diff options
| author | Dipankar Bhardwaj <dipankarb@google.com> | 2022-08-15 11:08:39 +0000 |
|---|---|---|
| committer | Dipankar Bhardwaj <dipankarb@google.com> | 2022-08-19 09:08:35 +0000 |
| commit | 45d41e1ee9e2d47c8b34c8cfb1b9af47b07118d5 (patch) | |
| tree | cda1290f5bb1ffae3da668f86b54131d2c1f87be | |
| parent | 69a34a013d866dcfb34ca810b33342f736c606b4 (diff) | |
| download | leveldb-45d41e1ee9e2d47c8b34c8cfb1b9af47b07118d5.tar.gz | |
Add leveldb files for version 1.23
Test: No dependency on module yet
Bug: 239414225
Change-Id: Iebb25f3bc23545a8300774375935d968e0275dc4
149 files changed, 30791 insertions, 0 deletions
@@ -0,0 +1,12 @@ +# Names should be added to this file like so: +# Name or Organization <email address> + +Google Inc. + +# Initial version authors: +Jeffrey Dean <jeff@google.com> +Sanjay Ghemawat <sanjay@google.com> + +# Partial list of contributors: +Kevin Regan <kevin.d.regan@gmail.com> +Johan Bilien <jobi@litl.com> diff --git a/CMakeLists.txt b/CMakeLists.txt new file mode 100644 index 0000000..f8285b8 --- /dev/null +++ b/CMakeLists.txt @@ -0,0 +1,495 @@ +# 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 1.23.0 LANGUAGES C CXX) + +# C standard can be overridden when this is used as a sub-project. +if(NOT CMAKE_C_STANDARD) + # 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) +endif(NOT CMAKE_C_STANDARD) + +# C++ standard can be overridden when this is used as a sub-project. +if(NOT CMAKE_CXX_STANDARD) + # This project requires C++11. + set(CMAKE_CXX_STANDARD 11) + set(CMAKE_CXX_STANDARD_REQUIRED ON) + set(CMAKE_CXX_EXTENSIONS OFF) +endif(NOT CMAKE_CXX_STANDARD) + +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(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 <unistd.h> 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) + +if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC") + # Disable C++ exceptions. + string(REGEX REPLACE "/EH[a-z]+" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /EHs-c-") + add_definitions(-D_HAS_EXCEPTIONS=0) + + # Disable RTTI. + string(REGEX REPLACE "/GR" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /GR-") +else(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC") + # Enable strict prototype warnings for C code in clang and gcc. + if(NOT CMAKE_C_FLAGS MATCHES "-Wstrict-prototypes") + set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wstrict-prototypes") + endif(NOT CMAKE_C_FLAGS MATCHES "-Wstrict-prototypes") + + # Disable C++ exceptions. + string(REGEX REPLACE "-fexceptions" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-exceptions") + + # Disable RTTI. + string(REGEX REPLACE "-frtti" "" CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-rtti") +endif(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC") + +# Test whether -Wthread-safety is available. See +# https://clang.llvm.org/docs/ThreadSafetyAnalysis.html +include(CheckCXXCompilerFlag) +check_cxx_compiler_flag(-Wthread-safety HAVE_CLANG_THREAD_SAFETY) + +# Used by googletest. +check_cxx_compiler_flag(-Wno-missing-field-initializers + LEVELDB_HAVE_NO_MISSING_FIELD_INITIALIZERS) + +include(CheckCXXSourceCompiles) + +# Test whether C++17 __has_include is available. +check_cxx_source_compiles(" +#if defined(__has_include) && __has_include(<string>) +#include <string> +#endif +int main() { std::string str; return 0; } +" HAVE_CXX17_HAS_INCLUDE) + +set(LEVELDB_PUBLIC_INCLUDE_DIR "include/leveldb") +set(LEVELDB_PORT_CONFIG_DIR "include/port") + +configure_file( + "port/port_config.h.in" + "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" +) + +include_directories( + "${PROJECT_BINARY_DIR}/include" + "." +) + +if(BUILD_SHARED_LIBS) + # Only export LEVELDB_EXPORT symbols from the shared library. + add_compile_options(-fvisibility=hidden) +endif(BUILD_SHARED_LIBS) + +# Must be included before CMAKE_INSTALL_INCLUDEDIR is used. +include(GNUInstallDirs) + +add_library(leveldb "") +target_sources(leveldb + PRIVATE + "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" + "db/builder.cc" + "db/builder.h" + "db/c.cc" + "db/db_impl.cc" + "db/db_impl.h" + "db/db_iter.cc" + "db/db_iter.h" + "db/dbformat.cc" + "db/dbformat.h" + "db/dumpfile.cc" + "db/filename.cc" + "db/filename.h" + "db/log_format.h" + "db/log_reader.cc" + "db/log_reader.h" + "db/log_writer.cc" + "db/log_writer.h" + "db/memtable.cc" + "db/memtable.h" + "db/repair.cc" + "db/skiplist.h" + "db/snapshot.h" + "db/table_cache.cc" + "db/table_cache.h" + "db/version_edit.cc" + "db/version_edit.h" + "db/version_set.cc" + "db/version_set.h" + "db/write_batch_internal.h" + "db/write_batch.cc" + "port/port_stdcxx.h" + "port/port.h" + "port/thread_annotations.h" + "table/block_builder.cc" + "table/block_builder.h" + "table/block.cc" + "table/block.h" + "table/filter_block.cc" + "table/filter_block.h" + "table/format.cc" + "table/format.h" + "table/iterator_wrapper.h" + "table/iterator.cc" + "table/merger.cc" + "table/merger.h" + "table/table_builder.cc" + "table/table.cc" + "table/two_level_iterator.cc" + "table/two_level_iterator.h" + "util/arena.cc" + "util/arena.h" + "util/bloom.cc" + "util/cache.cc" + "util/coding.cc" + "util/coding.h" + "util/comparator.cc" + "util/crc32c.cc" + "util/crc32c.h" + "util/env.cc" + "util/filter_policy.cc" + "util/hash.cc" + "util/hash.h" + "util/logging.cc" + "util/logging.h" + "util/mutexlock.h" + "util/no_destructor.h" + "util/options.cc" + "util/random.h" + "util/status.cc" + + # Only CMake 3.3+ supports PUBLIC sources in targets exported by "install". + $<$<VERSION_GREATER:CMAKE_VERSION,3.2>: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 + "util/env_windows.cc" + "util/windows_logger.h" + ) +else (WIN32) + target_sources(leveldb + PRIVATE + "util/env_posix.cc" + "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 + "helpers/memenv/memenv.cc" + "helpers/memenv/memenv.h" +) + +target_include_directories(leveldb + PUBLIC + $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/include> + $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}> +) + +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. + ${LEVELDB_PLATFORM_NAME}=1 +) +if (NOT HAVE_CXX17_HAS_INCLUDE) + target_compile_definitions(leveldb + PRIVATE + LEVELDB_HAS_PORT_CONFIG_H=1 + ) +endif(NOT HAVE_CXX17_HAS_INCLUDE) + +if(BUILD_SHARED_LIBS) + target_compile_definitions(leveldb + PUBLIC + # Used by include/export.h. + LEVELDB_SHARED_LIBRARY + ) +endif(BUILD_SHARED_LIBS) + +if(HAVE_CLANG_THREAD_SAFETY) + target_compile_options(leveldb + PUBLIC + -Werror -Wthread-safety) +endif(HAVE_CLANG_THREAD_SAFETY) + +if(HAVE_CRC32C) + target_link_libraries(leveldb crc32c) +endif(HAVE_CRC32C) +if(HAVE_SNAPPY) + target_link_libraries(leveldb snappy) +endif(HAVE_SNAPPY) +if(HAVE_TCMALLOC) + target_link_libraries(leveldb tcmalloc) +endif(HAVE_TCMALLOC) + +# Needed by port_stdcxx.h +find_package(Threads REQUIRED) +target_link_libraries(leveldb Threads::Threads) + +add_executable(leveldbutil + "db/leveldbutil.cc" +) +target_link_libraries(leveldbutil leveldb) + +if(LEVELDB_BUILD_TESTS) + enable_testing() + + # Prevent overriding the parent project's compiler/linker settings on Windows. + set(gtest_force_shared_crt ON CACHE BOOL "" FORCE) + set(install_gtest OFF) + set(install_gmock OFF) + set(build_gmock ON) + + # This project is tested using GoogleTest. + add_subdirectory("third_party/googletest") + + # This project uses Google benchmark for benchmarking. + set(BENCHMARK_ENABLE_TESTING OFF CACHE BOOL "" FORCE) + set(BENCHMARK_ENABLE_EXCEPTIONS OFF CACHE BOOL "" FORCE) + add_subdirectory("third_party/benchmark") + + # GoogleTest triggers a missing field initializers warning. + if(LEVELDB_HAVE_NO_MISSING_FIELD_INITIALIZERS) + set_property(TARGET gtest + APPEND PROPERTY COMPILE_OPTIONS -Wno-missing-field-initializers) + set_property(TARGET gmock + APPEND PROPERTY COMPILE_OPTIONS -Wno-missing-field-initializers) + endif(LEVELDB_HAVE_NO_MISSING_FIELD_INITIALIZERS) + + function(leveldb_test test_file) + get_filename_component(test_target_name "${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" + "util/testutil.cc" + "util/testutil.h" + + "${test_file}" + ) + target_link_libraries("${test_target_name}" leveldb gmock gtest benchmark) + 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("db/c_test.c") + leveldb_test("db/fault_injection_test.cc") + + leveldb_test("issues/issue178_test.cc") + leveldb_test("issues/issue200_test.cc") + leveldb_test("issues/issue320_test.cc") + + leveldb_test("util/env_test.cc") + leveldb_test("util/status_test.cc") + leveldb_test("util/no_destructor_test.cc") + + if(NOT BUILD_SHARED_LIBS) + leveldb_test("db/autocompact_test.cc") + leveldb_test("db/corruption_test.cc") + leveldb_test("db/db_test.cc") + leveldb_test("db/dbformat_test.cc") + leveldb_test("db/filename_test.cc") + leveldb_test("db/log_test.cc") + leveldb_test("db/recovery_test.cc") + leveldb_test("db/skiplist_test.cc") + leveldb_test("db/version_edit_test.cc") + leveldb_test("db/version_set_test.cc") + leveldb_test("db/write_batch_test.cc") + + leveldb_test("helpers/memenv/memenv_test.cc") + + leveldb_test("table/filter_block_test.cc") + leveldb_test("table/table_test.cc") + + leveldb_test("util/arena_test.cc") + leveldb_test("util/bloom_test.cc") + leveldb_test("util/cache_test.cc") + leveldb_test("util/coding_test.cc") + leveldb_test("util/crc32c_test.cc") + leveldb_test("util/hash_test.cc") + leveldb_test("util/logging_test.cc") + + # TODO(costan): This test also uses + # "util/env_{posix|windows}_test_helper.h" + if (WIN32) + leveldb_test("util/env_windows_test.cc") + else (WIN32) + leveldb_test("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}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h" + "util/histogram.cc" + "util/histogram.h" + "util/testutil.cc" + "util/testutil.h" + + "${bench_file}" + ) + target_link_libraries("${bench_target_name}" leveldb gmock gtest) + target_compile_definitions("${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("benchmarks/db_bench.cc") + endif(NOT BUILD_SHARED_LIBS) + + check_library_exists(sqlite3 sqlite3_open "" HAVE_SQLITE3) + if(HAVE_SQLITE3) + leveldb_benchmark("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 <kcpolydb.h> + +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("benchmarks/db_bench_tree_db.cc") + target_link_libraries(db_bench_tree_db kyotocabinet) + endif(HAVE_KYOTOCABINET) +endif(LEVELDB_BUILD_BENCHMARKS) + +if(LEVELDB_INSTALL) + install(TARGETS leveldb + EXPORT leveldbTargets + RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR} + LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR} + ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR} + ) + install( + FILES + "${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" + DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}/leveldb" + ) + + include(CMakePackageConfigHelpers) + configure_package_config_file( + "cmake/${PROJECT_NAME}Config.cmake.in" + "${PROJECT_BINARY_DIR}/cmake/${PROJECT_NAME}Config.cmake" + INSTALL_DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME}" + ) + write_basic_package_version_file( + "${PROJECT_BINARY_DIR}/cmake/${PROJECT_NAME}ConfigVersion.cmake" + COMPATIBILITY SameMajorVersion + ) + install( + EXPORT leveldbTargets + NAMESPACE leveldb:: + DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME}" + ) + install( + FILES + "${PROJECT_BINARY_DIR}/cmake/${PROJECT_NAME}Config.cmake" + "${PROJECT_BINARY_DIR}/cmake/${PROJECT_NAME}ConfigVersion.cmake" + DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME}" + ) +endif(LEVELDB_INSTALL) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md new file mode 100644 index 0000000..7ede021 --- /dev/null +++ b/CONTRIBUTING.md @@ -0,0 +1,36 @@ +# Contributing + +We'd love to accept your code patches! However, before we can take them, we +have to jump a couple of legal hurdles. + +## Contributor License Agreements + +Please fill out either the individual or corporate Contributor License +Agreement as appropriate. + +* If you are an individual writing original source code and you're sure you +own the intellectual property, then sign an [individual CLA](https://developers.google.com/open-source/cla/individual). +* If you work for a company that wants to allow you to contribute your work, +then sign a [corporate CLA](https://developers.google.com/open-source/cla/corporate). + +Follow either of the two links above to access the appropriate CLA and +instructions for how to sign and return it. + +## Submitting a Patch + +1. Sign the contributors license agreement above. +2. Decide which code you want to submit. A submission should be a set of changes +that addresses one issue in the [issue tracker](https://github.com/google/leveldb/issues). +Please don't mix more than one logical change per submission, because it makes +the history hard to follow. If you want to make a change +(e.g. add a sample or feature) that doesn't have a corresponding issue in the +issue tracker, please create one. +3. **Submitting**: When you are ready to submit, send us a Pull Request. Be +sure to include the issue number you fixed and the name you used to sign +the CLA. + +## Writing Code ## + +If your contribution contains code, please make sure that it follows +[the style guide](https://google.github.io/styleguide/cppguide.html). +Otherwise we will have to ask you to make changes, and that's no fun for anyone. @@ -0,0 +1,27 @@ +Copyright (c) 2011 The LevelDB Authors. All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are +met: + + * Redistributions of source code must retain the above copyright +notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above +copyright notice, this list of conditions and the following disclaimer +in the documentation and/or other materials provided with the +distribution. + * Neither the name of Google Inc. nor the names of its +contributors may be used to endorse or promote products derived from +this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. diff --git a/METADATA b/METADATA new file mode 100644 index 0000000..e7fc4a6 --- /dev/null +++ b/METADATA @@ -0,0 +1,18 @@ +name: "leveldb" +description: + "LevelDB is a fast key-value storage library written at Google that " + "provides an ordered mapping from string keys to string values." + +third_party { + url { + type: HOMEPAGE + value: "https://github.com/google/leveldb/" + } + url { + type: GIT + value: "https://github.com/google/leveldb" + } + version: "1.23" + last_upgrade_date { year: 2022 month: 7 day: 18 } + license_type: NOTICE +} diff --git a/MODULE_LICENSE_BSD b/MODULE_LICENSE_BSD new file mode 100644 index 0000000..c59d9b6 --- /dev/null +++ b/MODULE_LICENSE_BSD @@ -0,0 +1 @@ +d
\ No newline at end of file @@ -0,0 +1,17 @@ +Release 1.2 2011-05-16 +---------------------- + +Fixes for larger databases (tested up to one billion 100-byte entries, +i.e., ~100GB). + +(1) Place hard limit on number of level-0 files. This fixes errors +of the form "too many open files". + +(2) Fixed memtable management. Before the fix, a heavy write burst +could cause unbounded memory usage. + +A fix for a logging bug where the reader would incorrectly complain +about corruption. + +Allow public access to WriteBatch contents so that users can easily +wrap a DB. diff --git a/README.md b/README.md new file mode 100644 index 0000000..81144dd --- /dev/null +++ b/README.md @@ -0,0 +1,231 @@ +**LevelDB is a fast key-value storage library written at Google that provides an ordered mapping from string keys to string values.** + +[](https://travis-ci.org/google/leveldb) +[](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. + * The basic operations are `Put(key,value)`, `Get(key)`, `Delete(key)`. + * 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](https://google.github.io/snappy/). + * External activity (file system operations etc.) is relayed through a virtual interface so users can customize the operating system interactions. + +# 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. + +# Getting the Source + +```bash +git clone --recurse-submodules https://github.com/google/leveldb.git +``` + +# 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 <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 +included db_bench program. The results are somewhat noisy, but should +be enough to get a ballpark performance estimate. + +## Setup + +We use a database with a million entries. Each entry has a 16 byte +key, and a 100 byte value. Values used by the benchmark compress to +about half their original size. + + LevelDB: version 1.1 + Date: Sun May 1 12:11:26 2011 + CPU: 4 x Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz + CPUCache: 4096 KB + Keys: 16 bytes each + Values: 100 bytes each (50 bytes after compression) + Entries: 1000000 + Raw Size: 110.6 MB (estimated) + File Size: 62.9 MB (estimated) + +## Write performance + +The "fill" benchmarks create a brand new database, in either +sequential, or random order. The "fillsync" benchmark flushes data +from the operating system to the disk after every operation; the other +write operations leave the data sitting in the operating system buffer +cache for a while. The "overwrite" benchmark does random writes that +update existing keys in the database. + + fillseq : 1.765 micros/op; 62.7 MB/s + fillsync : 268.409 micros/op; 0.4 MB/s (10000 ops) + fillrandom : 2.460 micros/op; 45.0 MB/s + overwrite : 2.380 micros/op; 46.5 MB/s + +Each "op" above corresponds to a write of a single key/value pair. +I.e., a random write benchmark goes at approximately 400,000 writes per second. + +Each "fillsync" operation costs much less (0.3 millisecond) +than a disk seek (typically 10 milliseconds). We suspect that this is +because the hard disk itself is buffering the update in its memory and +responding before the data has been written to the platter. This may +or may not be safe based on whether or not the hard disk has enough +power to save its memory in the event of a power failure. + +## Read performance + +We list the performance of reading sequentially in both the forward +and reverse direction, and also the performance of a random lookup. +Note that the database created by the benchmark is quite small. +Therefore the report characterizes the performance of leveldb when the +working set fits in memory. The cost of reading a piece of data that +is not present in the operating system buffer cache will be dominated +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 + +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 + +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) + +## Repository contents + +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/leveldb/*.h. Callers should not include or +rely on the details of any other header files in this package. Those +internal APIs may be changed without warning. + +Guide to header files: + +* **include/leveldb/db.h**: Main interface to the DB: Start here. + +* **include/leveldb/options.h**: Control over the behavior of an entire database, +and also control over the behavior of individual reads and writes. + +* **include/leveldb/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.). + +* **include/leveldb/iterator.h**: Interface for iterating over data. You can get +an iterator from a DB object. + +* **include/leveldb/write_batch.h**: Interface for atomically applying multiple +updates to a database. + +* **include/leveldb/slice.h**: A simple module for maintaining a pointer and a +length into some other byte array. + +* **include/leveldb/status.h**: Status is returned from many of the public interfaces +and is used to report success and various kinds of errors. + +* **include/leveldb/env.h**: +Abstraction of the OS environment. A posix implementation of this interface is +in util/env_posix.cc. + +* **include/leveldb/table.h, include/leveldb/table_builder.h**: Lower-level modules that most +clients probably won't use directly. @@ -0,0 +1,14 @@ +ss +- Stats + +db +- Maybe implement DB::BulkDeleteForRange(start_key, end_key) + that would blow away files whose ranges are entirely contained + within [start_key..end_key]? For Chrome, deletion of obsolete + object stores, etc. can be done in the background anyway, so + probably not that important. +- There have been requests for MultiGet. + +After a range is completely deleted, what gets rid of the +corresponding files if we do no future changes to that range. Make +the conditions for triggering compactions fire in more situations? diff --git a/benchmarks/db_bench.cc b/benchmarks/db_bench.cc new file mode 100644 index 0000000..429a61a --- /dev/null +++ b/benchmarks/db_bench.cc @@ -0,0 +1,1090 @@ +// 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 <sys/types.h> + +#include <atomic> +#include <cstdio> +#include <cstdlib> + +#include "leveldb/cache.h" +#include "leveldb/comparator.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" +#include "util/histogram.h" +#include "util/mutexlock.h" +#include "util/random.h" +#include "util/testutil.h" + +// Comma-separated list of operations to run in the specified order +// Actual benchmarks: +// fillseq -- write N values in sequential key order in async mode +// fillrandom -- write N values in random key order in async mode +// overwrite -- overwrite N values in random key order in async mode +// fillsync -- write N/100 values in random key order in sync mode +// fill100K -- write N/1000 100K values in random order in async mode +// deleteseq -- delete N keys in sequential order +// deleterandom -- delete N keys in random order +// readseq -- read N times sequentially +// readreverse -- read N times in reverse order +// readrandom -- read N times in random order +// readmissing -- read N missing keys in random order +// readhot -- read N times in random order from 1% section of DB +// seekrandom -- N random seeks +// seekordered -- N ordered seeks +// open -- cost of opening a DB +// crc32c -- repeated crc32c of 4K of data +// Meta operations: +// compact -- Compact the entire DB +// stats -- Print DB stats +// sstables -- Print sstable info +// heapprofile -- Dump a heap profile (if supported by this port) +static const char* FLAGS_benchmarks = + "fillseq," + "fillsync," + "fillrandom," + "overwrite," + "readrandom," + "readrandom," // Extra run to allow previous compactions to quiesce + "readseq," + "readreverse," + "compact," + "readrandom," + "readseq," + "readreverse," + "fill100K," + "crc32c," + "snappycomp," + "snappyuncomp,"; + +// Number of key/values to place in database +static int FLAGS_num = 1000000; + +// Number of read operations to do. If negative, do FLAGS_num reads. +static int FLAGS_reads = -1; + +// Number of concurrent threads to run. +static int FLAGS_threads = 1; + +// Size of each value +static int FLAGS_value_size = 100; + +// Arrange to generate values that shrink to this fraction of +// their original size after compression +static double FLAGS_compression_ratio = 0.5; + +// Print histogram of operation timings +static bool FLAGS_histogram = false; + +// Count the number of string comparisons performed +static bool FLAGS_comparisons = false; + +// Number of bytes to buffer in memtable before compacting +// (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; + +// Maximum number of files to keep open at the same time (use default if == 0) +static int FLAGS_open_files = 0; + +// Bloom filter bits per key. +// Negative means use default settings. +static int FLAGS_bloom_bits = -1; + +// Common key prefix length. +static int FLAGS_key_prefix = 0; + +// If true, do not destroy the existing database. If you set this +// flag and also specify a benchmark that wants a fresh database, that +// benchmark will fail. +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 = nullptr; + +namespace leveldb { + +namespace { +leveldb::Env* g_env = nullptr; + +class CountComparator : public Comparator { + public: + CountComparator(const Comparator* wrapped) : wrapped_(wrapped) {} + ~CountComparator() override {} + int Compare(const Slice& a, const Slice& b) const override { + count_.fetch_add(1, std::memory_order_relaxed); + return wrapped_->Compare(a, b); + } + const char* Name() const override { return wrapped_->Name(); } + void FindShortestSeparator(std::string* start, + const Slice& limit) const override { + wrapped_->FindShortestSeparator(start, limit); + } + + void FindShortSuccessor(std::string* key) const override { + return wrapped_->FindShortSuccessor(key); + } + + size_t comparisons() const { return count_.load(std::memory_order_relaxed); } + + void reset() { count_.store(0, std::memory_order_relaxed); } + + private: + mutable std::atomic<size_t> count_{0}; + const Comparator* const wrapped_; +}; + +// Helper for quickly generating random data. +class RandomGenerator { + private: + std::string data_; + int pos_; + + public: + RandomGenerator() { + // We use a limited amount of data over and over again and ensure + // that it is larger than the compression window (32KB), and also + // large enough to serve all typical value sizes we want to write. + Random rnd(301); + std::string piece; + while (data_.size() < 1048576) { + // Add a short fragment that is as compressible as specified + // by FLAGS_compression_ratio. + test::CompressibleString(&rnd, FLAGS_compression_ratio, 100, &piece); + data_.append(piece); + } + pos_ = 0; + } + + Slice Generate(size_t len) { + if (pos_ + len > data_.size()) { + pos_ = 0; + assert(len < data_.size()); + } + pos_ += len; + return Slice(data_.data() + pos_ - len, len); + } +}; + +class KeyBuffer { + public: + KeyBuffer() { + assert(FLAGS_key_prefix < sizeof(buffer_)); + memset(buffer_, 'a', FLAGS_key_prefix); + } + KeyBuffer& operator=(KeyBuffer& other) = delete; + KeyBuffer(KeyBuffer& other) = delete; + + void Set(int k) { + std::snprintf(buffer_ + FLAGS_key_prefix, + sizeof(buffer_) - FLAGS_key_prefix, "%016d", k); + } + + Slice slice() const { return Slice(buffer_, FLAGS_key_prefix + 16); } + + private: + char buffer_[1024]; +}; + +#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])) { + limit--; + } + return Slice(s.data() + start, limit - start); +} +#endif + +static void AppendWithSpace(std::string* str, Slice msg) { + if (msg.empty()) return; + if (!str->empty()) { + str->push_back(' '); + } + str->append(msg.data(), msg.size()); +} + +class Stats { + private: + double start_; + double finish_; + double seconds_; + int done_; + int next_report_; + int64_t bytes_; + double last_op_finish_; + Histogram hist_; + std::string message_; + + public: + Stats() { Start(); } + + void Start() { + next_report_ = 100; + hist_.Clear(); + done_ = 0; + bytes_ = 0; + seconds_ = 0; + message_.clear(); + start_ = finish_ = last_op_finish_ = g_env->NowMicros(); + } + + void Merge(const Stats& other) { + hist_.Merge(other.hist_); + done_ += other.done_; + bytes_ += other.bytes_; + seconds_ += other.seconds_; + if (other.start_ < start_) start_ = other.start_; + if (other.finish_ > finish_) finish_ = other.finish_; + + // Just keep the messages from one thread + if (message_.empty()) message_ = other.message_; + } + + void Stop() { + finish_ = g_env->NowMicros(); + seconds_ = (finish_ - start_) * 1e-6; + } + + void AddMessage(Slice msg) { AppendWithSpace(&message_, msg); } + + void FinishedSingleOp() { + if (FLAGS_histogram) { + double now = g_env->NowMicros(); + double micros = now - last_op_finish_; + hist_.Add(micros); + if (micros > 20000) { + std::fprintf(stderr, "long op: %.1f micros%30s\r", micros, ""); + std::fflush(stderr); + } + last_op_finish_ = now; + } + + 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; + std::fprintf(stderr, "... finished %d ops%30s\r", done_, ""); + std::fflush(stderr); + } + } + + void AddBytes(int64_t n) { bytes_ += n; } + + void Report(const Slice& name) { + // Pretend at least one op was done in case we are running a benchmark + // that does not call FinishedSingleOp(). + if (done_ < 1) done_ = 1; + + std::string extra; + if (bytes_ > 0) { + // Rate is computed on actual elapsed time, not the sum of per-thread + // elapsed times. + double elapsed = (finish_ - start_) * 1e-6; + char rate[100]; + std::snprintf(rate, sizeof(rate), "%6.1f MB/s", + (bytes_ / 1048576.0) / elapsed); + extra = rate; + } + AppendWithSpace(&extra, message_); + + std::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) { + std::fprintf(stdout, "Microseconds per op:\n%s\n", + hist_.ToString().c_str()); + } + std::fflush(stdout); + } +}; + +// State shared by all concurrent executions of the same benchmark. +struct SharedState { + port::Mutex mu; + port::CondVar cv GUARDED_BY(mu); + int total GUARDED_BY(mu); + + // Each thread goes through the following states: + // (1) initializing + // (2) waiting for others to be initialized + // (3) running + // (4) done + + int num_initialized GUARDED_BY(mu); + int num_done GUARDED_BY(mu); + bool start GUARDED_BY(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 + Stats stats; + SharedState* shared; + + ThreadState(int index, int seed) : tid(index), rand(seed), shared(nullptr) {} +}; + +} // namespace + +class Benchmark { + private: + Cache* cache_; + const FilterPolicy* filter_policy_; + DB* db_; + int num_; + int value_size_; + int entries_per_batch_; + WriteOptions write_options_; + int reads_; + int heap_counter_; + CountComparator count_comparator_; + int total_thread_count_; + + void PrintHeader() { + const int kKeySize = 16 + FLAGS_key_prefix; + PrintEnvironment(); + std::fprintf(stdout, "Keys: %d bytes each\n", kKeySize); + std::fprintf( + stdout, "Values: %d bytes each (%d bytes after compression)\n", + FLAGS_value_size, + static_cast<int>(FLAGS_value_size * FLAGS_compression_ratio + 0.5)); + std::fprintf(stdout, "Entries: %d\n", num_); + std::fprintf(stdout, "RawSize: %.1f MB (estimated)\n", + ((static_cast<int64_t>(kKeySize + FLAGS_value_size) * num_) / + 1048576.0)); + std::fprintf( + stdout, "FileSize: %.1f MB (estimated)\n", + (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) / + 1048576.0)); + PrintWarnings(); + std::fprintf(stdout, "------------------------------------------------\n"); + } + + void PrintWarnings() { +#if defined(__GNUC__) && !defined(__OPTIMIZE__) + std::fprintf( + stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); +#endif +#ifndef NDEBUG + std::fprintf( + stdout, + "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); +#endif + + // See if snappy is working by attempting to compress a compressible string + const char text[] = "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"; + std::string compressed; + if (!port::Snappy_Compress(text, sizeof(text), &compressed)) { + std::fprintf(stdout, "WARNING: Snappy compression is not enabled\n"); + } else if (compressed.size() >= sizeof(text)) { + std::fprintf(stdout, "WARNING: Snappy compression is not effective\n"); + } + } + + void PrintEnvironment() { + std::fprintf(stderr, "LevelDB: version %d.%d\n", kMajorVersion, + kMinorVersion); + +#if defined(__linux) + time_t now = time(nullptr); + std::fprintf(stderr, "Date: %s", + ctime(&now)); // ctime() adds newline + + FILE* cpuinfo = std::fopen("/proc/cpuinfo", "r"); + if (cpuinfo != nullptr) { + char line[1000]; + int num_cpus = 0; + std::string cpu_type; + std::string cache_size; + while (fgets(line, sizeof(line), cpuinfo) != nullptr) { + const char* sep = strchr(line, ':'); + if (sep == nullptr) { + continue; + } + Slice key = TrimSpace(Slice(line, sep - 1 - line)); + Slice val = TrimSpace(Slice(sep + 1)); + if (key == "model name") { + ++num_cpus; + cpu_type = val.ToString(); + } else if (key == "cache size") { + cache_size = val.ToString(); + } + } + std::fclose(cpuinfo); + std::fprintf(stderr, "CPU: %d * %s\n", num_cpus, cpu_type.c_str()); + std::fprintf(stderr, "CPUCache: %s\n", cache_size.c_str()); + } +#endif + } + + public: + Benchmark() + : 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), + count_comparator_(BytewiseComparator()), + total_thread_count_(0) { + std::vector<std::string> files; + g_env->GetChildren(FLAGS_db, &files); + for (size_t i = 0; i < files.size(); i++) { + if (Slice(files[i]).starts_with("heap-")) { + g_env->RemoveFile(std::string(FLAGS_db) + "/" + files[i]); + } + } + if (!FLAGS_use_existing_db) { + DestroyDB(FLAGS_db, Options()); + } + } + + ~Benchmark() { + delete db_; + delete cache_; + delete filter_policy_; + } + + void Run() { + PrintHeader(); + Open(); + + const char* benchmarks = FLAGS_benchmarks; + while (benchmarks != nullptr) { + const char* sep = strchr(benchmarks, ','); + Slice name; + if (sep == nullptr) { + name = benchmarks; + benchmarks = nullptr; + } else { + name = Slice(benchmarks, sep - benchmarks); + benchmarks = sep + 1; + } + + // Reset parameters that may be overridden below + num_ = FLAGS_num; + reads_ = (FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads); + value_size_ = FLAGS_value_size; + entries_per_batch_ = 1; + write_options_ = WriteOptions(); + + void (Benchmark::*method)(ThreadState*) = nullptr; + bool fresh_db = false; + int num_threads = FLAGS_threads; + + if (name == Slice("open")) { + method = &Benchmark::OpenBench; + num_ /= 10000; + if (num_ < 1) num_ = 1; + } else if (name == Slice("fillseq")) { + fresh_db = true; + method = &Benchmark::WriteSeq; + } else if (name == Slice("fillbatch")) { + fresh_db = true; + entries_per_batch_ = 1000; + method = &Benchmark::WriteSeq; + } else if (name == Slice("fillrandom")) { + fresh_db = true; + method = &Benchmark::WriteRandom; + } else if (name == Slice("overwrite")) { + fresh_db = false; + method = &Benchmark::WriteRandom; + } else if (name == Slice("fillsync")) { + fresh_db = true; + num_ /= 1000; + write_options_.sync = true; + method = &Benchmark::WriteRandom; + } else if (name == Slice("fill100K")) { + fresh_db = true; + num_ /= 1000; + value_size_ = 100 * 1000; + method = &Benchmark::WriteRandom; + } else if (name == Slice("readseq")) { + method = &Benchmark::ReadSequential; + } else if (name == Slice("readreverse")) { + method = &Benchmark::ReadReverse; + } else if (name == Slice("readrandom")) { + method = &Benchmark::ReadRandom; + } else if (name == Slice("readmissing")) { + method = &Benchmark::ReadMissing; + } else if (name == Slice("seekrandom")) { + method = &Benchmark::SeekRandom; + } else if (name == Slice("seekordered")) { + method = &Benchmark::SeekOrdered; + } else if (name == Slice("readhot")) { + method = &Benchmark::ReadHot; + } else if (name == Slice("readrandomsmall")) { + reads_ /= 1000; + method = &Benchmark::ReadRandom; + } else if (name == Slice("deleteseq")) { + method = &Benchmark::DeleteSeq; + } else if (name == Slice("deleterandom")) { + method = &Benchmark::DeleteRandom; + } else if (name == Slice("readwhilewriting")) { + num_threads++; // Add extra thread for writing + method = &Benchmark::ReadWhileWriting; + } else if (name == Slice("compact")) { + method = &Benchmark::Compact; + } else if (name == Slice("crc32c")) { + method = &Benchmark::Crc32c; + } else if (name == Slice("snappycomp")) { + method = &Benchmark::SnappyCompress; + } else if (name == Slice("snappyuncomp")) { + method = &Benchmark::SnappyUncompress; + } else if (name == Slice("heapprofile")) { + HeapProfile(); + } else if (name == Slice("stats")) { + PrintStats("leveldb.stats"); + } else if (name == Slice("sstables")) { + PrintStats("leveldb.sstables"); + } else { + if (!name.empty()) { // No error message for empty name + std::fprintf(stderr, "unknown benchmark '%s'\n", + name.ToString().c_str()); + } + } + + if (fresh_db) { + if (FLAGS_use_existing_db) { + std::fprintf(stdout, "%-12s : skipped (--use_existing_db is true)\n", + name.ToString().c_str()); + method = nullptr; + } else { + delete db_; + db_ = nullptr; + DestroyDB(FLAGS_db, Options()); + Open(); + } + } + + if (method != nullptr) { + RunBenchmark(num_threads, name, method); + } + } + } + + private: + struct ThreadArg { + Benchmark* bm; + SharedState* shared; + ThreadState* thread; + void (Benchmark::*method)(ThreadState*); + }; + + static void ThreadBody(void* v) { + ThreadArg* arg = reinterpret_cast<ThreadArg*>(v); + SharedState* shared = arg->shared; + ThreadState* thread = arg->thread; + { + MutexLock l(&shared->mu); + shared->num_initialized++; + if (shared->num_initialized >= shared->total) { + shared->cv.SignalAll(); + } + while (!shared->start) { + shared->cv.Wait(); + } + } + + thread->stats.Start(); + (arg->bm->*(arg->method))(thread); + thread->stats.Stop(); + + { + MutexLock l(&shared->mu); + shared->num_done++; + if (shared->num_done >= shared->total) { + shared->cv.SignalAll(); + } + } + } + + void RunBenchmark(int n, Slice name, + void (Benchmark::*method)(ThreadState*)) { + SharedState shared(n); + + ThreadArg* arg = new ThreadArg[n]; + for (int i = 0; i < n; i++) { + arg[i].bm = this; + arg[i].method = method; + arg[i].shared = &shared; + ++total_thread_count_; + // Seed the thread's random state deterministically based upon thread + // creation across all benchmarks. This ensures that the seeds are unique + // but reproducible when rerunning the same set of benchmarks. + arg[i].thread = new ThreadState(i, /*seed=*/1000 + total_thread_count_); + arg[i].thread->shared = &shared; + g_env->StartThread(ThreadBody, &arg[i]); + } + + shared.mu.Lock(); + while (shared.num_initialized < n) { + shared.cv.Wait(); + } + + shared.start = true; + shared.cv.SignalAll(); + while (shared.num_done < n) { + shared.cv.Wait(); + } + shared.mu.Unlock(); + + for (int i = 1; i < n; i++) { + arg[0].thread->stats.Merge(arg[i].thread->stats); + } + arg[0].thread->stats.Report(name); + if (FLAGS_comparisons) { + fprintf(stdout, "Comparisons: %zu\n", count_comparator_.comparisons()); + count_comparator_.reset(); + fflush(stdout); + } + + for (int i = 0; i < n; i++) { + delete arg[i].thread; + } + delete[] arg; + } + + void Crc32c(ThreadState* thread) { + // Checksum about 500MB of data total + const int size = 4096; + const char* label = "(4K per op)"; + std::string data(size, 'x'); + int64_t bytes = 0; + uint32_t crc = 0; + while (bytes < 500 * 1048576) { + crc = crc32c::Value(data.data(), size); + thread->stats.FinishedSingleOp(); + bytes += size; + } + // Print so result is not dead + std::fprintf(stderr, "... crc=0x%x\r", static_cast<unsigned int>(crc)); + + thread->stats.AddBytes(bytes); + thread->stats.AddMessage(label); + } + + void SnappyCompress(ThreadState* thread) { + RandomGenerator gen; + Slice input = gen.Generate(Options().block_size); + int64_t bytes = 0; + int64_t produced = 0; + bool ok = true; + std::string compressed; + while (ok && bytes < 1024 * 1048576) { // Compress 1G + ok = port::Snappy_Compress(input.data(), input.size(), &compressed); + produced += compressed.size(); + bytes += input.size(); + thread->stats.FinishedSingleOp(); + } + + if (!ok) { + thread->stats.AddMessage("(snappy failure)"); + } else { + char buf[100]; + std::snprintf(buf, sizeof(buf), "(output: %.1f%%)", + (produced * 100.0) / bytes); + thread->stats.AddMessage(buf); + thread->stats.AddBytes(bytes); + } + } + + void SnappyUncompress(ThreadState* thread) { + RandomGenerator gen; + Slice input = gen.Generate(Options().block_size); + std::string compressed; + bool ok = port::Snappy_Compress(input.data(), input.size(), &compressed); + 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); + bytes += input.size(); + thread->stats.FinishedSingleOp(); + } + delete[] uncompressed; + + if (!ok) { + thread->stats.AddMessage("(snappy failure)"); + } else { + thread->stats.AddBytes(bytes); + } + } + + void Open() { + 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; + if (FLAGS_comparisons) { + options.comparator = &count_comparator_; + } + 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()) { + std::fprintf(stderr, "open error: %s\n", s.ToString().c_str()); + std::exit(1); + } + } + + void OpenBench(ThreadState* thread) { + for (int i = 0; i < num_; i++) { + delete db_; + Open(); + thread->stats.FinishedSingleOp(); + } + } + + void WriteSeq(ThreadState* thread) { DoWrite(thread, true); } + + void WriteRandom(ThreadState* thread) { DoWrite(thread, false); } + + void DoWrite(ThreadState* thread, bool seq) { + if (num_ != FLAGS_num) { + char msg[100]; + std::snprintf(msg, sizeof(msg), "(%d ops)", num_); + thread->stats.AddMessage(msg); + } + + RandomGenerator gen; + WriteBatch batch; + Status s; + int64_t bytes = 0; + KeyBuffer key; + 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.Uniform(FLAGS_num); + key.Set(k); + batch.Put(key.slice(), gen.Generate(value_size_)); + bytes += value_size_ + key.slice().size(); + thread->stats.FinishedSingleOp(); + } + s = db_->Write(write_options_, &batch); + if (!s.ok()) { + std::fprintf(stderr, "put error: %s\n", s.ToString().c_str()); + std::exit(1); + } + } + thread->stats.AddBytes(bytes); + } + + void ReadSequential(ThreadState* thread) { + Iterator* iter = db_->NewIterator(ReadOptions()); + int i = 0; + int64_t bytes = 0; + for (iter->SeekToFirst(); i < reads_ && iter->Valid(); iter->Next()) { + bytes += iter->key().size() + iter->value().size(); + thread->stats.FinishedSingleOp(); + ++i; + } + delete iter; + thread->stats.AddBytes(bytes); + } + + void ReadReverse(ThreadState* thread) { + Iterator* iter = db_->NewIterator(ReadOptions()); + int i = 0; + int64_t bytes = 0; + for (iter->SeekToLast(); i < reads_ && iter->Valid(); iter->Prev()) { + bytes += iter->key().size() + iter->value().size(); + thread->stats.FinishedSingleOp(); + ++i; + } + delete iter; + thread->stats.AddBytes(bytes); + } + + void ReadRandom(ThreadState* thread) { + ReadOptions options; + std::string value; + int found = 0; + KeyBuffer key; + for (int i = 0; i < reads_; i++) { + const int k = thread->rand.Uniform(FLAGS_num); + key.Set(k); + if (db_->Get(options, key.slice(), &value).ok()) { + found++; + } + thread->stats.FinishedSingleOp(); + } + char msg[100]; + std::snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_); + thread->stats.AddMessage(msg); + } + + void ReadMissing(ThreadState* thread) { + ReadOptions options; + std::string value; + KeyBuffer key; + for (int i = 0; i < reads_; i++) { + const int k = thread->rand.Uniform(FLAGS_num); + key.Set(k); + Slice s = Slice(key.slice().data(), key.slice().size() - 1); + db_->Get(options, s, &value); + thread->stats.FinishedSingleOp(); + } + } + + void ReadHot(ThreadState* thread) { + ReadOptions options; + std::string value; + const int range = (FLAGS_num + 99) / 100; + KeyBuffer key; + for (int i = 0; i < reads_; i++) { + const int k = thread->rand.Uniform(range); + key.Set(k); + db_->Get(options, key.slice(), &value); + thread->stats.FinishedSingleOp(); + } + } + + void SeekRandom(ThreadState* thread) { + ReadOptions options; + int found = 0; + KeyBuffer key; + for (int i = 0; i < reads_; i++) { + Iterator* iter = db_->NewIterator(options); + const int k = thread->rand.Uniform(FLAGS_num); + key.Set(k); + iter->Seek(key.slice()); + if (iter->Valid() && iter->key() == key.slice()) found++; + delete iter; + thread->stats.FinishedSingleOp(); + } + char msg[100]; + snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_); + thread->stats.AddMessage(msg); + } + + void SeekOrdered(ThreadState* thread) { + ReadOptions options; + Iterator* iter = db_->NewIterator(options); + int found = 0; + int k = 0; + KeyBuffer key; + for (int i = 0; i < reads_; i++) { + k = (k + (thread->rand.Uniform(100))) % FLAGS_num; + key.Set(k); + iter->Seek(key.slice()); + if (iter->Valid() && iter->key() == key.slice()) found++; + thread->stats.FinishedSingleOp(); + } + delete iter; + char msg[100]; + std::snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_); + thread->stats.AddMessage(msg); + } + + void DoDelete(ThreadState* thread, bool seq) { + RandomGenerator gen; + WriteBatch batch; + Status s; + KeyBuffer key; + 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.Uniform(FLAGS_num)); + key.Set(k); + batch.Delete(key.slice()); + thread->stats.FinishedSingleOp(); + } + s = db_->Write(write_options_, &batch); + if (!s.ok()) { + std::fprintf(stderr, "del error: %s\n", s.ToString().c_str()); + std::exit(1); + } + } + } + + void DeleteSeq(ThreadState* thread) { DoDelete(thread, true); } + + void DeleteRandom(ThreadState* thread) { DoDelete(thread, false); } + + void ReadWhileWriting(ThreadState* thread) { + if (thread->tid > 0) { + ReadRandom(thread); + } else { + // Special thread that keeps writing until other threads are done. + RandomGenerator gen; + KeyBuffer key; + while (true) { + { + MutexLock l(&thread->shared->mu); + if (thread->shared->num_done + 1 >= thread->shared->num_initialized) { + // Other threads have finished + break; + } + } + + const int k = thread->rand.Uniform(FLAGS_num); + key.Set(k); + Status s = + db_->Put(write_options_, key.slice(), gen.Generate(value_size_)); + if (!s.ok()) { + std::fprintf(stderr, "put error: %s\n", s.ToString().c_str()); + std::exit(1); + } + } + + // Do not count any of the preceding work/delay in stats. + thread->stats.Start(); + } + } + + void Compact(ThreadState* thread) { db_->CompactRange(nullptr, nullptr); } + + void PrintStats(const char* key) { + std::string stats; + if (!db_->GetProperty(key, &stats)) { + stats = "(failed)"; + } + std::fprintf(stdout, "\n%s\n", stats.c_str()); + } + + static void WriteToFile(void* arg, const char* buf, int n) { + reinterpret_cast<WritableFile*>(arg)->Append(Slice(buf, n)); + } + + void HeapProfile() { + char fname[100]; + std::snprintf(fname, sizeof(fname), "%s/heap-%04d", FLAGS_db, + ++heap_counter_); + WritableFile* file; + Status s = g_env->NewWritableFile(fname, &file); + if (!s.ok()) { + std::fprintf(stderr, "%s\n", s.ToString().c_str()); + return; + } + bool ok = port::GetHeapProfile(WriteToFile, file); + delete file; + if (!ok) { + std::fprintf(stderr, "heap profiling not supported\n"); + g_env->RemoveFile(fname); + } + } +}; + +} // namespace leveldb + +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; + + for (int i = 1; i < argc; i++) { + double d; + int n; + char junk; + if (leveldb::Slice(argv[i]).starts_with("--benchmarks=")) { + FLAGS_benchmarks = argv[i] + strlen("--benchmarks="); + } else if (sscanf(argv[i], "--compression_ratio=%lf%c", &d, &junk) == 1) { + FLAGS_compression_ratio = d; + } else if (sscanf(argv[i], "--histogram=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_histogram = n; + } else if (sscanf(argv[i], "--comparisons=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_comparisons = n; + } 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) { + FLAGS_reads = n; + } else if (sscanf(argv[i], "--threads=%d%c", &n, &junk) == 1) { + FLAGS_threads = n; + } else if (sscanf(argv[i], "--value_size=%d%c", &n, &junk) == 1) { + 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], "--key_prefix=%d%c", &n, &junk) == 1) { + FLAGS_key_prefix = 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) { + FLAGS_bloom_bits = n; + } else if (sscanf(argv[i], "--open_files=%d%c", &n, &junk) == 1) { + FLAGS_open_files = n; + } else if (strncmp(argv[i], "--db=", 5) == 0) { + FLAGS_db = argv[i] + 5; + } else { + std::fprintf(stderr, "Invalid flag '%s'\n", argv[i]); + std::exit(1); + } + } + + leveldb::g_env = leveldb::Env::Default(); + + // Choose a location for the test database if none given with --db=<path> + 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; + benchmark.Run(); + return 0; +} diff --git a/benchmarks/db_bench_sqlite3.cc b/benchmarks/db_bench_sqlite3.cc new file mode 100644 index 0000000..c9be652 --- /dev/null +++ b/benchmarks/db_bench_sqlite3.cc @@ -0,0 +1,726 @@ +// 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 <sqlite3.h> + +#include <cstdio> +#include <cstdlib> + +#include "util/histogram.h" +#include "util/random.h" +#include "util/testutil.h" + +// Comma-separated list of operations to run in the specified order +// Actual benchmarks: +// +// fillseq -- write N values in sequential key order in async mode +// fillseqsync -- write N/100 values in sequential key order in sync mode +// fillseqbatch -- batch write N values in sequential key order in async mode +// fillrandom -- write N values in random key order in async mode +// fillrandsync -- write N/100 values in random key order in sync mode +// fillrandbatch -- batch write N values in sequential key order in async mode +// overwrite -- overwrite N values in random key order in async mode +// fillrand100K -- write N/1000 100K values in random order in async mode +// fillseq100K -- write N/1000 100K values in sequential order in async mode +// readseq -- read N times sequentially +// readrandom -- read N times in random order +// readrand100K -- read N/1000 100K values in sequential order in async mode +static const char* FLAGS_benchmarks = + "fillseq," + "fillseqsync," + "fillseqbatch," + "fillrandom," + "fillrandsync," + "fillrandbatch," + "overwrite," + "overwritebatch," + "readrandom," + "readseq," + "fillrand100K," + "fillseq100K," + "readseq," + "readrand100K,"; + +// Number of key/values to place in database +static int FLAGS_num = 1000000; + +// Number of read operations to do. If negative, do FLAGS_num reads. +static int FLAGS_reads = -1; + +// Size of each value +static int FLAGS_value_size = 100; + +// Print histogram of operation timings +static bool FLAGS_histogram = false; + +// Arrange to generate values that shrink to this fraction of +// their original size after compression +static double FLAGS_compression_ratio = 0.5; + +// Page size. Default 1 KB. +static int FLAGS_page_size = 1024; + +// Number of pages. +// Default cache size = FLAGS_page_size * FLAGS_num_pages = 4 MB. +static int FLAGS_num_pages = 4096; + +// If true, do not destroy the existing database. If you set this +// flag and also specify a benchmark that wants a fresh database, that +// benchmark will fail. +static bool FLAGS_use_existing_db = false; + +// If true, the SQLite table has ROWIDs. +static bool FLAGS_use_rowids = false; + +// If true, we allow batch writes to occur +static bool FLAGS_transaction = true; + +// If true, we enable Write-Ahead Logging +static bool FLAGS_WAL_enabled = true; + +// Use the db with the following name. +static const char* FLAGS_db = nullptr; + +inline static void ExecErrorCheck(int status, char* err_msg) { + if (status != SQLITE_OK) { + std::fprintf(stderr, "SQL error: %s\n", err_msg); + sqlite3_free(err_msg); + std::exit(1); + } +} + +inline static void StepErrorCheck(int status) { + if (status != SQLITE_DONE) { + std::fprintf(stderr, "SQL step error: status = %d\n", status); + std::exit(1); + } +} + +inline static void ErrorCheck(int status) { + if (status != SQLITE_OK) { + std::fprintf(stderr, "sqlite3 error: status = %d\n", status); + std::exit(1); + } +} + +inline static void WalCheckpoint(sqlite3* db_) { + // Flush all writes to disk + if (FLAGS_WAL_enabled) { + sqlite3_wal_checkpoint_v2(db_, nullptr, SQLITE_CHECKPOINT_FULL, nullptr, + nullptr); + } +} + +namespace leveldb { + +// Helper for quickly generating random data. +namespace { +class RandomGenerator { + private: + std::string data_; + int pos_; + + public: + RandomGenerator() { + // We use a limited amount of data over and over again and ensure + // that it is larger than the compression window (32KB), and also + // large enough to serve all typical value sizes we want to write. + Random rnd(301); + std::string piece; + while (data_.size() < 1048576) { + // Add a short fragment that is as compressible as specified + // by FLAGS_compression_ratio. + test::CompressibleString(&rnd, FLAGS_compression_ratio, 100, &piece); + data_.append(piece); + } + pos_ = 0; + } + + Slice Generate(int len) { + if (pos_ + len > data_.size()) { + pos_ = 0; + assert(len < data_.size()); + } + pos_ += len; + return Slice(data_.data() + pos_ - len, len); + } +}; + +static Slice TrimSpace(Slice s) { + int start = 0; + while (start < s.size() && isspace(s[start])) { + start++; + } + int limit = s.size(); + while (limit > start && isspace(s[limit - 1])) { + limit--; + } + return Slice(s.data() + start, limit - start); +} + +} // namespace + +class Benchmark { + private: + sqlite3* db_; + int db_num_; + int num_; + int reads_; + double start_; + double last_op_finish_; + int64_t bytes_; + std::string message_; + Histogram hist_; + RandomGenerator gen_; + Random rand_; + + // State kept for progress messages + int done_; + int next_report_; // When to report next + + void PrintHeader() { + const int kKeySize = 16; + PrintEnvironment(); + std::fprintf(stdout, "Keys: %d bytes each\n", kKeySize); + std::fprintf(stdout, "Values: %d bytes each\n", FLAGS_value_size); + std::fprintf(stdout, "Entries: %d\n", num_); + std::fprintf(stdout, "RawSize: %.1f MB (estimated)\n", + ((static_cast<int64_t>(kKeySize + FLAGS_value_size) * num_) / + 1048576.0)); + PrintWarnings(); + std::fprintf(stdout, "------------------------------------------------\n"); + } + + void PrintWarnings() { +#if defined(__GNUC__) && !defined(__OPTIMIZE__) + std::fprintf( + stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); +#endif +#ifndef NDEBUG + std::fprintf( + stdout, + "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); +#endif + } + + void PrintEnvironment() { + std::fprintf(stderr, "SQLite: version %s\n", SQLITE_VERSION); + +#if defined(__linux) + time_t now = time(nullptr); + std::fprintf(stderr, "Date: %s", + ctime(&now)); // ctime() adds newline + + FILE* cpuinfo = std::fopen("/proc/cpuinfo", "r"); + if (cpuinfo != nullptr) { + char line[1000]; + int num_cpus = 0; + std::string cpu_type; + std::string cache_size; + while (fgets(line, sizeof(line), cpuinfo) != nullptr) { + const char* sep = strchr(line, ':'); + if (sep == nullptr) { + continue; + } + Slice key = TrimSpace(Slice(line, sep - 1 - line)); + Slice val = TrimSpace(Slice(sep + 1)); + if (key == "model name") { + ++num_cpus; + cpu_type = val.ToString(); + } else if (key == "cache size") { + cache_size = val.ToString(); + } + } + std::fclose(cpuinfo); + std::fprintf(stderr, "CPU: %d * %s\n", num_cpus, cpu_type.c_str()); + std::fprintf(stderr, "CPUCache: %s\n", cache_size.c_str()); + } +#endif + } + + void Start() { + start_ = Env::Default()->NowMicros() * 1e-6; + bytes_ = 0; + message_.clear(); + last_op_finish_ = start_; + hist_.Clear(); + done_ = 0; + next_report_ = 100; + } + + void FinishedSingleOp() { + if (FLAGS_histogram) { + double now = Env::Default()->NowMicros() * 1e-6; + double micros = (now - last_op_finish_) * 1e6; + hist_.Add(micros); + if (micros > 20000) { + std::fprintf(stderr, "long op: %.1f micros%30s\r", micros, ""); + std::fflush(stderr); + } + last_op_finish_ = now; + } + + 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; + std::fprintf(stderr, "... finished %d ops%30s\r", done_, ""); + std::fflush(stderr); + } + } + + void Stop(const Slice& name) { + double finish = Env::Default()->NowMicros() * 1e-6; + + // Pretend at least one op was done in case we are running a benchmark + // that does not call FinishedSingleOp(). + if (done_ < 1) done_ = 1; + + if (bytes_ > 0) { + char rate[100]; + std::snprintf(rate, sizeof(rate), "%6.1f MB/s", + (bytes_ / 1048576.0) / (finish - start_)); + if (!message_.empty()) { + message_ = std::string(rate) + " " + message_; + } else { + message_ = rate; + } + } + + std::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) { + std::fprintf(stdout, "Microseconds per op:\n%s\n", + hist_.ToString().c_str()); + } + std::fflush(stdout); + } + + public: + enum Order { SEQUENTIAL, RANDOM }; + enum DBState { FRESH, EXISTING }; + + Benchmark() + : db_(nullptr), + db_num_(0), + num_(FLAGS_num), + reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), + bytes_(0), + rand_(301) { + std::vector<std::string> files; + std::string test_dir; + Env::Default()->GetTestDirectory(&test_dir); + Env::Default()->GetChildren(test_dir, &files); + if (!FLAGS_use_existing_db) { + for (int i = 0; i < files.size(); i++) { + if (Slice(files[i]).starts_with("dbbench_sqlite3")) { + std::string file_name(test_dir); + file_name += "/"; + file_name += files[i]; + Env::Default()->RemoveFile(file_name.c_str()); + } + } + } + } + + ~Benchmark() { + int status = sqlite3_close(db_); + ErrorCheck(status); + } + + void Run() { + PrintHeader(); + Open(); + + const char* benchmarks = FLAGS_benchmarks; + while (benchmarks != nullptr) { + const char* sep = strchr(benchmarks, ','); + Slice name; + if (sep == nullptr) { + name = benchmarks; + benchmarks = nullptr; + } else { + name = Slice(benchmarks, sep - benchmarks); + benchmarks = sep + 1; + } + + bytes_ = 0; + Start(); + + bool known = true; + bool write_sync = false; + if (name == Slice("fillseq")) { + Write(write_sync, SEQUENTIAL, FRESH, num_, FLAGS_value_size, 1); + WalCheckpoint(db_); + } else if (name == Slice("fillseqbatch")) { + Write(write_sync, SEQUENTIAL, FRESH, num_, FLAGS_value_size, 1000); + WalCheckpoint(db_); + } else if (name == Slice("fillrandom")) { + Write(write_sync, RANDOM, FRESH, num_, FLAGS_value_size, 1); + WalCheckpoint(db_); + } else if (name == Slice("fillrandbatch")) { + Write(write_sync, RANDOM, FRESH, num_, FLAGS_value_size, 1000); + WalCheckpoint(db_); + } else if (name == Slice("overwrite")) { + Write(write_sync, RANDOM, EXISTING, num_, FLAGS_value_size, 1); + WalCheckpoint(db_); + } else if (name == Slice("overwritebatch")) { + Write(write_sync, RANDOM, EXISTING, num_, FLAGS_value_size, 1000); + WalCheckpoint(db_); + } else if (name == Slice("fillrandsync")) { + write_sync = true; + Write(write_sync, RANDOM, FRESH, num_ / 100, FLAGS_value_size, 1); + WalCheckpoint(db_); + } else if (name == Slice("fillseqsync")) { + write_sync = true; + Write(write_sync, SEQUENTIAL, FRESH, num_ / 100, FLAGS_value_size, 1); + WalCheckpoint(db_); + } else if (name == Slice("fillrand100K")) { + Write(write_sync, RANDOM, FRESH, num_ / 1000, 100 * 1000, 1); + WalCheckpoint(db_); + } else if (name == Slice("fillseq100K")) { + Write(write_sync, SEQUENTIAL, FRESH, num_ / 1000, 100 * 1000, 1); + WalCheckpoint(db_); + } else if (name == Slice("readseq")) { + ReadSequential(); + } else if (name == Slice("readrandom")) { + Read(RANDOM, 1); + } else if (name == Slice("readrand100K")) { + int n = reads_; + reads_ /= 1000; + Read(RANDOM, 1); + reads_ = n; + } else { + known = false; + if (name != Slice()) { // No error message for empty name + std::fprintf(stderr, "unknown benchmark '%s'\n", + name.ToString().c_str()); + } + } + if (known) { + Stop(name); + } + } + } + + void Open() { + assert(db_ == nullptr); + + int status; + char file_name[100]; + char* err_msg = nullptr; + db_num_++; + + // Open database + std::string tmp_dir; + Env::Default()->GetTestDirectory(&tmp_dir); + std::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) { + std::fprintf(stderr, "open error: %s\n", sqlite3_errmsg(db_)); + std::exit(1); + } + + // Change SQLite cache size + char cache_size[100]; + std::snprintf(cache_size, sizeof(cache_size), "PRAGMA cache_size = %d", + FLAGS_num_pages); + status = sqlite3_exec(db_, cache_size, nullptr, nullptr, &err_msg); + ExecErrorCheck(status, err_msg); + + // FLAGS_page_size is defaulted to 1024 + if (FLAGS_page_size != 1024) { + char page_size[100]; + std::snprintf(page_size, sizeof(page_size), "PRAGMA page_size = %d", + FLAGS_page_size); + status = sqlite3_exec(db_, page_size, nullptr, nullptr, &err_msg); + ExecErrorCheck(status, err_msg); + } + + // Change journal mode to WAL if WAL enabled flag is on + if (FLAGS_WAL_enabled) { + std::string WAL_stmt = "PRAGMA journal_mode = WAL"; + + // 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(), nullptr, nullptr, &err_msg); + ExecErrorCheck(status, 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))"; + if (!FLAGS_use_rowids) create_stmt += " WITHOUT ROWID"; + 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(), 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) { + // Create new database if state == FRESH + if (state == FRESH) { + if (FLAGS_use_existing_db) { + message_ = "skipping (--use_existing_db is true)"; + return; + } + sqlite3_close(db_); + db_ = nullptr; + Open(); + Start(); + } + + if (num_entries != num_) { + char msg[100]; + std::snprintf(msg, sizeof(msg), "(%d ops)", num_entries); + message_ = msg; + } + + char* err_msg = nullptr; + int status; + + sqlite3_stmt *replace_stmt, *begin_trans_stmt, *end_trans_stmt; + std::string replace_str = "REPLACE INTO test (key, value) VALUES (?, ?)"; + std::string begin_trans_str = "BEGIN TRANSACTION;"; + 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(), nullptr, nullptr, &err_msg); + ExecErrorCheck(status, err_msg); + + // Preparing sqlite3 statements + 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, nullptr); + ErrorCheck(status); + status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1, &end_trans_stmt, + nullptr); + ErrorCheck(status); + + bool transaction = (entries_per_batch > 1); + for (int i = 0; i < num_entries; i += entries_per_batch) { + // Begin write transaction + if (FLAGS_transaction && transaction) { + status = sqlite3_step(begin_trans_stmt); + StepErrorCheck(status); + status = sqlite3_reset(begin_trans_stmt); + ErrorCheck(status); + } + + // Create and execute SQL statements + for (int j = 0; j < entries_per_batch; j++) { + 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); + char key[100]; + std::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); + ErrorCheck(status); + + // Execute replace_stmt + bytes_ += value_size + strlen(key); + status = sqlite3_step(replace_stmt); + StepErrorCheck(status); + + // Reset SQLite statement for another use + status = sqlite3_clear_bindings(replace_stmt); + ErrorCheck(status); + status = sqlite3_reset(replace_stmt); + ErrorCheck(status); + + FinishedSingleOp(); + } + + // End write transaction + if (FLAGS_transaction && transaction) { + status = sqlite3_step(end_trans_stmt); + StepErrorCheck(status); + status = sqlite3_reset(end_trans_stmt); + ErrorCheck(status); + } + } + + status = sqlite3_finalize(replace_stmt); + ErrorCheck(status); + status = sqlite3_finalize(begin_trans_stmt); + ErrorCheck(status); + status = sqlite3_finalize(end_trans_stmt); + ErrorCheck(status); + } + + void Read(Order order, int entries_per_batch) { + int status; + sqlite3_stmt *read_stmt, *begin_trans_stmt, *end_trans_stmt; + + std::string read_str = "SELECT * FROM test WHERE key = ?"; + std::string begin_trans_str = "BEGIN TRANSACTION;"; + std::string end_trans_str = "END TRANSACTION;"; + + // Preparing sqlite3 statements + status = sqlite3_prepare_v2(db_, begin_trans_str.c_str(), -1, + &begin_trans_stmt, nullptr); + ErrorCheck(status); + 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, nullptr); + ErrorCheck(status); + + bool transaction = (entries_per_batch > 1); + for (int i = 0; i < reads_; i += entries_per_batch) { + // Begin read transaction + if (FLAGS_transaction && transaction) { + status = sqlite3_step(begin_trans_stmt); + StepErrorCheck(status); + status = sqlite3_reset(begin_trans_stmt); + ErrorCheck(status); + } + + // Create and execute SQL statements + for (int j = 0; j < entries_per_batch; j++) { + // Create key value + char key[100]; + int k = (order == SEQUENTIAL) ? i + j : (rand_.Next() % reads_); + std::snprintf(key, sizeof(key), "%016d", k); + + // Bind key value into read_stmt + status = sqlite3_bind_blob(read_stmt, 1, key, 16, SQLITE_STATIC); + ErrorCheck(status); + + // Execute read statement + while ((status = sqlite3_step(read_stmt)) == SQLITE_ROW) { + } + StepErrorCheck(status); + + // Reset SQLite statement for another use + status = sqlite3_clear_bindings(read_stmt); + ErrorCheck(status); + status = sqlite3_reset(read_stmt); + ErrorCheck(status); + FinishedSingleOp(); + } + + // End read transaction + if (FLAGS_transaction && transaction) { + status = sqlite3_step(end_trans_stmt); + StepErrorCheck(status); + status = sqlite3_reset(end_trans_stmt); + ErrorCheck(status); + } + } + + status = sqlite3_finalize(read_stmt); + ErrorCheck(status); + status = sqlite3_finalize(begin_trans_stmt); + ErrorCheck(status); + status = sqlite3_finalize(end_trans_stmt); + ErrorCheck(status); + } + + void ReadSequential() { + int status; + sqlite3_stmt* pStmt; + std::string read_str = "SELECT * FROM test ORDER BY key"; + + 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); + FinishedSingleOp(); + } + + status = sqlite3_finalize(pStmt); + ErrorCheck(status); + } +}; + +} // namespace leveldb + +int main(int argc, char** argv) { + std::string default_db_path; + for (int i = 1; i < argc; i++) { + double d; + int n; + char junk; + if (leveldb::Slice(argv[i]).starts_with("--benchmarks=")) { + FLAGS_benchmarks = argv[i] + strlen("--benchmarks="); + } else if (sscanf(argv[i], "--histogram=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_histogram = n; + } else if (sscanf(argv[i], "--compression_ratio=%lf%c", &d, &junk) == 1) { + FLAGS_compression_ratio = d; + } 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], "--use_rowids=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_use_rowids = 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) { + FLAGS_reads = n; + } else if (sscanf(argv[i], "--value_size=%d%c", &n, &junk) == 1) { + FLAGS_value_size = n; + } else if (leveldb::Slice(argv[i]) == leveldb::Slice("--no_transaction")) { + FLAGS_transaction = false; + } else if (sscanf(argv[i], "--page_size=%d%c", &n, &junk) == 1) { + FLAGS_page_size = n; + } else if (sscanf(argv[i], "--num_pages=%d%c", &n, &junk) == 1) { + FLAGS_num_pages = n; + } else if (sscanf(argv[i], "--WAL_enabled=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_WAL_enabled = n; + } else if (strncmp(argv[i], "--db=", 5) == 0) { + FLAGS_db = argv[i] + 5; + } else { + std::fprintf(stderr, "Invalid flag '%s'\n", argv[i]); + std::exit(1); + } + } + + // Choose a location for the test database if none given with --db=<path> + 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; + benchmark.Run(); + return 0; +} diff --git a/benchmarks/db_bench_tree_db.cc b/benchmarks/db_bench_tree_db.cc new file mode 100644 index 0000000..533600b --- /dev/null +++ b/benchmarks/db_bench_tree_db.cc @@ -0,0 +1,531 @@ +// 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 <kcpolydb.h> + +#include <cstdio> +#include <cstdlib> + +#include "util/histogram.h" +#include "util/random.h" +#include "util/testutil.h" + +// Comma-separated list of operations to run in the specified order +// Actual benchmarks: +// +// fillseq -- write N values in sequential key order in async mode +// fillrandom -- write N values in random key order in async mode +// overwrite -- overwrite N values in random key order in async mode +// fillseqsync -- write N/100 values in sequential key order in sync mode +// fillrandsync -- write N/100 values in random key order in sync mode +// fillrand100K -- write N/1000 100K values in random order in async mode +// fillseq100K -- write N/1000 100K values in seq order in async mode +// readseq -- read N times sequentially +// readseq100K -- read N/1000 100K values in sequential order in async mode +// readrand100K -- read N/1000 100K values in sequential order in async mode +// readrandom -- read N times in random order +static const char* FLAGS_benchmarks = + "fillseq," + "fillseqsync," + "fillrandsync," + "fillrandom," + "overwrite," + "readrandom," + "readseq," + "fillrand100K," + "fillseq100K," + "readseq100K," + "readrand100K,"; + +// Number of key/values to place in database +static int FLAGS_num = 1000000; + +// Number of read operations to do. If negative, do FLAGS_num reads. +static int FLAGS_reads = -1; + +// Size of each value +static int FLAGS_value_size = 100; + +// Arrange to generate values that shrink to this fraction of +// their original size after compression +static double FLAGS_compression_ratio = 0.5; + +// Print histogram of operation timings +static bool FLAGS_histogram = false; + +// Cache size. Default 4 MB +static int FLAGS_cache_size = 4194304; + +// Page size. Default 1 KB +static int FLAGS_page_size = 1024; + +// If true, do not destroy the existing database. If you set this +// flag and also specify a benchmark that wants a fresh database, that +// benchmark will fail. +static bool FLAGS_use_existing_db = false; + +// Compression flag. If true, compression is on. If false, compression +// is off. +static bool FLAGS_compression = true; + +// Use the db with the following name. +static const char* FLAGS_db = nullptr; + +inline static void DBSynchronize(kyotocabinet::TreeDB* db_) { + // Synchronize will flush writes to disk + if (!db_->synchronize()) { + std::fprintf(stderr, "synchronize error: %s\n", db_->error().name()); + } +} + +namespace leveldb { + +// Helper for quickly generating random data. +namespace { +class RandomGenerator { + private: + std::string data_; + int pos_; + + public: + RandomGenerator() { + // We use a limited amount of data over and over again and ensure + // that it is larger than the compression window (32KB), and also + // large enough to serve all typical value sizes we want to write. + Random rnd(301); + std::string piece; + while (data_.size() < 1048576) { + // Add a short fragment that is as compressible as specified + // by FLAGS_compression_ratio. + test::CompressibleString(&rnd, FLAGS_compression_ratio, 100, &piece); + data_.append(piece); + } + pos_ = 0; + } + + Slice Generate(int len) { + if (pos_ + len > data_.size()) { + pos_ = 0; + assert(len < data_.size()); + } + pos_ += len; + return Slice(data_.data() + pos_ - len, len); + } +}; + +static Slice TrimSpace(Slice s) { + int start = 0; + while (start < s.size() && isspace(s[start])) { + start++; + } + int limit = s.size(); + while (limit > start && isspace(s[limit - 1])) { + limit--; + } + return Slice(s.data() + start, limit - start); +} + +} // namespace + +class Benchmark { + private: + kyotocabinet::TreeDB* db_; + int db_num_; + int num_; + int reads_; + double start_; + double last_op_finish_; + int64_t bytes_; + std::string message_; + Histogram hist_; + RandomGenerator gen_; + Random rand_; + kyotocabinet::LZOCompressor<kyotocabinet::LZO::RAW> comp_; + + // State kept for progress messages + int done_; + int next_report_; // When to report next + + void PrintHeader() { + const int kKeySize = 16; + PrintEnvironment(); + std::fprintf(stdout, "Keys: %d bytes each\n", kKeySize); + std::fprintf( + stdout, "Values: %d bytes each (%d bytes after compression)\n", + FLAGS_value_size, + static_cast<int>(FLAGS_value_size * FLAGS_compression_ratio + 0.5)); + std::fprintf(stdout, "Entries: %d\n", num_); + std::fprintf(stdout, "RawSize: %.1f MB (estimated)\n", + ((static_cast<int64_t>(kKeySize + FLAGS_value_size) * num_) / + 1048576.0)); + std::fprintf( + stdout, "FileSize: %.1f MB (estimated)\n", + (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_) / + 1048576.0)); + PrintWarnings(); + std::fprintf(stdout, "------------------------------------------------\n"); + } + + void PrintWarnings() { +#if defined(__GNUC__) && !defined(__OPTIMIZE__) + std::fprintf( + stdout, + "WARNING: Optimization is disabled: benchmarks unnecessarily slow\n"); +#endif +#ifndef NDEBUG + std::fprintf( + stdout, + "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n"); +#endif + } + + void PrintEnvironment() { + std::fprintf( + stderr, "Kyoto Cabinet: version %s, lib ver %d, lib rev %d\n", + kyotocabinet::VERSION, kyotocabinet::LIBVER, kyotocabinet::LIBREV); + +#if defined(__linux) + time_t now = time(nullptr); + std::fprintf(stderr, "Date: %s", + ctime(&now)); // ctime() adds newline + + FILE* cpuinfo = std::fopen("/proc/cpuinfo", "r"); + if (cpuinfo != nullptr) { + char line[1000]; + int num_cpus = 0; + std::string cpu_type; + std::string cache_size; + while (fgets(line, sizeof(line), cpuinfo) != nullptr) { + const char* sep = strchr(line, ':'); + if (sep == nullptr) { + continue; + } + Slice key = TrimSpace(Slice(line, sep - 1 - line)); + Slice val = TrimSpace(Slice(sep + 1)); + if (key == "model name") { + ++num_cpus; + cpu_type = val.ToString(); + } else if (key == "cache size") { + cache_size = val.ToString(); + } + } + std::fclose(cpuinfo); + std::fprintf(stderr, "CPU: %d * %s\n", num_cpus, + cpu_type.c_str()); + std::fprintf(stderr, "CPUCache: %s\n", cache_size.c_str()); + } +#endif + } + + void Start() { + start_ = Env::Default()->NowMicros() * 1e-6; + bytes_ = 0; + message_.clear(); + last_op_finish_ = start_; + hist_.Clear(); + done_ = 0; + next_report_ = 100; + } + + void FinishedSingleOp() { + if (FLAGS_histogram) { + double now = Env::Default()->NowMicros() * 1e-6; + double micros = (now - last_op_finish_) * 1e6; + hist_.Add(micros); + if (micros > 20000) { + std::fprintf(stderr, "long op: %.1f micros%30s\r", micros, ""); + std::fflush(stderr); + } + last_op_finish_ = now; + } + + 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; + std::fprintf(stderr, "... finished %d ops%30s\r", done_, ""); + std::fflush(stderr); + } + } + + void Stop(const Slice& name) { + double finish = Env::Default()->NowMicros() * 1e-6; + + // Pretend at least one op was done in case we are running a benchmark + // that does not call FinishedSingleOp(). + if (done_ < 1) done_ = 1; + + if (bytes_ > 0) { + char rate[100]; + std::snprintf(rate, sizeof(rate), "%6.1f MB/s", + (bytes_ / 1048576.0) / (finish - start_)); + if (!message_.empty()) { + message_ = std::string(rate) + " " + message_; + } else { + message_ = rate; + } + } + + std::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) { + std::fprintf(stdout, "Microseconds per op:\n%s\n", + hist_.ToString().c_str()); + } + std::fflush(stdout); + } + + public: + enum Order { SEQUENTIAL, RANDOM }; + enum DBState { FRESH, EXISTING }; + + Benchmark() + : db_(nullptr), + num_(FLAGS_num), + reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads), + bytes_(0), + rand_(301) { + std::vector<std::string> files; + std::string test_dir; + Env::Default()->GetTestDirectory(&test_dir); + Env::Default()->GetChildren(test_dir.c_str(), &files); + if (!FLAGS_use_existing_db) { + for (int i = 0; i < files.size(); i++) { + if (Slice(files[i]).starts_with("dbbench_polyDB")) { + std::string file_name(test_dir); + file_name += "/"; + file_name += files[i]; + Env::Default()->RemoveFile(file_name.c_str()); + } + } + } + } + + ~Benchmark() { + if (!db_->close()) { + std::fprintf(stderr, "close error: %s\n", db_->error().name()); + } + } + + void Run() { + PrintHeader(); + Open(false); + + const char* benchmarks = FLAGS_benchmarks; + while (benchmarks != nullptr) { + const char* sep = strchr(benchmarks, ','); + Slice name; + if (sep == nullptr) { + name = benchmarks; + benchmarks = nullptr; + } else { + name = Slice(benchmarks, sep - benchmarks); + benchmarks = sep + 1; + } + + Start(); + + bool known = true; + bool write_sync = false; + if (name == Slice("fillseq")) { + Write(write_sync, SEQUENTIAL, FRESH, num_, FLAGS_value_size, 1); + DBSynchronize(db_); + } else if (name == Slice("fillrandom")) { + Write(write_sync, RANDOM, FRESH, num_, FLAGS_value_size, 1); + DBSynchronize(db_); + } else if (name == Slice("overwrite")) { + Write(write_sync, RANDOM, EXISTING, num_, FLAGS_value_size, 1); + DBSynchronize(db_); + } else if (name == Slice("fillrandsync")) { + write_sync = true; + Write(write_sync, RANDOM, FRESH, num_ / 100, FLAGS_value_size, 1); + DBSynchronize(db_); + } else if (name == Slice("fillseqsync")) { + write_sync = true; + Write(write_sync, SEQUENTIAL, FRESH, num_ / 100, FLAGS_value_size, 1); + DBSynchronize(db_); + } else if (name == Slice("fillrand100K")) { + Write(write_sync, RANDOM, FRESH, num_ / 1000, 100 * 1000, 1); + DBSynchronize(db_); + } else if (name == Slice("fillseq100K")) { + Write(write_sync, SEQUENTIAL, FRESH, num_ / 1000, 100 * 1000, 1); + DBSynchronize(db_); + } else if (name == Slice("readseq")) { + ReadSequential(); + } else if (name == Slice("readrandom")) { + ReadRandom(); + } else if (name == Slice("readrand100K")) { + int n = reads_; + reads_ /= 1000; + ReadRandom(); + reads_ = n; + } else if (name == Slice("readseq100K")) { + int n = reads_; + reads_ /= 1000; + ReadSequential(); + reads_ = n; + } else { + known = false; + if (name != Slice()) { // No error message for empty name + std::fprintf(stderr, "unknown benchmark '%s'\n", + name.ToString().c_str()); + } + } + if (known) { + Stop(name); + } + } + } + + private: + void Open(bool sync) { + assert(db_ == nullptr); + + // Initialize db_ + db_ = new kyotocabinet::TreeDB(); + char file_name[100]; + db_num_++; + std::string test_dir; + Env::Default()->GetTestDirectory(&test_dir); + std::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; + if (FLAGS_compression) { + tune_options |= kyotocabinet::TreeDB::TCOMPRESS; + db_->tune_compressor(&comp_); + } + db_->tune_options(tune_options); + db_->tune_page_cache(FLAGS_cache_size); + db_->tune_page(FLAGS_page_size); + db_->tune_map(256LL << 20); + if (sync) { + open_options |= kyotocabinet::PolyDB::OAUTOSYNC; + } + if (!db_->open(file_name, open_options)) { + std::fprintf(stderr, "open error: %s\n", db_->error().name()); + } + } + + 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) { + message_ = "skipping (--use_existing_db is true)"; + return; + } + delete db_; + db_ = nullptr; + Open(sync); + Start(); // Do not count time taken to destroy/open + } + + if (num_entries != num_) { + char msg[100]; + std::snprintf(msg, sizeof(msg), "(%d ops)", num_entries); + message_ = msg; + } + + // Write to database + for (int i = 0; i < num_entries; i++) { + const int k = (order == SEQUENTIAL) ? i : (rand_.Next() % num_entries); + char key[100]; + std::snprintf(key, sizeof(key), "%016d", k); + bytes_ += value_size + strlen(key); + std::string cpp_key = key; + if (!db_->set(cpp_key, gen_.Generate(value_size).ToString())) { + std::fprintf(stderr, "set error: %s\n", db_->error().name()); + } + FinishedSingleOp(); + } + } + + void ReadSequential() { + kyotocabinet::DB::Cursor* cur = db_->cursor(); + cur->jump(); + std::string ckey, cvalue; + while (cur->get(&ckey, &cvalue, true)) { + bytes_ += ckey.size() + cvalue.size(); + FinishedSingleOp(); + } + delete cur; + } + + void ReadRandom() { + std::string value; + for (int i = 0; i < reads_; i++) { + char key[100]; + const int k = rand_.Next() % reads_; + std::snprintf(key, sizeof(key), "%016d", k); + db_->get(key, &value); + FinishedSingleOp(); + } + } +}; + +} // namespace leveldb + +int main(int argc, char** argv) { + std::string default_db_path; + for (int i = 1; i < argc; i++) { + double d; + int n; + char junk; + if (leveldb::Slice(argv[i]).starts_with("--benchmarks=")) { + FLAGS_benchmarks = argv[i] + strlen("--benchmarks="); + } else if (sscanf(argv[i], "--compression_ratio=%lf%c", &d, &junk) == 1) { + FLAGS_compression_ratio = d; + } else if (sscanf(argv[i], "--histogram=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_histogram = 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) { + FLAGS_reads = n; + } else if (sscanf(argv[i], "--value_size=%d%c", &n, &junk) == 1) { + FLAGS_value_size = n; + } else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) { + FLAGS_cache_size = n; + } else if (sscanf(argv[i], "--page_size=%d%c", &n, &junk) == 1) { + FLAGS_page_size = n; + } else if (sscanf(argv[i], "--compression=%d%c", &n, &junk) == 1 && + (n == 0 || n == 1)) { + FLAGS_compression = (n == 1) ? true : false; + } else if (strncmp(argv[i], "--db=", 5) == 0) { + FLAGS_db = argv[i] + 5; + } else { + std::fprintf(stderr, "Invalid flag '%s'\n", argv[i]); + std::exit(1); + } + } + + // Choose a location for the test database if none given with --db=<path> + 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; + benchmark.Run(); + return 0; +} diff --git a/cmake/leveldbConfig.cmake.in b/cmake/leveldbConfig.cmake.in new file mode 100644 index 0000000..2572728 --- /dev/null +++ b/cmake/leveldbConfig.cmake.in @@ -0,0 +1,9 @@ +# Copyright 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. + +@PACKAGE_INIT@ + +include("${CMAKE_CURRENT_LIST_DIR}/leveldbTargets.cmake") + +check_required_components(leveldb)
