Test cmd markdown

cout/endl were missing the std namespace. Also std::make_unique
was used inadvertently which is part of C++14 and only C++11
is currently supported.

PiperOrigin-RevId: 243221310

.appveyor.yml

@@ -0,0 +1,35 @@
+# Build matrix / environment variables are explained on:
+# https://www.appveyor.com/docs/appveyor-yml/
+# This file can be validated on: https://ci.appveyor.com/tools/validate-yaml
+
+version: "{build}"
+
+environment:
+  matrix:
+    # AppVeyor currently has no custom job name feature.
+    # http://help.appveyor.com/discussions/questions/1623-can-i-provide-a-friendly-name-for-jobs
+    - JOB: Visual Studio 2017
+      APPVEYOR_BUILD_WORKER_IMAGE: Visual Studio 2017
+      CMAKE_GENERATOR: Visual Studio 15 2017
+
+platform:
+  - x86
+  - x64
+
+configuration:
+  - RelWithDebInfo
+  - Debug
+
+build_script:
+  - git submodule update --init --recursive
+  - mkdir build
+  - cd build
+  - if "%platform%"=="x64" set CMAKE_GENERATOR=%CMAKE_GENERATOR% Win64
+  - cmake --version
+  - cmake .. -G "%CMAKE_GENERATOR%"
+      -DCMAKE_CONFIGURATION_TYPES="%CONFIGURATION%"
+  - cmake --build . --config "%CONFIGURATION%"
+  - cd ..
+
+test_script:
+  - cd build && ctest --verbose --build-config "%CONFIGURATION%" && cd ..

.gitignore

@@ -1,9 +1,8 @@
-build_config.mk
-*.a
-*.o
-*.dylib*
-*.so
-*.so.*
-*_test
-db_bench
-leveldbutil
+# Editors.
+*.sw*
+.vscode
+.DS_Store
+
+# Build directory.
+build/
+out/

.travis.yml

@@ -0,0 +1,76 @@
+# Build matrix / environment variable are explained on:
+# http://about.travis-ci.org/docs/user/build-configuration/
+# This file can be validated on: http://lint.travis-ci.org/
+
+dist: xenial
+language: cpp
+
+compiler:
+  - gcc
+  - clang
+os:
+  - linux
+  - osx
+
+env:
+  - BUILD_TYPE=Debug
+  - BUILD_TYPE=RelWithDebInfo
+
+matrix:
+  exclude:
+    # GCC fails on recent Travis OSX images.
+    # https://github.com/travis-ci/travis-ci/issues/9640
+    - compiler: gcc
+      os: osx
+
+addons:
+  apt:
+    sources:
+    - llvm-toolchain-xenial-7
+    - ubuntu-toolchain-r-test
+    packages:
+    - clang-7
+    - cmake
+    - gcc-8
+    - g++-8
+    - libgoogle-perftools-dev
+    - libkyotocabinet-dev
+    - libsnappy-dev
+    - libsqlite3-dev
+    - ninja-build
+  homebrew:
+    packages:
+    - crc32c
+    - gperftools
+    - kyotocabinet
+    - gcc@7
+    - ninja
+    - snappy
+    - sqlite3
+
+before_install:
+# The Travis VM image for Mac already has a link at /usr/local/include/c++,
+# causing Homebrew's gcc installation to error out. This was reported to
+# Homebrew maintainers at https://github.com/Homebrew/brew/issues/1742 and
+# removing the link emerged as a workaround.
+- if [ "$TRAVIS_OS_NAME" == "osx" ]; then rm -f /usr/local/include/c++ ; fi
+
+install:
+# /usr/bin/gcc is stuck to old versions on both Linux and OSX.
+- if [ "$CXX" = "g++" ]; then export CXX="g++-8" CC="gcc-8"; fi
+- echo ${CC}
+- echo ${CXX}
+- ${CXX} --version
+- cmake --version
+
+before_script:
+- mkdir -p build && cd build
+- cmake .. -G Ninja -DCMAKE_BUILD_TYPE=$BUILD_TYPE
+- cmake --build .
+- cd ..
+
+script:
+- cd build && ctest --verbose && cd ..
+- "if [ -f build/db_bench ] ; then build/db_bench ; fi"
+- "if [ -f build/db_bench_sqlite3 ] ; then build/db_bench_sqlite3 ; fi"
+- "if [ -f build/db_bench_tree_db ] ; then build/db_bench_tree_db ; fi"

CMakeLists.txt

@@ -0,0 +1,451 @@
+# 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 1.21.0 LANGUAGES C CXX)
+
+# This project can use C11, but will gracefully decay down to C89.
+set(CMAKE_C_STANDARD 11)
+set(CMAKE_C_STANDARD_REQUIRED OFF)
+set(CMAKE_C_EXTENSIONS OFF)
+
+# This project requires C++11.
+set(CMAKE_CXX_STANDARD 11)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
+
+if (WIN32)
+  set(LEVELDB_PLATFORM_NAME LEVELDB_PLATFORM_WINDOWS)
+  # TODO(cmumford): Make UNICODE configurable for Windows.
+  add_definitions(-D_UNICODE -DUNICODE)
+else (WIN32)
+  set(LEVELDB_PLATFORM_NAME LEVELDB_PLATFORM_POSIX)
+endif (WIN32)
+
+option(LEVELDB_BUILD_TESTS "Build LevelDB's unit tests" ON)
+option(LEVELDB_BUILD_BENCHMARKS "Build LevelDB's benchmarks" ON)
+option(LEVELDB_INSTALL "Install LevelDB's header and library" ON)
+
+include(TestBigEndian)
+test_big_endian(LEVELDB_IS_BIG_ENDIAN)
+
+include(CheckIncludeFile)
+check_include_file("unistd.h" HAVE_UNISTD_H)
+
+include(CheckLibraryExists)
+check_library_exists(crc32c crc32c_value "" HAVE_CRC32C)
+check_library_exists(snappy snappy_compress "" HAVE_SNAPPY)
+check_library_exists(tcmalloc malloc "" HAVE_TCMALLOC)
+
+include(CheckCXXSymbolExists)
+# Using check_cxx_symbol_exists() instead of check_c_symbol_exists() because
+# we're including the header from C++, and feature detection should use the same
+# compiler language that the project will use later. Principles aside, some
+# versions of do not expose fdatasync() in <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)
+
+include(CheckCXXSourceCompiles)
+
+# Test whether -Wthread-safety is available. See
+# https://clang.llvm.org/docs/ThreadSafetyAnalysis.html
+# -Werror is necessary because unknown attributes only generate warnings.
+set(OLD_CMAKE_REQUIRED_FLAGS ${CMAKE_REQUIRED_FLAGS})
+list(APPEND CMAKE_REQUIRED_FLAGS -Werror -Wthread-safety)
+check_cxx_source_compiles("
+struct __attribute__((lockable)) Lock {
+  void Acquire() __attribute__((exclusive_lock_function()));
+  void Release() __attribute__((unlock_function()));
+};
+struct ThreadSafeType {
+  Lock lock_;
+  int data_ __attribute__((guarded_by(lock_)));
+};
+int main() { return 0; }
+"  HAVE_CLANG_THREAD_SAFETY)
+set(CMAKE_REQUIRED_FLAGS ${OLD_CMAKE_REQUIRED_FLAGS})
+
+# Test whether C++17 __has_include is available.
+check_cxx_source_compiles("
+#if defined(__has_include) &&  __has_include(<string>)
+#include <string>
+#endif
+int main() { std::string str; return 0; }
+" HAVE_CXX17_HAS_INCLUDE)
+
+set(LEVELDB_PUBLIC_INCLUDE_DIR "include/leveldb")
+set(LEVELDB_PORT_CONFIG_DIR "include/port")
+
+configure_file(
+  "${PROJECT_SOURCE_DIR}/port/port_config.h.in"
+  "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h"
+)
+
+include_directories(
+  "${PROJECT_BINARY_DIR}/include"
+  "${PROJECT_SOURCE_DIR}"
+)
+
+if(BUILD_SHARED_LIBS)
+  # Only export LEVELDB_EXPORT symbols from the shared library.
+  add_compile_options(-fvisibility=hidden)
+endif(BUILD_SHARED_LIBS)
+
+add_library(leveldb "")
+target_sources(leveldb
+  PRIVATE
+    "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h"
+    "${PROJECT_SOURCE_DIR}/db/builder.cc"
+    "${PROJECT_SOURCE_DIR}/db/builder.h"
+    "${PROJECT_SOURCE_DIR}/db/c.cc"
+    "${PROJECT_SOURCE_DIR}/db/db_impl.cc"
+    "${PROJECT_SOURCE_DIR}/db/db_impl.h"
+    "${PROJECT_SOURCE_DIR}/db/db_iter.cc"
+    "${PROJECT_SOURCE_DIR}/db/db_iter.h"
+    "${PROJECT_SOURCE_DIR}/db/dbformat.cc"
+    "${PROJECT_SOURCE_DIR}/db/dbformat.h"
+    "${PROJECT_SOURCE_DIR}/db/dumpfile.cc"
+    "${PROJECT_SOURCE_DIR}/db/filename.cc"
+    "${PROJECT_SOURCE_DIR}/db/filename.h"
+    "${PROJECT_SOURCE_DIR}/db/log_format.h"
+    "${PROJECT_SOURCE_DIR}/db/log_reader.cc"
+    "${PROJECT_SOURCE_DIR}/db/log_reader.h"
+    "${PROJECT_SOURCE_DIR}/db/log_writer.cc"
+    "${PROJECT_SOURCE_DIR}/db/log_writer.h"
+    "${PROJECT_SOURCE_DIR}/db/memtable.cc"
+    "${PROJECT_SOURCE_DIR}/db/memtable.h"
+    "${PROJECT_SOURCE_DIR}/db/repair.cc"
+    "${PROJECT_SOURCE_DIR}/db/skiplist.h"
+    "${PROJECT_SOURCE_DIR}/db/snapshot.h"
+    "${PROJECT_SOURCE_DIR}/db/table_cache.cc"
+    "${PROJECT_SOURCE_DIR}/db/table_cache.h"
+    "${PROJECT_SOURCE_DIR}/db/version_edit.cc"
+    "${PROJECT_SOURCE_DIR}/db/version_edit.h"
+    "${PROJECT_SOURCE_DIR}/db/version_set.cc"
+    "${PROJECT_SOURCE_DIR}/db/version_set.h"
+    "${PROJECT_SOURCE_DIR}/db/write_batch_internal.h"
+    "${PROJECT_SOURCE_DIR}/db/write_batch.cc"
+    "${PROJECT_SOURCE_DIR}/port/port_stdcxx.h"
+    "${PROJECT_SOURCE_DIR}/port/port.h"
+    "${PROJECT_SOURCE_DIR}/port/thread_annotations.h"
+    "${PROJECT_SOURCE_DIR}/table/block_builder.cc"
+    "${PROJECT_SOURCE_DIR}/table/block_builder.h"
+    "${PROJECT_SOURCE_DIR}/table/block.cc"
+    "${PROJECT_SOURCE_DIR}/table/block.h"
+    "${PROJECT_SOURCE_DIR}/table/filter_block.cc"
+    "${PROJECT_SOURCE_DIR}/table/filter_block.h"
+    "${PROJECT_SOURCE_DIR}/table/format.cc"
+    "${PROJECT_SOURCE_DIR}/table/format.h"
+    "${PROJECT_SOURCE_DIR}/table/iterator_wrapper.h"
+    "${PROJECT_SOURCE_DIR}/table/iterator.cc"
+    "${PROJECT_SOURCE_DIR}/table/merger.cc"
+    "${PROJECT_SOURCE_DIR}/table/merger.h"
+    "${PROJECT_SOURCE_DIR}/table/table_builder.cc"
+    "${PROJECT_SOURCE_DIR}/table/table.cc"
+    "${PROJECT_SOURCE_DIR}/table/two_level_iterator.cc"
+    "${PROJECT_SOURCE_DIR}/table/two_level_iterator.h"
+    "${PROJECT_SOURCE_DIR}/util/arena.cc"
+    "${PROJECT_SOURCE_DIR}/util/arena.h"
+    "${PROJECT_SOURCE_DIR}/util/bloom.cc"
+    "${PROJECT_SOURCE_DIR}/util/cache.cc"
+    "${PROJECT_SOURCE_DIR}/util/coding.cc"
+    "${PROJECT_SOURCE_DIR}/util/coding.h"
+    "${PROJECT_SOURCE_DIR}/util/comparator.cc"
+    "${PROJECT_SOURCE_DIR}/util/crc32c.cc"
+    "${PROJECT_SOURCE_DIR}/util/crc32c.h"
+    "${PROJECT_SOURCE_DIR}/util/env.cc"
+    "${PROJECT_SOURCE_DIR}/util/filter_policy.cc"
+    "${PROJECT_SOURCE_DIR}/util/hash.cc"
+    "${PROJECT_SOURCE_DIR}/util/hash.h"
+    "${PROJECT_SOURCE_DIR}/util/logging.cc"
+    "${PROJECT_SOURCE_DIR}/util/logging.h"
+    "${PROJECT_SOURCE_DIR}/util/mutexlock.h"
+    "${PROJECT_SOURCE_DIR}/util/no_destructor.h"
+    "${PROJECT_SOURCE_DIR}/util/options.cc"
+    "${PROJECT_SOURCE_DIR}/util/random.h"
+    "${PROJECT_SOURCE_DIR}/util/status.cc"
+
+  # Only CMake 3.3+ supports PUBLIC sources in targets exported by "install".
+  $<$<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"
+    "${LEVELDB_PUBLIC_INCLUDE_DIR}/write_batch.h"
+)
+
+if (WIN32)
+  target_sources(leveldb
+    PRIVATE
+      "${PROJECT_SOURCE_DIR}/util/env_windows.cc"
+      "${PROJECT_SOURCE_DIR}/util/windows_logger.h"
+  )
+else (WIN32)
+  target_sources(leveldb
+    PRIVATE
+      "${PROJECT_SOURCE_DIR}/util/env_posix.cc"
+      "${PROJECT_SOURCE_DIR}/util/posix_logger.h"
+  )
+endif (WIN32)
+
+# MemEnv is not part of the interface and could be pulled to a separate library.
+target_sources(leveldb
+  PRIVATE
+    "${PROJECT_SOURCE_DIR}/helpers/memenv/memenv.cc"
+    "${PROJECT_SOURCE_DIR}/helpers/memenv/memenv.h"
+)
+
+target_include_directories(leveldb
+  PUBLIC
+    $<BUILD_INTERFACE:${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.
+    ${LEVELDB_PLATFORM_NAME}=1
+)
+if (NOT HAVE_CXX17_HAS_INCLUDE)
+  target_compile_definitions(leveldb
+    PRIVATE
+      LEVELDB_HAS_PORT_CONFIG_H=1
+  )
+endif(NOT HAVE_CXX17_HAS_INCLUDE)
+
+if(BUILD_SHARED_LIBS)
+  target_compile_definitions(leveldb
+    PUBLIC
+      # Used by include/export.h.
+      LEVELDB_SHARED_LIBRARY
+  )
+endif(BUILD_SHARED_LIBS)
+
+if(HAVE_CLANG_THREAD_SAFETY)
+  target_compile_options(leveldb
+    PUBLIC
+      -Werror -Wthread-safety)
+endif(HAVE_CLANG_THREAD_SAFETY)
+
+if(HAVE_CRC32C)
+  target_link_libraries(leveldb crc32c)
+endif(HAVE_CRC32C)
+if(HAVE_SNAPPY)
+  target_link_libraries(leveldb snappy)
+endif(HAVE_SNAPPY)
+if(HAVE_TCMALLOC)
+  target_link_libraries(leveldb tcmalloc)
+endif(HAVE_TCMALLOC)
+
+# Needed by port_stdcxx.h
+find_package(Threads REQUIRED)
+target_link_libraries(leveldb Threads::Threads)
+
+add_executable(leveldbutil
+  "${PROJECT_SOURCE_DIR}/db/leveldbutil.cc"
+)
+target_link_libraries(leveldbutil leveldb)
+
+if(LEVELDB_BUILD_TESTS)
+  enable_testing()
+
+  function(leveldb_test test_file)
+    get_filename_component(test_target_name "${test_file}" NAME_WE)
+
+    add_executable("${test_target_name}" "")
+    target_sources("${test_target_name}"
+      PRIVATE
+        "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h"
+        "${PROJECT_SOURCE_DIR}/util/testharness.cc"
+        "${PROJECT_SOURCE_DIR}/util/testharness.h"
+        "${PROJECT_SOURCE_DIR}/util/testutil.cc"
+        "${PROJECT_SOURCE_DIR}/util/testutil.h"
+
+        "${test_file}"
+    )
+    target_link_libraries("${test_target_name}" leveldb)
+    target_compile_definitions("${test_target_name}"
+      PRIVATE
+        ${LEVELDB_PLATFORM_NAME}=1
+    )
+    if (NOT HAVE_CXX17_HAS_INCLUDE)
+      target_compile_definitions("${test_target_name}"
+        PRIVATE
+          LEVELDB_HAS_PORT_CONFIG_H=1
+      )
+    endif(NOT HAVE_CXX17_HAS_INCLUDE)
+
+    add_test(NAME "${test_target_name}" COMMAND "${test_target_name}")
+  endfunction(leveldb_test)
+
+  leveldb_test("${PROJECT_SOURCE_DIR}/db/c_test.c")
+  leveldb_test("${PROJECT_SOURCE_DIR}/db/fault_injection_test.cc")
+
+  leveldb_test("${PROJECT_SOURCE_DIR}/issues/issue178_test.cc")
+  leveldb_test("${PROJECT_SOURCE_DIR}/issues/issue200_test.cc")
+  leveldb_test("${PROJECT_SOURCE_DIR}/issues/issue320_test.cc")
+
+  leveldb_test("${PROJECT_SOURCE_DIR}/util/env_test.cc")
+  leveldb_test("${PROJECT_SOURCE_DIR}/util/status_test.cc")
+  leveldb_test("${PROJECT_SOURCE_DIR}/util/no_destructor_test.cc")
+
+  if(NOT BUILD_SHARED_LIBS)
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/autocompact_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/corruption_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/db_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/dbformat_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/filename_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/log_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/recovery_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/skiplist_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/version_edit_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/version_set_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/db/write_batch_test.cc")
+
+    leveldb_test("${PROJECT_SOURCE_DIR}/helpers/memenv/memenv_test.cc")
+
+    leveldb_test("${PROJECT_SOURCE_DIR}/table/filter_block_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/table/table_test.cc")
+
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/arena_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/bloom_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/cache_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/coding_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/crc32c_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/hash_test.cc")
+    leveldb_test("${PROJECT_SOURCE_DIR}/util/logging_test.cc")
+
+    # TODO(costan): This test also uses
+    #               "${PROJECT_SOURCE_DIR}/util/env_{posix|windows}_test_helper.h"
+    if (WIN32)
+      leveldb_test("${PROJECT_SOURCE_DIR}/util/env_windows_test.cc")
+    else (WIN32)
+      leveldb_test("${PROJECT_SOURCE_DIR}/util/env_posix_test.cc")
+    endif (WIN32)
+  endif(NOT BUILD_SHARED_LIBS)
+endif(LEVELDB_BUILD_TESTS)
+
+if(LEVELDB_BUILD_BENCHMARKS)
+  function(leveldb_benchmark bench_file)
+    get_filename_component(bench_target_name "${bench_file}" NAME_WE)
+
+    add_executable("${bench_target_name}" "")
+    target_sources("${bench_target_name}"
+      PRIVATE
+        "${PROJECT_BINARY_DIR}/${LEVELDB_PORT_CONFIG_DIR}/port_config.h"
+        "${PROJECT_SOURCE_DIR}/util/histogram.cc"
+        "${PROJECT_SOURCE_DIR}/util/histogram.h"
+        "${PROJECT_SOURCE_DIR}/util/testharness.cc"
+        "${PROJECT_SOURCE_DIR}/util/testharness.h"
+        "${PROJECT_SOURCE_DIR}/util/testutil.cc"
+        "${PROJECT_SOURCE_DIR}/util/testutil.h"
+
+        "${bench_file}"
+    )
+    target_link_libraries("${bench_target_name}" leveldb)
+    target_compile_definitions("${bench_target_name}"
+      PRIVATE
+        ${LEVELDB_PLATFORM_NAME}=1
+    )
+    if (NOT HAVE_CXX17_HAS_INCLUDE)
+      target_compile_definitions("${bench_target_name}"
+        PRIVATE
+          LEVELDB_HAS_PORT_CONFIG_H=1
+      )
+    endif(NOT HAVE_CXX17_HAS_INCLUDE)
+  endfunction(leveldb_benchmark)
+
+  if(NOT BUILD_SHARED_LIBS)
+    leveldb_benchmark("${PROJECT_SOURCE_DIR}/db/db_bench.cc")
+  endif(NOT BUILD_SHARED_LIBS)
+
+  check_library_exists(sqlite3 sqlite3_open "" HAVE_SQLITE3)
+  if(HAVE_SQLITE3)
+    leveldb_benchmark("${PROJECT_SOURCE_DIR}/doc/bench/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 ${OLD_CMAKE_REQURED_LIBRARIES})
+  if(HAVE_KYOTOCABINET)
+    leveldb_benchmark("${PROJECT_SOURCE_DIR}/doc/bench/db_bench_tree_db.cc")
+    target_link_libraries(db_bench_tree_db kyotocabinet)
+  endif(HAVE_KYOTOCABINET)
+endif(LEVELDB_BUILD_BENCHMARKS)
+
+if(LEVELDB_INSTALL)
+  include(GNUInstallDirs)
+  install(TARGETS leveldb
+    EXPORT leveldbTargets
+    RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
+    LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+    ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+  )
+  install(
+    FILES
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/c.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/cache.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/comparator.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/db.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/dumpfile.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/env.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/export.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/filter_policy.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/iterator.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/options.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/slice.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/status.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/table_builder.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/table.h"
+      "${PROJECT_SOURCE_DIR}/${LEVELDB_PUBLIC_INCLUDE_DIR}/write_batch.h"
+    DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/leveldb
+  )
+
+  include(CMakePackageConfigHelpers)
+  write_basic_package_version_file(
+      "${PROJECT_BINARY_DIR}/leveldbConfigVersion.cmake"
+      COMPATIBILITY SameMajorVersion
+  )
+  install(
+    EXPORT leveldbTargets
+    NAMESPACE leveldb::
+    DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/leveldb"
+  )
+  install(
+    FILES
+      "${PROJECT_SOURCE_DIR}/cmake/leveldbConfig.cmake"
+      "${PROJECT_BINARY_DIR}/leveldbConfigVersion.cmake"
+    DESTINATION "${CMAKE_INSTALL_LIBDIR}/cmake/leveldb"
+  )
+endif(LEVELDB_INSTALL)

CONTRIBUTING.md

@@ -31,6 +31,6 @@ the CLA.
 
 ## Writing Code ##
 
-If your contribution contains code, please make sure that it follows 
-[the style guide](http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml).
+If your contribution contains code, please make sure that it follows
+[the style guide](http://google.github.io/styleguide/cppguide.html).
 Otherwise we will have to ask you to make changes, and that's no fun for anyone.

Makefile

@@ -1,227 +0,0 @@
-# Copyright (c) 2011 The LevelDB Authors. All rights reserved.
-# Use of this source code is governed by a BSD-style license that can be
-# found in the LICENSE file. See the AUTHORS file for names of contributors.
-
-#-----------------------------------------------
-# Uncomment exactly one of the lines labelled (A), (B), and (C) below
-# to switch between compilation modes.
-
-# (A) Production use (optimized mode)
-OPT ?= -O2 -DNDEBUG
-# (B) Debug mode, w/ full line-level debugging symbols
-# OPT ?= -g2
-# (C) Profiling mode: opt, but w/debugging symbols
-# OPT ?= -O2 -g2 -DNDEBUG
-#-----------------------------------------------
-
-# detect what platform we're building on
-$(shell CC="$(CC)" CXX="$(CXX)" TARGET_OS="$(TARGET_OS)" \
-    ./build_detect_platform build_config.mk ./)
-# this file is generated by the previous line to set build flags and sources
-include build_config.mk
-
-CFLAGS += -I. -I./include $(PLATFORM_CCFLAGS) $(OPT)
-CXXFLAGS += -I. -I./include $(PLATFORM_CXXFLAGS) $(OPT)
-
-LDFLAGS += $(PLATFORM_LDFLAGS)
-LIBS += $(PLATFORM_LIBS)
-
-LIBOBJECTS = $(SOURCES:.cc=.o)
-MEMENVOBJECTS = $(MEMENV_SOURCES:.cc=.o)
-
-TESTUTIL = ./util/testutil.o
-TESTHARNESS = ./util/testharness.o $(TESTUTIL)
-
-# Note: iOS should probably be using libtool, not ar.
-ifeq ($(PLATFORM), IOS)
-AR=xcrun ar
-endif
-
-TESTS = \
-	arena_test \
-	autocompact_test \
-	bloom_test \
-	c_test \
-	cache_test \
-	coding_test \
-	corruption_test \
-	crc32c_test \
-	db_test \
-	dbformat_test \
-	env_test \
-	filename_test \
-	filter_block_test \
-	hash_test \
-	issue178_test \
-	issue200_test \
-	log_test \
-	memenv_test \
-	skiplist_test \
-	table_test \
-	version_edit_test \
-	version_set_test \
-	write_batch_test
-
-PROGRAMS = db_bench leveldbutil $(TESTS)
-BENCHMARKS = db_bench_sqlite3 db_bench_tree_db
-
-LIBRARY = libleveldb.a
-MEMENVLIBRARY = libmemenv.a
-
-default: all
-
-# Should we build shared libraries?
-ifneq ($(PLATFORM_SHARED_EXT),)
-
-ifneq ($(PLATFORM_SHARED_VERSIONED),true)
-SHARED1 = libleveldb.$(PLATFORM_SHARED_EXT)
-SHARED2 = $(SHARED1)
-SHARED3 = $(SHARED1)
-SHARED = $(SHARED1)
-else
-# Update db.h if you change these.
-SHARED_MAJOR = 1
-SHARED_MINOR = 18
-SHARED1 = libleveldb.$(PLATFORM_SHARED_EXT)
-SHARED2 = $(SHARED1).$(SHARED_MAJOR)
-SHARED3 = $(SHARED1).$(SHARED_MAJOR).$(SHARED_MINOR)
-SHARED = $(SHARED1) $(SHARED2) $(SHARED3)
-$(SHARED1): $(SHARED3)
-	ln -fs $(SHARED3) $(SHARED1)
-$(SHARED2): $(SHARED3)
-	ln -fs $(SHARED3) $(SHARED2)
-endif
-
-$(SHARED3):
-	$(CXX) $(LDFLAGS) $(PLATFORM_SHARED_LDFLAGS)$(SHARED2) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(SOURCES) -o $(SHARED3) $(LIBS)
-
-endif  # PLATFORM_SHARED_EXT
-
-all: $(SHARED) $(LIBRARY)
-
-check: all $(PROGRAMS) $(TESTS)
-	for t in $(TESTS); do echo "***** Running $$t"; ./$$t || exit 1; done
-
-clean:
-	-rm -f $(PROGRAMS) $(BENCHMARKS) $(LIBRARY) $(SHARED) $(MEMENVLIBRARY) */*.o */*/*.o ios-x86/*/*.o ios-arm/*/*.o build_config.mk
-	-rm -rf ios-x86/* ios-arm/*
-
-$(LIBRARY): $(LIBOBJECTS)
-	rm -f $@
-	$(AR) -rs $@ $(LIBOBJECTS)
-
-db_bench: db/db_bench.o $(LIBOBJECTS) $(TESTUTIL)
-	$(CXX) $(LDFLAGS) db/db_bench.o $(LIBOBJECTS) $(TESTUTIL) -o $@ $(LIBS)
-
-db_bench_sqlite3: doc/bench/db_bench_sqlite3.o $(LIBOBJECTS) $(TESTUTIL)
-	$(CXX) $(LDFLAGS) doc/bench/db_bench_sqlite3.o $(LIBOBJECTS) $(TESTUTIL) -o $@ -lsqlite3 $(LIBS)
-
-db_bench_tree_db: doc/bench/db_bench_tree_db.o $(LIBOBJECTS) $(TESTUTIL)
-	$(CXX) $(LDFLAGS) doc/bench/db_bench_tree_db.o $(LIBOBJECTS) $(TESTUTIL) -o $@ -lkyotocabinet $(LIBS)
-
-leveldbutil: db/leveldb_main.o $(LIBOBJECTS)
-	$(CXX) $(LDFLAGS) db/leveldb_main.o $(LIBOBJECTS) -o $@ $(LIBS)
-
-arena_test: util/arena_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/arena_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-autocompact_test: db/autocompact_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/autocompact_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-bloom_test: util/bloom_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/bloom_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-c_test: db/c_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/c_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-cache_test: util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-corruption_test: db/corruption_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/corruption_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-crc32c_test: util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-db_test: db/db_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/db_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-dbformat_test: db/dbformat_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/dbformat_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-env_test: util/env_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/env_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-filename_test: db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/filename_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-filter_block_test: table/filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) table/filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-hash_test: util/hash_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) util/hash_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-issue178_test: issues/issue178_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) issues/issue178_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-issue200_test: issues/issue200_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) issues/issue200_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-log_test: db/log_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/log_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-table_test: table/table_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) table/table_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-skiplist_test: db/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-version_edit_test: db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-version_set_test: db/version_set_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/version_set_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-write_batch_test: db/write_batch_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) db/write_batch_test.o $(LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
-
-$(MEMENVLIBRARY) : $(MEMENVOBJECTS)
-	rm -f $@
-	$(AR) -rs $@ $(MEMENVOBJECTS)
-
-memenv_test : helpers/memenv/memenv_test.o $(MEMENVLIBRARY) $(LIBRARY) $(TESTHARNESS)
-	$(CXX) $(LDFLAGS) helpers/memenv/memenv_test.o $(MEMENVLIBRARY) $(LIBRARY) $(TESTHARNESS) -o $@ $(LIBS)
-
-ifeq ($(PLATFORM), IOS)
-# For iOS, create universal object files to be used on both the simulator and
-# a device.
-PLATFORMSROOT=/Applications/Xcode.app/Contents/Developer/Platforms
-SIMULATORROOT=$(PLATFORMSROOT)/iPhoneSimulator.platform/Developer
-DEVICEROOT=$(PLATFORMSROOT)/iPhoneOS.platform/Developer
-IOSVERSION=$(shell defaults read $(PLATFORMSROOT)/iPhoneOS.platform/version CFBundleShortVersionString)
-IOSARCH=-arch armv6 -arch armv7 -arch armv7s -arch arm64
-
-.cc.o:
-	mkdir -p ios-x86/$(dir $@)
-	xcrun -sdk iphonesimulator $(CXX) $(CXXFLAGS) -isysroot $(SIMULATORROOT)/SDKs/iPhoneSimulator$(IOSVERSION).sdk -arch i686 -arch x86_64 -c $< -o ios-x86/$@
-	mkdir -p ios-arm/$(dir $@)
-	xcrun -sdk iphoneos $(CXX) $(CXXFLAGS) -isysroot $(DEVICEROOT)/SDKs/iPhoneOS$(IOSVERSION).sdk $(IOSARCH) -c $< -o ios-arm/$@
-	xcrun lipo ios-x86/$@ ios-arm/$@ -create -output $@
-
-.c.o:
-	mkdir -p ios-x86/$(dir $@)
-	xcrun -sdk iphonesimulator $(CC) $(CFLAGS) -isysroot $(SIMULATORROOT)/SDKs/iPhoneSimulator$(IOSVERSION).sdk -arch i686 -arch x86_64 -c $< -o ios-x86/$@
-	mkdir -p ios-arm/$(dir $@)
-	xcrun -sdk iphoneos $(CC) $(CFLAGS) -isysroot $(DEVICEROOT)/SDKs/iPhoneOS$(IOSVERSION).sdk $(IOSARCH) -c $< -o ios-arm/$@
-	xcrun lipo ios-x86/$@ ios-arm/$@ -create -output $@
-
-else
-.cc.o:
-	$(CXX) $(CXXFLAGS) -c $< -o $@
-
-.c.o:
-	$(CC) $(CFLAGS) -c $< -o $@
-endif

README

@@ -1,51 +0,0 @@
-leveldb: A key-value store
-Authors: Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com)
-
-The code under this directory implements a system for maintaining a
-persistent key/value store.
-
-See doc/index.html for more explanation.
-See doc/impl.html for a brief overview of the implementation.
-
-The public interface is in include/*.h.  Callers should not include or
-rely on the details of any other header files in this package.  Those
-internal APIs may be changed without warning.
-
-Guide to header files:
-
-include/db.h
-    Main interface to the DB: Start here
-
-include/options.h
-    Control over the behavior of an entire database, and also
-    control over the behavior of individual reads and writes.
-
-include/comparator.h
-    Abstraction for user-specified comparison function.  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/iterator.h
-    Interface for iterating over data. You can get an iterator
-    from a DB object.
-
-include/write_batch.h
-    Interface for atomically applying multiple updates to a database.
-
-include/slice.h
-    A simple module for maintaining a pointer and a length into some
-    other byte array.
-
-include/status.h
-    Status is returned from many of the public interfaces and is used
-    to report success and various kinds of errors.
-
-include/env.h
-    Abstraction of the OS environment.  A posix implementation of
-    this interface is in util/env_posix.cc
-
-include/table.h
-include/table_builder.h
-    Lower-level modules that most clients probably won't use directly

README.md

@@ -1,8 +1,12 @@
 **LevelDB is a fast key-value storage library written at Google that provides an ordered mapping from string keys to string values.**
 
+[![Build Status](https://travis-ci.org/google/leveldb.svg?branch=master)](https://travis-ci.org/google/leveldb)
+[![Build status](https://ci.appveyor.com/api/projects/status/g2j5j4rfkda6eyw5/branch/master?svg=true)](https://ci.appveyor.com/project/pwnall/leveldb)
+
 Authors: Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com)
 
 # Features
+
   * Keys and values are arbitrary byte arrays.
   * Data is stored sorted by key.
   * Callers can provide a custom comparison function to override the sort order.
@@ -10,16 +14,90 @@ Authors: Sanjay Ghemawat (sanjay@google.com) and Jeff Dean (jeff@google.com)
   * Multiple changes can be made in one atomic batch.
   * Users can create a transient snapshot to get a consistent view of data.
   * Forward and backward iteration is supported over the data.
-  * Data is automatically compressed using the [Snappy compression library](http://code.google.com/p/snappy).
+  * Data is automatically compressed using the [Snappy compression library](http://google.github.io/snappy/).
   * External activity (file system operations etc.) is relayed through a virtual interface so users can customize the operating system interactions.
-  * [Detailed documentation](http://htmlpreview.github.io/?https://github.com/google/leveldb/blob/master/doc/index.html) about how to use the library is included with the source code.
 
+# Documentation
+
+  [LevelDB library documentation](https://github.com/google/leveldb/blob/master/doc/index.md) is online and bundled with the source code.
 
 # Limitations
+
   * This is not a SQL database.  It does not have a relational data model, it does not support SQL queries, and it has no support for indexes.
   * Only a single process (possibly multi-threaded) can access a particular database at a time.
   * There is no client-server support builtin to the library.  An application that needs such support will have to wrap their own server around the library.
 
+# Building
+
+This project supports [CMake](https://cmake.org/) out of the box.
+
+### Build for POSIX
+
+Quick start:
+
+```bash
+mkdir -p build && cd build
+cmake -DCMAKE_BUILD_TYPE=Release .. && cmake --build .
+```
+
+### Building for Windows
+
+First generate the Visual Studio 2017 project/solution files:
+
+```cmd
+mkdir build
+cd build
+cmake -G "Visual Studio 15" ..
+```
+The default default will build for x86. For 64-bit run:
+
+```cmd
+cmake -G "Visual Studio 15 Win64" ..
+```
+
+To compile the Windows solution from the command-line:
+
+```cmd
+devenv /build Debug leveldb.sln
+```
+
+or open leveldb.sln in Visual Studio and build from within.
+
+Please see the CMake documentation and `CMakeLists.txt` for more advanced usage.
+
+# Contributing to the leveldb Project
+
+The leveldb project welcomes contributions. leveldb's primary goal is to be
+a reliable and fast key/value store. Changes that are in line with the
+features/limitations outlined above, and meet the requirements below,
+will be considered.
+
+Contribution requirements:
+
+1. **Tested platforms only**. We _generally_ will only accept changes for
+   platforms that are compiled and tested. This means POSIX (for Linux and
+   macOS) or Windows. Very small changes will sometimes be accepted, but
+   consider that more of an exception than the rule.
+
+2. **Stable API**. We strive very hard to maintain a stable API. Changes that
+   require changes for projects using leveldb _might_ be rejected without
+   sufficient benefit to the project.
+
+3. **Tests**: All changes must be accompanied by a new (or changed) test, or
+   a sufficient explanation as to why a new (or changed) test is not required.
+
+## Submitting a Pull Request
+
+Before any pull request will be accepted the author must first sign a
+Contributor License Agreement (CLA) at https://cla.developers.google.com/.
+
+In order to keep the commit timeline linear
+[squash](https://git-scm.com/book/en/v2/Git-Tools-Rewriting-History#Squashing-Commits)
+your changes down to a single commit and [rebase](https://git-scm.com/docs/git-rebase)
+on google/leveldb/master. This keeps the commit timeline linear and more easily sync'ed
+with the internal repository at Google. More information at GitHub's
+[About Git rebase](https://help.github.com/articles/about-git-rebase/) page.
+
 # Performance
 
 Here is a performance report (with explanations) from the run of the
@@ -78,29 +156,30 @@ by the one or two disk seeks needed to fetch the data from disk.
 Write performance will be mostly unaffected by whether or not the
 working set fits in memory.
 
-    readrandom   :      16.677 micros/op;  (approximately 60,000 reads per second)
-    readseq      :       0.476 micros/op;  232.3 MB/s
-    readreverse  :       0.724 micros/op;  152.9 MB/s
+    readrandom  : 16.677 micros/op;  (approximately 60,000 reads per second)
+    readseq     :  0.476 micros/op;  232.3 MB/s
+    readreverse :  0.724 micros/op;  152.9 MB/s
 
 LevelDB compacts its underlying storage data in the background to
 improve read performance.  The results listed above were done
 immediately after a lot of random writes.  The results after
 compactions (which are usually triggered automatically) are better.
 
-    readrandom   :      11.602 micros/op;  (approximately 85,000 reads per second)
-    readseq      :       0.423 micros/op;  261.8 MB/s
-    readreverse  :       0.663 micros/op;  166.9 MB/s
+    readrandom  : 11.602 micros/op;  (approximately 85,000 reads per second)
+    readseq     :  0.423 micros/op;  261.8 MB/s
+    readreverse :  0.663 micros/op;  166.9 MB/s
 
 Some of the high cost of reads comes from repeated decompression of blocks
 read from disk.  If we supply enough cache to the leveldb so it can hold the
 uncompressed blocks in memory, the read performance improves again:
 
-    readrandom   :       9.775 micros/op;  (approximately 100,000 reads per second before compaction)
-    readrandom   :       5.215 micros/op;  (approximately 190,000 reads per second after compaction)
+    readrandom  : 9.775 micros/op;  (approximately 100,000 reads per second before compaction)
+    readrandom  : 5.215 micros/op;  (approximately 190,000 reads per second after compaction)
 
 ## Repository contents
 
-See doc/index.html for more explanation. See doc/impl.html for a brief overview of the implementation.
+See [doc/index.md](doc/index.md) for more explanation. See
+[doc/impl.md](doc/impl.md) for a brief overview of the implementation.
 
 The public interface is in include/*.h.  Callers should not include or
 rely on the details of any other header files in this package.  Those
@@ -113,7 +192,7 @@ Guide to header files:
 * **include/options.h**: Control over the behavior of an entire database,
 and also control over the behavior of individual reads and writes.
 
-* **include/comparator.h**: Abstraction for user-specified comparison function. 
+* **include/comparator.h**: Abstraction for user-specified comparison function.
 If you want just bytewise comparison of keys, you can use the default
 comparator, but clients can write their own comparator implementations if they
 want custom ordering (e.g. to handle different character encodings, etc.)
@@ -130,7 +209,7 @@ length into some other byte array.
 * **include/status.h**: Status is returned from many of the public interfaces
 and is used to report success and various kinds of errors.
 
-* **include/env.h**: 
+* **include/env.h**:
 Abstraction of the OS environment.  A posix implementation of this interface is
 in util/env_posix.cc
 

build_detect_platform

@@ -1,228 +0,0 @@
-#!/bin/sh
-#
-# Detects OS we're compiling on and outputs a file specified by the first
-# argument, which in turn gets read while processing Makefile.
-#
-# The output will set the following variables:
-#   CC                          C Compiler path
-#   CXX                         C++ Compiler path
-#   PLATFORM_LDFLAGS            Linker flags
-#   PLATFORM_LIBS               Libraries flags
-#   PLATFORM_SHARED_EXT         Extension for shared libraries
-#   PLATFORM_SHARED_LDFLAGS     Flags for building shared library
-#                               This flag is embedded just before the name
-#                               of the shared library without intervening spaces
-#   PLATFORM_SHARED_CFLAGS      Flags for compiling objects for shared library
-#   PLATFORM_CCFLAGS            C compiler flags
-#   PLATFORM_CXXFLAGS           C++ compiler flags.  Will contain:
-#   PLATFORM_SHARED_VERSIONED   Set to 'true' if platform supports versioned
-#                               shared libraries, empty otherwise.
-#
-# The PLATFORM_CCFLAGS and PLATFORM_CXXFLAGS might include the following:
-#
-#       -DLEVELDB_ATOMIC_PRESENT     if <atomic> is present
-#       -DLEVELDB_PLATFORM_POSIX     for Posix-based platforms
-#       -DSNAPPY                     if the Snappy library is present
-#
-
-OUTPUT=$1
-PREFIX=$2
-if test -z "$OUTPUT" || test -z "$PREFIX"; then
-  echo "usage: $0 <output-filename> <directory_prefix>" >&2
-  exit 1
-fi
-
-# Delete existing output, if it exists
-rm -f $OUTPUT
-touch $OUTPUT
-
-if test -z "$CC"; then
-    CC=cc
-fi
-
-if test -z "$CXX"; then
-    CXX=g++
-fi
-
-if test -z "$TMPDIR"; then
-    TMPDIR=/tmp
-fi
-
-# Detect OS
-if test -z "$TARGET_OS"; then
-    TARGET_OS=`uname -s`
-fi
-
-COMMON_FLAGS=
-CROSS_COMPILE=
-PLATFORM_CCFLAGS=
-PLATFORM_CXXFLAGS=
-PLATFORM_LDFLAGS=
-PLATFORM_LIBS=
-PLATFORM_SHARED_EXT="so"
-PLATFORM_SHARED_LDFLAGS="-shared -Wl,-soname -Wl,"
-PLATFORM_SHARED_CFLAGS="-fPIC"
-PLATFORM_SHARED_VERSIONED=true
-
-MEMCMP_FLAG=
-if [ "$CXX" = "g++" ]; then
-    # Use libc's memcmp instead of GCC's memcmp.  This results in ~40%
-    # performance improvement on readrandom under gcc 4.4.3 on Linux/x86.
-    MEMCMP_FLAG="-fno-builtin-memcmp"
-fi
-
-case "$TARGET_OS" in
-    CYGWIN_*)
-        PLATFORM=OS_LINUX
-        COMMON_FLAGS="$MEMCMP_FLAG -lpthread -DOS_LINUX -DCYGWIN"
-        PLATFORM_LDFLAGS="-lpthread"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    Darwin)
-        PLATFORM=OS_MACOSX
-        COMMON_FLAGS="$MEMCMP_FLAG -DOS_MACOSX"
-        PLATFORM_SHARED_EXT=dylib
-        [ -z "$INSTALL_PATH" ] && INSTALL_PATH=`pwd`
-        PLATFORM_SHARED_LDFLAGS="-dynamiclib -install_name $INSTALL_PATH/"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    Linux)
-        PLATFORM=OS_LINUX
-        COMMON_FLAGS="$MEMCMP_FLAG -pthread -DOS_LINUX"
-        PLATFORM_LDFLAGS="-pthread"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    SunOS)
-        PLATFORM=OS_SOLARIS
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_SOLARIS"
-        PLATFORM_LIBS="-lpthread -lrt"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    FreeBSD)
-        PLATFORM=OS_FREEBSD
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_FREEBSD"
-        PLATFORM_LIBS="-lpthread"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    NetBSD)
-        PLATFORM=OS_NETBSD
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_NETBSD"
-        PLATFORM_LIBS="-lpthread -lgcc_s"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    OpenBSD)
-        PLATFORM=OS_OPENBSD
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_OPENBSD"
-        PLATFORM_LDFLAGS="-pthread"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    DragonFly)
-        PLATFORM=OS_DRAGONFLYBSD
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_DRAGONFLYBSD"
-        PLATFORM_LIBS="-lpthread"
-        PORT_FILE=port/port_posix.cc
-        ;;
-    OS_ANDROID_CROSSCOMPILE)
-        PLATFORM=OS_ANDROID
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_ANDROID -DLEVELDB_PLATFORM_POSIX"
-        PLATFORM_LDFLAGS=""  # All pthread features are in the Android C library
-        PORT_FILE=port/port_posix.cc
-        CROSS_COMPILE=true
-        ;;
-    HP-UX)
-        PLATFORM=OS_HPUX
-        COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_HPUX"
-        PLATFORM_LDFLAGS="-pthread"
-        PORT_FILE=port/port_posix.cc
-        # man ld: +h internal_name
-        PLATFORM_SHARED_LDFLAGS="-shared -Wl,+h -Wl,"
-        ;;
-    IOS)
-        PLATFORM=IOS
-        COMMON_FLAGS="$MEMCMP_FLAG -DOS_MACOSX"
-        [ -z "$INSTALL_PATH" ] && INSTALL_PATH=`pwd`
-        PORT_FILE=port/port_posix.cc
-        PLATFORM_SHARED_EXT=
-        PLATFORM_SHARED_LDFLAGS=
-        PLATFORM_SHARED_CFLAGS=
-        PLATFORM_SHARED_VERSIONED=
-        ;;
-    *)
-        echo "Unknown platform!" >&2
-        exit 1
-esac
-
-# We want to make a list of all cc files within util, db, table, and helpers
-# except for the test and benchmark files. By default, find will output a list
-# of all files matching either rule, so we need to append -print to make the
-# prune take effect.
-DIRS="$PREFIX/db $PREFIX/util $PREFIX/table"
-
-set -f # temporarily disable globbing so that our patterns aren't expanded
-PRUNE_TEST="-name *test*.cc -prune"
-PRUNE_BENCH="-name *_bench.cc -prune"
-PRUNE_TOOL="-name leveldb_main.cc -prune"
-PORTABLE_FILES=`find $DIRS $PRUNE_TEST -o $PRUNE_BENCH -o $PRUNE_TOOL -o -name '*.cc' -print | sort | sed "s,^$PREFIX/,," | tr "\n" " "`
-
-set +f # re-enable globbing
-
-# The sources consist of the portable files, plus the platform-specific port
-# file.
-echo "SOURCES=$PORTABLE_FILES $PORT_FILE" >> $OUTPUT
-echo "MEMENV_SOURCES=helpers/memenv/memenv.cc" >> $OUTPUT
-
-if [ "$CROSS_COMPILE" = "true" ]; then
-    # Cross-compiling; do not try any compilation tests.
-    true
-else
-    CXXOUTPUT="${TMPDIR}/leveldb_build_detect_platform-cxx.$$"
-
-    # If -std=c++0x works, use <atomic> as fallback for when memory barriers
-    # are not available.
-    $CXX $CXXFLAGS -std=c++0x -x c++ - -o $CXXOUTPUT 2>/dev/null  <<EOF
-      #include <atomic>
-      int main() {}
-EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DLEVELDB_PLATFORM_POSIX -DLEVELDB_ATOMIC_PRESENT"
-        PLATFORM_CXXFLAGS="-std=c++0x"
-    else
-        COMMON_FLAGS="$COMMON_FLAGS -DLEVELDB_PLATFORM_POSIX"
-    fi
-
-    # Test whether Snappy library is installed
-    # http://code.google.com/p/snappy/
-    $CXX $CXXFLAGS -x c++ - -o $CXXOUTPUT 2>/dev/null  <<EOF
-      #include <snappy.h>
-      int main() {}
-EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DSNAPPY"
-        PLATFORM_LIBS="$PLATFORM_LIBS -lsnappy"
-    fi
-
-    # Test whether tcmalloc is available
-    $CXX $CXXFLAGS -x c++ - -o $CXXOUTPUT -ltcmalloc 2>/dev/null  <<EOF
-      int main() {}
-EOF
-    if [ "$?" = 0 ]; then
-        PLATFORM_LIBS="$PLATFORM_LIBS -ltcmalloc"
-    fi
-
-    rm -f $CXXOUTPUT 2>/dev/null
-fi
-
-PLATFORM_CCFLAGS="$PLATFORM_CCFLAGS $COMMON_FLAGS"
-PLATFORM_CXXFLAGS="$PLATFORM_CXXFLAGS $COMMON_FLAGS"
-
-echo "CC=$CC" >> $OUTPUT
-echo "CXX=$CXX" >> $OUTPUT
-echo "PLATFORM=$PLATFORM" >> $OUTPUT
-echo "PLATFORM_LDFLAGS=$PLATFORM_LDFLAGS" >> $OUTPUT
-echo "PLATFORM_LIBS=$PLATFORM_LIBS" >> $OUTPUT
-echo "PLATFORM_CCFLAGS=$PLATFORM_CCFLAGS" >> $OUTPUT
-echo "PLATFORM_CXXFLAGS=$PLATFORM_CXXFLAGS" >> $OUTPUT
-echo "PLATFORM_SHARED_CFLAGS=$PLATFORM_SHARED_CFLAGS" >> $OUTPUT
-echo "PLATFORM_SHARED_EXT=$PLATFORM_SHARED_EXT" >> $OUTPUT
-echo "PLATFORM_SHARED_LDFLAGS=$PLATFORM_SHARED_LDFLAGS" >> $OUTPUT
-echo "PLATFORM_SHARED_VERSIONED=$PLATFORM_SHARED_VERSIONED" >> $OUTPUT

cmake/leveldbConfig.cmake

@@ -0,0 +1 @@
+include("${CMAKE_CURRENT_LIST_DIR}/leveldbTargets.cmake")

db/builder.cc

@@ -41,14 +41,10 @@ Status BuildTable(const std::string& dbname,
     }
 
     // Finish and check for builder errors
+    s = builder->Finish();
     if (s.ok()) {
-      s = builder->Finish();
-      if (s.ok()) {
-        meta->file_size = builder->FileSize();
-        assert(meta->file_size > 0);
-      }
-    } else {
-      builder->Abandon();
+      meta->file_size = builder->FileSize();
+      assert(meta->file_size > 0);
     }
     delete builder;
 
