB_EXPORT void leveldb_writebatch_delete(leveldb_writebatch_t*, + const char* key, size_t klen); +LEVELDB_EXPORT void leveldb_writebatch_iterate( + const leveldb_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)); +LEVELDB_EXPORT void leveldb_writebatch_append( + leveldb_writebatch_t* destination, const leveldb_writebatch_t* source); /* Options */ -extern leveldb_options_t* leveldb_options_create(); -extern void leveldb_options_destroy(leveldb_options_t*); -extern void leveldb_options_set_comparator( - leveldb_options_t*, - leveldb_comparator_t*); -extern void leveldb_options_set_filter_policy( - leveldb_options_t*, - leveldb_filterpolicy_t*); -extern void leveldb_options_set_create_if_missing( - leveldb_options_t*, unsigned char); -extern void leveldb_options_set_error_if_exists( - leveldb_options_t*, unsigned char); -extern void leveldb_options_set_paranoid_checks( - leveldb_options_t*, unsigned char); -extern void leveldb_options_set_env(leveldb_options_t*, leveldb_env_t*); -extern void leveldb_options_set_info_log(leveldb_options_t*, leveldb_logger_t*); -extern void leveldb_options_set_write_buffer_size(leveldb_options_t*, size_t); -extern void leveldb_options_set_max_open_files(leveldb_options_t*, int); -extern void leveldb_options_set_cache(leveldb_options_t*, leveldb_cache_t*); -extern void leveldb_options_set_block_size(leveldb_options_t*, size_t); -extern void leveldb_options_set_block_restart_interval(leveldb_options_t*, int); +LEVELDB_EXPORT leveldb_options_t* leveldb_options_create(void); +LEVELDB_EXPORT void leveldb_options_destroy(leveldb_options_t*); +LEVELDB_EXPORT void leveldb_options_set_comparator(leveldb_options_t*, + leveldb_comparator_t*); +LEVELDB_EXPORT void leveldb_options_set_filter_policy(leveldb_options_t*, + leveldb_filterpolicy_t*); +LEVELDB_EXPORT void leveldb_options_set_create_if_missing(leveldb_options_t*, + unsigned char); +LEVELDB_EXPORT void leveldb_options_set_error_if_exists(leveldb_options_t*, + unsigned char); +LEVELDB_EXPORT void leveldb_options_set_paranoid_checks(leveldb_options_t*, + unsigned char); +LEVELDB_EXPORT void leveldb_options_set_env(leveldb_options_t*, leveldb_env_t*); +LEVELDB_EXPORT void leveldb_options_set_info_log(leveldb_options_t*, + leveldb_logger_t*); +LEVELDB_EXPORT void leveldb_options_set_write_buffer_size(leveldb_options_t*, + size_t); +LEVELDB_EXPORT void leveldb_options_set_max_open_files(leveldb_options_t*, int); +LEVELDB_EXPORT void leveldb_options_set_cache(leveldb_options_t*, + leveldb_cache_t*); +LEVELDB_EXPORT void leveldb_options_set_block_size(leveldb_options_t*, size_t); +LEVELDB_EXPORT void leveldb_options_set_block_restart_interval( + leveldb_options_t*, int); +LEVELDB_EXPORT void leveldb_options_set_max_file_size(leveldb_options_t*, + size_t); -enum { - leveldb_no_compression = 0, - leveldb_snappy_compression = 1 -}; -extern void leveldb_options_set_compression(leveldb_options_t*, int); +enum { leveldb_no_compression = 0, leveldb_snappy_compression = 1 }; +LEVELDB_EXPORT void leveldb_options_set_compression(leveldb_options_t*, int); /* Comparator */ -extern leveldb_comparator_t* leveldb_comparator_create( - void* state, - void (*destructor)(void*), - int (*compare)( - void*, - const char* a, size_t alen, - const char* b, size_t blen), +LEVELDB_EXPORT leveldb_comparator_t* leveldb_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 void leveldb_comparator_destroy(leveldb_comparator_t*); +LEVELDB_EXPORT void leveldb_comparator_destroy(leveldb_comparator_t*); /* Filter policy */ -extern leveldb_filterpolicy_t* leveldb_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), +LEVELDB_EXPORT leveldb_filterpolicy_t* leveldb_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), const char* (*name)(void*)); -extern void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t*); +LEVELDB_EXPORT void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t*); -extern leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom( +LEVELDB_EXPORT leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom( int bits_per_key); /* Read options */ -extern leveldb_readoptions_t* leveldb_readoptions_create(); -extern void leveldb_readoptions_destroy(leveldb_readoptions_t*); -extern void leveldb_readoptions_set_verify_checksums( - leveldb_readoptions_t*, - unsigned char); -extern void leveldb_readoptions_set_fill_cache( +LEVELDB_EXPORT leveldb_readoptions_t* leveldb_readoptions_create(void); +LEVELDB_EXPORT void leveldb_readoptions_destroy(leveldb_readoptions_t*); +LEVELDB_EXPORT void leveldb_readoptions_set_verify_checksums( leveldb_readoptions_t*, unsigned char); -extern void leveldb_readoptions_set_snapshot( - leveldb_readoptions_t*, - const leveldb_snapshot_t*); +LEVELDB_EXPORT void leveldb_readoptions_set_fill_cache(leveldb_readoptions_t*, + unsigned char); +LEVELDB_EXPORT void leveldb_readoptions_set_snapshot(leveldb_readoptions_t*, + const leveldb_snapshot_t*); /* Write options */ -extern leveldb_writeoptions_t* leveldb_writeoptions_create(); -extern void leveldb_writeoptions_destroy(leveldb_writeoptions_t*); -extern void leveldb_writeoptions_set_sync( - leveldb_writeoptions_t*, unsigned char); +LEVELDB_EXPORT leveldb_writeoptions_t* leveldb_writeoptions_create(void); +LEVELDB_EXPORT void leveldb_writeoptions_destroy(leveldb_writeoptions_t*); +LEVELDB_EXPORT void leveldb_writeoptions_set_sync(leveldb_writeoptions_t*, + unsigned char); /* Cache */ -extern leveldb_cache_t* leveldb_cache_create_lru(size_t capacity); -extern void leveldb_cache_destroy(leveldb_cache_t* cache); +LEVELDB_EXPORT leveldb_cache_t* leveldb_cache_create_lru(size_t capacity); +LEVELDB_EXPORT void leveldb_cache_destroy(leveldb_cache_t* cache); /* Env */ -extern leveldb_env_t* leveldb_create_default_env(); -extern void leveldb_env_destroy(leveldb_env_t*); +LEVELDB_EXPORT leveldb_env_t* leveldb_create_default_env(void); +LEVELDB_EXPORT void leveldb_env_destroy(leveldb_env_t*); + +/* If not NULL, the returned buffer must be released using leveldb_free(). */ +LEVELDB_EXPORT char* leveldb_env_get_test_directory(leveldb_env_t*); /* Utility */ @@ -275,16 +255,16 @@ extern void leveldb_env_destroy(leveldb_env_t*); in this file. Note that in certain cases (typically on Windows), you may need to call this routine instead of free(ptr) to dispose of malloc()-ed memory returned by this library. */ -extern void leveldb_free(void* ptr); +LEVELDB_EXPORT void leveldb_free(void* ptr); /* Return the major version number for this release. */ -extern int leveldb_major_version(); +LEVELDB_EXPORT int leveldb_major_version(void); /* Return the minor version number for this release. */ -extern int leveldb_minor_version(); +LEVELDB_EXPORT int leveldb_minor_version(void); #ifdef __cplusplus -} /* end extern "C" */ +} /* end extern "C" */ #endif -#endif /* STORAGE_LEVELDB_INCLUDE_C_H_ */ +#endif /* STORAGE_LEVELDB_INCLUDE_C_H_ */ diff --git a/include/leveldb/cache.h b/include/leveldb/cache.h index 1a201e5..7d1a221 100644 --- a/include/leveldb/cache.h +++ b/include/leveldb/cache.h @@ -19,26 +19,31 @@ #define STORAGE_LEVELDB_INCLUDE_CACHE_H_ #include + +#include "leveldb/export.h" #include "leveldb/slice.h" namespace leveldb { -class Cache; +class LEVELDB_EXPORT Cache; // Create a new cache with a fixed size capacity. This implementation // of Cache uses a least-recently-used eviction policy. -extern Cache* NewLRUCache(size_t capacity); +LEVELDB_EXPORT Cache* NewLRUCache(size_t capacity); -class Cache { +class LEVELDB_EXPORT Cache { public: - Cache() { } + Cache() = default; + + Cache(const Cache&) = delete; + Cache& operator=(const Cache&) = delete; // Destroys all existing entries by calling the "deleter" // function that was passed to the constructor. virtual ~Cache(); // Opaque handle to an entry stored in the cache. - struct Handle { }; + struct Handle {}; // Insert a mapping from key->value into the cache and assign it // the specified charge against the total cache capacity. @@ -52,7 +57,7 @@ class Cache { 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 NULL. + // 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 @@ -81,6 +86,17 @@ class Cache { // its cache keys. virtual uint64_t NewId() = 0; + // Remove all cache entries that are not actively in use. Memory-constrained + // applications may wish to call this method to reduce memory usage. + // Default implementation of Prune() does nothing. Subclasses are strongly + // encouraged to override the default implementation. A future release of + // leveldb may change Prune() to a pure abstract method. + virtual void Prune() {} + + // Return an estimate of the combined charges of all elements stored in the + // cache. + virtual size_t TotalCharge() const = 0; + private: void LRU_Remove(Handle* e); void LRU_Append(Handle* e); @@ -88,10 +104,6 @@ class Cache { struct Rep; Rep* rep_; - - // No copying allowed - Cache(const Cache&); - void operator=(const Cache&); }; } // namespace leveldb diff --git a/include/leveldb/comparator.h b/include/leveldb/comparator.h index 556b984..a85b51e 100644 --- a/include/leveldb/comparator.h +++ b/include/leveldb/comparator.h @@ -7,6 +7,8 @@ #include +#include "leveldb/export.h" + namespace leveldb { class Slice; @@ -15,7 +17,7 @@ class Slice; // used as keys in an sstable or a database. A Comparator implementation // must be thread-safe since leveldb may invoke its methods concurrently // from multiple threads. -class Comparator { +class LEVELDB_EXPORT Comparator { public: virtual ~Comparator(); @@ -43,9 +45,8 @@ class Comparator { // 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; + 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, @@ -56,7 +57,7 @@ class Comparator { // 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(); +LEVELDB_EXPORT const Comparator* BytewiseComparator(); } // namespace leveldb diff --git a/include/leveldb/db.h b/include/leveldb/db.h index 4c169bf..b73014a 100644 --- a/include/leveldb/db.h +++ b/include/leveldb/db.h @@ -7,14 +7,16 @@ #include #include + +#include "leveldb/export.h" #include "leveldb/iterator.h" #include "leveldb/options.h" namespace leveldb { -// Update Makefile if you change these +// Update CMakeLists.txt if you change these static const int kMajorVersion = 1; -static const int kMinorVersion = 18; +static const int kMinorVersion = 22; struct Options; struct ReadOptions; @@ -24,42 +26,44 @@ class WriteBatch; // 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. -class Snapshot { +class LEVELDB_EXPORT Snapshot { protected: virtual ~Snapshot(); }; // A range of keys -struct Range { - Slice start; // Included in the range - Slice limit; // Not included in the range +struct LEVELDB_EXPORT Range { + Range() = default; + Range(const Slice& s, const Slice& l) : start(s), limit(l) {} - Range() { } - Range(const Slice& s, const Slice& l) : start(s), limit(l) { } + Slice start; // Included in the range + Slice limit; // Not included in the range }; // 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 { +class LEVELDB_EXPORT 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 NULL in *dbptr and returns a non-OK status on error. + // 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, + static Status Open(const Options& options, const std::string& name, DB** dbptr); - DB() { } + DB() = default; + + DB(const DB&) = delete; + DB& operator=(const DB&) = delete; + virtual ~DB(); // Set the database entry for "key" to "value". Returns OK on success, // and a non-OK status on error. // Note: consider setting options.sync = true. - virtual Status Put(const WriteOptions& options, - const Slice& key, + virtual Status Put(const WriteOptions& options, const Slice& key, const Slice& value) = 0; // Remove the database entry (if any) for "key". Returns OK on @@ -80,8 +84,8 @@ class DB { // a status for which Status::IsNotFound() returns true. // // May return some other Status on an error. - virtual Status Get(const ReadOptions& options, - const Slice& key, std::string* value) = 0; + virtual Status Get(const ReadOptions& options, const Slice& key, + std::string* value) = 0; // Return a heap-allocated iterator over the contents of the database. // The result of NewIterator() is initially invalid (caller must @@ -115,6 +119,8 @@ class DB { // about the internal operation of the DB. // "leveldb.sstables" - returns a multi-line string that describes all // of the sstables that make up the db contents. + // "leveldb.approximate-memory-usage" - returns the approximate number of + // bytes of memory in use by the DB. virtual bool GetProperty(const Slice& property, std::string* value) = 0; // For each i in [0,n-1], store in "sizes[i]", the approximate @@ -134,27 +140,27 @@ class DB { // needed to access the data. This operation should typically only // be invoked by users who understand the underlying implementation. // - // begin==NULL is treated as a key before all keys in the database. - // end==NULL is treated as a key after all keys in the database. + // 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(NULL, NULL); + // db->CompactRange(nullptr, nullptr); virtual void CompactRange(const Slice* begin, const Slice* end) = 0; - - 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); +// +// Note: For backwards compatibility, if DestroyDB is unable to list the +// database files, Status::OK() will still be returned masking this failure. +LEVELDB_EXPORT Status DestroyDB(const std::string& name, + const Options& options); // 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); +LEVELDB_EXPORT Status RepairDB(const std::string& dbname, + const Options& options); } // namespace leveldb diff --git a/include/leveldb/dumpfile.h b/include/leveldb/dumpfile.h index 3f97fda..a58bc6b 100644 --- a/include/leveldb/dumpfile.h +++ b/include/leveldb/dumpfile.h @@ -6,7 +6,9 @@ #define STORAGE_LEVELDB_INCLUDE_DUMPFILE_H_ #include + #include "leveldb/env.h" +#include "leveldb/export.h" #include "leveldb/status.h" namespace leveldb { @@ -18,7 +20,8 @@ namespace leveldb { // // Returns a non-OK result if fname does not name a leveldb storage // file, or if the file cannot be read. -Status DumpFile(Env* env, const std::string& fname, WritableFile* dst); +LEVELDB_EXPORT Status DumpFile(Env* env, const std::string& fname, + WritableFile* dst); } // namespace leveldb diff --git a/include/leveldb/env.h b/include/leveldb/env.h index f709514..112fe96 100644 --- a/include/leveldb/env.h +++ b/include/leveldb/env.h @@ -13,12 +13,36 @@ #ifndef STORAGE_LEVELDB_INCLUDE_ENV_H_ #define STORAGE_LEVELDB_INCLUDE_ENV_H_ -#include -#include #include #include + +#include +#include + +#include "leveldb/export.h" #include "leveldb/status.h" +#if defined(_WIN32) +// The leveldb::Env class below contains a DeleteFile method. +// At the same time, , a fairly popular header +// file for Windows applications, defines a DeleteFile macro. +// +// Without any intervention on our part, the result of this +// unfortunate coincidence is that the name of the +// leveldb::Env::DeleteFile method seen by the compiler depends on +// whether was included before or after the LevelDB +// headers. +// +// To avoid headaches, we undefined DeleteFile (if defined) and +// redefine it at the bottom of this file. This way +// can be included before this file (or not at all) and the +// exported method will always be leveldb::Env::DeleteFile. +#if defined(DeleteFile) +#undef DeleteFile +#define LEVELDB_DELETEFILE_UNDEFINED +#endif // defined(DeleteFile) +#endif // defined(_WIN32) + namespace leveldb { class FileLock; @@ -28,9 +52,13 @@ class SequentialFile; class Slice; class WritableFile; -class Env { +class LEVELDB_EXPORT Env { public: - Env() { } + Env() = default; + + Env(const Env&) = delete; + Env& operator=(const Env&) = delete; + virtual ~Env(); // Return a default environment suitable for the current operating @@ -40,20 +68,22 @@ class Env { // The result of Default() belongs to leveldb and must never be deleted. static Env* Default(); - // Create a brand new sequentially-readable file with the specified name. + // Create an object that sequentially reads the file with the specified name. // On success, stores a pointer to the new file in *result and returns OK. - // On failure stores NULL in *result and returns non-OK. If the file does - // not exist, returns a non-OK status. + // On failure stores nullptr in *result and returns non-OK. If the file does + // not exist, returns a non-OK status. Implementations should return a + // NotFound status when the file does not exist. // // The returned file will only be accessed by one thread at a time. virtual Status NewSequentialFile(const std::string& fname, SequentialFile** result) = 0; - // Create a brand new random access read-only file with the + // Create an object supporting random-access reads from the file with the // specified name. On success, stores a pointer to the new file in - // *result and returns OK. On failure stores NULL in *result and + // *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. + // status. Implementations should return a NotFound status when the file does + // not exist. // // The returned file may be concurrently accessed by multiple threads. virtual Status NewRandomAccessFile(const std::string& fname, @@ -62,13 +92,28 @@ class Env { // 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 NULL in *result and + // *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, WritableFile** result) = 0; + // Create an object that either appends to an existing file, or + // writes to a new file (if the file does not exist to begin with). + // 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. + // + // May return an IsNotSupportedError error if this Env does + // not allow appending to an existing file. Users of Env (including + // the leveldb implementation) must be prepared to deal with + // an Env that does not support appending. + virtual Status NewAppendableFile(const std::string& fname, + WritableFile** result); + // Returns true iff the named file exists. virtual bool FileExists(const std::string& fname) = 0; @@ -95,7 +140,7 @@ class Env { const std::string& target) = 0; // Lock the specified file. Used to prevent concurrent access to - // the same db by multiple processes. On failure, stores NULL in + // 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 @@ -121,16 +166,14 @@ class Env { // added to the same Env may run concurrently in different threads. // I.e., the caller may not assume that background work items are // serialized. - virtual void Schedule( - void (*function)(void* arg), - void* arg) = 0; + virtual void Schedule(void (*function)(void* arg), void* arg) = 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; // *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 + // or may 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; @@ -144,17 +187,16 @@ class Env { // Sleep/delay the thread for the prescribed number of micro-seconds. virtual void SleepForMicroseconds(int micros) = 0; - - private: - // No copying allowed - Env(const Env&); - void operator=(const Env&); }; // A file abstraction for reading sequentially through a file -class SequentialFile { +class LEVELDB_EXPORT SequentialFile { public: - SequentialFile() { } + SequentialFile() = default; + + SequentialFile(const SequentialFile&) = delete; + SequentialFile& operator=(const SequentialFile&) = delete; + virtual ~SequentialFile(); // Read up to "n" bytes from the file. "scratch[0..n-1]" may be @@ -175,17 +217,16 @@ class SequentialFile { // // REQUIRES: External synchronization virtual Status Skip(uint64_t n) = 0; - - private: - // No copying allowed - SequentialFile(const SequentialFile&); - void operator=(const SequentialFile&); }; // A file abstraction for randomly reading the contents of a file. -class RandomAccessFile { +class LEVELDB_EXPORT RandomAccessFile { public: - RandomAccessFile() { } + RandomAccessFile() = default; + + RandomAccessFile(const RandomAccessFile&) = delete; + RandomAccessFile& operator=(const RandomAccessFile&) = delete; + virtual ~RandomAccessFile(); // Read up to "n" bytes from the file starting at "offset". @@ -199,135 +240,148 @@ class RandomAccessFile { // Safe for concurrent use by multiple threads. virtual Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const = 0; - - private: - // No copying allowed - RandomAccessFile(const RandomAccessFile&); - void operator=(const RandomAccessFile&); }; // 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 { +class LEVELDB_EXPORT WritableFile { public: - WritableFile() { } + WritableFile() = default; + + WritableFile(const WritableFile&) = delete; + WritableFile& operator=(const WritableFile&) = delete; + virtual ~WritableFile(); virtual Status Append(const Slice& data) = 0; virtual Status Close() = 0; virtual Status Flush() = 0; virtual Status Sync() = 0; - - private: - // No copying allowed - WritableFile(const WritableFile&); - void operator=(const WritableFile&); }; // An interface for writing log messages. -class Logger { +class LEVELDB_EXPORT Logger { public: - Logger() { } + Logger() = default; + + Logger(const Logger&) = delete; + Logger& operator=(const Logger&) = delete; + virtual ~Logger(); // Write an entry to the log file with the specified format. virtual void Logv(const char* format, va_list ap) = 0; - - private: - // No copying allowed - Logger(const Logger&); - void operator=(const Logger&); }; - // Identifies a locked file. -class FileLock { +class LEVELDB_EXPORT FileLock { public: - FileLock() { } + FileLock() = default; + + FileLock(const FileLock&) = delete; + FileLock& operator=(const FileLock&) = delete; + virtual ~FileLock(); - private: - // No copying allowed - FileLock(const FileLock&); - void operator=(const FileLock&); }; -// Log the specified data to *info_log if info_log is non-NULL. -extern void Log(Logger* info_log, const char* format, ...) -# if defined(__GNUC__) || defined(__clang__) - __attribute__((__format__ (__printf__, 2, 3))) -# endif +// Log the specified data to *info_log if info_log is non-null. +void Log(Logger* info_log, const char* format, ...) +#if defined(__GNUC__) || defined(__clang__) + __attribute__((__format__(__printf__, 2, 3))) +#endif ; // A utility routine: write "data" to the named file. -extern Status WriteStringToFile(Env* env, const Slice& data, - const std::string& fname); +LEVELDB_EXPORT Status WriteStringToFile(Env* env, const Slice& data, + const std::string& fname); // A utility routine: read contents of named file into *data -extern Status ReadFileToString(Env* env, const std::string& fname, - std::string* data); +LEVELDB_EXPORT 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 { +class LEVELDB_EXPORT EnvWrapper : public Env { public: - // Initialize an EnvWrapper that delegates all calls to *t - explicit EnvWrapper(Env* t) : target_(t) { } + // 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 + // 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, SequentialFile** r) { + // The following text is boilerplate that forwards all methods to target(). + Status NewSequentialFile(const std::string& f, SequentialFile** r) override { return target_->NewSequentialFile(f, r); } - Status NewRandomAccessFile(const std::string& f, RandomAccessFile** r) { + Status NewRandomAccessFile(const std::string& f, + RandomAccessFile** r) override { return target_->NewRandomAccessFile(f, r); } - Status NewWritableFile(const std::string& f, WritableFile** r) { + Status NewWritableFile(const std::string& f, WritableFile** r) override { return target_->NewWritableFile(f, r); } - bool FileExists(const std::string& f) { return target_->FileExists(f); } - Status GetChildren(const std::string& dir, std::vector* r) { + Status NewAppendableFile(const std::string& f, WritableFile** r) override { + return target_->NewAppendableFile(f, r); + } + bool 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) { return target_->DeleteFile(f); } - Status CreateDir(const std::string& d) { return target_->CreateDir(d); } - Status DeleteDir(const std::string& d) { return target_->DeleteDir(d); } - Status GetFileSize(const std::string& f, uint64_t* s) { + Status DeleteFile(const std::string& f) override { + return target_->DeleteFile(f); + } + Status CreateDir(const std::string& d) override { + return target_->CreateDir(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 RenameFile(const std::string& s, const std::string& t) { + Status RenameFile(const std::string& s, const std::string& t) override { return target_->RenameFile(s, t); } - Status LockFile(const std::string& f, FileLock** l) { + Status LockFile(const std::string& f, FileLock** l) override { return target_->LockFile(f, l); } - Status UnlockFile(FileLock* l) { return target_->UnlockFile(l); } - void Schedule(void (*f)(void*), void* a) { + Status UnlockFile(FileLock* l) override { return target_->UnlockFile(l); } + void Schedule(void (*f)(void*), void* a) override { return target_->Schedule(f, a); } - void StartThread(void (*f)(void*), void* a) { + void StartThread(void (*f)(void*), void* a) override { return target_->StartThread(f, a); } - virtual Status GetTestDirectory(std::string* path) { + Status GetTestDirectory(std::string* path) override { return target_->GetTestDirectory(path); } - virtual Status NewLogger(const std::string& fname, Logger** result) { + Status NewLogger(const std::string& fname, Logger** result) override { return target_->NewLogger(fname, result); } - uint64_t NowMicros() { - return target_->NowMicros(); - } - void SleepForMicroseconds(int micros) { + uint64_t NowMicros() override { return target_->NowMicros(); } + void SleepForMicroseconds(int micros) override { target_->SleepForMicroseconds(micros); } + private: Env* target_; }; } // namespace leveldb +// Redefine DeleteFile if necessary. +#if defined(_WIN32) && defined(LEVELDB_DELETEFILE_UNDEFINED) +#if defined(UNICODE) +#define DeleteFile DeleteFileW +#else +#define DeleteFile DeleteFileA +#endif // defined(UNICODE) +#endif // defined(_WIN32) && defined(LEVELDB_DELETEFILE_UNDEFINED) + #endif // STORAGE_LEVELDB_INCLUDE_ENV_H_ diff --git a/include/leveldb/export.h b/include/leveldb/export.h new file mode 100644 index 0000000..6ba9b18 --- /dev/null +++ b/include/leveldb/export.h @@ -0,0 +1,33 @@ +// Copyright (c) 2017 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_INCLUDE_EXPORT_H_ +#define STORAGE_LEVELDB_INCLUDE_EXPORT_H_ + +#if !defined(LEVELDB_EXPORT) + +#if defined(LEVELDB_SHARED_LIBRARY) +#if defined(_WIN32) + +#if defined(LEVELDB_COMPILE_LIBRARY) +#define LEVELDB_EXPORT __declspec(dllexport) +#else +#define LEVELDB_EXPORT __declspec(dllimport) +#endif // defined(LEVELDB_COMPILE_LIBRARY) + +#else // defined(_WIN32) +#if defined(LEVELDB_COMPILE_LIBRARY) +#define LEVELDB_EXPORT __attribute__((visibility("default"))) +#else +#define LEVELDB_EXPORT +#endif +#endif // defined(_WIN32) + +#else // defined(LEVELDB_SHARED_LIBRARY) +#define LEVELDB_EXPORT +#endif + +#endif // !defined(LEVELDB_EXPORT) + +#endif // STORAGE_LEVELDB_INCLUDE_EXPORT_H_ diff --git a/include/leveldb/filter_policy.h b/include/leveldb/filter_policy.h index 1fba080..49c8eda 100644 --- a/include/leveldb/filter_policy.h +++ b/include/leveldb/filter_policy.h @@ -18,11 +18,13 @@ #include +#include "leveldb/export.h" + namespace leveldb { class Slice; -class FilterPolicy { +class LEVELDB_EXPORT FilterPolicy { public: virtual ~FilterPolicy(); @@ -38,8 +40,8 @@ class FilterPolicy { // // 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; + 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 @@ -63,8 +65,8 @@ class FilterPolicy { // 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); +LEVELDB_EXPORT const FilterPolicy* NewBloomFilterPolicy(int bits_per_key); -} +} // namespace leveldb #endif // STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_ diff --git a/include/leveldb/iterator.h b/include/leveldb/iterator.h index da631ed..bb9a5df 100644 --- a/include/leveldb/iterator.h +++ b/include/leveldb/iterator.h @@ -15,14 +15,19 @@ #ifndef STORAGE_LEVELDB_INCLUDE_ITERATOR_H_ #define STORAGE_LEVELDB_INCLUDE_ITERATOR_H_ +#include "leveldb/export.h" #include "leveldb/slice.h" #include "leveldb/status.h" namespace leveldb { -class Iterator { +class LEVELDB_EXPORT Iterator { public: Iterator(); + + Iterator(const Iterator&) = delete; + Iterator& operator=(const Iterator&) = delete; + virtual ~Iterator(); // An iterator is either positioned at a key/value pair, or @@ -72,28 +77,35 @@ class Iterator { // // 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); + using CleanupFunction = void (*)(void* arg1, void* arg2); void RegisterCleanup(CleanupFunction function, void* arg1, void* arg2); private: - struct Cleanup { + // Cleanup functions are stored in a single-linked list. + // The list's head node is inlined in the iterator. + struct CleanupNode { + // True if the node is not used. Only head nodes might be unused. + bool IsEmpty() const { return function == nullptr; } + // Invokes the cleanup function. + void Run() { + assert(function != nullptr); + (*function)(arg1, arg2); + } + + // The head node is used if the function pointer is not null. CleanupFunction function; void* arg1; void* arg2; - Cleanup* next; + CleanupNode* next; }; - Cleanup cleanup_; - - // No copying allowed - Iterator(const Iterator&); - void operator=(const Iterator&); + CleanupNode cleanup_head_; }; // Return an empty iterator (yields nothing). -extern Iterator* NewEmptyIterator(); +LEVELDB_EXPORT Iterator* NewEmptyIterator(); // Return an empty iterator with the specified status. -extern Iterator* NewErrorIterator(const Status& status); +LEVELDB_EXPORT Iterator* NewErrorIterator(const Status& status); } // namespace leveldb diff --git a/include/leveldb/options.h b/include/leveldb/options.h index 7c9b973..b748772 100644 --- a/include/leveldb/options.h +++ b/include/leveldb/options.h @@ -7,6 +7,8 @@ #include +#include "leveldb/export.h" + namespace leveldb { class Cache; @@ -23,12 +25,15 @@ class Snapshot; enum CompressionType { // NOTE: do not change the values of existing entries, as these are // part of the persistent format on disk. - kNoCompression = 0x0, + kNoCompression = 0x0, kSnappyCompression = 0x1 }; // Options to control the behavior of a database (passed to DB::Open) -struct Options { +struct LEVELDB_EXPORT Options { + // Create an Options object with default values for all fields. + Options(); + // ------------------- // Parameters that affect behavior @@ -41,20 +46,17 @@ struct Options { const Comparator* comparator; // If true, the database will be created if it is missing. - // Default: false - bool create_if_missing; + bool create_if_missing = false; // If true, an error is raised if the database already exists. - // Default: false - bool error_if_exists; + bool error_if_exists = false; // 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; + bool paranoid_checks = false; // Use the specified object to interact with the environment, // e.g. to read/write files, schedule background work, etc. @@ -62,10 +64,9 @@ struct Options { 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; + // 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. + Logger* info_log = nullptr; // ------------------- // Parameters that affect performance @@ -78,39 +79,40 @@ struct Options { // 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; + size_t write_buffer_size = 4 * 1024 * 1024; // 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; + int max_open_files = 1000; // 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; + // If non-null, use the specified cache for blocks. + // If null, leveldb will automatically create and use an 8MB internal cache. + Cache* block_cache = 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. - // - // Default: 4K - size_t block_size; + size_t block_size = 4 * 1024; // 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; + int block_restart_interval = 16; + + // Leveldb will write up to this amount of bytes to a file before + // switching to a new one. + // Most clients should leave this parameter alone. However if your + // filesystem is more efficient with larger files, you could + // consider increasing the value. The downside will be longer + // compactions and hence longer latency/performance hiccups. + // Another reason to increase this parameter might be when you are + // initially populating a large database. + size_t max_file_size = 2 * 1024 * 1024; // Compress blocks using the specified compression algorithm. This // parameter can be changed dynamically. @@ -126,47 +128,43 @@ struct Options { // 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; + CompressionType compression = kSnappyCompression; - // If non-NULL, use the specified filter policy to reduce disk reads. + // EXPERIMENTAL: If true, append to existing MANIFEST and log files + // when a database is opened. This can significantly speed up open. + // + // Default: currently false, but may become true later. + bool reuse_logs = false; + + // 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 an Options object with default values for all fields. - Options(); + const FilterPolicy* filter_policy = nullptr; }; // Options that control read operations -struct ReadOptions { +struct LEVELDB_EXPORT ReadOptions { + ReadOptions() = default; + // If true, all data read from underlying storage will be // verified against corresponding checksums. - // Default: false - bool verify_checksums; + bool verify_checksums = false; // Should the data 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; + bool fill_cache = true; - // If "snapshot" is non-NULL, read as of the supplied snapshot + // If "snapshot" is non-null, 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 NULL, use an implicit + // not have been released). If "snapshot" is null, use an implicit // snapshot of the state at the beginning of this read operation. - // Default: NULL - const Snapshot* snapshot; - - ReadOptions() - : verify_checksums(false), - fill_cache(true), - snapshot(NULL) { - } + const Snapshot* snapshot = nullptr; }; // Options that control write operations -struct WriteOptions { +struct LEVELDB_EXPORT WriteOptions { + WriteOptions() = default; + // 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 @@ -181,13 +179,7 @@ struct WriteOptions { // crash semantics as the "write()" system call. A DB write // with sync==true has similar crash semantics to a "write()" // system call followed by "fsync()". - // - // Default: false - bool sync; - - WriteOptions() - : sync(false) { - } + bool sync = false; }; } // namespace leveldb diff --git a/include/leveldb/slice.h b/include/leveldb/slice.h index bc36798..2df417d 100644 --- a/include/leveldb/slice.h +++ b/include/leveldb/slice.h @@ -18,23 +18,30 @@ #include #include #include + #include +#include "leveldb/export.h" + namespace leveldb { -class Slice { +class LEVELDB_EXPORT Slice { public: // Create an empty slice. - Slice() : data_(""), size_(0) { } + 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) { } + Slice(const char* d, size_t n) : data_(d), size_(n) {} // Create a slice that refers to the contents of "s" - Slice(const std::string& s) : data_(s.data()), size_(s.size()) { } + Slice(const std::string& s) : data_(s.data()), size_(s.size()) {} // Create a slice that refers to s[0,strlen(s)-1] - Slice(const char* s) : data_(s), size_(strlen(s)) { } + Slice(const char* s) : data_(s), size_(strlen(s)) {} + + // Intentionally copyable. + Slice(const Slice&) = default; + Slice& operator=(const Slice&) = default; // Return a pointer to the beginning of the referenced data const char* data() const { return data_; } @@ -53,7 +60,10 @@ class Slice { } // Change this slice to refer to an empty array - void clear() { data_ = ""; size_ = 0; } + void clear() { + data_ = ""; + size_ = 0; + } // Drop the first "n" bytes from this slice. void remove_prefix(size_t n) { @@ -73,15 +83,12 @@ class Slice { // 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)); + return ((size_ >= x.size_) && (memcmp(data_, x.data_, x.size_) == 0)); } private: const char* data_; size_t size_; - - // Intentionally copyable }; inline bool operator==(const Slice& x, const Slice& y) { @@ -89,21 +96,20 @@ inline bool operator==(const Slice& x, const Slice& y) { (memcmp(x.data(), y.data(), x.size()) == 0)); } -inline bool operator!=(const Slice& x, const Slice& y) { - return !