\ No newline at end of file diff --git a/db/autocompact_test.cc b/db/autocompact_test.cc new file mode 100644 index 0000000..3b7241b --- /dev/null +++ b/db/autocompact_test.cc @@ -0,0 +1,115 @@ +// Copyright (c) 2013 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "gtest/gtest.h" +#include "db/db_impl.h" +#include "leveldb/cache.h" +#include "leveldb/db.h" +#include "util/testutil.h" + +namespace leveldb { + +class AutoCompactTest : public testing::Test { + public: + AutoCompactTest() { + dbname_ = testing::TempDir() + "autocompact_test"; + tiny_cache_ = NewLRUCache(100); + options_.block_cache = tiny_cache_; + DestroyDB(dbname_, options_); + options_.create_if_missing = true; + options_.compression = kNoCompression; + EXPECT_LEVELDB_OK(DB::Open(options_, dbname_, &db_)); + } + + ~AutoCompactTest() { + delete db_; + DestroyDB(dbname_, Options()); + delete tiny_cache_; + } + + std::string Key(int i) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "key%06d", i); + return std::string(buf); + } + + uint64_t Size(const Slice& start, const Slice& limit) { + Range r(start, limit); + uint64_t size; + db_->GetApproximateSizes(&r, 1, &size); + return size; + } + + void DoReads(int n); + + private: + std::string dbname_; + Cache* tiny_cache_; + Options options_; + DB* db_; +}; + +static const int kValueSize = 200 * 1024; +static const int kTotalSize = 100 * 1024 * 1024; +static const int kCount = kTotalSize / kValueSize; + +// Read through the first n keys repeatedly and check that they get +// compacted (verified by checking the size of the key space). +void AutoCompactTest::DoReads(int n) { + std::string value(kValueSize, 'x'); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + + // Fill database + for (int i = 0; i < kCount; i++) { + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), Key(i), value)); + } + ASSERT_LEVELDB_OK(dbi->TEST_CompactMemTable()); + + // Delete everything + for (int i = 0; i < kCount; i++) { + ASSERT_LEVELDB_OK(db_->Delete(WriteOptions(), Key(i))); + } + ASSERT_LEVELDB_OK(dbi->TEST_CompactMemTable()); + + // Get initial measurement of the space we will be reading. + const int64_t initial_size = Size(Key(0), Key(n)); + const int64_t initial_other_size = Size(Key(n), Key(kCount)); + + // Read until size drops significantly. + std::string limit_key = Key(n); + for (int read = 0; true; read++) { + 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()) { + // 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)); + std::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; + } + } + + // Verify that the size of the key space not touched by the reads + // 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); +} + +TEST_F(AutoCompactTest, ReadAll) { DoReads(kCount); } + +TEST_F(AutoCompactTest, ReadHalf) { DoReads(kCount / 2); } + +} // namespace leveldb + +int main(int argc, char** argv) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/db/builder.cc b/db/builder.cc new file mode 100644 index 0000000..e6329e0 --- /dev/null +++ b/db/builder.cc @@ -0,0 +1,82 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/builder.h" + +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/table_cache.h" +#include "db/version_edit.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" + +namespace leveldb { + +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(); + + std::string fname = TableFileName(dbname, meta->number); + if (iter->Valid()) { + WritableFile* file; + s = env->NewWritableFile(fname, &file); + if (!s.ok()) { + return s; + } + + TableBuilder* builder = new TableBuilder(options, file); + meta->smallest.DecodeFrom(iter->key()); + Slice key; + for (; iter->Valid(); iter->Next()) { + key = iter->key(); + builder->Add(key, iter->value()); + } + if (!key.empty()) { + meta->largest.DecodeFrom(key); + } + + // Finish and check for builder errors + s = builder->Finish(); + if (s.ok()) { + meta->file_size = builder->FileSize(); + assert(meta->file_size > 0); + } + delete builder; + + // Finish and check for file errors + if (s.ok()) { + s = file->Sync(); + } + if (s.ok()) { + s = file->Close(); + } + delete file; + file = nullptr; + + if (s.ok()) { + // Verify that the table is usable + Iterator* it = table_cache->NewIterator(ReadOptions(), meta->number, + meta->file_size); + s = it->status(); + delete it; + } + } + + // Check for input iterator errors + if (!iter->status().ok()) { + s = iter->status(); + } + + if (s.ok() && meta->file_size > 0) { + // Keep it + } else { + env->RemoveFile(fname); + } + return s; +} + +} // namespace leveldb diff --git a/db/builder.h b/db/builder.h new file mode 100644 index 0000000..7bd0b80 --- /dev/null +++ b/db/builder.h @@ -0,0 +1,30 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_BUILDER_H_ +#define STORAGE_LEVELDB_DB_BUILDER_H_ + +#include "leveldb/status.h" + +namespace leveldb { + +struct Options; +struct FileMetaData; + +class Env; +class Iterator; +class TableCache; +class VersionEdit; + +// Build a Table file from the contents of *iter. The generated file +// will be named according to meta->number. On success, the rest of +// *meta will be filled with metadata about the generated table. +// If no data is present in *iter, meta->file_size will be set to +// zero, and no Table file will be produced. +Status BuildTable(const std::string& dbname, Env* env, const Options& options, + TableCache* table_cache, Iterator* iter, FileMetaData* meta); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_BUILDER_H_ @@ -0,0 +1,565 @@ +// 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 "leveldb/c.h" + +#include <string.h> + +#include <cstdint> +#include <cstdlib> + +#include "leveldb/cache.h" +#include "leveldb/comparator.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/filter_policy.h" +#include "leveldb/iterator.h" +#include "leveldb/options.h" +#include "leveldb/status.h" +#include "leveldb/write_batch.h" + +using leveldb::Cache; +using leveldb::Comparator; +using leveldb::CompressionType; +using leveldb::DB; +using leveldb::Env; +using leveldb::FileLock; +using leveldb::FilterPolicy; +using leveldb::Iterator; +using leveldb::kMajorVersion; +using leveldb::kMinorVersion; +using leveldb::Logger; +using leveldb::NewBloomFilterPolicy; +using leveldb::NewLRUCache; +using leveldb::Options; +using leveldb::RandomAccessFile; +using leveldb::Range; +using leveldb::ReadOptions; +using leveldb::SequentialFile; +using leveldb::Slice; +using leveldb::Snapshot; +using leveldb::Status; +using leveldb::WritableFile; +using leveldb::WriteBatch; +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_comparator_t : public Comparator { + ~leveldb_comparator_t() override { (*destructor_)(state_); } + + int Compare(const Slice& a, const Slice& b) const override { + return (*compare_)(state_, a.data(), a.size(), b.data(), b.size()); + } + + const char* Name() const override { return (*name_)(state_); } + + // No-ops since the C binding does not support key shortening methods. + 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 { + ~leveldb_filterpolicy_t() override { (*destructor_)(state_); } + + const char* Name() const override { return (*name_)(state_); } + + void CreateFilter(const Slice* keys, int n, std::string* dst) const override { + std::vector<const char*> key_pointers(n); + std::vector<size_t> key_sizes(n); + for (int i = 0; i < n; i++) { + key_pointers[i] = keys[i].data(); + key_sizes[i] = keys[i].size(); + } + size_t len; + char* filter = (*create_)(state_, &key_pointers[0], &key_sizes[0], n, &len); + dst->append(filter, len); + std::free(filter); + } + + 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); + uint8_t (*key_match_)(void*, const char* key, size_t length, + const char* filter, size_t filter_length); +}; + +struct leveldb_env_t { + Env* rep; + bool is_default; +}; + +static bool SaveError(char** errptr, const Status& s) { + assert(errptr != nullptr); + if (s.ok()) { + return false; + } else if (*errptr == nullptr) { + *errptr = strdup(s.ToString().c_str()); + } else { + // TODO(sanjay): Merge with existing error? + std::free(*errptr); + *errptr = strdup(s.ToString().c_str()); + } + return true; +} + +static char* CopyString(const std::string& str) { + char* result = + reinterpret_cast<char*>(std::malloc(sizeof(char) * str.size())); + std::memcpy(result, str.data(), sizeof(char) * str.size()); + return result; +} + +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 nullptr; + } + leveldb_t* result = new leveldb_t; + result->rep = db; + return result; +} + +void leveldb_close(leveldb_t* db) { + delete db->rep; + 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) { + 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) { + 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) { + 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 = nullptr; + std::string tmp; + Status s = db->rep->Get(options->rep, Slice(key, keylen), &tmp); + if (s.ok()) { + *vallen = tmp.size(); + result = CopyString(tmp); + } else { + *vallen = 0; + if (!s.IsNotFound()) { + SaveError(errptr, s); + } + } + return result; +} + +leveldb_iterator_t* leveldb_create_iterator( + 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) { + 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) { + db->rep->ReleaseSnapshot(snapshot->rep); + delete snapshot; +} + +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 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) { + Range* ranges = new Range[num_ranges]; + for (int i = 0; i < num_ranges; i++) { + ranges[i].start = Slice(range_start_key[i], range_start_key_len[i]); + ranges[i].limit = Slice(range_limit_key[i], range_limit_key_len[i]); + } + db->rep->GetApproximateSizes(ranges, num_ranges, sizes); + delete[] ranges; +} + +void 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) : 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) { + SaveError(errptr, DestroyDB(name, options->rep)); +} + +void leveldb_repair_db(const leveldb_options_t* options, const char* name, + char** errptr) { + SaveError(errptr, RepairDB(name, options->rep)); +} + +void leveldb_iter_destroy(leveldb_iterator_t* iter) { + delete iter->rep; + delete iter; +} + +uint8_t leveldb_iter_valid(const leveldb_iterator_t* iter) { + return iter->rep->Valid(); +} + +void leveldb_iter_seek_to_first(leveldb_iterator_t* iter) { + iter->rep->SeekToFirst(); +} + +void leveldb_iter_seek_to_last(leveldb_iterator_t* iter) { + iter->rep->SeekToLast(); +} + +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_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(); + *klen = s.size(); + return s.data(); +} + +const char* leveldb_iter_value(const leveldb_iterator_t* iter, size_t* vlen) { + Slice s = iter->rep->value(); + *vlen = s.size(); + return s.data(); +} + +void leveldb_iter_get_error(const leveldb_iterator_t* iter, char** errptr) { + SaveError(errptr, iter->rep->status()); +} + +leveldb_writebatch_t* leveldb_writebatch_create() { + return new leveldb_writebatch_t; +} + +void leveldb_writebatch_destroy(leveldb_writebatch_t* b) { delete b; } + +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) { + b->rep.Put(Slice(key, klen), Slice(val, vlen)); +} + +void leveldb_writebatch_delete(leveldb_writebatch_t* b, const char* key, + size_t klen) { + b->rep.Delete(Slice(key, 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); + void Put(const Slice& key, const Slice& value) override { + (*put_)(state_, key.data(), key.size(), value.data(), value.size()); + } + void Delete(const Slice& key) override { + (*deleted_)(state_, key.data(), key.size()); + } + }; + H handler; + handler.state_ = state; + handler.put_ = put; + handler.deleted_ = deleted; + b->rep.Iterate(&handler); +} + +void leveldb_writebatch_append(leveldb_writebatch_t* destination, + const leveldb_writebatch_t* source) { + destination->rep.Append(source->rep); +} + +leveldb_options_t* leveldb_options_create() { return new leveldb_options_t; } + +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) { + opt->rep.filter_policy = policy; +} + +void leveldb_options_set_create_if_missing(leveldb_options_t* opt, uint8_t v) { + opt->rep.create_if_missing = v; +} + +void leveldb_options_set_error_if_exists(leveldb_options_t* opt, uint8_t v) { + opt->rep.error_if_exists = v; +} + +void leveldb_options_set_paranoid_checks(leveldb_options_t* opt, uint8_t 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 : nullptr); +} + +void leveldb_options_set_info_log(leveldb_options_t* opt, leveldb_logger_t* l) { + opt->rep.info_log = (l ? l->rep : nullptr); +} + +void leveldb_options_set_write_buffer_size(leveldb_options_t* opt, size_t s) { + opt->rep.write_buffer_size = s; +} + +void leveldb_options_set_max_open_files(leveldb_options_t* opt, int n) { + opt->rep.max_open_files = n; +} + +void leveldb_options_set_cache(leveldb_options_t* opt, leveldb_cache_t* c) { + opt->rep.block_cache = c->rep; +} + +void leveldb_options_set_block_size(leveldb_options_t* opt, size_t s) { + opt->rep.block_size = s; +} + +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<CompressionType>(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), + const char* (*name)(void*)) { + leveldb_comparator_t* result = new leveldb_comparator_t; + result->state_ = state; + result->destructor_ = destructor; + result->compare_ = compare; + result->name_ = name; + return result; +} + +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), + uint8_t (*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; + result->destructor_ = destructor; + result->create_ = create_filter; + result->key_match_ = key_may_match; + result->name_ = name; + return result; +} + +void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t* filter) { + delete filter; +} + +leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(int bits_per_key) { + // Make a leveldb_filterpolicy_t, but override all of its methods so + // they delegate to a NewBloomFilterPolicy() instead of user + // supplied C functions. + struct Wrapper : public leveldb_filterpolicy_t { + 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 { + return rep_->CreateFilter(keys, n, dst); + } + bool KeyMayMatch(const Slice& key, const Slice& filter) const { + return rep_->KeyMayMatch(key, filter); + } + + const FilterPolicy* rep_; + }; + Wrapper* wrapper = new Wrapper; + wrapper->rep_ = NewBloomFilterPolicy(bits_per_key); + wrapper->state_ = nullptr; + wrapper->destructor_ = &Wrapper::DoNothing; + return wrapper; +} + +leveldb_readoptions_t* leveldb_readoptions_create() { + return new leveldb_readoptions_t; +} + +void leveldb_readoptions_destroy(leveldb_readoptions_t* opt) { delete opt; } + +void leveldb_readoptions_set_verify_checksums(leveldb_readoptions_t* opt, + uint8_t v) { + opt->rep.verify_checksums = v; +} + +void leveldb_readoptions_set_fill_cache(leveldb_readoptions_t* opt, uint8_t 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 : 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_set_sync(leveldb_writeoptions_t* opt, uint8_t v) { + opt->rep.sync = v; +} + +leveldb_cache_t* leveldb_cache_create_lru(size_t capacity) { + leveldb_cache_t* c = new leveldb_cache_t; + c->rep = NewLRUCache(capacity); + return c; +} + +void leveldb_cache_destroy(leveldb_cache_t* cache) { + delete cache->rep; + delete cache; +} + +leveldb_env_t* leveldb_create_default_env() { + leveldb_env_t* result = new leveldb_env_t; + result->rep = Env::Default(); + result->is_default = true; + return result; +} + +void leveldb_env_destroy(leveldb_env_t* env) { + if (!env->is_default) delete env->rep; + delete env; +} + +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<char*>(std::malloc(result.size() + 1)); + std::memcpy(buffer, result.data(), result.size()); + buffer[result.size()] = '\0'; + return buffer; +} + +void leveldb_free(void* ptr) { std::free(ptr); } + +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 new file mode 100644 index 0000000..16c77ee --- /dev/null +++ b/db/c_test.c @@ -0,0 +1,384 @@ +/* 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 "leveldb/c.h" + +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +const char* phase = ""; + +static void StartPhase(const char* name) { + fprintf(stderr, "=== Test %s\n", name); + phase = name; +} + +#define CheckNoError(err) \ + if ((err) != NULL) { \ + fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, (err)); \ + abort(); \ + } + +#define CheckCondition(cond) \ + if (!(cond)) { \ + fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, #cond); \ + abort(); \ + } + +static void CheckEqual(const char* expected, const char* v, size_t n) { + if (expected == NULL && v == NULL) { + // ok + } else if (expected != NULL && v != NULL && n == strlen(expected) && + memcmp(expected, v, n) == 0) { + // ok + return; + } else { + fprintf(stderr, "%s: expected '%s', got '%s'\n", + phase, + (expected ? expected : "(null)"), + (v ? v : "(null")); + abort(); + } +} + +static void Free(char** ptr) { + if (*ptr) { + free(*ptr); + *ptr = NULL; + } +} + +static void CheckGet( + leveldb_t* db, + const leveldb_readoptions_t* options, + const char* key, + const char* expected) { + char* err = NULL; + size_t val_len; + char* val; + val = leveldb_get(db, options, key, strlen(key), &val_len, &err); + CheckNoError(err); + CheckEqual(expected, val, val_len); + Free(&val); +} + +static void CheckIter(leveldb_iterator_t* iter, + const char* key, const char* val) { + size_t len; + const char* str; + str = leveldb_iter_key(iter, &len); + CheckEqual(key, str, len); + str = leveldb_iter_value(iter, &len); + CheckEqual(val, str, len); +} + +// Callback from leveldb_writebatch_iterate() +static void CheckPut(void* ptr, + const char* k, size_t klen, + const char* v, size_t vlen) { + int* state = (int*) ptr; + CheckCondition(*state < 2); + switch (*state) { + case 0: + CheckEqual("bar", k, klen); + CheckEqual("b", v, vlen); + break; + case 1: + CheckEqual("box", k, klen); + CheckEqual("c", v, vlen); + break; + } + (*state)++; +} + +// Callback from leveldb_writebatch_iterate() +static void CheckDel(void* ptr, const char* k, size_t klen) { + int* state = (int*) ptr; + CheckCondition(*state == 2); + CheckEqual("bar", k, klen); + (*state)++; +} + +static void CmpDestroy(void* arg) { } + +static int CmpCompare(void* arg, const char* a, size_t alen, + const char* b, size_t blen) { + int n = (alen < blen) ? alen : blen; + int r = memcmp(a, b, n); + if (r == 0) { + if (alen < blen) r = -1; + else if (alen > blen) r = +1; + } + return r; +} + +static const char* CmpName(void* arg) { + return "foo"; +} + +// Custom filter policy +static uint8_t fake_filter_result = 1; +static void FilterDestroy(void* arg) { } +static const char* FilterName(void* arg) { + return "TestFilter"; +} +static char* FilterCreate( + void* arg, + const char* const* key_array, const size_t* key_length_array, + int num_keys, + size_t* filter_length) { + *filter_length = 4; + char* result = malloc(4); + memcpy(result, "fake", 4); + return result; +} +uint8_t FilterKeyMatch(void* arg, const char* key, size_t length, + const char* filter, size_t filter_length) { + CheckCondition(filter_length == 4); + CheckCondition(memcmp(filter, "fake", 4) == 0); + return fake_filter_result; +} + +int main(int argc, char** argv) { + leveldb_t* db; + leveldb_comparator_t* cmp; + leveldb_cache_t* cache; + leveldb_env_t* env; + 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); + + 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); + leveldb_options_set_error_if_exists(options, 1); + leveldb_options_set_cache(options, cache); + leveldb_options_set_env(options, env); + leveldb_options_set_info_log(options, NULL); + leveldb_options_set_write_buffer_size(options, 100000); + leveldb_options_set_paranoid_checks(options, 1); + 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(); + leveldb_readoptions_set_verify_checksums(roptions, 1); + leveldb_readoptions_set_fill_cache(roptions, 0); + + woptions = leveldb_writeoptions_create(); + leveldb_writeoptions_set_sync(woptions, 1); + + StartPhase("destroy"); + leveldb_destroy_db(options, dbname, &err); + Free(&err); + + StartPhase("open_error"); + db = leveldb_open(options, dbname, &err); + CheckCondition(err != NULL); + Free(&err); + + StartPhase("leveldb_free"); + db = leveldb_open(options, dbname, &err); + CheckCondition(err != NULL); + leveldb_free(err); + err = NULL; + + StartPhase("open"); + leveldb_options_set_create_if_missing(options, 1); + db = leveldb_open(options, dbname, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", NULL); + + StartPhase("put"); + leveldb_put(db, woptions, "foo", 3, "hello", 5, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", "hello"); + + StartPhase("compactall"); + leveldb_compact_range(db, NULL, 0, NULL, 0); + CheckGet(db, roptions, "foo", "hello"); + + StartPhase("compactrange"); + leveldb_compact_range(db, "a", 1, "z", 1); + CheckGet(db, roptions, "foo", "hello"); + + StartPhase("writebatch"); + { + leveldb_writebatch_t* wb = leveldb_writebatch_create(); + leveldb_writebatch_put(wb, "foo", 3, "a", 1); + leveldb_writebatch_clear(wb); + leveldb_writebatch_put(wb, "bar", 3, "b", 1); + leveldb_writebatch_put(wb, "box", 3, "c", 1); + + 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); + leveldb_writebatch_destroy(wb); + } + + StartPhase("iter"); + { + leveldb_iterator_t* iter = leveldb_create_iterator(db, roptions); + CheckCondition(!leveldb_iter_valid(iter)); + leveldb_iter_seek_to_first(iter); + CheckCondition(leveldb_iter_valid(iter)); + CheckIter(iter, "box", "c"); + leveldb_iter_next(iter); + CheckIter(iter, "foo", "hello"); + leveldb_iter_prev(iter); + CheckIter(iter, "box", "c"); + leveldb_iter_prev(iter); + CheckCondition(!leveldb_iter_valid(iter)); + leveldb_iter_seek_to_last(iter); + CheckIter(iter, "foo", "hello"); + leveldb_iter_seek(iter, "b", 1); + CheckIter(iter, "box", "c"); + leveldb_iter_get_error(iter, &err); + CheckNoError(err); + leveldb_iter_destroy(iter); + } + + StartPhase("approximate_sizes"); + { + int i; + int n = 20000; + char keybuf[100]; + char valbuf[100]; + uint64_t sizes[2]; + const char* start[2] = { "a", "k00000000000000010000" }; + size_t start_len[2] = { 1, 21 }; + const char* limit[2] = { "k00000000000000010000", "z" }; + size_t limit_len[2] = { 21, 1 }; + leveldb_writeoptions_set_sync(woptions, 0); + for (i = 0; i < n; i++) { + snprintf(keybuf, sizeof(keybuf), "k%020d", i); + snprintf(valbuf, sizeof(valbuf), "v%020d", i); + leveldb_put(db, woptions, keybuf, strlen(keybuf), valbuf, strlen(valbuf), + &err); + CheckNoError(err); + } + leveldb_approximate_sizes(db, 2, start, start_len, limit, limit_len, sizes); + CheckCondition(sizes[0] > 0); + CheckCondition(sizes[1] > 0); + } + + StartPhase("property"); + { + char* prop = leveldb_property_value(db, "nosuchprop"); + CheckCondition(prop == NULL); + prop = leveldb_property_value(db, "leveldb.stats"); + CheckCondition(prop != NULL); + Free(&prop); + } + + StartPhase("snapshot"); + { + const leveldb_snapshot_t* snap; + snap = leveldb_create_snapshot(db); + leveldb_delete(db, woptions, "foo", 3, &err); + CheckNoError(err); + leveldb_readoptions_set_snapshot(roptions, snap); + CheckGet(db, roptions, "foo", "hello"); + leveldb_readoptions_set_snapshot(roptions, NULL); + CheckGet(db, roptions, "foo", NULL); + leveldb_release_snapshot(db, snap); + } + + StartPhase("repair"); + { + leveldb_close(db); + leveldb_options_set_create_if_missing(options, 0); + leveldb_options_set_error_if_exists(options, 0); + leveldb_repair_db(options, dbname, &err); + CheckNoError(err); + db = leveldb_open(options, dbname, &err); + CheckNoError(err); + CheckGet(db, roptions, "foo", NULL); + CheckGet(db, roptions, "bar", NULL); + CheckGet(db, roptions, "box", "c"); + leveldb_options_set_create_if_missing(options, 1); + leveldb_options_set_error_if_exists(options, 1); + } + + StartPhase("filter"); + for (run = 0; run < 2; run++) { + // First run uses custom filter, second run uses bloom filter + CheckNoError(err); + leveldb_filterpolicy_t* policy; + if (run == 0) { + policy = leveldb_filterpolicy_create( + NULL, FilterDestroy, FilterCreate, FilterKeyMatch, FilterName); + } else { + policy = leveldb_filterpolicy_create_bloom(10); + } + + // Create new database + leveldb_close(db); + leveldb_destroy_db(options, dbname, &err); + leveldb_options_set_filter_policy(options, policy); + db = leveldb_open(options, dbname, &err); + CheckNoError(err); + leveldb_put(db, woptions, "foo", 3, "foovalue", 8, &err); + CheckNoError(err); + leveldb_put(db, woptions, "bar", 3, "barvalue", 8, &err); + CheckNoError(err); + leveldb_compact_range(db, NULL, 0, NULL, 0); + + fake_filter_result = 1; + CheckGet(db, roptions, "foo", "foovalue"); + CheckGet(db, roptions, "bar", "barvalue"); + if (phase == 0) { + // Must not find value when custom filter returns false + fake_filter_result = 0; + CheckGet(db, roptions, "foo", NULL); + CheckGet(db, roptions, "bar", NULL); + fake_filter_result = 1; + + CheckGet(db, roptions, "foo", "foovalue"); + CheckGet(db, roptions, "bar", "barvalue"); + } + leveldb_options_set_filter_policy(options, NULL); + leveldb_filterpolicy_destroy(policy); + } + + StartPhase("cleanup"); + leveldb_close(db); + 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); + + fprintf(stderr, "PASS\n"); + return 0; +} diff --git a/db/corruption_test.cc b/db/corruption_test.cc new file mode 100644 index 0000000..5bb2c90 --- /dev/null +++ b/db/corruption_test.cc @@ -0,0 +1,363 @@ +// 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 <sys/types.h> + +#include "gtest/gtest.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/testutil.h" + +namespace leveldb { + +static const int kValueSize = 1000; + +class CorruptionTest : public testing::Test { + public: + CorruptionTest() + : db_(nullptr), + dbname_("/memenv/corruption_test"), + tiny_cache_(NewLRUCache(100)) { + options_.env = &env_; + options_.block_cache = tiny_cache_; + DestroyDB(dbname_, options_); + + options_.create_if_missing = true; + Reopen(); + options_.create_if_missing = false; + } + + ~CorruptionTest() { + delete db_; + delete tiny_cache_; + } + + Status TryReopen() { + delete db_; + db_ = nullptr; + return DB::Open(options_, dbname_, &db_); + } + + void Reopen() { ASSERT_LEVELDB_OK(TryReopen()); } + + void RepairDB() { + delete db_; + db_ = nullptr; + ASSERT_LEVELDB_OK(::leveldb::RepairDB(dbname_, options_)); + } + + void Build(int n) { + std::string key_space, value_space; + WriteBatch batch; + for (int i = 0; i < n; i++) { + // if ((i % 100) == 0) std::fprintf(stderr, "@ %d of %d\n", i, n); + Slice key = Key(i, &key_space); + batch.Clear(); + batch.Put(key, Value(i, &value_space)); + WriteOptions options; + // Corrupt() doesn't work without this sync on windows; stat reports 0 for + // the file size. + if (i == n - 1) { + options.sync = true; + } + ASSERT_LEVELDB_OK(db_->Write(options, &batch)); + } + } + + void Check(int min_expected, int max_expected) { + int next_expected = 0; + int missed = 0; + int bad_keys = 0; + int bad_values = 0; + int correct = 0; + std::string value_space; + Iterator* iter = db_->NewIterator(ReadOptions()); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + uint64_t key; + Slice in(iter->key()); + if (in == "" || in == "~") { + // Ignore boundary keys. + continue; + } + if (!ConsumeDecimalNumber(&in, &key) || !in.empty() || + key < next_expected) { + bad_keys++; + continue; + } + missed += (key - next_expected); + next_expected = key + 1; + if (iter->value() != Value(key, &value_space)) { + bad_values++; + } else { + correct++; + } + } + delete iter; + + std::fprintf( + stderr, + "expected=%d..%d; got=%d; bad_keys=%d; bad_values=%d; missed=%d\n", + min_expected, max_expected, correct, bad_keys, bad_values, missed); + ASSERT_LE(min_expected, correct); + ASSERT_GE(max_expected, correct); + } + + void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) { + // Pick file to corrupt + std::vector<std::string> filenames; + ASSERT_LEVELDB_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 && + int(number) > picked_number) { // Pick latest file + fname = dbname_ + "/" + filenames[i]; + picked_number = number; + } + } + ASSERT_TRUE(!fname.empty()) << filetype; + + uint64_t file_size; + ASSERT_LEVELDB_OK(env_.target()->GetFileSize(fname, &file_size)); + + if (offset < 0) { + // Relative to end of file; make it absolute + if (-offset > file_size) { + offset = 0; + } else { + offset = file_size + offset; + } + } + if (offset > file_size) { + offset = file_size; + } + if (offset + bytes_to_corrupt > file_size) { + bytes_to_corrupt = file_size - offset; + } + + // Do it + std::string 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_.target(), contents, fname); + ASSERT_TRUE(s.ok()) << s.ToString(); + } + + int Property(const std::string& name) { + std::string property; + int result; + if (db_->GetProperty(name, &property) && + sscanf(property.c_str(), "%d", &result) == 1) { + return result; + } else { + return -1; + } + } + + // Return the ith key + Slice Key(int i, std::string* storage) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "%016d", i); + storage->assign(buf, strlen(buf)); + return Slice(*storage); + } + + // Return the value to associate with the specified key + Slice Value(int k, std::string* storage) { + Random r(k); + return test::RandomString(&r, kValueSize, storage); + } + + test::ErrorEnv env_; + Options options_; + DB* db_; + + private: + std::string dbname_; + Cache* tiny_cache_; +}; + +TEST_F(CorruptionTest, Recovery) { + Build(100); + Check(100, 100); + Corrupt(kLogFile, 19, 1); // WriteBatch tag for first record + Corrupt(kLogFile, log::kBlockSize + 1000, 1); // Somewhere in second block + Reopen(); + + // The 64 records in the first two log blocks are completely lost. + Check(36, 36); +} + +TEST_F(CorruptionTest, RecoverWriteError) { + env_.writable_file_error_ = true; + Status s = TryReopen(); + ASSERT_TRUE(!s.ok()); +} + +TEST_F(CorruptionTest, NewFileErrorDuringWrite) { + // Do enough writing to force minor compaction + env_.writable_file_error_ = true; + const int num = 3 + (Options().write_buffer_size / kValueSize); + std::string value_storage; + Status s; + for (int i = 0; s.ok() && i < num; i++) { + WriteBatch batch; + batch.Put("a", Value(100, &value_storage)); + s = db_->Write(WriteOptions(), &batch); + } + ASSERT_TRUE(!s.ok()); + ASSERT_GE(env_.num_writable_file_errors_, 1); + env_.writable_file_error_ = false; + Reopen(); +} + +TEST_F(CorruptionTest, TableFile) { + Build(100); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + dbi->TEST_CompactRange(0, nullptr, nullptr); + dbi->TEST_CompactRange(1, nullptr, nullptr); + + Corrupt(kTableFile, 100, 1); + Check(90, 99); +} + +TEST_F(CorruptionTest, TableFileRepair) { + options_.block_size = 2 * kValueSize; // Limit scope of corruption + options_.paranoid_checks = true; + Reopen(); + Build(100); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + dbi->TEST_CompactRange(0, nullptr, nullptr); + dbi->TEST_CompactRange(1, nullptr, nullptr); + + Corrupt(kTableFile, 100, 1); + RepairDB(); + Reopen(); + Check(95, 99); +} + +TEST_F(CorruptionTest, TableFileIndexData) { + Build(10000); // Enough to build multiple Tables + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + + Corrupt(kTableFile, -2000, 500); + Reopen(); + Check(5000, 9999); +} + +TEST_F(CorruptionTest, MissingDescriptor) { + Build(1000); + RepairDB(); + Reopen(); + Check(1000, 1000); +} + +TEST_F(CorruptionTest, SequenceNumberRecovery) { + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v1")); + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v2")); + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v3")); + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v4")); + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v5")); + RepairDB(); + Reopen(); + std::string v; + ASSERT_LEVELDB_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("v5", v); + // Write something. If sequence number was not recovered properly, + // it will be hidden by an earlier write. + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v6")); + ASSERT_LEVELDB_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("v6", v); + Reopen(); + ASSERT_LEVELDB_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("v6", v); +} + +TEST_F(CorruptionTest, CorruptedDescriptor) { + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "hello")); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + dbi->TEST_CompactRange(0, nullptr, nullptr); + + Corrupt(kDescriptorFile, 0, 1000); + Status s = TryReopen(); + ASSERT_TRUE(!s.ok()); + + RepairDB(); + Reopen(); + std::string v; + ASSERT_LEVELDB_OK(db_->Get(ReadOptions(), "foo", &v)); + ASSERT_EQ("hello", v); +} + +TEST_F(CorruptionTest, CompactionInputError) { + Build(10); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + const int last = config::kMaxMemCompactLevel; + ASSERT_EQ(1, Property("leveldb.num-files-at-level" + NumberToString(last))); + + Corrupt(kTableFile, 100, 1); + Check(5, 9); + + // Force compactions by writing lots of values + Build(10000); + Check(10000, 10000); +} + +TEST_F(CorruptionTest, CompactionInputErrorParanoid) { + options_.paranoid_checks = true; + options_.write_buffer_size = 512 << 10; + Reopen(); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + + // Make multiple inputs so we need to compact. + for (int i = 0; i < 2; i++) { + Build(10); + dbi->TEST_CompactMemTable(); + Corrupt(kTableFile, 100, 1); + env_.SleepForMicroseconds(100000); + } + dbi->CompactRange(nullptr, nullptr); + + // Write must fail because of corrupted table + std::string tmp1, tmp2; + Status s = db_->Put(WriteOptions(), Key(5, &tmp1), Value(5, &tmp2)); + ASSERT_TRUE(!s.ok()) << "write did not fail in corrupted paranoid db"; +} + +TEST_F(CorruptionTest, UnrelatedKeys) { + Build(10); + DBImpl* dbi = reinterpret_cast<DBImpl*>(db_); + dbi->TEST_CompactMemTable(); + Corrupt(kTableFile, 100, 1); + + std::string tmp1, tmp2; + ASSERT_LEVELDB_OK( + db_->Put(WriteOptions(), Key(1000, &tmp1), Value(1000, &tmp2))); + std::string v; + ASSERT_LEVELDB_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v)); + ASSERT_EQ(Value(1000, &tmp2).ToString(), v); + dbi->TEST_CompactMemTable(); + ASSERT_LEVELDB_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v)); + ASSERT_EQ(Value(1000, &tmp2).ToString(), v); +} + +} // namespace leveldb + diff --git a/db/db_impl.cc b/db/db_impl.cc new file mode 100644 index 0000000..1a4e459 --- /dev/null +++ b/db/db_impl.cc @@ -0,0 +1,1559 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/db_impl.h" + +#include <algorithm> +#include <atomic> +#include <cstdint> +#include <cstdio> +#include <set> +#include <string> +#include <vector> + +#include "db/builder.h" +#include "db/db_iter.h" +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "db/memtable.h" +#include "db/table_cache.h" +#include "db/version_set.h" +#include "db/write_batch_internal.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/status.h" +#include "leveldb/table.h" +#include "leveldb/table_builder.h" +#include "port/port.h" +#include "table/block.h" +#include "table/merger.h" +#include "table/two_level_iterator.h" +#include "util/coding.h" +#include "util/logging.h" +#include "util/mutexlock.h" + +namespace leveldb { + +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; +}; + +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 + // will never have to service a snapshot below smallest_snapshot. + // Therefore if we have seen a sequence number S <= smallest_snapshot, + // we can drop all entries for the same key with sequence numbers < S. + SequenceNumber smallest_snapshot; + + std::vector<Output> outputs; + + // State kept for output being generated + WritableFile* outfile; + TableBuilder* builder; + + uint64_t total_bytes; +}; + +// Fix user-supplied options to be reasonable +template <class T, class V> +static void ClipToRange(T* ptr, V minvalue, V maxvalue) { + if (static_cast<V>(*ptr) > maxvalue) *ptr = maxvalue; + if (static_cast<V>(*ptr) < minvalue) *ptr = minvalue; +} +Options SanitizeOptions(const std::string& dbname, + const InternalKeyComparator* icmp, + const InternalFilterPolicy* ipolicy, + const Options& src) { + Options result = src; + result.comparator = icmp; + 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 = nullptr; + } + } + 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), + internal_filter_policy_(raw_options.filter_policy), + options_(SanitizeOptions(dbname, &internal_comparator_, + &internal_filter_policy_, raw_options)), + owns_info_log_(options_.info_log != raw_options.info_log), + owns_cache_(options_.block_cache != raw_options.block_cache), + dbname_(dbname), + 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_(nullptr), + seed_(0), + tmp_batch_(new WriteBatch), + background_compaction_scheduled_(false), + manual_compaction_(nullptr), + versions_(new VersionSet(dbname_, &options_, table_cache_, + &internal_comparator_)) {} + +DBImpl::~DBImpl() { + // Wait for background work to finish. + mutex_.Lock(); + shutting_down_.store(true, std::memory_order_release); + while (background_compaction_scheduled_) { + background_work_finished_signal_.Wait(); + } + mutex_.Unlock(); + + if (db_lock_ != nullptr) { + env_->UnlockFile(db_lock_); + } + + delete versions_; + if (mem_ != nullptr) mem_->Unref(); + if (imm_ != nullptr) imm_->Unref(); + delete tmp_batch_; + delete log_; + delete logfile_; + delete table_cache_; + + if (owns_info_log_) { + delete options_.info_log; + } + if (owns_cache_) { + delete options_.block_cache; + } +} + +Status DBImpl::NewDB() { + VersionEdit new_db; + new_db.SetComparatorName(user_comparator()->Name()); + new_db.SetLogNumber(0); + new_db.SetNextFile(2); + new_db.SetLastSequence(0); + + const std::string manifest = DescriptorFileName(dbname_, 1); + WritableFile* file; + Status s = env_->NewWritableFile(manifest, &file); + if (!s.ok()) { + return s; + } + { + log::Writer log(file); + std::string record; + new_db.EncodeTo(&record); + s = log.AddRecord(record); + if (s.ok()) { + s = file->Sync(); + } + if (s.ok()) { + s = file->Close(); + } + } + delete file; + if (s.ok()) { + // Make "CURRENT" file that points to the new manifest file. + s = SetCurrentFile(env_, dbname_, 1); + } else { + env_->RemoveFile(manifest); + } + return s; +} + +void DBImpl::MaybeIgnoreError(Status* s) const { + if (s->ok() || options_.paranoid_checks) { + // No change needed + } else { + Log(options_.info_log, "Ignoring error %s", s->ToString().c_str()); + *s = Status::OK(); + } +} + +void DBImpl::RemoveObsoleteFiles() { + mutex_.AssertHeld(); + + if (!bg_error_.ok()) { + // After a background error, we don't know whether a new version may + // or may not have been committed, so we cannot safely garbage collect. + return; + } + + // Make a set of all of the live files + std::set<uint64_t> live = pending_outputs_; + versions_->AddLiveFiles(&live); + + std::vector<std::string> filenames; + env_->GetChildren(dbname_, &filenames); // Ignoring errors on purpose + uint64_t number; + FileType type; + std::vector<std::string> files_to_delete; + for (std::string& filename : filenames) { + if (ParseFileName(filename, &number, &type)) { + bool keep = true; + switch (type) { + case kLogFile: + keep = ((number >= versions_->LogNumber()) || + (number == versions_->PrevLogNumber())); + break; + case kDescriptorFile: + // Keep my manifest file, and any newer incarnations' + // (in case there is a race that allows other incarnations) + keep = (number >= versions_->ManifestFileNumber()); + break; + case kTableFile: + keep = (live.find(number) != live.end()); + break; + case kTempFile: + // Any temp files that are currently being written to must + // be recorded in pending_outputs_, which is inserted into "live" + keep = (live.find(number) != live.end()); + break; + case kCurrentFile: + case kDBLockFile: + case kInfoLogFile: + keep = true; + break; + } + + if (!keep) { + files_to_delete.push_back(std::move(filename)); + if (type == kTableFile) { + table_cache_->Evict(number); + } + Log(options_.info_log, "Delete type=%d #%lld\n", static_cast<int>(type), + static_cast<unsigned long long>(number)); + } + } + } + + // While deleting all files unblock other threads. All files being deleted + // have unique names which will not collide with newly created files and + // are therefore safe to delete while allowing other threads to proceed. + mutex_.Unlock(); + for (const std::string& filename : files_to_delete) { + env_->RemoveFile(dbname_ + "/" + filename); + } + mutex_.Lock(); +} + +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_ == nullptr); + Status s = env_->LockFile(LockFileName(dbname_), &db_lock_); + if (!s.ok()) { + return s; + } + + if (!env_->FileExists(CurrentFileName(dbname_))) { + if (options_.create_if_missing) { + Log(options_.info_log, "Creating DB %s since it was missing.", + dbname_.c_str()); + s = NewDB(); + if (!s.ok()) { + return s; + } + } else { + return Status::InvalidArgument( + dbname_, "does not exist (create_if_missing is false)"); + } + } else { + if (options_.error_if_exists) { + return Status::InvalidArgument(dbname_, + "exists (error_if_exists is true)"); + } + } + + 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<std::string> filenames; + s = env_->GetChildren(dbname_, &filenames); + if (!s.ok()) { + return s; + } + std::set<uint64_t> expected; + versions_->AddLiveFiles(&expected); + uint64_t number; + FileType type; + std::vector<uint64_t> 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]; + std::snprintf(buf, sizeof(buf), "%d missing files; e.g.", + static_cast<int>(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; + } + + // 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 (versions_->LastSequence() < max_sequence) { + versions_->SetLastSequence(max_sequence); + } + + return Status::OK(); +} + +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 + void Corruption(size_t bytes, const Status& s) override { + Log(info_log, "%s%s: dropping %d bytes; %s", + (this->status == nullptr ? "(ignoring error) " : ""), fname, + static_cast<int>(bytes), s.ToString().c_str()); + if (this->status != nullptr && this->status->ok()) *this->status = s; + } + }; + + mutex_.AssertHeld(); + + // Open the log file + std::string fname = LogFileName(dbname_, log_number); + SequentialFile* file; + Status status = env_->NewSequentialFile(fname, &file); + if (!status.ok()) { + MaybeIgnoreError(&status); + return status; + } + + // Create the log reader. + LogReporter reporter; + reporter.env = env_; + reporter.info_log = options_.info_log; + reporter.fname = fname.c_str(); + 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(options_.info_log, "Recovering log #%llu", + (unsigned long long)log_number); + + // Read all the records and add to a memtable + std::string scratch; + Slice record; + WriteBatch batch; + 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")); + continue; + } + WriteBatchInternal::SetContents(&batch, record); + + if (mem == nullptr) { + mem = new MemTable(internal_comparator_); + mem->Ref(); + } + status = WriteBatchInternal::InsertInto(&batch, mem); + MaybeIgnoreError(&status); + if (!status.ok()) { + break; + } + 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) { + 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; + } + } + } + + 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(); + } + + return status; +} + +Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit, + Version* base) { + mutex_.AssertHeld(); + const uint64_t start_micros = env_->NowMicros(); + FileMetaData meta; + meta.number = versions_->NewFileNumber(); + pending_outputs_.insert(meta.number); + Iterator* iter = mem->NewIterator(); + Log(options_.info_log, "Level-0 table #%llu: started", + (unsigned long long)meta.number); + + Status s; + { + mutex_.Unlock(); + s = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta); + mutex_.Lock(); + } + + Log(options_.info_log, "Level-0 table #%llu: %lld bytes %s", + (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 != nullptr) { + level = base->PickLevelForMemTableOutput(min_user_key, max_user_key); + } + edit->AddFile(level, meta.number, meta.file_size, meta.smallest, + meta.largest); + } + + CompactionStats stats; + stats.micros = env_->NowMicros() - start_micros; + stats.bytes_written = meta.file_size; + stats_[level].Add(stats); + return s; +} + +void DBImpl::CompactMemTable() { + mutex_.AssertHeld(); + assert(imm_ != nullptr); + + // Save the contents of the memtable as a new Table + VersionEdit edit; + Version* base = versions_->current(); + base->Ref(); + Status s = WriteLevel0Table(imm_, &edit, base); + base->Unref(); + + if (s.ok() && shutting_down_.load(std::memory_order_acquire)) { + s = Status::IOError("Deleting DB during memtable compaction"); + } + + // Replace immutable memtable with the generated Table + if (s.ok()) { + edit.SetPrevLogNumber(0); + edit.SetLogNumber(logfile_number_); // Earlier logs no longer needed + s = versions_->LogAndApply(&edit, &mutex_); + } + + if (s.ok()) { + // Commit to the new state + imm_->Unref(); + imm_ = nullptr; + has_imm_.store(false, std::memory_order_release); + RemoveObsoleteFiles(); + } else { + RecordBackgroundError(s); + } +} + +void DBImpl::CompactRange(const Slice* begin, const Slice* end) { + int max_level_with_files = 1; + { + MutexLock l(&mutex_); + Version* base = versions_->current(); + for (int level = 1; level < config::kNumLevels; level++) { + if (base->OverlapInLevel(level, begin, end)) { + max_level_with_files = level; + } + } + } + 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) { + assert(level >= 0); + assert(level + 1 < config::kNumLevels); + + InternalKey begin_storage, end_storage; + + ManualCompaction manual; + manual.level = level; + manual.done = false; + if (begin == nullptr) { + manual.begin = nullptr; + } else { + begin_storage = InternalKey(*begin, kMaxSequenceNumber, kValueTypeForSeek); + manual.begin = &begin_storage; + } + if (end == nullptr) { + manual.end = nullptr; + } else { + end_storage = InternalKey(*end, 0, static_cast<ValueType>(0)); + manual.end = &end_storage; + } + + MutexLock l(&mutex_); + 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. + background_work_finished_signal_.Wait(); + } + } + if (manual_compaction_ == &manual) { + // Cancel my manual compaction since we aborted early for some reason. + manual_compaction_ = nullptr; + } +} + +Status DBImpl::TEST_CompactMemTable() { + // 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_ != nullptr && bg_error_.ok()) { + background_work_finished_signal_.Wait(); + } + if (imm_ != nullptr) { + s = bg_error_; + } + } + return s; +} + +void DBImpl::RecordBackgroundError(const Status& s) { + mutex_.AssertHeld(); + if (bg_error_.ok()) { + bg_error_ = s; + background_work_finished_signal_.SignalAll(); + } +} + +void DBImpl::MaybeScheduleCompaction() { + mutex_.AssertHeld(); + if (background_compaction_scheduled_) { + // Already scheduled + } 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_ == nullptr && manual_compaction_ == nullptr && + !versions_->NeedsCompaction()) { + // No work to be done + } else { + background_compaction_scheduled_ = true; + env_->Schedule(&DBImpl::BGWork, this); + } +} + +void DBImpl::BGWork(void* db) { + reinterpret_cast<DBImpl*>(db)->BackgroundCall(); +} + +void DBImpl::BackgroundCall() { + MutexLock l(&mutex_); + 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. + } else { + BackgroundCompaction(); + } + + background_compaction_scheduled_ = false; + + // Previous compaction may have produced too many files in a level, + // so reschedule another compaction if needed. + MaybeScheduleCompaction(); + background_work_finished_signal_.SignalAll(); +} + +void DBImpl::BackgroundCompaction() { + mutex_.AssertHeld(); + + if (imm_ != nullptr) { + CompactMemTable(); + return; + } + + Compaction* c; + 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 == 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->end ? m->end->DebugString().c_str() : "(end)"), + (m->done ? "(end)" : manual_end.DebugString().c_str())); + } else { + c = versions_->PickCompaction(); + } + + Status status; + 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()->RemoveFile(c->level(), f->number); + 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<unsigned long long>(f->number), c->level() + 1, + static_cast<unsigned long long>(f->file_size), + status.ToString().c_str(), versions_->LevelSummary(&tmp)); + } else { + CompactionState* compact = new CompactionState(c); + status = DoCompactionWork(compact); + if (!status.ok()) { + RecordBackgroundError(status); + } + CleanupCompaction(compact); + c->ReleaseInputs(); + RemoveObsoleteFiles(); + } + delete c; + + if (status.ok()) { + // Done + } 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()); + } + + if (is_manual) { + ManualCompaction* m = manual_compaction_; + if (!status.ok()) { + m->done = true; + } + if (!m->done) { + // We only compacted part of the requested range. Update *m + // to the range that is left to be compacted. + m->tmp_storage = manual_end; + m->begin = &m->tmp_storage; + } + manual_compaction_ = nullptr; + } +} + +void DBImpl::CleanupCompaction(CompactionState* compact) { + mutex_.AssertHeld(); + 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 == nullptr); + } + delete compact->outfile; + for (size_t i = 0; i < compact->outputs.size(); i++) { + const CompactionState::Output& out = compact->outputs[i]; + pending_outputs_.erase(out.number); + } + delete compact; +} + +Status DBImpl::OpenCompactionOutputFile(CompactionState* compact) { + assert(compact != nullptr); + assert(compact->builder == nullptr); + uint64_t file_number; + { + mutex_.Lock(); + file_number = versions_->NewFileNumber(); + pending_outputs_.insert(file_number); + CompactionState::Output out; + out.number = file_number; + out.smallest.Clear(); + out.largest.Clear(); + compact->outputs.push_back(out); + mutex_.Unlock(); + } + + // Make the output file + std::string fname = TableFileName(dbname_, file_number); + Status s = env_->NewWritableFile(fname, &compact->outfile); + if (s.ok()) { + compact->builder = new TableBuilder(options_, compact->outfile); + } + return s; +} + +Status DBImpl::FinishCompactionOutputFile(CompactionState* compact, + Iterator* input) { + assert(compact != nullptr); + assert(compact->outfile != nullptr); + assert(compact->builder != nullptr); + + const uint64_t output_number = compact->current_output()->number; + assert(output_number != 0); + + // Check for iterator errors + Status s = input->status(); + const uint64_t current_entries = compact->builder->NumEntries(); + if (s.ok()) { + s = compact->builder->Finish(); + } else { + compact->builder->Abandon(); + } + const uint64_t current_bytes = compact->builder->FileSize(); + compact->current_output()->file_size = current_bytes; + compact->total_bytes += current_bytes; + delete compact->builder; + compact->builder = nullptr; + + // Finish and check for file errors + if (s.ok()) { + s = compact->outfile->Sync(); + } + if (s.ok()) { + s = compact->outfile->Close(); + } + delete compact->outfile; + 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); + s = iter->status(); + delete iter; + if (s.ok()) { + 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, + static_cast<long long>(compact->total_bytes)); + + // Add compaction outputs + compact->compaction->AddInputDeletions(compact->compaction->edit()); + 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); + } + return versions_->LogAndApply(compact->compaction->edit(), &mutex_); +} + +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(), + compact->compaction->num_input_files(1), + compact->compaction->level() + 1); + + assert(versions_->NumLevelFiles(compact->compaction->level()) > 0); + assert(compact->builder == nullptr); + assert(compact->outfile == nullptr); + if (snapshots_.empty()) { + compact->smallest_snapshot = versions_->LastSequence(); + } else { + 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(); + + input->SeekToFirst(); + Status status; + ParsedInternalKey ikey; + std::string current_user_key; + bool has_current_user_key = false; + SequenceNumber last_sequence_for_key = kMaxSequenceNumber; + while (input->Valid() && !shutting_down_.load(std::memory_order_acquire)) { + // Prioritize immutable compaction work + if (has_imm_.load(std::memory_order_relaxed)) { + const uint64_t imm_start = env_->NowMicros(); + mutex_.Lock(); + if (imm_ != nullptr) { + CompactMemTable(); + // Wake up MakeRoomForWrite() if necessary. + background_work_finished_signal_.SignalAll(); + } + mutex_.Unlock(); + imm_micros += (env_->NowMicros() - imm_start); + } + + Slice key = input->key(); + if (compact->compaction->ShouldStopBefore(key) && + compact->builder != nullptr) { + status = FinishCompactionOutputFile(compact, input); + if (!status.ok()) { + break; + } + } + + // Handle key/value, add to state, etc. + bool drop = false; + if (!ParseInternalKey(key, &ikey)) { + // Do not hide error keys + current_user_key.clear(); + has_current_user_key = false; + last_sequence_for_key = kMaxSequenceNumber; + } else { + if (!has_current_user_key || + 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; + last_sequence_for_key = kMaxSequenceNumber; + } + + if (last_sequence_for_key <= compact->smallest_snapshot) { + // Hidden by an newer entry for same user key + drop = true; // (A) + } else if (ikey.type == kTypeDeletion && + ikey.sequence <= compact->smallest_snapshot && + compact->compaction->IsBaseLevelForKey(ikey.user_key)) { + // For this user key: + // (1) there is no data in higher levels + // (2) data in lower levels will have larger sequence numbers + // (3) data in layers that are being compacted here and have + // smaller sequence numbers will be dropped in the next + // few iterations of this loop (by rule (A) above). + // Therefore this deletion marker is obsolete and can be dropped. + drop = true; + } + + last_sequence_for_key = ikey.sequence; + } +#if 0 + Log(options_.info_log, + " Compact: %s, seq %d, type: %d %d, drop: %d, is_base: %d, " + "%d smallest_snapshot: %d", + ikey.user_key.ToString().c_str(), + (int)ikey.sequence, ikey.type, kTypeValue, drop, + compact->compaction->IsBaseLevelForKey(ikey.user_key), + (int)last_sequence_for_key, (int)compact->smallest_snapshot); +#endif + + if (!drop) { + // Open output file if necessary + if (compact->builder == nullptr) { + status = OpenCompactionOutputFile(compact); + if (!status.ok()) { + break; + } + } + if (compact->builder->NumEntries() == 0) { + compact->current_output()->smallest.DecodeFrom(key); + } + compact->current_output()->largest.DecodeFrom(key); + compact->builder->Add(key, input->value()); + + // Close output file if it is big enough + if (compact->builder->FileSize() >= + compact->compaction->MaxOutputFileSize()) { + status = FinishCompactionOutputFile(compact, input); + if (!status.ok()) { + break; + } + } + } + + input->Next(); + } + + if (status.ok() && shutting_down_.load(std::memory_order_acquire)) { + status = Status::IOError("Deleting DB during compaction"); + } + if (status.ok() && compact->builder != nullptr) { + status = FinishCompactionOutputFile(compact, input); + } + if (status.ok()) { + status = input->status(); + } + delete input; + input = nullptr; + + CompactionStats stats; + stats.micros = env_->NowMicros() - start_micros - imm_micros; + for (int which = 0; which < 2; which++) { + for (int i = 0; i < compact->compaction->num_input_files(which); i++) { + stats.bytes_read += compact->compaction->input(which, i)->file_size; + } + } + for (size_t i = 0; i < compact->outputs.size(); i++) { + stats.bytes_written += compact->outputs[i].file_size; + } + + mutex_.Lock(); + stats_[compact->compaction->level() + 1].Add(stats); + + if (status.ok()) { + status = InstallCompactionResults(compact); + } + if (!status.ok()) { + RecordBackgroundError(status); + } + VersionSet::LevelSummaryStorage tmp; + Log(options_.info_log, "compacted to: %s", versions_->LevelSummary(&tmp)); + return status; +} + +namespace { + +struct IterState { + 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<IterState*>(arg1); + state->mu->Lock(); + state->mem->Unref(); + if (state->imm != nullptr) state->imm->Unref(); + state->version->Unref(); + state->mu->Unlock(); + delete state; +} + +} // anonymous namespace + +Iterator* DBImpl::NewInternalIterator(const ReadOptions& options, + SequenceNumber* latest_snapshot, + uint32_t* seed) { + mutex_.Lock(); + *latest_snapshot = versions_->LastSequence(); + + // Collect together all needed child iterators + std::vector<Iterator*> list; + list.push_back(mem_->NewIterator()); + mem_->Ref(); + if (imm_ != nullptr) { + list.push_back(imm_->NewIterator()); + imm_->Ref(); + } + versions_->current()->AddIterators(options, &list); + Iterator* internal_iter = + NewMergingIterator(&internal_comparator_, &list[0], list.size()); + versions_->current()->Ref(); + + IterState* cleanup = new IterState(&mutex_, mem_, imm_, versions_->current()); + internal_iter->RegisterCleanup(CleanupIteratorState, cleanup, nullptr); + + *seed = ++seed_; + mutex_.Unlock(); + return internal_iter; +} + +Iterator* DBImpl::TEST_NewInternalIterator() { + SequenceNumber ignored; + uint32_t ignored_seed; + return NewInternalIterator(ReadOptions(), &ignored, &ignored_seed); +} + +int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes() { + MutexLock l(&mutex_); + return versions_->MaxNextLevelOverlappingBytes(); +} + +Status DBImpl::Get(const ReadOptions& options, const Slice& key, + std::string* value) { + Status s; + MutexLock l(&mutex_); + SequenceNumber snapshot; + if (options.snapshot != nullptr) { + snapshot = + static_cast<const SnapshotImpl*>(options.snapshot)->sequence_number(); + } else { + snapshot = versions_->LastSequence(); + } + + MemTable* mem = mem_; + MemTable* imm = imm_; + Version* current = versions_->current(); + mem->Ref(); + if (imm != nullptr) imm->Ref(); + current->Ref(); + + bool have_stat_update = false; + Version::GetStats stats; + + // Unlock while reading from files and memtables + { + mutex_.Unlock(); + // First look in the memtable, then in the immutable memtable (if any). + LookupKey lkey(key, snapshot); + if (mem->Get(lkey, value, &s)) { + // Done + } else if (imm != nullptr && imm->Get(lkey, value, &s)) { + // Done + } else { + s = current->Get(options, lkey, value, &stats); + have_stat_update = true; + } + mutex_.Lock(); + } + + if (have_stat_update && current->UpdateStats(stats)) { + MaybeScheduleCompaction(); + } + mem->Unref(); + if (imm != nullptr) imm->Unref(); + current->Unref(); + return s; +} + +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 != nullptr + ? static_cast<const SnapshotImpl*>(options.snapshot) + ->sequence_number() + : latest_snapshot), + seed); +} + +void DBImpl::RecordReadSample(Slice key) { + MutexLock l(&mutex_); + if (versions_->current()->RecordReadSample(key)) { + MaybeScheduleCompaction(); + } +} + +const Snapshot* DBImpl::GetSnapshot() { + MutexLock l(&mutex_); + return snapshots_.New(versions_->LastSequence()); +} + +void DBImpl::ReleaseSnapshot(const Snapshot* snapshot) { + MutexLock l(&mutex_); + snapshots_.Delete(static_cast<const SnapshotImpl*>(snapshot)); +} + +// Convenience methods +Status DBImpl::Put(const WriteOptions& o, const Slice& key, const Slice& val) { + return DB::Put(o, key, val); +} + +Status DBImpl::Delete(const WriteOptions& options, const Slice& key) { + return DB::Delete(options, key); +} + +Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) { + Writer w(&mutex_); + w.batch = updates; + w.sync = options.sync; + w.done = false; + + MutexLock l(&mutex_); + writers_.push_back(&w); + while (!w.done && &w != writers_.front()) { + w.cv.Wait(); + } + if (w.done) { + return w.status; + } + + // May temporarily unlock and wait. + Status status = MakeRoomForWrite(updates == nullptr); + uint64_t last_sequence = versions_->LastSequence(); + Writer* last_writer = &w; + if (status.ok() && updates != nullptr) { // nullptr batch is for compactions + WriteBatch* write_batch = BuildBatchGroup(&last_writer); + WriteBatchInternal::SetSequence(write_batch, last_sequence + 1); + last_sequence += WriteBatchInternal::Count(write_batch); + + // Add to log and apply to memtable. We can release the lock + // during this phase since &w is currently responsible for logging + // and protects against concurrent loggers and concurrent writes + // into mem_. + { + mutex_.Unlock(); + status = log_->AddRecord(WriteBatchInternal::Contents(write_batch)); + bool sync_error = false; + if (status.ok() && options.sync) { + status = logfile_->Sync(); + if (!status.ok()) { + sync_error = true; + } + } + if (status.ok()) { + status = WriteBatchInternal::InsertInto(write_batch, mem_); + } + mutex_.Lock(); + if (sync_error) { + // The state of the log file is indeterminate: the log record we + // just added may or may not show up when the DB is re-opened. + // So we force the DB into a mode where all future writes fail. + RecordBackgroundError(status); + } + } + if (write_batch == tmp_batch_) tmp_batch_->Clear(); + + versions_->SetLastSequence(last_sequence); + } + + while (true) { + Writer* ready = writers_.front(); + writers_.pop_front(); + if (ready != &w) { + ready->status = status; + ready->done = true; + ready->cv.Signal(); + } + if (ready == last_writer) break; + } + + // Notify new head of write queue + if (!writers_.empty()) { + writers_.front()->cv.Signal(); + } + + return status; +} + +// REQUIRES: Writer list must be non-empty +// 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 != nullptr); + + size_t size = WriteBatchInternal::ByteSize(first->batch); + + // Allow the group to grow up to a maximum size, but if the + // original write is small, limit the growth so we do not slow + // down the small write too much. + size_t max_size = 1 << 20; + if (size <= (128 << 10)) { + max_size = size + (128 << 10); + } + + *last_writer = first; + std::deque<Writer*>::iterator iter = writers_.begin(); + ++iter; // Advance past "first" + for (; iter != writers_.end(); ++iter) { + Writer* w = *iter; + if (w->sync && !first->sync) { + // Do not include a sync write into a batch handled by a non-sync write. + break; + } + + if (w->batch != nullptr) { + size += WriteBatchInternal::ByteSize(w->batch); + if (size > max_size) { + // Do not make batch too big + break; + } + + // Append to *result + if (result == first->batch) { + // Switch to temporary batch instead of disturbing caller's batch + result = tmp_batch_; + assert(WriteBatchInternal::Count(result) == 0); + WriteBatchInternal::Append(result, first->batch); + } + WriteBatchInternal::Append(result, w->batch); + } + *last_writer = w; + } + return result; +} + +// REQUIRES: mutex_ is held +// REQUIRES: this thread is currently at the front of the writer queue +Status DBImpl::MakeRoomForWrite(bool force) { + mutex_.AssertHeld(); + assert(!writers_.empty()); + bool allow_delay = !force; + Status s; + while (true) { + if (!bg_error_.ok()) { + // Yield previous error + s = bg_error_; + break; + } 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 + // individual write by 1ms to reduce latency variance. Also, + // this delay hands over some CPU to the compaction thread in + // case it is sharing the same core as the writer. + mutex_.Unlock(); + env_->SleepForMicroseconds(1000); + allow_delay = false; // Do not delay a single write more than once + mutex_.Lock(); + } else if (!force && + (mem_->ApproximateMemoryUsage() <= options_.write_buffer_size)) { + // There is room in current memtable + break; + } 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"); + 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"); + 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 = nullptr; + s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile); + if (!s.ok()) { + // Avoid chewing through file number space in a tight loop. + versions_->ReuseFileNumber(new_log_number); + break; + } + delete log_; + delete logfile_; + logfile_ = lfile; + logfile_number_ = new_log_number; + log_ = new log::Writer(lfile); + imm_ = mem_; + 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 + MaybeScheduleCompaction(); + } + } + return s; +} + +bool DBImpl::GetProperty(const Slice& property, std::string* value) { + value->clear(); + + MutexLock l(&mutex_); + Slice in = property; + Slice prefix("leveldb."); + if (!in.starts_with(prefix)) return false; + in.remove_prefix(prefix.size()); + + if (in.starts_with("num-files-at-level")) { + in.remove_prefix(strlen("num-files-at-level")); + uint64_t level; + bool ok = ConsumeDecimalNumber(&in, &level) && in.empty(); + if (!ok || level >= config::kNumLevels) { + return false; + } else { + char buf[100]; + std::snprintf(buf, sizeof(buf), "%d", + versions_->NumLevelFiles(static_cast<int>(level))); + *value = buf; + return true; + } + } else if (in == "stats") { + char buf[200]; + std::snprintf(buf, sizeof(buf), + " Compactions\n" + "Level Files Size(MB) Time(sec) Read(MB) Write(MB)\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) { + std::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); + } + } + return true; + } 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]; + std::snprintf(buf, sizeof(buf), "%llu", + static_cast<unsigned long long>(total_usage)); + value->append(buf); + return true; + } + + return false; +} + +void DBImpl::GetApproximateSizes(const Range* range, int n, uint64_t* sizes) { + // TODO(opt): better implementation + MutexLock l(&mutex_); + Version* v = versions_->current(); + v->Ref(); + + for (int i = 0; i < n; i++) { + // Convert user_key into a corresponding internal key. + InternalKey k1(range[i].start, kMaxSequenceNumber, kValueTypeForSeek); + InternalKey k2(range[i].limit, kMaxSequenceNumber, kValueTypeForSeek); + uint64_t start = versions_->ApproximateOffsetOf(v, k1); + uint64_t limit = versions_->ApproximateOffsetOf(v, k2); + sizes[i] = (limit >= start ? limit - start : 0); + } + + v->Unref(); +} + +// Default implementations of convenience methods that subclasses of DB +// can call if they wish +Status DB::Put(const WriteOptions& opt, const Slice& key, const Slice& value) { + WriteBatch batch; + batch.Put(key, value); + return Write(opt, &batch); +} + +Status DB::Delete(const WriteOptions& opt, const Slice& key) { + WriteBatch batch; + batch.Delete(key); + return Write(opt, &batch); +} + +DB::~DB() = default; + +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; + // 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), + &lfile); + if (s.ok()) { + edit.SetLogNumber(new_log_number); + impl->logfile_ = lfile; + impl->logfile_number_ = new_log_number; + impl->log_ = new log::Writer(lfile); + 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->RemoveObsoleteFiles(); + impl->MaybeScheduleCompaction(); + } + impl->mutex_.Unlock(); + if (s.ok()) { + assert(impl->mem_ != nullptr); + *dbptr = impl; + } else { + delete impl; + } + return s; +} + +Snapshot::~Snapshot() = default; + +Status DestroyDB(const std::string& dbname, const Options& options) { + Env* env = options.env; + std::vector<std::string> filenames; + 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); + result = env->LockFile(lockname, &lock); + if (result.ok()) { + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && + type != kDBLockFile) { // Lock file will be deleted at end + Status del = env->RemoveFile(dbname + "/" + filenames[i]); + if (result.ok() && !del.ok()) { + result = del; + } + } + } + env->UnlockFile(lock); // Ignore error since state is already gone + env->RemoveFile(lockname); + env->RemoveDir(dbname); // Ignore error in case dir contains other files + } + return result; +} + +} // namespace leveldb diff --git a/db/db_impl.h b/db/db_impl.h new file mode 100644 index 0000000..c7b0172 --- /dev/null +++ b/db/db_impl.h @@ -0,0 +1,217 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_DB_IMPL_H_ +#define STORAGE_LEVELDB_DB_DB_IMPL_H_ + +#include <atomic> +#include <deque> +#include <set> +#include <string> + +#include "db/dbformat.h" +#include "db/log_writer.h" +#include "db/snapshot.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "port/port.h" +#include "port/thread_annotations.h" + +namespace leveldb { + +class MemTable; +class TableCache; +class Version; +class VersionEdit; +class VersionSet; + +class DBImpl : public DB { + public: + DBImpl(const Options& options, const std::string& dbname); + + DBImpl(const DBImpl&) = delete; + DBImpl& operator=(const DBImpl&) = delete; + + ~DBImpl() override; + + // Implementations of the DB interface + 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 + + // Compact any files in the named level that overlap [*begin,*end] + void TEST_CompactRange(int level, const Slice* begin, const Slice* end); + + // Force current memtable contents to be compacted. + Status TEST_CompactMemTable(); + + // Return an internal iterator over the current state of the database. + // The keys of this iterator are internal keys (see format.h). + // The returned iterator should be deleted when no longer needed. + Iterator* TEST_NewInternalIterator(); + + // Return the maximum overlapping data (in bytes) at next level for any + // file at a level >= 1. + int64_t TEST_MaxNextLevelOverlappingBytes(); + + // Record a sample of bytes read at the specified internal key. + // Samples are taken approximately once every config::kReadBytesPeriod + // bytes. + void RecordReadSample(Slice key); + + private: + friend class 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); + + Status NewDB(); + + // Recover the descriptor from persistent storage. May do a significant + // amount of work to recover recently logged updates. Any changes to + // be made to the descriptor are added to *edit. + Status Recover(VersionEdit* edit, bool* save_manifest) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + void MaybeIgnoreError(Status* s) const; + + // Delete any unneeded files and stale in-memory entries. + void RemoveObsoleteFiles() 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, bool last_log, bool* save_manifest, + VersionEdit* edit, SequenceNumber* max_sequence) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status WriteLevel0Table(MemTable* mem, VersionEdit* edit, Version* base) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status MakeRoomForWrite(bool force /* compact even if there is room? */) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + 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 CleanupCompaction(CompactionState* compact) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + Status DoCompactionWork(CompactionState* compact) + EXCLUSIVE_LOCKS_REQUIRED(mutex_); + + Status OpenCompactionOutputFile(CompactionState* compact); + Status FinishCompactionOutputFile(CompactionState* compact, Iterator* input); + 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_ + const bool owns_info_log_; + const bool owns_cache_; + const std::string dbname_; + + // table_cache_ provides its own synchronization + TableCache* const table_cache_; + + // Lock over the persistent DB state. Non-null iff successfully acquired. + FileLock* db_lock_; + + // State below is protected by mutex_ + port::Mutex mutex_; + std::atomic<bool> shutting_down_; + port::CondVar background_work_finished_signal_ GUARDED_BY(mutex_); + MemTable* mem_; + MemTable* imm_ GUARDED_BY(mutex_); // Memtable being compacted + std::atomic<bool> has_imm_; // So bg thread can detect non-null imm_ + WritableFile* logfile_; + uint64_t logfile_number_ GUARDED_BY(mutex_); + log::Writer* log_; + uint32_t seed_ GUARDED_BY(mutex_); // For sampling. + + // Queue of writers. + std::deque<Writer*> writers_ GUARDED_BY(mutex_); + WriteBatch* tmp_batch_ GUARDED_BY(mutex_); + + SnapshotList snapshots_ GUARDED_BY(mutex_); + + // Set of table files to protect from deletion because they are + // part of ongoing compactions. + std::set<uint64_t> pending_outputs_ GUARDED_BY(mutex_); + + // Has a background compaction been scheduled or is running? + bool background_compaction_scheduled_ GUARDED_BY(mutex_); + + ManualCompaction* manual_compaction_ GUARDED_BY(mutex_); + + VersionSet* const versions_ GUARDED_BY(mutex_); + + // Have we encountered a background error in paranoid mode? + Status bg_error_ GUARDED_BY(mutex_); + + 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. +Options SanitizeOptions(const std::string& db, + const InternalKeyComparator* icmp, + const InternalFilterPolicy* ipolicy, + const Options& src); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_DB_IMPL_H_ diff --git a/db/db_iter.cc b/db/db_iter.cc new file mode 100644 index 0000000..532c2db --- /dev/null +++ b/db/db_iter.cc @@ -0,0 +1,318 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/db_iter.h" + +#include "db/db_impl.h" +#include "db/dbformat.h" +#include "db/filename.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" +#include "port/port.h" +#include "util/logging.h" +#include "util/mutexlock.h" +#include "util/random.h" + +namespace leveldb { + +#if 0 +static void DumpInternalIter(Iterator* iter) { + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + ParsedInternalKey k; + if (!ParseInternalKey(iter->key(), &k)) { + std::fprintf(stderr, "Corrupt '%s'\n", EscapeString(iter->key()).c_str()); + } else { + std::fprintf(stderr, "@ '%s'\n", k.DebugString().c_str()); + } + } +} +#endif + +namespace { + +// Memtables and sstables that make the DB representation contain +// (userkey,seq,type) => uservalue entries. DBIter +// combines multiple entries for the same userkey found in the DB +// representation into a single entry while accounting for sequence +// numbers, deletion markers, overwrites, etc. +class DBIter : public Iterator { + public: + // Which direction is the iterator currently moving? + // (1) When moving forward, the internal iterator is positioned at + // the exact entry that yields this->key(), this->value() + // (2) When moving backwards, the internal iterator is positioned + // just before all entries whose user key == this->key(). + enum Direction { kForward, kReverse }; + + DBIter(DBImpl* db, const Comparator* cmp, Iterator* iter, SequenceNumber s, + uint32_t seed) + : db_(db), + user_comparator_(cmp), + iter_(iter), + sequence_(s), + direction_(kForward), + valid_(false), + rnd_(seed), + 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_; + } + Slice value() const override { + assert(valid_); + return (direction_ == kForward) ? iter_->value() : saved_value_; + } + Status status() const override { + if (status_.ok()) { + return iter_->status(); + } else { + return status_; + } + } + + 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); + void FindPrevUserEntry(); + bool ParseKey(ParsedInternalKey* key); + + inline void SaveKey(const Slice& k, std::string* dst) { + dst->assign(k.data(), k.size()); + } + + inline void ClearSavedValue() { + if (saved_value_.capacity() > 1048576) { + std::string empty; + swap(empty, saved_value_); + } else { + saved_value_.clear(); + } + } + + // 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 + Direction direction_; + bool valid_; + Random rnd_; + size_t bytes_until_read_sampling_; +}; + +inline bool DBIter::ParseKey(ParsedInternalKey* ikey) { + Slice k = iter_->key(); + + 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; + } else { + return true; + } +} + +void DBIter::Next() { + assert(valid_); + + if (direction_ == kReverse) { // Switch directions? + direction_ = kForward; + // iter_ is pointing just before the entries for this->key(), + // so advance into the range of entries for this->key() and then + // use the normal skipping code below. + if (!iter_->Valid()) { + iter_->SeekToFirst(); + } else { + iter_->Next(); + } + if (!iter_->Valid()) { + valid_ = false; + saved_key_.clear(); + return; + } + // saved_key_ already contains the key to skip past. + } 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_); +} + +void DBIter::FindNextUserEntry(bool skipping, std::string* skip) { + // Loop until we hit an acceptable entry to yield + assert(iter_->Valid()); + assert(direction_ == kForward); + do { + ParsedInternalKey ikey; + if (ParseKey(&ikey) && ikey.sequence <= sequence_) { + switch (ikey.type) { + case kTypeDeletion: + // Arrange to skip all upcoming entries for this key since + // they are hidden by this deletion. + SaveKey(ikey.user_key, skip); + skipping = true; + break; + case kTypeValue: + if (skipping && + user_comparator_->Compare(ikey.user_key, *skip) <= 0) { + // Entry hidden + } else { + valid_ = true; + saved_key_.clear(); + return; + } + break; + } + } + iter_->Next(); + } while (iter_->Valid()); + saved_key_.clear(); + valid_ = false; +} + +void DBIter::Prev() { + assert(valid_); + + if (direction_ == kForward) { // Switch directions? + // iter_ is pointing at the current entry. Scan backwards until + // the key changes so we can use the normal reverse scanning code. + assert(iter_->Valid()); // Otherwise valid_ would have been false + SaveKey(ExtractUserKey(iter_->key()), &saved_key_); + while (true) { + iter_->Prev(); + if (!iter_->Valid()) { + valid_ = false; + saved_key_.clear(); + ClearSavedValue(); + return; + } + if (user_comparator_->Compare(ExtractUserKey(iter_->key()), saved_key_) < + 0) { + break; + } + } + direction_ = kReverse; + } + + FindPrevUserEntry(); +} + +void DBIter::FindPrevUserEntry() { + assert(direction_ == kReverse); + + ValueType value_type = kTypeDeletion; + if (iter_->Valid()) { + do { + ParsedInternalKey ikey; + if (ParseKey(&ikey) && ikey.sequence <= sequence_) { + if ((value_type != kTypeDeletion) && + user_comparator_->Compare(ikey.user_key, saved_key_) < 0) { + // We encountered a non-deleted value in entries for previous keys, + break; + } + value_type = ikey.type; + if (value_type == kTypeDeletion) { + saved_key_.clear(); + ClearSavedValue(); + } else { + Slice raw_value = iter_->value(); + if (saved_value_.capacity() > raw_value.size() + 1048576) { + std::string empty; + swap(empty, saved_value_); + } + SaveKey(ExtractUserKey(iter_->key()), &saved_key_); + saved_value_.assign(raw_value.data(), raw_value.size()); + } + } + iter_->Prev(); + } while (iter_->Valid()); + } + + if (value_type == kTypeDeletion) { + // End + valid_ = false; + saved_key_.clear(); + ClearSavedValue(); + direction_ = kForward; + } else { + valid_ = true; + } +} + +void DBIter::Seek(const Slice& target) { + direction_ = kForward; + ClearSavedValue(); + saved_key_.clear(); + AppendInternalKey(&saved_key_, + ParsedInternalKey(target, sequence_, kValueTypeForSeek)); + iter_->Seek(saved_key_); + if (iter_->Valid()) { + FindNextUserEntry(false, &saved_key_ /* temporary storage */); + } else { + valid_ = false; + } +} + +void DBIter::SeekToFirst() { + direction_ = kForward; + ClearSavedValue(); + iter_->SeekToFirst(); + if (iter_->Valid()) { + FindNextUserEntry(false, &saved_key_ /* temporary storage */); + } else { + valid_ = false; + } +} + +void DBIter::SeekToLast() { + direction_ = kReverse; + ClearSavedValue(); + iter_->SeekToLast(); + FindPrevUserEntry(); +} + +} // anonymous namespace + +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); +} + +} // namespace leveldb diff --git a/db/db_iter.h b/db/db_iter.h new file mode 100644 index 0000000..5977fc8 --- /dev/null +++ b/db/db_iter.h @@ -0,0 +1,26 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_DB_ITER_H_ +#define STORAGE_LEVELDB_DB_DB_ITER_H_ + +#include <cstdint> + +#include "db/dbformat.h" +#include "leveldb/db.h" + +namespace leveldb { + +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. +Iterator* NewDBIterator(DBImpl* db, const Comparator* user_key_comparator, + Iterator* internal_iter, SequenceNumber sequence, + uint32_t seed); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_DB_ITER_H_ diff --git a/db/db_test.cc b/db/db_test.cc new file mode 100644 index 0000000..fd60d38 --- /dev/null +++ b/db/db_test.cc @@ -0,0 +1,2361 @@ +// 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 "leveldb/db.h" + +#include <atomic> +#include <cinttypes> +#include <string> + +#include "gtest/gtest.h" +#include "benchmark/benchmark.h" +#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" +#include "util/testutil.h" + +namespace leveldb { + +static std::string RandomString(Random* rnd, int len) { + std::string r; + test::RandomString(rnd, len, &r); + 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 { + public: + AtomicCounter() : count_(0) {} + void Increment() { IncrementBy(1); } + void IncrementBy(int count) LOCKS_EXCLUDED(mu_) { + MutexLock l(&mu_); + count_ += count; + } + int Read() LOCKS_EXCLUDED(mu_) { + MutexLock l(&mu_); + return count_; + } + 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 + +// 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<std::string>* result) override { + Status s = target()->GetChildren(dir, result); + if (!s.ok() || !ignore_dot_files_) { + return s; + } + + std::vector<std::string>::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. + std::atomic<bool> delay_data_sync_; + + // sstable/log Sync() calls return an error. + std::atomic<bool> data_sync_error_; + + // Simulate no-space errors while this pointer is non-null. + std::atomic<bool> no_space_; + + // Simulate non-writable file system while this pointer is non-null. + std::atomic<bool> non_writable_; + + // Force sync of manifest files to fail while this pointer is non-null. + std::atomic<bool> manifest_sync_error_; + + // Force write to manifest files to fail while this pointer is non-null. + std::atomic<bool> manifest_write_error_; + + bool count_random_reads_; + AtomicCounter random_read_counter_; + + 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* const env_; + WritableFile* const base_; + + public: + DataFile(SpecialEnv* env, WritableFile* base) : env_(env), base_(base) {} + ~DataFile() { delete base_; } + Status Append(const Slice& data) { + if (env_->no_space_.load(std::memory_order_acquire)) { + // Drop writes on the floor + return Status::OK(); + } else { + return base_->Append(data); + } + } + Status Close() { return base_->Close(); } + Status Flush() { return base_->Flush(); } + Status Sync() { + if (env_->data_sync_error_.load(std::memory_order_acquire)) { + return Status::IOError("simulated data sync error"); + } + while (env_->delay_data_sync_.load(std::memory_order_acquire)) { + DelayMilliseconds(100); + } + return base_->Sync(); + } + }; + class ManifestFile : public WritableFile { + private: + SpecialEnv* env_; + WritableFile* base_; + + public: + ManifestFile(SpecialEnv* env, WritableFile* b) : env_(env), base_(b) {} + ~ManifestFile() { delete base_; } + Status Append(const Slice& data) { + if (env_->manifest_write_error_.load(std::memory_order_acquire)) { + return Status::IOError("simulated writer error"); + } else { + return base_->Append(data); + } + } + Status Close() { return base_->Close(); } + Status Flush() { return base_->Flush(); } + Status Sync() { + if (env_->manifest_sync_error_.load(std::memory_order_acquire)) { + return Status::IOError("simulated sync error"); + } else { + return base_->Sync(); + } + } + }; + + 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") != nullptr || + strstr(f.c_str(), ".log") != nullptr) { + *r = new DataFile(this, *r); + } else if (strstr(f.c_str(), "MANIFEST") != nullptr) { + *r = new ManifestFile(this, *r); + } + } + return s; + } + + Status NewRandomAccessFile(const std::string& f, RandomAccessFile** r) { + class CountingFile : public RandomAccessFile { + private: + RandomAccessFile* target_; + AtomicCounter* counter_; + + public: + CountingFile(RandomAccessFile* target, AtomicCounter* counter) + : 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); + } + }; + + Status s = target()->NewRandomAccessFile(f, r); + if (s.ok() && count_random_reads_) { + *r = new CountingFile(*r, &random_read_counter_); + } + return s; + } +}; + +class DBTest : public testing::Test { + public: + std::string dbname_; + SpecialEnv* env_; + DB* db_; + + Options last_options_; + + DBTest() : env_(new SpecialEnv(Env::Default())), option_config_(kDefault) { + filter_policy_ = NewBloomFilterPolicy(10); + dbname_ = testing::TempDir() + "db_test"; + DestroyDB(dbname_, Options()); + db_ = nullptr; + Reopen(); + } + + ~DBTest() { + delete db_; + DestroyDB(dbname_, Options()); + delete env_; + delete filter_policy_; + } + + // Switch to a fresh database with the next option configuration to + // test. Return false if there are no more configurations to test. + bool ChangeOptions() { + option_config_++; + if (option_config_ >= kEnd) { + return false; + } else { + DestroyAndReopen(); + return true; + } + } + + // 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; + case kUncompressed: + options.compression = kNoCompression; + break; + default: + break; + } + return options; + } + + DBImpl* dbfull() { return reinterpret_cast<DBImpl*>(db_); } + + void Reopen(Options* options = nullptr) { + ASSERT_LEVELDB_OK(TryReopen(options)); + } + + void Close() { + delete db_; + db_ = nullptr; + } + + void DestroyAndReopen(Options* options = nullptr) { + delete db_; + db_ = nullptr; + DestroyDB(dbname_, Options()); + ASSERT_LEVELDB_OK(TryReopen(options)); + } + + Status TryReopen(Options* options) { + delete db_; + db_ = nullptr; + Options opts; + if (options != nullptr) { + opts = *options; + } else { + opts = CurrentOptions(); + opts.create_if_missing = true; + } + last_options_ = opts; + + return DB::Open(opts, dbname_, &db_); + } + + Status Put(const std::string& k, const std::string& v) { + return db_->Put(WriteOptions(), k, v); + } + + Status Delete(const std::string& k) { return db_->Delete(WriteOptions(), k); } + + std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) { + ReadOptions options; + options.snapshot = snapshot; + std::string result; + Status s = db_->Get(options, k, &result); + if (s.IsNotFound()) { + result = "NOT_FOUND"; + } else if (!s.ok()) { + result = s.ToString(); + } + return result; + } + + // Return a string that contains all key,value pairs in order, + // formatted like "(k1->v1)(k2->v2)". + std::string Contents() { + std::vector<std::string> forward; + std::string result; + Iterator* iter = db_->NewIterator(ReadOptions()); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + std::string s = IterStatus(iter); + result.push_back('('); + result.append(s); + result.push_back(')'); + forward.push_back(s); + } + + // Check reverse iteration results are the reverse of forward results + size_t matched = 0; + for (iter->SeekToLast(); iter->Valid(); iter->Prev()) { + EXPECT_LT(matched, forward.size()); + EXPECT_EQ(IterStatus(iter), forward[forward.size() - matched - 1]); + matched++; + } + EXPECT_EQ(matched, forward.size()); + + delete iter; + return result; + } + + std::string AllEntriesFor(const Slice& user_key) { + Iterator* iter = dbfull()->TEST_NewInternalIterator(); + InternalKey target(user_key, kMaxSequenceNumber, kTypeValue); + iter->Seek(target.Encode()); + std::string result; + if (!iter->status().ok()) { + result = iter->status().ToString(); + } else { + result = "[ "; + bool first = true; + while (iter->Valid()) { + ParsedInternalKey ikey; + if (!ParseInternalKey(iter->key(), &ikey)) { + result += "CORRUPTED"; + } else { + if (last_options_.comparator->Compare(ikey.user_key, user_key) != 0) { + break; + } + if (!first) { + result += ", "; + } + first = false; + switch (ikey.type) { + case kTypeValue: + result += iter->value().ToString(); + break; + case kTypeDeletion: + result += "DEL"; + break; + } + } + iter->Next(); + } + if (!first) { + result += " "; + } + result += "]"; + } + delete iter; + return result; + } + + int NumTableFilesAtLevel(int level) { + std::string property; + EXPECT_TRUE(db_->GetProperty( + "leveldb.num-files-at-level" + NumberToString(level), &property)); + return std::stoi(property); + } + + int TotalTableFiles() { + int result = 0; + for (int level = 0; level < config::kNumLevels; level++) { + result += NumTableFilesAtLevel(level); + } + return result; + } + + // Return spread of files per level + std::string FilesPerLevel() { + std::string result; + int last_non_zero_offset = 0; + for (int level = 0; level < config::kNumLevels; level++) { + int f = NumTableFilesAtLevel(level); + char buf[100]; + std::snprintf(buf, sizeof(buf), "%s%d", (level ? "," : ""), f); + result += buf; + if (f > 0) { + last_non_zero_offset = result.size(); + } + } + result.resize(last_non_zero_offset); + return result; + } + + int CountFiles() { + std::vector<std::string> files; + env_->GetChildren(dbname_, &files); + return static_cast<int>(files.size()); + } + + uint64_t Size(const Slice& start, const Slice& limit) { + Range r(start, limit); + uint64_t size; + db_->GetApproximateSizes(&r, 1, &size); + return size; + } + + void Compact(const Slice& start, const Slice& limit) { + db_->CompactRange(&start, &limit); + } + + // Do n memtable compactions, each of which produces an sstable + // 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_key, "begin"); + Put(large_key, "end"); + dbfull()->TEST_CompactMemTable(); + } + } + + // Prevent pushing of new sstables into deeper levels by adding + // tables that cover a specified range to all levels. + void FillLevels(const std::string& smallest, const std::string& largest) { + MakeTables(config::kNumLevels, smallest, largest); + } + + void DumpFileCounts(const char* label) { + std::fprintf(stderr, "---\n%s:\n", label); + std::fprintf( + stderr, "maxoverlap: %lld\n", + static_cast<long long>(dbfull()->TEST_MaxNextLevelOverlappingBytes())); + for (int level = 0; level < config::kNumLevels; level++) { + int num = NumTableFilesAtLevel(level); + if (num > 0) { + std::fprintf(stderr, " level %3d : %d files\n", level, num); + } + } + } + + std::string DumpSSTableList() { + std::string property; + db_->GetProperty("leveldb.sstables", &property); + return property; + } + + std::string IterStatus(Iterator* iter) { + std::string result; + if (iter->Valid()) { + result = iter->key().ToString() + "->" + iter->value().ToString(); + } else { + result = "(invalid)"; + } + return result; + } + + bool DeleteAnSSTFile() { + std::vector<std::string> filenames; + EXPECT_LEVELDB_OK(env_->GetChildren(dbname_, &filenames)); + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && type == kTableFile) { + EXPECT_LEVELDB_OK(env_->RemoveFile(TableFileName(dbname_, number))); + return true; + } + } + return false; + } + + // Returns number of files renamed. + int RenameLDBToSST() { + std::vector<std::string> filenames; + EXPECT_LEVELDB_OK(env_->GetChildren(dbname_, &filenames)); + uint64_t number; + FileType type; + int files_renamed = 0; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type) && type == kTableFile) { + const std::string from = TableFileName(dbname_, number); + const std::string to = SSTTableFileName(dbname_, number); + EXPECT_LEVELDB_OK(env_->RenameFile(from, to)); + files_renamed++; + } + } + 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_F(DBTest, Empty) { + do { + ASSERT_TRUE(db_ != nullptr); + ASSERT_EQ("NOT_FOUND", Get("foo")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, EmptyKey) { + do { + ASSERT_LEVELDB_OK(Put("", "v1")); + ASSERT_EQ("v1", Get("")); + ASSERT_LEVELDB_OK(Put("", "v2")); + ASSERT_EQ("v2", Get("")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, EmptyValue) { + do { + ASSERT_LEVELDB_OK(Put("key", "v1")); + ASSERT_EQ("v1", Get("key")); + ASSERT_LEVELDB_OK(Put("key", "")); + ASSERT_EQ("", Get("key")); + ASSERT_LEVELDB_OK(Put("key", "v2")); + ASSERT_EQ("v2", Get("key")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, ReadWrite) { + do { + ASSERT_LEVELDB_OK(Put("foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + ASSERT_LEVELDB_OK(Put("bar", "v2")); + ASSERT_LEVELDB_OK(Put("foo", "v3")); + ASSERT_EQ("v3", Get("foo")); + ASSERT_EQ("v2", Get("bar")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, PutDeleteGet) { + do { + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), "foo", "v2")); + ASSERT_EQ("v2", Get("foo")); + ASSERT_LEVELDB_OK(db_->Delete(WriteOptions(), "foo")); + ASSERT_EQ("NOT_FOUND", Get("foo")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, GetFromImmutableLayer) { + do { + Options options = CurrentOptions(); + options.env = env_; + options.write_buffer_size = 100000; // Small write buffer + Reopen(&options); + + ASSERT_LEVELDB_OK(Put("foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + + // 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")); + // Release sync calls. + env_->delay_data_sync_.store(false, std::memory_order_release); + } while (ChangeOptions()); +} + +TEST_F(DBTest, GetFromVersions) { + do { + ASSERT_LEVELDB_OK(Put("foo", "v1")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v1", Get("foo")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, GetMemUsage) { + do { + ASSERT_LEVELDB_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_F(DBTest, GetSnapshot) { + 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_LEVELDB_OK(Put(key, "v1")); + const Snapshot* s1 = db_->GetSnapshot(); + ASSERT_LEVELDB_OK(Put(key, "v2")); + ASSERT_EQ("v2", Get(key)); + ASSERT_EQ("v1", Get(key, s1)); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get(key)); + ASSERT_EQ("v1", Get(key, s1)); + db_->ReleaseSnapshot(s1); + } + } while (ChangeOptions()); +} + +TEST_F(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_LEVELDB_OK(Put(key, "v1")); + const Snapshot* s1 = db_->GetSnapshot(); + const Snapshot* s2 = db_->GetSnapshot(); + const Snapshot* s3 = db_->GetSnapshot(); + ASSERT_LEVELDB_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_F(DBTest, IterateOverEmptySnapshot) { + do { + const Snapshot* snapshot = db_->GetSnapshot(); + ReadOptions read_options; + read_options.snapshot = snapshot; + ASSERT_LEVELDB_OK(Put("foo", "v1")); + ASSERT_LEVELDB_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_F(DBTest, GetLevel0Ordering) { + do { + // Check that we process level-0 files in correct order. The code + // below generates two level-0 files where the earlier one comes + // before the later one in the level-0 file list since the earlier + // one has a smaller "smallest" key. + ASSERT_LEVELDB_OK(Put("bar", "b")); + ASSERT_LEVELDB_OK(Put("foo", "v1")); + dbfull()->TEST_CompactMemTable(); + ASSERT_LEVELDB_OK(Put("foo", "v2")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get("foo")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, GetOrderedByLevels) { + do { + ASSERT_LEVELDB_OK(Put("foo", "v1")); + Compact("a", "z"); + ASSERT_EQ("v1", Get("foo")); + ASSERT_LEVELDB_OK(Put("foo", "v2")); + ASSERT_EQ("v2", Get("foo")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("v2", Get("foo")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, GetPicksCorrectFile) { + do { + // Arrange to have multiple files in a non-level-0 level. + ASSERT_LEVELDB_OK(Put("a", "va")); + Compact("a", "b"); + ASSERT_LEVELDB_OK(Put("x", "vx")); + Compact("x", "y"); + ASSERT_LEVELDB_OK(Put("f", "vf")); + Compact("f", "g"); + ASSERT_EQ("va", Get("a")); + ASSERT_EQ("vf", Get("f")); + ASSERT_EQ("vx", Get("x")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, GetEncountersEmptyLevel) { + do { + // Arrange for the following to happen: + // * sstable A in level 0 + // * nothing in level 1 + // * sstable B in level 2 + // Then do enough Get() calls to arrange for an automatic compaction + // of sstable A. A bug would cause the compaction to be marked as + // occurring at level 1 (instead of the correct level 0). + + // Step 1: First place sstables in levels 0 and 2 + int compaction_count = 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"); + Put("z", "end"); + dbfull()->TEST_CompactMemTable(); + } + + // Step 2: clear level 1 if necessary. + dbfull()->TEST_CompactRange(1, nullptr, nullptr); + ASSERT_EQ(NumTableFilesAtLevel(0), 1); + ASSERT_EQ(NumTableFilesAtLevel(1), 0); + ASSERT_EQ(NumTableFilesAtLevel(2), 1); + + // Step 3: read a bunch of times + for (int i = 0; i < 1000; i++) { + ASSERT_EQ("NOT_FOUND", Get("missing")); + } + + // Step 4: Wait for compaction to finish + DelayMilliseconds(1000); + + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + } while (ChangeOptions()); +} + +TEST_F(DBTest, IterEmpty) { + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek("foo"); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST_F(DBTest, IterSingle) { + ASSERT_LEVELDB_OK(Put("a", "va")); + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek(""); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek("a"); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek("b"); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST_F(DBTest, IterMulti) { + ASSERT_LEVELDB_OK(Put("a", "va")); + ASSERT_LEVELDB_OK(Put("b", "vb")); + ASSERT_LEVELDB_OK(Put("c", "vc")); + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->Seek(""); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Seek("a"); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Seek("ax"); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Seek("b"); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Seek("z"); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + // Switch from reverse to forward + iter->SeekToLast(); + iter->Prev(); + iter->Prev(); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + + // Switch from forward to reverse + iter->SeekToFirst(); + iter->Next(); + iter->Next(); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + + // Make sure iter stays at snapshot + ASSERT_LEVELDB_OK(Put("a", "va2")); + ASSERT_LEVELDB_OK(Put("a2", "va3")); + ASSERT_LEVELDB_OK(Put("b", "vb2")); + ASSERT_LEVELDB_OK(Put("c", "vc2")); + ASSERT_LEVELDB_OK(Delete("b")); + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->vb"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST_F(DBTest, IterSmallAndLargeMix) { + ASSERT_LEVELDB_OK(Put("a", "va")); + ASSERT_LEVELDB_OK(Put("b", std::string(100000, 'b'))); + ASSERT_LEVELDB_OK(Put("c", "vc")); + ASSERT_LEVELDB_OK(Put("d", std::string(100000, 'd'))); + ASSERT_LEVELDB_OK(Put("e", std::string(100000, 'e'))); + + Iterator* iter = db_->NewIterator(ReadOptions()); + + iter->SeekToFirst(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "b->" + std::string(100000, 'b')); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "d->" + std::string(100000, 'd')); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "e->" + std::string(100000, 'e')); + iter->Next(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + iter->SeekToLast(); + ASSERT_EQ(IterStatus(iter), "e->" + std::string(100000, 'e')); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "d->" + std::string(100000, 'd')); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "b->" + std::string(100000, 'b')); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "(invalid)"); + + delete iter; +} + +TEST_F(DBTest, IterMultiWithDelete) { + do { + ASSERT_LEVELDB_OK(Put("a", "va")); + ASSERT_LEVELDB_OK(Put("b", "vb")); + ASSERT_LEVELDB_OK(Put("c", "vc")); + ASSERT_LEVELDB_OK(Delete("b")); + ASSERT_EQ("NOT_FOUND", Get("b")); + + Iterator* iter = db_->NewIterator(ReadOptions()); + iter->Seek("c"); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + delete iter; + } while (ChangeOptions()); +} + +TEST_F(DBTest, IterMultiWithDeleteAndCompaction) { + do { + ASSERT_LEVELDB_OK(Put("b", "vb")); + ASSERT_LEVELDB_OK(Put("c", "vc")); + ASSERT_LEVELDB_OK(Put("a", "va")); + dbfull()->TEST_CompactMemTable(); + ASSERT_LEVELDB_OK(Delete("b")); + ASSERT_EQ("NOT_FOUND", Get("b")); + + Iterator* iter = db_->NewIterator(ReadOptions()); + iter->Seek("c"); + ASSERT_EQ(IterStatus(iter), "c->vc"); + iter->Prev(); + ASSERT_EQ(IterStatus(iter), "a->va"); + iter->Seek("b"); + ASSERT_EQ(IterStatus(iter), "c->vc"); + delete iter; + } while (ChangeOptions()); +} + +TEST_F(DBTest, Recover) { + do { + ASSERT_LEVELDB_OK(Put("foo", "v1")); + ASSERT_LEVELDB_OK(Put("baz", "v5")); + + Reopen(); + ASSERT_EQ("v1", Get("foo")); + + ASSERT_EQ("v1", Get("foo")); + ASSERT_EQ("v5", Get("baz")); + ASSERT_LEVELDB_OK(Put("bar", "v2")); + ASSERT_LEVELDB_OK(Put("foo", "v3")); + + Reopen(); + ASSERT_EQ("v3", Get("foo")); + ASSERT_LEVELDB_OK(Put("foo", "v4")); + ASSERT_EQ("v4", Get("foo")); + ASSERT_EQ("v2", Get("bar")); + ASSERT_EQ("v5", Get("baz")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, RecoveryWithEmptyLog) { + do { + ASSERT_LEVELDB_OK(Put("foo", "v1")); + ASSERT_LEVELDB_OK(Put("foo", "v2")); + Reopen(); + Reopen(); + ASSERT_LEVELDB_OK(Put("foo", "v3")); + Reopen(); + ASSERT_EQ("v3", Get("foo")); + } while (ChangeOptions()); +} + +// Check that writes done during a memtable compaction are recovered +// if the database is shutdown during the memtable compaction. +TEST_F(DBTest, RecoverDuringMemtableCompaction) { + do { + Options options = CurrentOptions(); + options.env = env_; + options.write_buffer_size = 1000000; + Reopen(&options); + + // Trigger a long memtable compaction and reopen the database during it + ASSERT_LEVELDB_OK(Put("foo", "v1")); // Goes to 1st log file + ASSERT_LEVELDB_OK( + Put("big1", std::string(10000000, 'x'))); // Fills memtable + ASSERT_LEVELDB_OK( + Put("big2", std::string(1000, 'y'))); // Triggers compaction + ASSERT_LEVELDB_OK(Put("bar", "v2")); // Goes to new log file + + Reopen(&options); + ASSERT_EQ("v1", Get("foo")); + ASSERT_EQ("v2", Get("bar")); + ASSERT_EQ(std::string(10000000, 'x'), Get("big1")); + ASSERT_EQ(std::string(1000, 'y'), Get("big2")); + } while (ChangeOptions()); +} + +static std::string Key(int i) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "key%06d", i); + return std::string(buf); +} + +TEST_F(DBTest, MinorCompactionsHappen) { + Options options = CurrentOptions(); + options.write_buffer_size = 10000; + Reopen(&options); + + const int N = 500; + + int starting_num_tables = TotalTableFiles(); + for (int i = 0; i < N; i++) { + ASSERT_LEVELDB_OK(Put(Key(i), Key(i) + std::string(1000, 'v'))); + } + int ending_num_tables = TotalTableFiles(); + ASSERT_GT(ending_num_tables, starting_num_tables); + + for (int i = 0; i < N; i++) { + ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(Key(i))); + } + + Reopen(); + + for (int i = 0; i < N; i++) { + ASSERT_EQ(Key(i) + std::string(1000, 'v'), Get(Key(i))); + } +} + +TEST_F(DBTest, RecoverWithLargeLog) { + { + Options options = CurrentOptions(); + Reopen(&options); + ASSERT_LEVELDB_OK(Put("big1", std::string(200000, '1'))); + ASSERT_LEVELDB_OK(Put("big2", std::string(200000, '2'))); + ASSERT_LEVELDB_OK(Put("small3", std::string(10, '3'))); + ASSERT_LEVELDB_OK(Put("small4", std::string(10, '4'))); + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + } + + // Make sure that if we re-open with a small write buffer size that + // we flush table files in the middle of a large log file. + Options options = CurrentOptions(); + options.write_buffer_size = 100000; + Reopen(&options); + ASSERT_EQ(NumTableFilesAtLevel(0), 3); + ASSERT_EQ(std::string(200000, '1'), Get("big1")); + ASSERT_EQ(std::string(200000, '2'), Get("big2")); + ASSERT_EQ(std::string(10, '3'), Get("small3")); + ASSERT_EQ(std::string(10, '4'), Get("small4")); + ASSERT_GT(NumTableFilesAtLevel(0), 1); +} + +TEST_F(DBTest, CompactionsGenerateMultipleFiles) { + Options options = CurrentOptions(); + options.write_buffer_size = 100000000; // Large write buffer + Reopen(&options); + + Random rnd(301); + + // Write 8MB (80 values, each 100K) + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + std::vector<std::string> values; + for (int i = 0; i < 80; i++) { + values.push_back(RandomString(&rnd, 100000)); + ASSERT_LEVELDB_OK(Put(Key(i), values[i])); + } + + // Reopening moves updates to level-0 + Reopen(&options); + dbfull()->TEST_CompactRange(0, nullptr, nullptr); + + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + ASSERT_GT(NumTableFilesAtLevel(1), 1); + for (int i = 0; i < 80; i++) { + ASSERT_EQ(Get(Key(i)), values[i]); + } +} + +TEST_F(DBTest, RepeatedWritesToSameKey) { + Options options = CurrentOptions(); + options.env = env_; + options.write_buffer_size = 100000; // Small write buffer + Reopen(&options); + + // We must have at most one file per level except for level-0, + // which may have up to kL0_StopWritesTrigger files. + const int kMaxFiles = config::kNumLevels + config::kL0_StopWritesTrigger; + + Random rnd(301); + std::string value = RandomString(&rnd, 2 * options.write_buffer_size); + for (int i = 0; i < 5 * kMaxFiles; i++) { + Put("key", value); + ASSERT_LE(TotalTableFiles(), kMaxFiles); + std::fprintf(stderr, "after %d: %d files\n", i + 1, TotalTableFiles()); + } +} + +TEST_F(DBTest, SparseMerge) { + Options options = CurrentOptions(); + options.compression = kNoCompression; + Reopen(&options); + + FillLevels("A", "Z"); + + // Suppose there is: + // small amount of data with prefix A + // large amount of data with prefix B + // small amount of data with prefix C + // and that recent updates have made small changes to all three prefixes. + // Check that we do not do a compaction that merges all of B in one shot. + const std::string value(1000, 'x'); + Put("A", "va"); + // Write approximately 100MB of "B" values + for (int i = 0; i < 100000; i++) { + char key[100]; + std::snprintf(key, sizeof(key), "B%010d", i); + Put(key, value); + } + Put("C", "vc"); + dbfull()->TEST_CompactMemTable(); + dbfull()->TEST_CompactRange(0, nullptr, nullptr); + + // Make sparse update + Put("A", "va2"); + Put("B100", "bvalue2"); + 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, 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) { + std::fprintf(stderr, "Value %llu is not in range [%llu, %llu]\n", + (unsigned long long)(val), (unsigned long long)(low), + (unsigned long long)(high)); + } + return result; +} + +TEST_F(DBTest, ApproximateSizes) { + do { + Options options = CurrentOptions(); + options.write_buffer_size = 100000000; // Large write buffer + options.compression = kNoCompression; + DestroyAndReopen(); + + ASSERT_TRUE(Between(Size("", "xyz"), 0, 0)); + Reopen(&options); + ASSERT_TRUE(Between(Size("", "xyz"), 0, 0)); + + // Write 8MB (80 values, each 100K) + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + const int N = 80; + static const int S1 = 100000; + static const int S2 = 105000; // Allow some expansion from metadata + Random rnd(301); + for (int i = 0; i < N; i++) { + ASSERT_LEVELDB_OK(Put(Key(i), RandomString(&rnd, S1))); + } + + // 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(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); + Slice cstart = cstart_str; + Slice cend = cend_str; + dbfull()->TEST_CompactRange(0, &cstart, &cend); + } + + ASSERT_EQ(NumTableFilesAtLevel(0), 0); + ASSERT_GT(NumTableFilesAtLevel(1), 0); + } + } while (ChangeOptions()); +} + +TEST_F(DBTest, ApproximateSizes_MixOfSmallAndLarge) { + do { + Options options = CurrentOptions(); + options.compression = kNoCompression; + Reopen(); + + Random rnd(301); + std::string big1 = RandomString(&rnd, 100000); + ASSERT_LEVELDB_OK(Put(Key(0), RandomString(&rnd, 10000))); + ASSERT_LEVELDB_OK(Put(Key(1), RandomString(&rnd, 10000))); + ASSERT_LEVELDB_OK(Put(Key(2), big1)); + ASSERT_LEVELDB_OK(Put(Key(3), RandomString(&rnd, 10000))); + ASSERT_LEVELDB_OK(Put(Key(4), big1)); + ASSERT_LEVELDB_OK(Put(Key(5), RandomString(&rnd, 10000))); + ASSERT_LEVELDB_OK(Put(Key(6), RandomString(&rnd, 300000))); + ASSERT_LEVELDB_OK(Put(Key(7), RandomString(&rnd, 10000))); + + if (options.reuse_logs) { + // Need to force a memtable compaction since recovery does not do so. + ASSERT_LEVELDB_OK(dbfull()->TEST_CompactMemTable()); + } + + // Check sizes across recovery by reopening a few times + for (int run = 0; run < 3; run++) { + Reopen(&options); + + ASSERT_TRUE(Between(Size("", Key(0)), 0, 0)); + ASSERT_TRUE(Between(Size("", Key(1)), 10000, 11000)); + ASSERT_TRUE(Between(Size("", Key(2)), 20000, 21000)); + ASSERT_TRUE(Between(Size("", Key(3)), 120000, 121000)); + ASSERT_TRUE(Between(Size("", Key(4)), 130000, 131000)); + ASSERT_TRUE(Between(Size("", Key(5)), 230000, 231000)); + ASSERT_TRUE(Between(Size("", Key(6)), 240000, 241000)); + ASSERT_TRUE(Between(Size("", Key(7)), 540000, 541000)); + ASSERT_TRUE(Between(Size("", Key(8)), 550000, 560000)); + + ASSERT_TRUE(Between(Size(Key(3), Key(5)), 110000, 111000)); + + dbfull()->TEST_CompactRange(0, nullptr, nullptr); + } + } while (ChangeOptions()); +} + +TEST_F(DBTest, IteratorPinsRef) { + Put("foo", "hello"); + + // Get iterator that will yield the current contents of the DB. + Iterator* iter = db_->NewIterator(ReadOptions()); + + // Write to force compactions + Put("foo", "newvalue1"); + for (int i = 0; i < 100; i++) { + ASSERT_LEVELDB_OK( + Put(Key(i), Key(i) + std::string(100000, 'v'))); // 100K values + } + Put("foo", "newvalue2"); + + iter->SeekToFirst(); + ASSERT_TRUE(iter->Valid()); + ASSERT_EQ("foo", iter->key().ToString()); + ASSERT_EQ("hello", iter->value().ToString()); + iter->Next(); + ASSERT_TRUE(!iter->Valid()); + delete iter; +} + +TEST_F(DBTest, Snapshot) { + do { + Put("foo", "v1"); + const Snapshot* s1 = db_->GetSnapshot(); + Put("foo", "v2"); + const Snapshot* s2 = db_->GetSnapshot(); + Put("foo", "v3"); + const Snapshot* s3 = db_->GetSnapshot(); + + Put("foo", "v4"); + ASSERT_EQ("v1", Get("foo", s1)); + ASSERT_EQ("v2", Get("foo", s2)); + ASSERT_EQ("v3", Get("foo", s3)); + ASSERT_EQ("v4", Get("foo")); + + db_->ReleaseSnapshot(s3); + ASSERT_EQ("v1", Get("foo", s1)); + ASSERT_EQ("v2", Get("foo", s2)); + ASSERT_EQ("v4", Get("foo")); + + db_->ReleaseSnapshot(s1); + ASSERT_EQ("v2", Get("foo", s2)); + ASSERT_EQ("v4", Get("foo")); + + db_->ReleaseSnapshot(s2); + ASSERT_EQ("v4", Get("foo")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, HiddenValuesAreRemoved) { + do { + Random rnd(301); + FillLevels("a", "z"); + + std::string big = RandomString(&rnd, 50000); + Put("foo", big); + Put("pastfoo", "v"); + const Snapshot* snapshot = db_->GetSnapshot(); + Put("foo", "tiny"); + Put("pastfoo2", "v2"); // Advance sequence number one more + + ASSERT_LEVELDB_OK(dbfull()->TEST_CompactMemTable()); + ASSERT_GT(NumTableFilesAtLevel(0), 0); + + ASSERT_EQ(big, Get("foo", snapshot)); + ASSERT_TRUE(Between(Size("", "pastfoo"), 50000, 60000)); + db_->ReleaseSnapshot(snapshot); + ASSERT_EQ(AllEntriesFor("foo"), "[ tiny, " + big + " ]"); + Slice x("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, nullptr, &x); + ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]"); + + ASSERT_TRUE(Between(Size("", "pastfoo"), 0, 1000)); + } while (ChangeOptions()); +} + +TEST_F(DBTest, DeletionMarkers1) { + Put("foo", "v1"); + ASSERT_LEVELDB_OK(dbfull()->TEST_CompactMemTable()); + const int last = config::kMaxMemCompactLevel; + 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); + + Delete("foo"); + Put("foo", "v2"); + ASSERT_EQ(AllEntriesFor("foo"), "[ v2, DEL, v1 ]"); + ASSERT_LEVELDB_OK(dbfull()->TEST_CompactMemTable()); // Moves to level last-2 + ASSERT_EQ(AllEntriesFor("foo"), "[ v2, DEL, v1 ]"); + Slice z("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, 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 ]"); +} + +TEST_F(DBTest, DeletionMarkers2) { + Put("foo", "v1"); + ASSERT_LEVELDB_OK(dbfull()->TEST_CompactMemTable()); + const int last = config::kMaxMemCompactLevel; + 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); + + Delete("foo"); + ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); + ASSERT_LEVELDB_OK(dbfull()->TEST_CompactMemTable()); // Moves to level last-2 + ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); + dbfull()->TEST_CompactRange(last - 2, nullptr, nullptr); + // DEL kept: "last" file overlaps + ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]"); + 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"), "[ ]"); +} + +TEST_F(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. + ASSERT_LEVELDB_OK(Put("100", "v100")); + ASSERT_LEVELDB_OK(Put("999", "v999")); + dbfull()->TEST_CompactMemTable(); + ASSERT_LEVELDB_OK(Delete("100")); + ASSERT_LEVELDB_OK(Delete("999")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("0,1,1", FilesPerLevel()); + + // Make files spanning the following ranges in level-0: + // files[0] 200 .. 900 + // files[1] 300 .. 500 + // Note that files are sorted by smallest key. + ASSERT_LEVELDB_OK(Put("300", "v300")); + ASSERT_LEVELDB_OK(Put("500", "v500")); + dbfull()->TEST_CompactMemTable(); + ASSERT_LEVELDB_OK(Put("200", "v200")); + ASSERT_LEVELDB_OK(Put("600", "v600")); + ASSERT_LEVELDB_OK(Put("900", "v900")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("2,1,1", FilesPerLevel()); + + // Compact away the placeholder files we created initially + 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 + // not detect the overlap with level-0 files and would incorrectly place + // the deletion in a deeper level. + ASSERT_LEVELDB_OK(Delete("600")); + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("3", FilesPerLevel()); + ASSERT_EQ("NOT_FOUND", Get("600")); + } while (ChangeOptions()); +} + +TEST_F(DBTest, L0_CompactionBug_Issue44_a) { + Reopen(); + ASSERT_LEVELDB_OK(Put("b", "v")); + Reopen(); + ASSERT_LEVELDB_OK(Delete("b")); + ASSERT_LEVELDB_OK(Delete("a")); + Reopen(); + ASSERT_LEVELDB_OK(Delete("a")); + Reopen(); + ASSERT_LEVELDB_OK(Put("a", "v")); + Reopen(); + Reopen(); + ASSERT_EQ("(a->v)", Contents()); + DelayMilliseconds(1000); // Wait for compaction to finish + ASSERT_EQ("(a->v)", Contents()); +} + +TEST_F(DBTest, L0_CompactionBug_Issue44_b) { + Reopen(); + Put("", ""); + Reopen(); + Delete("e"); + Put("", ""); + Reopen(); + Put("c", "cv"); + Reopen(); + Put("", ""); + Reopen(); + Put("", ""); + DelayMilliseconds(1000); // Wait for compaction to finish + Reopen(); + Put("d", "dv"); + Reopen(); + Put("", ""); + Reopen(); + Delete("d"); + Delete("b"); + Reopen(); + ASSERT_EQ("(->)(c->cv)", Contents()); + DelayMilliseconds(1000); // Wait for compaction to finish + ASSERT_EQ("(->)(c->cv)", Contents()); +} + +TEST_F(DBTest, Fflush_Issue474) { + static const int kNum = 100000; + Random rnd(test::RandomSeed()); + for (int i = 0; i < kNum; i++) { + std::fflush(nullptr); + ASSERT_LEVELDB_OK(Put(RandomKey(&rnd), RandomString(&rnd, 100))); + } +} + +TEST_F(DBTest, ComparatorCheck) { + class NewComparator : public Comparator { + public: + const char* Name() const override { return "leveldb.NewComparator"; } + int Compare(const Slice& a, const Slice& b) const override { + return BytewiseComparator()->Compare(a, b); + } + void FindShortestSeparator(std::string* s, const Slice& l) const override { + BytewiseComparator()->FindShortestSeparator(s, l); + } + void FindShortSuccessor(std::string* key) const override { + BytewiseComparator()->FindShortSuccessor(key); + } + }; + NewComparator cmp; + Options new_options = CurrentOptions(); + new_options.comparator = &cmp; + Status s = TryReopen(&new_options); + ASSERT_TRUE(!s.ok()); + ASSERT_TRUE(s.ToString().find("comparator") != std::string::npos) + << s.ToString(); +} + +TEST_F(DBTest, CustomComparator) { + class NumberComparator : public Comparator { + public: + const char* Name() const override { return "test.NumberComparator"; } + int Compare(const Slice& a, const Slice& b) const override { + return ToNumber(a) - ToNumber(b); + } + void FindShortestSeparator(std::string* s, const Slice& l) const override { + ToNumber(*s); // Check format + ToNumber(l); // 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. + EXPECT_TRUE(x.size() >= 2 && x[0] == '[' && x[x.size() - 1] == ']') + << EscapeString(x); + int val; + char ignored; + EXPECT_TRUE(sscanf(x.ToString().c_str(), "[%i]%c", &val, &ignored) == 1) + << EscapeString(x); + return val; + } + }; + NumberComparator cmp; + Options new_options = CurrentOptions(); + new_options.create_if_missing = true; + new_options.comparator = &cmp; + new_options.filter_policy = nullptr; // Cannot use bloom filters + new_options.write_buffer_size = 1000; // Compact more often + DestroyAndReopen(&new_options); + ASSERT_LEVELDB_OK(Put("[10]", "ten")); + ASSERT_LEVELDB_OK(Put("[0x14]", "twenty")); + for (int i = 0; i < 2; i++) { + ASSERT_EQ("ten", Get("[10]")); + ASSERT_EQ("ten", Get("[0xa]")); + ASSERT_EQ("twenty", Get("[20]")); + ASSERT_EQ("twenty", Get("[0x14]")); + ASSERT_EQ("NOT_FOUND", Get("[15]")); + ASSERT_EQ("NOT_FOUND", Get("[0xf]")); + Compact("[0]", "[9999]"); + } + + for (int run = 0; run < 2; run++) { + for (int i = 0; i < 1000; i++) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "[%d]", i * 10); + ASSERT_LEVELDB_OK(Put(buf, buf)); + } + Compact("[0]", "[1000000]"); + } +} + +TEST_F(DBTest, ManualCompaction) { + ASSERT_EQ(config::kMaxMemCompactLevel, 2) + << "Need to update this test to match kMaxMemCompactLevel"; + + MakeTables(3, "p", "q"); + ASSERT_EQ("1,1,1", FilesPerLevel()); + + // Compaction range falls before files + Compact("", "c"); + ASSERT_EQ("1,1,1", FilesPerLevel()); + + // Compaction range falls after files + Compact("r", "z"); + ASSERT_EQ("1,1,1", FilesPerLevel()); + + // Compaction range overlaps files + Compact("p1", "p9"); + ASSERT_EQ("0,0,1", FilesPerLevel()); + + // Populate a different range + MakeTables(3, "c", "e"); + ASSERT_EQ("1,1,2", FilesPerLevel()); + + // Compact just the new range + Compact("b", "f"); + ASSERT_EQ("0,0,2", FilesPerLevel()); + + // Compact all + MakeTables(1, "a", "z"); + ASSERT_EQ("0,1,2", FilesPerLevel()); + db_->CompactRange(nullptr, nullptr); + ASSERT_EQ("0,0,1", FilesPerLevel()); +} + +TEST_F(DBTest, DBOpen_Options) { + std::string dbname = testing::TempDir() + "db_options_test"; + DestroyDB(dbname, Options()); + + // Does not exist, and create_if_missing == false: error + 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") != 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_LEVELDB_OK(s); + ASSERT_TRUE(db != nullptr); + + delete db; + 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") != 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_LEVELDB_OK(s); + ASSERT_TRUE(db != nullptr); + + delete db; + db = nullptr; +} + +TEST_F(DBTest, DestroyEmptyDir) { + std::string dbname = testing::TempDir() + "db_empty_dir"; + TestEnv env(Env::Default()); + env.RemoveDir(dbname); + ASSERT_TRUE(!env.FileExists(dbname)); + + Options opts; + opts.env = &env; + + ASSERT_LEVELDB_OK(env.CreateDir(dbname)); + ASSERT_TRUE(env.FileExists(dbname)); + std::vector<std::string> children; + ASSERT_LEVELDB_OK(env.GetChildren(dbname, &children)); + // The stock Env's do not filter out '.' and '..' special files. + ASSERT_EQ(2, children.size()); + ASSERT_LEVELDB_OK(DestroyDB(dbname, opts)); + ASSERT_TRUE(!env.FileExists(dbname)); + + // Should also be destroyed if Env is filtering out dot files. + env.SetIgnoreDotFiles(true); + ASSERT_LEVELDB_OK(env.CreateDir(dbname)); + ASSERT_TRUE(env.FileExists(dbname)); + ASSERT_LEVELDB_OK(env.GetChildren(dbname, &children)); + ASSERT_EQ(0, children.size()); + ASSERT_LEVELDB_OK(DestroyDB(dbname, opts)); + ASSERT_TRUE(!env.FileExists(dbname)); +} + +TEST_F(DBTest, DestroyOpenDB) { + std::string dbname = testing::TempDir() + "open_db_dir"; + env_->RemoveDir(dbname); + ASSERT_TRUE(!env_->FileExists(dbname)); + + Options opts; + opts.create_if_missing = true; + DB* db = nullptr; + ASSERT_LEVELDB_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_LEVELDB_OK(DestroyDB(dbname, Options())); + ASSERT_TRUE(!env_->FileExists(dbname)); +} + +TEST_F(DBTest, Locking) { + DB* db2 = nullptr; + Status s = DB::Open(CurrentOptions(), dbname_, &db2); + ASSERT_TRUE(!s.ok()) << "Locking did not prevent re-opening db"; +} + +// Check that number of files does not grow when we are out of space +TEST_F(DBTest, NoSpace) { + Options options = CurrentOptions(); + options.env = env_; + Reopen(&options); + + ASSERT_LEVELDB_OK(Put("foo", "v1")); + ASSERT_EQ("v1", Get("foo")); + Compact("a", "z"); + const int num_files = CountFiles(); + // 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, nullptr, nullptr); + } + } + env_->no_space_.store(false, std::memory_order_release); + ASSERT_LT(CountFiles(), num_files + 3); +} + +TEST_F(DBTest, NonWritableFileSystem) { + Options options = CurrentOptions(); + options.write_buffer_size = 1000; + options.env = env_; + Reopen(&options); + ASSERT_LEVELDB_OK(Put("foo", "v1")); + // 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++) { + std::fprintf(stderr, "iter %d; errors %d\n", i, errors); + if (!Put("foo", big).ok()) { + errors++; + DelayMilliseconds(100); + } + } + ASSERT_GT(errors, 0); + env_->non_writable_.store(false, std::memory_order_release); +} + +TEST_F(DBTest, WriteSyncError) { + // Check that log sync errors cause the DB to disallow future writes. + + // (a) Cause log sync calls to fail + Options options = CurrentOptions(); + options.env = env_; + Reopen(&options); + env_->data_sync_error_.store(true, std::memory_order_release); + + // (b) Normal write should succeed + WriteOptions w; + ASSERT_LEVELDB_OK(db_->Put(w, "k1", "v1")); + ASSERT_EQ("v1", Get("k1")); + + // (c) Do a sync write; should fail + w.sync = true; + ASSERT_TRUE(!db_->Put(w, "k2", "v2").ok()); + ASSERT_EQ("v1", Get("k1")); + ASSERT_EQ("NOT_FOUND", Get("k2")); + + // (d) make sync behave normally + env_->data_sync_error_.store(false, std::memory_order_release); + + // (e) Do a non-sync write; should fail + w.sync = false; + ASSERT_TRUE(!db_->Put(w, "k3", "v3").ok()); + ASSERT_EQ("v1", Get("k1")); + ASSERT_EQ("NOT_FOUND", Get("k2")); + ASSERT_EQ("NOT_FOUND", Get("k3")); +} + +TEST_F(DBTest, ManifestWriteError) { + // Test for the following problem: + // (a) Compaction produces file F + // (b) Log record containing F is written to MANIFEST file, but Sync() fails + // (c) GC deletes F + // (d) After reopening DB, reads fail since deleted F is named in log record + + // We iterate twice. In the second iteration, everything is the + // same except the log record never makes it to the MANIFEST file. + for (int iter = 0; iter < 2; iter++) { + std::atomic<bool>* error_type = (iter == 0) ? &env_->manifest_sync_error_ + : &env_->manifest_write_error_; + + // Insert foo=>bar mapping + Options options = CurrentOptions(); + options.env = env_; + options.create_if_missing = true; + options.error_if_exists = false; + DestroyAndReopen(&options); + ASSERT_LEVELDB_OK(Put("foo", "bar")); + ASSERT_EQ("bar", Get("foo")); + + // Memtable compaction (will succeed) + 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 + + // Merging compaction (will fail) + error_type->store(true, std::memory_order_release); + dbfull()->TEST_CompactRange(last, nullptr, nullptr); // Should fail + ASSERT_EQ("bar", Get("foo")); + + // Recovery: should not lose data + error_type->store(false, std::memory_order_release); + Reopen(&options); + ASSERT_EQ("bar", Get("foo")); + } +} + +TEST_F(DBTest, MissingSSTFile) { + ASSERT_LEVELDB_OK(Put("foo", "bar")); + ASSERT_EQ("bar", Get("foo")); + + // Dump the memtable to disk. + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("bar", Get("foo")); + + Close(); + ASSERT_TRUE(DeleteAnSSTFile()); + Options options = CurrentOptions(); + options.paranoid_checks = true; + Status s = TryReopen(&options); + ASSERT_TRUE(!s.ok()); + ASSERT_TRUE(s.ToString().find("issing") != std::string::npos) << s.ToString(); +} + +TEST_F(DBTest, StillReadSST) { + ASSERT_LEVELDB_OK(Put("foo", "bar")); + ASSERT_EQ("bar", Get("foo")); + + // Dump the memtable to disk. + dbfull()->TEST_CompactMemTable(); + ASSERT_EQ("bar", Get("foo")); + Close(); + ASSERT_GT(RenameLDBToSST(), 0); + Options options = CurrentOptions(); + options.paranoid_checks = true; + Status s = TryReopen(&options); + ASSERT_TRUE(s.ok()); + ASSERT_EQ("bar", Get("foo")); +} + +TEST_F(DBTest, FilesDeletedAfterCompaction) { + ASSERT_LEVELDB_OK(Put("foo", "v2")); + Compact("a", "z"); + const int num_files = CountFiles(); + for (int i = 0; i < 10; i++) { + ASSERT_LEVELDB_OK(Put("foo", "v2")); + Compact("a", "z"); + } + ASSERT_EQ(CountFiles(), num_files); +} + +TEST_F(DBTest, BloomFilter) { + env_->count_random_reads_ = true; + Options options = CurrentOptions(); + options.env = env_; + options.block_cache = NewLRUCache(0); // Prevent cache hits + options.filter_policy = NewBloomFilterPolicy(10); + Reopen(&options); + + // Populate multiple layers + const int N = 10000; + for (int i = 0; i < N; i++) { + ASSERT_LEVELDB_OK(Put(Key(i), Key(i))); + } + Compact("a", "z"); + for (int i = 0; i < N; i += 100) { + ASSERT_LEVELDB_OK(Put(Key(i), Key(i))); + } + dbfull()->TEST_CompactMemTable(); + + // Prevent auto compactions triggered by seeks + env_->delay_data_sync_.store(true, std::memory_order_release); + + // Lookup present keys. Should rarely read from small sstable. + env_->random_read_counter_.Reset(); + for (int i = 0; i < N; i++) { + ASSERT_EQ(Key(i), Get(Key(i))); + } + int reads = env_->random_read_counter_.Read(); + std::fprintf(stderr, "%d present => %d reads\n", N, reads); + ASSERT_GE(reads, N); + ASSERT_LE(reads, N + 2 * N / 100); + + // Lookup present keys. Should rarely read from either sstable. + env_->random_read_counter_.Reset(); + for (int i = 0; i < N; i++) { + ASSERT_EQ("NOT_FOUND", Get(Key(i) + ".missing")); + } + reads = env_->random_read_counter_.Read(); + std::fprintf(stderr, "%d missing => %d reads\n", N, reads); + ASSERT_LE(reads, 3 * N / 100); + + env_->delay_data_sync_.store(false, std::memory_order_release); + Close(); + delete options.block_cache; + delete options.filter_policy; +} + +// Multi-threaded test: +namespace { + +static const int kNumThreads = 4; +static const int kTestSeconds = 10; +static const int kNumKeys = 1000; + +struct MTState { + DBTest* test; + std::atomic<bool> stop; + std::atomic<int> counter[kNumThreads]; + std::atomic<bool> thread_done[kNumThreads]; +}; + +struct MTThread { + MTState* state; + int id; +}; + +static void MTThreadBody(void* arg) { + MTThread* t = reinterpret_cast<MTThread*>(arg); + int id = t->id; + DB* db = t->state->test->db_; + int counter = 0; + std::fprintf(stderr, "... starting thread %d\n", id); + Random rnd(1000 + id); + std::string value; + char valbuf[1500]; + 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]; + std::snprintf(keybuf, sizeof(keybuf), "%016d", key); + + if (rnd.OneIn(2)) { + // Write values of the form <key, my id, counter>. + // We add some padding for force compactions. + std::snprintf(valbuf, sizeof(valbuf), "%d.%d.%-1000d", key, id, + static_cast<int>(counter)); + ASSERT_LEVELDB_OK(db->Put(WriteOptions(), Slice(keybuf), Slice(valbuf))); + } else { + // Read a value and verify that it matches the pattern written above. + Status s = db->Get(ReadOptions(), Slice(keybuf), &value); + if (s.IsNotFound()) { + // Key has not yet been written + } else { + // Check that the writer thread counter is >= the counter in the value + ASSERT_LEVELDB_OK(s); + int k, w, c; + ASSERT_EQ(3, sscanf(value.c_str(), "%d.%d.%d", &k, &w, &c)) << value; + ASSERT_EQ(k, key); + ASSERT_GE(w, 0); + ASSERT_LT(w, kNumThreads); + ASSERT_LE(c, t->state->counter[w].load(std::memory_order_acquire)); + } + } + counter++; + } + t->state->thread_done[id].store(true, std::memory_order_release); + std::fprintf(stderr, "... stopping thread %d after %d ops\n", id, counter); +} + +} // namespace + +TEST_F(DBTest, MultiThreaded) { + do { + // Initialize state + MTState mt; + mt.test = this; + mt.stop.store(false, std::memory_order_release); + for (int id = 0; id < kNumThreads; id++) { + mt.counter[id].store(false, std::memory_order_release); + mt.thread_done[id].store(false, std::memory_order_release); + } + + // Start threads + MTThread thread[kNumThreads]; + for (int id = 0; id < kNumThreads; id++) { + thread[id].state = &mt; + thread[id].id = id; + env_->StartThread(MTThreadBody, &thread[id]); + } + + // Let them run for a while + DelayMilliseconds(kTestSeconds * 1000); + + // Stop the threads and wait for them to finish + mt.stop.store(true, std::memory_order_release); + for (int id = 0; id < kNumThreads; id++) { + while (!mt.thread_done[id].load(std::memory_order_acquire)) { + DelayMilliseconds(100); + } + } + } while (ChangeOptions()); +} + +namespace { +typedef std::map<std::string, std::string> KVMap; +} + +class ModelDB : public DB { + public: + class ModelSnapshot : public Snapshot { + public: + KVMap map_; + }; + + 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); + } + Status Delete(const WriteOptions& o, const Slice& key) override { + return DB::Delete(o, key); + } + Status Get(const ReadOptions& options, const Slice& key, + std::string* value) override { + assert(false); // Not implemented + return Status::NotFound(key); + } + Iterator* NewIterator(const ReadOptions& options) override { + if (options.snapshot == nullptr) { + KVMap* saved = new KVMap; + *saved = map_; + return new ModelIter(saved, true); + } else { + const KVMap* snapshot_state = + &(reinterpret_cast<const ModelSnapshot*>(options.snapshot)->map_); + return new ModelIter(snapshot_state, false); + } + } + const Snapshot* GetSnapshot() override { + ModelSnapshot* snapshot = new ModelSnapshot; + snapshot->map_ = map_; + return snapshot; + } + + void ReleaseSnapshot(const Snapshot* snapshot) override { + delete reinterpret_cast<const ModelSnapshot*>(snapshot); + } + Status Write(const WriteOptions& options, WriteBatch* batch) override { + class Handler : public WriteBatch::Handler { + public: + KVMap* map_; + void Put(const Slice& key, const Slice& value) override { + (*map_)[key.ToString()] = value.ToString(); + } + void Delete(const Slice& key) override { map_->erase(key.ToString()); } + }; + Handler handler; + handler.map_ = &map_; + return batch->Iterate(&handler); + } + + bool GetProperty(const Slice& property, std::string* value) override { + return false; + } + void GetApproximateSizes(const Range* r, int n, uint64_t* sizes) override { + for (int i = 0; i < n; i++) { + sizes[i] = 0; + } + } + void CompactRange(const Slice* start, const Slice* end) override {} + + private: + class ModelIter : public Iterator { + public: + ModelIter(const KVMap* map, bool owned) + : map_(map), owned_(owned), iter_(map_->end()) {} + ~ModelIter() override { + if (owned_) delete map_; + } + 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); + } + } + void Seek(const Slice& k) override { + iter_ = map_->lower_bound(k.ToString()); + } + 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_ + KVMap::const_iterator iter_; + }; + const Options options_; + KVMap map_; +}; + +static bool CompareIterators(int step, DB* model, DB* db, + const Snapshot* model_snap, + const Snapshot* db_snap) { + ReadOptions options; + options.snapshot = model_snap; + Iterator* miter = model->NewIterator(options); + options.snapshot = db_snap; + Iterator* dbiter = db->NewIterator(options); + bool ok = true; + int count = 0; + std::vector<std::string> seek_keys; + // Compare equality of all elements using Next(). Save some of the keys for + // comparing Seek equality. + for (miter->SeekToFirst(), dbiter->SeekToFirst(); + ok && miter->Valid() && dbiter->Valid(); miter->Next(), dbiter->Next()) { + count++; + if (miter->key().compare(dbiter->key()) != 0) { + std::fprintf(stderr, "step %d: Key mismatch: '%s' vs. '%s'\n", step, + EscapeString(miter->key()).c_str(), + EscapeString(dbiter->key()).c_str()); + ok = false; + break; + } + + if (miter->value().compare(dbiter->value()) != 0) { + std::fprintf(stderr, + "step %d: Value mismatch for key '%s': '%s' vs. '%s'\n", + step, EscapeString(miter->key()).c_str(), + EscapeString(miter->value()).c_str(), + EscapeString(miter->value()).c_str()); + ok = false; + break; + } + + if (count % 10 == 0) { + seek_keys.push_back(miter->key().ToString()); + } + } + + if (ok) { + if (miter->Valid() != dbiter->Valid()) { + std::fprintf(stderr, "step %d: Mismatch at end of iterators: %d vs. %d\n", + step, miter->Valid(), dbiter->Valid()); + ok = false; + } + } + + if (ok) { + // Validate iterator equality when performing seeks. + for (auto kiter = seek_keys.begin(); ok && kiter != seek_keys.end(); + ++kiter) { + miter->Seek(*kiter); + dbiter->Seek(*kiter); + if (!miter->Valid() || !dbiter->Valid()) { + std::fprintf(stderr, "step %d: Seek iterators invalid: %d vs. %d\n", + step, miter->Valid(), dbiter->Valid()); + ok = false; + } + if (miter->key().compare(dbiter->key()) != 0) { + std::fprintf(stderr, "step %d: Seek key mismatch: '%s' vs. '%s'\n", + step, EscapeString(miter->key()).c_str(), + EscapeString(dbiter->key()).c_str()); + ok = false; + break; + } + + if (miter->value().compare(dbiter->value()) != 0) { + std::fprintf( + stderr, + "step %d: Seek value mismatch for key '%s': '%s' vs. '%s'\n", step, + EscapeString(miter->key()).c_str(), + EscapeString(miter->value()).c_str(), + EscapeString(miter->value()).c_str()); + ok = false; + break; + } + } + } + + std::fprintf(stderr, "%d entries compared: ok=%d\n", count, ok); + delete miter; + delete dbiter; + return ok; +} + +TEST_F(DBTest, Randomized) { + Random rnd(test::RandomSeed()); + do { + ModelDB model(CurrentOptions()); + const int N = 10000; + const Snapshot* model_snap = nullptr; + const Snapshot* db_snap = nullptr; + std::string k, v; + for (int step = 0; step < N; step++) { + if (step % 100 == 0) { + std::fprintf(stderr, "Step %d of %d\n", step, N); + } + // TODO(sanjay): Test Get() works + int p = rnd.Uniform(100); + if (p < 45) { // Put + k = RandomKey(&rnd); + v = RandomString( + &rnd, rnd.OneIn(20) ? 100 + rnd.Uniform(100) : rnd.Uniform(8)); + ASSERT_LEVELDB_OK(model.Put(WriteOptions(), k, v)); + ASSERT_LEVELDB_OK(db_->Put(WriteOptions(), k, v)); + + } else if (p < 90) { // Delete + k = RandomKey(&rnd); + ASSERT_LEVELDB_OK(model.Delete(WriteOptions(), k)); + ASSERT_LEVELDB_OK(db_->Delete(WriteOptions(), k)); + + } else { // Multi-element batch + WriteBatch b; + const int num = rnd.Uniform(8); + for (int i = 0; i < num; i++) { + if (i == 0 || !rnd.OneIn(10)) { + k = RandomKey(&rnd); + } else { + // Periodically re-use the same key from the previous iter, so + // we have multiple entries in the write batch for the same key + } + if (rnd.OneIn(2)) { + v = RandomString(&rnd, rnd.Uniform(10)); + b.Put(k, v); + } else { + b.Delete(k); + } + } + ASSERT_LEVELDB_OK(model.Write(WriteOptions(), &b)); + ASSERT_LEVELDB_OK(db_->Write(WriteOptions(), &b)); + } + + if ((step % 100) == 0) { + ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr)); + ASSERT_TRUE(CompareIterators(step, &model, db_, model_snap, db_snap)); + // Save a snapshot from each DB this time that we'll use next + // time we compare things, to make sure the current state is + // preserved with the snapshot + if (model_snap != nullptr) model.ReleaseSnapshot(model_snap); + if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap); + + Reopen(); + ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr)); + + model_snap = model.GetSnapshot(); + db_snap = db_->GetSnapshot(); + } + } + if (model_snap != nullptr) model.ReleaseSnapshot(model_snap); + if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap); + } while (ChangeOptions()); +} + +std::string MakeKey(unsigned int num) { + char buf[30]; + std::snprintf(buf, sizeof(buf), "%016u", num); + return std::string(buf); +} + +static void BM_LogAndApply(benchmark::State& state) { + const int num_base_files = state.range(0); + + std::string dbname = testing::TempDir() + "leveldb_test_benchmark"; + DestroyDB(dbname, Options()); + + DB* db = nullptr; + Options opts; + opts.create_if_missing = true; + Status s = DB::Open(opts, dbname, &db); + ASSERT_LEVELDB_OK(s); + ASSERT_TRUE(db != nullptr); + + delete db; + db = nullptr; + + Env* env = Env::Default(); + + port::Mutex mu; + MutexLock l(&mu); + + InternalKeyComparator cmp(BytewiseComparator()); + Options options; + VersionSet vset(dbname, &options, nullptr, &cmp); + bool save_manifest; + ASSERT_LEVELDB_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); + vbase.AddFile(2, fnum++, 1 /* file size */, start, limit); + } + ASSERT_LEVELDB_OK(vset.LogAndApply(&vbase, &mu)); + + uint64_t start_micros = env->NowMicros(); + + for (auto st : state) { + VersionEdit vedit; + vedit.RemoveFile(2, fnum); + 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); + } + uint64_t stop_micros = env->NowMicros(); + unsigned int us = stop_micros - start_micros; + char buf[16]; + std::snprintf(buf, sizeof(buf), "%d", num_base_files); + std::fprintf(stderr, + "BM_LogAndApply/%-6s %8" PRIu64 + " iters : %9u us (%7.0f us / iter)\n", + buf, state.iterations(), us, ((float)us) / state.iterations()); +} + +BENCHMARK(BM_LogAndApply)->Arg(1)->Arg(100)->Arg(10000)->Arg(100000); +} // namespace leveldb + diff --git a/db/dbformat.cc b/db/dbformat.cc new file mode 100644 index 0000000..2a5749f --- /dev/null +++ b/db/dbformat.cc @@ -0,0 +1,136 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/dbformat.h" + +#include <cstdio> +#include <sstream> + +#include "port/port.h" +#include "util/coding.h" + +namespace leveldb { + +static uint64_t PackSequenceAndType(uint64_t seq, ValueType t) { + assert(seq <= kMaxSequenceNumber); + assert(t <= kValueTypeForSeek); + return (seq << 8) | t; +} + +void AppendInternalKey(std::string* result, const ParsedInternalKey& key) { + result->append(key.user_key.data(), key.user_key.size()); + PutFixed64(result, PackSequenceAndType(key.sequence, key.type)); +} + +std::string ParsedInternalKey::DebugString() const { + std::ostringstream ss; + ss << '\'' << EscapeString(user_key.ToString()) << "' @ " << sequence << " : " + << static_cast<int>(type); + return ss.str(); +} + +std::string InternalKey::DebugString() const { + ParsedInternalKey parsed; + if (ParseInternalKey(rep_, &parsed)) { + return parsed.DebugString(); + } + std::ostringstream ss; + ss << "(bad)" << EscapeString(rep_); + return ss.str(); +} + +const char* InternalKeyComparator::Name() const { + return "leveldb.InternalKeyComparator"; +} + +int InternalKeyComparator::Compare(const Slice& akey, const Slice& bkey) const { + // Order by: + // increasing user key (according to user-supplied comparator) + // decreasing sequence number + // decreasing type (though sequence# should be enough to disambiguate) + int r = user_comparator_->Compare(ExtractUserKey(akey), ExtractUserKey(bkey)); + if (r == 0) { + const uint64_t anum = DecodeFixed64(akey.data() + akey.size() - 8); + const uint64_t bnum = DecodeFixed64(bkey.data() + bkey.size() - 8); + if (anum > bnum) { + r = -1; + } else if (anum < bnum) { + r = +1; + } + } + return r; +} + +void InternalKeyComparator::FindShortestSeparator(std::string* start, + const Slice& limit) const { + // Attempt to shorten the user portion of the key + Slice user_start = ExtractUserKey(*start); + Slice user_limit = ExtractUserKey(limit); + std::string tmp(user_start.data(), user_start.size()); + user_comparator_->FindShortestSeparator(&tmp, user_limit); + if (tmp.size() < user_start.size() && + user_comparator_->Compare(user_start, tmp) < 0) { + // User key has become shorter physically, but larger logically. + // Tack on the earliest possible number to the shortened user key. + PutFixed64(&tmp, + PackSequenceAndType(kMaxSequenceNumber, kValueTypeForSeek)); + assert(this->Compare(*start, tmp) < 0); + assert(this->Compare(tmp, limit) < 0); + start->swap(tmp); + } +} + +void InternalKeyComparator::FindShortSuccessor(std::string* key) const { + Slice user_key = ExtractUserKey(*key); + std::string tmp(user_key.data(), user_key.size()); + user_comparator_->FindShortSuccessor(&tmp); + if (tmp.size() < user_key.size() && + user_comparator_->Compare(user_key, tmp) < 0) { + // User key has become shorter physically, but larger logically. + // Tack on the earliest possible number to the shortened user key. + PutFixed64(&tmp, + PackSequenceAndType(kMaxSequenceNumber, kValueTypeForSeek)); + assert(this->Compare(*key, tmp) < 0); + key->swap(tmp); + } +} + +const char* InternalFilterPolicy::Name() const { return user_policy_->Name(); } + +void InternalFilterPolicy::CreateFilter(const Slice* keys, int n, + std::string* dst) const { + // We rely on the fact that the code in table.cc does not mind us + // adjusting keys[]. + Slice* mkey = const_cast<Slice*>(keys); + for (int i = 0; i < n; i++) { + mkey[i] = ExtractUserKey(keys[i]); + // TODO(sanjay): Suppress dups? + } + user_policy_->CreateFilter(keys, n, dst); +} + +bool InternalFilterPolicy::KeyMayMatch(const Slice& key, const Slice& f) const { + return user_policy_->KeyMayMatch(ExtractUserKey(key), f); +} + +LookupKey::LookupKey(const Slice& user_key, SequenceNumber s) { + size_t usize = user_key.size(); + size_t needed = usize + 13; // A conservative estimate + char* dst; + if (needed <= sizeof(space_)) { + dst = space_; + } else { + dst = new char[needed]; + } + start_ = dst; + dst = EncodeVarint32(dst, usize + 8); + kstart_ = dst; + std::memcpy(dst, user_key.data(), usize); + dst += usize; + EncodeFixed64(dst, PackSequenceAndType(s, kValueTypeForSeek)); + dst += 8; + end_ = dst; +} + +} // namespace leveldb diff --git a/db/dbformat.h b/db/dbformat.h new file mode 100644 index 0000000..a1c30ed --- /dev/null +++ b/db/dbformat.h @@ -0,0 +1,224 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_DBFORMAT_H_ +#define STORAGE_LEVELDB_DB_DBFORMAT_H_ + +#include <cstddef> +#include <cstdint> +#include <string> + +#include "leveldb/comparator.h" +#include "leveldb/db.h" +#include "leveldb/filter_policy.h" +#include "leveldb/slice.h" +#include "leveldb/table_builder.h" +#include "util/coding.h" +#include "util/logging.h" + +namespace leveldb { + +// Grouping of constants. We may want to make some of these +// parameters set via options. +namespace config { +static const int kNumLevels = 7; + +// Level-0 compaction is started when we hit this many files. +static const int kL0_CompactionTrigger = 4; + +// Soft limit on number of level-0 files. We slow down writes at this point. +static const int kL0_SlowdownWritesTrigger = 8; + +// Maximum number of level-0 files. We stop writes at this point. +static const int kL0_StopWritesTrigger = 12; + +// Maximum level to which a new compacted memtable is pushed if it +// does not create overlap. We try to push to level 2 to avoid the +// relatively expensive level 0=>1 compactions and to avoid some +// expensive manifest file operations. We do not push all the way to +// the largest level since that can generate a lot of wasted disk +// space if the same key space is being repeatedly overwritten. +static const int kMaxMemCompactLevel = 2; + +// Approximate gap in bytes between samples of data read during iteration. +static const int kReadBytesPeriod = 1048576; + +} // namespace config + +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 }; +// kValueTypeForSeek defines the ValueType that should be passed when +// constructing a ParsedInternalKey object for seeking to a particular +// sequence number (since we sort sequence numbers in decreasing order +// and the value type is embedded as the low 8 bits in the sequence +// number in internal keys, we need to use the highest-numbered +// ValueType, not the lowest). +static const ValueType kValueTypeForSeek = kTypeValue; + +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); + +struct ParsedInternalKey { + Slice user_key; + SequenceNumber sequence; + ValueType type; + + ParsedInternalKey() {} // Intentionally left uninitialized (for speed) + ParsedInternalKey(const Slice& u, const SequenceNumber& seq, ValueType t) + : user_key(u), sequence(seq), type(t) {} + std::string DebugString() const; +}; + +// Return the length of the encoding of "key". +inline size_t InternalKeyEncodingLength(const ParsedInternalKey& key) { + return key.user_key.size() + 8; +} + +// Append the serialization of "key" to *result. +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. +bool ParseInternalKey(const Slice& internal_key, ParsedInternalKey* result); + +// Returns the user key portion of an internal key. +inline Slice ExtractUserKey(const Slice& internal_key) { + assert(internal_key.size() >= 8); + return Slice(internal_key.data(), internal_key.size() - 8); +} + +// 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) {} + 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_; } + + int Compare(const InternalKey& a, const InternalKey& b) const; +}; + +// Filter policy wrapper that converts from internal keys to user keys +class InternalFilterPolicy : public FilterPolicy { + private: + const FilterPolicy* const user_policy_; + + public: + 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 +// the following class instead of plain strings so that we do not +// incorrectly use string comparisons instead of an InternalKeyComparator. +class InternalKey { + private: + std::string rep_; + + public: + InternalKey() {} // Leave rep_ as empty to indicate it is invalid + InternalKey(const Slice& user_key, SequenceNumber s, ValueType t) { + AppendInternalKey(&rep_, ParsedInternalKey(user_key, s, t)); + } + + bool DecodeFrom(const Slice& s) { + rep_.assign(s.data(), s.size()); + return !rep_.empty(); + } + + Slice Encode() const { + assert(!rep_.empty()); + return rep_; + } + + Slice user_key() const { return ExtractUserKey(rep_); } + + void SetFrom(const ParsedInternalKey& p) { + rep_.clear(); + AppendInternalKey(&rep_, p); + } + + void Clear() { rep_.clear(); } + + std::string DebugString() const; +}; + +inline int InternalKeyComparator::Compare(const InternalKey& a, + const InternalKey& b) const { + return Compare(a.Encode(), b.Encode()); +} + +inline bool ParseInternalKey(const Slice& internal_key, + ParsedInternalKey* result) { + const size_t n = internal_key.size(); + if (n < 8) return false; + uint64_t num = DecodeFixed64(internal_key.data() + n - 8); + uint8_t c = num & 0xff; + result->sequence = num >> 8; + result->type = static_cast<ValueType>(c); + result->user_key = Slice(internal_key.data(), n - 8); + return (c <= static_cast<uint8_t>(kTypeValue)); +} + +// A helper class useful for DBImpl::Get() +class LookupKey { + public: + // Initialize *this for looking up user_key at a snapshot with + // the specified sequence number. + LookupKey(const Slice& user_key, SequenceNumber sequence); + + LookupKey(const LookupKey&) = delete; + LookupKey& operator=(const LookupKey&) = delete; + + ~LookupKey(); + + // Return a key suitable for lookup in a MemTable. + Slice memtable_key() const { return Slice(start_, end_ - start_); } + + // Return an internal key (suitable for passing to an internal iterator) + Slice internal_key() const { return Slice(kstart_, end_ - kstart_); } + + // Return the user key + Slice user_key() const { return Slice(kstart_, end_ - kstart_ - 8); } + + private: + // We construct a char array of the form: + // klength varint32 <-- start_ + // userkey char[klength] <-- kstart_ + // tag uint64 + // <-- end_ + // The array is a suitable MemTable key. + // The suffix starting with "userkey" can be used as an InternalKey. + const char* start_; + const char* kstart_; + const char* end_; + char space_[200]; // Avoid allocation for short keys +}; + +inline LookupKey::~LookupKey() { + if (start_ != space_) delete[] start_; +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_DBFORMAT_H_ diff --git a/db/dbformat_test.cc b/db/dbformat_test.cc new file mode 100644 index 0000000..36dd220 --- /dev/null +++ b/db/dbformat_test.cc @@ -0,0 +1,129 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/dbformat.h" + +#include "gtest/gtest.h" +#include "util/logging.h" + +namespace leveldb { + +static std::string IKey(const std::string& user_key, uint64_t seq, + ValueType vt) { + std::string encoded; + AppendInternalKey(&encoded, ParsedInternalKey(user_key, seq, vt)); + return encoded; +} + +static std::string Shorten(const std::string& s, const std::string& l) { + std::string result = s; + InternalKeyComparator(BytewiseComparator()).FindShortestSeparator(&result, l); + return result; +} + +static std::string ShortSuccessor(const std::string& s) { + std::string result = s; + InternalKeyComparator(BytewiseComparator()).FindShortSuccessor(&result); + return result; +} + +static void TestKey(const std::string& key, uint64_t seq, ValueType vt) { + std::string encoded = IKey(key, seq, vt); + + Slice in(encoded); + ParsedInternalKey decoded("", 0, kTypeValue); + + ASSERT_TRUE(ParseInternalKey(in, &decoded)); + ASSERT_EQ(key, decoded.user_key.ToString()); + ASSERT_EQ(seq, decoded.sequence); + ASSERT_EQ(vt, decoded.type); + + ASSERT_TRUE(!ParseInternalKey(Slice("bar"), &decoded)); +} + +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}; + 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); + TestKey("hello", 1, kTypeDeletion); + } + } +} + +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))); + + // When user keys are misordered + ASSERT_EQ(IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), IKey("bar", 99, kTypeValue))); + + // When user keys are different, but correctly ordered + ASSERT_EQ( + IKey("g", kMaxSequenceNumber, kValueTypeForSeek), + Shorten(IKey("foo", 100, kTypeValue), IKey("hello", 200, kTypeValue))); + + // When start user key is prefix of limit user key + ASSERT_EQ( + IKey("foo", 100, kTypeValue), + Shorten(IKey("foo", 100, kTypeValue), IKey("foobar", 200, kTypeValue))); + + // When limit user key is prefix of start user key + ASSERT_EQ( + IKey("foobar", 100, kTypeValue), + Shorten(IKey("foobar", 100, kTypeValue), IKey("foo", 200, kTypeValue))); +} + +TEST(FormatTest, InternalKeyShortestSuccessor) { + ASSERT_EQ(IKey("g", kMaxSequenceNumber, kValueTypeForSeek), + ShortSuccessor(IKey("foo", 100, kTypeValue))); + ASSERT_EQ(IKey("\xff\xff", 100, kTypeValue), + ShortSuccessor(IKey("\xff\xff", 100, kTypeValue))); +} + +TEST(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 + diff --git a/db/dumpfile.cc b/db/dumpfile.cc new file mode 100644 index 0000000..6085475 --- /dev/null +++ b/db/dumpfile.cc @@ -0,0 +1,232 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "leveldb/dumpfile.h" + +#include <cstdio> + +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/version_edit.h" +#include "db/write_batch_internal.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" +#include "leveldb/options.h" +#include "leveldb/status.h" +#include "leveldb/table.h" +#include "leveldb/write_batch.h" +#include "util/logging.h" + +namespace leveldb { + +namespace { + +bool GuessType(const std::string& fname, FileType* type) { + size_t pos = fname.rfind('/'); + std::string basename; + if (pos == std::string::npos) { + basename = fname; + } else { + basename = std::string(fname.data() + pos + 1, fname.size() - pos - 1); + } + uint64_t ignored; + return ParseFileName(basename, &ignored, type); +} + +// Notified when log reader encounters corruption. +class CorruptionReporter : public log::Reader::Reporter { + public: + void Corruption(size_t bytes, const Status& status) override { + std::string r = "corruption: "; + AppendNumberTo(&r, bytes); + r += " bytes; "; + r += status.ToString(); + r.push_back('\n'); + dst_->Append(r); + } + + WritableFile* dst_; +}; + +// Print contents of a log file. (*func)() is called on every record. +Status PrintLogContents(Env* env, const std::string& fname, + void (*func)(uint64_t, Slice, WritableFile*), + WritableFile* dst) { + SequentialFile* file; + Status s = env->NewSequentialFile(fname, &file); + if (!s.ok()) { + return s; + } + CorruptionReporter reporter; + reporter.dst_ = dst; + log::Reader reader(file, &reporter, true, 0); + Slice record; + std::string scratch; + while (reader.ReadRecord(&record, &scratch)) { + (*func)(reader.LastRecordOffset(), record, dst); + } + delete file; + return Status::OK(); +} + +// Called on every item found in a WriteBatch. +class WriteBatchItemPrinter : public WriteBatch::Handler { + public: + void Put(const Slice& key, const Slice& value) override { + std::string r = " put '"; + AppendEscapedStringTo(&r, key); + r += "' '"; + AppendEscapedStringTo(&r, value); + r += "'\n"; + dst_->Append(r); + } + 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. +static void WriteBatchPrinter(uint64_t pos, Slice record, WritableFile* dst) { + std::string r = "--- offset "; + AppendNumberTo(&r, pos); + r += "; "; + if (record.size() < 12) { + r += "log record length "; + AppendNumberTo(&r, record.size()); + r += " is too small\n"; + dst->Append(r); + return; + } + WriteBatch batch; + WriteBatchInternal::SetContents(&batch, record); + r += "sequence "; + AppendNumberTo(&r, WriteBatchInternal::Sequence(&batch)); + r.push_back('\n'); + dst->Append(r); + WriteBatchItemPrinter batch_item_printer; + batch_item_printer.dst_ = dst; + Status s = batch.Iterate(&batch_item_printer); + if (!s.ok()) { + dst->Append(" error: " + s.ToString() + "\n"); + } +} + +Status DumpLog(Env* env, const std::string& fname, WritableFile* dst) { + return PrintLogContents(env, fname, WriteBatchPrinter, dst); +} + +// Called on every log record (each one of which is a WriteBatch) +// found in a kDescriptorFile. +static void VersionEditPrinter(uint64_t pos, Slice record, WritableFile* dst) { + std::string r = "--- offset "; + AppendNumberTo(&r, pos); + r += "; "; + VersionEdit edit; + Status s = edit.DecodeFrom(record); + if (!s.ok()) { + r += s.ToString(); + r.push_back('\n'); + } else { + r += edit.DebugString(); + } + dst->Append(r); +} + +Status DumpDescriptor(Env* env, const std::string& fname, WritableFile* dst) { + return PrintLogContents(env, fname, VersionEditPrinter, dst); +} + +Status DumpTable(Env* env, const std::string& fname, WritableFile* dst) { + uint64_t file_size; + RandomAccessFile* file = nullptr; + Table* table = nullptr; + Status s = env->GetFileSize(fname, &file_size); + if (s.ok()) { + s = env->NewRandomAccessFile(fname, &file); + } + if (s.ok()) { + // We use the default comparator, which may or may not match the + // comparator used in this database. However this should not cause + // problems since we only use Table operations that do not require + // any comparisons. In particular, we do not call Seek or Prev. + s = Table::Open(Options(), file, file_size, &table); + } + if (!s.ok()) { + delete table; + delete file; + return s; + } + + ReadOptions ro; + ro.fill_cache = false; + Iterator* iter = table->NewIterator(ro); + std::string r; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + r.clear(); + ParsedInternalKey key; + if (!ParseInternalKey(iter->key(), &key)) { + r = "badkey '"; + AppendEscapedStringTo(&r, iter->key()); + r += "' => '"; + AppendEscapedStringTo(&r, iter->value()); + r += "'\n"; + dst->Append(r); + } else { + r = "'"; + AppendEscapedStringTo(&r, key.user_key); + r += "' @ "; + AppendNumberTo(&r, key.sequence); + r += " : "; + if (key.type == kTypeDeletion) { + r += "del"; + } else if (key.type == kTypeValue) { + r += "val"; + } else { + AppendNumberTo(&r, key.type); + } + r += " => '"; + AppendEscapedStringTo(&r, iter->value()); + r += "'\n"; + dst->Append(r); + } + } + s = iter->status(); + if (!s.ok()) { + dst->Append("iterator error: " + s.ToString() + "\n"); + } + + delete iter; + delete table; + delete file; + return Status::OK(); +} + +} // namespace + +Status DumpFile(Env* env, const std::string& fname, WritableFile* dst) { + FileType ftype; + if (!GuessType(fname, &ftype)) { + 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); + default: + break; + } + return Status::InvalidArgument(fname + ": not a dump-able file type"); +} + +} // namespace leveldb diff --git a/db/fault_injection_test.cc b/db/fault_injection_test.cc new file mode 100644 index 0000000..0522762 --- /dev/null +++ b/db/fault_injection_test.cc @@ -0,0 +1,551 @@ +// Copyright 2014 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +// This test uses a custom Env to keep track of the state of a filesystem as of +// the last "sync". It then checks for data loss errors by purposely dropping +// file data (or entire files) not protected by a "sync". + +#include <map> +#include <set> + +#include "gtest/gtest.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/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/testutil.h" + +namespace leveldb { + +static const int kValueSize = 1000; +static const int kMaxNumValues = 2000; +static const size_t kNumIterations = 3; + +class FaultInjectionTestEnv; + +namespace { + +// Assume a filename, and not a directory name like "/foo/bar/" +static std::string GetDirName(const std::string& filename) { + size_t found = filename.find_last_of("/\\"); + if (found == std::string::npos) { + return ""; + } else { + return filename.substr(0, found); + } +} + +Status SyncDir(const std::string& dir) { + // As this is a test it isn't required to *actually* sync this directory. + return Status::OK(); +} + +// A basic file truncation function suitable for this test. +Status Truncate(const std::string& filename, uint64_t length) { + leveldb::Env* env = leveldb::Env::Default(); + + SequentialFile* orig_file; + Status s = env->NewSequentialFile(filename, &orig_file); + if (!s.ok()) return s; + + char* scratch = new char[length]; + leveldb::Slice result; + s = orig_file->Read(length, &result, scratch); + delete orig_file; + if (s.ok()) { + std::string tmp_name = GetDirName(filename) + "/truncate.tmp"; + WritableFile* tmp_file; + s = env->NewWritableFile(tmp_name, &tmp_file); + if (s.ok()) { + s = tmp_file->Append(result); + delete tmp_file; + if (s.ok()) { + s = env->RenameFile(tmp_name, filename); + } else { + env->RemoveFile(tmp_name); + } + } + } + + delete[] scratch; + + return s; +} + +struct FileState { + std::string filename_; + 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) {} + + FileState() : pos_(-1), pos_at_last_sync_(-1), pos_at_last_flush_(-1) {} + + bool IsFullySynced() const { return pos_ <= 0 || pos_ == pos_at_last_sync_; } + + Status DropUnsyncedData() const; +}; + +} // anonymous namespace + +// A wrapper around WritableFile which informs another Env whenever this file +// is written to or sync'ed. +class TestWritableFile : public WritableFile { + public: + TestWritableFile(const FileState& state, WritableFile* f, + FaultInjectionTestEnv* env); + ~TestWritableFile() override; + Status Append(const Slice& data) override; + Status Close() override; + Status Flush() override; + Status Sync() override; + + private: + FileState state_; + WritableFile* target_; + bool writable_file_opened_; + FaultInjectionTestEnv* env_; + + Status SyncParent(); +}; + +class FaultInjectionTestEnv : public EnvWrapper { + public: + 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 RemoveFile(const std::string& f) override; + Status RenameFile(const std::string& s, const std::string& t) override; + + void WritableFileClosed(const FileState& state); + Status DropUnsyncedFileData(); + Status RemoveFilesCreatedAfterLastDirSync(); + void DirWasSynced(); + bool IsFileCreatedSinceLastDirSync(const std::string& filename); + void ResetState(); + void UntrackFile(const std::string& f); + // Setting the filesystem to inactive is the test equivalent to simulating a + // system reset. Setting to inactive will freeze our saved filesystem state so + // that it will stop being recorded. It can then be reset back to the state at + // the time of the reset. + bool IsFilesystemActive() 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<std::string, FileState> db_file_state_ GUARDED_BY(mutex_); + std::set<std::string> new_files_since_last_dir_sync_ GUARDED_BY(mutex_); + bool filesystem_active_ GUARDED_BY(mutex_); // Record flushes, syncs, writes +}; + +TestWritableFile::TestWritableFile(const FileState& state, WritableFile* f, + FaultInjectionTestEnv* env) + : state_(state), target_(f), writable_file_opened_(true), env_(env) { + assert(f != nullptr); +} + +TestWritableFile::~TestWritableFile() { + if (writable_file_opened_) { + Close(); + } + delete target_; +} + +Status TestWritableFile::Append(const Slice& data) { + Status s = target_->Append(data); + if (s.ok() && env_->IsFilesystemActive()) { + state_.pos_ += data.size(); + } + return s; +} + +Status TestWritableFile::Close() { + writable_file_opened_ = false; + Status s = target_->Close(); + if (s.ok()) { + env_->WritableFileClosed(state_); + } + return s; +} + +Status TestWritableFile::Flush() { + Status s = target_->Flush(); + if (s.ok() && env_->IsFilesystemActive()) { + state_.pos_at_last_flush_ = state_.pos_; + } + return s; +} + +Status TestWritableFile::SyncParent() { + Status s = SyncDir(GetDirName(state_.filename_)); + if (s.ok()) { + env_->DirWasSynced(); + } + return s; +} + +Status TestWritableFile::Sync() { + if (!env_->IsFilesystemActive()) { + return Status::OK(); + } + // Ensure new files referred to by the manifest are in the filesystem. + Status s = target_->Sync(); + if (s.ok()) { + state_.pos_at_last_sync_ = state_.pos_; + } + if (env_->IsFileCreatedSinceLastDirSync(state_.filename_)) { + Status ps = SyncParent(); + if (s.ok() && !ps.ok()) { + s = ps; + } + } + return s; +} + +Status FaultInjectionTestEnv::NewWritableFile(const std::string& fname, + WritableFile** result) { + WritableFile* actual_writable_file; + Status s = target()->NewWritableFile(fname, &actual_writable_file); + if (s.ok()) { + 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); + } + 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 (const auto& kvp : db_file_state_) { + if (!s.ok()) { + break; + } + const FileState& state = kvp.second; + if (!state.IsFullySynced()) { + s = state.DropUnsyncedData(); + } + } + return s; +} + +void FaultInjectionTestEnv::DirWasSynced() { + MutexLock l(&mutex_); + new_files_since_last_dir_sync_.clear(); +} + +bool FaultInjectionTestEnv::IsFileCreatedSinceLastDirSync( + const std::string& filename) { + MutexLock l(&mutex_); + return new_files_since_last_dir_sync_.find(filename) != + new_files_since_last_dir_sync_.end(); +} + +void FaultInjectionTestEnv::UntrackFile(const std::string& f) { + MutexLock l(&mutex_); + db_file_state_.erase(f); + new_files_since_last_dir_sync_.erase(f); +} + +Status FaultInjectionTestEnv::RemoveFile(const std::string& f) { + Status s = EnvWrapper::RemoveFile(f); + EXPECT_LEVELDB_OK(s); + if (s.ok()) { + UntrackFile(f); + } + return s; +} + +Status FaultInjectionTestEnv::RenameFile(const std::string& s, + const std::string& t) { + Status ret = EnvWrapper::RenameFile(s, t); + + if (ret.ok()) { + MutexLock l(&mutex_); + if (db_file_state_.find(s) != db_file_state_.end()) { + db_file_state_[t] = db_file_state_[s]; + db_file_state_.erase(s); + } + + if (new_files_since_last_dir_sync_.erase(s) != 0) { + assert(new_files_since_last_dir_sync_.find(t) == + new_files_since_last_dir_sync_.end()); + new_files_since_last_dir_sync_.insert(t); + } + } + + return ret; +} + +void FaultInjectionTestEnv::ResetState() { + // 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); +} + +Status FaultInjectionTestEnv::RemoveFilesCreatedAfterLastDirSync() { + // Because RemoveFile access this container make a copy to avoid deadlock + mutex_.Lock(); + std::set<std::string> new_files(new_files_since_last_dir_sync_.begin(), + new_files_since_last_dir_sync_.end()); + mutex_.Unlock(); + Status status; + for (const auto& new_file : new_files) { + Status remove_status = RemoveFile(new_file); + if (!remove_status.ok() && status.ok()) { + status = std::move(remove_status); + } + } + return status; +} + +void FaultInjectionTestEnv::WritableFileClosed(const FileState& state) { + MutexLock l(&mutex_); + db_file_state_[state.filename_] = state; +} + +Status FileState::DropUnsyncedData() const { + int64_t sync_pos = pos_at_last_sync_ == -1 ? 0 : pos_at_last_sync_; + return Truncate(filename_, sync_pos); +} + +class FaultInjectionTest : public testing::Test { + public: + enum ExpectedVerifResult { VAL_EXPECT_NO_ERROR, VAL_EXPECT_ERROR }; + enum ResetMethod { RESET_DROP_UNSYNCED_DATA, RESET_DELETE_UNSYNCED_FILES }; + + FaultInjectionTestEnv* env_; + std::string dbname_; + Cache* tiny_cache_; + Options options_; + DB* db_; + + FaultInjectionTest() + : env_(new FaultInjectionTestEnv), + tiny_cache_(NewLRUCache(100)), + db_(nullptr) { + dbname_ = testing::TempDir() + "fault_test"; + DestroyDB(dbname_, Options()); // Destroy any db from earlier run + options_.reuse_logs = true; + options_.env = env_; + options_.paranoid_checks = true; + options_.block_cache = tiny_cache_; + options_.create_if_missing = true; + } + + ~FaultInjectionTest() { + CloseDB(); + DestroyDB(dbname_, Options()); + delete tiny_cache_; + delete env_; + } + + void ReuseLogs(bool reuse) { options_.reuse_logs = reuse; } + + void Build(int start_idx, int num_vals) { + std::string key_space, value_space; + WriteBatch batch; + for (int i = start_idx; i < start_idx + num_vals; i++) { + Slice key = Key(i, &key_space); + batch.Clear(); + batch.Put(key, Value(i, &value_space)); + WriteOptions options; + ASSERT_LEVELDB_OK(db_->Write(options, &batch)); + } + } + + Status ReadValue(int i, std::string* val) const { + std::string key_space, value_space; + Slice key = Key(i, &key_space); + Value(i, &value_space); + ReadOptions options; + return db_->Get(options, key, val); + } + + Status Verify(int start_idx, int num_vals, + ExpectedVerifResult expected) const { + std::string val; + std::string value_space; + Status s; + for (int i = start_idx; i < start_idx + num_vals && s.ok(); i++) { + Value(i, &value_space); + s = ReadValue(i, &val); + if (expected == VAL_EXPECT_NO_ERROR) { + if (s.ok()) { + EXPECT_EQ(value_space, val); + } + } else if (s.ok()) { + std::fprintf(stderr, "Expected an error at %d, but was OK\n", i); + s = Status::IOError(dbname_, "Expected value error:"); + } else { + s = Status::OK(); // An expected error + } + } + return s; + } + + // Return the ith key + Slice Key(int i, std::string* storage) const { + char buf[100]; + std::snprintf(buf, sizeof(buf), "%016d", i); + storage->assign(buf, strlen(buf)); + return Slice(*storage); + } + + // Return the value to associate with the specified key + Slice Value(int k, std::string* storage) const { + Random r(k); + return test::RandomString(&r, kValueSize, storage); + } + + Status OpenDB() { + delete db_; + db_ = nullptr; + env_->ResetState(); + return DB::Open(options_, dbname_, &db_); + } + + void CloseDB() { + delete db_; + db_ = nullptr; + } + + void DeleteAllData() { + Iterator* iter = db_->NewIterator(ReadOptions()); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + ASSERT_LEVELDB_OK(db_->Delete(WriteOptions(), iter->key())); + } + + delete iter; + } + + void ResetDBState(ResetMethod reset_method) { + switch (reset_method) { + case RESET_DROP_UNSYNCED_DATA: + ASSERT_LEVELDB_OK(env_->DropUnsyncedFileData()); + break; + case RESET_DELETE_UNSYNCED_FILES: + ASSERT_LEVELDB_OK(env_->RemoveFilesCreatedAfterLastDirSync()); + break; + default: + assert(false); + } + } + + void PartialCompactTestPreFault(int num_pre_sync, int num_post_sync) { + DeleteAllData(); + Build(0, num_pre_sync); + db_->CompactRange(nullptr, nullptr); + Build(num_pre_sync, num_post_sync); + } + + void PartialCompactTestReopenWithFault(ResetMethod reset_method, + int num_pre_sync, int num_post_sync) { + env_->SetFilesystemActive(false); + CloseDB(); + ResetDBState(reset_method); + ASSERT_LEVELDB_OK(OpenDB()); + ASSERT_LEVELDB_OK( + Verify(0, num_pre_sync, FaultInjectionTest::VAL_EXPECT_NO_ERROR)); + ASSERT_LEVELDB_OK(Verify(num_pre_sync, num_post_sync, + FaultInjectionTest::VAL_EXPECT_ERROR)); + } + + void NoWriteTestPreFault() {} + + void NoWriteTestReopenWithFault(ResetMethod reset_method) { + CloseDB(); + ResetDBState(reset_method); + ASSERT_LEVELDB_OK(OpenDB()); + } + + void DoTest() { + Random rnd(0); + ASSERT_LEVELDB_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_F(FaultInjectionTest, FaultTestNoLogReuse) { + ReuseLogs(false); + DoTest(); +} + +TEST_F(FaultInjectionTest, FaultTestWithLogReuse) { + ReuseLogs(true); + DoTest(); +} + +} // namespace leveldb + diff --git a/db/filename.cc b/db/filename.cc new file mode 100644 index 0000000..e526249 --- /dev/null +++ b/db/filename.cc @@ -0,0 +1,141 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/filename.h" + +#include <cassert> +#include <cstdio> + +#include "db/dbformat.h" +#include "leveldb/env.h" +#include "util/logging.h" + +namespace leveldb { + +// A utility routine: write "data" to the named file and Sync() it. +Status WriteStringToFileSync(Env* env, const Slice& data, + const std::string& fname); + +static std::string MakeFileName(const std::string& dbname, uint64_t number, + const char* suffix) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "/%06llu.%s", + static_cast<unsigned long long>(number), suffix); + return dbname + buf; +} + +std::string LogFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + return MakeFileName(dbname, number, "log"); +} + +std::string TableFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + return MakeFileName(dbname, number, "ldb"); +} + +std::string SSTTableFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + return MakeFileName(dbname, number, "sst"); +} + +std::string DescriptorFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + char buf[100]; + std::snprintf(buf, sizeof(buf), "/MANIFEST-%06llu", + static_cast<unsigned long long>(number)); + return dbname + buf; +} + +std::string CurrentFileName(const std::string& dbname) { + return dbname + "/CURRENT"; +} + +std::string LockFileName(const std::string& dbname) { return dbname + "/LOCK"; } + +std::string TempFileName(const std::string& dbname, uint64_t number) { + assert(number > 0); + return MakeFileName(dbname, number, "dbtmp"); +} + +std::string InfoLogFileName(const std::string& dbname) { + return dbname + "/LOG"; +} + +// Return the name of the old info log file for "dbname". +std::string OldInfoLogFileName(const std::string& dbname) { + return dbname + "/LOG.old"; +} + +// Owned filenames have the form: +// dbname/CURRENT +// dbname/LOCK +// dbname/LOG +// dbname/LOG.old +// dbname/MANIFEST-[0-9]+ +// dbname/[0-9]+.