@@ -60,7 +56,7 @@ Status BuildTable(const std::string& dbname,
       s = file->Close();
     }
     delete file;
-    file = NULL;
+    file = nullptr;
 
     if (s.ok()) {
       // Verify that the table is usable

db/builder.h

@@ -22,12 +22,12 @@ class VersionEdit;
 // *meta will be filled with metadata about the generated table.
 // If no data is present in *iter, meta->file_size will be set to
 // zero, and no Table file will be produced.
-extern Status BuildTable(const std::string& dbname,
-                         Env* env,
-                         const Options& options,
-                         TableCache* table_cache,
-                         Iterator* iter,
-                         FileMetaData* meta);
+Status BuildTable(const std::string& dbname,
+                  Env* env,
+                  const Options& options,
+                  TableCache* table_cache,
+                  Iterator* iter,
+                  FileMetaData* meta);
 
 }  // namespace leveldb
 

db/c.cc

@@ -5,7 +5,6 @@
 #include "leveldb/c.h"
 
 #include <stdlib.h>
-#include <unistd.h>
 #include "leveldb/cache.h"
 #include "leveldb/comparator.h"
 #include "leveldb/db.h"
@@ -130,10 +129,10 @@ struct leveldb_env_t {
 };
 
 static bool SaveError(char** errptr, const Status& s) {
-  assert(errptr != NULL);
+  assert(errptr != nullptr);
   if (s.ok()) {
     return false;
-  } else if (*errptr == NULL) {
+  } else if (*errptr == nullptr) {
     *errptr = strdup(s.ToString().c_str());
   } else {
     // TODO(sanjay): Merge with existing error?
@@ -155,7 +154,7 @@ leveldb_t* leveldb_open(
     char** errptr) {
   DB* db;
   if (SaveError(errptr, DB::Open(options->rep, std::string(name), &db))) {
-    return NULL;
+    return nullptr;
   }
   leveldb_t* result = new leveldb_t;
   result->rep = db;
@@ -200,7 +199,7 @@ char* leveldb_get(
     const char* key, size_t keylen,
     size_t* vallen,
     char** errptr) {
-  char* result = NULL;
+  char* result = nullptr;
   std::string tmp;
   Status s = db->rep->Get(options->rep, Slice(key, keylen), &tmp);
   if (s.ok()) {
@@ -245,7 +244,7 @@ char* leveldb_property_value(
     // We use strdup() since we expect human readable output.
     return strdup(tmp.c_str());
   } else {
-    return NULL;
+    return nullptr;
   }
 }
 
@@ -270,9 +269,9 @@ void leveldb_compact_range(
     const char* limit_key, size_t limit_key_len) {
   Slice a, b;
   db->rep->CompactRange(
-      // Pass NULL Slice if corresponding "const char*" is NULL
-      (start_key ? (a = Slice(start_key, start_key_len), &a) : NULL),
-      (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : NULL));
+      // Pass null Slice if corresponding "const char*" is null
+      (start_key ? (a = Slice(start_key, start_key_len), &a) : nullptr),
+      (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : nullptr));
 }
 
 void leveldb_destroy_db(
@@ -360,7 +359,7 @@ void leveldb_writebatch_delete(
 }
 
 void leveldb_writebatch_iterate(
-    leveldb_writebatch_t* b,
+    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)) {
@@ -383,6 +382,11 @@ void leveldb_writebatch_iterate(
   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;
 }
@@ -419,11 +423,11 @@ void leveldb_options_set_paranoid_checks(
 }
 
 void leveldb_options_set_env(leveldb_options_t* opt, leveldb_env_t* env) {
-  opt->rep.env = (env ? env->rep : NULL);
+  opt->rep.env = (env ? env->rep : nullptr);
 }
 
 void leveldb_options_set_info_log(leveldb_options_t* opt, leveldb_logger_t* l) {
-  opt->rep.info_log = (l ? l->rep : NULL);
+  opt->rep.info_log = (l ? l->rep : nullptr);
 }
 
 void leveldb_options_set_write_buffer_size(leveldb_options_t* opt, size_t s) {
@@ -446,6 +450,10 @@ 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);
 }
@@ -514,7 +522,7 @@ leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(int bits_per_key) {
   };
   Wrapper* wrapper = new Wrapper;
   wrapper->rep_ = NewBloomFilterPolicy(bits_per_key);
-  wrapper->state_ = NULL;
+  wrapper->state_ = nullptr;
   wrapper->destructor_ = &Wrapper::DoNothing;
   return wrapper;
 }
@@ -541,7 +549,7 @@ void leveldb_readoptions_set_fill_cache(
 void leveldb_readoptions_set_snapshot(
     leveldb_readoptions_t* opt,
     const leveldb_snapshot_t* snap) {
-  opt->rep.snapshot = (snap ? snap->rep : NULL);
+  opt->rep.snapshot = (snap ? snap->rep : nullptr);
 }
 
 leveldb_writeoptions_t* leveldb_writeoptions_create() {
@@ -580,6 +588,18 @@ void leveldb_env_destroy(leveldb_env_t* env) {
   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*>(malloc(result.size() + 1));
+  memcpy(buffer, result.data(), result.size());
+  buffer[result.size()] = '\0';
+  return buffer;
+}
+
 void leveldb_free(void* ptr) {
   free(ptr);
 }

db/c_test.c

@@ -8,24 +8,14 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <sys/types.h>
-#include <unistd.h>
 
 const char* phase = "";
-static char dbname[200];
 
 static void StartPhase(const char* name) {
   fprintf(stderr, "=== Test %s\n", name);
   phase = name;
 }
 
-static const char* GetTempDir(void) {
-    const char* ret = getenv("TEST_TMPDIR");
-    if (ret == NULL || ret[0] == '\0')
-        ret = "/tmp";
-    return ret;
-}
-
 #define CheckNoError(err)                                               \
   if ((err) != NULL) {                                                  \
     fprintf(stderr, "%s:%d: %s: %s\n", __FILE__, __LINE__, phase, (err)); \
@@ -162,21 +152,19 @@ int main(int argc, char** argv) {
   leveldb_options_t* options;
   leveldb_readoptions_t* roptions;
   leveldb_writeoptions_t* woptions;
+  char* dbname;
   char* err = NULL;
   int run = -1;
 
   CheckCondition(leveldb_major_version() >= 1);
   CheckCondition(leveldb_minor_version() >= 1);
 
-  snprintf(dbname, sizeof(dbname),
-           "%s/leveldb_c_test-%d",
-           GetTempDir(),
-           ((int) geteuid()));
-
   StartPhase("create_objects");
   cmp = leveldb_comparator_create(NULL, CmpDestroy, CmpCompare, CmpName);
   env = leveldb_create_default_env();
   cache = leveldb_cache_create_lru(100000);
+  dbname = leveldb_env_get_test_directory(env);
+  CheckCondition(dbname != NULL);
 
   options = leveldb_options_create();
   leveldb_options_set_comparator(options, cmp);
@@ -189,6 +177,7 @@ int main(int argc, char** argv) {
   leveldb_options_set_max_open_files(options, 10);
   leveldb_options_set_block_size(options, 1024);
   leveldb_options_set_block_restart_interval(options, 8);
+  leveldb_options_set_max_file_size(options, 3 << 20);
   leveldb_options_set_compression(options, leveldb_no_compression);
 
   roptions = leveldb_readoptions_create();
@@ -239,12 +228,18 @@ int main(int argc, char** argv) {
     leveldb_writebatch_clear(wb);
     leveldb_writebatch_put(wb, "bar", 3, "b", 1);
     leveldb_writebatch_put(wb, "box", 3, "c", 1);
-    leveldb_writebatch_delete(wb, "bar", 3);
+
+    leveldb_writebatch_t* wb2 = leveldb_writebatch_create();
+    leveldb_writebatch_delete(wb2, "bar", 3);
+    leveldb_writebatch_append(wb, wb2);
+    leveldb_writebatch_destroy(wb2);
+
     leveldb_write(db, woptions, wb, &err);
     CheckNoError(err);
     CheckGet(db, roptions, "foo", "hello");
     CheckGet(db, roptions, "bar", NULL);
     CheckGet(db, roptions, "box", "c");
+
     int pos = 0;
     leveldb_writebatch_iterate(wb, &pos, CheckPut, CheckDel);
     CheckCondition(pos == 3);
@@ -381,6 +376,7 @@ int main(int argc, char** argv) {
   leveldb_options_destroy(options);
   leveldb_readoptions_destroy(roptions);
   leveldb_writeoptions_destroy(woptions);
+  leveldb_free(dbname);
   leveldb_cache_destroy(cache);
   leveldb_comparator_destroy(cmp);
   leveldb_env_destroy(env);

db/corruption_test.cc

@@ -4,12 +4,8 @@
 
 #include "leveldb/db.h"
 
-#include <errno.h>
-#include <fcntl.h>
-#include <sys/stat.h>
 #include <sys/types.h>
 #include "leveldb/cache.h"
-#include "leveldb/env.h"
 #include "leveldb/table.h"
 #include "leveldb/write_batch.h"
 #include "db/db_impl.h"
@@ -36,10 +32,10 @@ class CorruptionTest {
     tiny_cache_ = NewLRUCache(100);
     options_.env = &env_;
     options_.block_cache = tiny_cache_;
-    dbname_ = test::TmpDir() + "/db_test";
+    dbname_ = "/memenv/corruption_test";
     DestroyDB(dbname_, options_);
 
-    db_ = NULL;
+    db_ = nullptr;
     options_.create_if_missing = true;
     Reopen();
     options_.create_if_missing = false;
@@ -47,13 +43,12 @@ class CorruptionTest {
 
   ~CorruptionTest() {
      delete db_;
-     DestroyDB(dbname_, Options());
      delete tiny_cache_;
   }
 
   Status TryReopen() {
     delete db_;
-    db_ = NULL;
+    db_ = nullptr;
     return DB::Open(options_, dbname_, &db_);
   }
 
@@ -63,7 +58,7 @@ class CorruptionTest {
 
   void RepairDB() {
     delete db_;
-    db_ = NULL;
+    db_ = nullptr;
     ASSERT_OK(::leveldb::RepairDB(dbname_, options_));
   }
 
@@ -126,7 +121,7 @@ class CorruptionTest {
   void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) {
     // Pick file to corrupt
     std::vector<std::string> filenames;
-    ASSERT_OK(env_.GetChildren(dbname_, &filenames));
+    ASSERT_OK(env_.target()->GetChildren(dbname_, &filenames));
     uint64_t number;
     FileType type;
     std::string fname;
@@ -141,35 +136,32 @@ class CorruptionTest {
     }
     ASSERT_TRUE(!fname.empty()) << filetype;
 
-    struct stat sbuf;
-    if (stat(fname.c_str(), &sbuf) != 0) {
-      const char* msg = strerror(errno);
-      ASSERT_TRUE(false) << fname << ": " << msg;
-    }
+    uint64_t file_size;
+    ASSERT_OK(env_.target()->GetFileSize(fname, &file_size));
 
     if (offset < 0) {
       // Relative to end of file; make it absolute
-      if (-offset > sbuf.st_size) {
+      if (-offset > file_size) {
         offset = 0;
       } else {
-        offset = sbuf.st_size + offset;
+        offset = file_size + offset;
       }
     }
-    if (offset > sbuf.st_size) {
-      offset = sbuf.st_size;
+    if (offset > file_size) {
+      offset = file_size;
     }
-    if (offset + bytes_to_corrupt > sbuf.st_size) {
-      bytes_to_corrupt = sbuf.st_size - offset;
+    if (offset + bytes_to_corrupt > file_size) {
+      bytes_to_corrupt = file_size - offset;
     }
 
     // Do it
     std::string contents;
-    Status s = ReadFileToString(Env::Default(), fname, &contents);
+    Status s = ReadFileToString(env_.target(), fname, &contents);
     ASSERT_TRUE(s.ok()) << s.ToString();
     for (int i = 0; i < bytes_to_corrupt; i++) {
       contents[i + offset] ^= 0x80;
     }
-    s = WriteStringToFile(Env::Default(), contents, fname);
+    s = WriteStringToFile(env_.target(), contents, fname);
     ASSERT_TRUE(s.ok()) << s.ToString();
   }
 
@@ -237,8 +229,8 @@ TEST(CorruptionTest, TableFile) {
   Build(100);
   DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
   dbi->TEST_CompactMemTable();
-  dbi->TEST_CompactRange(0, NULL, NULL);
-  dbi->TEST_CompactRange(1, NULL, NULL);
+  dbi->TEST_CompactRange(0, nullptr, nullptr);
+  dbi->TEST_CompactRange(1, nullptr, nullptr);
 
   Corrupt(kTableFile, 100, 1);
   Check(90, 99);
@@ -251,8 +243,8 @@ TEST(CorruptionTest, TableFileRepair) {
   Build(100);
   DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
   dbi->TEST_CompactMemTable();
-  dbi->TEST_CompactRange(0, NULL, NULL);
-  dbi->TEST_CompactRange(1, NULL, NULL);
+  dbi->TEST_CompactRange(0, nullptr, nullptr);
+  dbi->TEST_CompactRange(1, nullptr, nullptr);
 
   Corrupt(kTableFile, 100, 1);
   RepairDB();
@@ -302,7 +294,7 @@ TEST(CorruptionTest, CorruptedDescriptor) {
   ASSERT_OK(db_->Put(WriteOptions(), "foo", "hello"));
   DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
   dbi->TEST_CompactMemTable();
-  dbi->TEST_CompactRange(0, NULL, NULL);
+  dbi->TEST_CompactRange(0, nullptr, nullptr);
 
   Corrupt(kDescriptorFile, 0, 1000);
   Status s = TryReopen();
@@ -343,7 +335,7 @@ TEST(CorruptionTest, CompactionInputErrorParanoid) {
     Corrupt(kTableFile, 100, 1);
     env_.SleepForMicroseconds(100000);
   }
-  dbi->CompactRange(NULL, NULL);
+  dbi->CompactRange(nullptr, nullptr);
 
   // Write must fail because of corrupted table
   std::string tmp1, tmp2;

db/db_bench.cc

@@ -5,11 +5,10 @@
 #include <sys/types.h>
 #include <stdio.h>
 #include <stdlib.h>
-#include "db/db_impl.h"
-#include "db/version_set.h"
 #include "leveldb/cache.h"
 #include "leveldb/db.h"
 #include "leveldb/env.h"
+#include "leveldb/filter_policy.h"
 #include "leveldb/write_batch.h"
 #include "port/port.h"
 #include "util/crc32c.h"
@@ -33,8 +32,8 @@
 //      readmissing   -- read N missing keys in random order
 //      readhot       -- read N times in random order from 1% section of DB
 //      seekrandom    -- N random seeks
+//      open          -- cost of opening a DB
 //      crc32c        -- repeated crc32c of 4K of data
-//      acquireload   -- load N*1000 times
 //   Meta operations:
 //      compact     -- Compact the entire DB
 //      stats       -- Print DB stats
@@ -57,7 +56,6 @@ static const char* FLAGS_benchmarks =
     "crc32c,"
     "snappycomp,"
     "snappyuncomp,"
-    "acquireload,"
     ;
 
 // Number of key/values to place in database
@@ -83,6 +81,14 @@ static bool FLAGS_histogram = false;
 // (initialized to default value by "main")
 static int FLAGS_write_buffer_size = 0;
 
+// Number of bytes written to each file.
+// (initialized to default value by "main")
+static int FLAGS_max_file_size = 0;
+
+// Approximate size of user data packed per block (before compression.
+// (initialized to default value by "main")
+static int FLAGS_block_size = 0;
+
 // Number of bytes to use as a cache of uncompressed data.
 // Negative means use default settings.
 static int FLAGS_cache_size = -1;
@@ -99,12 +105,16 @@ static int FLAGS_bloom_bits = -1;
 // benchmark will fail.
 static bool FLAGS_use_existing_db = false;
 
+// If true, reuse existing log/MANIFEST files when re-opening a database.
+static bool FLAGS_reuse_logs = false;
+
 // Use the db with the following name.
-static const char* FLAGS_db = NULL;
+static const char* FLAGS_db = nullptr;
 
 namespace leveldb {
 
 namespace {
+leveldb::Env* g_env = nullptr;
 
 // Helper for quickly generating random data.
 class RandomGenerator {
@@ -138,6 +148,7 @@ class RandomGenerator {
   }
 };
 
+#if defined(__linux)
 static Slice TrimSpace(Slice s) {
   size_t start = 0;
   while (start < s.size() && isspace(s[start])) {
@@ -149,6 +160,7 @@ static Slice TrimSpace(Slice s) {
   }
   return Slice(s.data() + start, limit - start);
 }
+#endif
 
 static void AppendWithSpace(std::string* str, Slice msg) {
   if (msg.empty()) return;
@@ -180,7 +192,7 @@ class Stats {
     done_ = 0;
     bytes_ = 0;
     seconds_ = 0;
-    start_ = Env::Default()->NowMicros();
+    start_ = g_env->NowMicros();
     finish_ = start_;
     message_.clear();
   }
@@ -198,7 +210,7 @@ class Stats {
   }
 
   void Stop() {
-    finish_ = Env::Default()->NowMicros();
+    finish_ = g_env->NowMicros();
     seconds_ = (finish_ - start_) * 1e-6;
   }
 
@@ -208,7 +220,7 @@ class Stats {
 
   void FinishedSingleOp() {
     if (FLAGS_histogram) {
-      double now = Env::Default()->NowMicros();
+      double now = g_env->NowMicros();
       double micros = now - last_op_finish_;
       hist_.Add(micros);
       if (micros > 20000) {
@@ -268,8 +280,8 @@ class Stats {
 // State shared by all concurrent executions of the same benchmark.
 struct SharedState {
   port::Mutex mu;
-  port::CondVar cv;
-  int total;
+  port::CondVar cv GUARDED_BY(mu);
+  int total GUARDED_BY(mu);
 
   // Each thread goes through the following states:
   //    (1) initializing
@@ -277,11 +289,12 @@ struct SharedState {
   //    (3) running
   //    (4) done
 
-  int num_initialized;
-  int num_done;
-  bool start;
+  int num_initialized GUARDED_BY(mu);
+  int num_done GUARDED_BY(mu);
+  bool start GUARDED_BY(mu);
 
-  SharedState() : cv(&mu) { }
+  SharedState(int total)
+      : cv(&mu), total(total), num_initialized(0), num_done(0), start(false) { }
 };
 
 // Per-thread state for concurrent executions of the same benchmark.
@@ -355,18 +368,18 @@ class Benchmark {
             kMajorVersion, kMinorVersion);
 
 #if defined(__linux)
-    time_t now = time(NULL);
+    time_t now = time(nullptr);
     fprintf(stderr, "Date:       %s", ctime(&now));  // ctime() adds newline
 
     FILE* cpuinfo = fopen("/proc/cpuinfo", "r");
-    if (cpuinfo != NULL) {
+    if (cpuinfo != nullptr) {
       char line[1000];
       int num_cpus = 0;
       std::string cpu_type;
       std::string cache_size;
-      while (fgets(line, sizeof(line), cpuinfo) != NULL) {
+      while (fgets(line, sizeof(line), cpuinfo) != nullptr) {
         const char* sep = strchr(line, ':');
-        if (sep == NULL) {
+        if (sep == nullptr) {
           continue;
         }
         Slice key = TrimSpace(Slice(line, sep - 1 - line));
@@ -387,21 +400,21 @@ class Benchmark {
 
  public:
   Benchmark()
-  : cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : NULL),
+  : cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : nullptr),
     filter_policy_(FLAGS_bloom_bits >= 0
                    ? NewBloomFilterPolicy(FLAGS_bloom_bits)
-                   : NULL),
-    db_(NULL),
+                   : nullptr),
+    db_(nullptr),
     num_(FLAGS_num),
     value_size_(FLAGS_value_size),
     entries_per_batch_(1),
     reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads),
     heap_counter_(0) {
     std::vector<std::string> files;
-    Env::Default()->GetChildren(FLAGS_db, &files);
+    g_env->GetChildren(FLAGS_db, &files);
     for (size_t i = 0; i < files.size(); i++) {
       if (Slice(files[i]).starts_with("heap-")) {
-        Env::Default()->DeleteFile(std::string(FLAGS_db) + "/" + files[i]);
+        g_env->DeleteFile(std::string(FLAGS_db) + "/" + files[i]);
       }
     }
     if (!FLAGS_use_existing_db) {
@@ -420,12 +433,12 @@ class Benchmark {
     Open();
 
     const char* benchmarks = FLAGS_benchmarks;
-    while (benchmarks != NULL) {
+    while (benchmarks != nullptr) {
       const char* sep = strchr(benchmarks, ',');
       Slice name;
-      if (sep == NULL) {
+      if (sep == nullptr) {
         name = benchmarks;
-        benchmarks = NULL;
+        benchmarks = nullptr;
       } else {
         name = Slice(benchmarks, sep - benchmarks);
         benchmarks = sep + 1;
@@ -438,11 +451,15 @@ class Benchmark {
       entries_per_batch_ = 1;
       write_options_ = WriteOptions();
 
-      void (Benchmark::*method)(ThreadState*) = NULL;
+      void (Benchmark::*method)(ThreadState*) = nullptr;
       bool fresh_db = false;
       int num_threads = FLAGS_threads;
 
-      if (name == Slice("fillseq")) {
+      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")) {
@@ -491,8 +508,6 @@ class Benchmark {
         method = &Benchmark::Compact;
       } else if (name == Slice("crc32c")) {
         method = &Benchmark::Crc32c;
-      } else if (name == Slice("acquireload")) {
-        method = &Benchmark::AcquireLoad;
       } else if (name == Slice("snappycomp")) {
         method = &Benchmark::SnappyCompress;
       } else if (name == Slice("snappyuncomp")) {
@@ -513,16 +528,16 @@ class Benchmark {
         if (FLAGS_use_existing_db) {
           fprintf(stdout, "%-12s : skipped (--use_existing_db is true)\n",
                   name.ToString().c_str());
-          method = NULL;
+          method = nullptr;
         } else {
           delete db_;
-          db_ = NULL;
+          db_ = nullptr;
           DestroyDB(FLAGS_db, Options());
           Open();
         }
       }
 
-      if (method != NULL) {
+      if (method != nullptr) {
         RunBenchmark(num_threads, name, method);
       }
     }
@@ -566,11 +581,7 @@ class Benchmark {
 
   void RunBenchmark(int n, Slice name,
                     void (Benchmark::*method)(ThreadState*)) {
-    SharedState shared;
-    shared.total = n;
-    shared.num_initialized = 0;
-    shared.num_done = 0;
-    shared.start = false;
+    SharedState shared(n);
 
     ThreadArg* arg = new ThreadArg[n];
     for (int i = 0; i < n; i++) {
@@ -579,7 +590,7 @@ class Benchmark {
       arg[i].shared = &shared;
       arg[i].thread = new ThreadState(i);
       arg[i].thread->shared = &shared;
-      Env::Default()->StartThread(ThreadBody, &arg[i]);
+      g_env->StartThread(ThreadBody, &arg[i]);
     }
 
     shared.mu.Lock();
@@ -624,22 +635,6 @@ class Benchmark {
     thread->stats.AddMessage(label);
   }
 
-  void AcquireLoad(ThreadState* thread) {
-    int dummy;
-    port::AtomicPointer ap(&dummy);
-    int count = 0;
-    void *ptr = NULL;
-    thread->stats.AddMessage("(each op is 1000 loads)");
-    while (count < 100000) {
-      for (int i = 0; i < 1000; i++) {
-        ptr = ap.Acquire_Load();
-      }
-      count++;
-      thread->stats.FinishedSingleOp();
-    }
-    if (ptr == NULL) exit(1); // Disable unused variable warning.
-  }
-
   void SnappyCompress(ThreadState* thread) {
     RandomGenerator gen;
     Slice input = gen.Generate(Options().block_size);
@@ -688,13 +683,17 @@ class Benchmark {
   }
 
   void Open() {
-    assert(db_ == NULL);
+    assert(db_ == nullptr);
     Options options;
+    options.env = g_env;
     options.create_if_missing = !FLAGS_use_existing_db;
     options.block_cache = cache_;
     options.write_buffer_size = FLAGS_write_buffer_size;
+    options.max_file_size = FLAGS_max_file_size;
+    options.block_size = FLAGS_block_size;
     options.max_open_files = FLAGS_open_files;
     options.filter_policy = filter_policy_;
+    options.reuse_logs = FLAGS_reuse_logs;
     Status s = DB::Open(options, FLAGS_db, &db_);
     if (!s.ok()) {
       fprintf(stderr, "open error: %s\n", s.ToString().c_str());
@@ -702,6 +701,14 @@ class Benchmark {
     }
   }
 
+  void OpenBench(ThreadState* thread) {
+    for (int i = 0; i < num_; i++) {
+      delete db_;
+      Open();
+      thread->stats.FinishedSingleOp();
+    }
+  }
+
   void WriteSeq(ThreadState* thread) {
     DoWrite(thread, true);
   }
@@ -887,7 +894,7 @@ class Benchmark {
   }
 
   void Compact(ThreadState* thread) {
-    db_->CompactRange(NULL, NULL);
+    db_->CompactRange(nullptr, nullptr);
   }
 
   void PrintStats(const char* key) {
@@ -906,7 +913,7 @@ class Benchmark {
     char fname[100];
     snprintf(fname, sizeof(fname), "%s/heap-%04d", FLAGS_db, ++heap_counter_);
     WritableFile* file;
-    Status s = Env::Default()->NewWritableFile(fname, &file);
+    Status s = g_env->NewWritableFile(fname, &file);
     if (!s.ok()) {
       fprintf(stderr, "%s\n", s.ToString().c_str());
       return;
@@ -915,7 +922,7 @@ class Benchmark {
     delete file;
     if (!ok) {
       fprintf(stderr, "heap profiling not supported\n");
-      Env::Default()->DeleteFile(fname);
+      g_env->DeleteFile(fname);
     }
   }
 };
@@ -924,6 +931,8 @@ class Benchmark {
 
 int main(int argc, char** argv) {
   FLAGS_write_buffer_size = leveldb::Options().write_buffer_size;
+  FLAGS_max_file_size = leveldb::Options().max_file_size;
+  FLAGS_block_size = leveldb::Options().block_size;
   FLAGS_open_files = leveldb::Options().max_open_files;
   std::string default_db_path;
 
@@ -941,6 +950,9 @@ int main(int argc, char** argv) {
     } else if (sscanf(argv[i], "--use_existing_db=%d%c", &n, &junk) == 1 &&
                (n == 0 || n == 1)) {
       FLAGS_use_existing_db = n;
+    } else if (sscanf(argv[i], "--reuse_logs=%d%c", &n, &junk) == 1 &&
+               (n == 0 || n == 1)) {
+      FLAGS_reuse_logs = n;
     } else if (sscanf(argv[i], "--num=%d%c", &n, &junk) == 1) {
       FLAGS_num = n;
     } else if (sscanf(argv[i], "--reads=%d%c", &n, &junk) == 1) {
@@ -951,6 +963,10 @@ int main(int argc, char** argv) {
       FLAGS_value_size = n;
     } else if (sscanf(argv[i], "--write_buffer_size=%d%c", &n, &junk) == 1) {
       FLAGS_write_buffer_size = n;
+    } else if (sscanf(argv[i], "--max_file_size=%d%c", &n, &junk) == 1) {
+      FLAGS_max_file_size = n;
+    } else if (sscanf(argv[i], "--block_size=%d%c", &n, &junk) == 1) {
+      FLAGS_block_size = n;
     } else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) {
       FLAGS_cache_size = n;
     } else if (sscanf(argv[i], "--bloom_bits=%d%c", &n, &junk) == 1) {
@@ -965,9 +981,11 @@ int main(int argc, char** argv) {
     }
   }
 
+  leveldb::g_env = leveldb::Env::Default();
+
   // Choose a location for the test database if none given with --db=<path>
-  if (FLAGS_db == NULL) {
-      leveldb::Env::Default()->GetTestDirectory(&default_db_path);
+  if (FLAGS_db == nullptr) {
+      leveldb::g_env->GetTestDirectory(&default_db_path);
       default_db_path += "/dbbench";
       FLAGS_db = default_db_path.c_str();
   }

db/db_impl.h

@@ -5,8 +5,11 @@
 #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"
@@ -78,21 +81,21 @@ class DBImpl : public DB {
   // Recover the descriptor from persistent storage.  May do a significant
   // amount of work to recover recently logged updates.  Any changes to
   // be made to the descriptor are added to *edit.
-  Status Recover(VersionEdit* edit) EXCLUSIVE_LOCKS_REQUIRED(mutex_);
+  Status Recover(VersionEdit* edit, bool* save_manifest)
+      EXCLUSIVE_LOCKS_REQUIRED(mutex_);
 
   void MaybeIgnoreError(Status* s) const;
 
   // Delete any unneeded files and stale in-memory entries.
-  void DeleteObsoleteFiles();
+  void DeleteObsoleteFiles() EXCLUSIVE_LOCKS_REQUIRED(mutex_);
 
   // Compact the in-memory write buffer to disk.  Switches to a new
   // log-file/memtable and writes a new descriptor iff successful.
   // Errors are recorded in bg_error_.
   void CompactMemTable() EXCLUSIVE_LOCKS_REQUIRED(mutex_);
 
-  Status RecoverLogFile(uint64_t log_number,
-                        VersionEdit* edit,
-                        SequenceNumber* max_sequence)
+  Status RecoverLogFile(uint64_t log_number, bool last_log, bool* save_manifest,
+                        VersionEdit* edit, SequenceNumber* max_sequence)
       EXCLUSIVE_LOCKS_REQUIRED(mutex_);
 
   Status WriteLevel0Table(MemTable* mem, VersionEdit* edit, Version* base)
@@ -100,14 +103,15 @@ class DBImpl : public DB {
 
   Status MakeRoomForWrite(bool force /* compact even if there is room? */)
       EXCLUSIVE_LOCKS_REQUIRED(mutex_);
-  WriteBatch* BuildBatchGroup(Writer** last_writer);
+  WriteBatch* BuildBatchGroup(Writer** last_writer)
+      EXCLUSIVE_LOCKS_REQUIRED(mutex_);
 
   void RecordBackgroundError(const Status& s);
 
   void MaybeScheduleCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_);
   static void BGWork(void* db);
   void BackgroundCall();
-  void  BackgroundCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_);
+  void BackgroundCompaction() EXCLUSIVE_LOCKS_REQUIRED(mutex_);
   void CleanupCompaction(CompactionState* compact)
       EXCLUSIVE_LOCKS_REQUIRED(mutex_);
   Status DoCompactionWork(CompactionState* compact)
@@ -123,55 +127,55 @@ class DBImpl : public DB {
   const InternalKeyComparator internal_comparator_;
   const InternalFilterPolicy internal_filter_policy_;
   const Options options_;  // options_.comparator == &internal_comparator_
-  bool owns_info_log_;
-  bool owns_cache_;
+  const bool owns_info_log_;
+  const bool owns_cache_;
   const std::string dbname_;
 
   // table_cache_ provides its own synchronization
-  TableCache* table_cache_;
+  TableCache* const table_cache_;
 
-  // Lock over the persistent DB state.  Non-NULL iff successfully acquired.
+  // Lock over the persistent DB state.  Non-null iff successfully acquired.
   FileLock* db_lock_;
 
   // State below is protected by mutex_
   port::Mutex mutex_;
-  port::AtomicPointer shutting_down_;
-  port::CondVar bg_cv_;          // Signalled when background work finishes
+  std::atomic<bool> shutting_down_;
+  port::CondVar background_work_finished_signal_ GUARDED_BY(mutex_);
   MemTable* mem_;
-  MemTable* imm_;                // Memtable being compacted
-  port::AtomicPointer has_imm_;  // So bg thread can detect non-NULL imm_
+  MemTable* imm_ GUARDED_BY(mutex_);  // Memtable being compacted
+  std::atomic<bool> has_imm_;         // So bg thread can detect non-null imm_
   WritableFile* logfile_;
-  uint64_t logfile_number_;
+  uint64_t logfile_number_ GUARDED_BY(mutex_);
   log::Writer* log_;
-  uint32_t seed_;                // For sampling.
+  uint32_t seed_ GUARDED_BY(mutex_);  // For sampling.
 
   // Queue of writers.
-  std::deque<Writer*> writers_;
-  WriteBatch* tmp_batch_;
+  std::deque<Writer*> writers_ GUARDED_BY(mutex_);
+  WriteBatch* tmp_batch_ GUARDED_BY(mutex_);
 
-  SnapshotList snapshots_;
+  SnapshotList snapshots_ GUARDED_BY(mutex_);
 
   // Set of table files to protect from deletion because they are
   // part of ongoing compactions.
-  std::set<uint64_t> pending_outputs_;
+  std::set<uint64_t> pending_outputs_ GUARDED_BY(mutex_);
 
   // Has a background compaction been scheduled or is running?
-  bool bg_compaction_scheduled_;
+  bool background_compaction_scheduled_ GUARDED_BY(mutex_);
 
   // Information for a manual compaction
   struct ManualCompaction {
     int level;
     bool done;
-    const InternalKey* begin;   // NULL means beginning of key range
-    const InternalKey* end;     // NULL means end of key range
+    const InternalKey* begin;   // null means beginning of key range
+    const InternalKey* end;     // null means end of key range
     InternalKey tmp_storage;    // Used to keep track of compaction progress
   };
-  ManualCompaction* manual_compaction_;
+  ManualCompaction* manual_compaction_ GUARDED_BY(mutex_);
 
-  VersionSet* versions_;
+  VersionSet* const versions_;
 
   // Have we encountered a background error in paranoid mode?
-  Status bg_error_;
+  Status bg_error_ GUARDED_BY(mutex_);
 
   // Per level compaction stats.  stats_[level] stores the stats for
   // compactions that produced data for the specified "level".
@@ -188,7 +192,7 @@ class DBImpl : public DB {
       this->bytes_written += c.bytes_written;
     }
   };
-  CompactionStats stats_[config::kNumLevels];
+  CompactionStats stats_[config::kNumLevels] GUARDED_BY(mutex_);
 
   // No copying allowed
   DBImpl(const DBImpl&);
@@ -201,10 +205,10 @@ class DBImpl : public DB {
 
 // Sanitize db options.  The caller should delete result.info_log if
 // it is not equal to src.info_log.
-extern Options SanitizeOptions(const std::string& db,
-                               const InternalKeyComparator* icmp,
-                               const InternalFilterPolicy* ipolicy,
-                               const Options& src);
+Options SanitizeOptions(const std::string& db,
+                        const InternalKeyComparator* icmp,
+                        const InternalFilterPolicy* ipolicy,
+                        const Options& src);
 
 }  // namespace leveldb
 

db/db_iter.cc

@@ -57,7 +57,7 @@ class DBIter: public Iterator {
         direction_(kForward),
         valid_(false),
         rnd_(seed),
-        bytes_counter_(RandomPeriod()) {
+        bytes_until_read_sampling_(RandomCompactionPeriod()) {
   }
   virtual ~DBIter() {
     delete iter_;
@@ -103,8 +103,8 @@ class DBIter: public Iterator {
     }
   }
 
-  // Pick next gap with average value of config::kReadBytesPeriod.
-  ssize_t RandomPeriod() {
+  // Picks the number of bytes that can be read until a compaction is scheduled.
+  size_t RandomCompactionPeriod() {
     return rnd_.Uniform(2*config::kReadBytesPeriod);
   }
 
@@ -120,7 +120,7 @@ class DBIter: public Iterator {
   bool valid_;
 
   Random rnd_;
-  ssize_t bytes_counter_;
+  size_t bytes_until_read_sampling_;
 
   // No copying allowed
   DBIter(const DBIter&);
@@ -129,12 +129,15 @@ class DBIter: public Iterator {
 
 inline bool DBIter::ParseKey(ParsedInternalKey* ikey) {
   Slice k = iter_->key();
-  ssize_t n = k.size() + iter_->value().size();
-  bytes_counter_ -= n;
-  while (bytes_counter_ < 0) {
-    bytes_counter_ += RandomPeriod();
+
+  size_t bytes_read = k.size() + iter_->value().size();
+  while (bytes_until_read_sampling_ < bytes_read) {
+    bytes_until_read_sampling_ += RandomCompactionPeriod();
     db_->RecordReadSample(k);
   }
+  assert(bytes_until_read_sampling_ >= bytes_read);
+  bytes_until_read_sampling_ -= bytes_read;
+
   if (!ParseInternalKey(k, ikey)) {
     status_ = Status::Corruption("corrupted internal key in DBIter");
     return false;

db/db_iter.h

@@ -16,12 +16,11 @@ class DBImpl;
 // Return a new iterator that converts internal keys (yielded by
 // "*internal_iter") that were live at the specified "sequence" number
 // into appropriate user keys.
-extern Iterator* NewDBIterator(
-    DBImpl* db,
-    const Comparator* user_key_comparator,
-    Iterator* internal_iter,
-    SequenceNumber sequence,
-    uint32_t seed);
+Iterator* NewDBIterator(DBImpl* db,
+                        const Comparator* user_key_comparator,
+                        Iterator* internal_iter,
+                        SequenceNumber sequence,
+                        uint32_t seed);
 
 }  // namespace leveldb
 

db/db_test.cc

@@ -2,6 +2,9 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 
+#include <atomic>
+#include <string>
+
 #include "leveldb/db.h"
 #include "leveldb/filter_policy.h"
 #include "db/db_impl.h"
@@ -11,6 +14,8 @@
 #include "leveldb/cache.h"
 #include "leveldb/env.h"
 #include "leveldb/table.h"
+#include "port/port.h"
+#include "port/thread_annotations.h"
 #include "util/hash.h"
 #include "util/logging.h"
 #include "util/mutexlock.h"
@@ -25,25 +30,32 @@ static std::string RandomString(Random* rnd, int len) {
   return r;
 }
 
+static std::string RandomKey(Random* rnd) {
+  int len = (rnd->OneIn(3)
+             ? 1                // Short sometimes to encourage collisions
+             : (rnd->OneIn(100) ? rnd->Skewed(10) : rnd->Uniform(10)));
+  return test::RandomKey(rnd, len);
+}
+
 namespace {
 class AtomicCounter {
  private:
   port::Mutex mu_;
-  int count_;
+  int count_ GUARDED_BY(mu_);
  public:
   AtomicCounter() : count_(0) { }
   void Increment() {
     IncrementBy(1);
   }
-  void IncrementBy(int count) {
+  void IncrementBy(int count) LOCKS_EXCLUDED(mu_) {
     MutexLock l(&mu_);
     count_ += count;
   }
-  int Read() {
+  int Read() LOCKS_EXCLUDED(mu_) {
     MutexLock l(&mu_);
     return count_;
   }
-  void Reset() {
+  void Reset() LOCKS_EXCLUDED(mu_) {
     MutexLock l(&mu_);
     count_ = 0;
   }
@@ -52,47 +64,77 @@ class AtomicCounter {
 void DelayMilliseconds(int millis) {
   Env::Default()->SleepForMicroseconds(millis * 1000);
 }
-}
+}  // namespace
 
-// Special Env used to delay background operations
+// Test Env to override default Env behavior for testing.
+class TestEnv : public EnvWrapper {
+ public:
+  explicit TestEnv(Env* base) : EnvWrapper(base), ignore_dot_files_(false) {}
+
+  void SetIgnoreDotFiles(bool ignored) { ignore_dot_files_ = ignored; }
+
+  Status GetChildren(const std::string& dir,
+                     std::vector<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.
-  port::AtomicPointer delay_data_sync_;
+  // 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.
-  port::AtomicPointer data_sync_error_;
+  std::atomic<bool> data_sync_error_;
 
-  // Simulate no-space errors while this pointer is non-NULL.
-  port::AtomicPointer no_space_;
+  // Simulate no-space errors while this pointer is non-null.
+  std::atomic<bool> no_space_;
 
-  // Simulate non-writable file system while this pointer is non-NULL
-  port::AtomicPointer non_writable_;
+  // Simulate non-writable file system while this pointer is non-null.
+  std::atomic<bool> non_writable_;
 
-  // Force sync of manifest files to fail while this pointer is non-NULL
-  port::AtomicPointer manifest_sync_error_;
+  // Force sync of manifest files to fail while this pointer is non-null.
+  std::atomic<bool> manifest_sync_error_;
 
-  // Force write to manifest files to fail while this pointer is non-NULL
-  port::AtomicPointer manifest_write_error_;
+  // Force write to manifest files to fail while this pointer is non-null.
+  std::atomic<bool> manifest_write_error_;
 
   bool count_random_reads_;
   AtomicCounter random_read_counter_;
 
-  explicit SpecialEnv(Env* base) : EnvWrapper(base) {
-    delay_data_sync_.Release_Store(NULL);
-    data_sync_error_.Release_Store(NULL);
-    no_space_.Release_Store(NULL);
-    non_writable_.Release_Store(NULL);
-    count_random_reads_ = false;
-    manifest_sync_error_.Release_Store(NULL);
-    manifest_write_error_.Release_Store(NULL);
+  explicit SpecialEnv(Env* base) : EnvWrapper(base),
+    delay_data_sync_(false),
+    data_sync_error_(false),
+    no_space_(false),
+    non_writable_(false),
+    manifest_sync_error_(false),
+    manifest_write_error_(false),
+    count_random_reads_(false) {
   }
 
   Status NewWritableFile(const std::string& f, WritableFile** r) {
     class DataFile : public WritableFile {
      private:
-      SpecialEnv* env_;
-      WritableFile* base_;
+      SpecialEnv* const env_;
+      WritableFile* const base_;
 
      public:
       DataFile(SpecialEnv* env, WritableFile* base)
@@ -101,7 +143,7 @@ class SpecialEnv : public EnvWrapper {
       }
       ~DataFile() { delete base_; }
       Status Append(const Slice& data) {
-        if (env_->no_space_.Acquire_Load() != NULL) {
+        if (env_->no_space_.load(std::memory_order_acquire)) {
           // Drop writes on the floor
           return Status::OK();
         } else {
@@ -111,10 +153,10 @@ class SpecialEnv : public EnvWrapper {
       Status Close() { return base_->Close(); }
       Status Flush() { return base_->Flush(); }
       Status Sync() {
-        if (env_->data_sync_error_.Acquire_Load() != NULL) {
+        if (env_->data_sync_error_.load(std::memory_order_acquire)) {
           return Status::IOError("simulated data sync error");
         }
-        while (env_->delay_data_sync_.Acquire_Load() != NULL) {
+        while (env_->delay_data_sync_.load(std::memory_order_acquire)) {
           DelayMilliseconds(100);
         }
         return base_->Sync();
@@ -128,7 +170,7 @@ class SpecialEnv : public EnvWrapper {
       ManifestFile(SpecialEnv* env, WritableFile* b) : env_(env), base_(b) { }
       ~ManifestFile() { delete base_; }
       Status Append(const Slice& data) {
-        if (env_->manifest_write_error_.Acquire_Load() != NULL) {
+        if (env_->manifest_write_error_.load(std::memory_order_acquire)) {
           return Status::IOError("simulated writer error");
         } else {
           return base_->Append(data);
@@ -137,7 +179,7 @@ class SpecialEnv : public EnvWrapper {
       Status Close() { return base_->Close(); }
       Status Flush() { return base_->Flush(); }
       Status Sync() {
-        if (env_->manifest_sync_error_.Acquire_Load() != NULL) {
+        if (env_->manifest_sync_error_.load(std::memory_order_acquire)) {
           return Status::IOError("simulated sync error");
         } else {
           return base_->Sync();
@@ -145,16 +187,16 @@ class SpecialEnv : public EnvWrapper {
       }
     };
 
-    if (non_writable_.Acquire_Load() != NULL) {
+    if (non_writable_.load(std::memory_order_acquire)) {
       return Status::IOError("simulated write error");
     }
 