(x == y); -} +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; + if (size_ < b.size_) + r = -1; + else if (size_ > b.size_) + r = +1; } return r; } } // namespace leveldb - #endif // STORAGE_LEVELDB_INCLUDE_SLICE_H_ diff --git a/include/leveldb/status.h b/include/leveldb/status.h index 11dbd4b..e327314 100644 --- a/include/leveldb/status.h +++ b/include/leveldb/status.h @@ -13,20 +13,25 @@ #ifndef STORAGE_LEVELDB_INCLUDE_STATUS_H_ #define STORAGE_LEVELDB_INCLUDE_STATUS_H_ +#include #include + +#include "leveldb/export.h" #include "leveldb/slice.h" namespace leveldb { -class Status { +class LEVELDB_EXPORT Status { public: // Create a success status. - Status() : state_(NULL) { } + Status() noexcept : state_(nullptr) {} ~Status() { delete[] state_; } - // Copy the specified status. - Status(const Status& s); - void operator=(const Status& s); + Status(const Status& rhs); + Status& operator=(const Status& rhs); + + Status(Status&& rhs) noexcept : state_(rhs.state_) { rhs.state_ = nullptr; } + Status& operator=(Status&& rhs) noexcept; // Return a success status. static Status OK() { return Status(); } @@ -49,7 +54,7 @@ class Status { } // Returns true iff the status indicates success. - bool ok() const { return (state_ == NULL); } + bool ok() const { return (state_ == nullptr); } // Returns true iff the status indicates a NotFound error. bool IsNotFound() const { return code() == kNotFound; } @@ -60,18 +65,17 @@ class Status { // Returns true iff the status indicates an IOError. bool IsIOError() const { return code() == kIOError; } + // Returns true iff the status indicates a NotSupportedError. + bool IsNotSupportedError() const { return code() == kNotSupported; } + + // Returns true iff the status indicates an InvalidArgument. + bool IsInvalidArgument() const { return code() == kInvalidArgument; } + // Return a string representation of this status suitable for printing. // Returns the string "OK" for success. std::string ToString() const; private: - // OK status has a NULL state_. Otherwise, state_ is a new[] array - // of the following form: - // state_[0..3] == length of message - // state_[4] == code - // state_[5..] == message - const char* state_; - enum Code { kOk = 0, kNotFound = 1, @@ -82,23 +86,35 @@ class Status { }; Code code() const { - return (state_ == NULL) ? kOk : static_cast(state_[4]); + return (state_ == nullptr) ? kOk : static_cast(state_[4]); } Status(Code code, const Slice& msg, const Slice& msg2); static const char* CopyState(const char* s); + + // OK status has a null state_. Otherwise, state_ is a new[] array + // of the following form: + // state_[0..3] == length of message + // state_[4] == code + // state_[5..] == message + const char* state_; }; -inline Status::Status(const Status& s) { - state_ = (s.state_ == NULL) ? NULL : CopyState(s.state_); +inline Status::Status(const Status& rhs) { + state_ = (rhs.state_ == nullptr) ? nullptr : CopyState(rhs.state_); } -inline void 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 (state_ != s.state_) { +inline Status& Status::operator=(const Status& rhs) { + // The following condition catches both aliasing (when this == &rhs), + // and the common case where both rhs and *this are ok. + if (state_ != rhs.state_) { delete[] state_; - state_ = (s.state_ == NULL) ? NULL : CopyState(s.state_); + state_ = (rhs.state_ == nullptr) ? nullptr : CopyState(rhs.state_); } + return *this; +} +inline Status& Status::operator=(Status&& rhs) noexcept { + std::swap(state_, rhs.state_); + return *this; } } // namespace leveldb diff --git a/include/leveldb/table.h b/include/leveldb/table.h index a9746c3..25c6013 100644 --- a/include/leveldb/table.h +++ b/include/leveldb/table.h @@ -6,6 +6,8 @@ #define STORAGE_LEVELDB_INCLUDE_TABLE_H_ #include + +#include "leveldb/export.h" #include "leveldb/iterator.h" namespace leveldb { @@ -21,7 +23,7 @@ class TableCache; // 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 Table { +class LEVELDB_EXPORT Table { public: // Attempt to open the table that is stored in bytes [0..file_size) // of "file", and read the metadata entries necessary to allow @@ -30,15 +32,16 @@ class Table { // If successful, returns ok and sets "*table" to the newly opened // table. The client should delete "*table" when no longer needed. // If there was an error while initializing the table, sets "*table" - // to NULL and returns a non-ok status. Does not take ownership of + // to nullptr and returns a non-ok status. Does not take ownership of // "*source", but the client must ensure that "source" remains live // for the duration of the returned table's lifetime. // // *file must remain live while this Table is in use. - static Status Open(const Options& options, - RandomAccessFile* file, - uint64_t file_size, - Table** table); + static Status Open(const Options& options, RandomAccessFile* file, + uint64_t file_size, Table** table); + + Table(const Table&) = delete; + Table& operator=(const Table&) = delete; ~Table(); @@ -56,28 +59,24 @@ class Table { uint64_t ApproximateOffsetOf(const Slice& key) const; private: + friend class TableCache; struct Rep; - Rep* rep_; - explicit Table(Rep* rep) { rep_ = rep; } static Iterator* BlockReader(void*, const ReadOptions&, const Slice&); + explicit Table(Rep* rep) : rep_(rep) {} + // Calls (*handle_result)(arg, ...) with the entry found after a call // to Seek(key). May not make such a call if filter policy says // that key is not present. - friend class TableCache; - Status InternalGet( - const ReadOptions&, const Slice& key, - void* arg, - void (*handle_result)(void* arg, const Slice& k, const Slice& v)); - + Status InternalGet(const ReadOptions&, const Slice& key, void* arg, + void (*handle_result)(void* arg, const Slice& k, + const Slice& v)); void ReadMeta(const Footer& footer); void ReadFilter(const Slice& filter_handle_value); - // No copying allowed - Table(const Table&); - void operator=(const Table&); + Rep* const rep_; }; } // namespace leveldb diff --git a/include/leveldb/table_builder.h b/include/leveldb/table_builder.h index 5fd1dc7..7d8896b 100644 --- a/include/leveldb/table_builder.h +++ b/include/leveldb/table_builder.h @@ -14,6 +14,8 @@ #define STORAGE_LEVELDB_INCLUDE_TABLE_BUILDER_H_ #include + +#include "leveldb/export.h" #include "leveldb/options.h" #include "leveldb/status.h" @@ -23,13 +25,16 @@ class BlockBuilder; class BlockHandle; class WritableFile; -class TableBuilder { +class LEVELDB_EXPORT 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(). TableBuilder(const Options& options, WritableFile* file); + TableBuilder(const TableBuilder&) = delete; + TableBuilder& operator=(const TableBuilder&) = delete; + // REQUIRES: Either Finish() or Abandon() has been called. ~TableBuilder(); @@ -81,10 +86,6 @@ class TableBuilder { struct Rep; Rep* rep_; - - // No copying allowed - TableBuilder(const TableBuilder&); - void operator=(const TableBuilder&); }; } // namespace leveldb diff --git a/include/leveldb/write_batch.h b/include/leveldb/write_batch.h index ee9aab6..94d4115 100644 --- a/include/leveldb/write_batch.h +++ b/include/leveldb/write_batch.h @@ -22,15 +22,29 @@ #define STORAGE_LEVELDB_INCLUDE_WRITE_BATCH_H_ #include + +#include "leveldb/export.h" #include "leveldb/status.h" namespace leveldb { class Slice; -class WriteBatch { +class LEVELDB_EXPORT WriteBatch { public: + class LEVELDB_EXPORT Handler { + public: + virtual ~Handler(); + virtual void Put(const Slice& key, const Slice& value) = 0; + virtual void Delete(const Slice& key) = 0; + }; + WriteBatch(); + + // Intentionally copyable. + WriteBatch(const WriteBatch&) = default; + WriteBatch& operator=(const WriteBatch&) = default; + ~WriteBatch(); // Store the mapping "key->value" in the database. @@ -42,21 +56,26 @@ class WriteBatch { // Clear all updates buffered in this batch. void Clear(); + // The size of the database changes caused by this batch. + // + // This number is tied to implementation details, and may change across + // releases. It is intended for LevelDB usage metrics. + size_t ApproximateSize() const; + + // Copies the operations in "source" to this batch. + // + // This runs in O(source size) time. However, the constant factor is better + // than calling Iterate() over the source batch with a Handler that replicates + // the operations into this batch. + void Append(const WriteBatch& source); + // Support for iterating over the contents of a batch. - class Handler { - public: - virtual ~Handler(); - virtual void Put(const Slice& key, const Slice& value) = 0; - virtual void Delete(const Slice& key) = 0; - }; Status Iterate(Handler* handler) const; private: friend class WriteBatchInternal; std::string rep_; // See comment in write_batch.cc for the format of rep_ - - // Intentionally copyable }; } // namespace leveldb diff --git a/issues/issue178_test.cc b/issues/issue178_test.cc index 1b1cf8b..d50ffeb 100644 --- a/issues/issue178_test.cc +++ b/issues/issue178_test.cc @@ -3,9 +3,9 @@ // found in the LICENSE file. See the AUTHORS file for names of contributors. // Test for issue 178: a manual compaction causes deleted data to reappear. +#include #include #include -#include #include "leveldb/db.h" #include "leveldb/write_batch.h" @@ -21,11 +21,9 @@ std::string Key1(int i) { return buf; } -std::string Key2(int i) { - return Key1(i) + "_xxx"; -} +std::string Key2(int i) { return Key1(i) + "_xxx"; } -class Issue178 { }; +class Issue178 {}; TEST(Issue178, Test) { // Get rid of any state from an old run. @@ -87,6 +85,4 @@ TEST(Issue178, Test) { } // anonymous namespace -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/issues/issue200_test.cc b/issues/issue200_test.cc index 1cec79f..877b2af 100644 --- a/issues/issue200_test.cc +++ b/issues/issue200_test.cc @@ -11,14 +11,14 @@ namespace leveldb { -class Issue200 { }; +class Issue200 {}; TEST(Issue200, Test) { // Get rid of any state from an old run. std::string dbpath = test::TmpDir() + "/leveldb_issue200_test"; DestroyDB(dbpath, Options()); - DB *db; + DB* db; Options options; options.create_if_missing = true; ASSERT_OK(DB::Open(options, dbpath, &db)); @@ -31,7 +31,7 @@ TEST(Issue200, Test) { ASSERT_OK(db->Put(write_options, "5", "f")); ReadOptions read_options; - Iterator *iter = db->NewIterator(read_options); + Iterator* iter = db->NewIterator(read_options); // Add an element that should not be reflected in the iterator. ASSERT_OK(db->Put(write_options, "25", "cd")); @@ -54,6 +54,4 @@ TEST(Issue200, Test) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/issues/issue320_test.cc b/issues/issue320_test.cc new file mode 100644 index 0000000..c5fcbfc --- /dev/null +++ b/issues/issue320_test.cc @@ -0,0 +1,128 @@ +// Copyright (c) 2019 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 "leveldb/db.h" +#include "leveldb/write_batch.h" +#include "util/testharness.h" + +namespace leveldb { + +namespace { + +// Creates a random number in the range of [0, max). +int GenerateRandomNumber(int max) { return std::rand() % max; } + +std::string CreateRandomString(int32_t index) { + static const size_t len = 1024; + char bytes[len]; + size_t i = 0; + while (i < 8) { + bytes[i] = 'a' + ((index >> (4 * i)) & 0xf); + ++i; + } + while (i < sizeof(bytes)) { + bytes[i] = 'a' + GenerateRandomNumber(26); + ++i; + } + return std::string(bytes, sizeof(bytes)); +} + +} // namespace + +class Issue320 {}; + +TEST(Issue320, Test) { + std::srand(0); + + bool delete_before_put = false; + bool keep_snapshots = true; + + std::vector>> test_map( + 10000); + std::vector snapshots(100, nullptr); + + DB* db; + Options options; + options.create_if_missing = true; + + std::string dbpath = test::TmpDir() + "/leveldb_issue320_test"; + ASSERT_OK(DB::Open(options, dbpath, &db)); + + uint32_t target_size = 10000; + uint32_t num_items = 0; + uint32_t count = 0; + std::string key; + std::string value, old_value; + + WriteOptions writeOptions; + ReadOptions readOptions; + while (count < 200000) { + if ((++count % 1000) == 0) { + std::cout << "count: " << count << std::endl; + } + + int index = GenerateRandomNumber(test_map.size()); + WriteBatch batch; + + if (test_map[index] == nullptr) { + num_items++; + test_map[index].reset(new std::pair( + CreateRandomString(index), CreateRandomString(index))); + batch.Put(test_map[index]->first, test_map[index]->second); + } else { + ASSERT_OK(db->Get(readOptions, test_map[index]->first, &old_value)); + if (old_value != test_map[index]->second) { + std::cout << "ERROR incorrect value returned by Get" << std::endl; + std::cout << " count=" << count << std::endl; + std::cout << " old value=" << old_value << std::endl; + std::cout << " test_map[index]->second=" << test_map[index]->second + << std::endl; + std::cout << " test_map[index]->first=" << test_map[index]->first + << std::endl; + std::cout << " index=" << index << std::endl; + ASSERT_EQ(old_value, test_map[index]->second); + } + + if (num_items >= target_size && GenerateRandomNumber(100) > 30) { + batch.Delete(test_map[index]->first); + test_map[index] = nullptr; + --num_items; + } else { + test_map[index]->second = CreateRandomString(index); + if (delete_before_put) batch.Delete(test_map[index]->first); + batch.Put(test_map[index]->first, test_map[index]->second); + } + } + + ASSERT_OK(db->Write(writeOptions, &batch)); + + if (keep_snapshots && GenerateRandomNumber(10) == 0) { + int i = GenerateRandomNumber(snapshots.size()); + if (snapshots[i] != nullptr) { + db->ReleaseSnapshot(snapshots[i]); + } + snapshots[i] = db->GetSnapshot(); + } + } + + for (Snapshot const* snapshot : snapshots) { + if (snapshot) { + db->ReleaseSnapshot(snapshot); + } + } + + delete db; + DestroyDB(dbpath, options); +} + +} // namespace leveldb + +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/port/README b/port/README.md similarity index 82% rename from port/README rename to port/README.md index 422563e..8b17153 100644 --- a/port/README +++ b/port/README.md @@ -5,6 +5,6 @@ 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 +See port_stdcxx.h for an example of what must be provided in a platform specific header file. diff --git a/port/atomic_pointer.h b/port/atomic_pointer.h deleted file mode 100644 index 0f6f05e..0000000 --- a/port/atomic_pointer.h +++ /dev/null @@ -1,233 +0,0 @@ -// 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. - -// AtomicPointer provides storage for a lock-free pointer. -// Platform-dependent implementation of AtomicPointer: -// - If the platform provides a cheap barrier, we use it with raw pointers -// - If is present (on newer versions of gcc, it is), we use -// a -based AtomicPointer. However we prefer the memory -// barrier based version, because at least on a gcc 4.4 32-bit build -// on linux, we have encountered a buggy implementation. -// Also, some implementations are much slower than a memory-barrier -// based implementation (~16ns for based acquire-load vs. ~1ns for -// a barrier based acquire-load). -// This code is based on atomicops-internals-* in Google's perftools: -// http://code.google.com/p/google-perftools/source/browse/#svn%2Ftrunk%2Fsrc%2Fbase - -#ifndef PORT_ATOMIC_POINTER_H_ -#define PORT_ATOMIC_POINTER_H_ - -#include -#ifdef LEVELDB_ATOMIC_PRESENT -#include -#endif -#ifdef OS_WIN -#include -#endif -#ifdef OS_MACOSX -#include -#endif - -#if defined(_M_X64) || defined(__x86_64__) -#define ARCH_CPU_X86_FAMILY 1 -#elif defined(_M_IX86) || defined(__i386__) || defined(__i386) -#define ARCH_CPU_X86_FAMILY 1 -#elif defined(__ARMEL__) -#define ARCH_CPU_ARM_FAMILY 1 -#elif defined(__aarch64__) -#define ARCH_CPU_ARM64_FAMILY 1 -#elif defined(__ppc__) || defined(__powerpc__) || defined(__powerpc64__) -#define ARCH_CPU_PPC_FAMILY 1 -#endif - -namespace leveldb { -namespace port { - -// Define MemoryBarrier() if available -// Windows on x86 -#if defined(OS_WIN) && defined(COMPILER_MSVC) && defined(ARCH_CPU_X86_FAMILY) -// windows.h already provides a MemoryBarrier(void) macro -// http://msdn.microsoft.com/en-us/library/ms684208(v=vs.85).aspx -#define LEVELDB_HAVE_MEMORY_BARRIER - -// Mac OS -#elif defined(OS_MACOSX) -inline void MemoryBarrier() { - OSMemoryBarrier(); -} -#define LEVELDB_HAVE_MEMORY_BARRIER - -// Gcc on x86 -#elif defined(ARCH_CPU_X86_FAMILY) && defined(__GNUC__) -inline void MemoryBarrier() { - // See http://gcc.gnu.org/ml/gcc/2003-04/msg01180.html for a discussion on - // this idiom. Also see http://en.wikipedia.org/wiki/Memory_ordering. - __asm__ __volatile__("" : : : "memory"); -} -#define LEVELDB_HAVE_MEMORY_BARRIER - -// Sun Studio -#elif defined(ARCH_CPU_X86_FAMILY) && defined(__SUNPRO_CC) -inline void MemoryBarrier() { - // See http://gcc.gnu.org/ml/gcc/2003-04/msg01180.html for a discussion on - // this idiom. Also see http://en.wikipedia.org/wiki/Memory_ordering. - asm volatile("" : : : "memory"); -} -#define LEVELDB_HAVE_MEMORY_BARRIER - -// ARM Linux -#elif defined(ARCH_CPU_ARM_FAMILY) && defined(__linux__) -typedef void (*LinuxKernelMemoryBarrierFunc)(void); -// The Linux ARM kernel provides a highly optimized device-specific memory -// barrier function at a fixed memory address that is mapped in every -// user-level process. -// -// This beats using CPU-specific instructions which are, on single-core -// devices, un-necessary and very costly (e.g. ARMv7-A "dmb" takes more -// than 180ns on a Cortex-A8 like the one on a Nexus One). Benchmarking -// shows that the extra function call cost is completely negligible on -// multi-core devices. -// -inline void MemoryBarrier() { - (*(LinuxKernelMemoryBarrierFunc)0xffff0fa0)(); -} -#define LEVELDB_HAVE_MEMORY_BARRIER - -// ARM64 -#elif defined(ARCH_CPU_ARM64_FAMILY) -inline void MemoryBarrier() { - asm volatile("dmb sy" : : : "memory"); -} -#define LEVELDB_HAVE_MEMORY_BARRIER - -// PPC -#elif defined(ARCH_CPU_PPC_FAMILY) && defined(__GNUC__) -inline void MemoryBarrier() { - // TODO for some powerpc expert: is there a cheaper suitable variant? - // Perhaps by having separate barriers for acquire and release ops. - asm volatile("sync" : : : "memory"); -} -#define LEVELDB_HAVE_MEMORY_BARRIER - -#endif - -// AtomicPointer built using platform-specific MemoryBarrier() -#if defined(LEVELDB_HAVE_MEMORY_BARRIER) -class AtomicPointer { - private: - void* rep_; - public: - AtomicPointer() { } - explicit AtomicPointer(void* p) : rep_(p) {} - inline void* NoBarrier_Load() const { return rep_; } - inline void NoBarrier_Store(void* v) { rep_ = v; } - inline void* Acquire_Load() const { - void* result = rep_; - MemoryBarrier(); - return result; - } - inline void Release_Store(void* v) { - MemoryBarrier(); - rep_ = v; - } -}; - -// AtomicPointer based on -#elif defined(LEVELDB_ATOMIC_PRESENT) -class AtomicPointer { - private: - std::atomic rep_; - public: - AtomicPointer() { } - explicit AtomicPointer(void* v) : rep_(v) { } - inline void* Acquire_Load() const { - return rep_.load(std::memory_order_acquire); - } - inline void Release_Store(void* v) { - rep_.store(v, std::memory_order_release); - } - inline void* NoBarrier_Load() const { - return rep_.load(std::memory_order_relaxed); - } - inline void NoBarrier_Store(void* v) { - rep_.store(v, std::memory_order_relaxed); - } -}; - -// Atomic pointer based on sparc memory barriers -#elif defined(__sparcv9) && defined(__GNUC__) -class AtomicPointer { - private: - void* rep_; - public: - AtomicPointer() { } - explicit AtomicPointer(void* v) : rep_(v) { } - inline void* Acquire_Load() const { - void* val; - __asm__ __volatile__ ( - "ldx [%[rep_]], %[val] \n\t" - "membar #LoadLoad|#LoadStore \n\t" - : [val] "=r" (val) - : [rep_] "r" (&rep_) - : "memory"); - return val; - } - inline void Release_Store(void* v) { - __asm__ __volatile__ ( - "membar #LoadStore|#StoreStore \n\t" - "stx %[v], [%[rep_]] \n\t" - : - : [rep_] "r" (&rep_), [v] "r" (v) - : "memory"); - } - inline void* NoBarrier_Load() const { return rep_; } - inline void NoBarrier_Store(void* v) { rep_ = v; } -}; - -// Atomic pointer based on ia64 acq/rel -#elif defined(__ia64) && defined(__GNUC__) -class AtomicPointer { - private: - void* rep_; - public: - AtomicPointer() { } - explicit AtomicPointer(void* v) : rep_(v) { } - inline void* Acquire_Load() const { - void* val ; - __asm__ __volatile__ ( - "ld8.acq %[val] = [%[rep_]] \n\t" - : [val] "=r" (val) - : [rep_] "r" (&rep_) - : "memory" - ); - return val; - } - inline void Release_Store(void* v) { - __asm__ __volatile__ ( - "st8.rel [%[rep_]] = %[v] \n\t" - : - : [rep_] "r" (&rep_), [v] "r" (v) - : "memory" - ); - } - inline void* NoBarrier_Load() const { return rep_; } - inline void NoBarrier_Store(void* v) { rep_ = v; } -}; - -// We have neither MemoryBarrier(), nor -#else -#error Please implement AtomicPointer for this platform. - -#endif - -#undef LEVELDB_HAVE_MEMORY_BARRIER -#undef ARCH_CPU_X86_FAMILY -#undef ARCH_CPU_ARM_FAMILY -#undef ARCH_CPU_ARM64_FAMILY -#undef ARCH_CPU_PPC_FAMILY - -} // namespace port -} // namespace leveldb - -#endif // PORT_ATOMIC_POINTER_H_ diff --git a/port/port.h b/port/port.h index e667db4..4b247f7 100644 --- a/port/port.h +++ b/port/port.h @@ -10,10 +10,10 @@ // 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(LEVELDB_PLATFORM_POSIX) -# include "port/port_posix.h" +#if defined(LEVELDB_PLATFORM_POSIX) || defined(LEVELDB_PLATFORM_WINDOWS) +#include "port/port_stdcxx.h" #elif defined(LEVELDB_PLATFORM_CHROMIUM) -# include "port/port_chromium.h" +#include "port/port_chromium.h" #endif #endif // STORAGE_LEVELDB_PORT_PORT_H_ diff --git a/port/port_config.h.in b/port/port_config.h.in new file mode 100644 index 0000000..2127315 --- /dev/null +++ b/port/port_config.h.in @@ -0,0 +1,39 @@ +// Copyright 2017 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_PORT_CONFIG_H_ +#define STORAGE_LEVELDB_PORT_PORT_CONFIG_H_ + +// Define to 1 if you have a definition for fdatasync() in . +#if !defined(HAVE_FDATASYNC) +#cmakedefine01 HAVE_FDATASYNC +#endif // !defined(HAVE_FDATASYNC) + +// Define to 1 if you have a definition for F_FULLFSYNC in . +#if !defined(HAVE_FULLFSYNC) +#cmakedefine01 HAVE_FULLFSYNC +#endif // !defined(HAVE_FULLFSYNC) + +// Define to 1 if you have a definition for O_CLOEXEC in . +#if !defined(HAVE_O_CLOEXEC) +#cmakedefine01 HAVE_O_CLOEXEC +#endif // !defined(HAVE_O_CLOEXEC) + +// Define to 1 if you have Google CRC32C. +#if !defined(HAVE_CRC32C) +#cmakedefine01 HAVE_CRC32C +#endif // !defined(HAVE_CRC32C) + +// Define to 1 if you have Google Snappy. +#if !defined(HAVE_SNAPPY) +#cmakedefine01 HAVE_SNAPPY +#endif // !defined(HAVE_SNAPPY) + +// Define to 1 if your processor stores words with the most significant byte +// first (like Motorola and SPARC, unlike Intel and VAX). +#if !defined(LEVELDB_IS_BIG_ENDIAN) +#cmakedefine01 LEVELDB_IS_BIG_ENDIAN +#endif // !defined(LEVELDB_IS_BIG_ENDIAN) + +#endif // STORAGE_LEVELDB_PORT_PORT_CONFIG_H_ \ No newline at end of file diff --git a/port/port_example.h b/port/port_example.h index ab9e489..1a8fca2 100644 --- a/port/port_example.h +++ b/port/port_example.h @@ -10,6 +10,8 @@ #ifndef STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ #define STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ +#include "port/thread_annotations.h" + namespace leveldb { namespace port { @@ -23,23 +25,23 @@ static const bool kLittleEndian = true /* or some other expression */; // ------------------ Threading ------------------- // A Mutex represents an exclusive lock. -class Mutex { +class LOCKABLE 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(); + void Lock() EXCLUSIVE_LOCK_FUNCTION(); // Unlock the mutex. // REQUIRES: This mutex was locked by this thread. - void Unlock(); + void Unlock() UNLOCK_FUNCTION(); // 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(); + void AssertHeld() ASSERT_EXCLUSIVE_LOCK(); }; class CondVar { @@ -60,57 +62,18 @@ class CondVar { 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)()); - -// A type that holds a pointer that can be read or written atomically -// (i.e., without word-tearing.) -class AtomicPointer { - private: - intptr_t rep_; - public: - // Initialize to arbitrary value - AtomicPointer(); - - // Initialize to hold v - explicit AtomicPointer(void* v) : rep_(v) { } - - // Read and return the stored pointer with the guarantee that no - // later memory access (read or write) by this thread can be - // reordered ahead of this read. - void* Acquire_Load() const; - - // Set v as the stored pointer with the guarantee that no earlier - // memory access (read or write) by this thread can be reordered - // after this store. - void Release_Store(void* v); - - // Read the stored pointer with no ordering guarantees. - void* NoBarrier_Load() const; - - // Set va as the stored pointer with no ordering guarantees. - void NoBarrier_Store(void* v); -}; - // ------------------ 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); +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); +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 @@ -119,15 +82,21 @@ extern bool Snappy_GetUncompressedLength(const char* input, size_t length, // 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); +bool Snappy_Uncompress(const char* input_data, size_t input_length, + char* output); // ------------------ Miscellaneous ------------------- // If heap profiling is not supported, returns false. // Else repeatedly calls (*func)(arg, data, n) and then returns true. // The concatenation of all "data[0,n-1]" fragments is the heap profile. -extern bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg); +bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg); + +// Extend the CRC to include the first n bytes of buf. +// +// Returns zero if the CRC cannot be extended using acceleration, else returns +// the newly extended CRC value (which may also be zero). +uint32_t AcceleratedCRC32C(uint32_t crc, const char* buf, size_t size); } // namespace port } // namespace leveldb diff --git a/port/port_posix.cc b/port/port_posix.cc deleted file mode 100644 index 5ba127a..0000000 --- a/port/port_posix.cc +++ /dev/null @@ -1,54 +0,0 @@ -// 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 -#include -#include -#include "util/logging.h" - -namespace leveldb { -namespace port { - -static void PthreadCall(const char* label, int result) { - if (result != 0) { - fprintf(stderr, "pthread %s: %s\n", label, strerror(result)); - abort(); - } -} - -Mutex::Mutex() { PthreadCall("init mutex", pthread_mutex_init(&mu_, NULL)); } - -Mutex::~Mutex() { PthreadCall("destroy mutex", pthread_mutex_destroy(&mu_)); } - -void Mutex::Lock() { PthreadCall("lock", pthread_mutex_lock(&mu_)); } - -void Mutex::Unlock() { PthreadCall("unlock", pthread_mutex_unlock(&mu_)); } - -CondVar::CondVar(Mutex* mu) - : mu_(mu) { - PthreadCall("init cv", pthread_cond_init(&cv_, NULL)); -} - -CondVar::~CondVar() { PthreadCall("destroy cv", pthread_cond_destroy(&cv_)); } - -void CondVar::Wait() { - PthreadCall("wait", pthread_cond_wait(&cv_, &mu_->mu_)); -} - -void CondVar::Signal() { - PthreadCall("signal", pthread_cond_signal(&cv_)); -} - -void CondVar::SignalAll() { - PthreadCall("broadcast", pthread_cond_broadcast(&cv_)); -} - -void InitOnce(OnceType* once, void (*initializer)()) { - PthreadCall("once", pthread_once(once, initializer)); -} - -} // namespace port -} // namespace leveldb diff --git a/port/port_posix.h b/port/port_posix.h deleted file mode 100644 index 89fc222..0000000 --- a/port/port_posix.h +++ /dev/null @@ -1,154 +0,0 @@ -// 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_POSIX_H_ -#define STORAGE_LEVELDB_PORT_PORT_POSIX_H_ - -#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) - #include - #include - #define PLATFORM_IS_LITTLE_ENDIAN (_BYTE_ORDER == _LITTLE_ENDIAN) -#elif defined(OS_HPUX) - #define PLATFORM_IS_LITTLE_ENDIAN false -#elif defined(OS_ANDROID) - // Due to a bug in the NDK x86 definition, - // _BYTE_ORDER must be used instead of __BYTE_ORDER on Android. - // See http://code.google.com/p/android/issues/detail?id=39824 - #include - #define PLATFORM_IS_LITTLE_ENDIAN (_BYTE_ORDER == _LITTLE_ENDIAN) -#else - #include -#endif - -#include -#ifdef SNAPPY -#include -#endif -#include -#include -#include "port/atomic_pointer.h" - -#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(OS_HPUX) || 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 targetting older platforms. -#define fdatasync fsync -#endif - -namespace leveldb { -namespace port { - -static const bool kLittleEndian = PLATFORM_IS_LITTLE_ENDIAN; -#undef PLATFORM_IS_LITTLE_ENDIAN - -class CondVar; - -class Mutex { - public: - Mutex(); - ~Mutex(); - - void Lock(); - void Unlock(); - void AssertHeld() { } - - private: - friend class CondVar; - pthread_mutex_t mu_; - - // No copying - Mutex(const Mutex&); - void operator=(const Mutex&); -}; - -class CondVar { - public: - explicit CondVar(Mutex* mu); - ~CondVar(); - void Wait(); - void Signal(); - void SignalAll(); - private: - pthread_cond_t cv_; - Mutex* mu_; -}; - -typedef pthread_once_t OnceType; -#define LEVELDB_ONCE_INIT PTHREAD_ONCE_INIT -extern void InitOnce(OnceType* once, void (*initializer)()); - -inline bool Snappy_Compress(const char* input, size_t length, - ::std::string* output) { -#ifdef SNAPPY - output->resize(snappy::MaxCompressedLength(length)); - size_t outlen; - snappy::RawCompress(input, length, &(*output)[0], &outlen); - output->resize(outlen); - return true; -#endif - - return false; -} - -inline bool Snappy_GetUncompressedLength(const char* input, size_t length, - size_t* result) { -#ifdef SNAPPY - return snappy::GetUncompressedLength(input, length, result); -#else - return false; -#endif -} - -inline bool Snappy_Uncompress(const char* input, size_t length, - char* output) { -#ifdef SNAPPY - return snappy::RawUncompress(input, length, output); -#else - return false; -#endif -} - -inline bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg) { - return false; -} - -} // namespace port -} // namespace leveldb - -#endif // STORAGE_LEVELDB_PORT_PORT_POSIX_H_ diff --git a/port/port_stdcxx.h b/port/port_stdcxx.h new file mode 100644 index 0000000..e9cb0e5 --- /dev/null +++ b/port/port_stdcxx.h @@ -0,0 +1,153 @@ +// Copyright (c) 2018 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_PORT_STDCXX_H_ +#define STORAGE_LEVELDB_PORT_PORT_STDCXX_H_ + +// port/port_config.h availability is automatically detected via __has_include +// in newer compilers. If LEVELDB_HAS_PORT_CONFIG_H is defined, it overrides the +// configuration detection. +#if defined(LEVELDB_HAS_PORT_CONFIG_H) + +#if LEVELDB_HAS_PORT_CONFIG_H +#include "port/port_config.h" +#endif // LEVELDB_HAS_PORT_CONFIG_H + +#elif defined(__has_include) + +#if __has_include("port/port_config.h") +#include "port/port_config.h" +#endif // __has_include("port/port_config.h") + +#endif // defined(LEVELDB_HAS_PORT_CONFIG_H) + +#if HAVE_CRC32C +#include +#endif // HAVE_CRC32C +#if HAVE_SNAPPY +#include +#endif // HAVE_SNAPPY + +#include +#include // NOLINT +#include +#include +#include // NOLINT +#include + +#include "port/thread_annotations.h" + +namespace leveldb { +namespace port { + +static const bool kLittleEndian = !LEVELDB_IS_BIG_ENDIAN; + +class CondVar; + +// Thinly wraps std::mutex. +class LOCKABLE Mutex { + public: + Mutex() = default; + ~Mutex() = default; + + Mutex(const Mutex&) = delete; + Mutex& operator=(const Mutex&) = delete; + + void Lock() EXCLUSIVE_LOCK_FUNCTION() { mu_.lock(); } + void Unlock() UNLOCK_FUNCTION() { mu_.unlock(); } + void AssertHeld() ASSERT_EXCLUSIVE_LOCK() {} + + private: + friend class CondVar; + std::mutex mu_; +}; + +// Thinly wraps std::condition_variable. +class CondVar { + public: + explicit CondVar(Mutex* mu) : mu_(mu) { assert(mu != nullptr); } + ~CondVar() = default; + + CondVar(const CondVar&) = delete; + CondVar& operator=(const CondVar&) = delete; + + void Wait() { + std::unique_lock lock(mu_->mu_, std::adopt_lock); + cv_.wait(lock); + lock.release(); + } + void Signal() { cv_.notify_one(); } + void SignalAll() { cv_.notify_all(); } + + private: + std::condition_variable cv_; + Mutex* const mu_; +}; + +inline bool Snappy_Compress(const char* input, size_t length, + std::string* output) { +#if HAVE_SNAPPY + output->resize(snappy::MaxCompressedLength(length)); + size_t outlen; + snappy::RawCompress(input, length, &(*output)[0], &outlen); + output->resize(outlen); + return true; +#else + // Silence compiler warnings about unused arguments. + (void)input; + (void)length; + (void)output; +#endif // HAVE_SNAPPY + + return false; +} + +inline bool Snappy_GetUncompressedLength(const char* input, size_t length, + size_t* result) { +#if HAVE_SNAPPY + return snappy::GetUncompressedLength(input, length, result); +#else + // Silence compiler warnings about unused arguments. + (void)input; + (void)length; + (void)result; + return false; +#endif // HAVE_SNAPPY +} + +inline bool Snappy_Uncompress(const char* input, size_t length, char* output) { +#if HAVE_SNAPPY + return snappy::RawUncompress(input, length, output); +#else + // Silence compiler warnings about unused arguments. + (void)input; + (void)length; + (void)output; + return false; +#endif // HAVE_SNAPPY +} + +inline bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg) { + // Silence compiler warnings about unused arguments. + (void)func; + (void)arg; + return false; +} + +inline uint32_t AcceleratedCRC32C(uint32_t crc, const char* buf, size_t size) { +#if HAVE_CRC32C + return ::crc32c::Extend(crc, reinterpret_cast(buf), size); +#else + // Silence compiler warnings about unused arguments. + (void)crc; + (void)buf; + (void)size; + return 0; +#endif // HAVE_CRC32C +} + +} // namespace port +} // namespace leveldb + +#endif // STORAGE_LEVELDB_PORT_PORT_STDCXX_H_ diff --git a/port/thread_annotations.h b/port/thread_annotations.h index 9470ef5..1547df9 100644 --- a/port/thread_annotations.h +++ b/port/thread_annotations.h @@ -5,56 +5,104 @@ #ifndef STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_ #define STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_ -// Some environments provide custom macros to aid in static thread-safety -// analysis. Provide empty definitions of such macros unless they are already -// defined. +// Use Clang's thread safety analysis annotations when available. In other +// environments, the macros receive empty definitions. +// Usage documentation: https://clang.llvm.org/docs/ThreadSafetyAnalysis.html + +#if !defined(THREAD_ANNOTATION_ATTRIBUTE__) + +#if defined(__clang__) + +#define THREAD_ANNOTATION_ATTRIBUTE__(x) __attribute__((x)) +#else +#define THREAD_ANNOTATION_ATTRIBUTE__(x) // no-op +#endif + +#endif // !defined(THREAD_ANNOTATION_ATTRIBUTE__) + +#ifndef GUARDED_BY +#define GUARDED_BY(x) THREAD_ANNOTATION_ATTRIBUTE__(guarded_by(x)) +#endif + +#ifndef PT_GUARDED_BY +#define PT_GUARDED_BY(x) THREAD_ANNOTATION_ATTRIBUTE__(pt_guarded_by(x)) +#endif + +#ifndef ACQUIRED_AFTER +#define ACQUIRED_AFTER(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(acquired_after(__VA_ARGS__)) +#endif + +#ifndef ACQUIRED_BEFORE +#define ACQUIRED_BEFORE(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(acquired_before(__VA_ARGS__)) +#endif #ifndef EXCLUSIVE_LOCKS_REQUIRED -#define EXCLUSIVE_LOCKS_REQUIRED(...) +#define EXCLUSIVE_LOCKS_REQUIRED(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(exclusive_locks_required(__VA_ARGS__)) #endif #ifndef SHARED_LOCKS_REQUIRED -#define SHARED_LOCKS_REQUIRED(...) +#define SHARED_LOCKS_REQUIRED(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(shared_locks_required(__VA_ARGS__)) #endif #ifndef LOCKS_EXCLUDED -#define LOCKS_EXCLUDED(...) +#define LOCKS_EXCLUDED(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(locks_excluded(__VA_ARGS__)) #endif #ifndef LOCK_RETURNED -#define LOCK_RETURNED(x) +#define LOCK_RETURNED(x) THREAD_ANNOTATION_ATTRIBUTE__(lock_returned(x)) #endif #ifndef LOCKABLE -#define LOCKABLE +#define LOCKABLE THREAD_ANNOTATION_ATTRIBUTE__(lockable) #endif #ifndef SCOPED_LOCKABLE -#define SCOPED_LOCKABLE +#define SCOPED_LOCKABLE THREAD_ANNOTATION_ATTRIBUTE__(scoped_lockable) #endif #ifndef EXCLUSIVE_LOCK_FUNCTION -#define EXCLUSIVE_LOCK_FUNCTION(...) +#define EXCLUSIVE_LOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(exclusive_lock_function(__VA_ARGS__)) #endif #ifndef SHARED_LOCK_FUNCTION -#define SHARED_LOCK_FUNCTION(...) +#define SHARED_LOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(shared_lock_function(__VA_ARGS__)) #endif #ifndef EXCLUSIVE_TRYLOCK_FUNCTION -#define EXCLUSIVE_TRYLOCK_FUNCTION(...) +#define EXCLUSIVE_TRYLOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(exclusive_trylock_function(__VA_ARGS__)) #endif #ifndef SHARED_TRYLOCK_FUNCTION -#define SHARED_TRYLOCK_FUNCTION(...) +#define SHARED_TRYLOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(shared_trylock_function(__VA_ARGS__)) #endif #ifndef UNLOCK_FUNCTION -#define UNLOCK_FUNCTION(...) +#define UNLOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(unlock_function(__VA_ARGS__)) #endif #ifndef NO_THREAD_SAFETY_ANALYSIS -#define NO_THREAD_SAFETY_ANALYSIS +#define NO_THREAD_SAFETY_ANALYSIS \ + THREAD_ANNOTATION_ATTRIBUTE__(no_thread_safety_analysis) +#endif + +#ifndef ASSERT_EXCLUSIVE_LOCK +#define ASSERT_EXCLUSIVE_LOCK(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(assert_exclusive_lock(__VA_ARGS__)) +#endif + +#ifndef ASSERT_SHARED_LOCK +#define ASSERT_SHARED_LOCK(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(assert_shared_lock(__VA_ARGS__)) #endif #endif // STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_ diff --git a/port/win/stdint.h b/port/win/stdint.h deleted file mode 100644 index 39edd0d..0000000 --- a/port/win/stdint.h +++ /dev/null @@ -1,24 +0,0 @@ -// 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. - -// MSVC didn't ship with this file until the 2010 version. - -#ifndef STORAGE_LEVELDB_PORT_WIN_STDINT_H_ -#define STORAGE_LEVELDB_PORT_WIN_STDINT_H_ - -#if !defined(_MSC_VER) -#error This file should only be included when compiling with MSVC. -#endif - -// Define C99 equivalent types. -typedef signed char int8_t; -typedef signed short int16_t; -typedef signed int int32_t; -typedef signed long long int64_t; -typedef unsigned char uint8_t; -typedef unsigned short uint16_t; -typedef unsigned int uint32_t; -typedef unsigned long long uint64_t; - -#endif // STORAGE_LEVELDB_PORT_WIN_STDINT_H_ diff --git a/table/block.cc b/table/block.cc index 43e402c..05c600f 100644 --- a/table/block.cc +++ b/table/block.cc @@ -6,8 +6,9 @@ #include "table/block.h" -#include #include +#include + #include "leveldb/comparator.h" #include "table/format.h" #include "util/coding.h" @@ -27,7 +28,7 @@ Block::Block(const BlockContents& contents) if (size_ < sizeof(uint32_t)) { size_ = 0; // Error marker } else { - size_t max_restarts_allowed = (size_-sizeof(uint32_t)) / sizeof(uint32_t); + size_t max_restarts_allowed = (size_ - sizeof(uint32_t)) / sizeof(uint32_t); if (NumRestarts() > max_restarts_allowed) { // The size is too small for NumRestarts() size_ = 0; @@ -48,13 +49,12 @@ Block::~Block() { // and the length of the value in "*shared", "*non_shared", and // "*value_length", respectively. Will not dereference past "limit". // -// If any errors are detected, returns NULL. Otherwise, returns a +// 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* shared, uint32_t* non_shared, uint32_t* value_length) { - if (limit - p < 3) return NULL; + if (limit - p < 3) return nullptr; *shared = reinterpret_cast(p)[0]; *non_shared = reinterpret_cast(p)[1]; *value_length = reinterpret_cast(p)[2]; @@ -62,13 +62,13 @@ static inline const char* DecodeEntry(const char* p, const char* limit, // Fast path: all three values are encoded in one byte each p += 3; } else { - if ((p = GetVarint32Ptr(p, limit, shared)) == NULL) return NULL; - if ((p = GetVarint32Ptr(p, limit, non_shared)) == NULL) return NULL; - if ((p = GetVarint32Ptr(p, limit, value_length)) == NULL) return NULL; + 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 NULL; + return nullptr; } return p; } @@ -76,9 +76,9 @@ static inline const char* DecodeEntry(const char* p, const char* limit, class Block::Iter : public Iterator { private: const Comparator* const comparator_; - const char* const data_; // underlying block contents - uint32_t const restarts_; // Offset of restart array (list of fixed32) - uint32_t const num_restarts_; // Number of uint32_t entries in restart array + const char* const data_; // underlying block contents + uint32_t const restarts_; // Offset of restart array (list of fixed32) + uint32_t const num_restarts_; // Number of uint32_t entries in restart array // current_ is offset in data_ of current entry. >= restarts_ if !Valid uint32_t current_; @@ -112,9 +112,7 @@ class Block::Iter : public Iterator { } public: - Iter(const Comparator* comparator, - const char* data, - uint32_t restarts, + Iter(const Comparator* comparator, const char* data, uint32_t restarts, uint32_t num_restarts) : comparator_(comparator), data_(data), @@ -125,23 +123,23 @@ class Block::Iter : public Iterator { assert(num_restarts_ > 0); } - virtual bool Valid() const { return current_ < restarts_; } - virtual Status status() const { return status_; } - virtual Slice key() const { + bool Valid() const override { return current_ < restarts_; } + Status status() const override { return status_; } + Slice key() const override { assert(Valid()); return key_; } - virtual Slice value() const { + Slice value() const override { assert(Valid()); return value_; } - virtual void Next() { + void Next() override { assert(Valid()); ParseNextKey(); } - virtual void Prev() { + void Prev() override { assert(Valid()); // Scan backwards to a restart point before current_ @@ -162,7 +160,7 @@ class Block::Iter : public Iterator { } while (ParseNextKey() && NextEntryOffset() < original); } - virtual void Seek(const Slice& target) { + void Seek(const Slice& target) override { // Binary search in restart array to find the last restart point // with a key < target uint32_t left = 0; @@ -171,10 +169,10 @@ class Block::Iter : public Iterator { 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 == NULL || (shared != 0)) { + const char* key_ptr = + DecodeEntry(data_ + region_offset, data_ + restarts_, &shared, + &non_shared, &value_length); + if (key_ptr == nullptr || (shared != 0)) { CorruptionError(); return; } @@ -202,12 +200,12 @@ class Block::Iter : public Iterator { } } - virtual void SeekToFirst() { + void SeekToFirst() override { SeekToRestartPoint(0); ParseNextKey(); } - virtual void SeekToLast() { + void SeekToLast() override { SeekToRestartPoint(num_restarts_ - 1); while (ParseNextKey() && NextEntryOffset() < restarts_) { // Keep skipping @@ -237,7 +235,7 @@ class Block::Iter : public Iterator { // Decode next entry uint32_t shared, non_shared, value_length; p = DecodeEntry(p, limit, &shared, &non_shared, &value_length); - if (p == NULL || key_.size() < shared) { + if (p == nullptr || key_.size() < shared) { CorruptionError(); return false; } else { @@ -253,7 +251,7 @@ class Block::Iter : public Iterator { } }; -Iterator* Block::NewIterator(const Comparator* cmp) { +Iterator* Block::NewIterator(const Comparator* comparator) { if (size_ < sizeof(uint32_t)) { return NewErrorIterator(Status::Corruption("bad block contents")); } @@ -261,7 +259,7 @@ Iterator* Block::NewIterator(const Comparator* cmp) { if (num_restarts == 0) { return NewEmptyIterator(); } else { - return new Iter(cmp, data_, restart_offset_, num_restarts); + return new Iter(comparator, data_, restart_offset_, num_restarts); } } diff --git a/table/block.h b/table/block.h index 2493eb9..c8f1f7b 100644 --- a/table/block.h +++ b/table/block.h @@ -7,6 +7,7 @@ #include #include + #include "leveldb/iterator.h" namespace leveldb { @@ -19,24 +20,23 @@ class Block { // Initialize the block with the specified contents. explicit Block(const BlockContents& contents); + Block(const Block&) = delete; + Block& operator=(const Block&) = delete; + ~Block(); size_t size() const { return size_; } Iterator* NewIterator(const Comparator* comparator); private: + class Iter; + uint32_t NumRestarts() const; const char* data_; size_t size_; - uint32_t restart_offset_; // Offset in data_ of restart array - bool owned_; // Block owns data_[] - - // No copying allowed - Block(const Block&); - void operator=(const Block&); - - class Iter; + uint32_t restart_offset_; // Offset in data_ of restart array + bool owned_; // Block owns data_[] }; } // namespace leveldb diff --git a/table/block_builder.cc b/table/block_builder.cc index db660cd..919cff5 100644 --- a/table/block_builder.cc +++ b/table/block_builder.cc @@ -28,36 +28,35 @@ #include "table/block_builder.h" -#include #include + +#include + #include "leveldb/comparator.h" -#include "leveldb/table_builder.h" +#include "leveldb/options.h" #include "util/coding.h" namespace leveldb { BlockBuilder::BlockBuilder(const Options* options) - : options_(options), - restarts_(), - counter_(0), - finished_(false) { + : options_(options), restarts_(), counter_(0), finished_(false) { assert(options->block_restart_interval >= 1); - restarts_.push_back(0); // First restart point is at offset 0 + 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 + 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 + return (buffer_.size() + // Raw data buffer + restarts_.size() * sizeof(uint32_t) + // Restart array + sizeof(uint32_t)); // Restart array length } Slice BlockBuilder::Finish() { @@ -74,7 +73,7 @@ void BlockBuilder::Add(const Slice& key, const Slice& value) { Slice last_key_piece(last_key_); assert(!finished_); assert(counter_ <= options_->block_restart_interval); - assert(buffer_.empty() // No values yet? + assert(buffer_.empty() // No values yet? || options_->comparator->Compare(key, last_key_piece) > 0); size_t shared = 0; if (counter_ < options_->block_restart_interval) { diff --git a/table/block_builder.h b/table/block_builder.h index 4fbcb33..f91f5e6 100644 --- a/table/block_builder.h +++ b/table/block_builder.h @@ -5,9 +5,10 @@ #ifndef STORAGE_LEVELDB_TABLE_BLOCK_BUILDER_H_ #define STORAGE_LEVELDB_TABLE_BLOCK_BUILDER_H_ +#include + #include -#include #include "leveldb/slice.h" namespace leveldb { @@ -18,6 +19,9 @@ class BlockBuilder { public: explicit BlockBuilder(const Options* options); + BlockBuilder(const BlockBuilder&) = delete; + BlockBuilder& operator=(const BlockBuilder&) = delete; + // Reset the contents as if the BlockBuilder was just constructed. void Reset(); @@ -35,21 +39,15 @@ class BlockBuilder { size_t CurrentSizeEstimate() const; // Return true iff no entries have been added since the last Reset() - bool empty() const { - return buffer_.empty(); - } + bool empty() const { return buffer_.empty(); } private: - const Options* options_; - 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_; - - // No copying allowed - BlockBuilder(const BlockBuilder&); - void operator=(const BlockBuilder&); + const Options* options_; + 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 leveldb diff --git a/table/filter_block.cc b/table/filter_block.cc index 203e15c..09ec009 100644 --- a/table/filter_block.cc +++ b/table/filter_block.cc @@ -9,15 +9,14 @@ namespace leveldb { -// See doc/table_format.txt for an explanation of the filter block format. +// See doc/table_format.md 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; FilterBlockBuilder::FilterBlockBuilder(const FilterPolicy* policy) - : policy_(policy) { -} + : policy_(policy) {} void FilterBlockBuilder::StartBlock(uint64_t block_offset) { uint64_t filter_index = (block_offset / kFilterBase); @@ -62,13 +61,13 @@ void FilterBlockBuilder::GenerateFilter() { tmp_keys_.resize(num_keys); for (size_t i = 0; i < num_keys; i++) { const char* base = keys_.data() + start_[i]; - size_t length = start_[i+1] - start_[i]; + size_t length = start_[i + 1] - start_[i]; tmp_keys_[i] = Slice(base, length); } // Generate filter for current set of keys and append to result_. filter_offsets_.push_back(result_.size()); - policy_->CreateFilter(&tmp_keys_[0], num_keys, &result_); + policy_->CreateFilter(&tmp_keys_[0], static_cast(num_keys), &result_); tmp_keys_.clear(); keys_.clear(); @@ -77,14 +76,10 @@ void FilterBlockBuilder::GenerateFilter() { FilterBlockReader::FilterBlockReader(const FilterPolicy* policy, const Slice& contents) - : policy_(policy), - data_(NULL), - offset_(NULL), - num_(0), - base_lg_(0) { + : policy_(policy), data_(nullptr), offset_(nullptr), num_(0), base_lg_(0) { size_t n = contents.size(); if (n < 5) return; // 1 byte for base_lg_ and 4 for start of offset array - base_lg_ = contents[n-1]; + base_lg_ = contents[n - 1]; uint32_t last_word = DecodeFixed32(contents.data() + n - 5); if (last_word > n - 5) return; data_ = contents.data(); @@ -95,9 +90,9 @@ FilterBlockReader::FilterBlockReader(const FilterPolicy* policy, bool FilterBlockReader::KeyMayMatch(uint64_t block_offset, const Slice& key) { 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 <= (offset_ - data_)) { + uint32_t start = DecodeFixed32(offset_ + index * 4); + uint32_t limit = DecodeFixed32(offset_ + index * 4 + 4); + if (start <= limit && limit <= static_cast(offset_ - data_)) { Slice filter = Slice(data_ + start, limit - start); return policy_->KeyMayMatch(key, filter); } else if (start == limit) { @@ -108,4 +103,4 @@ bool FilterBlockReader::KeyMayMatch(uint64_t block_offset, const Slice& key) { return true; // Errors are treated as potential matches } -} +} // namespace leveldb diff --git a/table/filter_block.h b/table/filter_block.h index c67d010..73b5399 100644 --- a/table/filter_block.h +++ b/table/filter_block.h @@ -11,8 +11,10 @@ #include #include + #include #include + #include "leveldb/slice.h" #include "util/hash.h" @@ -30,6 +32,9 @@ class FilterBlockBuilder { public: explicit FilterBlockBuilder(const FilterPolicy*); + FilterBlockBuilder(const FilterBlockBuilder&) = delete; + FilterBlockBuilder& operator=(const FilterBlockBuilder&) = delete; + void StartBlock(uint64_t block_offset); void AddKey(const Slice& key); Slice Finish(); @@ -38,20 +43,16 @@ class FilterBlockBuilder { void GenerateFilter(); const FilterPolicy* policy_; - std::string keys_; // Flattened key contents - std::vector start_; // Starting index in keys_ of each key - std::string result_; // Filter data computed so far - std::vector tmp_keys_; // policy_->CreateFilter() argument + std::string keys_; // Flattened key contents + std::vector start_; // Starting index in keys_ of each key + std::string result_; // Filter data computed so far + std::vector tmp_keys_; // policy_->CreateFilter() argument std::vector filter_offsets_; - - // No copying allowed - FilterBlockBuilder(const FilterBlockBuilder&); - void operator=(const FilterBlockBuilder&); }; class FilterBlockReader { public: - // REQUIRES: "contents" and *policy must stay live while *this is live. + // REQUIRES: "contents" and *policy must stay live while *this is live. FilterBlockReader(const FilterPolicy* policy, const Slice& contents); bool KeyMayMatch(uint64_t block_offset, const Slice& key); @@ -63,6 +64,6 @@ class FilterBlockReader { size_t base_lg_; // Encoding parameter (see kFilterBaseLg in .cc file) }; -} +} // namespace leveldb #endif // STORAGE_LEVELDB_TABLE_FILTER_BLOCK_H_ diff --git a/table/filter_block_test.cc b/table/filter_block_test.cc index 8c4a474..8b33bbd 100644 --- a/table/filter_block_test.cc +++ b/table/filter_block_test.cc @@ -16,18 +16,16 @@ namespace leveldb { // For testing: emit an array with one hash value per key class TestHashFilter : public FilterPolicy { public: - virtual const char* Name() const { - return "TestHashFilter"; - } + const char* Name() const override { return "TestHashFilter"; } - virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const { + 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 { + bool KeyMayMatch(const Slice& key, const Slice& filter) const override { uint32_t h = Hash(key.data(), key.size(), 1); for (size_t i = 0; i + 4 <= filter.size(); i += 4) { if (h == DecodeFixed32(filter.data() + i)) { @@ -69,8 +67,8 @@ TEST(FilterBlockTest, SingleChunk) { ASSERT_TRUE(reader.KeyMayMatch(100, "box")); ASSERT_TRUE(reader.KeyMayMatch(100, "hello")); ASSERT_TRUE(reader.KeyMayMatch(100, "foo")); - ASSERT_TRUE(! reader.KeyMayMatch(100, "missing")); - ASSERT_TRUE(! reader.KeyMayMatch(100, "other")); + ASSERT_TRUE(!reader.KeyMayMatch(100, "missing")); + ASSERT_TRUE(!reader.KeyMayMatch(100, "other")); } TEST(FilterBlockTest, MultiChunk) { @@ -99,30 +97,28 @@ TEST(FilterBlockTest, MultiChunk) { // Check first filter ASSERT_TRUE(reader.KeyMayMatch(0, "foo")); ASSERT_TRUE(reader.KeyMayMatch(2000, "bar")); - ASSERT_TRUE(! reader.KeyMayMatch(0, "box")); - ASSERT_TRUE(! reader.KeyMayMatch(0, "hello")); + ASSERT_TRUE(!reader.KeyMayMatch(0, "box")); + ASSERT_TRUE(!reader.KeyMayMatch(0, "hello")); // Check second filter ASSERT_TRUE(reader.KeyMayMatch(3100, "box")); - ASSERT_TRUE(! reader.KeyMayMatch(3100, "foo")); - ASSERT_TRUE(! reader.KeyMayMatch(3100, "bar")); - ASSERT_TRUE(! reader.KeyMayMatch(3100, "hello")); + ASSERT_TRUE(!reader.KeyMayMatch(3100, "foo")); + ASSERT_TRUE(!reader.KeyMayMatch(3100, "bar")); + ASSERT_TRUE(!reader.KeyMayMatch(3100, "hello")); // Check third filter (empty) - ASSERT_TRUE(! reader.KeyMayMatch(4100, "foo")); - ASSERT_TRUE(! reader.KeyMayMatch(4100, "bar")); - ASSERT_TRUE(! reader.KeyMayMatch(4100, "box")); - ASSERT_TRUE(! reader.KeyMayMatch(4100, "hello")); + ASSERT_TRUE(!reader.KeyMayMatch(4100, "foo")); + ASSERT_TRUE(!reader.KeyMayMatch(4100, "bar")); + ASSERT_TRUE(!reader.KeyMayMatch(4100, "box")); + ASSERT_TRUE(!reader.KeyMayMatch(4100, "hello")); // Check last filter ASSERT_TRUE(reader.KeyMayMatch(9000, "box")); ASSERT_TRUE(reader.KeyMayMatch(9000, "hello")); - ASSERT_TRUE(! reader.KeyMayMatch(9000, "foo")); - ASSERT_TRUE(! reader.KeyMayMatch(9000, "bar")); + ASSERT_TRUE(!reader.KeyMayMatch(9000, "foo")); + ASSERT_TRUE(!reader.KeyMayMatch(9000, "bar")); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/table/format.cc b/table/format.cc index aa63144..e183977 100644 --- a/table/format.cc +++ b/table/format.cc @@ -21,8 +21,7 @@ void BlockHandle::EncodeTo(std::string* dst) const { } Status BlockHandle::DecodeFrom(Slice* input) { - if (GetVarint64(input, &offset_) && - GetVarint64(input, &size_)) { + if (GetVarint64(input, &offset_) && GetVarint64(input, &size_)) { return Status::OK(); } else { return Status::Corruption("bad block handle"); @@ -30,15 +29,14 @@ Status BlockHandle::DecodeFrom(Slice* input) { } void Footer::EncodeTo(std::string* dst) const { -#ifndef NDEBUG const size_t original_size = dst->size(); -#endif metaindex_handle_.EncodeTo(dst); index_handle_.EncodeTo(dst); dst->resize(2 * BlockHandle::kMaxEncodedLength); // Padding PutFixed32(dst, static_cast(kTableMagicNumber & 0xffffffffu)); PutFixed32(dst, static_cast(kTableMagicNumber >> 32)); assert(dst->size() == original_size + kEncodedLength); + (void)original_size; // Disable unused variable warning. } Status Footer::DecodeFrom(Slice* input) { @@ -63,10 +61,8 @@ Status Footer::DecodeFrom(Slice* input) { return result; } -Status ReadBlock(RandomAccessFile* file, - const ReadOptions& options, - const BlockHandle& handle, - BlockContents* result) { +Status ReadBlock(RandomAccessFile* file, const ReadOptions& options, + const BlockHandle& handle, BlockContents* result) { result->data = Slice(); result->cachable = false; result->heap_allocated = false; @@ -87,7 +83,7 @@ Status ReadBlock(RandomAccessFile* file, } // Check the crc of the type and the block contents - const char* data = contents.data(); // Pointer to where Read put the data + const char* data = contents.data(); // Pointer to where Read put the data if (options.verify_checksums) { const uint32_t crc = crc32c::Unmask(DecodeFixed32(data + n + 1)); const uint32_t actual = crc32c::Value(data, n + 1); diff --git a/table/format.h b/table/format.h index 6c0b80c..e49dfdc 100644 --- a/table/format.h +++ b/table/format.h @@ -5,8 +5,10 @@ #ifndef STORAGE_LEVELDB_TABLE_FORMAT_H_ #define STORAGE_LEVELDB_TABLE_FORMAT_H_ -#include #include + +#include + #include "leveldb/slice.h" #include "leveldb/status.h" #include "leveldb/table_builder.h" @@ -21,6 +23,9 @@ struct ReadOptions; // block or a meta block. class BlockHandle { public: + // Maximum encoding length of a BlockHandle + enum { kMaxEncodedLength = 10 + 10 }; + BlockHandle(); // The offset of the block in the file. @@ -34,9 +39,6 @@ class BlockHandle { void EncodeTo(std::string* dst) const; Status DecodeFrom(Slice* input); - // Maximum encoding length of a BlockHandle - enum { kMaxEncodedLength = 10 + 10 }; - private: uint64_t offset_; uint64_t size_; @@ -46,30 +48,24 @@ class BlockHandle { // end of every table file. class Footer { public: - Footer() { } + // Encoded length of a Footer. Note that the serialization of a + // Footer will always occupy exactly this many bytes. It consists + // of two block handles and a magic number. + enum { kEncodedLength = 2 * BlockHandle::kMaxEncodedLength + 8 }; + + Footer() = default; // 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; - } + const BlockHandle& index_handle() const { return index_handle_; } + void set_index_handle(const BlockHandle& h) { index_handle_ = h; } void EncodeTo(std::string* dst) const; Status DecodeFrom(Slice* input); - // Encoded length of a Footer. Note that the serialization of a - // Footer will always occupy exactly this many bytes. It consists - // of two block handles and a magic number. - enum { - kEncodedLength = 2*BlockHandle::kMaxEncodedLength + 8 - }; - private: BlockHandle metaindex_handle_; BlockHandle index_handle_; @@ -91,17 +87,13 @@ struct BlockContents { // Read the block identified by "handle" from "file". On failure // return non-OK. On success fill *result and return OK. -extern Status ReadBlock(RandomAccessFile* file, - const ReadOptions& options, - const BlockHandle& handle, - BlockContents* result); +Status ReadBlock(RandomAccessFile* file, const ReadOptions& options, + const BlockHandle& handle, BlockContents* result); // Implementation details follow. Clients should ignore, inline BlockHandle::BlockHandle() - : offset_(~static_cast(0)), - size_(~static_cast(0)) { -} + : offset_(~static_cast(0)), size_(~static_cast(0)) {} } // namespace leveldb diff --git a/table/iterator.cc b/table/iterator.cc index 3d1c87f..dfef083 100644 --- a/table/iterator.cc +++ b/table/iterator.cc @@ -7,58 +7,67 @@ namespace leveldb { Iterator::Iterator() { - cleanup_.function = NULL; - cleanup_.next = NULL; + cleanup_head_.function = nullptr; + cleanup_head_.next = nullptr; } Iterator::~Iterator() { - if (cleanup_.function != NULL) { - (*cleanup_.function)(cleanup_.arg1, cleanup_.arg2); - for (Cleanup* c = cleanup_.next; c != NULL; ) { - (*c->function)(c->arg1, c->arg2); - Cleanup* next = c->next; - delete c; - c = next; + if (!cleanup_head_.IsEmpty()) { + cleanup_head_.Run(); + for (CleanupNode* node = cleanup_head_.next; node != nullptr;) { + node->Run(); + CleanupNode* next_node = node->next; + delete node; + node = next_node; } } } void Iterator::RegisterCleanup(CleanupFunction func, void* arg1, void* arg2) { - assert(func != NULL); - Cleanup* c; - if (cleanup_.function == NULL) { - c = &cleanup_; + assert(func != nullptr); + CleanupNode* node; + if (cleanup_head_.IsEmpty()) { + node = &cleanup_head_; } else { - c = new Cleanup; - c->next = cleanup_.next; - cleanup_.next = c; + node = new CleanupNode(); + node->next = cleanup_head_.next; + cleanup_head_.next = node; } - c->function = func; - c->arg1 = arg1; - c->arg2 = arg2; + node->function = func; + node->arg1 = arg1; + node->arg2 = arg2; } namespace { + class EmptyIterator : public Iterator { public: - EmptyIterator(const Status& s) : status_(s) { } - virtual bool Valid() const { return false; } - virtual void Seek(const Slice& target) { } - virtual void SeekToFirst() { } - virtual void SeekToLast() { } - virtual void Next() { assert(false); } - virtual void Prev() { assert(false); } - Slice key() const { assert(false); return Slice(); } - Slice value() const { assert(false); return Slice(); } - virtual Status status() const { return status_; } + EmptyIterator(const Status& s) : status_(s) {} + ~EmptyIterator() override = default; + + bool Valid() const override { return false; } + void Seek(const Slice& target) override {} + void SeekToFirst() override {} + void SeekToLast() override {} + void Next() override { assert(false); } + void Prev() override { assert(false); } + Slice key() const override { + assert(false); + return Slice(); + } + Slice value() const override { + assert(false); + return Slice(); + } + Status status() const override { return status_; } + private: Status status_; }; -} // namespace -Iterator* NewEmptyIterator() { - return new EmptyIterator(Status::OK()); -} +} // anonymous namespace + +Iterator* NewEmptyIterator() { return new EmptyIterator(Status::OK()); } Iterator* NewErrorIterator(const Status& status) { return new EmptyIterator(status); diff --git a/table/iterator_wrapper.h b/table/iterator_wrapper.h index f410c3f..c230572 100644 --- a/table/iterator_wrapper.h +++ b/table/iterator_wrapper.h @@ -16,10 +16,8 @@ namespace leveldb { // cache locality. class IteratorWrapper { public: - IteratorWrapper(): iter_(NULL), valid_(false) { } - explicit IteratorWrapper(Iterator* iter): iter_(NULL) { - Set(iter); - } + IteratorWrapper() : iter_(nullptr), valid_(false) {} + explicit IteratorWrapper(Iterator* iter) : iter_(nullptr) { Set(iter); } ~IteratorWrapper() { delete iter_; } Iterator* iter() const { return iter_; } @@ -28,25 +26,53 @@ class IteratorWrapper { void Set(Iterator* iter) { delete iter_; iter_ = iter; - if (iter_ == NULL) { + if (iter_ == nullptr) { valid_ = false; } else { Update(); } } - // 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() != NULL - 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(); } + 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() { diff --git a/table/merger.cc b/table/merger.cc index 2dde4dc..76441b1 100644 --- a/table/merger.cc +++ b/table/merger.cc @@ -17,22 +17,18 @@ class MergingIterator : public Iterator { : comparator_(comparator), children_(new IteratorWrapper[n]), n_(n), - current_(NULL), + current_(nullptr), direction_(kForward) { for (int i = 0; i < n; i++) { children_[i].Set(children[i]); } } - virtual ~MergingIterator() { - delete[] children_; - } + ~MergingIterator() override { delete[] children_; } - virtual bool Valid() const { - return (current_ != NULL); - } + bool Valid() const override { return (current_ != nullptr); } - virtual void SeekToFirst() { + void SeekToFirst() override { for (int i = 0; i < n_; i++) { children_[i].SeekToFirst(); } @@ -40,7 +36,7 @@ class MergingIterator : public Iterator { direction_ = kForward; } - virtual void SeekToLast() { + void SeekToLast() override { for (int i = 0; i < n_; i++) { children_[i].SeekToLast(); } @@ -48,7 +44,7 @@ class MergingIterator : public Iterator { direction_ = kReverse; } - virtual void Seek(const Slice& target) { + void Seek(const Slice& target) override { for (int i = 0; i < n_; i++) { children_[i].Seek(target); } @@ -56,7 +52,7 @@ class MergingIterator : public Iterator { direction_ = kForward; } - virtual void Next() { + void Next() override { assert(Valid()); // Ensure that all children are positioned after key(). @@ -82,7 +78,7 @@ class MergingIterator : public Iterator { FindSmallest(); } - virtual void Prev() { + void Prev() override { assert(Valid()); // Ensure that all children are positioned before key(). @@ -111,17 +107,17 @@ class MergingIterator : public Iterator { FindLargest(); } - virtual Slice key() const { + Slice key() const override { assert(Valid()); return current_->key(); } - virtual Slice value() const { + Slice value() const override { assert(Valid()); return current_->value(); } - virtual Status status() const { + Status status() const override { Status status; for (int i = 0; i < n_; i++) { status = children_[i].status(); @@ -133,6 +129,9 @@ class MergingIterator : public Iterator { } private: + // Which direction is the iterator moving? + enum Direction { kForward, kReverse }; + void FindSmallest(); void FindLargest(); @@ -143,21 +142,15 @@ class MergingIterator : public Iterator { IteratorWrapper* children_; int n_; IteratorWrapper* current_; - - // Which direction is the iterator moving? - enum Direction { - kForward, - kReverse - }; Direction direction_; }; void MergingIterator::FindSmallest() { - IteratorWrapper* smallest = NULL; + IteratorWrapper* smallest = nullptr; for (int i = 0; i < n_; i++) { IteratorWrapper* child = &children_[i]; if (child->Valid()) { - if (smallest == NULL) { + if (smallest == nullptr) { smallest = child; } else if (comparator_->Compare(child->key(), smallest->key()) < 0) { smallest = child; @@ -168,11 +161,11 @@ void MergingIterator::FindSmallest() { } void MergingIterator::FindLargest() { - IteratorWrapper* largest = NULL; - for (int i = n_-1; i >= 0; i--) { + IteratorWrapper* largest = nullptr; + for (int i = n_ - 1; i >= 0; i--) { IteratorWrapper* child = &children_[i]; if (child->Valid()) { - if (largest == NULL) { + if (largest == nullptr) { largest = child; } else if (comparator_->Compare(child->key(), largest->key()) > 0) { largest = child; @@ -183,14 +176,15 @@ void MergingIterator::FindLargest() { } } // namespace -Iterator* NewMergingIterator(const Comparator* cmp, Iterator** list, int n) { +Iterator* NewMergingIterator(const Comparator* comparator, Iterator** children, + int n) { assert(n >= 0); if (n == 0) { return NewEmptyIterator(); } else if (n == 1) { - return list[0]; + return children[0]; } else { - return new MergingIterator(cmp, list, n); + return new MergingIterator(comparator, children, n); } } diff --git a/table/merger.h b/table/merger.h index 91ddd80..41cedc5 100644 --- a/table/merger.h +++ b/table/merger.h @@ -18,8 +18,8 @@ class Iterator; // key is present in K child iterators, it will be yielded K times. // // REQUIRES: n >= 0 -extern Iterator* NewMergingIterator( - const Comparator* comparator, Iterator** children, int n); +Iterator* NewMergingIterator(const Comparator* comparator, Iterator** children, + int n); } // namespace leveldb diff --git a/table/table.cc b/table/table.cc index dff8a82..b07bc88 100644 --- a/table/table.cc +++ b/table/table.cc @@ -20,7 +20,7 @@ namespace leveldb { struct Table::Rep { ~Rep() { delete filter; - delete [] filter_data; + delete[] filter_data; delete index_block; } @@ -35,11 +35,9 @@ struct Table::Rep { Block* index_block; }; -Status Table::Open(const Options& options, - RandomAccessFile* file, - uint64_t size, - Table** table) { - *table = NULL; +Status Table::Open(const Options& options, RandomAccessFile* file, + uint64_t size, Table** table) { + *table = nullptr; if (size < Footer::kEncodedLength) { return Status::Corruption("file is too short to be an sstable"); } @@ -55,41 +53,36 @@ Status Table::Open(const Options& options, if (!s.ok()) return s; // Read the index block - BlockContents contents; - Block* index_block = NULL; + BlockContents index_block_contents; if (s.ok()) { ReadOptions opt; if (options.paranoid_checks) { opt.verify_checksums = true; } - s = ReadBlock(file, opt, footer.index_handle(), &contents); - if (s.ok()) { - index_block = new Block(contents); - } + s = ReadBlock(file, opt, footer.index_handle(), &index_block_contents); } if (s.ok()) { // We've successfully read the footer and the index block: we're // ready to serve requests. + Block* index_block = new Block(index_block_contents); Rep* rep = new Table::Rep; rep->options = options; rep->file = file; rep->metaindex_handle = footer.metaindex_handle(); rep->index_block = index_block; rep->cache_id = (options.block_cache ? options.block_cache->NewId() : 0); - rep->filter_data = NULL; - rep->filter = NULL; + rep->filter_data = nullptr; + rep->filter = nullptr; *table = new Table(rep); (*table)->ReadMeta(footer); - } else { - if (index_block) delete index_block; } return s; } void Table::ReadMeta(const Footer& footer) { - if (rep_->options.filter_policy == NULL) { + if (rep_->options.filter_policy == nullptr) { return; // Do not need any metadata } @@ -135,14 +128,12 @@ void Table::ReadFilter(const Slice& filter_handle_value) { return; } if (block.heap_allocated) { - rep_->filter_data = block.data.data(); // Will need to delete later + rep_->filter_data = block.data.data(); // Will need to delete later } rep_->filter = new FilterBlockReader(rep_->options.filter_policy, block.data); } -Table::~Table() { - delete rep_; -} +Table::~Table() { delete rep_; } static void DeleteBlock(void* arg, void* ignored) { delete reinterpret_cast(arg); @@ -161,13 +152,12 @@ static void ReleaseBlock(void* arg, void* h) { // Convert an index iterator value (i.e., an encoded BlockHandle) // into an iterator over the contents of the corresponding block. -Iterator* Table::BlockReader(void* arg, - const ReadOptions& options, +Iterator* Table::BlockReader(void* arg, const ReadOptions& options, const Slice& index_value) { Table* table = reinterpret_cast(arg); Cache* block_cache = table->rep_->options.block_cache; - Block* block = NULL; - Cache::Handle* cache_handle = NULL; + Block* block = nullptr; + Cache::Handle* cache_handle = nullptr; BlockHandle handle; Slice input = index_value; @@ -177,21 +167,21 @@ Iterator* Table::BlockReader(void* arg, if (s.ok()) { BlockContents contents; - if (block_cache != NULL) { + if (block_cache != nullptr) { char cache_key_buffer[16]; EncodeFixed64(cache_key_buffer, table->rep_->cache_id); - EncodeFixed64(cache_key_buffer+8, handle.offset()); + EncodeFixed64(cache_key_buffer + 8, handle.offset()); Slice key(cache_key_buffer, sizeof(cache_key_buffer)); cache_handle = block_cache->Lookup(key); - if (cache_handle != NULL) { + if (cache_handle != nullptr) { block = reinterpret_cast(block_cache->Value(cache_handle)); } else { s = ReadBlock(table->rep_->file, options, handle, &contents); if (s.ok()) { block = new Block(contents); if (contents.cachable && options.fill_cache) { - cache_handle = block_cache->Insert( - key, block, block->size(), &DeleteCachedBlock); + cache_handle = block_cache->Insert(key, block, block->size(), + &DeleteCachedBlock); } } } @@ -204,10 +194,10 @@ Iterator* Table::BlockReader(void* arg, } Iterator* iter; - if (block != NULL) { + if (block != nullptr) { iter = block->NewIterator(table->rep_->options.comparator); - if (cache_handle == NULL) { - iter->RegisterCleanup(&DeleteBlock, block, NULL); + if (cache_handle == nullptr) { + iter->RegisterCleanup(&DeleteBlock, block, nullptr); } else { iter->RegisterCleanup(&ReleaseBlock, block_cache, cache_handle); } @@ -223,9 +213,9 @@ Iterator* Table::NewIterator(const ReadOptions& options) const { &Table::BlockReader, const_cast(this), options); } -Status Table::InternalGet(const ReadOptions& options, const Slice& k, - void* arg, - void (*saver)(void*, const Slice&, const Slice&)) { +Status Table::InternalGet(const ReadOptions& options, const Slice& k, void* arg, + void (*handle_result)(void*, const Slice&, + const Slice&)) { Status s; Iterator* iiter = rep_->index_block->NewIterator(rep_->options.comparator); iiter->Seek(k); @@ -233,15 +223,14 @@ Status Table::InternalGet(const ReadOptions& options, const Slice& k, Slice handle_value = iiter->value(); FilterBlockReader* filter = rep_->filter; BlockHandle handle; - if (filter != NULL && - handle.DecodeFrom(&handle_value).ok() && + if (filter != nullptr && handle.DecodeFrom(&handle_value).ok() && !filter->KeyMayMatch(handle.offset(), k)) { // Not found } else { Iterator* block_iter = BlockReader(this, options, iiter->value()); block_iter->Seek(k); if (block_iter->Valid()) { - (*saver)(arg, block_iter->key(), block_iter->value()); + (*handle_result)(arg, block_iter->key(), block_iter->value()); } s = block_iter->status(); delete block_iter; @@ -254,7 +243,6 @@ Status Table::InternalGet(const ReadOptions& options, const Slice& k, return s; } - uint64_t Table::ApproximateOffsetOf(const Slice& key) const { Iterator* index_iter = rep_->index_block->NewIterator(rep_->options.comparator); diff --git a/table/table_builder.cc b/table/table_builder.cc index 62002c8..278febf 100644 --- a/table/table_builder.cc +++ b/table/table_builder.cc @@ -5,6 +5,7 @@ #include "leveldb/table_builder.h" #include + #include "leveldb/comparator.h" #include "leveldb/env.h" #include "leveldb/filter_policy.h" @@ -18,6 +19,22 @@ namespace leveldb { struct TableBuilder::Rep { + Rep(const Options& opt, WritableFile* f) + : options(opt), + index_block_options(opt), + file(f), + offset(0), + data_block(&options), + index_block(&index_block_options), + num_entries(0), + closed(false), + filter_block(opt.filter_policy == nullptr + ? nullptr + : new FilterBlockBuilder(opt.filter_policy)), + pending_index_entry(false) { + index_block_options.block_restart_interval = 1; + } + Options options; Options index_block_options; WritableFile* file; @@ -27,7 +44,7 @@ struct TableBuilder::Rep { BlockBuilder index_block; std::string last_key; int64_t num_entries; - bool closed; // Either Finish() or Abandon() has been called. + bool closed; // Either Finish() or Abandon() has been called. FilterBlockBuilder* filter_block; // We do not emit the index entry for a block until we have seen the @@ -43,26 +60,11 @@ struct TableBuilder::Rep { BlockHandle pending_handle; // Handle to add to index block std::string compressed_output; - - Rep(const Options& opt, WritableFile* f) - : options(opt), - index_block_options(opt), - file(f), - offset(0), - data_block(&options), - index_block(&index_block_options), - num_entries(0), - closed(false), - filter_block(opt.filter_policy == NULL ? NULL - : new FilterBlockBuilder(opt.filter_policy)), - pending_index_entry(false) { - index_block_options.block_restart_interval = 1; - } }; TableBuilder::TableBuilder(const Options& options, WritableFile* file) : rep_(new Rep(options, file)) { - if (rep_->filter_block != NULL) { + if (rep_->filter_block != nullptr) { rep_->filter_block->StartBlock(0); } } @@ -106,7 +108,7 @@ void TableBuilder::Add(const Slice& key, const Slice& value) { r->pending_index_entry = false; } - if (r->filter_block != NULL) { + if (r->filter_block != nullptr) { r->filter_block->AddKey(key); } @@ -131,7 +133,7 @@ void TableBuilder::Flush() { r->pending_index_entry = true; r->status = r->file->Flush(); } - if (r->filter_block != NULL) { + if (r->filter_block != nullptr) { r->filter_block->StartBlock(r->offset); } } @@ -173,8 +175,7 @@ void TableBuilder::WriteBlock(BlockBuilder* block, BlockHandle* handle) { } void TableBuilder::WriteRawBlock(const Slice& block_contents, - CompressionType type, - BlockHandle* handle) { + CompressionType type, BlockHandle* handle) { Rep* r = rep_; handle->set_offset(r->offset); handle->set_size(block_contents.size()); @@ -184,7 +185,7 @@ void TableBuilder::WriteRawBlock(const Slice& block_contents, trailer[0] = type; uint32_t crc = crc32c::Value(block_contents.data(), block_contents.size()); crc = crc32c::Extend(crc, trailer, 1); // Extend crc to cover block type - EncodeFixed32(trailer+1, crc32c::Mask(crc)); + EncodeFixed32(trailer + 1, crc32c::Mask(crc)); r->status = r->file->Append(Slice(trailer, kBlockTrailerSize)); if (r->status.ok()) { r->offset += block_contents.size() + kBlockTrailerSize; @@ -192,9 +193,7 @@ void TableBuilder::WriteRawBlock(const Slice& block_contents, } } -Status TableBuilder::status() const { - return rep_->status; -} +Status TableBuilder::status() const { return rep_->status; } Status TableBuilder::Finish() { Rep* r = rep_; @@ -205,7 +204,7 @@ Status TableBuilder::Finish() { BlockHandle filter_block_handle, metaindex_block_handle, index_block_handle; // Write filter block - if (ok() && r->filter_block != NULL) { + if (ok() && r->filter_block != nullptr) { WriteRawBlock(r->filter_block->Finish(), kNoCompression, &filter_block_handle); } @@ -213,7 +212,7 @@ Status TableBuilder::Finish() { // Write metaindex block if (ok()) { BlockBuilder meta_index_block(&r->options); - if (r->filter_block != NULL) { + if (r->filter_block != nullptr) { // Add mapping from "filter.Name" to location of filter data std::string key = "filter."; key.append(r->options.filter_policy->Name()); @@ -259,12 +258,8 @@ void TableBuilder::Abandon() { r->closed = true; } -uint64_t TableBuilder::NumEntries() const { - return rep_->num_entries; -} +uint64_t TableBuilder::NumEntries() const { return rep_->num_entries; } -uint64_t TableBuilder::FileSize() const { - return rep_->offset; -} +uint64_t TableBuilder::FileSize() const { return rep_->offset; } } // namespace leveldb diff --git a/table/table_test.cc b/table/table_test.cc index c723bf8..f689a27 100644 --- a/table/table_test.cc +++ b/table/table_test.cc @@ -6,6 +6,7 @@ #include #include + #include "db/dbformat.h" #include "db/memtable.h" #include "db/write_batch_internal.h" @@ -27,8 +28,8 @@ namespace leveldb { static std::string Reverse(const Slice& key) { std::string str(key.ToString()); std::string rev(""); - for (std::string::reverse_iterator rit = str.rbegin(); - rit != str.rend(); ++rit) { + for (std::string::reverse_iterator rit = str.rbegin(); rit != str.rend(); + ++rit) { rev.push_back(*rit); } return rev; @@ -37,24 +38,23 @@ static std::string Reverse(const Slice& key) { namespace { class ReverseKeyComparator : public Comparator { public: - virtual const char* Name() const { + const char* Name() const override { return "leveldb.ReverseBytewiseComparator"; } - virtual int Compare(const Slice& a, const Slice& b) const { + 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 { + 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 { + void FindShortSuccessor(std::string* key) const override { std::string s = Reverse(*key); BytewiseComparator()->FindShortSuccessor(&s); *key = Reverse(s); @@ -79,25 +79,25 @@ namespace { struct STLLessThan { const Comparator* cmp; - STLLessThan() : cmp(BytewiseComparator()) { } - STLLessThan(const Comparator* c) : cmp(c) { } + STLLessThan() : cmp(BytewiseComparator()) {} + STLLessThan(const Comparator* c) : cmp(c) {} bool operator()(const std::string& a, const std::string& b) const { return cmp->Compare(Slice(a), Slice(b)) < 0; } }; } // namespace -class StringSink: public WritableFile { +class StringSink : public WritableFile { public: - ~StringSink() { } + ~StringSink() override = default; const std::string& contents() const { return contents_; } - virtual Status Close() { return Status::OK(); } - virtual Status Flush() { return Status::OK(); } - virtual Status Sync() { return Status::OK(); } + Status Close() override { return Status::OK(); } + Status Flush() override { return Status::OK(); } + Status Sync() override { return Status::OK(); } - virtual Status Append(const Slice& data) { + Status Append(const Slice& data) override { contents_.append(data.data(), data.size()); return Status::OK(); } @@ -106,20 +106,18 @@ class StringSink: public WritableFile { std::string contents_; }; - -class StringSource: public RandomAccessFile { +class StringSource : public RandomAccessFile { public: StringSource(const Slice& contents) - : contents_(contents.data(), contents.size()) { - } + : contents_(contents.data(), contents.size()) {} - virtual ~StringSource() { } + ~StringSource() override = default; uint64_t Size() const { return contents_.size(); } - virtual Status Read(uint64_t offset, size_t n, Slice* result, - char* scratch) const { - if (offset > contents_.size()) { + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + if (offset >= contents_.size()) { return Status::InvalidArgument("invalid Read offset"); } if (offset + n > contents_.size()) { @@ -140,8 +138,8 @@ typedef std::map KVMap; // BlockBuilder/TableBuilder and Block/Table. class Constructor { public: - explicit Constructor(const Comparator* cmp) : data_(STLLessThan(cmp)) { } - virtual ~Constructor() { } + explicit Constructor(const Comparator* cmp) : data_(STLLessThan(cmp)) {} + virtual ~Constructor() = default; void Add(const std::string& key, const Slice& value) { data_[key] = value.ToString(); @@ -150,15 +148,12 @@ class Constructor { // 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, - std::vector* keys, + void Finish(const Options& options, std::vector* keys, KVMap* kvmap) { *kvmap = data_; keys->clear(); - for (KVMap::const_iterator it = data_.begin(); - it != data_.end(); - ++it) { - keys->push_back(it->first); + for (const auto& kvp : data_) { + keys->push_back(kvp.first); } data_.clear(); Status s = FinishImpl(options, *kvmap); @@ -170,32 +165,26 @@ class Constructor { virtual Iterator* NewIterator() const = 0; - virtual const KVMap& data() { return data_; } + const KVMap& data() const { return data_; } - virtual DB* db() const { return NULL; } // Overridden in DBConstructor + virtual DB* db() const { return nullptr; } // Overridden in DBConstructor private: KVMap data_; }; -class BlockConstructor: public Constructor { +class BlockConstructor : public Constructor { public: explicit BlockConstructor(const Comparator* cmp) - : Constructor(cmp), - comparator_(cmp), - block_(NULL) { } - ~BlockConstructor() { + : Constructor(cmp), comparator_(cmp), block_(nullptr) {} + ~BlockConstructor() override { delete block_; } + Status FinishImpl(const Options& options, const KVMap& data) override { delete block_; - } - virtual Status FinishImpl(const Options& options, const KVMap& data) { - delete block_; - block_ = NULL; + block_ = nullptr; BlockBuilder builder(&options); - for (KVMap::const_iterator it = data.begin(); - it != data.end(); - ++it) { - builder.Add(it->first, it->second); + for (const auto& kvp : data) { + builder.Add(kvp.first, kvp.second); } // Open the block data_ = builder.Finish().ToString(); @@ -206,36 +195,30 @@ class BlockConstructor: public Constructor { block_ = new Block(contents); return Status::OK(); } - virtual Iterator* NewIterator() const { + Iterator* NewIterator() const override { return block_->NewIterator(comparator_); } private: - const Comparator* comparator_; + const Comparator* const comparator_; std::string data_; Block* block_; BlockConstructor(); }; -class TableConstructor: public Constructor { +class TableConstructor : public Constructor { public: TableConstructor(const Comparator* cmp) - : Constructor(cmp), - source_(NULL), table_(NULL) { - } - ~TableConstructor() { - Reset(); - } - virtual Status FinishImpl(const Options& options, const KVMap& data) { + : Constructor(cmp), source_(nullptr), table_(nullptr) {} + ~TableConstructor() override { Reset(); } + Status FinishImpl(const Options& options, const KVMap& data) override { Reset(); StringSink sink; TableBuilder builder(options, &sink); - for (KVMap::const_iterator it = data.begin(); - it != data.end(); - ++it) { - builder.Add(it->first, it->second); + for (const auto& kvp : data) { + builder.Add(kvp.first, kvp.second); ASSERT_TRUE(builder.status().ok()); } Status s = builder.Finish(); @@ -250,7 +233,7 @@ class TableConstructor: public Constructor { return Table::Open(table_options, source_, sink.contents().size(), &table_); } - virtual Iterator* NewIterator() const { + Iterator* NewIterator() const override { return table_->NewIterator(ReadOptions()); } @@ -262,8 +245,8 @@ class TableConstructor: public Constructor { void Reset() { delete table_; delete source_; - table_ = NULL; - source_ = NULL; + table_ = nullptr; + source_ = nullptr; } StringSource* source_; @@ -273,23 +256,28 @@ class TableConstructor: public Constructor { }; // A helper class that converts internal format keys into user keys -class KeyConvertingIterator: public Iterator { +class KeyConvertingIterator : public Iterator { public: - explicit KeyConvertingIterator(Iterator* iter) : iter_(iter) { } - virtual ~KeyConvertingIterator() { delete iter_; } - virtual bool Valid() const { return iter_->Valid(); } - virtual void Seek(const Slice& target) { + explicit KeyConvertingIterator(Iterator* iter) : iter_(iter) {} + + KeyConvertingIterator(const KeyConvertingIterator&) = delete; + KeyConvertingIterator& operator=(const KeyConvertingIterator&) = delete; + + ~KeyConvertingIterator() override { delete iter_; } + + bool Valid() const override { return iter_->Valid(); } + void Seek(const Slice& target) override { ParsedInternalKey ikey(target, kMaxSequenceNumber, kTypeValue); std::string encoded; AppendInternalKey(&encoded, ikey); iter_->Seek(encoded); } - virtual void SeekToFirst() { iter_->SeekToFirst(); } - virtual void SeekToLast() { iter_->SeekToLast(); } - virtual void Next() { iter_->Next(); } - virtual void Prev() { iter_->Prev(); } + void SeekToFirst() override { iter_->SeekToFirst(); } + void SeekToLast() override { iter_->SeekToLast(); } + void Next() override { iter_->Next(); } + void Prev() override { iter_->Prev(); } - virtual Slice key() const { + Slice key() const override { assert(Valid()); ParsedInternalKey key; if (!ParseInternalKey(iter_->key(), &key)) { @@ -299,82 +287,68 @@ class KeyConvertingIterator: public Iterator { return key.user_key; } - virtual Slice value() const { return iter_->value(); } - virtual Status status() const { + Slice value() const override { return iter_->value(); } + Status status() const override { return status_.ok() ? iter_->status() : status_; } private: mutable Status status_; Iterator* iter_; - - // No copying allowed - KeyConvertingIterator(const KeyConvertingIterator&); - void operator=(const KeyConvertingIterator&); }; -class MemTableConstructor: public Constructor { +class MemTableConstructor : public Constructor { public: explicit MemTableConstructor(const Comparator* cmp) - : Constructor(cmp), - internal_comparator_(cmp) { + : Constructor(cmp), internal_comparator_(cmp) { memtable_ = new MemTable(internal_comparator_); memtable_->Ref(); } - ~MemTableConstructor() { - memtable_->Unref(); - } - virtual Status FinishImpl(const Options& options, const KVMap& data) { + ~MemTableConstructor() override { memtable_->Unref(); } + Status FinishImpl(const Options& options, const KVMap& data) override { memtable_->Unref(); memtable_ = new MemTable(internal_comparator_); memtable_->Ref(); int seq = 1; - for (KVMap::const_iterator it = data.begin(); - it != data.end(); - ++it) { - memtable_->Add(seq, kTypeValue, it->first, it->second); + for (const auto& kvp : data) { + memtable_->Add(seq, kTypeValue, kvp.first, kvp.second); seq++; } return Status::OK(); } - virtual Iterator* NewIterator() const { + Iterator* NewIterator() const override { return new KeyConvertingIterator(memtable_->NewIterator()); } private: - InternalKeyComparator internal_comparator_; + const InternalKeyComparator internal_comparator_; MemTable* memtable_; }; -class DBConstructor: public Constructor { +class DBConstructor : public Constructor { public: explicit DBConstructor(const Comparator* cmp) - : Constructor(cmp), - comparator_(cmp) { - db_ = NULL; + : Constructor(cmp), comparator_(cmp) { + db_ = nullptr; NewDB(); } - ~DBConstructor() { + ~DBConstructor() override { delete db_; } + Status FinishImpl(const Options& options, const KVMap& data) override { delete db_; - } - virtual Status FinishImpl(const Options& options, const KVMap& data) { - delete db_; - db_ = NULL; + db_ = nullptr; NewDB(); - for (KVMap::const_iterator it = data.begin(); - it != data.end(); - ++it) { + for (const auto& kvp : data) { WriteBatch batch; - batch.Put(it->first, it->second); + batch.Put(kvp.first, kvp.second); ASSERT_TRUE(db_->Write(WriteOptions(), &batch).ok()); } return Status::OK(); } - virtual Iterator* NewIterator() const { + Iterator* NewIterator() const override { return db_->NewIterator(ReadOptions()); } - virtual DB* db() const { return db_; } + DB* db() const override { return db_; } private: void NewDB() { @@ -392,16 +366,11 @@ class DBConstructor: public Constructor { ASSERT_TRUE(status.ok()) << status.ToString(); } - const Comparator* comparator_; + const Comparator* const comparator_; DB* db_; }; -enum TestType { - TABLE_TEST, - BLOCK_TEST, - MEMTABLE_TEST, - DB_TEST -}; +enum TestType { TABLE_TEST, BLOCK_TEST, MEMTABLE_TEST, DB_TEST }; struct TestArgs { TestType type; @@ -410,37 +379,37 @@ struct TestArgs { }; static const TestArgs kTestArgList[] = { - { TABLE_TEST, false, 16 }, - { TABLE_TEST, false, 1 }, - { TABLE_TEST, false, 1024 }, - { TABLE_TEST, true, 16 }, - { TABLE_TEST, true, 1 }, - { TABLE_TEST, true, 1024 }, + {TABLE_TEST, false, 16}, + {TABLE_TEST, false, 1}, + {TABLE_TEST, false, 1024}, + {TABLE_TEST, true, 16}, + {TABLE_TEST, true, 1}, + {TABLE_TEST, true, 1024}, - { BLOCK_TEST, false, 16 }, - { BLOCK_TEST, false, 1 }, - { BLOCK_TEST, false, 1024 }, - { BLOCK_TEST, true, 16 }, - { BLOCK_TEST, true, 1 }, - { BLOCK_TEST, true, 1024 }, + {BLOCK_TEST, false, 16}, + {BLOCK_TEST, false, 1}, + {BLOCK_TEST, false, 1024}, + {BLOCK_TEST, true, 16}, + {BLOCK_TEST, true, 1}, + {BLOCK_TEST, true, 1024}, - // Restart interval does not matter for memtables - { MEMTABLE_TEST, false, 16 }, - { MEMTABLE_TEST, true, 16 }, + // Restart interval does not matter for memtables + {MEMTABLE_TEST, false, 16}, + {MEMTABLE_TEST, true, 16}, - // Do not bother with restart interval variations for DB - { DB_TEST, false, 16 }, - { DB_TEST, true, 16 }, + // Do not bother with restart interval variations for DB + {DB_TEST, false, 16}, + {DB_TEST, true, 16}, }; static const int kNumTestArgs = sizeof(kTestArgList) / sizeof(kTestArgList[0]); class Harness { public: - Harness() : constructor_(NULL) { } + Harness() : constructor_(nullptr) {} void Init(const TestArgs& args) { delete constructor_; - constructor_ = NULL; + constructor_ = nullptr; options_ = Options(); options_.block_restart_interval = args.restart_interval; @@ -466,9 +435,7 @@ class Harness { } } - ~Harness() { - delete constructor_; - } + ~Harness() { delete constructor_; } void Add(const std::string& key, const std::string& value) { constructor_->Add(key, value); @@ -490,8 +457,7 @@ class Harness { ASSERT_TRUE(!iter->Valid()); iter->SeekToFirst(); for (KVMap::const_iterator model_iter = data.begin(); - model_iter != data.end(); - ++model_iter) { + model_iter != data.end(); ++model_iter) { ASSERT_EQ(ToString(data, model_iter), ToString(iter)); iter->Next(); } @@ -505,8 +471,7 @@ class Harness { ASSERT_TRUE(!iter->Valid()); iter->SeekToLast(); for (KVMap::const_reverse_iterator model_iter = data.rbegin(); - model_iter != data.rend(); - ++model_iter) { + model_iter != data.rend(); ++model_iter) { ASSERT_EQ(ToString(data, model_iter), ToString(iter)); iter->Prev(); } @@ -514,8 +479,7 @@ class Harness { delete iter; } - void TestRandomAccess(Random* rnd, - const std::vector& keys, + void TestRandomAccess(Random* rnd, const std::vector& keys, const KVMap& data) { static const bool kVerbose = false; Iterator* iter = constructor_->NewIterator(); @@ -546,8 +510,8 @@ class Harness { 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()); + if (kVerbose) + fprintf(stderr, "Seek '%s'\n", EscapeString(key).c_str()); iter->Seek(Slice(key)); ASSERT_EQ(ToString(data, model_iter), ToString(iter)); break; @@ -558,7 +522,7 @@ class Harness { if (kVerbose) fprintf(stderr, "Prev\n"); iter->Prev(); if (model_iter == data.begin()) { - model_iter = data.end(); // Wrap around to invalid value + model_iter = data.end(); // Wrap around to invalid value } else { --model_iter; } @@ -621,8 +585,8 @@ class Harness { break; case 1: { // Attempt to return something smaller than an existing key - if (result.size() > 0 && result[result.size()-1] > '\0') { - result[result.size()-1]--; + if (!result.empty() && result[result.size() - 1] > '\0') { + result[result.size() - 1]--; } break; } @@ -636,7 +600,7 @@ class Harness { } } - // Returns NULL if not running against a DB + // Returns nullptr if not running against a DB DB* db() const { return constructor_->db(); } private: @@ -720,8 +684,8 @@ TEST(Harness, Randomized) { 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), int(kNumTestArgs), num_entries); + fprintf(stderr, "case %d of %d: num_entries = %d\n", (i + 1), + int(kNumTestArgs), num_entries); } for (int e = 0; e < num_entries; e++) { std::string v; @@ -735,7 +699,7 @@ TEST(Harness, Randomized) { TEST(Harness, RandomizedLongDB) { Random rnd(test::RandomSeed()); - TestArgs args = { DB_TEST, false, 16 }; + TestArgs args = {DB_TEST, false, 16}; Init(args); int num_entries = 100000; for (int e = 0; e < num_entries; e++) { @@ -757,7 +721,7 @@ TEST(Harness, RandomizedLongDB) { ASSERT_GT(files, 0); } -class MemTableTest { }; +class MemTableTest {}; TEST(MemTableTest, Simple) { InternalKeyComparator cmp(BytewiseComparator()); @@ -774,8 +738,7 @@ TEST(MemTableTest, Simple) { Iterator* iter = memtable->NewIterator(); iter->SeekToFirst(); while (iter->Valid()) { - fprintf(stderr, "key: '%s' -> '%s'\n", - iter->key().ToString().c_str(), + fprintf(stderr, "key: '%s' -> '%s'\n", iter->key().ToString().c_str(), iter->value().ToString().c_str()); iter->Next(); } @@ -788,14 +751,13 @@ 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)(val), (unsigned long long)(low), (unsigned long long)(high)); } return result; } -class TableTest { }; +class TableTest {}; TEST(TableTest, ApproximateOffsetOfPlain) { TableConstructor c(BytewiseComparator()); @@ -813,18 +775,17 @@ TEST(TableTest, ApproximateOffsetOfPlain) { options.compression = kNoCompression; c.Finish(options, &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("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)); - + 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 bool SnappyCompressionSupported() { @@ -853,16 +814,22 @@ TEST(TableTest, ApproximateOffsetOfCompressed) { options.compression = kSnappyCompression; c.Finish(options, &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, 6000)); + // Expected upper and lower bounds of space used by compressible strings. + static const int kSlop = 1000; // Compressor effectiveness varies. + const int expected = 2500; // 10000 * compression ratio (0.25) + const int min_z = expected - kSlop; + const int max_z = expected + kSlop; + + ASSERT_TRUE(Between(c.ApproximateOffsetOf("abc"), 0, kSlop)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k01"), 0, kSlop)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k02"), 0, kSlop)); + // Have now emitted a large compressible string, so adjust expected offset. + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k03"), min_z, max_z)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k04"), min_z, max_z)); + // Have now emitted two large compressible strings, so adjust expected offset. + ASSERT_TRUE(Between(c.ApproximateOffsetOf("xyz"), 2 * min_z, 2 * max_z)); } } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/table/two_level_iterator.cc b/table/two_level_iterator.cc index 7822eba..144790d 100644 --- a/table/two_level_iterator.cc +++ b/table/two_level_iterator.cc @@ -15,38 +15,33 @@ namespace { typedef Iterator* (*BlockFunction)(void*, const ReadOptions&, const Slice&); -class TwoLevelIterator: public Iterator { +class TwoLevelIterator : public Iterator { public: - TwoLevelIterator( - Iterator* index_iter, - BlockFunction block_function, - void* arg, - const ReadOptions& options); + TwoLevelIterator(Iterator* index_iter, BlockFunction block_function, + void* arg, const ReadOptions& options); - virtual ~TwoLevelIterator(); + ~TwoLevelIterator() override; - virtual void Seek(const Slice& target); - virtual void SeekToFirst(); - virtual void SeekToLast(); - virtual void Next(); - virtual void Prev(); + void Seek(const Slice& target) override; + void SeekToFirst() override; + void SeekToLast() override; + void Next() override; + void Prev() override; - virtual bool Valid() const { - return data_iter_.Valid(); - } - virtual Slice key() const { + bool Valid() const override { return data_iter_.Valid(); } + Slice key() const override { assert(Valid()); return data_iter_.key(); } - virtual Slice value() const { + Slice value() const override { assert(Valid()); return data_iter_.value(); } - virtual Status status() const { + Status status() const override { // It'd be nice if status() returned a const Status& instead of a Status if (!index_iter_.status().ok()) { return index_iter_.status(); - } else if (data_iter_.iter() != NULL && !data_iter_.status().ok()) { + } else if (data_iter_.iter() != nullptr && !data_iter_.status().ok()) { return data_iter_.status(); } else { return status_; @@ -67,45 +62,41 @@ class TwoLevelIterator: public Iterator { const ReadOptions options_; Status status_; IteratorWrapper index_iter_; - IteratorWrapper data_iter_; // May be NULL - // If data_iter_ is non-NULL, then "data_block_handle_" holds the + IteratorWrapper data_iter_; // May be nullptr + // If data_iter_ is non-null, then "data_block_handle_" holds the // "index_value" passed to block_function_ to create the data_iter_. std::string data_block_handle_; }; -TwoLevelIterator::TwoLevelIterator( - Iterator* index_iter, - BlockFunction block_function, - void* arg, - const ReadOptions& options) +TwoLevelIterator::TwoLevelIterator(Iterator* index_iter, + BlockFunction block_function, void* arg, + const ReadOptions& options) : block_function_(block_function), arg_(arg), options_(options), index_iter_(index_iter), - data_iter_(NULL) { -} + data_iter_(nullptr) {} -TwoLevelIterator::~TwoLevelIterator() { -} +TwoLevelIterator::~TwoLevelIterator() = default; void TwoLevelIterator::Seek(const Slice& target) { index_iter_.Seek(target); InitDataBlock(); - if (data_iter_.iter() != NULL) data_iter_.Seek(target); + if (data_iter_.iter() != nullptr) data_iter_.Seek(target); SkipEmptyDataBlocksForward(); } void TwoLevelIterator::SeekToFirst() { index_iter_.SeekToFirst(); InitDataBlock(); - if (data_iter_.iter() != NULL) data_iter_.SeekToFirst(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToFirst(); SkipEmptyDataBlocksForward(); } void TwoLevelIterator::SeekToLast() { index_iter_.SeekToLast(); InitDataBlock(); - if (data_iter_.iter() != NULL) data_iter_.SeekToLast(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToLast(); SkipEmptyDataBlocksBackward(); } @@ -121,44 +112,44 @@ void TwoLevelIterator::Prev() { SkipEmptyDataBlocksBackward(); } - void TwoLevelIterator::SkipEmptyDataBlocksForward() { - while (data_iter_.iter() == NULL || !data_iter_.Valid()) { + while (data_iter_.iter() == nullptr || !data_iter_.Valid()) { // Move to next block if (!index_iter_.Valid()) { - SetDataIterator(NULL); + SetDataIterator(nullptr); return; } index_iter_.Next(); InitDataBlock(); - if (data_iter_.iter() != NULL) data_iter_.SeekToFirst(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToFirst(); } } void TwoLevelIterator::SkipEmptyDataBlocksBackward() { - while (data_iter_.iter() == NULL || !data_iter_.Valid()) { + while (data_iter_.iter() == nullptr || !data_iter_.Valid()) { // Move to next block if (!index_iter_.Valid()) { - SetDataIterator(NULL); + SetDataIterator(nullptr); return; } index_iter_.Prev(); InitDataBlock(); - if (data_iter_.iter() != NULL) data_iter_.SeekToLast(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToLast(); } } void TwoLevelIterator::SetDataIterator(Iterator* data_iter) { - if (data_iter_.iter() != NULL) SaveError(data_iter_.status()); + if (data_iter_.iter() != nullptr) SaveError(data_iter_.status()); data_iter_.Set(data_iter); } void TwoLevelIterator::InitDataBlock() { if (!index_iter_.Valid()) { - SetDataIterator(NULL); + SetDataIterator(nullptr); } else { Slice handle = index_iter_.value(); - if (data_iter_.iter() != NULL && handle.compare(data_block_handle_) == 0) { + if (data_iter_.iter() != nullptr && + handle.compare(data_block_handle_) == 0) { // data_iter_ is already constructed with this iterator, so // no need to change anything } else { @@ -171,11 +162,9 @@ void TwoLevelIterator::InitDataBlock() { } // namespace -Iterator* NewTwoLevelIterator( - Iterator* index_iter, - BlockFunction block_function, - void* arg, - const ReadOptions& options) { +Iterator* NewTwoLevelIterator(Iterator* index_iter, + BlockFunction block_function, void* arg, + const ReadOptions& options) { return new TwoLevelIterator(index_iter, block_function, arg, options); } diff --git a/table/two_level_iterator.h b/table/two_level_iterator.h index 629ca34..81ffe80 100644 --- a/table/two_level_iterator.h +++ b/table/two_level_iterator.h @@ -20,14 +20,11 @@ struct ReadOptions; // // Uses a supplied function to convert an index_iter value into // an iterator over the contents of the corresponding block. -extern Iterator* NewTwoLevelIterator( +Iterator* NewTwoLevelIterator( Iterator* index_iter, - Iterator* (*block_function)( - void* arg, - const ReadOptions& options, - const Slice& index_value), - void* arg, - const ReadOptions& options); + Iterator* (*block_function)(void* arg, const ReadOptions& options, + const Slice& index_value), + void* arg, const ReadOptions& options); } // namespace leveldb diff --git a/util/arena.cc b/util/arena.cc index 9367f71..46e3b2e 100644 --- a/util/arena.cc +++ b/util/arena.cc @@ -3,17 +3,13 @@ // found in the LICENSE file. See the AUTHORS file for names of contributors. #include "util/arena.h" -#include namespace leveldb { static const int kBlockSize = 4096; -Arena::Arena() { - blocks_memory_ = 0; - alloc_ptr_ = NULL; // First allocation will allocate a block - alloc_bytes_remaining_ = 0; -} +Arena::Arena() + : alloc_ptr_(nullptr), alloc_bytes_remaining_(0), memory_usage_(0) {} Arena::~Arena() { for (size_t i = 0; i < blocks_.size(); i++) { @@ -41,8 +37,9 @@ char* Arena::AllocateFallback(size_t bytes) { char* Arena::AllocateAligned(size_t bytes) { const int align = (sizeof(void*) > 8) ? sizeof(void*) : 8; - assert((align & (align-1)) == 0); // Pointer size should be a power of 2 - size_t current_mod = reinterpret_cast(alloc_ptr_) & (align-1); + static_assert((align & (align - 1)) == 0, + "Pointer size should be a power of 2"); + size_t current_mod = reinterpret_cast(alloc_ptr_) & (align - 1); size_t slop = (current_mod == 0 ? 0 : align - current_mod); size_t needed = bytes + slop; char* result; @@ -54,14 +51,15 @@ char* Arena::AllocateAligned(size_t bytes) { // AllocateFallback always returned aligned memory result = AllocateFallback(bytes); } - assert((reinterpret_cast(result) & (align-1)) == 0); + assert((reinterpret_cast(result) & (align - 1)) == 0); return result; } char* Arena::AllocateNewBlock(size_t block_bytes) { char* result = new char[block_bytes]; - blocks_memory_ += block_bytes; blocks_.push_back(result); + memory_usage_.fetch_add(block_bytes + sizeof(char*), + std::memory_order_relaxed); return result; } diff --git a/util/arena.h b/util/arena.h index 73bbf1c..68fc55d 100644 --- a/util/arena.h +++ b/util/arena.h @@ -5,29 +5,33 @@ #ifndef STORAGE_LEVELDB_UTIL_ARENA_H_ #define STORAGE_LEVELDB_UTIL_ARENA_H_ +#include +#include +#include +#include #include -#include -#include -#include namespace leveldb { class Arena { public: Arena(); + + Arena(const Arena&) = delete; + Arena& operator=(const Arena&) = delete; + ~Arena(); // Return a pointer to a newly allocated memory block of "bytes" bytes. char* Allocate(size_t bytes); - // Allocate memory with the normal alignment guarantees provided by malloc + // Allocate memory with the normal alignment guarantees provided by malloc. char* AllocateAligned(size_t bytes); // Returns an estimate of the total memory usage of data allocated - // by the arena (including space allocated but not yet used for user - // allocations). + // by the arena. size_t MemoryUsage() const { - return blocks_memory_ + blocks_.capacity() * sizeof(char*); + return memory_usage_.load(std::memory_order_relaxed); } private: @@ -41,12 +45,11 @@ class Arena { // Array of new[] allocated memory blocks std::vector blocks_; - // Bytes of memory in blocks allocated so far - size_t blocks_memory_; - - // No copying allowed - Arena(const Arena&); - void operator=(const Arena&); + // Total memory usage of the arena. + // + // TODO(costan): This member is accessed via atomics, but the others are + // accessed without any locking. Is this OK? + std::atomic memory_usage_; }; inline char* Arena::Allocate(size_t bytes) { diff --git a/util/arena_test.cc b/util/arena_test.cc index 58e870e..f34095c 100644 --- a/util/arena_test.cc +++ b/util/arena_test.cc @@ -9,11 +9,9 @@ namespace leveldb { -class ArenaTest { }; +class ArenaTest {}; -TEST(ArenaTest, Empty) { - Arena arena; -} +TEST(ArenaTest, Empty) { Arena arena; } TEST(ArenaTest, Simple) { std::vector > allocated; @@ -26,8 +24,9 @@ TEST(ArenaTest, Simple) { if (i % (N / 10) == 0) { s = i; } else { - s = rnd.OneIn(4000) ? rnd.Uniform(6000) : - (rnd.OneIn(10) ? rnd.Uniform(100) : rnd.Uniform(20)); + s = rnd.OneIn(4000) + ? rnd.Uniform(6000) + : (rnd.OneIn(10) ? rnd.Uniform(100) : rnd.Uniform(20)); } if (s == 0) { // Our arena disallows size 0 allocations. @@ -47,7 +46,7 @@ TEST(ArenaTest, Simple) { bytes += s; allocated.push_back(std::make_pair(s, r)); ASSERT_GE(arena.MemoryUsage(), bytes); - if (i > N/10) { + if (i > N / 10) { ASSERT_LE(arena.MemoryUsage(), bytes * 1.10); } } @@ -63,6 +62,4 @@ TEST(ArenaTest, Simple) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/bloom.cc b/util/bloom.cc index a27a2ac..87547a7 100644 --- a/util/bloom.cc +++ b/util/bloom.cc @@ -15,24 +15,17 @@ static uint32_t BloomHash(const Slice& key) { } class BloomFilterPolicy : public FilterPolicy { - private: - size_t bits_per_key_; - size_t k_; - public: - explicit BloomFilterPolicy(int bits_per_key) - : bits_per_key_(bits_per_key) { + explicit BloomFilterPolicy(int bits_per_key) : bits_per_key_(bits_per_key) { // We intentionally round down to reduce probing cost a little bit k_ = static_cast(bits_per_key * 0.69); // 0.69 =~ ln(2) if (k_ < 1) k_ = 1; if (k_ > 30) k_ = 30; } - virtual const char* Name() const { - return "leveldb.BuiltinBloomFilter2"; - } + const char* Name() const override { return "leveldb.BuiltinBloomFilter2"; } - virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const { + void CreateFilter(const Slice* keys, int n, std::string* dst) const override { // Compute bloom filter size (in both bits and bytes) size_t bits = n * bits_per_key_; @@ -47,20 +40,20 @@ class BloomFilterPolicy : public FilterPolicy { dst->resize(init_size + bytes, 0); dst->push_back(static_cast(k_)); // Remember # of probes in filter char* array = &(*dst)[init_size]; - for (size_t i = 0; i < n; i++) { + for (int i = 0; i < n; i++) { // Use double-hashing to generate a sequence of hash values. // See analysis in [Kirsch,Mitzenmacher 2006]. uint32_t h = BloomHash(keys[i]); const uint32_t delta = (h >> 17) | (h << 15); // Rotate right 17 bits for (size_t j = 0; j < k_; j++) { const uint32_t bitpos = h % bits; - array[bitpos/8] |= (1 << (bitpos % 8)); + array[bitpos / 8] |= (1 << (bitpos % 8)); h += delta; } } } - virtual bool KeyMayMatch(const Slice& key, const Slice& bloom_filter) const { + bool KeyMayMatch(const Slice& key, const Slice& bloom_filter) const override { const size_t len = bloom_filter.size(); if (len < 2) return false; @@ -69,7 +62,7 @@ class BloomFilterPolicy : public FilterPolicy { // Use the encoded k so that we can read filters generated by // bloom filters created using different parameters. - const size_t k = array[len-1]; + const size_t k = array[len - 1]; if (k > 30) { // Reserved for potentially new encodings for short bloom filters. // Consider it a match. @@ -80,13 +73,17 @@ class BloomFilterPolicy : public FilterPolicy { const uint32_t delta = (h >> 17) | (h << 15); // Rotate right 17 bits for (size_t j = 0; j < k; j++) { const uint32_t bitpos = h % bits; - if ((array[bitpos/8] & (1 << (bitpos % 8))) == 0) return false; + if ((array[bitpos / 8] & (1 << (bitpos % 8))) == 0) return false; h += delta; } return true; } + + private: + size_t bits_per_key_; + size_t k_; }; -} +} // namespace const FilterPolicy* NewBloomFilterPolicy(int bits_per_key) { return new BloomFilterPolicy(bits_per_key); diff --git a/util/bloom_test.cc b/util/bloom_test.cc index 77fb1b3..436daa9 100644 --- a/util/bloom_test.cc +++ b/util/bloom_test.cc @@ -19,26 +19,17 @@ static Slice Key(int i, char* buffer) { } class BloomTest { - private: - const FilterPolicy* policy_; - std::string filter_; - std::vector keys_; - public: - BloomTest() : policy_(NewBloomFilterPolicy(10)) { } + BloomTest() : policy_(NewBloomFilterPolicy(10)) {} - ~BloomTest() { - delete policy_; - } + ~BloomTest() { delete policy_; } void Reset() { keys_.clear(); filter_.clear(); } - void Add(const Slice& s) { - keys_.push_back(s.ToString()); - } + void Add(const Slice& s) { keys_.push_back(s.ToString()); } void Build() { std::vector key_slices; @@ -46,21 +37,20 @@ class BloomTest { key_slices.push_back(Slice(keys_[i])); } filter_.clear(); - policy_->CreateFilter(&key_slices[0], key_slices.size(), &filter_); + policy_->CreateFilter(&key_slices[0], static_cast(key_slices.size()), + &filter_); keys_.clear(); if (kVerbose >= 2) DumpFilter(); } - size_t FilterSize() const { - return filter_.size(); - } + size_t FilterSize() const { return filter_.size(); } void DumpFilter() { fprintf(stderr, "F("); - for (size_t i = 0; i+1 < filter_.size(); i++) { + for (size_t i = 0; i + 1 < filter_.size(); i++) { const unsigned int c = static_cast(filter_[i]); for (int j = 0; j < 8; j++) { - fprintf(stderr, "%c", (c & (1 < keys_; }; TEST(BloomTest, EmptyFilter) { - ASSERT_TRUE(! Matches("hello")); - ASSERT_TRUE(! Matches("world")); + ASSERT_TRUE(!Matches("hello")); + ASSERT_TRUE(!Matches("world")); } TEST(BloomTest, Small) { @@ -95,8 +90,8 @@ TEST(BloomTest, Small) { Add("world"); ASSERT_TRUE(Matches("hello")); ASSERT_TRUE(Matches("world")); - ASSERT_TRUE(! Matches("x")); - ASSERT_TRUE(! Matches("foo")); + ASSERT_TRUE(!Matches("x")); + ASSERT_TRUE(!Matches("foo")); } static int NextLength(int length) { @@ -139,23 +134,23 @@ TEST(BloomTest, VaryingLengths) { double rate = FalsePositiveRate(); if (kVerbose >= 1) { fprintf(stderr, "False positives: %5.2f%% @ length = %6d ; bytes = %6d\n", - rate*100.0, length, static_cast(FilterSize())); + rate * 100.0, length, static_cast(FilterSize())); } - ASSERT_LE(rate, 0.02); // Must not be over 2% - if (rate > 0.0125) mediocre_filters++; // Allowed, but not too often - else good_filters++; + ASSERT_LE(rate, 0.02); // Must not be over 2% + if (rate > 0.0125) + mediocre_filters++; // Allowed, but not too often + else + good_filters++; } if (kVerbose >= 1) { - fprintf(stderr, "Filters: %d good, %d mediocre\n", - good_filters, mediocre_filters); + fprintf(stderr, "Filters: %d good, %d mediocre\n", good_filters, + mediocre_filters); } - ASSERT_LE(mediocre_filters, good_filters/5); + ASSERT_LE(mediocre_filters, good_filters / 5); } // Different bits-per-byte } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/cache.cc b/util/cache.cc index 8b197bc..12de306 100644 --- a/util/cache.cc +++ b/util/cache.cc @@ -8,17 +8,34 @@ #include "leveldb/cache.h" #include "port/port.h" +#include "port/thread_annotations.h" #include "util/hash.h" #include "util/mutexlock.h" namespace leveldb { -Cache::~Cache() { -} +Cache::~Cache() {} namespace { // LRU cache implementation +// +// Cache entries have an "in_cache" boolean indicating whether the cache has a +// reference on the entry. The only ways that this can become false without the +// entry being passed to its "deleter" are via Erase(), via Insert() when +// an element with a duplicate key is inserted, or on destruction of the cache. +// +// The cache keeps two linked lists of items in the cache. All items in the +// cache are in one list or the other, and never both. Items still referenced +// by clients but erased from the cache are in neither list. The lists are: +// - in-use: contains the items currently referenced by clients, in no +// particular order. (This list is used for invariant checking. If we +// removed the check, elements that would otherwise be on this list could be +// left as disconnected singleton lists.) +// - LRU: contains the items not currently referenced by clients, in LRU order +// Elements are moved between these lists by the Ref() and Unref() methods, +// when they detect an element in the cache acquiring or losing its only +// external reference. // An entry is a variable length heap-allocated structure. Entries // are kept in a circular doubly linked list ordered by access time. @@ -28,20 +45,19 @@ struct LRUHandle { LRUHandle* next_hash; LRUHandle* next; LRUHandle* prev; - size_t charge; // TODO(opt): Only allow uint32_t? + size_t charge; // TODO(opt): Only allow uint32_t? size_t key_length; - uint32_t refs; - uint32_t hash; // Hash of key(); used for fast sharding and comparisons - char key_data[1]; // Beginning of key + bool in_cache; // Whether entry is in the cache. + uint32_t refs; // References, including cache reference, if present. + uint32_t hash; // Hash of key(); used for fast sharding and comparisons + char key_data[1]; // Beginning of key Slice key() const { - // For cheaper lookups, we allow a temporary Handle object - // to store a pointer to a key in "value". - if (next == this) { - return *(reinterpret_cast(value)); - } else { - return Slice(key_data, key_length); - } + // next_ is only equal to this if the LRU handle is the list head of an + // empty list. List heads never have meaningful keys. + assert(next != this); + + return Slice(key_data, key_length); } }; @@ -52,7 +68,7 @@ struct LRUHandle { // 4.4.3's builtin hashtable. class HandleTable { public: - HandleTable() : length_(0), elems_(0), list_(NULL) { Resize(); } + HandleTable() : length_(0), elems_(0), list_(nullptr) { Resize(); } ~HandleTable() { delete[] list_; } LRUHandle* Lookup(const Slice& key, uint32_t hash) { @@ -62,9 +78,9 @@ class HandleTable { LRUHandle* Insert(LRUHandle* h) { LRUHandle** ptr = FindPointer(h->key(), h->hash); LRUHandle* old = *ptr; - h->next_hash = (old == NULL ? NULL : old->next_hash); + h->next_hash = (old == nullptr ? nullptr : old->next_hash); *ptr = h; - if (old == NULL) { + if (old == nullptr) { ++elems_; if (elems_ > length_) { // Since each cache entry is fairly large, we aim for a small @@ -78,7 +94,7 @@ class HandleTable { LRUHandle* Remove(const Slice& key, uint32_t hash) { LRUHandle** ptr = FindPointer(key, hash); LRUHandle* result = *ptr; - if (result != NULL) { + if (result != nullptr) { *ptr = result->next_hash; --elems_; } @@ -97,8 +113,7 @@ class HandleTable { // pointer to the trailing slot in the corresponding linked list. LRUHandle** FindPointer(const Slice& key, uint32_t hash) { LRUHandle** ptr = &list_[hash & (length_ - 1)]; - while (*ptr != NULL && - ((*ptr)->hash != hash || key != (*ptr)->key())) { + while (*ptr != nullptr && ((*ptr)->hash != hash || key != (*ptr)->key())) { ptr = &(*ptr)->next_hash; } return ptr; @@ -114,7 +129,7 @@ class HandleTable { uint32_t count = 0; for (uint32_t i = 0; i < length_; i++) { LRUHandle* h = list_[i]; - while (h != NULL) { + while (h != nullptr) { LRUHandle* next = h->next_hash; uint32_t hash = h->hash; LRUHandle** ptr = &new_list[hash & (new_length - 1)]; @@ -141,55 +156,83 @@ class LRUCache { void SetCapacity(size_t capacity) { capacity_ = capacity; } // Like Cache methods, but with an extra "hash" parameter. - Cache::Handle* Insert(const Slice& key, uint32_t hash, - void* value, size_t charge, + Cache::Handle* Insert(const Slice& key, uint32_t hash, void* value, + size_t charge, void (*deleter)(const Slice& key, void* value)); Cache::Handle* Lookup(const Slice& key, uint32_t hash); void Release(Cache::Handle* handle); void Erase(const Slice& key, uint32_t hash); + void Prune(); + size_t TotalCharge() const { + MutexLock l(&mutex_); + return usage_; + } private: void LRU_Remove(LRUHandle* e); - void LRU_Append(LRUHandle* e); + void LRU_Append(LRUHandle* list, LRUHandle* e); + void Ref(LRUHandle* e); void Unref(LRUHandle* e); + bool FinishErase(LRUHandle* e) EXCLUSIVE_LOCKS_REQUIRED(mutex_); // Initialized before use. size_t capacity_; // mutex_ protects the following state. - port::Mutex mutex_; - size_t usage_; + mutable port::Mutex mutex_; + size_t usage_ GUARDED_BY(mutex_); // Dummy head of LRU list. // lru.prev is newest entry, lru.next is oldest entry. - LRUHandle lru_; + // Entries have refs==1 and in_cache==true. + LRUHandle lru_ GUARDED_BY(mutex_); - HandleTable table_; + // Dummy head of in-use list. + // Entries are in use by clients, and have refs >= 2 and in_cache==true. + LRUHandle in_use_ GUARDED_BY(mutex_); + + HandleTable table_ GUARDED_BY(mutex_); }; -LRUCache::LRUCache() - : usage_(0) { - // Make empty circular linked list +LRUCache::LRUCache() : capacity_(0), usage_(0) { + // Make empty circular linked lists. lru_.next = &lru_; lru_.prev = &lru_; + in_use_.next = &in_use_; + in_use_.prev = &in_use_; } LRUCache::~LRUCache() { - for (LRUHandle* e = lru_.next; e != &lru_; ) { + assert(in_use_.next == &in_use_); // Error if caller has an unreleased handle + for (LRUHandle* e = lru_.next; e != &lru_;) { LRUHandle* next = e->next; - assert(e->refs == 1); // Error if caller has an unreleased handle + assert(e->in_cache); + e->in_cache = false; + assert(e->refs == 1); // Invariant of lru_ list. Unref(e); e = next; } } +void LRUCache::Ref(LRUHandle* e) { + if (e->refs == 1 && e->in_cache) { // If on lru_ list, move to in_use_ list. + LRU_Remove(e); + LRU_Append(&in_use_, e); + } + e->refs++; +} + void LRUCache::Unref(LRUHandle* e) { assert(e->refs > 0); e->refs--; - if (e->refs <= 0) { - usage_ -= e->charge; + if (e->refs == 0) { // Deallocate. + assert(!e->in_cache); (*e->deleter)(e->key(), e->value); free(e); + } else if (e->in_cache && e->refs == 1) { + // No longer in use; move to lru_ list. + LRU_Remove(e); + LRU_Append(&lru_, e); } } @@ -198,10 +241,10 @@ void LRUCache::LRU_Remove(LRUHandle* e) { e->prev->next = e->next; } -void LRUCache::LRU_Append(LRUHandle* e) { - // Make "e" newest entry by inserting just before lru_ - e->next = &lru_; - e->prev = lru_.prev; +void LRUCache::LRU_Append(LRUHandle* list, LRUHandle* e) { + // Make "e" newest entry by inserting just before *list + e->next = list; + e->prev = list->prev; e->prev->next = e; e->next->prev = e; } @@ -209,10 +252,8 @@ void LRUCache::LRU_Append(LRUHandle* e) { Cache::Handle* LRUCache::Lookup(const Slice& key, uint32_t hash) { MutexLock l(&mutex_); LRUHandle* e = table_.Lookup(key, hash); - if (e != NULL) { - e->refs++; - LRU_Remove(e); - LRU_Append(e); + if (e != nullptr) { + Ref(e); } return reinterpret_cast(e); } @@ -222,45 +263,72 @@ void LRUCache::Release(Cache::Handle* handle) { Unref(reinterpret_cast(handle)); } -Cache::Handle* LRUCache::Insert( - const Slice& key, uint32_t hash, void* value, size_t charge, - void (*deleter)(const Slice& key, void* value)) { +Cache::Handle* LRUCache::Insert(const Slice& key, uint32_t hash, void* value, + size_t charge, + void (*deleter)(const Slice& key, + void* value)) { MutexLock l(&mutex_); - LRUHandle* e = reinterpret_cast( - malloc(sizeof(LRUHandle)-1 + key.size())); + LRUHandle* e = + reinterpret_cast(malloc(sizeof(LRUHandle) - 1 + key.size())); e->value = value; e->deleter = deleter; e->charge = charge; e->key_length = key.size(); e->hash = hash; - e->refs = 2; // One from LRUCache, one for the returned handle + e->in_cache = false; + e->refs = 1; // for the returned handle. memcpy(e->key_data, key.data(), key.size()); - LRU_Append(e); - usage_ += charge; - LRUHandle* old = table_.Insert(e); - if (old != NULL) { - LRU_Remove(old); - Unref(old); + if (capacity_ > 0) { + e->refs++; // for the cache's reference. + e->in_cache = true; + LRU_Append(&in_use_, e); + usage_ += charge; + FinishErase(table_.Insert(e)); + } else { // don't cache. (capacity_==0 is supported and turns off caching.) + // next is read by key() in an assert, so it must be initialized + e->next = nullptr; } - while (usage_ > capacity_ && lru_.next != &lru_) { LRUHandle* old = lru_.next; - LRU_Remove(old); - table_.Remove(old->key(), old->hash); - Unref(old); + assert(old->refs == 1); + bool erased = FinishErase(table_.Remove(old->key(), old->hash)); + if (!erased) { // to avoid unused variable when compiled NDEBUG + assert(erased); + } } return reinterpret_cast(e); } +// If e != nullptr, finish removing *e from the cache; it has already been +// removed from the hash table. Return whether e != nullptr. +bool LRUCache::FinishErase(LRUHandle* e) { + if (e != nullptr) { + assert(e->in_cache); + LRU_Remove(e); + e->in_cache = false; + usage_ -= e->charge; + Unref(e); + } + return e != nullptr; +} + void LRUCache::Erase(const Slice& key, uint32_t hash) { MutexLock l(&mutex_); - LRUHandle* e = table_.Remove(key, hash); - if (e != NULL) { - LRU_Remove(e); - Unref(e); + FinishErase(table_.Remove(key, hash)); +} + +void LRUCache::Prune() { + MutexLock l(&mutex_); + while (lru_.next != &lru_) { + LRUHandle* e = lru_.next; + assert(e->refs == 1); + bool erased = FinishErase(table_.Remove(e->key(), e->hash)); + if (!erased) { // to avoid unused variable when compiled NDEBUG + assert(erased); + } } } @@ -277,49 +345,56 @@ class ShardedLRUCache : public Cache { return Hash(s.data(), s.size(), 0); } - static uint32_t Shard(uint32_t hash) { - return hash >> (32 - kNumShardBits); - } + static uint32_t Shard(uint32_t hash) { return hash >> (32 - kNumShardBits); } public: - explicit ShardedLRUCache(size_t capacity) - : last_id_(0) { + explicit ShardedLRUCache(size_t capacity) : last_id_(0) { const size_t per_shard = (capacity + (kNumShards - 1)) / kNumShards; for (int s = 0; s < kNumShards; s++) { shard_[s].SetCapacity(per_shard); } } - virtual ~ShardedLRUCache() { } - virtual Handle* Insert(const Slice& key, void* value, size_t charge, - void (*deleter)(const Slice& key, void* value)) { + ~ShardedLRUCache() override {} + Handle* Insert(const Slice& key, void* value, size_t charge, + void (*deleter)(const Slice& key, void* value)) override { const uint32_t hash = HashSlice(key); return shard_[Shard(hash)].Insert(key, hash, value, charge, deleter); } - virtual Handle* Lookup(const Slice& key) { + Handle* Lookup(const Slice& key) override { const uint32_t hash = HashSlice(key); return shard_[Shard(hash)].Lookup(key, hash); } - virtual void Release(Handle* handle) { + void Release(Handle* handle) override { LRUHandle* h = reinterpret_cast(handle); shard_[Shard(h->hash)].Release(handle); } - virtual void Erase(const Slice& key) { + void Erase(const Slice& key) override { const uint32_t hash = HashSlice(key); shard_[Shard(hash)].Erase(key, hash); } - virtual void* Value(Handle* handle) { + void* Value(Handle* handle) override { return reinterpret_cast(handle)->value; } - virtual uint64_t NewId() { + uint64_t NewId() override { MutexLock l(&id_mutex_); return ++(last_id_); } + void Prune() override { + for (int s = 0; s < kNumShards; s++) { + shard_[s].Prune(); + } + } + size_t TotalCharge() const override { + size_t total = 0; + for (int s = 0; s < kNumShards; s++) { + total += shard_[s].TotalCharge(); + } + return total; + } }; } // end anonymous namespace -Cache* NewLRUCache(size_t capacity) { - return new ShardedLRUCache(capacity); -} +Cache* NewLRUCache(size_t capacity) { return new ShardedLRUCache(capacity); } } // namespace leveldb diff --git a/util/cache_test.cc b/util/cache_test.cc index 4371671..974334b 100644 --- a/util/cache_test.cc +++ b/util/cache_test.cc @@ -25,8 +25,6 @@ static int DecodeValue(void* v) { return reinterpret_cast(v); } class CacheTest { public: - static CacheTest* current_; - static void Deleter(const Slice& key, void* v) { current_->deleted_keys_.push_back(DecodeKey(key)); current_->deleted_values_.push_back(DecodeValue(v)); @@ -37,18 +35,14 @@ class CacheTest { std::vector deleted_values_; Cache* cache_; - CacheTest() : cache_(NewLRUCache(kCacheSize)) { - current_ = this; - } + CacheTest() : cache_(NewLRUCache(kCacheSize)) { current_ = this; } - ~CacheTest() { - delete cache_; - } + ~CacheTest() { delete cache_; } int Lookup(int key) { Cache::Handle* handle = cache_->Lookup(EncodeKey(key)); - const int r = (handle == NULL) ? -1 : DecodeValue(cache_->Value(handle)); - if (handle != NULL) { + const int r = (handle == nullptr) ? -1 : DecodeValue(cache_->Value(handle)); + if (handle != nullptr) { cache_->Release(handle); } return r; @@ -59,9 +53,14 @@ class CacheTest { &CacheTest::Deleter)); } - void Erase(int key) { - cache_->Erase(EncodeKey(key)); + Cache::Handle* InsertAndReturnHandle(int key, int value, int charge = 1) { + return cache_->Insert(EncodeKey(key), EncodeValue(value), charge, + &CacheTest::Deleter); } + + void Erase(int key) { cache_->Erase(EncodeKey(key)); } + + static CacheTest* current_; }; CacheTest* CacheTest::current_; @@ -70,18 +69,18 @@ TEST(CacheTest, HitAndMiss) { Insert(100, 101); ASSERT_EQ(101, Lookup(100)); - ASSERT_EQ(-1, Lookup(200)); - ASSERT_EQ(-1, Lookup(300)); + ASSERT_EQ(-1, Lookup(200)); + ASSERT_EQ(-1, Lookup(300)); Insert(200, 201); ASSERT_EQ(101, Lookup(100)); ASSERT_EQ(201, Lookup(200)); - ASSERT_EQ(-1, Lookup(300)); + ASSERT_EQ(-1, Lookup(300)); Insert(100, 102); ASSERT_EQ(102, Lookup(100)); ASSERT_EQ(201, Lookup(200)); - ASSERT_EQ(-1, Lookup(300)); + ASSERT_EQ(-1, Lookup(300)); ASSERT_EQ(1, deleted_keys_.size()); ASSERT_EQ(100, deleted_keys_[0]); @@ -95,14 +94,14 @@ TEST(CacheTest, Erase) { Insert(100, 101); Insert(200, 201); Erase(100); - ASSERT_EQ(-1, Lookup(100)); + ASSERT_EQ(-1, Lookup(100)); ASSERT_EQ(201, Lookup(200)); ASSERT_EQ(1, deleted_keys_.size()); ASSERT_EQ(100, deleted_keys_[0]); ASSERT_EQ(101, deleted_values_[0]); Erase(100); - ASSERT_EQ(-1, Lookup(100)); + ASSERT_EQ(-1, Lookup(100)); ASSERT_EQ(201, Lookup(200)); ASSERT_EQ(1, deleted_keys_.size()); } @@ -135,15 +134,37 @@ TEST(CacheTest, EntriesArePinned) { TEST(CacheTest, EvictionPolicy) { Insert(100, 101); Insert(200, 201); + Insert(300, 301); + Cache::Handle* h = cache_->Lookup(EncodeKey(300)); - // Frequently used entry must be kept around + // Frequently used entry must be kept around, + // as must things that are still in use. for (int i = 0; i < kCacheSize + 100; i++) { - Insert(1000+i, 2000+i); - ASSERT_EQ(2000+i, Lookup(1000+i)); + Insert(1000 + i, 2000 + i); + ASSERT_EQ(2000 + i, Lookup(1000 + i)); ASSERT_EQ(101, Lookup(100)); } ASSERT_EQ(101, Lookup(100)); ASSERT_EQ(-1, Lookup(200)); + ASSERT_EQ(301, Lookup(300)); + cache_->Release(h); +} + +TEST(CacheTest, UseExceedsCacheSize) { + // Overfill the cache, keeping handles on all inserted entries. + std::vector h; + for (int i = 0; i < kCacheSize + 100; i++) { + h.push_back(InsertAndReturnHandle(1000 + i, 2000 + i)); + } + + // Check that all the entries can be found in the cache. + for (int i = 0; i < h.size(); i++) { + ASSERT_EQ(2000 + i, Lookup(1000 + i)); + } + + for (int i = 0; i < h.size(); i++) { + cache_->Release(h[i]); + } } TEST(CacheTest, HeavyEntries) { @@ -154,9 +175,9 @@ TEST(CacheTest, HeavyEntries) { const int kHeavy = 10; int added = 0; int index = 0; - while (added < 2*kCacheSize) { + while (added < 2 * kCacheSize) { const int weight = (index & 1) ? kLight : kHeavy; - Insert(index, 1000+index, weight); + Insert(index, 1000 + index, weight); added += weight; index++; } @@ -167,10 +188,10 @@ TEST(CacheTest, HeavyEntries) { int r = Lookup(i); if (r >= 0) { cached_weight += weight; - ASSERT_EQ(1000+i, r); + ASSERT_EQ(1000 + i, r); } } - ASSERT_LE(cached_weight, kCacheSize + kCacheSize/10); + ASSERT_LE(cached_weight, kCacheSize + kCacheSize / 10); } TEST(CacheTest, NewId) { @@ -179,8 +200,27 @@ TEST(CacheTest, NewId) { ASSERT_NE(a, b); } +TEST(CacheTest, Prune) { + Insert(1, 100); + Insert(2, 200); + + Cache::Handle* handle = cache_->Lookup(EncodeKey(1)); + ASSERT_TRUE(handle); + cache_->Prune(); + cache_->Release(handle); + + ASSERT_EQ(100, Lookup(1)); + ASSERT_EQ(-1, Lookup(2)); +} + +TEST(CacheTest, ZeroSizeCache) { + delete cache_; + cache_ = NewLRUCache(0); + + Insert(1, 100); + ASSERT_EQ(-1, Lookup(1)); +} + } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/coding.cc b/util/coding.cc index 21e3186..55be020 100644 --- a/util/coding.cc +++ b/util/coding.cc @@ -6,32 +6,6 @@ namespace leveldb { -void EncodeFixed32(char* buf, uint32_t value) { - if (port::kLittleEndian) { - memcpy(buf, &value, sizeof(value)); - } else { - buf[0] = value & 0xff; - buf[1] = (value >> 8) & 0xff; - buf[2] = (value >> 16) & 0xff; - buf[3] = (value >> 24) & 0xff; - } -} - -void EncodeFixed64(char* buf, uint64_t value) { - if (port::kLittleEndian) { - memcpy(buf, &value, sizeof(value)); - } else { - buf[0] = value & 0xff; - buf[1] = (value >> 8) & 0xff; - buf[2] = (value >> 16) & 0xff; - buf[3] = (value >> 24) & 0xff; - buf[4] = (value >> 32) & 0xff; - buf[5] = (value >> 40) & 0xff; - buf[6] = (value >> 48) & 0xff; - buf[7] = (value >> 56) & 0xff; - } -} - void PutFixed32(std::string* dst, uint32_t value) { char buf[sizeof(value)]; EncodeFixed32(buf, value); @@ -48,26 +22,26 @@ char* EncodeVarint32(char* dst, uint32_t v) { // Operate on characters as unsigneds unsigned char* ptr = reinterpret_cast(dst); static const int B = 128; - if (v < (1<<7)) { + if (v < (1 << 7)) { *(ptr++) = v; - } else if (v < (1<<14)) { + } else if (v < (1 << 14)) { *(ptr++) = v | B; - *(ptr++) = v>>7; - } else if (v < (1<<21)) { + *(ptr++) = v >> 7; + } else if (v < (1 << 21)) { *(ptr++) = v | B; - *(ptr++) = (v>>7) | B; - *(ptr++) = v>>14; - } else if (v < (1<<28)) { + *(ptr++) = (v >> 7) | B; + *(ptr++) = v >> 14; + } else if (v < (1 << 28)) { *(ptr++) = v | B; - *(ptr++) = (v>>7) | B; - *(ptr++) = (v>>14) | B; - *(ptr++) = v>>21; + *(ptr++) = (v >> 7) | B; + *(ptr++) = (v >> 14) | B; + *(ptr++) = v >> 21; } else { *(ptr++) = v | B; - *(ptr++) = (v>>7) | B; - *(ptr++) = (v>>14) | B; - *(ptr++) = (v>>21) | B; - *(ptr++) = v>>28; + *(ptr++) = (v >> 7) | B; + *(ptr++) = (v >> 14) | B; + *(ptr++) = (v >> 21) | B; + *(ptr++) = v >> 28; } return reinterpret_cast(ptr); } @@ -82,7 +56,7 @@ char* EncodeVarint64(char* dst, uint64_t v) { static const int B = 128; unsigned char* ptr = reinterpret_cast(dst); while (v >= B) { - *(ptr++) = (v & (B-1)) | B; + *(ptr++) = v | B; v >>= 7; } *(ptr++) = static_cast(v); @@ -109,8 +83,7 @@ int VarintLength(uint64_t v) { return len; } -const char* GetVarint32PtrFallback(const char* p, - const char* limit, +const char* GetVarint32PtrFallback(const char* p, const char* limit, uint32_t* value) { uint32_t result = 0; for (uint32_t shift = 0; shift <= 28 && p < limit; shift += 7) { @@ -125,14 +98,14 @@ const char* GetVarint32PtrFallback(const char* p, return reinterpret_cast(p); } } - return NULL; + return nullptr; } bool GetVarint32(Slice* input, uint32_t* value) { const char* p = input->data(); const char* limit = p + input->size(); const char* q = GetVarint32Ptr(p, limit, value); - if (q == NULL) { + if (q == nullptr) { return false; } else { *input = Slice(q, limit - q); @@ -154,14 +127,14 @@ const char* GetVarint64Ptr(const char* p, const char* limit, uint64_t* value) { return reinterpret_cast(p); } } - return NULL; + return nullptr; } bool GetVarint64(Slice* input, uint64_t* value) { const char* p = input->data(); const char* limit = p + input->size(); const char* q = GetVarint64Ptr(p, limit, value); - if (q == NULL) { + if (q == nullptr) { return false; } else { *input = Slice(q, limit - q); @@ -173,16 +146,15 @@ const char* GetLengthPrefixedSlice(const char* p, const char* limit, Slice* result) { uint32_t len; p = GetVarint32Ptr(p, limit, &len); - if (p == NULL) return NULL; - if (p + len > limit) return NULL; + if (p == nullptr) return nullptr; + if (p + len > limit) return nullptr; *result = Slice(p, len); return p + len; } bool GetLengthPrefixedSlice(Slice* input, Slice* result) { uint32_t len; - if (GetVarint32(input, &len) && - input->size() >= len) { + if (GetVarint32(input, &len) && input->size() >= len) { *result = Slice(input->data(), len); input->remove_prefix(len); return true; diff --git a/util/coding.h b/util/coding.h index 3993c4a..92a961f 100644 --- a/util/coding.h +++ b/util/coding.h @@ -10,84 +10,144 @@ #ifndef STORAGE_LEVELDB_UTIL_CODING_H_ #define STORAGE_LEVELDB_UTIL_CODING_H_ -#include -#include +#include +#include #include + #include "leveldb/slice.h" #include "port/port.h" namespace leveldb { // Standard Put... routines append to a string -extern void PutFixed32(std::string* dst, uint32_t value); -extern void PutFixed64(std::string* dst, uint64_t value); -extern void PutVarint32(std::string* dst, uint32_t value); -extern void PutVarint64(std::string* dst, uint64_t value); -extern void PutLengthPrefixedSlice(std::string* dst, const Slice& value); +void PutFixed32(std::string* dst, uint32_t value); +void PutFixed64(std::string* dst, uint64_t value); +void PutVarint32(std::string* dst, uint32_t value); +void PutVarint64(std::string* dst, uint64_t value); +void PutLengthPrefixedSlice(std::string* dst, const Slice& value); // Standard Get... routines parse a value from the beginning of a Slice // and advance the slice past the parsed value. -extern bool GetVarint32(Slice* input, uint32_t* value); -extern bool GetVarint64(Slice* input, uint64_t* value); -extern bool GetLengthPrefixedSlice(Slice* input, Slice* result); +bool GetVarint32(Slice* input, uint32_t* value); +bool GetVarint64(Slice* input, uint64_t* value); +bool GetLengthPrefixedSlice(Slice* input, Slice* result); // Pointer-based variants of GetVarint... These either store a value // in *v and return a pointer just past the parsed value, or return -// NULL on error. These routines only look at bytes in the range +// nullptr on error. These routines only look at bytes in the range // [p..limit-1] -extern const char* GetVarint32Ptr(const char* p,const char* limit, uint32_t* v); -extern const char* GetVarint64Ptr(const char* p,const char* limit, uint64_t* v); +const char* GetVarint32Ptr(const