(log|sst|ldb) +bool ParseFileName(const std::string& filename, uint64_t* number, + FileType* type) { + Slice rest(filename); + if (rest == "CURRENT") { + *number = 0; + *type = kCurrentFile; + } else if (rest == "LOCK") { + *number = 0; + *type = kDBLockFile; + } else if (rest == "LOG" || rest == "LOG.old") { + *number = 0; + *type = kInfoLogFile; + } else if (rest.starts_with("MANIFEST-")) { + rest.remove_prefix(strlen("MANIFEST-")); + uint64_t num; + if (!ConsumeDecimalNumber(&rest, &num)) { + return false; + } + if (!rest.empty()) { + return false; + } + *type = kDescriptorFile; + *number = num; + } else { + // Avoid strtoull() to keep filename format independent of the + // current locale + uint64_t num; + if (!ConsumeDecimalNumber(&rest, &num)) { + return false; + } + Slice suffix = rest; + if (suffix == Slice(".log")) { + *type = kLogFile; + } else if (suffix == Slice(".sst") || suffix == Slice(".ldb")) { + *type = kTableFile; + } else if (suffix == Slice(".dbtmp")) { + *type = kTempFile; + } else { + return false; + } + *number = num; + } + return true; +} + +Status SetCurrentFile(Env* env, const std::string& dbname, + uint64_t descriptor_number) { + // Remove leading "dbname/" and add newline to manifest file name + std::string manifest = DescriptorFileName(dbname, descriptor_number); + Slice contents = manifest; + assert(contents.starts_with(dbname + "/")); + contents.remove_prefix(dbname.size() + 1); + std::string tmp = TempFileName(dbname, descriptor_number); + Status s = WriteStringToFileSync(env, contents.ToString() + "\n", tmp); + if (s.ok()) { + s = env->RenameFile(tmp, CurrentFileName(dbname)); + } + if (!s.ok()) { + env->RemoveFile(tmp); + } + return s; +} + +} // namespace leveldb diff --git a/db/filename.h b/db/filename.h new file mode 100644 index 0000000..563c6d8 --- /dev/null +++ b/db/filename.h @@ -0,0 +1,83 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// File names used by DB code + +#ifndef STORAGE_LEVELDB_DB_FILENAME_H_ +#define STORAGE_LEVELDB_DB_FILENAME_H_ + +#include <cstdint> +#include <string> + +#include "leveldb/slice.h" +#include "leveldb/status.h" +#include "port/port.h" + +namespace leveldb { + +class Env; + +enum FileType { + kLogFile, + kDBLockFile, + kTableFile, + kDescriptorFile, + kCurrentFile, + kTempFile, + kInfoLogFile // Either the current one, or an old one +}; + +// Return the name of the log file with the specified number +// in the db named by "dbname". The result will be prefixed with +// "dbname". +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". +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". +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". +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". +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". +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". +std::string TempFileName(const std::string& dbname, uint64_t number); + +// Return the name of the info log file for "dbname". +std::string InfoLogFileName(const std::string& dbname); + +// Return the name of the old info log file for "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. +bool ParseFileName(const std::string& filename, uint64_t* number, + FileType* type); + +// Make the CURRENT file point to the descriptor file with the +// specified number. +Status SetCurrentFile(Env* env, const std::string& dbname, + uint64_t descriptor_number); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_FILENAME_H_ diff --git a/db/filename_test.cc b/db/filename_test.cc new file mode 100644 index 0000000..5731ac3 --- /dev/null +++ b/db/filename_test.cc @@ -0,0 +1,128 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/filename.h" + +#include "gtest/gtest.h" +#include "db/dbformat.h" +#include "port/port.h" +#include "util/logging.h" + +namespace leveldb { + +TEST(FileNameTest, Parse) { + Slice db; + FileType type; + uint64_t number; + + // Successful parses + static struct { + const char* fname; + 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}, + }; + for (int i = 0; i < sizeof(cases) / sizeof(cases[0]); i++) { + std::string f = cases[i].fname; + ASSERT_TRUE(ParseFileName(f, &number, &type)) << f; + ASSERT_EQ(cases[i].type, type) << f; + ASSERT_EQ(cases[i].number, number) << f; + } + + // Errors + static const char* errors[] = {"", + "foo", + "foo-dx-100.log", + ".log", + "", + "manifest", + "CURREN", + "CURRENTX", + "MANIFES", + "MANIFEST", + "MANIFEST-", + "XMANIFEST-3", + "MANIFEST-3x", + "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; + } +} + +TEST(FileNameTest, Construction) { + uint64_t number; + FileType type; + std::string fname; + + fname = CurrentFileName("foo"); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(0, number); + ASSERT_EQ(kCurrentFile, type); + + fname = LockFileName("foo"); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(0, number); + ASSERT_EQ(kDBLockFile, type); + + fname = LogFileName("foo", 192); + ASSERT_EQ("foo/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(192, number); + ASSERT_EQ(kLogFile, type); + + fname = TableFileName("bar", 200); + ASSERT_EQ("bar/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(200, number); + ASSERT_EQ(kTableFile, type); + + fname = DescriptorFileName("bar", 100); + ASSERT_EQ("bar/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(100, number); + ASSERT_EQ(kDescriptorFile, type); + + fname = TempFileName("tmp", 999); + ASSERT_EQ("tmp/", std::string(fname.data(), 4)); + ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type)); + ASSERT_EQ(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 + diff --git a/db/leveldbutil.cc b/db/leveldbutil.cc new file mode 100644 index 0000000..95ee897 --- /dev/null +++ b/db/leveldbutil.cc @@ -0,0 +1,64 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include <cstdio> + +#include "leveldb/dumpfile.h" +#include "leveldb/env.h" +#include "leveldb/status.h" + +namespace leveldb { +namespace { + +class StdoutPrinter : public WritableFile { + public: + Status Append(const Slice& data) override { + fwrite(data.data(), 1, data.size(), stdout); + 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) { + StdoutPrinter printer; + bool ok = true; + for (int i = 0; i < num; i++) { + Status s = DumpFile(env, files[i], &printer); + if (!s.ok()) { + std::fprintf(stderr, "%s\n", s.ToString().c_str()); + ok = false; + } + } + return ok; +} + +} // namespace +} // namespace leveldb + +static void Usage() { + std::fprintf( + stderr, + "Usage: leveldbutil command...\n" + " dump files... -- dump contents of specified files\n"); +} + +int main(int argc, char** argv) { + leveldb::Env* env = leveldb::Env::Default(); + bool ok = true; + if (argc < 2) { + Usage(); + ok = false; + } else { + std::string command = argv[1]; + if (command == "dump") { + ok = leveldb::HandleDumpCommand(env, argv + 2, argc - 2); + } else { + Usage(); + ok = false; + } + } + return (ok ? 0 : 1); +} diff --git a/db/log_format.h b/db/log_format.h new file mode 100644 index 0000000..356e69f --- /dev/null +++ b/db/log_format.h @@ -0,0 +1,35 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// Log format information shared by reader and writer. +// See ../doc/log_format.md for more detail. + +#ifndef STORAGE_LEVELDB_DB_LOG_FORMAT_H_ +#define STORAGE_LEVELDB_DB_LOG_FORMAT_H_ + +namespace leveldb { +namespace log { + +enum RecordType { + // Zero is reserved for preallocated files + kZeroType = 0, + + kFullType = 1, + + // For fragments + kFirstType = 2, + kMiddleType = 3, + kLastType = 4 +}; +static const int kMaxRecordType = kLastType; + +static const int kBlockSize = 32768; + +// Header is checksum (4 bytes), length (2 bytes), type (1 byte). +static const int kHeaderSize = 4 + 2 + 1; + +} // namespace log +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_LOG_FORMAT_H_ diff --git a/db/log_reader.cc b/db/log_reader.cc new file mode 100644 index 0000000..9880279 --- /dev/null +++ b/db/log_reader.cc @@ -0,0 +1,274 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/log_reader.h" + +#include <cstdio> + +#include "leveldb/env.h" +#include "util/coding.h" +#include "util/crc32c.h" + +namespace leveldb { +namespace log { + +Reader::Reporter::~Reporter() = default; + +Reader::Reader(SequentialFile* file, Reporter* reporter, bool checksum, + uint64_t initial_offset) + : file_(file), + reporter_(reporter), + checksum_(checksum), + backing_store_(new char[kBlockSize]), + buffer_(), + eof_(false), + last_record_offset_(0), + end_of_buffer_offset_(0), + initial_offset_(initial_offset), + resyncing_(initial_offset > 0) {} + +Reader::~Reader() { delete[] backing_store_; } + +bool Reader::SkipToInitialBlock() { + 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) { + block_start_location += kBlockSize; + } + + end_of_buffer_offset_ = block_start_location; + + // Skip to start of first block that can contain the initial record + if (block_start_location > 0) { + Status skip_status = file_->Skip(block_start_location); + if (!skip_status.ok()) { + ReportDrop(block_start_location, skip_status); + return false; + } + } + + return true; +} + +bool Reader::ReadRecord(Slice* record, std::string* scratch) { + if (last_record_offset_ < initial_offset_) { + if (!SkipToInitialBlock()) { + return false; + } + } + + scratch->clear(); + record->clear(); + bool in_fragmented_record = false; + // Record offset of the logical record that we're reading + // 0 is a dummy value to make compilers happy + uint64_t prospective_record_offset = 0; + + Slice fragment; + while (true) { + 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) { + // Handle bug in earlier versions of log::Writer where + // it could emit an empty kFirstType record at the tail end + // of a block followed by a kFullType or kFirstType record + // at the beginning of the next block. + if (!scratch->empty()) { + ReportCorruption(scratch->size(), "partial record without end(1)"); + } + } + prospective_record_offset = physical_record_offset; + scratch->clear(); + *record = fragment; + last_record_offset_ = prospective_record_offset; + return true; + + case kFirstType: + if (in_fragmented_record) { + // Handle bug in earlier versions of log::Writer where + // it could emit an empty kFirstType record at the tail end + // of a block followed by a kFullType or kFirstType record + // at the beginning of the next block. + if (!scratch->empty()) { + ReportCorruption(scratch->size(), "partial record without end(2)"); + } + } + prospective_record_offset = physical_record_offset; + scratch->assign(fragment.data(), fragment.size()); + in_fragmented_record = true; + break; + + case kMiddleType: + if (!in_fragmented_record) { + ReportCorruption(fragment.size(), + "missing start of fragmented record(1)"); + } else { + scratch->append(fragment.data(), fragment.size()); + } + break; + + case kLastType: + if (!in_fragmented_record) { + ReportCorruption(fragment.size(), + "missing start of fragmented record(2)"); + } else { + scratch->append(fragment.data(), fragment.size()); + *record = Slice(*scratch); + last_record_offset_ = prospective_record_offset; + return true; + } + break; + + case kEof: + if (in_fragmented_record) { + // This can be caused by the writer dying immediately after + // writing a physical record but before completing the next; don't + // treat it as a corruption, just ignore the entire logical record. + scratch->clear(); + } + return false; + + case kBadRecord: + if (in_fragmented_record) { + ReportCorruption(scratch->size(), "error in middle of record"); + in_fragmented_record = false; + scratch->clear(); + } + break; + + default: { + char buf[40]; + std::snprintf(buf, sizeof(buf), "unknown record type %u", record_type); + ReportCorruption( + (fragment.size() + (in_fragmented_record ? scratch->size() : 0)), + buf); + in_fragmented_record = false; + scratch->clear(); + break; + } + } + } + return false; +} + +uint64_t Reader::LastRecordOffset() { return last_record_offset_; } + +void Reader::ReportCorruption(uint64_t bytes, const char* reason) { + ReportDrop(bytes, Status::Corruption(reason)); +} + +void Reader::ReportDrop(uint64_t bytes, const Status& reason) { + if (reporter_ != nullptr && + end_of_buffer_offset_ - buffer_.size() - bytes >= initial_offset_) { + reporter_->Corruption(static_cast<size_t>(bytes), reason); + } +} + +unsigned int Reader::ReadPhysicalRecord(Slice* result) { + while (true) { + if (buffer_.size() < kHeaderSize) { + if (!eof_) { + // Last read was a full read, so this is a trailer to skip + buffer_.clear(); + Status status = file_->Read(kBlockSize, &buffer_, backing_store_); + end_of_buffer_offset_ += buffer_.size(); + if (!status.ok()) { + buffer_.clear(); + ReportDrop(kBlockSize, status); + eof_ = true; + return kEof; + } else if (buffer_.size() < kBlockSize) { + eof_ = true; + } + continue; + } else { + // Note that if buffer_ is non-empty, we have a truncated header at the + // end of the file, which can be caused by the writer crashing in the + // middle of writing the header. Instead of considering this an error, + // just report EOF. + buffer_.clear(); + return kEof; + } + } + + // Parse the header + const char* header = buffer_.data(); + const uint32_t a = static_cast<uint32_t>(header[4]) & 0xff; + const uint32_t b = static_cast<uint32_t>(header[5]) & 0xff; + const unsigned int type = header[6]; + const uint32_t length = a | (b << 8); + if (kHeaderSize + length > buffer_.size()) { + size_t drop_size = buffer_.size(); + buffer_.clear(); + if (!eof_) { + ReportCorruption(drop_size, "bad record length"); + return kBadRecord; + } + // If the end of the file has been reached without reading |length| bytes + // of payload, assume the writer died in the middle of writing the record. + // Don't report a corruption. + return kEof; + } + + if (type == kZeroType && length == 0) { + // Skip zero length record without reporting any drops since + // such records are produced by the mmap based writing code in + // env_posix.cc that preallocates file regions. + buffer_.clear(); + return kBadRecord; + } + + // Check crc + if (checksum_) { + uint32_t expected_crc = crc32c::Unmask(DecodeFixed32(header)); + uint32_t actual_crc = crc32c::Value(header + 6, 1 + length); + if (actual_crc != expected_crc) { + // Drop the rest of the buffer since "length" itself may have + // been corrupted and if we trust it, we could find some + // fragment of a real log record that just happens to look + // like a valid log record. + size_t drop_size = buffer_.size(); + buffer_.clear(); + ReportCorruption(drop_size, "checksum mismatch"); + return kBadRecord; + } + } + + buffer_.remove_prefix(kHeaderSize + length); + + // Skip physical record that started before initial_offset_ + if (end_of_buffer_offset_ - buffer_.size() - kHeaderSize - length < + initial_offset_) { + result->clear(); + return kBadRecord; + } + + *result = Slice(header + kHeaderSize, length); + return type; + } +} + +} // namespace log +} // namespace leveldb diff --git a/db/log_reader.h b/db/log_reader.h new file mode 100644 index 0000000..75d53f7 --- /dev/null +++ b/db/log_reader.h @@ -0,0 +1,112 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_LOG_READER_H_ +#define STORAGE_LEVELDB_DB_LOG_READER_H_ + +#include <cstdint> + +#include "db/log_format.h" +#include "leveldb/slice.h" +#include "leveldb/status.h" + +namespace leveldb { + +class SequentialFile; + +namespace log { + +class Reader { + public: + // Interface for reporting errors. + class Reporter { + public: + virtual ~Reporter(); + + // Some corruption was detected. "size" is the approximate number + // of bytes dropped due to the corruption. + virtual void Corruption(size_t bytes, const Status& status) = 0; + }; + + // Create a reader that will return log records from "*file". + // "*file" must remain live while this Reader is in use. + // + // If "reporter" is non-null, it is notified whenever some data is + // dropped due to a detected corruption. "*reporter" must remain + // live while this Reader is in use. + // + // If "checksum" is true, verify checksums if available. + // + // The Reader will start reading at the first record located at physical + // position >= initial_offset within the file. + Reader(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 + // successfully, false if we hit end of the input. May use + // "*scratch" as temporary storage. The contents filled in *record + // will only be valid until the next mutating operation on this + // reader or the next mutation to *scratch. + bool ReadRecord(Slice* record, std::string* scratch); + + // Returns the physical offset of the last record returned by ReadRecord. + // + // Undefined before the first call to ReadRecord. + uint64_t LastRecordOffset(); + + private: + // Extend record types with the following special values + enum { + kEof = kMaxRecordType + 1, + // Returned whenever we find an invalid physical record. + // Currently there are three situations in which this happens: + // * The record has an invalid CRC (ReadPhysicalRecord reports a drop) + // * The record is a 0-length record (No drop is reported) + // * The record is below constructor's initial_offset (No drop is reported) + kBadRecord = kMaxRecordType + 2 + }; + + // Skips all blocks that are completely before "initial_offset_". + // + // Returns true on success. Handles reporting. + bool SkipToInitialBlock(); + + // Return type, or one of the preceding special values + unsigned int ReadPhysicalRecord(Slice* result); + + // Reports dropped bytes to the reporter. + // buffer_ must be updated to remove the dropped bytes prior to invocation. + void ReportCorruption(uint64_t bytes, const char* reason); + void ReportDrop(uint64_t bytes, const Status& reason); + + 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 +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_LOG_READER_H_ diff --git a/db/log_test.cc b/db/log_test.cc new file mode 100644 index 0000000..346b19c --- /dev/null +++ b/db/log_test.cc @@ -0,0 +1,563 @@ +// 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 "gtest/gtest.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "leveldb/env.h" +#include "util/coding.h" +#include "util/crc32c.h" +#include "util/random.h" + +namespace leveldb { +namespace log { + +// Construct a string of the specified length made out of the supplied +// partial string. +static std::string BigString(const std::string& partial_string, size_t n) { + std::string result; + while (result.size() < n) { + result.append(partial_string); + } + result.resize(n); + return result; +} + +// Construct a string from a number +static std::string NumberString(int n) { + char buf[50]; + std::snprintf(buf, sizeof(buf), "%d.", n); + return std::string(buf); +} + +// Return a skewed potentially long string +static std::string RandomSkewedString(int i, Random* rnd) { + return BigString(NumberString(i), rnd->Skewed(17)); +} + +class LogTest : public testing::Test { + public: + 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)); + } + + size_t WrittenBytes() const { return dest_.contents_.size(); } + + std::string Read() { + if (!reading_) { + reading_ = true; + source_.contents_ = Slice(dest_.contents_); + } + std::string scratch; + Slice record; + if (reader_->ReadRecord(&record, &scratch)) { + return record.ToString(); + } else { + return "EOF"; + } + } + + void IncrementByte(int offset, int delta) { + dest_.contents_[offset] += delta; + } + + void SetByte(int offset, char new_byte) { + dest_.contents_[offset] = new_byte; + } + + void ShrinkSize(int bytes) { + dest_.contents_.resize(dest_.contents_.size() - bytes); + } + + 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); + crc = crc32c::Mask(crc); + EncodeFixed32(&dest_.contents_[header_offset], crc); + } + + void ForceError() { source_.force_error_ = true; } + + size_t DroppedBytes() const { return report_.dropped_bytes_; } + + std::string ReportMessage() const { return report_.message_; } + + // Returns OK iff recorded error message contains "msg" + std::string MatchError(const std::string& msg) const { + if (report_.message_.find(msg) == std::string::npos) { + return report_.message_; + } else { + return "OK"; + } + } + + void WriteInitialOffsetLog() { + for (int i = 0; i < num_initial_offset_records_; i++) { + std::string record(initial_offset_record_sizes_[i], + static_cast<char>('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*/, + WrittenBytes() + offset_past_end); + Slice record; + std::string scratch; + ASSERT_TRUE(!offset_reader->ReadRecord(&record, &scratch)); + delete offset_reader; + } + + void CheckInitialOffsetRecord(uint64_t initial_offset, + int expected_record_offset) { + WriteInitialOffsetLog(); + reading_ = true; + source_.contents_ = Slice(dest_.contents_); + 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 { + EXPECT_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, + 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, + 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_F(LogTest, Empty) { ASSERT_EQ("EOF", Read()); } + +TEST_F(LogTest, ReadWrite) { + Write("foo"); + Write("bar"); + Write(""); + Write("xxxx"); + ASSERT_EQ("foo", Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("", Read()); + ASSERT_EQ("xxxx", Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ("EOF", Read()); // Make sure reads at eof work +} + +TEST_F(LogTest, ManyBlocks) { + for (int i = 0; i < 100000; i++) { + Write(NumberString(i)); + } + for (int i = 0; i < 100000; i++) { + ASSERT_EQ(NumberString(i), Read()); + } + ASSERT_EQ("EOF", Read()); +} + +TEST_F(LogTest, Fragmentation) { + Write("small"); + Write(BigString("medium", 50000)); + Write(BigString("large", 100000)); + ASSERT_EQ("small", Read()); + ASSERT_EQ(BigString("medium", 50000), Read()); + ASSERT_EQ(BigString("large", 100000), Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST_F(LogTest, MarginalTrailer) { + // Make a trailer that is exactly the same length as an empty record. + const int n = kBlockSize - 2 * kHeaderSize; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize, WrittenBytes()); + Write(""); + Write("bar"); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("", Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST_F(LogTest, MarginalTrailer2) { + // Make a trailer that is exactly the same length as an empty record. + const int n = kBlockSize - 2 * kHeaderSize; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize, WrittenBytes()); + Write("bar"); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("", ReportMessage()); +} + +TEST_F(LogTest, ShortTrailer) { + const int n = kBlockSize - 2 * kHeaderSize + 4; + Write(BigString("foo", n)); + ASSERT_EQ(kBlockSize - kHeaderSize + 4, WrittenBytes()); + Write(""); + Write("bar"); + ASSERT_EQ(BigString("foo", n), Read()); + ASSERT_EQ("", Read()); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST_F(LogTest, AlignedEof) { + 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_F(LogTest, OpenForAppend) { + Write("hello"); + ReopenForAppend(); + Write("world"); + ASSERT_EQ("hello", Read()); + ASSERT_EQ("world", Read()); + ASSERT_EQ("EOF", Read()); +} + +TEST_F(LogTest, RandomRead) { + const int N = 500; + Random write_rnd(301); + for (int i = 0; i < N; i++) { + Write(RandomSkewedString(i, &write_rnd)); + } + Random read_rnd(301); + for (int i = 0; i < N; i++) { + ASSERT_EQ(RandomSkewedString(i, &read_rnd), Read()); + } + ASSERT_EQ("EOF", Read()); +} + +// Tests of all the error paths in log_reader.cc follow: + +TEST_F(LogTest, ReadError) { + Write("foo"); + ForceError(); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(kBlockSize, DroppedBytes()); + ASSERT_EQ("OK", MatchError("read error")); +} + +TEST_F(LogTest, BadRecordType) { + Write("foo"); + // Type is stored in header[6] + IncrementByte(6, 100); + FixChecksum(0, 3); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("unknown record type")); +} + +TEST_F(LogTest, TruncatedTrailingRecordIsIgnored) { + Write("foo"); + ShrinkSize(4); // Drop all payload as well as a header byte + ASSERT_EQ("EOF", Read()); + // Truncated last record is ignored, not treated as an error. + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("", ReportMessage()); +} + +TEST_F(LogTest, BadLength) { + const int kPayloadSize = kBlockSize - kHeaderSize; + Write(BigString("bar", kPayloadSize)); + Write("foo"); + // Least significant size byte is stored in header[4]. + IncrementByte(4, 1); + ASSERT_EQ("foo", Read()); + ASSERT_EQ(kBlockSize, DroppedBytes()); + ASSERT_EQ("OK", MatchError("bad record length")); +} + +TEST_F(LogTest, BadLengthAtEndIsIgnored) { + Write("foo"); + ShrinkSize(1); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(0, DroppedBytes()); + ASSERT_EQ("", ReportMessage()); +} + +TEST_F(LogTest, ChecksumMismatch) { + Write("foo"); + IncrementByte(0, 10); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(10, DroppedBytes()); + ASSERT_EQ("OK", MatchError("checksum mismatch")); +} + +TEST_F(LogTest, UnexpectedMiddleType) { + Write("foo"); + SetByte(6, kMiddleType); + FixChecksum(0, 3); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("missing start")); +} + +TEST_F(LogTest, UnexpectedLastType) { + Write("foo"); + SetByte(6, kLastType); + FixChecksum(0, 3); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("missing start")); +} + +TEST_F(LogTest, UnexpectedFullType) { + Write("foo"); + Write("bar"); + SetByte(6, kFirstType); + FixChecksum(0, 3); + ASSERT_EQ("bar", Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("partial record without end")); +} + +TEST_F(LogTest, UnexpectedFirstType) { + Write("foo"); + Write(BigString("bar", 100000)); + SetByte(6, kFirstType); + FixChecksum(0, 3); + ASSERT_EQ(BigString("bar", 100000), Read()); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ(3, DroppedBytes()); + ASSERT_EQ("OK", MatchError("partial record without end")); +} + +TEST_F(LogTest, MissingLastIsIgnored) { + Write(BigString("bar", kBlockSize)); + // Remove the LAST block, including header. + ShrinkSize(14); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ("", ReportMessage()); + ASSERT_EQ(0, DroppedBytes()); +} + +TEST_F(LogTest, PartialLastIsIgnored) { + Write(BigString("bar", kBlockSize)); + // Cause a bad record length in the LAST block. + ShrinkSize(1); + ASSERT_EQ("EOF", Read()); + ASSERT_EQ("", ReportMessage()); + ASSERT_EQ(0, DroppedBytes()); +} + +TEST_F(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_F(LogTest, ErrorJoinsRecords) { + // Consider two fragmented records: + // first(R1) last(R1) first(R2) last(R2) + // where the middle two fragments disappear. We do not want + // first(R1),last(R2) to get joined and returned as a valid record. + + // Write records that span two blocks + Write(BigString("foo", kBlockSize)); + Write(BigString("bar", kBlockSize)); + Write("correct"); + + // Wipe the middle block + for (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); +} + +TEST_F(LogTest, ReadStart) { CheckInitialOffsetRecord(0, 0); } + +TEST_F(LogTest, ReadSecondOneOff) { CheckInitialOffsetRecord(1, 1); } + +TEST_F(LogTest, ReadSecondTenThousand) { CheckInitialOffsetRecord(10000, 1); } + +TEST_F(LogTest, ReadSecondStart) { CheckInitialOffsetRecord(10007, 1); } + +TEST_F(LogTest, ReadThirdOneOff) { CheckInitialOffsetRecord(10008, 2); } + +TEST_F(LogTest, ReadThirdStart) { CheckInitialOffsetRecord(20014, 2); } + +TEST_F(LogTest, ReadFourthOneOff) { CheckInitialOffsetRecord(20015, 3); } + +TEST_F(LogTest, ReadFourthFirstBlockTrailer) { + CheckInitialOffsetRecord(log::kBlockSize - 4, 3); +} + +TEST_F(LogTest, ReadFourthMiddleBlock) { + CheckInitialOffsetRecord(log::kBlockSize + 1, 3); +} + +TEST_F(LogTest, ReadFourthLastBlock) { + CheckInitialOffsetRecord(2 * log::kBlockSize + 1, 3); +} + +TEST_F(LogTest, ReadFourthStart) { + CheckInitialOffsetRecord( + 2 * (kHeaderSize + 1000) + (2 * log::kBlockSize - 1000) + 3 * kHeaderSize, + 3); +} + +TEST_F(LogTest, ReadInitialOffsetIntoBlockPadding) { + CheckInitialOffsetRecord(3 * log::kBlockSize - 3, 5); +} + +TEST_F(LogTest, ReadEnd) { CheckOffsetPastEndReturnsNoRecords(0); } + +TEST_F(LogTest, ReadPastEnd) { CheckOffsetPastEndReturnsNoRecords(5); } + +} // namespace log +} // namespace leveldb + +int main(int argc, char** argv) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/db/log_writer.cc b/db/log_writer.cc new file mode 100644 index 0000000..ad66bfb --- /dev/null +++ b/db/log_writer.cc @@ -0,0 +1,111 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/log_writer.h" + +#include <cstdint> + +#include "leveldb/env.h" +#include "util/coding.h" +#include "util/crc32c.h" + +namespace leveldb { +namespace log { + +static void InitTypeCrc(uint32_t* type_crc) { + for (int i = 0; i <= kMaxRecordType; i++) { + char t = static_cast<char>(i); + type_crc[i] = crc32c::Value(&t, 1); + } +} + +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(); + + // Fragment the record if necessary and emit it. Note that if slice + // is empty, we still want to iterate once to emit a single + // zero-length record + Status s; + bool begin = true; + do { + const int leftover = kBlockSize - block_offset_; + assert(leftover >= 0); + if (leftover < kHeaderSize) { + // Switch to a new block + if (leftover > 0) { + // Fill the trailer (literal below relies on kHeaderSize being 7) + static_assert(kHeaderSize == 7, ""); + dest_->Append(Slice("\x00\x00\x00\x00\x00\x00", leftover)); + } + block_offset_ = 0; + } + + // Invariant: we never leave < kHeaderSize bytes in a block. + assert(kBlockSize - block_offset_ - kHeaderSize >= 0); + + const size_t avail = kBlockSize - block_offset_ - kHeaderSize; + const size_t fragment_length = (left < avail) ? left : avail; + + RecordType type; + const bool end = (left == fragment_length); + if (begin && end) { + type = kFullType; + } else if (begin) { + type = kFirstType; + } else if (end) { + type = kLastType; + } else { + type = kMiddleType; + } + + s = EmitPhysicalRecord(type, ptr, fragment_length); + ptr += fragment_length; + left -= fragment_length; + begin = false; + } while (s.ok() && left > 0); + return s; +} + +Status Writer::EmitPhysicalRecord(RecordType t, const char* ptr, + size_t 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<char>(length & 0xff); + buf[5] = static_cast<char>(length >> 8); + buf[6] = static_cast<char>(t); + + // Compute the crc of the record type and the payload. + 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, length)); + if (s.ok()) { + s = dest_->Flush(); + } + } + block_offset_ += kHeaderSize + length; + return s; +} + +} // namespace log +} // namespace leveldb diff --git a/db/log_writer.h b/db/log_writer.h new file mode 100644 index 0000000..ad36794 --- /dev/null +++ b/db/log_writer.h @@ -0,0 +1,54 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_LOG_WRITER_H_ +#define STORAGE_LEVELDB_DB_LOG_WRITER_H_ + +#include <cstdint> + +#include "db/log_format.h" +#include "leveldb/slice.h" +#include "leveldb/status.h" + +namespace leveldb { + +class WritableFile; + +namespace log { + +class Writer { + public: + // Create a writer that will append data to "*dest". + // "*dest" must be initially empty. + // "*dest" must remain live while this Writer is in use. + explicit Writer(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 + + // 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]; +}; + +} // namespace log +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_LOG_WRITER_H_ diff --git a/db/memtable.cc b/db/memtable.cc new file mode 100644 index 0000000..f42774d --- /dev/null +++ b/db/memtable.cc @@ -0,0 +1,137 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/memtable.h" +#include "db/dbformat.h" +#include "leveldb/comparator.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" +#include "util/coding.h" + +namespace leveldb { + +static Slice GetLengthPrefixedSlice(const char* data) { + uint32_t len; + const char* p = data; + p = GetVarint32Ptr(p, p + 5, &len); // +5: we assume "p" is not corrupted + return Slice(p, len); +} + +MemTable::MemTable(const InternalKeyComparator& comparator) + : comparator_(comparator), refs_(0), table_(comparator_, &arena_) {} + +MemTable::~MemTable() { assert(refs_ == 0); } + +size_t MemTable::ApproximateMemoryUsage() { return arena_.MemoryUsage(); } + +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); + return comparator.Compare(a, b); +} + +// Encode a suitable internal key target for "target" and return it. +// Uses *scratch as scratch space, and the returned pointer will point +// into this scratch space. +static const char* EncodeKey(std::string* scratch, const Slice& target) { + scratch->clear(); + PutVarint32(scratch, target.size()); + scratch->append(target.data(), target.size()); + return scratch->data(); +} + +class MemTableIterator : public Iterator { + public: + explicit MemTableIterator(MemTable::Table* table) : iter_(table) {} + + 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()); + } + + Status status() const override { return Status::OK(); } + + private: + MemTable::Table::Iterator iter_; + std::string tmp_; // For passing to EncodeKey +}; + +Iterator* MemTable::NewIterator() { return new MemTableIterator(&table_); } + +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() + // key bytes : char[internal_key.size()] + // value_size : varint32 of value.size() + // value bytes : char[value.size()] + 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; + char* buf = arena_.Allocate(encoded_len); + char* p = EncodeVarint32(buf, internal_key_size); + std::memcpy(p, key.data(), key_size); + p += key_size; + EncodeFixed64(p, (s << 8) | type); + p += 8; + p = EncodeVarint32(p, val_size); + std::memcpy(p, value.data(), val_size); + assert(p + val_size == buf + encoded_len); + table_.Insert(buf); +} + +bool MemTable::Get(const LookupKey& key, std::string* value, Status* s) { + Slice memkey = key.memtable_key(); + Table::Iterator iter(&table_); + iter.Seek(memkey.data()); + if (iter.Valid()) { + // entry format is: + // klength varint32 + // userkey char[klength] + // tag uint64 + // vlength varint32 + // value char[vlength] + // Check that it belongs to same user key. We do not check the + // sequence number since the Seek() call above should have skipped + // all entries with overly large sequence numbers. + const char* entry = iter.key(); + uint32_t key_length; + 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) { + // Correct user key + const uint64_t tag = DecodeFixed64(key_ptr + key_length - 8); + switch (static_cast<ValueType>(tag & 0xff)) { + case kTypeValue: { + Slice v = GetLengthPrefixedSlice(key_ptr + key_length); + value->assign(v.data(), v.size()); + return true; + } + case kTypeDeletion: + *s = Status::NotFound(Slice()); + return true; + } + } + } + return false; +} + +} // namespace leveldb diff --git a/db/memtable.h b/db/memtable.h new file mode 100644 index 0000000..9d986b1 --- /dev/null +++ b/db/memtable.h @@ -0,0 +1,87 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_MEMTABLE_H_ +#define STORAGE_LEVELDB_DB_MEMTABLE_H_ + +#include <string> + +#include "db/dbformat.h" +#include "db/skiplist.h" +#include "leveldb/db.h" +#include "util/arena.h" + +namespace leveldb { + +class InternalKeyComparator; +class MemTableIterator; + +class MemTable { + public: + // MemTables are reference counted. The initial reference count + // 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_; } + + // Drop reference count. Delete if no more references exist. + void Unref() { + --refs_; + assert(refs_ >= 0); + if (refs_ <= 0) { + delete this; + } + } + + // Returns an estimate of the number of bytes of data in use by this + // 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. + // + // The caller must ensure that the underlying MemTable remains live + // while the returned iterator is live. The keys returned by this + // iterator are internal keys encoded by AppendInternalKey in the + // db/format.{h,cc} module. + Iterator* NewIterator(); + + // 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, + const Slice& value); + + // If memtable contains a value for key, store it in *value and return true. + // If memtable contains a deletion for key, store a NotFound() error + // in *status and return true. + // Else, return false. + bool Get(const LookupKey& key, std::string* value, Status* s); + + private: + 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<const char*, KeyComparator> Table; + + ~MemTable(); // Private since only Unref() should be used to delete it + + KeyComparator comparator_; + int refs_; + Arena arena_; + Table table_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_MEMTABLE_H_ diff --git a/db/recovery_test.cc b/db/recovery_test.cc new file mode 100644 index 0000000..097dac4 --- /dev/null +++ b/db/recovery_test.cc @@ -0,0 +1,335 @@ +// 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 "gtest/gtest.h" +#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/testutil.h" + +namespace leveldb { + +class RecoveryTest : public testing::Test { + public: + RecoveryTest() : env_(Env::Default()), db_(nullptr) { + dbname_ = testing::TempDir() + "/recovery_test"; + DestroyDB(dbname_, Options()); + Open(); + } + + ~RecoveryTest() { + Close(); + DestroyDB(dbname_, Options()); + } + + DBImpl* dbfull() const { return reinterpret_cast<DBImpl*>(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_LEVELDB_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; + EXPECT_LEVELDB_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 RemoveLogFiles() { + // Linux allows unlinking open files, but Windows does not. + // Closing the db allows for file deletion. + Close(); + std::vector<uint64_t> logs = GetFiles(kLogFile); + for (size_t i = 0; i < logs.size(); i++) { + EXPECT_LEVELDB_OK(env_->RemoveFile(LogName(logs[i]))) << LogName(logs[i]); + } + return logs.size(); + } + + void RemoveManifestFile() { + ASSERT_LEVELDB_OK(env_->RemoveFile(ManifestFileName())); + } + + uint64_t FirstLogFile() { return GetFiles(kLogFile)[0]; } + + std::vector<uint64_t> GetFiles(FileType t) { + std::vector<std::string> filenames; + EXPECT_LEVELDB_OK(env_->GetChildren(dbname_, &filenames)); + std::vector<uint64_t> 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; + EXPECT_LEVELDB_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_LEVELDB_OK(env_->NewWritableFile(fname, &file)); + log::Writer writer(file); + WriteBatch batch; + batch.Put(key, val); + WriteBatchInternal::SetSequence(&batch, seq); + ASSERT_LEVELDB_OK(writer.AddRecord(WriteBatchInternal::Contents(&batch))); + ASSERT_LEVELDB_OK(file->Flush()); + delete file; + } + + private: + std::string dbname_; + Env* env_; + DB* db_; +}; + +TEST_F(RecoveryTest, ManifestReused) { + if (!CanAppend()) { + std::fprintf(stderr, + "skipping test because env does not support appending\n"); + return; + } + ASSERT_LEVELDB_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_F(RecoveryTest, LargeManifestCompacted) { + if (!CanAppend()) { + std::fprintf(stderr, + "skipping test because env does not support appending\n"); + return; + } + ASSERT_LEVELDB_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_LEVELDB_OK(env()->NewAppendableFile(old_manifest, &file)); + std::string zeroes(3 * 1048576 - static_cast<size_t>(len), 0); + ASSERT_LEVELDB_OK(file->Append(zeroes)); + ASSERT_LEVELDB_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_F(RecoveryTest, NoLogFiles) { + ASSERT_LEVELDB_OK(Put("foo", "bar")); + ASSERT_EQ(1, RemoveLogFiles()); + Open(); + ASSERT_EQ("NOT_FOUND", Get("foo")); + Open(); + ASSERT_EQ("NOT_FOUND", Get("foo")); +} + +TEST_F(RecoveryTest, LogFileReuse) { + if (!CanAppend()) { + std::fprintf(stderr, + "skipping test because env does not support appending\n"); + return; + } + for (int i = 0; i < 2; i++) { + ASSERT_LEVELDB_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_F(RecoveryTest, MultipleMemTables) { + // Make a large log. + const int kNum = 1000; + for (int i = 0; i < kNum; i++) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "%050d", i); + ASSERT_LEVELDB_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]; + std::snprintf(buf, sizeof(buf), "%050d", i); + ASSERT_EQ(buf, Get(buf)); + } +} + +TEST_F(RecoveryTest, MultipleLogFiles) { + ASSERT_LEVELDB_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_F(RecoveryTest, ManifestMissing) { + ASSERT_LEVELDB_OK(Put("foo", "bar")); + Close(); + RemoveManifestFile(); + + Status status = OpenWithStatus(); + ASSERT_TRUE(status.IsCorruption()); +} + +} // namespace leveldb + diff --git a/db/repair.cc b/db/repair.cc new file mode 100644 index 0000000..97a27c6 --- /dev/null +++ b/db/repair.cc @@ -0,0 +1,451 @@ +// 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. +// +// We recover the contents of the descriptor from the other files we find. +// (1) Any log files are first converted to tables +// (2) We scan every table to compute +// (a) smallest/largest for the table +// (b) largest sequence number in the table +// (3) We generate descriptor contents: +// - log number is set to zero +// - next-file-number is set to 1 + largest file number we found +// - last-sequence-number is set to largest sequence# found across +// all tables (see 2c) +// - compaction pointers are cleared +// - every table file is added at level 0 +// +// Possible optimization 1: +// (a) Compute total size and use to pick appropriate max-level M +// (b) Sort tables by largest sequence# in the table +// (c) For each table: if it overlaps earlier table, place in level-0, +// else place in level-M. +// Possible optimization 2: +// Store per-table metadata (smallest, largest, largest-seq#, ...) +// in the table's meta section to speed up ScanTable. + +#include "db/builder.h" +#include "db/db_impl.h" +#include "db/dbformat.h" +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "db/memtable.h" +#include "db/table_cache.h" +#include "db/version_edit.h" +#include "db/write_batch_internal.h" +#include "leveldb/comparator.h" +#include "leveldb/db.h" +#include "leveldb/env.h" + +namespace leveldb { + +namespace { + +class Repairer { + public: + Repairer(const std::string& dbname, const Options& options) + : dbname_(dbname), + env_(options.env), + icmp_(options.comparator), + ipolicy_(options.filter_policy), + options_(SanitizeOptions(dbname, &icmp_, &ipolicy_, options)), + owns_info_log_(options_.info_log != options.info_log), + 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); + } + + ~Repairer() { + delete table_cache_; + if (owns_info_log_) { + delete options_.info_log; + } + if (owns_cache_) { + delete options_.block_cache; + } + } + + Status Run() { + Status status = FindFiles(); + if (status.ok()) { + ConvertLogFilesToTables(); + ExtractMetaData(); + status = WriteDescriptor(); + } + if (status.ok()) { + unsigned long long bytes = 0; + for (size_t i = 0; i < tables_.size(); i++) { + bytes += tables_[i].meta.file_size; + } + Log(options_.info_log, + "**** Repaired leveldb %s; " + "recovered %d files; %llu bytes. " + "Some data may have been lost. " + "****", + dbname_.c_str(), static_cast<int>(tables_.size()), bytes); + } + return status; + } + + private: + struct TableInfo { + FileMetaData meta; + SequenceNumber max_sequence; + }; + + Status FindFiles() { + std::vector<std::string> filenames; + Status status = env_->GetChildren(dbname_, &filenames); + if (!status.ok()) { + return status; + } + if (filenames.empty()) { + return Status::IOError(dbname_, "repair found no files"); + } + + uint64_t number; + FileType type; + for (size_t i = 0; i < filenames.size(); i++) { + if (ParseFileName(filenames[i], &number, &type)) { + if (type == kDescriptorFile) { + manifests_.push_back(filenames[i]); + } else { + if (number + 1 > next_file_number_) { + next_file_number_ = number + 1; + } + if (type == kLogFile) { + logs_.push_back(number); + } else if (type == kTableFile) { + table_numbers_.push_back(number); + } else { + // Ignore other files + } + } + } + } + return status; + } + + void ConvertLogFilesToTables() { + for (size_t i = 0; i < logs_.size(); i++) { + std::string logname = LogFileName(dbname_, logs_[i]); + Status status = ConvertLogToTable(logs_[i]); + if (!status.ok()) { + Log(options_.info_log, "Log #%llu: ignoring conversion error: %s", + (unsigned long long)logs_[i], status.ToString().c_str()); + } + ArchiveFile(logname); + } + } + + Status ConvertLogToTable(uint64_t log) { + struct LogReporter : public log::Reader::Reporter { + Env* env; + Logger* info_log; + uint64_t lognum; + 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<int>(bytes), + s.ToString().c_str()); + } + }; + + // Open the log file + std::string logname = LogFileName(dbname_, log); + SequentialFile* lfile; + Status status = env_->NewSequentialFile(logname, &lfile); + if (!status.ok()) { + return status; + } + + // Create the log reader. + LogReporter reporter; + reporter.env = env_; + reporter.info_log = options_.info_log; + reporter.lognum = log; + // We intentionally make log::Reader do checksumming so that + // corruptions cause entire commits to be skipped instead of + // propagating bad information (like overly large sequence + // numbers). + log::Reader reader(lfile, &reporter, false /*do not checksum*/, + 0 /*initial_offset*/); + + // Read all the records and add to a memtable + std::string scratch; + Slice record; + WriteBatch batch; + MemTable* mem = new MemTable(icmp_); + mem->Ref(); + int counter = 0; + while (reader.ReadRecord(&record, &scratch)) { + if (record.size() < 12) { + reporter.Corruption(record.size(), + Status::Corruption("log record too small")); + continue; + } + WriteBatchInternal::SetContents(&batch, record); + status = WriteBatchInternal::InsertInto(&batch, mem); + if (status.ok()) { + counter += WriteBatchInternal::Count(&batch); + } else { + Log(options_.info_log, "Log #%llu: ignoring %s", + (unsigned long long)log, status.ToString().c_str()); + status = Status::OK(); // Keep going with rest of file + } + } + delete lfile; + + // Do not record a version edit for this conversion to a Table + // since ExtractMetaData() will also generate edits. + FileMetaData meta; + meta.number = next_file_number_++; + Iterator* iter = mem->NewIterator(); + status = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta); + delete iter; + mem->Unref(); + 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, + status.ToString().c_str()); + return status; + } + + void ExtractMetaData() { + for (size_t i = 0; i < table_numbers_.size(); i++) { + ScanTable(table_numbers_[i]); + } + } + + Iterator* NewTableIterator(const FileMetaData& meta) { + // Same as compaction iterators: if paranoid_checks are on, turn + // on checksum verification. + ReadOptions r; + r.verify_checksums = options_.paranoid_checks; + return table_cache_->NewIterator(r, meta.number, meta.file_size); + } + + void ScanTable(uint64_t number) { + TableInfo t; + t.meta.number = number; + std::string fname = TableFileName(dbname_, number); + Status status = env_->GetFileSize(fname, &t.meta.file_size); + if (!status.ok()) { + // Try alternate file name. + fname = SSTTableFileName(dbname_, number); + Status s2 = env_->GetFileSize(fname, &t.meta.file_size); + if (s2.ok()) { + status = Status::OK(); + } + } + if (!status.ok()) { + 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()); + return; + } + + // Extract metadata by scanning through table. + int counter = 0; + Iterator* iter = NewTableIterator(t.meta); + bool empty = true; + ParsedInternalKey parsed; + t.max_sequence = 0; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + 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()); + continue; + } + + counter++; + if (empty) { + empty = false; + t.meta.smallest.DecodeFrom(key); + } + t.meta.largest.DecodeFrom(key); + if (parsed.sequence > t.max_sequence) { + t.max_sequence = parsed.sequence; + } + } + if (!iter->status().ok()) { + status = iter->status(); + } + delete iter; + Log(options_.info_log, "Table #%llu: %d entries %s", + (unsigned long long)t.meta.number, counter, status.ToString().c_str()); + + if (status.ok()) { + tables_.push_back(t); + } else { + RepairTable(fname, t); // RepairTable archives input file. + } + } + + void RepairTable(const std::string& src, TableInfo t) { + // We will copy src contents to a new table and then rename the + // new table over the source. + + // Create builder. + std::string copy = TableFileName(dbname_, next_file_number_++); + WritableFile* file; + Status s = env_->NewWritableFile(copy, &file); + if (!s.ok()) { + return; + } + TableBuilder* builder = new TableBuilder(options_, file); + + // Copy data. + Iterator* iter = NewTableIterator(t.meta); + int counter = 0; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + builder->Add(iter->key(), iter->value()); + counter++; + } + delete iter; + + ArchiveFile(src); + if (counter == 0) { + builder->Abandon(); // Nothing to save + } else { + s = builder->Finish(); + if (s.ok()) { + t.meta.file_size = builder->FileSize(); + } + } + delete builder; + builder = nullptr; + + if (s.ok()) { + s = file->Close(); + } + delete file; + 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); + tables_.push_back(t); + } + } + if (!s.ok()) { + env_->RemoveFile(copy); + } + } + + Status WriteDescriptor() { + std::string tmp = TempFileName(dbname_, 1); + WritableFile* file; + Status status = env_->NewWritableFile(tmp, &file); + if (!status.ok()) { + return status; + } + + SequenceNumber max_sequence = 0; + for (size_t i = 0; i < tables_.size(); i++) { + if (max_sequence < tables_[i].max_sequence) { + max_sequence = tables_[i].max_sequence; + } + } + + edit_.SetComparatorName(icmp_.user_comparator()->Name()); + edit_.SetLogNumber(0); + edit_.SetNextFile(next_file_number_); + edit_.SetLastSequence(max_sequence); + + for (size_t i = 0; i < tables_.size(); i++) { + // TODO(opt): separate out into multiple levels + const TableInfo& t = tables_[i]; + edit_.AddFile(0, t.meta.number, t.meta.file_size, t.meta.smallest, + t.meta.largest); + } + + // std::fprintf(stderr, + // "NewDescriptor:\n%s\n", edit_.DebugString().c_str()); + { + log::Writer log(file); + std::string record; + edit_.EncodeTo(&record); + status = log.AddRecord(record); + } + if (status.ok()) { + status = file->Close(); + } + delete file; + file = nullptr; + + if (!status.ok()) { + env_->RemoveFile(tmp); + } else { + // Discard older manifests + for (size_t i = 0; i < manifests_.size(); i++) { + ArchiveFile(dbname_ + "/" + manifests_[i]); + } + + // Install new manifest + status = env_->RenameFile(tmp, DescriptorFileName(dbname_, 1)); + if (status.ok()) { + status = SetCurrentFile(env_, dbname_, 1); + } else { + env_->RemoveFile(tmp); + } + } + return status; + } + + void ArchiveFile(const std::string& fname) { + // Move into another directory. E.g., for + // dir/foo + // rename to + // dir/lost/foo + const char* slash = strrchr(fname.c_str(), '/'); + std::string new_dir; + if (slash != nullptr) { + new_dir.assign(fname.data(), slash - fname.data()); + } + new_dir.append("/lost"); + env_->CreateDir(new_dir); // Ignore error + std::string new_file = new_dir; + new_file.append("/"); + new_file.append((slash == nullptr) ? fname.c_str() : slash + 1); + Status s = env_->RenameFile(fname, new_file); + Log(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<std::string> manifests_; + std::vector<uint64_t> table_numbers_; + std::vector<uint64_t> logs_; + std::vector<TableInfo> tables_; + uint64_t next_file_number_; +}; +} // namespace + +Status RepairDB(const std::string& dbname, const Options& options) { + Repairer repairer(dbname, options); + return repairer.Run(); +} + +} // namespace leveldb diff --git a/db/skiplist.h b/db/skiplist.h new file mode 100644 index 0000000..a59b45b --- /dev/null +++ b/db/skiplist.h @@ -0,0 +1,382 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_SKIPLIST_H_ +#define STORAGE_LEVELDB_DB_SKIPLIST_H_ + +// Thread safety +// ------------- +// +// Writes require external synchronization, most likely a mutex. +// Reads require a guarantee that the SkipList will not be destroyed +// while the read is in progress. Apart from that, reads progress +// without any internal locking or synchronization. +// +// Invariants: +// +// (1) Allocated nodes are never deleted until the SkipList is +// destroyed. This is trivially guaranteed by the code since we +// never delete any skip list nodes. +// +// (2) The contents of a Node except for the next/prev pointers are +// immutable after the Node has been linked into the SkipList. +// Only Insert() modifies the list, and it is careful to initialize +// a node and use release-stores to publish the nodes in one or +// more lists. +// +// ... prev vs. next pointer ordering ... + +#include <atomic> +#include <cassert> +#include <cstdlib> + +#include "util/arena.h" +#include "util/random.h" + +namespace leveldb { + +class Arena; + +template <typename Key, class Comparator> +class SkipList { + private: + struct Node; + + public: + // Create a new SkipList object that will use "cmp" for comparing keys, + // and will allocate memory using "*arena". Objects allocated in the arena + // 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); + + // Returns true iff an entry that compares equal to key is in the list. + bool Contains(const Key& key) const; + + // Iteration over the contents of a skip list + class Iterator { + public: + // Initialize an iterator over the specified list. + // The returned iterator is not valid. + explicit Iterator(const SkipList* list); + + // Returns true iff the iterator is positioned at a valid node. + bool Valid() const; + + // Returns the key at the current position. + // REQUIRES: Valid() + const Key& key() const; + + // Advances to the next position. + // REQUIRES: Valid() + void Next(); + + // Advances to the previous position. + // REQUIRES: Valid() + void Prev(); + + // Advance to the first entry with a key >= target + void Seek(const Key& target); + + // Position at the first entry in list. + // Final state of iterator is Valid() iff list is not empty. + void SeekToFirst(); + + // Position at the last entry in list. + // Final state of iterator is Valid() iff list is not empty. + void SeekToLast(); + + private: + const SkipList* list_; + Node* node_; + // Intentionally copyable + }; + + private: + enum { kMaxHeight = 12 }; + + inline int GetMaxHeight() const { + return max_height_.load(std::memory_order_relaxed); + } + + Node* NewNode(const Key& key, int height); + int RandomHeight(); + bool Equal(const Key& a, const Key& b) const { return (compare_(a, b) == 0); } + + // Return true if key is greater than the data stored in "n" + bool KeyIsAfterNode(const Key& key, Node* n) const; + + // Return the earliest node that comes at or after key. + // Return nullptr if there is no such node. + // + // 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; + + // Return the latest node with a key < key. + // Return head_ if there is no such node. + Node* FindLessThan(const Key& key) const; + + // Return the last node in the list. + // Return head_ if list is empty. + Node* FindLast() const; + + // 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<int> max_height_; // Height of the entire list + + // Read/written only by Insert(). + Random rnd_; +}; + +// Implementation details follow +template <typename Key, class Comparator> +struct SkipList<Key, Comparator>::Node { + explicit Node(const Key& k) : key(k) {} + + Key const key; + + // Accessors/mutators for links. Wrapped in methods so we can + // add the appropriate barriers as necessary. + Node* Next(int n) { + assert(n >= 0); + // Use an 'acquire load' so that we observe a fully initialized + // version of the returned Node. + return next_[n].load(std::memory_order_acquire); + } + void SetNext(int n, Node* x) { + assert(n >= 0); + // Use a 'release store' so that anybody who reads through this + // pointer observes a fully initialized version of the inserted node. + next_[n].store(x, std::memory_order_release); + } + + // No-barrier variants that can be safely used in a few locations. + Node* NoBarrier_Next(int n) { + assert(n >= 0); + return next_[n].load(std::memory_order_relaxed); + } + void NoBarrier_SetNext(int n, Node* x) { + assert(n >= 0); + next_[n].store(x, std::memory_order_relaxed); + } + + private: + // Array of length equal to the node height. next_[0] is lowest level link. + std::atomic<Node*> next_[1]; +}; + +template <typename Key, class Comparator> +typename SkipList<Key, Comparator>::Node* SkipList<Key, Comparator>::NewNode( + const Key& key, int height) { + char* const node_memory = arena_->AllocateAligned( + sizeof(Node) + sizeof(std::atomic<Node*>) * (height - 1)); + return new (node_memory) Node(key); +} + +template <typename Key, class Comparator> +inline SkipList<Key, Comparator>::Iterator::Iterator(const SkipList* list) { + list_ = list; + node_ = nullptr; +} + +template <typename Key, class Comparator> +inline bool SkipList<Key, Comparator>::Iterator::Valid() const { + return node_ != nullptr; +} + +template <typename Key, class Comparator> +inline const Key& SkipList<Key, Comparator>::Iterator::key() const { + assert(Valid()); + return node_->key; +} + +template <typename Key, class Comparator> +inline void SkipList<Key, Comparator>::Iterator::Next() { + assert(Valid()); + node_ = node_->Next(0); +} + +template <typename Key, class Comparator> +inline void SkipList<Key, Comparator>::Iterator::Prev() { + // Instead of using explicit "prev" links, we just search for the + // last node that falls before key. + assert(Valid()); + node_ = list_->FindLessThan(node_->key); + if (node_ == list_->head_) { + node_ = nullptr; + } +} + +template <typename Key, class Comparator> +inline void SkipList<Key, Comparator>::Iterator::Seek(const Key& target) { + node_ = list_->FindGreaterOrEqual(target, nullptr); +} + +template <typename Key, class Comparator> +inline void SkipList<Key, Comparator>::Iterator::SeekToFirst() { + node_ = list_->head_->Next(0); +} + +template <typename Key, class Comparator> +inline void SkipList<Key, Comparator>::Iterator::SeekToLast() { + node_ = list_->FindLast(); + if (node_ == list_->head_) { + node_ = nullptr; + } +} + +template <typename Key, class Comparator> +int SkipList<Key, Comparator>::RandomHeight() { + // Increase height with probability 1 in kBranching + static const unsigned int kBranching = 4; + int height = 1; + while (height < kMaxHeight && ((rnd_.Next() % kBranching) == 0)) { + height++; + } + assert(height > 0); + assert(height <= kMaxHeight); + return height; +} + +template <typename Key, class Comparator> +bool SkipList<Key, Comparator>::KeyIsAfterNode(const Key& key, Node* n) const { + // null n is considered infinite + return (n != nullptr) && (compare_(n->key, key) < 0); +} + +template <typename Key, class Comparator> +typename SkipList<Key, Comparator>::Node* +SkipList<Key, Comparator>::FindGreaterOrEqual(const Key& key, + Node** prev) const { + Node* x = head_; + int level = GetMaxHeight() - 1; + while (true) { + Node* next = x->Next(level); + if (KeyIsAfterNode(key, next)) { + // Keep searching in this list + x = next; + } else { + if (prev != nullptr) prev[level] = x; + if (level == 0) { + return next; + } else { + // Switch to next list + level--; + } + } + } +} + +template <typename Key, class Comparator> +typename SkipList<Key, Comparator>::Node* +SkipList<Key, Comparator>::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 == nullptr || compare_(next->key, key) >= 0) { + if (level == 0) { + return x; + } else { + // Switch to next list + level--; + } + } else { + x = next; + } + } +} + +template <typename Key, class Comparator> +typename SkipList<Key, Comparator>::Node* SkipList<Key, Comparator>::FindLast() + const { + Node* x = head_; + int level = GetMaxHeight() - 1; + while (true) { + Node* next = x->Next(level); + if (next == nullptr) { + if (level == 0) { + return x; + } else { + // Switch to next list + level--; + } + } else { + x = next; + } + } +} + +template <typename Key, class Comparator> +SkipList<Key, Comparator>::SkipList(Comparator cmp, Arena* arena) + : compare_(cmp), + arena_(arena), + head_(NewNode(0 /* any key will do */, kMaxHeight)), + max_height_(1), + rnd_(0xdeadbeef) { + for (int i = 0; i < kMaxHeight; i++) { + head_->SetNext(i, nullptr); + } +} + +template <typename Key, class Comparator> +void SkipList<Key, Comparator>::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 == nullptr || !Equal(key, x->key)); + + int height = RandomHeight(); + if (height > GetMaxHeight()) { + for (int i = GetMaxHeight(); i < height; i++) { + prev[i] = head_; + } + // It is ok to mutate max_height_ without any synchronization + // with concurrent readers. A concurrent reader that observes + // the new value of max_height_ will see either the old value of + // new level pointers from head_ (nullptr), or a new value set in + // the loop below. In the former case the reader will + // immediately drop to the next level since nullptr sorts after all + // keys. In the latter case the reader will use the new node. + max_height_.store(height, std::memory_order_relaxed); + } + + x = NewNode(key, height); + for (int i = 0; i < height; i++) { + // NoBarrier_SetNext() suffices since we will add a barrier when + // we publish a pointer to "x" in prev[i]. + x->NoBarrier_SetNext(i, prev[i]->NoBarrier_Next(i)); + prev[i]->SetNext(i, x); + } +} + +template <typename Key, class Comparator> +bool SkipList<Key, Comparator>::Contains(const Key& key) const { + Node* x = FindGreaterOrEqual(key, nullptr); + if (x != nullptr && Equal(key, x->key)) { + return true; + } else { + return false; + } +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_SKIPLIST_H_ diff --git a/db/skiplist_test.cc b/db/skiplist_test.cc new file mode 100644 index 0000000..360f238 --- /dev/null +++ b/db/skiplist_test.cc @@ -0,0 +1,369 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/skiplist.h" + +#include <atomic> +#include <set> + +#include "gtest/gtest.h" +#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" +#include "util/testutil.h" + +namespace leveldb { + +typedef uint64_t Key; + +struct Comparator { + int operator()(const Key& a, const Key& b) const { + if (a < b) { + return -1; + } else if (a > b) { + return +1; + } else { + return 0; + } + } +}; + +TEST(SkipTest, Empty) { + Arena arena; + Comparator cmp; + SkipList<Key, Comparator> list(cmp, &arena); + ASSERT_TRUE(!list.Contains(10)); + + SkipList<Key, Comparator>::Iterator iter(&list); + ASSERT_TRUE(!iter.Valid()); + iter.SeekToFirst(); + ASSERT_TRUE(!iter.Valid()); + iter.Seek(100); + ASSERT_TRUE(!iter.Valid()); + iter.SeekToLast(); + ASSERT_TRUE(!iter.Valid()); +} + +TEST(SkipTest, InsertAndLookup) { + const int N = 2000; + const int R = 5000; + Random rnd(1000); + std::set<Key> keys; + Arena arena; + Comparator cmp; + SkipList<Key, Comparator> list(cmp, &arena); + for (int i = 0; i < N; i++) { + Key key = rnd.Next() % R; + if (keys.insert(key).second) { + list.Insert(key); + } + } + + for (int i = 0; i < R; i++) { + if (list.Contains(i)) { + ASSERT_EQ(keys.count(i), 1); + } else { + ASSERT_EQ(keys.count(i), 0); + } + } + + // Simple iterator tests + { + SkipList<Key, Comparator>::Iterator iter(&list); + ASSERT_TRUE(!iter.Valid()); + + iter.Seek(0); + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*(keys.begin()), iter.key()); + + iter.SeekToFirst(); + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*(keys.begin()), iter.key()); + + iter.SeekToLast(); + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*(keys.rbegin()), iter.key()); + } + + // Forward iteration test + for (int i = 0; i < R; i++) { + SkipList<Key, Comparator>::Iterator iter(&list); + iter.Seek(i); + + // Compare against model iterator + std::set<Key>::iterator model_iter = keys.lower_bound(i); + for (int j = 0; j < 3; j++) { + if (model_iter == keys.end()) { + ASSERT_TRUE(!iter.Valid()); + break; + } else { + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*model_iter, iter.key()); + ++model_iter; + iter.Next(); + } + } + } + + // Backward iteration test + { + SkipList<Key, Comparator>::Iterator iter(&list); + iter.SeekToLast(); + + // Compare against model iterator + for (std::set<Key>::reverse_iterator model_iter = keys.rbegin(); + model_iter != keys.rend(); ++model_iter) { + ASSERT_TRUE(iter.Valid()); + ASSERT_EQ(*model_iter, iter.key()); + iter.Prev(); + } + ASSERT_TRUE(!iter.Valid()); + } +} + +// We want to make sure that with a single writer and multiple +// concurrent readers (with no synchronization other than when a +// reader's iterator is created), the reader always observes all the +// data that was present in the skip list when the iterator was +// 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. +// +// We generate multi-part keys: +// <key,gen,hash> +// where: +// key is in range [0..K-1] +// gen is a generation number for key +// hash is hash(key,gen) +// +// The insertion code picks a random key, sets gen to be 1 + the last +// generation number inserted for that key, and sets hash to Hash(key,gen). +// +// At the beginning of a read, we snapshot the last inserted +// generation number for each key. We then iterate, including random +// calls to Next() and Seek(). For every key we encounter, we +// check that it is either expected given the initial snapshot or has +// been concurrently added since the iterator started. +class ConcurrentTest { + private: + static constexpr uint32_t K = 4; + + static uint64_t key(Key key) { return (key >> 40); } + static uint64_t gen(Key key) { return (key >> 8) & 0xffffffffu; } + static uint64_t hash(Key key) { return key & 0xff; } + + static uint64_t HashNumbers(uint64_t k, uint64_t g) { + uint64_t data[2] = {k, g}; + return Hash(reinterpret_cast<char*>(data), sizeof(data), 0); + } + + static Key MakeKey(uint64_t k, uint64_t g) { + 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)); + } + + static bool IsValidKey(Key k) { + return hash(k) == (HashNumbers(key(k), gen(k)) & 0xff); + } + + static Key RandomTarget(Random* rnd) { + switch (rnd->Next() % 10) { + case 0: + // Seek to beginning + return MakeKey(0, 0); + case 1: + // Seek to end + return MakeKey(K, 0); + default: + // Seek to middle + return MakeKey(rnd->Next() % K, 0); + } + } + + // Per-key generation + struct State { + std::atomic<int> 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++) { + Set(k, 0); + } + } + }; + + // Current state of the test + State current_; + + Arena arena_; + + // SkipList is not protected by mu_. We just use a single writer + // thread to modify it. + SkipList<Key, Comparator> list_; + + public: + ConcurrentTest() : list_(Comparator(), &arena_) {} + + // REQUIRES: External synchronization + void WriteStep(Random* rnd) { + const uint32_t k = rnd->Next() % K; + const intptr_t g = current_.Get(k) + 1; + const Key key = MakeKey(k, g); + list_.Insert(key); + current_.Set(k, g); + } + + void ReadStep(Random* rnd) { + // Remember the initial committed state of the skiplist. + State initial_state; + for (int k = 0; k < K; k++) { + initial_state.Set(k, current_.Get(k)); + } + + Key pos = RandomTarget(rnd); + SkipList<Key, Comparator>::Iterator iter(&list_); + iter.Seek(pos); + while (true) { + Key current; + if (!iter.Valid()) { + current = MakeKey(K, 0); + } else { + current = iter.key(); + ASSERT_TRUE(IsValidKey(current)) << current; + } + ASSERT_LE(pos, current) << "should not go backwards"; + + // Verify that everything in [pos,current) was not present in + // initial_state. + while (pos < current) { + ASSERT_LT(key(pos), K) << pos; + + // Note that generation 0 is never inserted, so it is ok if + // <*,0,*> is missing. + ASSERT_TRUE((gen(pos) == 0) || + (gen(pos) > static_cast<Key>(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)) { + pos = MakeKey(key(pos) + 1, 0); + } else { + pos = MakeKey(key(pos), gen(pos) + 1); + } + } + + if (!iter.Valid()) { + break; + } + + if (rnd->Next() % 2) { + iter.Next(); + pos = MakeKey(key(pos), gen(pos) + 1); + } else { + Key new_target = RandomTarget(rnd); + if (new_target > pos) { + pos = new_target; + iter.Seek(new_target); + } + } + } + } +}; + +// Needed when building in C++11 mode. +constexpr uint32_t ConcurrentTest::K; + +// Simple test that does single-threaded testing of the ConcurrentTest +// scaffolding. +TEST(SkipTest, ConcurrentWithoutThreads) { + ConcurrentTest test; + Random rnd(test::RandomSeed()); + for (int i = 0; i < 10000; i++) { + test.ReadStep(&rnd); + test.WriteStep(&rnd); + } +} + +class TestState { + public: + ConcurrentTest t_; + int seed_; + std::atomic<bool> quit_flag_; + + enum ReaderState { STARTING, RUNNING, DONE }; + + explicit TestState(int s) + : seed_(s), quit_flag_(false), state_(STARTING), state_cv_(&mu_) {} + + void Wait(ReaderState s) LOCKS_EXCLUDED(mu_) { + mu_.Lock(); + while (state_ != s) { + state_cv_.Wait(); + } + mu_.Unlock(); + } + + void Change(ReaderState s) LOCKS_EXCLUDED(mu_) { + mu_.Lock(); + state_ = s; + state_cv_.Signal(); + mu_.Unlock(); + } + + private: + port::Mutex mu_; + ReaderState state_ GUARDED_BY(mu_); + port::CondVar state_cv_ GUARDED_BY(mu_); +}; + +static void ConcurrentReader(void* arg) { + TestState* state = reinterpret_cast<TestState*>(arg); + Random rnd(state->seed_); + int64_t reads = 0; + state->Change(TestState::RUNNING); + while (!state->quit_flag_.load(std::memory_order_acquire)) { + state->t_.ReadStep(&rnd); + ++reads; + } + state->Change(TestState::DONE); +} + +static void RunConcurrent(int run) { + const int seed = test::RandomSeed() + (run * 100); + Random rnd(seed); + const int N = 1000; + const int kSize = 1000; + for (int i = 0; i < N; i++) { + if ((i % 100) == 0) { + std::fprintf(stderr, "Run %d of %d\n", i, N); + } + TestState state(seed + 1); + Env::Default()->Schedule(ConcurrentReader, &state); + state.Wait(TestState::RUNNING); + for (int i = 0; i < kSize; i++) { + state.t_.WriteStep(&rnd); + } + state.quit_flag_.store(true, std::memory_order_release); + state.Wait(TestState::DONE); + } +} + +TEST(SkipTest, Concurrent1) { RunConcurrent(1); } +TEST(SkipTest, Concurrent2) { RunConcurrent(2); } +TEST(SkipTest, Concurrent3) { RunConcurrent(3); } +TEST(SkipTest, Concurrent4) { RunConcurrent(4); } +TEST(SkipTest, Concurrent5) { RunConcurrent(5); } + +} // namespace leveldb + diff --git a/db/snapshot.h b/db/snapshot.h new file mode 100644 index 0000000..9f1d664 --- /dev/null +++ b/db/snapshot.h @@ -0,0 +1,95 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_SNAPSHOT_H_ +#define STORAGE_LEVELDB_DB_SNAPSHOT_H_ + +#include "db/dbformat.h" +#include "leveldb/db.h" + +namespace leveldb { + +class SnapshotList; + +// Snapshots are kept in a doubly-linked list in the DB. +// Each SnapshotImpl corresponds to a particular sequence number. +class SnapshotImpl : public Snapshot { + public: + 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. The SnapshotList + // implementation operates on the next/previous fields direcly. + SnapshotImpl* prev_; + SnapshotImpl* next_; + + const SequenceNumber sequence_number_; + +#if !defined(NDEBUG) + SnapshotList* list_ = nullptr; +#endif // !defined(NDEBUG) +}; + +class SnapshotList { + public: + SnapshotList() : head_(0) { + head_.prev_ = &head_; + head_.next_ = &head_; + } + + bool empty() const { return head_.next_ == &head_; } + SnapshotImpl* oldest() const { + assert(!empty()); + return head_.next_; + } + SnapshotImpl* newest() const { + assert(!empty()); + return head_.prev_; + } + + // 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 head_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_SNAPSHOT_H_ diff --git a/db/table_cache.cc b/db/table_cache.cc new file mode 100644 index 0000000..73f05fd --- /dev/null +++ b/db/table_cache.cc @@ -0,0 +1,120 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/table_cache.h" + +#include "db/filename.h" +#include "leveldb/env.h" +#include "leveldb/table.h" +#include "util/coding.h" + +namespace leveldb { + +struct TableAndFile { + RandomAccessFile* file; + Table* table; +}; + +static void DeleteEntry(const Slice& key, void* value) { + TableAndFile* tf = reinterpret_cast<TableAndFile*>(value); + delete tf->table; + delete tf->file; + delete tf; +} + +static void UnrefEntry(void* arg1, void* arg2) { + Cache* cache = reinterpret_cast<Cache*>(arg1); + Cache::Handle* h = reinterpret_cast<Cache::Handle*>(arg2); + cache->Release(h); +} + +TableCache::TableCache(const std::string& dbname, const Options& options, + int entries) + : env_(options.env), + dbname_(dbname), + options_(options), + cache_(NewLRUCache(entries)) {} + +TableCache::~TableCache() { delete cache_; } + +Status TableCache::FindTable(uint64_t file_number, uint64_t file_size, + Cache::Handle** handle) { + Status s; + char buf[sizeof(file_number)]; + EncodeFixed64(buf, file_number); + Slice key(buf, sizeof(buf)); + *handle = cache_->Lookup(key); + if (*handle == nullptr) { + std::string fname = TableFileName(dbname_, file_number); + RandomAccessFile* file = nullptr; + Table* table = nullptr; + s = env_->NewRandomAccessFile(fname, &file); + if (!s.ok()) { + std::string old_fname = SSTTableFileName(dbname_, file_number); + if (env_->NewRandomAccessFile(old_fname, &file).ok()) { + s = Status::OK(); + } + } + if (s.ok()) { + s = Table::Open(options_, file, file_size, &table); + } + + if (!s.ok()) { + 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. + } else { + TableAndFile* tf = new TableAndFile; + tf->file = file; + tf->table = table; + *handle = cache_->Insert(key, tf, 1, &DeleteEntry); + } + } + return s; +} + +Iterator* TableCache::NewIterator(const ReadOptions& options, + uint64_t file_number, uint64_t file_size, + Table** tableptr) { + if (tableptr != nullptr) { + *tableptr = nullptr; + } + + Cache::Handle* handle = nullptr; + Status s = FindTable(file_number, file_size, &handle); + if (!s.ok()) { + return NewErrorIterator(s); + } + + Table* table = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table; + Iterator* result = table->NewIterator(options); + result->RegisterCleanup(&UnrefEntry, cache_, handle); + 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 (*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<TableAndFile*>(cache_->Value(handle))->table; + s = t->InternalGet(options, k, arg, handle_result); + cache_->Release(handle); + } + return s; +} + +void TableCache::Evict(uint64_t file_number) { + char buf[sizeof(file_number)]; + EncodeFixed64(buf, file_number); + cache_->Erase(Slice(buf, sizeof(buf))); +} + +} // namespace leveldb diff --git a/db/table_cache.h b/db/table_cache.h new file mode 100644 index 0000000..aac9bfc --- /dev/null +++ b/db/table_cache.h @@ -0,0 +1,57 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// Thread-safe (provides internal synchronization) + +#ifndef STORAGE_LEVELDB_DB_TABLE_CACHE_H_ +#define STORAGE_LEVELDB_DB_TABLE_CACHE_H_ + +#include <cstdint> +#include <string> + +#include "db/dbformat.h" +#include "leveldb/cache.h" +#include "leveldb/table.h" +#include "port/port.h" + +namespace leveldb { + +class Env; + +class TableCache { + public: + 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 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 = 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, + 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_; + Cache* cache_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_TABLE_CACHE_H_ diff --git a/db/version_edit.cc b/db/version_edit.cc new file mode 100644 index 0000000..3e9012f --- /dev/null +++ b/db/version_edit.cc @@ -0,0 +1,257 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/version_edit.h" + +#include "db/version_set.h" +#include "util/coding.h" + +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, + // 8 was used for large value refs + kPrevLogNumber = 9 +}; + +void VersionEdit::Clear() { + comparator_.clear(); + log_number_ = 0; + prev_log_number_ = 0; + last_sequence_ = 0; + next_file_number_ = 0; + has_comparator_ = false; + has_log_number_ = false; + has_prev_log_number_ = false; + has_next_file_number_ = false; + has_last_sequence_ = false; + deleted_files_.clear(); + new_files_.clear(); +} + +void VersionEdit::EncodeTo(std::string* dst) const { + if (has_comparator_) { + PutVarint32(dst, kComparator); + PutLengthPrefixedSlice(dst, comparator_); + } + if (has_log_number_) { + PutVarint32(dst, kLogNumber); + PutVarint64(dst, log_number_); + } + if (has_prev_log_number_) { + PutVarint32(dst, kPrevLogNumber); + PutVarint64(dst, prev_log_number_); + } + if (has_next_file_number_) { + PutVarint32(dst, kNextFileNumber); + PutVarint64(dst, next_file_number_); + } + if (has_last_sequence_) { + PutVarint32(dst, kLastSequence); + PutVarint64(dst, last_sequence_); + } + + for (size_t i = 0; i < compact_pointers_.size(); i++) { + PutVarint32(dst, kCompactPointer); + PutVarint32(dst, compact_pointers_[i].first); // level + PutLengthPrefixedSlice(dst, compact_pointers_[i].second.Encode()); + } + + for (const auto& deleted_file_kvp : deleted_files_) { + PutVarint32(dst, kDeletedFile); + 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++) { + const FileMetaData& f = new_files_[i].second; + PutVarint32(dst, kNewFile); + PutVarint32(dst, new_files_[i].first); // level + PutVarint64(dst, f.number); + PutVarint64(dst, f.file_size); + PutLengthPrefixedSlice(dst, f.smallest.Encode()); + PutLengthPrefixedSlice(dst, f.largest.Encode()); + } +} + +static bool GetInternalKey(Slice* input, InternalKey* dst) { + Slice str; + if (GetLengthPrefixedSlice(input, &str)) { + return dst->DecodeFrom(str); + } else { + return false; + } +} + +static bool GetLevel(Slice* input, int* level) { + uint32_t v; + if (GetVarint32(input, &v) && v < config::kNumLevels) { + *level = v; + return true; + } else { + return false; + } +} + +Status VersionEdit::DecodeFrom(const Slice& src) { + Clear(); + Slice input = src; + const char* msg = nullptr; + uint32_t tag; + + // Temporary storage for parsing + int level; + uint64_t number; + FileMetaData f; + Slice str; + InternalKey key; + + while (msg == nullptr && GetVarint32(&input, &tag)) { + switch (tag) { + case kComparator: + if (GetLengthPrefixedSlice(&input, &str)) { + comparator_ = str.ToString(); + has_comparator_ = true; + } else { + msg = "comparator name"; + } + break; + + case kLogNumber: + if (GetVarint64(&input, &log_number_)) { + has_log_number_ = true; + } else { + msg = "log number"; + } + break; + + case kPrevLogNumber: + if (GetVarint64(&input, &prev_log_number_)) { + has_prev_log_number_ = true; + } else { + msg = "previous log number"; + } + break; + + case kNextFileNumber: + if (GetVarint64(&input, &next_file_number_)) { + has_next_file_number_ = true; + } else { + msg = "next file number"; + } + break; + + case kLastSequence: + if (GetVarint64(&input, &last_sequence_)) { + has_last_sequence_ = true; + } else { + msg = "last sequence number"; + } + break; + + case kCompactPointer: + if (GetLevel(&input, &level) && GetInternalKey(&input, &key)) { + compact_pointers_.push_back(std::make_pair(level, key)); + } else { + msg = "compaction pointer"; + } + break; + + case kDeletedFile: + if (GetLevel(&input, &level) && GetVarint64(&input, &number)) { + deleted_files_.insert(std::make_pair(level, number)); + } else { + msg = "deleted file"; + } + break; + + case kNewFile: + if (GetLevel(&input, &level) && GetVarint64(&input, &f.number) && + GetVarint64(&input, &f.file_size) && + GetInternalKey(&input, &f.smallest) && + GetInternalKey(&input, &f.largest)) { + new_files_.push_back(std::make_pair(level, f)); + } else { + msg = "new-file entry"; + } + break; + + default: + msg = "unknown tag"; + break; + } + } + + if (msg == nullptr && !input.empty()) { + msg = "invalid tag"; + } + + Status result; + if (msg != nullptr) { + result = Status::Corruption("VersionEdit", msg); + } + return result; +} + +std::string VersionEdit::DebugString() const { + std::string r; + r.append("VersionEdit {"); + if (has_comparator_) { + r.append("\n Comparator: "); + r.append(comparator_); + } + if (has_log_number_) { + r.append("\n LogNumber: "); + AppendNumberTo(&r, log_number_); + } + if (has_prev_log_number_) { + r.append("\n PrevLogNumber: "); + AppendNumberTo(&r, prev_log_number_); + } + if (has_next_file_number_) { + r.append("\n NextFile: "); + AppendNumberTo(&r, next_file_number_); + } + if (has_last_sequence_) { + r.append("\n LastSeq: "); + AppendNumberTo(&r, last_sequence_); + } + for (size_t i = 0; i < compact_pointers_.size(); i++) { + r.append("\n CompactPointer: "); + AppendNumberTo(&r, compact_pointers_[i].first); + r.append(" "); + r.append(compact_pointers_[i].second.DebugString()); + } + for (const auto& deleted_files_kvp : deleted_files_) { + r.append("\n RemoveFile: "); + AppendNumberTo(&r, deleted_files_kvp.first); + r.append(" "); + AppendNumberTo(&r, deleted_files_kvp.second); + } + for (size_t i = 0; i < new_files_.size(); i++) { + const FileMetaData& f = new_files_[i].second; + r.append("\n AddFile: "); + AppendNumberTo(&r, new_files_[i].first); + r.append(" "); + AppendNumberTo(&r, f.number); + r.append(" "); + AppendNumberTo(&r, f.file_size); + r.append(" "); + r.append(f.smallest.DebugString()); + r.append(" .. "); + r.append(f.largest.DebugString()); + } + r.append("\n}\n"); + return r; +} + +} // namespace leveldb diff --git a/db/version_edit.h b/db/version_edit.h new file mode 100644 index 0000000..137b4b1 --- /dev/null +++ b/db/version_edit.h @@ -0,0 +1,106 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_VERSION_EDIT_H_ +#define STORAGE_LEVELDB_DB_VERSION_EDIT_H_ + +#include <set> +#include <utility> +#include <vector> + +#include "db/dbformat.h" + +namespace leveldb { + +class VersionSet; + +struct FileMetaData { + 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() = default; + + void Clear(); + + void SetComparatorName(const Slice& name) { + has_comparator_ = true; + comparator_ = name.ToString(); + } + void SetLogNumber(uint64_t num) { + has_log_number_ = true; + log_number_ = num; + } + void SetPrevLogNumber(uint64_t num) { + has_prev_log_number_ = true; + prev_log_number_ = num; + } + void SetNextFile(uint64_t num) { + has_next_file_number_ = true; + next_file_number_ = num; + } + void SetLastSequence(SequenceNumber seq) { + has_last_sequence_ = true; + last_sequence_ = seq; + } + void SetCompactPointer(int level, const InternalKey& key) { + compact_pointers_.push_back(std::make_pair(level, key)); + } + + // 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) { + FileMetaData f; + f.number = file; + f.file_size = file_size; + f.smallest = smallest; + f.largest = largest; + new_files_.push_back(std::make_pair(level, f)); + } + + // Delete the specified "file" from the specified "level". + void RemoveFile(int level, uint64_t file) { + deleted_files_.insert(std::make_pair(level, file)); + } + + void EncodeTo(std::string* dst) const; + Status DecodeFrom(const Slice& src); + + std::string DebugString() const; + + private: + friend class VersionSet; + + typedef std::set<std::pair<int, uint64_t>> DeletedFileSet; + + std::string comparator_; + uint64_t log_number_; + uint64_t prev_log_number_; + uint64_t next_file_number_; + SequenceNumber last_sequence_; + bool has_comparator_; + bool has_log_number_; + bool has_prev_log_number_; + bool has_next_file_number_; + bool has_last_sequence_; + + std::vector<std::pair<int, InternalKey>> compact_pointers_; + DeletedFileSet deleted_files_; + std::vector<std::pair<int, FileMetaData>> new_files_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_VERSION_EDIT_H_ diff --git a/db/version_edit_test.cc b/db/version_edit_test.cc new file mode 100644 index 0000000..acafab0 --- /dev/null +++ b/db/version_edit_test.cc @@ -0,0 +1,46 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/version_edit.h" + +#include "gtest/gtest.h" + +namespace leveldb { + +static void TestEncodeDecode(const VersionEdit& edit) { + std::string encoded, encoded2; + edit.EncodeTo(&encoded); + VersionEdit parsed; + Status s = parsed.DecodeFrom(encoded); + ASSERT_TRUE(s.ok()) << s.ToString(); + parsed.EncodeTo(&encoded2); + ASSERT_EQ(encoded, encoded2); +} + +TEST(VersionEditTest, EncodeDecode) { + static const uint64_t kBig = 1ull << 50; + + VersionEdit edit; + for (int i = 0; i < 4; i++) { + TestEncodeDecode(edit); + edit.AddFile(3, kBig + 300 + i, kBig + 400 + i, + InternalKey("foo", kBig + 500 + i, kTypeValue), + InternalKey("zoo", kBig + 600 + i, kTypeDeletion)); + edit.RemoveFile(4, kBig + 700 + i); + edit.SetCompactPointer(i, InternalKey("x", kBig + 900 + i, kTypeValue)); + } + + edit.SetComparatorName("foo"); + edit.SetLogNumber(kBig + 100); + edit.SetNextFile(kBig + 200); + edit.SetLastSequence(kBig + 1000); + TestEncodeDecode(edit); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/db/version_set.cc b/db/version_set.cc new file mode 100644 index 0000000..1963353 --- /dev/null +++ b/db/version_set.cc @@ -0,0 +1,1568 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/version_set.h" + +#include <algorithm> +#include <cstdio> + +#include "db/filename.h" +#include "db/log_reader.h" +#include "db/log_writer.h" +#include "db/memtable.h" +#include "db/table_cache.h" +#include "leveldb/env.h" +#include "leveldb/table_builder.h" +#include "table/merger.h" +#include "table/two_level_iterator.h" +#include "util/coding.h" +#include "util/logging.h" + +namespace leveldb { + +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 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 int64_t ExpandedCompactionByteSizeLimit(const Options* options) { + return 25 * TargetFileSize(options); +} + +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. + + // Result for both level-0 and level-1 + double result = 10. * 1048576.0; + while (level > 1) { + result *= 10; + level--; + } + return result; +} + +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<FileMetaData*>& files) { + int64_t sum = 0; + for (size_t i = 0; i < files.size(); i++) { + sum += files[i]->file_size; + } + return sum; +} + +Version::~Version() { + assert(refs_ == 0); + + // Remove from linked list + prev_->next_ = next_; + next_->prev_ = prev_; + + // Drop references to files + for (int level = 0; level < config::kNumLevels; level++) { + for (size_t i = 0; i < files_[level].size(); i++) { + FileMetaData* f = files_[level][i]; + assert(f->refs > 0); + f->refs--; + if (f->refs <= 0) { + delete f; + } + } + } +} + +int FindFile(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>& files, const Slice& key) { + uint32_t left = 0; + uint32_t right = files.size(); + while (left < right) { + uint32_t mid = (left + right) / 2; + const FileMetaData* f = files[mid]; + if (icmp.InternalKeyComparator::Compare(f->largest.Encode(), key) < 0) { + // Key at "mid.largest" is < "target". Therefore all + // files at or before "mid" are uninteresting. + left = mid + 1; + } else { + // Key at "mid.largest" is >= "target". Therefore all files + // after "mid" are uninteresting. + right = mid; + } + } + return right; +} + +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 != nullptr && + ucmp->Compare(*user_key, f->smallest.user_key()) < 0); +} + +bool SomeFileOverlapsRange(const InternalKeyComparator& icmp, + bool disjoint_sorted_files, + const std::vector<FileMetaData*>& 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 + for (size_t i = 0; i < files.size(); i++) { + const FileMetaData* f = files[i]; + if (AfterFile(ucmp, smallest_user_key, f) || + BeforeFile(ucmp, largest_user_key, f)) { + // No overlap + } else { + return true; // Overlap + } + } + return false; + } + + // Binary search over file list + uint32_t index = 0; + if (smallest_user_key != nullptr) { + // Find the earliest possible internal key for smallest_user_key + InternalKey small_key(*smallest_user_key, kMaxSequenceNumber, + kValueTypeForSeek); + index = FindFile(icmp, files, small_key.Encode()); + } + + if (index >= files.size()) { + // beginning of range is after all files, so no overlap. + return false; + } + + return !BeforeFile(ucmp, largest_user_key, files[index]); +} + +// An internal iterator. For a given version/level pair, yields +// information about the files in the level. For a given entry, key() +// is the largest key that occurs in the file, and value() is an +// 16-byte value containing the file number and file size, both +// encoded using EncodeFixed64. +class Version::LevelFileNumIterator : public Iterator { + public: + LevelFileNumIterator(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>* flist) + : icmp_(icmp), flist_(flist), index_(flist->size()) { // Marks as invalid + } + bool Valid() const override { return index_ < flist_->size(); } + void Seek(const Slice& target) override { + index_ = FindFile(icmp_, *flist_, target); + } + void SeekToFirst() override { index_ = 0; } + void SeekToLast() override { + index_ = flist_->empty() ? 