     Status s = target()->NewWritableFile(f, r);
     if (s.ok()) {
-      if (strstr(f.c_str(), ".ldb") != NULL ||
-          strstr(f.c_str(), ".log") != NULL) {
+      if (strstr(f.c_str(), ".ldb") != nullptr ||
+          strstr(f.c_str(), ".log") != nullptr) {
         *r = new DataFile(this, *r);
-      } else if (strstr(f.c_str(), "MANIFEST") != NULL) {
+      } else if (strstr(f.c_str(), "MANIFEST") != nullptr) {
         *r = new ManifestFile(this, *r);
       }
     }
@@ -193,6 +235,7 @@ class DBTest {
   // Sequence of option configurations to try
   enum OptionConfig {
     kDefault,
+    kReuse,
     kFilter,
     kUncompressed,
     kEnd
@@ -211,7 +254,7 @@ class DBTest {
     filter_policy_ = NewBloomFilterPolicy(10);
     dbname_ = test::TmpDir() + "/db_test";
     DestroyDB(dbname_, Options());
-    db_ = NULL;
+    db_ = nullptr;
     Reopen();
   }
 
@@ -237,7 +280,11 @@ class DBTest {
   // Return the current option configuration.
   Options CurrentOptions() {
     Options options;
+    options.reuse_logs = false;
     switch (option_config_) {
+      case kReuse:
+        options.reuse_logs = true;
+        break;
       case kFilter:
         options.filter_policy = filter_policy_;
         break;
@@ -254,27 +301,27 @@ class DBTest {
     return reinterpret_cast<DBImpl*>(db_);
   }
 
-  void Reopen(Options* options = NULL) {
+  void Reopen(Options* options = nullptr) {
     ASSERT_OK(TryReopen(options));
   }
 
   void Close() {
     delete db_;
-    db_ = NULL;
+    db_ = nullptr;
   }
 
-  void DestroyAndReopen(Options* options = NULL) {
+  void DestroyAndReopen(Options* options = nullptr) {
     delete db_;
-    db_ = NULL;
+    db_ = nullptr;
     DestroyDB(dbname_, Options());
     ASSERT_OK(TryReopen(options));
   }
 
   Status TryReopen(Options* options) {
     delete db_;
-    db_ = NULL;
+    db_ = nullptr;
     Options opts;
-    if (options != NULL) {
+    if (options != nullptr) {
       opts = *options;
     } else {
       opts = CurrentOptions();
@@ -293,7 +340,7 @@ class DBTest {
     return db_->Delete(WriteOptions(), k);
   }
 
-  std::string Get(const std::string& k, const Snapshot* snapshot = NULL) {
+  std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) {
     ReadOptions options;
     options.snapshot = snapshot;
     std::string result;
@@ -380,7 +427,7 @@ class DBTest {
     ASSERT_TRUE(
         db_->GetProperty("leveldb.num-files-at-level" + NumberToString(level),
                          &property));
-    return atoi(property.c_str());
+    return std::stoi(property);
   }
 
   int TotalTableFiles() {
@@ -426,11 +473,12 @@ class DBTest {
   }
 
   // Do n memtable compactions, each of which produces an sstable
-  // covering the range [small,large].
-  void MakeTables(int n, const std::string& small, const std::string& large) {
+  // covering the range [small_key,large_key].
+  void MakeTables(int n, const std::string& small_key,
+                  const std::string& large_key) {
     for (int i = 0; i < n; i++) {
-      Put(small, "begin");
-      Put(large, "end");
+      Put(small_key, "begin");
+      Put(large_key, "end");
       dbfull()->TEST_CompactMemTable();
     }
   }
@@ -505,11 +553,31 @@ class DBTest {
 
 TEST(DBTest, Empty) {
   do {
-    ASSERT_TRUE(db_ != NULL);
+    ASSERT_TRUE(db_ != nullptr);
     ASSERT_EQ("NOT_FOUND", Get("foo"));
   } while (ChangeOptions());
 }
 
+TEST(DBTest, EmptyKey) {
+  do {
+    ASSERT_OK(Put("", "v1"));
+    ASSERT_EQ("v1", Get(""));
+    ASSERT_OK(Put("", "v2"));
+    ASSERT_EQ("v2", Get(""));
+  } while (ChangeOptions());
+}
+
+TEST(DBTest, EmptyValue) {
+  do {
+    ASSERT_OK(Put("key", "v1"));
+    ASSERT_EQ("v1", Get("key"));
+    ASSERT_OK(Put("key", ""));
+    ASSERT_EQ("", Get("key"));
+    ASSERT_OK(Put("key", "v2"));
+    ASSERT_EQ("v2", Get("key"));
+  } while (ChangeOptions());
+}
+
 TEST(DBTest, ReadWrite) {
   do {
     ASSERT_OK(Put("foo", "v1"));
@@ -542,11 +610,13 @@ TEST(DBTest, GetFromImmutableLayer) {
     ASSERT_OK(Put("foo", "v1"));
     ASSERT_EQ("v1", Get("foo"));
 
-    env_->delay_data_sync_.Release_Store(env_);      // Block sync calls
-    Put("k1", std::string(100000, 'x'));             // Fill memtable
-    Put("k2", std::string(100000, 'y'));             // Trigger compaction
+    // Block sync calls.
+    env_->delay_data_sync_.store(true, std::memory_order_release);
+    Put("k1", std::string(100000, 'x'));             // Fill memtable.
+    Put("k2", std::string(100000, 'y'));             // Trigger compaction.
     ASSERT_EQ("v1", Get("foo"));
-    env_->delay_data_sync_.Release_Store(NULL);      // Release sync calls
+    // Release sync calls.
+    env_->delay_data_sync_.store(false, std::memory_order_release);
   } while (ChangeOptions());
 }
 
@@ -558,6 +628,17 @@ TEST(DBTest, GetFromVersions) {
   } while (ChangeOptions());
 }
 
+TEST(DBTest, GetMemUsage) {
+  do {
+    ASSERT_OK(Put("foo", "v1"));
+    std::string val;
+    ASSERT_TRUE(db_->GetProperty("leveldb.approximate-memory-usage", &val));
+    int mem_usage = std::stoi(val);
+    ASSERT_GT(mem_usage, 0);
+    ASSERT_LT(mem_usage, 5*1024*1024);
+  } while (ChangeOptions());
+}
+
 TEST(DBTest, GetSnapshot) {
   do {
     // Try with both a short key and a long key
@@ -576,6 +657,55 @@ TEST(DBTest, GetSnapshot) {
   } while (ChangeOptions());
 }
 
+TEST(DBTest, GetIdenticalSnapshots) {
+  do {
+    // Try with both a short key and a long key
+    for (int i = 0; i < 2; i++) {
+      std::string key = (i == 0) ? std::string("foo") : std::string(200, 'x');
+      ASSERT_OK(Put(key, "v1"));
+      const Snapshot* s1 = db_->GetSnapshot();
+      const Snapshot* s2 = db_->GetSnapshot();
+      const Snapshot* s3 = db_->GetSnapshot();
+      ASSERT_OK(Put(key, "v2"));
+      ASSERT_EQ("v2", Get(key));
+      ASSERT_EQ("v1", Get(key, s1));
+      ASSERT_EQ("v1", Get(key, s2));
+      ASSERT_EQ("v1", Get(key, s3));
+      db_->ReleaseSnapshot(s1);
+      dbfull()->TEST_CompactMemTable();
+      ASSERT_EQ("v2", Get(key));
+      ASSERT_EQ("v1", Get(key, s2));
+      db_->ReleaseSnapshot(s2);
+      ASSERT_EQ("v1", Get(key, s3));
+      db_->ReleaseSnapshot(s3);
+    }
+  } while (ChangeOptions());
+}
+
+TEST(DBTest, IterateOverEmptySnapshot) {
+  do {
+    const Snapshot* snapshot = db_->GetSnapshot();
+    ReadOptions read_options;
+    read_options.snapshot = snapshot;
+    ASSERT_OK(Put("foo", "v1"));
+    ASSERT_OK(Put("foo", "v2"));
+
+    Iterator* iterator1 = db_->NewIterator(read_options);
+    iterator1->SeekToFirst();
+    ASSERT_TRUE(!iterator1->Valid());
+    delete iterator1;
+
+    dbfull()->TEST_CompactMemTable();
+
+    Iterator* iterator2 = db_->NewIterator(read_options);
+    iterator2->SeekToFirst();
+    ASSERT_TRUE(!iterator2->Valid());
+    delete iterator2;
+
+    db_->ReleaseSnapshot(snapshot);
+  } while (ChangeOptions());
+}
+
 TEST(DBTest, GetLevel0Ordering) {
   do {
     // Check that we process level-0 files in correct order.  The code
@@ -640,7 +770,7 @@ TEST(DBTest, GetEncountersEmptyLevel) {
     }
 
     // Step 2: clear level 1 if necessary.
-    dbfull()->TEST_CompactRange(1, NULL, NULL);
+    dbfull()->TEST_CompactRange(1, nullptr, nullptr);
     ASSERT_EQ(NumTableFilesAtLevel(0), 1);
     ASSERT_EQ(NumTableFilesAtLevel(1), 0);
     ASSERT_EQ(NumTableFilesAtLevel(2), 1);
@@ -977,7 +1107,7 @@ TEST(DBTest, CompactionsGenerateMultipleFiles) {
 
   // Reopening moves updates to level-0
   Reopen(&options);
-  dbfull()->TEST_CompactRange(0, NULL, NULL);
+  dbfull()->TEST_CompactRange(0, nullptr, nullptr);
 
   ASSERT_EQ(NumTableFilesAtLevel(0), 0);
   ASSERT_GT(NumTableFilesAtLevel(1), 1);
@@ -1001,7 +1131,7 @@ TEST(DBTest, RepeatedWritesToSameKey) {
   for (int i = 0; i < 5 * kMaxFiles; i++) {
     Put("key", value);
     ASSERT_LE(TotalTableFiles(), kMaxFiles);
-    fprintf(stderr, "after %d: %d files\n", int(i+1), TotalTableFiles());
+    fprintf(stderr, "after %d: %d files\n", i + 1, TotalTableFiles());
   }
 }
 
@@ -1028,7 +1158,7 @@ TEST(DBTest, SparseMerge) {
   }
   Put("C", "vc");
   dbfull()->TEST_CompactMemTable();
-  dbfull()->TEST_CompactRange(0, NULL, NULL);
+  dbfull()->TEST_CompactRange(0, nullptr, nullptr);
 
   // Make sparse update
   Put("A",    "va2");
@@ -1039,9 +1169,9 @@ TEST(DBTest, SparseMerge) {
   // Compactions should not cause us to create a situation where
   // a file overlaps too much data at the next level.
   ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576);
-  dbfull()->TEST_CompactRange(0, NULL, NULL);
+  dbfull()->TEST_CompactRange(0, nullptr, nullptr);
   ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576);
-  dbfull()->TEST_CompactRange(1, NULL, NULL);
+  dbfull()->TEST_CompactRange(1, nullptr, nullptr);
   ASSERT_LE(dbfull()->TEST_MaxNextLevelOverlappingBytes(), 20*1048576);
 }
 
@@ -1080,6 +1210,14 @@ TEST(DBTest, ApproximateSizes) {
     // 0 because GetApproximateSizes() does not account for memtable space
     ASSERT_TRUE(Between(Size("", Key(50)), 0, 0));
 
+    if (options.reuse_logs) {
+      // Recovery will reuse memtable, and GetApproximateSizes() does not
+      // account for memtable usage;
+      Reopen(&options);
+      ASSERT_TRUE(Between(Size("", Key(50)), 0, 0));
+      continue;
+    }
+
     // Check sizes across recovery by reopening a few times
     for (int run = 0; run < 3; run++) {
       Reopen(&options);
@@ -1123,6 +1261,11 @@ TEST(DBTest, ApproximateSizes_MixOfSmallAndLarge) {
     ASSERT_OK(Put(Key(6), RandomString(&rnd, 300000)));
     ASSERT_OK(Put(Key(7), RandomString(&rnd, 10000)));
 
+    if (options.reuse_logs) {
+      // Need to force a memtable compaction since recovery does not do so.
+      ASSERT_OK(dbfull()->TEST_CompactMemTable());
+    }
+
     // Check sizes across recovery by reopening a few times
     for (int run = 0; run < 3; run++) {
       Reopen(&options);
@@ -1139,7 +1282,7 @@ TEST(DBTest, ApproximateSizes_MixOfSmallAndLarge) {
 
       ASSERT_TRUE(Between(Size(Key(3), Key(5)), 110000, 111000));
 
-      dbfull()->TEST_CompactRange(0, NULL, NULL);
+      dbfull()->TEST_CompactRange(0, nullptr, nullptr);
     }
   } while (ChangeOptions());
 }
@@ -1153,7 +1296,7 @@ TEST(DBTest, IteratorPinsRef) {
   // Write to force compactions
   Put("foo", "newvalue1");
   for (int i = 0; i < 100; i++) {
-    ASSERT_OK(Put(Key(i), Key(i) + std::string(100000, 'v'))); // 100K values
+    ASSERT_OK(Put(Key(i), Key(i) + std::string(100000, 'v')));  // 100K values
   }
   Put("foo", "newvalue2");
 
@@ -1215,11 +1358,11 @@ TEST(DBTest, HiddenValuesAreRemoved) {
     db_->ReleaseSnapshot(snapshot);
     ASSERT_EQ(AllEntriesFor("foo"), "[ tiny, " + big + " ]");
     Slice x("x");
-    dbfull()->TEST_CompactRange(0, NULL, &x);
+    dbfull()->TEST_CompactRange(0, nullptr, &x);
     ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]");
     ASSERT_EQ(NumTableFilesAtLevel(0), 0);
     ASSERT_GE(NumTableFilesAtLevel(1), 1);
-    dbfull()->TEST_CompactRange(1, NULL, &x);
+    dbfull()->TEST_CompactRange(1, nullptr, &x);
     ASSERT_EQ(AllEntriesFor("foo"), "[ tiny ]");
 
     ASSERT_TRUE(Between(Size("", "pastfoo"), 0, 1000));
@@ -1245,11 +1388,11 @@ TEST(DBTest, DeletionMarkers1) {
   ASSERT_OK(dbfull()->TEST_CompactMemTable());  // Moves to level last-2
   ASSERT_EQ(AllEntriesFor("foo"), "[ v2, DEL, v1 ]");
   Slice z("z");
-  dbfull()->TEST_CompactRange(last-2, NULL, &z);
+  dbfull()->TEST_CompactRange(last-2, nullptr, &z);
   // DEL eliminated, but v1 remains because we aren't compacting that level
   // (DEL can be eliminated because v2 hides v1).
   ASSERT_EQ(AllEntriesFor("foo"), "[ v2, v1 ]");
-  dbfull()->TEST_CompactRange(last-1, NULL, NULL);
+  dbfull()->TEST_CompactRange(last-1, nullptr, nullptr);
   // Merging last-1 w/ last, so we are the base level for "foo", so
   // DEL is removed.  (as is v1).
   ASSERT_EQ(AllEntriesFor("foo"), "[ v2 ]");
@@ -1272,10 +1415,10 @@ TEST(DBTest, DeletionMarkers2) {
   ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]");
   ASSERT_OK(dbfull()->TEST_CompactMemTable());  // Moves to level last-2
   ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]");
-  dbfull()->TEST_CompactRange(last-2, NULL, NULL);
+  dbfull()->TEST_CompactRange(last-2, nullptr, nullptr);
   // DEL kept: "last" file overlaps
   ASSERT_EQ(AllEntriesFor("foo"), "[ DEL, v1 ]");
-  dbfull()->TEST_CompactRange(last-1, NULL, NULL);
+  dbfull()->TEST_CompactRange(last-1, nullptr, nullptr);
   // Merging last-1 w/ last, so we are the base level for "foo", so
   // DEL is removed.  (as is v1).
   ASSERT_EQ(AllEntriesFor("foo"), "[ ]");
@@ -1308,8 +1451,8 @@ TEST(DBTest, OverlapInLevel0) {
     ASSERT_EQ("2,1,1", FilesPerLevel());
 
     // Compact away the placeholder files we created initially
-    dbfull()->TEST_CompactRange(1, NULL, NULL);
-    dbfull()->TEST_CompactRange(2, NULL, NULL);
+    dbfull()->TEST_CompactRange(1, nullptr, nullptr);
+    dbfull()->TEST_CompactRange(2, nullptr, nullptr);
     ASSERT_EQ("2", FilesPerLevel());
 
     // Do a memtable compaction.  Before bug-fix, the compaction would
@@ -1341,21 +1484,21 @@ TEST(DBTest, L0_CompactionBug_Issue44_a) {
 
 TEST(DBTest, L0_CompactionBug_Issue44_b) {
   Reopen();
-  Put("","");
+  Put("", "");
   Reopen();
   Delete("e");
-  Put("","");
+  Put("", "");
   Reopen();
   Put("c", "cv");
   Reopen();
-  Put("","");
+  Put("", "");
   Reopen();
-  Put("","");
+  Put("", "");
   DelayMilliseconds(1000);  // Wait for compaction to finish
   Reopen();
-  Put("d","dv");
+  Put("d", "dv");
   Reopen();
-  Put("","");
+  Put("", "");
   Reopen();
   Delete("d");
   Delete("b");
@@ -1365,6 +1508,15 @@ TEST(DBTest, L0_CompactionBug_Issue44_b) {
   ASSERT_EQ("(->)(c->cv)", Contents());
 }
 
+TEST(DBTest, Fflush_Issue474) {
+  static const int kNum = 100000;
+  Random rnd(test::RandomSeed());
+  for (int i = 0; i < kNum; i++) {
+    fflush(nullptr);
+    ASSERT_OK(Put(RandomKey(&rnd), RandomString(&rnd, 100)));
+  }
+}
+
 TEST(DBTest, ComparatorCheck) {
   class NewComparator : public Comparator {
    public:
@@ -1418,7 +1570,7 @@ TEST(DBTest, CustomComparator) {
   Options new_options = CurrentOptions();
   new_options.create_if_missing = true;
   new_options.comparator = &cmp;
-  new_options.filter_policy = NULL;     // Cannot use bloom filters
+  new_options.filter_policy = nullptr;     // Cannot use bloom filters
   new_options.write_buffer_size = 1000;  // Compact more often
   DestroyAndReopen(&new_options);
   ASSERT_OK(Put("[10]", "ten"));
@@ -1473,7 +1625,7 @@ TEST(DBTest, ManualCompaction) {
   // Compact all
   MakeTables(1, "a", "z");
   ASSERT_EQ("0,1,2", FilesPerLevel());
-  db_->CompactRange(NULL, NULL);
+  db_->CompactRange(nullptr, nullptr);
   ASSERT_EQ("0,0,1", FilesPerLevel());
 }
 
@@ -1482,42 +1634,94 @@ TEST(DBTest, DBOpen_Options) {
   DestroyDB(dbname, Options());
 
   // Does not exist, and create_if_missing == false: error
-  DB* db = NULL;
+  DB* db = nullptr;
   Options opts;
   opts.create_if_missing = false;
   Status s = DB::Open(opts, dbname, &db);
-  ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != NULL);
-  ASSERT_TRUE(db == NULL);
+  ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != nullptr);
+  ASSERT_TRUE(db == nullptr);
 
   // Does not exist, and create_if_missing == true: OK
   opts.create_if_missing = true;
   s = DB::Open(opts, dbname, &db);
   ASSERT_OK(s);
-  ASSERT_TRUE(db != NULL);
+  ASSERT_TRUE(db != nullptr);
 
   delete db;
-  db = NULL;
+  db = nullptr;
 
   // Does exist, and error_if_exists == true: error
   opts.create_if_missing = false;
   opts.error_if_exists = true;
   s = DB::Open(opts, dbname, &db);
-  ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != NULL);
-  ASSERT_TRUE(db == NULL);
+  ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != nullptr);
+  ASSERT_TRUE(db == nullptr);
 
   // Does exist, and error_if_exists == false: OK
   opts.create_if_missing = true;
   opts.error_if_exists = false;
   s = DB::Open(opts, dbname, &db);
   ASSERT_OK(s);
-  ASSERT_TRUE(db != NULL);
+  ASSERT_TRUE(db != nullptr);
 
   delete db;
-  db = NULL;
+  db = nullptr;
+}
+
+TEST(DBTest, DestroyEmptyDir) {
+  std::string dbname = test::TmpDir() + "/db_empty_dir";
+  TestEnv env(Env::Default());
+  env.DeleteDir(dbname);
+  ASSERT_TRUE(!env.FileExists(dbname));
+
+  Options opts;
+  opts.env = &env;
+
+  ASSERT_OK(env.CreateDir(dbname));
+  ASSERT_TRUE(env.FileExists(dbname));
+  std::vector<std::string> children;
+  ASSERT_OK(env.GetChildren(dbname, &children));
+  // The stock Env's do not filter out '.' and '..' special files.
+  ASSERT_EQ(2, children.size());
+  ASSERT_OK(DestroyDB(dbname, opts));
+  ASSERT_TRUE(!env.FileExists(dbname));
+
+  // Should also be destroyed if Env is filtering out dot files.
+  env.SetIgnoreDotFiles(true);
+  ASSERT_OK(env.CreateDir(dbname));
+  ASSERT_TRUE(env.FileExists(dbname));
+  ASSERT_OK(env.GetChildren(dbname, &children));
+  ASSERT_EQ(0, children.size());
+  ASSERT_OK(DestroyDB(dbname, opts));
+  ASSERT_TRUE(!env.FileExists(dbname));
+}
+
+TEST(DBTest, DestroyOpenDB) {
+  std::string dbname = test::TmpDir() + "/open_db_dir";
+  env_->DeleteDir(dbname);
+  ASSERT_TRUE(!env_->FileExists(dbname));
+
+  Options opts;
+  opts.create_if_missing = true;
+  DB* db = nullptr;
+  ASSERT_OK(DB::Open(opts, dbname, &db));
+  ASSERT_TRUE(db != nullptr);
+
+  // Must fail to destroy an open db.
+  ASSERT_TRUE(env_->FileExists(dbname));
+  ASSERT_TRUE(!DestroyDB(dbname, Options()).ok());
+  ASSERT_TRUE(env_->FileExists(dbname));
+
+  delete db;
+  db = nullptr;
+
+  // Should succeed destroying a closed db.
+  ASSERT_OK(DestroyDB(dbname, Options()));
+  ASSERT_TRUE(!env_->FileExists(dbname));
 }
 
 TEST(DBTest, Locking) {
-  DB* db2 = NULL;
+  DB* db2 = nullptr;
   Status s = DB::Open(CurrentOptions(), dbname_, &db2);
   ASSERT_TRUE(!s.ok()) << "Locking did not prevent re-opening db";
 }
@@ -1532,13 +1736,14 @@ TEST(DBTest, NoSpace) {
   ASSERT_EQ("v1", Get("foo"));
   Compact("a", "z");
   const int num_files = CountFiles();
-  env_->no_space_.Release_Store(env_);   // Force out-of-space errors
+  // Force out-of-space errors.
+  env_->no_space_.store(true, std::memory_order_release);
   for (int i = 0; i < 10; i++) {
     for (int level = 0; level < config::kNumLevels-1; level++) {
-      dbfull()->TEST_CompactRange(level, NULL, NULL);
+      dbfull()->TEST_CompactRange(level, nullptr, nullptr);
     }
   }
-  env_->no_space_.Release_Store(NULL);
+  env_->no_space_.store(false, std::memory_order_release);
   ASSERT_LT(CountFiles(), num_files + 3);
 }
 
@@ -1548,7 +1753,8 @@ TEST(DBTest, NonWritableFileSystem) {
   options.env = env_;
   Reopen(&options);
   ASSERT_OK(Put("foo", "v1"));
-  env_->non_writable_.Release_Store(env_);  // Force errors for new files
+  // Force errors for new files.
+  env_->non_writable_.store(true, std::memory_order_release);
   std::string big(100000, 'x');
   int errors = 0;
   for (int i = 0; i < 20; i++) {
@@ -1559,7 +1765,7 @@ TEST(DBTest, NonWritableFileSystem) {
     }
   }
   ASSERT_GT(errors, 0);
-  env_->non_writable_.Release_Store(NULL);
+  env_->non_writable_.store(false, std::memory_order_release);
 }
 
 TEST(DBTest, WriteSyncError) {
@@ -1569,7 +1775,7 @@ TEST(DBTest, WriteSyncError) {
   Options options = CurrentOptions();
   options.env = env_;
   Reopen(&options);
-  env_->data_sync_error_.Release_Store(env_);
+  env_->data_sync_error_.store(true, std::memory_order_release);
 
   // (b) Normal write should succeed
   WriteOptions w;
@@ -1583,7 +1789,7 @@ TEST(DBTest, WriteSyncError) {
   ASSERT_EQ("NOT_FOUND", Get("k2"));
 
   // (d) make sync behave normally
-  env_->data_sync_error_.Release_Store(NULL);
+  env_->data_sync_error_.store(false, std::memory_order_release);
 
   // (e) Do a non-sync write; should fail
   w.sync = false;
@@ -1603,7 +1809,7 @@ TEST(DBTest, ManifestWriteError) {
   // We iterate twice.  In the second iteration, everything is the
   // same except the log record never makes it to the MANIFEST file.
   for (int iter = 0; iter < 2; iter++) {
-    port::AtomicPointer* error_type = (iter == 0)
+    std::atomic<bool>* error_type = (iter == 0)
         ? &env_->manifest_sync_error_
         : &env_->manifest_write_error_;
 
@@ -1623,12 +1829,12 @@ TEST(DBTest, ManifestWriteError) {
     ASSERT_EQ(NumTableFilesAtLevel(last), 1);   // foo=>bar is now in last level
 
     // Merging compaction (will fail)
-    error_type->Release_Store(env_);
-    dbfull()->TEST_CompactRange(last, NULL, NULL);  // Should fail
+    error_type->store(true, std::memory_order_release);
+    dbfull()->TEST_CompactRange(last, nullptr, nullptr);  // Should fail
     ASSERT_EQ("bar", Get("foo"));
 
     // Recovery: should not lose data
-    error_type->Release_Store(NULL);
+    error_type->store(false, std::memory_order_release);
     Reopen(&options);
     ASSERT_EQ("bar", Get("foo"));
   }
@@ -1699,7 +1905,7 @@ TEST(DBTest, BloomFilter) {
   dbfull()->TEST_CompactMemTable();
 
   // Prevent auto compactions triggered by seeks
-  env_->delay_data_sync_.Release_Store(env_);
+  env_->delay_data_sync_.store(true, std::memory_order_release);
 
   // Lookup present keys.  Should rarely read from small sstable.
   env_->random_read_counter_.Reset();
@@ -1720,7 +1926,7 @@ TEST(DBTest, BloomFilter) {
   fprintf(stderr, "%d missing => %d reads\n", N, reads);
   ASSERT_LE(reads, 3*N/100);
 
-  env_->delay_data_sync_.Release_Store(NULL);
+  env_->delay_data_sync_.store(false, std::memory_order_release);
   Close();
   delete options.block_cache;
   delete options.filter_policy;
@@ -1735,9 +1941,9 @@ static const int kNumKeys = 1000;
 
 struct MTState {
   DBTest* test;
-  port::AtomicPointer stop;
-  port::AtomicPointer counter[kNumThreads];
-  port::AtomicPointer thread_done[kNumThreads];
+  std::atomic<bool> stop;
+  std::atomic<int> counter[kNumThreads];
+  std::atomic<bool> thread_done[kNumThreads];
 };
 
 struct MTThread {
@@ -1749,13 +1955,13 @@ static void MTThreadBody(void* arg) {
   MTThread* t = reinterpret_cast<MTThread*>(arg);
   int id = t->id;
   DB* db = t->state->test->db_;
-  uintptr_t counter = 0;
+  int counter = 0;
   fprintf(stderr, "... starting thread %d\n", id);
   Random rnd(1000 + id);
   std::string value;
   char valbuf[1500];
-  while (t->state->stop.Acquire_Load() == NULL) {
-    t->state->counter[id].Release_Store(reinterpret_cast<void*>(counter));
+  while (!t->state->stop.load(std::memory_order_acquire)) {
+    t->state->counter[id].store(counter, std::memory_order_release);
 
     int key = rnd.Uniform(kNumKeys);
     char keybuf[20];
@@ -1780,14 +1986,13 @@ static void MTThreadBody(void* arg) {
         ASSERT_EQ(k, key);
         ASSERT_GE(w, 0);
         ASSERT_LT(w, kNumThreads);
-        ASSERT_LE(static_cast<uintptr_t>(c), reinterpret_cast<uintptr_t>(
-            t->state->counter[w].Acquire_Load()));
+        ASSERT_LE(c, t->state->counter[w].load(std::memory_order_acquire));
       }
     }
     counter++;
   }
-  t->state->thread_done[id].Release_Store(t);
-  fprintf(stderr, "... stopping thread %d after %d ops\n", id, int(counter));
+  t->state->thread_done[id].store(true, std::memory_order_release);
+  fprintf(stderr, "... stopping thread %d after %d ops\n", id, counter);
 }
 
 }  // namespace
@@ -1797,10 +2002,10 @@ TEST(DBTest, MultiThreaded) {
     // Initialize state
     MTState mt;
     mt.test = this;
-    mt.stop.Release_Store(0);
+    mt.stop.store(false, std::memory_order_release);
     for (int id = 0; id < kNumThreads; id++) {
-      mt.counter[id].Release_Store(0);
-      mt.thread_done[id].Release_Store(0);
+      mt.counter[id].store(false, std::memory_order_release);
+      mt.thread_done[id].store(false, std::memory_order_release);
     }
 
     // Start threads
@@ -1815,9 +2020,9 @@ TEST(DBTest, MultiThreaded) {
     DelayMilliseconds(kTestSeconds * 1000);
 
     // Stop the threads and wait for them to finish
-    mt.stop.Release_Store(&mt);
+    mt.stop.store(true, std::memory_order_release);
     for (int id = 0; id < kNumThreads; id++) {
-      while (mt.thread_done[id].Acquire_Load() == NULL) {
+      while (!mt.thread_done[id].load(std::memory_order_acquire)) {
         DelayMilliseconds(100);
       }
     }
@@ -1849,7 +2054,7 @@ class ModelDB: public DB {
     return Status::NotFound(key);
   }
   virtual Iterator* NewIterator(const ReadOptions& options) {
-    if (options.snapshot == NULL) {
+    if (options.snapshot == nullptr) {
       KVMap* saved = new KVMap;
       *saved = map_;
       return new ModelIter(saved, true);
@@ -1921,6 +2126,7 @@ class ModelDB: public DB {
     virtual Slice key() const { return iter_->first; }
     virtual Slice value() const { return iter_->second; }
     virtual Status status() const { return Status::OK(); }
+
    private:
     const KVMap* const map_;
     const bool owned_;  // Do we own map_
@@ -1930,13 +2136,6 @@ class ModelDB: public DB {
   KVMap map_;
 };
 
-static std::string RandomKey(Random* rnd) {
-  int len = (rnd->OneIn(3)
-             ? 1                // Short sometimes to encourage collisions
-             : (rnd->OneIn(100) ? rnd->Skewed(10) : rnd->Uniform(10)));
-  return test::RandomKey(rnd, len);
-}
-
 static bool CompareIterators(int step,
                              DB* model,
                              DB* db,
@@ -1990,8 +2189,8 @@ TEST(DBTest, Randomized) {
   do {
     ModelDB model(CurrentOptions());
     const int N = 10000;
-    const Snapshot* model_snap = NULL;
-    const Snapshot* db_snap = NULL;
+    const Snapshot* model_snap = nullptr;
+    const Snapshot* db_snap = nullptr;
     std::string k, v;
     for (int step = 0; step < N; step++) {
       if (step % 100 == 0) {
@@ -2036,23 +2235,23 @@ TEST(DBTest, Randomized) {
       }
 
       if ((step % 100) == 0) {
-        ASSERT_TRUE(CompareIterators(step, &model, db_, NULL, NULL));
+        ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr));
         ASSERT_TRUE(CompareIterators(step, &model, db_, model_snap, db_snap));
         // Save a snapshot from each DB this time that we'll use next
         // time we compare things, to make sure the current state is
         // preserved with the snapshot
-        if (model_snap != NULL) model.ReleaseSnapshot(model_snap);
-        if (db_snap != NULL) db_->ReleaseSnapshot(db_snap);
+        if (model_snap != nullptr) model.ReleaseSnapshot(model_snap);
+        if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap);
 
         Reopen();
-        ASSERT_TRUE(CompareIterators(step, &model, db_, NULL, NULL));
+        ASSERT_TRUE(CompareIterators(step, &model, db_, nullptr, nullptr));
 
         model_snap = model.GetSnapshot();
         db_snap = db_->GetSnapshot();
       }
     }
-    if (model_snap != NULL) model.ReleaseSnapshot(model_snap);
-    if (db_snap != NULL) db_->ReleaseSnapshot(db_snap);
+    if (model_snap != nullptr) model.ReleaseSnapshot(model_snap);
+    if (db_snap != nullptr) db_->ReleaseSnapshot(db_snap);
   } while (ChangeOptions());
 }
 
@@ -2066,15 +2265,15 @@ void BM_LogAndApply(int iters, int num_base_files) {
   std::string dbname = test::TmpDir() + "/leveldb_test_benchmark";
   DestroyDB(dbname, Options());
 
-  DB* db = NULL;
+  DB* db = nullptr;
   Options opts;
   opts.create_if_missing = true;
   Status s = DB::Open(opts, dbname, &db);
   ASSERT_OK(s);
-  ASSERT_TRUE(db != NULL);
+  ASSERT_TRUE(db != nullptr);
 
   delete db;
-  db = NULL;
+  db = nullptr;
 
   Env* env = Env::Default();
 
@@ -2083,8 +2282,9 @@ void BM_LogAndApply(int iters, int num_base_files) {
 
   InternalKeyComparator cmp(BytewiseComparator());
   Options options;
-  VersionSet vset(dbname, &options, NULL, &cmp);
-  ASSERT_OK(vset.Recover());
+  VersionSet vset(dbname, &options, nullptr, &cmp);
+  bool save_manifest;
+  ASSERT_OK(vset.Recover(&save_manifest));
   VersionEdit vbase;
   uint64_t fnum = 1;
   for (int i = 0; i < num_base_files; i++) {

db/dbformat.h

@@ -84,15 +84,13 @@ inline size_t InternalKeyEncodingLength(const ParsedInternalKey& key) {
 }
 
 // Append the serialization of "key" to *result.
-extern void AppendInternalKey(std::string* result,
-                              const ParsedInternalKey& key);
+void AppendInternalKey(std::string* result, const ParsedInternalKey& key);
 
 // Attempt to parse an internal key from "internal_key".  On success,
 // stores the parsed data in "*result", and returns true.
 //
 // On error, returns false, leaves "*result" in an undefined state.
-extern bool ParseInternalKey(const Slice& internal_key,
-                             ParsedInternalKey* result);
+bool ParseInternalKey(const Slice& internal_key, ParsedInternalKey* result);
 
 // Returns the user key portion of an internal key.
 inline Slice ExtractUserKey(const Slice& internal_key) {
@@ -100,14 +98,6 @@ inline Slice ExtractUserKey(const Slice& internal_key) {
   return Slice(internal_key.data(), internal_key.size() - 8);
 }
 
-inline ValueType ExtractValueType(const Slice& internal_key) {
-  assert(internal_key.size() >= 8);
-  const size_t n = internal_key.size();
-  uint64_t num = DecodeFixed64(internal_key.data() + n - 8);
-  unsigned char c = num & 0xff;
-  return static_cast<ValueType>(c);
-}
-
 // A comparator for internal keys that uses a specified comparator for
 // the user key portion and breaks ties by decreasing sequence number.
 class InternalKeyComparator : public Comparator {

db/dumpfile.cc

@@ -2,7 +2,10 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 
+#include "leveldb/dumpfile.h"
+
 #include <stdio.h>
+
 #include "db/dbformat.h"
 #include "db/filename.h"
 #include "db/log_reader.h"
@@ -142,8 +145,8 @@ Status DumpDescriptor(Env* env, const std::string& fname, WritableFile* dst) {
 
 Status DumpTable(Env* env, const std::string& fname, WritableFile* dst) {
   uint64_t file_size;
-  RandomAccessFile* file = NULL;
-  Table* table = NULL;
+  RandomAccessFile* file = nullptr;
+  Table* table = nullptr;
   Status s = env->GetFileSize(fname, &file_size);
   if (s.ok()) {
     s = env->NewRandomAccessFile(fname, &file);

db/fault_injection_test.cc

@@ -0,0 +1,561 @@
+// 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 "leveldb/db.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/env.h"
+#include "leveldb/table.h"
+#include "leveldb/write_batch.h"
+#include "port/port.h"
+#include "port/thread_annotations.h"
+#include "util/logging.h"
+#include "util/mutexlock.h"
+#include "util/testharness.h"
+#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->DeleteFile(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);
+  virtual ~TestWritableFile();
+  virtual Status Append(const Slice& data);
+  virtual Status Close();
+  virtual Status Flush();
+  virtual Status Sync();
+
+ 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) {}
+  virtual ~FaultInjectionTestEnv() { }
+  virtual Status NewWritableFile(const std::string& fname,
+                                 WritableFile** result);
+  virtual Status NewAppendableFile(const std::string& fname,
+                                   WritableFile** result);
+  virtual Status DeleteFile(const std::string& f);
+  virtual Status RenameFile(const std::string& s, const std::string& t);
+
+  void WritableFileClosed(const FileState& state);
+  Status DropUnsyncedFileData();
+  Status DeleteFilesCreatedAfterLastDirSync();
+  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 (std::map<std::string, FileState>::const_iterator it =
+           db_file_state_.begin();
+       s.ok() && it != db_file_state_.end(); ++it) {
+    const FileState& state = it->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::DeleteFile(const std::string& f) {
+  Status s = EnvWrapper::DeleteFile(f);
+  ASSERT_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::DeleteFilesCreatedAfterLastDirSync() {
+  // Because DeleteFile 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 s;
+  std::set<std::string>::const_iterator it;
+  for (it = new_files.begin(); s.ok() && it != new_files.end(); ++it) {
+    s = DeleteFile(*it);
+  }
+  return s;
+}
+
+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:
+  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_ = test::TmpDir() + "/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_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()) {
+          ASSERT_EQ(value_space, val);
+        }
+      } else if (s.ok()) {
+        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];
+    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_OK(db_->Delete(WriteOptions(), iter->key()));
+    }
+
+    delete iter;
+  }
+
+  void ResetDBState(ResetMethod reset_method) {
+    switch (reset_method) {
+      case RESET_DROP_UNSYNCED_DATA:
+        ASSERT_OK(env_->DropUnsyncedFileData());
+        break;
+      case RESET_DELETE_UNSYNCED_FILES:
+        ASSERT_OK(env_->DeleteFilesCreatedAfterLastDirSync());
+        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_OK(OpenDB());
+    ASSERT_OK(Verify(0, num_pre_sync, FaultInjectionTest::VAL_EXPECT_NO_ERROR));
+    ASSERT_OK(Verify(num_pre_sync, num_post_sync, FaultInjectionTest::VAL_EXPECT_ERROR));
+  }
+
+  void NoWriteTestPreFault() {
+  }
+
+  void NoWriteTestReopenWithFault(ResetMethod reset_method) {
+    CloseDB();
+    ResetDBState(reset_method);
+    ASSERT_OK(OpenDB());
+  }
+
+  void DoTest() {
+    Random rnd(0);
+    ASSERT_OK(OpenDB());
+    for (size_t idx = 0; idx < kNumIterations; idx++) {
+      int num_pre_sync = rnd.Uniform(kMaxNumValues);
+      int num_post_sync = rnd.Uniform(kMaxNumValues);
+
+      PartialCompactTestPreFault(num_pre_sync, num_post_sync);
+      PartialCompactTestReopenWithFault(RESET_DROP_UNSYNCED_DATA,
+                                        num_pre_sync,
+                                        num_post_sync);
+
+      NoWriteTestPreFault();
+      NoWriteTestReopenWithFault(RESET_DROP_UNSYNCED_DATA);
+
+      PartialCompactTestPreFault(num_pre_sync, num_post_sync);
+      // No new files created so we expect all values since no files will be
+      // dropped.
+      PartialCompactTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES,
+                                        num_pre_sync + num_post_sync,
+                                        0);
+
+      NoWriteTestPreFault();
+      NoWriteTestReopenWithFault(RESET_DELETE_UNSYNCED_FILES);
+    }
+  }
+};
+
+TEST(FaultInjectionTest, FaultTestNoLogReuse) {
+  ReuseLogs(false);
+  DoTest();
+}
+
+TEST(FaultInjectionTest, FaultTestWithLogReuse) {
+  ReuseLogs(true);
+  DoTest();
+}
+
+}  // namespace leveldb
+
+int main(int argc, char** argv) {
+  return leveldb::test::RunAllTests();
+}

db/filename.cc

@@ -12,31 +12,31 @@
 namespace leveldb {
 
 // A utility routine: write "data" to the named file and Sync() it.
-extern Status WriteStringToFileSync(Env* env, const Slice& data,
-                                    const std::string& fname);
+Status WriteStringToFileSync(Env* env, const Slice& data,
+                             const std::string& fname);
 
-static std::string MakeFileName(const std::string& name, uint64_t number,
+static std::string MakeFileName(const std::string& dbname, uint64_t number,
                                 const char* suffix) {
   char buf[100];
   snprintf(buf, sizeof(buf), "/%06llu.%s",
            static_cast<unsigned long long>(number),
            suffix);
-  return name + buf;
+  return dbname + buf;
 }
 
-std::string LogFileName(const std::string& name, uint64_t number) {
+std::string LogFileName(const std::string& dbname, uint64_t number) {
   assert(number > 0);
-  return MakeFileName(name, number, "log");
+  return MakeFileName(dbname, number, "log");
 }
 
-std::string TableFileName(const std::string& name, uint64_t number) {
+std::string TableFileName(const std::string& dbname, uint64_t number) {
   assert(number > 0);
-  return MakeFileName(name, number, "ldb");
+  return MakeFileName(dbname, number, "ldb");
 }
 
-std::string SSTTableFileName(const std::string& name, uint64_t number) {
+std::string SSTTableFileName(const std::string& dbname, uint64_t number) {
   assert(number > 0);
-  return MakeFileName(name, number, "sst");
+  return MakeFileName(dbname, number, "sst");
 }
 
 std::string DescriptorFileName(const std::string& dbname, uint64_t number) {
@@ -77,10 +77,10 @@ std::string OldInfoLogFileName(const std::string& dbname) {
 //    dbname/LOG.old
 //    dbname/MANIFEST-[0-9]+
 //    dbname/[0-9]+.(log|sst|ldb)
-bool ParseFileName(const std::string& fname,
+bool ParseFileName(const std::string& filename,
                    uint64_t* number,
                    FileType* type) {
-  Slice rest(fname);
+  Slice rest(filename);
   if (rest == "CURRENT") {
     *number = 0;
     *type = kCurrentFile;

db/filename.h

@@ -30,55 +30,53 @@ enum FileType {
 // Return the name of the log file with the specified number
 // in the db named by "dbname".  The result will be prefixed with
 // "dbname".
-extern std::string LogFileName(const std::string& dbname, uint64_t number);
+std::string LogFileName(const std::string& dbname, uint64_t number);
 
 // Return the name of the sstable with the specified number
 // in the db named by "dbname".  The result will be prefixed with
 // "dbname".
-extern std::string TableFileName(const std::string& dbname, uint64_t number);
+std::string TableFileName(const std::string& dbname, uint64_t number);
 
 // Return the legacy file name for an sstable with the specified number
 // in the db named by "dbname". The result will be prefixed with
 // "dbname".
-extern std::string SSTTableFileName(const std::string& dbname, uint64_t number);
+std::string SSTTableFileName(const std::string& dbname, uint64_t number);
 
 // Return the name of the descriptor file for the db named by
 // "dbname" and the specified incarnation number.  The result will be
 // prefixed with "dbname".
-extern std::string DescriptorFileName(const std::string& dbname,
-                                      uint64_t number);
+std::string DescriptorFileName(const std::string& dbname, uint64_t number);
 
 // Return the name of the current file.  This file contains the name
 // of the current manifest file.  The result will be prefixed with
 // "dbname".
-extern std::string CurrentFileName(const std::string& dbname);
+std::string CurrentFileName(const std::string& dbname);
 
 // Return the name of the lock file for the db named by
 // "dbname".  The result will be prefixed with "dbname".
-extern std::string LockFileName(const std::string& dbname);
+std::string LockFileName(const std::string& dbname);
 
 // Return the name of a temporary file owned by the db named "dbname".
 // The result will be prefixed with "dbname".
-extern std::string TempFileName(const std::string& dbname, uint64_t number);
+std::string TempFileName(const std::string& dbname, uint64_t number);
 
 // Return the name of the info log file for "dbname".
-extern std::string InfoLogFileName(const std::string& dbname);
+std::string InfoLogFileName(const std::string& dbname);
 
 // Return the name of the old info log file for "dbname".
-extern std::string OldInfoLogFileName(const std::string& dbname);
+std::string OldInfoLogFileName(const std::string& dbname);
 
 // If filename is a leveldb file, store the type of the file in *type.
 // The number encoded in the filename is stored in *number.  If the
 // filename was successfully parsed, returns true.  Else return false.
-extern bool ParseFileName(const std::string& filename,
-                          uint64_t* number,
-                          FileType* type);
+bool ParseFileName(const std::string& filename,
+                   uint64_t* number,
+                   FileType* type);
 
 // Make the CURRENT file point to the descriptor file with the
 // specified number.
-extern Status SetCurrentFile(Env* env, const std::string& dbname,
-                             uint64_t descriptor_number);
-
+Status SetCurrentFile(Env* env, const std::string& dbname,
+                      uint64_t descriptor_number);
 
 }  // namespace leveldb
 

db/filename_test.cc

@@ -114,6 +114,18 @@ TEST(FileNameTest, Construction) {
   ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type));
   ASSERT_EQ(999, number);
   ASSERT_EQ(kTempFile, type);
+
+  fname = InfoLogFileName("foo");
+  ASSERT_EQ("foo/", std::string(fname.data(), 4));
+  ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type));
+  ASSERT_EQ(0, number);
+  ASSERT_EQ(kInfoLogFile, type);
+
+  fname = OldInfoLogFileName("foo");
+  ASSERT_EQ("foo/", std::string(fname.data(), 4));
+  ASSERT_TRUE(ParseFileName(fname.c_str() + 4, &number, &type));
+  ASSERT_EQ(0, number);
+  ASSERT_EQ(kInfoLogFile, type);
 }
 
 }  // namespace leveldb

db/log_format.h

@@ -3,7 +3,7 @@
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 //
 // Log format information shared by reader and writer.
-// See ../doc/log_format.txt for more detail.
+// See ../doc/log_format.md for more detail.
 