char* p, const char* limit, uint32_t* v); +const char* GetVarint64Ptr(const char* p, const char* limit, uint64_t* v); // Returns the length of the varint32 or varint64 encoding of "v" -extern int VarintLength(uint64_t v); - -// Lower-level versions of Put... that write directly into a character buffer -// REQUIRES: dst has enough space for the value being written -extern void EncodeFixed32(char* dst, uint32_t value); -extern void EncodeFixed64(char* dst, uint64_t value); +int VarintLength(uint64_t v); // Lower-level versions of Put... that write directly into a character buffer // and return a pointer just past the last byte written. // REQUIRES: dst has enough space for the value being written -extern char* EncodeVarint32(char* dst, uint32_t value); -extern char* EncodeVarint64(char* dst, uint64_t value); +char* EncodeVarint32(char* dst, uint32_t value); +char* EncodeVarint64(char* dst, uint64_t value); + +// TODO(costan): Remove port::kLittleEndian and the fast paths based on +// std::memcpy when clang learns to optimize the generic code, as +// described in https://bugs.llvm.org/show_bug.cgi?id=41761 +// +// The platform-independent code in DecodeFixed{32,64}() gets optimized to mov +// on x86 and ldr on ARM64, by both clang and gcc. However, only gcc optimizes +// the platform-independent code in EncodeFixed{32,64}() to mov / str. + +// Lower-level versions of Put... that write directly into a character buffer +// REQUIRES: dst has enough space for the value being written + +inline void EncodeFixed32(char* dst, uint32_t value) { + uint8_t* const buffer = reinterpret_cast(dst); + + if (port::kLittleEndian) { + // Fast path for little-endian CPUs. All major compilers optimize this to a + // single mov (x86_64) / str (ARM) instruction. + std::memcpy(buffer, &value, sizeof(uint32_t)); + return; + } + + // Platform-independent code. + // Currently, only gcc optimizes this to a single mov / str instruction. + buffer[0] = static_cast(value); + buffer[1] = static_cast(value >> 8); + buffer[2] = static_cast(value >> 16); + buffer[3] = static_cast(value >> 24); +} + +inline void EncodeFixed64(char* dst, uint64_t value) { + uint8_t* const buffer = reinterpret_cast(dst); + + if (port::kLittleEndian) { + // Fast path for little-endian CPUs. All major compilers optimize this to a + // single mov (x86_64) / str (ARM) instruction. + std::memcpy(buffer, &value, sizeof(uint64_t)); + return; + } + + // Platform-independent code. + // Currently, only gcc optimizes this to a single mov / str instruction. + buffer[0] = static_cast(value); + buffer[1] = static_cast(value >> 8); + buffer[2] = static_cast(value >> 16); + buffer[3] = static_cast(value >> 24); + buffer[4] = static_cast(value >> 32); + buffer[5] = static_cast(value >> 40); + buffer[6] = static_cast(value >> 48); + buffer[7] = static_cast(value >> 56); +} // Lower-level versions of Get... that read directly from a character buffer // without any bounds checking. inline uint32_t DecodeFixed32(const char* ptr) { + const uint8_t* const buffer = reinterpret_cast(ptr); + if (port::kLittleEndian) { - // Load the raw bytes + // Fast path for little-endian CPUs. All major compilers optimize this to a + // single mov (x86_64) / ldr (ARM) instruction. uint32_t result; - memcpy(&result, ptr, sizeof(result)); // gcc optimizes this to a plain load + std::memcpy(&result, buffer, sizeof(uint32_t)); return result; - } else { - return ((static_cast(static_cast(ptr[0]))) - | (static_cast(static_cast(ptr[1])) << 8) - | (static_cast(static_cast(ptr[2])) << 16) - | (static_cast(static_cast(ptr[3])) << 24)); } + + // Platform-independent code. + // Clang and gcc optimize this to a single mov / ldr instruction. + return (static_cast(buffer[0])) | + (static_cast(buffer[1]) << 8) | + (static_cast(buffer[2]) << 16) | + (static_cast(buffer[3]) << 24); } inline uint64_t DecodeFixed64(const char* ptr) { + const uint8_t* const buffer = reinterpret_cast(ptr); + if (port::kLittleEndian) { - // Load the raw bytes + // Fast path for little-endian CPUs. All major compilers optimize this to a + // single mov (x86_64) / ldr (ARM) instruction. uint64_t result; - memcpy(&result, ptr, sizeof(result)); // gcc optimizes this to a plain load + std::memcpy(&result, buffer, sizeof(uint64_t)); return result; - } else { - uint64_t lo = DecodeFixed32(ptr); - uint64_t hi = DecodeFixed32(ptr + 4); - return (hi << 32) | lo; } + + // Platform-independent code. + // Clang and gcc optimize this to a single mov / ldr instruction. + return (static_cast(buffer[0])) | + (static_cast(buffer[1]) << 8) | + (static_cast(buffer[2]) << 16) | + (static_cast(buffer[3]) << 24) | + (static_cast(buffer[4]) << 32) | + (static_cast(buffer[5]) << 40) | + (static_cast(buffer[6]) << 48) | + (static_cast(buffer[7]) << 56); } // Internal routine for use by fallback path of GetVarint32Ptr -extern const char* GetVarint32PtrFallback(const char* p, - const char* limit, - uint32_t* value); -inline const char* GetVarint32Ptr(const char* p, - const char* limit, +const char* GetVarint32PtrFallback(const char* p, const char* limit, + uint32_t* value); +inline const char* GetVarint32Ptr(const char* p, const char* limit, uint32_t* value) { if (p < limit) { uint32_t result = *(reinterpret_cast(p)); diff --git a/util/coding_test.cc b/util/coding_test.cc index 521541e..0d2a0c5 100644 --- a/util/coding_test.cc +++ b/util/coding_test.cc @@ -2,13 +2,14 @@ // 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 "util/coding.h" +#include +#include "util/coding.h" #include "util/testharness.h" namespace leveldb { -class Coding { }; +class Coding {}; TEST(Coding, Fixed32) { std::string s; @@ -38,15 +39,15 @@ TEST(Coding, Fixed64) { uint64_t v = static_cast(1) << power; uint64_t actual; actual = DecodeFixed64(p); - ASSERT_EQ(v-1, actual); + ASSERT_EQ(v - 1, actual); p += sizeof(uint64_t); actual = DecodeFixed64(p); - ASSERT_EQ(v+0, actual); + ASSERT_EQ(v + 0, actual); p += sizeof(uint64_t); actual = DecodeFixed64(p); - ASSERT_EQ(v+1, actual); + ASSERT_EQ(v + 1, actual); p += sizeof(uint64_t); } } @@ -88,7 +89,7 @@ TEST(Coding, Varint32) { uint32_t actual; const char* start = p; p = GetVarint32Ptr(p, limit, &actual); - ASSERT_TRUE(p != NULL); + ASSERT_TRUE(p != nullptr); ASSERT_EQ(expected, actual); ASSERT_EQ(VarintLength(actual), p - start); } @@ -107,8 +108,8 @@ TEST(Coding, Varint64) { // Test values near powers of two const uint64_t power = 1ull << k; values.push_back(power); - values.push_back(power-1); - values.push_back(power+1); + values.push_back(power - 1); + values.push_back(power + 1); } std::string s; @@ -123,19 +124,18 @@ TEST(Coding, Varint64) { uint64_t actual; const char* start = p; p = GetVarint64Ptr(p, limit, &actual); - ASSERT_TRUE(p != NULL); + ASSERT_TRUE(p != nullptr); ASSERT_EQ(values[i], actual); ASSERT_EQ(VarintLength(actual), p - start); } ASSERT_EQ(p, limit); - } TEST(Coding, Varint32Overflow) { uint32_t result; std::string input("\x81\x82\x83\x84\x85\x11"); - ASSERT_TRUE(GetVarint32Ptr(input.data(), input.data() + input.size(), &result) - == NULL); + ASSERT_TRUE(GetVarint32Ptr(input.data(), input.data() + input.size(), + &result) == nullptr); } TEST(Coding, Varint32Truncation) { @@ -144,17 +144,18 @@ TEST(Coding, Varint32Truncation) { PutVarint32(&s, large_value); uint32_t result; for (size_t len = 0; len < s.size() - 1; len++) { - ASSERT_TRUE(GetVarint32Ptr(s.data(), s.data() + len, &result) == NULL); + ASSERT_TRUE(GetVarint32Ptr(s.data(), s.data() + len, &result) == nullptr); } - ASSERT_TRUE(GetVarint32Ptr(s.data(), s.data() + s.size(), &result) != NULL); + ASSERT_TRUE(GetVarint32Ptr(s.data(), s.data() + s.size(), &result) != + nullptr); ASSERT_EQ(large_value, result); } TEST(Coding, Varint64Overflow) { uint64_t result; std::string input("\x81\x82\x83\x84\x85\x81\x82\x83\x84\x85\x11"); - ASSERT_TRUE(GetVarint64Ptr(input.data(), input.data() + input.size(), &result) - == NULL); + ASSERT_TRUE(GetVarint64Ptr(input.data(), input.data() + input.size(), + &result) == nullptr); } TEST(Coding, Varint64Truncation) { @@ -163,9 +164,10 @@ TEST(Coding, Varint64Truncation) { PutVarint64(&s, large_value); uint64_t result; for (size_t len = 0; len < s.size() - 1; len++) { - ASSERT_TRUE(GetVarint64Ptr(s.data(), s.data() + len, &result) == NULL); + ASSERT_TRUE(GetVarint64Ptr(s.data(), s.data() + len, &result) == nullptr); } - ASSERT_TRUE(GetVarint64Ptr(s.data(), s.data() + s.size(), &result) != NULL); + ASSERT_TRUE(GetVarint64Ptr(s.data(), s.data() + s.size(), &result) != + nullptr); ASSERT_EQ(large_value, result); } @@ -191,6 +193,4 @@ TEST(Coding, Strings) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/comparator.cc b/util/comparator.cc index 4b7b572..c5766e9 100644 --- a/util/comparator.cc +++ b/util/comparator.cc @@ -2,33 +2,34 @@ // 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 "leveldb/comparator.h" + +#include +#include +#include +#include + #include "leveldb/slice.h" -#include "port/port.h" #include "util/logging.h" +#include "util/no_destructor.h" namespace leveldb { -Comparator::~Comparator() { } +Comparator::~Comparator() = default; namespace { class BytewiseComparatorImpl : public Comparator { public: - BytewiseComparatorImpl() { } + BytewiseComparatorImpl() = default; - virtual const char* Name() const { - return "leveldb.BytewiseComparator"; - } + const char* Name() const override { return "leveldb.BytewiseComparator"; } - virtual int Compare(const Slice& a, const Slice& b) const { + int Compare(const Slice& a, const Slice& b) const override { return a.compare(b); } - virtual void FindShortestSeparator( - std::string* start, - const Slice& limit) const { + void FindShortestSeparator(std::string* start, + const Slice& limit) const override { // Find length of common prefix size_t min_length = std::min(start->size(), limit.size()); size_t diff_index = 0; @@ -50,14 +51,14 @@ class BytewiseComparatorImpl : public Comparator { } } - virtual void FindShortSuccessor(std::string* key) const { + void FindShortSuccessor(std::string* key) const override { // Find first character that can be incremented size_t n = key->size(); for (size_t i = 0; i < n; i++) { const uint8_t byte = (*key)[i]; if (byte != static_cast(0xff)) { (*key)[i] = byte + 1; - key->resize(i+1); + key->resize(i + 1); return; } } @@ -66,16 +67,9 @@ class BytewiseComparatorImpl : public Comparator { }; } // namespace -static port::OnceType once = LEVELDB_ONCE_INIT; -static const Comparator* bytewise; - -static void InitModule() { - bytewise = new BytewiseComparatorImpl; -} - const Comparator* BytewiseComparator() { - port::InitOnce(&once, InitModule); - return bytewise; + static NoDestructor singleton; + return singleton.get(); } } // namespace leveldb diff --git a/util/crc32c.cc b/util/crc32c.cc index 6db9e77..c2e61f7 100644 --- a/util/crc32c.cc +++ b/util/crc32c.cc @@ -2,330 +2,378 @@ // 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 portable implementation of crc32c, optimized to handle -// four bytes at a time. +// A portable implementation of crc32c. #include "util/crc32c.h" +#include #include + +#include "port/port.h" #include "util/coding.h" namespace leveldb { namespace crc32c { -static const uint32_t table0_[256] = { - 0x00000000, 0xf26b8303, 0xe13b70f7, 0x1350f3f4, - 0xc79a971f, 0x35f1141c, 0x26a1e7e8, 0xd4ca64eb, - 0x8ad958cf, 0x78b2dbcc, 0x6be22838, 0x9989ab3b, - 0x4d43cfd0, 0xbf284cd3, 0xac78bf27, 0x5e133c24, - 0x105ec76f, 0xe235446c, 0xf165b798, 0x030e349b, - 0xd7c45070, 0x25afd373, 0x36ff2087, 0xc494a384, - 0x9a879fa0, 0x68ec1ca3, 0x7bbcef57, 0x89d76c54, - 0x5d1d08bf, 0xaf768bbc, 0xbc267848, 0x4e4dfb4b, - 0x20bd8ede, 0xd2d60ddd, 0xc186fe29, 0x33ed7d2a, - 0xe72719c1, 0x154c9ac2, 0x061c6936, 0xf477ea35, - 0xaa64d611, 0x580f5512, 0x4b5fa6e6, 0xb93425e5, - 0x6dfe410e, 0x9f95c20d, 0x8cc531f9, 0x7eaeb2fa, - 0x30e349b1, 0xc288cab2, 0xd1d83946, 0x23b3ba45, - 0xf779deae, 0x05125dad, 0x1642ae59, 0xe4292d5a, - 0xba3a117e, 0x4851927d, 0x5b016189, 0xa96ae28a, - 0x7da08661, 0x8fcb0562, 0x9c9bf696, 0x6ef07595, - 0x417b1dbc, 0xb3109ebf, 0xa0406d4b, 0x522bee48, - 0x86e18aa3, 0x748a09a0, 0x67dafa54, 0x95b17957, - 0xcba24573, 0x39c9c670, 0x2a993584, 0xd8f2b687, - 0x0c38d26c, 0xfe53516f, 0xed03a29b, 0x1f682198, - 0x5125dad3, 0xa34e59d0, 0xb01eaa24, 0x42752927, - 0x96bf4dcc, 0x64d4cecf, 0x77843d3b, 0x85efbe38, - 0xdbfc821c, 0x2997011f, 0x3ac7f2eb, 0xc8ac71e8, - 0x1c661503, 0xee0d9600, 0xfd5d65f4, 0x0f36e6f7, - 0x61c69362, 0x93ad1061, 0x80fde395, 0x72966096, - 0xa65c047d, 0x5437877e, 0x4767748a, 0xb50cf789, - 0xeb1fcbad, 0x197448ae, 0x0a24bb5a, 0xf84f3859, - 0x2c855cb2, 0xdeeedfb1, 0xcdbe2c45, 0x3fd5af46, - 0x7198540d, 0x83f3d70e, 0x90a324fa, 0x62c8a7f9, - 0xb602c312, 0x44694011, 0x5739b3e5, 0xa55230e6, - 0xfb410cc2, 0x092a8fc1, 0x1a7a7c35, 0xe811ff36, - 0x3cdb9bdd, 0xceb018de, 0xdde0eb2a, 0x2f8b6829, - 0x82f63b78, 0x709db87b, 0x63cd4b8f, 0x91a6c88c, - 0x456cac67, 0xb7072f64, 0xa457dc90, 0x563c5f93, - 0x082f63b7, 0xfa44e0b4, 0xe9141340, 0x1b7f9043, - 0xcfb5f4a8, 0x3dde77ab, 0x2e8e845f, 0xdce5075c, - 0x92a8fc17, 0x60c37f14, 0x73938ce0, 0x81f80fe3, - 0x55326b08, 0xa759e80b, 0xb4091bff, 0x466298fc, - 0x1871a4d8, 0xea1a27db, 0xf94ad42f, 0x0b21572c, - 0xdfeb33c7, 0x2d80b0c4, 0x3ed04330, 0xccbbc033, - 0xa24bb5a6, 0x502036a5, 0x4370c551, 0xb11b4652, - 0x65d122b9, 0x97baa1ba, 0x84ea524e, 0x7681d14d, - 0x2892ed69, 0xdaf96e6a, 0xc9a99d9e, 0x3bc21e9d, - 0xef087a76, 0x1d63f975, 0x0e330a81, 0xfc588982, - 0xb21572c9, 0x407ef1ca, 0x532e023e, 0xa145813d, - 0x758fe5d6, 0x87e466d5, 0x94b49521, 0x66df1622, - 0x38cc2a06, 0xcaa7a905, 0xd9f75af1, 0x2b9cd9f2, - 0xff56bd19, 0x0d3d3e1a, 0x1e6dcdee, 0xec064eed, - 0xc38d26c4, 0x31e6a5c7, 0x22b65633, 0xd0ddd530, - 0x0417b1db, 0xf67c32d8, 0xe52cc12c, 0x1747422f, - 0x49547e0b, 0xbb3ffd08, 0xa86f0efc, 0x5a048dff, - 0x8ecee914, 0x7ca56a17, 0x6ff599e3, 0x9d9e1ae0, - 0xd3d3e1ab, 0x21b862a8, 0x32e8915c, 0xc083125f, - 0x144976b4, 0xe622f5b7, 0xf5720643, 0x07198540, - 0x590ab964, 0xab613a67, 0xb831c993, 0x4a5a4a90, - 0x9e902e7b, 0x6cfbad78, 0x7fab5e8c, 0x8dc0dd8f, - 0xe330a81a, 0x115b2b19, 0x020bd8ed, 0xf0605bee, - 0x24aa3f05, 0xd6c1bc06, 0xc5914ff2, 0x37faccf1, - 0x69e9f0d5, 0x9b8273d6, 0x88d28022, 0x7ab90321, - 0xae7367ca, 0x5c18e4c9, 0x4f48173d, 0xbd23943e, - 0xf36e6f75, 0x0105ec76, 0x12551f82, 0xe03e9c81, - 0x34f4f86a, 0xc69f7b69, 0xd5cf889d, 0x27a40b9e, - 0x79b737ba, 0x8bdcb4b9, 0x988c474d, 0x6ae7c44e, - 0xbe2da0a5, 0x4c4623a6, 0x5f16d052, 0xad7d5351 -}; -static const uint32_t table1_[256] = { - 0x00000000, 0x13a29877, 0x274530ee, 0x34e7a899, - 0x4e8a61dc, 0x5d28f9ab, 0x69cf5132, 0x7a6dc945, - 0x9d14c3b8, 0x8eb65bcf, 0xba51f356, 0xa9f36b21, - 0xd39ea264, 0xc03c3a13, 0xf4db928a, 0xe7790afd, - 0x3fc5f181, 0x2c6769f6, 0x1880c16f, 0x0b225918, - 0x714f905d, 0x62ed082a, 0x560aa0b3, 0x45a838c4, - 0xa2d13239, 0xb173aa4e, 0x859402d7, 0x96369aa0, - 0xec5b53e5, 0xfff9cb92, 0xcb1e630b, 0xd8bcfb7c, - 0x7f8be302, 0x6c297b75, 0x58ced3ec, 0x4b6c4b9b, - 0x310182de, 0x22a31aa9, 0x1644b230, 0x05e62a47, - 0xe29f20ba, 0xf13db8cd, 0xc5da1054, 0xd6788823, - 0xac154166, 0xbfb7d911, 0x8b507188, 0x98f2e9ff, - 0x404e1283, 0x53ec8af4, 0x670b226d, 0x74a9ba1a, - 0x0ec4735f, 0x1d66eb28, 0x298143b1, 0x3a23dbc6, - 0xdd5ad13b, 0xcef8494c, 0xfa1fe1d5, 0xe9bd79a2, - 0x93d0b0e7, 0x80722890, 0xb4958009, 0xa737187e, - 0xff17c604, 0xecb55e73, 0xd852f6ea, 0xcbf06e9d, - 0xb19da7d8, 0xa23f3faf, 0x96d89736, 0x857a0f41, - 0x620305bc, 0x71a19dcb, 0x45463552, 0x56e4ad25, - 0x2c896460, 0x3f2bfc17, 0x0bcc548e, 0x186eccf9, - 0xc0d23785, 0xd370aff2, 0xe797076b, 0xf4359f1c, - 0x8e585659, 0x9dface2e, 0xa91d66b7, 0xbabffec0, - 0x5dc6f43d, 0x4e646c4a, 0x7a83c4d3, 0x69215ca4, - 0x134c95e1, 0x00ee0d96, 0x3409a50f, 0x27ab3d78, - 0x809c2506, 0x933ebd71, 0xa7d915e8, 0xb47b8d9f, - 0xce1644da, 0xddb4dcad, 0xe9537434, 0xfaf1ec43, - 0x1d88e6be, 0x0e2a7ec9, 0x3acdd650, 0x296f4e27, - 0x53028762, 0x40a01f15, 0x7447b78c, 0x67e52ffb, - 0xbf59d487, 0xacfb4cf0, 0x981ce469, 0x8bbe7c1e, - 0xf1d3b55b, 0xe2712d2c, 0xd69685b5, 0xc5341dc2, - 0x224d173f, 0x31ef8f48, 0x050827d1, 0x16aabfa6, - 0x6cc776e3, 0x7f65ee94, 0x4b82460d, 0x5820de7a, - 0xfbc3faf9, 0xe861628e, 0xdc86ca17, 0xcf245260, - 0xb5499b25, 0xa6eb0352, 0x920cabcb, 0x81ae33bc, - 0x66d73941, 0x7575a136, 0x419209af, 0x523091d8, - 0x285d589d, 0x3bffc0ea, 0x0f186873, 0x1cbaf004, - 0xc4060b78, 0xd7a4930f, 0xe3433b96, 0xf0e1a3e1, - 0x8a8c6aa4, 0x992ef2d3, 0xadc95a4a, 0xbe6bc23d, - 0x5912c8c0, 0x4ab050b7, 0x7e57f82e, 0x6df56059, - 0x1798a91c, 0x043a316b, 0x30dd99f2, 0x237f0185, - 0x844819fb, 0x97ea818c, 0xa30d2915, 0xb0afb162, - 0xcac27827, 0xd960e050, 0xed8748c9, 0xfe25d0be, - 0x195cda43, 0x0afe4234, 0x3e19eaad, 0x2dbb72da, - 0x57d6bb9f, 0x447423e8, 0x70938b71, 0x63311306, - 0xbb8de87a, 0xa82f700d, 0x9cc8d894, 0x8f6a40e3, - 0xf50789a6, 0xe6a511d1, 0xd242b948, 0xc1e0213f, - 0x26992bc2, 0x353bb3b5, 0x01dc1b2c, 0x127e835b, - 0x68134a1e, 0x7bb1d269, 0x4f567af0, 0x5cf4e287, - 0x04d43cfd, 0x1776a48a, 0x23910c13, 0x30339464, - 0x4a5e5d21, 0x59fcc556, 0x6d1b6dcf, 0x7eb9f5b8, - 0x99c0ff45, 0x8a626732, 0xbe85cfab, 0xad2757dc, - 0xd74a9e99, 0xc4e806ee, 0xf00fae77, 0xe3ad3600, - 0x3b11cd7c, 0x28b3550b, 0x1c54fd92, 0x0ff665e5, - 0x759baca0, 0x663934d7, 0x52de9c4e, 0x417c0439, - 0xa6050ec4, 0xb5a796b3, 0x81403e2a, 0x92e2a65d, - 0xe88f6f18, 0xfb2df76f, 0xcfca5ff6, 0xdc68c781, - 0x7b5fdfff, 0x68fd4788, 0x5c1aef11, 0x4fb87766, - 0x35d5be23, 0x26772654, 0x12908ecd, 0x013216ba, - 0xe64b1c47, 0xf5e98430, 0xc10e2ca9, 0xd2acb4de, - 0xa8c17d9b, 0xbb63e5ec, 0x8f844d75, 0x9c26d502, - 0x449a2e7e, 0x5738b609, 0x63df1e90, 0x707d86e7, - 0x0a104fa2, 0x19b2d7d5, 0x2d557f4c, 0x3ef7e73b, - 0xd98eedc6, 0xca2c75b1, 0xfecbdd28, 0xed69455f, - 0x97048c1a, 0x84a6146d, 0xb041bcf4, 0xa3e32483 -}; -static const uint32_t table2_[256] = { - 0x00000000, 0xa541927e, 0x4f6f520d, 0xea2ec073, - 0x9edea41a, 0x3b9f3664, 0xd1b1f617, 0x74f06469, - 0x38513ec5, 0x9d10acbb, 0x773e6cc8, 0xd27ffeb6, - 0xa68f9adf, 0x03ce08a1, 0xe9e0c8d2, 0x4ca15aac, - 0x70a27d8a, 0xd5e3eff4, 0x3fcd2f87, 0x9a8cbdf9, - 0xee7cd990, 0x4b3d4bee, 0xa1138b9d, 0x045219e3, - 0x48f3434f, 0xedb2d131, 0x079c1142, 0xa2dd833c, - 0xd62de755, 0x736c752b, 0x9942b558, 0x3c032726, - 0xe144fb14, 0x4405696a, 0xae2ba919, 0x0b6a3b67, - 0x7f9a5f0e, 0xdadbcd70, 0x30f50d03, 0x95b49f7d, - 0xd915c5d1, 0x7c5457af, 0x967a97dc, 0x333b05a2, - 0x47cb61cb, 0xe28af3b5, 0x08a433c6, 0xade5a1b8, - 0x91e6869e, 0x34a714e0, 0xde89d493, 0x7bc846ed, - 0x0f382284, 0xaa79b0fa, 0x40577089, 0xe516e2f7, - 0xa9b7b85b, 0x0cf62a25, 0xe6d8ea56, 0x43997828, - 0x37691c41, 0x92288e3f, 0x78064e4c, 0xdd47dc32, - 0xc76580d9, 0x622412a7, 0x880ad2d4, 0x2d4b40aa, - 0x59bb24c3, 0xfcfab6bd, 0x16d476ce, 0xb395e4b0, - 0xff34be1c, 0x5a752c62, 0xb05bec11, 0x151a7e6f, - 0x61ea1a06, 0xc4ab8878, 0x2e85480b, 0x8bc4da75, - 0xb7c7fd53, 0x12866f2d, 0xf8a8af5e, 0x5de93d20, - 0x29195949, 0x8c58cb37, 0x66760b44, 0xc337993a, - 0x8f96c396, 0x2ad751e8, 0xc0f9919b, 0x65b803e5, - 0x1148678c, 0xb409f5f2, 0x5e273581, 0xfb66a7ff, - 0x26217bcd, 0x8360e9b3, 0x694e29c0, 0xcc0fbbbe, - 0xb8ffdfd7, 0x1dbe4da9, 0xf7908dda, 0x52d11fa4, - 0x1e704508, 0xbb31d776, 0x511f1705, 0xf45e857b, - 0x80aee112, 0x25ef736c, 0xcfc1b31f, 0x6a802161, - 0x56830647, 0xf3c29439, 0x19ec544a, 0xbcadc634, - 0xc85da25d, 0x6d1c3023, 0x8732f050, 0x2273622e, - 0x6ed23882, 0xcb93aafc, 0x21bd6a8f, 0x84fcf8f1, - 0xf00c9c98, 0x554d0ee6, 0xbf63ce95, 0x1a225ceb, - 0x8b277743, 0x2e66e53d, 0xc448254e, 0x6109b730, - 0x15f9d359, 0xb0b84127, 0x5a968154, 0xffd7132a, - 0xb3764986, 0x1637dbf8, 0xfc191b8b, 0x595889f5, - 0x2da8ed9c, 0x88e97fe2, 0x62c7bf91, 0xc7862def, - 0xfb850ac9, 0x5ec498b7, 0xb4ea58c4, 0x11abcaba, - 0x655baed3, 0xc01a3cad, 0x2a34fcde, 0x8f756ea0, - 0xc3d4340c, 0x6695a672, 0x8cbb6601, 0x29faf47f, - 0x5d0a9016, 0xf84b0268, 0x1265c21b, 0xb7245065, - 0x6a638c57, 0xcf221e29, 0x250cde5a, 0x804d4c24, - 0xf4bd284d, 0x51fcba33, 0xbbd27a40, 0x1e93e83e, - 0x5232b292, 0xf77320ec, 0x1d5de09f, 0xb81c72e1, - 0xccec1688, 0x69ad84f6, 0x83834485, 0x26c2d6fb, - 0x1ac1f1dd, 0xbf8063a3, 0x55aea3d0, 0xf0ef31ae, - 0x841f55c7, 0x215ec7b9, 0xcb7007ca, 0x6e3195b4, - 0x2290cf18, 0x87d15d66, 0x6dff9d15, 0xc8be0f6b, - 0xbc4e6b02, 0x190ff97c, 0xf321390f, 0x5660ab71, - 0x4c42f79a, 0xe90365e4, 0x032da597, 0xa66c37e9, - 0xd29c5380, 0x77ddc1fe, 0x9df3018d, 0x38b293f3, - 0x7413c95f, 0xd1525b21, 0x3b7c9b52, 0x9e3d092c, - 0xeacd6d45, 0x4f8cff3b, 0xa5a23f48, 0x00e3ad36, - 0x3ce08a10, 0x99a1186e, 0x738fd81d, 0xd6ce4a63, - 0xa23e2e0a, 0x077fbc74, 0xed517c07, 0x4810ee79, - 0x04b1b4d5, 0xa1f026ab, 0x4bdee6d8, 0xee9f74a6, - 0x9a6f10cf, 0x3f2e82b1, 0xd50042c2, 0x7041d0bc, - 0xad060c8e, 0x08479ef0, 0xe2695e83, 0x4728ccfd, - 0x33d8a894, 0x96993aea, 0x7cb7fa99, 0xd9f668e7, - 0x9557324b, 0x3016a035, 0xda386046, 0x7f79f238, - 0x0b899651, 0xaec8042f, 0x44e6c45c, 0xe1a75622, - 0xdda47104, 0x78e5e37a, 0x92cb2309, 0x378ab177, - 0x437ad51e, 0xe63b4760, 0x0c158713, 0xa954156d, - 0xe5f54fc1, 0x40b4ddbf, 0xaa9a1dcc, 0x0fdb8fb2, - 0x7b2bebdb, 0xde6a79a5, 0x3444b9d6, 0x91052ba8 -}; -static const uint32_t table3_[256] = { - 0x00000000, 0xdd45aab8, 0xbf672381, 0x62228939, - 0x7b2231f3, 0xa6679b4b, 0xc4451272, 0x1900b8ca, - 0xf64463e6, 0x2b01c95e, 0x49234067, 0x9466eadf, - 0x8d665215, 0x5023f8ad, 0x32017194, 0xef44db2c, - 0xe964b13d, 0x34211b85, 0x560392bc, 0x8b463804, - 0x924680ce, 0x4f032a76, 0x2d21a34f, 0xf06409f7, - 0x1f20d2db, 0xc2657863, 0xa047f15a, 0x7d025be2, - 0x6402e328, 0xb9474990, 0xdb65c0a9, 0x06206a11, - 0xd725148b, 0x0a60be33, 0x6842370a, 0xb5079db2, - 0xac072578, 0x71428fc0, 0x136006f9, 0xce25ac41, - 0x2161776d, 0xfc24ddd5, 0x9e0654ec, 0x4343fe54, - 0x5a43469e, 0x8706ec26, 0xe524651f, 0x3861cfa7, - 0x3e41a5b6, 0xe3040f0e, 0x81268637, 0x5c632c8f, - 0x45639445, 0x98263efd, 0xfa04b7c4, 0x27411d7c, - 0xc805c650, 0x15406ce8, 0x7762e5d1, 0xaa274f69, - 0xb327f7a3, 0x6e625d1b, 0x0c40d422, 0xd1057e9a, - 0xaba65fe7, 0x76e3f55f, 0x14c17c66, 0xc984d6de, - 0xd0846e14, 0x0dc1c4ac, 0x6fe34d95, 0xb2a6e72d, - 0x5de23c01, 0x80a796b9, 0xe2851f80, 0x3fc0b538, - 0x26c00df2, 0xfb85a74a, 0x99a72e73, 0x44e284cb, - 0x42c2eeda, 0x9f874462, 0xfda5cd5b, 0x20e067e3, - 0x39e0df29, 0xe4a57591, 0x8687fca8, 0x5bc25610, - 0xb4868d3c, 0x69c32784, 0x0be1aebd, 0xd6a40405, - 0xcfa4bccf, 0x12e11677, 0x70c39f4e, 0xad8635f6, - 0x7c834b6c, 0xa1c6e1d4, 0xc3e468ed, 0x1ea1c255, - 0x07a17a9f, 0xdae4d027, 0xb8c6591e, 0x6583f3a6, - 0x8ac7288a, 0x57828232, 0x35a00b0b, 0xe8e5a1b3, - 0xf1e51979, 0x2ca0b3c1, 0x4e823af8, 0x93c79040, - 0x95e7fa51, 0x48a250e9, 0x2a80d9d0, 0xf7c57368, - 0xeec5cba2, 0x3380611a, 0x51a2e823, 0x8ce7429b, - 0x63a399b7, 0xbee6330f, 0xdcc4ba36, 0x0181108e, - 0x1881a844, 0xc5c402fc, 0xa7e68bc5, 0x7aa3217d, - 0x52a0c93f, 0x8fe56387, 0xedc7eabe, 0x30824006, - 0x2982f8cc, 0xf4c75274, 0x96e5db4d, 0x4ba071f5, - 0xa4e4aad9, 0x79a10061, 0x1b838958, 0xc6c623e0, - 0xdfc69b2a, 0x02833192, 0x60a1b8ab, 0xbde41213, - 0xbbc47802, 0x6681d2ba, 0x04a35b83, 0xd9e6f13b, - 0xc0e649f1, 0x1da3e349, 0x7f816a70, 0xa2c4c0c8, - 0x4d801be4, 0x90c5b15c, 0xf2e73865, 0x2fa292dd, - 0x36a22a17, 0xebe780af, 0x89c50996, 0x5480a32e, - 0x8585ddb4, 0x58c0770c, 0x3ae2fe35, 0xe7a7548d, - 0xfea7ec47, 0x23e246ff, 0x41c0cfc6, 0x9c85657e, - 0x73c1be52, 0xae8414ea, 0xcca69dd3, 0x11e3376b, - 0x08e38fa1, 0xd5a62519, 0xb784ac20, 0x6ac10698, - 0x6ce16c89, 0xb1a4c631, 0xd3864f08, 0x0ec3e5b0, - 0x17c35d7a, 0xca86f7c2, 0xa8a47efb, 0x75e1d443, - 0x9aa50f6f, 0x47e0a5d7, 0x25c22cee, 0xf8878656, - 0xe1873e9c, 0x3cc29424, 0x5ee01d1d, 0x83a5b7a5, - 0xf90696d8, 0x24433c60, 0x4661b559, 0x9b241fe1, - 0x8224a72b, 0x5f610d93, 0x3d4384aa, 0xe0062e12, - 0x0f42f53e, 0xd2075f86, 0xb025d6bf, 0x6d607c07, - 0x7460c4cd, 0xa9256e75, 0xcb07e74c, 0x16424df4, - 0x106227e5, 0xcd278d5d, 0xaf050464, 0x7240aedc, - 0x6b401616, 0xb605bcae, 0xd4273597, 0x09629f2f, - 0xe6264403, 0x3b63eebb, 0x59416782, 0x8404cd3a, - 0x9d0475f0, 0x4041df48, 0x22635671, 0xff26fcc9, - 0x2e238253, 0xf36628eb, 0x9144a1d2, 0x4c010b6a, - 0x5501b3a0, 0x88441918, 0xea669021, 0x37233a99, - 0xd867e1b5, 0x05224b0d, 0x6700c234, 0xba45688c, - 0xa345d046, 0x7e007afe, 0x1c22f3c7, 0xc167597f, - 0xc747336e, 0x1a0299d6, 0x782010ef, 0xa565ba57, - 0xbc65029d, 0x6120a825, 0x0302211c, 0xde478ba4, - 0x31035088, 0xec46fa30, 0x8e647309, 0x5321d9b1, - 0x4a21617b, 0x9764cbc3, 0xf54642fa, 0x2803e842 -}; +namespace { -// Used to fetch a naturally-aligned 32-bit word in little endian byte-order -static inline uint32_t LE_LOAD32(const uint8_t *p) { - return DecodeFixed32(reinterpret_cast(p)); +const uint32_t kByteExtensionTable[256] = { + 0x00000000, 0xf26b8303, 0xe13b70f7, 0x1350f3f4, 0xc79a971f, 0x35f1141c, + 0x26a1e7e8, 0xd4ca64eb, 0x8ad958cf, 0x78b2dbcc, 0x6be22838, 0x9989ab3b, + 0x4d43cfd0, 0xbf284cd3, 0xac78bf27, 0x5e133c24, 0x105ec76f, 0xe235446c, + 0xf165b798, 0x030e349b, 0xd7c45070, 0x25afd373, 0x36ff2087, 0xc494a384, + 0x9a879fa0, 0x68ec1ca3, 0x7bbcef57, 0x89d76c54, 0x5d1d08bf, 0xaf768bbc, + 0xbc267848, 0x4e4dfb4b, 0x20bd8ede, 0xd2d60ddd, 0xc186fe29, 0x33ed7d2a, + 0xe72719c1, 0x154c9ac2, 0x061c6936, 0xf477ea35, 0xaa64d611, 0x580f5512, + 0x4b5fa6e6, 0xb93425e5, 0x6dfe410e, 0x9f95c20d, 0x8cc531f9, 0x7eaeb2fa, + 0x30e349b1, 0xc288cab2, 0xd1d83946, 0x23b3ba45, 0xf779deae, 0x05125dad, + 0x1642ae59, 0xe4292d5a, 0xba3a117e, 0x4851927d, 0x5b016189, 0xa96ae28a, + 0x7da08661, 0x8fcb0562, 0x9c9bf696, 0x6ef07595, 0x417b1dbc, 0xb3109ebf, + 0xa0406d4b, 0x522bee48, 0x86e18aa3, 0x748a09a0, 0x67dafa54, 0x95b17957, + 0xcba24573, 0x39c9c670, 0x2a993584, 0xd8f2b687, 0x0c38d26c, 0xfe53516f, + 0xed03a29b, 0x1f682198, 0x5125dad3, 0xa34e59d0, 0xb01eaa24, 0x42752927, + 0x96bf4dcc, 0x64d4cecf, 0x77843d3b, 0x85efbe38, 0xdbfc821c, 0x2997011f, + 0x3ac7f2eb, 0xc8ac71e8, 0x1c661503, 0xee0d9600, 0xfd5d65f4, 0x0f36e6f7, + 0x61c69362, 0x93ad1061, 0x80fde395, 0x72966096, 0xa65c047d, 0x5437877e, + 0x4767748a, 0xb50cf789, 0xeb1fcbad, 0x197448ae, 0x0a24bb5a, 0xf84f3859, + 0x2c855cb2, 0xdeeedfb1, 0xcdbe2c45, 0x3fd5af46, 0x7198540d, 0x83f3d70e, + 0x90a324fa, 0x62c8a7f9, 0xb602c312, 0x44694011, 0x5739b3e5, 0xa55230e6, + 0xfb410cc2, 0x092a8fc1, 0x1a7a7c35, 0xe811ff36, 0x3cdb9bdd, 0xceb018de, + 0xdde0eb2a, 0x2f8b6829, 0x82f63b78, 0x709db87b, 0x63cd4b8f, 0x91a6c88c, + 0x456cac67, 0xb7072f64, 0xa457dc90, 0x563c5f93, 0x082f63b7, 0xfa44e0b4, + 0xe9141340, 0x1b7f9043, 0xcfb5f4a8, 0x3dde77ab, 0x2e8e845f, 0xdce5075c, + 0x92a8fc17, 0x60c37f14, 0x73938ce0, 0x81f80fe3, 0x55326b08, 0xa759e80b, + 0xb4091bff, 0x466298fc, 0x1871a4d8, 0xea1a27db, 0xf94ad42f, 0x0b21572c, + 0xdfeb33c7, 0x2d80b0c4, 0x3ed04330, 0xccbbc033, 0xa24bb5a6, 0x502036a5, + 0x4370c551, 0xb11b4652, 0x65d122b9, 0x97baa1ba, 0x84ea524e, 0x7681d14d, + 0x2892ed69, 0xdaf96e6a, 0xc9a99d9e, 0x3bc21e9d, 0xef087a76, 0x1d63f975, + 0x0e330a81, 0xfc588982, 0xb21572c9, 0x407ef1ca, 0x532e023e, 0xa145813d, + 0x758fe5d6, 0x87e466d5, 0x94b49521, 0x66df1622, 0x38cc2a06, 0xcaa7a905, + 0xd9f75af1, 0x2b9cd9f2, 0xff56bd19, 0x0d3d3e1a, 0x1e6dcdee, 0xec064eed, + 0xc38d26c4, 0x31e6a5c7, 0x22b65633, 0xd0ddd530, 0x0417b1db, 0xf67c32d8, + 0xe52cc12c, 0x1747422f, 0x49547e0b, 0xbb3ffd08, 0xa86f0efc, 0x5a048dff, + 0x8ecee914, 0x7ca56a17, 0x6ff599e3, 0x9d9e1ae0, 0xd3d3e1ab, 0x21b862a8, + 0x32e8915c, 0xc083125f, 0x144976b4, 0xe622f5b7, 0xf5720643, 0x07198540, + 0x590ab964, 0xab613a67, 0xb831c993, 0x4a5a4a90, 0x9e902e7b, 0x6cfbad78, + 0x7fab5e8c, 0x8dc0dd8f, 0xe330a81a, 0x115b2b19, 0x020bd8ed, 0xf0605bee, + 0x24aa3f05, 0xd6c1bc06, 0xc5914ff2, 0x37faccf1, 0x69e9f0d5, 0x9b8273d6, + 0x88d28022, 0x7ab90321, 0xae7367ca, 0x5c18e4c9, 0x4f48173d, 0xbd23943e, + 0xf36e6f75, 0x0105ec76, 0x12551f82, 0xe03e9c81, 0x34f4f86a, 0xc69f7b69, + 0xd5cf889d, 0x27a40b9e, 0x79b737ba, 0x8bdcb4b9, 0x988c474d, 0x6ae7c44e, + 0xbe2da0a5, 0x4c4623a6, 0x5f16d052, 0xad7d5351}; + +const uint32_t kStrideExtensionTable0[256] = { + 0x00000000, 0x30d23865, 0x61a470ca, 0x517648af, 0xc348e194, 0xf39ad9f1, + 0xa2ec915e, 0x923ea93b, 0x837db5d9, 0xb3af8dbc, 0xe2d9c513, 0xd20bfd76, + 0x4035544d, 0x70e76c28, 0x21912487, 0x11431ce2, 0x03171d43, 0x33c52526, + 0x62b36d89, 0x526155ec, 0xc05ffcd7, 0xf08dc4b2, 0xa1fb8c1d, 0x9129b478, + 0x806aa89a, 0xb0b890ff, 0xe1ced850, 0xd11ce035, 0x4322490e, 0x73f0716b, + 0x228639c4, 0x125401a1, 0x062e3a86, 0x36fc02e3, 0x678a4a4c, 0x57587229, + 0xc566db12, 0xf5b4e377, 0xa4c2abd8, 0x941093bd, 0x85538f5f, 0xb581b73a, + 0xe4f7ff95, 0xd425c7f0, 0x461b6ecb, 0x76c956ae, 0x27bf1e01, 0x176d2664, + 0x053927c5, 0x35eb1fa0, 0x649d570f, 0x544f6f6a, 0xc671c651, 0xf6a3fe34, + 0xa7d5b69b, 0x97078efe, 0x8644921c, 0xb696aa79, 0xe7e0e2d6, 0xd732dab3, + 0x450c7388, 0x75de4bed, 0x24a80342, 0x147a3b27, 0x0c5c750c, 0x3c8e4d69, + 0x6df805c6, 0x5d2a3da3, 0xcf149498, 0xffc6acfd, 0xaeb0e452, 0x9e62dc37, + 0x8f21c0d5, 0xbff3f8b0, 0xee85b01f, 0xde57887a, 0x4c692141, 0x7cbb1924, + 0x2dcd518b, 0x1d1f69ee, 0x0f4b684f, 0x3f99502a, 0x6eef1885, 0x5e3d20e0, + 0xcc0389db, 0xfcd1b1be, 0xada7f911, 0x9d75c174, 0x8c36dd96, 0xbce4e5f3, + 0xed92ad5c, 0xdd409539, 0x4f7e3c02, 0x7fac0467, 0x2eda4cc8, 0x1e0874ad, + 0x0a724f8a, 0x3aa077ef, 0x6bd63f40, 0x5b040725, 0xc93aae1e, 0xf9e8967b, + 0xa89eded4, 0x984ce6b1, 0x890ffa53, 0xb9ddc236, 0xe8ab8a99, 0xd879b2fc, + 0x4a471bc7, 0x7a9523a2, 0x2be36b0d, 0x1b315368, 0x096552c9, 0x39b76aac, + 0x68c12203, 0x58131a66, 0xca2db35d, 0xfaff8b38, 0xab89c397, 0x9b5bfbf2, + 0x8a18e710, 0xbacadf75, 0xebbc97da, 0xdb6eafbf, 0x49500684, 0x79823ee1, + 0x28f4764e, 0x18264e2b, 0x18b8ea18, 0x286ad27d, 0x791c9ad2, 0x49cea2b7, + 0xdbf00b8c, 0xeb2233e9, 0xba547b46, 0x8a864323, 0x9bc55fc1, 0xab1767a4, + 0xfa612f0b, 0xcab3176e, 0x588dbe55, 0x685f8630, 0x3929ce9f, 0x09fbf6fa, + 0x1baff75b, 0x2b7dcf3e, 0x7a0b8791, 0x4ad9bff4, 0xd8e716cf, 0xe8352eaa, + 0xb9436605, 0x89915e60, 0x98d24282, 0xa8007ae7, 0xf9763248, 0xc9a40a2d, + 0x5b9aa316, 0x6b489b73, 0x3a3ed3dc, 0x0aecebb9, 0x1e96d09e, 0x2e44e8fb, + 0x7f32a054, 0x4fe09831, 0xddde310a, 0xed0c096f, 0xbc7a41c0, 0x8ca879a5, + 0x9deb6547, 0xad395d22, 0xfc4f158d, 0xcc9d2de8, 0x5ea384d3, 0x6e71bcb6, + 0x3f07f419, 0x0fd5cc7c, 0x1d81cddd, 0x2d53f5b8, 0x7c25bd17, 0x4cf78572, + 0xdec92c49, 0xee1b142c, 0xbf6d5c83, 0x8fbf64e6, 0x9efc7804, 0xae2e4061, + 0xff5808ce, 0xcf8a30ab, 0x5db49990, 0x6d66a1f5, 0x3c10e95a, 0x0cc2d13f, + 0x14e49f14, 0x2436a771, 0x7540efde, 0x4592d7bb, 0xd7ac7e80, 0xe77e46e5, + 0xb6080e4a, 0x86da362f, 0x97992acd, 0xa74b12a8, 0xf63d5a07, 0xc6ef6262, + 0x54d1cb59, 0x6403f33c, 0x3575bb93, 0x05a783f6, 0x17f38257, 0x2721ba32, + 0x7657f29d, 0x4685caf8, 0xd4bb63c3, 0xe4695ba6, 0xb51f1309, 0x85cd2b6c, + 0x948e378e, 0xa45c0feb, 0xf52a4744, 0xc5f87f21, 0x57c6d61a, 0x6714ee7f, + 0x3662a6d0, 0x06b09eb5, 0x12caa592, 0x22189df7, 0x736ed558, 0x43bced3d, + 0xd1824406, 0xe1507c63, 0xb02634cc, 0x80f40ca9, 0x91b7104b, 0xa165282e, + 0xf0136081, 0xc0c158e4, 0x52fff1df, 0x622dc9ba, 0x335b8115, 0x0389b970, + 0x11ddb8d1, 0x210f80b4, 0x7079c81b, 0x40abf07e, 0xd2955945, 0xe2476120, + 0xb331298f, 0x83e311ea, 0x92a00d08, 0xa272356d, 0xf3047dc2, 0xc3d645a7, + 0x51e8ec9c, 0x613ad4f9, 0x304c9c56, 0x009ea433}; + +const uint32_t kStrideExtensionTable1[256] = { + 0x00000000, 0x54075546, 0xa80eaa8c, 0xfc09ffca, 0x55f123e9, 0x01f676af, + 0xfdff8965, 0xa9f8dc23, 0xabe247d2, 0xffe51294, 0x03eced5e, 0x57ebb818, + 0xfe13643b, 0xaa14317d, 0x561dceb7, 0x021a9bf1, 0x5228f955, 0x062fac13, + 0xfa2653d9, 0xae21069f, 0x07d9dabc, 0x53de8ffa, 0xafd77030, 0xfbd02576, + 0xf9cabe87, 0xadcdebc1, 0x51c4140b, 0x05c3414d, 0xac3b9d6e, 0xf83cc828, + 0x043537e2, 0x503262a4, 0xa451f2aa, 0xf056a7ec, 0x0c5f5826, 0x58580d60, + 0xf1a0d143, 0xa5a78405, 0x59ae7bcf, 0x0da92e89, 0x0fb3b578, 0x5bb4e03e, + 0xa7bd1ff4, 0xf3ba4ab2, 0x5a429691, 0x0e45c3d7, 0xf24c3c1d, 0xa64b695b, + 0xf6790bff, 0xa27e5eb9, 0x5e77a173, 0x0a70f435, 0xa3882816, 0xf78f7d50, + 0x0b86829a, 0x5f81d7dc, 0x5d9b4c2d, 0x099c196b, 0xf595e6a1, 0xa192b3e7, + 0x086a6fc4, 0x5c6d3a82, 0xa064c548, 0xf463900e, 0x4d4f93a5, 0x1948c6e3, + 0xe5413929, 0xb1466c6f, 0x18beb04c, 0x4cb9e50a, 0xb0b01ac0, 0xe4b74f86, + 0xe6add477, 0xb2aa8131, 0x4ea37efb, 0x1aa42bbd, 0xb35cf79e, 0xe75ba2d8, + 0x1b525d12, 0x4f550854, 0x1f676af0, 0x4b603fb6, 0xb769c07c, 0xe36e953a, + 0x4a964919, 0x1e911c5f, 0xe298e395, 0xb69fb6d3, 0xb4852d22, 0xe0827864, + 0x1c8b87ae, 0x488cd2e8, 0xe1740ecb, 0xb5735b8d, 0x497aa447, 0x1d7df101, + 0xe91e610f, 0xbd193449, 0x4110cb83, 0x15179ec5, 0xbcef42e6, 0xe8e817a0, + 0x14e1e86a, 0x40e6bd2c, 0x42fc26dd, 0x16fb739b, 0xeaf28c51, 0xbef5d917, + 0x170d0534, 0x430a5072, 0xbf03afb8, 0xeb04fafe, 0xbb36985a, 0xef31cd1c, + 0x133832d6, 0x473f6790, 0xeec7bbb3, 0xbac0eef5, 0x46c9113f, 0x12ce4479, + 0x10d4df88, 0x44d38ace, 0xb8da7504, 0xecdd2042, 0x4525fc61, 0x1122a927, + 0xed2b56ed, 0xb92c03ab, 0x9a9f274a, 0xce98720c, 0x32918dc6, 0x6696d880, + 0xcf6e04a3, 0x9b6951e5, 0x6760ae2f, 0x3367fb69, 0x317d6098, 0x657a35de, + 0x9973ca14, 0xcd749f52, 0x648c4371, 0x308b1637, 0xcc82e9fd, 0x9885bcbb, + 0xc8b7de1f, 0x9cb08b59, 0x60b97493, 0x34be21d5, 0x9d46fdf6, 0xc941a8b0, + 0x3548577a, 0x614f023c, 0x635599cd, 0x3752cc8b, 0xcb5b3341, 0x9f5c6607, + 0x36a4ba24, 0x62a3ef62, 0x9eaa10a8, 0xcaad45ee, 0x3eced5e0, 0x6ac980a6, + 0x96c07f6c, 0xc2c72a2a, 0x6b3ff609, 0x3f38a34f, 0xc3315c85, 0x973609c3, + 0x952c9232, 0xc12bc774, 0x3d2238be, 0x69256df8, 0xc0ddb1db, 0x94dae49d, + 0x68d31b57, 0x3cd44e11, 0x6ce62cb5, 0x38e179f3, 0xc4e88639, 0x90efd37f, + 0x39170f5c, 0x6d105a1a, 0x9119a5d0, 0xc51ef096, 0xc7046b67, 0x93033e21, + 0x6f0ac1eb, 0x3b0d94ad, 0x92f5488e, 0xc6f21dc8, 0x3afbe202, 0x6efcb744, + 0xd7d0b4ef, 0x83d7e1a9, 0x7fde1e63, 0x2bd94b25, 0x82219706, 0xd626c240, + 0x2a2f3d8a, 0x7e2868cc, 0x7c32f33d, 0x2835a67b, 0xd43c59b1, 0x803b0cf7, + 0x29c3d0d4, 0x7dc48592, 0x81cd7a58, 0xd5ca2f1e, 0x85f84dba, 0xd1ff18fc, + 0x2df6e736, 0x79f1b270, 0xd0096e53, 0x840e3b15, 0x7807c4df, 0x2c009199, + 0x2e1a0a68, 0x7a1d5f2e, 0x8614a0e4, 0xd213f5a2, 0x7beb2981, 0x2fec7cc7, + 0xd3e5830d, 0x87e2d64b, 0x73814645, 0x27861303, 0xdb8fecc9, 0x8f88b98f, + 0x267065ac, 0x727730ea, 0x8e7ecf20, 0xda799a66, 0xd8630197, 0x8c6454d1, + 0x706dab1b, 0x246afe5d, 0x8d92227e, 0xd9957738, 0x259c88f2, 0x719bddb4, + 0x21a9bf10, 0x75aeea56, 0x89a7159c, 0xdda040da, 0x74589cf9, 0x205fc9bf, + 0xdc563675, 0x88516333, 0x8a4bf8c2, 0xde4cad84, 0x2245524e, 0x76420708, + 0xdfbadb2b, 0x8bbd8e6d, 0x77b471a7, 0x23b324e1}; + +const uint32_t kStrideExtensionTable2[256] = { + 0x00000000, 0x678efd01, 0xcf1dfa02, 0xa8930703, 0x9bd782f5, 0xfc597ff4, + 0x54ca78f7, 0x334485f6, 0x3243731b, 0x55cd8e1a, 0xfd5e8919, 0x9ad07418, + 0xa994f1ee, 0xce1a0cef, 0x66890bec, 0x0107f6ed, 0x6486e636, 0x03081b37, + 0xab9b1c34, 0xcc15e135, 0xff5164c3, 0x98df99c2, 0x304c9ec1, 0x57c263c0, + 0x56c5952d, 0x314b682c, 0x99d86f2f, 0xfe56922e, 0xcd1217d8, 0xaa9cead9, + 0x020fedda, 0x658110db, 0xc90dcc6c, 0xae83316d, 0x0610366e, 0x619ecb6f, + 0x52da4e99, 0x3554b398, 0x9dc7b49b, 0xfa49499a, 0xfb4ebf77, 0x9cc04276, + 0x34534575, 0x53ddb874, 0x60993d82, 0x0717c083, 0xaf84c780, 0xc80a3a81, + 0xad8b2a5a, 0xca05d75b, 0x6296d058, 0x05182d59, 0x365ca8af, 0x51d255ae, + 0xf94152ad, 0x9ecfafac, 0x9fc85941, 0xf846a440, 0x50d5a343, 0x375b5e42, + 0x041fdbb4, 0x639126b5, 0xcb0221b6, 0xac8cdcb7, 0x97f7ee29, 0xf0791328, + 0x58ea142b, 0x3f64e92a, 0x0c206cdc, 0x6bae91dd, 0xc33d96de, 0xa4b36bdf, + 0xa5b49d32, 0xc23a6033, 0x6aa96730, 0x0d279a31, 0x3e631fc7, 0x59ede2c6, + 0xf17ee5c5, 0x96f018c4, 0xf371081f, 0x94fff51e, 0x3c6cf21d, 0x5be20f1c, + 0x68a68aea, 0x0f2877eb, 0xa7bb70e8, 0xc0358de9, 0xc1327b04, 0xa6bc8605, + 0x0e2f8106, 0x69a17c07, 0x5ae5f9f1, 0x3d6b04f0, 0x95f803f3, 0xf276fef2, + 0x5efa2245, 0x3974df44, 0x91e7d847, 0xf6692546, 0xc52da0b0, 0xa2a35db1, + 0x0a305ab2, 0x6dbea7b3, 0x6cb9515e, 0x0b37ac5f, 0xa3a4ab5c, 0xc42a565d, + 0xf76ed3ab, 0x90e02eaa, 0x387329a9, 0x5ffdd4a8, 0x3a7cc473, 0x5df23972, + 0xf5613e71, 0x92efc370, 0xa1ab4686, 0xc625bb87, 0x6eb6bc84, 0x09384185, + 0x083fb768, 0x6fb14a69, 0xc7224d6a, 0xa0acb06b, 0x93e8359d, 0xf466c89c, + 0x5cf5cf9f, 0x3b7b329e, 0x2a03aaa3, 0x4d8d57a2, 0xe51e50a1, 0x8290ada0, + 0xb1d42856, 0xd65ad557, 0x7ec9d254, 0x19472f55, 0x1840d9b8, 0x7fce24b9, + 0xd75d23ba, 0xb0d3debb, 0x83975b4d, 0xe419a64c, 0x4c8aa14f, 0x2b045c4e, + 0x4e854c95, 0x290bb194, 0x8198b697, 0xe6164b96, 0xd552ce60, 0xb2dc3361, + 0x1a4f3462, 0x7dc1c963, 0x7cc63f8e, 0x1b48c28f, 0xb3dbc58c, 0xd455388d, + 0xe711bd7b, 0x809f407a, 0x280c4779, 0x4f82ba78, 0xe30e66cf, 0x84809bce, + 0x2c139ccd, 0x4b9d61cc, 0x78d9e43a, 0x1f57193b, 0xb7c41e38, 0xd04ae339, + 0xd14d15d4, 0xb6c3e8d5, 0x1e50efd6, 0x79de12d7, 0x4a9a9721, 0x2d146a20, + 0x85876d23, 0xe2099022, 0x878880f9, 0xe0067df8, 0x48957afb, 0x2f1b87fa, + 0x1c5f020c, 0x7bd1ff0d, 0xd342f80e, 0xb4cc050f, 0xb5cbf3e2, 0xd2450ee3, + 0x7ad609e0, 0x1d58f4e1, 0x2e1c7117, 0x49928c16, 0xe1018b15, 0x868f7614, + 0xbdf4448a, 0xda7ab98b, 0x72e9be88, 0x15674389, 0x2623c67f, 0x41ad3b7e, + 0xe93e3c7d, 0x8eb0c17c, 0x8fb73791, 0xe839ca90, 0x40aacd93, 0x27243092, + 0x1460b564, 0x73ee4865, 0xdb7d4f66, 0xbcf3b267, 0xd972a2bc, 0xbefc5fbd, + 0x166f58be, 0x71e1a5bf, 0x42a52049, 0x252bdd48, 0x8db8da4b, 0xea36274a, + 0xeb31d1a7, 0x8cbf2ca6, 0x242c2ba5, 0x43a2d6a4, 0x70e65352, 0x1768ae53, + 0xbffba950, 0xd8755451, 0x74f988e6, 0x137775e7, 0xbbe472e4, 0xdc6a8fe5, + 0xef2e0a13, 0x88a0f712, 0x2033f011, 0x47bd0d10, 0x46bafbfd, 0x213406fc, + 0x89a701ff, 0xee29fcfe, 0xdd6d7908, 0xbae38409, 0x1270830a, 0x75fe7e0b, + 0x107f6ed0, 0x77f193d1, 0xdf6294d2, 0xb8ec69d3, 0x8ba8ec25, 0xec261124, + 0x44b51627, 0x233beb26, 0x223c1dcb, 0x45b2e0ca, 0xed21e7c9, 0x8aaf1ac8, + 0xb9eb9f3e, 0xde65623f, 0x76f6653c, 0x1178983d}; + +const uint32_t kStrideExtensionTable3[256] = { + 0x00000000, 0xf20c0dfe, 0xe1f46d0d, 0x13f860f3, 0xc604aceb, 0x3408a115, + 0x27f0c1e6, 0xd5fccc18, 0x89e52f27, 0x7be922d9, 0x6811422a, 0x9a1d4fd4, + 0x4fe183cc, 0xbded8e32, 0xae15eec1, 0x5c19e33f, 0x162628bf, 0xe42a2541, + 0xf7d245b2, 0x05de484c, 0xd0228454, 0x222e89aa, 0x31d6e959, 0xc3dae4a7, + 0x9fc30798, 0x6dcf0a66, 0x7e376a95, 0x8c3b676b, 0x59c7ab73, 0xabcba68d, + 0xb833c67e, 0x4a3fcb80, 0x2c4c517e, 0xde405c80, 0xcdb83c73, 0x3fb4318d, + 0xea48fd95, 0x1844f06b, 0x0bbc9098, 0xf9b09d66, 0xa5a97e59, 0x57a573a7, + 0x445d1354, 0xb6511eaa, 0x63add2b2, 0x91a1df4c, 0x8259bfbf, 0x7055b241, + 0x3a6a79c1, 0xc866743f, 0xdb9e14cc, 0x29921932, 0xfc6ed52a, 0x0e62d8d4, + 0x1d9ab827, 0xef96b5d9, 0xb38f56e6, 0x41835b18, 0x527b3beb, 0xa0773615, + 0x758bfa0d, 0x8787f7f3, 0x947f9700, 0x66739afe, 0x5898a2fc, 0xaa94af02, + 0xb96ccff1, 0x4b60c20f, 0x9e9c0e17, 0x6c9003e9, 0x7f68631a, 0x8d646ee4, + 0xd17d8ddb, 0x23718025, 0x3089e0d6, 0xc285ed28, 0x17792130, 0xe5752cce, + 0xf68d4c3d, 0x048141c3, 0x4ebe8a43, 0xbcb287bd, 0xaf4ae74e, 0x5d46eab0, + 0x88ba26a8, 0x7ab62b56, 0x694e4ba5, 0x9b42465b, 0xc75ba564, 0x3557a89a, + 0x26afc869, 0xd4a3c597, 0x015f098f, 0xf3530471, 0xe0ab6482, 0x12a7697c, + 0x74d4f382, 0x86d8fe7c, 0x95209e8f, 0x672c9371, 0xb2d05f69, 0x40dc5297, + 0x53243264, 0xa1283f9a, 0xfd31dca5, 0x0f3dd15b, 0x1cc5b1a8, 0xeec9bc56, + 0x3b35704e, 0xc9397db0, 0xdac11d43, 0x28cd10bd, 0x62f2db3d, 0x90fed6c3, + 0x8306b630, 0x710abbce, 0xa4f677d6, 0x56fa7a28, 0x45021adb, 0xb70e1725, + 0xeb17f41a, 0x191bf9e4, 0x0ae39917, 0xf8ef94e9, 0x2d1358f1, 0xdf1f550f, + 0xcce735fc, 0x3eeb3802, 0xb13145f8, 0x433d4806, 0x50c528f5, 0xa2c9250b, + 0x7735e913, 0x8539e4ed, 0x96c1841e, 0x64cd89e0, 0x38d46adf, 0xcad86721, + 0xd92007d2, 0x2b2c0a2c, 0xfed0c634, 0x0cdccbca, 0x1f24ab39, 0xed28a6c7, + 0xa7176d47, 0x551b60b9, 0x46e3004a, 0xb4ef0db4, 0x6113c1ac, 0x931fcc52, + 0x80e7aca1, 0x72eba15f, 0x2ef24260, 0xdcfe4f9e, 0xcf062f6d, 0x3d0a2293, + 0xe8f6ee8b, 0x1afae375, 0x09028386, 0xfb0e8e78, 0x9d7d1486, 0x6f711978, + 0x7c89798b, 0x8e857475, 0x5b79b86d, 0xa975b593, 0xba8dd560, 0x4881d89e, + 0x14983ba1, 0xe694365f, 0xf56c56ac, 0x07605b52, 0xd29c974a, 0x20909ab4, + 0x3368fa47, 0xc164f7b9, 0x8b5b3c39, 0x795731c7, 0x6aaf5134, 0x98a35cca, + 0x4d5f90d2, 0xbf539d2c, 0xacabfddf, 0x5ea7f021, 0x02be131e, 0xf0b21ee0, + 0xe34a7e13, 0x114673ed, 0xc4babff5, 0x36b6b20b, 0x254ed2f8, 0xd742df06, + 0xe9a9e704, 0x1ba5eafa, 0x085d8a09, 0xfa5187f7, 0x2fad4bef, 0xdda14611, + 0xce5926e2, 0x3c552b1c, 0x604cc823, 0x9240c5dd, 0x81b8a52e, 0x73b4a8d0, + 0xa64864c8, 0x54446936, 0x47bc09c5, 0xb5b0043b, 0xff8fcfbb, 0x0d83c245, + 0x1e7ba2b6, 0xec77af48, 0x398b6350, 0xcb876eae, 0xd87f0e5d, 0x2a7303a3, + 0x766ae09c, 0x8466ed62, 0x979e8d91, 0x6592806f, 0xb06e4c77, 0x42624189, + 0x519a217a, 0xa3962c84, 0xc5e5b67a, 0x37e9bb84, 0x2411db77, 0xd61dd689, + 0x03e11a91, 0xf1ed176f, 0xe215779c, 0x10197a62, 0x4c00995d, 0xbe0c94a3, + 0xadf4f450, 0x5ff8f9ae, 0x8a0435b6, 0x78083848, 0x6bf058bb, 0x99fc5545, + 0xd3c39ec5, 0x21cf933b, 0x3237f3c8, 0xc03bfe36, 0x15c7322e, 0xe7cb3fd0, + 0xf4335f23, 0x063f52dd, 0x5a26b1e2, 0xa82abc1c, 0xbbd2dcef, 0x49ded111, + 0x9c221d09, 0x6e2e10f7, 0x7dd67004, 0x8fda7dfa}; + +// CRCs are pre- and post- conditioned by xoring with all ones. +static constexpr const uint32_t kCRC32Xor = static_cast(0xffffffffU); + +// Reads a little-endian 32-bit integer from a 32-bit-aligned buffer. +inline uint32_t ReadUint32LE(const uint8_t* buffer) { + return DecodeFixed32(reinterpret_cast(buffer)); } -uint32_t Extend(uint32_t crc, const char* buf, size_t size) { - const uint8_t *p = reinterpret_cast(buf); - const uint8_t *e = p + size; - uint32_t l = crc ^ 0xffffffffu; +// Returns the smallest address >= the given address that is aligned to N bytes. +// +// N must be a power of two. +template +constexpr inline const uint8_t* RoundUp(const uint8_t* pointer) { + return reinterpret_cast( + (reinterpret_cast(pointer) + (N - 1)) & + ~static_cast(N - 1)); +} -#define STEP1 do { \ - int c = (l & 0xff) ^ *p++; \ - l = table0_[c] ^ (l >> 8); \ -} while (0) -#define STEP4 do { \ - uint32_t c = l ^ LE_LOAD32(p); \ - p += 4; \ - l = table3_[c & 0xff] ^ \ - table2_[(c >> 8) & 0xff] ^ \ - table1_[(c >> 16) & 0xff] ^ \ - table0_[c >> 24]; \ -} while (0) +} // namespace - // Point x at first 4-byte aligned byte in string. This might be - // just past the end of the string. - const uintptr_t pval = reinterpret_cast(p); - const uint8_t* x = reinterpret_cast(((pval + 3) >> 2) << 2); +// Determine if the CPU running this program can accelerate the CRC32C +// calculation. +static bool CanAccelerateCRC32C() { + // port::AcceleretedCRC32C returns zero when unable to accelerate. + static const char kTestCRCBuffer[] = "TestCRCBuffer"; + static const char kBufSize = sizeof(kTestCRCBuffer) - 1; + static const uint32_t kTestCRCValue = 0xdcbc59fa; + + return port::AcceleratedCRC32C(0, kTestCRCBuffer, kBufSize) == kTestCRCValue; +} + +uint32_t Extend(uint32_t crc, const char* data, size_t n) { + static bool accelerate = CanAccelerateCRC32C(); + if (accelerate) { + return port::AcceleratedCRC32C(crc, data, n); + } + + const uint8_t* p = reinterpret_cast(data); + const uint8_t* e = p + n; + uint32_t l = crc ^ kCRC32Xor; + +// Process one byte at a time. +#define