0 : flist_->size() - 1; + } + void Next() override { + assert(Valid()); + index_++; + } + void Prev() override { + assert(Valid()); + if (index_ == 0) { + index_ = flist_->size(); // Marks as invalid + } else { + index_--; + } + } + Slice key() const override { + assert(Valid()); + return (*flist_)[index_]->largest.Encode(); + } + Slice value() const override { + assert(Valid()); + EncodeFixed64(value_buf_, (*flist_)[index_]->number); + EncodeFixed64(value_buf_ + 8, (*flist_)[index_]->file_size); + return Slice(value_buf_, sizeof(value_buf_)); + } + Status status() const override { return Status::OK(); } + + private: + const InternalKeyComparator icmp_; + const std::vector<FileMetaData*>* const flist_; + uint32_t index_; + + // Backing store for value(). Holds the file number and size. + mutable char value_buf_[16]; +}; + +static Iterator* GetFileIterator(void* arg, const ReadOptions& options, + const Slice& file_value) { + TableCache* cache = reinterpret_cast<TableCache*>(arg); + if (file_value.size() != 16) { + return NewErrorIterator( + Status::Corruption("FileReader invoked with unexpected value")); + } else { + return cache->NewIterator(options, DecodeFixed64(file_value.data()), + DecodeFixed64(file_value.data() + 8)); + } +} + +Iterator* Version::NewConcatenatingIterator(const ReadOptions& options, + int level) const { + return NewTwoLevelIterator( + new LevelFileNumIterator(vset_->icmp_, &files_[level]), &GetFileIterator, + vset_->table_cache_, options); +} + +void Version::AddIterators(const ReadOptions& options, + std::vector<Iterator*>* 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)); + } + + // For levels > 0, we can use a concatenating iterator that sequentially + // walks through the non-overlapping files in the level, opening them + // lazily. + for (int level = 1; level < config::kNumLevels; level++) { + if (!files_[level].empty()) { + iters->push_back(NewConcatenatingIterator(options, level)); + } + } +} + +// Callback from TableCache::Get() +namespace { +enum SaverState { + kNotFound, + kFound, + kDeleted, + kCorrupt, +}; +struct Saver { + SaverState state; + const Comparator* ucmp; + Slice user_key; + std::string* value; +}; +} // namespace +static void SaveValue(void* arg, const Slice& ikey, const Slice& v) { + Saver* s = reinterpret_cast<Saver*>(arg); + ParsedInternalKey parsed_key; + if (!ParseInternalKey(ikey, &parsed_key)) { + s->state = kCorrupt; + } else { + if (s->ucmp->Compare(parsed_key.user_key, s->user_key) == 0) { + s->state = (parsed_key.type == kTypeValue) ? kFound : kDeleted; + if (s->state == kFound) { + s->value->assign(v.data(), v.size()); + } + } + } +} + +static bool NewestFirst(FileMetaData* a, FileMetaData* b) { + return a->number > b->number; +} + +void Version::ForEachOverlapping(Slice user_key, Slice internal_key, void* arg, + bool (*func)(void*, int, FileMetaData*)) { + const Comparator* ucmp = vset_->icmp_.user_comparator(); + + // Search level-0 in order from newest to oldest. + std::vector<FileMetaData*> tmp; + tmp.reserve(files_[0].size()); + for (uint32_t i = 0; i < files_[0].size(); i++) { + FileMetaData* f = files_[0][i]; + if (ucmp->Compare(user_key, f->smallest.user_key()) >= 0 && + ucmp->Compare(user_key, f->largest.user_key()) <= 0) { + tmp.push_back(f); + } + } + if (!tmp.empty()) { + std::sort(tmp.begin(), tmp.end(), NewestFirst); + for (uint32_t i = 0; i < tmp.size(); i++) { + if (!(*func)(arg, 0, tmp[i])) { + return; + } + } + } + + // Search other levels. + for (int level = 1; level < config::kNumLevels; level++) { + size_t num_files = files_[level].size(); + if (num_files == 0) continue; + + // Binary search to find earliest index whose largest key >= internal_key. + uint32_t index = FindFile(vset_->icmp_, files_[level], internal_key); + if (index < num_files) { + FileMetaData* f = files_[level][index]; + if (ucmp->Compare(user_key, f->smallest.user_key()) < 0) { + // All of "f" is past any data for user_key + } else { + if (!(*func)(arg, level, f)) { + return; + } + } + } + } +} + +Status Version::Get(const ReadOptions& options, const LookupKey& k, + std::string* value, GetStats* stats) { + stats->seek_file = nullptr; + stats->seek_file_level = -1; + + struct State { + Saver saver; + GetStats* stats; + const ReadOptions* options; + Slice ikey; + FileMetaData* last_file_read; + int last_file_read_level; + + VersionSet* vset; + Status s; + bool found; + + static bool Match(void* arg, int level, FileMetaData* f) { + State* state = reinterpret_cast<State*>(arg); + + if (state->stats->seek_file == nullptr && + state->last_file_read != nullptr) { + // We have had more than one seek for this read. Charge the 1st file. + state->stats->seek_file = state->last_file_read; + state->stats->seek_file_level = state->last_file_read_level; + } + + state->last_file_read = f; + state->last_file_read_level = level; + + state->s = state->vset->table_cache_->Get(*state->options, f->number, + f->file_size, state->ikey, + &state->saver, SaveValue); + if (!state->s.ok()) { + state->found = true; + return false; + } + switch (state->saver.state) { + case kNotFound: + return true; // Keep searching in other files + case kFound: + state->found = true; + return false; + case kDeleted: + return false; + case kCorrupt: + state->s = + Status::Corruption("corrupted key for ", state->saver.user_key); + state->found = true; + return false; + } + + // Not reached. Added to avoid false compilation warnings of + // "control reaches end of non-void function". + return false; + } + }; + + State state; + state.found = false; + state.stats = stats; + state.last_file_read = nullptr; + state.last_file_read_level = -1; + + state.options = &options; + state.ikey = k.internal_key(); + state.vset = vset_; + + state.saver.state = kNotFound; + state.saver.ucmp = vset_->icmp_.user_comparator(); + state.saver.user_key = k.user_key(); + state.saver.value = value; + + ForEachOverlapping(state.saver.user_key, state.ikey, &state, &State::Match); + + return state.found ? state.s : Status::NotFound(Slice()); +} + +bool Version::UpdateStats(const GetStats& stats) { + FileMetaData* f = stats.seek_file; + if (f != nullptr) { + f->allowed_seeks--; + if (f->allowed_seeks <= 0 && file_to_compact_ == nullptr) { + file_to_compact_ = f; + file_to_compact_level_ = stats.seek_file_level; + return true; + } + } + return false; +} + +bool Version::RecordReadSample(Slice internal_key) { + ParsedInternalKey ikey; + if (!ParseInternalKey(internal_key, &ikey)) { + return false; + } + + struct State { + GetStats stats; // Holds first matching file + int matches; + + static bool Match(void* arg, int level, FileMetaData* f) { + State* state = reinterpret_cast<State*>(arg); + state->matches++; + if (state->matches == 1) { + // Remember first match. + state->stats.seek_file = f; + state->stats.seek_file_level = level; + } + // We can stop iterating once we have a second match. + return state->matches < 2; + } + }; + + State state; + state.matches = 0; + ForEachOverlapping(ikey.user_key, internal_key, &state, &State::Match); + + // Must have at least two matches since we want to merge across + // files. But what if we have a single file that contains many + // overwrites and deletions? Should we have another mechanism for + // finding such files? + if (state.matches >= 2) { + // 1MB cost is about 1 seek (see comment in Builder::Apply). + return UpdateStats(state.stats); + } + return false; +} + +void Version::Ref() { ++refs_; } + +void Version::Unref() { + assert(this != &vset_->dummy_versions_); + assert(refs_ >= 1); + --refs_; + if (refs_ == 0) { + delete this; + } +} + +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 level = 0; + if (!OverlapInLevel(0, &smallest_user_key, &largest_user_key)) { + // Push to next level if there is no overlap in next level, + // and the #bytes overlapping in the level after that are limited. + InternalKey start(smallest_user_key, kMaxSequenceNumber, kValueTypeForSeek); + InternalKey limit(largest_user_key, 0, static_cast<ValueType>(0)); + std::vector<FileMetaData*> overlaps; + while (level < config::kMaxMemCompactLevel) { + if (OverlapInLevel(level + 1, &smallest_user_key, &largest_user_key)) { + break; + } + if (level + 2 < config::kNumLevels) { + // Check that file does not overlap too many grandparent bytes. + GetOverlappingInputs(level + 2, &start, &limit, &overlaps); + const int64_t sum = TotalFileSize(overlaps); + if (sum > MaxGrandParentOverlapBytes(vset_->options_)) { + break; + } + } + level++; + } + } + return level; +} + +// Store in "*inputs" all files in "level" that overlap [begin,end] +void Version::GetOverlappingInputs(int level, const InternalKey* begin, + const InternalKey* end, + std::vector<FileMetaData*>* inputs) { + assert(level >= 0); + assert(level < config::kNumLevels); + inputs->clear(); + Slice user_begin, user_end; + if (begin != nullptr) { + user_begin = begin->user_key(); + } + 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();) { + FileMetaData* f = files_[level][i++]; + const Slice file_start = f->smallest.user_key(); + const Slice file_limit = f->largest.user_key(); + if (begin != nullptr && user_cmp->Compare(file_limit, user_begin) < 0) { + // "f" is completely before specified range; skip it + } else if (end != nullptr && user_cmp->Compare(file_start, user_end) > 0) { + // "f" is completely after specified range; skip it + } else { + inputs->push_back(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 != nullptr && user_cmp->Compare(file_start, user_begin) < 0) { + user_begin = file_start; + inputs->clear(); + i = 0; + } else if (end != nullptr && + user_cmp->Compare(file_limit, user_end) > 0) { + user_end = file_limit; + inputs->clear(); + i = 0; + } + } + } + } +} + +std::string Version::DebugString() const { + std::string r; + for (int level = 0; level < config::kNumLevels; level++) { + // E.g., + // --- level 1 --- + // 17:123['a' .. 'd'] + // 20:43['e' .. 'g'] + r.append("--- level "); + AppendNumberTo(&r, level); + r.append(" ---\n"); + const std::vector<FileMetaData*>& files = files_[level]; + for (size_t i = 0; i < files.size(); i++) { + r.push_back(' '); + AppendNumberTo(&r, files[i]->number); + r.push_back(':'); + AppendNumberTo(&r, files[i]->file_size); + r.append("["); + r.append(files[i]->smallest.DebugString()); + r.append(" .. "); + r.append(files[i]->largest.DebugString()); + r.append("]\n"); + } + } + return r; +} + +// A helper class so we can efficiently apply a whole sequence +// of edits to a particular state without creating intermediate +// Versions that contain full copies of the intermediate state. +class VersionSet::Builder { + private: + // Helper to sort by v->files_[file_number].smallest + struct BySmallestKey { + const InternalKeyComparator* internal_comparator; + + bool operator()(FileMetaData* f1, FileMetaData* f2) const { + int r = internal_comparator->Compare(f1->smallest, f2->smallest); + if (r != 0) { + return (r < 0); + } else { + // Break ties by file number + return (f1->number < f2->number); + } + } + }; + + typedef std::set<FileMetaData*, BySmallestKey> FileSet; + struct LevelState { + std::set<uint64_t> deleted_files; + FileSet* added_files; + }; + + VersionSet* vset_; + Version* base_; + LevelState levels_[config::kNumLevels]; + + public: + // Initialize a builder with the files from *base and other info from *vset + Builder(VersionSet* vset, Version* base) : vset_(vset), base_(base) { + base_->Ref(); + BySmallestKey cmp; + cmp.internal_comparator = &vset_->icmp_; + for (int level = 0; level < config::kNumLevels; level++) { + levels_[level].added_files = new FileSet(cmp); + } + } + + ~Builder() { + for (int level = 0; level < config::kNumLevels; level++) { + const FileSet* added = levels_[level].added_files; + std::vector<FileMetaData*> to_unref; + to_unref.reserve(added->size()); + for (FileSet::const_iterator it = added->begin(); it != added->end(); + ++it) { + to_unref.push_back(*it); + } + delete added; + for (uint32_t i = 0; i < to_unref.size(); i++) { + FileMetaData* f = to_unref[i]; + f->refs--; + if (f->refs <= 0) { + delete f; + } + } + } + base_->Unref(); + } + + // Apply all of the edits in *edit to the current state. + void Apply(VersionEdit* edit) { + // Update compaction pointers + for (size_t i = 0; i < edit->compact_pointers_.size(); i++) { + const int level = edit->compact_pointers_[i].first; + vset_->compact_pointer_[level] = + edit->compact_pointers_[i].second.Encode().ToString(); + } + + // Delete files + 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); + } + + // Add new files + for (size_t i = 0; i < edit->new_files_.size(); i++) { + const int level = edit->new_files_[i].first; + FileMetaData* f = new FileMetaData(edit->new_files_[i].second); + f->refs = 1; + + // We arrange to automatically compact this file after + // a certain number of seeks. Let's assume: + // (1) One seek costs 10ms + // (2) Writing or reading 1MB costs 10ms (100MB/s) + // (3) A compaction of 1MB does 25MB of IO: + // 1MB read from this level + // 10-12MB read from next level (boundaries may be misaligned) + // 10-12MB written to next level + // This implies that 25 seeks cost the same as the compaction + // of 1MB of data. I.e., one seek costs approximately the + // 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 = static_cast<int>((f->file_size / 16384U)); + if (f->allowed_seeks < 100) f->allowed_seeks = 100; + + levels_[level].deleted_files.erase(f->number); + levels_[level].added_files->insert(f); + } + } + + // Save the current state in *v. + void SaveTo(Version* v) { + BySmallestKey cmp; + cmp.internal_comparator = &vset_->icmp_; + for (int level = 0; level < config::kNumLevels; level++) { + // Merge the set of added files with the set of pre-existing files. + // Drop any deleted files. Store the result in *v. + const std::vector<FileMetaData*>& base_files = base_->files_[level]; + std::vector<FileMetaData*>::const_iterator base_iter = base_files.begin(); + std::vector<FileMetaData*>::const_iterator base_end = base_files.end(); + 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<FileMetaData*>::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_file); + } + + // Add remaining base files + for (; base_iter != base_end; ++base_iter) { + MaybeAddFile(v, level, *base_iter); + } + +#ifndef NDEBUG + // 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& this_begin = v->files_[level][i]->smallest; + if (vset_->icmp_.Compare(prev_end, this_begin) >= 0) { + std::fprintf(stderr, "overlapping ranges in same level %s vs. %s\n", + prev_end.DebugString().c_str(), + this_begin.DebugString().c_str()); + std::abort(); + } + } + } +#endif + } + } + + void MaybeAddFile(Version* v, int level, FileMetaData* f) { + if (levels_[level].deleted_files.count(f->number) > 0) { + // File is deleted: do nothing + } else { + std::vector<FileMetaData*>* files = &v->files_[level]; + if (level > 0 && !files->empty()) { + // Must not overlap + assert(vset_->icmp_.Compare((*files)[files->size() - 1]->largest, + f->smallest) < 0); + } + f->refs++; + files->push_back(f); + } + } +}; + +VersionSet::VersionSet(const std::string& dbname, const Options* options, + TableCache* table_cache, + const InternalKeyComparator* cmp) + : env_(options->env), + dbname_(dbname), + options_(options), + table_cache_(table_cache), + icmp_(*cmp), + next_file_number_(2), + manifest_file_number_(0), // Filled by Recover() + last_sequence_(0), + log_number_(0), + prev_log_number_(0), + descriptor_file_(nullptr), + descriptor_log_(nullptr), + dummy_versions_(this), + current_(nullptr) { + AppendVersion(new Version(this)); +} + +VersionSet::~VersionSet() { + current_->Unref(); + assert(dummy_versions_.next_ == &dummy_versions_); // List must be empty + delete descriptor_log_; + delete descriptor_file_; +} + +void VersionSet::AppendVersion(Version* v) { + // Make "v" current + assert(v->refs_ == 0); + assert(v != current_); + if (current_ != nullptr) { + current_->Unref(); + } + current_ = v; + v->Ref(); + + // Append to linked list + v->prev_ = dummy_versions_.prev_; + v->next_ = &dummy_versions_; + v->prev_->next_ = v; + v->next_->prev_ = v; +} + +Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) { + if (edit->has_log_number_) { + assert(edit->log_number_ >= log_number_); + assert(edit->log_number_ < next_file_number_); + } else { + edit->SetLogNumber(log_number_); + } + + if (!edit->has_prev_log_number_) { + edit->SetPrevLogNumber(prev_log_number_); + } + + edit->SetNextFile(next_file_number_); + edit->SetLastSequence(last_sequence_); + + Version* v = new Version(this); + { + Builder builder(this, current_); + builder.Apply(edit); + builder.SaveTo(v); + } + Finalize(v); + + // Initialize new descriptor log file if necessary by creating + // a temporary file that contains a snapshot of the current version. + std::string new_manifest_file; + Status s; + 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_ == nullptr); + new_manifest_file = DescriptorFileName(dbname_, manifest_file_number_); + edit->SetNextFile(next_file_number_); + s = env_->NewWritableFile(new_manifest_file, &descriptor_file_); + if (s.ok()) { + descriptor_log_ = new log::Writer(descriptor_file_); + s = WriteSnapshot(descriptor_log_); + } + } + + // Unlock during expensive MANIFEST log write + { + mu->Unlock(); + + // Write new record to MANIFEST log + if (s.ok()) { + std::string record; + edit->EncodeTo(&record); + s = descriptor_log_->AddRecord(record); + if (s.ok()) { + s = descriptor_file_->Sync(); + } + if (!s.ok()) { + Log(options_->info_log, "MANIFEST write: %s\n", s.ToString().c_str()); + } + } + + // If we just created a new descriptor file, install it by writing a + // new CURRENT file that points to it. + if (s.ok() && !new_manifest_file.empty()) { + s = SetCurrentFile(env_, dbname_, manifest_file_number_); + } + + mu->Lock(); + } + + // Install the new version + if (s.ok()) { + AppendVersion(v); + log_number_ = edit->log_number_; + prev_log_number_ = edit->prev_log_number_; + } else { + delete v; + if (!new_manifest_file.empty()) { + delete descriptor_log_; + delete descriptor_file_; + descriptor_log_ = nullptr; + descriptor_file_ = nullptr; + env_->RemoveFile(new_manifest_file); + } + } + + return s; +} + +Status VersionSet::Recover(bool* save_manifest) { + struct LogReporter : public log::Reader::Reporter { + Status* status; + void Corruption(size_t bytes, const Status& s) override { + if (this->status->ok()) *this->status = s; + } + }; + + // Read "CURRENT" file, which contains a pointer to the current manifest file + std::string current; + Status s = ReadFileToString(env_, CurrentFileName(dbname_), ¤t); + if (!s.ok()) { + return s; + } + if (current.empty() || current[current.size() - 1] != '\n') { + return Status::Corruption("CURRENT file does not end with newline"); + } + current.resize(current.size() - 1); + + std::string dscname = dbname_ + "/" + current; + 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; + } + + bool have_log_number = false; + bool have_prev_log_number = false; + bool have_next_file = false; + bool have_last_sequence = false; + uint64_t next_file = 0; + uint64_t last_sequence = 0; + uint64_t log_number = 0; + uint64_t prev_log_number = 0; + Builder builder(this, current_); + int read_records = 0; + + { + LogReporter reporter; + reporter.status = &s; + log::Reader reader(file, &reporter, true /*checksum*/, + 0 /*initial_offset*/); + Slice record; + std::string scratch; + while (reader.ReadRecord(&record, &scratch) && s.ok()) { + ++read_records; + VersionEdit edit; + s = edit.DecodeFrom(record); + if (s.ok()) { + if (edit.has_comparator_ && + edit.comparator_ != icmp_.user_comparator()->Name()) { + s = Status::InvalidArgument( + edit.comparator_ + " does not match existing comparator ", + icmp_.user_comparator()->Name()); + } + } + + if (s.ok()) { + builder.Apply(&edit); + } + + if (edit.has_log_number_) { + log_number = edit.log_number_; + have_log_number = true; + } + + if (edit.has_prev_log_number_) { + prev_log_number = edit.prev_log_number_; + have_prev_log_number = true; + } + + if (edit.has_next_file_number_) { + next_file = edit.next_file_number_; + have_next_file = true; + } + + if (edit.has_last_sequence_) { + last_sequence = edit.last_sequence_; + have_last_sequence = true; + } + } + } + delete file; + file = nullptr; + + if (s.ok()) { + if (!have_next_file) { + s = Status::Corruption("no meta-nextfile entry in descriptor"); + } else if (!have_log_number) { + s = Status::Corruption("no meta-lognumber entry in descriptor"); + } else if (!have_last_sequence) { + s = Status::Corruption("no last-sequence-number entry in descriptor"); + } + + if (!have_prev_log_number) { + prev_log_number = 0; + } + + MarkFileNumberUsed(prev_log_number); + MarkFileNumberUsed(log_number); + } + + if (s.ok()) { + Version* v = new Version(this); + builder.SaveTo(v); + // Install recovered version + Finalize(v); + AppendVersion(v); + manifest_file_number_ = next_file; + next_file_number_ = next_file + 1; + 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; + } + } else { + std::string error = s.ToString(); + Log(options_->info_log, "Error recovering version set with %d records: %s", + read_records, error.c_str()); + } + + 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; + } +} + +void VersionSet::Finalize(Version* v) { + // Precomputed best level for next compaction + int best_level = -1; + double best_score = -1; + + 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 + // instead of number of bytes for two reasons: + // + // (1) With larger write-buffer sizes, it is nice not to do too + // many level-0 compactions. + // + // (2) The files in level-0 are merged on every read and + // therefore we wish to avoid too many files when the individual + // file size is small (perhaps because of a small write-buffer + // setting, or very high compression ratios, or lots of + // overwrites/deletions). + score = v->files_[level].size() / + static_cast<double>(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<double>(level_bytes) / MaxBytesForLevel(options_, level); + } + + if (score > best_score) { + best_level = level; + best_score = score; + } + } + + v->compaction_level_ = best_level; + v->compaction_score_ = best_score; +} + +Status VersionSet::WriteSnapshot(log::Writer* log) { + // TODO: Break up into multiple records to reduce memory usage on recovery? + + // Save metadata + VersionEdit edit; + edit.SetComparatorName(icmp_.user_comparator()->Name()); + + // Save compaction pointers + for (int level = 0; level < config::kNumLevels; level++) { + if (!compact_pointer_[level].empty()) { + InternalKey key; + key.DecodeFrom(compact_pointer_[level]); + edit.SetCompactPointer(level, key); + } + } + + // Save files + for (int level = 0; level < config::kNumLevels; level++) { + const std::vector<FileMetaData*>& files = current_->files_[level]; + for (size_t i = 0; i < files.size(); i++) { + const FileMetaData* f = files[i]; + edit.AddFile(level, f->number, f->file_size, f->smallest, f->largest); + } + } + + std::string record; + edit.EncodeTo(&record); + return log->AddRecord(record); +} + +int VersionSet::NumLevelFiles(int level) const { + assert(level >= 0); + assert(level < config::kNumLevels); + return current_->files_[level].size(); +} + +const char* VersionSet::LevelSummary(LevelSummaryStorage* scratch) const { + // Update code if kNumLevels changes + static_assert(config::kNumLevels == 7, ""); + std::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())); + return scratch->buffer; +} + +uint64_t VersionSet::ApproximateOffsetOf(Version* v, const InternalKey& ikey) { + uint64_t result = 0; + for (int level = 0; level < config::kNumLevels; level++) { + const std::vector<FileMetaData*>& files = v->files_[level]; + for (size_t i = 0; i < files.size(); i++) { + if (icmp_.Compare(files[i]->largest, ikey) <= 0) { + // Entire file is before "ikey", so just add the file size + result += files[i]->file_size; + } else if (icmp_.Compare(files[i]->smallest, ikey) > 0) { + // Entire file is after "ikey", so ignore + if (level > 0) { + // Files other than level 0 are sorted by meta->smallest, so + // no further files in this level will contain data for + // "ikey". + break; + } + } else { + // "ikey" falls in the range for this table. Add the + // approximate offset of "ikey" within the table. + Table* tableptr; + Iterator* iter = table_cache_->NewIterator( + ReadOptions(), files[i]->number, files[i]->file_size, &tableptr); + if (tableptr != nullptr) { + result += tableptr->ApproximateOffsetOf(ikey.Encode()); + } + delete iter; + } + } + } + return result; +} + +void VersionSet::AddLiveFiles(std::set<uint64_t>* live) { + for (Version* v = dummy_versions_.next_; v != &dummy_versions_; + v = v->next_) { + for (int level = 0; level < config::kNumLevels; level++) { + const std::vector<FileMetaData*>& files = v->files_[level]; + for (size_t i = 0; i < files.size(); i++) { + live->insert(files[i]->number); + } + } + } +} + +int64_t VersionSet::NumLevelBytes(int level) const { + assert(level >= 0); + assert(level < config::kNumLevels); + return TotalFileSize(current_->files_[level]); +} + +int64_t VersionSet::MaxNextLevelOverlappingBytes() { + int64_t result = 0; + std::vector<FileMetaData*> overlaps; + 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, + &overlaps); + const int64_t sum = TotalFileSize(overlaps); + if (sum > result) { + result = sum; + } + } + } + return result; +} + +// Stores the minimal range that covers all entries in inputs in +// *smallest, *largest. +// REQUIRES: inputs is not empty +void VersionSet::GetRange(const std::vector<FileMetaData*>& inputs, + InternalKey* smallest, InternalKey* largest) { + assert(!inputs.empty()); + smallest->Clear(); + largest->Clear(); + for (size_t i = 0; i < inputs.size(); i++) { + FileMetaData* f = inputs[i]; + if (i == 0) { + *smallest = f->smallest; + *largest = f->largest; + } else { + if (icmp_.Compare(f->smallest, *smallest) < 0) { + *smallest = f->smallest; + } + if (icmp_.Compare(f->largest, *largest) > 0) { + *largest = f->largest; + } + } + } +} + +// Stores the minimal range that covers all entries in inputs1 and inputs2 +// in *smallest, *largest. +// REQUIRES: inputs is not empty +void VersionSet::GetRange2(const std::vector<FileMetaData*>& inputs1, + const std::vector<FileMetaData*>& inputs2, + InternalKey* smallest, InternalKey* largest) { + std::vector<FileMetaData*> all = inputs1; + all.insert(all.end(), inputs2.begin(), inputs2.end()); + GetRange(all, smallest, largest); +} + +Iterator* VersionSet::MakeInputIterator(Compaction* c) { + ReadOptions options; + options.verify_checksums = options_->paranoid_checks; + options.fill_cache = false; + + // Level-0 files have to be merged together. For other levels, + // we will make a concatenating iterator per level. + // TODO(opt): use concatenating iterator for level-0 if there is no overlap + const int space = (c->level() == 0 ? c->inputs_[0].size() + 1 : 2); + Iterator** list = new Iterator*[space]; + int num = 0; + for (int which = 0; which < 2; which++) { + if (!c->inputs_[which].empty()) { + if (c->level() + which == 0) { + const std::vector<FileMetaData*>& 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); + } + } else { + // Create concatenating iterator for the files from this level + list[num++] = NewTwoLevelIterator( + new Version::LevelFileNumIterator(icmp_, &c->inputs_[which]), + &GetFileIterator, table_cache_, options); + } + } + } + assert(num <= space); + Iterator* result = NewMergingIterator(&icmp_, list, num); + delete[] list; + return result; +} + +Compaction* VersionSet::PickCompaction() { + Compaction* c; + int level; + + // 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_ != nullptr); + if (size_compaction) { + level = current_->compaction_level_; + assert(level >= 0); + 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++) { + FileMetaData* f = current_->files_[level][i]; + if (compact_pointer_[level].empty() || + icmp_.Compare(f->largest.Encode(), compact_pointer_[level]) > 0) { + c->inputs_[0].push_back(f); + break; + } + } + if (c->inputs_[0].empty()) { + // Wrap-around to the beginning of the key space + c->inputs_[0].push_back(current_->files_[level][0]); + } + } else if (seek_compaction) { + level = current_->file_to_compact_level_; + c = new Compaction(options_, level); + c->inputs_[0].push_back(current_->file_to_compact_); + } else { + return nullptr; + } + + c->input_version_ = current_; + c->input_version_->Ref(); + + // Files in level 0 may overlap each other, so pick up all overlapping ones + if (level == 0) { + InternalKey smallest, largest; + GetRange(c->inputs_[0], &smallest, &largest); + // Note that the next call will discard the file we placed in + // c->inputs_[0] earlier and replace it with an overlapping set + // which will include the picked file. + current_->GetOverlappingInputs(0, &smallest, &largest, &c->inputs_[0]); + assert(!c->inputs_[0].empty()); + } + + SetupOtherInputs(c); + + 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<FileMetaData*>& 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<FileMetaData*>& 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<FileMetaData*>& level_files, + std::vector<FileMetaData*>* 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]); + + // Get entire range covered by compaction + InternalKey all_start, all_limit; + GetRange2(c->inputs_[0], c->inputs_[1], &all_start, &all_limit); + + // See if we can grow the number of inputs in "level" without + // changing the number of "level+1" files we pick up. + if (!c->inputs_[1].empty()) { + std::vector<FileMetaData*> 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 < + ExpandedCompactionByteSizeLimit(options_)) { + InternalKey new_start, new_limit; + GetRange(expanded0, &new_start, &new_limit); + std::vector<FileMetaData*> expanded1; + 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)); + smallest = new_start; + largest = new_limit; + c->inputs_[0] = expanded0; + c->inputs_[1] = expanded1; + GetRange2(c->inputs_[0], c->inputs_[1], &all_start, &all_limit); + } + } + } + + // Compute the set of grandparent files that overlap this compaction + // (parent == level+1; grandparent == level+2) + if (level + 2 < config::kNumLevels) { + current_->GetOverlappingInputs(level + 2, &all_start, &all_limit, + &c->grandparents_); + } + + // 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 + // key range next time. + compact_pointer_[level] = largest.Encode().ToString(); + c->edit_.SetCompactPointer(level, largest); +} + +Compaction* VersionSet::CompactRange(int level, const InternalKey* begin, + const InternalKey* end) { + std::vector<FileMetaData*> inputs; + current_->GetOverlappingInputs(level, begin, end, &inputs); + if (inputs.empty()) { + return nullptr; + } + + // Avoid compacting too much in one shot in case the range is large. + // But we cannot do this for level-0 since level-0 files can overlap + // and we must not pick one file and drop another older file if the + // two files overlap. + if (level > 0) { + 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; + total += s; + if (total >= limit) { + inputs.resize(i + 1); + break; + } + } + } + + Compaction* c = new Compaction(options_, level); + c->input_version_ = current_; + c->input_version_->Ref(); + c->inputs_[0] = inputs; + SetupOtherInputs(c); + return c; +} + +Compaction::Compaction(const Options* options, int level) + : level_(level), + max_output_file_size_(MaxFileSizeForLevel(options, level)), + input_version_(nullptr), + grandparent_index_(0), + seen_key_(false), + overlapped_bytes_(0) { + for (int i = 0; i < config::kNumLevels; i++) { + level_ptrs_[i] = 0; + } +} + +Compaction::~Compaction() { + 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_) <= + MaxGrandParentOverlapBytes(vset->options_)); +} + +void Compaction::AddInputDeletions(VersionEdit* edit) { + for (int which = 0; which < 2; which++) { + for (size_t i = 0; i < inputs_[which].size(); i++) { + edit->RemoveFile(level_ + which, inputs_[which][i]->number); + } + } +} + +bool Compaction::IsBaseLevelForKey(const Slice& user_key) { + // Maybe use binary search to find right entry instead of linear search? + const Comparator* user_cmp = input_version_->vset_->icmp_.user_comparator(); + for (int lvl = level_ + 2; lvl < config::kNumLevels; lvl++) { + const std::vector<FileMetaData*>& files = input_version_->files_[lvl]; + 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 + if (user_cmp->Compare(user_key, f->smallest.user_key()) >= 0) { + // Key falls in this file's range, so definitely not base level + return false; + } + break; + } + level_ptrs_[lvl]++; + } + } + return true; +} + +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 = &vset->icmp_; + while (grandparent_index_ < grandparents_.size() && + icmp->Compare(internal_key, + grandparents_[grandparent_index_]->largest.Encode()) > + 0) { + if (seen_key_) { + overlapped_bytes_ += grandparents_[grandparent_index_]->file_size; + } + grandparent_index_++; + } + seen_key_ = true; + + if (overlapped_bytes_ > MaxGrandParentOverlapBytes(vset->options_)) { + // Too much overlap for current output; start new output + overlapped_bytes_ = 0; + return true; + } else { + return false; + } +} + +void Compaction::ReleaseInputs() { + if (input_version_ != nullptr) { + input_version_->Unref(); + input_version_ = nullptr; + } +} + +} // namespace leveldb diff --git a/db/version_set.h b/db/version_set.h new file mode 100644 index 0000000..69f3d70 --- /dev/null +++ b/db/version_set.h @@ -0,0 +1,393 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// The representation of a DBImpl consists of a set of Versions. The +// newest version is called "current". Older versions may be kept +// around to provide a consistent view to live iterators. +// +// Each Version keeps track of a set of Table files per level. The +// entire set of versions is maintained in a VersionSet. +// +// Version,VersionSet are thread-compatible, but require external +// synchronization on all accesses. + +#ifndef STORAGE_LEVELDB_DB_VERSION_SET_H_ +#define STORAGE_LEVELDB_DB_VERSION_SET_H_ + +#include <map> +#include <set> +#include <vector> + +#include "db/dbformat.h" +#include "db/version_edit.h" +#include "port/port.h" +#include "port/thread_annotations.h" + +namespace leveldb { + +namespace log { +class Writer; +} + +class Compaction; +class Iterator; +class MemTable; +class TableBuilder; +class TableCache; +class Version; +class VersionSet; +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. +int FindFile(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>& files, const Slice& key); + +// Returns true iff some file in "files" overlaps the user key range +// [*smallest,*largest]. +// smallest==nullptr represents a key smaller than all keys in the DB. +// largest==nullptr represents a key largest than all keys in the DB. +// REQUIRES: If disjoint_sorted_files, files[] contains disjoint ranges +// in sorted order. +bool SomeFileOverlapsRange(const InternalKeyComparator& icmp, + bool disjoint_sorted_files, + const std::vector<FileMetaData*>& files, + const Slice* smallest_user_key, + const Slice* largest_user_key); + +class Version { + public: + // 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 + struct GetStats { + 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<Iterator*>* iters); + + Status Get(const ReadOptions&, const LookupKey& key, std::string* val, + GetStats* stats); + + // Adds "stats" into the current state. Returns true if a new + // compaction may need to be triggered, false otherwise. + // REQUIRES: lock is held + bool UpdateStats(const GetStats& stats); + + // Record a sample of bytes read at the specified internal key. + // Samples are taken approximately once every config::kReadBytesPeriod + // bytes. Returns true if a new compaction may need to be triggered. + // REQUIRES: lock is held + bool RecordReadSample(Slice key); + + // Reference count management (so Versions do not disappear out from + // under live iterators) + void Ref(); + void Unref(); + + void GetOverlappingInputs( + int level, + const InternalKey* begin, // nullptr means before all keys + const InternalKey* end, // nullptr means after all keys + std::vector<FileMetaData*>* inputs); + + // Returns true iff some file in the specified level overlaps + // some part of [*smallest_user_key,*largest_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 + // result that covers the range [smallest_user_key,largest_user_key]. + int PickLevelForMemTableOutput(const Slice& smallest_user_key, + const Slice& largest_user_key); + + int NumFiles(int level) const { return files_[level].size(); } + + // Return a human readable string that describes this version's contents. + std::string DebugString() const; + + private: + friend class Compaction; + 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 + // order from newest to oldest. If an invocation of func returns + // false, makes no more calls. + // + // REQUIRES: user portion of internal_key == user_key. + 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 + + // List of files per level + std::vector<FileMetaData*> files_[config::kNumLevels]; + + // Next file to compact based on seek stats. + FileMetaData* file_to_compact_; + int file_to_compact_level_; + + // Level that should be compacted next and its compaction score. + // Score < 1 means compaction is not strictly needed. These fields + // are initialized by Finalize(). + double compaction_score_; + int compaction_level_; +}; + +class VersionSet { + public: + 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 + // is both saved to persistent state and installed as the new + // current version. Will release *mu while actually writing to the file. + // REQUIRES: *mu is held on entry. + // REQUIRES: no other thread concurrently calls LogAndApply() + Status LogAndApply(VersionEdit* edit, port::Mutex* mu) + EXCLUSIVE_LOCKS_REQUIRED(mu); + + // Recover the last saved descriptor from persistent storage. + Status Recover(bool* save_manifest); + + // Return the current version. + Version* current() const { return current_; } + + // Return the current manifest file number + uint64_t ManifestFileNumber() const { return manifest_file_number_; } + + // Allocate and return a new file number + uint64_t NewFileNumber() { return next_file_number_++; } + + // Arrange to reuse "file_number" unless a newer file number has + // already been allocated. + // REQUIRES: "file_number" was returned by a call to NewFileNumber(). + void ReuseFileNumber(uint64_t file_number) { + if (next_file_number_ == file_number + 1) { + next_file_number_ = file_number; + } + } + + // Return the number of Table files at the specified level. + int NumLevelFiles(int level) const; + + // Return the combined file size of all files at the specified level. + int64_t NumLevelBytes(int level) const; + + // Return the last sequence number. + uint64_t LastSequence() const { return last_sequence_; } + + // Set the last sequence number to s. + void SetLastSequence(uint64_t s) { + assert(s >= last_sequence_); + last_sequence_ = s; + } + + // Mark the specified file number as used. + void MarkFileNumberUsed(uint64_t number); + + // Return the current log file number. + uint64_t LogNumber() const { return log_number_; } + + // Return the log file number for the log file that is currently + // being compacted, or zero if there is no such log file. + uint64_t PrevLogNumber() const { return prev_log_number_; } + + // Pick level and inputs for a new compaction. + // Returns nullptr if there is no compaction to be done. + // Otherwise returns a pointer to a heap-allocated object that + // describes the compaction. Caller should delete the result. + Compaction* PickCompaction(); + + // Return a compaction object for compacting the range [begin,end] in + // the specified level. Returns nullptr if there is nothing in that + // level that overlaps the specified range. Caller should delete + // the result. + 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. + int64_t MaxNextLevelOverlappingBytes(); + + // Create an iterator that reads over the compaction inputs for "*c". + // The caller should delete the iterator when no longer needed. + Iterator* MakeInputIterator(Compaction* c); + + // Returns true iff some level needs a compaction. + bool NeedsCompaction() const { + Version* v = current_; + return (v->compaction_score_ >= 1) || (v->file_to_compact_ != nullptr); + } + + // Add all files listed in any live version to *live. + // May also mutate some internal state. + void AddLiveFiles(std::set<uint64_t>* live); + + // Return the approximate offset in the database of the data for + // "key" as of version "v". + uint64_t ApproximateOffsetOf(Version* v, const InternalKey& key); + + // Return a human-readable short (single-line) summary of the number + // of files per level. Uses *scratch as backing store. + struct LevelSummaryStorage { + char buffer[100]; + }; + const char* LevelSummary(LevelSummaryStorage* scratch) const; + + private: + class Builder; + + 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<FileMetaData*>& inputs, InternalKey* smallest, + InternalKey* largest); + + void GetRange2(const std::vector<FileMetaData*>& inputs1, + const std::vector<FileMetaData*>& inputs2, + InternalKey* smallest, InternalKey* largest); + + void SetupOtherInputs(Compaction* c); + + // Save current contents to *log + Status WriteSnapshot(log::Writer* log); + + void AppendVersion(Version* v); + + Env* const env_; + const std::string dbname_; + const Options* const options_; + TableCache* const table_cache_; + const InternalKeyComparator icmp_; + uint64_t next_file_number_; + uint64_t manifest_file_number_; + uint64_t last_sequence_; + uint64_t log_number_; + uint64_t prev_log_number_; // 0 or backing store for memtable being compacted + + // Opened lazily + WritableFile* descriptor_file_; + log::Writer* descriptor_log_; + Version dummy_versions_; // Head of circular doubly-linked list of versions. + Version* current_; // == dummy_versions_.prev_ + + // 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]; +}; + +// A Compaction encapsulates information about a compaction. +class Compaction { + public: + ~Compaction(); + + // Return the level that is being compacted. Inputs from "level" + // and "level+1" will be merged to produce a set of "level+1" files. + int level() const { return level_; } + + // Return the object that holds the edits to the descriptor done + // by this compaction. + VersionEdit* edit() { return &edit_; } + + // "which" must be either 0 or 1 + int num_input_files(int which) const { return inputs_[which].size(); } + + // Return the ith input file at "level()+which" ("which" must be 0 or 1). + FileMetaData* input(int which, int i) const { return inputs_[which][i]; } + + // Maximum size of files to build during this compaction. + uint64_t MaxOutputFileSize() const { return max_output_file_size_; } + + // Is this a trivial compaction that can be implemented by just + // moving a single input file to the next level (no merging or splitting) + bool IsTrivialMove() const; + + // Add all inputs to this compaction as delete operations to *edit. + void AddInputDeletions(VersionEdit* edit); + + // Returns true if the information we have available guarantees that + // the compaction is producing data in "level+1" for which no data exists + // in levels greater than "level+1". + bool IsBaseLevelForKey(const Slice& user_key); + + // Returns true iff we should stop building the current output + // before processing "internal_key". + bool ShouldStopBefore(const Slice& internal_key); + + // Release the input version for the compaction, once the compaction + // is successful. + void ReleaseInputs(); + + private: + friend class Version; + friend class VersionSet; + + Compaction(const Options* options, int level); + + int level_; + uint64_t max_output_file_size_; + Version* input_version_; + VersionEdit edit_; + + // Each compaction reads inputs from "level_" and "level_+1" + std::vector<FileMetaData*> inputs_[2]; // The two sets of inputs + + // State used to check for number of overlapping grandparent files + // (parent == level_ + 1, grandparent == level_ + 2) + std::vector<FileMetaData*> grandparents_; + size_t grandparent_index_; // Index in grandparent_starts_ + bool seen_key_; // Some output key has been seen + int64_t overlapped_bytes_; // Bytes of overlap between current output + // and grandparent files + + // State for implementing IsBaseLevelForKey + + // level_ptrs_ holds indices into input_version_->levels_: our state + // is that we are positioned at one of the file ranges for each + // higher level than the ones involved in this compaction (i.e. for + // all L >= level_ + 2). + size_t level_ptrs_[config::kNumLevels]; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_VERSION_SET_H_ diff --git a/db/version_set_test.cc b/db/version_set_test.cc new file mode 100644 index 0000000..694ddfa --- /dev/null +++ b/db/version_set_test.cc @@ -0,0 +1,332 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "db/version_set.h" + +#include "gtest/gtest.h" +#include "util/logging.h" +#include "util/testutil.h" + +namespace leveldb { + +class FindFileTest : public testing::Test { + public: + FindFileTest() : disjoint_sorted_files_(true) {} + + ~FindFileTest() { + for (int i = 0; i < files_.size(); i++) { + delete files_[i]; + } + } + + void Add(const char* smallest, const char* largest, + SequenceNumber smallest_seq = 100, + SequenceNumber largest_seq = 100) { + FileMetaData* f = new FileMetaData; + f->number = files_.size() + 1; + f->smallest = InternalKey(smallest, smallest_seq, kTypeValue); + f->largest = InternalKey(largest, largest_seq, kTypeValue); + files_.push_back(f); + } + + int Find(const char* key) { + InternalKey target(key, 100, kTypeValue); + InternalKeyComparator cmp(BytewiseComparator()); + return FindFile(cmp, files_, target.Encode()); + } + + bool Overlaps(const char* smallest, const char* largest) { + InternalKeyComparator cmp(BytewiseComparator()); + Slice s(smallest != nullptr ? smallest : ""); + Slice l(largest != nullptr ? largest : ""); + return SomeFileOverlapsRange(cmp, disjoint_sorted_files_, files_, + (smallest != nullptr ? &s : nullptr), + (largest != nullptr ? &l : nullptr)); + } + + bool disjoint_sorted_files_; + + private: + std::vector<FileMetaData*> files_; +}; + +TEST_F(FindFileTest, Empty) { + ASSERT_EQ(0, Find("foo")); + ASSERT_TRUE(!Overlaps("a", "z")); + ASSERT_TRUE(!Overlaps(nullptr, "z")); + ASSERT_TRUE(!Overlaps("a", nullptr)); + ASSERT_TRUE(!Overlaps(nullptr, nullptr)); +} + +TEST_F(FindFileTest, Single) { + Add("p", "q"); + ASSERT_EQ(0, Find("a")); + ASSERT_EQ(0, Find("p")); + ASSERT_EQ(0, Find("p1")); + ASSERT_EQ(0, Find("q")); + ASSERT_EQ(1, Find("q1")); + ASSERT_EQ(1, Find("z")); + + ASSERT_TRUE(!Overlaps("a", "b")); + ASSERT_TRUE(!Overlaps("z1", "z2")); + ASSERT_TRUE(Overlaps("a", "p")); + ASSERT_TRUE(Overlaps("a", "q")); + ASSERT_TRUE(Overlaps("a", "z")); + ASSERT_TRUE(Overlaps("p", "p1")); + ASSERT_TRUE(Overlaps("p", "q")); + ASSERT_TRUE(Overlaps("p", "z")); + ASSERT_TRUE(Overlaps("p1", "p2")); + ASSERT_TRUE(Overlaps("p1", "z")); + ASSERT_TRUE(Overlaps("q", "q")); + ASSERT_TRUE(Overlaps("q", "q1")); + + ASSERT_TRUE(!Overlaps(nullptr, "j")); + ASSERT_TRUE(!Overlaps("r", nullptr)); + ASSERT_TRUE(Overlaps(nullptr, "p")); + ASSERT_TRUE(Overlaps(nullptr, "p1")); + ASSERT_TRUE(Overlaps("q", nullptr)); + ASSERT_TRUE(Overlaps(nullptr, nullptr)); +} + +TEST_F(FindFileTest, Multiple) { + Add("150", "200"); + Add("200", "250"); + Add("300", "350"); + Add("400", "450"); + ASSERT_EQ(0, Find("100")); + ASSERT_EQ(0, Find("150")); + ASSERT_EQ(0, Find("151")); + ASSERT_EQ(0, Find("199")); + ASSERT_EQ(0, Find("200")); + ASSERT_EQ(1, Find("201")); + ASSERT_EQ(1, Find("249")); + ASSERT_EQ(1, Find("250")); + ASSERT_EQ(2, Find("251")); + ASSERT_EQ(2, Find("299")); + ASSERT_EQ(2, Find("300")); + ASSERT_EQ(2, Find("349")); + ASSERT_EQ(2, Find("350")); + ASSERT_EQ(3, Find("351")); + ASSERT_EQ(3, Find("400")); + ASSERT_EQ(3, Find("450")); + ASSERT_EQ(4, Find("451")); + + ASSERT_TRUE(!Overlaps("100", "149")); + ASSERT_TRUE(!Overlaps("251", "299")); + ASSERT_TRUE(!Overlaps("451", "500")); + ASSERT_TRUE(!Overlaps("351", "399")); + + ASSERT_TRUE(Overlaps("100", "150")); + ASSERT_TRUE(Overlaps("100", "200")); + ASSERT_TRUE(Overlaps("100", "300")); + ASSERT_TRUE(Overlaps("100", "400")); + ASSERT_TRUE(Overlaps("100", "500")); + ASSERT_TRUE(Overlaps("375", "400")); + ASSERT_TRUE(Overlaps("450", "450")); + ASSERT_TRUE(Overlaps("450", "500")); +} + +TEST_F(FindFileTest, MultipleNullBoundaries) { + Add("150", "200"); + Add("200", "250"); + Add("300", "350"); + Add("400", "450"); + ASSERT_TRUE(!Overlaps(nullptr, "149")); + ASSERT_TRUE(!Overlaps("451", nullptr)); + ASSERT_TRUE(Overlaps(nullptr, nullptr)); + ASSERT_TRUE(Overlaps(nullptr, "150")); + ASSERT_TRUE(Overlaps(nullptr, "199")); + ASSERT_TRUE(Overlaps(nullptr, "200")); + ASSERT_TRUE(Overlaps(nullptr, "201")); + ASSERT_TRUE(Overlaps(nullptr, "400")); + ASSERT_TRUE(Overlaps(nullptr, "800")); + ASSERT_TRUE(Overlaps("100", nullptr)); + ASSERT_TRUE(Overlaps("200", nullptr)); + ASSERT_TRUE(Overlaps("449", nullptr)); + ASSERT_TRUE(Overlaps("450", nullptr)); +} + +TEST_F(FindFileTest, OverlapSequenceChecks) { + Add("200", "200", 5000, 3000); + ASSERT_TRUE(!Overlaps("199", "199")); + ASSERT_TRUE(!Overlaps("201", "300")); + ASSERT_TRUE(Overlaps("200", "200")); + ASSERT_TRUE(Overlaps("190", "200")); + ASSERT_TRUE(Overlaps("200", "210")); +} + +TEST_F(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", "150")); + ASSERT_TRUE(Overlaps("100", "200")); + ASSERT_TRUE(Overlaps("100", "300")); + ASSERT_TRUE(Overlaps("100", "400")); + ASSERT_TRUE(Overlaps("100", "500")); + ASSERT_TRUE(Overlaps("375", "400")); + ASSERT_TRUE(Overlaps("450", "450")); + ASSERT_TRUE(Overlaps("450", "500")); + ASSERT_TRUE(Overlaps("450", "700")); + ASSERT_TRUE(Overlaps("600", "700")); +} + +void AddBoundaryInputs(const InternalKeyComparator& icmp, + const std::vector<FileMetaData*>& level_files, + std::vector<FileMetaData*>* compaction_files); + +class AddBoundaryInputsTest : public testing::Test { + public: + std::vector<FileMetaData*> level_files_; + std::vector<FileMetaData*> compaction_files_; + std::vector<FileMetaData*> 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_F(AddBoundaryInputsTest, TestEmptyFileSets) { + AddBoundaryInputs(icmp_, level_files_, &compaction_files_); + ASSERT_TRUE(compaction_files_.empty()); + ASSERT_TRUE(level_files_.empty()); +} + +TEST_F(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_F(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_F(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_F(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_F(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_F(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 + diff --git a/db/write_batch.cc b/db/write_batch.cc new file mode 100644 index 0000000..b54313c --- /dev/null +++ b/db/write_batch.cc @@ -0,0 +1,150 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// WriteBatch::rep_ := +// sequence: fixed64 +// count: fixed32 +// data: record[count] +// record := +// kTypeValue varstring varstring | +// kTypeDeletion varstring +// varstring := +// len: varint32 +// data: uint8[len] + +#include "leveldb/write_batch.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 { + +// 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() = default; + +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) { + return Status::Corruption("malformed WriteBatch (too small)"); + } + + input.remove_prefix(kHeader); + Slice key, value; + int found = 0; + while (!input.empty()) { + found++; + char tag = input[0]; + input.remove_prefix(1); + switch (tag) { + case kTypeValue: + if (GetLengthPrefixedSlice(&input, &key) && + GetLengthPrefixedSlice(&input, &value)) { + handler->Put(key, value); + } else { + return Status::Corruption("bad WriteBatch Put"); + } + break; + case kTypeDeletion: + if (GetLengthPrefixedSlice(&input, &key)) { + handler->Delete(key); + } else { + return Status::Corruption("bad WriteBatch Delete"); + } + break; + default: + return Status::Corruption("unknown WriteBatch tag"); + } + } + if (found != WriteBatchInternal::Count(this)) { + return Status::Corruption("WriteBatch has wrong count"); + } else { + return Status::OK(); + } +} + +int WriteBatchInternal::Count(const WriteBatch* b) { + return DecodeFixed32(b->rep_.data() + 8); +} + +void WriteBatchInternal::SetCount(WriteBatch* b, int n) { + EncodeFixed32(&b->rep_[8], n); +} + +SequenceNumber WriteBatchInternal::Sequence(const WriteBatch* b) { + return SequenceNumber(DecodeFixed64(b->rep_.data())); +} + +void WriteBatchInternal::SetSequence(WriteBatch* b, SequenceNumber seq) { + EncodeFixed64(&b->rep_[0], seq); +} + +void WriteBatch::Put(const Slice& key, const Slice& value) { + WriteBatchInternal::SetCount(this, WriteBatchInternal::Count(this) + 1); + rep_.push_back(static_cast<char>(kTypeValue)); + PutLengthPrefixedSlice(&rep_, key); + PutLengthPrefixedSlice(&rep_, value); +} + +void WriteBatch::Delete(const Slice& key) { + WriteBatchInternal::SetCount(this, WriteBatchInternal::Count(this) + 1); + rep_.push_back(static_cast<char>(kTypeDeletion)); + PutLengthPrefixedSlice(&rep_, key); +} + +void WriteBatch::Append(const WriteBatch& source) { + WriteBatchInternal::Append(this, &source); +} + +namespace { +class MemTableInserter : public WriteBatch::Handler { + public: + SequenceNumber sequence_; + MemTable* mem_; + + void Put(const Slice& key, const Slice& value) override { + mem_->Add(sequence_, kTypeValue, key, value); + sequence_++; + } + void Delete(const Slice& key) override { + mem_->Add(sequence_, kTypeDeletion, key, Slice()); + sequence_++; + } +}; +} // namespace + +Status WriteBatchInternal::InsertInto(const WriteBatch* b, MemTable* memtable) { + MemTableInserter inserter; + inserter.sequence_ = WriteBatchInternal::Sequence(b); + inserter.mem_ = memtable; + return b->Iterate(&inserter); +} + +void WriteBatchInternal::SetContents(WriteBatch* b, const Slice& contents) { + assert(contents.size() >= kHeader); + b->rep_.assign(contents.data(), contents.size()); +} + +void WriteBatchInternal::Append(WriteBatch* dst, const WriteBatch* src) { + SetCount(dst, Count(dst) + Count(src)); + assert(src->rep_.size() >= kHeader); + dst->rep_.append(src->rep_.data() + kHeader, src->rep_.size() - kHeader); +} + +} // namespace leveldb diff --git a/db/write_batch_internal.h b/db/write_batch_internal.h new file mode 100644 index 0000000..fce86e3 --- /dev/null +++ b/db/write_batch_internal.h @@ -0,0 +1,45 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_ +#define STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_ + +#include "db/dbformat.h" +#include "leveldb/write_batch.h" + +namespace leveldb { + +class MemTable; + +// WriteBatchInternal provides static methods for manipulating a +// WriteBatch that we don't want in the public WriteBatch interface. +class WriteBatchInternal { + public: + // Return the number of entries in the batch. + static int Count(const WriteBatch* batch); + + // Set the count for the number of entries in the batch. + static void SetCount(WriteBatch* batch, int n); + + // Return the sequence number for the start of this batch. + static SequenceNumber Sequence(const WriteBatch* batch); + + // Store the specified number as the sequence number for the start of + // this batch. + static void SetSequence(WriteBatch* batch, SequenceNumber seq); + + static Slice Contents(const WriteBatch* batch) { return Slice(batch->rep_); } + + static size_t ByteSize(const WriteBatch* batch) { return batch->rep_.size(); } + + static void SetContents(WriteBatch* batch, const Slice& contents); + + static Status InsertInto(const WriteBatch* batch, MemTable* memtable); + + static void Append(WriteBatch* dst, const WriteBatch* src); +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_ diff --git a/db/write_batch_test.cc b/db/write_batch_test.cc new file mode 100644 index 0000000..baabaa2 --- /dev/null +++ b/db/write_batch_test.cc @@ -0,0 +1,133 @@ +// 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 "gtest/gtest.h" +#include "db/memtable.h" +#include "db/write_batch_internal.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "util/logging.h" + +namespace leveldb { + +static std::string PrintContents(WriteBatch* b) { + InternalKeyComparator cmp(BytewiseComparator()); + MemTable* mem = new MemTable(cmp); + mem->Ref(); + std::string state; + Status s = WriteBatchInternal::InsertInto(b, mem); + int count = 0; + Iterator* iter = mem->NewIterator(); + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + ParsedInternalKey ikey; + EXPECT_TRUE(ParseInternalKey(iter->key(), &ikey)); + switch (ikey.type) { + case kTypeValue: + state.append("Put("); + state.append(ikey.user_key.ToString()); + state.append(", "); + state.append(iter->value().ToString()); + state.append(")"); + count++; + break; + case kTypeDeletion: + state.append("Delete("); + state.append(ikey.user_key.ToString()); + state.append(")"); + count++; + break; + } + state.append("@"); + state.append(NumberToString(ikey.sequence)); + } + delete iter; + if (!s.ok()) { + state.append("ParseError()"); + } else if (count != WriteBatchInternal::Count(b)) { + state.append("CountMismatch()"); + } + mem->Unref(); + return state; +} + +TEST(WriteBatchTest, Empty) { + WriteBatch batch; + ASSERT_EQ("", PrintContents(&batch)); + ASSERT_EQ(0, WriteBatchInternal::Count(&batch)); +} + +TEST(WriteBatchTest, Multiple) { + WriteBatch batch; + batch.Put(Slice("foo"), Slice("bar")); + batch.Delete(Slice("box")); + batch.Put(Slice("baz"), Slice("boo")); + WriteBatchInternal::SetSequence(&batch, 100); + ASSERT_EQ(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)); +} + +TEST(WriteBatchTest, Corruption) { + WriteBatch batch; + batch.Put(Slice("foo"), Slice("bar")); + batch.Delete(Slice("box")); + WriteBatchInternal::SetSequence(&batch, 200); + Slice contents = WriteBatchInternal::Contents(&batch); + WriteBatchInternal::SetContents(&batch, + Slice(contents.data(), contents.size() - 1)); + ASSERT_EQ( + "Put(foo, bar)@200" + "ParseError()", + PrintContents(&batch)); +} + +TEST(WriteBatchTest, Append) { + WriteBatch b1, b2; + WriteBatchInternal::SetSequence(&b1, 200); + WriteBatchInternal::SetSequence(&b2, 300); + b1.Append(b2); + ASSERT_EQ("", PrintContents(&b1)); + b2.Put("a", "va"); + b1.Append(b2); + ASSERT_EQ("Put(a, va)@200", PrintContents(&b1)); + b2.Clear(); + b2.Put("b", "vb"); + b1.Append(b2); + ASSERT_EQ( + "Put(a, va)@200" + "Put(b, vb)@201", + PrintContents(&b1)); + b2.Delete("foo"); + 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 + diff --git a/doc/benchmark.html b/doc/benchmark.html new file mode 100644 index 0000000..1e0b4ef --- /dev/null +++ b/doc/benchmark.html @@ -0,0 +1,459 @@ +<!DOCTYPE html> +<html> +<head> +<title>LevelDB Benchmarks</title> +<meta http-equiv="Content-Type" content="text/html; charset=utf-8"> +<style> +body { + font-family:Helvetica,sans-serif; + padding:20px; +} + +h2 { + padding-top:30px; +} + +table.bn { + width:800px; + border-collapse:collapse; + border:0; + padding:0; +} + +table.bnbase { + width:650px; +} + +table.bn td { + padding:2px 0; +} + +table.bn td.c1 { + font-weight:bold; + width:150px; +} + +table.bn td.c1 div.e { + float:right; + font-weight:normal; +} + +table.bn td.c2 { + width:150px; + text-align:right; + padding:2px; +} + +table.bn td.c3 { + width:350px; +} + +table.bn td.c4 { + width:150px; + font-size:small; + padding-left:4px; +} + +/* chart bars */ +div.bldb { + background-color:#0255df; +} + +div.bkct { + background-color:#df5555; +} + +div.bsql { + background-color:#aadf55; +} + +.code { + font-family:monospace; + font-size:large; +} + +.todo { + color: red; +} + +</style> +</head> +<body> +<h1>LevelDB Benchmarks</h1> +<p>Google, July 2011</p> +<hr> + +<p>In order to test LevelDB's performance, we benchmark it against other well-established database implementations. We compare LevelDB (revision 39) against <a href="https://www.sqlite.org/">SQLite3</a> (version 3.7.6.3) and <a href="https://dbmx.net/kyotocabinet/spex.html">Kyoto Cabinet's</a> (version 1.2.67) TreeDB (a B+Tree based key-value store). We would like to acknowledge Scott Hess and Mikio Hirabayashi for their suggestions and contributions to the SQLite3 and Kyoto Cabinet benchmarks, respectively.</p> + +<p>Benchmarks were all performed on a six-core Intel(R) Xeon(R) CPU X5650 @ 2.67GHz, with 12288 KB of total L3 cache and 12 GB of DDR3 RAM at 1333 MHz. (Note that LevelDB uses at most two CPUs since the benchmarks are single threaded: one to run the benchmark, and one for background compactions.) We ran the benchmarks on two machines (with identical processors), one with an Ext3 file system and one with an Ext4 file system. The machine with the Ext3 file system has a SATA Hitachi HDS721050CLA362 hard drive. The machine with the Ext4 file system has a SATA Samsung HD502HJ hard drive. Both hard drives spin at 7200 RPM and have hard drive write-caching enabled (using `hdparm -W 1 [device]`). The numbers reported below are the median of three measurements.</p> + +<h4>Benchmark Source Code</h4> +<p>We wrote benchmark tools for SQLite and Kyoto TreeDB based on LevelDB's <span class="code">db_bench</span>. The code for each of the benchmarks resides here:</p> +<ul> + <li> <b>LevelDB:</b> <a href="https://github.com/google/leveldb/blob/master/benchmarks/db_bench.cc">benchmarks/db_bench.cc</a>.</li> + <li> <b>SQLite:</b> <a href="https://github.com/google/leveldb/blob/master/benchmarks/db_bench_sqlite3.cc">benchmarks/db_bench_sqlite3.cc</a>.</li> + <li> <b>Kyoto TreeDB:</b> <a href="https://github.com/google/leveldb/blob/master/benchmarks/db_bench_tree_db.cc">benchmarks/db_bench_tree_db.cc</a>.</li> +</ul> + +<h4>Custom Build Specifications</h4> +<ul> +<li>LevelDB: LevelDB was compiled with the <a href="https://github.com/gperftools/gperftools">tcmalloc</a> library and the <a href="https://github.com/google/snappy">Snappy</a> compression library (revision 33). Assertions were disabled.</li> +<li>TreeDB: TreeDB was compiled using the <a href="https://www.oberhumer.com/opensource/lzo/">LZO</a> compression library (version 2.03). Furthermore, we enabled the TSMALL and TLINEAR options when opening the database in order to reduce the footprint of each record.</li> +<li>SQLite: We tuned SQLite's performance, by setting its locking mode to exclusive. We also enabled SQLite's <a href="https://www.sqlite.org/draft/wal.html">write-ahead logging</a>.</li> +</ul> + +<h2>1. Baseline Performance</h2> +<p>This section gives the baseline performance of all the +databases. Following sections show how performance changes as various +parameters are varied. For the baseline:</p> +<ul> + <li> Each database is allowed 4 MB of cache memory.</li> + <li> Databases are opened in <em>asynchronous</em> write mode. + (LevelDB's sync option, TreeDB's OAUTOSYNC option, and + SQLite3's synchronous options are all turned off). I.e., + every write is pushed to the operating system, but the + benchmark does not wait for the write to reach the disk.</li> + <li> Keys are 16 bytes each.</li> + <li> Value are 100 bytes each (with enough redundancy so that + a simple compressor shrinks them to 50% of their original + size).</li> + <li> Sequential reads/writes traverse the key space in increasing order.</li> + <li> Random reads/writes traverse the key space in random order.</li> +</ul> + +<h3>A. Sequential Reads</h3> +<table class="bn bnbase"> +<tr><td class="c1">LevelDB</td> + <td class="c2">4,030,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">1,010,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:95px"> </div></td> +<tr><td class="c1">SQLite3</td> + <td class="c2">383,000 ops/sec</td> + <td class="c3"><div class="bsql" style="width:33px"> </div></td> +</table> +<h3>B. Random Reads</h3> +<table class="bn bnbase"> +<tr><td class="c1">LevelDB</td> + <td class="c2">129,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:298px"> </div></td> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">151,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:350px"> </div></td> +<tr><td class="c1">SQLite3</td> + <td class="c2">134,000 ops/sec</td> + <td class="c3"><div class="bsql" style="width:310px"> </div></td> +</table> +<h3>C. Sequential Writes</h3> +<table class="bn bnbase"> +<tr><td class="c1">LevelDB</td> + <td class="c2">779,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">342,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:154px"> </div></td> +<tr><td class="c1">SQLite3</td> + <td class="c2">48,600 ops/sec</td> + <td class="c3"><div class="bsql" style="width:22px"> </div></td> +</table> +<h3>D. Random Writes</h3> +<table class="bn bnbase"> +<tr><td class="c1">LevelDB</td> + <td class="c2">164,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">88,500 ops/sec</td> + <td class="c3"><div class="bkct" style="width:188px"> </div></td> +<tr><td class="c1">SQLite3</td> + <td class="c2">9,860 ops/sec</td> + <td class="c3"><div class="bsql" style="width:21px"> </div></td> +</table> + +<p>LevelDB outperforms both SQLite3 and TreeDB in sequential and random write operations and sequential read operations. Kyoto Cabinet has the fastest random read operations.</p> + +<h2>2. Write Performance under Different Configurations</h2> +<h3>A. Large Values </h3> +<p>For this benchmark, we start with an empty database, and write 100,000 byte values (~50% compressible). To keep the benchmark running time reasonable, we stop after writing 1000 values.</p> +<h4>Sequential Writes</h4> +<table class="bn bnbase"> +<tr><td class="c1">LevelDB</td> + <td class="c2">1,100 ops/sec</td> + <td class="c3"><div class="bldb" style="width:234px"> </div></td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">1,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:224px"> </div></td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">1,600 ops/sec</td> + <td class="c3"><div class="bsql" style="width:350px"> </div></td></tr> +</table> +<h4>Random Writes</h4> +<table class="bn bnbase"> +<tr><td class="c1">LevelDB</td> + <td class="c2">480 ops/sec</td> + <td class="c3"><div class="bldb" style="width:105px"> </div></td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">1,100 ops/sec</td> + <td class="c3"><div class="bkct" style="width:240px"> </div></td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">1,600 ops/sec</td> + <td class="c3"><div class="bsql" style="width:350px"> </div></td></tr> +</table> +<p>LevelDB doesn't perform as well with large values of 100,000 bytes each. This is because LevelDB writes keys and values at least twice: first time to the transaction log, and second time (during a compaction) to a sorted file. +With larger values, LevelDB's per-operation efficiency is swamped by the +cost of extra copies of large values.</p> +<h3>B. Batch Writes</h3> +<p>A batch write is a set of writes that are applied atomically to the underlying database. A single batch of N writes may be significantly faster than N individual writes. The following benchmark writes one thousand batches where each batch contains one thousand 100-byte values. TreeDB does not support batch writes and is omitted from this benchmark.</p> +<h4>Sequential Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">840,000 entries/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(1.08x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">124,000 entries/sec</td> + <td class="c3"><div class="bsql" style="width:52px"> </div></td> + <td class="c4">(2.55x baseline)</td></tr> +</table> +<h4>Random Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">221,000 entries/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(1.35x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">22,000 entries/sec</td> + <td class="c3"><div class="bsql" style="width:34px"> </div></td> + <td class="c4">(2.23x baseline)</td></tr> +</table> + +<p>Because of the way LevelDB persistent storage is organized, batches of +random writes are not much slower (only a factor of 4x) than batches +of sequential writes.</p> + +<h3>C. Synchronous Writes</h3> +<p>In the following benchmark, we enable the synchronous writing modes +of all of the databases. Since this change significantly slows down the +benchmark, we stop after 10,000 writes. For synchronous write tests, we've +disabled hard drive write-caching (using `hdparm -W 0 [device]`).</p> +<ul> + <li>For LevelDB, we set WriteOptions.sync = true.</li> + <li>In TreeDB, we enabled TreeDB's OAUTOSYNC option.</li> + <li>For SQLite3, we set "PRAGMA synchronous = FULL".</li> +</ul> +<h4>Sequential Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">100 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(0.003x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">7 ops/sec</td> + <td class="c3"><div class="bkct" style="width:27px"> </div></td> + <td class="c4">(0.0004x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">88 ops/sec</td> + <td class="c3"><div class="bsql" style="width:315px"> </div></td> + <td class="c4">(0.002x baseline)</td></tr> +</table> +<h4>Random Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">100 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(0.015x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">8 ops/sec</td> + <td class="c3"><div class="bkct" style="width:29px"> </div></td> + <td class="c4">(0.001x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">88 ops/sec</td> + <td class="c3"><div class="bsql" style="width:314px"> </div></td> + <td class="c4">(0.009x baseline)</td></tr> +</table> + +<p>Also see the <code>ext4</code> performance numbers below +since synchronous writes behave significantly differently +on <code>ext3</code> and <code>ext4</code>.</p> + +<h3>D. Turning Compression Off</h3> + +<p>In the baseline measurements, LevelDB and TreeDB were using +light-weight compression +(<a href="http://code.google.com/p/snappy/">Snappy</a> for LevelDB, +and <a href="http://www.oberhumer.com/opensource/lzo/">LZO</a> for +TreeDB). SQLite3, by default does not use compression. The +experiments below show what happens when compression is disabled in +all of the databases (the SQLite3 numbers are just a copy of +its baseline measurements):</p> + +<h4>Sequential Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">594,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(0.76x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">485,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:239px"> </div></td> + <td class="c4">(1.42x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">48,600 ops/sec</td> + <td class="c3"><div class="bsql" style="width:29px"> </div></td> + <td class="c4">(1.00x baseline)</td></tr> +</table> +<h4>Random Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">135,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:296px"> </div></td> + <td class="c4">(0.82x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">159,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:350px"> </div></td> + <td class="c4">(1.80x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">9,860 ops/sec</td> + <td class="c3"><div class="bsql" style="width:22px"> </div></td> + <td class="c4">(1.00x baseline)</td></tr> +</table> + +<p>LevelDB's write performance is better with compression than without +since compression decreases the amount of data that has to be written +to disk. Therefore LevelDB users can leave compression enabled in +most scenarios without having worry about a tradeoff between space +usage and performance. TreeDB's performance on the other hand is +better without compression than with compression. Presumably this is +because TreeDB's compression library (LZO) is more expensive than +LevelDB's compression library (Snappy).<p> + +<h3>E. Using More Memory</h3> +<p>We increased the overall cache size for each database to 128 MB. For LevelDB, we partitioned 128 MB into a 120 MB write buffer and 8 MB of cache (up from 2 MB of write buffer and 2 MB of cache). For SQLite3, we kept the page size at 1024 bytes, but increased the number of pages to 131,072 (up from 4096). For TreeDB, we also kept the page size at 1024 bytes, but increased the cache size to 128 MB (up from 4 MB).</p> +<h4>Sequential Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">812,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(1.04x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">321,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:138px"> </div></td> + <td class="c4">(0.94x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">48,500 ops/sec</td> + <td class="c3"><div class="bsql" style="width:21px"> </div></td> + <td class="c4">(1.00x baseline)</td></tr> +</table> +<h4>Random Writes</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">355,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(2.16x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">284,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:280px"> </div></td> + <td class="c4">(3.21x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">9,670 ops/sec</td> + <td class="c3"><div class="bsql" style="width:10px"> </div></td> + <td class="c4">(0.98x baseline)</td></tr> +</table> + +<p>SQLite's performance does not change substantially when compared to +the baseline, but the random write performance for both LevelDB and +TreeDB increases significantly. LevelDB's performance improves +because a larger write buffer reduces the need to merge sorted files +(since it creates a smaller number of larger sorted files). TreeDB's +performance goes up because the entire database is available in memory +for fast in-place updates.</p> + + <h2>3. Read Performance under Different Configurations</h2> +<h3>A. Larger Caches</h3> +<p>We increased the overall memory usage to 128 MB for each database. +For LevelDB, we allocated 8 MB to LevelDB's write buffer and 120 MB +to LevelDB's cache. The other databases don't differentiate between a +write buffer and a cache, so we simply set their cache size to 128 +MB.</p> +<h4>Sequential Reads</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">5,210,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(1.29x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">1,070,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:72px"> </div></td> + <td class="c4">(1.06x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">609,000 ops/sec</td> + <td class="c3"><div class="bsql" style="width:41px"> </div></td> + <td class="c4">(1.59x baseline)</td></tr> +</table> + +<h4>Random Reads</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">190,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:144px"> </div></td> + <td class="c4">(1.47x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">463,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:350px"> </div></td> + <td class="c4">(3.07x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">186,000 ops/sec</td> + <td class="c3"><div class="bsql" style="width:141px"> </div></td> + <td class="c4">(1.39x baseline)</td></tr> +</table> + +<p>As expected, the read performance of all of the databases increases +when the caches are enlarged. In particular, TreeDB seems to make +very effective use of a cache that is large enough to hold the entire +database.</p> + +<h3>B. No Compression Reads </h3> +<p>For this benchmark, we populated a database with 1 million entries consisting of 16 byte keys and 100 byte values. We compiled LevelDB and Kyoto Cabinet without compression support, so results that are read out from the database are already uncompressed. We've listed the SQLite3 baseline read performance as a point of comparison.</p> +<h4>Sequential Reads</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">4,880,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:350px"> </div></td> + <td class="c4">(1.21x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">1,230,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:88px"> </div></td> + <td class="c4">(3.60x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">383,000 ops/sec</td> + <td class="c3"><div class="bsql" style="width:27px"> </div></td> + <td class="c4">(1.00x baseline)</td></tr> +</table> +<h4>Random Reads</h4> +<table class="bn"> +<tr><td class="c1">LevelDB</td> + <td class="c2">149,000 ops/sec</td> + <td class="c3"><div class="bldb" style="width:300px"> </div></td> + <td class="c4">(1.16x baseline)</td></tr> +<tr><td class="c1">Kyoto TreeDB</td> + <td class="c2">175,000 ops/sec</td> + <td class="c3"><div class="bkct" style="width:350px"> </div></td> + <td class="c4">(1.16x baseline)</td></tr> +<tr><td class="c1">SQLite3</td> + <td class="c2">134,000 ops/sec</td> + <td class="c3"><div class="bsql" style="width:268px"> </div></td> + <td class="c4">(1.00x baseline)</td></tr> +</table> + +<p>Performance of both LevelDB and TreeDB improves a small amount when +compression is disabled. Note however that under different workloads, +performance may very well be better with compression if it allows more +of the working set to fit in memory.</p> + +<h2>Note about Ext4 Filesystems</h2> +<p>The preceding numbers are for an ext3 file system. Synchronous writes are much slower under <a href="https://en.wikipedia.org/wiki/Ext4">ext4</a> (LevelDB drops to ~31 writes / second and TreeDB drops to ~5 writes / second; SQLite3's synchronous writes do not noticeably drop) due to ext4's different handling of <span class="code">fsync</span> / <span class="code">msync</span> calls. Even LevelDB's asynchronous write performance drops somewhat since it spreads its storage across multiple files and issues <span class="code">fsync</span> calls when switching to a new file.