 #ifndef STORAGE_LEVELDB_DB_LOG_FORMAT_H_
 #define STORAGE_LEVELDB_DB_LOG_FORMAT_H_

db/log_reader.cc

@@ -25,7 +25,8 @@ Reader::Reader(SequentialFile* file, Reporter* reporter, bool checksum,
       eof_(false),
       last_record_offset_(0),
       end_of_buffer_offset_(0),
-      initial_offset_(initial_offset) {
+      initial_offset_(initial_offset),
+      resyncing_(initial_offset > 0) {
 }
 
 Reader::~Reader() {
@@ -33,12 +34,11 @@ Reader::~Reader() {
 }
 
 bool Reader::SkipToInitialBlock() {
-  size_t offset_in_block = initial_offset_ % kBlockSize;
+  const size_t offset_in_block = initial_offset_ % kBlockSize;
   uint64_t block_start_location = initial_offset_ - offset_in_block;
 
   // Don't search a block if we'd be in the trailer
   if (offset_in_block > kBlockSize - 6) {
-    offset_in_block = 0;
     block_start_location += kBlockSize;
   }
 
@@ -72,8 +72,25 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) {
 
   Slice fragment;
   while (true) {
-    uint64_t physical_record_offset = end_of_buffer_offset_ - buffer_.size();
     const unsigned int record_type = ReadPhysicalRecord(&fragment);
+
+    // ReadPhysicalRecord may have only had an empty trailer remaining in its
+    // internal buffer. Calculate the offset of the next physical record now
+    // that it has returned, properly accounting for its header size.
+    uint64_t physical_record_offset =
+        end_of_buffer_offset_ - buffer_.size() - kHeaderSize - fragment.size();
+
+    if (resyncing_) {
+      if (record_type == kMiddleType) {
+        continue;
+      } else if (record_type == kLastType) {
+        resyncing_ = false;
+        continue;
+      } else {
+        resyncing_ = false;
+      }
+    }
+
     switch (record_type) {
       case kFullType:
         if (in_fragmented_record) {
@@ -81,9 +98,7 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) {
           // it could emit an empty kFirstType record at the tail end
           // of a block followed by a kFullType or kFirstType record
           // at the beginning of the next block.
-          if (scratch->empty()) {
-            in_fragmented_record = false;
-          } else {
+          if (!scratch->empty()) {
             ReportCorruption(scratch->size(), "partial record without end(1)");
           }
         }
@@ -99,9 +114,7 @@ bool Reader::ReadRecord(Slice* record, std::string* scratch) {
           // it could emit an empty kFirstType record at the tail end
           // of a block followed by a kFullType or kFirstType record
           // at the beginning of the next block.
-          if (scratch->empty()) {
-            in_fragmented_record = false;
-          } else {
+          if (!scratch->empty()) {
             ReportCorruption(scratch->size(), "partial record without end(2)");
           }
         }
@@ -172,7 +185,7 @@ void Reader::ReportCorruption(uint64_t bytes, const char* reason) {
 }
 
 void Reader::ReportDrop(uint64_t bytes, const Status& reason) {
-  if (reporter_ != NULL &&
+  if (reporter_ != nullptr &&
       end_of_buffer_offset_ - buffer_.size() - bytes >= initial_offset_) {
     reporter_->Corruption(static_cast<size_t>(bytes), reason);
   }

db/log_reader.h

@@ -32,7 +32,7 @@ class Reader {
   // Create a reader that will return log records from "*file".
   // "*file" must remain live while this Reader is in use.
   //
-  // If "reporter" is non-NULL, it is notified whenever some data is
+  // If "reporter" is non-null, it is notified whenever some data is
   // dropped due to a detected corruption.  "*reporter" must remain
   // live while this Reader is in use.
   //
@@ -73,6 +73,11 @@ class Reader {
   // 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_;
+
   // Extend record types with the following special values
   enum {
     kEof = kMaxRecordType + 1,

db/log_test.cc

@@ -79,7 +79,7 @@ class LogTest {
     virtual Status Skip(uint64_t n) {
       if (n > contents_.size()) {
         contents_.clear();
-        return Status::NotFound("in-memory file skipepd past end");
+        return Status::NotFound("in-memory file skipped past end");
       }
 
       contents_.remove_prefix(n);
@@ -104,23 +104,34 @@ class LogTest {
   StringSource source_;
   ReportCollector report_;
   bool reading_;
-  Writer writer_;
-  Reader reader_;
+  Writer* writer_;
+  Reader* reader_;
 
   // Record metadata for testing initial offset functionality
   static size_t initial_offset_record_sizes_[];
   static uint64_t initial_offset_last_record_offsets_[];
+  static int num_initial_offset_records_;
 
  public:
   LogTest() : reading_(false),
-              writer_(&dest_),
-              reader_(&source_, &report_, true/*checksum*/,
-                      0/*initial_offset*/) {
+              writer_(new Writer(&dest_)),
+              reader_(new Reader(&source_, &report_, true/*checksum*/,
+                      0/*initial_offset*/)) {
+  }
+
+  ~LogTest() {
+    delete writer_;
+    delete reader_;
+  }
+
+  void ReopenForAppend() {
+    delete writer_;
+    writer_ = new Writer(&dest_, dest_.contents_.size());
   }
 
   void Write(const std::string& msg) {
     ASSERT_TRUE(!reading_) << "Write() after starting to read";
-    writer_.AddRecord(Slice(msg));
+    writer_->AddRecord(Slice(msg));
   }
 
   size_t WrittenBytes() const {
@@ -134,7 +145,7 @@ class LogTest {
     }
     std::string scratch;
     Slice record;
-    if (reader_.ReadRecord(&record, &scratch)) {
+    if (reader_->ReadRecord(&record, &scratch)) {
       return record.ToString();
     } else {
       return "EOF";
@@ -182,13 +193,18 @@ class LogTest {
   }
 
   void WriteInitialOffsetLog() {
-    for (int i = 0; i < 4; i++) {
+    for (int i = 0; i < num_initial_offset_records_; i++) {
       std::string record(initial_offset_record_sizes_[i],
                          static_cast<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;
@@ -208,32 +224,48 @@ class LogTest {
     source_.contents_ = Slice(dest_.contents_);
     Reader* offset_reader = new Reader(&source_, &report_, true/*checksum*/,
                                        initial_offset);
-    Slice record;
-    std::string scratch;
-    ASSERT_TRUE(offset_reader->ReadRecord(&record, &scratch));
-    ASSERT_EQ(initial_offset_record_sizes_[expected_record_offset],
-              record.size());
-    ASSERT_EQ(initial_offset_last_record_offsets_[expected_record_offset],
-              offset_reader->LastRecordOffset());
-    ASSERT_EQ((char)('a' + expected_record_offset), record.data()[0]);
+
+    // 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;
   }
-
 };
 
 size_t LogTest::initial_offset_record_sizes_[] =
     {10000,  // Two sizable records in first block
      10000,
      2 * log::kBlockSize - 1000,  // Span three blocks
-     1};
+     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 * log::kBlockSize - 1000) + 3 * kHeaderSize,
+     2 * (kHeaderSize + 10000) +
+         (2 * log::kBlockSize - 1000) + 3 * kHeaderSize
+         + kHeaderSize + 1,
+     3 * log::kBlockSize,
+    };
 
+// LogTest::initial_offset_last_record_offsets_ must be defined before this.
+int LogTest::num_initial_offset_records_ =
+    sizeof(LogTest::initial_offset_last_record_offsets_)/sizeof(uint64_t);
 
 TEST(LogTest, Empty) {
   ASSERT_EQ("EOF", Read());
@@ -318,6 +350,15 @@ TEST(LogTest, AlignedEof) {
   ASSERT_EQ("EOF", Read());
 }
 
+TEST(LogTest, OpenForAppend) {
+  Write("hello");
+  ReopenForAppend();
+  Write("world");
+  ASSERT_EQ("hello", Read());
+  ASSERT_EQ("world", Read());
+  ASSERT_EQ("EOF", Read());
+}
+
 TEST(LogTest, RandomRead) {
   const int N = 500;
   Random write_rnd(301);
@@ -445,6 +486,22 @@ TEST(LogTest, PartialLastIsIgnored) {
   ASSERT_EQ(0, DroppedBytes());
 }
 
+TEST(LogTest, SkipIntoMultiRecord) {
+  // Consider a fragmented record:
+  //    first(R1), middle(R1), last(R1), first(R2)
+  // If initial_offset points to a record after first(R1) but before first(R2)
+  // incomplete fragment errors are not actual errors, and must be suppressed
+  // until a new first or full record is encountered.
+  Write(BigString("foo", 3*kBlockSize));
+  Write("correct");
+  StartReadingAt(kBlockSize);
+
+  ASSERT_EQ("correct", Read());
+  ASSERT_EQ("", ReportMessage());
+  ASSERT_EQ(0, DroppedBytes());
+  ASSERT_EQ("EOF", Read());
+}
+
 TEST(LogTest, ErrorJoinsRecords) {
   // Consider two fragmented records:
   //    first(R1) last(R1) first(R2) last(R2)
@@ -514,6 +571,10 @@ TEST(LogTest, ReadFourthStart) {
       3);
 }
 
+TEST(LogTest, ReadInitialOffsetIntoBlockPadding) {
+  CheckInitialOffsetRecord(3 * log::kBlockSize - 3, 5);
+}
+
 TEST(LogTest, ReadEnd) {
   CheckOffsetPastEndReturnsNoRecords(0);
 }

db/log_writer.cc

@@ -12,13 +12,22 @@
 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) {
-  for (int i = 0; i <= kMaxRecordType; i++) {
-    char t = static_cast<char>(i);
-    type_crc_[i] = crc32c::Value(&t, 1);
-  }
+  InitTypeCrc(type_crc_);
+}
+
+Writer::Writer(WritableFile* dest, uint64_t dest_length)
+    : dest_(dest), block_offset_(dest_length % kBlockSize) {
+  InitTypeCrc(type_crc_);
 }
 
 Writer::~Writer() {

db/log_writer.h

@@ -22,6 +22,12 @@ class Writer {
   // "*dest" must be initially empty.
   // "*dest" must remain live while this Writer is in use.
   explicit Writer(WritableFile* dest);
+
+  // Create a writer that will append data to "*dest".
+  // "*dest" must have initial length "dest_length".
+  // "*dest" must remain live while this Writer is in use.
+  Writer(WritableFile* dest, uint64_t dest_length);
+
   ~Writer();
 
   Status AddRecord(const Slice& slice);

db/memtable.cc

@@ -101,7 +101,7 @@ void MemTable::Add(SequenceNumber s, ValueType type,
   p += 8;
   p = EncodeVarint32(p, val_size);
   memcpy(p, value.data(), val_size);
-  assert((p + val_size) - buf == encoded_len);
+  assert(p + val_size == buf + encoded_len);
   table_.Insert(buf);
 }
 

db/memtable.h

@@ -14,7 +14,6 @@
 namespace leveldb {
 
 class InternalKeyComparator;
-class Mutex;
 class MemTableIterator;
 
 class MemTable {
@@ -36,10 +35,7 @@ class MemTable {
   }
 
   // Returns an estimate of the number of bytes of data in use by this
-  // data structure.
-  //
-  // REQUIRES: external synchronization to prevent simultaneous
-  // operations on the same MemTable.
+  // data structure. It is safe to call when MemTable is being modified.
   size_t ApproximateMemoryUsage();
 
   // Return an iterator that yields the contents of the memtable.

db/recovery_test.cc

@@ -0,0 +1,344 @@
+// Copyright (c) 2014 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "db/db_impl.h"
+#include "db/filename.h"
+#include "db/version_set.h"
+#include "db/write_batch_internal.h"
+#include "leveldb/db.h"
+#include "leveldb/env.h"
+#include "leveldb/write_batch.h"
+#include "util/logging.h"
+#include "util/testharness.h"
+#include "util/testutil.h"
+
+namespace leveldb {
+
+class RecoveryTest {
+ public:
+  RecoveryTest() : env_(Env::Default()), db_(nullptr) {
+    dbname_ = test::TmpDir() + "/recovery_test";
+    DestroyDB(dbname_, Options());
+    Open();
+  }
+
+  ~RecoveryTest() {
+    Close();
+    DestroyDB(dbname_, Options());
+  }
+
+  DBImpl* dbfull() const { return reinterpret_cast<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_OK(OpenWithStatus(options));
+    ASSERT_EQ(1, NumLogs());
+  }
+
+  Status Put(const std::string& k, const std::string& v) {
+    return db_->Put(WriteOptions(), k, v);
+  }
+
+  std::string Get(const std::string& k, const Snapshot* snapshot = nullptr) {
+    std::string result;
+    Status s = db_->Get(ReadOptions(), k, &result);
+    if (s.IsNotFound()) {
+      result = "NOT_FOUND";
+    } else if (!s.ok()) {
+      result = s.ToString();
+    }
+    return result;
+  }
+
+  std::string ManifestFileName() {
+    std::string current;
+    ASSERT_OK(ReadFileToString(env_, CurrentFileName(dbname_), &current));
+    size_t len = current.size();
+    if (len > 0 && current[len-1] == '\n') {
+      current.resize(len - 1);
+    }
+    return dbname_ + "/" + current;
+  }
+
+  std::string LogName(uint64_t number) {
+    return LogFileName(dbname_, number);
+  }
+
+  size_t DeleteLogFiles() {
+    // Linux allows unlinking open files, but Windows does not.
+    // Closing the db allows for file deletion.
+    Close();
+    std::vector<uint64_t> logs = GetFiles(kLogFile);
+    for (size_t i = 0; i < logs.size(); i++) {
+      ASSERT_OK(env_->DeleteFile(LogName(logs[i]))) << LogName(logs[i]);
+    }
+    return logs.size();
+  }
+
+  void DeleteManifestFile() {
+    ASSERT_OK(env_->DeleteFile(ManifestFileName()));
+  }
+
+  uint64_t FirstLogFile() {
+    return GetFiles(kLogFile)[0];
+  }
+
+  std::vector<uint64_t> GetFiles(FileType t) {
+    std::vector<std::string> filenames;
+    ASSERT_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;
+    ASSERT_OK(env_->GetFileSize(fname, &result)) << fname;
+    return result;
+  }
+
+  void CompactMemTable() {
+    dbfull()->TEST_CompactMemTable();
+  }
+
+  // Directly construct a log file that sets key to val.
+  void MakeLogFile(uint64_t lognum, SequenceNumber seq, Slice key, Slice val) {
+    std::string fname = LogFileName(dbname_, lognum);
+    WritableFile* file;
+    ASSERT_OK(env_->NewWritableFile(fname, &file));
+    log::Writer writer(file);
+    WriteBatch batch;
+    batch.Put(key, val);
+    WriteBatchInternal::SetSequence(&batch, seq);
+    ASSERT_OK(writer.AddRecord(WriteBatchInternal::Contents(&batch)));
+    ASSERT_OK(file->Flush());
+    delete file;
+  }
+
+ private:
+  std::string dbname_;
+  Env* env_;
+  DB* db_;
+};
+
+TEST(RecoveryTest, ManifestReused) {
+  if (!CanAppend()) {
+    fprintf(stderr, "skipping test because env does not support appending\n");
+    return;
+  }
+  ASSERT_OK(Put("foo", "bar"));
+  Close();
+  std::string old_manifest = ManifestFileName();
+  Open();
+  ASSERT_EQ(old_manifest, ManifestFileName());
+  ASSERT_EQ("bar", Get("foo"));
+  Open();
+  ASSERT_EQ(old_manifest, ManifestFileName());
+  ASSERT_EQ("bar", Get("foo"));
+}
+
+TEST(RecoveryTest, LargeManifestCompacted) {
+  if (!CanAppend()) {
+    fprintf(stderr, "skipping test because env does not support appending\n");
+    return;
+  }
+  ASSERT_OK(Put("foo", "bar"));
+  Close();
+  std::string old_manifest = ManifestFileName();
+
+  // Pad with zeroes to make manifest file very big.
+  {
+    uint64_t len = FileSize(old_manifest);
+    WritableFile* file;
+    ASSERT_OK(env()->NewAppendableFile(old_manifest, &file));
+    std::string zeroes(3*1048576 - static_cast<size_t>(len), 0);
+    ASSERT_OK(file->Append(zeroes));
+    ASSERT_OK(file->Flush());
+    delete file;
+  }
+
+  Open();
+  std::string new_manifest = ManifestFileName();
+  ASSERT_NE(old_manifest, new_manifest);
+  ASSERT_GT(10000, FileSize(new_manifest));
+  ASSERT_EQ("bar", Get("foo"));
+
+  Open();
+  ASSERT_EQ(new_manifest, ManifestFileName());
+  ASSERT_EQ("bar", Get("foo"));
+}
+
+TEST(RecoveryTest, NoLogFiles) {
+  ASSERT_OK(Put("foo", "bar"));
+  ASSERT_EQ(1, DeleteLogFiles());
+  Open();
+  ASSERT_EQ("NOT_FOUND", Get("foo"));
+  Open();
+  ASSERT_EQ("NOT_FOUND", Get("foo"));
+}
+
+TEST(RecoveryTest, LogFileReuse) {
+  if (!CanAppend()) {
+    fprintf(stderr, "skipping test because env does not support appending\n");
+    return;
+  }
+  for (int i = 0; i < 2; i++) {
+    ASSERT_OK(Put("foo", "bar"));
+    if (i == 0) {
+      // Compact to ensure current log is empty
+      CompactMemTable();
+    }
+    Close();
+    ASSERT_EQ(1, NumLogs());
+    uint64_t number = FirstLogFile();
+    if (i == 0) {
+      ASSERT_EQ(0, FileSize(LogName(number)));
+    } else {
+      ASSERT_LT(0, FileSize(LogName(number)));
+    }
+    Open();
+    ASSERT_EQ(1, NumLogs());
+    ASSERT_EQ(number, FirstLogFile()) << "did not reuse log file";
+    ASSERT_EQ("bar", Get("foo"));
+    Open();
+    ASSERT_EQ(1, NumLogs());
+    ASSERT_EQ(number, FirstLogFile()) << "did not reuse log file";
+    ASSERT_EQ("bar", Get("foo"));
+  }
+}
+
+TEST(RecoveryTest, MultipleMemTables) {
+  // Make a large log.
+  const int kNum = 1000;
+  for (int i = 0; i < kNum; i++) {
+    char buf[100];
+    snprintf(buf, sizeof(buf), "%050d", i);
+    ASSERT_OK(Put(buf, buf));
+  }
+  ASSERT_EQ(0, NumTables());
+  Close();
+  ASSERT_EQ(0, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  uint64_t old_log_file = FirstLogFile();
+
+  // Force creation of multiple memtables by reducing the write buffer size.
+  Options opt;
+  opt.reuse_logs = true;
+  opt.write_buffer_size = (kNum*100) / 2;
+  Open(&opt);
+  ASSERT_LE(2, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  ASSERT_NE(old_log_file, FirstLogFile()) << "must not reuse log";
+  for (int i = 0; i < kNum; i++) {
+    char buf[100];
+    snprintf(buf, sizeof(buf), "%050d", i);
+    ASSERT_EQ(buf, Get(buf));
+  }
+}
+
+TEST(RecoveryTest, MultipleLogFiles) {
+  ASSERT_OK(Put("foo", "bar"));
+  Close();
+  ASSERT_EQ(1, NumLogs());
+
+  // Make a bunch of uncompacted log files.
+  uint64_t old_log = FirstLogFile();
+  MakeLogFile(old_log+1, 1000, "hello", "world");
+  MakeLogFile(old_log+2, 1001, "hi", "there");
+  MakeLogFile(old_log+3, 1002, "foo", "bar2");
+
+  // Recover and check that all log files were processed.
+  Open();
+  ASSERT_LE(1, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  uint64_t new_log = FirstLogFile();
+  ASSERT_LE(old_log+3, new_log);
+  ASSERT_EQ("bar2", Get("foo"));
+  ASSERT_EQ("world", Get("hello"));
+  ASSERT_EQ("there", Get("hi"));
+
+  // Test that previous recovery produced recoverable state.
+  Open();
+  ASSERT_LE(1, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  if (CanAppend()) {
+    ASSERT_EQ(new_log, FirstLogFile());
+  }
+  ASSERT_EQ("bar2", Get("foo"));
+  ASSERT_EQ("world", Get("hello"));
+  ASSERT_EQ("there", Get("hi"));
+
+  // Check that introducing an older log file does not cause it to be re-read.
+  Close();
+  MakeLogFile(old_log+1, 2000, "hello", "stale write");
+  Open();
+  ASSERT_LE(1, NumTables());
+  ASSERT_EQ(1, NumLogs());
+  if (CanAppend()) {
+    ASSERT_EQ(new_log, FirstLogFile());
+  }
+  ASSERT_EQ("bar2", Get("foo"));
+  ASSERT_EQ("world", Get("hello"));
+  ASSERT_EQ("there", Get("hi"));
+}
+
+TEST(RecoveryTest, ManifestMissing) {
+  ASSERT_OK(Put("foo", "bar"));
+  Close();
+  DeleteManifestFile();
+
+  Status status = OpenWithStatus();
+  ASSERT_TRUE(status.IsCorruption());
+}
+
+}  // namespace leveldb
+
+int main(int argc, char** argv) {
+  return leveldb::test::RunAllTests();
+}

db/repair.cc

@@ -54,7 +54,7 @@ class Repairer {
         owns_cache_(options_.block_cache != options.block_cache),
         next_file_number_(1) {
     // TableCache can be small since we expect each table to be opened once.
-    table_cache_ = new TableCache(dbname_, &options_, 10);
+    table_cache_ = new TableCache(dbname_, options_, 10);
   }
 
   ~Repairer() {
@@ -227,7 +227,7 @@ class Repairer {
     status = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta);
     delete iter;
     mem->Unref();
-    mem = NULL;
+    mem = nullptr;
     if (status.ok()) {
       if (meta.file_size > 0) {
         table_numbers_.push_back(meta.number);
@@ -350,13 +350,13 @@ class Repairer {
       }
     }
     delete builder;
-    builder = NULL;
+    builder = nullptr;
 
     if (s.ok()) {
       s = file->Close();
     }
     delete file;
-    file = NULL;
+    file = nullptr;
 
     if (counter > 0 && s.ok()) {
       std::string orig = TableFileName(dbname_, t.meta.number);
@@ -410,7 +410,7 @@ class Repairer {
       status = file->Close();
     }
     delete file;
-    file = NULL;
+    file = nullptr;
 
     if (!status.ok()) {
       env_->DeleteFile(tmp);
@@ -438,14 +438,14 @@ class Repairer {
     //    dir/lost/foo
     const char* slash = strrchr(fname.c_str(), '/');
     std::string new_dir;
-    if (slash != NULL) {
+    if (slash != nullptr) {
       new_dir.assign(fname.data(), slash - fname.data());
     }
     new_dir.append("/lost");
     env_->CreateDir(new_dir);  // Ignore error
     std::string new_file = new_dir;
     new_file.append("/");
-    new_file.append((slash == NULL) ? fname.c_str() : slash + 1);
+    new_file.append((slash == nullptr) ? fname.c_str() : slash + 1);
     Status s = env_->RenameFile(fname, new_file);
     Log(options_.info_log, "Archiving %s: %s\n",
         fname.c_str(), s.ToString().c_str());

db/skiplist.h

@@ -1,10 +1,10 @@
-#ifndef STORAGE_LEVELDB_DB_SKIPLIST_H_
-#define STORAGE_LEVELDB_DB_SKIPLIST_H_
-
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source 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
 // -------------
 //
@@ -27,9 +27,10 @@
 //
 // ... prev vs. next pointer ordering ...
 
-#include <assert.h>
-#include <stdlib.h>
-#include "port/port.h"
+#include <atomic>
+#include <cassert>
+#include <cstdlib>
+
 #include "util/arena.h"
 #include "util/random.h"
 
@@ -105,11 +106,10 @@ class SkipList {
 
   // Modified only by Insert().  Read racily by readers, but stale
   // values are ok.
-  port::AtomicPointer max_height_;   // Height of the entire list
+  std::atomic<int> max_height_;   // Height of the entire list
 
   inline int GetMaxHeight() const {
-    return static_cast<int>(
-        reinterpret_cast<intptr_t>(max_height_.NoBarrier_Load()));
+    return max_height_.load(std::memory_order_relaxed);
   }
 
   // Read/written only by Insert().
@@ -123,9 +123,9 @@ class SkipList {
   bool KeyIsAfterNode(const Key& key, Node* n) const;
 
   // Return the earliest node that comes at or after key.
-  // Return NULL if there is no such node.
+  // Return nullptr if there is no such node.
   //
-  // If prev is non-NULL, fills prev[level] with pointer to previous
+  // If prev is non-null, fills prev[level] with pointer to previous
   // node at "level" for every level in [0..max_height_-1].
   Node* FindGreaterOrEqual(const Key& key, Node** prev) const;
 
@@ -144,7 +144,7 @@ class SkipList {
 
 // Implementation details follow
 template<typename Key, class Comparator>
-struct SkipList<Key,Comparator>::Node {
+struct SkipList<Key, Comparator>::Node {
   explicit Node(const Key& k) : key(k) { }
 
   Key const key;
@@ -155,92 +155,92 @@ struct SkipList<Key,Comparator>::Node {
     assert(n >= 0);
     // Use an 'acquire load' so that we observe a fully initialized
     // version of the returned Node.
-    return reinterpret_cast<Node*>(next_[n].Acquire_Load());
+    return next_[n].load(std::memory_order_acquire);
   }
   void SetNext(int n, Node* x) {
     assert(n >= 0);
     // Use a 'release store' so that anybody who reads through this
     // pointer observes a fully initialized version of the inserted node.
-    next_[n].Release_Store(x);
+    next_[n].store(x, std::memory_order_release);
   }
 
   // No-barrier variants that can be safely used in a few locations.
   Node* NoBarrier_Next(int n) {
     assert(n >= 0);
-    return reinterpret_cast<Node*>(next_[n].NoBarrier_Load());
+    return next_[n].load(std::memory_order_relaxed);
   }
   void NoBarrier_SetNext(int n, Node* x) {
     assert(n >= 0);
-    next_[n].NoBarrier_Store(x);
+    next_[n].store(x, std::memory_order_relaxed);
   }
 
  private:
   // Array of length equal to the node height.  next_[0] is lowest level link.
-  port::AtomicPointer next_[1];
+  std::atomic<Node*> next_[1];
 };
 
 template<typename Key, class Comparator>
-typename SkipList<Key,Comparator>::Node*
-SkipList<Key,Comparator>::NewNode(const Key& key, int height) {
-  char* mem = arena_->AllocateAligned(
-      sizeof(Node) + sizeof(port::AtomicPointer) * (height - 1));
-  return new (mem) Node(key);
+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) {
+inline SkipList<Key, Comparator>::Iterator::Iterator(const SkipList* list) {
   list_ = list;
-  node_ = NULL;
+  node_ = nullptr;
 }
 
 template<typename Key, class Comparator>
-inline bool SkipList<Key,Comparator>::Iterator::Valid() const {
-  return node_ != NULL;
+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 {
+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() {
+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() {
+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_ = NULL;
+    node_ = nullptr;
   }
 }
 
 template<typename Key, class Comparator>
-inline void SkipList<Key,Comparator>::Iterator::Seek(const Key& target) {
-  node_ = list_->FindGreaterOrEqual(target, NULL);
+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() {
+inline void SkipList<Key, Comparator>::Iterator::SeekToFirst() {
   node_ = list_->head_->Next(0);
 }
 
 template<typename Key, class Comparator>
-inline void SkipList<Key,Comparator>::Iterator::SeekToLast() {
+inline void SkipList<Key, Comparator>::Iterator::SeekToLast() {
   node_ = list_->FindLast();
   if (node_ == list_->head_) {
-    node_ = NULL;
+    node_ = nullptr;
   }
 }
 
 template<typename Key, class Comparator>
-int SkipList<Key,Comparator>::RandomHeight() {
+int SkipList<Key, Comparator>::RandomHeight() {
   // Increase height with probability 1 in kBranching
   static const unsigned int kBranching = 4;
   int height = 1;
@@ -253,14 +253,15 @@ int SkipList<Key,Comparator>::RandomHeight() {
 }
 
 template<typename Key, class Comparator>
-bool SkipList<Key,Comparator>::KeyIsAfterNode(const Key& key, Node* n) const {
-  // NULL n is considered infinite
-  return (n != NULL) && (compare_(n->key, key) < 0);
+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 {
+typename SkipList<Key, Comparator>::Node*
+SkipList<Key, Comparator>::FindGreaterOrEqual(const Key& key,
+                                              Node** prev) const {
   Node* x = head_;
   int level = GetMaxHeight() - 1;
   while (true) {
@@ -269,7 +270,7 @@ typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindGreaterOr
       // Keep searching in this list
       x = next;
     } else {
-      if (prev != NULL) prev[level] = x;
+      if (prev != nullptr) prev[level] = x;
       if (level == 0) {
         return next;
       } else {
@@ -281,14 +282,14 @@ typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindGreaterOr
 }
 
 template<typename Key, class Comparator>
-typename SkipList<Key,Comparator>::Node*
-SkipList<Key,Comparator>::FindLessThan(const Key& key) const {
+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 == NULL || compare_(next->key, key) >= 0) {
+    if (next == nullptr || compare_(next->key, key) >= 0) {
       if (level == 0) {
         return x;
       } else {
@@ -302,13 +303,13 @@ SkipList<Key,Comparator>::FindLessThan(const Key& key) const {
 }
 
 template<typename Key, class Comparator>
-typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindLast()
+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 == NULL) {
+    if (next == nullptr) {
       if (level == 0) {
         return x;
       } else {
@@ -322,42 +323,40 @@ typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindLast()
 }
 
 template<typename Key, class Comparator>
-SkipList<Key,Comparator>::SkipList(Comparator cmp, Arena* arena)
+SkipList<Key, Comparator>::SkipList(Comparator cmp, Arena* arena)
     : compare_(cmp),
       arena_(arena),
       head_(NewNode(0 /* any key will do */, kMaxHeight)),
-      max_height_(reinterpret_cast<void*>(1)),
+      max_height_(1),
       rnd_(0xdeadbeef) {
   for (int i = 0; i < kMaxHeight; i++) {
-    head_->SetNext(i, NULL);
+    head_->SetNext(i, nullptr);
   }
 }
 
 template<typename Key, class Comparator>
-void SkipList<Key,Comparator>::Insert(const Key& key) {
+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 == NULL || !Equal(key, x->key));
+  assert(x == nullptr || !Equal(key, x->key));
 
   int height = RandomHeight();
   if (height > GetMaxHeight()) {
     for (int i = GetMaxHeight(); i < height; i++) {
       prev[i] = head_;
     }
-    //fprintf(stderr, "Change height from %d to %d\n", max_height_, height);
-
     // It is ok to mutate max_height_ without any synchronization
     // with concurrent readers.  A concurrent reader that observes
     // the new value of max_height_ will see either the old value of
-    // new level pointers from head_ (NULL), or a new value set in
+    // new level pointers from head_ (nullptr), or a new value set in
     // the loop below.  In the former case the reader will
-    // immediately drop to the next level since NULL sorts after all
+    // immediately drop to the next level since nullptr sorts after all
     // keys.  In the latter case the reader will use the new node.
-    max_height_.NoBarrier_Store(reinterpret_cast<void*>(height));
+    max_height_.store(height, std::memory_order_relaxed);
   }
 
   x = NewNode(key, height);
@@ -370,9 +369,9 @@ void SkipList<Key,Comparator>::Insert(const Key& key) {
 }
 
 template<typename Key, class Comparator>
-bool SkipList<Key,Comparator>::Contains(const Key& key) const {
-  Node* x = FindGreaterOrEqual(key, NULL);
-  if (x != NULL && Equal(key, x->key)) {
+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;

db/skiplist_test.cc

@@ -3,8 +3,13 @@
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 
 #include "db/skiplist.h"
+
+#include <atomic>
 #include <set>
+
 #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"
@@ -126,7 +131,7 @@ TEST(SkipTest, InsertAndLookup) {
 // concurrent readers (with no synchronization other than when a
 // reader's iterator is created), the reader always observes all the
 // data that was present in the skip list when the iterator was
-// constructor.  Because insertions are happening concurrently, we may
+// constructed.  Because insertions are happening concurrently, we may
 // also observe new values that were inserted since the iterator was
 // constructed, but we should never miss any values that were present
 // at iterator construction time.
@@ -186,12 +191,12 @@ class ConcurrentTest {
 
   // Per-key generation
   struct State {
-    port::AtomicPointer generation[K];
-    void Set(int k, intptr_t v) {
-      generation[k].Release_Store(reinterpret_cast<void*>(v));
+    std::atomic<int> generation[K];
+    void Set(int k, int v) {
+      generation[k].store(v, std::memory_order_release);
     }
-    intptr_t Get(int k) {
-      return reinterpret_cast<intptr_t>(generation[k].Acquire_Load());
+    int Get(int k) {
+      return generation[k].load(std::memory_order_acquire);
     }
 
     State() {
@@ -250,7 +255,7 @@ class ConcurrentTest {
         // Note that generation 0 is never inserted, so it is ok if
         // <*,0,*> is missing.
         ASSERT_TRUE((gen(pos) == 0) ||
-                    (gen(pos) > initial_state.Get(key(pos)))
+                    (gen(pos) > static_cast<Key>(initial_state.Get(key(pos))))
                     ) << "key: " << key(pos)
                       << "; gen: " << gen(pos)
                       << "; initgen: "
@@ -298,7 +303,7 @@ class TestState {
  public:
   ConcurrentTest t_;
   int seed_;
-  port::AtomicPointer quit_flag_;
+  std::atomic<bool> quit_flag_;
 
   enum ReaderState {
     STARTING,
@@ -308,11 +313,11 @@ class TestState {
 
   explicit TestState(int s)
       : seed_(s),
-        quit_flag_(NULL),
+        quit_flag_(false),
         state_(STARTING),
         state_cv_(&mu_) {}
 
-  void Wait(ReaderState s) {
+  void Wait(ReaderState s) LOCKS_EXCLUDED(mu_) {
     mu_.Lock();
     while (state_ != s) {
       state_cv_.Wait();
@@ -320,7 +325,7 @@ class TestState {
     mu_.Unlock();
   }
 
-  void Change(ReaderState s) {
+  void Change(ReaderState s) LOCKS_EXCLUDED(mu_) {
     mu_.Lock();
     state_ = s;
     state_cv_.Signal();
@@ -329,8 +334,8 @@ class TestState {
 
  private:
   port::Mutex mu_;
-  ReaderState state_;
-  port::CondVar state_cv_;
+  ReaderState state_ GUARDED_BY(mu_);
+  port::CondVar state_cv_ GUARDED_BY(mu_);
 };
 
 static void ConcurrentReader(void* arg) {
@@ -338,7 +343,7 @@ static void ConcurrentReader(void* arg) {
   Random rnd(state->seed_);
   int64_t reads = 0;
   state->Change(TestState::RUNNING);
-  while (!state->quit_flag_.Acquire_Load()) {
+  while (!state->quit_flag_.load(std::memory_order_acquire)) {
     state->t_.ReadStep(&rnd);
     ++reads;
   }
@@ -360,7 +365,7 @@ static void RunConcurrent(int run) {
     for (int i = 0; i < kSize; i++) {
       state.t_.WriteStep(&rnd);
     }
-    state.quit_flag_.Release_Store(&state);  // Any non-NULL arg will do
+    state.quit_flag_.store(true, std::memory_order_release);
     state.Wait(TestState::DONE);
   }
 }

db/snapshot.h

@@ -5,6 +5,7 @@
 #ifndef STORAGE_LEVELDB_DB_SNAPSHOT_H_
 #define STORAGE_LEVELDB_DB_SNAPSHOT_H_
 
+#include "db/dbformat.h"
 #include "leveldb/db.h"
 
 namespace leveldb {
@@ -15,50 +16,72 @@ class SnapshotList;
 // Each SnapshotImpl corresponds to a particular sequence number.
 class SnapshotImpl : public Snapshot {
  public:
-  SequenceNumber number_;  // const after creation
+  SnapshotImpl(SequenceNumber sequence_number)
+      : sequence_number_(sequence_number) {}
+
+  SequenceNumber sequence_number() const { return sequence_number_; }
 
  private:
   friend class SnapshotList;
 
-  // SnapshotImpl is kept in a doubly-linked circular list
+  // SnapshotImpl is kept in a doubly-linked circular list. The SnapshotList
+  // implementation operates on the next/previous fields direcly.
   SnapshotImpl* prev_;
   SnapshotImpl* next_;
 
-  SnapshotList* list_;                 // just for sanity checks
+  const SequenceNumber sequence_number_;
+
+#if !defined(NDEBUG)
+  SnapshotList* list_ = nullptr;
+#endif  // !defined(NDEBUG)
 };
 
 class SnapshotList {
  public:
-  SnapshotList() {
-    list_.prev_ = &list_;
-    list_.next_ = &list_;
+  SnapshotList() : head_(0) {
+    head_.prev_ = &head_;
+    head_.next_ = &head_;
   }
 
-  bool empty() const { return list_.next_ == &list_; }
-  SnapshotImpl* oldest() const { assert(!empty()); return list_.next_; }
-  SnapshotImpl* newest() const { assert(!empty()); return list_.prev_; }
+  bool empty() const { return head_.next_ == &head_; }
+  SnapshotImpl* oldest() const { assert(!empty()); return head_.next_; }
+  SnapshotImpl* newest() const { assert(!empty()); return head_.prev_; }
 
-  const SnapshotImpl* New(SequenceNumber seq) {
-    SnapshotImpl* s = new SnapshotImpl;
-    s->number_ = seq;
-    s->list_ = this;
-    s->next_ = &list_;
-    s->prev_ = list_.prev_;
-    s->prev_->next_ = s;
-    s->next_->prev_ = s;
-    return s;
+  // 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;
   }
 
-  void Delete(const SnapshotImpl* s) {
-    assert(s->list_ == this);
-    s->prev_->next_ = s->next_;
-    s->next_->prev_ = s->prev_;
-    delete s;
+  // Removes a SnapshotImpl from this list.
+  //
+  // The snapshot must have been created by calling New() on this list.
+  //
+  // The snapshot pointer should not be const, because its memory is
+  // deallocated. However, that would force us to change DB::ReleaseSnapshot(),
+  // which is in the API, and currently takes a const Snapshot.
+  void Delete(const SnapshotImpl* snapshot) {
+#if !defined(NDEBUG)
+    assert(snapshot->list_ == this);
+#endif  // !defined(NDEBUG)
+    snapshot->prev_->next_ = snapshot->next_;
+    snapshot->next_->prev_ = snapshot->prev_;
+    delete snapshot;
   }
 
  private:
   // Dummy head of doubly-linked list of snapshots
-  SnapshotImpl list_;
+  SnapshotImpl head_;
 };
 
 }  // namespace leveldb

db/table_cache.cc

@@ -30,9 +30,9 @@ static void UnrefEntry(void* arg1, void* arg2) {
 }
 
 TableCache::TableCache(const std::string& dbname,
-                       const Options* options,
+                       const Options& options,
                        int entries)
-    : env_(options->env),
+    : env_(options.env),
       dbname_(dbname),
       options_(options),
       cache_(NewLRUCache(entries)) {
@@ -49,10 +49,10 @@ Status TableCache::FindTable(uint64_t file_number, uint64_t file_size,
   EncodeFixed64(buf, file_number);
   Slice key(buf, sizeof(buf));
   *handle = cache_->Lookup(key);
-  if (*handle == NULL) {
+  if (*handle == nullptr) {
     std::string fname = TableFileName(dbname_, file_number);
-    RandomAccessFile* file = NULL;
-    Table* table = NULL;
+    RandomAccessFile* file = nullptr;
+    Table* table = nullptr;
     s = env_->NewRandomAccessFile(fname, &file);
     if (!s.ok()) {
       std::string old_fname = SSTTableFileName(dbname_, file_number);
@@ -61,11 +61,11 @@ Status TableCache::FindTable(uint64_t file_number, uint64_t file_size,
       }
     }
     if (s.ok()) {
-      s = Table::Open(*options_, file, file_size, &table);
+      s = Table::Open(options_, file, file_size, &table);
     }
 
     if (!s.ok()) {
-      assert(table == NULL);
+      assert(table == nullptr);
       delete file;
       // We do not cache error results so that if the error is transient,
       // or somebody repairs the file, we recover automatically.
@@ -83,11 +83,11 @@ Iterator* TableCache::NewIterator(const ReadOptions& options,
                                   uint64_t file_number,
                                   uint64_t file_size,
                                   Table** tableptr) {
-  if (tableptr != NULL) {
-    *tableptr = NULL;
+  if (tableptr != nullptr) {
+    *tableptr = nullptr;
   }
 
-  Cache::Handle* handle = NULL;
+  Cache::Handle* handle = nullptr;
   Status s = FindTable(file_number, file_size, &handle);
   if (!s.ok()) {
     return NewErrorIterator(s);
@@ -96,7 +96,7 @@ Iterator* TableCache::NewIterator(const ReadOptions& options,
   Table* table = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table;
   Iterator* result = table->NewIterator(options);
   result->RegisterCleanup(&UnrefEntry, cache_, handle);
-  if (tableptr != NULL) {
+  if (tableptr != nullptr) {
     *tableptr = table;
   }
   return result;
@@ -108,7 +108,7 @@ Status TableCache::Get(const ReadOptions& options,
                        const Slice& k,
                        void* arg,
                        void (*saver)(void*, const Slice&, const Slice&)) {
-  Cache::Handle* handle = NULL;
+  Cache::Handle* handle = nullptr;
   Status s = FindTable(file_number, file_size, &handle);
   if (s.ok()) {
     Table* t = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table;

db/table_cache.h

@@ -20,20 +20,20 @@ class Env;
 
 class TableCache {
  public:
-  TableCache(const std::string& dbname, const Options* options, int entries);
+  TableCache(const std::string& dbname, const Options& options, int entries);
   ~TableCache();
 
   // Return an iterator for the specified file number (the corresponding
   // file length must be exactly "file_size" bytes).  If "tableptr" is
-  // non-NULL, also sets "*tableptr" to point to the Table object
-  // underlying the returned iterator, or NULL if no Table object underlies
-  // the returned iterator.  The returned "*tableptr" object is owned by
-  // the cache and should not be deleted, and is valid for as long as the
+  // non-null, also sets "*tableptr" to point to the Table object
+  // underlying the returned iterator, or to nullptr if no Table object
+  // underlies the returned iterator.  The returned "*tableptr" object is owned
+  // by the cache and should not be deleted, and is valid for as long as the
   // returned iterator is live.
   Iterator* NewIterator(const ReadOptions& options,
                         uint64_t file_number,
                         uint64_t file_size,
-                        Table** tableptr = NULL);
+                        Table** tableptr = nullptr);
 
   // If a seek to internal key "k" in specified file finds an entry,
   // call (*handle_result)(arg, found_key, found_value).
@@ -50,7 +50,7 @@ class TableCache {
  private:
   Env* const env_;
   const std::string dbname_;
-  const Options* options_;
+  const Options& options_;
   Cache* cache_;
 
   Status FindTable(uint64_t file_number, uint64_t file_size, Cache::Handle**);

db/version_edit.cc

@@ -109,7 +109,7 @@ static bool GetLevel(Slice* input, int* level) {
 Status VersionEdit::DecodeFrom(const Slice& src) {
   Clear();
   Slice input = src;
-  const char* msg = NULL;
+  const char* msg = nullptr;
   uint32_t tag;
 
   // Temporary storage for parsing
@@ -119,7 +119,7 @@ Status VersionEdit::DecodeFrom(const Slice& src) {
   Slice str;
   InternalKey key;
 
-  while (msg == NULL && GetVarint32(&input, &tag)) {
+  while (msg == nullptr && GetVarint32(&input, &tag)) {
     switch (tag) {
       case kComparator:
         if (GetLengthPrefixedSlice(&input, &str)) {
@@ -198,12 +198,12 @@ Status VersionEdit::DecodeFrom(const Slice& src) {
     }
   }
 
-  if (msg == NULL && !input.empty()) {
+  if (msg == nullptr && !input.empty()) {
     msg = "invalid tag";
   }
 
   Status result;
-  if (msg != NULL) {
+  if (msg != nullptr) {
     result = Status::Corruption("VersionEdit", msg);
   }
   return result;

db/version_set.cc

@@ -20,21 +20,29 @@
 
 namespace leveldb {
 
-static const int kTargetFileSize = 2 * 1048576;
+static size_t TargetFileSize(const Options* options) {
+  return options->max_file_size;
+}
 
 // Maximum bytes of overlaps in grandparent (i.e., level+2) before we
 // stop building a single file in a level->level+1 compaction.
-static const int64_t kMaxGrandParentOverlapBytes = 10 * kTargetFileSize;
+static int64_t MaxGrandParentOverlapBytes(const Options* options) {
+  return 10 * TargetFileSize(options);
+}
 
 // Maximum number of bytes in all compacted files.  We avoid expanding
 // the lower level file set of a compaction if it would make the
 // total compaction cover more than this many bytes.
-static const int64_t kExpandedCompactionByteSizeLimit = 25 * kTargetFileSize;
+static int64_t ExpandedCompactionByteSizeLimit(const Options* options) {
+  return 25 * TargetFileSize(options);
+}
 
-static double MaxBytesForLevel(int level) {
+static double MaxBytesForLevel(const Options* options, int level) {
   // Note: the result for level zero is not really used since we set
   // the level-0 compaction threshold based on number of files.
-  double result = 10 * 1048576.0;  // Result for both level-0 and level-1
+
+  // Result for both level-0 and level-1
+  double result = 10. * 1048576.0;
   while (level > 1) {
     result *= 10;
     level--;
@@ -42,8 +50,9 @@ static double MaxBytesForLevel(int level) {
   return result;
 }
 
-static uint64_t MaxFileSizeForLevel(int level) {
-  return kTargetFileSize;  // We could vary per level to reduce number of files?
+static uint64_t MaxFileSizeForLevel(const Options* options, int level) {
+  // We could vary per level to reduce number of files?
+  return TargetFileSize(options);
 }
 
 static int64_t TotalFileSize(const std::vector<FileMetaData*>& files) {
@@ -97,15 +106,15 @@ int FindFile(const InternalKeyComparator& icmp,
 
 static bool AfterFile(const Comparator* ucmp,
                       const Slice* user_key, const FileMetaData* f) {
-  // NULL user_key occurs before all keys and is therefore never after *f
-  return (user_key != NULL &&
+  // null user_key occurs before all keys and is therefore never after *f
+  return (user_key != nullptr &&
           ucmp->Compare(*user_key, f->largest.user_key()) > 0);
 }
 
 static bool BeforeFile(const Comparator* ucmp,
                        const Slice* user_key, const FileMetaData* f) {
-  // NULL user_key occurs after all keys and is therefore never before *f
-  return (user_key != NULL &&
+  // 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);
 }
 
@@ -132,10 +141,10 @@ bool SomeFileOverlapsRange(
 
   // Binary search over file list
   uint32_t index = 0;
-  if (smallest_user_key != NULL) {
+  if (smallest_user_key != nullptr) {
     // Find the earliest possible internal key for smallest_user_key
-    InternalKey small(*smallest_user_key, kMaxSequenceNumber,kValueTypeForSeek);
-    index = FindFile(icmp, files, small.Encode());
+    InternalKey small_key(*smallest_user_key, kMaxSequenceNumber,kValueTypeForSeek);
+    index = FindFile(icmp, files, small_key.Encode());
   }
 
   if (index >= files.size()) {
@@ -329,14 +338,14 @@ Status Version::Get(const ReadOptions& options,
   const Comparator* ucmp = vset_->icmp_.user_comparator();
   Status s;
 
-  stats->seek_file = NULL;
+  stats->seek_file = nullptr;
   stats->seek_file_level = -1;
-  FileMetaData* last_file_read = NULL;
+  FileMetaData* last_file_read = nullptr;
   int last_file_read_level = -1;
 