STEP1 \ + do { \ + int c = (l & 0xff) ^ *p++; \ + l = kByteExtensionTable[c] ^ (l >> 8); \ + } while (0) + +// Process one of the 4 strides of 4-byte data. +#define STEP4(s) \ + do { \ + crc##s = ReadUint32LE(p + s * 4) ^ kStrideExtensionTable3[crc##s & 0xff] ^ \ + kStrideExtensionTable2[(crc##s >> 8) & 0xff] ^ \ + kStrideExtensionTable1[(crc##s >> 16) & 0xff] ^ \ + kStrideExtensionTable0[crc##s >> 24]; \ + } while (0) + +// Process a 16-byte swath of 4 strides, each of which has 4 bytes of data. +#define STEP16 \ + do { \ + STEP4(0); \ + STEP4(1); \ + STEP4(2); \ + STEP4(3); \ + p += 16; \ + } while (0) + +// Process 4 bytes that were already loaded into a word. +#define STEP4W(w) \ + do { \ + w ^= l; \ + for (size_t i = 0; i < 4; ++i) { \ + w = (w >> 8) ^ kByteExtensionTable[w & 0xff]; \ + } \ + l = w; \ + } while (0) + + // Point x at first 4-byte aligned byte in the buffer. This might be past the + // end of the buffer. + const uint8_t* x = RoundUp<4>(p); if (x <= e) { - // Process bytes until finished or p is 4-byte aligned + // Process bytes p is 4-byte aligned. while (p != x) { STEP1; } } - // Process bytes 16 at a time - while ((e-p) >= 16) { - STEP4; STEP4; STEP4; STEP4; + + if ((e - p) >= 16) { + // Load a 16-byte swath into the stride partial results. + uint32_t crc0 = ReadUint32LE(p + 0 * 4) ^ l; + uint32_t crc1 = ReadUint32LE(p + 1 * 4); + uint32_t crc2 = ReadUint32LE(p + 2 * 4); + uint32_t crc3 = ReadUint32LE(p + 3 * 4); + p += 16; + + // It is possible to get better speeds (at least on x86) by interleaving + // prefetching 256 bytes ahead with processing 64 bytes at a time. See the + // portable implementation in https://github.com/google/crc32c/. + + // Process one 16-byte swath at a time. + while ((e - p) >= 16) { + STEP16; + } + + // Advance one word at a time as far as possible. + while ((e - p) >= 4) { + STEP4(0); + uint32_t tmp = crc0; + crc0 = crc1; + crc1 = crc2; + crc2 = crc3; + crc3 = tmp; + p += 4; + } + + // Combine the 4 partial stride results. + l = 0; + STEP4W(crc0); + STEP4W(crc1); + STEP4W(crc2); + STEP4W(crc3); } - // Process bytes 4 at a time - while ((e-p) >= 4) { - STEP4; - } - // Process the last few bytes + + // Process the last few bytes. while (p != e) { STEP1; } +#undef STEP4W +#undef STEP16 #undef STEP4 #undef STEP1 - return l ^ 0xffffffffu; + return l ^ kCRC32Xor; } } // namespace crc32c diff --git a/util/crc32c.h b/util/crc32c.h index 1d7e5c0..98fabb0 100644 --- a/util/crc32c.h +++ b/util/crc32c.h @@ -14,12 +14,10 @@ namespace crc32c { // Return the crc32c of concat(A, data[0,n-1]) where init_crc is the // crc32c of some string A. Extend() is often used to maintain the // crc32c of a stream of data. -extern uint32_t Extend(uint32_t init_crc, const char* data, size_t n); +uint32_t Extend(uint32_t init_crc, const char* data, size_t n); // Return the crc32c of data[0,n-1] -inline uint32_t Value(const char* data, size_t n) { - return Extend(0, data, n); -} +inline uint32_t Value(const char* data, size_t n) { return Extend(0, data, n); } static const uint32_t kMaskDelta = 0xa282ead8ul; diff --git a/util/crc32c_test.cc b/util/crc32c_test.cc index 4b957ee..dbd2ba4 100644 --- a/util/crc32c_test.cc +++ b/util/crc32c_test.cc @@ -8,7 +8,7 @@ namespace leveldb { namespace crc32c { -class CRC { }; +class CRC {}; TEST(CRC, StandardResults) { // From rfc3720 section B.4. @@ -31,29 +31,18 @@ TEST(CRC, StandardResults) { ASSERT_EQ(0x113fdb5c, Value(buf, sizeof(buf))); unsigned char data[48] = { - 0x01, 0xc0, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x14, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x04, 0x00, - 0x00, 0x00, 0x00, 0x14, - 0x00, 0x00, 0x00, 0x18, - 0x28, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x02, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, + 0x01, 0xc0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, + 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x18, 0x28, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, }; ASSERT_EQ(0xd9963a56, Value(reinterpret_cast(data), sizeof(data))); } -TEST(CRC, Values) { - ASSERT_NE(Value("a", 1), Value("foo", 3)); -} +TEST(CRC, Values) { ASSERT_NE(Value("a", 1), Value("foo", 3)); } TEST(CRC, Extend) { - ASSERT_EQ(Value("hello world", 11), - Extend(Value("hello ", 6), "world", 5)); + ASSERT_EQ(Value("hello world", 11), Extend(Value("hello ", 6), "world", 5)); } TEST(CRC, Mask) { @@ -67,6 +56,4 @@ TEST(CRC, Mask) { } // namespace crc32c } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/env.cc b/util/env.cc index c2600e9..d2f0aef 100644 --- a/util/env.cc +++ b/util/env.cc @@ -6,26 +6,24 @@ namespace leveldb { -Env::~Env() { +Env::~Env() = default; + +Status Env::NewAppendableFile(const std::string& fname, WritableFile** result) { + return Status::NotSupported("NewAppendableFile", fname); } -SequentialFile::~SequentialFile() { -} +SequentialFile::~SequentialFile() = default; -RandomAccessFile::~RandomAccessFile() { -} +RandomAccessFile::~RandomAccessFile() = default; -WritableFile::~WritableFile() { -} +WritableFile::~WritableFile() = default; -Logger::~Logger() { -} +Logger::~Logger() = default; -FileLock::~FileLock() { -} +FileLock::~FileLock() = default; void Log(Logger* info_log, const char* format, ...) { - if (info_log != NULL) { + if (info_log != nullptr) { va_list ap; va_start(ap, format); info_log->Logv(format, ap); @@ -34,8 +32,7 @@ void Log(Logger* info_log, const char* format, ...) { } static Status DoWriteStringToFile(Env* env, const Slice& data, - const std::string& fname, - bool should_sync) { + const std::string& fname, bool should_sync) { WritableFile* file; Status s = env->NewWritableFile(fname, &file); if (!s.ok()) { @@ -90,7 +87,6 @@ Status ReadFileToString(Env* env, const std::string& fname, std::string* data) { return s; } -EnvWrapper::~EnvWrapper() { -} +EnvWrapper::~EnvWrapper() {} } // namespace leveldb diff --git a/util/env_posix.cc b/util/env_posix.cc index ad56131..420e709 100644 --- a/util/env_posix.cc +++ b/util/env_posix.cc @@ -3,603 +3,891 @@ // found in the LICENSE file. See the AUTHORS file for names of contributors. #include -#include #include #include -#include -#include -#include #include +#include #include #include #include -#include #include -#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include #include +#include +#include +#include +#include + #include "leveldb/env.h" #include "leveldb/slice.h" +#include "leveldb/status.h" #include "port/port.h" -#include "util/logging.h" -#include "util/mutexlock.h" +#include "port/thread_annotations.h" +#include "util/env_posix_test_helper.h" #include "util/posix_logger.h" namespace leveldb { namespace { -static Status IOError(const std::string& context, int err_number) { - return Status::IOError(context, strerror(err_number)); +// Set by EnvPosixTestHelper::SetReadOnlyMMapLimit() and MaxOpenFiles(). +int g_open_read_only_file_limit = -1; + +// Up to 1000 mmap regions for 64-bit binaries; none for 32-bit. +constexpr const int kDefaultMmapLimit = (sizeof(void*) >= 8) ? 1000 : 0; + +// Can be set using EnvPosixTestHelper::SetReadOnlyMMapLimit. +int g_mmap_limit = kDefaultMmapLimit; + +// Common flags defined for all posix open operations +#if defined(HAVE_O_CLOEXEC) +constexpr const int kOpenBaseFlags = O_CLOEXEC; +#else +constexpr const int kOpenBaseFlags = 0; +#endif // defined(HAVE_O_CLOEXEC) + +constexpr const size_t kWritableFileBufferSize = 65536; + +Status PosixError(const std::string& context, int error_number) { + if (error_number == ENOENT) { + return Status::NotFound(context, std::strerror(error_number)); + } else { + return Status::IOError(context, std::strerror(error_number)); + } } -class PosixSequentialFile: public SequentialFile { - private: - std::string filename_; - FILE* file_; - +// Helper class to limit resource usage to avoid exhaustion. +// Currently used to limit read-only file descriptors and mmap file usage +// so that we do not run out of file descriptors or virtual memory, or run into +// kernel performance problems for very large databases. +class Limiter { public: - PosixSequentialFile(const std::string& fname, FILE* f) - : filename_(fname), file_(f) { } - virtual ~PosixSequentialFile() { fclose(file_); } + // Limit maximum number of resources to |max_acquires|. + Limiter(int max_acquires) : acquires_allowed_(max_acquires) {} - virtual Status Read(size_t n, Slice* result, char* scratch) { - Status s; - size_t r = fread_unlocked(scratch, 1, n, file_); - *result = Slice(scratch, r); - if (r < n) { - if (feof(file_)) { - // We leave status as ok if we hit the end of the file - } else { - // A partial read with an error: return a non-ok status - s = IOError(filename_, errno); - } - } - return s; - } + Limiter(const Limiter&) = delete; + Limiter operator=(const Limiter&) = delete; - virtual Status Skip(uint64_t n) { - if (fseek(file_, n, SEEK_CUR)) { - return IOError(filename_, errno); - } - return Status::OK(); - } -}; - -// pread() based random-access -class PosixRandomAccessFile: public RandomAccessFile { - private: - std::string filename_; - int fd_; - - public: - PosixRandomAccessFile(const std::string& fname, int fd) - : filename_(fname), fd_(fd) { } - virtual ~PosixRandomAccessFile() { close(fd_); } - - virtual Status Read(uint64_t offset, size_t n, Slice* result, - char* scratch) const { - Status s; - ssize_t r = pread(fd_, scratch, n, static_cast(offset)); - *result = Slice(scratch, (r < 0) ? 0 : r); - if (r < 0) { - // An error: return a non-ok status - s = IOError(filename_, errno); - } - return s; - } -}; - -// Helper class to limit mmap file usage so that we do not end up -// running out virtual memory or running into kernel performance -// problems for very large databases. -class MmapLimiter { - public: - // Up to 1000 mmaps for 64-bit binaries; none for smaller pointer sizes. - MmapLimiter() { - SetAllowed(sizeof(void*) >= 8 ? 1000 : 0); - } - - // If another mmap slot is available, acquire it and return true. + // If another resource is available, acquire it and return true. // Else return false. bool Acquire() { - if (GetAllowed() <= 0) { - return false; - } - MutexLock l(&mu_); - intptr_t x = GetAllowed(); - if (x <= 0) { - return false; - } else { - SetAllowed(x - 1); - return true; - } + int old_acquires_allowed = + acquires_allowed_.fetch_sub(1, std::memory_order_relaxed); + + if (old_acquires_allowed > 0) return true; + + acquires_allowed_.fetch_add(1, std::memory_order_relaxed); + return false; } - // Release a slot acquired by a previous call to Acquire() that returned true. - void Release() { - MutexLock l(&mu_); - SetAllowed(GetAllowed() + 1); - } + // Release a resource acquired by a previous call to Acquire() that returned + // true. + void Release() { acquires_allowed_.fetch_add(1, std::memory_order_relaxed); } private: - port::Mutex mu_; - port::AtomicPointer allowed_; - - intptr_t GetAllowed() const { - return reinterpret_cast(allowed_.Acquire_Load()); - } - - // REQUIRES: mu_ must be held - void SetAllowed(intptr_t v) { - allowed_.Release_Store(reinterpret_cast(v)); - } - - MmapLimiter(const MmapLimiter&); - void operator=(const MmapLimiter&); + // The number of available resources. + // + // This is a counter and is not tied to the invariants of any other class, so + // it can be operated on safely using std::memory_order_relaxed. + std::atomic acquires_allowed_; }; -// mmap() based random-access -class PosixMmapReadableFile: public RandomAccessFile { - private: - std::string filename_; - void* mmapped_region_; - size_t length_; - MmapLimiter* limiter_; - +// Implements sequential read access in a file using read(). +// +// Instances of this class are thread-friendly but not thread-safe, as required +// by the SequentialFile API. +class PosixSequentialFile final : public SequentialFile { public: - // base[0,length-1] contains the mmapped contents of the file. - PosixMmapReadableFile(const std::string& fname, void* base, size_t length, - MmapLimiter* limiter) - : filename_(fname), mmapped_region_(base), length_(length), - limiter_(limiter) { - } + PosixSequentialFile(std::string filename, int fd) + : fd_(fd), filename_(filename) {} + ~PosixSequentialFile() override { close(fd_); } - virtual ~PosixMmapReadableFile() { - munmap(mmapped_region_, length_); - limiter_->Release(); - } - - virtual Status Read(uint64_t offset, size_t n, Slice* result, - char* scratch) const { - Status s; - if (offset + n > length_) { - *result = Slice(); - s = IOError(filename_, EINVAL); - } else { - *result = Slice(reinterpret_cast(mmapped_region_) + offset, n); - } - return s; - } -}; - -class PosixWritableFile : public WritableFile { - private: - std::string filename_; - FILE* file_; - - public: - PosixWritableFile(const std::string& fname, FILE* f) - : filename_(fname), file_(f) { } - - ~PosixWritableFile() { - if (file_ != NULL) { - // Ignoring any potential errors - fclose(file_); + Status Read(size_t n, Slice* result, char* scratch) override { + Status status; + while (true) { + ::ssize_t read_size = ::read(fd_, scratch, n); + if (read_size < 0) { // Read error. + if (errno == EINTR) { + continue; // Retry + } + status = PosixError(filename_, errno); + break; + } + *result = Slice(scratch, read_size); + break; } + return status; } - virtual Status Append(const Slice& data) { - size_t r = fwrite_unlocked(data.data(), 1, data.size(), file_); - if (r != data.size()) { - return IOError(filename_, errno); + Status Skip(uint64_t n) override { + if (::lseek(fd_, n, SEEK_CUR) == static_cast(-1)) { + return PosixError(filename_, errno); } return Status::OK(); } - virtual Status Close() { - Status result; - if (fclose(file_) != 0) { - result = IOError(filename_, errno); + private: + const int fd_; + const std::string filename_; +}; + +// Implements random read access in a file using pread(). +// +// Instances of this class are thread-safe, as required by the RandomAccessFile +// API. Instances are immutable and Read() only calls thread-safe library +// functions. +class PosixRandomAccessFile final : public RandomAccessFile { + public: + // The new instance takes ownership of |fd|. |fd_limiter| must outlive this + // instance, and will be used to determine if . + PosixRandomAccessFile(std::string filename, int fd, Limiter* fd_limiter) + : has_permanent_fd_(fd_limiter->Acquire()), + fd_(has_permanent_fd_ ? fd : -1), + fd_limiter_(fd_limiter), + filename_(std::move(filename)) { + if (!has_permanent_fd_) { + assert(fd_ == -1); + ::close(fd); // The file will be opened on every read. } - file_ = NULL; - return result; } - virtual Status Flush() { - if (fflush_unlocked(file_) != 0) { - return IOError(filename_, errno); + ~PosixRandomAccessFile() override { + if (has_permanent_fd_) { + assert(fd_ != -1); + ::close(fd_); + fd_limiter_->Release(); + } + } + + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + int fd = fd_; + if (!has_permanent_fd_) { + fd = ::open(filename_.c_str(), O_RDONLY | kOpenBaseFlags); + if (fd < 0) { + return PosixError(filename_, errno); + } + } + + assert(fd != -1); + + Status status; + ssize_t read_size = ::pread(fd, scratch, n, static_cast(offset)); + *result = Slice(scratch, (read_size < 0) ? 0 : read_size); + if (read_size < 0) { + // An error: return a non-ok status. + status = PosixError(filename_, errno); + } + if (!has_permanent_fd_) { + // Close the temporary file descriptor opened earlier. + assert(fd != fd_); + ::close(fd); + } + return status; + } + + private: + const bool has_permanent_fd_; // If false, the file is opened on every read. + const int fd_; // -1 if has_permanent_fd_ is false. + Limiter* const fd_limiter_; + const std::string filename_; +}; + +// Implements random read access in a file using mmap(). +// +// Instances of this class are thread-safe, as required by the RandomAccessFile +// API. Instances are immutable and Read() only calls thread-safe library +// functions. +class PosixMmapReadableFile final : public RandomAccessFile { + public: + // mmap_base[0, length-1] points to the memory-mapped contents of the file. It + // must be the result of a successful call to mmap(). This instances takes + // over the ownership of the region. + // + // |mmap_limiter| must outlive this instance. The caller must have already + // aquired the right to use one mmap region, which will be released when this + // instance is destroyed. + PosixMmapReadableFile(std::string filename, char* mmap_base, size_t length, + Limiter* mmap_limiter) + : mmap_base_(mmap_base), + length_(length), + mmap_limiter_(mmap_limiter), + filename_(std::move(filename)) {} + + ~PosixMmapReadableFile() override { + ::munmap(static_cast(mmap_base_), length_); + mmap_limiter_->Release(); + } + + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + if (offset + n > length_) { + *result = Slice(); + return PosixError(filename_, EINVAL); + } + + *result = Slice(mmap_base_ + offset, n); + return Status::OK(); + } + + private: + char* const mmap_base_; + const size_t length_; + Limiter* const mmap_limiter_; + const std::string filename_; +}; + +class PosixWritableFile final : public WritableFile { + public: + PosixWritableFile(std::string filename, int fd) + : pos_(0), + fd_(fd), + is_manifest_(IsManifest(filename)), + filename_(std::move(filename)), + dirname_(Dirname(filename_)) {} + + ~PosixWritableFile() override { + if (fd_ >= 0) { + // Ignoring any potential errors + Close(); + } + } + + Status Append(const Slice& data) override { + size_t write_size = data.size(); + const char* write_data = data.data(); + + // Fit as much as possible into buffer. + size_t copy_size = std::min(write_size, kWritableFileBufferSize - pos_); + std::memcpy(buf_ + pos_, write_data, copy_size); + write_data += copy_size; + write_size -= copy_size; + pos_ += copy_size; + if (write_size == 0) { + return Status::OK(); + } + + // Can't fit in buffer, so need to do at least one write. + Status status = FlushBuffer(); + if (!status.ok()) { + return status; + } + + // Small writes go to buffer, large writes are written directly. + if (write_size < kWritableFileBufferSize) { + std::memcpy(buf_, write_data, write_size); + pos_ = write_size; + return Status::OK(); + } + return WriteUnbuffered(write_data, write_size); + } + + Status Close() override { + Status status = FlushBuffer(); + const int close_result = ::close(fd_); + if (close_result < 0 && status.ok()) { + status = PosixError(filename_, errno); + } + fd_ = -1; + return status; + } + + Status Flush() override { return FlushBuffer(); } + + Status Sync() override { + // Ensure new files referred to by the manifest are in the filesystem. + // + // This needs to happen before the manifest file is flushed to disk, to + // avoid crashing in a state where the manifest refers to files that are not + // yet on disk. + Status status = SyncDirIfManifest(); + if (!status.ok()) { + return status; + } + + status = FlushBuffer(); + if (!status.ok()) { + return status; + } + + return SyncFd(fd_, filename_); + } + + private: + Status FlushBuffer() { + Status status = WriteUnbuffered(buf_, pos_); + pos_ = 0; + return status; + } + + Status WriteUnbuffered(const char* data, size_t size) { + while (size > 0) { + ssize_t write_result = ::write(fd_, data, size); + if (write_result < 0) { + if (errno == EINTR) { + continue; // Retry + } + return PosixError(filename_, errno); + } + data += write_result; + size -= write_result; } return Status::OK(); } Status SyncDirIfManifest() { - const char* f = filename_.c_str(); - const char* sep = strrchr(f, '/'); - Slice basename; - std::string dir; - if (sep == NULL) { - dir = "."; - basename = f; + Status status; + if (!is_manifest_) { + return status; + } + + int fd = ::open(dirname_.c_str(), O_RDONLY | kOpenBaseFlags); + if (fd < 0) { + status = PosixError(dirname_, errno); } else { - dir = std::string(f, sep - f); - basename = sep + 1; + status = SyncFd(fd, dirname_); + ::close(fd); } - Status s; - if (basename.starts_with("MANIFEST")) { - int fd = open(dir.c_str(), O_RDONLY); - if (fd < 0) { - s = IOError(dir, errno); - } else { - if (fsync(fd) < 0) { - s = IOError(dir, errno); - } - close(fd); - } - } - return s; + return status; } - virtual Status Sync() { - // Ensure new files referred to by the manifest are in the filesystem. - Status s = SyncDirIfManifest(); - if (!s.ok()) { - return s; + // Ensures that all the caches associated with the given file descriptor's + // data are flushed all the way to durable media, and can withstand power + // failures. + // + // The path argument is only used to populate the description string in the + // returned Status if an error occurs. + static Status SyncFd(int fd, const std::string& fd_path) { +#if HAVE_FULLFSYNC + // On macOS and iOS, fsync() doesn't guarantee durability past power + // failures. fcntl(F_FULLFSYNC) is required for that purpose. Some + // filesystems don't support fcntl(F_FULLFSYNC), and require a fallback to + // fsync(). + if (::fcntl(fd, F_FULLFSYNC) == 0) { + return Status::OK(); } - if (fflush_unlocked(file_) != 0 || - fdatasync(fileno(file_)) != 0) { - s = Status::IOError(filename_, strerror(errno)); +#endif // HAVE_FULLFSYNC + +#if HAVE_FDATASYNC + bool sync_success = ::fdatasync(fd) == 0; +#else + bool sync_success = ::fsync(fd) == 0; +#endif // HAVE_FDATASYNC + + if (sync_success) { + return Status::OK(); } - return s; + return PosixError(fd_path, errno); } + + // Returns the directory name in a path pointing to a file. + // + // Returns "." if the path does not contain any directory separator. + static std::string Dirname(const std::string& filename) { + std::string::size_type separator_pos = filename.rfind('/'); + if (separator_pos == std::string::npos) { + return std::string("."); + } + // The filename component should not contain a path separator. If it does, + // the splitting was done incorrectly. + assert(filename.find('/', separator_pos + 1) == std::string::npos); + + return filename.substr(0, separator_pos); + } + + // Extracts the file name from a path pointing to a file. + // + // The returned Slice points to |filename|'s data buffer, so it is only valid + // while |filename| is alive and unchanged. + static Slice Basename(const std::string& filename) { + std::string::size_type separator_pos = filename.rfind('/'); + if (separator_pos == std::string::npos) { + return Slice(filename); + } + // The filename component should not contain a path separator. If it does, + // the splitting was done incorrectly. + assert(filename.find('/', separator_pos + 1) == std::string::npos); + + return Slice(filename.data() + separator_pos + 1, + filename.length() - separator_pos - 1); + } + + // True if the given file is a manifest file. + static bool IsManifest(const std::string& filename) { + return Basename(filename).starts_with("MANIFEST"); + } + + // buf_[0, pos_ - 1] contains data to be written to fd_. + char buf_[kWritableFileBufferSize]; + size_t pos_; + int fd_; + + const bool is_manifest_; // True if the file's name starts with MANIFEST. + const std::string filename_; + const std::string dirname_; // The directory of filename_. }; -static int LockOrUnlock(int fd, bool lock) { +int LockOrUnlock(int fd, bool lock) { errno = 0; - struct flock f; - memset(&f, 0, sizeof(f)); - f.l_type = (lock ? F_WRLCK : F_UNLCK); - f.l_whence = SEEK_SET; - f.l_start = 0; - f.l_len = 0; // Lock/unlock entire file - return fcntl(fd, F_SETLK, &f); + struct ::flock file_lock_info; + std::memset(&file_lock_info, 0, sizeof(file_lock_info)); + file_lock_info.l_type = (lock ? F_WRLCK : F_UNLCK); + file_lock_info.l_whence = SEEK_SET; + file_lock_info.l_start = 0; + file_lock_info.l_len = 0; // Lock/unlock entire file. + return ::fcntl(fd, F_SETLK, &file_lock_info); } +// Instances are thread-safe because they are immutable. class PosixFileLock : public FileLock { public: - int fd_; - std::string name_; + PosixFileLock(int fd, std::string filename) + : fd_(fd), filename_(std::move(filename)) {} + + int fd() const { return fd_; } + const std::string& filename() const { return filename_; } + + private: + const int fd_; + const std::string filename_; }; -// Set of locked files. We keep a separate set instead of just -// relying on fcntrl(F_SETLK) since fcntl(F_SETLK) does not provide -// any protection against multiple uses from the same process. +// Tracks the files locked by PosixEnv::LockFile(). +// +// We maintain a separate set instead of relying on fcntl(F_SETLK) because +// fcntl(F_SETLK) does not provide any protection against multiple uses from the +// same process. +// +// Instances are thread-safe because all member data is guarded by a mutex. class PosixLockTable { + public: + bool Insert(const std::string& fname) LOCKS_EXCLUDED(mu_) { + mu_.Lock(); + bool succeeded = locked_files_.insert(fname).second; + mu_.Unlock(); + return succeeded; + } + void Remove(const std::string& fname) LOCKS_EXCLUDED(mu_) { + mu_.Lock(); + locked_files_.erase(fname); + mu_.Unlock(); + } + private: port::Mutex mu_; - std::set locked_files_; - public: - bool Insert(const std::string& fname) { - MutexLock l(&mu_); - return locked_files_.insert(fname).second; - } - void Remove(const std::string& fname) { - MutexLock l(&mu_); - locked_files_.erase(fname); - } + std::set locked_files_ GUARDED_BY(mu_); }; class PosixEnv : public Env { public: PosixEnv(); - virtual ~PosixEnv() { - char msg[] = "Destroying Env::Default()\n"; - fwrite(msg, 1, sizeof(msg), stderr); - abort(); + ~PosixEnv() override { + static char msg[] = "PosixEnv singleton destroyed. Unsupported behavior!\n"; + std::fwrite(msg, 1, sizeof(msg), stderr); + std::abort(); } - virtual Status NewSequentialFile(const std::string& fname, - SequentialFile** result) { - FILE* f = fopen(fname.c_str(), "r"); - if (f == NULL) { - *result = NULL; - return IOError(fname, errno); - } else { - *result = new PosixSequentialFile(fname, f); - return Status::OK(); - } - } - - virtual Status NewRandomAccessFile(const std::string& fname, - RandomAccessFile** result) { - *result = NULL; - Status s; - int fd = open(fname.c_str(), O_RDONLY); + Status NewSequentialFile(const std::string& filename, + SequentialFile** result) override { + int fd = ::open(filename.c_str(), O_RDONLY | kOpenBaseFlags); if (fd < 0) { - s = IOError(fname, errno); - } else if (mmap_limit_.Acquire()) { - uint64_t size; - s = GetFileSize(fname, &size); - if (s.ok()) { - void* base = mmap(NULL, size, PROT_READ, MAP_SHARED, fd, 0); - if (base != MAP_FAILED) { - *result = new PosixMmapReadableFile(fname, base, size, &mmap_limit_); - } else { - s = IOError(fname, errno); - } - } - close(fd); - if (!s.ok()) { - mmap_limit_.Release(); - } - } else { - *result = new PosixRandomAccessFile(fname, fd); + *result = nullptr; + return PosixError(filename, errno); } - return s; - } - virtual Status NewWritableFile(const std::string& fname, - WritableFile** result) { - Status s; - FILE* f = fopen(fname.c_str(), "w"); - if (f == NULL) { - *result = NULL; - s = IOError(fname, errno); - } else { - *result = new PosixWritableFile(fname, f); - } - return s; - } - - virtual bool FileExists(const std::string& fname) { - return access(fname.c_str(), F_OK) == 0; - } - - virtual Status GetChildren(const std::string& dir, - std::vector* result) { - result->clear(); - DIR* d = opendir(dir.c_str()); - if (d == NULL) { - return IOError(dir, errno); - } - struct dirent* entry; - while ((entry = readdir(d)) != NULL) { - result->push_back(entry->d_name); - } - closedir(d); + *result = new PosixSequentialFile(filename, fd); return Status::OK(); } - virtual Status DeleteFile(const std::string& fname) { - Status result; - if (unlink(fname.c_str()) != 0) { - result = IOError(fname, errno); - } - return result; - } - - virtual Status CreateDir(const std::string& name) { - Status result; - if (mkdir(name.c_str(), 0755) != 0) { - result = IOError(name, errno); - } - return result; - } - - virtual Status DeleteDir(const std::string& name) { - Status result; - if (rmdir(name.c_str()) != 0) { - result = IOError(name, errno); - } - return result; - } - - virtual Status GetFileSize(const std::string& fname, uint64_t* size) { - Status s; - struct stat sbuf; - if (stat(fname.c_str(), &sbuf) != 0) { - *size = 0; - s = IOError(fname, errno); - } else { - *size = sbuf.st_size; - } - return s; - } - - virtual Status RenameFile(const std::string& src, const std::string& target) { - Status result; - if (rename(src.c_str(), target.c_str()) != 0) { - result = IOError(src, errno); - } - return result; - } - - virtual Status LockFile(const std::string& fname, FileLock** lock) { - *lock = NULL; - Status result; - int fd = open(fname.c_str(), O_RDWR | O_CREAT, 0644); + Status NewRandomAccessFile(const std::string& filename, + RandomAccessFile** result) override { + *result = nullptr; + int fd = ::open(filename.c_str(), O_RDONLY | kOpenBaseFlags); if (fd < 0) { - result = IOError(fname, errno); - } else if (!locks_.Insert(fname)) { - close(fd); - result = Status::IOError("lock " + fname, "already held by process"); - } else if (LockOrUnlock(fd, true) == -1) { - result = IOError("lock " + fname, errno); - close(fd); - locks_.Remove(fname); - } else { - PosixFileLock* my_lock = new PosixFileLock; - my_lock->fd_ = fd; - my_lock->name_ = fname; - *lock = my_lock; + return PosixError(filename, errno); } - return result; + + if (!mmap_limiter_.Acquire()) { + *result = new PosixRandomAccessFile(filename, fd, &fd_limiter_); + return Status::OK(); + } + + uint64_t file_size; + Status status = GetFileSize(filename, &file_size); + if (status.ok()) { + void* mmap_base = + ::mmap(/*addr=*/nullptr, file_size, PROT_READ, MAP_SHARED, fd, 0); + if (mmap_base != MAP_FAILED) { + *result = new PosixMmapReadableFile(filename, + reinterpret_cast(mmap_base), + file_size, &mmap_limiter_); + } else { + status = PosixError(filename, errno); + } + } + ::close(fd); + if (!status.ok()) { + mmap_limiter_.Release(); + } + return status; } - virtual Status UnlockFile(FileLock* lock) { - PosixFileLock* my_lock = reinterpret_cast(lock); - Status result; - if (LockOrUnlock(my_lock->fd_, false) == -1) { - result = IOError("unlock", errno); + Status NewWritableFile(const std::string& filename, + WritableFile** result) override { + int fd = ::open(filename.c_str(), + O_TRUNC | O_WRONLY | O_CREAT | kOpenBaseFlags, 0644); + if (fd < 0) { + *result = nullptr; + return PosixError(filename, errno); } - locks_.Remove(my_lock->name_); - close(my_lock->fd_); - delete my_lock; - return result; + + *result = new PosixWritableFile(filename, fd); + return Status::OK(); } - virtual void Schedule(void (*function)(void*), void* arg); + Status NewAppendableFile(const std::string& filename, + WritableFile** result) override { + int fd = ::open(filename.c_str(), + O_APPEND | O_WRONLY | O_CREAT | kOpenBaseFlags, 0644); + if (fd < 0) { + *result = nullptr; + return PosixError(filename, errno); + } - virtual void StartThread(void (*function)(void* arg), void* arg); + *result = new PosixWritableFile(filename, fd); + return Status::OK(); + } - virtual Status GetTestDirectory(std::string* result) { - const char* env = getenv("TEST_TMPDIR"); + bool FileExists(const std::string& filename) override { + return ::access(filename.c_str(), F_OK) == 0; + } + + Status GetChildren(const std::string& directory_path, + std::vector* result) override { + result->clear(); + ::DIR* dir = ::opendir(directory_path.c_str()); + if (dir == nullptr) { + return PosixError(directory_path, errno); + } + struct ::dirent* entry; + while ((entry = ::readdir(dir)) != nullptr) { + result->emplace_back(entry->d_name); + } + ::closedir(dir); + return Status::OK(); + } + + Status DeleteFile(const std::string& filename) override { + if (::unlink(filename.c_str()) != 0) { + return PosixError(filename, errno); + } + return Status::OK(); + } + + Status CreateDir(const std::string& dirname) override { + if (::mkdir(dirname.c_str(), 0755) != 0) { + return PosixError(dirname, errno); + } + return Status::OK(); + } + + Status DeleteDir(const std::string& dirname) override { + if (::rmdir(dirname.c_str()) != 0) { + return PosixError(dirname, errno); + } + return Status::OK(); + } + + Status GetFileSize(const std::string& filename, uint64_t* size) override { + struct ::stat file_stat; + if (::stat(filename.c_str(), &file_stat) != 0) { + *size = 0; + return PosixError(filename, errno); + } + *size = file_stat.st_size; + return Status::OK(); + } + + Status RenameFile(const std::string& from, const std::string& to) override { + if (std::rename(from.c_str(), to.c_str()) != 0) { + return PosixError(from, errno); + } + return Status::OK(); + } + + Status LockFile(const std::string& filename, FileLock** lock) override { + *lock = nullptr; + + int fd = ::open(filename.c_str(), O_RDWR | O_CREAT | kOpenBaseFlags, 0644); + if (fd < 0) { + return PosixError(filename, errno); + } + + if (!locks_.Insert(filename)) { + ::close(fd); + return Status::IOError("lock " + filename, "already held by process"); + } + + if (LockOrUnlock(fd, true) == -1) { + int lock_errno = errno; + ::close(fd); + locks_.Remove(filename); + return PosixError("lock " + filename, lock_errno); + } + + *lock = new PosixFileLock(fd, filename); + return Status::OK(); + } + + Status UnlockFile(FileLock* lock) override { + PosixFileLock* posix_file_lock = static_cast(lock); + if (LockOrUnlock(posix_file_lock->fd(), false) == -1) { + return PosixError("unlock " + posix_file_lock->filename(), errno); + } + locks_.Remove(posix_file_lock->filename()); + ::close(posix_file_lock->fd()); + delete posix_file_lock; + return Status::OK(); + } + + void Schedule(void (*background_work_function)(void* background_work_arg), + void* background_work_arg) override; + + void StartThread(void (*thread_main)(void* thread_main_arg), + void* thread_main_arg) override; + + Status GetTestDirectory(std::string* result) override { + const char* env = std::getenv("TEST_TMPDIR"); if (env && env[0] != '\0') { *result = env; } else { char buf[100]; - snprintf(buf, sizeof(buf), "/tmp/leveldbtest-%d", int(geteuid())); + std::snprintf(buf, sizeof(buf), "/tmp/leveldbtest-%d", + static_cast(::geteuid())); *result = buf; } - // Directory may already exist + + // The CreateDir status is ignored because the directory may already exist. CreateDir(*result); + return Status::OK(); } - static uint64_t gettid() { - pthread_t tid = pthread_self(); - uint64_t thread_id = 0; - memcpy(&thread_id, &tid, std::min(sizeof(thread_id), sizeof(tid))); - return thread_id; - } + Status NewLogger(const std::string& filename, Logger** result) override { + int fd = ::open(filename.c_str(), + O_APPEND | O_WRONLY | O_CREAT | kOpenBaseFlags, 0644); + if (fd < 0) { + *result = nullptr; + return PosixError(filename, errno); + } - virtual Status NewLogger(const std::string& fname, Logger** result) { - FILE* f = fopen(fname.c_str(), "w"); - if (f == NULL) { - *result = NULL; - return IOError(fname, errno); + std::FILE* fp = ::fdopen(fd, "w"); + if (fp == nullptr) { + ::close(fd); + *result = nullptr; + return PosixError(filename, errno); } else { - *result = new PosixLogger(f, &PosixEnv::gettid); + *result = new PosixLogger(fp); return Status::OK(); } } - virtual uint64_t NowMicros() { - struct timeval tv; - gettimeofday(&tv, NULL); - return static_cast(tv.tv_sec) * 1000000 + tv.tv_usec; + uint64_t NowMicros() override { + static constexpr uint64_t kUsecondsPerSecond = 1000000; + struct ::timeval tv; + ::gettimeofday(&tv, nullptr); + return static_cast(tv.tv_sec) * kUsecondsPerSecond + tv.tv_usec; } - virtual void SleepForMicroseconds(int micros) { - usleep(micros); - } + void SleepForMicroseconds(int micros) override { ::usleep(micros); } private: - void PthreadCall(const char* label, int result) { - if (result != 0) { - fprintf(stderr, "pthread %s: %s\n", label, strerror(result)); - abort(); - } + void BackgroundThreadMain(); + + static void BackgroundThreadEntryPoint(PosixEnv* env) { + env->BackgroundThreadMain(); } - // BGThread() is the body of the background thread - void BGThread(); - static void* BGThreadWrapper(void* arg) { - reinterpret_cast(arg)->BGThread(); - return NULL; - } + // Stores the work item data in a Schedule() call. + // + // Instances are constructed on the thread calling Schedule() and used on the + // background thread. + // + // This structure is thread-safe beacuse it is immutable. + struct BackgroundWorkItem { + explicit BackgroundWorkItem(void (*function)(void* arg), void* arg) + : function(function), arg(arg) {} - pthread_mutex_t mu_; - pthread_cond_t bgsignal_; - pthread_t bgthread_; - bool started_bgthread_; + void (*const function)(void*); + void* const arg; + }; - // Entry per Schedule() call - struct BGItem { void* arg; void (*function)(void*); }; - typedef std::deque BGQueue; - BGQueue queue_; + port::Mutex background_work_mutex_; + port::CondVar background_work_cv_ GUARDED_BY(background_work_mutex_); + bool started_background_thread_ GUARDED_BY(background_work_mutex_); - PosixLockTable locks_; - MmapLimiter mmap_limit_; + std::queue background_work_queue_ + GUARDED_BY(background_work_mutex_); + + PosixLockTable locks_; // Thread-safe. + Limiter mmap_limiter_; // Thread-safe. + Limiter fd_limiter_; // Thread-safe. }; -PosixEnv::PosixEnv() : started_bgthread_(false) { - PthreadCall("mutex_init", pthread_mutex_init(&mu_, NULL)); - PthreadCall("cvar_init", pthread_cond_init(&bgsignal_, NULL)); -} +// Return the maximum number of concurrent mmaps. +int MaxMmaps() { return g_mmap_limit; } -void PosixEnv::Schedule(void (*function)(void*), void* arg) { - PthreadCall("lock", pthread_mutex_lock(&mu_)); - - // Start background thread if necessary - if (!started_bgthread_) { - started_bgthread_ = true; - PthreadCall( - "create thread", - pthread_create(&bgthread_, NULL, &PosixEnv::BGThreadWrapper, this)); +// Return the maximum number of read-only files to keep open. +int MaxOpenFiles() { + if (g_open_read_only_file_limit >= 0) { + return g_open_read_only_file_limit; } - - // If the queue is currently empty, the background thread may currently be - // waiting. - if (queue_.empty()) { - PthreadCall("signal", pthread_cond_signal(&bgsignal_)); + struct ::rlimit rlim; + if (::getrlimit(RLIMIT_NOFILE, &rlim)) { + // getrlimit failed, fallback to hard-coded default. + g_open_read_only_file_limit = 50; + } else if (rlim.rlim_cur == RLIM_INFINITY) { + g_open_read_only_file_limit = std::numeric_limits::max(); + } else { + // Allow use of 20% of available file descriptors for read-only files. + g_open_read_only_file_limit = rlim.rlim_cur / 5; } - - // Add to priority queue - queue_.push_back(BGItem()); - queue_.back().function = function; - queue_.back().arg = arg; - - PthreadCall("unlock", pthread_mutex_unlock(&mu_)); -} - -void PosixEnv::BGThread() { - while (true) { - // Wait until there is an item that is ready to run - PthreadCall("lock", pthread_mutex_lock(&mu_)); - while (queue_.empty()) { - PthreadCall("wait", pthread_cond_wait(&bgsignal_, &mu_)); - } - - void (*function)(void*) = queue_.front().function; - void* arg = queue_.front().arg; - queue_.pop_front(); - - PthreadCall("unlock", pthread_mutex_unlock(&mu_)); - (*function)(arg); - } -} - -namespace { -struct StartThreadState { - void (*user_function)(void*); - void* arg; -}; -} -static void* StartThreadWrapper(void* arg) { - StartThreadState* state = reinterpret_cast(arg); - state->user_function(state->arg); - delete state; - return NULL; -} - -void PosixEnv::StartThread(void (*function)(void* arg), void* arg) { - pthread_t t; - StartThreadState* state = new StartThreadState; - state->user_function = function; - state->arg = arg; - PthreadCall("start thread", - pthread_create(&t, NULL, &StartThreadWrapper, state)); + return g_open_read_only_file_limit; } } // namespace -static pthread_once_t once = PTHREAD_ONCE_INIT; -static Env* default_env; -static void InitDefaultEnv() { default_env = new PosixEnv; } +PosixEnv::PosixEnv() + : background_work_cv_(&background_work_mutex_), + started_background_thread_(false), + mmap_limiter_(MaxMmaps()), + fd_limiter_(MaxOpenFiles()) {} + +void PosixEnv::Schedule( + void (*background_work_function)(void* background_work_arg), + void* background_work_arg) { + background_work_mutex_.Lock(); + + // Start the background thread, if we haven't done so already. + if (!started_background_thread_) { + started_background_thread_ = true; + std::thread background_thread(PosixEnv::BackgroundThreadEntryPoint, this); + background_thread.detach(); + } + + // If the queue is empty, the background thread may be waiting for work. + if (background_work_queue_.empty()) { + background_work_cv_.Signal(); + } + + background_work_queue_.emplace(background_work_function, background_work_arg); + background_work_mutex_.Unlock(); +} + +void PosixEnv::BackgroundThreadMain() { + while (true) { + background_work_mutex_.Lock(); + + // Wait until there is work to be done. + while (background_work_queue_.empty()) { + background_work_cv_.Wait(); + } + + assert(!background_work_queue_.empty()); + auto background_work_function = background_work_queue_.front().function; + void* background_work_arg = background_work_queue_.front().arg; + background_work_queue_.pop(); + + background_work_mutex_.Unlock(); + background_work_function(background_work_arg); + } +} + +namespace { + +// Wraps an Env instance whose destructor is never created. +// +// Intended usage: +// using PlatformSingletonEnv = SingletonEnv; +// void ConfigurePosixEnv(int param) { +// PlatformSingletonEnv::AssertEnvNotInitialized(); +// // set global configuration flags. +// } +// Env* Env::Default() { +// static PlatformSingletonEnv default_env; +// return default_env.env(); +// } +template +class SingletonEnv { + public: + SingletonEnv() { +#if !defined(NDEBUG) + env_initialized_.store(true, std::memory_order::memory_order_relaxed); +#endif // !defined(NDEBUG) + static_assert(sizeof(env_storage_) >= sizeof(EnvType), + "env_storage_ will not fit the Env"); + static_assert(alignof(decltype(env_storage_)) >= alignof(EnvType), + "env_storage_ does not meet the Env's alignment needs"); + new (&env_storage_) EnvType(); + } + ~SingletonEnv() = default; + + SingletonEnv(const SingletonEnv&) = delete; + SingletonEnv& operator=(const SingletonEnv&) = delete; + + Env* env() { return reinterpret_cast(&env_storage_); } + + static void AssertEnvNotInitialized() { +#if !defined(NDEBUG) + assert(!env_initialized_.load(std::memory_order::memory_order_relaxed)); +#endif // !defined(NDEBUG) + } + + private: + typename