</p> + +<h2>Acknowledgements</h2> +<p>Jeff Dean and Sanjay Ghemawat wrote LevelDB. Kevin Tseng wrote and compiled these benchmarks. Mikio Hirabayashi, Scott Hess, and Gabor Cselle provided help and advice.</p> +</body> +</html> diff --git a/doc/impl.md b/doc/impl.md new file mode 100644 index 0000000..c9bb621 --- /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)](https://research.google/pubs/pub27898/). +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 + +`RemoveObsoleteFiles()` 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.md b/doc/index.md new file mode 100644 index 0000000..01693ad --- /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 <cassert> +#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/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<Slice> trimmed(n); + for (int i = 0; i < n; i++) { + trimmed[i] = RemoveTrailingSpaces(keys[i]); + } + return builtin_policy_->CreateFilter(trimmed.data(), 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]; +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/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 +=================== + + <beginning_of_file> + [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)) + <end_of_file> + +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.<N>` to the BlockHandle for the filter +block where `<N>` 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 new file mode 100644 index 0000000..e476613 --- /dev/null +++ b/helpers/memenv/memenv.cc @@ -0,0 +1,390 @@ +// 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 "helpers/memenv/memenv.h" + +#include <cstring> +#include <limits> +#include <map> +#include <string> +#include <vector> + +#include "leveldb/env.h" +#include "leveldb/status.h" +#include "port/port.h" +#include "port/thread_annotations.h" +#include "util/mutexlock.h" + +namespace leveldb { + +namespace { + +class FileState { + public: + // FileStates are reference counted. The initial reference count is zero + // 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_); + ++refs_; + } + + // Decrease the reference count. Delete if this is the last reference. + void Unref() { + bool do_delete = false; + + { + MutexLock lock(&refs_mutex_); + --refs_; + assert(refs_ >= 0); + if (refs_ <= 0) { + do_delete = true; + } + } + + if (do_delete) { + delete this; + } + } + + 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."); + } + const uint64_t available = size_ - offset; + if (n > available) { + n = static_cast<size_t>(available); + } + if (n == 0) { + *result = Slice(); + return Status::OK(); + } + + assert(offset / kBlockSize <= std::numeric_limits<size_t>::max()); + size_t block = static_cast<size_t>(offset / kBlockSize); + size_t block_offset = offset % kBlockSize; + size_t bytes_to_copy = n; + char* dst = scratch; + + while (bytes_to_copy > 0) { + size_t avail = kBlockSize - block_offset; + if (avail > bytes_to_copy) { + avail = bytes_to_copy; + } + std::memcpy(dst, blocks_[block] + block_offset, avail); + + bytes_to_copy -= avail; + dst += avail; + block++; + block_offset = 0; + } + + *result = Slice(scratch, n); + return Status::OK(); + } + + Status Append(const Slice& data) { + 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; + + if (offset != 0) { + // There is some room in the last block. + avail = kBlockSize - offset; + } else { + // No room in the last block; push new one. + blocks_.push_back(new char[kBlockSize]); + avail = kBlockSize; + } + + if (avail > src_len) { + avail = src_len; + } + std::memcpy(blocks_.back() + offset, src, avail); + src_len -= avail; + src += avail; + size_ += avail; + } + + return Status::OK(); + } + + private: + enum { kBlockSize = 8 * 1024 }; + + // Private since only Unref() should be used to delete it. + ~FileState() { Truncate(); } + + port::Mutex refs_mutex_; + int refs_ GUARDED_BY(refs_mutex_); + + mutable port::Mutex blocks_mutex_; + std::vector<char*> blocks_ GUARDED_BY(blocks_mutex_); + uint64_t size_ GUARDED_BY(blocks_mutex_); +}; + +class SequentialFileImpl : public SequentialFile { + public: + explicit SequentialFileImpl(FileState* file) : file_(file), pos_(0) { + file_->Ref(); + } + + ~SequentialFileImpl() override { file_->Unref(); } + + Status Read(size_t n, Slice* result, char* scratch) override { + Status s = file_->Read(pos_, n, result, scratch); + if (s.ok()) { + pos_ += result->size(); + } + return s; + } + + Status Skip(uint64_t n) override { + if (pos_ > file_->Size()) { + return Status::IOError("pos_ > file_->Size()"); + } + const uint64_t available = file_->Size() - pos_; + if (n > available) { + n = available; + } + pos_ += n; + return Status::OK(); + } + + private: + FileState* file_; + uint64_t pos_; +}; + +class RandomAccessFileImpl : public RandomAccessFile { + public: + explicit RandomAccessFileImpl(FileState* file) : file_(file) { file_->Ref(); } + + ~RandomAccessFileImpl() override { file_->Unref(); } + + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + return file_->Read(offset, n, result, scratch); + } + + private: + FileState* file_; +}; + +class WritableFileImpl : public WritableFile { + public: + WritableFileImpl(FileState* file) : file_(file) { file_->Ref(); } + + ~WritableFileImpl() override { file_->Unref(); } + + Status Append(const Slice& data) override { return file_->Append(data); } + + Status Close() override { return Status::OK(); } + Status Flush() override { return Status::OK(); } + Status Sync() override { return Status::OK(); } + + private: + FileState* file_; +}; + +class NoOpLogger : public Logger { + public: + void Logv(const char* format, std::va_list ap) override {} +}; + +class InMemoryEnv : public EnvWrapper { + public: + explicit InMemoryEnv(Env* base_env) : EnvWrapper(base_env) {} + + ~InMemoryEnv() override { + for (const auto& kvp : file_map_) { + kvp.second->Unref(); + } + } + + // Partial implementation of the Env interface. + Status NewSequentialFile(const std::string& fname, + SequentialFile** result) override { + MutexLock lock(&mutex_); + if (file_map_.find(fname) == file_map_.end()) { + *result = nullptr; + return Status::IOError(fname, "File not found"); + } + + *result = new SequentialFileImpl(file_map_[fname]); + return Status::OK(); + } + + Status NewRandomAccessFile(const std::string& fname, + RandomAccessFile** result) override { + MutexLock lock(&mutex_); + if (file_map_.find(fname) == file_map_.end()) { + *result = nullptr; + return Status::IOError(fname, "File not found"); + } + + *result = new RandomAccessFileImpl(file_map_[fname]); + return Status::OK(); + } + + Status NewWritableFile(const std::string& fname, + WritableFile** result) override { + MutexLock lock(&mutex_); + FileSystem::iterator it = file_map_.find(fname); + + 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(); + } + + 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(); + } + + Status GetChildren(const std::string& dir, + std::vector<std::string>* result) override { + MutexLock lock(&mutex_); + result->clear(); + + 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))) { + result->push_back(filename.substr(dir.size() + 1)); + } + } + + return Status::OK(); + } + + void RemoveFileInternal(const std::string& fname) + EXCLUSIVE_LOCKS_REQUIRED(mutex_) { + if (file_map_.find(fname) == file_map_.end()) { + return; + } + + file_map_[fname]->Unref(); + file_map_.erase(fname); + } + + Status RemoveFile(const std::string& fname) override { + MutexLock lock(&mutex_); + if (file_map_.find(fname) == file_map_.end()) { + return Status::IOError(fname, "File not found"); + } + + RemoveFileInternal(fname); + return Status::OK(); + } + + Status CreateDir(const std::string& dirname) override { return Status::OK(); } + + Status RemoveDir(const std::string& dirname) override { return Status::OK(); } + + 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"); + } + + *file_size = file_map_[fname]->Size(); + return Status::OK(); + } + + 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"); + } + + RemoveFileInternal(target); + file_map_[target] = file_map_[src]; + file_map_.erase(src); + return Status::OK(); + } + + Status LockFile(const std::string& fname, FileLock** lock) override { + *lock = new FileLock; + return Status::OK(); + } + + Status UnlockFile(FileLock* lock) override { + delete lock; + return Status::OK(); + } + + Status GetTestDirectory(std::string* path) override { + *path = "/test"; + return Status::OK(); + } + + Status NewLogger(const std::string& fname, Logger** result) override { + *result = new NoOpLogger; + return Status::OK(); + } + + private: + // Map from filenames to FileState objects, representing a simple file system. + typedef std::map<std::string, FileState*> FileSystem; + + port::Mutex mutex_; + FileSystem file_map_ GUARDED_BY(mutex_); +}; + +} // namespace + +Env* NewMemEnv(Env* base_env) { return new InMemoryEnv(base_env); } + +} // namespace leveldb diff --git a/helpers/memenv/memenv.h b/helpers/memenv/memenv.h new file mode 100644 index 0000000..3d929e4 --- /dev/null +++ b/helpers/memenv/memenv.h @@ -0,0 +1,22 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_ +#define STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_ + +#include "leveldb/export.h" + +namespace leveldb { + +class Env; + +// Returns a new environment that stores its data in memory and delegates +// all non-file-storage tasks to base_env. The caller must delete the result +// when it is no longer needed. +// *base_env must remain live while the result is in use. +LEVELDB_EXPORT Env* NewMemEnv(Env* base_env); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_ diff --git a/helpers/memenv/memenv_test.cc b/helpers/memenv/memenv_test.cc new file mode 100644 index 0000000..3f03cb6 --- /dev/null +++ b/helpers/memenv/memenv_test.cc @@ -0,0 +1,264 @@ +// 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 "helpers/memenv/memenv.h" + +#include <string> +#include <vector> + +#include "gtest/gtest.h" +#include "db/db_impl.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "util/testutil.h" + +namespace leveldb { + +class MemEnvTest : public testing::Test { + public: + MemEnvTest() : env_(NewMemEnv(Env::Default())) {} + ~MemEnvTest() { delete env_; } + + Env* env_; +}; + +TEST_F(MemEnvTest, Basics) { + uint64_t file_size; + WritableFile* writable_file; + std::vector<std::string> children; + + ASSERT_LEVELDB_OK(env_->CreateDir("/dir")); + + // Check that the directory is empty. + ASSERT_TRUE(!env_->FileExists("/dir/non_existent")); + ASSERT_TRUE(!env_->GetFileSize("/dir/non_existent", &file_size).ok()); + ASSERT_LEVELDB_OK(env_->GetChildren("/dir", &children)); + ASSERT_EQ(0, children.size()); + + // Create a file. + ASSERT_LEVELDB_OK(env_->NewWritableFile("/dir/f", &writable_file)); + ASSERT_LEVELDB_OK(env_->GetFileSize("/dir/f", &file_size)); + ASSERT_EQ(0, file_size); + delete writable_file; + + // Check that the file exists. + ASSERT_TRUE(env_->FileExists("/dir/f")); + ASSERT_LEVELDB_OK(env_->GetFileSize("/dir/f", &file_size)); + ASSERT_EQ(0, file_size); + ASSERT_LEVELDB_OK(env_->GetChildren("/dir", &children)); + ASSERT_EQ(1, children.size()); + ASSERT_EQ("f", children[0]); + + // Write to the file. + ASSERT_LEVELDB_OK(env_->NewWritableFile("/dir/f", &writable_file)); + ASSERT_LEVELDB_OK(writable_file->Append("abc")); + delete writable_file; + + // Check that append works. + ASSERT_LEVELDB_OK(env_->NewAppendableFile("/dir/f", &writable_file)); + ASSERT_LEVELDB_OK(env_->GetFileSize("/dir/f", &file_size)); + ASSERT_EQ(3, file_size); + ASSERT_LEVELDB_OK(writable_file->Append("hello")); + delete writable_file; + + // Check for expected size. + ASSERT_LEVELDB_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()); + ASSERT_LEVELDB_OK(env_->RenameFile("/dir/f", "/dir/g")); + ASSERT_TRUE(!env_->FileExists("/dir/f")); + ASSERT_TRUE(env_->FileExists("/dir/g")); + ASSERT_LEVELDB_OK(env_->GetFileSize("/dir/g", &file_size)); + ASSERT_EQ(8, file_size); + + // Check that opening non-existent file fails. + SequentialFile* seq_file; + RandomAccessFile* rand_file; + ASSERT_TRUE(!env_->NewSequentialFile("/dir/non_existent", &seq_file).ok()); + ASSERT_TRUE(!seq_file); + ASSERT_TRUE(!env_->NewRandomAccessFile("/dir/non_existent", &rand_file).ok()); + ASSERT_TRUE(!rand_file); + + // Check that deleting works. + ASSERT_TRUE(!env_->RemoveFile("/dir/non_existent").ok()); + ASSERT_LEVELDB_OK(env_->RemoveFile("/dir/g")); + ASSERT_TRUE(!env_->FileExists("/dir/g")); + ASSERT_LEVELDB_OK(env_->GetChildren("/dir", &children)); + ASSERT_EQ(0, children.size()); + ASSERT_LEVELDB_OK(env_->RemoveDir("/dir")); +} + +TEST_F(MemEnvTest, ReadWrite) { + WritableFile* writable_file; + SequentialFile* seq_file; + RandomAccessFile* rand_file; + Slice result; + char scratch[100]; + + ASSERT_LEVELDB_OK(env_->CreateDir("/dir")); + + ASSERT_LEVELDB_OK(env_->NewWritableFile("/dir/f", &writable_file)); + ASSERT_LEVELDB_OK(writable_file->Append("hello ")); + ASSERT_LEVELDB_OK(writable_file->Append("world")); + delete writable_file; + + // Read sequentially. + ASSERT_LEVELDB_OK(env_->NewSequentialFile("/dir/f", &seq_file)); + ASSERT_LEVELDB_OK(seq_file->Read(5, &result, scratch)); // Read "hello". + ASSERT_EQ(0, result.compare("hello")); + ASSERT_LEVELDB_OK(seq_file->Skip(1)); + ASSERT_LEVELDB_OK(seq_file->Read(1000, &result, scratch)); // Read "world". + ASSERT_EQ(0, result.compare("world")); + ASSERT_LEVELDB_OK( + seq_file->Read(1000, &result, scratch)); // Try reading past EOF. + ASSERT_EQ(0, result.size()); + ASSERT_LEVELDB_OK(seq_file->Skip(100)); // Try to skip past end of file. + ASSERT_LEVELDB_OK(seq_file->Read(1000, &result, scratch)); + ASSERT_EQ(0, result.size()); + delete seq_file; + + // Random reads. + ASSERT_LEVELDB_OK(env_->NewRandomAccessFile("/dir/f", &rand_file)); + ASSERT_LEVELDB_OK(rand_file->Read(6, 5, &result, scratch)); // Read "world". + ASSERT_EQ(0, result.compare("world")); + ASSERT_LEVELDB_OK(rand_file->Read(0, 5, &result, scratch)); // Read "hello". + ASSERT_EQ(0, result.compare("hello")); + ASSERT_LEVELDB_OK(rand_file->Read(10, 100, &result, scratch)); // Read "d". + ASSERT_EQ(0, result.compare("d")); + + // Too high offset. + ASSERT_TRUE(!rand_file->Read(1000, 5, &result, scratch).ok()); + delete rand_file; +} + +TEST_F(MemEnvTest, Locks) { + FileLock* lock; + + // These are no-ops, but we test they return success. + ASSERT_LEVELDB_OK(env_->LockFile("some file", &lock)); + ASSERT_LEVELDB_OK(env_->UnlockFile(lock)); +} + +TEST_F(MemEnvTest, Misc) { + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + ASSERT_TRUE(!test_dir.empty()); + + WritableFile* writable_file; + ASSERT_LEVELDB_OK(env_->NewWritableFile("/a/b", &writable_file)); + + // These are no-ops, but we test they return success. + ASSERT_LEVELDB_OK(writable_file->Sync()); + ASSERT_LEVELDB_OK(writable_file->Flush()); + ASSERT_LEVELDB_OK(writable_file->Close()); + delete writable_file; +} + +TEST_F(MemEnvTest, LargeWrite) { + const size_t kWriteSize = 300 * 1024; + char* scratch = new char[kWriteSize * 2]; + + std::string write_data; + for (size_t i = 0; i < kWriteSize; ++i) { + write_data.append(1, static_cast<char>(i)); + } + + WritableFile* writable_file; + ASSERT_LEVELDB_OK(env_->NewWritableFile("/dir/f", &writable_file)); + ASSERT_LEVELDB_OK(writable_file->Append("foo")); + ASSERT_LEVELDB_OK(writable_file->Append(write_data)); + delete writable_file; + + SequentialFile* seq_file; + Slice result; + ASSERT_LEVELDB_OK(env_->NewSequentialFile("/dir/f", &seq_file)); + ASSERT_LEVELDB_OK(seq_file->Read(3, &result, scratch)); // Read "foo". + ASSERT_EQ(0, result.compare("foo")); + + size_t read = 0; + std::string read_data; + while (read < kWriteSize) { + ASSERT_LEVELDB_OK(seq_file->Read(kWriteSize - read, &result, scratch)); + read_data.append(result.data(), result.size()); + read += result.size(); + } + ASSERT_TRUE(write_data == read_data); + delete seq_file; + delete[] scratch; +} + +TEST_F(MemEnvTest, OverwriteOpenFile) { + const char kWrite1Data[] = "Write #1 data"; + const size_t kFileDataLen = sizeof(kWrite1Data) - 1; + const std::string kTestFileName = testing::TempDir() + "leveldb-TestFile.dat"; + + ASSERT_LEVELDB_OK(WriteStringToFile(env_, kWrite1Data, kTestFileName)); + + RandomAccessFile* rand_file; + ASSERT_LEVELDB_OK(env_->NewRandomAccessFile(kTestFileName, &rand_file)); + + const char kWrite2Data[] = "Write #2 data"; + ASSERT_LEVELDB_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_LEVELDB_OK(rand_file->Read(0, kFileDataLen, &result, scratch)); + ASSERT_EQ(0, result.compare(kWrite2Data)); + + delete rand_file; +} + +TEST_F(MemEnvTest, DBTest) { + Options options; + options.create_if_missing = true; + options.env = env_; + DB* db; + + const Slice keys[] = {Slice("aaa"), Slice("bbb"), Slice("ccc")}; + const Slice vals[] = {Slice("foo"), Slice("bar"), Slice("baz")}; + + ASSERT_LEVELDB_OK(DB::Open(options, "/dir/db", &db)); + for (size_t i = 0; i < 3; ++i) { + ASSERT_LEVELDB_OK(db->Put(WriteOptions(), keys[i], vals[i])); + } + + for (size_t i = 0; i < 3; ++i) { + std::string res; + ASSERT_LEVELDB_OK(db->Get(ReadOptions(), keys[i], &res)); + ASSERT_TRUE(res == vals[i]); + } + + Iterator* iterator = db->NewIterator(ReadOptions()); + iterator->SeekToFirst(); + for (size_t i = 0; i < 3; ++i) { + ASSERT_TRUE(iterator->Valid()); + ASSERT_TRUE(keys[i] == iterator->key()); + ASSERT_TRUE(vals[i] == iterator->value()); + iterator->Next(); + } + ASSERT_TRUE(!iterator->Valid()); + delete iterator; + + DBImpl* dbi = reinterpret_cast<DBImpl*>(db); + ASSERT_LEVELDB_OK(dbi->TEST_CompactMemTable()); + + for (size_t i = 0; i < 3; ++i) { + std::string res; + ASSERT_LEVELDB_OK(db->Get(ReadOptions(), keys[i], &res)); + ASSERT_TRUE(res == vals[i]); + } + + delete db; +} + +} // namespace leveldb + +int main(int argc, char** argv) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/include/leveldb/c.h b/include/leveldb/c.h new file mode 100644 index 0000000..02c79ba --- /dev/null +++ b/include/leveldb/c.h @@ -0,0 +1,270 @@ +/* Copyright (c) 2011 The LevelDB Authors. All rights reserved. + Use of this source code is governed by a BSD-style license that can be + found in the LICENSE file. See the AUTHORS file for names of contributors. + + C bindings for leveldb. May be useful as a stable ABI that can be + used by programs that keep leveldb in a shared library, or for + a JNI api. + + Does not support: + . getters for the option types + . custom comparators that implement key shortening + . custom iter, db, env, cache implementations using just the C bindings + + Some conventions: + + (1) We expose just opaque struct pointers and functions to clients. + This allows us to change internal representations without having to + recompile clients. + + (2) For simplicity, there is no equivalent to the Slice type. Instead, + the caller has to pass the pointer and length as separate + arguments. + + (3) Errors are represented by a null-terminated c string. NULL + means no error. All operations that can raise an error are passed + a "char** errptr" as the last argument. One of the following must + be true on entry: + *errptr == NULL + *errptr points to a malloc()ed null-terminated error message + (On Windows, *errptr must have been malloc()-ed by this library.) + On success, a leveldb routine leaves *errptr unchanged. + On failure, leveldb frees the old value of *errptr and + set *errptr to a malloc()ed error message. + + (4) Bools have the type uint8_t (0 == false; rest == true) + + (5) All of the pointer arguments must be non-NULL. +*/ + +#ifndef STORAGE_LEVELDB_INCLUDE_C_H_ +#define STORAGE_LEVELDB_INCLUDE_C_H_ + +#ifdef __cplusplus +extern "C" { +#endif + +#include <stdarg.h> +#include <stddef.h> +#include <stdint.h> + +#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; + +/* DB operations */ + +LEVELDB_EXPORT leveldb_t* leveldb_open(const leveldb_options_t* options, + const char* name, char** errptr); + +LEVELDB_EXPORT void leveldb_close(leveldb_t* db); + +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); + +LEVELDB_EXPORT void leveldb_delete(leveldb_t* db, + const leveldb_writeoptions_t* options, + const char* key, size_t keylen, + 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. */ +LEVELDB_EXPORT char* leveldb_get(leveldb_t* db, + const leveldb_readoptions_t* options, + const char* key, size_t keylen, size_t* vallen, + char** errptr); + +LEVELDB_EXPORT leveldb_iterator_t* leveldb_create_iterator( + leveldb_t* db, const leveldb_readoptions_t* options); + +LEVELDB_EXPORT const leveldb_snapshot_t* leveldb_create_snapshot(leveldb_t* db); + +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. */ +LEVELDB_EXPORT char* leveldb_property_value(leveldb_t* db, + const char* propname); + +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); + +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 */ + +LEVELDB_EXPORT void leveldb_destroy_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 */ + +LEVELDB_EXPORT void leveldb_iter_destroy(leveldb_iterator_t*); +LEVELDB_EXPORT uint8_t 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 */ + +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 */ + +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*, + uint8_t); +LEVELDB_EXPORT void leveldb_options_set_error_if_exists(leveldb_options_t*, + uint8_t); +LEVELDB_EXPORT void leveldb_options_set_paranoid_checks(leveldb_options_t*, + uint8_t); +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 }; +LEVELDB_EXPORT void leveldb_options_set_compression(leveldb_options_t*, int); + +/* Comparator */ + +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*)); +LEVELDB_EXPORT void leveldb_comparator_destroy(leveldb_comparator_t*); + +/* Filter policy */ + +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), + uint8_t (*key_may_match)(void*, const char* key, size_t length, + const char* filter, size_t filter_length), + const char* (*name)(void*)); +LEVELDB_EXPORT void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t*); + +LEVELDB_EXPORT leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom( + int bits_per_key); + +/* Read options */ + +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*, uint8_t); +LEVELDB_EXPORT void leveldb_readoptions_set_fill_cache(leveldb_readoptions_t*, + uint8_t); +LEVELDB_EXPORT void leveldb_readoptions_set_snapshot(leveldb_readoptions_t*, + const leveldb_snapshot_t*); + +/* Write options */ + +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*, + uint8_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 */ + +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 */ + +/* Calls free(ptr). + REQUIRES: ptr was malloc()-ed and returned by one of the routines + 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. */ +LEVELDB_EXPORT void leveldb_free(void* ptr); + +/* Return the major version number for this release. */ +LEVELDB_EXPORT int leveldb_major_version(void); + +/* Return the minor version number for this release. */ +LEVELDB_EXPORT int leveldb_minor_version(void); + +#ifdef __cplusplus +} /* end extern "C" */ +#endif + +#endif /* STORAGE_LEVELDB_INCLUDE_C_H_ */ diff --git a/include/leveldb/cache.h b/include/leveldb/cache.h new file mode 100644 index 0000000..98c95ac --- /dev/null +++ b/include/leveldb/cache.h @@ -0,0 +1,111 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// A Cache is an interface that maps keys to values. It has internal +// synchronization and may be safely accessed concurrently from +// multiple threads. It may automatically evict entries to make room +// for new entries. Values have a specified charge against the cache +// capacity. For example, a cache where the values are variable +// length strings, may use the length of the string as the charge for +// the string. +// +// A builtin cache implementation with a least-recently-used eviction +// policy is provided. Clients may use their own implementations if +// they want something more sophisticated (like scan-resistance, a +// custom eviction policy, variable cache sizing, etc.) + +#ifndef STORAGE_LEVELDB_INCLUDE_CACHE_H_ +#define STORAGE_LEVELDB_INCLUDE_CACHE_H_ + +#include <cstdint> + +#include "leveldb/export.h" +#include "leveldb/slice.h" + +namespace leveldb { + +class LEVELDB_EXPORT Cache; + +// Create a new cache with a fixed size capacity. This implementation +// of Cache uses a least-recently-used eviction policy. +LEVELDB_EXPORT Cache* NewLRUCache(size_t capacity); + +class LEVELDB_EXPORT Cache { + public: + 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 {}; + + // Insert a mapping from key->value into the cache and assign it + // the specified charge against the total cache capacity. + // + // Returns a handle that corresponds to the mapping. The caller + // must call this->Release(handle) when the returned mapping is no + // longer needed. + // + // When the inserted entry is no longer needed, the key and + // value will be passed to "deleter". + virtual Handle* Insert(const Slice& key, void* value, size_t charge, + void (*deleter)(const Slice& key, void* value)) = 0; + + // If the cache has no mapping for "key", returns nullptr. + // + // Else return a handle that corresponds to the mapping. The caller + // must call this->Release(handle) when the returned mapping is no + // longer needed. + virtual Handle* Lookup(const Slice& key) = 0; + + // Release a mapping returned by a previous Lookup(). + // REQUIRES: handle must not have been released yet. + // REQUIRES: handle must have been returned by a method on *this. + virtual void Release(Handle* handle) = 0; + + // Return the value encapsulated in a handle returned by a + // successful Lookup(). + // REQUIRES: handle must not have been released yet. + // REQUIRES: handle must have been returned by a method on *this. + virtual void* Value(Handle* handle) = 0; + + // If the cache contains entry for key, erase it. Note that the + // underlying entry will be kept around until all existing handles + // to it have been released. + virtual void Erase(const Slice& key) = 0; + + // Return a new numeric id. May be used by multiple clients who are + // sharing the same cache to partition the key space. Typically the + // client will allocate a new id at startup and prepend the id to + // its cache keys. + virtual uint64_t NewId() = 0; + + // 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); + void Unref(Handle* e); + + struct Rep; + Rep* rep_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_CACHE_H_ diff --git a/include/leveldb/comparator.h b/include/leveldb/comparator.h new file mode 100644 index 0000000..a85b51e --- /dev/null +++ b/include/leveldb/comparator.h @@ -0,0 +1,64 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_ +#define STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_ + +#include <string> + +#include "leveldb/export.h" + +namespace leveldb { + +class Slice; + +// A Comparator object provides a total order across slices that are +// used as keys in an sstable or a database. A Comparator implementation +// must be thread-safe since leveldb may invoke its methods concurrently +// from multiple threads. +class LEVELDB_EXPORT Comparator { + public: + virtual ~Comparator(); + + // Three-way comparison. Returns value: + // < 0 iff "a" < "b", + // == 0 iff "a" == "b", + // > 0 iff "a" > "b" + virtual int Compare(const Slice& a, const Slice& b) const = 0; + + // The name of the comparator. Used to check for comparator + // mismatches (i.e., a DB created with one comparator is + // accessed using a different comparator. + // + // The client of this package should switch to a new name whenever + // the comparator implementation changes in a way that will cause + // the relative ordering of any two keys to change. + // + // Names starting with "leveldb." are reserved and should not be used + // by any clients of this package. + virtual const char* Name() const = 0; + + // Advanced functions: these are used to reduce the space requirements + // for internal data structures like index blocks. + + // If *start < limit, changes *start to a short string in [start,limit). + // Simple comparator implementations may return with *start unchanged, + // i.e., an implementation of this method that does nothing is correct. + virtual void FindShortestSeparator(std::string* start, + const Slice& limit) const = 0; + + // Changes *key to a short string >= *key. + // Simple comparator implementations may return with *key unchanged, + // i.e., an implementation of this method that does nothing is correct. + virtual void FindShortSuccessor(std::string* key) const = 0; +}; + +// Return a builtin comparator that uses lexicographic byte-wise +// ordering. The result remains the property of this module and +// must not be deleted. +LEVELDB_EXPORT const Comparator* BytewiseComparator(); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_ diff --git a/include/leveldb/db.h b/include/leveldb/db.h new file mode 100644 index 0000000..a13d147 --- /dev/null +++ b/include/leveldb/db.h @@ -0,0 +1,167 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_INCLUDE_DB_H_ +#define STORAGE_LEVELDB_INCLUDE_DB_H_ + +#include <cstdint> +#include <cstdio> + +#include "leveldb/export.h" +#include "leveldb/iterator.h" +#include "leveldb/options.h" + +namespace leveldb { + +// Update CMakeLists.txt if you change these +static const int kMajorVersion = 1; +static const int kMinorVersion = 23; + +struct Options; +struct ReadOptions; +struct WriteOptions; +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 LEVELDB_EXPORT Snapshot { + protected: + virtual ~Snapshot(); +}; + +// A range of keys +struct LEVELDB_EXPORT Range { + Range() = default; + 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 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 nullptr in *dbptr and returns a non-OK status on error. + // Caller should delete *dbptr when it is no longer needed. + static Status Open(const Options& options, const std::string& name, + DB** dbptr); + + 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, + const Slice& value) = 0; + + // Remove the database entry (if any) for "key". Returns OK on + // success, and a non-OK status on error. It is not an error if "key" + // did not exist in the database. + // Note: consider setting options.sync = true. + virtual Status Delete(const WriteOptions& options, const Slice& key) = 0; + + // Apply the specified updates to the database. + // Returns OK on success, non-OK on failure. + // Note: consider setting options.sync = true. + virtual Status Write(const WriteOptions& options, WriteBatch* updates) = 0; + + // If the database contains an entry for "key" store the + // corresponding value in *value and return OK. + // + // If there is no entry for "key" leave *value unchanged and return + // a status for which Status::IsNotFound() returns true. + // + // May return some other Status on an error. + virtual Status Get(const ReadOptions& options, 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 + // call one of the Seek methods on the iterator before using it). + // + // Caller should delete the iterator when it is no longer needed. + // The returned iterator should be deleted before this db is deleted. + virtual Iterator* NewIterator(const ReadOptions& options) = 0; + + // Return a handle to the current DB state. Iterators created with + // this handle will all observe a stable snapshot of the current DB + // state. The caller must call ReleaseSnapshot(result) when the + // snapshot is no longer needed. + virtual const Snapshot* GetSnapshot() = 0; + + // Release a previously acquired snapshot. The caller must not + // use "snapshot" after this call. + virtual void ReleaseSnapshot(const Snapshot* snapshot) = 0; + + // DB implementations can export properties about their state + // via this method. If "property" is a valid property understood by this + // DB implementation, fills "*value" with its current value and returns + // true. Otherwise returns false. + // + // + // Valid property names include: + // + // "leveldb.num-files-at-level<N>" - return the number of files at level <N>, + // where <N> is an ASCII representation of a level number (e.g. "0"). + // "leveldb.stats" - returns a multi-line string that describes statistics + // 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 + // file system space used by keys in "[range[i].start .. range[i].limit)". + // + // Note that the returned sizes measure file system space usage, so + // if the user data compresses by a factor of ten, the returned + // sizes will be one-tenth the size of the corresponding user data size. + // + // The results may not include the sizes of recently written data. + virtual void GetApproximateSizes(const Range* range, int n, + uint64_t* sizes) = 0; + + // Compact the underlying storage for the key range [*begin,*end]. + // In particular, deleted and overwritten versions are discarded, + // and the data is rearranged to reduce the cost of operations + // needed to access the data. This operation should typically only + // be invoked by users who understand the underlying implementation. + // + // begin==nullptr is treated as a key before all keys in the database. + // end==nullptr is treated as a key after all keys in the database. + // Therefore the following call will compact the entire database: + // db->CompactRange(nullptr, nullptr); + virtual void CompactRange(const Slice* begin, const Slice* end) = 0; +}; + +// Destroy the contents of the specified database. +// Be very careful using this method. +// +// 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. +LEVELDB_EXPORT Status RepairDB(const std::string& dbname, + const Options& options); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_DB_H_ diff --git a/include/leveldb/dumpfile.h b/include/leveldb/dumpfile.h new file mode 100644 index 0000000..a58bc6b --- /dev/null +++ b/include/leveldb/dumpfile.h @@ -0,0 +1,28 @@ +// 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. + +#ifndef STORAGE_LEVELDB_INCLUDE_DUMPFILE_H_ +#define STORAGE_LEVELDB_INCLUDE_DUMPFILE_H_ + +#include <string> + +#include "leveldb/env.h" +#include "leveldb/export.h" +#include "leveldb/status.h" + +namespace leveldb { + +// Dump the contents of the file named by fname in text format to +// *dst. Makes a sequence of dst->Append() calls; each call is passed +// the newline-terminated text corresponding to a single item found +// in the file. +// +// Returns a non-OK result if fname does not name a leveldb storage +// file, or if the file cannot be read. +LEVELDB_EXPORT Status DumpFile(Env* env, const std::string& fname, + WritableFile* dst); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_DUMPFILE_H_ diff --git a/include/leveldb/env.h b/include/leveldb/env.h new file mode 100644 index 0000000..e00895a --- /dev/null +++ b/include/leveldb/env.h @@ -0,0 +1,417 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// An Env is an interface used by the leveldb implementation to access +// operating system functionality like the filesystem etc. Callers +// may wish to provide a custom Env object when opening a database to +// get fine gain control; e.g., to rate limit file system operations. +// +// All Env implementations are safe for concurrent access from +// multiple threads without any external synchronization. + +#ifndef STORAGE_LEVELDB_INCLUDE_ENV_H_ +#define STORAGE_LEVELDB_INCLUDE_ENV_H_ + +#include <cstdarg> +#include <cstdint> +#include <string> +#include <vector> + +#include "leveldb/export.h" +#include "leveldb/status.h" + +// This workaround can be removed when leveldb::Env::DeleteFile is removed. +#if defined(_WIN32) +// On Windows, the method name DeleteFile (below) introduces the risk of +// triggering undefined behavior by exposing the compiler to different +// declarations of the Env class in different translation units. +// +// This is because <windows.h>, a fairly popular header file for Windows +// applications, defines a DeleteFile macro. So, files that include the Windows +// header before this header will contain an altered Env declaration. +// +// This workaround ensures that the compiler sees the same Env declaration, +// independently of whether <windows.h> was included. +#if defined(DeleteFile) +#undef DeleteFile +#define LEVELDB_DELETEFILE_UNDEFINED +#endif // defined(DeleteFile) +#endif // defined(_WIN32) + +namespace leveldb { + +class FileLock; +class Logger; +class RandomAccessFile; +class SequentialFile; +class Slice; +class WritableFile; + +class LEVELDB_EXPORT Env { + public: + Env(); + + Env(const Env&) = delete; + Env& operator=(const Env&) = delete; + + virtual ~Env(); + + // Return a default environment suitable for the current operating + // system. Sophisticated users may wish to provide their own Env + // implementation instead of relying on this default environment. + // + // The result of Default() belongs to leveldb and must never be deleted. + static Env* Default(); + + // 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 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 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 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 may be concurrently accessed by multiple threads. + virtual Status NewRandomAccessFile(const std::string& fname, + RandomAccessFile** result) = 0; + + // Create an object that writes to a new file with the specified + // name. Deletes any existing file with the same name and creates a + // new file. On success, stores a pointer to the new file in + // *result and returns OK. On failure stores nullptr in *result and + // returns non-OK. + // + // The returned file will only be accessed by one thread at a time. + virtual Status NewWritableFile(const std::string& fname, + 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; + + // Store in *result the names of the children of the specified directory. + // The names are relative to "dir". + // Original contents of *results are dropped. + virtual Status GetChildren(const std::string& dir, + std::vector<std::string>* result) = 0; + // Delete the named file. + // + // The default implementation calls DeleteFile, to support legacy Env + // implementations. Updated Env implementations must override RemoveFile and + // ignore the existence of DeleteFile. Updated code calling into the Env API + // must call RemoveFile instead of DeleteFile. + // + // A future release will remove DeleteDir and the default implementation of + // RemoveDir. + virtual Status RemoveFile(const std::string& fname); + + // DEPRECATED: Modern Env implementations should override RemoveFile instead. + // + // The default implementation calls RemoveFile, to support legacy Env user + // code that calls this method on modern Env implementations. Modern Env user + // code should call RemoveFile. + // + // A future release will remove this method. + virtual Status DeleteFile(const std::string& fname); + + // Create the specified directory. + virtual Status CreateDir(const std::string& dirname) = 0; + + // Delete the specified directory. + // + // The default implementation calls DeleteDir, to support legacy Env + // implementations. Updated Env implementations must override RemoveDir and + // ignore the existence of DeleteDir. Modern code calling into the Env API + // must call RemoveDir instead of DeleteDir. + // + // A future release will remove DeleteDir and the default implementation of + // RemoveDir. + virtual Status RemoveDir(const std::string& dirname); + + // DEPRECATED: Modern Env implementations should override RemoveDir instead. + // + // The default implementation calls RemoveDir, to support legacy Env user + // code that calls this method on modern Env implementations. Modern Env user + // code should call RemoveDir. + // + // A future release will remove this method. + virtual Status DeleteDir(const std::string& dirname); + + // Store the size of fname in *file_size. + virtual Status GetFileSize(const std::string& fname, uint64_t* file_size) = 0; + + // Rename file src to target. + virtual Status RenameFile(const std::string& src, + const std::string& target) = 0; + + // Lock the specified file. Used to prevent concurrent access to + // the same db by multiple processes. On failure, stores nullptr in + // *lock and returns non-OK. + // + // On success, stores a pointer to the object that represents the + // acquired lock in *lock and returns OK. The caller should call + // UnlockFile(*lock) to release the lock. If the process exits, + // the lock will be automatically released. + // + // If somebody else already holds the lock, finishes immediately + // with a failure. I.e., this call does not wait for existing locks + // to go away. + // + // May create the named file if it does not already exist. + virtual Status LockFile(const std::string& fname, FileLock** lock) = 0; + + // Release the lock acquired by a previous successful call to LockFile. + // REQUIRES: lock was returned by a successful LockFile() call + // REQUIRES: lock has not already been unlocked. + virtual Status UnlockFile(FileLock* lock) = 0; + + // Arrange to run "(*function)(arg)" once in a background thread. + // + // "function" may run in an unspecified thread. Multiple functions + // added to the same Env may run concurrently in different threads. + // I.e., the caller may not assume that background work items are + // serialized. + 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 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; + + // Create and return a log file for storing informational messages. + virtual Status NewLogger(const std::string& fname, Logger** result) = 0; + + // Returns the number of micro-seconds since some fixed point in time. Only + // useful for computing deltas of time. + virtual uint64_t NowMicros() = 0; + + // Sleep/delay the thread for the prescribed number of micro-seconds. + virtual void SleepForMicroseconds(int micros) = 0; +}; + +// A file abstraction for reading sequentially through a file +class LEVELDB_EXPORT SequentialFile { + public: + 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 + // written by this routine. Sets "*result" to the data that was + // read (including if fewer than "n" bytes were successfully read). + // May set "*result" to point at data in "scratch[0..n-1]", so + // "scratch[0..n-1]" must be live when "*result" is used. + // If an error was encountered, returns a non-OK status. + // + // REQUIRES: External synchronization + virtual Status Read(size_t n, Slice* result, char* scratch) = 0; + + // Skip "n" bytes from the file. This is guaranteed to be no + // slower that reading the same data, but may be faster. + // + // If end of file is reached, skipping will stop at the end of the + // file, and Skip will return OK. + // + // REQUIRES: External synchronization + virtual Status Skip(uint64_t n) = 0; +}; + +// A file abstraction for randomly reading the contents of a file. +class LEVELDB_EXPORT RandomAccessFile { + public: + RandomAccessFile() = default; + + RandomAccessFile(const RandomAccessFile&) = delete; + RandomAccessFile& operator=(const RandomAccessFile&) = delete; + + virtual ~RandomAccessFile(); + + // Read up to "n" bytes from the file starting at "offset". + // "scratch[0..n-1]" may be written by this routine. Sets "*result" + // to the data that was read (including if fewer than "n" bytes were + // successfully read). May set "*result" to point at data in + // "scratch[0..n-1]", so "scratch[0..n-1]" must be live when + // "*result" is used. If an error was encountered, returns a non-OK + // status. + // + // Safe for concurrent use by multiple threads. + virtual Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const = 0; +}; + +// A file abstraction for sequential writing. The implementation +// must provide buffering since callers may append small fragments +// at a time to the file. +class LEVELDB_EXPORT WritableFile { + public: + 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; +}; + +// An interface for writing log messages. +class LEVELDB_EXPORT Logger { + public: + 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, std::va_list ap) = 0; +}; + +// Identifies a locked file. +class LEVELDB_EXPORT FileLock { + public: + FileLock() = default; + + FileLock(const FileLock&) = delete; + FileLock& operator=(const FileLock&) = delete; + + virtual ~FileLock(); +}; + +// 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. +LEVELDB_EXPORT Status WriteStringToFile(Env* env, const Slice& data, + const std::string& fname); + +// A utility routine: read contents of named file into *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 LEVELDB_EXPORT EnvWrapper : public Env { + public: + // Initialize an EnvWrapper that delegates all calls to *t. + explicit EnvWrapper(Env* t) : target_(t) {} + virtual ~EnvWrapper(); + + // Return the target to which this Env forwards all calls. + Env* target() const { return target_; } + + // The following text is boilerplate that forwards all methods to target(). + Status NewSequentialFile(const std::string& f, SequentialFile** r) override { + return target_->NewSequentialFile(f, r); + } + Status NewRandomAccessFile(const std::string& f, + RandomAccessFile** r) override { + return target_->NewRandomAccessFile(f, r); + } + Status NewWritableFile(const std::string& f, WritableFile** r) override { + return target_->NewWritableFile(f, 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<std::string>* r) override { + return target_->GetChildren(dir, r); + } + Status RemoveFile(const std::string& f) override { + return target_->RemoveFile(f); + } + Status CreateDir(const std::string& d) override { + return target_->CreateDir(d); + } + Status RemoveDir(const std::string& d) override { + return target_->RemoveDir(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) override { + return target_->RenameFile(s, t); + } + Status LockFile(const std::string& f, FileLock** l) override { + return target_->LockFile(f, l); + } + Status UnlockFile(FileLock* l) override { return target_->UnlockFile(l); } + void Schedule(void (*f)(void*), void* a) override { + return target_->Schedule(f, a); + } + void StartThread(void (*f)(void*), void* a) override { + return target_->StartThread(f, a); + } + Status GetTestDirectory(std::string* path) override { + return target_->GetTestDirectory(path); + } + Status NewLogger(const std::string& fname, Logger** result) override { + return target_->NewLogger(fname, result); + } + uint64_t NowMicros() override { return target_->NowMicros(); } + void SleepForMicroseconds(int micros) override { + target_->SleepForMicroseconds(micros); + } + + private: + Env* target_; +}; + +} // namespace leveldb + +// This workaround can be removed when leveldb::Env::DeleteFile is removed. +// Redefine DeleteFile if it was undefined earlier. +#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 new file mode 100644 index 0000000..49c8eda --- /dev/null +++ b/include/leveldb/filter_policy.h @@ -0,0 +1,72 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// A database can be configured with a custom FilterPolicy object. +// This object is responsible for creating a small filter from a set +// of keys. These filters are stored in leveldb and are consulted +// automatically by leveldb to decide whether or not to read some +// information from disk. In many cases, a filter can cut down the +// number of disk seeks form a handful to a single disk seek per +// DB::Get() call. +// +// Most people will want to use the builtin bloom filter support (see +// NewBloomFilterPolicy() below). + +#ifndef STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_ +#define STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_ + +#include <string> + +#include "leveldb/export.h" + +namespace leveldb { + +class Slice; + +class LEVELDB_EXPORT FilterPolicy { + public: + virtual ~FilterPolicy(); + + // Return the name of this policy. Note that if the filter encoding + // changes in an incompatible way, the name returned by this method + // must be changed. Otherwise, old incompatible filters may be + // passed to methods of this type. + virtual const char* Name() const = 0; + + // keys[0,n-1] contains a list of keys (potentially with duplicates) + // that are ordered according to the user supplied comparator. + // Append a filter that summarizes keys[0,n-1] to *dst. + // + // Warning: do not change the initial contents of *dst. Instead, + // append the newly constructed filter to *dst. + virtual void CreateFilter(const Slice* keys, int n, + std::string* dst) const = 0; + + // "filter" contains the data appended by a preceding call to + // CreateFilter() on this class. This method must return true if + // the key was in the list of keys passed to CreateFilter(). + // This method may return true or false if the key was not on the + // list, but it should aim to return false with a high probability. + virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const = 0; +}; + +// Return a new filter policy that uses a bloom filter with approximately +// the specified number of bits per key. A good value for bits_per_key +// is 10, which yields a filter with ~ 1% false positive rate. +// +// Callers must delete the result after any database that is using the +// result has been closed. +// +// Note: if you are using a custom comparator that ignores some parts +// of the keys being compared, you must not use NewBloomFilterPolicy() +// and must provide your own FilterPolicy that also ignores the +// corresponding parts of the keys. For example, if the comparator +// ignores trailing spaces, it would be incorrect to use a +// FilterPolicy (like NewBloomFilterPolicy) that does not ignore +// trailing spaces in keys. +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 new file mode 100644 index 0000000..bb9a5df --- /dev/null +++ b/include/leveldb/iterator.h @@ -0,0 +1,112 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// An iterator yields a sequence of key/value pairs from a source. +// The following class defines the interface. Multiple implementations +// are provided by this library. In particular, iterators are provided +// to access the contents of a Table or a DB. +// +// Multiple threads can invoke const methods on an Iterator without +// external synchronization, but if any of the threads may call a +// non-const method, all threads accessing the same Iterator must use +// external synchronization. + +#ifndef STORAGE_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 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 + // not valid. This method returns true iff the iterator is valid. + virtual bool Valid() const = 0; + + // Position at the first key in the source. The iterator is Valid() + // after this call iff the source is not empty. + virtual void SeekToFirst() = 0; + + // Position at the last key in the source. The iterator is + // Valid() after this call iff the source is not empty. + virtual void SeekToLast() = 0; + + // Position at the first key in the source that is at or past target. + // The iterator is Valid() after this call iff the source contains + // an entry that comes at or past target. + virtual void Seek(const Slice& target) = 0; + + // Moves to the next entry in the source. After this call, Valid() is + // true iff the iterator was not positioned at the last entry in the source. + // REQUIRES: Valid() + virtual void Next() = 0; + + // Moves to the previous entry in the source. After this call, Valid() is + // true iff the iterator was not positioned at the first entry in source. + // REQUIRES: Valid() + virtual void Prev() = 0; + + // Return the key for the current entry. The underlying storage for + // the returned slice is valid only until the next modification of + // the iterator. + // REQUIRES: Valid() + virtual Slice key() const = 0; + + // Return the value for the current entry. The underlying storage for + // the returned slice is valid only until the next modification of + // the iterator. + // REQUIRES: Valid() + virtual Slice value() const = 0; + + // If an error has occurred, return it. Else return an ok status. + virtual Status status() const = 0; + + // Clients are allowed to register function/arg1/arg2 triples that + // will be invoked when this iterator is destroyed. + // + // Note that unlike all of the preceding methods, this method is + // not abstract and therefore clients should not override it. + using CleanupFunction = void (*)(void* arg1, void* arg2); + void RegisterCleanup(CleanupFunction function, void* arg1, void* arg2); + + private: + // 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; + CleanupNode* next; + }; + CleanupNode cleanup_head_; +}; + +// Return an empty iterator (yields nothing). +LEVELDB_EXPORT Iterator* NewEmptyIterator(); + +// Return an empty iterator with the specified status. +LEVELDB_EXPORT Iterator* NewErrorIterator(const Status& status); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_ITERATOR_H_ diff --git a/include/leveldb/options.h b/include/leveldb/options.h new file mode 100644 index 0000000..0f285bc --- /dev/null +++ b/include/leveldb/options.h @@ -0,0 +1,187 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_INCLUDE_OPTIONS_H_ +#define STORAGE_LEVELDB_INCLUDE_OPTIONS_H_ + +#include <cstddef> + +#include "leveldb/export.h" + +namespace leveldb { + +class Cache; +class Comparator; +class Env; +class FilterPolicy; +class Logger; +class Snapshot; + +// DB contents are stored in a set of blocks, each of which holds a +// sequence of key,value pairs. Each block may be compressed before +// being stored in a file. The following enum describes which +// compression method (if any) is used to compress a block. +enum CompressionType { + // NOTE: do not change the values of existing entries, as these are + // part of the persistent format on disk. + kNoCompression = 0x0, + kSnappyCompression = 0x1 +}; + +// Options to control the behavior of a database (passed to DB::Open) +struct LEVELDB_EXPORT Options { + // Create an Options object with default values for all fields. + Options(); + + // ------------------- + // Parameters that affect behavior + + // Comparator used to define the order of keys in the table. + // Default: a comparator that uses lexicographic byte-wise ordering + // + // REQUIRES: The client must ensure that the comparator supplied + // here has the same name and orders keys *exactly* the same as the + // comparator provided to previous open calls on the same DB. + const Comparator* comparator; + + // If true, the database will be created if it is missing. + bool create_if_missing = false; + + // If true, an error is raised if the database already 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. + bool paranoid_checks = false; + + // Use the specified object to interact with the environment, + // e.g. to read/write files, schedule background work, etc. + // Default: Env::Default() + Env* env; + + // Any internal progress/error information generated by the db will + // be written to info_log if it is non-null, or to a file stored + // in the same directory as the DB contents if info_log is null. + Logger* info_log = nullptr; + + // ------------------- + // Parameters that affect performance + + // Amount of data to build up in memory (backed by an unsorted log + // on disk) before converting to a sorted on-disk file. + // + // Larger values increase performance, especially during bulk loads. + // Up to two write buffers may be held in memory at the same time, + // so you may wish to adjust this parameter to control memory usage. + // Also, a larger write buffer will result in a longer recovery time + // the next time the database is opened. + 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). + 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. + 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. + 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. + 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. + // + // Default: kSnappyCompression, which gives lightweight but fast + // compression. + // + // Typical speeds of kSnappyCompression on an Intel(R) Core(TM)2 2.4GHz: + // ~200-500MB/s compression + // ~400-800MB/s decompression + // Note that these speeds are significantly faster than most + // persistent storage speeds, and therefore it is typically never + // worth switching to kNoCompression. Even if the input data is + // incompressible, the kSnappyCompression implementation will + // efficiently detect that and will switch to uncompressed mode. + CompressionType compression = kSnappyCompression; + + // 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. + const FilterPolicy* filter_policy = nullptr; +}; + +// Options that control read operations +struct LEVELDB_EXPORT ReadOptions { + ReadOptions() = default; + + // If true, all data read from underlying storage will be + // verified against corresponding 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. + bool fill_cache = true; + + // 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 + // snapshot of the state at the beginning of this read operation. + const Snapshot* snapshot = nullptr; +}; + +// Options that control write operations +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 + // slower. + // + // If this flag is false, and the machine crashes, some recent + // writes may be lost. Note that if it is just the process that + // crashes (i.e., the machine does not reboot), no writes will be + // lost even if sync==false. + // + // In other words, a DB write with sync==false has similar + // crash semantics as the "write()" system call. A DB write + // with sync==true has similar crash semantics to a "write()" + // system call followed by "fsync()". + bool sync = false; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_OPTIONS_H_ diff --git a/include/leveldb/slice.h b/include/leveldb/slice.h new file mode 100644 index 0000000..37cb821 --- /dev/null +++ b/include/leveldb/slice.h @@ -0,0 +1,114 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// Slice is a simple structure containing a pointer into some external +// storage and a size. The user of a Slice must ensure that the slice +// is not used after the corresponding external storage has been +// deallocated. +// +// Multiple threads can invoke const methods on a Slice without +// external synchronization, but if any of the threads may call a +// non-const method, all threads accessing the same Slice must use +// external synchronization. + +#ifndef STORAGE_LEVELDB_INCLUDE_SLICE_H_ +#define STORAGE_LEVELDB_INCLUDE_SLICE_H_ + +#include <cassert> +#include <cstddef> +#include <cstring> +#include <string> + +#include "leveldb/export.h" + +namespace leveldb { + +class LEVELDB_EXPORT Slice { + public: + // Create an empty slice. + Slice() : data_(""), size_(0) {} + + // Create a slice that refers to d[0,n-1]. + Slice(const char* d, size_t n) : data_(d), size_(n) {} + + // Create a slice that refers to the contents of "s" + 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)) {} + + // 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_; } + + // Return the length (in bytes) of the referenced data + size_t size() const { return size_; } + + // Return true iff the length of the referenced data is zero + bool empty() const { return size_ == 0; } + + // Return the ith byte in the referenced data. + // REQUIRES: n < size() + char operator[](size_t n) const { + assert(n < size()); + return data_[n]; + } + + // Change this slice to refer to an empty array + void clear() { + data_ = ""; + size_ = 0; + } + + // Drop the first "n" bytes from this slice. + void remove_prefix(size_t n) { + assert(n <= size()); + data_ += n; + size_ -= n; + } + + // Return a string that contains the copy of the referenced data. + std::string ToString() const { return std::string(data_, size_); } + + // Three-way comparison. Returns value: + // < 0 iff "*this" < "b", + // == 0 iff "*this" == "b", + // > 0 iff "*this" > "b" + int compare(const Slice& b) const; + + // Return true iff "x" is a prefix of "*this" + bool starts_with(const Slice& x) const { + return ((size_ >= x.size_) && (memcmp(data_, x.data_, x.size_) == 0)); + } + + private: + const char* data_; + size_t size_; +}; + +inline bool operator==(const Slice& x, const Slice& y) { + return ((x.size() == y.size()) && + (memcmp(x.data(), y.data(), x.size()) == 0)); +} + +inline bool operator!=(const Slice& x, const Slice& y) { return !(x == y); } + +inline int Slice::compare(const Slice& b) const { + const size_t min_len = (size_ < b.size_) ? size_ : b.size_; + int r = memcmp(data_, b.data_, min_len); + if (r == 0) { + if (size_ < b.size_) + r = -1; + else if (size_ > b.size_) + r = +1; + } + return r; +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_SLICE_H_ diff --git a/include/leveldb/status.h b/include/leveldb/status.h new file mode 100644 index 0000000..e327314 --- /dev/null +++ b/include/leveldb/status.h @@ -0,0 +1,122 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// A Status encapsulates the result of an operation. It may indicate success, +// or it may indicate an error with an associated error message. +// +// Multiple threads can invoke const methods on a Status without +// external synchronization, but if any of the threads may call a +// non-const method, all threads accessing the same Status must use +// external synchronization. + +#ifndef STORAGE_LEVELDB_INCLUDE_STATUS_H_ +#define STORAGE_LEVELDB_INCLUDE_STATUS_H_ + +#include <algorithm> +#include <string> + +#include "leveldb/export.h" +#include "leveldb/slice.h" + +namespace leveldb { + +class LEVELDB_EXPORT Status { + public: + // Create a success status. + Status() noexcept : state_(nullptr) {} + ~Status() { delete[] state_; } + + 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(); } + + // Return error status of an appropriate type. + static Status NotFound(const Slice& msg, const Slice& msg2 = Slice()) { + return Status(kNotFound, msg, msg2); + } + static Status Corruption(const Slice& msg, const Slice& msg2 = Slice()) { + return Status(kCorruption, msg, msg2); + } + static Status NotSupported(const Slice& msg, const Slice& msg2 = Slice()) { + return Status(kNotSupported, msg, msg2); + } + static Status InvalidArgument(const Slice& msg, const Slice& msg2 = Slice()) { + return Status(kInvalidArgument, msg, msg2); + } + static Status IOError(const Slice& msg, const Slice& msg2 = Slice()) { + return Status(kIOError, msg, msg2); + } + + // Returns true iff the status indicates success. + bool ok() const { return (state_ == nullptr); } + + // Returns true iff the status indicates a NotFound error. + bool IsNotFound() const { return code() == kNotFound; } + + // Returns true iff the status indicates a Corruption error. + bool IsCorruption() const { return code() == kCorruption; } + + // Returns true iff the status indicates 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: + enum Code { + kOk = 0, + kNotFound = 1, + kCorruption = 2, + kNotSupported = 3, + kInvalidArgument = 4, + kIOError = 5 + }; + + Code code() const { + return (state_ == nullptr) ? kOk : static_cast<Code>(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& rhs) { + state_ = (rhs.state_ == nullptr) ? nullptr : CopyState(rhs.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_ = (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 + +#endif // STORAGE_LEVELDB_INCLUDE_STATUS_H_ diff --git a/include/leveldb/table.h b/include/leveldb/table.h new file mode 100644 index 0000000..a30e903 --- /dev/null +++ b/include/leveldb/table.h @@ -0,0 +1,84 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_INCLUDE_TABLE_H_ +#define STORAGE_LEVELDB_INCLUDE_TABLE_H_ + +#include <cstdint> + +#include "leveldb/export.h" +#include "leveldb/iterator.h" + +namespace leveldb { + +class Block; +class BlockHandle; +class Footer; +struct Options; +class RandomAccessFile; +struct ReadOptions; +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 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 + // retrieving data from the 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 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); + + Table(const Table&) = delete; + Table& operator=(const Table&) = delete; + + ~Table(); + + // Returns a new iterator over the table contents. + // The result of NewIterator() is initially invalid (caller must + // call one of the Seek methods on the iterator before using it). + Iterator* NewIterator(const ReadOptions&) const; + + // Given a key, return an approximate byte offset in the file where + // the data for that key begins (or would begin if the key were + // present in the file). The returned value is in terms of file + // bytes, and so includes effects like compression of the underlying data. + // E.g., the approximate offset of the last key in the table will + // be close to the file length. + uint64_t ApproximateOffsetOf(const Slice& key) const; + + private: + friend class TableCache; + struct 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. + 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); + + Rep* const rep_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_TABLE_H_ diff --git a/include/leveldb/table_builder.h b/include/leveldb/table_builder.h new file mode 100644 index 0000000..85710c3 --- /dev/null +++ b/include/leveldb/table_builder.h @@ -0,0 +1,93 @@ +// 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. +// +// TableBuilder provides the interface used to build a Table +// (an immutable and sorted map from keys to values). +// +// Multiple threads can invoke const methods on a TableBuilder without +// external synchronization, but if any of the threads may call a +// non-const method, all threads accessing the same TableBuilder must use +// external synchronization. + +#ifndef STORAGE_LEVELDB_INCLUDE_TABLE_BUILDER_H_ +#define STORAGE_LEVELDB_INCLUDE_TABLE_BUILDER_H_ + +#include <cstdint> + +#include "leveldb/export.h" +#include "leveldb/options.h" +#include "leveldb/status.h" + +namespace leveldb { + +class BlockBuilder; +class BlockHandle; +class WritableFile; + +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(); + + // Change the options used by this builder. Note: only some of the + // option fields can be changed after construction. If a field is + // not allowed to change dynamically and its value in the structure + // passed to the constructor is different from its value in the + // structure passed to this method, this method will return an error + // without changing any fields. + Status ChangeOptions(const Options& options); + + // Add key,value to the table being constructed. + // REQUIRES: key is after any previously added key according to comparator. + // REQUIRES: Finish(), Abandon() have not been called + void Add(const Slice& key, const Slice& value); + + // Advanced operation: flush any buffered key/value pairs to file. + // Can be used to ensure that two adjacent entries never live in + // the same data block. Most clients should not need to use this method. + // REQUIRES: Finish(), Abandon() have not been called + void Flush(); + + // Return non-ok iff some error has been detected. + Status status() const; + + // Finish building the table. Stops using the file passed to the + // constructor after this function returns. + // REQUIRES: Finish(), Abandon() have not been called + Status Finish(); + + // Indicate that the contents of this builder should be abandoned. Stops + // using the file passed to the constructor after this function returns. + // If the caller is not going to call Finish(), it must call Abandon() + // before destroying this builder. + // REQUIRES: Finish(), Abandon() have not been called + void Abandon(); + + // Number of calls to Add() so far. + uint64_t NumEntries() const; + + // Size of the file generated so far. If invoked after a successful + // Finish() call, returns the size of the final generated file. + uint64_t FileSize() const; + + private: + bool ok() const { return status().ok(); } + void WriteBlock(BlockBuilder* block, BlockHandle* handle); + void WriteRawBlock(const Slice& data, CompressionType, BlockHandle* handle); + + struct Rep; + Rep* rep_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_TABLE_BUILDER_H_ diff --git a/include/leveldb/write_batch.h b/include/leveldb/write_batch.h new file mode 100644 index 0000000..94d4115 --- /dev/null +++ b/include/leveldb/write_batch.h @@ -0,0 +1,83 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// WriteBatch holds a collection of updates to apply atomically to a DB. +// +// The updates are applied in the order in which they are added +// to the WriteBatch. For example, the value of "key" will be "v3" +// after the following batch is written: +// +// batch.Put("key", "v1"); +// batch.Delete("key"); +// batch.Put("key", "v2"); +// batch.Put("key", "v3"); +// +// Multiple threads can invoke const methods on a WriteBatch without +// external synchronization, but if any of the threads may call a +// non-const method, all threads accessing the same WriteBatch must use +// external synchronization. + +#ifndef STORAGE_LEVELDB_INCLUDE_WRITE_BATCH_H_ +#define STORAGE_LEVELDB_INCLUDE_WRITE_BATCH_H_ + +#include <string> + +#include "leveldb/export.h" +#include "leveldb/status.h" + +namespace leveldb { + +class Slice; + +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. + void Put(const Slice& key, const Slice& value); + + // If the database contains a mapping for "key", erase it. Else do nothing. + void Delete(const Slice& key); + + // 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. + Status Iterate(Handler* handler) const; + + private: + friend class WriteBatchInternal; + + std::string rep_; // See comment in write_batch.cc for the format of rep_ +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_INCLUDE_WRITE_BATCH_H_ diff --git a/issues/issue178_test.cc b/issues/issue178_test.cc new file mode 100644 index 0000000..8fa5bb9 --- /dev/null +++ b/issues/issue178_test.cc @@ -0,0 +1,90 @@ +// Copyright (c) 2013 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +// Test for issue 178: a manual compaction causes deleted data to reappear. +#include <cstdlib> +#include <iostream> +#include <sstream> + +#include "gtest/gtest.h" +#include "leveldb/db.h" +#include "leveldb/write_batch.h" +#include "util/testutil.h" + +namespace { + +const int kNumKeys = 1100000; + +std::string Key1(int i) { + char buf[100]; + std::snprintf(buf, sizeof(buf), "my_key_%d", i); + return buf; +} + +std::string Key2(int i) { return Key1(i) + "_xxx"; } + +TEST(Issue178, Test) { + // Get rid of any state from an old run. + std::string dbpath = testing::TempDir() + "leveldb_cbug_test"; + DestroyDB(dbpath, leveldb::Options()); + + // Open database. Disable compression since it affects the creation + // of layers and the code below is trying to test against a very + // specific scenario. + leveldb::DB* db; + leveldb::Options db_options; + db_options.create_if_missing = true; + db_options.compression = leveldb::kNoCompression; + ASSERT_LEVELDB_OK(leveldb::DB::Open(db_options, dbpath, &db)); + + // create first key range + leveldb::WriteBatch batch; + for (size_t i = 0; i < kNumKeys; i++) { + batch.Put(Key1(i), "value for range 1 key"); + } + ASSERT_LEVELDB_OK(db->Write(leveldb::WriteOptions(), &batch)); + + // create second key range + batch.Clear(); + for (size_t i = 0; i < kNumKeys; i++) { + batch.Put(Key2(i), "value for range 2 key"); + } + ASSERT_LEVELDB_OK(db->Write(leveldb::WriteOptions(), &batch)); + + // delete second key range + batch.Clear(); + for (size_t i = 0; i < kNumKeys; i++) { + batch.Delete(Key2(i)); + } + ASSERT_LEVELDB_OK(db->Write(leveldb::WriteOptions(), &batch)); + + // compact database + std::string start_key = Key1(0); + std::string end_key = Key1(kNumKeys - 1); + leveldb::Slice least(start_key.data(), start_key.size()); + leveldb::Slice greatest(end_key.data(), end_key.size()); + + // commenting out the line below causes the example to work correctly + db->CompactRange(&least, &greatest); + + // count the keys + leveldb::Iterator* iter = db->NewIterator(leveldb::ReadOptions()); + size_t num_keys = 0; + for (iter->SeekToFirst(); iter->Valid(); iter->Next()) { + num_keys++; + } + delete iter; + ASSERT_EQ(kNumKeys, num_keys) << "Bad number of keys"; + + // close database + delete db; + DestroyDB(dbpath, leveldb::Options()); +} + +} // anonymous namespace + +int main(int argc, char** argv) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/issues/issue200_test.cc b/issues/issue200_test.cc new file mode 100644 index 0000000..4eba23a --- /dev/null +++ b/issues/issue200_test.cc @@ -0,0 +1,59 @@ +// Copyright (c) 2013 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +// Test for issue 200: when iterator switches direction from backward +// to forward, the current key can be yielded unexpectedly if a new +// mutation has been added just before the current key. + +#include "gtest/gtest.h" +#include "leveldb/db.h" +#include "util/testutil.h" + +namespace leveldb { + +TEST(Issue200, Test) { + // Get rid of any state from an old run. + std::string dbpath = testing::TempDir() + "leveldb_issue200_test"; + DestroyDB(dbpath, Options()); + + DB* db; + Options options; + options.create_if_missing = true; + ASSERT_LEVELDB_OK(DB::Open(options, dbpath, &db)); + + WriteOptions write_options; + ASSERT_LEVELDB_OK(db->Put(write_options, "1", "b")); + ASSERT_LEVELDB_OK(db->Put(write_options, "2", "c")); + ASSERT_LEVELDB_OK(db->Put(write_options, "3", "d")); + ASSERT_LEVELDB_OK(db->Put(write_options, "4", "e")); + ASSERT_LEVELDB_OK(db->Put(write_options, "5", "f")); + + ReadOptions read_options; + Iterator* iter = db->NewIterator(read_options); + + // Add an element that should not be reflected in the iterator. + ASSERT_LEVELDB_OK(db->Put(write_options, "25", "cd")); + + iter->Seek("5"); + ASSERT_EQ(iter->key().ToString(), "5"); + iter->Prev(); + ASSERT_EQ(iter->key().ToString(), "4"); + iter->Prev(); + ASSERT_EQ(iter->key().ToString(), "3"); + iter->Next(); + ASSERT_EQ(iter->key().ToString(), "4"); + iter->Next(); + ASSERT_EQ(iter->key().ToString(), "5"); + + delete iter; + delete db; + DestroyDB(dbpath, options); +} + +} // namespace leveldb + +int main(int argc, char** argv) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/issues/issue320_test.cc b/issues/issue320_test.cc new file mode 100644 index 0000000..c08296a --- /dev/null +++ b/issues/issue320_test.cc @@ -0,0 +1,131 @@ +// 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 <cstdint> +#include <cstdlib> +#include <iostream> +#include <memory> +#include <string> +#include <vector> + +#include "gtest/gtest.h" +#include "leveldb/db.h" +#include "leveldb/write_batch.h" +#include "util/testutil.