   // We can search level-by-level since entries never hop across
   // levels.  Therefore we are guaranteed that if we find data
-  // in an smaller level, later levels are irrelevant.
+  // in a smaller level, later levels are irrelevant.
   std::vector<FileMetaData*> tmp;
   FileMetaData* tmp2;
   for (int level = 0; level < config::kNumLevels; level++) {
@@ -365,13 +374,13 @@ Status Version::Get(const ReadOptions& options,
       // Binary search to find earliest index whose largest key >= ikey.
       uint32_t index = FindFile(vset_->icmp_, files_[level], ikey);
       if (index >= num_files) {
-        files = NULL;
+        files = nullptr;
         num_files = 0;
       } else {
         tmp2 = files[index];
         if (ucmp->Compare(user_key, tmp2->smallest.user_key()) < 0) {
           // All of "tmp2" is past any data for user_key
-          files = NULL;
+          files = nullptr;
           num_files = 0;
         } else {
           files = &tmp2;
@@ -381,7 +390,7 @@ Status Version::Get(const ReadOptions& options,
     }
 
     for (uint32_t i = 0; i < num_files; ++i) {
-      if (last_file_read != NULL && stats->seek_file == NULL) {
+      if (last_file_read != nullptr && stats->seek_file == nullptr) {
         // We have had more than one seek for this read.  Charge the 1st file.
         stats->seek_file = last_file_read;
         stats->seek_file_level = last_file_read_level;
@@ -421,9 +430,9 @@ Status Version::Get(const ReadOptions& options,
 
 bool Version::UpdateStats(const GetStats& stats) {
   FileMetaData* f = stats.seek_file;
-  if (f != NULL) {
+  if (f != nullptr) {
     f->allowed_seeks--;
-    if (f->allowed_seeks <= 0 && file_to_compact_ == NULL) {
+    if (f->allowed_seeks <= 0 && file_to_compact_ == nullptr) {
       file_to_compact_ = f;
       file_to_compact_level_ = stats.seek_file_level;
       return true;
@@ -508,7 +517,7 @@ int Version::PickLevelForMemTableOutput(
         // Check that file does not overlap too many grandparent bytes.
         GetOverlappingInputs(level + 2, &start, &limit, &overlaps);
         const int64_t sum = TotalFileSize(overlaps);
-        if (sum > kMaxGrandParentOverlapBytes) {
+        if (sum > MaxGrandParentOverlapBytes(vset_->options_)) {
           break;
         }
       }
@@ -528,10 +537,10 @@ void Version::GetOverlappingInputs(
   assert(level < config::kNumLevels);
   inputs->clear();
   Slice user_begin, user_end;
-  if (begin != NULL) {
+  if (begin != nullptr) {
     user_begin = begin->user_key();
   }
-  if (end != NULL) {
+  if (end != nullptr) {
     user_end = end->user_key();
   }
   const Comparator* user_cmp = vset_->icmp_.user_comparator();
@@ -539,20 +548,21 @@ void Version::GetOverlappingInputs(
     FileMetaData* f = files_[level][i++];
     const Slice file_start = f->smallest.user_key();
     const Slice file_limit = f->largest.user_key();
-    if (begin != NULL && user_cmp->Compare(file_limit, user_begin) < 0) {
+    if (begin != nullptr && user_cmp->Compare(file_limit, user_begin) < 0) {
       // "f" is completely before specified range; skip it
-    } else if (end != NULL && user_cmp->Compare(file_start, user_end) > 0) {
+    } else if (end != nullptr && user_cmp->Compare(file_start, user_end) > 0) {
       // "f" is completely after specified range; skip it
     } else {
       inputs->push_back(f);
       if (level == 0) {
         // Level-0 files may overlap each other.  So check if the newly
         // added file has expanded the range.  If so, restart search.
-        if (begin != NULL && user_cmp->Compare(file_start, user_begin) < 0) {
+        if (begin != nullptr && user_cmp->Compare(file_start, user_begin) < 0) {
           user_begin = file_start;
           inputs->clear();
           i = 0;
-        } else if (end != NULL && user_cmp->Compare(file_limit, user_end) > 0) {
+        } else if (end != nullptr && user_cmp->Compare(file_limit,
+                                                       user_end) > 0) {
           user_end = file_limit;
           inputs->clear();
           i = 0;
@@ -690,7 +700,7 @@ class VersionSet::Builder {
       // same as the compaction of 40KB of data.  We are a little
       // conservative and allow approximately one seek for every 16KB
       // of data before triggering a compaction.
-      f->allowed_seeks = (f->file_size / 16384);
+      f->allowed_seeks = static_cast<int>((f->file_size / 16384U));
       if (f->allowed_seeks < 100) f->allowed_seeks = 100;
 
       levels_[level].deleted_files.erase(f->number);
@@ -777,10 +787,10 @@ VersionSet::VersionSet(const std::string& dbname,
       last_sequence_(0),
       log_number_(0),
       prev_log_number_(0),
-      descriptor_file_(NULL),
-      descriptor_log_(NULL),
+      descriptor_file_(nullptr),
+      descriptor_log_(nullptr),
       dummy_versions_(this),
-      current_(NULL) {
+      current_(nullptr) {
   AppendVersion(new Version(this));
 }
 
@@ -795,7 +805,7 @@ void VersionSet::AppendVersion(Version* v) {
   // Make "v" current
   assert(v->refs_ == 0);
   assert(v != current_);
-  if (current_ != NULL) {
+  if (current_ != nullptr) {
     current_->Unref();
   }
   current_ = v;
@@ -835,10 +845,10 @@ Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) {
   // a temporary file that contains a snapshot of the current version.
   std::string new_manifest_file;
   Status s;
-  if (descriptor_log_ == NULL) {
+  if (descriptor_log_ == nullptr) {
     // No reason to unlock *mu here since we only hit this path in the
     // first call to LogAndApply (when opening the database).
-    assert(descriptor_file_ == NULL);
+    assert(descriptor_file_ == nullptr);
     new_manifest_file = DescriptorFileName(dbname_, manifest_file_number_);
     edit->SetNextFile(next_file_number_);
     s = env_->NewWritableFile(new_manifest_file, &descriptor_file_);
@@ -884,8 +894,8 @@ Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) {
     if (!new_manifest_file.empty()) {
       delete descriptor_log_;
       delete descriptor_file_;
-      descriptor_log_ = NULL;
-      descriptor_file_ = NULL;
+      descriptor_log_ = nullptr;
+      descriptor_file_ = nullptr;
       env_->DeleteFile(new_manifest_file);
     }
   }
@@ -893,7 +903,7 @@ Status VersionSet::LogAndApply(VersionEdit* edit, port::Mutex* mu) {
   return s;
 }
 
-Status VersionSet::Recover() {
+Status VersionSet::Recover(bool *save_manifest) {
   struct LogReporter : public log::Reader::Reporter {
     Status* status;
     virtual void Corruption(size_t bytes, const Status& s) {
@@ -916,6 +926,10 @@ Status VersionSet::Recover() {
   SequentialFile* file;
   s = env_->NewSequentialFile(dscname, &file);
   if (!s.ok()) {
+    if (s.IsNotFound()) {
+      return Status::Corruption(
+            "CURRENT points to a non-existent file", s.ToString());
+    }
     return s;
   }
 
@@ -973,7 +987,7 @@ Status VersionSet::Recover() {
     }
   }
   delete file;
-  file = NULL;
+  file = nullptr;
 
   if (s.ok()) {
     if (!have_next_file) {
@@ -1003,11 +1017,49 @@ Status VersionSet::Recover() {
     last_sequence_ = last_sequence;
     log_number_ = log_number;
     prev_log_number_ = prev_log_number;
+
+    // See if we can reuse the existing MANIFEST file.
+    if (ReuseManifest(dscname, current)) {
+      // No need to save new manifest
+    } else {
+      *save_manifest = true;
+    }
   }
 
   return s;
 }
 
+bool VersionSet::ReuseManifest(const std::string& dscname,
+                               const std::string& dscbase) {
+  if (!options_->reuse_logs) {
+    return false;
+  }
+  FileType manifest_type;
+  uint64_t manifest_number;
+  uint64_t manifest_size;
+  if (!ParseFileName(dscbase, &manifest_number, &manifest_type) ||
+      manifest_type != kDescriptorFile ||
+      !env_->GetFileSize(dscname, &manifest_size).ok() ||
+      // Make new compacted MANIFEST if old one is too big
+      manifest_size >= TargetFileSize(options_)) {
+    return false;
+  }
+
+  assert(descriptor_file_ == nullptr);
+  assert(descriptor_log_ == nullptr);
+  Status r = env_->NewAppendableFile(dscname, &descriptor_file_);
+  if (!r.ok()) {
+    Log(options_->info_log, "Reuse MANIFEST: %s\n", r.ToString().c_str());
+    assert(descriptor_file_ == nullptr);
+    return false;
+  }
+
+  Log(options_->info_log, "Reusing MANIFEST %s\n", dscname.c_str());
+  descriptor_log_ = new log::Writer(descriptor_file_, manifest_size);
+  manifest_file_number_ = manifest_number;
+  return true;
+}
+
 void VersionSet::MarkFileNumberUsed(uint64_t number) {
   if (next_file_number_ <= number) {
     next_file_number_ = number + 1;
@@ -1038,7 +1090,8 @@ void VersionSet::Finalize(Version* v) {
     } 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(level);
+      score =
+          static_cast<double>(level_bytes) / MaxBytesForLevel(options_, level);
     }
 
     if (score > best_score) {
@@ -1124,7 +1177,7 @@ uint64_t VersionSet::ApproximateOffsetOf(Version* v, const InternalKey& ikey) {
         Table* tableptr;
         Iterator* iter = table_cache_->NewIterator(
             ReadOptions(), files[i]->number, files[i]->file_size, &tableptr);
-        if (tableptr != NULL) {
+        if (tableptr != nullptr) {
           result += tableptr->ApproximateOffsetOf(ikey.Encode());
         }
         delete iter;
@@ -1247,12 +1300,12 @@ Compaction* VersionSet::PickCompaction() {
   // We prefer compactions triggered by too much data in a level over
   // the compactions triggered by seeks.
   const bool size_compaction = (current_->compaction_score_ >= 1);
-  const bool seek_compaction = (current_->file_to_compact_ != NULL);
+  const bool seek_compaction = (current_->file_to_compact_ != nullptr);
   if (size_compaction) {
     level = current_->compaction_level_;
     assert(level >= 0);
     assert(level+1 < config::kNumLevels);
-    c = new Compaction(level);
+    c = new Compaction(options_, level);
 
     // Pick the first file that comes after compact_pointer_[level]
     for (size_t i = 0; i < current_->files_[level].size(); i++) {
@@ -1269,10 +1322,10 @@ Compaction* VersionSet::PickCompaction() {
     }
   } else if (seek_compaction) {
     level = current_->file_to_compact_level_;
-    c = new Compaction(level);
+    c = new Compaction(options_, level);
     c->inputs_[0].push_back(current_->file_to_compact_);
   } else {
-    return NULL;
+    return nullptr;
   }
 
   c->input_version_ = current_;
@@ -1294,9 +1347,90 @@ Compaction* VersionSet::PickCompaction() {
   return c;
 }
 
+// Finds the largest key in a vector of files. Returns true if files it not
+// empty.
+bool FindLargestKey(const InternalKeyComparator& icmp,
+                    const std::vector<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]);
@@ -1310,11 +1444,13 @@ void VersionSet::SetupOtherInputs(Compaction* c) {
   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 < kExpandedCompactionByteSizeLimit) {
+        inputs1_size + expanded0_size <
+            ExpandedCompactionByteSizeLimit(options_)) {
       InternalKey new_start, new_limit;
       GetRange(expanded0, &new_start, &new_limit);
       std::vector<FileMetaData*> expanded1;
@@ -1346,13 +1482,6 @@ void VersionSet::SetupOtherInputs(Compaction* c) {
                                    &c->grandparents_);
   }
 
-  if (false) {
-    Log(options_->info_log, "Compacting %d '%s' .. '%s'",
-        level,
-        smallest.DebugString().c_str(),
-        largest.DebugString().c_str());
-  }
-
   // Update the place where we will do the next compaction for this level.
   // We update this immediately instead of waiting for the VersionEdit
   // to be applied so that if the compaction fails, we will try a different
@@ -1368,7 +1497,7 @@ Compaction* VersionSet::CompactRange(
   std::vector<FileMetaData*> inputs;
   current_->GetOverlappingInputs(level, begin, end, &inputs);
   if (inputs.empty()) {
-    return NULL;
+    return nullptr;
   }
 
   // Avoid compacting too much in one shot in case the range is large.
@@ -1376,7 +1505,7 @@ Compaction* VersionSet::CompactRange(
   // and we must not pick one file and drop another older file if the
   // two files overlap.
   if (level > 0) {
-    const uint64_t limit = MaxFileSizeForLevel(level);
+    const uint64_t limit = MaxFileSizeForLevel(options_, level);
     uint64_t total = 0;
     for (size_t i = 0; i < inputs.size(); i++) {
       uint64_t s = inputs[i]->file_size;
@@ -1388,7 +1517,7 @@ Compaction* VersionSet::CompactRange(
     }
   }
 
-  Compaction* c = new Compaction(level);
+  Compaction* c = new Compaction(options_, level);
   c->input_version_ = current_;
   c->input_version_->Ref();
   c->inputs_[0] = inputs;
@@ -1396,10 +1525,10 @@ Compaction* VersionSet::CompactRange(
   return c;
 }
 
-Compaction::Compaction(int level)
+Compaction::Compaction(const Options* options, int level)
     : level_(level),
-      max_output_file_size_(MaxFileSizeForLevel(level)),
-      input_version_(NULL),
+      max_output_file_size_(MaxFileSizeForLevel(options, level)),
+      input_version_(nullptr),
       grandparent_index_(0),
       seen_key_(false),
       overlapped_bytes_(0) {
@@ -1409,18 +1538,19 @@ Compaction::Compaction(int level)
 }
 
 Compaction::~Compaction() {
-  if (input_version_ != NULL) {
+  if (input_version_ != nullptr) {
     input_version_->Unref();
   }
 }
 
 bool Compaction::IsTrivialMove() const {
+  const VersionSet* vset = input_version_->vset_;
   // Avoid a move if there is lots of overlapping grandparent data.
   // Otherwise, the move could create a parent file that will require
   // a very expensive merge later on.
-  return (num_input_files(0) == 1 &&
-          num_input_files(1) == 0 &&
-          TotalFileSize(grandparents_) <= kMaxGrandParentOverlapBytes);
+  return (num_input_files(0) == 1 && num_input_files(1) == 0 &&
+          TotalFileSize(grandparents_) <=
+              MaxGrandParentOverlapBytes(vset->options_));
 }
 
 void Compaction::AddInputDeletions(VersionEdit* edit) {
@@ -1453,8 +1583,9 @@ bool Compaction::IsBaseLevelForKey(const Slice& user_key) {
 }
 
 bool Compaction::ShouldStopBefore(const Slice& internal_key) {
+  const VersionSet* vset = input_version_->vset_;
   // Scan to find earliest grandparent file that contains key.
-  const InternalKeyComparator* icmp = &input_version_->vset_->icmp_;
+  const InternalKeyComparator* icmp = &vset->icmp_;
   while (grandparent_index_ < grandparents_.size() &&
       icmp->Compare(internal_key,
                     grandparents_[grandparent_index_]->largest.Encode()) > 0) {
@@ -1465,7 +1596,7 @@ bool Compaction::ShouldStopBefore(const Slice& internal_key) {
   }
   seen_key_ = true;
 
-  if (overlapped_bytes_ > kMaxGrandParentOverlapBytes) {
+  if (overlapped_bytes_ > MaxGrandParentOverlapBytes(vset->options_)) {
     // Too much overlap for current output; start new output
     overlapped_bytes_ = 0;
     return true;
@@ -1475,9 +1606,9 @@ bool Compaction::ShouldStopBefore(const Slice& internal_key) {
 }
 
 void Compaction::ReleaseInputs() {
-  if (input_version_ != NULL) {
+  if (input_version_ != nullptr) {
     input_version_->Unref();
-    input_version_ = NULL;
+    input_version_ = nullptr;
   }
 }
 

db/version_set.h

@@ -39,22 +39,21 @@ class WritableFile;
 // Return the smallest index i such that files[i]->largest >= key.
 // Return files.size() if there is no such file.
 // REQUIRES: "files" contains a sorted list of non-overlapping files.
-extern int FindFile(const InternalKeyComparator& icmp,
-                    const std::vector<FileMetaData*>& files,
-                    const Slice& key);
+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==NULL represents a key smaller than all keys in the DB.
-// largest==NULL represents a key largest than all keys in the DB.
+// smallest==nullptr represents a key smaller than all keys in the DB.
+// largest==nullptr represents a key largest than all keys in the DB.
 // REQUIRES: If disjoint_sorted_files, files[] contains disjoint ranges
 //           in sorted order.
-extern bool SomeFileOverlapsRange(
-    const InternalKeyComparator& icmp,
-    bool disjoint_sorted_files,
-    const std::vector<FileMetaData*>& files,
-    const Slice* smallest_user_key,
-    const Slice* largest_user_key);
+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:
@@ -91,14 +90,14 @@ class Version {
 
   void GetOverlappingInputs(
       int level,
-      const InternalKey* begin,         // NULL means before all keys
-      const InternalKey* end,           // NULL means after all keys
+      const InternalKey* begin,         // nullptr means before all keys
+      const InternalKey* end,           // nullptr means after all keys
       std::vector<FileMetaData*>* inputs);
 
   // Returns true iff some file in the specified level overlaps
   // some part of [*smallest_user_key,*largest_user_key].
-  // smallest_user_key==NULL represents a key smaller than all keys in the DB.
-  // largest_user_key==NULL represents a key largest than all keys in the DB.
+  // 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);
@@ -149,7 +148,7 @@ class Version {
 
   explicit Version(VersionSet* vset)
       : vset_(vset), next_(this), prev_(this), refs_(0),
-        file_to_compact_(NULL),
+        file_to_compact_(nullptr),
         file_to_compact_level_(-1),
         compaction_score_(-1),
         compaction_level_(-1) {
@@ -179,7 +178,7 @@ class VersionSet {
       EXCLUSIVE_LOCKS_REQUIRED(mu);
 
   // Recover the last saved descriptor from persistent storage.
-  Status Recover();
+  Status Recover(bool *save_manifest);
 
   // Return the current version.
   Version* current() const { return current_; }
@@ -225,13 +224,13 @@ class VersionSet {
   uint64_t PrevLogNumber() const { return prev_log_number_; }
 
   // Pick level and inputs for a new compaction.
-  // Returns NULL if there is no compaction to be done.
+  // Returns nullptr if there is no compaction to be done.
   // Otherwise returns a pointer to a heap-allocated object that
   // describes the compaction.  Caller should delete the result.
   Compaction* PickCompaction();
 
   // Return a compaction object for compacting the range [begin,end] in
-  // the specified level.  Returns NULL if there is nothing in that
+  // the specified level.  Returns nullptr if there is nothing in that
   // level that overlaps the specified range.  Caller should delete
   // the result.
   Compaction* CompactRange(
@@ -250,7 +249,7 @@ class VersionSet {
   // Returns true iff some level needs a compaction.
   bool NeedsCompaction() const {
     Version* v = current_;
-    return (v->compaction_score_ >= 1) || (v->file_to_compact_ != NULL);
+    return (v->compaction_score_ >= 1) || (v->file_to_compact_ != nullptr);
   }
 
   // Add all files listed in any live version to *live.
@@ -274,6 +273,8 @@ class VersionSet {
   friend class Compaction;
   friend class Version;
 
+  bool ReuseManifest(const std::string& dscname, const std::string& dscbase);
+
   void Finalize(Version* v);
 
   void GetRange(const std::vector<FileMetaData*>& inputs,
@@ -364,7 +365,7 @@ class Compaction {
   friend class Version;
   friend class VersionSet;
 
-  explicit Compaction(int level);
+  Compaction(const Options* options, int level);
 
   int level_;
   uint64_t max_output_file_size_;
@@ -374,7 +375,7 @@ class Compaction {
   // 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 of overlapping grandparent files
+  // 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_

db/version_set_test.cc

@@ -40,20 +40,20 @@ class FindFileTest {
 
   bool Overlaps(const char* smallest, const char* largest) {
     InternalKeyComparator cmp(BytewiseComparator());
-    Slice s(smallest != NULL ? smallest : "");
-    Slice l(largest != NULL ? largest : "");
+    Slice s(smallest != nullptr ? smallest : "");
+    Slice l(largest != nullptr ? largest : "");
     return SomeFileOverlapsRange(cmp, disjoint_sorted_files_, files_,
-                                 (smallest != NULL ? &s : NULL),
-                                 (largest != NULL ? &l : NULL));
+                                 (smallest != nullptr ? &s : nullptr),
+                                 (largest != nullptr ? &l : nullptr));
   }
 };
 
 TEST(FindFileTest, Empty) {
   ASSERT_EQ(0, Find("foo"));
   ASSERT_TRUE(! Overlaps("a", "z"));
-  ASSERT_TRUE(! Overlaps(NULL, "z"));
-  ASSERT_TRUE(! Overlaps("a", NULL));
-  ASSERT_TRUE(! Overlaps(NULL, NULL));
+  ASSERT_TRUE(! Overlaps(nullptr, "z"));
+  ASSERT_TRUE(! Overlaps("a", nullptr));
+  ASSERT_TRUE(! Overlaps(nullptr, nullptr));
 }
 
 TEST(FindFileTest, Single) {
@@ -78,12 +78,12 @@ TEST(FindFileTest, Single) {
   ASSERT_TRUE(Overlaps("q", "q"));
   ASSERT_TRUE(Overlaps("q", "q1"));
 
-  ASSERT_TRUE(! Overlaps(NULL, "j"));
-  ASSERT_TRUE(! Overlaps("r", NULL));
-  ASSERT_TRUE(Overlaps(NULL, "p"));
-  ASSERT_TRUE(Overlaps(NULL, "p1"));
-  ASSERT_TRUE(Overlaps("q", NULL));
-  ASSERT_TRUE(Overlaps(NULL, NULL));
+  ASSERT_TRUE(! Overlaps(nullptr, "j"));
+  ASSERT_TRUE(! Overlaps("r", nullptr));
+  ASSERT_TRUE(Overlaps(nullptr, "p"));
+  ASSERT_TRUE(Overlaps(nullptr, "p1"));
+  ASSERT_TRUE(Overlaps("q", nullptr));
+  ASSERT_TRUE(Overlaps(nullptr, nullptr));
 }
 
 
@@ -130,19 +130,19 @@ TEST(FindFileTest, MultipleNullBoundaries) {
   Add("200", "250");
   Add("300", "350");
   Add("400", "450");
-  ASSERT_TRUE(! Overlaps(NULL, "149"));
-  ASSERT_TRUE(! Overlaps("451", NULL));
-  ASSERT_TRUE(Overlaps(NULL, NULL));
-  ASSERT_TRUE(Overlaps(NULL, "150"));
-  ASSERT_TRUE(Overlaps(NULL, "199"));
-  ASSERT_TRUE(Overlaps(NULL, "200"));
-  ASSERT_TRUE(Overlaps(NULL, "201"));
-  ASSERT_TRUE(Overlaps(NULL, "400"));
-  ASSERT_TRUE(Overlaps(NULL, "800"));
-  ASSERT_TRUE(Overlaps("100", NULL));
-  ASSERT_TRUE(Overlaps("200", NULL));
-  ASSERT_TRUE(Overlaps("449", NULL));
-  ASSERT_TRUE(Overlaps("450", NULL));
+  ASSERT_TRUE(! Overlaps(nullptr, "149"));
+  ASSERT_TRUE(! Overlaps("451", nullptr));
+  ASSERT_TRUE(Overlaps(nullptr, nullptr));
+  ASSERT_TRUE(Overlaps(nullptr, "150"));
+  ASSERT_TRUE(Overlaps(nullptr, "199"));
+  ASSERT_TRUE(Overlaps(nullptr, "200"));
+  ASSERT_TRUE(Overlaps(nullptr, "201"));
+  ASSERT_TRUE(Overlaps(nullptr, "400"));
+  ASSERT_TRUE(Overlaps(nullptr, "800"));
+  ASSERT_TRUE(Overlaps("100", nullptr));
+  ASSERT_TRUE(Overlaps("200", nullptr));
+  ASSERT_TRUE(Overlaps("449", nullptr));
+  ASSERT_TRUE(Overlaps("450", nullptr));
 }
 
 TEST(FindFileTest, OverlapSequenceChecks) {
@@ -172,8 +172,160 @@ TEST(FindFileTest, OverlappingFiles) {
   ASSERT_TRUE(Overlaps("600", "700"));
 }
 
+void AddBoundaryInputs(const InternalKeyComparator& icmp,
+                       const std::vector<FileMetaData*>& level_files,
+                       std::vector<FileMetaData*>* compaction_files);
+
+class AddBoundaryInputsTest {
+ 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(AddBoundaryInputsTest, TestEmptyFileSets) {
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_TRUE(compaction_files_.empty());
+  ASSERT_TRUE(level_files_.empty());
+}
+
+TEST(AddBoundaryInputsTest, TestEmptyLevelFiles) {
+  FileMetaData* f1 =
+      CreateFileMetaData(1, InternalKey("100", 2, kTypeValue),
+                         InternalKey(InternalKey("100", 1, kTypeValue)));
+  compaction_files_.push_back(f1);
+
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_EQ(1, compaction_files_.size());
+  ASSERT_EQ(f1, compaction_files_[0]);
+  ASSERT_TRUE(level_files_.empty());
+}
+
+TEST(AddBoundaryInputsTest, TestEmptyCompactionFiles) {
+  FileMetaData* f1 =
+      CreateFileMetaData(1, InternalKey("100", 2, kTypeValue),
+                         InternalKey(InternalKey("100", 1, kTypeValue)));
+  level_files_.push_back(f1);
+
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_TRUE(compaction_files_.empty());
+  ASSERT_EQ(1, level_files_.size());
+  ASSERT_EQ(f1, level_files_[0]);
+}
+
+TEST(AddBoundaryInputsTest, TestNoBoundaryFiles) {
+  FileMetaData* f1 =
+      CreateFileMetaData(1, InternalKey("100", 2, kTypeValue),
+                         InternalKey(InternalKey("100", 1, kTypeValue)));
+  FileMetaData* f2 =
+      CreateFileMetaData(1, InternalKey("200", 2, kTypeValue),
+                         InternalKey(InternalKey("200", 1, kTypeValue)));
+  FileMetaData* f3 =
+      CreateFileMetaData(1, InternalKey("300", 2, kTypeValue),
+                         InternalKey(InternalKey("300", 1, kTypeValue)));
+
+  level_files_.push_back(f3);
+  level_files_.push_back(f2);
+  level_files_.push_back(f1);
+  compaction_files_.push_back(f2);
+  compaction_files_.push_back(f3);
+
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_EQ(2, compaction_files_.size());
+}
+
+TEST(AddBoundaryInputsTest, TestOneBoundaryFiles) {
+  FileMetaData* f1 =
+      CreateFileMetaData(1, InternalKey("100", 3, kTypeValue),
+                         InternalKey(InternalKey("100", 2, kTypeValue)));
+  FileMetaData* f2 =
+      CreateFileMetaData(1, InternalKey("100", 1, kTypeValue),
+                         InternalKey(InternalKey("200", 3, kTypeValue)));
+  FileMetaData* f3 =
+      CreateFileMetaData(1, InternalKey("300", 2, kTypeValue),
+                         InternalKey(InternalKey("300", 1, kTypeValue)));
+
+  level_files_.push_back(f3);
+  level_files_.push_back(f2);
+  level_files_.push_back(f1);
+  compaction_files_.push_back(f1);
+
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_EQ(2, compaction_files_.size());
+  ASSERT_EQ(f1, compaction_files_[0]);
+  ASSERT_EQ(f2, compaction_files_[1]);
+}
+
+TEST(AddBoundaryInputsTest, TestTwoBoundaryFiles) {
+  FileMetaData* f1 =
+      CreateFileMetaData(1, InternalKey("100", 6, kTypeValue),
+                         InternalKey(InternalKey("100", 5, kTypeValue)));
+  FileMetaData* f2 =
+      CreateFileMetaData(1, InternalKey("100", 2, kTypeValue),
+                         InternalKey(InternalKey("300", 1, kTypeValue)));
+  FileMetaData* f3 =
+      CreateFileMetaData(1, InternalKey("100", 4, kTypeValue),
+                         InternalKey(InternalKey("100", 3, kTypeValue)));
+
+  level_files_.push_back(f2);
+  level_files_.push_back(f3);
+  level_files_.push_back(f1);
+  compaction_files_.push_back(f1);
+
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_EQ(3, compaction_files_.size());
+  ASSERT_EQ(f1, compaction_files_[0]);
+  ASSERT_EQ(f3, compaction_files_[1]);
+  ASSERT_EQ(f2, compaction_files_[2]);
+}
+
+TEST(AddBoundaryInputsTest, TestDisjoinFilePointers) {
+  FileMetaData* f1 =
+      CreateFileMetaData(1, InternalKey("100", 6, kTypeValue),
+                         InternalKey(InternalKey("100", 5, kTypeValue)));
+  FileMetaData* f2 =
+      CreateFileMetaData(1, InternalKey("100", 6, kTypeValue),
+                         InternalKey(InternalKey("100", 5, kTypeValue)));
+  FileMetaData* f3 =
+      CreateFileMetaData(1, InternalKey("100", 2, kTypeValue),
+                         InternalKey(InternalKey("300", 1, kTypeValue)));
+  FileMetaData* f4 =
+      CreateFileMetaData(1, InternalKey("100", 4, kTypeValue),
+                         InternalKey(InternalKey("100", 3, kTypeValue)));
+
+  level_files_.push_back(f2);
+  level_files_.push_back(f3);
+  level_files_.push_back(f4);
+
+  compaction_files_.push_back(f1);
+
+  AddBoundaryInputs(icmp_, level_files_, &compaction_files_);
+  ASSERT_EQ(3, compaction_files_.size());
+  ASSERT_EQ(f1, compaction_files_[0]);
+  ASSERT_EQ(f4, compaction_files_[1]);
+  ASSERT_EQ(f3, compaction_files_[2]);
+}
+
 }  // namespace leveldb
 
-int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
-}
+int main(int argc, char** argv) { return leveldb::test::RunAllTests(); }

db/write_batch.cc

@@ -39,6 +39,10 @@ void WriteBatch::Clear() {
   rep_.resize(kHeader);
 }
 
+size_t WriteBatch::ApproximateSize() const {
+  return rep_.size();
+}
+
 Status WriteBatch::Iterate(Handler* handler) const {
   Slice input(rep_);
   if (input.size() < kHeader) {
@@ -108,6 +112,10 @@ void WriteBatch::Delete(const Slice& key) {
   PutLengthPrefixedSlice(&rep_, key);
 }
 
+void WriteBatch::Append(const WriteBatch &source) {
+  WriteBatchInternal::Append(this, &source);
+}
+
 namespace {
 class MemTableInserter : public WriteBatch::Handler {
  public:

db/write_batch_internal.h

@@ -5,6 +5,7 @@
 #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 {

db/write_batch_test.cc

@@ -91,21 +91,21 @@ TEST(WriteBatchTest, Append) {
   WriteBatch b1, b2;
   WriteBatchInternal::SetSequence(&b1, 200);
   WriteBatchInternal::SetSequence(&b2, 300);
-  WriteBatchInternal::Append(&b1, &b2);
+  b1.Append(b2);
   ASSERT_EQ("",
             PrintContents(&b1));
   b2.Put("a", "va");
-  WriteBatchInternal::Append(&b1, &b2);
+  b1.Append(b2);
   ASSERT_EQ("Put(a, va)@200",
             PrintContents(&b1));
   b2.Clear();
   b2.Put("b", "vb");
-  WriteBatchInternal::Append(&b1, &b2);
+  b1.Append(b2);
   ASSERT_EQ("Put(a, va)@200"
             "Put(b, vb)@201",
             PrintContents(&b1));
   b2.Delete("foo");
-  WriteBatchInternal::Append(&b1, &b2);
+  b1.Append(b2);
   ASSERT_EQ("Put(a, va)@200"
             "Put(b, vb)@202"
             "Put(b, vb)@201"
@@ -113,6 +113,23 @@ TEST(WriteBatchTest, Append) {
             PrintContents(&b1));
 }
 
+TEST(WriteBatchTest, ApproximateSize) {
+  WriteBatch batch;
+  size_t empty_size = batch.ApproximateSize();
+
+  batch.Put(Slice("foo"), Slice("bar"));
+  size_t one_key_size = batch.ApproximateSize();
+  ASSERT_LT(empty_size, one_key_size);
+
+  batch.Put(Slice("baz"), Slice("boo"));
+  size_t two_keys_size = batch.ApproximateSize();
+  ASSERT_LT(one_key_size, two_keys_size);
+
+  batch.Delete(Slice("box"));
+  size_t post_delete_size = batch.ApproximateSize();
+  ASSERT_LT(two_keys_size, post_delete_size);
+}
+
 }  // namespace leveldb
 
 int main(int argc, char** argv) {

doc/bench/db_bench_sqlite3.cc

@@ -76,7 +76,7 @@ static bool FLAGS_transaction = true;
 static bool FLAGS_WAL_enabled = true;
 
 // Use the db with the following name.
-static const char* FLAGS_db = NULL;
+static const char* FLAGS_db = nullptr;
 
 inline
 static void ExecErrorCheck(int status, char *err_msg) {
@@ -107,7 +107,8 @@ inline
 static void WalCheckpoint(sqlite3* db_) {
   // Flush all writes to disk
   if (FLAGS_WAL_enabled) {
-    sqlite3_wal_checkpoint_v2(db_, NULL, SQLITE_CHECKPOINT_FULL, NULL, NULL);
+    sqlite3_wal_checkpoint_v2(db_, nullptr, SQLITE_CHECKPOINT_FULL, nullptr,
+                              nullptr);
   }
 }
 
@@ -207,18 +208,18 @@ class Benchmark {
     fprintf(stderr, "SQLite:     version %s\n", SQLITE_VERSION);
 
 #if defined(__linux)
-    time_t now = time(NULL);
+    time_t now = time(nullptr);
     fprintf(stderr, "Date:       %s", ctime(&now));  // ctime() adds newline
 
     FILE* cpuinfo = fopen("/proc/cpuinfo", "r");
-    if (cpuinfo != NULL) {
+    if (cpuinfo != nullptr) {
       char line[1000];
       int num_cpus = 0;
       std::string cpu_type;
       std::string cache_size;
-      while (fgets(line, sizeof(line), cpuinfo) != NULL) {
+      while (fgets(line, sizeof(line), cpuinfo) != nullptr) {
         const char* sep = strchr(line, ':');
-        if (sep == NULL) {
+        if (sep == nullptr) {
           continue;
         }
         Slice key = TrimSpace(Slice(line, sep - 1 - line));
@@ -313,7 +314,7 @@ class Benchmark {
   };
 
   Benchmark()
-  : db_(NULL),
+  : db_(nullptr),
     db_num_(0),
     num_(FLAGS_num),
     reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads),
@@ -345,12 +346,12 @@ class Benchmark {
     Open();
 
     const char* benchmarks = FLAGS_benchmarks;
-    while (benchmarks != NULL) {
+    while (benchmarks != nullptr) {
       const char* sep = strchr(benchmarks, ',');
       Slice name;
-      if (sep == NULL) {
+      if (sep == nullptr) {
         name = benchmarks;
-        benchmarks = NULL;
+        benchmarks = nullptr;
       } else {
         name = Slice(benchmarks, sep - benchmarks);
         benchmarks = sep + 1;
@@ -415,11 +416,11 @@ class Benchmark {
   }
 
   void Open() {
-    assert(db_ == NULL);
+    assert(db_ == nullptr);
 
     int status;
     char file_name[100];
-    char* err_msg = NULL;
+    char* err_msg = nullptr;
     db_num_++;
 
     // Open database
@@ -439,7 +440,7 @@ class Benchmark {
     char cache_size[100];
     snprintf(cache_size, sizeof(cache_size), "PRAGMA cache_size = %d",
              FLAGS_num_pages);
-    status = sqlite3_exec(db_, cache_size, NULL, NULL, &err_msg);
+    status = sqlite3_exec(db_, cache_size, nullptr, nullptr, &err_msg);
     ExecErrorCheck(status, err_msg);
 
     // FLAGS_page_size is defaulted to 1024
@@ -447,7 +448,7 @@ class Benchmark {
       char page_size[100];
       snprintf(page_size, sizeof(page_size), "PRAGMA page_size = %d",
                FLAGS_page_size);
-      status = sqlite3_exec(db_, page_size, NULL, NULL, &err_msg);
+      status = sqlite3_exec(db_, page_size, nullptr, nullptr, &err_msg);
       ExecErrorCheck(status, err_msg);
     }
 
@@ -457,9 +458,10 @@ class Benchmark {
 
       // LevelDB's default cache size is a combined 4 MB
       std::string WAL_checkpoint = "PRAGMA wal_autocheckpoint = 4096";
-      status = sqlite3_exec(db_, WAL_stmt.c_str(), NULL, NULL, &err_msg);
+      status = sqlite3_exec(db_, WAL_stmt.c_str(), nullptr, nullptr, &err_msg);
       ExecErrorCheck(status, err_msg);
-      status = sqlite3_exec(db_, WAL_checkpoint.c_str(), NULL, NULL, &err_msg);
+      status = sqlite3_exec(db_, WAL_checkpoint.c_str(), nullptr, nullptr,
+                            &err_msg);
       ExecErrorCheck(status, err_msg);
     }
 
@@ -470,7 +472,8 @@ class Benchmark {
     std::string stmt_array[] = { locking_stmt, create_stmt };
     int stmt_array_length = sizeof(stmt_array) / sizeof(std::string);
     for (int i = 0; i < stmt_array_length; i++) {
-      status = sqlite3_exec(db_, stmt_array[i].c_str(), NULL, NULL, &err_msg);
+      status = sqlite3_exec(db_, stmt_array[i].c_str(), nullptr, nullptr,
+                            &err_msg);
       ExecErrorCheck(status, err_msg);
     }
   }
@@ -484,7 +487,7 @@ class Benchmark {
         return;
       }
       sqlite3_close(db_);
-      db_ = NULL;
+      db_ = nullptr;
       Open();
       Start();
     }
@@ -495,7 +498,7 @@ class Benchmark {
       message_ = msg;
     }
 
-    char* err_msg = NULL;
+    char* err_msg = nullptr;
     int status;
 
     sqlite3_stmt *replace_stmt, *begin_trans_stmt, *end_trans_stmt;
@@ -506,18 +509,18 @@ class Benchmark {
     // Check for synchronous flag in options
     std::string sync_stmt = (write_sync) ? "PRAGMA synchronous = FULL" :
                                            "PRAGMA synchronous = OFF";
-    status = sqlite3_exec(db_, sync_stmt.c_str(), NULL, NULL, &err_msg);
+    status = sqlite3_exec(db_, sync_stmt.c_str(), nullptr, nullptr, &err_msg);
     ExecErrorCheck(status, err_msg);
 
     // Preparing sqlite3 statements
     status = sqlite3_prepare_v2(db_, replace_str.c_str(), -1,
-                                &replace_stmt, NULL);
+                                &replace_stmt, nullptr);
     ErrorCheck(status);
     status = sqlite3_prepare_v2(db_, begin_trans_str.c_str(), -1,
-                                &begin_trans_stmt, NULL);
+                                &begin_trans_stmt, nullptr);
     ErrorCheck(status);
     status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1,
-                                &end_trans_stmt, NULL);
+                                &end_trans_stmt, nullptr);
     ErrorCheck(status);
 
     bool transaction = (entries_per_batch > 1);
@@ -588,12 +591,12 @@ class Benchmark {
 
     // Preparing sqlite3 statements
     status = sqlite3_prepare_v2(db_, begin_trans_str.c_str(), -1,
-                                &begin_trans_stmt, NULL);
+                                &begin_trans_stmt, nullptr);
     ErrorCheck(status);
     status = sqlite3_prepare_v2(db_, end_trans_str.c_str(), -1,
-                                &end_trans_stmt, NULL);
+                                &end_trans_stmt, nullptr);
     ErrorCheck(status);
-    status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &read_stmt, NULL);
+    status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &read_stmt, nullptr);
     ErrorCheck(status);
 
     bool transaction = (entries_per_batch > 1);
@@ -651,7 +654,7 @@ class Benchmark {
     sqlite3_stmt *pStmt;
     std::string read_str = "SELECT * FROM test ORDER BY key";
 
-    status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &pStmt, NULL);
+    status = sqlite3_prepare_v2(db_, read_str.c_str(), -1, &pStmt, nullptr);
     ErrorCheck(status);
     for (int i = 0; i < reads_ && SQLITE_ROW == sqlite3_step(pStmt); i++) {
       bytes_ += sqlite3_column_bytes(pStmt, 1) + sqlite3_column_bytes(pStmt, 2);
@@ -706,7 +709,7 @@ int main(int argc, char** argv) {
   }
 
   // Choose a location for the test database if none given with --db=<path>
-  if (FLAGS_db == NULL) {
+  if (FLAGS_db == nullptr) {
       leveldb::Env::Default()->GetTestDirectory(&default_db_path);
       default_db_path += "/dbbench";
       FLAGS_db = default_db_path.c_str();

doc/bench/db_bench_tree_db.cc

@@ -69,7 +69,7 @@ static bool FLAGS_use_existing_db = false;
 static bool FLAGS_compression = true;
 
 // Use the db with the following name.
-static const char* FLAGS_db = NULL;
+static const char* FLAGS_db = nullptr;
 
 inline
 static void DBSynchronize(kyotocabinet::TreeDB* db_)
@@ -183,18 +183,18 @@ class Benchmark {
             kyotocabinet::VERSION, kyotocabinet::LIBVER, kyotocabinet::LIBREV);
 
 #if defined(__linux)
-    time_t now = time(NULL);
+    time_t now = time(nullptr);
     fprintf(stderr, "Date:           %s", ctime(&now));  // ctime() adds newline
 
     FILE* cpuinfo = fopen("/proc/cpuinfo", "r");
-    if (cpuinfo != NULL) {
+    if (cpuinfo != nullptr) {
       char line[1000];
       int num_cpus = 0;
       std::string cpu_type;
       std::string cache_size;
-      while (fgets(line, sizeof(line), cpuinfo) != NULL) {
+      while (fgets(line, sizeof(line), cpuinfo) != nullptr) {
         const char* sep = strchr(line, ':');
-        if (sep == NULL) {
+        if (sep == nullptr) {
           continue;
         }
         Slice key = TrimSpace(Slice(line, sep - 1 - line));
@@ -289,7 +289,7 @@ class Benchmark {
   };
 
   Benchmark()
-  : db_(NULL),
+  : db_(nullptr),
     num_(FLAGS_num),
     reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads),
     bytes_(0),
@@ -321,12 +321,12 @@ class Benchmark {
     Open(false);
 
     const char* benchmarks = FLAGS_benchmarks;
-    while (benchmarks != NULL) {
+    while (benchmarks != nullptr) {
       const char* sep = strchr(benchmarks, ',');
       Slice name;
-      if (sep == NULL) {
+      if (sep == nullptr) {
         name = benchmarks;
-        benchmarks = NULL;
+        benchmarks = nullptr;
       } else {
         name = Slice(benchmarks, sep - benchmarks);
         benchmarks = sep + 1;
@@ -387,7 +387,7 @@ class Benchmark {
 
  private:
     void Open(bool sync) {
-    assert(db_ == NULL);
+    assert(db_ == nullptr);
 
     // Initialize db_
     db_ = new kyotocabinet::TreeDB();
@@ -430,7 +430,7 @@ class Benchmark {
         return;
       }
       delete db_;
-      db_ = NULL;
+      db_ = nullptr;
       Open(sync);
       Start();  // Do not count time taken to destroy/open
     }
@@ -516,7 +516,7 @@ int main(int argc, char** argv) {
   }
 