std::aligned_storage::type + env_storage_; +#if !defined(NDEBUG) + static std::atomic env_initialized_; +#endif // !defined(NDEBUG) +}; + +#if !defined(NDEBUG) +template +std::atomic SingletonEnv::env_initialized_; +#endif // !defined(NDEBUG) + +using PosixDefaultEnv = SingletonEnv; + +} // namespace + +void PosixEnv::StartThread(void (*thread_main)(void* thread_main_arg), + void* thread_main_arg) { + std::thread new_thread(thread_main, thread_main_arg); + new_thread.detach(); +} + +void EnvPosixTestHelper::SetReadOnlyFDLimit(int limit) { + PosixDefaultEnv::AssertEnvNotInitialized(); + g_open_read_only_file_limit = limit; +} + +void EnvPosixTestHelper::SetReadOnlyMMapLimit(int limit) { + PosixDefaultEnv::AssertEnvNotInitialized(); + g_mmap_limit = limit; +} Env* Env::Default() { - pthread_once(&once, InitDefaultEnv); - return default_env; + static PosixDefaultEnv env_container; + return env_container.env(); } } // namespace leveldb diff --git a/util/env_posix_test.cc b/util/env_posix_test.cc new file mode 100644 index 0000000..9675d73 --- /dev/null +++ b/util/env_posix_test.cc @@ -0,0 +1,350 @@ +// 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 "leveldb/env.h" +#include "port/port.h" +#include "util/env_posix_test_helper.h" +#include "util/testharness.h" + +#if HAVE_O_CLOEXEC + +namespace { + +// Exit codes for the helper process spawned by TestCloseOnExec* tests. +// Useful for debugging test failures. +constexpr int kTextCloseOnExecHelperExecFailedCode = 61; +constexpr int kTextCloseOnExecHelperDup2FailedCode = 62; +constexpr int kTextCloseOnExecHelperFoundOpenFdCode = 63; + +// Global set by main() and read in TestCloseOnExec. +// +// The argv[0] value is stored in a std::vector instead of a std::string because +// std::string does not return a mutable pointer to its buffer until C++17. +// +// The vector stores the string pointed to by argv[0], plus the trailing null. +std::vector* GetArgvZero() { + static std::vector program_name; + return &program_name; +} + +// Command-line switch used to run this test as the CloseOnExecSwitch helper. +static const char kTestCloseOnExecSwitch[] = "--test-close-on-exec-helper"; + +// Executed in a separate process by TestCloseOnExec* tests. +// +// main() delegates to this function when the test executable is launched with +// a special command-line switch. TestCloseOnExec* tests fork()+exec() the test +// executable and pass the special command-line switch. +// + +// main() delegates to this function when the test executable is launched with +// a special command-line switch. TestCloseOnExec* tests fork()+exec() the test +// executable and pass the special command-line switch. +// +// When main() delegates to this function, the process probes whether a given +// file descriptor is open, and communicates the result via its exit code. +int TestCloseOnExecHelperMain(char* pid_arg) { + int fd = std::atoi(pid_arg); + // When given the same file descriptor twice, dup2() returns -1 if the + // file descriptor is closed, or the given file descriptor if it is open. + if (::dup2(fd, fd) == fd) { + std::fprintf(stderr, "Unexpected open fd %d\n", fd); + return kTextCloseOnExecHelperFoundOpenFdCode; + } + // Double-check that dup2() is saying the file descriptor is closed. + if (errno != EBADF) { + std::fprintf(stderr, "Unexpected errno after calling dup2 on fd %d: %s\n", + fd, std::strerror(errno)); + return kTextCloseOnExecHelperDup2FailedCode; + } + return 0; +} + +// File descriptors are small non-negative integers. +// +// Returns void so the implementation can use ASSERT_EQ. +void GetMaxFileDescriptor(int* result_fd) { + // Get the maximum file descriptor number. + ::rlimit fd_rlimit; + ASSERT_EQ(0, ::getrlimit(RLIMIT_NOFILE, &fd_rlimit)); + *result_fd = fd_rlimit.rlim_cur; +} + +// Iterates through all possible FDs and returns the currently open ones. +// +// Returns void so the implementation can use ASSERT_EQ. +void GetOpenFileDescriptors(std::unordered_set* open_fds) { + int max_fd = 0; + GetMaxFileDescriptor(&max_fd); + + for (int fd = 0; fd < max_fd; ++fd) { + if (::dup2(fd, fd) != fd) { + // When given the same file descriptor twice, dup2() returns -1 if the + // file descriptor is closed, or the given file descriptor if it is open. + // + // Double-check that dup2() is saying the fd is closed. + ASSERT_EQ(EBADF, errno) + << "dup2() should set errno to EBADF on closed file descriptors"; + continue; + } + open_fds->insert(fd); + } +} + +// Finds an FD open since a previous call to GetOpenFileDescriptors(). +// +// |baseline_open_fds| is the result of a previous GetOpenFileDescriptors() +// call. Assumes that exactly one FD was opened since that call. +// +// Returns void so the implementation can use ASSERT_EQ. +void GetNewlyOpenedFileDescriptor( + const std::unordered_set& baseline_open_fds, int* result_fd) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + for (int fd : baseline_open_fds) { + ASSERT_EQ(1, open_fds.count(fd)) + << "Previously opened file descriptor was closed during test setup"; + open_fds.erase(fd); + } + ASSERT_EQ(1, open_fds.size()) + << "Expected exactly one newly opened file descriptor during test setup"; + *result_fd = *open_fds.begin(); +} + +// Check that a fork()+exec()-ed child process does not have an extra open FD. +void CheckCloseOnExecDoesNotLeakFDs( + const std::unordered_set& baseline_open_fds) { + // Prepare the argument list for the child process. + // execv() wants mutable buffers. + char switch_buffer[sizeof(kTestCloseOnExecSwitch)]; + std::memcpy(switch_buffer, kTestCloseOnExecSwitch, + sizeof(kTestCloseOnExecSwitch)); + + int probed_fd; + GetNewlyOpenedFileDescriptor(baseline_open_fds, &probed_fd); + std::string fd_string = std::to_string(probed_fd); + std::vector fd_buffer(fd_string.begin(), fd_string.end()); + fd_buffer.emplace_back('\0'); + + // The helper process is launched with the command below. + // env_posix_tests --test-close-on-exec-helper 3 + char* child_argv[] = {GetArgvZero()->data(), switch_buffer, fd_buffer.data(), + nullptr}; + + constexpr int kForkInChildProcessReturnValue = 0; + int child_pid = fork(); + if (child_pid == kForkInChildProcessReturnValue) { + ::execv(child_argv[0], child_argv); + std::fprintf(stderr, "Error spawning child process: %s\n", strerror(errno)); + std::exit(kTextCloseOnExecHelperExecFailedCode); + } + + int child_status = 0; + ASSERT_EQ(child_pid, ::waitpid(child_pid, &child_status, 0)); + ASSERT_TRUE(WIFEXITED(child_status)) + << "The helper process did not exit with an exit code"; + ASSERT_EQ(0, WEXITSTATUS(child_status)) + << "The helper process encountered an error"; +} + +} // namespace + +#endif // HAVE_O_CLOEXEC + +namespace leveldb { + +static const int kReadOnlyFileLimit = 4; +static const int kMMapLimit = 4; + +class EnvPosixTest { + public: + static void SetFileLimits(int read_only_file_limit, int mmap_limit) { + EnvPosixTestHelper::SetReadOnlyFDLimit(read_only_file_limit); + EnvPosixTestHelper::SetReadOnlyMMapLimit(mmap_limit); + } + + EnvPosixTest() : env_(Env::Default()) {} + + Env* env_; +}; + +TEST(EnvPosixTest, TestOpenOnRead) { + // Write some test data to a single file that will be opened |n| times. + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file = test_dir + "/open_on_read.txt"; + + FILE* f = fopen(test_file.c_str(), "we"); + ASSERT_TRUE(f != nullptr); + const char kFileData[] = "abcdefghijklmnopqrstuvwxyz"; + fputs(kFileData, f); + fclose(f); + + // Open test file some number above the sum of the two limits to force + // open-on-read behavior of POSIX Env leveldb::RandomAccessFile. + const int kNumFiles = kReadOnlyFileLimit + kMMapLimit + 5; + leveldb::RandomAccessFile* files[kNumFiles] = {0}; + for (int i = 0; i < kNumFiles; i++) { + ASSERT_OK(env_->NewRandomAccessFile(test_file, &files[i])); + } + char scratch; + Slice read_result; + for (int i = 0; i < kNumFiles; i++) { + ASSERT_OK(files[i]->Read(i, 1, &read_result, &scratch)); + ASSERT_EQ(kFileData[i], read_result[0]); + } + for (int i = 0; i < kNumFiles; i++) { + delete files[i]; + } + ASSERT_OK(env_->DeleteFile(test_file)); +} + +#if HAVE_O_CLOEXEC + +TEST(EnvPosixTest, TestCloseOnExecSequentialFile) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_sequential.txt"; + ASSERT_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::SequentialFile* file = nullptr; + ASSERT_OK(env_->NewSequentialFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_OK(env_->DeleteFile(file_path)); +} + +TEST(EnvPosixTest, TestCloseOnExecRandomAccessFile) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_random_access.txt"; + ASSERT_OK(WriteStringToFile(env_, "0123456789", file_path)); + + // Exhaust the RandomAccessFile mmap limit. This way, the test + // RandomAccessFile instance below is backed by a file descriptor, not by an + // mmap region. + leveldb::RandomAccessFile* mmapped_files[kReadOnlyFileLimit] = {nullptr}; + for (int i = 0; i < kReadOnlyFileLimit; i++) { + ASSERT_OK(env_->NewRandomAccessFile(file_path, &mmapped_files[i])); + } + + leveldb::RandomAccessFile* file = nullptr; + ASSERT_OK(env_->NewRandomAccessFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + for (int i = 0; i < kReadOnlyFileLimit; i++) { + delete mmapped_files[i]; + } + ASSERT_OK(env_->DeleteFile(file_path)); +} + +TEST(EnvPosixTest, TestCloseOnExecWritableFile) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_writable.txt"; + ASSERT_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::WritableFile* file = nullptr; + ASSERT_OK(env_->NewWritableFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_OK(env_->DeleteFile(file_path)); +} + +TEST(EnvPosixTest, TestCloseOnExecAppendableFile) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_appendable.txt"; + ASSERT_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::WritableFile* file = nullptr; + ASSERT_OK(env_->NewAppendableFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_OK(env_->DeleteFile(file_path)); +} + +TEST(EnvPosixTest, TestCloseOnExecLockFile) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_lock.txt"; + ASSERT_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::FileLock* lock = nullptr; + ASSERT_OK(env_->LockFile(file_path, &lock)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + ASSERT_OK(env_->UnlockFile(lock)); + + ASSERT_OK(env_->DeleteFile(file_path)); +} + +TEST(EnvPosixTest, TestCloseOnExecLogger) { + std::unordered_set open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_logger.txt"; + ASSERT_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::Logger* file = nullptr; + ASSERT_OK(env_->NewLogger(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_OK(env_->DeleteFile(file_path)); +} + +#endif // HAVE_O_CLOEXEC + +} // namespace leveldb + +int main(int argc, char** argv) { +#if HAVE_O_CLOEXEC + // Check if we're invoked as a helper program, or as the test suite. + for (int i = 1; i < argc; ++i) { + if (!std::strcmp(argv[i], kTestCloseOnExecSwitch)) { + return TestCloseOnExecHelperMain(argv[i + 1]); + } + } + + // Save argv[0] early, because googletest may modify argv. + GetArgvZero()->assign(argv[0], argv[0] + std::strlen(argv[0]) + 1); +#endif // HAVE_O_CLOEXEC + + // All tests currently run with the same read-only file limits. + leveldb::EnvPosixTest::SetFileLimits(leveldb::kReadOnlyFileLimit, + leveldb::kMMapLimit); + return leveldb::test::RunAllTests(); +} diff --git a/util/env_posix_test_helper.h b/util/env_posix_test_helper.h new file mode 100644 index 0000000..0386960 --- /dev/null +++ b/util/env_posix_test_helper.h @@ -0,0 +1,28 @@ +// Copyright 2017 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_UTIL_ENV_POSIX_TEST_HELPER_H_ +#define STORAGE_LEVELDB_UTIL_ENV_POSIX_TEST_HELPER_H_ + +namespace leveldb { + +class EnvPosixTest; + +// A helper for the POSIX Env to facilitate testing. +class EnvPosixTestHelper { + private: + friend class EnvPosixTest; + + // Set the maximum number of read-only files that will be opened. + // Must be called before creating an Env. + static void SetReadOnlyFDLimit(int limit); + + // Set the maximum number of read-only files that will be mapped via mmap. + // Must be called before creating an Env. + static void SetReadOnlyMMapLimit(int limit); +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_ENV_POSIX_TEST_HELPER_H_ diff --git a/util/env_test.cc b/util/env_test.cc index b72cb44..9e2ad1e 100644 --- a/util/env_test.cc +++ b/util/env_test.cc @@ -4,70 +4,127 @@ #include "leveldb/env.h" +#include +#include + #include "port/port.h" +#include "port/thread_annotations.h" +#include "util/mutexlock.h" #include "util/testharness.h" +#include "util/testutil.h" namespace leveldb { static const int kDelayMicros = 100000; -class EnvPosixTest { - private: - port::Mutex mu_; - std::string events_; - +class EnvTest { public: + EnvTest() : env_(Env::Default()) {} + Env* env_; - EnvPosixTest() : env_(Env::Default()) { } }; -static void SetBool(void* ptr) { - reinterpret_cast(ptr)->NoBarrier_Store(ptr); +namespace { + +static void SetAtomicBool(void* atomic_bool_ptr) { + std::atomic* atomic_bool = + reinterpret_cast*>(atomic_bool_ptr); + atomic_bool->store(true, std::memory_order_relaxed); } -TEST(EnvPosixTest, RunImmediately) { - port::AtomicPointer called (NULL); - env_->Schedule(&SetBool, &called); - Env::Default()->SleepForMicroseconds(kDelayMicros); - ASSERT_TRUE(called.NoBarrier_Load() != NULL); +} // namespace + +TEST(EnvTest, ReadWrite) { + Random rnd(test::RandomSeed()); + + // Get file to use for testing. + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file_name = test_dir + "/open_on_read.txt"; + WritableFile* writable_file; + ASSERT_OK(env_->NewWritableFile(test_file_name, &writable_file)); + + // Fill a file with data generated via a sequence of randomly sized writes. + static const size_t kDataSize = 10 * 1048576; + std::string data; + while (data.size() < kDataSize) { + int len = rnd.Skewed(18); // Up to 2^18 - 1, but typically much smaller + std::string r; + test::RandomString(&rnd, len, &r); + ASSERT_OK(writable_file->Append(r)); + data += r; + if (rnd.OneIn(10)) { + ASSERT_OK(writable_file->Flush()); + } + } + ASSERT_OK(writable_file->Sync()); + ASSERT_OK(writable_file->Close()); + delete writable_file; + + // Read all data using a sequence of randomly sized reads. + SequentialFile* sequential_file; + ASSERT_OK(env_->NewSequentialFile(test_file_name, &sequential_file)); + std::string read_result; + std::string scratch; + while (read_result.size() < data.size()) { + int len = std::min(rnd.Skewed(18), data.size() - read_result.size()); + scratch.resize(std::max(len, 1)); // at least 1 so &scratch[0] is legal + Slice read; + ASSERT_OK(sequential_file->Read(len, &read, &scratch[0])); + if (len > 0) { + ASSERT_GT(read.size(), 0); + } + ASSERT_LE(read.size(), len); + read_result.append(read.data(), read.size()); + } + ASSERT_EQ(read_result, data); + delete sequential_file; } -TEST(EnvPosixTest, RunMany) { - port::AtomicPointer last_id (NULL); +TEST(EnvTest, RunImmediately) { + std::atomic called(false); + env_->Schedule(&SetAtomicBool, &called); + env_->SleepForMicroseconds(kDelayMicros); + ASSERT_TRUE(called.load(std::memory_order_relaxed)); +} - struct CB { - port::AtomicPointer* last_id_ptr; // Pointer to shared slot - uintptr_t id; // Order# for the execution of this callback +TEST(EnvTest, RunMany) { + std::atomic last_id(0); - CB(port::AtomicPointer* p, int i) : last_id_ptr(p), id(i) { } + struct Callback { + std::atomic* const last_id_ptr_; // Pointer to shared state. + const int id_; // Order# for the execution of this callback. - static void Run(void* v) { - CB* cb = reinterpret_cast(v); - void* cur = cb->last_id_ptr->NoBarrier_Load(); - ASSERT_EQ(cb->id-1, reinterpret_cast(cur)); - cb->last_id_ptr->Release_Store(reinterpret_cast(cb->id)); + Callback(std::atomic* last_id_ptr, int id) + : last_id_ptr_(last_id_ptr), id_(id) {} + + static void Run(void* arg) { + Callback* callback = reinterpret_cast(arg); + int current_id = callback->last_id_ptr_->load(std::memory_order_relaxed); + ASSERT_EQ(callback->id_ - 1, current_id); + callback->last_id_ptr_->store(callback->id_, std::memory_order_relaxed); } }; - // Schedule in different order than start time - CB cb1(&last_id, 1); - CB cb2(&last_id, 2); - CB cb3(&last_id, 3); - CB cb4(&last_id, 4); - env_->Schedule(&CB::Run, &cb1); - env_->Schedule(&CB::Run, &cb2); - env_->Schedule(&CB::Run, &cb3); - env_->Schedule(&CB::Run, &cb4); + Callback callback1(&last_id, 1); + Callback callback2(&last_id, 2); + Callback callback3(&last_id, 3); + Callback callback4(&last_id, 4); + env_->Schedule(&Callback::Run, &callback1); + env_->Schedule(&Callback::Run, &callback2); + env_->Schedule(&Callback::Run, &callback3); + env_->Schedule(&Callback::Run, &callback4); - Env::Default()->SleepForMicroseconds(kDelayMicros); - void* cur = last_id.Acquire_Load(); - ASSERT_EQ(4, reinterpret_cast(cur)); + env_->SleepForMicroseconds(kDelayMicros); + ASSERT_EQ(4, last_id.load(std::memory_order_relaxed)); } struct State { port::Mutex mu; - int val; - int num_running; + int val GUARDED_BY(mu); + int num_running GUARDED_BY(mu); + + State(int val, int num_running) : val(val), num_running(num_running) {} }; static void ThreadBody(void* arg) { @@ -78,10 +135,8 @@ static void ThreadBody(void* arg) { s->mu.Unlock(); } -TEST(EnvPosixTest, StartThread) { - State state; - state.val = 0; - state.num_running = 3; +TEST(EnvTest, StartThread) { + State state(0, 3); for (int i = 0; i < 3; i++) { env_->StartThread(&ThreadBody, &state); } @@ -92,13 +147,79 @@ TEST(EnvPosixTest, StartThread) { if (num == 0) { break; } - Env::Default()->SleepForMicroseconds(kDelayMicros); + env_->SleepForMicroseconds(kDelayMicros); } + + MutexLock l(&state.mu); ASSERT_EQ(state.val, 3); } +TEST(EnvTest, TestOpenNonExistentFile) { + // Write some test data to a single file that will be opened |n| times. + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + + std::string non_existent_file = test_dir + "/non_existent_file"; + ASSERT_TRUE(!env_->FileExists(non_existent_file)); + + RandomAccessFile* random_access_file; + Status status = + env_->NewRandomAccessFile(non_existent_file, &random_access_file); + ASSERT_TRUE(status.IsNotFound()); + + SequentialFile* sequential_file; + status = env_->NewSequentialFile(non_existent_file, &sequential_file); + ASSERT_TRUE(status.IsNotFound()); +} + +TEST(EnvTest, ReopenWritableFile) { + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file_name = test_dir + "/reopen_writable_file.txt"; + env_->DeleteFile(test_file_name); + + WritableFile* writable_file; + ASSERT_OK(env_->NewWritableFile(test_file_name, &writable_file)); + std::string data("hello world!"); + ASSERT_OK(writable_file->Append(data)); + ASSERT_OK(writable_file->Close()); + delete writable_file; + + ASSERT_OK(env_->NewWritableFile(test_file_name, &writable_file)); + data = "42"; + ASSERT_OK(writable_file->Append(data)); + ASSERT_OK(writable_file->Close()); + delete writable_file; + + ASSERT_OK(ReadFileToString(env_, test_file_name, &data)); + ASSERT_EQ(std::string("42"), data); + env_->DeleteFile(test_file_name); +} + +TEST(EnvTest, ReopenAppendableFile) { + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file_name = test_dir + "/reopen_appendable_file.txt"; + env_->DeleteFile(test_file_name); + + WritableFile* appendable_file; + ASSERT_OK(env_->NewAppendableFile(test_file_name, &appendable_file)); + std::string data("hello world!"); + ASSERT_OK(appendable_file->Append(data)); + ASSERT_OK(appendable_file->Close()); + delete appendable_file; + + ASSERT_OK(env_->NewAppendableFile(test_file_name, &appendable_file)); + data = "42"; + ASSERT_OK(appendable_file->Append(data)); + ASSERT_OK(appendable_file->Close()); + delete appendable_file; + + ASSERT_OK(ReadFileToString(env_, test_file_name, &data)); + ASSERT_EQ(std::string("hello world!42"), data); + env_->DeleteFile(test_file_name); +} + } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/env_windows.cc b/util/env_windows.cc new file mode 100644 index 0000000..09e3df6 --- /dev/null +++ b/util/env_windows.cc @@ -0,0 +1,711 @@ +// Copyright (c) 2018 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. + +// Prevent Windows headers from defining min/max macros and instead +// use STL. +#ifndef NOMINMAX +#define NOMINMAX +#endif // ifndef NOMINMAX +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "leveldb/env.h" +#include "leveldb/slice.h" +#include "port/port.h" +#include "port/thread_annotations.h" +#include "util/env_windows_test_helper.h" +#include "util/logging.h" +#include "util/mutexlock.h" +#include "util/windows_logger.h" + +#if defined(DeleteFile) +#undef DeleteFile +#endif // defined(DeleteFile) + +namespace leveldb { + +namespace { + +constexpr const size_t kWritableFileBufferSize = 65536; + +// Up to 1000 mmaps for 64-bit binaries; none for 32-bit. +constexpr int kDefaultMmapLimit = sizeof(void*) >= 8 ? 1000 : 0; + +// Modified by EnvWindowsTestHelper::SetReadOnlyMMapLimit(). +int g_mmap_limit = kDefaultMmapLimit; + +std::string GetWindowsErrorMessage(DWORD error_code) { + std::string message; + char* error_text = nullptr; + // Use MBCS version of FormatMessage to match return value. + size_t error_text_size = ::FormatMessageA( + FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_ALLOCATE_BUFFER | + FORMAT_MESSAGE_IGNORE_INSERTS, + nullptr, error_code, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), + reinterpret_cast(&error_text), 0, nullptr); + if (!error_text) { + return message; + } + message.assign(error_text, error_text_size); + ::LocalFree(error_text); + return message; +} + +Status WindowsError(const std::string& context, DWORD error_code) { + if (error_code == ERROR_FILE_NOT_FOUND || error_code == ERROR_PATH_NOT_FOUND) + return Status::NotFound(context, GetWindowsErrorMessage(error_code)); + return Status::IOError(context, GetWindowsErrorMessage(error_code)); +} + +class ScopedHandle { + public: + ScopedHandle(HANDLE handle) : handle_(handle) {} + ScopedHandle(ScopedHandle&& other) noexcept : handle_(other.Release()) {} + ~ScopedHandle() { Close(); } + + ScopedHandle& operator=(ScopedHandle&& rhs) noexcept { + if (this != &rhs) handle_ = rhs.Release(); + return *this; + } + + bool Close() { + if (!is_valid()) { + return true; + } + HANDLE h = handle_; + handle_ = INVALID_HANDLE_VALUE; + return ::CloseHandle(h); + } + + bool is_valid() const { + return handle_ != INVALID_HANDLE_VALUE && handle_ != nullptr; + } + + HANDLE get() const { return handle_; } + + HANDLE Release() { + HANDLE h = handle_; + handle_ = INVALID_HANDLE_VALUE; + return h; + } + + private: + HANDLE handle_; +}; + +// Helper class to limit resource usage to avoid exhaustion. +// Currently used to limit read-only file descriptors and mmap file usage +// so that we do not run out of file descriptors or virtual memory, or run into +// kernel performance problems for very large databases. +class Limiter { + public: + // Limit maximum number of resources to |max_acquires|. + Limiter(int max_acquires) : acquires_allowed_(max_acquires) {} + + Limiter(const Limiter&) = delete; + Limiter operator=(const Limiter&) = delete; + + // If another resource is available, acquire it and return true. + // Else return false. + bool Acquire() { + int old_acquires_allowed = + acquires_allowed_.fetch_sub(1, std::memory_order_relaxed); + + if (old_acquires_allowed > 0) return true; + + acquires_allowed_.fetch_add(1, std::memory_order_relaxed); + return false; + } + + // Release a resource acquired by a previous call to Acquire() that returned + // true. + void Release() { acquires_allowed_.fetch_add(1, std::memory_order_relaxed); } + + private: + // The number of available resources. + // + // This is a counter and is not tied to the invariants of any other class, so + // it can be operated on safely using std::memory_order_relaxed. + std::atomic acquires_allowed_; +}; + +class WindowsSequentialFile : public SequentialFile { + public: + WindowsSequentialFile(std::string fname, ScopedHandle file) + : filename_(fname), file_(std::move(file)) {} + ~WindowsSequentialFile() override {} + + Status Read(size_t n, Slice* result, char* scratch) override { + Status s; + DWORD bytes_read; + // DWORD is 32-bit, but size_t could technically be larger. However leveldb + // files are limited to leveldb::Options::max_file_size which is clamped to + // 1<<30 or 1 GiB. + assert(n <= std::numeric_limits::max()); + if (!::ReadFile(file_.get(), scratch, static_cast(n), &bytes_read, + nullptr)) { + s = WindowsError(filename_, ::GetLastError()); + } else { + *result = Slice(scratch, bytes_read); + } + return s; + } + + Status Skip(uint64_t n) override { + LARGE_INTEGER distance; + distance.QuadPart = n; + if (!::SetFilePointerEx(file_.get(), distance, nullptr, FILE_CURRENT)) { + return WindowsError(filename_, ::GetLastError()); + } + return Status::OK(); + } + + private: + std::string filename_; + ScopedHandle file_; +}; + +class WindowsRandomAccessFile : public RandomAccessFile { + public: + WindowsRandomAccessFile(std::string fname, ScopedHandle handle) + : filename_(fname), handle_(std::move(handle)) {} + + ~WindowsRandomAccessFile() override = default; + + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + DWORD bytes_read = 0; + OVERLAPPED overlapped = {0}; + + overlapped.OffsetHigh = static_cast(offset >> 32); + overlapped.Offset = static_cast(offset); + if (!::ReadFile(handle_.get(), scratch, static_cast(n), &bytes_read, + &overlapped)) { + DWORD error_code = ::GetLastError(); + if (error_code != ERROR_HANDLE_EOF) { + *result = Slice(scratch, 0); + return Status::IOError(filename_, GetWindowsErrorMessage(error_code)); + } + } + + *result = Slice(scratch, bytes_read); + return Status::OK(); + } + + private: + std::string filename_; + ScopedHandle handle_; +}; + +class WindowsMmapReadableFile : public RandomAccessFile { + public: + // base[0,length-1] contains the mmapped contents of the file. + WindowsMmapReadableFile(std::string fname, void* base, size_t length, + Limiter* limiter) + : filename_(std::move(fname)), + mmapped_region_(base), + length_(length), + limiter_(limiter) {} + + ~WindowsMmapReadableFile() override { + ::UnmapViewOfFile(mmapped_region_); + limiter_->Release(); + } + + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + Status s; + if (offset + n > length_) { + *result = Slice(); + s = WindowsError(filename_, ERROR_INVALID_PARAMETER); + } else { + *result = Slice(reinterpret_cast(mmapped_region_) + offset, n); + } + return s; + } + + private: + std::string filename_; + void* mmapped_region_; + size_t length_; + Limiter* limiter_; +}; + +class WindowsWritableFile : public WritableFile { + public: + WindowsWritableFile(std::string fname, ScopedHandle handle) + : filename_(std::move(fname)), handle_(std::move(handle)), pos_(0) {} + + ~WindowsWritableFile() override = default; + + Status Append(const Slice& data) override { + size_t n = data.size(); + const char* p = data.data(); + + // Fit as much as possible into buffer. + size_t copy = std::min(n, kWritableFileBufferSize - pos_); + memcpy(buf_ + pos_, p, copy); + p += copy; + n -= copy; + pos_ += copy; + if (n == 0) { + return Status::OK(); + } + + // Can't fit in buffer, so need to do at least one write. + Status s = FlushBuffered(); + if (!s.ok()) { + return s; + } + + // Small writes go to buffer, large writes are written directly. + if (n < kWritableFileBufferSize) { + memcpy(buf_, p, n); + pos_ = n; + return Status::OK(); + } + return WriteRaw(p, n); + } + + Status Close() override { + Status result = FlushBuffered(); + if (!handle_.Close() && result.ok()) { + result = WindowsError(filename_, ::GetLastError()); + } + return result; + } + + Status Flush() override { return FlushBuffered(); } + + Status Sync() override { + // On Windows no need to sync parent directory. It's metadata will be + // updated via the creation of the new file, without an explicit sync. + return FlushBuffered(); + } + + private: + Status FlushBuffered() { + Status s = WriteRaw(buf_, pos_); + pos_ = 0; + return s; + } + + Status WriteRaw(const char* p, size_t n) { + DWORD bytes_written; + if (!::WriteFile(handle_.get(), p, static_cast(n), &bytes_written, + nullptr)) { + return Status::IOError(filename_, + GetWindowsErrorMessage(::GetLastError())); + } + return Status::OK(); + } + + // buf_[0, pos_-1] contains data to be written to handle_. + const std::string filename_; + ScopedHandle handle_; + char buf_[kWritableFileBufferSize]; + size_t pos_; +}; + +// Lock or unlock the entire file as specified by |lock|. Returns true +// when successful, false upon failure. Caller should call ::GetLastError() +// to determine cause of failure +bool LockOrUnlock(HANDLE handle, bool lock) { + if (lock) { + return ::LockFile(handle, + /*dwFileOffsetLow=*/0, /*dwFileOffsetHigh=*/0, + /*nNumberOfBytesToLockLow=*/MAXDWORD, + /*nNumberOfBytesToLockHigh=*/MAXDWORD); + } else { + return ::UnlockFile(handle, + /*dwFileOffsetLow=*/0, /*dwFileOffsetHigh=*/0, + /*nNumberOfBytesToLockLow=*/MAXDWORD, + /*nNumberOfBytesToLockHigh=*/MAXDWORD); + } +} + +class WindowsFileLock : public FileLock { + public: + WindowsFileLock(ScopedHandle handle, std::string name) + : handle_(std::move(handle)), name_(std::move(name)) {} + + ScopedHandle& handle() { return handle_; } + const std::string& name() const { return name_; } + + private: + ScopedHandle handle_; + std::string name_; +}; + +class WindowsEnv : public Env { + public: + WindowsEnv(); + ~WindowsEnv() override { + static char msg[] = "Destroying Env::Default()\n"; + fwrite(msg, 1, sizeof(msg), stderr); + abort(); + } + + Status NewSequentialFile(const std::string& fname, + SequentialFile** result) override { + *result = nullptr; + DWORD desired_access = GENERIC_READ; + DWORD share_mode = FILE_SHARE_READ; + ScopedHandle handle = + ::CreateFileA(fname.c_str(), desired_access, share_mode, nullptr, + OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, nullptr); + if (!handle.is_valid()) { + return WindowsError(fname, ::GetLastError()); + } + *result = new WindowsSequentialFile(fname, std::move(handle)); + return Status::OK(); + } + + Status NewRandomAccessFile(const std::string& fname, + RandomAccessFile** result) override { + *result = nullptr; + DWORD desired_access = GENERIC_READ; + DWORD share_mode = FILE_SHARE_READ; + DWORD file_flags = FILE_ATTRIBUTE_READONLY; + + ScopedHandle handle = + ::CreateFileA(fname.c_str(), desired_access, share_mode, nullptr, + OPEN_EXISTING, file_flags, nullptr); + if (!handle.is_valid()) { + return WindowsError(fname, ::GetLastError()); + } + if (!mmap_limiter_.Acquire()) { + *result = new WindowsRandomAccessFile(fname, std::move(handle)); + return Status::OK(); + } + + LARGE_INTEGER file_size; + if (!::GetFileSizeEx(handle.get(), &file_size)) { + return WindowsError(fname, ::GetLastError()); + } + + ScopedHandle mapping = + ::CreateFileMappingA(handle.get(), + /*security attributes=*/nullptr, PAGE_READONLY, + /*dwMaximumSizeHigh=*/0, + /*dwMaximumSizeLow=*/0, nullptr); + if (mapping.is_valid()) { + void* base = MapViewOfFile(mapping.get(), FILE_MAP_READ, 0, 0, 0); + if (base) { + *result = new WindowsMmapReadableFile( + fname, base, static_cast(file_size.QuadPart), + &mmap_limiter_); + return Status::OK(); + } + } + Status s = WindowsError(fname, ::GetLastError()); + + if (!s.ok()) { + mmap_limiter_.Release(); + } + return s; + } + + Status NewWritableFile(const std::string& fname, + WritableFile** result) override { + DWORD desired_access = GENERIC_WRITE; + DWORD share_mode = 0; + + ScopedHandle handle = + ::CreateFileA(fname.c_str(), desired_access, share_mode, nullptr, + CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, nullptr); + if (!handle.is_valid()) { + *result = nullptr; + return WindowsError(fname, ::GetLastError()); + } + + *result = new WindowsWritableFile(fname, std::move(handle)); + return Status::OK(); + } + + Status NewAppendableFile(const std::string& fname, + WritableFile** result) override { + ScopedHandle handle = + ::CreateFileA(fname.c_str(), FILE_APPEND_DATA, 0, nullptr, OPEN_ALWAYS, + FILE_ATTRIBUTE_NORMAL, nullptr); + if (!handle.is_valid()) { + *result = nullptr; + return WindowsError(fname, ::GetLastError()); + } + + *result = new WindowsWritableFile(fname, std::move(handle)); + return Status::OK(); + } + + bool FileExists(const std::string& fname) override { + return GetFileAttributesA(fname.c_str()) != INVALID_FILE_ATTRIBUTES; + } + + Status GetChildren(const std::string& dir, + std::vector* result) override { + const std::string find_pattern = dir + "\\*"; + WIN32_FIND_DATAA find_data; + HANDLE dir_handle = ::FindFirstFileA(find_pattern.c_str(), &find_data); + if (dir_handle == INVALID_HANDLE_VALUE) { + DWORD last_error = ::GetLastError(); + if (last_error == ERROR_FILE_NOT_FOUND) { + return Status::OK(); + } + return WindowsError(dir, last_error); + } + do { + char base_name[_MAX_FNAME]; + char ext[_MAX_EXT]; + + if (!_splitpath_s(find_data.cFileName, nullptr, 0, nullptr, 0, base_name, + ARRAYSIZE(base_name), ext, ARRAYSIZE(ext))) { + result->emplace_back(std::string(base_name) + ext); + } + } while (::FindNextFileA(dir_handle, &find_data)); + DWORD last_error = ::GetLastError(); + ::FindClose(dir_handle); + if (last_error != ERROR_NO_MORE_FILES) { + return WindowsError(dir, last_error); + } + return Status::OK(); + } + + Status DeleteFile(const std::string& fname) override { + if (!::DeleteFileA(fname.c_str())) { + return WindowsError(fname, ::GetLastError()); + } + return Status::OK(); + } + + Status CreateDir(const std::string& name) override { + if (!::CreateDirectoryA(name.c_str(), nullptr)) { + return WindowsError(name, ::GetLastError()); + } + return Status::OK(); + } + + Status DeleteDir(const std::string& name) override { + if (!::RemoveDirectoryA(name.c_str())) { + return WindowsError(name, ::GetLastError()); + } + return Status::OK(); + } + + Status GetFileSize(const std::string& fname, uint64_t* size) override { + WIN32_FILE_ATTRIBUTE_DATA attrs; + if (!::GetFileAttributesExA(fname.c_str(), GetFileExInfoStandard, &attrs)) { + return WindowsError(fname, ::GetLastError()); + } + ULARGE_INTEGER file_size; + file_size.HighPart = attrs.nFileSizeHigh; + file_size.LowPart = attrs.nFileSizeLow; + *size = file_size.QuadPart; + return Status::OK(); + } + + Status RenameFile(const std::string& src, + const std::string& target) override { + // Try a simple move first. It will only succeed when |to_path| doesn't + // already exist. + if (::MoveFileA(src.c_str(), target.c_str())) { + return Status::OK(); + } + DWORD move_error = ::GetLastError(); + + // Try the full-blown replace if the move fails, as ReplaceFile will only + // succeed when |to_path| does exist. When writing to a network share, we + // may not be able to change the ACLs. Ignore ACL errors then + // (REPLACEFILE_IGNORE_MERGE_ERRORS). + if (::ReplaceFileA(target.c_str(), src.c_str(), nullptr, + REPLACEFILE_IGNORE_MERGE_ERRORS, nullptr, nullptr)) { + return Status::OK(); + } + DWORD replace_error = ::GetLastError(); + // In the case of FILE_ERROR_NOT_FOUND from ReplaceFile, it is likely + // that |to_path| does not exist. In this case, the more relevant error + // comes from the call to MoveFile. + if (replace_error == ERROR_FILE_NOT_FOUND || + replace_error == ERROR_PATH_NOT_FOUND) { + return WindowsError(src, move_error); + } else { + return WindowsError(src, replace_error); + } + } + + Status LockFile(const std::string& fname, FileLock** lock) override { + *lock = nullptr; + Status result; + ScopedHandle handle = ::CreateFileA( + fname.c_str(), GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ, + /*lpSecurityAttributes=*/nullptr, OPEN_ALWAYS, FILE_ATTRIBUTE_NORMAL, + nullptr); + if (!handle.is_valid()) { + result = WindowsError(fname, ::GetLastError()); + } else if (!LockOrUnlock(handle.get(), true)) { + result = WindowsError("lock " + fname, ::GetLastError()); + } else { + *lock = new WindowsFileLock(std::move(handle), std::move(fname)); + } + return result; + } + + Status UnlockFile(FileLock* lock) override { + std::unique_ptr my_lock( + reinterpret_cast(lock)); + Status result; + if (!LockOrUnlock(my_lock->handle().get(), false)) { + result = WindowsError("unlock", ::GetLastError()); + } + return result; + } + + void Schedule(void (*function)(void*), void* arg) override; + + void StartThread(void (*function)(void* arg), void* arg) override { + std::thread t(function, arg); + t.detach(); + } + + Status GetTestDirectory(std::string* result) override { + const char* env = getenv("TEST_TMPDIR"); + if (env && env[0] != '\0') { + *result = env; + return Status::OK(); + } + + char tmp_path[MAX_PATH]; + if (!GetTempPathA(ARRAYSIZE(tmp_path), tmp_path)) { + return WindowsError("GetTempPath", ::GetLastError()); + } + std::stringstream ss; + ss << tmp_path << "leveldbtest-" << std::this_thread::get_id(); + *result = ss.str(); + + // Directory may already exist + CreateDir(*result); + return Status::OK(); + } + + Status NewLogger(const std::string& filename, Logger** result) override { + std::FILE* fp = std::fopen(filename.c_str(), "w"); + if (fp == nullptr) { + *result = nullptr; + return WindowsError("NewLogger", ::GetLastError()); + } else { + *result = new WindowsLogger(fp); + return Status::OK(); + } + } + + uint64_t NowMicros() override { + // GetSystemTimeAsFileTime typically has a resolution of 10-20 msec. + // TODO(cmumford): Switch to GetSystemTimePreciseAsFileTime which is + // available in Windows 8 and later. + FILETIME ft; + ::GetSystemTimeAsFileTime(&ft); + // Each tick represents a 100-nanosecond intervals since January 1, 1601 + // (UTC). + uint64_t num_ticks = + (static_cast(ft.dwHighDateTime) << 32) + ft.dwLowDateTime; + return num_ticks / 10; + } + + void SleepForMicroseconds(int micros) override { + std::this_thread::sleep_for(std::chrono::microseconds(micros)); + } + + private: + // Entry per Schedule() call + struct BGItem { + void* arg; + void (*function)(void*); + }; + + // BGThread() is the body of the background thread + void BGThread(); + + std::mutex mu_; + std::condition_variable bgsignal_; + bool started_bgthread_; + std::deque queue_; + Limiter mmap_limiter_; +}; + +// Return the maximum number of concurrent mmaps. +int MaxMmaps() { + if (g_mmap_limit >= 0) { + return g_mmap_limit; + } + // Up to 1000 mmaps for 64-bit binaries; none for smaller pointer sizes. + g_mmap_limit = sizeof(void*) >= 8 ? 