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 + +TEST(Issue320, Test) { + std::srand(0); + + bool delete_before_put = false; + bool keep_snapshots = true; + + std::vector<std::unique_ptr<std::pair<std::string, std::string>>> test_map( + 10000); + std::vector<Snapshot const*> snapshots(100, nullptr); + + DB* db; + Options options; + options.create_if_missing = true; + + std::string dbpath = testing::TempDir() + "leveldb_issue320_test"; + ASSERT_LEVELDB_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<std::string, std::string>( + CreateRandomString(index), CreateRandomString(index))); + batch.Put(test_map[index]->first, test_map[index]->second); + } else { + ASSERT_LEVELDB_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_LEVELDB_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) { + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} diff --git a/port/README.md b/port/README.md new file mode 100644 index 0000000..8b17153 --- /dev/null +++ b/port/README.md @@ -0,0 +1,10 @@ +This directory contains interfaces and implementations that isolate the +rest of the package from platform details. + +Code in the rest of the package includes "port.h" from this directory. +"port.h" in turn includes a platform specific "port_<platform>.h" file +that provides the platform specific implementation. + +See port_stdcxx.h for an example of what must be provided in a platform +specific header file. + diff --git a/port/port.h b/port/port.h new file mode 100644 index 0000000..4b247f7 --- /dev/null +++ b/port/port.h @@ -0,0 +1,19 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_PORT_PORT_H_ +#define STORAGE_LEVELDB_PORT_PORT_H_ + +#include <string.h> + +// Include the appropriate platform specific file below. If you are +// porting to a new platform, see "port_example.h" for documentation +// of what the new port_<platform>.h file must provide. +#if defined(LEVELDB_PLATFORM_POSIX) || defined(LEVELDB_PLATFORM_WINDOWS) +#include "port/port_stdcxx.h" +#elif defined(LEVELDB_PLATFORM_CHROMIUM) +#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..272671d --- /dev/null +++ b/port/port_config.h.in @@ -0,0 +1,33 @@ +// 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 <unistd.h>. +#if !defined(HAVE_FDATASYNC) +#cmakedefine01 HAVE_FDATASYNC +#endif // !defined(HAVE_FDATASYNC) + +// Define to 1 if you have a definition for F_FULLFSYNC in <fcntl.h>. +#if !defined(HAVE_FULLFSYNC) +#cmakedefine01 HAVE_FULLFSYNC +#endif // !defined(HAVE_FULLFSYNC) + +// Define to 1 if you have a definition for O_CLOEXEC in <fcntl.h>. +#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) + +#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 new file mode 100644 index 0000000..a665910 --- /dev/null +++ b/port/port_example.h @@ -0,0 +1,100 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// This file contains the specification, but not the implementations, +// of the types/operations/etc. that should be defined by a platform +// specific port_<platform>.h file. Use this file as a reference for +// how to port this package to a new platform. + +#ifndef STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ +#define STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ + +#include "port/thread_annotations.h" + +namespace leveldb { +namespace port { + +// TODO(jorlow): Many of these belong more in the environment class rather than +// here. We should try moving them and see if it affects perf. + +// ------------------ Threading ------------------- + +// A Mutex represents an exclusive lock. +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() EXCLUSIVE_LOCK_FUNCTION(); + + // Unlock the mutex. + // REQUIRES: This mutex was locked by this thread. + 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() ASSERT_EXCLUSIVE_LOCK(); +}; + +class CondVar { + public: + explicit CondVar(Mutex* mu); + ~CondVar(); + + // Atomically release *mu and block on this condition variable until + // either a call to SignalAll(), or a call to Signal() that picks + // this thread to wakeup. + // REQUIRES: this thread holds *mu + void Wait(); + + // If there are some threads waiting, wake up at least one of them. + void Signal(); + + // Wake up all waiting threads. + void SignallAll(); +}; + +// ------------------ Compression ------------------- + +// Store the snappy compression of "input[0,input_length-1]" in *output. +// Returns false if snappy is not supported by this port. +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. +bool Snappy_GetUncompressedLength(const char* input, size_t length, + size_t* result); + +// Attempt to snappy uncompress input[0,input_length-1] into *output. +// Returns true if successful, false if the input is invalid lightweight +// compressed data. +// +// REQUIRES: at least the first "n" bytes of output[] must be writable +// where "n" is the result of a successful call to +// Snappy_GetUncompressedLength. +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. +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 + +#endif // STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_ diff --git a/port/port_stdcxx.h b/port/port_stdcxx.h new file mode 100644 index 0000000..2bda48d --- /dev/null +++ b/port/port_stdcxx.h @@ -0,0 +1,151 @@ +// 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 <crc32c/crc32c.h> +#endif // HAVE_CRC32C +#if HAVE_SNAPPY +#include <snappy.h> +#endif // HAVE_SNAPPY + +#include <cassert> +#include <condition_variable> // NOLINT +#include <cstddef> +#include <cstdint> +#include <mutex> // NOLINT +#include <string> + +#include "port/thread_annotations.h" + +namespace leveldb { +namespace port { + +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<std::mutex> 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<const uint8_t*>(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 new file mode 100644 index 0000000..1547df9 --- /dev/null +++ b/port/thread_annotations.h @@ -0,0 +1,108 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_ +#define STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_ + +// 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(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(exclusive_locks_required(__VA_ARGS__)) +#endif + +#ifndef SHARED_LOCKS_REQUIRED +#define SHARED_LOCKS_REQUIRED(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(shared_locks_required(__VA_ARGS__)) +#endif + +#ifndef LOCKS_EXCLUDED +#define LOCKS_EXCLUDED(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(locks_excluded(__VA_ARGS__)) +#endif + +#ifndef LOCK_RETURNED +#define LOCK_RETURNED(x) THREAD_ANNOTATION_ATTRIBUTE__(lock_returned(x)) +#endif + +#ifndef LOCKABLE +#define LOCKABLE THREAD_ANNOTATION_ATTRIBUTE__(lockable) +#endif + +#ifndef SCOPED_LOCKABLE +#define SCOPED_LOCKABLE THREAD_ANNOTATION_ATTRIBUTE__(scoped_lockable) +#endif + +#ifndef EXCLUSIVE_LOCK_FUNCTION +#define EXCLUSIVE_LOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(exclusive_lock_function(__VA_ARGS__)) +#endif + +#ifndef SHARED_LOCK_FUNCTION +#define SHARED_LOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(shared_lock_function(__VA_ARGS__)) +#endif + +#ifndef EXCLUSIVE_TRYLOCK_FUNCTION +#define EXCLUSIVE_TRYLOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(exclusive_trylock_function(__VA_ARGS__)) +#endif + +#ifndef SHARED_TRYLOCK_FUNCTION +#define SHARED_TRYLOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(shared_trylock_function(__VA_ARGS__)) +#endif + +#ifndef UNLOCK_FUNCTION +#define UNLOCK_FUNCTION(...) \ + THREAD_ANNOTATION_ATTRIBUTE__(unlock_function(__VA_ARGS__)) +#endif + +#ifndef 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/table/block.cc b/table/block.cc new file mode 100644 index 0000000..3b15257 --- /dev/null +++ b/table/block.cc @@ -0,0 +1,292 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// Decodes the blocks generated by block_builder.cc. + +#include "table/block.h" + +#include <algorithm> +#include <cstdint> +#include <vector> + +#include "leveldb/comparator.h" +#include "table/format.h" +#include "util/coding.h" +#include "util/logging.h" + +namespace leveldb { + +inline uint32_t Block::NumRestarts() const { + assert(size_ >= sizeof(uint32_t)); + return DecodeFixed32(data_ + size_ - sizeof(uint32_t)); +} + +Block::Block(const BlockContents& contents) + : data_(contents.data.data()), + size_(contents.data.size()), + owned_(contents.heap_allocated) { + if (size_ < sizeof(uint32_t)) { + size_ = 0; // Error marker + } else { + 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; + } else { + restart_offset_ = size_ - (1 + NumRestarts()) * sizeof(uint32_t); + } + } +} + +Block::~Block() { + if (owned_) { + delete[] data_; + } +} + +// Helper routine: decode the next block entry starting at "p", +// storing the number of shared key bytes, non_shared key bytes, +// and the length of the value in "*shared", "*non_shared", and +// "*value_length", respectively. Will not dereference past "limit". +// +// If any errors are detected, returns nullptr. Otherwise, returns a +// pointer to the key delta (just past the three decoded values). +static inline const char* DecodeEntry(const char* p, const char* limit, + uint32_t* shared, uint32_t* non_shared, + uint32_t* value_length) { + if (limit - p < 3) return nullptr; + *shared = reinterpret_cast<const uint8_t*>(p)[0]; + *non_shared = reinterpret_cast<const uint8_t*>(p)[1]; + *value_length = reinterpret_cast<const uint8_t*>(p)[2]; + if ((*shared | *non_shared | *value_length) < 128) { + // Fast path: all three values are encoded in one byte each + p += 3; + } else { + if ((p = GetVarint32Ptr(p, limit, shared)) == nullptr) return nullptr; + if ((p = GetVarint32Ptr(p, limit, non_shared)) == nullptr) return nullptr; + if ((p = GetVarint32Ptr(p, limit, value_length)) == nullptr) return nullptr; + } + + if (static_cast<uint32_t>(limit - p) < (*non_shared + *value_length)) { + return nullptr; + } + return p; +} + +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 + + // current_ is offset in data_ of current entry. >= restarts_ if !Valid + uint32_t current_; + uint32_t restart_index_; // Index of restart block in which current_ falls + std::string key_; + Slice value_; + Status status_; + + inline int Compare(const Slice& a, const Slice& b) const { + return comparator_->Compare(a, b); + } + + // Return the offset in data_ just past the end of the current entry. + inline uint32_t NextEntryOffset() const { + return (value_.data() + value_.size()) - data_; + } + + uint32_t GetRestartPoint(uint32_t index) { + assert(index < num_restarts_); + return DecodeFixed32(data_ + restarts_ + index * sizeof(uint32_t)); + } + + void SeekToRestartPoint(uint32_t index) { + key_.clear(); + restart_index_ = index; + // current_ will be fixed by ParseNextKey(); + + // ParseNextKey() starts at the end of value_, so set value_ accordingly + uint32_t offset = GetRestartPoint(index); + value_ = Slice(data_ + offset, 0); + } + + public: + Iter(const Comparator* comparator, const char* data, uint32_t restarts, + uint32_t num_restarts) + : comparator_(comparator), + data_(data), + restarts_(restarts), + num_restarts_(num_restarts), + current_(restarts_), + restart_index_(num_restarts_) { + assert(num_restarts_ > 0); + } + + bool Valid() const override { return current_ < restarts_; } + Status status() const override { return status_; } + Slice key() const override { + assert(Valid()); + return key_; + } + Slice value() const override { + assert(Valid()); + return value_; + } + + void Next() override { + assert(Valid()); + ParseNextKey(); + } + + void Prev() override { + assert(Valid()); + + // Scan backwards to a restart point before current_ + const uint32_t original = current_; + while (GetRestartPoint(restart_index_) >= original) { + if (restart_index_ == 0) { + // No more entries + current_ = restarts_; + restart_index_ = num_restarts_; + return; + } + restart_index_--; + } + + SeekToRestartPoint(restart_index_); + do { + // Loop until end of current entry hits the start of original entry + } while (ParseNextKey() && NextEntryOffset() < original); + } + + void Seek(const Slice& target) override { + // Binary search in restart array to find the last restart point + // with a key < target + uint32_t left = 0; + uint32_t right = num_restarts_ - 1; + int current_key_compare = 0; + + if (Valid()) { + // If we're already scanning, use the current position as a starting + // point. This is beneficial if the key we're seeking to is ahead of the + // current position. + current_key_compare = Compare(key_, target); + if (current_key_compare < 0) { + // key_ is smaller than target + left = restart_index_; + } else if (current_key_compare > 0) { + right = restart_index_; + } else { + // We're seeking to the key we're already at. + return; + } + } + + while (left < right) { + uint32_t mid = (left + right + 1) / 2; + uint32_t region_offset = GetRestartPoint(mid); + uint32_t shared, non_shared, value_length; + const char* key_ptr = + DecodeEntry(data_ + region_offset, data_ + restarts_, &shared, + &non_shared, &value_length); + if (key_ptr == nullptr || (shared != 0)) { + CorruptionError(); + return; + } + Slice mid_key(key_ptr, non_shared); + if (Compare(mid_key, target) < 0) { + // Key at "mid" is smaller than "target". Therefore all + // blocks before "mid" are uninteresting. + left = mid; + } else { + // Key at "mid" is >= "target". Therefore all blocks at or + // after "mid" are uninteresting. + right = mid - 1; + } + } + + // We might be able to use our current position within the restart block. + // This is true if we determined the key we desire is in the current block + // and is after than the current key. + assert(current_key_compare == 0 || Valid()); + bool skip_seek = left == restart_index_ && current_key_compare < 0; + if (!skip_seek) { + SeekToRestartPoint(left); + } + // Linear search (within restart block) for first key >= target + while (true) { + if (!ParseNextKey()) { + return; + } + if (Compare(key_, target) >= 0) { + return; + } + } + } + + void SeekToFirst() override { + SeekToRestartPoint(0); + ParseNextKey(); + } + + void SeekToLast() override { + SeekToRestartPoint(num_restarts_ - 1); + while (ParseNextKey() && NextEntryOffset() < restarts_) { + // Keep skipping + } + } + + private: + void CorruptionError() { + current_ = restarts_; + restart_index_ = num_restarts_; + status_ = Status::Corruption("bad entry in block"); + key_.clear(); + value_.clear(); + } + + bool ParseNextKey() { + current_ = NextEntryOffset(); + const char* p = data_ + current_; + const char* limit = data_ + restarts_; // Restarts come right after data + if (p >= limit) { + // No more entries to return. Mark as invalid. + current_ = restarts_; + restart_index_ = num_restarts_; + return false; + } + + // Decode next entry + uint32_t shared, non_shared, value_length; + p = DecodeEntry(p, limit, &shared, &non_shared, &value_length); + if (p == nullptr || key_.size() < shared) { + CorruptionError(); + return false; + } else { + key_.resize(shared); + key_.append(p, non_shared); + value_ = Slice(p + non_shared, value_length); + while (restart_index_ + 1 < num_restarts_ && + GetRestartPoint(restart_index_ + 1) < current_) { + ++restart_index_; + } + return true; + } + } +}; + +Iterator* Block::NewIterator(const Comparator* comparator) { + if (size_ < sizeof(uint32_t)) { + return NewErrorIterator(Status::Corruption("bad block contents")); + } + const uint32_t num_restarts = NumRestarts(); + if (num_restarts == 0) { + return NewEmptyIterator(); + } else { + return new Iter(comparator, data_, restart_offset_, num_restarts); + } +} + +} // namespace leveldb diff --git a/table/block.h b/table/block.h new file mode 100644 index 0000000..5224108 --- /dev/null +++ b/table/block.h @@ -0,0 +1,44 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_TABLE_BLOCK_H_ +#define STORAGE_LEVELDB_TABLE_BLOCK_H_ + +#include <cstddef> +#include <cstdint> + +#include "leveldb/iterator.h" + +namespace leveldb { + +struct BlockContents; +class Comparator; + +class Block { + public: + // 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_[] +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_BLOCK_H_ diff --git a/table/block_builder.cc b/table/block_builder.cc new file mode 100644 index 0000000..37d4008 --- /dev/null +++ b/table/block_builder.cc @@ -0,0 +1,107 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// BlockBuilder generates blocks where keys are prefix-compressed: +// +// When we store a key, we drop the prefix shared with the previous +// string. This helps reduce the space requirement significantly. +// Furthermore, once every K keys, we do not apply the prefix +// compression and store the entire key. We call this a "restart +// point". The tail end of the block stores the offsets of all of the +// restart points, and can be used to do a binary search when looking +// for a particular key. Values are stored as-is (without compression) +// immediately following the corresponding key. +// +// An entry for a particular key-value pair has the form: +// shared_bytes: varint32 +// unshared_bytes: varint32 +// value_length: varint32 +// key_delta: char[unshared_bytes] +// value: char[value_length] +// shared_bytes == 0 for restart points. +// +// The trailer of the block has the form: +// restarts: uint32[num_restarts] +// num_restarts: uint32 +// restarts[i] contains the offset within the block of the ith restart point. + +#include "table/block_builder.h" + +#include <algorithm> +#include <cassert> + +#include "leveldb/comparator.h" +#include "leveldb/options.h" +#include "util/coding.h" + +namespace leveldb { + +BlockBuilder::BlockBuilder(const Options* options) + : options_(options), restarts_(), counter_(0), finished_(false) { + assert(options->block_restart_interval >= 1); + restarts_.push_back(0); // First restart point is at offset 0 +} + +void BlockBuilder::Reset() { + buffer_.clear(); + restarts_.clear(); + restarts_.push_back(0); // First restart point is at offset 0 + counter_ = 0; + finished_ = false; + last_key_.clear(); +} + +size_t BlockBuilder::CurrentSizeEstimate() const { + return (buffer_.size() + // Raw data buffer + restarts_.size() * sizeof(uint32_t) + // Restart array + sizeof(uint32_t)); // Restart array length +} + +Slice BlockBuilder::Finish() { + // Append restart array + for (size_t i = 0; i < restarts_.size(); i++) { + PutFixed32(&buffer_, restarts_[i]); + } + PutFixed32(&buffer_, restarts_.size()); + finished_ = true; + return Slice(buffer_); +} + +void BlockBuilder::Add(const Slice& key, const Slice& value) { + Slice last_key_piece(last_key_); + assert(!finished_); + assert(counter_ <= options_->block_restart_interval); + assert(buffer_.empty() // No values yet? + || options_->comparator->Compare(key, last_key_piece) > 0); + size_t shared = 0; + if (counter_ < options_->block_restart_interval) { + // See how much sharing to do with previous string + const size_t min_length = std::min(last_key_piece.size(), key.size()); + while ((shared < min_length) && (last_key_piece[shared] == key[shared])) { + shared++; + } + } else { + // Restart compression + restarts_.push_back(buffer_.size()); + counter_ = 0; + } + const size_t non_shared = key.size() - shared; + + // Add "<shared><non_shared><value_size>" to buffer_ + PutVarint32(&buffer_, shared); + PutVarint32(&buffer_, non_shared); + PutVarint32(&buffer_, value.size()); + + // Add string delta to buffer_ followed by value + buffer_.append(key.data() + shared, non_shared); + buffer_.append(value.data(), value.size()); + + // Update state + last_key_.resize(shared); + last_key_.append(key.data() + shared, non_shared); + assert(Slice(last_key_) == key); + counter_++; +} + +} // namespace leveldb diff --git a/table/block_builder.h b/table/block_builder.h new file mode 100644 index 0000000..7a481cd --- /dev/null +++ b/table/block_builder.h @@ -0,0 +1,54 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_TABLE_BLOCK_BUILDER_H_ +#define STORAGE_LEVELDB_TABLE_BLOCK_BUILDER_H_ + +#include <cstdint> +#include <vector> + +#include "leveldb/slice.h" + +namespace leveldb { + +struct Options; + +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(); + + // REQUIRES: Finish() has not been called since the last call to Reset(). + // REQUIRES: key is larger than any previously added key + void Add(const Slice& key, const Slice& value); + + // Finish building the block and return a slice that refers to the + // block contents. The returned slice will remain valid for the + // lifetime of this builder or until Reset() is called. + Slice Finish(); + + // Returns an estimate of the current (uncompressed) size of the block + // we are building. + size_t CurrentSizeEstimate() const; + + // Return true iff no entries have been added since the last Reset() + bool empty() const { return buffer_.empty(); } + + private: + const Options* options_; + std::string buffer_; // Destination buffer + std::vector<uint32_t> restarts_; // Restart points + int counter_; // Number of entries emitted since restart + bool finished_; // Has Finish() been called? + std::string last_key_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_BLOCK_BUILDER_H_ diff --git a/table/filter_block.cc b/table/filter_block.cc new file mode 100644 index 0000000..09ec009 --- /dev/null +++ b/table/filter_block.cc @@ -0,0 +1,106 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "table/filter_block.h" + +#include "leveldb/filter_policy.h" +#include "util/coding.h" + +namespace leveldb { + +// 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) {} + +void FilterBlockBuilder::StartBlock(uint64_t block_offset) { + uint64_t filter_index = (block_offset / kFilterBase); + assert(filter_index >= filter_offsets_.size()); + while (filter_index > filter_offsets_.size()) { + GenerateFilter(); + } +} + +void FilterBlockBuilder::AddKey(const Slice& key) { + Slice k = key; + start_.push_back(keys_.size()); + keys_.append(k.data(), k.size()); +} + +Slice FilterBlockBuilder::Finish() { + if (!start_.empty()) { + GenerateFilter(); + } + + // Append array of per-filter offsets + const uint32_t array_offset = result_.size(); + for (size_t i = 0; i < filter_offsets_.size(); i++) { + PutFixed32(&result_, filter_offsets_[i]); + } + + PutFixed32(&result_, array_offset); + result_.push_back(kFilterBaseLg); // Save encoding parameter in result + return Slice(result_); +} + +void FilterBlockBuilder::GenerateFilter() { + const size_t num_keys = start_.size(); + if (num_keys == 0) { + // Fast path if there are no keys for this filter + filter_offsets_.push_back(result_.size()); + return; + } + + // Make list of keys from flattened key structure + start_.push_back(keys_.size()); // Simplify length computation + 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]; + 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], static_cast<int>(num_keys), &result_); + + tmp_keys_.clear(); + keys_.clear(); + start_.clear(); +} + +FilterBlockReader::FilterBlockReader(const FilterPolicy* policy, + const Slice& contents) + : 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]; + uint32_t last_word = DecodeFixed32(contents.data() + n - 5); + if (last_word > n - 5) return; + data_ = contents.data(); + offset_ = data_ + last_word; + num_ = (n - 5 - last_word) / 4; +} + +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 <= static_cast<size_t>(offset_ - data_)) { + Slice filter = Slice(data_ + start, limit - start); + return policy_->KeyMayMatch(key, filter); + } else if (start == limit) { + // Empty filters do not match any keys + return false; + } + } + return true; // Errors are treated as potential matches +} + +} // namespace leveldb diff --git a/table/filter_block.h b/table/filter_block.h new file mode 100644 index 0000000..25ab75b --- /dev/null +++ b/table/filter_block.h @@ -0,0 +1,68 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// A filter block is stored near the end of a Table file. It contains +// filters (e.g., bloom filters) for all data blocks in the table combined +// into a single filter block. + +#ifndef STORAGE_LEVELDB_TABLE_FILTER_BLOCK_H_ +#define STORAGE_LEVELDB_TABLE_FILTER_BLOCK_H_ + +#include <cstddef> +#include <cstdint> +#include <string> +#include <vector> + +#include "leveldb/slice.h" +#include "util/hash.h" + +namespace leveldb { + +class FilterPolicy; + +// A FilterBlockBuilder is used to construct all of the filters for a +// particular Table. It generates a single string which is stored as +// a special block in the Table. +// +// The sequence of calls to FilterBlockBuilder must match the regexp: +// (StartBlock AddKey*)* Finish +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(); + + private: + void GenerateFilter(); + + const FilterPolicy* policy_; + std::string keys_; // Flattened key contents + std::vector<size_t> start_; // Starting index in keys_ of each key + std::string result_; // Filter data computed so far + std::vector<Slice> tmp_keys_; // policy_->CreateFilter() argument + std::vector<uint32_t> filter_offsets_; +}; + +class FilterBlockReader { + public: + // 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); + + private: + const FilterPolicy* policy_; + const char* data_; // Pointer to filter data (at block-start) + const char* offset_; // Pointer to beginning of offset array (at block-end) + size_t num_; // Number of entries in offset array + size_t base_lg_; // Encoding parameter (see kFilterBaseLg in .cc file) +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_FILTER_BLOCK_H_ diff --git a/table/filter_block_test.cc b/table/filter_block_test.cc new file mode 100644 index 0000000..e76479e --- /dev/null +++ b/table/filter_block_test.cc @@ -0,0 +1,123 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "table/filter_block.h" + +#include "gtest/gtest.h" +#include "leveldb/filter_policy.h" +#include "util/coding.h" +#include "util/hash.h" +#include "util/logging.h" +#include "util/testutil.h" + +namespace leveldb { + +// For testing: emit an array with one hash value per key +class TestHashFilter : public FilterPolicy { + public: + const char* Name() const override { return "TestHashFilter"; } + + 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); + } + } + + 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)) { + return true; + } + } + return false; + } +}; + +class FilterBlockTest : public testing::Test { + public: + TestHashFilter policy_; +}; + +TEST_F(FilterBlockTest, EmptyBuilder) { + FilterBlockBuilder builder(&policy_); + Slice block = builder.Finish(); + ASSERT_EQ("\\x00\\x00\\x00\\x00\\x0b", EscapeString(block)); + FilterBlockReader reader(&policy_, block); + ASSERT_TRUE(reader.KeyMayMatch(0, "foo")); + ASSERT_TRUE(reader.KeyMayMatch(100000, "foo")); +} + +TEST_F(FilterBlockTest, SingleChunk) { + FilterBlockBuilder builder(&policy_); + builder.StartBlock(100); + builder.AddKey("foo"); + builder.AddKey("bar"); + builder.AddKey("box"); + builder.StartBlock(200); + builder.AddKey("box"); + builder.StartBlock(300); + builder.AddKey("hello"); + Slice block = builder.Finish(); + FilterBlockReader reader(&policy_, block); + ASSERT_TRUE(reader.KeyMayMatch(100, "foo")); + ASSERT_TRUE(reader.KeyMayMatch(100, "bar")); + 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")); +} + +TEST_F(FilterBlockTest, MultiChunk) { + FilterBlockBuilder builder(&policy_); + + // First filter + builder.StartBlock(0); + builder.AddKey("foo"); + builder.StartBlock(2000); + builder.AddKey("bar"); + + // Second filter + builder.StartBlock(3100); + builder.AddKey("box"); + + // Third filter is empty + + // Last filter + builder.StartBlock(9000); + builder.AddKey("box"); + builder.AddKey("hello"); + + Slice block = builder.Finish(); + FilterBlockReader reader(&policy_, block); + + // 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")); + + // 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")); + + // 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")); + + // 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")); +} + +} // namespace leveldb + diff --git a/table/format.cc b/table/format.cc new file mode 100644 index 0000000..e183977 --- /dev/null +++ b/table/format.cc @@ -0,0 +1,141 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "table/format.h" + +#include "leveldb/env.h" +#include "port/port.h" +#include "table/block.h" +#include "util/coding.h" +#include "util/crc32c.h" + +namespace leveldb { + +void BlockHandle::EncodeTo(std::string* dst) const { + // Sanity check that all fields have been set + assert(offset_ != ~static_cast<uint64_t>(0)); + assert(size_ != ~static_cast<uint64_t>(0)); + PutVarint64(dst, offset_); + PutVarint64(dst, size_); +} + +Status BlockHandle::DecodeFrom(Slice* input) { + if (GetVarint64(input, &offset_) && GetVarint64(input, &size_)) { + return Status::OK(); + } else { + return Status::Corruption("bad block handle"); + } +} + +void Footer::EncodeTo(std::string* dst) const { + const size_t original_size = dst->size(); + metaindex_handle_.EncodeTo(dst); + index_handle_.EncodeTo(dst); + dst->resize(2 * BlockHandle::kMaxEncodedLength); // Padding + PutFixed32(dst, static_cast<uint32_t>(kTableMagicNumber & 0xffffffffu)); + PutFixed32(dst, static_cast<uint32_t>(kTableMagicNumber >> 32)); + assert(dst->size() == original_size + kEncodedLength); + (void)original_size; // Disable unused variable warning. +} + +Status Footer::DecodeFrom(Slice* input) { + const char* magic_ptr = input->data() + kEncodedLength - 8; + const uint32_t magic_lo = DecodeFixed32(magic_ptr); + const uint32_t magic_hi = DecodeFixed32(magic_ptr + 4); + const uint64_t magic = ((static_cast<uint64_t>(magic_hi) << 32) | + (static_cast<uint64_t>(magic_lo))); + if (magic != kTableMagicNumber) { + return Status::Corruption("not an sstable (bad magic number)"); + } + + Status result = metaindex_handle_.DecodeFrom(input); + if (result.ok()) { + result = index_handle_.DecodeFrom(input); + } + if (result.ok()) { + // We skip over any leftover data (just padding for now) in "input" + const char* end = magic_ptr + 8; + *input = Slice(end, input->data() + input->size() - end); + } + return result; +} + +Status ReadBlock(RandomAccessFile* file, const ReadOptions& options, + const BlockHandle& handle, BlockContents* result) { + result->data = Slice(); + result->cachable = false; + result->heap_allocated = false; + + // Read the block contents as well as the type/crc footer. + // See table_builder.cc for the code that built this structure. + size_t n = static_cast<size_t>(handle.size()); + char* buf = new char[n + kBlockTrailerSize]; + Slice contents; + Status s = file->Read(handle.offset(), n + kBlockTrailerSize, &contents, buf); + if (!s.ok()) { + delete[] buf; + return s; + } + if (contents.size() != n + kBlockTrailerSize) { + delete[] buf; + return Status::Corruption("truncated block read"); + } + + // Check the crc of the type and the block contents + const char* data = contents.data(); // Pointer to where Read put the data + if (options.verify_checksums) { + const uint32_t crc = crc32c::Unmask(DecodeFixed32(data + n + 1)); + const uint32_t actual = crc32c::Value(data, n + 1); + if (actual != crc) { + delete[] buf; + s = Status::Corruption("block checksum mismatch"); + return s; + } + } + + switch (data[n]) { + case kNoCompression: + if (data != buf) { + // File implementation gave us pointer to some other data. + // Use it directly under the assumption that it will be live + // while the file is open. + delete[] buf; + result->data = Slice(data, n); + result->heap_allocated = false; + result->cachable = false; // Do not double-cache + } else { + result->data = Slice(buf, n); + result->heap_allocated = true; + result->cachable = true; + } + + // Ok + break; + case kSnappyCompression: { + size_t ulength = 0; + if (!port::Snappy_GetUncompressedLength(data, n, &ulength)) { + delete[] buf; + return Status::Corruption("corrupted compressed block contents"); + } + char* ubuf = new char[ulength]; + if (!port::Snappy_Uncompress(data, n, ubuf)) { + delete[] buf; + delete[] ubuf; + return Status::Corruption("corrupted compressed block contents"); + } + delete[] buf; + result->data = Slice(ubuf, ulength); + result->heap_allocated = true; + result->cachable = true; + break; + } + default: + delete[] buf; + return Status::Corruption("bad block type"); + } + + return Status::OK(); +} + +} // namespace leveldb diff --git a/table/format.h b/table/format.h new file mode 100644 index 0000000..f6ea304 --- /dev/null +++ b/table/format.h @@ -0,0 +1,99 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_TABLE_FORMAT_H_ +#define STORAGE_LEVELDB_TABLE_FORMAT_H_ + +#include <cstdint> +#include <string> + +#include "leveldb/slice.h" +#include "leveldb/status.h" +#include "leveldb/table_builder.h" + +namespace leveldb { + +class Block; +class RandomAccessFile; +struct ReadOptions; + +// BlockHandle is a pointer to the extent of a file that stores a data +// 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. + uint64_t offset() const { return offset_; } + void set_offset(uint64_t offset) { offset_ = offset; } + + // The size of the stored block + uint64_t size() const { return size_; } + void set_size(uint64_t size) { size_ = size; } + + void EncodeTo(std::string* dst) const; + Status DecodeFrom(Slice* input); + + private: + uint64_t offset_; + uint64_t size_; +}; + +// Footer encapsulates the fixed information stored at the tail +// end of every table file. +class Footer { + public: + // 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; } + + void EncodeTo(std::string* dst) const; + Status DecodeFrom(Slice* input); + + private: + BlockHandle metaindex_handle_; + BlockHandle index_handle_; +}; + +// kTableMagicNumber was picked by running +// echo http://code.google.com/p/leveldb/ | sha1sum +// and taking the leading 64 bits. +static const uint64_t kTableMagicNumber = 0xdb4775248b80fb57ull; + +// 1-byte type + 32-bit crc +static const size_t kBlockTrailerSize = 5; + +struct BlockContents { + Slice data; // Actual contents of data + bool cachable; // True iff data can be cached + bool heap_allocated; // True iff caller should delete[] data.data() +}; + +// Read the block identified by "handle" from "file". On failure +// return non-OK. On success fill *result and return OK. +Status ReadBlock(RandomAccessFile* file, const ReadOptions& options, + const BlockHandle& handle, BlockContents* result); + +// Implementation details follow. Clients should ignore, + +inline BlockHandle::BlockHandle() + : offset_(~static_cast<uint64_t>(0)), size_(~static_cast<uint64_t>(0)) {} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_FORMAT_H_ diff --git a/table/iterator.cc b/table/iterator.cc new file mode 100644 index 0000000..dfef083 --- /dev/null +++ b/table/iterator.cc @@ -0,0 +1,76 @@ +// 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 "leveldb/iterator.h" + +namespace leveldb { + +Iterator::Iterator() { + cleanup_head_.function = nullptr; + cleanup_head_.next = nullptr; +} + +Iterator::~Iterator() { + 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 != nullptr); + CleanupNode* node; + if (cleanup_head_.IsEmpty()) { + node = &cleanup_head_; + } else { + node = new CleanupNode(); + node->next = cleanup_head_.next; + cleanup_head_.next = node; + } + node->function = func; + node->arg1 = arg1; + node->arg2 = arg2; +} + +namespace { + +class EmptyIterator : public Iterator { + public: + 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_; +}; + +} // anonymous namespace + +Iterator* NewEmptyIterator() { return new EmptyIterator(Status::OK()); } + +Iterator* NewErrorIterator(const Status& status) { + return new EmptyIterator(status); +} + +} // namespace leveldb diff --git a/table/iterator_wrapper.h b/table/iterator_wrapper.h new file mode 100644 index 0000000..c230572 --- /dev/null +++ b/table/iterator_wrapper.h @@ -0,0 +1,92 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_TABLE_ITERATOR_WRAPPER_H_ +#define STORAGE_LEVELDB_TABLE_ITERATOR_WRAPPER_H_ + +#include "leveldb/iterator.h" +#include "leveldb/slice.h" + +namespace leveldb { + +// A internal wrapper class with an interface similar to Iterator that +// caches the valid() and key() results for an underlying iterator. +// This can help avoid virtual function calls and also gives better +// cache locality. +class IteratorWrapper { + public: + IteratorWrapper() : iter_(nullptr), valid_(false) {} + explicit IteratorWrapper(Iterator* iter) : iter_(nullptr) { Set(iter); } + ~IteratorWrapper() { delete iter_; } + Iterator* iter() const { return iter_; } + + // Takes ownership of "iter" and will delete it when destroyed, or + // when Set() is invoked again. + void Set(Iterator* iter) { + delete iter_; + iter_ = iter; + 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() != nullptr + Status status() const { + assert(iter_); + return iter_->status(); + } + void Next() { + assert(iter_); + iter_->Next(); + Update(); + } + void Prev() { + assert(iter_); + iter_->Prev(); + Update(); + } + void Seek(const Slice& k) { + assert(iter_); + iter_->Seek(k); + Update(); + } + void SeekToFirst() { + assert(iter_); + iter_->SeekToFirst(); + Update(); + } + void SeekToLast() { + assert(iter_); + iter_->SeekToLast(); + Update(); + } + + private: + void Update() { + valid_ = iter_->Valid(); + if (valid_) { + key_ = iter_->key(); + } + } + + Iterator* iter_; + bool valid_; + Slice key_; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_ITERATOR_WRAPPER_H_ diff --git a/table/merger.cc b/table/merger.cc new file mode 100644 index 0000000..76441b1 --- /dev/null +++ b/table/merger.cc @@ -0,0 +1,191 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "table/merger.h" + +#include "leveldb/comparator.h" +#include "leveldb/iterator.h" +#include "table/iterator_wrapper.h" + +namespace leveldb { + +namespace { +class MergingIterator : public Iterator { + public: + MergingIterator(const Comparator* comparator, Iterator** children, int n) + : comparator_(comparator), + children_(new IteratorWrapper[n]), + n_(n), + current_(nullptr), + direction_(kForward) { + for (int i = 0; i < n; i++) { + children_[i].Set(children[i]); + } + } + + ~MergingIterator() override { delete[] children_; } + + bool Valid() const override { return (current_ != nullptr); } + + void SeekToFirst() override { + for (int i = 0; i < n_; i++) { + children_[i].SeekToFirst(); + } + FindSmallest(); + direction_ = kForward; + } + + void SeekToLast() override { + for (int i = 0; i < n_; i++) { + children_[i].SeekToLast(); + } + FindLargest(); + direction_ = kReverse; + } + + void Seek(const Slice& target) override { + for (int i = 0; i < n_; i++) { + children_[i].Seek(target); + } + FindSmallest(); + direction_ = kForward; + } + + void Next() override { + assert(Valid()); + + // Ensure that all children are positioned after key(). + // If we are moving in the forward direction, it is already + // true for all of the non-current_ children since current_ is + // the smallest child and key() == current_->key(). Otherwise, + // we explicitly position the non-current_ children. + if (direction_ != kForward) { + for (int i = 0; i < n_; i++) { + IteratorWrapper* child = &children_[i]; + if (child != current_) { + child->Seek(key()); + if (child->Valid() && + comparator_->Compare(key(), child->key()) == 0) { + child->Next(); + } + } + } + direction_ = kForward; + } + + current_->Next(); + FindSmallest(); + } + + void Prev() override { + assert(Valid()); + + // Ensure that all children are positioned before key(). + // If we are moving in the reverse direction, it is already + // true for all of the non-current_ children since current_ is + // the largest child and key() == current_->key(). Otherwise, + // we explicitly position the non-current_ children. + if (direction_ != kReverse) { + for (int i = 0; i < n_; i++) { + IteratorWrapper* child = &children_[i]; + if (child != current_) { + child->Seek(key()); + if (child->Valid()) { + // Child is at first entry >= key(). Step back one to be < key() + child->Prev(); + } else { + // Child has no entries >= key(). Position at last entry. + child->SeekToLast(); + } + } + } + direction_ = kReverse; + } + + current_->Prev(); + FindLargest(); + } + + Slice key() const override { + assert(Valid()); + return current_->key(); + } + + Slice value() const override { + assert(Valid()); + return current_->value(); + } + + Status status() const override { + Status status; + for (int i = 0; i < n_; i++) { + status = children_[i].status(); + if (!status.ok()) { + break; + } + } + return status; + } + + private: + // Which direction is the iterator moving? + enum Direction { kForward, kReverse }; + + void FindSmallest(); + void FindLargest(); + + // We might want to use a heap in case there are lots of children. + // For now we use a simple array since we expect a very small number + // of children in leveldb. + const Comparator* comparator_; + IteratorWrapper* children_; + int n_; + IteratorWrapper* current_; + Direction direction_; +}; + +void MergingIterator::FindSmallest() { + IteratorWrapper* smallest = nullptr; + for (int i = 0; i < n_; i++) { + IteratorWrapper* child = &children_[i]; + if (child->Valid()) { + if (smallest == nullptr) { + smallest = child; + } else if (comparator_->Compare(child->key(), smallest->key()) < 0) { + smallest = child; + } + } + } + current_ = smallest; +} + +void MergingIterator::FindLargest() { + IteratorWrapper* largest = nullptr; + for (int i = n_ - 1; i >= 0; i--) { + IteratorWrapper* child = &children_[i]; + if (child->Valid()) { + if (largest == nullptr) { + largest = child; + } else if (comparator_->Compare(child->key(), largest->key()) > 0) { + largest = child; + } + } + } + current_ = largest; +} +} // namespace + +Iterator* NewMergingIterator(const Comparator* comparator, Iterator** children, + int n) { + assert(n >= 0); + if (n == 0) { + return NewEmptyIterator(); + } else if (n == 1) { + return children[0]; + } else { + return new MergingIterator(comparator, children, n); + } +} + +} // namespace leveldb diff --git a/table/merger.h b/table/merger.h new file mode 100644 index 0000000..41cedc5 --- /dev/null +++ b/table/merger.h @@ -0,0 +1,26 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_TABLE_MERGER_H_ +#define STORAGE_LEVELDB_TABLE_MERGER_H_ + +namespace leveldb { + +class Comparator; +class Iterator; + +// Return an iterator that provided the union of the data in +// children[0,n-1]. Takes ownership of the child iterators and +// will delete them when the result iterator is deleted. +// +// The result does no duplicate suppression. I.e., if a particular +// key is present in K child iterators, it will be yielded K times. +// +// REQUIRES: n >= 0 +Iterator* NewMergingIterator(const Comparator* comparator, Iterator** children, + int n); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_MERGER_H_ diff --git a/table/table.cc b/table/table.cc new file mode 100644 index 0000000..29e835f --- /dev/null +++ b/table/table.cc @@ -0,0 +1,271 @@ +// 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 "leveldb/table.h" + +#include "leveldb/cache.h" +#include "leveldb/comparator.h" +#include "leveldb/env.h" +#include "leveldb/filter_policy.h" +#include "leveldb/options.h" +#include "table/block.h" +#include "table/filter_block.h" +#include "table/format.h" +#include "table/two_level_iterator.h" +#include "util/coding.h" + +namespace leveldb { + +struct Table::Rep { + ~Rep() { + delete filter; + delete[] filter_data; + delete index_block; + } + + Options options; + Status status; + RandomAccessFile* file; + uint64_t cache_id; + FilterBlockReader* filter; + const char* filter_data; + + BlockHandle metaindex_handle; // Handle to metaindex_block: saved from footer + Block* index_block; +}; + +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"); + } + + char footer_space[Footer::kEncodedLength]; + Slice footer_input; + Status s = file->Read(size - Footer::kEncodedLength, Footer::kEncodedLength, + &footer_input, footer_space); + if (!s.ok()) return s; + + Footer footer; + s = footer.DecodeFrom(&footer_input); + if (!s.ok()) return s; + + // Read the index block + BlockContents index_block_contents; + ReadOptions opt; + if (options.paranoid_checks) { + opt.verify_checksums = true; + } + 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 = nullptr; + rep->filter = nullptr; + *table = new Table(rep); + (*table)->ReadMeta(footer); + } + + return s; +} + +void Table::ReadMeta(const Footer& footer) { + if (rep_->options.filter_policy == nullptr) { + return; // Do not need any metadata + } + + // TODO(sanjay): Skip this if footer.metaindex_handle() size indicates + // it is an empty block. + ReadOptions opt; + if (rep_->options.paranoid_checks) { + opt.verify_checksums = true; + } + BlockContents contents; + if (!ReadBlock(rep_->file, opt, footer.metaindex_handle(), &contents).ok()) { + // Do not propagate errors since meta info is not needed for operation + return; + } + Block* meta = new Block(contents); + + Iterator* iter = meta->NewIterator(BytewiseComparator()); + std::string key = "filter."; + key.append(rep_->options.filter_policy->Name()); + iter->Seek(key); + if (iter->Valid() && iter->key() == Slice(key)) { + ReadFilter(iter->value()); + } + delete iter; + delete meta; +} + +void Table::ReadFilter(const Slice& filter_handle_value) { + Slice v = filter_handle_value; + BlockHandle filter_handle; + if (!filter_handle.DecodeFrom(&v).ok()) { + return; + } + + // We might want to unify with ReadBlock() if we start + // requiring checksum verification in Table::Open. + ReadOptions opt; + if (rep_->options.paranoid_checks) { + opt.verify_checksums = true; + } + BlockContents block; + if (!ReadBlock(rep_->file, opt, filter_handle, &block).ok()) { + return; + } + if (block.heap_allocated) { + 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_; } + +static void DeleteBlock(void* arg, void* ignored) { + delete reinterpret_cast<Block*>(arg); +} + +static void DeleteCachedBlock(const Slice& key, void* value) { + Block* block = reinterpret_cast<Block*>(value); + delete block; +} + +static void ReleaseBlock(void* arg, void* h) { + Cache* cache = reinterpret_cast<Cache*>(arg); + Cache::Handle* handle = reinterpret_cast<Cache::Handle*>(h); + cache->Release(handle); +} + +// 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, + const Slice& index_value) { + Table* table = reinterpret_cast<Table*>(arg); + Cache* block_cache = table->rep_->options.block_cache; + Block* block = nullptr; + Cache::Handle* cache_handle = nullptr; + + BlockHandle handle; + Slice input = index_value; + Status s = handle.DecodeFrom(&input); + // We intentionally allow extra stuff in index_value so that we + // can add more features in the future. + + if (s.ok()) { + BlockContents contents; + if (block_cache != nullptr) { + char cache_key_buffer[16]; + EncodeFixed64(cache_key_buffer, table->rep_->cache_id); + 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 != nullptr) { + block = reinterpret_cast<Block*>(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); + } + } + } + } else { + s = ReadBlock(table->rep_->file, options, handle, &contents); + if (s.ok()) { + block = new Block(contents); + } + } + } + + Iterator* iter; + if (block != nullptr) { + iter = block->NewIterator(table->rep_->options.comparator); + if (cache_handle == nullptr) { + iter->RegisterCleanup(&DeleteBlock, block, nullptr); + } else { + iter->RegisterCleanup(&ReleaseBlock, block_cache, cache_handle); + } + } else { + iter = NewErrorIterator(s); + } + return iter; +} + +Iterator* Table::NewIterator(const ReadOptions& options) const { + return NewTwoLevelIterator( + rep_->index_block->NewIterator(rep_->options.comparator), + &Table::BlockReader, const_cast<Table*>(this), options); +} + +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); + if (iiter->Valid()) { + Slice handle_value = iiter->value(); + FilterBlockReader* filter = rep_->filter; + BlockHandle handle; + 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()) { + (*handle_result)(arg, block_iter->key(), block_iter->value()); + } + s = block_iter->status(); + delete block_iter; + } + } + if (s.ok()) { + s = iiter->status(); + } + delete iiter; + return s; +} + +uint64_t Table::ApproximateOffsetOf(const Slice& key) const { + Iterator* index_iter = + rep_->index_block->NewIterator(rep_->options.comparator); + index_iter->Seek(key); + uint64_t result; + if (index_iter->Valid()) { + BlockHandle handle; + Slice input = index_iter->value(); + Status s = handle.DecodeFrom(&input); + if (s.ok()) { + result = handle.offset(); + } else { + // Strange: we can't decode the block handle in the index block. + // We'll just return the offset of the metaindex block, which is + // close to the whole file size for this case. + result = rep_->metaindex_handle.offset(); + } + } else { + // key is past the last key in the file. Approximate the offset + // by returning the offset of the metaindex block (which is + // right near the end of the file). + result = rep_->metaindex_handle.offset(); + } + delete index_iter; + return result; +} + +} // namespace leveldb diff --git a/table/table_builder.cc b/table/table_builder.cc new file mode 100644 index 0000000..29a619d --- /dev/null +++ b/table/table_builder.cc @@ -0,0 +1,265 @@ +// 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 "leveldb/table_builder.h" + +#include <cassert> + +#include "leveldb/comparator.h" +#include "leveldb/env.h" +#include "leveldb/filter_policy.h" +#include "leveldb/options.h" +#include "table/block_builder.h" +#include "table/filter_block.h" +#include "table/format.h" +#include "util/coding.h" +#include "util/crc32c.h" + +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; + uint64_t offset; + Status status; + BlockBuilder data_block; + BlockBuilder index_block; + std::string last_key; + int64_t num_entries; + 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 + // first key for the next data block. This allows us to use shorter + // keys in the index block. For example, consider a block boundary + // between the keys "the quick brown fox" and "the who". We can use + // "the r" as the key for the index block entry since it is >= all + // entries in the first block and < all entries in subsequent + // blocks. + // + // Invariant: r->pending_index_entry is true only if data_block is empty. + bool pending_index_entry; + BlockHandle pending_handle; // Handle to add to index block + + std::string compressed_output; +}; + +TableBuilder::TableBuilder(const Options& options, WritableFile* file) + : rep_(new Rep(options, file)) { + if (rep_->filter_block != nullptr) { + rep_->filter_block->StartBlock(0); + } +} + +TableBuilder::~TableBuilder() { + assert(rep_->closed); // Catch errors where caller forgot to call Finish() + delete rep_->filter_block; + delete rep_; +} + +Status TableBuilder::ChangeOptions(const Options& options) { + // Note: if more fields are added to Options, update + // this function to catch changes that should not be allowed to + // change in the middle of building a Table. + if (options.comparator != rep_->options.comparator) { + return Status::InvalidArgument("changing comparator while building table"); + } + + // Note that any live BlockBuilders point to rep_->options and therefore + // will automatically pick up the updated options. + rep_->options = options; + rep_->index_block_options = options; + rep_->index_block_options.block_restart_interval = 1; + return Status::OK(); +} + +void TableBuilder::Add(const Slice& key, const Slice& value) { + Rep* r = rep_; + assert(!r->closed); + if (!ok()) return; + if (r->num_entries > 0) { + assert(r->options.comparator->Compare(key, Slice(r->last_key)) > 0); + } + + if (r->pending_index_entry) { + assert(r->data_block.empty()); + r->options.comparator->FindShortestSeparator(&r->last_key, key); + std::string handle_encoding; + r->pending_handle.EncodeTo(&handle_encoding); + r->index_block.Add(r->last_key, Slice(handle_encoding)); + r->pending_index_entry = false; + } + + if (r->filter_block != nullptr) { + r->filter_block->AddKey(key); + } + + r->last_key.assign(key.data(), key.size()); + r->num_entries++; + r->data_block.Add(key, value); + + const size_t estimated_block_size = r->data_block.CurrentSizeEstimate(); + if (estimated_block_size >= r->options.block_size) { + Flush(); + } +} + +void TableBuilder::Flush() { + Rep* r = rep_; + assert(!r->closed); + if (!ok()) return; + if (r->data_block.empty()) return; + assert(!r->pending_index_entry); + WriteBlock(&r->data_block, &r->pending_handle); + if (ok()) { + r->pending_index_entry = true; + r->status = r->file->Flush(); + } + if (r->filter_block != nullptr) { + r->filter_block->StartBlock(r->offset); + } +} + +void TableBuilder::WriteBlock(BlockBuilder* block, BlockHandle* handle) { + // File format contains a sequence of blocks where each block has: + // block_data: uint8[n] + // type: uint8 + // crc: uint32 + assert(ok()); + Rep* r = rep_; + Slice raw = block->Finish(); + + Slice block_contents; + CompressionType type = r->options.compression; + // TODO(postrelease): Support more compression options: zlib? + switch (type) { + case kNoCompression: + block_contents = raw; + break; + + case kSnappyCompression: { + std::string* compressed = &r->compressed_output; + if (port::Snappy_Compress(raw.data(), raw.size(), compressed) && + compressed->size() < raw.size() - (raw.size() / 8u)) { + block_contents = *compressed; + } else { + // Snappy not supported, or compressed less than 12.5%, so just + // store uncompressed form + block_contents = raw; + type = kNoCompression; + } + break; + } + } + WriteRawBlock(block_contents, type, handle); + r->compressed_output.clear(); + block->Reset(); +} + +void TableBuilder::WriteRawBlock(const Slice& block_contents, + CompressionType type, BlockHandle* handle) { + Rep* r = rep_; + handle->set_offset(r->offset); + handle->set_size(block_contents.size()); + r->status = r->file->Append(block_contents); + if (r->status.ok()) { + char trailer[kBlockTrailerSize]; + trailer[0] = type; + 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)); + r->status = r->file->Append(Slice(trailer, kBlockTrailerSize)); + if (r->status.ok()) { + r->offset += block_contents.size() + kBlockTrailerSize; + } + } +} + +Status TableBuilder::status() const { return rep_->status; } + +Status TableBuilder::Finish() { + Rep* r = rep_; + Flush(); + assert(!r->closed); + r->closed = true; + + BlockHandle filter_block_handle, metaindex_block_handle, index_block_handle; + + // Write filter block + if (ok() && r->filter_block != nullptr) { + WriteRawBlock(r->filter_block->Finish(), kNoCompression, + &filter_block_handle); + } + + // Write metaindex block + if (ok()) { + BlockBuilder meta_index_block(&r->options); + 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()); + std::string handle_encoding; + filter_block_handle.EncodeTo(&handle_encoding); + meta_index_block.Add(key, handle_encoding); + } + + // TODO(postrelease): Add stats and other meta blocks + WriteBlock(&meta_index_block, &metaindex_block_handle); + } + + // Write index block + if (ok()) { + if (r->pending_index_entry) { + r->options.comparator->FindShortSuccessor(&r->last_key); + std::string handle_encoding; + r->pending_handle.EncodeTo(&handle_encoding); + r->index_block.Add(r->last_key, Slice(handle_encoding)); + r->pending_index_entry = false; + } + WriteBlock(&r->index_block, &index_block_handle); + } + + // Write footer + if (ok()) { + Footer footer; + footer.set_metaindex_handle(metaindex_block_handle); + footer.set_index_handle(index_block_handle); + std::string footer_encoding; + footer.EncodeTo(&footer_encoding); + r->status = r->file->Append(footer_encoding); + if (r->status.ok()) { + r->offset += footer_encoding.size(); + } + } + return r->status; +} + +void TableBuilder::Abandon() { + Rep* r = rep_; + assert(!r->closed); + r->closed = true; +} + +uint64_t TableBuilder::NumEntries() const { return rep_->num_entries; } + +uint64_t TableBuilder::FileSize() const { return rep_->offset; } + +} // namespace leveldb diff --git a/table/table_test.cc b/table/table_test.cc new file mode 100644 index 0000000..311839e --- /dev/null +++ b/table/table_test.cc @@ -0,0 +1,830 @@ +// 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 "leveldb/table.h" + +#include <map> +#include <string> + +#include "gtest/gtest.h" +#include "db/dbformat.h" +#include "db/memtable.h" +#include "db/write_batch_internal.h" +#include "leveldb/db.h" +#include "leveldb/env.h" +#include "leveldb/iterator.h" +#include "leveldb/table_builder.h" +#include "table/block.h" +#include "table/block_builder.h" +#include "table/format.h" +#include "util/random.h" +#include "util/testutil.h" + +namespace leveldb { + +// Return reverse of "key". +// Used to test non-lexicographic comparators. +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) { + rev.push_back(*rit); + } + return rev; +} + +namespace { +class ReverseKeyComparator : public Comparator { + public: + const char* Name() const override { + return "leveldb.ReverseBytewiseComparator"; + } + + int Compare(const Slice& a, const Slice& b) const override { + return BytewiseComparator()->Compare(Reverse(a), Reverse(b)); + } + + 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); + } + + void FindShortSuccessor(std::string* key) const override { + std::string s = Reverse(*key); + BytewiseComparator()->FindShortSuccessor(&s); + *key = Reverse(s); + } +}; +} // namespace +static ReverseKeyComparator reverse_key_comparator; + +static void Increment(const Comparator* cmp, std::string* key) { + if (cmp == BytewiseComparator()) { + key->push_back('\0'); + } else { + assert(cmp == &reverse_key_comparator); + std::string rev = Reverse(*key); + rev.push_back('\0'); + *key = Reverse(rev); + } +} + +// An STL comparator that uses a Comparator +namespace { +struct STLLessThan { + const Comparator* cmp; + + 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 { + public: + ~StringSink() override = default; + + const std::string& contents() const { return contents_; } + + Status Close() override { return Status::OK(); } + Status Flush() override { return Status::OK(); } + Status Sync() override { return Status::OK(); } + + Status Append(const Slice& data) override { + contents_.append(data.data(), data.size()); + return Status::OK(); + } + + private: + std::string contents_; +}; + +class StringSource : public RandomAccessFile { + public: + StringSource(const Slice& contents) + : contents_(contents.data(), contents.size()) {} + + ~StringSource() override = default; + + uint64_t Size() const { return 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()) { + n = contents_.size() - offset; + } + std::memcpy(scratch, &contents_[offset], n); + *result = Slice(scratch, n); + return Status::OK(); + } + + private: + std::string contents_; +}; + +typedef std::map<std::string, std::string, STLLessThan> KVMap; + +// Helper class for tests to unify the interface between +// BlockBuilder/TableBuilder and Block/Table. +class Constructor { + public: + explicit Constructor(const Comparator* cmp) : data_(STLLessThan(cmp)) {} + virtual ~Constructor() = default; + + void Add(const std::string& key, const Slice& value) { + data_[key] = value.ToString(); + } + + // Finish constructing the data structure with all the keys that have + // been added so far. Returns the keys in sorted order in "*keys" + // and stores the key/value pairs in "*kvmap" + void Finish(const Options& options, std::vector<std::string>* keys, + KVMap* kvmap) { + *kvmap = data_; + keys->clear(); + for (const auto& kvp : data_) { + keys->push_back(kvp.first); + } + data_.clear(); + Status s = FinishImpl(options, *kvmap); + ASSERT_TRUE(s.ok()) << s.ToString(); + } + + // Construct the data structure from the data in "data" + virtual Status FinishImpl(const Options& options, const KVMap& data) = 0; + + virtual Iterator* NewIterator() const = 0; + + const KVMap& data() const { return data_; } + + virtual DB* db() const { return nullptr; } // Overridden in DBConstructor + + private: + KVMap data_; +}; + +class BlockConstructor : public Constructor { + public: + explicit BlockConstructor(const Comparator* cmp) + : Constructor(cmp), comparator_(cmp), block_(nullptr) {} + ~BlockConstructor() override { delete block_; } + Status FinishImpl(const Options& options, const KVMap& data) override { + delete block_; + block_ = nullptr; + BlockBuilder builder(&options); + + for (const auto& kvp : data) { + builder.Add(kvp.first, kvp.second); + } + // Open the block + data_ = builder.Finish().ToString(); + BlockContents contents; + contents.data = data_; + contents.cachable = false; + contents.heap_allocated = false; + block_ = new Block(contents); + return Status::OK(); + } + Iterator* NewIterator() const override { + return block_->NewIterator(comparator_); + } + + private: + const Comparator* const comparator_; + std::string data_; + Block* block_; + + BlockConstructor(); +}; + +class TableConstructor : public Constructor { + public: + TableConstructor(const Comparator* cmp) + : 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 (const auto& kvp : data) { + builder.Add(kvp.first, kvp.second); + EXPECT_LEVELDB_OK(builder.status()); + } + Status s = builder.Finish(); + EXPECT_LEVELDB_OK(s); + + EXPECT_EQ(sink.contents().size(), builder.FileSize()); + + // Open the table + source_ = new StringSource(sink.contents()); + Options table_options; + table_options.comparator = options.comparator; + return Table::Open(table_options, source_, sink.contents().size(), &table_); + } + + Iterator* NewIterator() const override { + return table_->NewIterator(ReadOptions()); + } + + uint64_t ApproximateOffsetOf(const Slice& key) const { + return table_->ApproximateOffsetOf(key); + } + + private: + void Reset() { + delete table_; + delete source_; + table_ = nullptr; + source_ = nullptr; + } + + StringSource* source_; + Table* table_; + + TableConstructor(); +}; + +// A helper class that converts internal format keys into user keys +class KeyConvertingIterator : public Iterator { + public: + 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); + } + void SeekToFirst() override { iter_->SeekToFirst(); } + void SeekToLast() override { iter_->SeekToLast(); } + void Next() override { iter_->Next(); } + void Prev() override { iter_->Prev(); } + + Slice key() const override { + assert(Valid()); + ParsedInternalKey key; + if (!ParseInternalKey(iter_->key(), &key)) { + status_ = Status::Corruption("malformed internal key"); + return Slice("corrupted key"); + } + return key.user_key; + } + + Slice value() const override { return iter_->value(); } + Status status() const override { + return status_.ok() ? iter_->status() : status_; + } + + private: + mutable Status status_; + Iterator* iter_; +}; + +class MemTableConstructor : public Constructor { + public: + explicit MemTableConstructor(const Comparator* cmp) + : Constructor(cmp), internal_comparator_(cmp) { + memtable_ = new MemTable(internal_comparator_); + memtable_->Ref(); + } + ~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 (const auto& kvp : data) { + memtable_->Add(seq, kTypeValue, kvp.first, kvp.second); + seq++; + } + return Status::OK(); + } + Iterator* NewIterator() const override { + return new KeyConvertingIterator(memtable_->NewIterator()); + } + + private: + const InternalKeyComparator internal_comparator_; + MemTable* memtable_; +}; + +class DBConstructor : public Constructor { + public: + explicit DBConstructor(const Comparator* cmp) + : Constructor(cmp), comparator_(cmp) { + db_ = nullptr; + NewDB(); + } + ~DBConstructor() override { delete db_; } + Status FinishImpl(const Options& options, const KVMap& data) override { + delete db_; + db_ = nullptr; + NewDB(); + for (const auto& kvp : data) { + WriteBatch batch; + batch.Put(kvp.first, kvp.second); + EXPECT_TRUE(db_->Write(WriteOptions(), &batch).ok()); + } + return Status::OK(); + } + Iterator* NewIterator() const override { + return db_->NewIterator(ReadOptions()); + } + + DB* db() const override { return db_; } + + private: + void NewDB() { + std::string name = testing::TempDir() + "table_testdb"; + + Options options; + options.comparator = comparator_; + Status status = DestroyDB(name, options); + ASSERT_TRUE(status.ok()) << status.ToString(); + + options.create_if_missing = true; + options.error_if_exists = true; + options.write_buffer_size = 10000; // Something small to force merging + status = DB::Open(options, name, &db_); + ASSERT_TRUE(status.ok()) << status.ToString(); + } + + const Comparator* const comparator_; + DB* db_; +}; + +enum TestType { TABLE_TEST, BLOCK_TEST, MEMTABLE_TEST, DB_TEST }; + +struct TestArgs { + TestType type; + bool reverse_compare; + int restart_interval; +}; + +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}, + + {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}, + + // 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 testing::Test { + public: + Harness() : constructor_(nullptr) {} + + void Init(const TestArgs& args) { + delete constructor_; + constructor_ = nullptr; + options_ = Options(); + + options_.block_restart_interval = args.restart_interval; + // Use shorter block size for tests to exercise block boundary + // conditions more. + options_.block_size = 256; + if (args.reverse_compare) { + options_.comparator = &reverse_key_comparator; + } + switch (args.type) { + case TABLE_TEST: + constructor_ = new TableConstructor(options_.comparator); + break; + case BLOCK_TEST: + constructor_ = new BlockConstructor(options_.comparator); + break; + case MEMTABLE_TEST: + constructor_ = new MemTableConstructor(options_.comparator); + break; + case DB_TEST: + constructor_ = new DBConstructor(options_.comparator); + break; + } + } + + ~Harness() { delete constructor_; } + + void Add(const std::string& key, const std::string& value) { + constructor_->Add(key, value); + } + + void Test(Random* rnd) { + std::vector<std::string> keys; + KVMap data; + constructor_->Finish(options_, &keys, &data); + + TestForwardScan(keys, data); + TestBackwardScan(keys, data); + TestRandomAccess(rnd, keys, data); + } + + void TestForwardScan(const std::vector<std::string>& keys, + const KVMap& data) { + Iterator* iter = constructor_->NewIterator(); + ASSERT_TRUE(!iter->Valid()); + iter->SeekToFirst(); + for (KVMap::const_iterator model_iter = data.begin(); + model_iter != data.end(); ++model_iter) { + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + iter->Next(); + } + ASSERT_TRUE(!iter->Valid()); + delete iter; + } + + void TestBackwardScan(const std::vector<std::string>& keys, + const KVMap& data) { + Iterator* iter = constructor_->NewIterator(); + ASSERT_TRUE(!iter->Valid()); + iter->SeekToLast(); + for (KVMap::const_reverse_iterator model_iter = data.rbegin(); + model_iter != data.rend(); ++model_iter) { + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + iter->Prev(); + } + ASSERT_TRUE(!iter->Valid()); + delete iter; + } + + void TestRandomAccess(Random* rnd, const std::vector<std::string>& keys, + const KVMap& data) { + static const bool kVerbose = false; + Iterator* iter = constructor_->NewIterator(); + ASSERT_TRUE(!iter->Valid()); + KVMap::const_iterator model_iter = data.begin(); + if (kVerbose) std::fprintf(stderr, "---\n"); + for (int i = 0; i < 200; i++) { + const int toss = rnd->Uniform(5); + switch (toss) { + case 0: { + if (iter->Valid()) { + if (kVerbose) std::fprintf(stderr, "Next\n"); + iter->Next(); + ++model_iter; + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + } + break; + } + + case 1: { + if (kVerbose) std::fprintf(stderr, "SeekToFirst\n"); + iter->SeekToFirst(); + model_iter = data.begin(); + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + break; + } + + case 2: { + std::string key = PickRandomKey(rnd, keys); + model_iter = data.lower_bound(key); + if (kVerbose) + std::fprintf(stderr, "Seek '%s'\n", EscapeString(key).c_str()); + iter->Seek(Slice(key)); + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + break; + } + + case 3: { + if (iter->Valid()) { + if (kVerbose) std::fprintf(stderr, "Prev\n"); + iter->Prev(); + if (model_iter == data.begin()) { + model_iter = data.end(); // Wrap around to invalid value + } else { + --model_iter; + } + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + } + break; + } + + case 4: { + if (kVerbose) std::fprintf(stderr, "SeekToLast\n"); + iter->SeekToLast(); + if (keys.empty()) { + model_iter = data.end(); + } else { + std::string last = data.rbegin()->first; + model_iter = data.lower_bound(last); + } + ASSERT_EQ(ToString(data, model_iter), ToString(iter)); + break; + } + } + } + delete iter; + } + + std::string ToString(const KVMap& data, const KVMap::const_iterator& it) { + if (it == data.end()) { + return "END"; + } else { + return "'" + it->first + "->" + it->second + "'"; + } + } + + std::string ToString(const KVMap& data, + const KVMap::const_reverse_iterator& it) { + if (it == data.rend()) { + return "END"; + } else { + return "'" + it->first + "->" + it->second + "'"; + } + } + + std::string ToString(const Iterator* it) { + if (!it->Valid()) { + return "END"; + } else { + return "'" + it->key().ToString() + "->" + it->value().ToString() + "'"; + } + } + + std::string PickRandomKey(Random* rnd, const std::vector<std::string>& keys) { + if (keys.empty()) { + return "foo"; + } else { + const int index = rnd->Uniform(keys.size()); + std::string result = keys[index]; + switch (rnd->Uniform(3)) { + case 0: + // Return an existing key + break; + case 1: { + // Attempt to return something smaller than an existing key + if (!result.empty() && result[result.size() - 1] > '\0') { + result[result.size() - 1]--; + } + break; + } + case 2: { + // Return something larger than an existing key + Increment(options_.comparator, &result); + break; + } + } + return result; + } + } + + // Returns nullptr if not running against a DB + DB* db() const { return constructor_->db(); } + + private: + Options options_; + Constructor* constructor_; +}; + +// Test empty table/block. +TEST_F(Harness, Empty) { + for (int i = 0; i < kNumTestArgs; i++) { + Init(kTestArgList[i]); + Random rnd(test::RandomSeed() + 1); + Test(&rnd); + } +} + +// Special test for a block with no restart entries. The C++ leveldb +// code never generates such blocks, but the Java version of leveldb +// seems to. +TEST_F(Harness, ZeroRestartPointsInBlock) { + char data[sizeof(uint32_t)]; + memset(data, 0, sizeof(data)); + BlockContents contents; + contents.data = Slice(data, sizeof(data)); + contents.cachable = false; + contents.heap_allocated = false; + Block block(contents); + Iterator* iter = block.NewIterator(BytewiseComparator()); + iter->SeekToFirst(); + ASSERT_TRUE(!iter->Valid()); + iter->SeekToLast(); + ASSERT_TRUE(!iter->Valid()); + iter->Seek("foo"); + ASSERT_TRUE(!iter->Valid()); + delete iter; +} + +// Test the empty key +TEST_F(Harness, SimpleEmptyKey) { + for (int i = 0; i < kNumTestArgs; i++) { + Init(kTestArgList[i]); + Random rnd(test::RandomSeed() + 1); + Add("", "v"); + Test(&rnd); + } +} + +TEST_F(Harness, SimpleSingle) { + for (int i = 0; i < kNumTestArgs; i++) { + Init(kTestArgList[i]); + Random rnd(test::RandomSeed() + 2); + Add("abc", "v"); + Test(&rnd); + } +} + +TEST_F(Harness, SimpleMulti) { + for (int i = 0; i < kNumTestArgs; i++) { + Init(kTestArgList[i]); + Random rnd(test::RandomSeed() + 3); + Add("abc", "v"); + Add("abcd", "v"); + Add("ac", "v2"); + Test(&rnd); + } +} + +TEST_F(Harness, SimpleSpecialKey) { + for (int i = 0; i < kNumTestArgs; i++) { + Init(kTestArgList[i]); + Random rnd(test::RandomSeed() + 4); + Add("\xff\xff", "v3"); + Test(&rnd); + } +} + +TEST_F(Harness, Randomized) { + for (int i = 0; i < kNumTestArgs; i++) { + Init(kTestArgList[i]); + Random rnd(test::RandomSeed() + 5); + for (int num_entries = 0; num_entries < 2000; + num_entries += (num_entries < 50 ? 1 : 200)) { + if ((num_entries % 10) == 0) { + std::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; + Add(test::RandomKey(&rnd, rnd.Skewed(4)), + test::RandomString(&rnd, rnd.Skewed(5), &v).ToString()); + } + Test(&rnd); + } + } +} + +TEST_F(Harness, RandomizedLongDB) { + Random rnd(test::RandomSeed()); + TestArgs args = {DB_TEST, false, 16}; + Init(args); + int num_entries = 100000; + for (int e = 0; e < num_entries; e++) { + std::string v; + Add(test::RandomKey(&rnd, rnd.Skewed(4)), + test::RandomString(&rnd, rnd.Skewed(5), &v).ToString()); + } + Test(&rnd); + + // We must have created enough data to force merging + int files = 0; + for (int level = 0; level < config::kNumLevels; level++) { + std::string value; + char name[100]; + std::snprintf(name, sizeof(name), "leveldb.num-files-at-level%d", level); + ASSERT_TRUE(db()->GetProperty(name, &value)); + files += atoi(value.c_str()); + } + ASSERT_GT(files, 0); +} + +TEST(MemTableTest, Simple) { + InternalKeyComparator cmp(BytewiseComparator()); + MemTable* memtable = new MemTable(cmp); + memtable->Ref(); + WriteBatch batch; + WriteBatchInternal::SetSequence(&batch, 100); + batch.Put(std::string("k1"), std::string("v1")); + batch.Put(std::string("k2"), std::string("v2")); + batch.Put(std::string("k3"), std::string("v3")); + batch.Put(std::string("largekey"), std::string("vlarge")); + ASSERT_TRUE(WriteBatchInternal::InsertInto(&batch, memtable).ok()); + + Iterator* iter = memtable->NewIterator(); + iter->SeekToFirst(); + while (iter->Valid()) { + std::fprintf(stderr, "key: '%s' -> '%s'\n", iter->key().ToString().c_str(), + iter->value().ToString().c_str()); + iter->Next(); + } + + delete iter; + memtable->Unref(); +} + +static bool Between(uint64_t val, uint64_t low, uint64_t high) { + bool result = (val >= low) && (val <= high); + if (!result) { + std::fprintf(stderr, "Value %llu is not in range [%llu, %llu]\n", + (unsigned long long)(val), (unsigned long long)(low), + (unsigned long long)(high)); + } + return result; +} + +TEST(TableTest, ApproximateOffsetOfPlain) { + TableConstructor c(BytewiseComparator()); + c.Add("k01", "hello"); + c.Add("k02", "hello2"); + c.Add("k03", std::string(10000, 'x')); + c.Add("k04", std::string(200000, 'x')); + c.Add("k05", std::string(300000, 'x')); + c.Add("k06", "hello3"); + c.Add("k07", std::string(100000, 'x')); + std::vector<std::string> keys; + KVMap kvmap; + Options options; + options.block_size = 1024; + 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("k04a"), 210000, 211000)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k05"), 210000, 211000)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k06"), 510000, 511000)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("k07"), 510000, 511000)); + ASSERT_TRUE(Between(c.ApproximateOffsetOf("xyz"), 610000, 612000)); +} + +static bool SnappyCompressionSupported() { + std::string out; + Slice in = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"; + return port::Snappy_Compress(in.data(), in.size(), &out); +} + +TEST(TableTest, ApproximateOffsetOfCompressed) { + if (!SnappyCompressionSupported()) { + std::fprintf(stderr, "skipping compression tests\n"); + return; + } + + Random rnd(301); + TableConstructor c(BytewiseComparator()); + std::string tmp; + c.Add("k01", "hello"); + c.Add("k02", test::CompressibleString(&rnd, 0.25, 10000, &tmp)); + c.Add("k03", "hello3"); + c.Add("k04", test::CompressibleString(&rnd, 0.25, 10000, &tmp)); + std::vector<std::string> keys; + KVMap kvmap; + Options options; + options.block_size = 1024; + options.compression = kSnappyCompression; + c.Finish(options, &keys, &kvmap); + + // 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 + diff --git a/table/two_level_iterator.cc b/table/two_level_iterator.cc new file mode 100644 index 0000000..144790d --- /dev/null +++ b/table/two_level_iterator.cc @@ -0,0 +1,171 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "table/two_level_iterator.h" + +#include "leveldb/table.h" +#include "table/block.h" +#include "table/format.h" +#include "table/iterator_wrapper.h" + +namespace leveldb { + +namespace { + +typedef Iterator* (*BlockFunction)(void*, const ReadOptions&, const Slice&); + +class TwoLevelIterator : public Iterator { + public: + TwoLevelIterator(Iterator* index_iter, BlockFunction block_function, + void* arg, const ReadOptions& options); + + ~TwoLevelIterator() override; + + void Seek(const Slice& target) override; + void SeekToFirst() override; + void SeekToLast() override; + void Next() override; + void Prev() override; + + bool Valid() const override { return data_iter_.Valid(); } + Slice key() const override { + assert(Valid()); + return data_iter_.key(); + } + Slice value() const override { + assert(Valid()); + return data_iter_.value(); + } + 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() != nullptr && !data_iter_.status().ok()) { + return data_iter_.status(); + } else { + return status_; + } + } + + private: + void SaveError(const Status& s) { + if (status_.ok() && !s.ok()) status_ = s; + } + void SkipEmptyDataBlocksForward(); + void SkipEmptyDataBlocksBackward(); + void SetDataIterator(Iterator* data_iter); + void InitDataBlock(); + + BlockFunction block_function_; + void* arg_; + const ReadOptions options_; + Status status_; + IteratorWrapper index_iter_; + 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) + : block_function_(block_function), + arg_(arg), + options_(options), + index_iter_(index_iter), + data_iter_(nullptr) {} + +TwoLevelIterator::~TwoLevelIterator() = default; + +void TwoLevelIterator::Seek(const Slice& target) { + index_iter_.Seek(target); + InitDataBlock(); + if (data_iter_.iter() != nullptr) data_iter_.Seek(target); + SkipEmptyDataBlocksForward(); +} + +void TwoLevelIterator::SeekToFirst() { + index_iter_.SeekToFirst(); + InitDataBlock(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToFirst(); + SkipEmptyDataBlocksForward(); +} + +void TwoLevelIterator::SeekToLast() { + index_iter_.SeekToLast(); + InitDataBlock(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToLast(); + SkipEmptyDataBlocksBackward(); +} + +void TwoLevelIterator::Next() { + assert(Valid()); + data_iter_.Next(); + SkipEmptyDataBlocksForward(); +} + +void TwoLevelIterator::Prev() { + assert(Valid()); + data_iter_.Prev(); + SkipEmptyDataBlocksBackward(); +} + +void TwoLevelIterator::SkipEmptyDataBlocksForward() { + while (data_iter_.iter() == nullptr || !data_iter_.Valid()) { + // Move to next block + if (!index_iter_.Valid()) { + SetDataIterator(nullptr); + return; + } + index_iter_.Next(); + InitDataBlock(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToFirst(); + } +} + +void TwoLevelIterator::SkipEmptyDataBlocksBackward() { + while (data_iter_.iter() == nullptr || !data_iter_.Valid()) { + // Move to next block + if (!index_iter_.Valid()) { + SetDataIterator(nullptr); + return; + } + index_iter_.Prev(); + InitDataBlock(); + if (data_iter_.iter() != nullptr) data_iter_.SeekToLast(); + } +} + +void TwoLevelIterator::SetDataIterator(Iterator* data_iter) { + if (data_iter_.iter() != nullptr) SaveError(data_iter_.status()); + data_iter_.Set(data_iter); +} + +void TwoLevelIterator::InitDataBlock() { + if (!index_iter_.Valid()) { + SetDataIterator(nullptr); + } else { + Slice handle = index_iter_.value(); + 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 { + Iterator* iter = (*block_function_)(arg_, options_, handle); + data_block_handle_.assign(handle.data(), handle.size()); + SetDataIterator(iter); + } + } +} + +} // namespace + +Iterator* NewTwoLevelIterator(Iterator* index_iter, + BlockFunction block_function, void* arg, + const ReadOptions& options) { + return new TwoLevelIterator(index_iter, block_function, arg, options); +} + +} // namespace leveldb diff --git a/table/two_level_iterator.h b/table/two_level_iterator.h new file mode 100644 index 0000000..81ffe80 --- /dev/null +++ b/table/two_level_iterator.h @@ -0,0 +1,31 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_TABLE_TWO_LEVEL_ITERATOR_H_ +#define STORAGE_LEVELDB_TABLE_TWO_LEVEL_ITERATOR_H_ + +#include "leveldb/iterator.h" + +namespace leveldb { + +struct ReadOptions; + +// Return a new two level iterator. A two-level iterator contains an +// index iterator whose values point to a sequence of blocks where +// each block is itself a sequence of key,value pairs. The returned +// two-level iterator yields the concatenation of all key/value pairs +// in the sequence of blocks. Takes ownership of "index_iter" and +// will delete it when no longer needed. +// +// Uses a supplied function to convert an index_iter value into +// an iterator over the contents of the corresponding block. +Iterator* NewTwoLevelIterator( + Iterator* index_iter, + Iterator* (*block_function)(void* arg, const ReadOptions& options, + const Slice& index_value), + void* arg, const ReadOptions& options); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_TABLE_TWO_LEVEL_ITERATOR_H_ diff --git a/util/arena.cc b/util/arena.cc new file mode 100644 index 0000000..46e3b2e --- /dev/null +++ b/util/arena.cc @@ -0,0 +1,66 @@ +// 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 "util/arena.h" + +namespace leveldb { + +static const int kBlockSize = 4096; + +Arena::Arena() + : alloc_ptr_(nullptr), alloc_bytes_remaining_(0), memory_usage_(0) {} + +Arena::~Arena() { + for (size_t i = 0; i < blocks_.size(); i++) { + delete[] blocks_[i]; + } +} + +char* Arena::AllocateFallback(size_t bytes) { + if (bytes > kBlockSize / 4) { + // Object is more than a quarter of our block size. Allocate it separately + // to avoid wasting too much space in leftover bytes. + char* result = AllocateNewBlock(bytes); + return result; + } + + // We waste the remaining space in the current block. + alloc_ptr_ = AllocateNewBlock(kBlockSize); + alloc_bytes_remaining_ = kBlockSize; + + char* result = alloc_ptr_; + alloc_ptr_ += bytes; + alloc_bytes_remaining_ -= bytes; + return result; +} + +char* Arena::AllocateAligned(size_t bytes) { + const int align = (sizeof(void*) > 8) ? sizeof(void*) : 8; + static_assert((align & (align - 1)) == 0, + "Pointer size should be a power of 2"); + size_t current_mod = reinterpret_cast<uintptr_t>(alloc_ptr_) & (align - 1); + size_t slop = (current_mod == 0 ? 