   // Choose a location for the test database if none given with --db=<path>
-  if (FLAGS_db == NULL) {
+  if (FLAGS_db == nullptr) {
       leveldb::Env::Default()->GetTestDirectory(&default_db_path);
       default_db_path += "/dbbench";
       FLAGS_db = default_db_path.c_str();

doc/doc.css

@@ -1,89 +0,0 @@
-body {
-  margin-left: 0.5in;
-  margin-right: 0.5in;
-  background: white;
-  color: black;
-}
-
-h1 {
-  margin-left: -0.2in;
-  font-size: 14pt;
-}
-h2 {
-  margin-left: -0in;
-  font-size: 12pt;
-}
-h3 {
-  margin-left: -0in;
-}
-h4 {
-  margin-left: -0in;
-}
-hr {
-  margin-left: -0in;
-}
-
-/* Definition lists: definition term bold */
-dt {
-  font-weight: bold;
-}
-
-address {
-  text-align: center;
-}
-code,samp,var {
-  color: blue;
-}
-kbd {
-  color: #600000;
-}
-div.note p {
-  float: right;
-  width: 3in;
-  margin-right: 0%;
-  padding: 1px;
-  border: 2px solid #6060a0;
-  background-color: #fffff0;
-}
-
-ul {
-  margin-top: -0em;
-  margin-bottom: -0em;
-}
-
-ol {
-  margin-top: -0em;
-  margin-bottom: -0em;
-}
-
-UL.nobullets {
-  list-style-type: none;
-  list-style-image: none;
-  margin-left: -1em;
-}
-
-p {
-  margin: 1em 0 1em 0;
-  padding: 0 0 0 0;
-}
-
-pre {
-  line-height: 1.3em;
-  padding: 0.4em 0 0.8em 0;
-  margin:  0 0 0 0;
-  border:  0 0 0 0;
-  color: blue;
-}
-
-.datatable {
-  margin-left: auto;
-  margin-right: auto;
-  margin-top: 2em;
-  margin-bottom: 2em;
-  border: 1px solid;
-}
-
-.datatable td,th {
-  padding: 0 0.5em 0 0.5em;
-  text-align: right;
-}

doc/impl.html

@@ -1,213 +0,0 @@
-<!DOCTYPE html>
-<html>
-<head>
-<link rel="stylesheet" type="text/css" href="doc.css" />
-<title>Leveldb file layout and compactions</title>
-</head>
-
-<body>
-
-<h1>Files</h1>
-
-The implementation of leveldb is similar in spirit to the
-representation of a single
-<a href="http://research.google.com/archive/bigtable.html">
-Bigtable tablet (section 5.3)</a>.
-However the organization of the files that make up the representation
-is somewhat different and is explained below.
-
-<p>
-Each database is represented by a set of files stored in a directory.
-There are several different types of files as documented below:
-<p>
-<h2>Log files</h2>
-<p>
-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.
-<p>
-A copy of the current log file is kept in an in-memory structure (the
-<code>memtable</code>).  This copy is consulted on every read so that read
-operations reflect all logged updates.
-<p>
-<h2>Sorted tables</h2>
-<p>
-A sorted table (*.sst) stores a sequence of entries sorted by key.
-Each entry is either a value for the key, or a deletion marker for the
-key.  (Deletion markers are kept around to hide obsolete values
-present in older sorted tables).
-<p>
-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 <code>young</code>
-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.)
-<p>
-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).
-
-<h2>Manifest</h2>
-<p>
-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.
-<p>
-<h2>Current</h2>
-<p>
-CURRENT is a simple text file that contains the name of the latest
-MANIFEST file.
-<p>
-<h2>Info logs</h2>
-<p>
-Informational messages are printed to files named LOG and LOG.old.
-<p>
-<h2>Others</h2>
-<p>
-Other files used for miscellaneous purposes may also be present
-(LOCK, *.dbtmp).
-
-<h1>Level 0</h1>
-When the log file grows above a certain size (1MB by default):
-<ul>
-<li>Create a brand new memtable and log file and direct future updates here
-<li>In the background:
-<ul>
-<li>Write the contents of the previous memtable to an sstable
-<li>Discard the memtable
-<li>Delete the old log file and the old memtable
-<li>Add the new sstable to the young (level-0) level.
-</ul>
-</ul>
-
-<h1>Compactions</h1>
-
-<p>
-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.
-
-<p>
-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).
-
-<p>
-The old files are discarded and the new files are added to the serving
-state.
-
-<p>
-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).
-
-<p>
-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.
-
-<h2>Timing</h2>
-
-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.
-
-<p>
-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.
-
-<p>
-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.
-
-<p>
-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.
-
-<p>
-Solution 2: We might want to decrease write rate artificially when the
-number of level-0 files goes up.
-
-<p>
-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.
-
-<h2>Number of files</h2>
-
-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.
-
-<p>
-An experiment on an <code>ext3</code> filesystem on Feb 04, 2011 shows
-the following timings to do 100K file opens in directories with
-varying number of files:
-<table class="datatable">
-<tr><th>Files in directory</th><th>Microseconds to open a file</th></tr>
-<tr><td>1000</td><td>9</td>
-<tr><td>10000</td><td>10</td>
-<tr><td>100000</td><td>16</td>
-</table>
-So maybe even the sharding is not necessary on modern filesystems?
-
-<h1>Recovery</h1>
-
-<ul>
-<li> Read CURRENT to find name of the latest committed MANIFEST
-<li> Read the named MANIFEST file
-<li> Clean up stale files
-<li> We could open all sstables here, but it is probably better to be lazy...
-<li> Convert log chunk to a new level-0 sstable
-<li> Start directing new writes to a new log file with recovered sequence#
-</ul>
-
-<h1>Garbage collection of files</h1>
-
-<code>DeleteObsoleteFiles()</code> 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.
-
-</body>
-</html>

doc/impl.md

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

doc/index.html

@@ -1,549 +0,0 @@
-<!DOCTYPE html>
-<html>
-<head>
-<link rel="stylesheet" type="text/css" href="doc.css" />
-<title>Leveldb</title>
-</head>
-
-<body>
-<h1>Leveldb</h1>
-<address>Jeff Dean, Sanjay Ghemawat</address>
-<p>
-The <code>leveldb</code> 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.
-
-<p>
-<h1>Opening A Database</h1>
-<p>
-A <code>leveldb</code> 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:
-<p>
-<pre>
-  #include &lt;assert&gt;
-  #include "leveldb/db.h"
-
-  leveldb::DB* db;
-  leveldb::Options options;
-  options.create_if_missing = true;
-  leveldb::Status status = leveldb::DB::Open(options, "/tmp/testdb", &amp;db);
-  assert(status.ok());
-  ...
-</pre>
-If you want to raise an error if the database already exists, add
-the following line before the <code>leveldb::DB::Open</code> call:
-<pre>
-  options.error_if_exists = true;
-</pre>
-<h1>Status</h1>
-<p>
-You may have noticed the <code>leveldb::Status</code> type above.  Values of this
-type are returned by most functions in <code>leveldb</code> that may encounter an
-error.  You can check if such a result is ok, and also print an
-associated error message:
-<p>
-<pre>
-   leveldb::Status s = ...;
-   if (!s.ok()) cerr &lt;&lt; s.ToString() &lt;&lt; endl;
-</pre>
-<h1>Closing A Database</h1>
-<p>
-When you are done with a database, just delete the database object.
-Example:
-<p>
-<pre>
-  ... open the db as described above ...
-  ... do something with db ...
-  delete db;
-</pre>
-<h1>Reads And Writes</h1>
-<p>
-The database provides <code>Put</code>, <code>Delete</code>, and <code>Get</code> methods to
-modify/query the database.  For example, the following code
-moves the value stored under key1 to key2.
-<pre>
-  std::string value;
-  leveldb::Status s = db-&gt;Get(leveldb::ReadOptions(), key1, &amp;value);
-  if (s.ok()) s = db-&gt;Put(leveldb::WriteOptions(), key2, value);
-  if (s.ok()) s = db-&gt;Delete(leveldb::WriteOptions(), key1);
-</pre>
-
-<h1>Atomic Updates</h1>
-<p>
-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 <code>WriteBatch</code> class to
-atomically apply a set of updates:
-<p>
-<pre>
-  #include "leveldb/write_batch.h"
-  ...
-  std::string value;
-  leveldb::Status s = db-&gt;Get(leveldb::ReadOptions(), key1, &amp;value);
-  if (s.ok()) {
-    leveldb::WriteBatch batch;
-    batch.Delete(key1);
-    batch.Put(key2, value);
-    s = db-&gt;Write(leveldb::WriteOptions(), &amp;batch);
-  }
-</pre>
-The <code>WriteBatch</code> 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 <code>Delete</code> before <code>Put</code> so that if <code>key1</code> is identical to <code>key2</code>,
-we do not end up erroneously dropping the value entirely.
-<p>
-Apart from its atomicity benefits, <code>WriteBatch</code> may also be used to
-speed up bulk updates by placing lots of individual mutations into the
-same batch.
-
-<h1>Synchronous Writes</h1>
-By default, each write to <code>leveldb</code> 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 <code>sync</code> 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 <code>fsync(...)</code> or <code>fdatasync(...)</code> or
-<code>msync(..., MS_SYNC)</code> before the write operation returns.)
-<pre>
-  leveldb::WriteOptions write_options;
-  write_options.sync = true;
-  db-&gt;Put(write_options, ...);
-</pre>
-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 <code>sync</code> is false, an update
-is pushed from the process memory into the operating system before it
-is considered done.
-
-<p>
-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.)
-
-<p>
-<code>WriteBatch</code> provides an alternative to asynchronous writes.
-Multiple updates may be placed in the same <code>WriteBatch</code> and
-applied together using a synchronous write (i.e.,
-<code>write_options.sync</code> is set to true).  The extra cost of
-the synchronous write will be amortized across all of the writes in
-the batch.
-
-<p>
-<h1>Concurrency</h1>
-<p>
-A database may only be opened by one process at a time.
-The <code>leveldb</code> implementation acquires a lock from the
-operating system to prevent misuse.  Within a single process, the
-same <code>leveldb::DB</code> object may be safely shared by multiple
-concurrent threads.  I.e., different threads may write into or fetch
-iterators or call <code>Get</code> 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.
-<p>
-<h1>Iteration</h1>
-<p>
-The following example demonstrates how to print all key,value pairs
-in a database.
-<p>
-<pre>
-  leveldb::Iterator* it = db-&gt;NewIterator(leveldb::ReadOptions());
-  for (it-&gt;SeekToFirst(); it-&gt;Valid(); it-&gt;Next()) {
-    cout &lt;&lt; it-&gt;key().ToString() &lt;&lt; ": "  &lt;&lt; it-&gt;value().ToString() &lt;&lt; endl;
-  }
-  assert(it-&gt;status().ok());  // Check for any errors found during the scan
-  delete it;
-</pre>
-The following variation shows how to process just the keys in the
-range <code>[start,limit)</code>:
-<p>
-<pre>
-  for (it-&gt;Seek(start);
-       it-&gt;Valid() &amp;&amp; it-&gt;key().ToString() &lt; limit;
-       it-&gt;Next()) {
-    ...
-  }
-</pre>
-You can also process entries in reverse order.  (Caveat: reverse
-iteration may be somewhat slower than forward iteration.)
-<p>
-<pre>
-  for (it-&gt;SeekToLast(); it-&gt;Valid(); it-&gt;Prev()) {
-    ...
-  }
-</pre>
-<h1>Snapshots</h1>
-<p>
-Snapshots provide consistent read-only views over the entire state of
-the key-value store.  <code>ReadOptions::snapshot</code> may be non-NULL to indicate
-that a read should operate on a particular version of the DB state.
-If <code>ReadOptions::snapshot</code> is NULL, the read will operate on an
-implicit snapshot of the current state.
-<p>
-Snapshots are created by the DB::GetSnapshot() method:
-<p>
-<pre>
-  leveldb::ReadOptions options;
-  options.snapshot = db-&gt;GetSnapshot();
-  ... apply some updates to db ...
-  leveldb::Iterator* iter = db-&gt;NewIterator(options);
-  ... read using iter to view the state when the snapshot was created ...
-  delete iter;
-  db-&gt;ReleaseSnapshot(options.snapshot);
-</pre>
-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.
-<h1>Slice</h1>
-<p>
-The return value of the <code>it->key()</code> and <code>it->value()</code> calls above
-are instances of the <code>leveldb::Slice</code> type.  <code>Slice</code> is a simple
-structure that contains a length and a pointer to an external byte
-array.  Returning a <code>Slice</code> is a cheaper alternative to returning a
-<code>std::string</code> since we do not need to copy potentially large keys and
-values.  In addition, <code>leveldb</code> methods do not return null-terminated
-C-style strings since <code>leveldb</code> keys and values are allowed to
-contain '\0' bytes.
-<p>
-C++ strings and null-terminated C-style strings can be easily converted
-to a Slice:
-<p>
-<pre>
-   leveldb::Slice s1 = "hello";
-
-   std::string str("world");
-   leveldb::Slice s2 = str;
-</pre>
-A Slice can be easily converted back to a C++ string:
-<pre>
-   std::string str = s1.ToString();
-   assert(str == std::string("hello"));
-</pre>
-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:
-<p>
-<pre>
-   leveldb::Slice slice;
-   if (...) {
-     std::string str = ...;
-     slice = str;
-   }
-   Use(slice);
-</pre>
-When the <code>if</code> statement goes out of scope, <code>str</code> will be destroyed and the
-backing storage for <code>slice</code> will disappear.
-<p>
-<h1>Comparators</h1>
-<p>
-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 <code>leveldb::Comparator</code> that expresses these rules:
-<p>
-<pre>
-  class TwoPartComparator : public leveldb::Comparator {
-   public:
-    // Three-way comparison function:
-    //   if a &lt; b: negative result
-    //   if a &gt; b: positive result
-    //   else: zero result
-    int Compare(const leveldb::Slice&amp; a, const leveldb::Slice&amp; b) const {
-      int a1, a2, b1, b2;
-      ParseKey(a, &amp;a1, &amp;a2);
-      ParseKey(b, &amp;b1, &amp;b2);
-      if (a1 &lt; b1) return -1;
-      if (a1 &gt; b1) return +1;
-      if (a2 &lt; b2) return -1;
-      if (a2 &gt; b2) return +1;
-      return 0;
-    }
-
-    // Ignore the following methods for now:
-    const char* Name() const { return "TwoPartComparator"; }
-    void FindShortestSeparator(std::string*, const leveldb::Slice&amp;) const { }
-    void FindShortSuccessor(std::string*) const { }
-  };
-</pre>
-Now create a database using this custom comparator:
-<p>
-<pre>
-  TwoPartComparator cmp;
-  leveldb::DB* db;
-  leveldb::Options options;
-  options.create_if_missing = true;
-  options.comparator = &amp;cmp;
-  leveldb::Status status = leveldb::DB::Open(options, "/tmp/testdb", &amp;db);
-  ...
-</pre>
-<h2>Backwards compatibility</h2>
-<p>
-The result of the comparator's <code>Name</code> method is attached to the
-database when it is created, and is checked on every subsequent
-database open.  If the name changes, the <code>leveldb::DB::Open</code> 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.
-<p>
-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 <code>TwoPartComparator</code>),
-(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.
-<p>
-<h1>Performance</h1>
-<p>
-Performance can be tuned by changing the default values of the
-types defined in <code>include/leveldb/options.h</code>.
-
-<p>
-<h2>Block size</h2>
-<p>
-<code>leveldb</code> 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.
-<p>
-<h2>Compression</h2>
-<p>
-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:
-<p>
-<pre>
-  leveldb::Options options;
-  options.compression = leveldb::kNoCompression;
-  ... leveldb::DB::Open(options, name, ...) ....
-</pre>
-<h2>Cache</h2>
-<p>
-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
-<code>options.cache</code> is non-NULL, it is used to cache frequently used
-uncompressed block contents.
-<p>
-<pre>
-  #include "leveldb/cache.h"
-
-  leveldb::Options options;
-  options.cache = leveldb::NewLRUCache(100 * 1048576);  // 100MB cache
-  leveldb::DB* db;
-  leveldb::DB::Open(options, name, &db);
-  ... use the db ...
-  delete db
-  delete options.cache;
-</pre>
-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 <code>Env</code>
-implementation provided by the client.)
-<p>
-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:
-<p>
-<pre>
-  leveldb::ReadOptions options;
-  options.fill_cache = false;
-  leveldb::Iterator* it = db-&gt;NewIterator(options);
-  for (it-&gt;SeekToFirst(); it-&gt;Valid(); it-&gt;Next()) {
-    ...
-  }
-</pre>
-<h2>Key Layout</h2>
-<p>
-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.
-<p>
-For example, suppose we are implementing a simple file system on top
-of <code>leveldb</code>.  The types of entries we might wish to store are:
-<p>
-<pre>
-   filename -&gt; permission-bits, length, list of file_block_ids
-   file_block_id -&gt; data
-</pre>
-We might want to prefix <code>filename</code> keys with one letter (say '/') and the
-<code>file_block_id</code> 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.
-<p>
-<h2>Filters</h2>
-<p>
-Because of the way <code>leveldb</code> data is organized on disk,
-a single <code>Get()</code> call may involve multiple reads from disk.
-The optional <code>FilterPolicy</code> mechanism can be used to reduce
-the number of disk reads substantially.
-<pre>
-   leveldb::Options options;
-   options.filter_policy = NewBloomFilterPolicy(10);
-   leveldb::DB* db;
-   leveldb::DB::Open(options, "/tmp/testdb", &amp;db);
-   ... use the database ...
-   delete db;
-   delete options.filter_policy;
-</pre>
-The preceding code associates a
-<a href="http://en.wikipedia.org/wiki/Bloom_filter">Bloom filter</a>
-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 <code>Get()</code> 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.
-<p>
-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.
-<code>NewBloomFilterPolicy</code> must not be used with such a comparator.
-Instead, the application should provide a custom filter policy that
-also ignores trailing spaces.  For example:
-<pre>
-  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&lt;Slice&gt; trimmed(n);
-      for (int i = 0; i &lt; n; i++) {
-        trimmed[i] = RemoveTrailingSpaces(keys[i]);
-      }
-      return builtin_policy_-&gt;CreateFilter(&amp;trimmed[i], n, dst);
-    }
-
-    bool KeyMayMatch(const Slice& key, const Slice& filter) const {
-      // Use builtin bloom filter code after removing trailing spaces
-      return builtin_policy_-&gt;KeyMayMatch(RemoveTrailingSpaces(key), filter);
-    }
-  };
-</pre>
-<p>
-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 <code>leveldb/filter_policy.h</code> for detail.
-<p>
-<h1>Checksums</h1>
-<p>
-<code>leveldb</code> associates checksums with all data it stores in the file system.
-There are two separate controls provided over how aggressively these
-checksums are verified:
-<p>
-<ul>
-<li> <code>ReadOptions::verify_checksums</code> 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.
-<p>
-<li> <code>Options::paranoid_checks</code> 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.
-<p>
-  If a database is corrupted (perhaps it cannot be opened when
-  paranoid checking is turned on), the <code>leveldb::RepairDB</code> function
-  may be used to recover as much of the data as possible
-<p>
-</ul>
-<h1>Approximate Sizes</h1>
-<p>
-The <code>GetApproximateSizes</code> method can used to get the approximate
-number of bytes of file system space used by one or more key ranges.
-<p>
-<pre>
-   leveldb::Range ranges[2];
-   ranges[0] = leveldb::Range("a", "c");
-   ranges[1] = leveldb::Range("x", "z");
-   uint64_t sizes[2];
-   leveldb::Status s = db-&gt;GetApproximateSizes(ranges, 2, sizes);
-</pre>
-The preceding call will set <code>sizes[0]</code> to the approximate number of
-bytes of file system space used by the key range <code>[a..c)</code> and
-<code>sizes[1]</code> to the approximate number of bytes used by the key range
-<code>[x..z)</code>.
-<p>
-<h1>Environment</h1>
-<p>
-All file operations (and other operating system calls) issued by the
-<code>leveldb</code> implementation are routed through a <code>leveldb::Env</code> object.
-Sophisticated clients may wish to provide their own <code>Env</code>
-implementation to get better control.  For example, an application may
-introduce artificial delays in the file IO paths to limit the impact
-of <code>leveldb</code> on other activities in the system.
-<p>
-<pre>
-  class SlowEnv : public leveldb::Env {
-    .. implementation of the Env interface ...
-  };
-
-  SlowEnv env;
-  leveldb::Options options;
-  options.env = &amp;env;
-  Status s = leveldb::DB::Open(options, ...);
-</pre>
-<h1>Porting</h1>
-<p>
-<code>leveldb</code> may be ported to a new platform by providing platform
-specific implementations of the types/methods/functions exported by
-<code>leveldb/port/port.h</code>.  See <code>leveldb/port/port_example.h</code> for more
-details.
-<p>
-In addition, the new platform may need a new default <code>leveldb::Env</code>
-implementation.  See <code>leveldb/util/env_posix.h</code> for an example.
-
-<h1>Other Information</h1>
-
-<p>
-Details about the <code>leveldb</code> implementation may be found in
-the following documents:
-<ul>
-<li> <a href="impl.html">Implementation notes</a>
-<li> <a href="table_format.txt">Format of an immutable Table file</a>
-<li> <a href="log_format.txt">Format of a log file</a>
-</ul>
-
-</body>
-</html>

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/leveldb/options.h`.
+
+### Block size
+
+leveldb groups adjacent keys together into the same block and such a block is
+the unit of transfer to and from persistent storage. The default block size is
+approximately 4096 uncompressed bytes.  Applications that mostly do bulk scans
+over the contents of the database may wish to increase this size. Applications
+that do a lot of point reads of small values may wish to switch to a smaller
+block size if performance measurements indicate an improvement. There isn't much
+benefit in using blocks smaller than one kilobyte, or larger than a few
+megabytes. Also note that compression will be more effective with larger block
+sizes.
+
+### Compression
+
+Each block is individually compressed before being written to persistent
+storage. Compression is on by default since the default compression method is
+very fast, and is automatically disabled for uncompressible data. In rare cases,
+applications may want to disable compression entirely, but should only do so if
+benchmarks show a performance improvement:
+
+```c++
+leveldb::Options options;
+options.compression = leveldb::kNoCompression;
+... leveldb::DB::Open(options, name, ...) ....
+```
+
+### Cache
+
+The contents of the database are stored in a set of files in the filesystem and
+each file stores a sequence of compressed blocks. If options.block_cache is
+non-NULL, it is used to cache frequently used uncompressed block contents.
+
+```c++
+#include "leveldb/cache.h"
+
+leveldb::Options options;
+options.block_cache = leveldb::NewLRUCache(100 * 1048576);  // 100MB cache
+leveldb::DB* db;
+leveldb::DB::Open(options, name, &db);
+... use the db ...
+delete db
+delete options.block_cache;
+```
+
+Note that the cache holds uncompressed data, and therefore it should be sized
+according to application level data sizes, without any reduction from
+compression. (Caching of compressed blocks is left to the operating system
+buffer cache, or any custom Env implementation provided by the client.)
+
+When performing a bulk read, the application may wish to disable caching so that
+the data processed by the bulk read does not end up displacing most of the
+cached contents. A per-iterator option can be used to achieve this:
+
+```c++
+leveldb::ReadOptions options;
+options.fill_cache = false;
+leveldb::Iterator* it = db->NewIterator(options);
+for (it->SeekToFirst(); it->Valid(); it->Next()) {
+  ...
+}
+```
+
+### Key Layout
+
+Note that the unit of disk transfer and caching is a block. Adjacent keys
+(according to the database sort order) will usually be placed in the same block.
+Therefore the application can improve its performance by placing keys that are
+accessed together near each other and placing infrequently used keys in a
+separate region of the key space.
+
+For example, suppose we are implementing a simple file system on top of leveldb.
+The types of entries we might wish to store are:
+
+    filename -> permission-bits, length, list of file_block_ids
+    file_block_id -> data
+
+We might want to prefix filename keys with one letter (say '/') and the
+`file_block_id` keys with a different letter (say '0') so that scans over just
+the metadata do not force us to fetch and cache bulky file contents.
+
+### Filters
+
+Because of the way leveldb data is organized on disk, a single `Get()` call may
+involve multiple reads from disk. The optional FilterPolicy mechanism can be
+used to reduce the number of disk reads substantially.
+
+```c++
+leveldb::Options options;
+options.filter_policy = NewBloomFilterPolicy(10);
+leveldb::DB* db;
+leveldb::DB::Open(options, "/tmp/testdb", &db);
+... use the database ...
+delete db;
+delete options.filter_policy;
+```
+
+The preceding code associates a Bloom filter based filtering policy with the
+database.  Bloom filter based filtering relies on keeping some number of bits of
+data in memory per key (in this case 10 bits per key since that is the argument
+we passed to `NewBloomFilterPolicy`). This filter will reduce the number of
+unnecessary disk reads needed for Get() calls by a factor of approximately
+a 100. Increasing the bits per key will lead to a larger reduction at the cost
+of more memory usage. We recommend that applications whose working set does not
+fit in memory and that do a lot of random reads set a filter policy.
+
+If you are using a custom comparator, you should ensure that the filter policy
+you are using is compatible with your comparator. For example, consider a
+comparator that ignores trailing spaces when comparing keys.
+`NewBloomFilterPolicy` must not be used with such a comparator. Instead, the
+application should provide a custom filter policy that also ignores trailing
+spaces. For example:
+
+```c++
+class CustomFilterPolicy : public leveldb::FilterPolicy {
+ private:
+  FilterPolicy* builtin_policy_;
+
+ public:
+  CustomFilterPolicy() : builtin_policy_(NewBloomFilterPolicy(10)) {}
+  ~CustomFilterPolicy() { delete builtin_policy_; }
+
+  const char* Name() const { return "IgnoreTrailingSpacesFilter"; }
+
+  void CreateFilter(const Slice* keys, int n, std::string* dst) const {
+    // Use builtin bloom filter code after removing trailing spaces
+    std::vector<Slice> trimmed(n);
+    for (int i = 0; i < n; i++) {
+      trimmed[i] = RemoveTrailingSpaces(keys[i]);
+    }
+    return builtin_policy_->CreateFilter(&trimmed[i], n, dst);
+  }
+};
+```
+
+Advanced applications may provide a filter policy that does not use a bloom
+filter but uses some other mechanism for summarizing a set of keys. See
+`leveldb/filter_policy.h` for detail.
+
+## Checksums
+
+leveldb associates checksums with all data it stores in the file system. There
+are two separate controls provided over how aggressively these checksums are
+verified:
+
+`ReadOptions::verify_checksums` may be set to true to force checksum
+verification of all data that is read from the file system on behalf of a
+particular read.  By default, no such verification is done.
+
+`Options::paranoid_checks` may be set to true before opening a database to make
+the database implementation raise an error as soon as it detects an internal
+corruption. Depending on which portion of the database has been corrupted, the
+error may be raised when the database is opened, or later by another database
+operation. By default, paranoid checking is off so that the database can be used
+even if parts of its persistent storage have been corrupted.
+
+If a database is corrupted (perhaps it cannot be opened when paranoid checking
+is turned on), the `leveldb::RepairDB` function may be used to recover as much
+of the data as possible
+
+## Approximate Sizes
+
+The `GetApproximateSizes` method can used to get the approximate number of bytes
+of file system space used by one or more key ranges.
+
+```c++
+leveldb::Range ranges[2];
+ranges[0] = leveldb::Range("a", "c");
+ranges[1] = leveldb::Range("x", "z");
+uint64_t sizes[2];
+leveldb::Status s = db->GetApproximateSizes(ranges, 2, sizes);
+```
+
+The preceding call will set `sizes[0]` to the approximate number of bytes of
+file system space used by the key range `[a..c)` and `sizes[1]` to the
+approximate number of bytes used by the key range `[x..z)`.
+
+## Environment
+
+All file operations (and other operating system calls) issued by the leveldb
+implementation are routed through a `leveldb::Env` object. Sophisticated clients
+may wish to provide their own Env implementation to get better control.
+For example, an application may introduce artificial delays in the file IO
+paths to limit the impact of leveldb on other activities in the system.
+
+```c++
+class SlowEnv : public leveldb::Env {
+  ... implementation of the Env interface ...
+};
+
+SlowEnv env;
+leveldb::Options options;
+options.env = &env;
+Status s = leveldb::DB::Open(options, ...);
+```
+
+## Porting
+
+leveldb may be ported to a new platform by providing platform specific
+implementations of the types/methods/functions exported by
+`leveldb/port/port.h`.  See `leveldb/port/port_example.h` for more details.
+
+In addition, the new platform may need a new default `leveldb::Env`
+implementation.  See `leveldb/util/env_posix.h` for an example.
+
+## Other Information
+
+Details about the leveldb implementation may be found in the following
+documents:
+
+1. [Implementation notes](impl.md)
+2. [Format of an immutable Table file](table_format.md)
+3. [Format of a log file](log_format.md)

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.

doc/log_format.txt

@@ -1,75 +0,0 @@
-The log file contents are a sequence of 32KB blocks.  The only
-exception is that the tail of the file may contain a partial block.
-
-Each block consists of a sequence of records:
-   block := record* trailer?
-   record :=
-	checksum: uint32	// crc32c of type and data[] ; little-endian
-	length: uint16		// little-endian
-	type: uint8		// One of FULL, FIRST, MIDDLE, LAST
-	data: uint8[length]
-
-A record never starts within the last six bytes of a block (since it
-won't fit).  Any leftover bytes here form the trailer, which must
-consist entirely of zero bytes and must be skipped by readers.
-
-Aside: if exactly seven bytes are left in the current block, and a new
-non-zero length record is added, the writer must emit a FIRST record
-(which contains zero bytes of user data) to fill up the trailing seven
-bytes of the block and then emit all of the user data in subsequent
-blocks.
-
-More types may be added in the future.  Some Readers may skip record
-types they do not understand, others may report that some data was
-skipped.
-
-FULL == 1
-FIRST == 2
-MIDDLE == 3
-LAST == 4
-
-The FULL record contains the contents of an entire user record.
-
-FIRST, MIDDLE, LAST are types used for user records that have been
-split into multiple fragments (typically because of block boundaries).
-FIRST is the type of the first fragment of a user record, LAST is the
-type of the last fragment of a user record, and MIDDLE is the type of
-all interior fragments of a user record.
-
-Example: consider a sequence of user records:
-   A: length 1000
-   B: length 97270
-   C: length 8000
-A will be stored as a FULL record in the first block.
-
-B will be split into three fragments: first fragment occupies the rest
-of the first block, second fragment occupies the entirety of the
-second block, and the third fragment occupies a prefix of the third
-block.  This will leave six bytes free in the third block, which will
-be left empty as the trailer.
-
-C will be stored as a FULL record in the fourth block.
-
-===================
-
-Some benefits over the recordio format:
-
-(1) We do not need any heuristics for resyncing - just go to next
-block boundary and scan.  If there is a corruption, skip to the next
-block.  As a side-benefit, we do not get confused when part of the
-contents of one log file are embedded as a record inside another log
-file.
-
-(2) Splitting at approximate boundaries (e.g., for mapreduce) is
-simple: find the next block boundary and skip records until we
-hit a FULL or FIRST record.
-
-(3) We do not need extra buffering for large records.
-
-Some downsides compared to recordio format:
-
-(1) No packing of tiny records.  This could be fixed by adding a new
-record type, so it is a shortcoming of the current implementation,
-not necessarily the format.
-
-(2) No compression.  Again, this could be fixed by adding new record types.

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

doc/table_format.txt

@@ -1,104 +0,0 @@
-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 https://developers.google.com/protocol-buffers/docs/encoding#varints
-for an explanation of varint64 format.
-
-(1) The sequence of key/value pairs in the file are stored in sorted
-order and partitioned into a sequence of data blocks.  These blocks
-come one after another at the beginning of the file.  Each data block
-is formatted according to the code in block_builder.cc, and then
-optionally compressed.
-
-(2) After the data blocks we store a bunch of meta blocks.  The
-supported meta block types are described below.  More meta block types
-may be added in the future.  Each meta block is again formatted using
-block_builder.cc and then optionally compressed.
-
-(3) A "metaindex" block.  It contains one entry for every other meta
-block where the key is the name of the meta block and the value is a
-BlockHandle pointing to that meta block.
-
-(4) An "index" block.  This block contains one entry per data block,
-where the key is a string >= last key in that data block and before
-the first key in the successive data block.  The value is the
-BlockHandle for the data block.
-
-(6) At the very end of the file is a fixed length footer that contains
-the BlockHandle of the metaindex and index blocks as well as a magic number.
-       metaindex_handle: char[p];    // Block handle for metaindex
-       index_handle:     char[q];    // Block handle for index
-       padding:          char[40-p-q]; // zeroed bytes to make fixed length
-                                       // (40==2*BlockHandle::kMaxEncodedLength)
-       magic:            fixed64;    // == 0xdb4775248b80fb57 (little-endian)
-
-"filter" Meta Block
--------------------
-
-If a "FilterPolicy" was specified when the database was opened, a
-filter block is stored in each table.  The "metaindex" block contains
-an entry that maps from "filter.<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

helpers/memenv/memenv.cc

@@ -4,14 +4,18 @@
 
 #include "helpers/memenv/memenv.h"
 
+#include <string.h>
+
+#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"
-#include <map>
-#include <string.h>
-#include <string>
-#include <vector>
 
 namespace leveldb {
 
@@ -47,9 +51,22 @@ class FileState {
     }
   }
 
-  uint64_t Size() const { return size_; }
+  uint64_t Size() const {
+    MutexLock lock(&blocks_mutex_);
+    return size_;
+  }
+
+  void Truncate() {
+    MutexLock lock(&blocks_mutex_);
+    for (char*& block : blocks_) {
+      delete[] block;
+    }
+    blocks_.clear();
+    size_ = 0;
+  }
 
   Status Read(uint64_t offset, size_t n, Slice* result, char* scratch) const {
+    MutexLock lock(&blocks_mutex_);
     if (offset > size_) {
       return Status::IOError("Offset greater than file size.");
     }
@@ -62,16 +79,9 @@ class FileState {
       return Status::OK();
     }
 
-    assert(offset / kBlockSize <= SIZE_MAX);
+    assert(offset / kBlockSize <= std::numeric_limits<size_t>::max());
     size_t block = static_cast<size_t>(offset / kBlockSize);
     size_t block_offset = offset % kBlockSize;
-
-    if (n <= kBlockSize - block_offset) {
-      // The requested bytes are all in the first block.
-      *result = Slice(blocks_[block] + block_offset, n);
-      return Status::OK();
-    }
-
     size_t bytes_to_copy = n;
     char* dst = scratch;
 
@@ -96,6 +106,7 @@ class FileState {
     const char* src = data.data();
     size_t src_len = data.size();
 
+    MutexLock lock(&blocks_mutex_);
     while (src_len > 0) {
       size_t avail;
       size_t offset = size_ % kBlockSize;
@@ -124,10 +135,7 @@ class FileState {
  private:
   // Private since only Unref() should be used to delete it.
   ~FileState() {
-    for (std::vector<char*>::iterator i = blocks_.begin(); i != blocks_.end();
-         ++i) {
-      delete [] *i;
-    }
+    Truncate();
   }
 
   // No copying allowed.
@@ -135,13 +143,11 @@ class FileState {
   void operator=(const FileState&);
 
   port::Mutex refs_mutex_;
-  int refs_;  // Protected by refs_mutex_;
+  int refs_ GUARDED_BY(refs_mutex_);
 
-  // The following fields are not protected by any mutex. They are only mutable
-  // while the file is being written, and concurrent access is not allowed
-  // to writable files.
-  std::vector<char*> blocks_;
-  uint64_t size_;
+  mutable port::Mutex blocks_mutex_;
+  std::vector<char*> blocks_ GUARDED_BY(blocks_mutex_);
+  uint64_t size_ GUARDED_BY(blocks_mutex_);
 
   enum { kBlockSize = 8 * 1024 };
 };
@@ -242,7 +248,7 @@ class InMemoryEnv : public EnvWrapper {
                                    SequentialFile** result) {
     MutexLock lock(&mutex_);
     if (file_map_.find(fname) == file_map_.end()) {
-      *result = NULL;
+      *result = nullptr;
       return Status::IOError(fname, "File not found");
     }
 
@@ -254,7 +260,7 @@ class InMemoryEnv : public EnvWrapper {
                                      RandomAccessFile** result) {
     MutexLock lock(&mutex_);
     if (file_map_.find(fname) == file_map_.end()) {
-      *result = NULL;
+      *result = nullptr;
       return Status::IOError(fname, "File not found");
     }
 
@@ -265,14 +271,32 @@ class InMemoryEnv : public EnvWrapper {
   virtual Status NewWritableFile(const std::string& fname,
                                  WritableFile** result) {
     MutexLock lock(&mutex_);
-    if (file_map_.find(fname) != file_map_.end()) {
-      DeleteFileInternal(fname);
+    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();
     }
 
-    FileState* file = new FileState();
-    file->Ref();
-    file_map_[fname] = file;
+    *result = new WritableFileImpl(file);
+    return Status::OK();
+  }
 
+  virtual Status NewAppendableFile(const std::string& fname,
+                                   WritableFile** result) {
+    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();
   }
@@ -299,7 +323,8 @@ class InMemoryEnv : public EnvWrapper {
     return Status::OK();
   }
 
-  void DeleteFileInternal(const std::string& fname) {
+  void DeleteFileInternal(const std::string& fname)
+      EXCLUSIVE_LOCKS_REQUIRED(mutex_) {
     if (file_map_.find(fname) == file_map_.end()) {
       return;
     }
@@ -373,7 +398,7 @@ class InMemoryEnv : public EnvWrapper {
   // Map from filenames to FileState objects, representing a simple file system.
   typedef std::map<std::string, FileState*> FileSystem;
   port::Mutex mutex_;
-  FileSystem file_map_;  // Protected by mutex_.
+  FileSystem file_map_ GUARDED_BY(mutex_);
 };
 
 }  // namespace

helpers/memenv/memenv.h

@@ -5,6 +5,8 @@
 #ifndef STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_
 #define STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_
 
+#include "leveldb/export.h"
+
 namespace leveldb {
 
 class Env;
@@ -13,7 +15,7 @@ class Env;
 // all non-file-storage tasks to base_env. The caller must delete the result
 // when it is no longer needed.
 // *base_env must remain live while the result is in use.
-Env* NewMemEnv(Env* base_env);
+LEVELDB_EXPORT Env* NewMemEnv(Env* base_env);
 
 }  // namespace leveldb
 

helpers/memenv/memenv_test.cc

@@ -40,6 +40,8 @@ TEST(MemEnvTest, Basics) {
 
   // Create a file.
   ASSERT_OK(env_->NewWritableFile("/dir/f", &writable_file));
+  ASSERT_OK(env_->GetFileSize("/dir/f", &file_size));
+  ASSERT_EQ(0, file_size);
   delete writable_file;
 
   // Check that the file exists.
@@ -55,9 +57,16 @@ TEST(MemEnvTest, Basics) {
   ASSERT_OK(writable_file->Append("abc"));
   delete writable_file;
 
-  // Check for expected size.
+  // Check that append works.
+  ASSERT_OK(env_->NewAppendableFile("/dir/f", &writable_file));
   ASSERT_OK(env_->GetFileSize("/dir/f", &file_size));
   ASSERT_EQ(3, file_size);
+  ASSERT_OK(writable_file->Append("hello"));
+  delete writable_file;
+
+  // Check for expected size.
+  ASSERT_OK(env_->GetFileSize("/dir/f", &file_size));
+  ASSERT_EQ(8, file_size);
 
   // Check that renaming works.
   ASSERT_TRUE(!env_->RenameFile("/dir/non_existent", "/dir/g").ok());
@@ -65,7 +74,7 @@ TEST(MemEnvTest, Basics) {
   ASSERT_TRUE(!env_->FileExists("/dir/f"));
   ASSERT_TRUE(env_->FileExists("/dir/g"));
   ASSERT_OK(env_->GetFileSize("/dir/g", &file_size));
-  ASSERT_EQ(3, file_size);
+  ASSERT_EQ(8, file_size);
 
   // Check that opening non-existent file fails.
   SequentialFile* seq_file;
@@ -182,6 +191,29 @@ TEST(MemEnvTest, LargeWrite) {
   delete [] scratch;
 }
 
+TEST(MemEnvTest, OverwriteOpenFile) {
+  const char kWrite1Data[] = "Write #1 data";
+  const size_t kFileDataLen = sizeof(kWrite1Data) - 1;
+  const std::string kTestFileName = test::TmpDir() + "/leveldb-TestFile.dat";
+
+  ASSERT_OK(WriteStringToFile(env_, kWrite1Data, kTestFileName));
+
+  RandomAccessFile* rand_file;
+  ASSERT_OK(env_->NewRandomAccessFile(kTestFileName, &rand_file));
+
+  const char kWrite2Data[] = "Write #2 data";
+  ASSERT_OK(WriteStringToFile(env_, kWrite2Data, kTestFileName));
+
+  // Verify that overwriting an open file will result in the new file data
+  // being read from files opened before the write.
+  Slice result;
+  char scratch[kFileDataLen];
+  ASSERT_OK(rand_file->Read(0, kFileDataLen, &result, scratch));
+  ASSERT_EQ(0, result.compare(kWrite2Data));
+
+  delete rand_file;
+}
+
 TEST(MemEnvTest, DBTest) {
   Options options;
   options.create_if_missing = true;

include/leveldb/c.h

@@ -47,6 +47,7 @@ extern "C" {
 #include <stdarg.h>
 #include <stddef.h>
 #include <stdint.h>
+#include "leveldb/export.h"
 
 /* Exported types */
 
@@ -69,204 +70,185 @@ typedef struct leveldb_writeoptions_t  leveldb_writeoptions_t;
 
 /* DB operations */
 
-extern leveldb_t* leveldb_open(
-    const leveldb_options_t* options,
-    const char* name,
-    char** errptr);
+LEVELDB_EXPORT leveldb_t* leveldb_open(const leveldb_options_t* options,
+                                       const char* name, char** errptr);
 
-extern void leveldb_close(leveldb_t* db);
+LEVELDB_EXPORT void leveldb_close(leveldb_t* db);
 
-extern void leveldb_put(
-    leveldb_t* db,
-    const leveldb_writeoptions_t* options,
-    const char* key, size_t keylen,
-    const char* val, size_t vallen,
-    char** errptr);
+LEVELDB_EXPORT void leveldb_put(leveldb_t* db,
+                                const leveldb_writeoptions_t* options,
+                                const char* key, size_t keylen, const char* val,
+                                size_t vallen, char** errptr);
 
-extern void leveldb_delete(
-    leveldb_t* db,
-    const leveldb_writeoptions_t* options,
-    const char* key, size_t keylen,
-    char** errptr);
+LEVELDB_EXPORT void leveldb_delete(leveldb_t* db,
+                                   const leveldb_writeoptions_t* options,
+                                   const char* key, size_t keylen,
+                                   char** errptr);
 
-extern void leveldb_write(
-    leveldb_t* db,
-    const leveldb_writeoptions_t* options,
-    leveldb_writebatch_t* batch,
-    char** errptr);
+LEVELDB_EXPORT void leveldb_write(leveldb_t* db,
+                                  const leveldb_writeoptions_t* options,
+                                  leveldb_writebatch_t* batch, char** errptr);
 
 /* Returns NULL if not found.  A malloc()ed array otherwise.
    Stores the length of the array in *vallen. */
-extern char* leveldb_get(
-    leveldb_t* db,
-    const leveldb_readoptions_t* options,
-    const char* key, size_t keylen,
-    size_t* vallen,
-    char** errptr);
+LEVELDB_EXPORT char* leveldb_get(leveldb_t* db,
+                                 const leveldb_readoptions_t* options,
+                                 const char* key, size_t keylen, size_t* vallen,
+                                 char** errptr);
 
-extern leveldb_iterator_t* leveldb_create_iterator(
-    leveldb_t* db,
-    const leveldb_readoptions_t* options);
+LEVELDB_EXPORT leveldb_iterator_t* leveldb_create_iterator(
+    leveldb_t* db, const leveldb_readoptions_t* options);
 
-extern const leveldb_snapshot_t* leveldb_create_snapshot(
-    leveldb_t* db);
+LEVELDB_EXPORT const leveldb_snapshot_t* leveldb_create_snapshot(leveldb_t* db);
 
-extern void leveldb_release_snapshot(
-    leveldb_t* db,
-    const leveldb_snapshot_t* snapshot);
+LEVELDB_EXPORT void leveldb_release_snapshot(
+    leveldb_t* db, const leveldb_snapshot_t* snapshot);
 
 /* Returns NULL if property name is unknown.
    Else returns a pointer to a malloc()-ed null-terminated value. */
-extern char* leveldb_property_value(
-    leveldb_t* db,
-    const char* propname);
+LEVELDB_EXPORT char* leveldb_property_value(leveldb_t* db,
+                                            const char* propname);
 
-extern void leveldb_approximate_sizes(
-    leveldb_t* db,
-    int num_ranges,
-    const char* const* range_start_key, const size_t* range_start_key_len,
-    const char* const* range_limit_key, const size_t* range_limit_key_len,
-    uint64_t* sizes);
+LEVELDB_EXPORT void leveldb_approximate_sizes(
+    leveldb_t* db, int num_ranges, const char* const* range_start_key,
+    const size_t* range_start_key_len, const char* const* range_limit_key,
+    const size_t* range_limit_key_len, uint64_t* sizes);
 
-extern void leveldb_compact_range(
-    leveldb_t* db,
-    const char* start_key, size_t start_key_len,
-    const char* limit_key, size_t limit_key_len);
+LEVELDB_EXPORT void leveldb_compact_range(leveldb_t* db, const char* start_key,
+                                          size_t start_key_len,
+                                          const char* limit_key,
+                                          size_t limit_key_len);
 
 /* Management operations */
 
-extern void leveldb_destroy_db(
-    const leveldb_options_t* options,
-    const char* name,
-    char** errptr);
+LEVELDB_EXPORT void leveldb_destroy_db(const leveldb_options_t* options,
+                                       const char* name, char** errptr);
 
-extern void leveldb_repair_db(
-    const leveldb_options_t* options,
-    const char* name,
-    char** errptr);
+LEVELDB_EXPORT void leveldb_repair_db(const leveldb_options_t* options,
+                                      const char* name, char** errptr);
 
 /* Iterator */
 
-extern void leveldb_iter_destroy(leveldb_iterator_t*);
-extern unsigned char leveldb_iter_valid(const leveldb_iterator_t*);
-extern void leveldb_iter_seek_to_first(leveldb_iterator_t*);
-extern void leveldb_iter_seek_to_last(leveldb_iterator_t*);
-extern void leveldb_iter_seek(leveldb_iterator_t*, const char* k, size_t klen);
-extern void leveldb_iter_next(leveldb_iterator_t*);
-extern void leveldb_iter_prev(leveldb_iterator_t*);
-extern const char* leveldb_iter_key(const leveldb_iterator_t*, size_t* klen);
-extern const char* leveldb_iter_value(const leveldb_iterator_t*, size_t* vlen);
-extern void leveldb_iter_get_error(const leveldb_iterator_t*, char** errptr);
+LEVELDB_EXPORT void leveldb_iter_destroy(leveldb_iterator_t*);
+LEVELDB_EXPORT unsigned char leveldb_iter_valid(const leveldb_iterator_t*);
+LEVELDB_EXPORT void leveldb_iter_seek_to_first(leveldb_iterator_t*);
+LEVELDB_EXPORT void leveldb_iter_seek_to_last(leveldb_iterator_t*);
+LEVELDB_EXPORT void leveldb_iter_seek(leveldb_iterator_t*, const char* k,
+                                      size_t klen);
+LEVELDB_EXPORT void leveldb_iter_next(leveldb_iterator_t*);
+LEVELDB_EXPORT void leveldb_iter_prev(leveldb_iterator_t*);
+LEVELDB_EXPORT const char* leveldb_iter_key(const leveldb_iterator_t*,
+                                            size_t* klen);
+LEVELDB_EXPORT const char* leveldb_iter_value(const leveldb_iterator_t*,
+                                              size_t* vlen);
+LEVELDB_EXPORT void leveldb_iter_get_error(const leveldb_iterator_t*,
+                                           char** errptr);
 
 /* Write batch */
 
-extern leveldb_writebatch_t* leveldb_writebatch_create();
-extern void leveldb_writebatch_destroy(leveldb_writebatch_t*);
-extern void leveldb_writebatch_clear(leveldb_writebatch_t*);
-extern void leveldb_writebatch_put(
-    leveldb_writebatch_t*,
-    const char* key, size_t klen,
-    const char* val, size_t vlen);
-extern void leveldb_writebatch_delete(
-    leveldb_writebatch_t*,
-    const char* key, size_t klen);
-extern void leveldb_writebatch_iterate(
-    leveldb_writebatch_t*,
-    void* state,
+LEVELDB_EXPORT leveldb_writebatch_t* leveldb_writebatch_create();
+LEVELDB_EXPORT void leveldb_writebatch_destroy(leveldb_writebatch_t*);
+LEVELDB_EXPORT void leveldb_writebatch_clear(leveldb_writebatch_t*);
+LEVELDB_EXPORT void leveldb_writebatch_put(leveldb_writebatch_t*,
+                                           const char* key, size_t klen,
+                                           const char* val, size_t vlen);
+LEVELDB_EXPORT void leveldb_writebatch_delete(leveldb_writebatch_t*,
+                                              const char* key, size_t klen);
+LEVELDB_EXPORT void leveldb_writebatch_iterate(
+    const leveldb_writebatch_t*, void* state,
     void (*put)(void*, const char* k, size_t klen, const char* v, size_t vlen),
     void (*deleted)(void*, const char* k, size_t klen));
+LEVELDB_EXPORT void leveldb_writebatch_append(
+    leveldb_writebatch_t* destination, const leveldb_writebatch_t* source);
 