1000 : 0; + return g_mmap_limit; +} + +WindowsEnv::WindowsEnv() + : started_bgthread_(false), mmap_limiter_(MaxMmaps()) {} + +void WindowsEnv::Schedule(void (*function)(void*), void* arg) { + std::lock_guard guard(mu_); + + // Start background thread if necessary + if (!started_bgthread_) { + started_bgthread_ = true; + std::thread t(&WindowsEnv::BGThread, this); + t.detach(); + } + + // If the queue is currently empty, the background thread may currently be + // waiting. + if (queue_.empty()) { + bgsignal_.notify_one(); + } + + // Add to priority queue + queue_.push_back(BGItem()); + queue_.back().function = function; + queue_.back().arg = arg; +} + +void WindowsEnv::BGThread() { + while (true) { + // Wait until there is an item that is ready to run + std::unique_lock lk(mu_); + bgsignal_.wait(lk, [this] { return !queue_.empty(); }); + + void (*function)(void*) = queue_.front().function; + void* arg = queue_.front().arg; + queue_.pop_front(); + + lk.unlock(); + (*function)(arg); + } +} + +} // namespace + +static std::once_flag once; +static Env* default_env; +static void InitDefaultEnv() { default_env = new WindowsEnv(); } + +void EnvWindowsTestHelper::SetReadOnlyMMapLimit(int limit) { + assert(default_env == nullptr); + g_mmap_limit = limit; +} + +Env* Env::Default() { + std::call_once(once, InitDefaultEnv); + return default_env; +} + +} // namespace leveldb diff --git a/util/env_windows_test.cc b/util/env_windows_test.cc new file mode 100644 index 0000000..3c22133 --- /dev/null +++ b/util/env_windows_test.cc @@ -0,0 +1,64 @@ +// Copyright (c) 2018 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 "leveldb/env.h" + +#include "port/port.h" +#include "util/env_windows_test_helper.h" +#include "util/testharness.h" + +namespace leveldb { + +static const int kMMapLimit = 4; + +class EnvWindowsTest { + public: + static void SetFileLimits(int mmap_limit) { + EnvWindowsTestHelper::SetReadOnlyMMapLimit(mmap_limit); + } + + EnvWindowsTest() : env_(Env::Default()) {} + + Env* env_; +}; + +TEST(EnvWindowsTest, TestOpenOnRead) { + // Write some test data to a single file that will be opened |n| times. + std::string test_dir; + ASSERT_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file = test_dir + "/open_on_read.txt"; + + FILE* f = fopen(test_file.c_str(), "w"); + ASSERT_TRUE(f != nullptr); + const char kFileData[] = "abcdefghijklmnopqrstuvwxyz"; + fputs(kFileData, f); + fclose(f); + + // Open test file some number above the sum of the two limits to force + // leveldb::WindowsEnv to switch from mapping the file into memory + // to basic file reading. + const int kNumFiles = kMMapLimit + 5; + leveldb::RandomAccessFile* files[kNumFiles] = {0}; + for (int i = 0; i < kNumFiles; i++) { + ASSERT_OK(env_->NewRandomAccessFile(test_file, &files[i])); + } + char scratch; + Slice read_result; + for (int i = 0; i < kNumFiles; i++) { + ASSERT_OK(files[i]->Read(i, 1, &read_result, &scratch)); + ASSERT_EQ(kFileData[i], read_result[0]); + } + for (int i = 0; i < kNumFiles; i++) { + delete files[i]; + } + ASSERT_OK(env_->DeleteFile(test_file)); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + // All tests currently run with the same read-only file limits. + leveldb::EnvWindowsTest::SetFileLimits(leveldb::kMMapLimit); + return leveldb::test::RunAllTests(); +} diff --git a/util/env_windows_test_helper.h b/util/env_windows_test_helper.h new file mode 100644 index 0000000..e6f6020 --- /dev/null +++ b/util/env_windows_test_helper.h @@ -0,0 +1,25 @@ +// Copyright 2018 (c) 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_UTIL_ENV_WINDOWS_TEST_HELPER_H_ +#define STORAGE_LEVELDB_UTIL_ENV_WINDOWS_TEST_HELPER_H_ + +namespace leveldb { + +class EnvWindowsTest; + +// A helper for the Windows Env to facilitate testing. +class EnvWindowsTestHelper { + private: + friend class CorruptionTest; + friend class EnvWindowsTest; + + // Set the maximum number of read-only files that will be mapped via mmap. + // Must be called before creating an Env. + static void SetReadOnlyMMapLimit(int limit); +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_ENV_WINDOWS_TEST_HELPER_H_ diff --git a/util/filter_policy.cc b/util/filter_policy.cc index 7b045c8..90fd754 100644 --- a/util/filter_policy.cc +++ b/util/filter_policy.cc @@ -6,6 +6,6 @@ namespace leveldb { -FilterPolicy::~FilterPolicy() { } +FilterPolicy::~FilterPolicy() {} } // namespace leveldb diff --git a/util/hash.cc b/util/hash.cc index ed439ce..67dc134 100644 --- a/util/hash.cc +++ b/util/hash.cc @@ -2,15 +2,19 @@ // 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 "util/coding.h" #include "util/hash.h" +#include + +#include "util/coding.h" + // The FALLTHROUGH_INTENDED macro can be used to annotate implicit fall-through // between switch labels. The real definition should be provided externally. // This one is a fallback version for unsupported compilers. #ifndef FALLTHROUGH_INTENDED -#define FALLTHROUGH_INTENDED do { } while (0) +#define FALLTHROUGH_INTENDED \ + do { \ + } while (0) #endif namespace leveldb { @@ -48,5 +52,4 @@ uint32_t Hash(const char* data, size_t n, uint32_t seed) { return h; } - } // namespace leveldb diff --git a/util/hash.h b/util/hash.h index 8889d56..74bdb6e 100644 --- a/util/hash.h +++ b/util/hash.h @@ -12,8 +12,8 @@ namespace leveldb { -extern uint32_t Hash(const char* data, size_t n, uint32_t seed); +uint32_t Hash(const char* data, size_t n, uint32_t seed); -} +} // namespace leveldb #endif // STORAGE_LEVELDB_UTIL_HASH_H_ diff --git a/util/hash_test.cc b/util/hash_test.cc index eaa1c92..8f579cc 100644 --- a/util/hash_test.cc +++ b/util/hash_test.cc @@ -7,7 +7,7 @@ namespace leveldb { -class HASH { }; +class HASH {}; TEST(HASH, SignedUnsignedIssue) { const unsigned char data1[1] = {0x62}; @@ -15,18 +15,10 @@ TEST(HASH, SignedUnsignedIssue) { const unsigned char data3[3] = {0xe2, 0x99, 0xa5}; const unsigned char data4[4] = {0xe1, 0x80, 0xb9, 0x32}; const unsigned char data5[48] = { - 0x01, 0xc0, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x14, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x04, 0x00, - 0x00, 0x00, 0x00, 0x14, - 0x00, 0x00, 0x00, 0x18, - 0x28, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, - 0x02, 0x00, 0x00, 0x00, - 0x00, 0x00, 0x00, 0x00, + 0x01, 0xc0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, + 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x18, 0x28, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, }; ASSERT_EQ(Hash(0, 0, 0xbc9f1d34), 0xbc9f1d34); @@ -49,6 +41,4 @@ TEST(HASH, SignedUnsignedIssue) { } // namespace leveldb -int main(int argc, char** argv) { - return leveldb::test::RunAllTests(); -} +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/histogram.cc b/util/histogram.cc index bb95f58..65092c8 100644 --- a/util/histogram.cc +++ b/util/histogram.cc @@ -2,36 +2,174 @@ // 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 "util/histogram.h" + #include #include + #include "port/port.h" -#include "util/histogram.h" namespace leveldb { const double Histogram::kBucketLimit[kNumBuckets] = { - 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 16, 18, 20, 25, 30, 35, 40, 45, - 50, 60, 70, 80, 90, 100, 120, 140, 160, 180, 200, 250, 300, 350, 400, 450, - 500, 600, 700, 800, 900, 1000, 1200, 1400, 1600, 1800, 2000, 2500, 3000, - 3500, 4000, 4500, 5000, 6000, 7000, 8000, 9000, 10000, 12000, 14000, - 16000, 18000, 20000, 25000, 30000, 35000, 40000, 45000, 50000, 60000, - 70000, 80000, 90000, 100000, 120000, 140000, 160000, 180000, 200000, - 250000, 300000, 350000, 400000, 450000, 500000, 600000, 700000, 800000, - 900000, 1000000, 1200000, 1400000, 1600000, 1800000, 2000000, 2500000, - 3000000, 3500000, 4000000, 4500000, 5000000, 6000000, 7000000, 8000000, - 9000000, 10000000, 12000000, 14000000, 16000000, 18000000, 20000000, - 25000000, 30000000, 35000000, 40000000, 45000000, 50000000, 60000000, - 70000000, 80000000, 90000000, 100000000, 120000000, 140000000, 160000000, - 180000000, 200000000, 250000000, 300000000, 350000000, 400000000, - 450000000, 500000000, 600000000, 700000000, 800000000, 900000000, - 1000000000, 1200000000, 1400000000, 1600000000, 1800000000, 2000000000, - 2500000000.0, 3000000000.0, 3500000000.0, 4000000000.0, 4500000000.0, - 5000000000.0, 6000000000.0, 7000000000.0, 8000000000.0, 9000000000.0, - 1e200, + 1, + 2, + 3, + 4, + 5, + 6, + 7, + 8, + 9, + 10, + 12, + 14, + 16, + 18, + 20, + 25, + 30, + 35, + 40, + 45, + 50, + 60, + 70, + 80, + 90, + 100, + 120, + 140, + 160, + 180, + 200, + 250, + 300, + 350, + 400, + 450, + 500, + 600, + 700, + 800, + 900, + 1000, + 1200, + 1400, + 1600, + 1800, + 2000, + 2500, + 3000, + 3500, + 4000, + 4500, + 5000, + 6000, + 7000, + 8000, + 9000, + 10000, + 12000, + 14000, + 16000, + 18000, + 20000, + 25000, + 30000, + 35000, + 40000, + 45000, + 50000, + 60000, + 70000, + 80000, + 90000, + 100000, + 120000, + 140000, + 160000, + 180000, + 200000, + 250000, + 300000, + 350000, + 400000, + 450000, + 500000, + 600000, + 700000, + 800000, + 900000, + 1000000, + 1200000, + 1400000, + 1600000, + 1800000, + 2000000, + 2500000, + 3000000, + 3500000, + 4000000, + 4500000, + 5000000, + 6000000, + 7000000, + 8000000, + 9000000, + 10000000, + 12000000, + 14000000, + 16000000, + 18000000, + 20000000, + 25000000, + 30000000, + 35000000, + 40000000, + 45000000, + 50000000, + 60000000, + 70000000, + 80000000, + 90000000, + 100000000, + 120000000, + 140000000, + 160000000, + 180000000, + 200000000, + 250000000, + 300000000, + 350000000, + 400000000, + 450000000, + 500000000, + 600000000, + 700000000, + 800000000, + 900000000, + 1000000000, + 1200000000, + 1400000000, + 1600000000, + 1800000000, + 2000000000, + 2500000000.0, + 3000000000.0, + 3500000000.0, + 4000000000.0, + 4500000000.0, + 5000000000.0, + 6000000000.0, + 7000000000.0, + 8000000000.0, + 9000000000.0, + 1e200, }; void Histogram::Clear() { - min_ = kBucketLimit[kNumBuckets-1]; + min_ = kBucketLimit[kNumBuckets - 1]; max_ = 0; num_ = 0; sum_ = 0; @@ -66,9 +204,7 @@ void Histogram::Merge(const Histogram& other) { } } -double Histogram::Median() const { - return Percentile(50.0); -} +double Histogram::Median() const { return Percentile(50.0); } double Histogram::Percentile(double p) const { double threshold = num_ * (p / 100.0); @@ -77,7 +213,7 @@ double Histogram::Percentile(double p) const { sum += buckets_[b]; if (sum >= threshold) { // Scale linearly within this bucket - double left_point = (b == 0) ? 0 : kBucketLimit[b-1]; + double left_point = (b == 0) ? 0 : kBucketLimit[b - 1]; double right_point = kBucketLimit[b]; double left_sum = sum - buckets_[b]; double right_sum = sum; @@ -105,12 +241,10 @@ double Histogram::StandardDeviation() const { std::string Histogram::ToString() const { std::string r; char buf[200]; - snprintf(buf, sizeof(buf), - "Count: %.0f Average: %.4f StdDev: %.2f\n", - num_, Average(), StandardDeviation()); + snprintf(buf, sizeof(buf), "Count: %.0f Average: %.4f StdDev: %.2f\n", num_, + Average(), StandardDeviation()); r.append(buf); - snprintf(buf, sizeof(buf), - "Min: %.4f Median: %.4f Max: %.4f\n", + snprintf(buf, sizeof(buf), "Min: %.4f Median: %.4f Max: %.4f\n", (num_ == 0.0 ? 0.0 : min_), Median(), max_); r.append(buf); r.append("------------------------------------------------------\n"); @@ -119,17 +253,16 @@ std::string Histogram::ToString() const { for (int b = 0; b < kNumBuckets; b++) { if (buckets_[b] <= 0.0) continue; sum += buckets_[b]; - snprintf(buf, sizeof(buf), - "[ %7.0f, %7.0f ) %7.0f %7.3f%% %7.3f%% ", - ((b == 0) ? 0.0 : kBucketLimit[b-1]), // left - kBucketLimit[b], // right - buckets_[b], // count - mult * buckets_[b], // percentage - mult * sum); // cumulative percentage + snprintf(buf, sizeof(buf), "[ %7.0f, %7.0f ) %7.0f %7.3f%% %7.3f%% ", + ((b == 0) ? 0.0 : kBucketLimit[b - 1]), // left + kBucketLimit[b], // right + buckets_[b], // count + mult * buckets_[b], // percentage + mult * sum); // cumulative percentage r.append(buf); // Add hash marks based on percentage; 20 marks for 100%. - int marks = static_cast(20*(buckets_[b] / num_) + 0.5); + int marks = static_cast(20 * (buckets_[b] / num_) + 0.5); r.append(marks, '#'); r.push_back('\n'); } diff --git a/util/histogram.h b/util/histogram.h index 1ef9f3c..4da60fb 100644 --- a/util/histogram.h +++ b/util/histogram.h @@ -11,8 +11,8 @@ namespace leveldb { class Histogram { public: - Histogram() { } - ~Histogram() { } + Histogram() {} + ~Histogram() {} void Clear(); void Add(double value); @@ -21,20 +21,22 @@ class Histogram { std::string ToString() const; private: + enum { kNumBuckets = 154 }; + + double Median() const; + double Percentile(double p) const; + double Average() const; + double StandardDeviation() const; + + static const double kBucketLimit[kNumBuckets]; + double min_; double max_; double num_; double sum_; double sum_squares_; - enum { kNumBuckets = 154 }; - static const double kBucketLimit[kNumBuckets]; double buckets_[kNumBuckets]; - - double Median() const; - double Percentile(double p) const; - double Average() const; - double StandardDeviation() const; }; } // namespace leveldb diff --git a/util/logging.cc b/util/logging.cc index ca6b324..1ad8f1c 100644 --- a/util/logging.cc +++ b/util/logging.cc @@ -8,6 +8,9 @@ #include #include #include + +#include + #include "leveldb/env.h" #include "leveldb/slice.h" @@ -15,7 +18,7 @@ namespace leveldb { void AppendNumberTo(std::string* str, uint64_t num) { char buf[30]; - snprintf(buf, sizeof(buf), "%llu", (unsigned long long) num); + snprintf(buf, sizeof(buf), "%llu", (unsigned long long)num); str->append(buf); } @@ -46,27 +49,37 @@ std::string EscapeString(const Slice& value) { } bool ConsumeDecimalNumber(Slice* in, uint64_t* val) { - uint64_t v = 0; - int digits = 0; - while (!in->empty()) { - char c = (*in)[0]; - if (c >= '0' && c <= '9') { - ++digits; - const int delta = (c - '0'); - static const uint64_t kMaxUint64 = ~static_cast(0); - if (v > kMaxUint64/10 || - (v == kMaxUint64/10 && delta > kMaxUint64%10)) { - // Overflow - return false; - } - v = (v * 10) + delta; - in->remove_prefix(1); - } else { - break; + // Constants that will be optimized away. + constexpr const uint64_t kMaxUint64 = std::numeric_limits::max(); + constexpr const char kLastDigitOfMaxUint64 = + '0' + static_cast(kMaxUint64 % 10); + + uint64_t value = 0; + + // reinterpret_cast-ing from char* to unsigned char* to avoid signedness. + const unsigned char* start = + reinterpret_cast(in->data()); + + const unsigned char* end = start + in->size(); + const unsigned char* current = start; + for (; current != end; ++current) { + const unsigned char ch = *current; + if (ch < '0' || ch > '9') break; + + // Overflow check. + // kMaxUint64 / 10 is also constant and will be optimized away. + if (value > kMaxUint64 / 10 || + (value == kMaxUint64 / 10 && ch > kLastDigitOfMaxUint64)) { + return false; } + + value = (value * 10) + (ch - '0'); } - *val = v; - return (digits > 0); + + *val = value; + const size_t digits_consumed = current - start; + in->remove_prefix(digits_consumed); + return digits_consumed != 0; } } // namespace leveldb diff --git a/util/logging.h b/util/logging.h index 1b450d2..8ff2da8 100644 --- a/util/logging.h +++ b/util/logging.h @@ -8,9 +8,11 @@ #ifndef STORAGE_LEVELDB_UTIL_LOGGING_H_ #define STORAGE_LEVELDB_UTIL_LOGGING_H_ -#include #include +#include + #include + #include "port/port.h" namespace leveldb { @@ -19,24 +21,24 @@ class Slice; class WritableFile; // Append a human-readable printout of "num" to *str -extern void AppendNumberTo(std::string* str, uint64_t num); +void AppendNumberTo(std::string* str, uint64_t num); // Append a human-readable printout of "value" to *str. // Escapes any non-printable characters found in "value". -extern void AppendEscapedStringTo(std::string* str, const Slice& value); +void AppendEscapedStringTo(std::string* str, const Slice& value); // Return a human-readable printout of "num" -extern std::string NumberToString(uint64_t num); +std::string NumberToString(uint64_t num); // Return a human-readable version of "value". // Escapes any non-printable characters found in "value". -extern std::string EscapeString(const Slice& value); +std::string EscapeString(const Slice& value); // Parse a human-readable number from "*in" into *value. On success, // advances "*in" past the consumed number and sets "*val" to the // numeric value. Otherwise, returns false and leaves *in in an // unspecified state. -extern bool ConsumeDecimalNumber(Slice* in, uint64_t* val); +bool ConsumeDecimalNumber(Slice* in, uint64_t* val); } // namespace leveldb diff --git a/util/logging_test.cc b/util/logging_test.cc new file mode 100644 index 0000000..389cbeb --- /dev/null +++ b/util/logging_test.cc @@ -0,0 +1,143 @@ +// Copyright (c) 2018 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 "leveldb/slice.h" +#include "util/logging.h" +#include "util/testharness.h" + +namespace leveldb { + +class Logging {}; + +TEST(Logging, NumberToString) { + ASSERT_EQ("0", NumberToString(0)); + ASSERT_EQ("1", NumberToString(1)); + ASSERT_EQ("9", NumberToString(9)); + + ASSERT_EQ("10", NumberToString(10)); + ASSERT_EQ("11", NumberToString(11)); + ASSERT_EQ("19", NumberToString(19)); + ASSERT_EQ("99", NumberToString(99)); + + ASSERT_EQ("100", NumberToString(100)); + ASSERT_EQ("109", NumberToString(109)); + ASSERT_EQ("190", NumberToString(190)); + ASSERT_EQ("123", NumberToString(123)); + ASSERT_EQ("12345678", NumberToString(12345678)); + + static_assert(std::numeric_limits::max() == 18446744073709551615U, + "Test consistency check"); + ASSERT_EQ("18446744073709551000", NumberToString(18446744073709551000U)); + ASSERT_EQ("18446744073709551600", NumberToString(18446744073709551600U)); + ASSERT_EQ("18446744073709551610", NumberToString(18446744073709551610U)); + ASSERT_EQ("18446744073709551614", NumberToString(18446744073709551614U)); + ASSERT_EQ("18446744073709551615", NumberToString(18446744073709551615U)); +} + +void ConsumeDecimalNumberRoundtripTest(uint64_t number, + const std::string& padding = "") { + std::string decimal_number = NumberToString(number); + std::string input_string = decimal_number + padding; + Slice input(input_string); + Slice output = input; + uint64_t result; + ASSERT_TRUE(ConsumeDecimalNumber(&output, &result)); + ASSERT_EQ(number, result); + ASSERT_EQ(decimal_number.size(), output.data() - input.data()); + ASSERT_EQ(padding.size(), output.size()); +} + +TEST(Logging, ConsumeDecimalNumberRoundtrip) { + ConsumeDecimalNumberRoundtripTest(0); + ConsumeDecimalNumberRoundtripTest(1); + ConsumeDecimalNumberRoundtripTest(9); + + ConsumeDecimalNumberRoundtripTest(10); + ConsumeDecimalNumberRoundtripTest(11); + ConsumeDecimalNumberRoundtripTest(19); + ConsumeDecimalNumberRoundtripTest(99); + + ConsumeDecimalNumberRoundtripTest(100); + ConsumeDecimalNumberRoundtripTest(109); + ConsumeDecimalNumberRoundtripTest(190); + ConsumeDecimalNumberRoundtripTest(123); + ASSERT_EQ("12345678", NumberToString(12345678)); + + for (uint64_t i = 0; i < 100; ++i) { + uint64_t large_number = std::numeric_limits::max() - i; + ConsumeDecimalNumberRoundtripTest(large_number); + } +} + +TEST(Logging, ConsumeDecimalNumberRoundtripWithPadding) { + ConsumeDecimalNumberRoundtripTest(0, " "); + ConsumeDecimalNumberRoundtripTest(1, "abc"); + ConsumeDecimalNumberRoundtripTest(9, "x"); + + ConsumeDecimalNumberRoundtripTest(10, "_"); + ConsumeDecimalNumberRoundtripTest(11, std::string("\0\0\0", 3)); + ConsumeDecimalNumberRoundtripTest(19, "abc"); + ConsumeDecimalNumberRoundtripTest(99, "padding"); + + ConsumeDecimalNumberRoundtripTest(100, " "); + + for (uint64_t i = 0; i < 100; ++i) { + uint64_t large_number = std::numeric_limits::max() - i; + ConsumeDecimalNumberRoundtripTest(large_number, "pad"); + } +} + +void ConsumeDecimalNumberOverflowTest(const std::string& input_string) { + Slice input(input_string); + Slice output = input; + uint64_t result; + ASSERT_EQ(false, ConsumeDecimalNumber(&output, &result)); +} + +TEST(Logging, ConsumeDecimalNumberOverflow) { + static_assert(std::numeric_limits::max() == 18446744073709551615U, + "Test consistency check"); + ConsumeDecimalNumberOverflowTest("18446744073709551616"); + ConsumeDecimalNumberOverflowTest("18446744073709551617"); + ConsumeDecimalNumberOverflowTest("18446744073709551618"); + ConsumeDecimalNumberOverflowTest("18446744073709551619"); + ConsumeDecimalNumberOverflowTest("18446744073709551620"); + ConsumeDecimalNumberOverflowTest("18446744073709551621"); + ConsumeDecimalNumberOverflowTest("18446744073709551622"); + ConsumeDecimalNumberOverflowTest("18446744073709551623"); + ConsumeDecimalNumberOverflowTest("18446744073709551624"); + ConsumeDecimalNumberOverflowTest("18446744073709551625"); + ConsumeDecimalNumberOverflowTest("18446744073709551626"); + + ConsumeDecimalNumberOverflowTest("18446744073709551700"); + + ConsumeDecimalNumberOverflowTest("99999999999999999999"); +} + +void ConsumeDecimalNumberNoDigitsTest(const std::string& input_string) { + Slice input(input_string); + Slice output = input; + uint64_t result; + ASSERT_EQ(false, ConsumeDecimalNumber(&output, &result)); + ASSERT_EQ(input.data(), output.data()); + ASSERT_EQ(input.size(), output.size()); +} + +TEST(Logging, ConsumeDecimalNumberNoDigits) { + ConsumeDecimalNumberNoDigitsTest(""); + ConsumeDecimalNumberNoDigitsTest(" "); + ConsumeDecimalNumberNoDigitsTest("a"); + ConsumeDecimalNumberNoDigitsTest(" 123"); + ConsumeDecimalNumberNoDigitsTest("a123"); + ConsumeDecimalNumberNoDigitsTest(std::string("\000123", 4)); + ConsumeDecimalNumberNoDigitsTest(std::string("\177123", 4)); + ConsumeDecimalNumberNoDigitsTest(std::string("\377123", 4)); +} + +} // namespace leveldb + +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/mutexlock.h b/util/mutexlock.h index 1ff5a9e..0cb2e25 100644 --- a/util/mutexlock.h +++ b/util/mutexlock.h @@ -22,20 +22,18 @@ namespace leveldb { class SCOPED_LOCKABLE MutexLock { public: - explicit MutexLock(port::Mutex *mu) EXCLUSIVE_LOCK_FUNCTION(mu) - : mu_(mu) { + explicit MutexLock(port::Mutex* mu) EXCLUSIVE_LOCK_FUNCTION(mu) : mu_(mu) { this->mu_->Lock(); } ~MutexLock() UNLOCK_FUNCTION() { this->mu_->Unlock(); } + MutexLock(const MutexLock&) = delete; + MutexLock& operator=(const MutexLock&) = delete; + private: - port::Mutex *const mu_; - // No copying allowed - MutexLock(const MutexLock&); - void operator=(const MutexLock&); + port::Mutex* const mu_; }; } // namespace leveldb - #endif // STORAGE_LEVELDB_UTIL_MUTEXLOCK_H_ diff --git a/util/no_destructor.h b/util/no_destructor.h new file mode 100644 index 0000000..a0d3b87 --- /dev/null +++ b/util/no_destructor.h @@ -0,0 +1,46 @@ +// Copyright (c) 2018 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_UTIL_NO_DESTRUCTOR_H_ +#define STORAGE_LEVELDB_UTIL_NO_DESTRUCTOR_H_ + +#include +#include + +namespace leveldb { + +// Wraps an instance whose destructor is never called. +// +// This is intended for use with function-level static variables. +template +class NoDestructor { + public: + template + explicit NoDestructor(ConstructorArgTypes&&... constructor_args) { + static_assert(sizeof(instance_storage_) >= sizeof(InstanceType), + "instance_storage_ is not large enough to hold the instance"); + static_assert( + alignof(decltype(instance_storage_)) >= alignof(InstanceType), + "instance_storage_ does not meet the instance's alignment requirement"); + new (&instance_storage_) + InstanceType(std::forward(constructor_args)...); + } + + ~NoDestructor() = default; + + NoDestructor(const NoDestructor&) = delete; + NoDestructor& operator=(const NoDestructor&) = delete; + + InstanceType* get() { + return reinterpret_cast(&instance_storage_); + } + + private: + typename std::aligned_storage::type instance_storage_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_NO_DESTRUCTOR_H_ diff --git a/util/no_destructor_test.cc b/util/no_destructor_test.cc new file mode 100644 index 0000000..b41caca --- /dev/null +++ b/util/no_destructor_test.cc @@ -0,0 +1,47 @@ +// Copyright (c) 2018 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 "util/no_destructor.h" +#include "util/testharness.h" + +namespace leveldb { + +namespace { + +struct DoNotDestruct { + public: + DoNotDestruct(uint32_t a, uint64_t b) : a(a), b(b) {} + ~DoNotDestruct() { std::abort(); } + + // Used to check constructor argument forwarding. + uint32_t a; + uint64_t b; +}; + +constexpr const uint32_t kGoldenA = 0xdeadbeef; +constexpr const uint64_t kGoldenB = 0xaabbccddeeffaabb; + +} // namespace + +class NoDestructorTest {}; + +TEST(NoDestructorTest, StackInstance) { + NoDestructor instance(kGoldenA, kGoldenB); + ASSERT_EQ(kGoldenA, instance.get()->a); + ASSERT_EQ(kGoldenB, instance.get()->b); +} + +TEST(NoDestructorTest, StaticInstance) { + static NoDestructor instance(kGoldenA, kGoldenB); + ASSERT_EQ(kGoldenA, instance.get()->a); + ASSERT_EQ(kGoldenB, instance.get()->b); +} + +} // namespace leveldb + +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/options.cc b/util/options.cc index 76af5b9..62de5bf 100644 --- a/util/options.cc +++ b/util/options.cc @@ -9,21 +9,6 @@ namespace leveldb { -Options::Options() - : comparator(BytewiseComparator()), - create_if_missing(false), - error_if_exists(false), - paranoid_checks(false), - env(Env::Default()), - info_log(NULL), - write_buffer_size(4<<20), - max_open_files(1000), - block_cache(NULL), - block_size(4096), - block_restart_interval(16), - compression(kSnappyCompression), - filter_policy(NULL) { -} - +Options::Options() : comparator(BytewiseComparator()), env(Env::Default()) {} } // namespace leveldb diff --git a/util/posix_logger.h b/util/posix_logger.h index 9741b1a..28e15d1 100644 --- a/util/posix_logger.h +++ b/util/posix_logger.h @@ -8,89 +8,121 @@ #ifndef STORAGE_LEVELDB_UTIL_POSIX_LOGGER_H_ #define STORAGE_LEVELDB_UTIL_POSIX_LOGGER_H_ -#include -#include #include -#include + +#include +#include +#include +#include +#include +#include + #include "leveldb/env.h" namespace leveldb { -class PosixLogger : public Logger { - private: - FILE* file_; - uint64_t (*gettid_)(); // Return the thread id for the current thread +class PosixLogger final : public Logger { public: - PosixLogger(FILE* f, uint64_t (*gettid)()) : file_(f), gettid_(gettid) { } - virtual ~PosixLogger() { - fclose(file_); - } - virtual void Logv(const char* format, va_list ap) { - const uint64_t thread_id = (*gettid_)(); + // Creates a logger that writes to the given file. + // + // The PosixLogger instance takes ownership of the file handle. + explicit PosixLogger(std::FILE* fp) : fp_(fp) { assert(fp != nullptr); } - // 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; + ~PosixLogger() override { std::fclose(fp_); } - struct timeval now_tv; - gettimeofday(&now_tv, NULL); - const time_t seconds = now_tv.tv_sec; - struct tm t; - localtime_r(&seconds, &t); - 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)); + void Logv(const char* format, va_list arguments) override { + // Record the time as close to the Logv() call as possible. + struct ::timeval now_timeval; + ::gettimeofday(&now_timeval, nullptr); + const std::time_t now_seconds = now_timeval.tv_sec; + struct std::tm now_components; + ::localtime_r(&now_seconds, &now_components); - // 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); - } + // Record the thread ID. + constexpr const int kMaxThreadIdSize = 32; + std::ostringstream thread_stream; + thread_stream << std::this_thread::get_id(); + std::string thread_id = thread_stream.str(); + if (thread_id.size() > kMaxThreadIdSize) { + thread_id.resize(kMaxThreadIdSize); + } - // Truncate to available space if necessary - if (p >= limit) { - if (iter == 0) { - continue; // Try again with larger buffer - } else { - p = limit - 1; + // We first attempt to print into a stack-allocated buffer. If this attempt + // fails, we make a second attempt with a dynamically allocated buffer. + constexpr const int kStackBufferSize = 512; + char stack_buffer[kStackBufferSize]; + static_assert(sizeof(stack_buffer) == static_cast(kStackBufferSize), + "sizeof(char) is expected to be 1 in C++"); + + int dynamic_buffer_size = 0; // Computed in the first iteration. + for (int iteration = 0; iteration < 2; ++iteration) { + const int buffer_size = + (iteration == 0) ? kStackBufferSize : dynamic_buffer_size; + char* const buffer = + (iteration == 0) ? stack_buffer : new char[dynamic_buffer_size]; + + // Print the header into the buffer. + int buffer_offset = snprintf( + buffer, buffer_size, "%04d/%02d/%02d-%02d:%02d:%02d.%06d %s ", + now_components.tm_year + 1900, now_components.tm_mon + 1, + now_components.tm_mday, now_components.tm_hour, now_components.tm_min, + now_components.tm_sec, static_cast(now_timeval.tv_usec), + thread_id.c_str()); + + // The header can be at most 28 characters (10 date + 15 time + + // 3 delimiters) plus the thread ID, which should fit comfortably into the + // static buffer. + assert(buffer_offset <= 28 + kMaxThreadIdSize); + static_assert(28 + kMaxThreadIdSize < kStackBufferSize, + "stack-allocated buffer may not fit the message header"); + assert(buffer_offset < buffer_size); + + // Print the message into the buffer. + std::va_list arguments_copy; + va_copy(arguments_copy, arguments); + buffer_offset += + std::vsnprintf(buffer + buffer_offset, buffer_size - buffer_offset, + format, arguments_copy); + va_end(arguments_copy); + + // The code below may append a newline at the end of the buffer, which + // requires an extra character. + if (buffer_offset >= buffer_size - 1) { + // The message did not fit into the buffer. + if (iteration == 0) { + // Re-run the loop and use a dynamically-allocated buffer. The buffer + // will be large enough for the log message, an extra newline and a + // null terminator. + dynamic_buffer_size = buffer_offset + 2; + continue; } + + // The dynamically-allocated buffer was incorrectly sized. This should + // not happen, assuming a correct implementation of (v)snprintf. Fail + // in tests, recover by truncating the log message in production. + assert(false); + buffer_offset = buffer_size - 1; } - // Add newline if necessary - if (p == base || p[-1] != '\n') { - *p++ = '\n'; + // Add a newline if necessary. + if (buffer[buffer_offset - 1] != '\n') { + buffer[buffer_offset] = '\n'; + ++buffer_offset; } - assert(p <= limit); - fwrite(base, 1, p - base, file_); - fflush(file_); - if (base != buffer) { - delete[] base; + assert(buffer_offset <= buffer_size); + std::fwrite(buffer, 1, buffer_offset, fp_); + std::fflush(fp_); + + if (iteration != 0) { + delete[] buffer; } break; } } + + private: + std::FILE* const fp_; }; } // namespace leveldb diff --git a/util/random.h b/util/random.h index ddd51b1..76f7daf 100644 --- a/util/random.h +++ b/util/random.h @@ -15,6 +15,7 @@ namespace leveldb { class Random { private: uint32_t seed_; + public: explicit Random(uint32_t s) : seed_(s & 0x7fffffffu) { // Avoid bad seeds. @@ -23,8 +24,8 @@ class Random { } } uint32_t Next() { - static const uint32_t M = 2147483647L; // 2^31-1 - static const uint64_t A = 16807; // bits 14, 8, 7, 5, 2, 1, 0 + static const uint32_t M = 2147483647L; // 2^31-1 + static const uint64_t A = 16807; // bits 14, 8, 7, 5, 2, 1, 0 // We are computing // seed_ = (seed_ * A) % M, where M = 2^31-1 // @@ -54,9 +55,7 @@ class Random { // Skewed: pick "base" uniformly from range [0,max_log] and then // return "base" random bits. The effect is to pick a number in the // range [0,2^max_log-1] with exponential bias towards smaller numbers. - uint32_t Skewed(int max_log) { - return Uniform(1 << Uniform(max_log + 1)); - } + uint32_t Skewed(int max_log) { return Uniform(1 << Uniform(max_log + 1)); } }; } // namespace leveldb diff --git a/util/status.cc b/util/status.cc index a44f35b..6ca8da6 100644 --- a/util/status.cc +++ b/util/status.cc @@ -2,10 +2,12 @@ // 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 "port/port.h" #include "leveldb/status.h" +#include + +#include "port/port.h" + namespace leveldb { const char* Status::CopyState(const char* state) { @@ -34,7 +36,7 @@ Status::Status(Code code, const Slice& msg, const Slice& msg2) { } std::string Status::ToString() const { - if (state_ == NULL) { + if (state_ == nullptr) { return "OK"; } else { char tmp[30]; @@ -59,8 +61,8 @@ std::string Status::ToString() const { type = "IO error: "; break; default: - snprintf(tmp, sizeof(tmp), "Unknown code(%d): ", - static_cast(code())); + snprintf(tmp, sizeof(tmp), + "Unknown code(%d): ", static_cast(code())); type = tmp; break; } diff --git a/util/status_test.cc b/util/status_test.cc new file mode 100644 index 0000000..2842319 --- /dev/null +++ b/util/status_test.cc @@ -0,0 +1,40 @@ +// Copyright (c) 2018 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 "leveldb/slice.h" +#include "leveldb/status.h" +#include "util/testharness.h" + +namespace leveldb { + +TEST(Status, MoveConstructor) { + { + Status ok = Status::OK(); + Status ok2 = std::move(ok); + + ASSERT_TRUE(ok2.ok()); + } + + { + Status status = Status::NotFound("custom NotFound status message"); + Status status2 = std::move(status); + + ASSERT_TRUE(status2.IsNotFound()); + ASSERT_EQ("NotFound: custom NotFound status message", status2.ToString()); + } + + { + Status self_moved = Status::IOError("custom IOError status message"); + + // Needed to bypass compiler warning about explicit move-assignment. + Status& self_moved_reference = self_moved; + self_moved_reference = std::move(self_moved); + } +} + +} // namespace leveldb + +int main(int argc, char** argv) { return leveldb::test::RunAllTests(); } diff --git a/util/testharness.cc b/util/testharness.cc index 402fab3..318ecfa 100644 --- a/util/testharness.cc +++ b/util/testharness.cc @@ -4,11 +4,15 @@ #include "util/testharness.h" -#include #include #include #include +#include +#include + +#include "leveldb/env.h" + namespace leveldb { namespace test { @@ -19,10 +23,10 @@ struct Test { void (*func)(); }; std::vector* tests; -} +} // namespace bool RegisterTest(const char* base, const char* name, void (*func)()) { - if (tests == NULL) { + if (tests == nullptr) { tests = new std::vector; } Test t; @@ -37,14 +41,14 @@ int RunAllTests() { const char* matcher = getenv("LEVELDB_TESTS"); int num = 0; - if (tests != NULL) { + if (tests != nullptr) { for (size_t i = 0; i < tests->size(); i++) { const Test& t = (*tests)[i]; - if (matcher != NULL) { + if (matcher != nullptr) { std::string name = t.base; name.push_back('.'); name.append(t.name); - if (strstr(name.c_str(), matcher) == NULL) { + if (strstr(name.c_str(), matcher) == nullptr) { continue; } } @@ -66,7 +70,7 @@ std::string TmpDir() { int RandomSeed() { const char* env = getenv("TEST_RANDOM_SEED"); - int result = (env != NULL ? atoi(env) : 301); + int result = (env != nullptr ? atoi(env) : 301); if (result <= 0) { result = 301; } diff --git a/util/testharness.h b/util/testharness.h index da4fe68..72cd162 100644 --- a/util/testharness.h +++ b/util/testharness.h @@ -7,10 +7,10 @@ #include #include + #include -#include "leveldb/env.h" -#include "leveldb/slice.h" -#include "util/random.h" + +#include "leveldb/status.h" namespace leveldb { namespace test { @@ -27,15 +27,15 @@ namespace test { // // Returns 0 if all tests pass. // Dies or returns a non-zero value if some test fails. -extern int RunAllTests(); +int RunAllTests(); // Return the directory to use for temporary storage. -extern std::string TmpDir(); +std::string TmpDir(); // Return a randomization seed for this run. Typically returns the // same number on repeated invocations of this binary, but automated // runs may be able to vary the seed. -extern int RandomSeed(); +int RandomSeed(); // An instance of Tester is allocated to hold temporary state during // the execution of an assertion. @@ -47,9 +47,7 @@ class Tester { std::stringstream ss_; public: - Tester(const char* f, int l) - : ok_(true), fname_(f), line_(l) { - } + Tester(const char* f, int l) : ok_(true), fname_(f), line_(l) {} ~Tester() { if (!ok_) { @@ -74,14 +72,14 @@ class Tester { return *this; } -#define BINARY_OP(name,op) \ - template \ - Tester& name(const X& x, const Y& y) { \ - if (! (x op y)) { \ - ss_ << " failed: " << x << (" " #op " ") << y; \ - ok_ = false; \ - } \ - return *this; \ +#define BINARY_OP(name, op) \ + template \ + Tester& name(const X& x, const Y& y) { \ + if (!(x op y)) { \ + ss_ << " failed: " << x << (" " #op " ") << y; \ + ok_ = false; \ + } \ + return *this; \ } BINARY_OP(IsEq, ==) @@ -104,33 +102,38 @@ class Tester { #define ASSERT_TRUE(c) ::leveldb::test::Tester(__FILE__, __LINE__).Is((c), #c) #define ASSERT_OK(s) ::leveldb::test::Tester(__FILE__, __LINE__).IsOk((s)) -#define ASSERT_EQ(a,b) ::leveldb::test::Tester(__FILE__, __LINE__).IsEq((a),(b)) -#define ASSERT_NE(a,b) ::leveldb::test::Tester(__FILE__, __LINE__).IsNe((a),(b)) -#define ASSERT_GE(a,b) ::leveldb::test::Tester(__FILE__, __LINE__).IsGe((a),(b)) -#define ASSERT_GT(a,b) ::leveldb::test::Tester(__FILE__, __LINE__).IsGt((a),(b)) -#define ASSERT_LE(a,b) ::leveldb::test::Tester(__FILE__, __LINE__).IsLe((a),(b)) -#define ASSERT_LT(a,b) ::leveldb::test::Tester(__FILE__, __LINE__).IsLt((a),(b)) +#define ASSERT_EQ(a, b) \ + ::leveldb::test::Tester(__FILE__, __LINE__).IsEq((a), (b)) +#define ASSERT_NE(a, b) \ + ::leveldb::test::Tester(__FILE__, __LINE__).IsNe((a), (b)) +#define ASSERT_GE(a, b) \ + ::leveldb::test::Tester(__FILE__, __LINE__).IsGe((a), (b)) +#define ASSERT_GT(a, b) \ + ::leveldb::test::Tester(__FILE__, __LINE__).IsGt((a), (b)) +#define ASSERT_LE(a, b) \ + ::leveldb::test::Tester(__FILE__, __LINE__).IsLe((a), (b)) +#define ASSERT_LT(a, b) \ + ::leveldb::test::Tester(__FILE__, __LINE__).IsLt((a), (b)) -#define TCONCAT(a,b) TCONCAT1(a,b) -#define TCONCAT1(a,b) a##b +#define TCONCAT(a, b) TCONCAT1(a, b) +#define TCONCAT1(a, b) a##b -#define TEST(base,name) \ -class TCONCAT(_Test_,name) : public base { \ - public: \ - void _Run(); \ - static void _RunIt() { \ - TCONCAT(_Test_,name) t; \ - t._Run(); \ - } \ -}; \ -bool TCONCAT(_Test_ignored_,name) = \ - ::leveldb::test::RegisterTest(#base, #name, &TCONCAT(_Test_,name)::_RunIt); \ -void TCONCAT(_Test_,name)::_Run() +#define TEST(base, name) \ + class TCONCAT(_Test_, name) : public base { \ + public: \ + void _Run(); \ + static void _RunIt() { \ + TCONCAT(_Test_, name) t; \ + t._Run(); \ + } \ + }; \ + bool TCONCAT(_Test_ignored_, name) = ::leveldb::test::RegisterTest( \ + #base, #name, &TCONCAT(_Test_, name)::_RunIt); \ + void TCONCAT(_Test_, name)::_Run() // Register the specified test. Typically not used directly, but // invoked via the macro expansion of TEST. -extern bool RegisterTest(const char* base, const char* name, void (*func)()); - +bool RegisterTest(const char* base, const char* name, void (*func)()); } // namespace test } // namespace leveldb diff --git a/util/testutil.cc b/util/testutil.cc index bee56bf..6b151b9 100644 --- a/util/testutil.cc +++ b/util/testutil.cc @@ -12,7 +12,7 @@ namespace test { Slice RandomString(Random* rnd, int len, std::string* dst) { dst->resize(len); for (int i = 0; i < len; i++) { - (*dst)[i] = static_cast(' ' + rnd->Uniform(95)); // ' ' .. '~' + (*dst)[i] = static_cast(' ' + rnd->Uniform(95)); // ' ' .. '~' } return Slice(*dst); } @@ -20,9 +20,8 @@ Slice RandomString(Random* rnd, int len, std::string* dst) { std::string RandomKey(Random* rnd, int len) { // Make sure to generate a wide variety of characters so we // test the boundary conditions for short-key optimizations. - static const char kTestChars[] = { - '\0', '\1', 'a', 'b', 'c', 'd', 'e', '\xfd', '\xfe', '\xff' - }; + static const char kTestChars[] = {'\0', '\1', 'a', 'b', 'c', + 'd', 'e', '\xfd', '\xfe', '\xff'}; std::string result; for (int i = 0; i < len; i++) { result += kTestChars[rnd->Uniform(sizeof(kTestChars))]; @@ -30,9 +29,8 @@ std::string RandomKey(Random* rnd, int len) { return result; } - -extern Slice CompressibleString(Random* rnd, double compressed_fraction, - size_t len, std::string* dst) { +Slice CompressibleString(Random* rnd, double compressed_fraction, size_t len, + std::string* dst) { int raw = static_cast(len * compressed_fraction); if (raw < 1) raw = 1; std::string raw_data; diff --git a/util/testutil.h b/util/testutil.h index adad3fc..bb4051b 100644 --- a/util/testutil.h +++ b/util/testutil.h @@ -5,6 +5,7 @@ #ifndef STORAGE_LEVELDB_UTIL_TESTUTIL_H_ #define STORAGE_LEVELDB_UTIL_TESTUTIL_H_ +#include "helpers/memenv/memenv.h" #include "leveldb/env.h" #include "leveldb/slice.h" #include "util/random.h" @@ -14,17 +15,17 @@ namespace test { // Store in *dst a random string of length "len" and return a Slice that // references the generated data. -extern Slice RandomString(Random* rnd, int len, std::string* dst); +Slice RandomString(Random* rnd, int len, std::string* dst); // Return a random key with the specified length that may contain interesting // characters (e.g. \x00, \xff, etc.). -extern std::string RandomKey(Random* rnd, int len); +std::string RandomKey(Random* rnd, int len); // Store in *dst a string of length "len" that will compress to // "N*compressed_fraction" bytes and return a Slice that references // the generated data. -extern Slice CompressibleString(Random* rnd, double compressed_fraction, - size_t len, std::string* dst); +Slice CompressibleString(Random* rnd, double compressed_fraction, size_t len, + std::string* dst); // A wrapper that allows injection of errors. class ErrorEnv : public EnvWrapper { @@ -32,19 +33,31 @@ class ErrorEnv : public EnvWrapper { bool writable_file_error_; int num_writable_file_errors_; - ErrorEnv() : EnvWrapper(Env::Default()), - writable_file_error_(false), - num_writable_file_errors_(0) { } + ErrorEnv() + : EnvWrapper(NewMemEnv(Env::Default())), + writable_file_error_(false), + num_writable_file_errors_(0) {} + ~ErrorEnv() override { delete target(); } - virtual Status NewWritableFile(const std::string& fname, - WritableFile** result) { + Status NewWritableFile(const std::string& fname, + WritableFile** result) override { if (writable_file_error_) { ++num_writable_file_errors_; - *result = NULL; + *result = nullptr; return Status::IOError(fname, "fake error"); } return target()->NewWritableFile(fname, result); } + + Status NewAppendableFile(const std::string& fname, + WritableFile** result) override { + if (writable_file_error_) { + ++num_writable_file_errors_; + *result = nullptr; + return Status::IOError(fname, "fake error"); + } + return target()->NewAppendableFile(fname, result); + } }; } // namespace test diff --git a/util/windows_logger.h b/util/windows_logger.h new file mode 100644 index 0000000..9296063 --- /dev/null +++ b/util/windows_logger.h @@ -0,0 +1,124 @@ +// Copyright (c) 2018 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 for the Windows platform. + +#ifndef STORAGE_LEVELDB_UTIL_WINDOWS_LOGGER_H_ +#define STORAGE_LEVELDB_UTIL_WINDOWS_LOGGER_H_ + +#include +#include +#include +#include +#include +#include + +#include "leveldb/env.h" + +namespace leveldb { + +class WindowsLogger final : public Logger { + public: + // Creates a logger that writes to the given file. + // + // The PosixLogger instance takes ownership of the file handle. + explicit WindowsLogger(std::FILE* fp) : fp_(fp) { assert(fp != nullptr); } + + ~WindowsLogger() override { std::fclose(fp_); } + + void Logv(const char* format, va_list arguments) override { + // Record the time as close to the Logv() call as possible. + SYSTEMTIME now_components; + ::GetLocalTime(&now_components); + + // Record the thread ID. + constexpr const int kMaxThreadIdSize = 32; + std::ostringstream thread_stream; + thread_stream << std::this_thread::get_id(); + std::string thread_id = thread_stream.str(); + if (thread_id.size() > kMaxThreadIdSize) { + thread_id.resize(kMaxThreadIdSize); + } + + // We first attempt to print into a stack-allocated buffer. If this attempt + // fails, we make a second attempt with a dynamically allocated buffer. + constexpr const int kStackBufferSize = 512; + char stack_buffer[kStackBufferSize]; + static_assert(sizeof(stack_buffer) == static_cast(kStackBufferSize), + "sizeof(char) is expected to be 1 in C++"); + + int dynamic_buffer_size = 0; // Computed in the first iteration. + for (int iteration = 0; iteration < 2; ++iteration) { + const int buffer_size = + (iteration == 0) ? kStackBufferSize : dynamic_buffer_size; + char* const buffer = + (iteration == 0) ? stack_buffer : new char[dynamic_buffer_size]; + + // Print the header into the buffer. + int buffer_offset = snprintf( + buffer, buffer_size, "%04d/%02d/%02d-%02d:%02d:%02d.%06d %s ", + now_components.wYear, now_components.wMonth, now_components.wDay, + now_components.wHour, now_components.wMinute, now_components.wSecond, + static_cast(now_components.wMilliseconds * 1000), + thread_id.c_str()); + + // The header can be at most 28 characters (10 date + 15 time + + // 3 delimiters) plus the thread ID, which should fit comfortably into the + // static buffer. + assert(buffer_offset <= 28 + kMaxThreadIdSize); + static_assert(28 + kMaxThreadIdSize < kStackBufferSize, + "stack-allocated buffer may not fit the message header"); + assert(buffer_offset < buffer_size); + + // Print the message into the buffer. + std::va_list arguments_copy; + va_copy(arguments_copy, arguments); + buffer_offset += + std::vsnprintf(buffer + buffer_offset, buffer_size - buffer_offset, + format, arguments_copy); + va_end(arguments_copy); + + // The code below may append a newline at the end of the buffer, which + // requires an extra character. + if (buffer_offset >= buffer_size - 1) { + // The message did not fit into the buffer. + if (iteration == 0) { + // Re-run the loop and use a dynamically-allocated buffer. The buffer + // will be large enough for the log message, an extra newline and a + // null terminator. + dynamic_buffer_size = buffer_offset + 2; + continue; + } + + // The dynamically-allocated buffer was incorrectly sized. This should + // not happen, assuming a correct implementation of (v)snprintf. Fail + // in tests, recover by truncating the log message in production. + assert(false); + buffer_offset = buffer_size - 1; + } + + // Add a newline if necessary. + if (buffer[buffer_offset - 1] != '\n') { + buffer[buffer_offset] = '\n'; + ++buffer_offset; + } + + assert(buffer_offset <= buffer_size); + std::fwrite(buffer, 1, buffer_offset, fp_); + std::fflush(fp_); + + if (iteration != 0) { + delete[] buffer; + } + break; + } + } + + private: + std::FILE* const fp_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_WINDOWS_LOGGER_H_