0 : align - current_mod); + size_t needed = bytes + slop; + char* result; + if (needed <= alloc_bytes_remaining_) { + result = alloc_ptr_ + slop; + alloc_ptr_ += needed; + alloc_bytes_remaining_ -= needed; + } else { + // AllocateFallback always returned aligned memory + result = AllocateFallback(bytes); + } + assert((reinterpret_cast<uintptr_t>(result) & (align - 1)) == 0); + return result; +} + +char* Arena::AllocateNewBlock(size_t block_bytes) { + char* result = new char[block_bytes]; + blocks_.push_back(result); + memory_usage_.fetch_add(block_bytes + sizeof(char*), + std::memory_order_relaxed); + return result; +} + +} // namespace leveldb diff --git a/util/arena.h b/util/arena.h new file mode 100644 index 0000000..68fc55d --- /dev/null +++ b/util/arena.h @@ -0,0 +1,71 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_UTIL_ARENA_H_ +#define STORAGE_LEVELDB_UTIL_ARENA_H_ + +#include <atomic> +#include <cassert> +#include <cstddef> +#include <cstdint> +#include <vector> + +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. + char* AllocateAligned(size_t bytes); + + // Returns an estimate of the total memory usage of data allocated + // by the arena. + size_t MemoryUsage() const { + return memory_usage_.load(std::memory_order_relaxed); + } + + private: + char* AllocateFallback(size_t bytes); + char* AllocateNewBlock(size_t block_bytes); + + // Allocation state + char* alloc_ptr_; + size_t alloc_bytes_remaining_; + + // Array of new[] allocated memory blocks + std::vector<char*> blocks_; + + // 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<size_t> memory_usage_; +}; + +inline char* Arena::Allocate(size_t bytes) { + // The semantics of what to return are a bit messy if we allow + // 0-byte allocations, so we disallow them here (we don't need + // them for our internal use). + assert(bytes > 0); + if (bytes <= alloc_bytes_remaining_) { + char* result = alloc_ptr_; + alloc_ptr_ += bytes; + alloc_bytes_remaining_ -= bytes; + return result; + } + return AllocateFallback(bytes); +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_ARENA_H_ diff --git a/util/arena_test.cc b/util/arena_test.cc new file mode 100644 index 0000000..583de78 --- /dev/null +++ b/util/arena_test.cc @@ -0,0 +1,62 @@ +// 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 "util/arena.h" + +#include "gtest/gtest.h" +#include "util/random.h" + +namespace leveldb { + +TEST(ArenaTest, Empty) { Arena arena; } + +TEST(ArenaTest, Simple) { + std::vector<std::pair<size_t, char*>> allocated; + Arena arena; + const int N = 100000; + size_t bytes = 0; + Random rnd(301); + for (int i = 0; i < N; i++) { + size_t s; + if (i % (N / 10) == 0) { + s = i; + } else { + 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. + s = 1; + } + char* r; + if (rnd.OneIn(10)) { + r = arena.AllocateAligned(s); + } else { + r = arena.Allocate(s); + } + + for (size_t b = 0; b < s; b++) { + // Fill the "i"th allocation with a known bit pattern + r[b] = i % 256; + } + bytes += s; + allocated.push_back(std::make_pair(s, r)); + ASSERT_GE(arena.MemoryUsage(), bytes); + if (i > N / 10) { + ASSERT_LE(arena.MemoryUsage(), bytes * 1.10); + } + } + for (size_t i = 0; i < allocated.size(); i++) { + size_t num_bytes = allocated[i].first; + const char* p = allocated[i].second; + for (size_t b = 0; b < num_bytes; b++) { + // Check the "i"th allocation for the known bit pattern + ASSERT_EQ(int(p[b]) & 0xff, i % 256); + } + } +} + +} // namespace leveldb + diff --git a/util/bloom.cc b/util/bloom.cc new file mode 100644 index 0000000..87547a7 --- /dev/null +++ b/util/bloom.cc @@ -0,0 +1,92 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "leveldb/filter_policy.h" + +#include "leveldb/slice.h" +#include "util/hash.h" + +namespace leveldb { + +namespace { +static uint32_t BloomHash(const Slice& key) { + return Hash(key.data(), key.size(), 0xbc9f1d34); +} + +class BloomFilterPolicy : public FilterPolicy { + public: + 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<size_t>(bits_per_key * 0.69); // 0.69 =~ ln(2) + if (k_ < 1) k_ = 1; + if (k_ > 30) k_ = 30; + } + + const char* Name() const override { return "leveldb.BuiltinBloomFilter2"; } + + 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_; + + // For small n, we can see a very high false positive rate. Fix it + // by enforcing a minimum bloom filter length. + if (bits < 64) bits = 64; + + size_t bytes = (bits + 7) / 8; + bits = bytes * 8; + + const size_t init_size = dst->size(); + dst->resize(init_size + bytes, 0); + dst->push_back(static_cast<char>(k_)); // Remember # of probes in filter + char* array = &(*dst)[init_size]; + 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)); + h += delta; + } + } + } + + bool KeyMayMatch(const Slice& key, const Slice& bloom_filter) const override { + const size_t len = bloom_filter.size(); + if (len < 2) return false; + + const char* array = bloom_filter.data(); + const size_t bits = (len - 1) * 8; + + // 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]; + if (k > 30) { + // Reserved for potentially new encodings for short bloom filters. + // Consider it a match. + return true; + } + + uint32_t h = BloomHash(key); + 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; + 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); +} + +} // namespace leveldb diff --git a/util/bloom_test.cc b/util/bloom_test.cc new file mode 100644 index 0000000..7c8c92b --- /dev/null +++ b/util/bloom_test.cc @@ -0,0 +1,155 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "gtest/gtest.h" +#include "leveldb/filter_policy.h" +#include "util/coding.h" +#include "util/logging.h" +#include "util/testutil.h" + +namespace leveldb { + +static const int kVerbose = 1; + +static Slice Key(int i, char* buffer) { + EncodeFixed32(buffer, i); + return Slice(buffer, sizeof(uint32_t)); +} + +class BloomTest : public testing::Test { + public: + BloomTest() : policy_(NewBloomFilterPolicy(10)) {} + + ~BloomTest() { delete policy_; } + + void Reset() { + keys_.clear(); + filter_.clear(); + } + + void Add(const Slice& s) { keys_.push_back(s.ToString()); } + + void Build() { + std::vector<Slice> key_slices; + for (size_t i = 0; i < keys_.size(); i++) { + key_slices.push_back(Slice(keys_[i])); + } + filter_.clear(); + policy_->CreateFilter(&key_slices[0], static_cast<int>(key_slices.size()), + &filter_); + keys_.clear(); + if (kVerbose >= 2) DumpFilter(); + } + + size_t FilterSize() const { return filter_.size(); } + + void DumpFilter() { + std::fprintf(stderr, "F("); + for (size_t i = 0; i + 1 < filter_.size(); i++) { + const unsigned int c = static_cast<unsigned int>(filter_[i]); + for (int j = 0; j < 8; j++) { + std::fprintf(stderr, "%c", (c & (1 << j)) ? '1' : '.'); + } + } + std::fprintf(stderr, ")\n"); + } + + bool Matches(const Slice& s) { + if (!keys_.empty()) { + Build(); + } + return policy_->KeyMayMatch(s, filter_); + } + + double FalsePositiveRate() { + char buffer[sizeof(int)]; + int result = 0; + for (int i = 0; i < 10000; i++) { + if (Matches(Key(i + 1000000000, buffer))) { + result++; + } + } + return result / 10000.0; + } + + private: + const FilterPolicy* policy_; + std::string filter_; + std::vector<std::string> keys_; +}; + +TEST_F(BloomTest, EmptyFilter) { + ASSERT_TRUE(!Matches("hello")); + ASSERT_TRUE(!Matches("world")); +} + +TEST_F(BloomTest, Small) { + Add("hello"); + Add("world"); + ASSERT_TRUE(Matches("hello")); + ASSERT_TRUE(Matches("world")); + ASSERT_TRUE(!Matches("x")); + ASSERT_TRUE(!Matches("foo")); +} + +static int NextLength(int length) { + if (length < 10) { + length += 1; + } else if (length < 100) { + length += 10; + } else if (length < 1000) { + length += 100; + } else { + length += 1000; + } + return length; +} + +TEST_F(BloomTest, VaryingLengths) { + char buffer[sizeof(int)]; + + // Count number of filters that significantly exceed the false positive rate + int mediocre_filters = 0; + int good_filters = 0; + + for (int length = 1; length <= 10000; length = NextLength(length)) { + Reset(); + for (int i = 0; i < length; i++) { + Add(Key(i, buffer)); + } + Build(); + + ASSERT_LE(FilterSize(), static_cast<size_t>((length * 10 / 8) + 40)) + << length; + + // All added keys must match + for (int i = 0; i < length; i++) { + ASSERT_TRUE(Matches(Key(i, buffer))) + << "Length " << length << "; key " << i; + } + + // Check false positive rate + double rate = FalsePositiveRate(); + if (kVerbose >= 1) { + std::fprintf(stderr, + "False positives: %5.2f%% @ length = %6d ; bytes = %6d\n", + rate * 100.0, length, static_cast<int>(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++; + } + if (kVerbose >= 1) { + std::fprintf(stderr, "Filters: %d good, %d mediocre\n", good_filters, + mediocre_filters); + } + ASSERT_LE(mediocre_filters, good_filters / 5); +} + +// Different bits-per-byte + +} // namespace leveldb + diff --git a/util/cache.cc b/util/cache.cc new file mode 100644 index 0000000..ad1e9a2 --- /dev/null +++ b/util/cache.cc @@ -0,0 +1,401 @@ +// 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 "leveldb/cache.h" + +#include <cassert> +#include <cstdio> +#include <cstdlib> + +#include "port/port.h" +#include "port/thread_annotations.h" +#include "util/hash.h" +#include "util/mutexlock.h" + +namespace leveldb { + +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. +struct LRUHandle { + void* value; + void (*deleter)(const Slice&, void* value); + LRUHandle* next_hash; + LRUHandle* next; + LRUHandle* prev; + size_t charge; // TODO(opt): Only allow uint32_t? + size_t key_length; + 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 { + // 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); + } +}; + +// We provide our own simple hash table since it removes a whole bunch +// of porting hacks and is also faster than some of the built-in hash +// table implementations in some of the compiler/runtime combinations +// we have tested. E.g., readrandom speeds up by ~5% over the g++ +// 4.4.3's builtin hashtable. +class HandleTable { + public: + HandleTable() : length_(0), elems_(0), list_(nullptr) { Resize(); } + ~HandleTable() { delete[] list_; } + + LRUHandle* Lookup(const Slice& key, uint32_t hash) { + return *FindPointer(key, hash); + } + + LRUHandle* Insert(LRUHandle* h) { + LRUHandle** ptr = FindPointer(h->key(), h->hash); + LRUHandle* old = *ptr; + h->next_hash = (old == nullptr ? nullptr : old->next_hash); + *ptr = h; + if (old == nullptr) { + ++elems_; + if (elems_ > length_) { + // Since each cache entry is fairly large, we aim for a small + // average linked list length (<= 1). + Resize(); + } + } + return old; + } + + LRUHandle* Remove(const Slice& key, uint32_t hash) { + LRUHandle** ptr = FindPointer(key, hash); + LRUHandle* result = *ptr; + if (result != nullptr) { + *ptr = result->next_hash; + --elems_; + } + return result; + } + + private: + // The table consists of an array of buckets where each bucket is + // a linked list of cache entries that hash into the bucket. + uint32_t length_; + uint32_t elems_; + LRUHandle** list_; + + // Return a pointer to slot that points to a cache entry that + // matches key/hash. If there is no such cache entry, return a + // 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 != nullptr && ((*ptr)->hash != hash || key != (*ptr)->key())) { + ptr = &(*ptr)->next_hash; + } + return ptr; + } + + void Resize() { + uint32_t new_length = 4; + while (new_length < elems_) { + new_length *= 2; + } + LRUHandle** new_list = new LRUHandle*[new_length]; + memset(new_list, 0, sizeof(new_list[0]) * new_length); + uint32_t count = 0; + for (uint32_t i = 0; i < length_; i++) { + LRUHandle* h = list_[i]; + while (h != nullptr) { + LRUHandle* next = h->next_hash; + uint32_t hash = h->hash; + LRUHandle** ptr = &new_list[hash & (new_length - 1)]; + h->next_hash = *ptr; + *ptr = h; + h = next; + count++; + } + } + assert(elems_ == count); + delete[] list_; + list_ = new_list; + length_ = new_length; + } +}; + +// A single shard of sharded cache. +class LRUCache { + public: + LRUCache(); + ~LRUCache(); + + // Separate from constructor so caller can easily make an array of 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, + 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* 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. + 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. + // Entries have refs==1 and in_cache==true. + LRUHandle lru_ GUARDED_BY(mutex_); + + // 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() : 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() { + 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->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) { // 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); + } +} + +void LRUCache::LRU_Remove(LRUHandle* e) { + e->next->prev = e->prev; + e->prev->next = e->next; +} + +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; +} + +Cache::Handle* LRUCache::Lookup(const Slice& key, uint32_t hash) { + MutexLock l(&mutex_); + LRUHandle* e = table_.Lookup(key, hash); + if (e != nullptr) { + Ref(e); + } + return reinterpret_cast<Cache::Handle*>(e); +} + +void LRUCache::Release(Cache::Handle* handle) { + MutexLock l(&mutex_); + Unref(reinterpret_cast<LRUHandle*>(handle)); +} + +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<LRUHandle*>(malloc(sizeof(LRUHandle) - 1 + key.size())); + e->value = value; + e->deleter = deleter; + e->charge = charge; + e->key_length = key.size(); + e->hash = hash; + e->in_cache = false; + e->refs = 1; // for the returned handle. + std::memcpy(e->key_data, key.data(), key.size()); + + 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; + 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<Cache::Handle*>(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_); + 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); + } + } +} + +static const int kNumShardBits = 4; +static const int kNumShards = 1 << kNumShardBits; + +class ShardedLRUCache : public Cache { + private: + LRUCache shard_[kNumShards]; + port::Mutex id_mutex_; + uint64_t last_id_; + + static inline uint32_t HashSlice(const Slice& s) { + return Hash(s.data(), s.size(), 0); + } + + static uint32_t Shard(uint32_t hash) { return hash >> (32 - kNumShardBits); } + + public: + 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); + } + } + ~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); + } + Handle* Lookup(const Slice& key) override { + const uint32_t hash = HashSlice(key); + return shard_[Shard(hash)].Lookup(key, hash); + } + void Release(Handle* handle) override { + LRUHandle* h = reinterpret_cast<LRUHandle*>(handle); + shard_[Shard(h->hash)].Release(handle); + } + void Erase(const Slice& key) override { + const uint32_t hash = HashSlice(key); + shard_[Shard(hash)].Erase(key, hash); + } + void* Value(Handle* handle) override { + return reinterpret_cast<LRUHandle*>(handle)->value; + } + 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); } + +} // namespace leveldb diff --git a/util/cache_test.cc b/util/cache_test.cc new file mode 100644 index 0000000..5b955c1 --- /dev/null +++ b/util/cache_test.cc @@ -0,0 +1,225 @@ +// 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 "leveldb/cache.h" + +#include <vector> + +#include "gtest/gtest.h" +#include "util/coding.h" + +namespace leveldb { + +// Conversions between numeric keys/values and the types expected by Cache. +static std::string EncodeKey(int k) { + std::string result; + PutFixed32(&result, k); + return result; +} +static int DecodeKey(const Slice& k) { + assert(k.size() == 4); + return DecodeFixed32(k.data()); +} +static void* EncodeValue(uintptr_t v) { return reinterpret_cast<void*>(v); } +static int DecodeValue(void* v) { return reinterpret_cast<uintptr_t>(v); } + +class CacheTest : public testing::Test { + public: + static void Deleter(const Slice& key, void* v) { + current_->deleted_keys_.push_back(DecodeKey(key)); + current_->deleted_values_.push_back(DecodeValue(v)); + } + + static constexpr int kCacheSize = 1000; + std::vector<int> deleted_keys_; + std::vector<int> deleted_values_; + Cache* cache_; + + CacheTest() : cache_(NewLRUCache(kCacheSize)) { current_ = this; } + + ~CacheTest() { delete cache_; } + + int Lookup(int key) { + Cache::Handle* handle = cache_->Lookup(EncodeKey(key)); + const int r = (handle == nullptr) ? -1 : DecodeValue(cache_->Value(handle)); + if (handle != nullptr) { + cache_->Release(handle); + } + return r; + } + + void Insert(int key, int value, int charge = 1) { + cache_->Release(cache_->Insert(EncodeKey(key), EncodeValue(value), charge, + &CacheTest::Deleter)); + } + + 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_; + +TEST_F(CacheTest, HitAndMiss) { + ASSERT_EQ(-1, Lookup(100)); + + Insert(100, 101); + ASSERT_EQ(101, Lookup(100)); + 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)); + + Insert(100, 102); + ASSERT_EQ(102, Lookup(100)); + ASSERT_EQ(201, Lookup(200)); + ASSERT_EQ(-1, Lookup(300)); + + ASSERT_EQ(1, deleted_keys_.size()); + ASSERT_EQ(100, deleted_keys_[0]); + ASSERT_EQ(101, deleted_values_[0]); +} + +TEST_F(CacheTest, Erase) { + Erase(200); + ASSERT_EQ(0, deleted_keys_.size()); + + Insert(100, 101); + Insert(200, 201); + Erase(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(201, Lookup(200)); + ASSERT_EQ(1, deleted_keys_.size()); +} + +TEST_F(CacheTest, EntriesArePinned) { + Insert(100, 101); + Cache::Handle* h1 = cache_->Lookup(EncodeKey(100)); + ASSERT_EQ(101, DecodeValue(cache_->Value(h1))); + + Insert(100, 102); + Cache::Handle* h2 = cache_->Lookup(EncodeKey(100)); + ASSERT_EQ(102, DecodeValue(cache_->Value(h2))); + ASSERT_EQ(0, deleted_keys_.size()); + + cache_->Release(h1); + 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, deleted_keys_.size()); + + cache_->Release(h2); + ASSERT_EQ(2, deleted_keys_.size()); + ASSERT_EQ(100, deleted_keys_[1]); + ASSERT_EQ(102, deleted_values_[1]); +} + +TEST_F(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, + // 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)); + ASSERT_EQ(101, Lookup(100)); + } + ASSERT_EQ(101, Lookup(100)); + ASSERT_EQ(-1, Lookup(200)); + ASSERT_EQ(301, Lookup(300)); + cache_->Release(h); +} + +TEST_F(CacheTest, UseExceedsCacheSize) { + // Overfill the cache, keeping handles on all inserted entries. + std::vector<Cache::Handle*> 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_F(CacheTest, HeavyEntries) { + // Add a bunch of light and heavy entries and then count the combined + // size of items still in the cache, which must be approximately the + // same as the total capacity. + const int kLight = 1; + const int kHeavy = 10; + int added = 0; + int index = 0; + while (added < 2 * kCacheSize) { + const int weight = (index & 1) ? kLight : kHeavy; + Insert(index, 1000 + index, weight); + added += weight; + index++; + } + + int cached_weight = 0; + for (int i = 0; i < index; i++) { + const int weight = (i & 1 ? kLight : kHeavy); + int r = Lookup(i); + if (r >= 0) { + cached_weight += weight; + ASSERT_EQ(1000 + i, r); + } + } + ASSERT_LE(cached_weight, kCacheSize + kCacheSize / 10); +} + +TEST_F(CacheTest, NewId) { + uint64_t a = cache_->NewId(); + uint64_t b = cache_->NewId(); + ASSERT_NE(a, b); +} + +TEST_F(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_F(CacheTest, ZeroSizeCache) { + delete cache_; + cache_ = NewLRUCache(0); + + Insert(1, 100); + ASSERT_EQ(-1, Lookup(1)); +} + +} // namespace leveldb + diff --git a/util/coding.cc b/util/coding.cc new file mode 100644 index 0000000..df3fa10 --- /dev/null +++ b/util/coding.cc @@ -0,0 +1,166 @@ +// 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 "util/coding.h" + +namespace leveldb { + +void PutFixed32(std::string* dst, uint32_t value) { + char buf[sizeof(value)]; + EncodeFixed32(buf, value); + dst->append(buf, sizeof(buf)); +} + +void PutFixed64(std::string* dst, uint64_t value) { + char buf[sizeof(value)]; + EncodeFixed64(buf, value); + dst->append(buf, sizeof(buf)); +} + +char* EncodeVarint32(char* dst, uint32_t v) { + // Operate on characters as unsigneds + uint8_t* ptr = reinterpret_cast<uint8_t*>(dst); + static const int B = 128; + if (v < (1 << 7)) { + *(ptr++) = v; + } else if (v < (1 << 14)) { + *(ptr++) = v | B; + *(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 | B; + *(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; + } + return reinterpret_cast<char*>(ptr); +} + +void PutVarint32(std::string* dst, uint32_t v) { + char buf[5]; + char* ptr = EncodeVarint32(buf, v); + dst->append(buf, ptr - buf); +} + +char* EncodeVarint64(char* dst, uint64_t v) { + static const int B = 128; + uint8_t* ptr = reinterpret_cast<uint8_t*>(dst); + while (v >= B) { + *(ptr++) = v | B; + v >>= 7; + } + *(ptr++) = static_cast<uint8_t>(v); + return reinterpret_cast<char*>(ptr); +} + +void PutVarint64(std::string* dst, uint64_t v) { + char buf[10]; + char* ptr = EncodeVarint64(buf, v); + dst->append(buf, ptr - buf); +} + +void PutLengthPrefixedSlice(std::string* dst, const Slice& value) { + PutVarint32(dst, value.size()); + dst->append(value.data(), value.size()); +} + +int VarintLength(uint64_t v) { + int len = 1; + while (v >= 128) { + v >>= 7; + len++; + } + return len; +} + +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) { + uint32_t byte = *(reinterpret_cast<const uint8_t*>(p)); + p++; + if (byte & 128) { + // More bytes are present + result |= ((byte & 127) << shift); + } else { + result |= (byte << shift); + *value = result; + return reinterpret_cast<const char*>(p); + } + } + 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 == nullptr) { + return false; + } else { + *input = Slice(q, limit - q); + return true; + } +} + +const char* GetVarint64Ptr(const char* p, const char* limit, uint64_t* value) { + uint64_t result = 0; + for (uint32_t shift = 0; shift <= 63 && p < limit; shift += 7) { + uint64_t byte = *(reinterpret_cast<const uint8_t*>(p)); + p++; + if (byte & 128) { + // More bytes are present + result |= ((byte & 127) << shift); + } else { + result |= (byte << shift); + *value = result; + return reinterpret_cast<const char*>(p); + } + } + 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 == nullptr) { + return false; + } else { + *input = Slice(q, limit - q); + return true; + } +} + +const char* GetLengthPrefixedSlice(const char* p, const char* limit, + Slice* result) { + uint32_t len; + p = GetVarint32Ptr(p, limit, &len); + 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) { + *result = Slice(input->data(), len); + input->remove_prefix(len); + return true; + } else { + return false; + } +} + +} // namespace leveldb diff --git a/util/coding.h b/util/coding.h new file mode 100644 index 0000000..f0bb57b --- /dev/null +++ b/util/coding.h @@ -0,0 +1,122 @@ +// 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. +// +// Endian-neutral encoding: +// * Fixed-length numbers are encoded with least-significant byte first +// * In addition we support variable length "varint" encoding +// * Strings are encoded prefixed by their length in varint format + +#ifndef STORAGE_LEVELDB_UTIL_CODING_H_ +#define STORAGE_LEVELDB_UTIL_CODING_H_ + +#include <cstdint> +#include <cstring> +#include <string> + +#include "leveldb/slice.h" +#include "port/port.h" + +namespace leveldb { + +// Standard Put... routines append to a string +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. +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 +// nullptr on error. These routines only look at bytes in the range +// [p..limit-1] +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" +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 +char* EncodeVarint32(char* dst, uint32_t value); +char* EncodeVarint64(char* dst, uint64_t value); + +// 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<uint8_t*>(dst); + + // Recent clang and gcc optimize this to a single mov / str instruction. + buffer[0] = static_cast<uint8_t>(value); + buffer[1] = static_cast<uint8_t>(value >> 8); + buffer[2] = static_cast<uint8_t>(value >> 16); + buffer[3] = static_cast<uint8_t>(value >> 24); +} + +inline void EncodeFixed64(char* dst, uint64_t value) { + uint8_t* const buffer = reinterpret_cast<uint8_t*>(dst); + + // Recent clang and gcc optimize this to a single mov / str instruction. + buffer[0] = static_cast<uint8_t>(value); + buffer[1] = static_cast<uint8_t>(value >> 8); + buffer[2] = static_cast<uint8_t>(value >> 16); + buffer[3] = static_cast<uint8_t>(value >> 24); + buffer[4] = static_cast<uint8_t>(value >> 32); + buffer[5] = static_cast<uint8_t>(value >> 40); + buffer[6] = static_cast<uint8_t>(value >> 48); + buffer[7] = static_cast<uint8_t>(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<const uint8_t*>(ptr); + + // Recent clang and gcc optimize this to a single mov / ldr instruction. + return (static_cast<uint32_t>(buffer[0])) | + (static_cast<uint32_t>(buffer[1]) << 8) | + (static_cast<uint32_t>(buffer[2]) << 16) | + (static_cast<uint32_t>(buffer[3]) << 24); +} + +inline uint64_t DecodeFixed64(const char* ptr) { + const uint8_t* const buffer = reinterpret_cast<const uint8_t*>(ptr); + + // Recent clang and gcc optimize this to a single mov / ldr instruction. + return (static_cast<uint64_t>(buffer[0])) | + (static_cast<uint64_t>(buffer[1]) << 8) | + (static_cast<uint64_t>(buffer[2]) << 16) | + (static_cast<uint64_t>(buffer[3]) << 24) | + (static_cast<uint64_t>(buffer[4]) << 32) | + (static_cast<uint64_t>(buffer[5]) << 40) | + (static_cast<uint64_t>(buffer[6]) << 48) | + (static_cast<uint64_t>(buffer[7]) << 56); +} + +// Internal routine for use by fallback path of GetVarint32Ptr +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<const uint8_t*>(p)); + if ((result & 128) == 0) { + *value = result; + return p + 1; + } + } + return GetVarint32PtrFallback(p, limit, value); +} + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_CODING_H_ diff --git a/util/coding_test.cc b/util/coding_test.cc new file mode 100644 index 0000000..3f5a214 --- /dev/null +++ b/util/coding_test.cc @@ -0,0 +1,194 @@ +// 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 "util/coding.h" + +#include <vector> + +#include "gtest/gtest.h" + +namespace leveldb { + +TEST(Coding, Fixed32) { + std::string s; + for (uint32_t v = 0; v < 100000; v++) { + PutFixed32(&s, v); + } + + const char* p = s.data(); + for (uint32_t v = 0; v < 100000; v++) { + uint32_t actual = DecodeFixed32(p); + ASSERT_EQ(v, actual); + p += sizeof(uint32_t); + } +} + +TEST(Coding, Fixed64) { + std::string s; + for (int power = 0; power <= 63; power++) { + uint64_t v = static_cast<uint64_t>(1) << power; + PutFixed64(&s, v - 1); + PutFixed64(&s, v + 0); + PutFixed64(&s, v + 1); + } + + const char* p = s.data(); + for (int power = 0; power <= 63; power++) { + uint64_t v = static_cast<uint64_t>(1) << power; + uint64_t actual; + actual = DecodeFixed64(p); + ASSERT_EQ(v - 1, actual); + p += sizeof(uint64_t); + + actual = DecodeFixed64(p); + ASSERT_EQ(v + 0, actual); + p += sizeof(uint64_t); + + actual = DecodeFixed64(p); + ASSERT_EQ(v + 1, actual); + p += sizeof(uint64_t); + } +} + +// Test that encoding routines generate little-endian encodings +TEST(Coding, EncodingOutput) { + std::string dst; + PutFixed32(&dst, 0x04030201); + ASSERT_EQ(4, dst.size()); + ASSERT_EQ(0x01, static_cast<int>(dst[0])); + ASSERT_EQ(0x02, static_cast<int>(dst[1])); + ASSERT_EQ(0x03, static_cast<int>(dst[2])); + ASSERT_EQ(0x04, static_cast<int>(dst[3])); + + dst.clear(); + PutFixed64(&dst, 0x0807060504030201ull); + ASSERT_EQ(8, dst.size()); + ASSERT_EQ(0x01, static_cast<int>(dst[0])); + ASSERT_EQ(0x02, static_cast<int>(dst[1])); + ASSERT_EQ(0x03, static_cast<int>(dst[2])); + ASSERT_EQ(0x04, static_cast<int>(dst[3])); + ASSERT_EQ(0x05, static_cast<int>(dst[4])); + ASSERT_EQ(0x06, static_cast<int>(dst[5])); + ASSERT_EQ(0x07, static_cast<int>(dst[6])); + ASSERT_EQ(0x08, static_cast<int>(dst[7])); +} + +TEST(Coding, Varint32) { + std::string s; + for (uint32_t i = 0; i < (32 * 32); i++) { + uint32_t v = (i / 32) << (i % 32); + PutVarint32(&s, v); + } + + const char* p = s.data(); + const char* limit = p + s.size(); + for (uint32_t i = 0; i < (32 * 32); i++) { + uint32_t expected = (i / 32) << (i % 32); + uint32_t actual; + const char* start = p; + p = GetVarint32Ptr(p, limit, &actual); + ASSERT_TRUE(p != nullptr); + ASSERT_EQ(expected, actual); + ASSERT_EQ(VarintLength(actual), p - start); + } + ASSERT_EQ(p, s.data() + s.size()); +} + +TEST(Coding, Varint64) { + // Construct the list of values to check + std::vector<uint64_t> values; + // Some special values + values.push_back(0); + values.push_back(100); + values.push_back(~static_cast<uint64_t>(0)); + values.push_back(~static_cast<uint64_t>(0) - 1); + for (uint32_t k = 0; k < 64; k++) { + // 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); + } + + std::string s; + for (size_t i = 0; i < values.size(); i++) { + PutVarint64(&s, values[i]); + } + + const char* p = s.data(); + const char* limit = p + s.size(); + for (size_t i = 0; i < values.size(); i++) { + ASSERT_TRUE(p < limit); + uint64_t actual; + const char* start = p; + p = GetVarint64Ptr(p, limit, &actual); + 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) == nullptr); +} + +TEST(Coding, Varint32Truncation) { + uint32_t large_value = (1u << 31) + 100; + std::string s; + 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) == nullptr); + } + 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) == nullptr); +} + +TEST(Coding, Varint64Truncation) { + uint64_t large_value = (1ull << 63) + 100ull; + std::string s; + 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) == nullptr); + } + ASSERT_TRUE(GetVarint64Ptr(s.data(), s.data() + s.size(), &result) != + nullptr); + ASSERT_EQ(large_value, result); +} + +TEST(Coding, Strings) { + std::string s; + PutLengthPrefixedSlice(&s, Slice("")); + PutLengthPrefixedSlice(&s, Slice("foo")); + PutLengthPrefixedSlice(&s, Slice("bar")); + PutLengthPrefixedSlice(&s, Slice(std::string(200, 'x'))); + + Slice input(s); + Slice v; + ASSERT_TRUE(GetLengthPrefixedSlice(&input, &v)); + ASSERT_EQ("", v.ToString()); + ASSERT_TRUE(GetLengthPrefixedSlice(&input, &v)); + ASSERT_EQ("foo", v.ToString()); + ASSERT_TRUE(GetLengthPrefixedSlice(&input, &v)); + ASSERT_EQ("bar", v.ToString()); + ASSERT_TRUE(GetLengthPrefixedSlice(&input, &v)); + ASSERT_EQ(std::string(200, 'x'), v.ToString()); + ASSERT_EQ("", input.ToString()); +} + +} // namespace leveldb + diff --git a/util/comparator.cc b/util/comparator.cc new file mode 100644 index 0000000..c5766e9 --- /dev/null +++ b/util/comparator.cc @@ -0,0 +1,75 @@ +// 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 "leveldb/comparator.h" + +#include <algorithm> +#include <cstdint> +#include <string> +#include <type_traits> + +#include "leveldb/slice.h" +#include "util/logging.h" +#include "util/no_destructor.h" + +namespace leveldb { + +Comparator::~Comparator() = default; + +namespace { +class BytewiseComparatorImpl : public Comparator { + public: + BytewiseComparatorImpl() = default; + + const char* Name() const override { return "leveldb.BytewiseComparator"; } + + int Compare(const Slice& a, const Slice& b) const override { + return a.compare(b); + } + + 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; + while ((diff_index < min_length) && + ((*start)[diff_index] == limit[diff_index])) { + diff_index++; + } + + if (diff_index >= min_length) { + // Do not shorten if one string is a prefix of the other + } else { + uint8_t diff_byte = static_cast<uint8_t>((*start)[diff_index]); + if (diff_byte < static_cast<uint8_t>(0xff) && + diff_byte + 1 < static_cast<uint8_t>(limit[diff_index])) { + (*start)[diff_index]++; + start->resize(diff_index + 1); + assert(Compare(*start, limit) < 0); + } + } + } + + 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<uint8_t>(0xff)) { + (*key)[i] = byte + 1; + key->resize(i + 1); + return; + } + } + // *key is a run of 0xffs. Leave it alone. + } +}; +} // namespace + +const Comparator* BytewiseComparator() { + static NoDestructor<BytewiseComparatorImpl> singleton; + return singleton.get(); +} + +} // namespace leveldb diff --git a/util/crc32c.cc b/util/crc32c.cc new file mode 100644 index 0000000..3f18908 --- /dev/null +++ b/util/crc32c.cc @@ -0,0 +1,380 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// A portable implementation of crc32c. + +#include "util/crc32c.h" + +#include <cstddef> +#include <cstdint> + +#include "port/port.h" +#include "util/coding.h" + +namespace leveldb { +namespace crc32c { + +namespace { + +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<uint32_t>(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<const char*>(buffer)); +} + +// Returns the smallest address >= the given address that is aligned to N bytes. +// +// N must be a power of two. +template <int N> +constexpr inline const uint8_t* RoundUp(const uint8_t* pointer) { + return reinterpret_cast<uint8_t*>( + (reinterpret_cast<uintptr_t>(pointer) + (N - 1)) & + ~static_cast<uintptr_t>(N - 1)); +} + +} // namespace + +// 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<const uint8_t*>(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 p is 4-byte aligned. + while (p != x) { + STEP1; + } + } + + 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 the last few bytes. + while (p != e) { + STEP1; + } +#undef STEP4W +#undef STEP16 +#undef STEP4 +#undef STEP1 + return l ^ kCRC32Xor; +} + +} // namespace crc32c +} // namespace leveldb diff --git a/util/crc32c.h b/util/crc32c.h new file mode 100644 index 0000000..b420b5f --- /dev/null +++ b/util/crc32c.h @@ -0,0 +1,43 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_UTIL_CRC32C_H_ +#define STORAGE_LEVELDB_UTIL_CRC32C_H_ + +#include <cstddef> +#include <cstdint> + +namespace leveldb { +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. +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); } + +static const uint32_t kMaskDelta = 0xa282ead8ul; + +// Return a masked representation of crc. +// +// Motivation: it is problematic to compute the CRC of a string that +// contains embedded CRCs. Therefore we recommend that CRCs stored +// somewhere (e.g., in files) should be masked before being stored. +inline uint32_t Mask(uint32_t crc) { + // Rotate right by 15 bits and add a constant. + return ((crc >> 15) | (crc << 17)) + kMaskDelta; +} + +// Return the crc whose masked representation is masked_crc. +inline uint32_t Unmask(uint32_t masked_crc) { + uint32_t rot = masked_crc - kMaskDelta; + return ((rot >> 17) | (rot << 15)); +} + +} // namespace crc32c +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_CRC32C_H_ diff --git a/util/crc32c_test.cc b/util/crc32c_test.cc new file mode 100644 index 0000000..adcd976 --- /dev/null +++ b/util/crc32c_test.cc @@ -0,0 +1,57 @@ +// 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 "util/crc32c.h" + +#include "gtest/gtest.h" + +namespace leveldb { +namespace crc32c { + +TEST(CRC, StandardResults) { + // From rfc3720 section B.4. + char buf[32]; + + memset(buf, 0, sizeof(buf)); + ASSERT_EQ(0x8a9136aa, Value(buf, sizeof(buf))); + + memset(buf, 0xff, sizeof(buf)); + ASSERT_EQ(0x62a8ab43, Value(buf, sizeof(buf))); + + for (int i = 0; i < 32; i++) { + buf[i] = i; + } + ASSERT_EQ(0x46dd794e, Value(buf, sizeof(buf))); + + for (int i = 0; i < 32; i++) { + buf[i] = 31 - i; + } + ASSERT_EQ(0x113fdb5c, Value(buf, sizeof(buf))); + + uint8_t 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, + }; + ASSERT_EQ(0xd9963a56, Value(reinterpret_cast<char*>(data), sizeof(data))); +} + +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)); +} + +TEST(CRC, Mask) { + uint32_t crc = Value("foo", 3); + ASSERT_NE(crc, Mask(crc)); + ASSERT_NE(crc, Mask(Mask(crc))); + ASSERT_EQ(crc, Unmask(Mask(crc))); + ASSERT_EQ(crc, Unmask(Unmask(Mask(Mask(crc))))); +} + +} // namespace crc32c +} // namespace leveldb + diff --git a/util/env.cc b/util/env.cc new file mode 100644 index 0000000..a53b230 --- /dev/null +++ b/util/env.cc @@ -0,0 +1,108 @@ +// 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 "leveldb/env.h" + +#include <cstdarg> + +// This workaround can be removed when leveldb::Env::DeleteFile is removed. +// See env.h for justification. +#if defined(_WIN32) && defined(LEVELDB_DELETEFILE_UNDEFINED) +#undef DeleteFile +#endif + +namespace leveldb { + +Env::Env() = default; + +Env::~Env() = default; + +Status Env::NewAppendableFile(const std::string& fname, WritableFile** result) { + return Status::NotSupported("NewAppendableFile", fname); +} + +Status Env::RemoveDir(const std::string& dirname) { return DeleteDir(dirname); } +Status Env::DeleteDir(const std::string& dirname) { return RemoveDir(dirname); } + +Status Env::RemoveFile(const std::string& fname) { return DeleteFile(fname); } +Status Env::DeleteFile(const std::string& fname) { return RemoveFile(fname); } + +SequentialFile::~SequentialFile() = default; + +RandomAccessFile::~RandomAccessFile() = default; + +WritableFile::~WritableFile() = default; + +Logger::~Logger() = default; + +FileLock::~FileLock() = default; + +void Log(Logger* info_log, const char* format, ...) { + if (info_log != nullptr) { + std::va_list ap; + va_start(ap, format); + info_log->Logv(format, ap); + va_end(ap); + } +} + +static Status DoWriteStringToFile(Env* env, const Slice& data, + const std::string& fname, bool should_sync) { + WritableFile* file; + Status s = env->NewWritableFile(fname, &file); + if (!s.ok()) { + return s; + } + s = file->Append(data); + if (s.ok() && should_sync) { + s = file->Sync(); + } + if (s.ok()) { + s = file->Close(); + } + delete file; // Will auto-close if we did not close above + if (!s.ok()) { + env->RemoveFile(fname); + } + return s; +} + +Status WriteStringToFile(Env* env, const Slice& data, + const std::string& fname) { + return DoWriteStringToFile(env, data, fname, false); +} + +Status WriteStringToFileSync(Env* env, const Slice& data, + const std::string& fname) { + return DoWriteStringToFile(env, data, fname, true); +} + +Status ReadFileToString(Env* env, const std::string& fname, std::string* data) { + data->clear(); + SequentialFile* file; + Status s = env->NewSequentialFile(fname, &file); + if (!s.ok()) { + return s; + } + static const int kBufferSize = 8192; + char* space = new char[kBufferSize]; + while (true) { + Slice fragment; + s = file->Read(kBufferSize, &fragment, space); + if (!s.ok()) { + break; + } + data->append(fragment.data(), fragment.size()); + if (fragment.empty()) { + break; + } + } + delete[] space; + delete file; + return s; +} + +EnvWrapper::~EnvWrapper() {} + +} // namespace leveldb diff --git a/util/env_posix.cc b/util/env_posix.cc new file mode 100644 index 0000000..d84cd1e --- /dev/null +++ b/util/env_posix.cc @@ -0,0 +1,893 @@ +// 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 <dirent.h> +#include <fcntl.h> +#include <pthread.h> +#include <sys/mman.h> +#include <sys/resource.h> +#include <sys/stat.h> +#include <sys/time.h> +#include <sys/types.h> +#include <unistd.h> + +#include <atomic> +#include <cerrno> +#include <cstddef> +#include <cstdint> +#include <cstdio> +#include <cstdlib> +#include <cstring> +#include <limits> +#include <queue> +#include <set> +#include <string> +#include <thread> +#include <type_traits> +#include <utility> + +#include "leveldb/env.h" +#include "leveldb/slice.h" +#include "leveldb/status.h" +#include "port/port.h" +#include "port/thread_annotations.h" +#include "util/env_posix_test_helper.h" +#include "util/posix_logger.h" + +namespace leveldb { + +namespace { + +// 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)); + } +} + +// 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<int> acquires_allowed_; +}; + +// 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: + PosixSequentialFile(std::string filename, int fd) + : fd_(fd), filename_(filename) {} + ~PosixSequentialFile() override { close(fd_); } + + 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; + } + + Status Skip(uint64_t n) override { + if (::lseek(fd_, n, SEEK_CUR) == static_cast<off_t>(-1)) { + return PosixError(filename_, errno); + } + return Status::OK(); + } + + 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. + } + } + + ~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<off_t>(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<void*>(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() { + Status status; + if (!is_manifest_) { + return status; + } + + int fd = ::open(dirname_.c_str(), O_RDONLY | kOpenBaseFlags); + if (fd < 0) { + status = PosixError(dirname_, errno); + } else { + status = SyncFd(fd, dirname_); + ::close(fd); + } + return status; + } + + // 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(); + } +#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 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_. +}; + +int LockOrUnlock(int fd, bool lock) { + errno = 0; + 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: + 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_; +}; + +// 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<std::string> locked_files_ GUARDED_BY(mu_); +}; + +class PosixEnv : public Env { + public: + PosixEnv(); + ~PosixEnv() override { + static const char msg[] = + "PosixEnv singleton destroyed. Unsupported behavior!\n"; + std::fwrite(msg, 1, sizeof(msg), stderr); + std::abort(); + } + + Status NewSequentialFile(const std::string& filename, + SequentialFile** result) override { + int fd = ::open(filename.c_str(), O_RDONLY | kOpenBaseFlags); + if (fd < 0) { + *result = nullptr; + return PosixError(filename, errno); + } + + *result = new PosixSequentialFile(filename, fd); + return Status::OK(); + } + + Status NewRandomAccessFile(const std::string& filename, + RandomAccessFile** result) override { + *result = nullptr; + int fd = ::open(filename.c_str(), O_RDONLY | kOpenBaseFlags); + if (fd < 0) { + return PosixError(filename, errno); + } + + 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<char*>(mmap_base), + file_size, &mmap_limiter_); + } else { + status = PosixError(filename, errno); + } + } + ::close(fd); + if (!status.ok()) { + mmap_limiter_.Release(); + } + return status; + } + + 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); + } + + *result = new PosixWritableFile(filename, fd); + return Status::OK(); + } + + 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); + } + + *result = new PosixWritableFile(filename, fd); + return Status::OK(); + } + + bool FileExists(const std::string& filename) override { + return ::access(filename.c_str(), F_OK) == 0; + } + + Status GetChildren(const std::string& directory_path, + std::vector<std::string>* 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 RemoveFile(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 RemoveDir(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<PosixFileLock*>(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 { + std::thread new_thread(thread_main, thread_main_arg); + new_thread.detach(); + } + + Status GetTestDirectory(std::string* result) override { + const char* env = std::getenv("TEST_TMPDIR"); + if (env && env[0] != '\0') { + *result = env; + } else { + char buf[100]; + std::snprintf(buf, sizeof(buf), "/tmp/leveldbtest-%d", + static_cast<int>(::geteuid())); + *result = buf; + } + + // The CreateDir status is ignored because the directory may already exist. + CreateDir(*result); + + return Status::OK(); + } + + 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); + } + + std::FILE* fp = ::fdopen(fd, "w"); + if (fp == nullptr) { + ::close(fd); + *result = nullptr; + return PosixError(filename, errno); + } else { + *result = new PosixLogger(fp); + return Status::OK(); + } + } + + uint64_t NowMicros() override { + static constexpr uint64_t kUsecondsPerSecond = 1000000; + struct ::timeval tv; + ::gettimeofday(&tv, nullptr); + return static_cast<uint64_t>(tv.tv_sec) * kUsecondsPerSecond + tv.tv_usec; + } + + void SleepForMicroseconds(int micros) override { + std::this_thread::sleep_for(std::chrono::microseconds(micros)); + } + + private: + void BackgroundThreadMain(); + + static void BackgroundThreadEntryPoint(PosixEnv* env) { + env->BackgroundThreadMain(); + } + + // 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) {} + + void (*const function)(void*); + void* const arg; + }; + + port::Mutex background_work_mutex_; + port::CondVar background_work_cv_ GUARDED_BY(background_work_mutex_); + bool started_background_thread_ GUARDED_BY(background_work_mutex_); + + std::queue<BackgroundWorkItem> background_work_queue_ + GUARDED_BY(background_work_mutex_); + + PosixLockTable locks_; // Thread-safe. + Limiter mmap_limiter_; // Thread-safe. + Limiter fd_limiter_; // Thread-safe. +}; + +// Return the maximum number of concurrent mmaps. +int MaxMmaps() { return g_mmap_limit; } + +// 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; + } + 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<int>::max(); + } else { + // Allow use of 20% of available file descriptors for read-only files. + g_open_read_only_file_limit = rlim.rlim_cur / 5; + } + return g_open_read_only_file_limit; +} + +} // namespace + +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<PlatformEnv>; +// void ConfigurePosixEnv(int param) { +// PlatformSingletonEnv::AssertEnvNotInitialized(); +// // set global configuration flags. +// } +// Env* Env::Default() { +// static PlatformSingletonEnv default_env; +// return default_env.env(); +// } +template <typename EnvType> +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*>(&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<sizeof(EnvType), alignof(EnvType)>::type + env_storage_; +#if !defined(NDEBUG) + static std::atomic<bool> env_initialized_; +#endif // !defined(NDEBUG) +}; + +#if !defined(NDEBUG) +template <typename EnvType> +std::atomic<bool> SingletonEnv<EnvType>::env_initialized_; +#endif // !defined(NDEBUG) + +using PosixDefaultEnv = SingletonEnv<PosixEnv>; + +} // namespace + +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() { + 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..fa96e2e --- /dev/null +++ b/util/env_posix_test.cc @@ -0,0 +1,333 @@ +// 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 <sys/resource.h> +#include <sys/wait.h> +#include <unistd.h> + +#include <cstdio> +#include <cstdlib> +#include <cstring> +#include <string> +#include <unordered_set> +#include <vector> + +#include "gtest/gtest.h" +#include "leveldb/env.h" +#include "port/port.h" +#include "util/env_posix_test_helper.h" +#include "util/testutil.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<char>* GetArgvZero() { + static std::vector<char> 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<int>* 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<int>& baseline_open_fds, int* result_fd) { + std::unordered_set<int> 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<int>& 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<char> 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 testing::Test { + 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_F(EnvPosixTest, TestOpenOnRead) { + // Write some test data to a single file that will be opened |n| times. + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file = test_dir + "/open_on_read.txt"; + + FILE* f = std::fopen(test_file.c_str(), "we"); + ASSERT_TRUE(f != nullptr); + const char kFileData[] = "abcdefghijklmnopqrstuvwxyz"; + fputs(kFileData, f); + std::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_LEVELDB_OK(env_->NewRandomAccessFile(test_file, &files[i])); + } + char scratch; + Slice read_result; + for (int i = 0; i < kNumFiles; i++) { + ASSERT_LEVELDB_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_LEVELDB_OK(env_->RemoveFile(test_file)); +} + +#if HAVE_O_CLOEXEC + +TEST_F(EnvPosixTest, TestCloseOnExecSequentialFile) { + std::unordered_set<int> open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_sequential.txt"; + ASSERT_LEVELDB_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::SequentialFile* file = nullptr; + ASSERT_LEVELDB_OK(env_->NewSequentialFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_LEVELDB_OK(env_->RemoveFile(file_path)); +} + +TEST_F(EnvPosixTest, TestCloseOnExecRandomAccessFile) { + std::unordered_set<int> open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_random_access.txt"; + ASSERT_LEVELDB_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_LEVELDB_OK(env_->NewRandomAccessFile(file_path, &mmapped_files[i])); + } + + leveldb::RandomAccessFile* file = nullptr; + ASSERT_LEVELDB_OK(env_->NewRandomAccessFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + for (int i = 0; i < kReadOnlyFileLimit; i++) { + delete mmapped_files[i]; + } + ASSERT_LEVELDB_OK(env_->RemoveFile(file_path)); +} + +TEST_F(EnvPosixTest, TestCloseOnExecWritableFile) { + std::unordered_set<int> open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_writable.txt"; + ASSERT_LEVELDB_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::WritableFile* file = nullptr; + ASSERT_LEVELDB_OK(env_->NewWritableFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_LEVELDB_OK(env_->RemoveFile(file_path)); +} + +TEST_F(EnvPosixTest, TestCloseOnExecAppendableFile) { + std::unordered_set<int> open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_appendable.txt"; + ASSERT_LEVELDB_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::WritableFile* file = nullptr; + ASSERT_LEVELDB_OK(env_->NewAppendableFile(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_LEVELDB_OK(env_->RemoveFile(file_path)); +} + +TEST_F(EnvPosixTest, TestCloseOnExecLockFile) { + std::unordered_set<int> open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_lock.txt"; + ASSERT_LEVELDB_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::FileLock* lock = nullptr; + ASSERT_LEVELDB_OK(env_->LockFile(file_path, &lock)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + ASSERT_LEVELDB_OK(env_->UnlockFile(lock)); + + ASSERT_LEVELDB_OK(env_->RemoveFile(file_path)); +} + +TEST_F(EnvPosixTest, TestCloseOnExecLogger) { + std::unordered_set<int> open_fds; + GetOpenFileDescriptors(&open_fds); + + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string file_path = test_dir + "/close_on_exec_logger.txt"; + ASSERT_LEVELDB_OK(WriteStringToFile(env_, "0123456789", file_path)); + + leveldb::Logger* file = nullptr; + ASSERT_LEVELDB_OK(env_->NewLogger(file_path, &file)); + CheckCloseOnExecDoesNotLeakFDs(open_fds); + delete file; + + ASSERT_LEVELDB_OK(env_->RemoveFile(file_path)); +} + +#endif // HAVE_O_CLOEXEC + +} // namespace leveldb + 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 new file mode 100644 index 0000000..97513f6 --- /dev/null +++ b/util/env_test.cc @@ -0,0 +1,236 @@ +// 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 "leveldb/env.h" + +#include <algorithm> + +#include "gtest/gtest.h" +#include "port/port.h" +#include "port/thread_annotations.h" +#include "util/mutexlock.h" +#include "util/testutil.h" + +namespace leveldb { + +static const int kDelayMicros = 100000; + +class EnvTest : public testing::Test { + public: + EnvTest() : env_(Env::Default()) {} + + Env* env_; +}; + +TEST_F(EnvTest, ReadWrite) { + Random rnd(test::RandomSeed()); + + // Get file to use for testing. + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file_name = test_dir + "/open_on_read.txt"; + WritableFile* writable_file; + ASSERT_LEVELDB_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_LEVELDB_OK(writable_file->Append(r)); + data += r; + if (rnd.OneIn(10)) { + ASSERT_LEVELDB_OK(writable_file->Flush()); + } + } + ASSERT_LEVELDB_OK(writable_file->Sync()); + ASSERT_LEVELDB_OK(writable_file->Close()); + delete writable_file; + + // Read all data using a sequence of randomly sized reads. + SequentialFile* sequential_file; + ASSERT_LEVELDB_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<int>(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_LEVELDB_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_F(EnvTest, RunImmediately) { + struct RunState { + port::Mutex mu; + port::CondVar cvar{&mu}; + bool called = false; + + static void Run(void* arg) { + RunState* state = reinterpret_cast<RunState*>(arg); + MutexLock l(&state->mu); + ASSERT_EQ(state->called, false); + state->called = true; + state->cvar.Signal(); + } + }; + + RunState state; + env_->Schedule(&RunState::Run, &state); + + MutexLock l(&state.mu); + while (!state.called) { + state.cvar.Wait(); + } +} + +TEST_F(EnvTest, RunMany) { + struct RunState { + port::Mutex mu; + port::CondVar cvar{&mu}; + int last_id = 0; + }; + + struct Callback { + RunState* state_; // Pointer to shared state. + const int id_; // Order# for the execution of this callback. + + Callback(RunState* s, int id) : state_(s), id_(id) {} + + static void Run(void* arg) { + Callback* callback = reinterpret_cast<Callback*>(arg); + RunState* state = callback->state_; + + MutexLock l(&state->mu); + ASSERT_EQ(state->last_id, callback->id_ - 1); + state->last_id = callback->id_; + state->cvar.Signal(); + } + }; + + RunState state; + Callback callback1(&state, 1); + Callback callback2(&state, 2); + Callback callback3(&state, 3); + Callback callback4(&state, 4); + env_->Schedule(&Callback::Run, &callback1); + env_->Schedule(&Callback::Run, &callback2); + env_->Schedule(&Callback::Run, &callback3); + env_->Schedule(&Callback::Run, &callback4); + + MutexLock l(&state.mu); + while (state.last_id != 4) { + state.cvar.Wait(); + } +} + +struct State { + port::Mutex mu; + port::CondVar cvar{&mu}; + + 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) { + State* s = reinterpret_cast<State*>(arg); + s->mu.Lock(); + s->val += 1; + s->num_running -= 1; + s->cvar.Signal(); + s->mu.Unlock(); +} + +TEST_F(EnvTest, StartThread) { + State state(0, 3); + for (int i = 0; i < 3; i++) { + env_->StartThread(&ThreadBody, &state); + } + + MutexLock l(&state.mu); + while (state.num_running != 0) { + state.cvar.Wait(); + } + ASSERT_EQ(state.val, 3); +} + +TEST_F(EnvTest, TestOpenNonExistentFile) { + // Write some test data to a single file that will be opened |n| times. + std::string test_dir; + ASSERT_LEVELDB_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_F(EnvTest, ReopenWritableFile) { + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file_name = test_dir + "/reopen_writable_file.txt"; + env_->RemoveFile(test_file_name); + + WritableFile* writable_file; + ASSERT_LEVELDB_OK(env_->NewWritableFile(test_file_name, &writable_file)); + std::string data("hello world!"); + ASSERT_LEVELDB_OK(writable_file->Append(data)); + ASSERT_LEVELDB_OK(writable_file->Close()); + delete writable_file; + + ASSERT_LEVELDB_OK(env_->NewWritableFile(test_file_name, &writable_file)); + data = "42"; + ASSERT_LEVELDB_OK(writable_file->Append(data)); + ASSERT_LEVELDB_OK(writable_file->Close()); + delete writable_file; + + ASSERT_LEVELDB_OK(ReadFileToString(env_, test_file_name, &data)); + ASSERT_EQ(std::string("42"), data); + env_->RemoveFile(test_file_name); +} + +TEST_F(EnvTest, ReopenAppendableFile) { + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file_name = test_dir + "/reopen_appendable_file.txt"; + env_->RemoveFile(test_file_name); + + WritableFile* appendable_file; + ASSERT_LEVELDB_OK(env_->NewAppendableFile(test_file_name, &appendable_file)); + std::string data("hello world!"); + ASSERT_LEVELDB_OK(appendable_file->Append(data)); + ASSERT_LEVELDB_OK(appendable_file->Close()); + delete appendable_file; + + ASSERT_LEVELDB_OK(env_->NewAppendableFile(test_file_name, &appendable_file)); + data = "42"; + ASSERT_LEVELDB_OK(appendable_file->Append(data)); + ASSERT_LEVELDB_OK(appendable_file->Close()); + delete appendable_file; + + ASSERT_LEVELDB_OK(ReadFileToString(env_, test_file_name, &data)); + ASSERT_EQ(std::string("hello world!42"), data); + env_->RemoveFile(test_file_name); +} + +} // namespace leveldb + diff --git a/util/env_windows.cc b/util/env_windows.cc new file mode 100644 index 0000000..489953e --- /dev/null +++ b/util/env_windows.cc @@ -0,0 +1,795 @@ +// 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 <algorithm> +#include <atomic> +#include <chrono> +#include <condition_variable> +#include <cstddef> +#include <cstdint> +#include <cstdlib> +#include <cstring> +#include <memory> +#include <mutex> +#include <queue> +#include <sstream> +#include <string> +#include <vector> + +#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" + +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; + +// Can be set by 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<char*>(&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(const ScopedHandle&) = delete; + ScopedHandle(ScopedHandle&& other) noexcept : handle_(other.Release()) {} + ~ScopedHandle() { Close(); } + + ScopedHandle& operator=(const ScopedHandle&) = delete; + + 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<int> acquires_allowed_; +}; + +class WindowsSequentialFile : public SequentialFile { + public: + WindowsSequentialFile(std::string filename, ScopedHandle handle) + : handle_(std::move(handle)), filename_(std::move(filename)) {} + ~WindowsSequentialFile() override {} + + Status Read(size_t n, Slice* result, char* scratch) override { + 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<DWORD>::max()); + if (!::ReadFile(handle_.get(), scratch, static_cast<DWORD>(n), &bytes_read, + nullptr)) { + return WindowsError(filename_, ::GetLastError()); + } + + *result = Slice(scratch, bytes_read); + return Status::OK(); + } + + Status Skip(uint64_t n) override { + LARGE_INTEGER distance; + distance.QuadPart = n; + if (!::SetFilePointerEx(handle_.get(), distance, nullptr, FILE_CURRENT)) { + return WindowsError(filename_, ::GetLastError()); + } + return Status::OK(); + } + + private: + const ScopedHandle handle_; + const std::string filename_; +}; + +class WindowsRandomAccessFile : public RandomAccessFile { + public: + WindowsRandomAccessFile(std::string filename, ScopedHandle handle) + : handle_(std::move(handle)), filename_(std::move(filename)) {} + + ~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<DWORD>(offset >> 32); + overlapped.Offset = static_cast<DWORD>(offset); + if (!::ReadFile(handle_.get(), scratch, static_cast<DWORD>(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: + const ScopedHandle handle_; + const std::string filename_; +}; + +class WindowsMmapReadableFile : public RandomAccessFile { + public: + // base[0,length-1] contains the mmapped contents of the file. + WindowsMmapReadableFile(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)) {} + + ~WindowsMmapReadableFile() override { + ::UnmapViewOfFile(mmap_base_); + mmap_limiter_->Release(); + } + + Status Read(uint64_t offset, size_t n, Slice* result, + char* scratch) const override { + if (offset + n > length_) { + *result = Slice(); + return WindowsError(filename_, ERROR_INVALID_PARAMETER); + } + + *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 WindowsWritableFile : public WritableFile { + public: + WindowsWritableFile(std::string filename, ScopedHandle handle) + : pos_(0), handle_(std::move(handle)), filename_(std::move(filename)) {} + + ~WindowsWritableFile() override = default; + + 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(); + if (!handle_.Close() && status.ok()) { + status = WindowsError(filename_, ::GetLastError()); + } + return status; + } + + Status Flush() override { return FlushBuffer(); } + + Status Sync() override { + // On Windows no need to sync parent directory. Its metadata will be updated + // via the creation of the new file, without an explicit sync. + + Status status = FlushBuffer(); + if (!status.ok()) { + return status; + } + + if (!::FlushFileBuffers(handle_.get())) { + return Status::IOError(filename_, + GetWindowsErrorMessage(::GetLastError())); + } + return Status::OK(); + } + + private: + Status FlushBuffer() { + Status status = WriteUnbuffered(buf_, pos_); + pos_ = 0; + return status; + } + + Status WriteUnbuffered(const char* data, size_t size) { + DWORD bytes_written; + if (!::WriteFile(handle_.get(), data, static_cast<DWORD>(size), + &bytes_written, nullptr)) { + return Status::IOError(filename_, + GetWindowsErrorMessage(::GetLastError())); + } + return Status::OK(); + } + + // buf_[0, pos_-1] contains data to be written to handle_. + char buf_[kWritableFileBufferSize]; + size_t pos_; + + ScopedHandle handle_; + const std::string filename_; +}; + +// 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 filename) + : handle_(std::move(handle)), filename_(std::move(filename)) {} + + const ScopedHandle& handle() const { return handle_; } + const std::string& filename() const { return filename_; } + + private: + const ScopedHandle handle_; + const std::string filename_; +}; + +class WindowsEnv : public Env { + public: + WindowsEnv(); + ~WindowsEnv() override { + static const char msg[] = + "WindowsEnv singleton destroyed. Unsupported behavior!\n"; + std::fwrite(msg, 1, sizeof(msg), stderr); + std::abort(); + } + + Status NewSequentialFile(const std::string& filename, + SequentialFile** result) override { + *result = nullptr; + DWORD desired_access = GENERIC_READ; + DWORD share_mode = FILE_SHARE_READ; + ScopedHandle handle = ::CreateFileA( + filename.c_str(), desired_access, share_mode, + /*lpSecurityAttributes=*/nullptr, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, + /*hTemplateFile=*/nullptr); + if (!handle.is_valid()) { + return WindowsError(filename, ::GetLastError()); + } + + *result = new WindowsSequentialFile(filename, std::move(handle)); + return Status::OK(); + } + + Status NewRandomAccessFile(const std::string& filename, + RandomAccessFile** result) override { + *result = nullptr; + DWORD desired_access = GENERIC_READ; + DWORD share_mode = FILE_SHARE_READ; + ScopedHandle handle = + ::CreateFileA(filename.c_str(), desired_access, share_mode, + /*lpSecurityAttributes=*/nullptr, OPEN_EXISTING, + FILE_ATTRIBUTE_READONLY, + /*hTemplateFile=*/nullptr); + if (!handle.is_valid()) { + return WindowsError(filename, ::GetLastError()); + } + if (!mmap_limiter_.Acquire()) { + *result = new WindowsRandomAccessFile(filename, std::move(handle)); + return Status::OK(); + } + + LARGE_INTEGER file_size; + Status status; + if (!::GetFileSizeEx(handle.get(), &file_size)) { + mmap_limiter_.Release(); + return WindowsError(filename, ::GetLastError()); + } + + ScopedHandle mapping = + ::CreateFileMappingA(handle.get(), + /*security attributes=*/nullptr, PAGE_READONLY, + /*dwMaximumSizeHigh=*/0, + /*dwMaximumSizeLow=*/0, + /*lpName=*/nullptr); + if (mapping.is_valid()) { + void* mmap_base = ::MapViewOfFile(mapping.get(), FILE_MAP_READ, + /*dwFileOffsetHigh=*/0, + /*dwFileOffsetLow=*/0, + /*dwNumberOfBytesToMap=*/0); + if (mmap_base) { + *result = new WindowsMmapReadableFile( + filename, reinterpret_cast<char*>(mmap_base), + static_cast<size_t>(file_size.QuadPart), &mmap_limiter_); + return Status::OK(); + } + } + mmap_limiter_.Release(); + return WindowsError(filename, ::GetLastError()); + } + + Status NewWritableFile(const std::string& filename, + WritableFile** result) override { + DWORD desired_access = GENERIC_WRITE; + DWORD share_mode = 0; // Exclusive access. + ScopedHandle handle = ::CreateFileA( + filename.c_str(), desired_access, share_mode, + /*lpSecurityAttributes=*/nullptr, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, + /*hTemplateFile=*/nullptr); + if (!handle.is_valid()) { + *result = nullptr; + return WindowsError(filename, ::GetLastError()); + } + + *result = new WindowsWritableFile(filename, std::move(handle)); + return Status::OK(); + } + + Status NewAppendableFile(const std::string& filename, + WritableFile** result) override { + DWORD desired_access = FILE_APPEND_DATA; + DWORD share_mode = 0; // Exclusive access. + ScopedHandle handle = ::CreateFileA( + filename.c_str(), desired_access, share_mode, + /*lpSecurityAttributes=*/nullptr, OPEN_ALWAYS, FILE_ATTRIBUTE_NORMAL, + /*hTemplateFile=*/nullptr); + if (!handle.is_valid()) { + *result = nullptr; + return WindowsError(filename, ::GetLastError()); + } + + *result = new WindowsWritableFile(filename, std::move(handle)); + return Status::OK(); + } + + bool FileExists(const std::string& filename) override { + return GetFileAttributesA(filename.c_str()) != INVALID_FILE_ATTRIBUTES; + } + + Status GetChildren(const std::string& directory_path, + std::vector<std::string>* result) override { + const std::string find_pattern = directory_path + "\\*"; + 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(directory_path, 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(directory_path, last_error); + } + return Status::OK(); + } + + Status RemoveFile(const std::string& filename) override { + if (!::DeleteFileA(filename.c_str())) { + return WindowsError(filename, ::GetLastError()); + } + return Status::OK(); + } + + Status CreateDir(const std::string& dirname) override { + if (!::CreateDirectoryA(dirname.c_str(), nullptr)) { + return WindowsError(dirname, ::GetLastError()); + } + return Status::OK(); + } + + Status RemoveDir(const std::string& dirname) override { + if (!::RemoveDirectoryA(dirname.c_str())) { + return WindowsError(dirname, ::GetLastError()); + } + return Status::OK(); + } + + Status GetFileSize(const std::string& filename, uint64_t* size) override { + WIN32_FILE_ATTRIBUTE_DATA file_attributes; + if (!::GetFileAttributesExA(filename.c_str(), GetFileExInfoStandard, + &file_attributes)) { + return WindowsError(filename, ::GetLastError()); + } + ULARGE_INTEGER file_size; + file_size.HighPart = file_attributes.nFileSizeHigh; + file_size.LowPart = file_attributes.nFileSizeLow; + *size = file_size.QuadPart; + return Status::OK(); + } + + Status RenameFile(const std::string& from, const std::string& to) override { + // Try a simple move first. It will only succeed when |to| doesn't already + // exist. + if (::MoveFileA(from.c_str(), to.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| 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(to.c_str(), from.c_str(), /*lpBackupFileName=*/nullptr, + REPLACEFILE_IGNORE_MERGE_ERRORS, + /*lpExclude=*/nullptr, /*lpReserved=*/nullptr)) { + return Status::OK(); + } + DWORD replace_error = ::GetLastError(); + // In the case of FILE_ERROR_NOT_FOUND from ReplaceFile, it is likely that + // |to| 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(from, move_error); + } else { + return WindowsError(from, replace_error); + } + } + + Status LockFile(const std::string& filename, FileLock** lock) override { + *lock = nullptr; + Status result; + ScopedHandle handle = ::CreateFileA( + filename.c_str(), GENERIC_READ | GENERIC_WRITE, FILE_SHARE_READ, + /*lpSecurityAttributes=*/nullptr, OPEN_ALWAYS, FILE_ATTRIBUTE_NORMAL, + nullptr); + if (!handle.is_valid()) { + result = WindowsError(filename, ::GetLastError()); + } else if (!LockOrUnlock(handle.get(), true)) { + result = WindowsError("lock " + filename, ::GetLastError()); + } else { + *lock = new WindowsFileLock(std::move(handle), filename); + } + return result; + } + + Status UnlockFile(FileLock* lock) override { + WindowsFileLock* windows_file_lock = + reinterpret_cast<WindowsFileLock*>(lock); + if (!LockOrUnlock(windows_file_lock->handle().get(), false)) { + return WindowsError("unlock " + windows_file_lock->filename(), + ::GetLastError()); + } + delete windows_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 { + std::thread new_thread(thread_main, thread_main_arg); + new_thread.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(filename, ::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<uint64_t>(ft.dwHighDateTime) << 32) + ft.dwLowDateTime; + return num_ticks / 10; + } + + void SleepForMicroseconds(int micros) override { + std::this_thread::sleep_for(std::chrono::microseconds(micros)); + } + + private: + void BackgroundThreadMain(); + + static void BackgroundThreadEntryPoint(WindowsEnv* env) { + env->BackgroundThreadMain(); + } + + // 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) {} + + void (*const function)(void*); + void* const arg; + }; + + port::Mutex background_work_mutex_; + port::CondVar background_work_cv_ GUARDED_BY(background_work_mutex_); + bool started_background_thread_ GUARDED_BY(background_work_mutex_); + + std::queue<BackgroundWorkItem> background_work_queue_ + GUARDED_BY(background_work_mutex_); + + Limiter mmap_limiter_; // Thread-safe. +}; + +// Return the maximum number of concurrent mmaps. +int MaxMmaps() { return g_mmap_limit; } + +WindowsEnv::WindowsEnv() + : background_work_cv_(&background_work_mutex_), + started_background_thread_(false), + mmap_limiter_(MaxMmaps()) {} + +void WindowsEnv::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(WindowsEnv::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 WindowsEnv::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); + } +} + +// Wraps an Env instance whose destructor is never created. +// +// Intended usage: +// using PlatformSingletonEnv = SingletonEnv<PlatformEnv>; +// void ConfigurePosixEnv(int param) { +// PlatformSingletonEnv::AssertEnvNotInitialized(); +// // set global configuration flags. +// } +// Env* Env::Default() { +// static PlatformSingletonEnv default_env; +// return default_env.env(); +// } +template <typename EnvType> +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*>(&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<sizeof(EnvType), alignof(EnvType)>::type + env_storage_; +#if !defined(NDEBUG) + static std::atomic<bool> env_initialized_; +#endif // !defined(NDEBUG) +}; + +#if !defined(NDEBUG) +template <typename EnvType> +std::atomic<bool> SingletonEnv<EnvType>::env_initialized_; +#endif // !defined(NDEBUG) + +using WindowsDefaultEnv = SingletonEnv<WindowsEnv>; + +} // namespace + +void EnvWindowsTestHelper::SetReadOnlyMMapLimit(int limit) { + WindowsDefaultEnv::AssertEnvNotInitialized(); + g_mmap_limit = limit; +} + +Env* Env::Default() { + static WindowsDefaultEnv env_container; + return env_container.env(); +} + +} // namespace leveldb diff --git a/util/env_windows_test.cc b/util/env_windows_test.cc new file mode 100644 index 0000000..e5dc490 --- /dev/null +++ b/util/env_windows_test.cc @@ -0,0 +1,59 @@ +// 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 "gtest/gtest.h" +#include "leveldb/env.h" +#include "port/port.h" +#include "util/env_windows_test_helper.h" +#include "util/testutil.h" + +namespace leveldb { + +static const int kMMapLimit = 4; + +class EnvWindowsTest : public testing::Test { + public: + static void SetFileLimits(int mmap_limit) { + EnvWindowsTestHelper::SetReadOnlyMMapLimit(mmap_limit); + } + + EnvWindowsTest() : env_(Env::Default()) {} + + Env* env_; +}; + +TEST_F(EnvWindowsTest, TestOpenOnRead) { + // Write some test data to a single file that will be opened |n| times. + std::string test_dir; + ASSERT_LEVELDB_OK(env_->GetTestDirectory(&test_dir)); + std::string test_file = test_dir + "/open_on_read.txt"; + + FILE* f = std::fopen(test_file.c_str(), "w"); + ASSERT_TRUE(f != nullptr); + const char kFileData[] = "abcdefghijklmnopqrstuvwxyz"; + fputs(kFileData, f); + std::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_LEVELDB_OK(env_->NewRandomAccessFile(test_file, &files[i])); + } + char scratch; + Slice read_result; + for (int i = 0; i < kNumFiles; i++) { + ASSERT_LEVELDB_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_LEVELDB_OK(env_->RemoveFile(test_file)); +} + +} // namespace leveldb + 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 new file mode 100644 index 0000000..90fd754 --- /dev/null +++ b/util/filter_policy.cc @@ -0,0 +1,11 @@ +// Copyright (c) 2012 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#include "leveldb/filter_policy.h" + +namespace leveldb { + +FilterPolicy::~FilterPolicy() {} + +} // namespace leveldb diff --git a/util/hash.cc b/util/hash.cc new file mode 100644 index 0000000..8122fa8 --- /dev/null +++ b/util/hash.cc @@ -0,0 +1,55 @@ +// 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 "util/hash.h" + +#include <cstring> + +#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) +#endif + +namespace leveldb { + +uint32_t Hash(const char* data, size_t n, uint32_t seed) { + // Similar to murmur hash + const uint32_t m = 0xc6a4a793; + const uint32_t r = 24; + const char* limit = data + n; + uint32_t h = seed ^ (n * m); + + // Pick up four bytes at a time + while (data + 4 <= limit) { + uint32_t w = DecodeFixed32(data); + data += 4; + h += w; + h *= m; + h ^= (h >> 16); + } + + // Pick up remaining bytes + switch (limit - data) { + case 3: + h += static_cast<uint8_t>(data[2]) << 16; + FALLTHROUGH_INTENDED; + case 2: + h += static_cast<uint8_t>(data[1]) << 8; + FALLTHROUGH_INTENDED; + case 1: + h += static_cast<uint8_t>(data[0]); + h *= m; + h ^= (h >> r); + break; + } + return h; +} + +} // namespace leveldb diff --git a/util/hash.h b/util/hash.h new file mode 100644 index 0000000..87ab279 --- /dev/null +++ b/util/hash.h @@ -0,0 +1,19 @@ +// 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. +// +// Simple hash function used for internal data structures + +#ifndef STORAGE_LEVELDB_UTIL_HASH_H_ +#define STORAGE_LEVELDB_UTIL_HASH_H_ + +#include <cstddef> +#include <cstdint> + +namespace leveldb { + +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 new file mode 100644 index 0000000..28eede9 --- /dev/null +++ b/util/hash_test.cc @@ -0,0 +1,42 @@ +// 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 "util/hash.h" + +#include "gtest/gtest.h" + +namespace leveldb { + +TEST(HASH, SignedUnsignedIssue) { + const uint8_t data1[1] = {0x62}; + const uint8_t data2[2] = {0xc3, 0x97}; + const uint8_t data3[3] = {0xe2, 0x99, 0xa5}; + const uint8_t data4[4] = {0xe1, 0x80, 0xb9, 0x32}; + const uint8_t 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, + }; + + ASSERT_EQ(Hash(0, 0, 0xbc9f1d34), 0xbc9f1d34); + ASSERT_EQ( + Hash(reinterpret_cast<const char*>(data1), sizeof(data1), 0xbc9f1d34), + 0xef1345c4); + ASSERT_EQ( + Hash(reinterpret_cast<const char*>(data2), sizeof(data2), 0xbc9f1d34), + 0x5b663814); + ASSERT_EQ( + Hash(reinterpret_cast<const char*>(data3), sizeof(data3), 0xbc9f1d34), + 0x323c078f); + ASSERT_EQ( + Hash(reinterpret_cast<const char*>(data4), sizeof(data4), 0xbc9f1d34), + 0xed21633a); + ASSERT_EQ( + Hash(reinterpret_cast<const char*>(data5), sizeof(data5), 0x12345678), + 0xf333dabb); +} + +} // namespace leveldb + diff --git a/util/histogram.cc b/util/histogram.cc new file mode 100644 index 0000000..7af4030 --- /dev/null +++ b/util/histogram.cc @@ -0,0 +1,272 @@ +// 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 "util/histogram.h" + +#include <cmath> +#include <cstdio> + +#include "port/port.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, +}; + +void Histogram::Clear() { + min_ = kBucketLimit[kNumBuckets - 1]; + max_ = 0; + num_ = 0; + sum_ = 0; + sum_squares_ = 0; + for (int i = 0; i < kNumBuckets; i++) { + buckets_[i] = 0; + } +} + +void Histogram::Add(double value) { + // Linear search is fast enough for our usage in db_bench + int b = 0; + while (b < kNumBuckets - 1 && kBucketLimit[b] <= value) { + b++; + } + buckets_[b] += 1.0; + if (min_ > value) min_ = value; + if (max_ < value) max_ = value; + num_++; + sum_ += value; + sum_squares_ += (value * value); +} + +void Histogram::Merge(const Histogram& other) { + if (other.min_ < min_) min_ = other.min_; + if (other.max_ > max_) max_ = other.max_; + num_ += other.num_; + sum_ += other.sum_; + sum_squares_ += other.sum_squares_; + for (int b = 0; b < kNumBuckets; b++) { + buckets_[b] += other.buckets_[b]; + } +} + +double Histogram::Median() const { return Percentile(50.0); } + +double Histogram::Percentile(double p) const { + double threshold = num_ * (p / 100.0); + double sum = 0; + for (int b = 0; b < kNumBuckets; b++) { + sum += buckets_[b]; + if (sum >= threshold) { + // Scale linearly within this bucket + double left_point = (b == 0) ? 0 : kBucketLimit[b - 1]; + double right_point = kBucketLimit[b]; + double left_sum = sum - buckets_[b]; + double right_sum = sum; + double pos = (threshold - left_sum) / (right_sum - left_sum); + double r = left_point + (right_point - left_point) * pos; + if (r < min_) r = min_; + if (r > max_) r = max_; + return r; + } + } + return max_; +} + +double Histogram::Average() const { + if (num_ == 0.0) return 0; + return sum_ / num_; +} + +double Histogram::StandardDeviation() const { + if (num_ == 0.0) return 0; + double variance = (sum_squares_ * num_ - sum_ * sum_) / (num_ * num_); + return sqrt(variance); +} + +std::string Histogram::ToString() const { + std::string r; + char buf[200]; + std::snprintf(buf, sizeof(buf), "Count: %.0f Average: %.4f StdDev: %.2f\n", + num_, Average(), StandardDeviation()); + r.append(buf); + std::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"); + const double mult = 100.0 / num_; + double sum = 0; + for (int b = 0; b < kNumBuckets; b++) { + if (buckets_[b] <= 0.0) continue; + sum += buckets_[b]; + std::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<int>(20 * (buckets_[b] / num_) + 0.5); + r.append(marks, '#'); + r.push_back('\n'); + } + return r; +} + +} // namespace leveldb diff --git a/util/histogram.h b/util/histogram.h new file mode 100644 index 0000000..4da60fb --- /dev/null +++ b/util/histogram.h @@ -0,0 +1,44 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_UTIL_HISTOGRAM_H_ +#define STORAGE_LEVELDB_UTIL_HISTOGRAM_H_ + +#include <string> + +namespace leveldb { + +class Histogram { + public: + Histogram() {} + ~Histogram() {} + + void Clear(); + void Add(double value); + void Merge(const Histogram& other); + + 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_; + + double buckets_[kNumBuckets]; +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_HISTOGRAM_H_ diff --git a/util/logging.cc b/util/logging.cc new file mode 100644 index 0000000..8d6fb5b --- /dev/null +++ b/util/logging.cc @@ -0,0 +1,82 @@ +// 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 "util/logging.h" + +#include <cstdarg> +#include <cstdio> +#include <cstdlib> +#include <limits> + +#include "leveldb/env.h" +#include "leveldb/slice.h" + +namespace leveldb { + +void AppendNumberTo(std::string* str, uint64_t num) { + char buf[30]; + std::snprintf(buf, sizeof(buf), "%llu", static_cast<unsigned long long>(num)); + str->append(buf); +} + +void AppendEscapedStringTo(std::string* str, const Slice& value) { + for (size_t i = 0; i < value.size(); i++) { + char c = value[i]; + if (c >= ' ' && c <= '~') { + str->push_back(c); + } else { + char buf[10]; + std::snprintf(buf, sizeof(buf), "\\x%02x", + static_cast<unsigned int>(c) & 0xff); + str->append(buf); + } + } +} + +std::string NumberToString(uint64_t num) { + std::string r; + AppendNumberTo(&r, num); + return r; +} + +std::string EscapeString(const Slice& value) { + std::string r; + AppendEscapedStringTo(&r, value); + return r; +} + +bool ConsumeDecimalNumber(Slice* in, uint64_t* val) { + // Constants that will be optimized away. + constexpr const uint64_t kMaxUint64 = std::numeric_limits<uint64_t>::max(); + constexpr const char kLastDigitOfMaxUint64 = + '0' + static_cast<char>(kMaxUint64 % 10); + + uint64_t value = 0; + + // reinterpret_cast-ing from char* to uint8_t* to avoid signedness. + const uint8_t* start = reinterpret_cast<const uint8_t*>(in->data()); + + const uint8_t* end = start + in->size(); + const uint8_t* current = start; + for (; current != end; ++current) { + const uint8_t 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 = 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 new file mode 100644 index 0000000..a0394b2 --- /dev/null +++ b/util/logging.h @@ -0,0 +1,44 @@ +// 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. +// +// Must not be included from any .h files to avoid polluting the namespace +// with macros. + +#ifndef STORAGE_LEVELDB_UTIL_LOGGING_H_ +#define STORAGE_LEVELDB_UTIL_LOGGING_H_ + +#include <cstdint> +#include <cstdio> +#include <string> + +#include "port/port.h" + +namespace leveldb { + +class Slice; +class WritableFile; + +// Append a human-readable printout of "num" to *str +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". +void AppendEscapedStringTo(std::string* str, const Slice& value); + +// Return a human-readable printout of "num" +std::string NumberToString(uint64_t num); + +// Return a human-readable version of "value". +// Escapes any non-printable characters found in "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. +bool ConsumeDecimalNumber(Slice* in, uint64_t* val); + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_LOGGING_H_ diff --git a/util/logging_test.cc b/util/logging_test.cc new file mode 100644 index 0000000..ee96047 --- /dev/null +++ b/util/logging_test.cc @@ -0,0 +1,141 @@ +// 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 "util/logging.h" + +#include <limits> +#include <string> + +#include "gtest/gtest.h" +#include "leveldb/slice.h" + +namespace leveldb { + +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<uint64_t>::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<uint64_t>::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<uint64_t>::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<uint64_t>::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 + diff --git a/util/mutexlock.h b/util/mutexlock.h new file mode 100644 index 0000000..0cb2e25 --- /dev/null +++ b/util/mutexlock.h @@ -0,0 +1,39 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_UTIL_MUTEXLOCK_H_ +#define STORAGE_LEVELDB_UTIL_MUTEXLOCK_H_ + +#include "port/port.h" +#include "port/thread_annotations.h" + +namespace leveldb { + +// Helper class that locks a mutex on construction and unlocks the mutex when +// the destructor of the MutexLock object is invoked. +// +// Typical usage: +// +// void MyClass::MyMethod() { +// MutexLock l(&mu_); // mu_ is an instance variable +// ... some complex code, possibly with multiple return paths ... +// } + +class SCOPED_LOCKABLE MutexLock { + public: + 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_; +}; + +} // 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 <type_traits> +#include <utility> + +namespace leveldb { + +// Wraps an instance whose destructor is never called. +// +// This is intended for use with function-level static variables. +template <typename InstanceType> +class NoDestructor { + public: + template <typename... ConstructorArgTypes> + 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<ConstructorArgTypes>(constructor_args)...); + } + + ~NoDestructor() = default; + + NoDestructor(const NoDestructor&) = delete; + NoDestructor& operator=(const NoDestructor&) = delete; + + InstanceType* get() { + return reinterpret_cast<InstanceType*>(&instance_storage_); + } + + private: + typename std::aligned_storage<sizeof(InstanceType), + alignof(InstanceType)>::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..8eacd5c --- /dev/null +++ b/util/no_destructor_test.cc @@ -0,0 +1,45 @@ +// 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 "util/no_destructor.h" + +#include <cstdint> +#include <cstdlib> +#include <utility> + +#include "gtest/gtest.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 + +TEST(NoDestructorTest, StackInstance) { + NoDestructor<DoNotDestruct> instance(kGoldenA, kGoldenB); + ASSERT_EQ(kGoldenA, instance.get()->a); + ASSERT_EQ(kGoldenB, instance.get()->b); +} + +TEST(NoDestructorTest, StaticInstance) { + static NoDestructor<DoNotDestruct> instance(kGoldenA, kGoldenB); + ASSERT_EQ(kGoldenA, instance.get()->a); + ASSERT_EQ(kGoldenB, instance.get()->b); +} + +} // namespace leveldb + diff --git a/util/options.cc b/util/options.cc new file mode 100644 index 0000000..62de5bf --- /dev/null +++ b/util/options.cc @@ -0,0 +1,14 @@ +// 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 "leveldb/options.h" + +#include "leveldb/comparator.h" +#include "leveldb/env.h" + +namespace leveldb { + +Options::Options() : comparator(BytewiseComparator()), env(Env::Default()) {} + +} // namespace leveldb diff --git a/util/posix_logger.h b/util/posix_logger.h new file mode 100644 index 0000000..6bbc1a0 --- /dev/null +++ b/util/posix_logger.h @@ -0,0 +1,130 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. +// +// Logger implementation that can be shared by all environments +// where enough posix functionality is available. + +#ifndef STORAGE_LEVELDB_UTIL_POSIX_LOGGER_H_ +#define STORAGE_LEVELDB_UTIL_POSIX_LOGGER_H_ + +#include <sys/time.h> + +#include <cassert> +#include <cstdarg> +#include <cstdio> +#include <ctime> +#include <sstream> +#include <thread> + +#include "leveldb/env.h" + +namespace leveldb { + +class PosixLogger final : public Logger { + public: + // 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); } + + ~PosixLogger() override { std::fclose(fp_); } + + void Logv(const char* format, std::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); + + // 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<size_t>(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 = std::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<int>(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 std::(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_POSIX_LOGGER_H_ diff --git a/util/random.h b/util/random.h new file mode 100644 index 0000000..fe76ab4 --- /dev/null +++ b/util/random.h @@ -0,0 +1,63 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_UTIL_RANDOM_H_ +#define STORAGE_LEVELDB_UTIL_RANDOM_H_ + +#include <cstdint> + +namespace leveldb { + +// A very simple random number generator. Not especially good at +// generating truly random bits, but good enough for our needs in this +// package. +class Random { + private: + uint32_t seed_; + + public: + explicit Random(uint32_t s) : seed_(s & 0x7fffffffu) { + // Avoid bad seeds. + if (seed_ == 0 || seed_ == 2147483647L) { + seed_ = 1; + } + } + 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 + // We are computing + // seed_ = (seed_ * A) % M, where M = 2^31-1 + // + // seed_ must not be zero or M, or else all subsequent computed values + // will be zero or M respectively. For all other values, seed_ will end + // up cycling through every number in [1,M-1] + uint64_t product = seed_ * A; + + // Compute (product % M) using the fact that ((x << 31) % M) == x. + seed_ = static_cast<uint32_t>((product >> 31) + (product & M)); + // The first reduction may overflow by 1 bit, so we may need to + // repeat. mod == M is not possible; using > allows the faster + // sign-bit-based test. + if (seed_ > M) { + seed_ -= M; + } + return seed_; + } + // Returns a uniformly distributed value in the range [0..n-1] + // REQUIRES: n > 0 + uint32_t Uniform(int n) { return Next() % n; } + + // Randomly returns true ~"1/n" of the time, and false otherwise. + // REQUIRES: n > 0 + bool OneIn(int n) { return (Next() % n) == 0; } + + // 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)); } +}; + +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_RANDOM_H_ diff --git a/util/status.cc b/util/status.cc new file mode 100644 index 0000000..0559f5b --- /dev/null +++ b/util/status.cc @@ -0,0 +1,77 @@ +// 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 "leveldb/status.h" + +#include <cstdio> + +#include "port/port.h" + +namespace leveldb { + +const char* Status::CopyState(const char* state) { + uint32_t size; + std::memcpy(&size, state, sizeof(size)); + char* result = new char[size + 5]; + std::memcpy(result, state, size + 5); + return result; +} + +Status::Status(Code code, const Slice& msg, const Slice& msg2) { + assert(code != kOk); + const uint32_t len1 = static_cast<uint32_t>(msg.size()); + const uint32_t len2 = static_cast<uint32_t>(msg2.size()); + const uint32_t size = len1 + (len2 ? (2 + len2) : 0); + char* result = new char[size + 5]; + std::memcpy(result, &size, sizeof(size)); + result[4] = static_cast<char>(code); + std::memcpy(result + 5, msg.data(), len1); + if (len2) { + result[5 + len1] = ':'; + result[6 + len1] = ' '; + std::memcpy(result + 7 + len1, msg2.data(), len2); + } + state_ = result; +} + +std::string Status::ToString() const { + if (state_ == nullptr) { + return "OK"; + } else { + char tmp[30]; + const char* type; + switch (code()) { + case kOk: + type = "OK"; + break; + case kNotFound: + type = "NotFound: "; + break; + case kCorruption: + type = "Corruption: "; + break; + case kNotSupported: + type = "Not implemented: "; + break; + case kInvalidArgument: + type = "Invalid argument: "; + break; + case kIOError: + type = "IO error: "; + break; + default: + std::snprintf(tmp, sizeof(tmp), + "Unknown code(%d): ", static_cast<int>(code())); + type = tmp; + break; + } + std::string result(type); + uint32_t length; + std::memcpy(&length, state_, sizeof(length)); + result.append(state_ + 5, length); + return result; + } +} + +} // namespace leveldb diff --git a/util/status_test.cc b/util/status_test.cc new file mode 100644 index 0000000..fa25e90 --- /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 "leveldb/status.h" + +#include <utility> + +#include "gtest/gtest.h" +#include "leveldb/slice.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 + diff --git a/util/testutil.cc b/util/testutil.cc new file mode 100644 index 0000000..5f77b08 --- /dev/null +++ b/util/testutil.cc @@ -0,0 +1,51 @@ +// 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 "util/testutil.h" + +#include <string> + +#include "util/random.h" + +namespace leveldb { +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<char>(' ' + rnd->Uniform(95)); // ' ' .. '~' + } + return Slice(*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'}; + std::string result; + for (int i = 0; i < len; i++) { + result += kTestChars[rnd->Uniform(sizeof(kTestChars))]; + } + return result; +} + +Slice CompressibleString(Random* rnd, double compressed_fraction, size_t len, + std::string* dst) { + int raw = static_cast<int>(len * compressed_fraction); + if (raw < 1) raw = 1; + std::string raw_data; + RandomString(rnd, raw, &raw_data); + + // Duplicate the random data until we have filled "len" bytes + dst->clear(); + while (dst->size() < len) { + dst->append(raw_data); + } + dst->resize(len); + return Slice(*dst); +} + +} // namespace test +} // namespace leveldb diff --git a/util/testutil.h b/util/testutil.h new file mode 100644 index 0000000..e0e2d64 --- /dev/null +++ b/util/testutil.h @@ -0,0 +1,82 @@ +// Copyright (c) 2011 The LevelDB Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. See the AUTHORS file for names of contributors. + +#ifndef STORAGE_LEVELDB_UTIL_TESTUTIL_H_ +#define STORAGE_LEVELDB_UTIL_TESTUTIL_H_ + +#include "gmock/gmock.h" +#include "gtest/gtest.h" +#include "helpers/memenv/memenv.h" +#include "leveldb/env.h" +#include "leveldb/slice.h" +#include "util/random.h" + +namespace leveldb { +namespace test { + +MATCHER(IsOK, "") { return arg.ok(); } + +// Macros for testing the results of functions that return leveldb::Status or +// absl::StatusOr<T> (for any type T). +#define EXPECT_LEVELDB_OK(expression) \ + EXPECT_THAT(expression, leveldb::test::IsOK()) +#define ASSERT_LEVELDB_OK(expression) \ + ASSERT_THAT(expression, leveldb::test::IsOK()) + +// Returns the random seed used at the start of the current test run. +inline int RandomSeed() { + return testing::UnitTest::GetInstance()->random_seed(); +} + +// Store in *dst a random string of length "len" and return a Slice that +// references the generated data. +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.). +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. +Slice CompressibleString(Random* rnd, double compressed_fraction, size_t len, + std::string* dst); + +// A wrapper that allows injection of errors. +class ErrorEnv : public EnvWrapper { + public: + bool writable_file_error_; + int num_writable_file_errors_; + + ErrorEnv() + : EnvWrapper(NewMemEnv(Env::Default())), + writable_file_error_(false), + num_writable_file_errors_(0) {} + ~ErrorEnv() override { delete target(); } + + Status NewWritableFile(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()->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 +} // namespace leveldb + +#endif // STORAGE_LEVELDB_UTIL_TESTUTIL_H_ diff --git a/util/windows_logger.h b/util/windows_logger.h new file mode 100644 index 0000000..26e6c7b --- /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 <cassert> +#include <cstdarg> +#include <cstdio> +#include <ctime> +#include <sstream> +#include <thread> + +#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, std::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<size_t>(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 = std::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<int>(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 std::(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_ |