 /* Options */
 
-extern leveldb_options_t* leveldb_options_create();
-extern void leveldb_options_destroy(leveldb_options_t*);
-extern void leveldb_options_set_comparator(
-    leveldb_options_t*,
-    leveldb_comparator_t*);
-extern void leveldb_options_set_filter_policy(
-    leveldb_options_t*,
-    leveldb_filterpolicy_t*);
-extern void leveldb_options_set_create_if_missing(
-    leveldb_options_t*, unsigned char);
-extern void leveldb_options_set_error_if_exists(
-    leveldb_options_t*, unsigned char);
-extern void leveldb_options_set_paranoid_checks(
-    leveldb_options_t*, unsigned char);
-extern void leveldb_options_set_env(leveldb_options_t*, leveldb_env_t*);
-extern void leveldb_options_set_info_log(leveldb_options_t*, leveldb_logger_t*);
-extern void leveldb_options_set_write_buffer_size(leveldb_options_t*, size_t);
-extern void leveldb_options_set_max_open_files(leveldb_options_t*, int);
-extern void leveldb_options_set_cache(leveldb_options_t*, leveldb_cache_t*);
-extern void leveldb_options_set_block_size(leveldb_options_t*, size_t);
-extern void leveldb_options_set_block_restart_interval(leveldb_options_t*, int);
+LEVELDB_EXPORT leveldb_options_t* leveldb_options_create();
+LEVELDB_EXPORT void leveldb_options_destroy(leveldb_options_t*);
+LEVELDB_EXPORT void leveldb_options_set_comparator(leveldb_options_t*,
+                                                   leveldb_comparator_t*);
+LEVELDB_EXPORT void leveldb_options_set_filter_policy(leveldb_options_t*,
+                                                      leveldb_filterpolicy_t*);
+LEVELDB_EXPORT void leveldb_options_set_create_if_missing(leveldb_options_t*,
+                                                          unsigned char);
+LEVELDB_EXPORT void leveldb_options_set_error_if_exists(leveldb_options_t*,
+                                                        unsigned char);
+LEVELDB_EXPORT void leveldb_options_set_paranoid_checks(leveldb_options_t*,
+                                                        unsigned char);
+LEVELDB_EXPORT void leveldb_options_set_env(leveldb_options_t*, leveldb_env_t*);
+LEVELDB_EXPORT void leveldb_options_set_info_log(leveldb_options_t*,
+                                                 leveldb_logger_t*);
+LEVELDB_EXPORT void leveldb_options_set_write_buffer_size(leveldb_options_t*,
+                                                          size_t);
+LEVELDB_EXPORT void leveldb_options_set_max_open_files(leveldb_options_t*, int);
+LEVELDB_EXPORT void leveldb_options_set_cache(leveldb_options_t*,
+                                              leveldb_cache_t*);
+LEVELDB_EXPORT void leveldb_options_set_block_size(leveldb_options_t*, size_t);
+LEVELDB_EXPORT void leveldb_options_set_block_restart_interval(
+    leveldb_options_t*, int);
+LEVELDB_EXPORT void leveldb_options_set_max_file_size(leveldb_options_t*,
+                                                      size_t);
 
 enum {
   leveldb_no_compression = 0,
   leveldb_snappy_compression = 1
 };
-extern void leveldb_options_set_compression(leveldb_options_t*, int);
+LEVELDB_EXPORT void leveldb_options_set_compression(leveldb_options_t*, int);
 
 /* Comparator */
 
-extern leveldb_comparator_t* leveldb_comparator_create(
-    void* state,
-    void (*destructor)(void*),
-    int (*compare)(
-        void*,
-        const char* a, size_t alen,
-        const char* b, size_t blen),
+LEVELDB_EXPORT leveldb_comparator_t* leveldb_comparator_create(
+    void* state, void (*destructor)(void*),
+    int (*compare)(void*, const char* a, size_t alen, const char* b,
+                   size_t blen),
     const char* (*name)(void*));
-extern void leveldb_comparator_destroy(leveldb_comparator_t*);
+LEVELDB_EXPORT void leveldb_comparator_destroy(leveldb_comparator_t*);
 
 /* Filter policy */
 
-extern leveldb_filterpolicy_t* leveldb_filterpolicy_create(
-    void* state,
-    void (*destructor)(void*),
-    char* (*create_filter)(
-        void*,
-        const char* const* key_array, const size_t* key_length_array,
-        int num_keys,
-        size_t* filter_length),
-    unsigned char (*key_may_match)(
-        void*,
-        const char* key, size_t length,
-        const char* filter, size_t filter_length),
+LEVELDB_EXPORT leveldb_filterpolicy_t* leveldb_filterpolicy_create(
+    void* state, void (*destructor)(void*),
+    char* (*create_filter)(void*, const char* const* key_array,
+                           const size_t* key_length_array, int num_keys,
+                           size_t* filter_length),
+    unsigned char (*key_may_match)(void*, const char* key, size_t length,
+                                   const char* filter, size_t filter_length),
     const char* (*name)(void*));
-extern void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t*);
+LEVELDB_EXPORT void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t*);
 
-extern leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(
+LEVELDB_EXPORT leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(
     int bits_per_key);
 
 /* Read options */
 
-extern leveldb_readoptions_t* leveldb_readoptions_create();
-extern void leveldb_readoptions_destroy(leveldb_readoptions_t*);
-extern void leveldb_readoptions_set_verify_checksums(
-    leveldb_readoptions_t*,
-    unsigned char);
-extern void leveldb_readoptions_set_fill_cache(
+LEVELDB_EXPORT leveldb_readoptions_t* leveldb_readoptions_create();
+LEVELDB_EXPORT void leveldb_readoptions_destroy(leveldb_readoptions_t*);
+LEVELDB_EXPORT void leveldb_readoptions_set_verify_checksums(
     leveldb_readoptions_t*, unsigned char);
-extern void leveldb_readoptions_set_snapshot(
-    leveldb_readoptions_t*,
-    const leveldb_snapshot_t*);
+LEVELDB_EXPORT void leveldb_readoptions_set_fill_cache(leveldb_readoptions_t*,
+                                                       unsigned char);
+LEVELDB_EXPORT void leveldb_readoptions_set_snapshot(leveldb_readoptions_t*,
+                                                     const leveldb_snapshot_t*);
 
 /* Write options */
 
-extern leveldb_writeoptions_t* leveldb_writeoptions_create();
-extern void leveldb_writeoptions_destroy(leveldb_writeoptions_t*);
-extern void leveldb_writeoptions_set_sync(
-    leveldb_writeoptions_t*, unsigned char);
+LEVELDB_EXPORT leveldb_writeoptions_t* leveldb_writeoptions_create();
+LEVELDB_EXPORT void leveldb_writeoptions_destroy(leveldb_writeoptions_t*);
+LEVELDB_EXPORT void leveldb_writeoptions_set_sync(leveldb_writeoptions_t*,
+                                                  unsigned char);
 
 /* Cache */
 
-extern leveldb_cache_t* leveldb_cache_create_lru(size_t capacity);
-extern void leveldb_cache_destroy(leveldb_cache_t* cache);
+LEVELDB_EXPORT leveldb_cache_t* leveldb_cache_create_lru(size_t capacity);
+LEVELDB_EXPORT void leveldb_cache_destroy(leveldb_cache_t* cache);
 
 /* Env */
 
-extern leveldb_env_t* leveldb_create_default_env();
-extern void leveldb_env_destroy(leveldb_env_t*);
+LEVELDB_EXPORT leveldb_env_t* leveldb_create_default_env();
+LEVELDB_EXPORT void leveldb_env_destroy(leveldb_env_t*);
+
+/* If not NULL, the returned buffer must be released using leveldb_free(). */
+LEVELDB_EXPORT char* leveldb_env_get_test_directory(leveldb_env_t*);
 
 /* Utility */
 
@@ -275,13 +257,13 @@ extern void leveldb_env_destroy(leveldb_env_t*);
    in this file.  Note that in certain cases (typically on Windows), you
    may need to call this routine instead of free(ptr) to dispose of
    malloc()-ed memory returned by this library. */
-extern void leveldb_free(void* ptr);
+LEVELDB_EXPORT void leveldb_free(void* ptr);
 
 /* Return the major version number for this release. */
-extern int leveldb_major_version();
+LEVELDB_EXPORT int leveldb_major_version();
 
 /* Return the minor version number for this release. */
-extern int leveldb_minor_version();
+LEVELDB_EXPORT int leveldb_minor_version();
 
 #ifdef __cplusplus
 }  /* end extern "C" */

include/leveldb/cache.h

@@ -19,19 +19,23 @@
 #define STORAGE_LEVELDB_INCLUDE_CACHE_H_
 
 #include <stdint.h>
+#include "leveldb/export.h"
 #include "leveldb/slice.h"
 
 namespace leveldb {
 
-class Cache;
+class LEVELDB_EXPORT Cache;
 
 // Create a new cache with a fixed size capacity.  This implementation
 // of Cache uses a least-recently-used eviction policy.
-extern Cache* NewLRUCache(size_t capacity);
+LEVELDB_EXPORT Cache* NewLRUCache(size_t capacity);
 
-class Cache {
+class LEVELDB_EXPORT Cache {
  public:
-  Cache() { }
+  Cache() = default;
+
+  Cache(const Cache&) = delete;
+  Cache& operator=(const Cache&) = delete;
 
   // Destroys all existing entries by calling the "deleter"
   // function that was passed to the constructor.
@@ -52,7 +56,7 @@ class Cache {
   virtual Handle* Insert(const Slice& key, void* value, size_t charge,
                          void (*deleter)(const Slice& key, void* value)) = 0;
 
-  // If the cache has no mapping for "key", returns NULL.
+  // If the cache has no mapping for "key", returns nullptr.
   //
   // Else return a handle that corresponds to the mapping.  The caller
   // must call this->Release(handle) when the returned mapping is no
@@ -81,6 +85,17 @@ class Cache {
   // its cache keys.
   virtual uint64_t NewId() = 0;
 
+  // Remove all cache entries that are not actively in use.  Memory-constrained
+  // applications may wish to call this method to reduce memory usage.
+  // Default implementation of Prune() does nothing.  Subclasses are strongly
+  // encouraged to override the default implementation.  A future release of
+  // leveldb may change Prune() to a pure abstract method.
+  virtual void Prune() {}
+
+  // Return an estimate of the combined charges of all elements stored in the
+  // cache.
+  virtual size_t TotalCharge() const = 0;
+
  private:
   void LRU_Remove(Handle* e);
   void LRU_Append(Handle* e);
@@ -88,10 +103,6 @@ class Cache {
 
   struct Rep;
   Rep* rep_;
-
-  // No copying allowed
-  Cache(const Cache&);
-  void operator=(const Cache&);
 };
 
 }  // namespace leveldb

include/leveldb/comparator.h

@@ -6,6 +6,7 @@
 #define STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_
 
 #include <string>
+#include "leveldb/export.h"
 
 namespace leveldb {
 
@@ -15,7 +16,7 @@ class Slice;
 // used as keys in an sstable or a database.  A Comparator implementation
 // must be thread-safe since leveldb may invoke its methods concurrently
 // from multiple threads.
-class Comparator {
+class LEVELDB_EXPORT Comparator {
  public:
   virtual ~Comparator();
 
@@ -56,7 +57,7 @@ class Comparator {
 // Return a builtin comparator that uses lexicographic byte-wise
 // ordering.  The result remains the property of this module and
 // must not be deleted.
-extern const Comparator* BytewiseComparator();
+LEVELDB_EXPORT const Comparator* BytewiseComparator();
 
 }  // namespace leveldb
 

include/leveldb/db.h

@@ -7,14 +7,15 @@
 
 #include <stdint.h>
 #include <stdio.h>
+#include "leveldb/export.h"
 #include "leveldb/iterator.h"
 #include "leveldb/options.h"
 
 namespace leveldb {
 
-// Update Makefile if you change these
+// Update CMakeLists.txt if you change these
 static const int kMajorVersion = 1;
-static const int kMinorVersion = 18;
+static const int kMinorVersion = 21;
 
 struct Options;
 struct ReadOptions;
@@ -24,13 +25,13 @@ class WriteBatch;
 // Abstract handle to particular state of a DB.
 // A Snapshot is an immutable object and can therefore be safely
 // accessed from multiple threads without any external synchronization.
-class Snapshot {
+class LEVELDB_EXPORT Snapshot {
  protected:
   virtual ~Snapshot();
 };
 
 // A range of keys
-struct Range {
+struct LEVELDB_EXPORT Range {
   Slice start;          // Included in the range
   Slice limit;          // Not included in the range
 
@@ -41,18 +42,22 @@ struct Range {
 // A DB is a persistent ordered map from keys to values.
 // A DB is safe for concurrent access from multiple threads without
 // any external synchronization.
-class DB {
+class LEVELDB_EXPORT DB {
  public:
   // Open the database with the specified "name".
   // Stores a pointer to a heap-allocated database in *dbptr and returns
   // OK on success.
-  // Stores NULL in *dbptr and returns a non-OK status on error.
+  // Stores nullptr in *dbptr and returns a non-OK status on error.
   // Caller should delete *dbptr when it is no longer needed.
   static Status Open(const Options& options,
                      const std::string& name,
                      DB** dbptr);
 
-  DB() { }
+  DB() = default;
+
+  DB(const DB&) = delete;
+  DB& operator=(const DB&) = delete;
+
   virtual ~DB();
 
   // Set the database entry for "key" to "value".  Returns OK on success,
@@ -115,6 +120,8 @@ class DB {
   //     about the internal operation of the DB.
   //  "leveldb.sstables" - returns a multi-line string that describes all
   //     of the sstables that make up the db contents.
+  //  "leveldb.approximate-memory-usage" - returns the approximate number of
+  //     bytes of memory in use by the DB.
   virtual bool GetProperty(const Slice& property, std::string* value) = 0;
 
   // For each i in [0,n-1], store in "sizes[i]", the approximate
@@ -134,27 +141,27 @@ class DB {
   // needed to access the data.  This operation should typically only
   // be invoked by users who understand the underlying implementation.
   //
-  // begin==NULL is treated as a key before all keys in the database.
-  // end==NULL is treated as a key after all keys in the database.
+  // begin==nullptr is treated as a key before all keys in the database.
+  // end==nullptr is treated as a key after all keys in the database.
   // Therefore the following call will compact the entire database:
-  //    db->CompactRange(NULL, NULL);
+  //    db->CompactRange(nullptr, nullptr);
   virtual void CompactRange(const Slice* begin, const Slice* end) = 0;
-
- private:
-  // No copying allowed
-  DB(const DB&);
-  void operator=(const DB&);
 };
 
 // Destroy the contents of the specified database.
 // Be very careful using this method.
-Status DestroyDB(const std::string& name, const Options& options);
+//
+// Note: For backwards compatibility, if DestroyDB is unable to list the
+// database files, Status::OK() will still be returned masking this failure.
+LEVELDB_EXPORT Status DestroyDB(const std::string& name,
+                                const Options& options);
 
 // If a DB cannot be opened, you may attempt to call this method to
 // resurrect as much of the contents of the database as possible.
 // Some data may be lost, so be careful when calling this function
 // on a database that contains important information.
-Status RepairDB(const std::string& dbname, const Options& options);
+LEVELDB_EXPORT Status RepairDB(const std::string& dbname,
+                               const Options& options);
 
 }  // namespace leveldb
 

include/leveldb/dumpfile.h

@@ -7,6 +7,7 @@
 
 #include <string>
 #include "leveldb/env.h"
+#include "leveldb/export.h"
 #include "leveldb/status.h"
 
 namespace leveldb {
@@ -18,7 +19,8 @@ namespace leveldb {
 //
 // Returns a non-OK result if fname does not name a leveldb storage
 // file, or if the file cannot be read.
-Status DumpFile(Env* env, const std::string& fname, WritableFile* dst);
+LEVELDB_EXPORT Status DumpFile(Env* env, const std::string& fname,
+                               WritableFile* dst);
 
 }  // namespace leveldb
 

include/leveldb/env.h

@@ -13,12 +13,34 @@
 #ifndef STORAGE_LEVELDB_INCLUDE_ENV_H_
 #define STORAGE_LEVELDB_INCLUDE_ENV_H_
 
-#include <string>
-#include <vector>
 #include <stdarg.h>
 #include <stdint.h>
+#include <string>
+#include <vector>
+#include "leveldb/export.h"
 #include "leveldb/status.h"
 
+#if defined(_WIN32)
+// The leveldb::Env class below contains a DeleteFile method.
+// At the same time, <windows.h>, a fairly popular header
+// file for Windows applications, defines a DeleteFile macro.
+//
+// Without any intervention on our part, the result of this
+// unfortunate coincidence is that the name of the
+// leveldb::Env::DeleteFile method seen by the compiler depends on
+// whether <windows.h> was included before or after the LevelDB
+// headers.
+//
+// To avoid headaches, we undefined DeleteFile (if defined) and
+// redefine it at the bottom of this file. This way <windows.h>
+// can be included before this file (or not at all) and the
+// exported method will always be leveldb::Env::DeleteFile.
+#if defined(DeleteFile)
+#undef DeleteFile
+#define LEVELDB_DELETEFILE_UNDEFINED
+#endif  // defined(DeleteFile)
+#endif  // defined(_WIN32)
+
 namespace leveldb {
 
 class FileLock;
@@ -28,9 +50,13 @@ class SequentialFile;
 class Slice;
 class WritableFile;
 
-class Env {
+class LEVELDB_EXPORT Env {
  public:
-  Env() { }
+  Env() = default;
+
+  Env(const Env&) = delete;
+  Env& operator=(const Env&) = delete;
+
   virtual ~Env();
 
   // Return a default environment suitable for the current operating
@@ -40,20 +66,22 @@ class Env {
   // The result of Default() belongs to leveldb and must never be deleted.
   static Env* Default();
 
-  // Create a brand new sequentially-readable file with the specified name.
+  // Create an object that sequentially reads the file with the specified name.
   // On success, stores a pointer to the new file in *result and returns OK.
-  // On failure stores NULL in *result and returns non-OK.  If the file does
-  // not exist, returns a non-OK status.
+  // On failure stores nullptr in *result and returns non-OK.  If the file does
+  // not exist, returns a non-OK status.  Implementations should return a
+  // NotFound status when the file does not exist.
   //
   // The returned file will only be accessed by one thread at a time.
   virtual Status NewSequentialFile(const std::string& fname,
                                    SequentialFile** result) = 0;
 
-  // Create a brand new random access read-only file with the
+  // Create an object supporting random-access reads from the file with the
   // specified name.  On success, stores a pointer to the new file in
-  // *result and returns OK.  On failure stores NULL in *result and
+  // *result and returns OK.  On failure stores nullptr in *result and
   // returns non-OK.  If the file does not exist, returns a non-OK
-  // status.
+  // status.  Implementations should return a NotFound status when the file does
+  // not exist.
   //
   // The returned file may be concurrently accessed by multiple threads.
   virtual Status NewRandomAccessFile(const std::string& fname,
@@ -62,13 +90,28 @@ class Env {
   // Create an object that writes to a new file with the specified
   // name.  Deletes any existing file with the same name and creates a
   // new file.  On success, stores a pointer to the new file in
-  // *result and returns OK.  On failure stores NULL in *result and
+  // *result and returns OK.  On failure stores nullptr in *result and
   // returns non-OK.
   //
   // The returned file will only be accessed by one thread at a time.
   virtual Status NewWritableFile(const std::string& fname,
                                  WritableFile** result) = 0;
 
+  // Create an object that either appends to an existing file, or
+  // writes to a new file (if the file does not exist to begin with).
+  // On success, stores a pointer to the new file in *result and
+  // returns OK.  On failure stores nullptr in *result and returns
+  // non-OK.
+  //
+  // The returned file will only be accessed by one thread at a time.
+  //
+  // May return an IsNotSupportedError error if this Env does
+  // not allow appending to an existing file.  Users of Env (including
+  // the leveldb implementation) must be prepared to deal with
+  // an Env that does not support appending.
+  virtual Status NewAppendableFile(const std::string& fname,
+                                   WritableFile** result);
+
   // Returns true iff the named file exists.
   virtual bool FileExists(const std::string& fname) = 0;
 
@@ -95,7 +138,7 @@ class Env {
                             const std::string& target) = 0;
 
   // Lock the specified file.  Used to prevent concurrent access to
-  // the same db by multiple processes.  On failure, stores NULL in
+  // the same db by multiple processes.  On failure, stores nullptr in
   // *lock and returns non-OK.
   //
   // On success, stores a pointer to the object that represents the
@@ -144,17 +187,16 @@ class Env {
 
   // Sleep/delay the thread for the prescribed number of micro-seconds.
   virtual void SleepForMicroseconds(int micros) = 0;
-
- private:
-  // No copying allowed
-  Env(const Env&);
-  void operator=(const Env&);
 };
 
 // A file abstraction for reading sequentially through a file
-class SequentialFile {
+class LEVELDB_EXPORT SequentialFile {
  public:
-  SequentialFile() { }
+  SequentialFile() = default;
+
+  SequentialFile(const SequentialFile&) = delete;
+  SequentialFile& operator=(const SequentialFile&) = delete;
+
   virtual ~SequentialFile();
 
   // Read up to "n" bytes from the file.  "scratch[0..n-1]" may be
@@ -175,17 +217,16 @@ class SequentialFile {
   //
   // REQUIRES: External synchronization
   virtual Status Skip(uint64_t n) = 0;
-
- private:
-  // No copying allowed
-  SequentialFile(const SequentialFile&);
-  void operator=(const SequentialFile&);
 };
 
 // A file abstraction for randomly reading the contents of a file.
-class RandomAccessFile {
+class LEVELDB_EXPORT RandomAccessFile {
  public:
-  RandomAccessFile() { }
+  RandomAccessFile() = default;
+
+  RandomAccessFile(const RandomAccessFile&) = delete;
+  RandomAccessFile& operator=(const RandomAccessFile&) = delete;
+
   virtual ~RandomAccessFile();
 
   // Read up to "n" bytes from the file starting at "offset".
@@ -199,135 +240,150 @@ class RandomAccessFile {
   // Safe for concurrent use by multiple threads.
   virtual Status Read(uint64_t offset, size_t n, Slice* result,
                       char* scratch) const = 0;
-
- private:
-  // No copying allowed
-  RandomAccessFile(const RandomAccessFile&);
-  void operator=(const RandomAccessFile&);
 };
 
 // A file abstraction for sequential writing.  The implementation
 // must provide buffering since callers may append small fragments
 // at a time to the file.
-class WritableFile {
+class LEVELDB_EXPORT WritableFile {
  public:
-  WritableFile() { }
+  WritableFile() = default;
+
+  WritableFile(const WritableFile&) = delete;
+  WritableFile& operator=(const WritableFile&) = delete;
+
   virtual ~WritableFile();
 
   virtual Status Append(const Slice& data) = 0;
   virtual Status Close() = 0;
   virtual Status Flush() = 0;
   virtual Status Sync() = 0;
-
- private:
-  // No copying allowed
-  WritableFile(const WritableFile&);
-  void operator=(const WritableFile&);
 };
 
 // An interface for writing log messages.
-class Logger {
+class LEVELDB_EXPORT Logger {
  public:
-  Logger() { }
+  Logger() = default;
+
+  Logger(const Logger&) = delete;
+  Logger& operator=(const Logger&) = delete;
+
   virtual ~Logger();
 
   // Write an entry to the log file with the specified format.
   virtual void Logv(const char* format, va_list ap) = 0;
-
- private:
-  // No copying allowed
-  Logger(const Logger&);
-  void operator=(const Logger&);
 };
 
-
 // Identifies a locked file.
-class FileLock {
+class LEVELDB_EXPORT FileLock {
  public:
-  FileLock() { }
+  FileLock() = default;
+
+  FileLock(const FileLock&) = delete;
+  FileLock& operator=(const FileLock&) = delete;
+
   virtual ~FileLock();
- private:
-  // No copying allowed
-  FileLock(const FileLock&);
-  void operator=(const FileLock&);
 };
 
-// Log the specified data to *info_log if info_log is non-NULL.
-extern void Log(Logger* info_log, const char* format, ...)
+// Log the specified data to *info_log if info_log is non-null.
+void Log(Logger* info_log, const char* format, ...)
 #   if defined(__GNUC__) || defined(__clang__)
     __attribute__((__format__ (__printf__, 2, 3)))
 #   endif
     ;
 
 // A utility routine: write "data" to the named file.
-extern Status WriteStringToFile(Env* env, const Slice& data,
-                                const std::string& fname);
+LEVELDB_EXPORT Status WriteStringToFile(Env* env, const Slice& data,
+                                        const std::string& fname);
 
 // A utility routine: read contents of named file into *data
-extern Status ReadFileToString(Env* env, const std::string& fname,
-                               std::string* data);
+LEVELDB_EXPORT Status ReadFileToString(Env* env, const std::string& fname,
+                                       std::string* data);
 
 // An implementation of Env that forwards all calls to another Env.
 // May be useful to clients who wish to override just part of the
 // functionality of another Env.
-class EnvWrapper : public Env {
+class LEVELDB_EXPORT EnvWrapper : public Env {
  public:
-  // Initialize an EnvWrapper that delegates all calls to *t
+  // Initialize an EnvWrapper that delegates all calls to *t.
   explicit EnvWrapper(Env* t) : target_(t) { }
   virtual ~EnvWrapper();
 
-  // Return the target to which this Env forwards all calls
+  // Return the target to which this Env forwards all calls.
   Env* target() const { return target_; }
 
-  // The following text is boilerplate that forwards all methods to target()
-  Status NewSequentialFile(const std::string& f, SequentialFile** r) {
+  // The following text is boilerplate that forwards all methods to target().
+  Status NewSequentialFile(const std::string& f, SequentialFile** r) override {
     return target_->NewSequentialFile(f, r);
   }
-  Status NewRandomAccessFile(const std::string& f, RandomAccessFile** r) {
+  Status NewRandomAccessFile(const std::string& f,
+                             RandomAccessFile** r) override {
     return target_->NewRandomAccessFile(f, r);
   }
-  Status NewWritableFile(const std::string& f, WritableFile** r) {
+  Status NewWritableFile(const std::string& f, WritableFile** r) override {
     return target_->NewWritableFile(f, r);
   }
-  bool FileExists(const std::string& f) { return target_->FileExists(f); }
-  Status GetChildren(const std::string& dir, std::vector<std::string>* 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 DeleteFile(const std::string& f) { return target_->DeleteFile(f); }
-  Status CreateDir(const std::string& d) { return target_->CreateDir(d); }
-  Status DeleteDir(const std::string& d) { return target_->DeleteDir(d); }
-  Status GetFileSize(const std::string& f, uint64_t* s) {
+  Status DeleteFile(const std::string& f) override {
+    return target_->DeleteFile(f);
+  }
+  Status CreateDir(const std::string& d) override {
+    return target_->CreateDir(d);
+  }
+  Status DeleteDir(const std::string& d) override {
+    return target_->DeleteDir(d);
+  }
+  Status GetFileSize(const std::string& f, uint64_t* s) override {
     return target_->GetFileSize(f, s);
   }
-  Status RenameFile(const std::string& s, const std::string& t) {
+  Status RenameFile(const std::string& s, const std::string& t) override {
     return target_->RenameFile(s, t);
   }
-  Status LockFile(const std::string& f, FileLock** l) {
+  Status LockFile(const std::string& f, FileLock** l) override {
     return target_->LockFile(f, l);
   }
-  Status UnlockFile(FileLock* l) { return target_->UnlockFile(l); }
-  void Schedule(void (*f)(void*), void* a) {
+  Status UnlockFile(FileLock* l) override { return target_->UnlockFile(l); }
+  void Schedule(void (*f)(void*), void* a) override {
     return target_->Schedule(f, a);
   }
-  void StartThread(void (*f)(void*), void* a) {
+  void StartThread(void (*f)(void*), void* a) override {
     return target_->StartThread(f, a);
   }
-  virtual Status GetTestDirectory(std::string* path) {
+  Status GetTestDirectory(std::string* path) override {
     return target_->GetTestDirectory(path);
   }
-  virtual Status NewLogger(const std::string& fname, Logger** result) {
+  Status NewLogger(const std::string& fname, Logger** result) override {
     return target_->NewLogger(fname, result);
   }
-  uint64_t NowMicros() {
+  uint64_t NowMicros() override {
     return target_->NowMicros();
   }
-  void SleepForMicroseconds(int micros) {
+  void SleepForMicroseconds(int micros) override {
     target_->SleepForMicroseconds(micros);
   }
+
  private:
   Env* target_;
 };
 
 }  // namespace leveldb
 
+// Redefine DeleteFile if necessary.
+#if defined(_WIN32) && defined(LEVELDB_DELETEFILE_UNDEFINED)
+#if defined(UNICODE)
+#define DeleteFile DeleteFileW
+#else
+#define DeleteFile DeleteFileA
+#endif  // defined(UNICODE)
+#endif  // defined(_WIN32) && defined(LEVELDB_DELETEFILE_UNDEFINED)
+
 #endif  // STORAGE_LEVELDB_INCLUDE_ENV_H_

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_

include/leveldb/filter_policy.h

@@ -17,12 +17,13 @@
 #define STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_
 
 #include <string>
+#include "leveldb/export.h"
 
 namespace leveldb {
 
 class Slice;
 
-class FilterPolicy {
+class LEVELDB_EXPORT FilterPolicy {
  public:
   virtual ~FilterPolicy();
 
@@ -63,8 +64,8 @@ class FilterPolicy {
 // ignores trailing spaces, it would be incorrect to use a
 // FilterPolicy (like NewBloomFilterPolicy) that does not ignore
 // trailing spaces in keys.
-extern const FilterPolicy* NewBloomFilterPolicy(int bits_per_key);
+LEVELDB_EXPORT const FilterPolicy* NewBloomFilterPolicy(int bits_per_key);
 
-}
+}  // namespace leveldb
 
 #endif  // STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_

include/leveldb/iterator.h

@@ -15,14 +15,19 @@
 #ifndef STORAGE_LEVELDB_INCLUDE_ITERATOR_H_
 #define STORAGE_LEVELDB_INCLUDE_ITERATOR_H_
 
+#include "leveldb/export.h"
 #include "leveldb/slice.h"
 #include "leveldb/status.h"
 
 namespace leveldb {
 
-class Iterator {
+class LEVELDB_EXPORT Iterator {
  public:
   Iterator();
+
+  Iterator(const Iterator&) = delete;
+  Iterator& operator=(const Iterator&) = delete;
+
   virtual ~Iterator();
 
   // An iterator is either positioned at a key/value pair, or
@@ -37,7 +42,7 @@ class Iterator {
   // Valid() after this call iff the source is not empty.
   virtual void SeekToLast() = 0;
 
-  // Position at the first key in the source that at or past target
+  // 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;
@@ -72,28 +77,32 @@ class Iterator {
   //
   // Note that unlike all of the preceding methods, this method is
   // not abstract and therefore clients should not override it.
-  typedef void (*CleanupFunction)(void* arg1, void* arg2);
+  using CleanupFunction = void (*)(void* arg1, void* arg2);
   void RegisterCleanup(CleanupFunction function, void* arg1, void* arg2);
 
  private:
-  struct Cleanup {
+  // Cleanup functions are stored in a single-linked list.
+  // The list's head node is inlined in the iterator.
+  struct CleanupNode {
+    // The head node is used if the function pointer is not null.
     CleanupFunction function;
     void* arg1;
     void* arg2;
-    Cleanup* next;
-  };
-  Cleanup cleanup_;
+    CleanupNode* next;
 
-  // No copying allowed
-  Iterator(const Iterator&);
-  void operator=(const Iterator&);
+    // 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); }
+  };
+  CleanupNode cleanup_head_;
 };
 
 // Return an empty iterator (yields nothing).
-extern Iterator* NewEmptyIterator();
+LEVELDB_EXPORT Iterator* NewEmptyIterator();
 
 // Return an empty iterator with the specified status.
-extern Iterator* NewErrorIterator(const Status& status);
+LEVELDB_EXPORT Iterator* NewErrorIterator(const Status& status);
 
 }  // namespace leveldb
 

include/leveldb/options.h

@@ -6,6 +6,7 @@
 #define STORAGE_LEVELDB_INCLUDE_OPTIONS_H_
 
 #include <stddef.h>
+#include "leveldb/export.h"
 
 namespace leveldb {
 
@@ -28,7 +29,7 @@ enum CompressionType {
 };
 
 // Options to control the behavior of a database (passed to DB::Open)
-struct Options {
+struct LEVELDB_EXPORT Options {
   // -------------------
   // Parameters that affect behavior
 
@@ -41,20 +42,17 @@ struct Options {
   const Comparator* comparator;
 
   // If true, the database will be created if it is missing.
-  // Default: false
-  bool create_if_missing;
+  bool create_if_missing = false;
 
   // If true, an error is raised if the database already exists.
-  // Default: false
-  bool error_if_exists;
+  bool error_if_exists = false;
 
   // If true, the implementation will do aggressive checking of the
   // data it is processing and will stop early if it detects any
   // errors.  This may have unforeseen ramifications: for example, a
   // corruption of one DB entry may cause a large number of entries to
   // become unreadable or for the entire DB to become unopenable.
-  // Default: false
-  bool paranoid_checks;
+  bool paranoid_checks = false;
 
   // Use the specified object to interact with the environment,
   // e.g. to read/write files, schedule background work, etc.
@@ -62,10 +60,9 @@ struct Options {
   Env* env;
 
   // Any internal progress/error information generated by the db will
-  // be written to info_log if it is non-NULL, or to a file stored
-  // in the same directory as the DB contents if info_log is NULL.
-  // Default: NULL
-  Logger* info_log;
+  // be written to info_log if it is non-null, or to a file stored
+  // in the same directory as the DB contents if info_log is null.
+  Logger* info_log = nullptr;
 
   // -------------------
   // Parameters that affect performance
@@ -78,39 +75,40 @@ struct Options {
   // so you may wish to adjust this parameter to control memory usage.
   // Also, a larger write buffer will result in a longer recovery time
   // the next time the database is opened.
-  //
-  // Default: 4MB
-  size_t write_buffer_size;
+  size_t write_buffer_size = 4 * 1024 * 1024;
 
   // Number of open files that can be used by the DB.  You may need to
   // increase this if your database has a large working set (budget
   // one open file per 2MB of working set).
-  //
-  // Default: 1000
-  int max_open_files;
+  int max_open_files = 1000;
 
   // Control over blocks (user data is stored in a set of blocks, and
   // a block is the unit of reading from disk).
 
-  // If non-NULL, use the specified cache for blocks.
-  // If NULL, leveldb will automatically create and use an 8MB internal cache.
-  // Default: NULL
-  Cache* block_cache;
+  // If non-null, use the specified cache for blocks.
+  // If null, leveldb will automatically create and use an 8MB internal cache.
+  Cache* block_cache = nullptr;
 
   // Approximate size of user data packed per block.  Note that the
   // block size specified here corresponds to uncompressed data.  The
   // actual size of the unit read from disk may be smaller if
   // compression is enabled.  This parameter can be changed dynamically.
-  //
-  // Default: 4K
-  size_t block_size;
+  size_t block_size = 4 * 1024;
 
   // Number of keys between restart points for delta encoding of keys.
   // This parameter can be changed dynamically.  Most clients should
   // leave this parameter alone.
-  //
-  // Default: 16
-  int block_restart_interval;
+  int block_restart_interval = 16;
+
+  // Leveldb will write up to this amount of bytes to a file before
+  // switching to a new one.
+  // Most clients should leave this parameter alone.  However if your
+  // filesystem is more efficient with larger files, you could
+  // consider increasing the value.  The downside will be longer
+  // compactions and hence longer latency/performance hiccups.
+  // Another reason to increase this parameter might be when you are
+  // initially populating a large database.
+  size_t max_file_size = 2 * 1024 * 1024;
 
   // Compress blocks using the specified compression algorithm.  This
   // parameter can be changed dynamically.
@@ -126,47 +124,44 @@ struct Options {
   // worth switching to kNoCompression.  Even if the input data is
   // incompressible, the kSnappyCompression implementation will
   // efficiently detect that and will switch to uncompressed mode.
-  CompressionType compression;
+  CompressionType compression = kSnappyCompression;
 
-  // If non-NULL, use the specified filter policy to reduce disk reads.
+  // EXPERIMENTAL: If true, append to existing MANIFEST and log files
+  // when a database is opened.  This can significantly speed up open.
+  //
+  // Default: currently false, but may become true later.
+  bool reuse_logs = false;
+
+  // If non-null, use the specified filter policy to reduce disk reads.
   // Many applications will benefit from passing the result of
   // NewBloomFilterPolicy() here.
-  //
-  // Default: NULL
-  const FilterPolicy* filter_policy;
+  const FilterPolicy* filter_policy = nullptr;
 
   // Create an Options object with default values for all fields.
   Options();
 };
 
 // Options that control read operations
-struct ReadOptions {
+struct LEVELDB_EXPORT ReadOptions {
   // If true, all data read from underlying storage will be
   // verified against corresponding checksums.
-  // Default: false
-  bool verify_checksums;
+  bool verify_checksums = false;
 
   // Should the data read for this iteration be cached in memory?
   // Callers may wish to set this field to false for bulk scans.
-  // Default: true
-  bool fill_cache;
+  bool fill_cache = true;
 
-  // If "snapshot" is non-NULL, read as of the supplied snapshot
+  // If "snapshot" is non-null, read as of the supplied snapshot
   // (which must belong to the DB that is being read and which must
-  // not have been released).  If "snapshot" is NULL, use an implicit
+  // not have been released).  If "snapshot" is null, use an implicit
   // snapshot of the state at the beginning of this read operation.
-  // Default: NULL
-  const Snapshot* snapshot;
+  const Snapshot* snapshot = nullptr;
 
-  ReadOptions()
-      : verify_checksums(false),
-        fill_cache(true),
-        snapshot(NULL) {
-  }
+  ReadOptions() = default;
 };
 
 // Options that control write operations
-struct WriteOptions {
+struct LEVELDB_EXPORT WriteOptions {
   // If true, the write will be flushed from the operating system
   // buffer cache (by calling WritableFile::Sync()) before the write
   // is considered complete.  If this flag is true, writes will be
@@ -181,13 +176,9 @@ struct WriteOptions {
   // crash semantics as the "write()" system call.  A DB write
   // with sync==true has similar crash semantics to a "write()"
   // system call followed by "fsync()".
-  //
-  // Default: false
-  bool sync;
+  bool sync = false;
 
-  WriteOptions()
-      : sync(false) {
-  }
+  WriteOptions() = default;
 };
 
 }  // namespace leveldb

include/leveldb/slice.h

@@ -19,10 +19,11 @@
 #include <stddef.h>
 #include <string.h>
 #include <string>
+#include "leveldb/export.h"
 
 namespace leveldb {
 
-class Slice {
+class LEVELDB_EXPORT Slice {
  public:
   // Create an empty slice.
   Slice() : data_(""), size_(0) { }
@@ -36,6 +37,10 @@ class Slice {
   // 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_; }
 
@@ -80,8 +85,6 @@ class Slice {
  private:
   const char* data_;
   size_t size_;
-
-  // Intentionally copyable
 };
 
 inline bool operator==(const Slice& x, const Slice& y) {

include/leveldb/status.h

@@ -13,20 +13,24 @@
 #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 Status {
+class LEVELDB_EXPORT Status {
  public:
   // Create a success status.
-  Status() : state_(NULL) { }
+  Status() noexcept : state_(nullptr) { }
   ~Status() { delete[] state_; }
 
-  // Copy the specified status.
-  Status(const Status& s);
-  void operator=(const Status& s);
+  Status(const Status& rhs);
+  Status& operator=(const Status& rhs);
+
+  Status(Status&& rhs) noexcept : state_(rhs.state_) { rhs.state_ = nullptr; }
+  Status& operator=(Status&& rhs) noexcept;
 
   // Return a success status.
   static Status OK() { return Status(); }
@@ -49,7 +53,7 @@ class Status {
   }
 
   // Returns true iff the status indicates success.
-  bool ok() const { return (state_ == NULL); }
+  bool ok() const { return (state_ == nullptr); }
 
   // Returns true iff the status indicates a NotFound error.
   bool IsNotFound() const { return code() == kNotFound; }
@@ -60,12 +64,18 @@ class Status {
   // Returns true iff the status indicates an IOError.
   bool IsIOError() const { return code() == kIOError; }
 
+  // Returns true iff the status indicates a NotSupportedError.
+  bool IsNotSupportedError() const { return code() == kNotSupported; }
+
+  // Returns true iff the status indicates an InvalidArgument.
+  bool IsInvalidArgument() const { return code() == kInvalidArgument; }
+
   // Return a string representation of this status suitable for printing.
   // Returns the string "OK" for success.
   std::string ToString() const;
 
  private:
-  // OK status has a NULL state_.  Otherwise, state_ is a new[] array
+  // 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
@@ -82,23 +92,28 @@ class Status {
   };
 
   Code code() const {
-    return (state_ == NULL) ? kOk : static_cast<Code>(state_[4]);
+    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);
 };
 
-inline Status::Status(const Status& s) {
-  state_ = (s.state_ == NULL) ? NULL : CopyState(s.state_);
+inline Status::Status(const Status& rhs) {
+  state_ = (rhs.state_ == nullptr) ? nullptr : CopyState(rhs.state_);
 }
-inline void Status::operator=(const Status& s) {
-  // The following condition catches both aliasing (when this == &s),
-  // and the common case where both s and *this are ok.
-  if (state_ != s.state_) {
+inline Status& Status::operator=(const Status& rhs) {
+  // The following condition catches both aliasing (when this == &rhs),
+  // and the common case where both rhs and *this are ok.
+  if (state_ != rhs.state_) {
     delete[] state_;
-    state_ = (s.state_ == NULL) ? NULL : CopyState(s.state_);
+    state_ = (rhs.state_ == nullptr) ? nullptr : CopyState(rhs.state_);
   }
+  return *this;
+}
+inline Status& Status::operator=(Status&& rhs) noexcept {
+  std::swap(state_, rhs.state_);
+  return *this;
 }
 
 }  // namespace leveldb

include/leveldb/table.h

@@ -6,6 +6,7 @@
 #define STORAGE_LEVELDB_INCLUDE_TABLE_H_
 
 #include <stdint.h>
+#include "leveldb/export.h"
 #include "leveldb/iterator.h"
 
 namespace leveldb {
@@ -21,7 +22,7 @@ class TableCache;
 // A Table is a sorted map from strings to strings.  Tables are
 // immutable and persistent.  A Table may be safely accessed from
 // multiple threads without external synchronization.
-class Table {
+class LEVELDB_EXPORT Table {
  public:
   // Attempt to open the table that is stored in bytes [0..file_size)
   // of "file", and read the metadata entries necessary to allow
@@ -30,7 +31,7 @@ class Table {
   // If successful, returns ok and sets "*table" to the newly opened
   // table.  The client should delete "*table" when no longer needed.
   // If there was an error while initializing the table, sets "*table"
-  // to NULL and returns a non-ok status.  Does not take ownership of
+  // to nullptr and returns a non-ok status.  Does not take ownership of
   // "*source", but the client must ensure that "source" remains live
   // for the duration of the returned table's lifetime.
   //
@@ -40,6 +41,9 @@ class Table {
                      uint64_t file_size,
                      Table** table);
 
+  Table(const Table&) = delete;
+  void operator=(const Table&) = delete;
+
   ~Table();
 
   // Returns a new iterator over the table contents.
@@ -74,10 +78,6 @@ class Table {
 
   void ReadMeta(const Footer& footer);
   void ReadFilter(const Slice& filter_handle_value);
-
-  // No copying allowed
-  Table(const Table&);
-  void operator=(const Table&);
 };
 
 }  // namespace leveldb

include/leveldb/table_builder.h

@@ -14,6 +14,7 @@
 #define STORAGE_LEVELDB_INCLUDE_TABLE_BUILDER_H_
 
 #include <stdint.h>
+#include "leveldb/export.h"
 #include "leveldb/options.h"
 #include "leveldb/status.h"
 
@@ -23,13 +24,16 @@ class BlockBuilder;
 class BlockHandle;
 class WritableFile;
 
-class TableBuilder {
+class LEVELDB_EXPORT TableBuilder {
  public:
   // Create a builder that will store the contents of the table it is
   // building in *file.  Does not close the file.  It is up to the
   // caller to close the file after calling Finish().
   TableBuilder(const Options& options, WritableFile* file);
 
+  TableBuilder(const TableBuilder&) = delete;
+  void operator=(const TableBuilder&) = delete;
+
   // REQUIRES: Either Finish() or Abandon() has been called.
   ~TableBuilder();
 
@@ -81,10 +85,6 @@ class TableBuilder {
 
   struct Rep;
   Rep* rep_;
-
-  // No copying allowed
-  TableBuilder(const TableBuilder&);
-  void operator=(const TableBuilder&);
 };
 
 }  // namespace leveldb

include/leveldb/write_batch.h

@@ -22,15 +22,21 @@
 #define STORAGE_LEVELDB_INCLUDE_WRITE_BATCH_H_
 
 #include <string>
+#include "leveldb/export.h"
 #include "leveldb/status.h"
 
 namespace leveldb {
 
 class Slice;
 
-class WriteBatch {
+class LEVELDB_EXPORT WriteBatch {
  public:
   WriteBatch();
+
+  // Intentionally copyable.
+  WriteBatch(const WriteBatch&) = default;
+  WriteBatch& operator =(const WriteBatch&) = default;
+
   ~WriteBatch();
 
   // Store the mapping "key->value" in the database.
@@ -42,6 +48,19 @@ class WriteBatch {
   // Clear all updates buffered in this batch.
   void Clear();
 
+  // The size of the database changes caused by this batch.
+  //
+  // This number is tied to implementation details, and may change across
+  // releases. It is intended for LevelDB usage metrics.
+  size_t ApproximateSize() const;
+
+  // Copies the operations in "source" to this batch.
+  //
+  // This runs in O(source size) time. However, the constant factor is better
+  // than calling Iterate() over the source batch with a Handler that replicates
+  // the operations into this batch.
+  void Append(const WriteBatch& source);
+
   // Support for iterating over the contents of a batch.
   class Handler {
    public:
@@ -55,8 +74,6 @@ class WriteBatch {
   friend class WriteBatchInternal;
 
   std::string rep_;  // See comment in write_batch.cc for the format of rep_
-
-  // Intentionally copyable
 };
 
 }  // namespace leveldb

issues/issue320_test.cc

@@ -0,0 +1,128 @@
+// Copyright (c) 2019 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include <cstdint>
+#include <cstdlib>
+#include <iostream>
+#include <memory>
+#include <string>
+#include <vector>
+
+#include "leveldb/db.h"
+#include "leveldb/write_batch.h"
+#include "util/testharness.h"
+
+namespace leveldb {
+
+namespace {
+
+// Creates a random number in the range of [0, max).
+int GenerateRandomNumber(int max) { return std::rand() % max; }
+
+std::string CreateRandomString(int32_t index) {
+  static const size_t len = 1024;
+  char bytes[len];
+  size_t i = 0;
+  while (i < 8) {
+    bytes[i] = 'a' + ((index >> (4 * i)) & 0xf);
+    ++i;
+  }
+  while (i < sizeof(bytes)) {
+    bytes[i] = 'a' + GenerateRandomNumber(26);
+    ++i;
+  }
+  return std::string(bytes, sizeof(bytes));
+}
+
+}  // namespace
+
+class Issue320 {};
+
+TEST(Issue320, Test) {
+  std::srand(0);
+
+  bool delete_before_put = false;
+  bool keep_snapshots = true;
+
+  std::vector<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 = test::TmpDir() + "/leveldb_issue320_test";
+  ASSERT_OK(DB::Open(options, dbpath, &db));
+
+  uint32_t target_size = 10000;
+  uint32_t num_items = 0;
+  uint32_t count = 0;
+  std::string key;
+  std::string value, old_value;
+
+  WriteOptions writeOptions;
+  ReadOptions readOptions;
+  while (count < 200000) {
+    if ((++count % 1000) == 0) {
+      std::cout << "count: " << count << std::endl;
+    }
+
+    int index = GenerateRandomNumber(test_map.size());
+    WriteBatch batch;
+
+    if (test_map[index] == nullptr) {
+      num_items++;
+      test_map[index].reset(new std::pair<std::string, std::string>(
+          CreateRandomString(index), CreateRandomString(index)));
+      batch.Put(test_map[index]->first, test_map[index]->second);
+    } else {
+      ASSERT_OK(db->Get(readOptions, test_map[index]->first, &old_value));
+      if (old_value != test_map[index]->second) {
+        std::cout << "ERROR incorrect value returned by Get" << std::endl;
+        std::cout << "  count=" << count << std::endl;
+        std::cout << "  old value=" << old_value << std::endl;
+        std::cout << "  test_map[index]->second=" << test_map[index]->second
+                  << std::endl;
+        std::cout << "  test_map[index]->first=" << test_map[index]->first
+                  << std::endl;
+        std::cout << "  index=" << index << std::endl;
+        ASSERT_EQ(old_value, test_map[index]->second);
+      }
+
+      if (num_items >= target_size && GenerateRandomNumber(100) > 30) {
+        batch.Delete(test_map[index]->first);
+        test_map[index] = nullptr;
+        --num_items;
+      } else {
+        test_map[index]->second = CreateRandomString(index);
+        if (delete_before_put) batch.Delete(test_map[index]->first);
+        batch.Put(test_map[index]->first, test_map[index]->second);
+      }
+    }
+
+    ASSERT_OK(db->Write(writeOptions, &batch));
+
+    if (keep_snapshots && GenerateRandomNumber(10) == 0) {
+      int i = GenerateRandomNumber(snapshots.size());
+      if (snapshots[i] != nullptr) {
+        db->ReleaseSnapshot(snapshots[i]);
+      }
+      snapshots[i] = db->GetSnapshot();
+    }
+  }
+
+  for (Snapshot const* snapshot : snapshots) {
+    if (snapshot) {
+      db->ReleaseSnapshot(snapshot);
+    }
+  }
+
+  delete db;
+  DestroyDB(dbpath, options);
+}
+
+}  // namespace leveldb
+
+int main(int argc, char** argv) { return leveldb::test::RunAllTests(); }

port/README

@@ -5,6 +5,6 @@ Code in the rest of the package includes "port.h" from this directory.
 "port.h" in turn includes a platform specific "port_<platform>.h" file
 that provides the platform specific implementation.
 
-See port_posix.h for an example of what must be provided in a platform
+See port_stdcxx.h for an example of what must be provided in a platform
 specific header file.
 

port/atomic_pointer.h

@@ -1,223 +0,0 @@
-// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file. See the AUTHORS file for names of contributors.
-
-// AtomicPointer provides storage for a lock-free pointer.
-// Platform-dependent implementation of AtomicPointer:
-// - If the platform provides a cheap barrier, we use it with raw pointers
-// - If <atomic> is present (on newer versions of gcc, it is), we use
-//   a <atomic>-based AtomicPointer.  However we prefer the memory
-//   barrier based version, because at least on a gcc 4.4 32-bit build
-//   on linux, we have encountered a buggy <atomic> implementation.
-//   Also, some <atomic> implementations are much slower than a memory-barrier
-//   based implementation (~16ns for <atomic> based acquire-load vs. ~1ns for
-//   a barrier based acquire-load).
-// This code is based on atomicops-internals-* in Google's perftools:
-// http://code.google.com/p/google-perftools/source/browse/#svn%2Ftrunk%2Fsrc%2Fbase
-
-#ifndef PORT_ATOMIC_POINTER_H_
-#define PORT_ATOMIC_POINTER_H_
-
-#include <stdint.h>
-#ifdef LEVELDB_ATOMIC_PRESENT
-#include <atomic>
-#endif
-#ifdef OS_WIN
-#include <windows.h>
-#endif
-#ifdef OS_MACOSX
-#include <libkern/OSAtomic.h>
-#endif
-
-#if defined(_M_X64) || defined(__x86_64__)
-#define ARCH_CPU_X86_FAMILY 1
-#elif defined(_M_IX86) || defined(__i386__) || defined(__i386)
-#define ARCH_CPU_X86_FAMILY 1
-#elif defined(__ARMEL__)
-#define ARCH_CPU_ARM_FAMILY 1
-#elif defined(__ppc__) || defined(__powerpc__) || defined(__powerpc64__)
-#define ARCH_CPU_PPC_FAMILY 1
-#endif
-
-namespace leveldb {
-namespace port {
-
-// Define MemoryBarrier() if available
-// Windows on x86
-#if defined(OS_WIN) && defined(COMPILER_MSVC) && defined(ARCH_CPU_X86_FAMILY)
-// windows.h already provides a MemoryBarrier(void) macro
-// http://msdn.microsoft.com/en-us/library/ms684208(v=vs.85).aspx
-#define LEVELDB_HAVE_MEMORY_BARRIER
-
-// Mac OS
-#elif defined(OS_MACOSX)
-inline void MemoryBarrier() {
-  OSMemoryBarrier();
-}
-#define LEVELDB_HAVE_MEMORY_BARRIER
-
-// Gcc on x86
-#elif defined(ARCH_CPU_X86_FAMILY) && defined(__GNUC__)
-inline void MemoryBarrier() {
-  // See http://gcc.gnu.org/ml/gcc/2003-04/msg01180.html for a discussion on
-  // this idiom. Also see http://en.wikipedia.org/wiki/Memory_ordering.
-  __asm__ __volatile__("" : : : "memory");
-}
-#define LEVELDB_HAVE_MEMORY_BARRIER
-
-// Sun Studio
-#elif defined(ARCH_CPU_X86_FAMILY) && defined(__SUNPRO_CC)
-inline void MemoryBarrier() {
-  // See http://gcc.gnu.org/ml/gcc/2003-04/msg01180.html for a discussion on
-  // this idiom. Also see http://en.wikipedia.org/wiki/Memory_ordering.
-  asm volatile("" : : : "memory");
-}
-#define LEVELDB_HAVE_MEMORY_BARRIER
-
-// ARM Linux
-#elif defined(ARCH_CPU_ARM_FAMILY) && defined(__linux__)
-typedef void (*LinuxKernelMemoryBarrierFunc)(void);
-// The Linux ARM kernel provides a highly optimized device-specific memory
-// barrier function at a fixed memory address that is mapped in every
-// user-level process.
-//
-// This beats using CPU-specific instructions which are, on single-core
-// devices, un-necessary and very costly (e.g. ARMv7-A "dmb" takes more
-// than 180ns on a Cortex-A8 like the one on a Nexus One). Benchmarking
-// shows that the extra function call cost is completely negligible on
-// multi-core devices.
-//
-inline void MemoryBarrier() {
-  (*(LinuxKernelMemoryBarrierFunc)0xffff0fa0)();
-}
-#define LEVELDB_HAVE_MEMORY_BARRIER
-
-// PPC
-#elif defined(ARCH_CPU_PPC_FAMILY) && defined(__GNUC__)
-inline void MemoryBarrier() {
-  // TODO for some powerpc expert: is there a cheaper suitable variant?
-  // Perhaps by having separate barriers for acquire and release ops.
-  asm volatile("sync" : : : "memory");
-}
-#define LEVELDB_HAVE_MEMORY_BARRIER
-
-#endif
-
-// AtomicPointer built using platform-specific MemoryBarrier()
-#if defined(LEVELDB_HAVE_MEMORY_BARRIER)
-class AtomicPointer {
- private:
-  void* rep_;
- public:
-  AtomicPointer() { }
-  explicit AtomicPointer(void* p) : rep_(p) {}
-  inline void* NoBarrier_Load() const { return rep_; }
-  inline void NoBarrier_Store(void* v) { rep_ = v; }
-  inline void* Acquire_Load() const {
-    void* result = rep_;
-    MemoryBarrier();
-    return result;
-  }
-  inline void Release_Store(void* v) {
-    MemoryBarrier();
-    rep_ = v;
-  }
-};
-
-// AtomicPointer based on <cstdatomic>
-#elif defined(LEVELDB_ATOMIC_PRESENT)
-class AtomicPointer {
- private:
-  std::atomic<void*> rep_;
- public:
-  AtomicPointer() { }
-  explicit AtomicPointer(void* v) : rep_(v) { }
-  inline void* Acquire_Load() const {
-    return rep_.load(std::memory_order_acquire);
-  }
-  inline void Release_Store(void* v) {
-    rep_.store(v, std::memory_order_release);
-  }
-  inline void* NoBarrier_Load() const {
-    return rep_.load(std::memory_order_relaxed);
-  }
-  inline void NoBarrier_Store(void* v) {
-    rep_.store(v, std::memory_order_relaxed);
-  }
-};
-
-// Atomic pointer based on sparc memory barriers
-#elif defined(__sparcv9) && defined(__GNUC__)
-class AtomicPointer {
- private:
-  void* rep_;
- public:
-  AtomicPointer() { }
-  explicit AtomicPointer(void* v) : rep_(v) { }
-  inline void* Acquire_Load() const {
-    void* val;
-    __asm__ __volatile__ (
-        "ldx [%[rep_]], %[val] \n\t"
-         "membar #LoadLoad|#LoadStore \n\t"
-        : [val] "=r" (val)
-        : [rep_] "r" (&rep_)
-        : "memory");
-    return val;
-  }
-  inline void Release_Store(void* v) {
-    __asm__ __volatile__ (
-        "membar #LoadStore|#StoreStore \n\t"
-        "stx %[v], [%[rep_]] \n\t"
-        :
-        : [rep_] "r" (&rep_), [v] "r" (v)
-        : "memory");
-  }
-  inline void* NoBarrier_Load() const { return rep_; }
-  inline void NoBarrier_Store(void* v) { rep_ = v; }
-};
-
-// Atomic pointer based on ia64 acq/rel
-#elif defined(__ia64) && defined(__GNUC__)
-class AtomicPointer {
- private:
-  void* rep_;
- public:
-  AtomicPointer() { }
-  explicit AtomicPointer(void* v) : rep_(v) { }
-  inline void* Acquire_Load() const {
-    void* val    ;
-    __asm__ __volatile__ (
-        "ld8.acq %[val] = [%[rep_]] \n\t"
-        : [val] "=r" (val)
-        : [rep_] "r" (&rep_)
-        : "memory"
-        );
-    return val;
-  }
-  inline void Release_Store(void* v) {
-    __asm__ __volatile__ (
-        "st8.rel [%[rep_]] = %[v]  \n\t"
-        :
-        : [rep_] "r" (&rep_), [v] "r" (v)
-        : "memory"
-        );
-  }
-  inline void* NoBarrier_Load() const { return rep_; }
-  inline void NoBarrier_Store(void* v) { rep_ = v; }
-};
-
-// We have neither MemoryBarrier(), nor <atomic>
-#else
-#error Please implement AtomicPointer for this platform.
-
-#endif
-
-#undef LEVELDB_HAVE_MEMORY_BARRIER
-#undef ARCH_CPU_X86_FAMILY
-#undef ARCH_CPU_ARM_FAMILY
-#undef ARCH_CPU_PPC_FAMILY
-
-}  // namespace port
-}  // namespace leveldb
-
-#endif  // PORT_ATOMIC_POINTER_H_

port/port.h

@@ -10,8 +10,8 @@
 // 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)
-#  include "port/port_posix.h"
+#if defined(LEVELDB_PLATFORM_POSIX) || defined(LEVELDB_PLATFORM_WINDOWS)
+#  include "port/port_stdcxx.h"
 #elif defined(LEVELDB_PLATFORM_CHROMIUM)
 #  include "port/port_chromium.h"
 #endif

port/port_config.h.in

@@ -0,0 +1,34 @@
+// 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 Google CRC32C.
+#if !defined(HAVE_CRC32C)
+#cmakedefine01 HAVE_CRC32C
+#endif  // !defined(HAVE_CRC32C)
+
+// Define to 1 if you have Google Snappy.
+#if !defined(HAVE_SNAPPY)
+#cmakedefine01 HAVE_SNAPPY
+#endif  // !defined(HAVE_SNAPPY)
+
+// Define to 1 if your processor stores words with the most significant byte
+// first (like Motorola and SPARC, unlike Intel and VAX).
+#if !defined(LEVELDB_IS_BIG_ENDIAN)
+#cmakedefine01 LEVELDB_IS_BIG_ENDIAN
+#endif  // !defined(LEVELDB_IS_BIG_ENDIAN)
+
+#endif  // STORAGE_LEVELDB_PORT_PORT_CONFIG_H_

port/port_example.h

@@ -10,6 +10,8 @@
 #ifndef STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_
 #define STORAGE_LEVELDB_PORT_PORT_EXAMPLE_H_
 
+#include "port/thread_annotations.h"
+
 namespace leveldb {
 namespace port {
 
@@ -23,23 +25,23 @@ static const bool kLittleEndian = true /* or some other expression */;
 // ------------------ Threading -------------------
 
 // A Mutex represents an exclusive lock.
-class Mutex {
+class LOCKABLE Mutex {
  public:
   Mutex();
   ~Mutex();
 
   // Lock the mutex.  Waits until other lockers have exited.
   // Will deadlock if the mutex is already locked by this thread.
-  void Lock();
+  void Lock() EXCLUSIVE_LOCK_FUNCTION();
 
   // Unlock the mutex.
   // REQUIRES: This mutex was locked by this thread.
-  void Unlock();
+  void Unlock() UNLOCK_FUNCTION();
 
   // Optionally crash if this thread does not hold this mutex.
   // The implementation must be fast, especially if NDEBUG is
   // defined.  The implementation is allowed to skip all checks.
-  void AssertHeld();
+  void AssertHeld() ASSERT_EXCLUSIVE_LOCK();
 };
 
 class CondVar {
@@ -60,57 +62,18 @@ class CondVar {
   void SignallAll();
 };
 
-// Thread-safe initialization.
-// Used as follows:
-//      static port::OnceType init_control = LEVELDB_ONCE_INIT;
-//      static void Initializer() { ... do something ...; }
-//      ...
-//      port::InitOnce(&init_control, &Initializer);
-typedef intptr_t OnceType;
-#define LEVELDB_ONCE_INIT 0
-extern void InitOnce(port::OnceType*, void (*initializer)());
-
-// A type that holds a pointer that can be read or written atomically
-// (i.e., without word-tearing.)
-class AtomicPointer {
- private:
-  intptr_t rep_;
- public:
-  // Initialize to arbitrary value
-  AtomicPointer();
-
-  // Initialize to hold v
-  explicit AtomicPointer(void* v) : rep_(v) { }
-
-  // Read and return the stored pointer with the guarantee that no
-  // later memory access (read or write) by this thread can be
-  // reordered ahead of this read.
-  void* Acquire_Load() const;
-
-  // Set v as the stored pointer with the guarantee that no earlier
-  // memory access (read or write) by this thread can be reordered
-  // after this store.
-  void Release_Store(void* v);
-
-  // Read the stored pointer with no ordering guarantees.
-  void* NoBarrier_Load() const;
-
-  // Set va as the stored pointer with no ordering guarantees.
-  void NoBarrier_Store(void* v);
-};
-
 // ------------------ Compression -------------------
 
 // Store the snappy compression of "input[0,input_length-1]" in *output.
 // Returns false if snappy is not supported by this port.
-extern bool Snappy_Compress(const char* input, size_t input_length,
-                            std::string* output);
+bool Snappy_Compress(const char* input, size_t input_length,
+                     std::string* output);
 
 // If input[0,input_length-1] looks like a valid snappy compressed
 // buffer, store the size of the uncompressed data in *result and
 // return true.  Else return false.
-extern bool Snappy_GetUncompressedLength(const char* input, size_t length,
-                                         size_t* result);
+bool Snappy_GetUncompressedLength(const char* input, size_t length,
+                                  size_t* result);
 
 // Attempt to snappy uncompress input[0,input_length-1] into *output.
 // Returns true if successful, false if the input is invalid lightweight
@@ -119,15 +82,21 @@ extern bool Snappy_GetUncompressedLength(const char* input, size_t length,
 // REQUIRES: at least the first "n" bytes of output[] must be writable
 // where "n" is the result of a successful call to
 // Snappy_GetUncompressedLength.
-extern bool Snappy_Uncompress(const char* input_data, size_t input_length,
-                              char* output);
+bool Snappy_Uncompress(const char* input_data, size_t input_length,
+                       char* output);
 
 // ------------------ Miscellaneous -------------------
 
 // If heap profiling is not supported, returns false.
 // Else repeatedly calls (*func)(arg, data, n) and then returns true.
 // The concatenation of all "data[0,n-1]" fragments is the heap profile.
-extern bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg);
+bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg);
+
+// Extend the CRC to include the first n bytes of buf.
+//
+// Returns zero if the CRC cannot be extended using acceleration, else returns
+// the newly extended CRC value (which may also be zero).
+uint32_t AcceleratedCRC32C(uint32_t crc, const char* buf, size_t size);
 
 }  // namespace port
 }  // namespace leveldb

port/port_posix.cc

@@ -1,54 +0,0 @@
-// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file. See the AUTHORS file for names of contributors.
-
-#include "port/port_posix.h"
-
-#include <cstdlib>
-#include <stdio.h>
-#include <string.h>
-#include "util/logging.h"
-
-namespace leveldb {
-namespace port {
-
-static void PthreadCall(const char* label, int result) {
-  if (result != 0) {
-    fprintf(stderr, "pthread %s: %s\n", label, strerror(result));
-    abort();
-  }
-}
-
-Mutex::Mutex() { PthreadCall("init mutex", pthread_mutex_init(&mu_, NULL)); }
-
-Mutex::~Mutex() { PthreadCall("destroy mutex", pthread_mutex_destroy(&mu_)); }
-
-void Mutex::Lock() { PthreadCall("lock", pthread_mutex_lock(&mu_)); }
-
-void Mutex::Unlock() { PthreadCall("unlock", pthread_mutex_unlock(&mu_)); }
-
-CondVar::CondVar(Mutex* mu)
-    : mu_(mu) {
-    PthreadCall("init cv", pthread_cond_init(&cv_, NULL));
-}
-
-CondVar::~CondVar() { PthreadCall("destroy cv", pthread_cond_destroy(&cv_)); }
-
-void CondVar::Wait() {
-  PthreadCall("wait", pthread_cond_wait(&cv_, &mu_->mu_));
-}
-
-void CondVar::Signal() {
-  PthreadCall("signal", pthread_cond_signal(&cv_));
-}
-
-void CondVar::SignalAll() {
-  PthreadCall("broadcast", pthread_cond_broadcast(&cv_));
-}
-
-void InitOnce(OnceType* once, void (*initializer)()) {
-  PthreadCall("once", pthread_once(once, initializer));
-}
-
-}  // namespace port
-}  // namespace leveldb

port/port_posix.h

@@ -1,154 +0,0 @@
-// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file. See the AUTHORS file for names of contributors.
-//
-// See port_example.h for documentation for the following types/functions.
-
-#ifndef STORAGE_LEVELDB_PORT_PORT_POSIX_H_
-#define STORAGE_LEVELDB_PORT_PORT_POSIX_H_
-
-#undef PLATFORM_IS_LITTLE_ENDIAN
-#if defined(OS_MACOSX)
-  #include <machine/endian.h>
-  #if defined(__DARWIN_LITTLE_ENDIAN) && defined(__DARWIN_BYTE_ORDER)
-    #define PLATFORM_IS_LITTLE_ENDIAN \
-        (__DARWIN_BYTE_ORDER == __DARWIN_LITTLE_ENDIAN)
-  #endif
-#elif defined(OS_SOLARIS)
-  #include <sys/isa_defs.h>
-  #ifdef _LITTLE_ENDIAN
-    #define PLATFORM_IS_LITTLE_ENDIAN true
-  #else
-    #define PLATFORM_IS_LITTLE_ENDIAN false
-  #endif
-#elif defined(OS_FREEBSD) || defined(OS_OPENBSD) ||\
-      defined(OS_NETBSD) || defined(OS_DRAGONFLYBSD)
-  #include <sys/types.h>
-  #include <sys/endian.h>
-  #define PLATFORM_IS_LITTLE_ENDIAN (_BYTE_ORDER == _LITTLE_ENDIAN)
-#elif defined(OS_HPUX)
-  #define PLATFORM_IS_LITTLE_ENDIAN false
-#elif defined(OS_ANDROID)
-  // Due to a bug in the NDK x86 <sys/endian.h> definition,
-  // _BYTE_ORDER must be used instead of __BYTE_ORDER on Android.
-  // See http://code.google.com/p/android/issues/detail?id=39824
-  #include <endian.h>
-  #define PLATFORM_IS_LITTLE_ENDIAN  (_BYTE_ORDER == _LITTLE_ENDIAN)
-#else
-  #include <endian.h>
-#endif
-
-#include <pthread.h>
-#ifdef SNAPPY
-#include <snappy.h>
-#endif
-#include <stdint.h>
-#include <string>
-#include "port/atomic_pointer.h"
-
-#ifndef PLATFORM_IS_LITTLE_ENDIAN
-#define PLATFORM_IS_LITTLE_ENDIAN (__BYTE_ORDER == __LITTLE_ENDIAN)
-#endif
-
-#if defined(OS_MACOSX) || defined(OS_SOLARIS) || defined(OS_FREEBSD) ||\
-    defined(OS_NETBSD) || defined(OS_OPENBSD) || defined(OS_DRAGONFLYBSD) ||\
-    defined(OS_ANDROID) || defined(OS_HPUX) || defined(CYGWIN)
-// Use fread/fwrite/fflush on platforms without _unlocked variants
-#define fread_unlocked fread
-#define fwrite_unlocked fwrite
-#define fflush_unlocked fflush
-#endif
-
-#if defined(OS_MACOSX) || defined(OS_FREEBSD) ||\
-    defined(OS_OPENBSD) || defined(OS_DRAGONFLYBSD)
-// Use fsync() on platforms without fdatasync()
-#define fdatasync fsync
-#endif
-
-#if defined(OS_ANDROID) && __ANDROID_API__ < 9
-// fdatasync() was only introduced in API level 9 on Android. Use fsync()
-// when targetting older platforms.
-#define fdatasync fsync
-#endif
-
-namespace leveldb {
-namespace port {
-
-static const bool kLittleEndian = PLATFORM_IS_LITTLE_ENDIAN;
-#undef PLATFORM_IS_LITTLE_ENDIAN
-
-class CondVar;
-
-class Mutex {
- public:
-  Mutex();
-  ~Mutex();
-
-  void Lock();
-  void Unlock();
-  void AssertHeld() { }
-
- private:
-  friend class CondVar;
-  pthread_mutex_t mu_;
-
-  // No copying
-  Mutex(const Mutex&);
-  void operator=(const Mutex&);
-};
-
-class CondVar {
- public:
-  explicit CondVar(Mutex* mu);
-  ~CondVar();
-  void Wait();
-  void Signal();
-  void SignalAll();
- private:
-  pthread_cond_t cv_;
-  Mutex* mu_;
-};
-
-typedef pthread_once_t OnceType;
-#define LEVELDB_ONCE_INIT PTHREAD_ONCE_INIT
-extern void InitOnce(OnceType* once, void (*initializer)());
-
-inline bool Snappy_Compress(const char* input, size_t length,
-                            ::std::string* output) {
-#ifdef SNAPPY
-  output->resize(snappy::MaxCompressedLength(length));
-  size_t outlen;
-  snappy::RawCompress(input, length, &(*output)[0], &outlen);
-  output->resize(outlen);
-  return true;
-#endif
-
-  return false;
-}
-
-inline bool Snappy_GetUncompressedLength(const char* input, size_t length,
-                                         size_t* result) {
-#ifdef SNAPPY
-  return snappy::GetUncompressedLength(input, length, result);
-#else
-  return false;
-#endif
-}
-
-inline bool Snappy_Uncompress(const char* input, size_t length,
-                              char* output) {
-#ifdef SNAPPY
-  return snappy::RawUncompress(input, length, output);
-#else
-  return false;
-#endif
-}
-
-inline bool GetHeapProfile(void (*func)(void*, const char*, int), void* arg) {
-  return false;
-}
-
-} // namespace port
-} // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_PORT_PORT_POSIX_H_

port/port_stdcxx.h

@@ -0,0 +1,143 @@
+// Copyright (c) 2018 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#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 <cstddef>
+#include <cstdint>
+#include <condition_variable>  // NOLINT
+#include <mutex>               // NOLINT
+#include <string>
+
+#include "port/thread_annotations.h"
+
+namespace leveldb {
+namespace port {
+
+static const bool kLittleEndian = !LEVELDB_IS_BIG_ENDIAN;
+
+class CondVar;
+
+// Thinly wraps std::mutex.
+class LOCKABLE Mutex {
+ public:
+  Mutex() = default;
+  ~Mutex() = default;
+
+  Mutex(const Mutex&) = delete;
+  Mutex& operator=(const Mutex&) = delete;
+
+  void Lock() EXCLUSIVE_LOCK_FUNCTION() { mu_.lock(); }
+  void Unlock() UNLOCK_FUNCTION() { mu_.unlock(); }
+  void AssertHeld() ASSERT_EXCLUSIVE_LOCK() { }
+
+ private:
+  friend class CondVar;
+  std::mutex mu_;
+};
+
+// Thinly wraps std::condition_variable.
+class CondVar {
+ public:
+  explicit CondVar(Mutex* mu) : mu_(mu) { assert(mu != nullptr); }
+  ~CondVar() = default;
+
+  CondVar(const CondVar&) = delete;
+  CondVar& operator=(const CondVar&) = delete;
+
+  void Wait() {
+    std::unique_lock<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_

port/thread_annotations.h

@@ -5,56 +5,107 @@
 #ifndef STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_
 #define STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_
 
-// Some environments provide custom macros to aid in static thread-safety
-// analysis.  Provide empty definitions of such macros unless they are already
-// defined.
+// Use Clang's thread safety analysis annotations when available. In other
+// environments, the macros receive empty definitions.
+// Usage documentation: https://clang.llvm.org/docs/ThreadSafetyAnalysis.html
+
+#if !defined(THREAD_ANNOTATION_ATTRIBUTE__)
+
+#if defined(__clang__)
+
+#define THREAD_ANNOTATION_ATTRIBUTE__(x)   __attribute__((x))
+#else
+#define THREAD_ANNOTATION_ATTRIBUTE__(x)   // no-op
+#endif
+
+#endif  // !defined(THREAD_ANNOTATION_ATTRIBUTE__)
+
+#ifndef GUARDED_BY
+#define GUARDED_BY(x) THREAD_ANNOTATION_ATTRIBUTE__(guarded_by(x))
+#endif
+
+#ifndef PT_GUARDED_BY
+#define PT_GUARDED_BY(x) THREAD_ANNOTATION_ATTRIBUTE__(pt_guarded_by(x))
+#endif
+
+#ifndef ACQUIRED_AFTER
+#define ACQUIRED_AFTER(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(acquired_after(__VA_ARGS__))
+#endif
+
+#ifndef ACQUIRED_BEFORE
+#define ACQUIRED_BEFORE(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(acquired_before(__VA_ARGS__))
+#endif
 
 #ifndef EXCLUSIVE_LOCKS_REQUIRED
-#define EXCLUSIVE_LOCKS_REQUIRED(...)
+#define EXCLUSIVE_LOCKS_REQUIRED(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(exclusive_locks_required(__VA_ARGS__))
 #endif
 
 #ifndef SHARED_LOCKS_REQUIRED
-#define SHARED_LOCKS_REQUIRED(...)
+#define SHARED_LOCKS_REQUIRED(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(shared_locks_required(__VA_ARGS__))
 #endif
 
 #ifndef LOCKS_EXCLUDED
-#define LOCKS_EXCLUDED(...)
+#define LOCKS_EXCLUDED(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(locks_excluded(__VA_ARGS__))
 #endif
 
 #ifndef LOCK_RETURNED
-#define LOCK_RETURNED(x)
+#define LOCK_RETURNED(x) \
+  THREAD_ANNOTATION_ATTRIBUTE__(lock_returned(x))
 #endif
 
 #ifndef LOCKABLE
-#define LOCKABLE
+#define LOCKABLE \
+  THREAD_ANNOTATION_ATTRIBUTE__(lockable)
 #endif
 
 #ifndef SCOPED_LOCKABLE
-#define SCOPED_LOCKABLE
+#define SCOPED_LOCKABLE \
+  THREAD_ANNOTATION_ATTRIBUTE__(scoped_lockable)
 #endif
 
 #ifndef EXCLUSIVE_LOCK_FUNCTION
-#define EXCLUSIVE_LOCK_FUNCTION(...)
+#define EXCLUSIVE_LOCK_FUNCTION(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(exclusive_lock_function(__VA_ARGS__))
 #endif
 
 #ifndef SHARED_LOCK_FUNCTION
-#define SHARED_LOCK_FUNCTION(...)
+#define SHARED_LOCK_FUNCTION(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(shared_lock_function(__VA_ARGS__))
 #endif
 
 #ifndef EXCLUSIVE_TRYLOCK_FUNCTION
-#define EXCLUSIVE_TRYLOCK_FUNCTION(...)
+#define EXCLUSIVE_TRYLOCK_FUNCTION(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(exclusive_trylock_function(__VA_ARGS__))
 #endif
 
 #ifndef SHARED_TRYLOCK_FUNCTION
-#define SHARED_TRYLOCK_FUNCTION(...)
+#define SHARED_TRYLOCK_FUNCTION(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(shared_trylock_function(__VA_ARGS__))
 #endif
 
 #ifndef UNLOCK_FUNCTION
-#define UNLOCK_FUNCTION(...)
+#define UNLOCK_FUNCTION(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(unlock_function(__VA_ARGS__))
 #endif
 
 #ifndef NO_THREAD_SAFETY_ANALYSIS
-#define NO_THREAD_SAFETY_ANALYSIS
+#define NO_THREAD_SAFETY_ANALYSIS \
+  THREAD_ANNOTATION_ATTRIBUTE__(no_thread_safety_analysis)
+#endif
+
+#ifndef ASSERT_EXCLUSIVE_LOCK
+#define ASSERT_EXCLUSIVE_LOCK(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(assert_exclusive_lock(__VA_ARGS__))
+#endif
+
+#ifndef ASSERT_SHARED_LOCK
+#define ASSERT_SHARED_LOCK(...) \
+  THREAD_ANNOTATION_ATTRIBUTE__(assert_shared_lock(__VA_ARGS__))
 #endif
 
 #endif  // STORAGE_LEVELDB_PORT_THREAD_ANNOTATIONS_H_

port/win/stdint.h

@@ -1,24 +0,0 @@
-// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file. See the AUTHORS file for names of contributors.
-
-// MSVC didn't ship with this file until the 2010 version.
-
-#ifndef STORAGE_LEVELDB_PORT_WIN_STDINT_H_
-#define STORAGE_LEVELDB_PORT_WIN_STDINT_H_
-
-#if !defined(_MSC_VER)
-#error This file should only be included when compiling with MSVC.
-#endif
-
-// Define C99 equivalent types.
-typedef signed char           int8_t;
-typedef signed short          int16_t;
-typedef signed int            int32_t;
-typedef signed long long      int64_t;
-typedef unsigned char         uint8_t;
-typedef unsigned short        uint16_t;
-typedef unsigned int          uint32_t;
-typedef unsigned long long    uint64_t;
-
-#endif  // STORAGE_LEVELDB_PORT_WIN_STDINT_H_

table/block.cc

@@ -48,13 +48,13 @@ Block::~Block() {
 // and the length of the value in "*shared", "*non_shared", and
 // "*value_length", respectively.  Will not dereference past "limit".
 //
-// If any errors are detected, returns NULL.  Otherwise, returns a
+// If any errors are detected, returns nullptr.  Otherwise, returns a
 // pointer to the key delta (just past the three decoded values).
 static inline const char* DecodeEntry(const char* p, const char* limit,
                                       uint32_t* shared,
                                       uint32_t* non_shared,
                                       uint32_t* value_length) {
-  if (limit - p < 3) return NULL;
+  if (limit - p < 3) return nullptr;
   *shared = reinterpret_cast<const unsigned char*>(p)[0];
   *non_shared = reinterpret_cast<const unsigned char*>(p)[1];
   *value_length = reinterpret_cast<const unsigned char*>(p)[2];
@@ -62,13 +62,13 @@ static inline const char* DecodeEntry(const char* p, const char* limit,
     // Fast path: all three values are encoded in one byte each
     p += 3;
   } else {
-    if ((p = GetVarint32Ptr(p, limit, shared)) == NULL) return NULL;
-    if ((p = GetVarint32Ptr(p, limit, non_shared)) == NULL) return NULL;
-    if ((p = GetVarint32Ptr(p, limit, value_length)) == NULL) return NULL;
+    if ((p = GetVarint32Ptr(p, limit, shared)) == nullptr) return nullptr;
+    if ((p = GetVarint32Ptr(p, limit, non_shared)) == nullptr) return nullptr;
+    if ((p = GetVarint32Ptr(p, limit, value_length)) == nullptr) return nullptr;
   }
 
   if (static_cast<uint32_t>(limit - p) < (*non_shared + *value_length)) {
-    return NULL;
+    return nullptr;
   }
   return p;
 }
@@ -174,7 +174,7 @@ class Block::Iter : public Iterator {
       const char* key_ptr = DecodeEntry(data_ + region_offset,
                                         data_ + restarts_,
                                         &shared, &non_shared, &value_length);
-      if (key_ptr == NULL || (shared != 0)) {
+      if (key_ptr == nullptr || (shared != 0)) {
         CorruptionError();
         return;
       }
@@ -237,7 +237,7 @@ class Block::Iter : public Iterator {
     // Decode next entry
     uint32_t shared, non_shared, value_length;
     p = DecodeEntry(p, limit, &shared, &non_shared, &value_length);
-    if (p == NULL || key_.size() < shared) {
+    if (p == nullptr || key_.size() < shared) {
       CorruptionError();
       return false;
     } else {

table/filter_block.cc

@@ -9,7 +9,7 @@
 
 namespace leveldb {
 
-// See doc/table_format.txt for an explanation of the filter block format.
+// See doc/table_format.md for an explanation of the filter block format.
 
 // Generate new filter every 2KB of data
 static const size_t kFilterBaseLg = 11;
@@ -68,7 +68,7 @@ void FilterBlockBuilder::GenerateFilter() {
 
   // Generate filter for current set of keys and append to result_.
   filter_offsets_.push_back(result_.size());
-  policy_->CreateFilter(&tmp_keys_[0], num_keys, &result_);
+  policy_->CreateFilter(&tmp_keys_[0], static_cast<int>(num_keys), &result_);
 
   tmp_keys_.clear();
   keys_.clear();
@@ -78,8 +78,8 @@ void FilterBlockBuilder::GenerateFilter() {
 FilterBlockReader::FilterBlockReader(const FilterPolicy* policy,
                                      const Slice& contents)
     : policy_(policy),
-      data_(NULL),
-      offset_(NULL),
+      data_(nullptr),
+      offset_(nullptr),
       num_(0),
       base_lg_(0) {
   size_t n = contents.size();
@@ -97,7 +97,7 @@ bool FilterBlockReader::KeyMayMatch(uint64_t block_offset, const Slice& key) {
   if (index < num_) {
     uint32_t start = DecodeFixed32(offset_ + index*4);
     uint32_t limit = DecodeFixed32(offset_ + index*4 + 4);
-    if (start <= limit && limit <= (offset_ - data_)) {
+    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) {

table/format.cc

@@ -30,15 +30,14 @@ Status BlockHandle::DecodeFrom(Slice* input) {
 }
 
 void Footer::EncodeTo(std::string* dst) const {
-#ifndef NDEBUG
   const size_t original_size = dst->size();
-#endif
   metaindex_handle_.EncodeTo(dst);
   index_handle_.EncodeTo(dst);
   dst->resize(2 * BlockHandle::kMaxEncodedLength);  // Padding
   PutFixed32(dst, static_cast<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) {

table/format.h

@@ -91,10 +91,10 @@ struct BlockContents {
 
 // Read the block identified by "handle" from "file".  On failure
 // return non-OK.  On success fill *result and return OK.
-extern Status ReadBlock(RandomAccessFile* file,
-                        const ReadOptions& options,
-                        const BlockHandle& handle,
-                        BlockContents* result);
+Status ReadBlock(RandomAccessFile* file,
+                 const ReadOptions& options,
+                 const BlockHandle& handle,
+                 BlockContents* result);
 
 // Implementation details follow.  Clients should ignore,
 

table/iterator.cc

@@ -7,54 +7,59 @@
 namespace leveldb {
 
 Iterator::Iterator() {
-  cleanup_.function = NULL;
-  cleanup_.next = NULL;
+  cleanup_head_.function = nullptr;
+  cleanup_head_.next = nullptr;
 }
 
 Iterator::~Iterator() {
-  if (cleanup_.function != NULL) {
-    (*cleanup_.function)(cleanup_.arg1, cleanup_.arg2);
-    for (Cleanup* c = cleanup_.next; c != NULL; ) {
-      (*c->function)(c->arg1, c->arg2);
-      Cleanup* next = c->next;
-      delete c;
-      c = next;
+  if (!cleanup_head_.IsEmpty()) {
+    cleanup_head_.Run();
+    for (CleanupNode* node = cleanup_head_.next; node != nullptr; ) {
+      node->Run();
+      CleanupNode* next_node = node->next;
+      delete node;
+      node = next_node;
     }
   }
 }
 
 void Iterator::RegisterCleanup(CleanupFunction func, void* arg1, void* arg2) {
-  assert(func != NULL);
-  Cleanup* c;
-  if (cleanup_.function == NULL) {
-    c = &cleanup_;
+  assert(func != nullptr);
+  CleanupNode* node;
+  if (cleanup_head_.IsEmpty()) {
+    node = &cleanup_head_;
   } else {
-    c = new Cleanup;
-    c->next = cleanup_.next;
-    cleanup_.next = c;
+    node = new CleanupNode();
+    node->next = cleanup_head_.next;
+    cleanup_head_.next = node;
   }
-  c->function = func;
-  c->arg1 = arg1;
-  c->arg2 = arg2;
+  node->function = func;
+  node->arg1 = arg1;
+  node->arg2 = arg2;
 }
 
 namespace {
+
 class EmptyIterator : public Iterator {
  public:
   EmptyIterator(const Status& s) : status_(s) { }
-  virtual bool Valid() const { return false; }
-  virtual void Seek(const Slice& target) { }
-  virtual void SeekToFirst() { }
-  virtual void SeekToLast() { }
-  virtual void Next() { assert(false); }
-  virtual void Prev() { assert(false); }
-  Slice key() const { assert(false); return Slice(); }
-  Slice value() const { assert(false); return Slice(); }
-  virtual Status status() const { return status_; }
+  ~EmptyIterator() override = default;
+
+  bool Valid() const override { return false; }
+  void Seek(const Slice& target) override { }
+  void SeekToFirst() override { }
+  void SeekToLast() override { }
+  void Next() override { assert(false); }
+  void Prev() override { assert(false); }
+  Slice key() const override { assert(false); return Slice(); }
+  Slice value() const override { assert(false); return Slice(); }
+  Status status() const override { return status_; }
+
  private:
   Status status_;
 };
-}  // namespace
+
+}  // anonymous namespace
 
 Iterator* NewEmptyIterator() {
   return new EmptyIterator(Status::OK());

table/iterator_wrapper.h

@@ -5,6 +5,9 @@
 #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
@@ -13,8 +16,8 @@ namespace leveldb {
 // cache locality.
 class IteratorWrapper {
  public:
-  IteratorWrapper(): iter_(NULL), valid_(false) { }
-  explicit IteratorWrapper(Iterator* iter): iter_(NULL) {
+  IteratorWrapper(): iter_(nullptr), valid_(false) { }
+  explicit IteratorWrapper(Iterator* iter): iter_(nullptr) {
     Set(iter);
   }
   ~IteratorWrapper() { delete iter_; }
@@ -25,7 +28,7 @@ class IteratorWrapper {
   void Set(Iterator* iter) {
     delete iter_;
     iter_ = iter;
-    if (iter_ == NULL) {
+    if (iter_ == nullptr) {
       valid_ = false;
     } else {
       Update();
@@ -37,7 +40,7 @@ class IteratorWrapper {
   bool Valid() const        { return valid_; }
   Slice key() const         { assert(Valid()); return key_; }
   Slice value() const       { assert(Valid()); return iter_->value(); }
-  // Methods below require iter() != NULL
+  // 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(); }

table/merger.cc

@@ -17,7 +17,7 @@ class MergingIterator : public Iterator {
       : comparator_(comparator),
         children_(new IteratorWrapper[n]),
         n_(n),
-        current_(NULL),
+        current_(nullptr),
         direction_(kForward) {
     for (int i = 0; i < n; i++) {
       children_[i].Set(children[i]);
@@ -29,7 +29,7 @@ class MergingIterator : public Iterator {
   }
 
   virtual bool Valid() const {
-    return (current_ != NULL);
+    return (current_ != nullptr);
   }
 
   virtual void SeekToFirst() {
@@ -153,11 +153,11 @@ class MergingIterator : public Iterator {
 };
 
 void MergingIterator::FindSmallest() {
-  IteratorWrapper* smallest = NULL;
+  IteratorWrapper* smallest = nullptr;
   for (int i = 0; i < n_; i++) {
     IteratorWrapper* child = &children_[i];
     if (child->Valid()) {
-      if (smallest == NULL) {
+      if (smallest == nullptr) {
         smallest = child;
       } else if (comparator_->Compare(child->key(), smallest->key()) < 0) {
         smallest = child;
@@ -168,11 +168,11 @@ void MergingIterator::FindSmallest() {
 }
 
 void MergingIterator::FindLargest() {
-  IteratorWrapper* largest = NULL;
+  IteratorWrapper* largest = nullptr;
   for (int i = n_-1; i >= 0; i--) {
     IteratorWrapper* child = &children_[i];
     if (child->Valid()) {
-      if (largest == NULL) {
+      if (largest == nullptr) {
         largest = child;
       } else if (comparator_->Compare(child->key(), largest->key()) > 0) {
         largest = child;

table/merger.h

@@ -18,7 +18,7 @@ class Iterator;
 // key is present in K child iterators, it will be yielded K times.
 //
 // REQUIRES: n >= 0
-extern Iterator* NewMergingIterator(
+Iterator* NewMergingIterator(
     const Comparator* comparator, Iterator** children, int n);
 
 }  // namespace leveldb

table/table.cc

@@ -39,7 +39,7 @@ Status Table::Open(const Options& options,
                    RandomAccessFile* file,
                    uint64_t size,
                    Table** table) {
-  *table = NULL;
+  *table = nullptr;
   if (size < Footer::kEncodedLength) {
     return Status::Corruption("file is too short to be an sstable");
   }
@@ -55,41 +55,36 @@ Status Table::Open(const Options& options,
   if (!s.ok()) return s;
 
   // Read the index block
-  BlockContents contents;
-  Block* index_block = NULL;
+  BlockContents index_block_contents;
   if (s.ok()) {
     ReadOptions opt;
     if (options.paranoid_checks) {
       opt.verify_checksums = true;
     }
-    s = ReadBlock(file, opt, footer.index_handle(), &contents);
-    if (s.ok()) {
-      index_block = new Block(contents);
-    }
+    s = ReadBlock(file, opt, footer.index_handle(), &index_block_contents);
   }
 
   if (s.ok()) {
     // We've successfully read the footer and the index block: we're
     // ready to serve requests.
+    Block* index_block = new Block(index_block_contents);
     Rep* rep = new Table::Rep;
     rep->options = options;
     rep->file = file;
     rep->metaindex_handle = footer.metaindex_handle();
     rep->index_block = index_block;
     rep->cache_id = (options.block_cache ? options.block_cache->NewId() : 0);
-    rep->filter_data = NULL;
-    rep->filter = NULL;
+    rep->filter_data = nullptr;
+    rep->filter = nullptr;
     *table = new Table(rep);
     (*table)->ReadMeta(footer);
-  } else {
-    if (index_block) delete index_block;
   }
 
   return s;
 }
 
 void Table::ReadMeta(const Footer& footer) {
-  if (rep_->options.filter_policy == NULL) {
+  if (rep_->options.filter_policy == nullptr) {
     return;  // Do not need any metadata
   }
 
@@ -166,8 +161,8 @@ Iterator* Table::BlockReader(void* arg,
                              const Slice& index_value) {
   Table* table = reinterpret_cast<Table*>(arg);
   Cache* block_cache = table->rep_->options.block_cache;
-  Block* block = NULL;
-  Cache::Handle* cache_handle = NULL;
+  Block* block = nullptr;
+  Cache::Handle* cache_handle = nullptr;
 
   BlockHandle handle;
   Slice input = index_value;
@@ -177,13 +172,13 @@ Iterator* Table::BlockReader(void* arg,
 
   if (s.ok()) {
     BlockContents contents;
-    if (block_cache != NULL) {
+    if (block_cache != nullptr) {
       char cache_key_buffer[16];
       EncodeFixed64(cache_key_buffer, table->rep_->cache_id);
       EncodeFixed64(cache_key_buffer+8, handle.offset());
       Slice key(cache_key_buffer, sizeof(cache_key_buffer));
       cache_handle = block_cache->Lookup(key);
-      if (cache_handle != NULL) {
+      if (cache_handle != nullptr) {
         block = reinterpret_cast<Block*>(block_cache->Value(cache_handle));
       } else {
         s = ReadBlock(table->rep_->file, options, handle, &contents);
@@ -204,10 +199,10 @@ Iterator* Table::BlockReader(void* arg,
   }
 
   Iterator* iter;
-  if (block != NULL) {
+  if (block != nullptr) {
     iter = block->NewIterator(table->rep_->options.comparator);
-    if (cache_handle == NULL) {
-      iter->RegisterCleanup(&DeleteBlock, block, NULL);
+    if (cache_handle == nullptr) {
+      iter->RegisterCleanup(&DeleteBlock, block, nullptr);
     } else {
       iter->RegisterCleanup(&ReleaseBlock, block_cache, cache_handle);
     }
@@ -233,7 +228,7 @@ Status Table::InternalGet(const ReadOptions& options, const Slice& k,
     Slice handle_value = iiter->value();
     FilterBlockReader* filter = rep_->filter;
     BlockHandle handle;
-    if (filter != NULL &&
+    if (filter != nullptr &&
         handle.DecodeFrom(&handle_value).ok() &&
         !filter->KeyMayMatch(handle.offset(), k)) {
       // Not found

table/table_builder.cc

@@ -53,7 +53,7 @@ struct TableBuilder::Rep {
         index_block(&index_block_options),
         num_entries(0),
         closed(false),
-        filter_block(opt.filter_policy == NULL ? NULL
+        filter_block(opt.filter_policy == nullptr ? nullptr
                      : new FilterBlockBuilder(opt.filter_policy)),
         pending_index_entry(false) {
     index_block_options.block_restart_interval = 1;
@@ -62,7 +62,7 @@ struct TableBuilder::Rep {
 
 TableBuilder::TableBuilder(const Options& options, WritableFile* file)
     : rep_(new Rep(options, file)) {
-  if (rep_->filter_block != NULL) {
+  if (rep_->filter_block != nullptr) {
     rep_->filter_block->StartBlock(0);
   }
 }
@@ -106,7 +106,7 @@ void TableBuilder::Add(const Slice& key, const Slice& value) {
     r->pending_index_entry = false;
   }
 
-  if (r->filter_block != NULL) {
+  if (r->filter_block != nullptr) {
     r->filter_block->AddKey(key);
   }
 
@@ -131,7 +131,7 @@ void TableBuilder::Flush() {
     r->pending_index_entry = true;
     r->status = r->file->Flush();
   }
-  if (r->filter_block != NULL) {
+  if (r->filter_block != nullptr) {
     r->filter_block->StartBlock(r->offset);
   }
 }
@@ -205,7 +205,7 @@ Status TableBuilder::Finish() {
   BlockHandle filter_block_handle, metaindex_block_handle, index_block_handle;
 
   // Write filter block
-  if (ok() && r->filter_block != NULL) {
+  if (ok() && r->filter_block != nullptr) {
     WriteRawBlock(r->filter_block->Finish(), kNoCompression,
                   &filter_block_handle);
   }
@@ -213,7 +213,7 @@ Status TableBuilder::Finish() {
   // Write metaindex block
   if (ok()) {
     BlockBuilder meta_index_block(&r->options);
-    if (r->filter_block != NULL) {
+    if (r->filter_block != nullptr) {
       // Add mapping from "filter.Name" to location of filter data
       std::string key = "filter.";
       key.append(r->options.filter_policy->Name());